Aku dapat materi ini sewaktu berkunjung (sebenarnya sih lebih k jalan-jalan daripada berkunjung ;) ). Jadi jika mau ngopi mohon sekiranya dipergunakan dengan baik dan benar ya. Aku g ingin kamu mendapat karma gara-gara daku yang hina ini..hehehe
Dan akhirnya monggo dinikmati dan sugeng berkreasi untuk kemajuan las di Indonesia..
Ooiya ada yang lupa,silahkan di perbanyak jika diperlukan..Absolutelly Free..
Vivat LAS..!!!
INDEX
Acicular ferrite, 66, 74, 88, 90, 233–239,
405
Alternating grain orientation, 291
Aluminum alloys,
Al-Li alloys, 80, 95, 188, 195, 354, 371,
374
designation, 355
filler metals, 183, 202, 251, 255, 285,
286, 291, 300, 303, 326, 328, 330, 332,
338, 339, 360, 362, 374
heat treatable, 353, 355, 359, 365–366,
369, 379
heat-affected zone softening, 352, 354,
373
partially melted zone cracking, 321,
334
partially melted zone liquation,
303–317, 330
porosity, 66, 80, 81, 95, 251, 259, 262,
353–354, 394
solidification cracking, 271, 273, 278,
281, 284–286, 288, 291, 294, 299, 300,
322, 330
typical welding problems, 354
Angular distortion, 5
Annealing, 343, 345, 346, 349, 369, 373,
374, 388, 408, 442, 447
Arc,
blow, 23, 34
efficiency, 37–41, 43, 46, 56, 63, 64, 319,
352, 373
fluid flow, 99–102
length, 113
oscillation, 188, 192–195, 210–212,
291–293, 300, 332, 338, 354
pulsation, 188, 193, 213, 300
stabilizers, 12
vibration, 192
Arc welding processes,
electroslag welding, 3, 6, 7, 24–27, 56,
393, 394, 406
flux-cored arc welding, 3, 6, 7, 22, 23,
36, 73, 79, 86, 88, 91
gas metal arc welding, 6, 19, 20, 32, 79
gas tungsten arc welding, 6, 13, 18, 28,
56, 67, 78
plasma arc welding, 3, 6, 17–19, 40, 42,
74, 80, 81, 354, 366
shielded metal arc welding, 3, 6, 11, 12,
66, 67, 75, 76
submerged arc welding, 3, 6, 22, 24, 90,
189, 238, 295
Argon, 14–16, 19, 31, 32, 70, 73, 193, 237
Artificial aging,
Al alloys, 353, 357, 359, 360, 363–367,
373
Ni-base alloys, 375, 376, 380
Austenite,
high-carbon, 401
retained, 399
stainless steel, 174
Autogenous welding, 16, 170, 285
Auto-tempered martensite, 406
Axial grains, 176, 178
Banding, 160, 163, 249–251
Basicity index, 85–89, 238
Bead shape, 294, 295
Bead tempering, 407, 408, 416
Bessel function, 50
Boundary layer, 153
Buoyancy force, 104, 107
Buttering, 289
Carbide-forming elements, 394
Carbide precipitation,
in austenitic stainless steels, 435–437,
440–444, 446
in ferritic steel, 421, 422
in Ni-base alloys, 378
Carbon equivalent, 394, 416, 417
Cell spacing, 165, 204–209, 213, 327 (see
also dendrite arm spacing)
Cellular solidification, 156, 160
Circular patch test, 323, 330, 386
Coarsening, 165
455
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
456 INDEX
Columnar dendrites, 156, 157, 159
Competitive growth, 174, 176, 204
Convection (see weld pool convection
and arc fluid flow)
Constitution diagrams, 223–226
Constitutional liquation, 306, 309–311,
330, 333
Constitutional supercooling, 156–160,
186, 187, 199, 200, 247, 318
Contact angle, 170, 171
Continuous cooling transformation
diagrams, 232, 236, 393, 402, 404,
406, 407
Contraction stresses, 284, 424
Cooling rate, 164, 204, 207, 212, 226–231,
249, 268, 274, 278, 291, 318, 325, 350,
351, 360, 398, 402–407, 425, 432, 437,
441, 442, 448
Crack susceptibility C-curves,
Ni-base alloys, 387–390
ferritic steels, 422
Cracking,
cold, 411, 423
delayed, 411
hot, 321, 353, 354, 376, 379, 384, 432
hydrogen, 12, 66, 75, 255, 294, 321, 328,
329, 393, 402, 406–408, 410–418, 428,
432 (see hydrogen cracking)
fatigue, 135, 139
intergranular, 263, 328, 419, 422, 430
liquation, 324–327, 335, 336
reheat, 376, 385, 390, 394, 418–422, 430
(see reheat cracking)
solidification, 66, 71, 170, 188, 192, 195,
212, 216, 243, 259, 263–300, 322, 330,
338, 392, 432 (see solidification
cracking)
strain-age, 384, 385, 392
stress corrosion, 125, 255, 445, 446
toe, 135, 136, 413
underbead, 412, 413, 425, 426, 450 (see
also lamellar tearing)
Critical transformation temperatures,
393, 394, 398
Current density distribution, 99, 101–103
Degree of restraint, 284, 332
DeLong diagram, 224
Dendrite arm spacing, 164, 165, 204–213,
274
Dendrite fragmentation, 180, 181, 189,
192
Dendritic solidification, 156, 159
Dendrite tip undercooling, 230, 248
Deoxidizer, 12, 66, 394
Deposition rate, 26
Diffusion of hydrogen, 76, 411, 412
Dihedral angle, 281, 282
Dilution ratio, 257, 285, 286, 288, 289,
330
Direct-current electrode positive, 15, 16,
19, 23
Direct-current electrode negative, 14, 17,
19, 259
Directional solidification, 147, 314
Dissimilar metal welds, 29, 33, 223, 252,
255, 257, 259
Distortion in weldments, 4, 11, 24, 25, 29,
32, 126–130, 294, 367, 389, 439
Ductile-brittle transition temperature,
406
Ductility curve, 277
Easy-growth directions, 175
Electrodes, 408, 423, 424
bias, 27
flux-core arc welding, 22–24
gas-metal arc welding, 19, 21–22,
41–43, 53, 65, 80
gas-tungsten arc welding, 15, 16, 41–43,
104
high hydrogen, 415
low carbon, 288
low hydrogen, 394, 402, 407, 409, 415,
432
plasma arc welding, 17, 18, 74
submerged arc welding, 66, 91, 92
shielded metal arc welding, 75, 76, 78,
84, 295, 296, 401, 417
tip angle, 45–47, 97–100, 102
Electrode coverings, 11–13, 22, 66, 75, 78,
84, 415
Electromagnetic pool stirring, 191
Electron beam welding, 3, 5, 6, 27–29, 43,
173, 207, 208, 227, 229, 279, 295, 332,
351, 354
Electroslag welding, 3, 6, 7, 24–27, 56,
394, 406
Epitaxial growth, 170–174, 184, 203 (see
also nonepitaxial growth)
Equiaxed dendrites, 157 (see also
nondendritic equiaxed zone)
Equiaxed grains, 181–186, 191, 193
Equilibrium partition ratio or
equilibrium segregation ratio, 145,
146
INDEX 457
Evaporation from weld pool, 28, 82, 91,
114, 115, 117
Fatigue, 131–140
beach marks, 132
extrusions, 131
intrusions, 131
joint design, 133
stress raisers, 134, 135
remedies, 135–137
Ferrite,
acicular, 66, 74, 88, 90, 233–239, 405
delta (d), 174, 216–223, 231, 232,
244–247, 279–281, 291, 295, 296, 431,
448, 449
grain boundary, 232, 233, 237, 239
side-plate, 233, 239, 398
Widmanstatten, 232, 233, 398
Ferrite content,
effect of cooling rate, 226–231
effect of multipass welding, 259
effect of nitrogen, 66, 71, 224
effect of reheating, 231
example of calculation, 290
prediction, 223–226
solidification cracking, 279, 289, 291
Ferrite morphology, 218–221, 259
Fluid flow (see weld pool convection and
arc fluid flow)
Flux core arc welding, 3, 6, 7, 22, 23, 36,
73, 79, 86, 88, 91
Fluxes, 22, 23, 67, 82–88, 90, 91, 94, 97,
116, 117
FNN-1999, 226
Free energy of nucleation, 170
Free energy of reactions, 68
Freezing range or solidification
temperature range, 158, 159,
268–271
Friction stir welding, 370
Gas metal arc welding, 6, 19, 20, 32, 79
Gas-metal reactions, 68–82
hydrogen-metal, 68, 75
nitrogen-metal, 68, 71
oxygen-metal, 68, 73
Sievert’s law, 68
Gas tungsten arc welding, 6, 13, 18, 28,
56, 67, 78
active flux, 116, 117
Gas welding, 3, 6, 7–11, 74
Gaussian distribution, 47, 57, 100, 101
Ghost grain boundary, 310, 311
Gleeble (thermal simulator), 58, 59, 184,
323, 419, 421
GP zone, 354–364
Grain boundary ferrite, 232, 233, 237, 239
Grain boundary liquid, 282
Grain boundary liquation, 303, 309, 313,
321, 325, 327, 332, 336
Grain boundary migration, 310, 314
Grain boundary segregation, 315, 316
Grain detachment, 180, 181
Grain growth, 236, 310, 343–352,
394–396, 405, 432, 448, 449
Grain growth inhibitors, 405
Grain refining,
in fusion zone, 170, 180, 187–193, 197,
291, 292, 394, 398, 402, 407
in heat-affected zone, 394, 397–399,
402, 403, 407
Grain size, 60, 189, 192, 235–239, 283,
333–335, 349, 398, 402, 405, 448
Grain structure, 170–195
effect of welding parameters, 174
control of, 187, 291
Growth rate, 166, 200, 201
Heat-affected zone softening,
Al alloys, 354, 373
Low alloy steels, 410
Ni-base alloys, 381–383
work-hardened materials, 343–351
Heat flow, 37–60
Adams equations, 51, 52
computer simulation, 54, 58
cooling rate, 55, 57
effect of preheating, 57
effect of welding parameters, 53
effect of weldment thickness, 57
Rosenthal’s equations, 48–51
Heat input, 4, 5, 41, 42, 332, 350, 351, 389
Heat source,
efficiency, 37–43
power-density distribution, 45, 57, 58,
107
current-density distribution, 47
Heat treatable alloy steels, 407
Heat treatable aluminum alloys, 353–371
Al-Cu-Mg, 359
Al-Mg-Si, 359
Al-Zn-Mg, 367
Heat treating of steels, 395
Helium, 14–16, 19, 31, 32
Heiple’s model, 109
458 INDEX
Heterogeneous nucleation, 173, 180–199,
292
Heterogeneous nuclei, 181–184, 190, 234,
235
High-energy density welding processes, 3
electron beam welding, 27
laser beam welding, 29
Hot cracking in partially melted zone,
321–336 (see also liquation
cracking)
Hot ductility test, 323, 324
Houldcroft test, 264, 265
HY-80 steel, 207, 212, 255, 328, 329, 406
Hydrogen cracking, 12, 66, 75
in martensitic stainless steels, 432, 450
in steels, 255, 328–329, 402, 406–408,
410–418
methods of reduction, 415
requirements, 411
test methods, 414
Hydrogen level,
effect on welds, 66
free energy of reaction, 68
solubility in weld pool, 70
measurement of, 76–78
methods of reduction, 78–82
Implant test, 414
Inclusions, 66, 250, 251
fatigue initiation, 131
fracture initiation, 88
lamellar tearing, 422, 423, 427
liquation, 307–309
nitride, 72
nucleation site for acicular ferrite, 74,
233–237
oxide, 73, 88–90, 237
tungsten, 16
Inoculation, 188–190
Intergranular corrosion, 433, 436, 440,
442, 444, 447
Intergranular cracking,
hot cracking in partially melted zone
or liquation cracking, 321–327
hydrogen cracking, 328
postweld heat-treatment cracking or
reheat cracking, 385
reheat cracking, 418–422
solidification cracking, 263, 264
Interpass temperature, 57, 255, 256, 354,
369, 402, 405–409, 415, 416
Ionization potential, 15, 16
Iron nitride, 72
Isothermal precipitation curves, 436, 439
Isothermal transformation diagrams,
410
Joint design, 7, 8, 251
Keyhole, 4, 17, 18, 27, 28, 43, 74, 80, 81,
127, 354
Knife-line attack, 432, 440–444
Lamellar tearing, 394, 422–425, 427
Laser,
CO2 laser, 30, 31
diode laser, 31
YAG laser, 30, 108
Laser assisted gas metal arc welding, 32
Laser beam welding, 3, 6, 29, 30, 31, 36,
37, 60, 110, 128, 366
Laves, 276, 310
Lehigh cantilever test, 424
Lehigh restraint test, 414, 415,
Liquation, 303–336
constitutional, 306, 309–311, 319, 320,
330, 333
cracking, 321–327, 330–336 (see also
hot cracking in partially melted
zone)
mechanisms, 304–314
temperature, 333
Liquidus surface, 217
Lorentz force, 104, 106, 107
Lorentz force field, 99, 101
Low hydrogen electrodes, 12, 75, 409
Macrosegregation, 255–259
Magnesium, 74, 115, 116
Magnetic field, 190 (see also arc
oscillation)
Manganese, 76, 84, 92, 116
Manganese/sulfur ratio, 288, 394
Maraging steel, 307, 309, 310
Marangoni convection, 109, 110 (see also
weld pool convection)
Martensite,
auto-tempered, 406
formation temperature, 405, 411
high carbon, 399–401
microstructure, 400, 403, 426, 448, 450
tempering, 406, 449
Melting efficiency, 44
INDEX 459
Mechanical properties
effect of annealing, 345
effect of ferrite content, 238, 279
effect of grain size, 187–188
effect o flux composition, 84
effect of inclusions and porosity, 250
effect of liquation, 329
effect of nitrogen content, 72
effect of oxygen content, 75, 89
effect of oxygen/acetylene ratio, 74
effect of porosity, 81
Mercury method, 77
Metal transfer, 19, 21, 22, 74, 116, 326
short-circuiting, 21
globular, 21
spray, 21
Microsegregation, 160–163, 232, 243–249,
268, 333, 334, 399, 401
Mushy zone, 179–182, 187, 275
Natural aging, 353, 359–365, 367–370
Ni-base alloys, 310, 375–390
compositions, 376
constitutional liquation, 309, 310
heat-affected zone softening, 376, 381,
383
partially-melted zone cracking or
liquation cracking, 335, 376
precipitation reaction, 376
reheat cracking, 376, 384–390
solidification cracking, 268–271,
273–276
typical welding problems, 376
Nitride formers, 66, 72, 405, 432
Nitrogen, 65–72, 224–226
Nondendritic equiaxed zone, 184, 185,
195
Nonepitaxial growth, 175
Nucleation in weld metal
mechanisms, 178–187
acicular ferrite, 233–235
heterogeneous (see also
heterogeneous nucleation)
Nuclei (see heterogeneous nuclei)
Oxyacetylene welding process, 3–11, 74
Oxygen/acetylene ratio, 74
Oxygen equivalence, 73
Oxygen equivalent, 237, 238
Oxygen level, 66–70, 73–76, 83, 87, 89, 91
effect of basicity index, 83, 87
effect on acicular ferrite, 235–238
effect on grain size, 236
effect of oxygen/acetylene ratio, 74
Partially grain refining, 396–403
Partially melted zone, 303–336
liquation cracking or hot cracking,
321–327
ductility loss, 328, 329, 354
hydrogen cracking, 328
liquation mechanism, 304–314
Partially mixed zone, 252
Phase diagrams,
Al-Cu, 306, 356
Al-Mg-Si, 331
Fe-Cr-Ni, 217, 220, 227, 434
Fe-C, 281, 318, 395, 434
Fe-Cr, 434
Fe-Cr-C, 447, 449
Ni-base, 377
304 stainless steel, 437
Phosphorus, 280
Planar solidification, 156, 159, 160, 316
Plasma arc welding, 3, 6, 17–19, 40, 42,
74, 80, 81, 354, 366
Polarity, 14, 15, 17, 19, 74, 80, 81, 91, 92,
354, 366
Pool shape, 54, 55
Porosity,
fatigue initiation, 131
in aluminum alloys, 66, 80, 81, 95, 252,
257, 259, 354
in copper, 82
in steel, 10, 28, 89, 90, 394
Post-solidification phase transformations,
ferrite-austenite, 216–232
austenite-ferrite, 232–239
Postweld heat treating,
Al alloys, 354, 363–370
effect on distortion, 130
steels, 79, 125, 127, 407–410, 416
Ni-base alloys, 384–386
stainless steels, 432, 439, 442, 450, 451
Postweld heat-treatment cracking,
384–390 (see also reheat cracking)
Powder metallurgy, 257
Power density, 3, 4, 11, 27, 28, 45 (see
also power density distribution)
Power density distribution, 45–47, 57, 58,
101–103, 107
Precipitation hardening, 353, 358, 359,
375, 379, 405
Preheat, 56, 57, 125, 130, 255, 256, 294,
394, 402, 404, 405–410, 415–417
460 INDEX
Primary solidification phase, 166,
216–219, 230, 246, 279, 281, 291
Quenched and tempered steels, 127,
406–408
Reactive and refractory metals, 29
Recrystallization, 343–345, 347
Reheat cracking,
ferritic steels, 394, 418–422
Ni-base alloys, 376, 384–390
test methods, 419–420
Reinforcement, 7, 132, 134, 136, 137
Residual stresses, 122–126, 132–136, 378,
384–388, 417, 422, 445, 446
Resistance spot welds, 309, 439
Reversion, 360–364, 381, 382
Rimmed steels, 28
Root cracks, 413
Schaeffler diagram, 223, 255
Scheil equation, 151, 245, 274
Self-shielded arc welding, 66, 67, 72
Sensitization,
austenitic stainless steels, 433–440, 451,
452
effect of carbon content, 438, 439
ferritic stainless steels, 447
location in welds, 438
remedies, 439, 440
stabilized austenitic stainless steels,
441–444
Shear stress, 104
Shielded metal arc welding, 3, 6, 11, 12,
66, 67, 75, 76
Shielding gas,
argon, 19, 40, 47, 65, 70, 120, 326
Ar-CO2, 237
Ar-H2, 77, 95, 250, 255
Ar-N2, 71, 223, 241
Ar-O2, 21–22, 237, 415
CO2, 21, 65, 73
flux-core arc welding, 23
He-10%Ar, 32
He, 20, 65, 117
hydrogen containing, 78, 79
laser beam welding, 31
nitrogen, 71
properties, 16
oxygen containing, 73, 237–239
Shot peening, 135, 136
Silicon, 76, 92
Slag inclusions, 251
Slag-metal reactions, 82–92
Solidification cracking, 263–296, 330
Al alloys, 271, 273, 277–287, 291–293
control of, 285–295
effect of bead shape, 294, 295
effect of composition, 272, 273,
285–291
effect of ductility of solidifying weld
metal, 276–279
effect of grain boundary liquid,
271–276, 281–283
effect of grain structure, 283–284
effect of primary solidification phase,
279–281
effect of solidification temperature
range, 268–271
stainless steels, 66, 71, 212, 216, 432
steels, 394, 432
test methods, 264–267, 322
theories, 263
Solidification modes, 155–156, 159, 199,
200, 202–206, 216, 316–317
Solidification paths, 166, 167, 271
Solidification temperature range,
268–271 (see also freezing range)
Solidification time, 164
Solidus surface, 217
Solidus temperature, 331
Solute redistribution, 145–155, 162
Solution heat treating, 185, 312, 313, 353,
362
Split-anode method, 100
Stabilized austenitic stainless steels,
440–445
Stainless steels, 431–452
austenitic, 59, 172, 175, 178, 207, 208,
216–232, 244–246, 253–255, 267,
279–281, 336, 408, 417, 432–446, 451,
452
classification, 431, 432
duplex, 71
ferritic, 173, 175, 183, 184, 243, 268,
431, 446–449
martensitic, 244, 431, 432, 449–451
typical welding problems, 432
Steels, (see also stainless steels)
alloy steels, 19, 21, 83–85, 88, 232–239,
252, 253, 258, 259, 268–269, 288, 394,
404–417, 425
carbon steels, 5, 6, 10, 21, 50, 90, 127,
134, 135, 175, 234, 255, 281, 288,
396–404
INDEX 461
cast steels, 401
ferritic steels, 418–422
heat treatable steels, 407–410
maraging steels, 307, 309, 310
quenched and tempered alloy steels,
127, 406–408
rimmed steels, 28
typical welding problems, 394
Stick welding, (see also shielded-metal
arc welding)
Stress corrosion cracking, 125, 445, 446
Stress raisers, 134–136, 294, 413
Stress relief, 125, 126, 136, 140, 385, 389,
409, 442, 446
Subgrain structure, 212
Submerged arc welding, 3, 6, 22, 24, 189,
238
Sulfur, 108–114, 280
Surface-active agent, 108–112
Surface nucleation, 181, 185, 193
Surface tension,
convection induced by, 104, 109, 110
(see also Maragoni convection)
of grain boundary liquid, 281, 282
temperature dependence of, 104, 105,
108–113, 117
Temperature gradient, 166, 186, 201
Tempering bead technique, 408
Titanium, 67, 73
Thermal cycles, 52–56
Thermal expansion coefficients, 284, 325,
419, 420, 445
Thermal properties, 50
Thermal simulator (Gleeble), 58, 59, 184,
323, 419, 421
Toe crack, 135, 413
Toughness, 72, 75, 89, 238
Underbead crack, 412, 425, 426, 450
Undercut, 20, 134–135, 207
Unmixed zone, 254
Vapor pressure, 82, 91, 115, 116
Varestraint test, 266, 267, 270, 276, 278,
321, 330, 332–334
Vinckier test, 419–420
Weld decay, 432–440 (see also
sensitization)
Weld pool,
evaporation, 114–116
electromagnetic stirring of, 188–192
shapes, 53–55, 112–114, 176 (see also
weld pool convection)
Weld pool convection, 103–114
driving forces, 104
effect on macrosegregation (weld pool
mixing), 243, 255, 257, 259
effect on nucleation, 180, 185
effect on penetration, 107–114
laminar flow, 114,
NaNO3, 109–111
turbulent flow, 114
Weld simulator, 58, 59
Welding positions, 9
Welding processes, 3, 6, 26, 66 (see also
arc welding processes)
Work-hardened materials, 343–354
WRC-1992 diagram, 225
WELDING
METALLURGY
SECOND EDITION
WELDING
METALLURGY
SECOND EDITION
Sindo Kou
Professor and Chair
Department of Materials Science and Engineering
University of Wisconsin
A JOHN WILEY & SONS, INC., PUBLICATION
Copyright © 2003 by John Wiley & Sons, Inc. All rights reserved.
Published by John Wiley & Sons, Inc., Hoboken, New Jersey.
Published simultaneously in Canada.
No part of this publication may be reproduced, stored in a retrieval system, or transmitted in
any form or by any means, electronic, mechanical, photocopying, recording, scanning, or
otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright
Act, without either the prior written permission of the Publisher, or authorization through
payment of the appropriate per-copy fee to the Copyright Clearance Center, Inc., 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400, fax 978-750-4470, or on the web at
www.copyright.com. Requests to the Publisher for permission should be addressed to the
Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201)
748-6011, fax (201) 748-6008, e-mail: permreq@wiley.com.
Limit of Liability/Disclaimer of Warranty:While the publisher and author have used their best
efforts in preparing this book, they make no representations or warranties with respect to the
accuracy or completeness of the contents of this book and specifically disclaim any implied
warranties of merchantability or fitness for a particular purpose. No warranty may be created
or extended by sales representatives or written sales materials. The advice and strategies
contained herein may not be suitable for your situation.You should consult with a professional
where appropriate. Neither the publisher nor author shall be liable for any loss of profit or any
other commercial damages, including but not limited to special, incidental, consequential, or
other damages.
For general information on our other products and services please contact our Customer Care
Department within the U.S. at 877-762-2974, outside the U.S. at 317-572-3993 or fax 317-572-
4002.
Wiley also publishes its books in a variety of electronic formats. Some content that appears in
print, however, may not be available in electronic format.
Library of Congress Cataloging-in-Publication Data
Kou, Sindo.
Welding metallurgy / Sindo Kou.–2nd ed.
p. cm.
“A Wiley-Interscience publication.”
Includes bibliographical references and index.
ISBN 0-471-43491-4
1. Welding. 2. Metallurgy. 3. Alloys. I. Title.
TS227 .K649 2002
671.5¢2–dc21
2002014327
Printed in the United States of America.
10 9 8 7 6 5 4 3 2 1
To Warren F. Savage
for his outstanding contributions to welding metallurgy
CONTENTS
Preface xiii
I INTRODUCTION 1
1 Fusion Welding Processes 3
1.1 Overview 3
1.2 Oxyacetylene Welding 7
1.3 Shielded Metal Arc Welding 11
1.4 Gas–Tungsten Arc Welding 13
1.5 Plasma Arc Welding 16
1.6 Gas–Metal Arc Welding 19
1.7 Flux-Core Arc Welding 22
1.8 Submerged Arc Welding 22
1.9 Electroslag Welding 24
1.10 Electron Beam Welding 27
1.11 Laser Beam Welding 29
References 33
Further Reading 34
Problems 34
2 Heat Flow in Welding 37
2.1 Heat Source 37
2.2 Analysis of Heat Flow in Welding 47
2.3 Effect of Welding Parameters 53
2.4 Weld Thermal Simulator 58
References 60
Further Reading 62
Problems 62
3 Chemical Reactions in Welding 65
3.1 Overview 65
3.2 Gas–Metal Reactions 68
3.3 Slag–Metal Reactions 82
References 92
vii
Further Reading 95
Problems 95
4 Fluid Flow and Metal Evaporation in Welding 97
4.1 Fluid Flow in Arcs 97
4.2 Fluid Flow in Weld Pools 103
4.3 Metal Evaporation 114
4.4 Active Flux GTAW 116
References 117
Further Reading 119
Problems 120
5 Residual Stresses, Distortion, and Fatigue 122
5.1 Residual Stresses 122
5.2 Distortion 126
5.3 Fatigue 131
5.4 Case Studies 137
References 140
Further Reading 141
Problems 141
II THE FUSION ZONE 143
6 Basic Solidification Concepts 145
6.1 Solute Redistribution during Solidification 145
6.2 Solidification Modes and Constitutional Supercooling 155
6.3 Microsegregation and Banding 160
6.4 Effect of Cooling Rate 163
6.5 Solidification Path 166
References 167
Further Reading 168
Problems 169
7 Weld Metal Solidification I: Grain Structure 170
7.1 Epitaxial Growth at Fusion Boundary 170
7.2 Nonepitaxial Growth at Fusion Boundary 172
7.3 Competitive Growth in Bulk Fusion Zone 174
7.4 Effect of Welding Parameters on Grain Structure 174
7.5 Weld Metal Nucleation Mechanisms 178
7.6 Grain Structure Control 187
viii CONTENTS
References 195
Further Reading 197
Problems 197
8 Weld Metal Solidification II: Microstructure within Grains 199
8.1 Solidification Modes 199
8.2 Dendrite and Cell Spacing 204
8.3 Effect of Welding Parameters 206
8.4 Refining Microstructure within Grains 209
References 213
Further Reading 213
Problems 214
9 Post-Solidification Phase Transformations 216
9.1 Ferrite-to-Austenite Transformation in Austenitic Stainless
Steel Welds 216
9.2 Austenite-to-Ferrite Transformation in Low-Carbon,
Low-Alloy Steel Welds 232
References 239
Further Reading 241
Problems 241
10 Weld Metal Chemical Inhomogeneities 243
10.1 Microsegregation 243
10.2 Banding 249
10.3 Inclusions and Gas Porosity 250
10.4 Inhomogeneities Near Fusion Boundary 252
10.5 Macrosegregation in Bulk Weld Metal 255
References 260
Further Reading 261
Problems 261
11 Weld Metal Solidification Cracking 263
11.1 Characteristics, Cause, and Testing 263
11.2 Metallurgical Factors 268
11.3 Mechanical Factors 284
11.4 Reducing Solidification Cracking 285
11.5 Case Study: Failure of a Large Exhaust Fan 295
References 296
Further Reading 299
Problems 299
CONTENTS ix
III THE PARTIALLY MELTED ZONE 301
12 Formation of the Partially Melted Zone 303
12.1 Evidence of Liquation 303
12.2 Liquation Mechanisms 304
12.3 Directional Solidification of Liquated Material 314
12.4 Grain Boundary Segregation 314
12.5 Grain Boundary Solidification Modes 316
12.6 Partially Melted Zone in Cast Irons 318
References 318
Problems 319
13 Difficulties Associated with the Partially Melted Zone 321
13.1 Liquation Cracking 321
13.2 Loss of Strength and Ductility 328
13.3 Hydrogen Cracking 328
13.4 Remedies 330
References 336
Problems 338
IV THE HEAT-AFFECTED ZONE 341
14 Work-Hardened Materials 343
14.1 Background 343
14.2 Recrystallization and Grain Growth in Welding 347
14.3 Effect of Welding Parameters and Process 349
References 351
Further Reading 352
Problems 352
15 Precipitation-Hardening Materials I: Aluminum Alloys 353
15.1 Background 353
15.2 Al–Cu–Mg and Al–Mg–Si Alloys 359
15.3 Al–Zn–Mg Alloys 367
15.4 Friction Stir Welding of Aluminum Alloys 370
References 371
Further Reading 372
Problems 372
16 Precipitation-Hardening Materials II: Nickel-Base Alloys 375
16.1 Background 375
x CONTENTS
16.2 Reversion of Precipitate and Loss of Strength 379
16.3 Postweld Heat Treatment Cracking 384
References 390
Further Reading 392
Problems 392
17 Transformation-Hardening Materials: Carbon and
Alloy Steels 393
17.1 Phase Diagram and CCT Diagrams 393
17.2 Carbon Steels 396
17.3 Low-Alloy Steels 404
17.4 Hydrogen Cracking 410
17.5 Reheat Cracking 418
17.6 Lamellar Tearing 422
17.7 Case Studies 425
References 427
Further Reading 429
Problems 430
18 Corrosion-Resistant Materials: Stainless Steels 431
18.1 Classification of Stainless Steels 431
18.2 Austenitic Stainless Steels 433
18.3 Ferritic Stainless Steels 446
18.4 Martensitic Stainless Steels 449
18.5 Case Study: Failure of a Pipe 451
References 452
Further Reading 453
Problems 454
Index 455
CONTENTS xi
PREFACE
Since the publication of the first edition of this book in 1987, there has been
much new progress made in welding metallurgy. The purpose for the second
edition is to update and improve the first edition. Examples of improvements
include (1) much sharper photomicrographs and line drawings; (2) integration
of the phase diagram, thermal cycles, and kinetics with the microstructure to
explain microstructural development and defect formation in welds; and (3)
additional exercise problems. Specific revisions are as follows.
In Chapter 1 the illustrations for all welding processes have been redrawn
to show both the overall process and the welding area. In Chapter
2 the heat source efficiency has been updated and the melting efficiency
added. Chapter 3 has been revised extensively, with the dissolution of
atomic nitrogen, oxygen, and hydrogen in the molten metal considered and
electrochemical reactions added. Chapter 4 has also been revised extensively,
with the arc added, and with flow visualization, arc plasma dragging, and
turbulence included in weld pool convection. Shot peening is added to
Chapter 5.
Chapter 6 has been revised extensively, with solute redistribution and
microsegregation expanded and the solidification path added. Chapter 7 now
includes nonepitaxial growth at the fusion boundary and formation of nondendritic
equiaxed grains. In Chapter 8 solidification modes are explained with
more illustrations. Chapter 9 has been expanded significantly to add ferrite
formation mechanisms, new ferrite prediction methods, the effect of cooling
rate, and factors affecting the austenite–ferrite transformation. Chapter 10
now includes the effect of both solid-state diffusion and dendrite tip undercooling
on microsegregation. Chapter 11 has been revised extensively to
include the effect of eutectic reactions, liquid distribution, and ductility of
the solidifying metal on solidification cracking and the calculation of fraction
of liquid in multicomponent alloys.
Chapter 12 has been rewritten completely to include six different liquation
mechanisms in the partially melted zone (PMZ), the direction and modes of
grain boundary (GB) solidification, and the resultant GB segregation. Chapter
13 has been revised extensively to include the mechanism of PMZ cracking
and the effect of the weld-metal composition on cracking.
Chapter 15 now includes the heat-affected zone (HAZ) in aluminum–
lithium–copper welds and friction stir welds and Chapter 16 the HAZ of
Inconel 718. Chapter 17 now includes the effect of multiple-pass welding on
xiii
reheat cracking and Chapter 18 the grain boundary chromium depletion in a
sensitized austenitic stainless steel.
The author thanks the National Science Foundation and NASA for
supporting his welding research, from which this book draws frequently.
He also thanks the American Welding Society and ASM International for permissions
to use numerous copyrighted materials. Finally, he thanks C. Huang,
G. Cao,C. Limmaneevichitr, H.D. Lu, K.W.Keehn, and T.Tantanawat for providing
technical material, requesting permissions, and proofreading.
Sindo Kou
Madison,Wisconsin
xiv PREFACE
PART I
Introduction
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
1 Fusion Welding Processes
Fusion welding processes will be described in this chapter, including gas
welding, arc welding, and high-energy beam welding. The advantages and disadvantages
of each process will be discussed.
1.1 OVERVIEW
1.1.1 Fusion Welding Processes
Fusion welding is a joining process that uses fusion of the base metal to make
the weld. The three major types of fusion welding processes are as follows:
1. Gas welding:
Oxyacetylene welding (OAW)
2. Arc welding:
Shielded metal arc welding (SMAW)
Gas–tungsten arc welding (GTAW)
Plasma arc welding (PAW)
Gas–metal arc welding (GMAW)
Flux-cored arc welding (FCAW)
Submerged arc welding (SAW)
Electroslag welding (ESW)
3. High-energy beam welding:
Electron beam welding (EBW)
Laser beam welding (LBW)
Since there is no arc involved in the electroslag welding process, it is not
exactly an arc welding process. For convenience of discussion, it is grouped
with arc welding processes.
1.1.2 Power Density of Heat Source
Consider directing a 1.5-kW hair drier very closely to a 304 stainless steel sheet
1.6mm (1/16 in.) thick. Obviously, the power spreads out over an area of roughly
3
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
50mm (2in.) diameter, and the sheet just heats up gradually but will not melt.
With GTAW at 1.5kW, however, the arc concentrates on a small area of about
6mm (1/4 in.) diameter and can easily produce a weld pool.This example clearly
demonstrates the importance of the power density of the heat source in
welding.
The heat sources for the gas, arc, and high-energy beam welding processes
are a gas flame, an electric arc, and a high-energy beam, respectively. The
power density increases from a gas flame to an electric arc and a high-energy
beam. As shown in Figure 1.1, as the power density of the heat source
increases, the heat input to the workpiece that is required for welding
decreases.The portion of the workpiece material exposed to a gas flame heats
up so slowly that, before any melting occurs, a large amount of heat is already
conducted away into the bulk of the workpiece. Excessive heating can cause
damage to the workpiece, including weakening and distortion. On the contrary,
the same material exposed to a sharply focused electron or laser beam
can melt or even vaporize to form a deep keyhole instantaneously, and before
much heat is conducted away into the bulk of the workpiece, welding is completed
(1).
Therefore, the advantages of increasing the power density of the heat
source are deeper weld penetration, higher welding speeds, and better weld
quality with less damage to the workpiece, as indicated in Figure 1.1. Figure
1.2 shows that the weld strength (of aluminum alloys) increases as the heat
input per unit length of the weld per unit thickness of the workpiece decreases
(2). Figure 1.3a shows that angular distortion is much smaller in EBW than in
4 FUSION WELDING PROCESSES
Increasing
damage to
workpiece
Increasing
penetration,
welding speed,
weld quality,
equipment cost
Power density of heat source
high energy
beam welding
arc
welding
gas
welding
Heat input to workpiece
Figure 1.1 Variation of heat input to the workpiece with power density of the heat
source.
GTAW (2). Unfortunately, as shown in Figure 1.3b, the costs of laser and electron
beam welding machines are very high (2).
1.1.3 Welding Processes and Materials
Table 1.1 summarizes the fusion welding processes recommended for carbon
steels, low-alloy steels, stainless steels, cast irons, nickel-base alloys, and
OVERVIEW 5
50
40
30
5 10 50 100 500
Heat input, kJ/in./in.
Strength, ksi
5083 7039
60
2219
6061
Figure 1.2 Variation of weld strength with heat input per unit length of weld per unit
thickness of workpiece. Reprinted from Mendez and Eagar (2).
Productivity, cm/s
Flame
Arc
Laser
electron
beam
Productivity, inch of weld/s
Capital equipment, dollars
Power density, W/m 2
0.1 1 10 100
0.04 0.4 4 40 400
103
105
107
103
105
107
(b)
t
EBW
GTAW
Weld thickness t, mm
0 20 40
Distortion angle degree
2
4
6
8
0
(a)
α
α
Figure 1.3 Comparisons between welding processes: (a) angular distortion; (b) capital
equipment cost. Reprinted from Mendez and Eagar (2).
TABLE 1.1 Overview of Welding Processesa
Material Thicknessb SMAW SAW GMAW FCAW GTAW PAW ESW OFW EBW LBW
Carbon S ✕✕ ✕ ✕ ✕ ✕ ✕
steels I
✕✕ ✕ ✕ ✕ ✕ ✕ ✕
M ✕✕ ✕ ✕ ✕ ✕ ✕
T ✕✕ ✕ ✕ ✕ ✕ ✕
Low-alloy S ✕✕ ✕ ✕ ✕ ✕ ✕
steels I
✕✕ ✕ ✕ ✕ ✕ ✕
M ✕✕ ✕ ✕ ✕ ✕
T ✕✕ ✕ ✕ ✕ ✕
Stainless S ✕✕ ✕ ✕✕ ✕ ✕ ✕
steels I
✕✕ ✕ ✕ ✕✕ ✕ ✕
M ✕✕ ✕ ✕ ✕ ✕ ✕
T ✕✕ ✕ ✕ ✕ ✕
Cast iron I
✕ ✕
M ✕✕ ✕ ✕ ✕
T ✕✕ ✕ ✕ ✕
Nickel S ✕ ✕ ✕ ✕ ✕ ✕
✕
and alloys I
✕✕ ✕M ✕✕ ✕ ✕ T ✕ ✕ ✕ ✕
Aluminum S ✕ and alloys I
✕
✕
M ✕✕ T ✕ ✕
a Process code: SMAW, shielded metal arc welding; SAW, submerged arc welding; GMAW, gas–metal arc welding; FCAW, flux-cored arc welding; GTAW,
gas–tungsten arc welding; PAW, plasma arc welding; ESW, electroslag welding; OFW, oxyfuel gas welding; EBW, electron beam welding; LBW, laser beam
welding.
b Abbreviations: S, sheet, up to 3mm (1/8 in.); I, intermediate, 3–6mm (1/8–1/4 in.); M, medium, 6–19mm (1/4–3/4 in.); T, thick, 19mm (3/4 in.) and up; X,
recommended.
Source: Welding Handbook (3).
6
aluminum alloys (3). For one example, GMAW can be used for all the materials
of almost all thickness ranges while GTAW is mostly for thinner workpieces.
For another example, any arc welding process that requires the use of a flux,
such as SMAW, SAW, FCAW, and ESW, is not applicable to aluminum alloys.
1.1.4 Types of Joints and Welding Positions
Figure 1.4 shows the basic weld joint designs in fusion welding: the butt, lap,
T-, edge, and corner joints. Figure 1.5 shows the transverse cross section of
some typical weld joint variations. The surface of the weld is called the face,
the two junctions between the face and the workpiece surface are called the
toes, and the portion of the weld beyond the workpiece surface is called the
reinforcement. Figure 1.6 shows four welding positions.
1.2 OXYACETYLENE WELDING
1.2.1 The Process
Gas welding is a welding process that melts and joins metals by heating them
with a flame caused by the reaction between a fuel gas and oxygen. Oxyacetylene
welding (OAW), shown in Figure 1.7, is the most commonly used
gas welding process because of its high flame temperature. A flux may be used
to deoxidize and cleanse the weld metal. The flux melts, solidifies, and forms
a slag skin on the resultant weld metal. Figure 1.8 shows three different types
of flames in oxyacetylene welding: neutral, reducing, and oxidizing (4), which
are described next.
1.2.2 Three Types of Flames
A. Neutral Flame This refers to the case where oxygen (O2) and acetylene
(C2H2) are mixed in equal amounts and burned at the tip of the welding torch.
A short inner cone and a longer outer envelope characterize a neutral flame
OXYACETYLENE WELDING 7
(a) butt joint
(c) T-joint
(b) lap joint
(d) edge joint
(e) corner joint
Figure 1.4 Five basic types of weld joint designs.
(Figure 1.8a). The inner cone is the area where the primary combustion takes
place through the chemical reaction between O2 and C2H2, as shown in Figure
1.9. The heat of this reaction accounts for about two-thirds of the total heat
generated.The products of the primary combustion, CO and H2, react with O2
from the surrounding air and form CO2 and H2O. This is the secondary combustion,
which accounts for about one-third of the total heat generated. The
area where this secondary combustion takes place is called the outer envelope.
It is also called the protection envelope since CO and H2 here consume
the O2 entering from the surrounding air, thereby protecting the weld metal
from oxidation. For most metals, a neutral flame is used.
B. Reducing Flame When excess acetylene is used, the resulting flame is
called a reducing flame.The combustion of acetylene is incomplete.As a result,
a greenish acetylene feather between the inert cone and the outer envelope
characterizes a reducing flame (Figure 1.8b). This flame is reducing in nature
and is desirable for welding aluminum alloys because aluminum oxidizes
easily. It is also good for welding high-carbon steels (also called carburizing
flame in this case) because excess oxygen can oxidize carbon and form CO gas
porosity in the weld metal.
8 FUSION WELDING PROCESSES
Toe
Toe
Reinforcement
T-joint;
fillet weld
Butt joint;
square weld
Toe
Reinforcement
Butt joint;
single-V-groove weld
Root
Toe
Lap joint;
fillet weld
Toe
Toe
(d)
(c)
(a)
(b)
Toe
T-joint;
single bevel weld
Toe
(e)
Figure 1.5 Typical weld joint variations.
OXYACETYLENE WELDING 9
(a) flat (b) horizontal
(c) vertical (d) overhead
Figure 1.6 Four welding positions.
Oxygen/acetylene
mixture
Filler rod
Protection
envelope
Metal
droplet
Base metal Weld pool
Weld
metal
Slag
Primary
combustion
Flow meter Regulator
Acetylene
Welding
direction
Gas torch
Workpiece
C2H2
O2
Valve
Oxygen
(a)
(b)
Figure 1.7 Oxyacetylene welding: (a) overall process; (b) welding area enlarged.
C. Oxidizing Flame When excess oxygen is used, the flame becomes oxidizing
because of the presence of unconsumed oxygen. A short white inner
cone characterizes an oxidizing flame (Figure 1.8c). This flame is preferred
when welding brass because copper oxide covers the weld pool and thus prevents
zinc from evaporating from the weld pool.
1.2.3 Advantages and Disadvantages
The main advantage of the oxyacetylene welding process is that the equipment
is simple, portable, and inexpensive.Therefore, it is convenient for maintenance
and repair applications. However, due to its limited power density, the
10 FUSION WELDING PROCESSES
inner cone
inner cone
acetylene feather
Reducing Flame
inner cone Oxidizing Flame
Neutral Flame
(a)
(b)
(c)
Figure 1.8 Three types of flames in oxyacetylene welding. Modified from Welding
Journal (4). Courtesy of American Welding Society.
Gas C2H2 + O2
Torch
2500 oC
1000 oC
2800 - 3500 oC
inner
cone
outer
envelope
2C2H2 + 2O2 (from cylinder)
Secondary combustion in outer
envelope (1/3 total heat) :
4CO + 2H2
4CO + 2O2 (from air) 4CO2
2H2 + O2 (from air) 2H2O
Primary combustion in inner
cone (2/3 total heat) :
Flame
Figure 1.9 Chemical reactions and temperature distribution in a neutral oxyacetylene
flame.
welding speed is very low and the total heat input per unit length of the weld
is rather high, resulting in large heat-affected zones and severe distortion.The
oxyacetylene welding process is not recommended for welding reactive metals
such as titanium and zirconium because of its limited protection power.
1.3 SHIELDED METAL ARC WELDING
1.3.1 The Process
Shielded metal arc welding (SMAW) is a process that melts and joins metals
by heating them with an arc established between a sticklike covered electrode
and the metals, as shown in Figure 1.10. It is often called stick welding.
The electrode holder is connected through a welding cable to one terminal
of the power source and the workpiece is connected through a second cable
to the other terminal of the power source (Figure 1.10a).
The core of the covered electrode, the core wire, conducts the electric
current to the arc and provides filler metal for the joint. For electrical contact,
the top 1.5 cm of the core wire is bare and held by the electrode holder. The
electrode holder is essentially a metal clamp with an electrically insulated
outside shell for the welder to hold safely.
The heat of the arc causes both the core wire and the flux covering at the
electrode tip to melt off as droplets (Figure 1.10b). The molten metal collects
in the weld pool and solidifies into the weld metal.The lighter molten flux, on
the other hand, floats on the pool surface and solidifies into a slag layer at the
top of the weld metal.
1.3.2 Functions of Electrode Covering
The covering of the electrode contains various chemicals and even metal
powder in order to perform one or more of the functions described below.
A. Protection It provides a gaseous shield to protect the molten metal from
air. For a cellulose-type electrode, the covering contains cellulose, (C6H10O5)x.
A large volume of gas mixture of H2, CO, H2O, and CO2 is produced when
cellulose in the electrode covering is heated and decomposes. For a limestone-
(CaCO3) type electrode, on the other hand,CO2 gas and CaO slag form when
the limestone decomposes. The limestone-type electrode is a low-hydrogentype
electrode because it produces a gaseous shield low in hydrogen. It is often
used for welding metals that are susceptible to hydrogen cracking, such as
high-strength steels.
B. Deoxidation It provides deoxidizers and fluxing agents to deoxidize and
cleanse the weld metal. The solid slag formed also protects the already solidified
but still hot weld metal from oxidation.
SHIELDED METAL ARC WELDING 11
C. Arc Stabilization It provides arc stabilizers to help maintain a stable
arc. The arc is an ionic gas (a plasma) that conducts the electric current.
Arc stabilizers are compounds that decompose readily into ions in the arc,
such as potassium oxalate and lithium carbonate. They increase the electrical
conductivity of the arc and help the arc conduct the electric current more
smoothly.
D. Metal Addition It provides alloying elements and/or metal powder to
the weld pool. The former helps control the composition of the weld metal
while the latter helps increase the deposition rate.
1.3.3 Advantages and Disadvantages
The welding equipment is relatively simple, portable, and inexpensive as compared
to other arc welding processes. For this reason, SMAW is often used for
maintenance, repair, and field construction. However, the gas shield in SMAW
is not clean enough for reactive metals such as aluminum and titanium. The
deposition rate is limited by the fact that the electrode covering tends to overheat
and fall off when excessively high welding currents are used.The limited
length of the electrode (about 35 cm) requires electrode changing, and this
further reduces the overall production rate.
12 FUSION WELDING PROCESSES
Gaseous shield
Core wire
Flux covering
Slag
Metal
droplet
Flux
droplet
Base metal Weld pool
Weld
metal
Arc
(a)
(b)
Power
Source
Cable 1
Electrode
holder
Stick
electrode
Welding
direction
Workpiece
Cable 2
Figure 1.10 Shielded metal arc welding: (a) overall process; (b) welding area enlarged.
1.4 GAS–TUNGSTEN ARC WELDING
1.4.1 The Process
Gas–tungsten arc welding (GTAW) is a process that melts and joins metals by
heating them with an arc established between a nonconsumable tungsten electrode
and the metals, as shown in Figure 1.11. The torch holding the tungsten
electrode is connected to a shielding gas cylinder as well as one terminal of
the power source, as shown in Figure 1.11a. The tungsten electrode is usually
in contact with a water-cooled copper tube, called the contact tube, as shown
in Figure 1.11b, which is connected to the welding cable (cable 1) from the
terminal. This allows both the welding current from the power source to
enter the electrode and the electrode to be cooled to prevent overheating.The
workpiece is connected to the other terminal of the power source through a
different cable (cable 2). The shielding gas goes through the torch body and
is directed by a nozzle toward the weld pool to protect it from the air. Protection
from the air is much better in GTAW than in SMAW because an inert
gas such as argon or helium is usually used as the shielding gas and because
the shielding gas is directed toward the weld pool. For this reason, GTAW is
GAS–TUNGSTEN ARC WELDING 13
Shielding gas
nozzle
Weld
metal
Metal
droplet
Shielding
gas
Base metal Weld pool
Arc
Filler
rod
Welding
direction
Filler rod
Torch
Cable 1
Workpiece
Shielding gas
cylinder
Flow
meter
Regulator
Tungsten electrode
(a)
(b)
Power
source
Contact tube
Shielding
gas
Cable 1
Cable 2
Figure 1.11 Gas–tungsten arc welding: (a) overall process; (b) welding area enlarged.
also called tungsten–inert gas (TIG) welding. However, in special occasions a
noninert gas (Chapter 3) can be added in a small quantity to the shielding gas.
Therefore, GTAW seems a more appropriate name for this welding process.
When a filler rod is needed, for instance, for joining thicker materials, it can
be fed either manually or automatically into the arc.
1.4.2 Polarity
Figure 1.12 shows three different polarities in GTAW (5), which are described
next.
A. Direct-Current Electrode Negative (DCEN) This, also called the straight
polarity, is the most common polarity in GTAW.The electrode is connected to
the negative terminal of the power supply. As shown in Figure 1.12a, electrons
are emitted from the tungsten electrode and accelerated while traveling
through the arc. A significant amount of energy, called the work function, is
required for an electron to be emitted from the electrode.When the electron
enters the workpiece, an amount of energy equivalent to the work function is
released.This is why in GTAW with DCEN more power (about two-thirds) is
located at the work end of the arc and less (about one-third) at the electrode
end. Consequently, a relatively narrow and deep weld is produced.
B. Direct-Current Electrode Positive (DCEP) This is also called the reverse
polarity. The electrode is connected to the positive terminal of the power
source. As shown in Figure 1.12b, the heating effect of electrons is now at the
tungsten electrode rather than at the workpiece. Consequently, a shallow weld
is produced. Furthermore, a large-diameter, water-cooled electrodes must be
used in order to prevent the electrode tip from melting. The positive ions of
the shielding gas bombard the workpiece, as shown in Figure 1.13, knocking
off oxide films and producing a clean weld surface. Therefore, DCEP can be
14 FUSION WELDING PROCESSES
DC electrode
negative
DC electrode
positive
AC
deep weld,
no surface cleaning
shallow weld,
surface cleaning
intermediate
pool
(a) (b) (c)
Figure 1.12 Three different polarities in GTAW.
used for welding thin sheets of strong oxide-forming materials such as aluminum
and magnesium, where deep penetration is not required.
C. Alternating Current (AC) Reasonably good penetration and oxide
cleaning action can both be obtained, as illustrated in Figure 1.12c.This is often
used for welding aluminum alloys.
1.4.3 Electrodes
Tungsten electrodes with 2% cerium or thorium have better electron
emissivity, current-carrying capacity, and resistance to contamination than
pure tungsten electrodes (3). As a result, arc starting is easier and the arc is
more stable. The electron emissivity refers to the ability of the electrode tip
to emit electrons. A lower electron emissivity implies a higher electrode tip
temperature required to emit electrons and hence a greater risk of melting the
tip.
1.4.4 Shielding Gases
Both argon and helium can be used. Table 1.2 lists the properties of some
shielding gases (6). As shown, the ionization potentials for argon and helium
are 15.7 and 24.5 eV (electron volts), respectively. Since it is easier to ionize
argon than helium, arc initiation is easier and the voltage drop across the arc
is lower with argon. Also, since argon is heavier than helium, it offers more
effective shielding and greater resistance to cross draft than helium. With
DCEP or AC, argon also has a greater oxide cleaning action than helium.
These advantages plus the lower cost of argon make it more attractive for
GTAW than helium.
GAS–TUNGSTEN ARC WELDING 15
Cleaning action (electrode positive)
knocked-off atoms
oxide film on surface
M O M O M
M O
Workpiece (negative)
Ar+
bombarding heavy ion
Figure 1.13 Surface cleaning action in GTAW with DC electrode positive.
Because of the greater voltage drop across a helium arc than an argon arc,
however, higher power inputs and greater sensitivity to variations in the arc
length can be obtained with helium. The former allows the welding of thicker
sections and the use of higher welding speeds. The latter, on the other hand,
allows a better control of the arc length during automatic GTAW.
1.4.5 Advantages and Disadvantages
Gas–tungsten arc welding is suitable for joining thin sections because of its
limited heat inputs. The feeding rate of the filler metal is somewhat independent
of the welding current, thus allowing a variation in the relative amount
of the fusion of the base metal and the fusion of the filler metal. Therefore,
the control of dilution and energy input to the weld can be achieved without
changing the size of the weld. It can also be used to weld butt joints of thin
sheets by fusion alone, that is, without the addition of filler metals or autogenous
welding. Since the GTAW process is a very clean welding process, it can
be used to weld reactive metals, such as titanium and zirconium, aluminum,
and magnesium.
However, the deposition rate in GTAW is low. Excessive welding currents
can cause melting of the tungsten electrode and results in brittle tungsten
inclusions in the weld metal. However, by using preheated filler metals, the
deposition rate can be improved. In the hot-wire GTAW process, the wire is
fed into and in contact with the weld pool so that resistance heating can be
obtained by passing an electric current through the wire.
1.5 PLASMA ARC WELDING
1.5.1 The Process
Plasma arc welding (PAW) is an arc welding process that melts and joins metals
by heating them with a constricted arc established between a tungsten elec-
16 FUSION WELDING PROCESSES
TABLE 1.2 Properties of Shielding Gases Used for Welding
Molecular Specific Gravity Ionization
Chemical Weight with Respect to Air Density Potential
Gas Symbol (g/mol) at 1atm and 0°C (g/L) (eV)
Argon Ar 39.95 1.38 1.784 15.7
Carbon dioxide CO2 44.01 1.53 1.978 14.4
Helium He 4.00 0.1368 0.178 24.5
Hydrogen H2 2.016 0.0695 0.090 13.5
Nitrogen N2 28.01 0.967 1.25 14.5
Oxygen O2 32.00 1.105 1.43 13.2
Source: Reprinted from Lyttle (6).
trode and the metals, as shown in Figure 1.14. It is similar to GTAW, but an
orifice gas as well as a shielding gas is used. As shown in Figure 1.15, the arc in
PAW is constricted or collimated because of the converging action of the orifice
gas nozzle, and the arc expands only slightly with increasing arc length (5).
Direct-current electrode negative is normally used, but a special variablepolarity
PAW machine has been developed for welding aluminum, where the
presence of aluminum oxide films prevents a keyhole from being established.
1.5.2 Arc Initiation
The tungsten electrode sticks out of the shielding gas nozzle in GTAW (Figure
1.11b) while it is recessed in the orifice gas nozzle in PAW (Figure 1.14b). Consequently,
arc initiation cannot be achieved by striking the electrode tip against
the workpiece as in GTAW. The control console (Figure 1.14a) allows a pilot
arc to be initiated, with the help of a high-frequency generator, between the
electrode tip and the water-cooled orifice gas nozzle.The arc is then gradually
PLASMA ARC WELDING 17
Shielding gas
Power
Source
Cables
Filler
rod
Control
console
Welding
direction
Torch
Orifice gas
Workpiece
(a)
(b) Tungsten
electrode
Weld
metal
Orifice gas
Shielding gas
nozzle
Shielding
gas
Orifice
Molten metal
Base metal
Keyhole
Arc plasma
Orifice gas nozzle
(water cooled)
Figure 1.14 Plasma arc welding: (a) overall process; (b) welding area enlarged and
shown with keyholing.
transferred from between the electrode tip and the orifice gas nozzle to
between the electrode tip and the workpiece.
1.5.3 Keyholing
In addition to the melt-in mode adopted in conventional arc welding processes
(such as GTAW), the keyholing mode can also be used in PAW in certain
ranges of metal thickness (e.g., 2.5–6.4mm).With proper combinations of the
orifice gas flow, the travel speed, and the welding current, keyholing is possible.
Keyholing is a positive indication of full penetration and it allows the
use of significantly higher welding speeds than GTAW. For example, it has
been reported (7) that PAW took one-fifth to one-tenth as long to complete
a 2.5-m-long weld in 6.4-mm-thick 410 stainless steel as GTAW. Gas–tungsten
arc welding requires multiple passes and is limited in welding speed.As shown
in Figure 1.16, 304 stainless steel up to 13mm (1/2 in.) thick can be welded in a
single pass (8). The wine-cup-shaped weld is common in keyholing PAW.
1.5.4 Advantages and Disadvantages
Plasma arc welding has several advantages over GTAW.With a collimated arc,
PAW is less sensitive to unintentional arc length variations during manual
welding and thus requires less operator skill than GTAW.The short arc length
in GTAW can cause a welder to unintentionally touch the weld pool with the
electrode tip and contaminate the weld metal with tungsten. However, PAW
does not have this problem since the electrode is recessed in the nozzle. As
already mentioned, the keyhole is a positive indication of full penetration, and
it allows higher welding speeds to be used in PAW.
However, the PAW torch is more complicated. It requires proper electrode
tip configuration and positioning, selection of correct orifice size for the application,
and setting of both orifice and shielding gas flow rates. Because of the
18 FUSION WELDING PROCESSES
Plasma arc Gas tungsten arc
Figure 1.15 Comparison between a gas–tungsten arc and a plasma arc. From Welding
Handbook (5). Courtesy of American Welding Society.
need for a control console, the equipment cost is higher in PAW than in GTAW.
The equipment for variable-polarity PAW is much more expensive than that
for GTAW.
1.6 GAS–METAL ARC WELDING
1.6.1 The Process
Gas–metal arc welding (GMAW) is a process that melts and joins metals by
heating them with an arc established between a continuously fed filler wire
electrode and the metals, as shown in Figure 1.17. Shielding of the arc and the
molten weld pool is often obtained by using inert gases such as argon and
helium, and this is why GMAW is also called the metal–inert gas (MIG)
welding process. Since noninert gases, particularly CO2, are also used,GMAW
seems a more appropriate name. This is the most widely used arc welding
process for aluminum alloys. Figure 1.18 shows gas–metal arc welds of 5083
aluminum, one made with Ar shielding and the other with 75% He–25% Ar
shielding (9). Unlike in GTAW, DCEP is used in GMAW. A stable arc, smooth
metal transfer with low spatter loss and good weld penetration can be
obtained.With DCEN or AC, however, metal transfer is erratic.
1.6.2 Shielding Gases
Argon, helium, and their mixtures are used for nonferrous metals as well as
stainless and alloy steels. The arc energy is less uniformly dispersed in an Ar
arc than in a He arc because of the lower thermal conductivity of Ar. Consequently,
the Ar arc plasma has a very high energy core and an outer mantle
of lesser thermal energy. This helps produce a stable, axial transfer of metal
GAS–METAL ARC WELDING 19
Figure 1.16 A plasma arc weld made in 13-mm-thick 304 stainless steel with keyholing.
From Lesnewich (8).
droplets through an Ar arc plasma.The resultant weld transverse cross section
is often characterized by a papillary- (nipple-) type penetration pattern (10)
such as that shown in Figure 1.18 (left).With pure He shielding, on the other
hand, a broad, parabolic-type penetration is often observed.
With ferrous metals, however, He shielding may produce spatter and Ar
shielding may cause undercutting at the fusion lines. Adding O2 (about 3%)
or CO2 (about 9%) to Ar reduces the problems. Carbon and low-alloy steels
are often welded with CO2 as the shielding gas, the advantages being higher
20 FUSION WELDING PROCESSES
(b)
(a)
Shielding gas
nozzle
Weld
metal
Metal
droplet
Shielding
gas
Base metal Weld pool
Arc
Shielding
gas
Flow meter Regulator
Wire drive
& control
Wire
reel
Wire
electrode
Workpiece
Gun
Power
Source
Shielding
gas
cylinder
Welding
direction
Wire electrode
Contact tube
Cable 1
Cable 2
Cable 1
Figure 1.17 Gas–metal arc welding: (a) overall process; (b) welding area enlarged.
Figure 1.18 Gas–metal arc welds in 6.4-mm-thick 5083 aluminum made with argon
(left) and 75% He–25% Ar (right). Reprinted from Gibbs (9). Courtesy of American
Welding Society.
welding speed, greater penetration, and lower cost. Since CO2 shielding produces
a high level of spatter, a relatively low voltage is used to maintain a short
buried arc to minimize spatter; that is, the electrode tip is actually below the
workpiece surface (10).
1.6.3 Modes of Metal Transfer
The molten metal at the electrode tip can be transferred to the weld pool by
three basic transfer modes: globular, spray, and short-circuiting.
A. Globular Transfer Discrete metal drops close to or larger than the
electrode diameter travel across the arc gap under the influence of gravity.
Figure 1.19a shows globular transfer during GMAW of steel at 180A and
with Ar–2% O2 shielding (11). Globular transfer often is not smooth and
produces spatter. At relatively low welding current globular transfer
occurs regardless of the type of the shielding gas.With CO2 and He, however,
it occurs at all usable welding currents. As already mentioned, a short
buried arc is used in CO2-shielded GMAW of carbon and low-alloy steels to
minimize spatter.
GAS–METAL ARC WELDING 21
Figure 1.19 Metal transfer during GMAW of steel with Ar–2% O2 shielding: (a)
globular transfer at 180A and 29V shown at every 3 ¥ 10-3 s; (b) spray transfer at
320A and 29V shown at every 2.5 ¥ 10-4 s. Reprinted from Jones et al. (11). Courtesy
of American Welding Society.
B. Spray Transfer Above a critical current level, small discrete metal drops
travel across the arc gap under the influence of the electromagnetic force at
much higher frequency and speed than in the globular mode. Figure 1.19b
shows spray transfer during GMAW of steel at 320A and with Ar–2% O2
shielding (11). Metal transfer is much more stable and spatter free. The critical
current level depends on the material and size of the electrode and the
composition of the shielding gas. In the case of Figure 1.19, the critical current
was found to be between 280 and 320 A (11).
C. Short-Circuiting Transfer The molten metal at the electrode tip is transferred
from the electrode to the weld pool when it touches the pool surface,
that is, when short circuiting occurs. Short-circuiting transfer encompasses the
lowest range of welding currents and electrode diameters. It produces a small
and fast-freezing weld pool that is desirable for welding thin sections, out-ofposition
welding (such as overhead-position welding), and bridging large root
openings.
1.6.4 Advantages and Disadvantages
Like GTAW,GMAW can be very clean when using an inert shielding gas.The
main advantage of GMAW over GTAW is the much higher deposition rate,
which allows thicker workpieces to be welded at higher welding speeds. The
dual-torch and twin-wire processes further increase the deposition rate of
GMAW (12). The skill to maintain a very short and yet stable arc in GTAW
is not required. However, GMAW guns can be bulky and difficult-to-reach
small areas or corners.
1.7 FLUX-CORE ARC WELDING
1.7.1 The Process
Flux-core arc welding (FCAW) is similar to GMAW, as shown in Figure 1.20a.
However, as shown in Figure 1.20b, the wire electrode is flux cored rather than
solid; that is, the electrode is a metal tube with flux wrapped inside. The functions
of the flux are similar to those of the electrode covering in SMAW, including
protecting the molten metal from air. The use of additional shielding gas
is optional.
1.8 SUBMERGED ARC WELDING
1.8.1 The Process
Submerged arc welding (SAW) is a process that melts and joins metals by
heating them with an arc established between a consumable wire electrode
22 FUSION WELDING PROCESSES
and the metals, with the arc being shielded by a molten slag and granular flux,
as shown in Figure 1.21. This process differs from the arc welding processes
discussed so far in that the arc is submerged and thus invisible.The flux is supplied
from a hopper (Figure 1.21a), which travels with the torch. No shielding
gas is needed because the molten metal is separated from the air by the molten
slag and granular flux (Figure 1.21b). Direct-current electrode positive is most
often used. However, at very high welding currents (e.g., above 900A) AC is
preferred in order to minimize arc blow. Arc blow is caused by the electromagnetic
(Lorentz) force as a result of the interaction between the electric
current itself and the magnetic field it induces.
1.8.2 Advantages and Disadvantages
The protecting and refining action of the slag helps produce clean welds in
SAW. Since the arc is submerged, spatter and heat losses to the surrounding
air are eliminated even at high welding currents. Both alloying elements and
metal powders can be added to the granular flux to control the weld metal
composition and increase the deposition rate, respectively. Using two or more
electrodes in tandem further increases the deposition rate. Because of its high
SUBMERGED ARC WELDING 23
(b)
(a)
Shielding gas
nozzle
Weld
metal
Base metal Weld pool
Arc
Shielding
gas
Flow meter Regulator
Wire drive
& control
Wire
reel
Wire
electrode
Workpiece
Gun
Power
Source
Shielding
gas
cylinder
Welding
direction
Wire electrode
Contact tube
Cable 1
Cable 2
Cable 1
Slag
Metal droplet
Flux droplet
Shielding gas
(optional)
Figure 1.20 Flux-core arc welding: (a) overall process; (b) welding area enlarged.
deposition rate, workpieces much thicker than that in GTAW and GMAW can
be welded by SAW. However, the relatively large volumes of molten slag and
metal pool often limit SAW to flat-position welding and circumferential
welding (of pipes). The relatively high heat input can reduce the weld quality
and increase distortions.
1.9 ELECTROSLAG WELDING
1.9.1 The Process
Electroslag welding (ESW) is a process that melts and joins metals by heating
them with a pool of molten slag held between the metals and continuously
feeding a filler wire electrode into it, as shown in Figure 1.22. The weld pool
is covered with molten slag and moves upward as welding progresses. A pair
of water-cooled copper shoes, one in the front of the workpiece and one
behind it, keeps the weld pool and the molten slag from breaking out. Similar
to SAW, the molten slag in ESW protects the weld metal from air and refines
it. Strictly speaking, however, ESW is not an arc welding process, because the
arc exists only during the initiation period of the process, that is, when the arc
24 FUSION WELDING PROCESSES
Wire reel
Wire electrode
Wire drive & control
Cables
Flux hopper
Workpiece
Granular
flux Wire electrode
Arc Molten slag
Solidified slag
Metal pool
(a)
(b)
Welding
direction
Base metal Weld metal
Power
Source
Droplet
Figure 1.21 Submerged arc welding: (a) overall process; (b) welding area enlarged.
heats up the flux and melts it. The arc is then extinguished, and the resistance
heating generated by the electric current passing through the slag keeps it
molten. In order to make heating more uniform, the electrode is often oscillated,
especially when welding thicker sections. Figure 1.23 is the transverse
cross section of an electroslag weld in a steel 7 cm thick (13).Typical examples
of the application of ESW include the welding of ship hulls, storage tanks, and
bridges.
1.9.2 Advantages and Disadvantages
Electroslag welding can have extremely high deposition rates, but only one
single pass is required no matter how thick the workpiece is. Unlike SAW or
other arc welding processes, there is no angular distortion in ESW because the
ELECTROSLAG WELDING 25
Wire electrode
Consumable
guide tube
Base metal
Molten slag
Weld pool
Weld metal
Cable
Watercooled
copper
shoes
Cable
Workpiece
Wire
reel
Power
Source
Wire feed
motor & control
Bottom
support
(a)
(b)
Figure 1.22 Electroslag welding: (a) overall process; (b) welding area enlarged.
weld is symmetrical with respect to its axis. However, the heat input is very
high and the weld quality can be rather poor, including low toughness caused
by the coarse grains in the fusion zone and the heat-affected zone. Electroslag
welding is restricted to vertical position welding because of the very large
pools of the molten metal and slag.
Figure 1.24 summarizes the deposition rates of the arc welding processes
discussed so far (14). As shown, the deposition rate increases in the order of
26 FUSION WELDING PROCESSES
Figure 1.23 Transverse cross section of electroslag weld in 70-mm-thick steel.
Reprinted from Eichhorn et al. (13). Courtesy of American Welding Society.
Deposition Rate, Lb/hr
Welding Process (100% Duty Cycle)
6010
6012
7018
7024
cold
wire
fine wire
spray
CO2 shielded
with iron powder added
1 electrode 2 electrodes
4 electrodes
5 electrodes
3 electrodes
SMAW
GTAW/PAW
GMAW
FCAW
SAW
ESW
0 40 80 120 160 200
Deposition Rate, Kg/hr
0 20 40 60 80
iron powder
Figure 1.24 Deposition rate in arc welding processes. Modified from Cary (14).
GTAW, SMAW, GMAW and FCAW, SAW, and ESW. The deposition rate can
be much increased by adding iron powder in SAW or using more than one
wire in SAW, ESW, and GMAW (not shown).
1.10 ELECTRON BEAM WELDING
1.10.1 The Process
Electron beam welding (EBW) is a process that melts and joins metals by
heating them with an electron beam. As shown in Figure 1.25a, the cathode of
the electron beam gun is a negatively charged filament (15).When heated up
to its thermionic emission temperature, this filament emits electrons. These
electrons are accelerated by the electric field between a negatively charged
bias electrode (located slightly below the cathode) and the anode. They pass
through the hole in the anode and are focused by an electromagnetic coil to
a point at the workpiece surface. The beam currents and the accelerating
voltages employed for typical EBW vary over the ranges of 50–1000mA and
30–175kV, respectively. An electron beam of very high intensity can vaporize
the metal and form a vapor hole during welding, that is, a keyhole, as depicted
in Figure 1.25b.
Figure 1.26 shows that the beam diameter decreases with decreasing
ambient pressure (1). Electrons are scattered when they hit air molecules, and
the lower the ambient pressure, the less they are scattered. This is the main
reason for EBW in a vacuum chamber.
The electron beam can be focused to diameters in the range of 0.3–0.8mm
and the resulting power density can be as high as 1010W/m2 (1).The very high
ELECTRON BEAM WELDING 27
welding
direction weld
bead
electron
beam
keyhole
weld pool
crosssection
of weld
(b)
moten
metal
electron
beam
specimen
bias
electrode
anode
focusing
coil
to
pump
high
cathode voltage
(a)
vacuum
chamber
Figure 1.25 Electron beam welding: (a) process; (b) keyhole. Modified from Arata
(15).
power density makes it possible to vaporize the material and produce a deeppenetrating
keyhole and hence weld. Figure 1.27 shows a single-pass electron
beam weld and a dual-pass gas–tungsten arc weld in a 13-mm-thick (0.5-in.)
2219 aluminum, the former being much narrower (16). The energy required
per unit length of the weld is much lower in the electron beam weld (1.5 kJ/cm,
or 3.8 kJ/in.) than in the gas–tungsten arc weld (22.7 kJ/cm, or 57.6 kJ/in.).
Electron beam welding is not intended for incompletely degassed materials
such as rimmed steels. Under high welding speeds gas bubbles that do not
have enough time to leave deep weld pools result in weld porosity. Materials
containing high-vapor-pressure constituents, such as Mg alloys and Pbcontaining
alloys, are not recommended for EBW because evaporation of
these elements tends to foul the pumps or contaminate the vacuum system.
1.10.2 Advantages and Disadvantages
With a very high power density in EBW, full-penetration keyholing is possible
even in thick workpieces. Joints that require multiple-pass arc welding can
28 FUSION WELDING PROCESSES
750 torr 500 torr 250 torr 50 torr 5 torr
Figure 1.26 Dispersion of electron beam at various ambient pressures (1). Reprinted
from Welding Handbook (1). Courtesy of American Welding Society.
13 mm
(0.5 in)
electron
beam
weld
gas
tungsten
arc weld
(a) (b)
Figure 1.27 Welds in 13-mm-thick 2219 aluminum: (a) electron beam weld; (b)
gas–tungsten arc weld. From Farrell (16).
be welded in a single pass at a high welding speed. Consequently, the total
heat input per unit length of the weld is much lower than that in arc welding,
resulting in a very narrow heat-affected zone and little distortion. Reactive
and refractory metals can be welded in vacuum where there is no air to cause
contamination. Some dissimilar metals can also be welded because the very
rapid cooling in EBW can prevent the formation of coarse, brittle intermetallic
compounds.When welding parts varying greatly in mass and size, the ability
of the electron beam to precisely locate the weld and form a favorably shaped
fusion zone helps prevent excessive melting of the smaller part.
However, the equipment cost for EBW is very high. The requirement of
high vacuum (10-3–10-6 torr) and x-ray shielding is inconvenient and time consuming.
For this reason, medium-vacuum (10-3–25torr) EBW and nonvacuum
(1 atm) EBW have also been developed. The fine beam size requires precise
fit-up of the joint and alignment of the joint with the gun. As shown in Figure
1.28, residual and dissimilar metal magnetism can cause beam deflection and
result in missed joints (17).
1.11 LASER BEAM WELDING
1.11.1 The Process
Laser beam welding (LBW) is a process that melts and joins metals by heating
them with a laser beam. The laser beam can be produced either by a solid-
LASER BEAM WELDING 29
Missed
joint
48356 47421
SB49 A387
Figure 1.28 Missed joints in electron beam welds in 150-mm-thick steels: (a)
2.25Cr–1Mo steel with a transverse flux density of 3.5 G parallel to joint plane; (b) SB
(C–Mn) steel and A387 (2.25Cr–1Mo) steel. Reprinted from Blakeley and Sanderson
(17). Courtesy of American Welding Society.
(a) (b)
state laser or a gas laser. In either case, the laser beam can be focused and
directed by optical means to achieve high power densities. In a solid-state
laser, a single crystal is doped with small concentrations of transition elements
or rare earth elements. For instance, in a YAG laser the crystal of yttrium–
aluminum–garnet (YAG) is doped with neodymium. The electrons of the
dopant element can be selectively excited to higher energy levels upon exposure
to high-intensity flash lamps, as shown in Figure 1.29a. Lasing occurs when
these excited electrons return to their normal energy state, as shown in Figure
1.29b.The power level of solid-state lasers has improved significantly, and continuous
YAG lasers of 3 or even 5 kW have been developed.
In a CO2 laser, a gas mixture of CO2, N2, and He is continuously excited
by electrodes connected to the power supply and lases continuously. Higher
30 FUSION WELDING PROCESSES
(a)
Power
source
Cooling
system
Reflecting mirror
Patially reflecting mirror
Focusing lens
Travel
direction
Pool Weld
Workpiece
Crystal
High
intensity
flash
lamp
(b)
energy absorbed
from flash lamp
energy emitted
as heat
energy
emitted as
light (photon)
nucleus
ground
intermediate
excited
inner
electrons
outer
electron
energy
levels
normal
electron
orbits
Figure 1.29 Laser beam welding with solid-state laser: (a) process; (b) energy absorption
and emission during laser action. Modified from Welding Handbook (1).
power can be achieved by a CO2 laser than a solid-state laser, for instance,
15kW. Figure 1.30a shows LBW in the keyholing mode. Figure 1.30b shows a
weld in a 13-mm-thick A633 steel made with a 15-kW CO2 laser at 20mm/s
(18).
Besides solid-state and gas lasers, semiconductor-based diode lasers have
also been developed. Diode lasers of 2.5kW power and 1mm focus diameter
have been demonstrated (19).While keyholing is not yet possible, conductionmode
(surface melting) welding has produced full-penetration welds with a
depth–width ratio of 3 : 1 or better in 3-mm-thick sheets.
1.11.2 Reflectivity
The very high reflectivity of a laser beam by the metal surface is a well-known
problem in LBW. As much as about 95% of the CO2 beam power can be
reflected by a polished metal surface. Reflectivity is slightly lower with a YAG
laser beam. Surface modifications such as roughening, oxidizing, and coating
can reduce reflectivity significantly (20). Once keyholing is established, absorption
is high because the beam is trapped inside the hole by internal reflection.
1.11.3 Shielding Gas
A plasma (an ionic gas) is produced during LBW, especially at high power
levels, due to ionization by the laser beam.The plasma can absorb and scatter
the laser beam and reduce the depth of penetration significantly. It is therefore
necessary to remove or suppress the plasma (21). The shielding gas for
protecting the molten metal can be directed sideways to blow and deflect the
plasma away from the beam path. Helium is often preferred to argon as the
shielding gas for high-power LBW because of greater penetration depth (22).
LASER BEAM WELDING 31
welding
direction weld
bead
laser
beam
keyhole
weld pool
crosssection
of weld
(a) (b)
molten
metal
2 mm
Figure 1.30 Laser beam welding with CO2 laser: (a) process; (b) weld in 13-mm-thick
A633 steel. (b) Courtesy of E.A. Metzbower.
Since the ionization energy of helium (24.5 eV) is higher than that of argon
(15.7 eV), helium is less likely to be ionized and become part of the plasma
than argon. However, helium is lighter than air and is thus less effective in displacing
air from the beam path. Helium–10% Ar shielding has been found
to improve penetration over pure He at high-speed welding where a light
shielding gas may not have enough time to displace air from the beam path
(23).
1.11.4 Lasers in Arc Welding
As shown in Figure 1.31, laser-assisted gas metal arc welding (LAGMAW) has
demonstrated significantly greater penetration than conventional GMAW
(24). In addition to direct heating, the laser beam acts to focus the arc by
heating its path through the arc. This increases ionization and hence the conductivity
of the arc along the beam path and helps focus the arc energy along
the path. It has been suggested that combining the arc power with a 5-kW CO2
laser, LAGMAW has the potential to achieve weld penetration in mild steel
equivalent to that of a 20–25-kW laser (24). Albright et al. (25) have shown
that a lower power CO (not CO2) laser of 7W and 1mm diameter can initiate,
guide, and focus an Ar–1% CO gas–tungsten arc.
1.11.5 Advantages and Disadvantages
Like EBW, LBW can produce deep and narrow welds at high welding speeds,
with a narrow heat-affected zone and little distortion of the workpiece. It can
32 FUSION WELDING PROCESSES
LAGMAW
GMAW
300 400 500 600 700
0
2
4
6
8
10
12
Welding current (A)
Penetration (mm)
Figure 1.31 Weld penetration in GMAW and laser-assisted GMAW using CO2 laser
at 5.7kW. Reprinted from Hyatt et al. (24). Courtesy of American Welding Society.
be used for welding dissimilar metals or parts varying greatly in mass and size.
Unlike EBW, however, vacuum and x-ray shielding are not required in LBW.
However, the very high reflectivity of a laser beam by the metal surface is a
major drawback, as already mentioned. Like EBW, the equipment cost is very
high, and precise joint fit-up and alignment are required.
REFERENCES
1. Welding Handbook, Vol. 3, 7th ed., American Welding Society, Miami, FL, 1980,
pp. 170–238.
2. Mendez, P. F., and Eagar, T. W., Advanced Materials and Processes, 159: 39,
2001.
3. Welding Handbook, Vol. 1, 7th ed., American Welding Society, Miami, FL, 1976,
pp. 2–32.
4. Welding Workbook, Data Sheet 212a, Weld. J., 77: 65, 1998.
5. Welding Handbook, Vol. 2, 7th ed., American Welding Society, Miami, FL, 1978,
pp. 78–112, 296–330.
6. Lyttle, K. A., in ASM Handbook, Vol. 6, ASM International, Materials Park, OH,
1993, p. 64.
7. Schwartz, M. M., Metals Joining Manual, McGraw-Hill, New York, 1979, pp. 2–1 to
3–40.
8. Lesnewich, A., in Weldability of Steels, 3rd ed., Eds. R. D. Stout and W. D. Doty,
Welding Research Council, New York, 1978, p. 5.
9. Gibbs, F. E., Weld. J., 59: 23, 1980.
10. Fact Sheet—Choosing Shielding for GMA Welding, Weld. J., 79: 18, 2000.
11. Jones, L. A., Eagar, T.W., and Lang, J. H., Weld. J., 77: 135s, 1998.
12. Blackman, S. A., and Dorling, D.V., Weld. J., 79: 39, 2000.
13. Eichhorn, F., Remmel, J., and Wubbels, B., Weld. J., 63: 37, 1984.
14. Cary, H. B., Modern Welding Technology, Prentice-Hall, Englewood Cliffs, NJ,
1979.
15. Arata, Y., Development of Ultra High Energy Density Heat Source and Its Application
to Heat Processing, Okada Memorial Japan Society, 1985.
16. Farrell,W. J., The Use of Electron Beam to Fabricate Structural Members, Creative
Manufacturing Seminars, ASTME Paper SP 63-208, 1962–1963.
17. Blakeley, P. J., and Sanderson, A., Weld. J., 63: 42, 1984.
18. Metzbower, E. A., private communication, Naval Research Laboratory,
Washington, DC.
19. Bliedtner, J., Heyse,Th., Jahn, D., Michel, G., Muller, H., and Wolff, D.,Weld. J., 80:
47, 2001.
20. Xie, J., and Kar, A., Weld. J., 78: 343s, 1999.
21. Mazumder, J., in ASM Handbook,Vol. 6, ASM International, Materials Park, OH,
1993, p. 874.
22. Rockstroh, T., and Mazumder, J., J. Appl. Phys., 61: 917, 1987.
REFERENCES 33
23. Seaman, F., Role of Shielding Gas in Laser Welding, Technical Paper MR77-982,
Society of Manufacturing Engineers, Dearborn, MI, 1977.
24. Hyatt, C.V., Magee, K. H., Porter, J. F., Merchant,V. E., and Matthews, J. R.,Weld.
J., 80: 163s, 2001.
25. Albright, C. E., Eastman, J., and Lempert,W., Weld. J., 80: 55, 2001.
26. Ushio, M., Matsuda, F., and Sadek, A. A., in “ International Trends in Welding
Science and Technology, Eds. S. A. David and J. M. Vitek, ASM International,
Materials Park, OH, March 1993, p. 408.
FURTHER READING
1. Arata,Y., Development of Ultra High Energy Density Heat Source and Its Application
to Heat Processing, Okada Memorial Society for the Promotion of Welding,
Japan, 1985.
2. Schwartz, M. M., Metals Joining Manual, McGraw-Hill, New York, 1979.
3. Welding Handbook, Vols. 1–3, 7th ed., American Welding Society, Miami, FL,
1980.
4. Duley,W.W., Laser Welding,Wiley, New York, 1999.
5. ASM Handbook, Vol. 6, ASM International, Materials Park, OH, 1993.
PROBLEMS
1.1 It has been suggested that compared to SMAW, the cooling rate is
higher in GMAW and it is, therefore, more likely for heat-affected zone
cracking to occur in hardenable steels.What is the main reason for the
cooling rate to be higher in GMAW than SMAW?
1.2 The diameter of the electrodes to be used in SMAW depends on factors
such as the workpiece thickness, the welding position, and the joint
design. Large electrodes, with their corresponding high currents, tend to
produce large weld pools. When welding in the overhead or vertical
position, do you prefer using larger or smaller electrodes?
1.3 In arc welding, the magnetic field induced by the welding current
passing through the electrode and the workpiece can interact with the
arc and cause “arc blow.” Severe arc blow can cause excessive weld
spatter and incomplete fusion.When arc blow is a problem in SMAW,
do you expect to minimize it by using DC or AC for welding?
1.4 In the hot-wire GTAW process, shown in Figure P1.4, the tip of the filler
metal wire is dipped in the weld pool and the wire itself is resistance
heated by means of a second power source between the contact tube of
the wire and the workpiece. In the case of steels, the deposition rate can
34 FUSION WELDING PROCESSES
be more than doubled this way. Do you prefer using an AC or a DC
power source for heating the wire? Do you expect to apply this process
to aluminum and copper alloys?
1.5 In GTAW the welding cable is connected to the tungsten electrode
through a water-cooled copper contact tube, as shown in Figure 1.11.
Why is the tube positioned near the lower end of the electrode instead
of the top?
1.6 Measurements of the axial temperature distribution along the GTAW
electrode have shown that the temperature drops sharply from the
electrode tip toward the contact tube. Why? For instance, with a
2.4-mm-diameter W–ThO2 electrode at 150A, the temperature drops
from about 3600 K at the tip to about 2000 K at 5mm above the tip.
Under the same condition but with a W–CeO2 electrode, the temperature
drops from about 2700 K at the tip to about 1800 K at 5mm above
the tip (26).Which electrode can carry more current before melting and
why?
1.7 Experimental results show that in EBW the penetration depth of the
weld decreases as the welding speed increases. Explain why. Under the
same power and welding speed, do you expect a much greater penetration
depth in aluminum or steel and why?
1.8 How does the working distance in EBW affect the depth–width ratio of
the resultant weld?
1.9 Consider EBW in the presence of a gas environment. Under the same
power and welding speed, rank and explain the weld penetration for Ar,
He, and air. The specific gravities of Ar, He, and air with respect to air
are 1.38, 0.137, and 1, respectively, at 1 atm, 0°C.
PROBLEMS 35
Shielding
gas nozzle
Weld
metal
Shielding
gas
Base metal Weld pool
Arc
Tungsten electrode
Contact tube
Cable 1
Wire feeder
Filler wire
2nd
power
source
Cable 3
Cable 4
Contact tube
Figure P1.4
1.10 Which arc welding process could have been used for joining the edge
weld of thin-gauge steel shown in Figure P1.10 and why?
1.11 Two 15-cm-thick steel plates were joined together in a single pass, as
shown in Figure P1.11.Which welding process could have been used and
why?
36 FUSION WELDING PROCESSES
weld
1 mm
1 mm
transverse
cross-section
Figure P1.10
15 cm
15 cm
weld transverse
cross-section
steel
steel
Figure P1.11
2 Heat Flow in Welding
Heat flow during welding, as will be shown throughout Parts II–IV of this
book, can strongly affect phase transformations during welding and thus the
resultant microstructure and properties of the weld. It is also responsible for
weld residual stresses and distortion, as will be discussed in Chapter 5.
2.1 HEAT SOURCE
2.1.1 Heat Source Efficiency
A. Definition The heat source efficiency h is defined as
(2.1)
where Q is the rate of heat transfer from the heat source to the workpiece,
Qnominal the nominal power of the heat source, and tweld the welding time. A
portion of the power provided by the heat source is transferred to the workpiece
and the remaining portion is lost to the surroundings. Consequently,
h < 1. If the heat source efficiency h is known, the heat transfer rate to theworkpiece, Q, can be easily determined from Equation (2.1).In arc welding with a constant voltage E and a constant current I, the arcefficiency can be expressed as(2.2)Equation (2.2) can also be applied to electron beam welding, where h is theheat source efficiency. In laser beam welding, Qnominal in Equation (2.1) is thepower of the laser beam, for instance, 2500W.It should be noted that in the welding community the term heat input oftenrefers to Qnominal, or EI in the case of arc welding, and the term heat input perunit length of weld often refers to the ratio Qnominal/V, or EI/V, where V is thewelding speed.B. Measurements The heat source efficiency can be measured with acalorimeter. The heat transferred from the heat source to the workpiece is inh= =QtEItQEIweldweldh= =QtQ tQQweldnominal weld nominal37Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4turn transferred from the workpiece to the calorimeter, which can be determinedas described below.Kou et al. (1, 2) used simple tubular calorimeters to determine the arcefficiency in GTAW of aluminum, as shown in Figure 2.1a. The calorimetercan be a round cross section if the workpiece is a pipe or a rectangularcross section if the workpiece is a sheet. The temperature rise in the coolingwater (Tout - Tin) can be measured using thermocouples or thermistors.Heat transfer from the workpiece to the calorimeter is as follows (1–3):(2.3)where W is the mass flow rate of water, C the specific heat of water, Tout theoutlet water temperature, Tin the inlet water temperature, and t time.The integralcorresponds to the shaded area in Figure 2.1b.The arc efficiency h can bedetermined from Equations (2.2) and (2.3).Giedt et al. (4) used the Seebeck envelop calorimeter shown in Figure 2.2ato measure the arc efficiency in GTAW. The name Seebeck came fromthe Seebeck thermoelectric effect of a thermocouple, namely, a voltage isproduced between two thermocouple junctions of different temperatures.The torch can be quickly withdrawn after welding, and the calorimeter lidcan be closed to minimize heat losses to the surrounding air. As shown inFigure 2.2b, heat transfer from the workpiece to the calorimeter can bedetermined by measuring the temperature difference DT and hence gradientQtweld= WC(Tout-Tin)dtªWC (Tout-Tin)dt • • Ú Ú 0 038 HEAT FLOW IN WELDINGGTAWPipe torchWater outThermocouple ThermocoupleWater inInsulationArcTime, secondsTemperature, oCTemperature, oF(a)(b)100 1201201008020304050ToutTinShaded area∞= ∫ 0[To u t- Tin ]dtFigure 2.1 Measurement of arc efficiency in GTAW: (a) calorimeter; (b) rise in coolingwater temperature as a function of time. Modified from Kou et al. (1, 2).across a “gradient layer” of material of known thermal conductivity k andthickness L:Qt A k (2.4)TLweld = dt • Ú D0HEAT SOURCE 39TLCooling water flowGradient layerThermocouplejunctions for sensingtemperaturedifferenceDirection ofheat flowTemperatureDistanceTemperatureΔ profileFigure 2.2 Measurement of arc efficiency in GTAW: (a) calorimeter; (b) layer of temperaturegradient. Reprinted from Giedt et al. (4). Courtesy of American WeldingSociety.(a)(b)where A is the area for heat flow and DT/L the temperature gradient.The arcefficiency h can be determined from Equations (2.4) and (2.2). This type ofcalorimeter was later used to determine the arc efficiencies in PAW, GMAW,and SAW (5–8).Figure 2.3 shows the results of arc efficiency measurements in GTAW andPAW (2, 5, 7, 9), and Figure 2.4 shows similar results in GMAW and SAW (7,10).These results were obtained using the two types of calorimeters describedabove except for the results of Lu and Kou for GMAW (10, 11), which aredescribed in what follows.In GMAW the arc, metal droplets, and cathode heating all contribute to theefficiency of the heat source. It has been observed in GMAW of aluminumand steel with Ar shielding that current flow or electron emission occurs not40 HEAT FLOW IN WELDING1.00.80.60.40.20.0Arc Efficiency0 10 20 30Travel Speed, mm/sPAW:GTAW:Fuerschbach et al. (1991)Dupont et al. (1995)Kou and Lu (1985)Fuerschbach et al. (1991)Dupont et al. (1995)GTAWEvans et al. (1998)PAWFigure 2.3 Arc efficiencies in GTAW and PAW.175 225 275 325 375 425Arc Efficiency1.00.80.60.40.275 125GMAW:SAW:Lu and Kou (1989)DuPont et al. (1995)SAWGMAWCurrent, ALu and Kou (1989)DuPont et al. (1995)Figure 2.4 Arc efficiencies in GMAW and SAW.uniformly over the workpiece surface but over localized areas on the workpiecesurface called cathode spots (12, 13). The localized heating, calledcathode heating, causes the surface oxide to dissociate and leaves a clean metalsurface (12). Cathode heating is attributed to field-type emission of electrons.Unlike thermionic emission at the tungsten electrode in DC electrodenegativeGTAW, field emission electrons do not cool the cathode (6).Lu and Kou (10, 11) used a combination of three calorimeters to estimatethe amounts of heat transfer from the arc, filler metal droplets, and cathodeheating to the workpiece in GMAW of aluminum. Figure 2.5a shows the measurementof heat transfer from droplets (11). The arc is established betweena GMAW torch and a GTA torch and the droplets collect in the calorimeterbelow. From the water temperature rise and the masses and specific heats ofthe water and the copper basin, the heat transfer from droplets can be deter-HEAT SOURCE 41Filler wireGMAW GTAW torchtorchDropletsWaterRadiation baffleThermistorInsulationCubasinFiller wireGMAW torchWeldO-ringWater out Water inDT DTWorkpieceDT: differential Insulationthermistor(b)Filler wireGMAW torchWeldO-ringWater out Water inDT DTWorkpieceInsulation(c)GTAWtorch(a)Figure 2.5 Calorimeter for measuring heat inputs in GMAW: (a) metal droplets; (b)total heat input; (c) combined heat inputs from arc and metal droplets. Reprinted fromLu and Kou (10, 11). Courtesy of American Welding Society.mined. Figure 2.5b shows the measurement of the total heat transfer to theworkpiece. Figure 2.5c shows the measurement of the combined heat transferfrom the arc and the droplets, with cathode heating shifted to the tungstenelectrode of a nearby GTAW torch (10).The results are shown in Figure 2.6a.By subtracting the combined heat transfer from the arc and droplets from thetotal heat transfer to the workpiece, heat transfer from cathode heating wasdetermined. Figure 2.6b shows the breakdown of the total heat transfer to theworkpiece into those from the arc, droplets, and cathode heating.Within therange of the power studied, the overall efficiency was about 80%, with about45% from the arc, 25% from droplets, and 10% from cathode heating.The heat source efficiency can be very low in LBW because of the highreflectivity of metal surfaces to a laser beam, for instance, 98% for CO2 laseron polished aluminum. The reflectivity can be found by determining the ratioof the reflected beam power to the incident beam power. Xie and Kar (14)show that roughening the surface with sandpapers and oxidizing the surfaceby brief exposure to high temperatures can reduce the reflectivity significantly.C. Heat Source Efficiencies in Various Welding Processes Figure 2.7 summarizesthe heat source efficiencies measured in several welding processes.Afew comments are made as follows:LBW: The heat source efficiency is very low because of the high reflectivityof metal surfaces but can be significantly improved by surface modifications,such as roughening, oxidizing, or coating.PAW: The heat source efficiency is much higher than LBW because reflectivityis not a problem.GTAW: Unlike in PAW there are no heat losses from the arc plasma to thewater-cooled constricting torch nozzle.With DCEN, the imparting electronsare a major source of heat transfer to the workpiece. They releasethe work function as heat and their kinetic energy is also converted into42 HEAT FLOW IN WELDINGGMAW1.6 mm (1/16 in)4043 filler metalTotalArc +dropletsDropletsCurrent x voltage, kW0 2 4 6 80Power inputs, kW246 GMAW1.6 mm (1/16 in)4043 filler metalTotalArcDropletsCathodeCurrent x voltage, kWPower inputs, %0 2 4 6 8020406080(a) (b)Figure 2.6 Power inputs during GMAW of aluminum: (a) measured results; (b) breakdownof total power input. Reprinted from Lu and Kou (10). Courtesy of AmericanWelding Society.heat at the workpiece. In AC GTAW, however, electrons bombard theworkpiece only during the electrode-negative half cycle, and the arc efficiencyis thus lower. In GTAW with DCEP, the arc efficiency is evenlower because the electrons bombard the electrode rather than theworkpiece.GMAW, SMAW: Unlike in GTAW, heat transfer to the electrode can betransferred back to the workpiece through metal droplets, thus improvingthe arc efficiency.SAW: Unlike in GMAW or SMAW, the arc is covered with a thermally insulatingblanket of molten slag and granular flux, thus reducing heat lossesto the surroundings and improving the arc efficiency.EBW: The keyhole in EBW acts like a “black body” trapping the energyfrom the electron beam. As a result, the efficiency of the electron beamis very high.2.1.2 Melting EfficiencyThe ability of the heat source to melt the base metal (as well as the filler metal)is of practical interest to the welder. Figure 2.8 shows the cross-sectional areaHEAT SOURCE 43GTAW GMAW SAW EBWdcenLBW PAWHeat Source Efficiency1.00.80.60.40.20SMAWsurfacemodificationsFigure 2.7 Heat source efficiencies in several welding processes.AfillerAbaseFigure 2.8 Transverse cross section of weld showing areas representing contributionsfrom base metal and filler metal.representing the portion of the weld metal contributed by the base metal,Abase,and that contributed by the filler metal, Afiller. One way to define the meltingefficiency of the welding arc, hm, is as follows:(2.5)where V is the welding speed, Hbase the energy required to raise a unit volumeof base metal to the melting point and melt it, and Hfiller the energy required toraise a unit volume of filler metal to the melting point and melt it.The quantityinside the parentheses represents the volume of material melted while thedenominator represents the heat transfer from the heat source to the workpiece.Figures 2.9a and b show the transverse cross section of two steel welds differingin the melting efficiency (7). Here, EI = 3825W and V = 10mm/s for theshallower weld of lower melting efficiency (Figure 2.9a) and EI = 10170W andV = 26mm/s for the deeper weld of higher melting efficiency (Figure 2.9b).Note that the ratio EI/V is equivalent in each case.hh mbase weld base filler weld fillerweld=(A Vt )H +(A Vt )HEIt44 HEAT FLOW IN WELDINGPAWGTAWGMAWSAWMelting efficiency0 1000 2000 3000 4000 50000.00.10.20.30.40.50.60.7(c)η = η− αν175m 0.50exp EIV/(H )ηEIV/(Hαν)Figure 2.9 Melting efficiency: (a) lower at lower heat input and welding speed; (b)higher at higher heat input and welding speed; (c) variation with dimensionless parameterhEIV/Ha. Reprinted from DuPont and Marder (7). Courtesy of AmericanWelding Society.Fuerschbach and Knorovsky (5) proposed the following empirical equationfor the melting efficiency:(2.6)where A and B are constants,H = Hbase + Hfiller, a is the thermal diffusivity, and is the kinematic viscosity of the weld pool.The results of DuPont and Marder(7) shown in Figure 2.9c confirms the validity of Equation (2.6). As thedimensionless parameter hEIV/Ha increases, the melting efficiency increasesrapidly first and then levels off. If the arc efficiency h is known, hEIV/Ha isalso known and the melting efficiency can be predicted from Figure 2.9.Withthe help of the following equation for determining Afiller, Abase can then becalculated from Equation (2.5):AfillerVtweld = pR2fillerVfillertweld (2.7)or(2.8)In the above equations Rfiller and Vfiller are the radius and feeding speed of thefiller metal, respectively.The left-hand side of Equation (2.7) is the volume ofthe weld metal contributed by the filler metal while the right-hand side is thevolume of filler metal used during welding.It should be noted that the melting efficiency cannot be increased indefinitelyby increasing the welding speed without increasing the power input. Todo so, the power input must be increased along with the welding speed. Itshould also be noted that in the presence of a surface-active agent suchas sulfur in steel, the weld pool can become much deeper even thoughthe welding parameters and physical properties in Equation (2.6) remainunchanged (Chapter 4).2.1.3 Power Density Distribution of Heat SourceA. Effect of Electrode Tip Angle In GTAW with DCEN, the shape of theelectrode tip affects both the shape and power density distribution of thearc. As the electrode tip becomes blunter, the diameter of the arc decreasesand the power density distribution increases, as illustrated in Figure 2.10.Glickstein (15) showed, in Figure 2.11, that the arc becomes more constrictedas the conical tip angle of the tungsten electrode increases.Savage et al. (16) observed that, under the same welding current, speed, andarc gap, the weld depth–width ratio increases with increasing vertex angle ofAR VV fillerfiller2filler =phh a m =- Êˈ¯ABEIV HexpnHEAT SOURCE 4546 HEAT FLOW IN WELDINGTungstenelectrodeArcPowerdensityRadialdistancePowerdensityRadialdistance(a)(b)(c)(d)Figure 2.10 Effect of electrode tip angle on shape and power density distribution ofgas–tungsten arc.Figure 2.11 Effect of electrode tip angle on shape of gas–tungsten arc. Reprinted fromGlickstein (15).the conical tip of the tungsten electrode. Key (17) reported a similar effect ofthe tip angle, at least with Ar shielding, as shown in Figure 2.12.B. Measurements Several investigators have measured the power densitydistribution (and current density distribution) in the gas tungsten arc by usingthe split-anode method (2, 18–20). Figure 2.13 shows the results of Tsai (20)and Lu and Kou (3).For simplicity, the Gaussian-type distribution is often usedas an approximation (21–23).2.2 ANALYSIS OF HEAT FLOW IN WELDINGFigure 2.14 is a schematic showing the welding of a stationary workpiece (24).The origin of the coordinate system moves with the heat source at a constantspeed V in the negative-x direction. Except for the initial and final transientsANALYSIS OF HEAT FLOW IN WELDING 4730o(0.125)60o(0.125)90o(0.500)180o(0)Electrode tip angle(truncation, mm)Figure 2.12 Effect of electrode tip geometry on shape of gas–tungsten arc welds instainless steel (pure Ar, 150A, 2.0 s, spot-on-plate). Reprinted from Key (17). Courtesyof American Welding Society.-0.1r20 2 4 6 80204060q(r) = 30ePower Density, W/mm 2Radius, mmLu and Kou (1988)Tsai (1983)electrode (+)100 A,14.2 V4.7 mm arc100% Ar3.2 mm electrode75 tip angleq(r) = 32e-0.1r2oFigure 2.13 Measured power density distributions. Reprinted from Lu and Kou (3).Courtesy of American Welding Society.of welding, heat flow in a workpiece of sufficient length is steady, or quasistationary,with respect to the moving heat source. In other words, for anobserver moving with the heat source, the temperature distribution and thepool geometry do not change with time.This steady-state assumption was firstused by Rosenthal (25) to simplify the mathematical treatment of heat flowduring welding.2.2.1 Rosenthal’s EquationsRosenthal (25) used the following simplifying assumptions to derive analyticalequations for heat flow during welding:1. steady-state heat flow,2. point heat source,3. negligible heat of fusion,4. constant thermal properties,5. no heat losses from the workpiece surface, and6. no convection in the weld pool.A. Rosenthal’s Two-Dimensional Equation Figure 2.15 is a schematicshowing the welding of thin sheets. Because of the small thickness of the workpiece,temperature variations in the thickness direction are assumed negligibleand heat flow is assumed two dimensional. Rosenthal (25) derived thefollowing equation for two-dimensional heat flow during the welding of thinsheets of infinite width:48 HEAT FLOW IN WELDINGFigure 2.14 Coordinate system (x, y, z) moving with heat source. From Kou and Le(24).(2.9)where T = temperatureT0 = workpiece temperature before weldingk = workpiece thermal conductivityg = workpiece thicknessQ = heat transferred from heat source to workpieceV = travel speeda = workpiece thermal diffusivity, namely, k/rC, where r and C aredensity and specific heat of the workpiece, respectivelyK0 = modified Bessel function of second kind and zero order, as shownin Figure 2.16 (26)r = radial distance from origin, namely, (x2 + y2)1/2Equation (2.9) can be used to calculate the temperature T(x, y) at any locationin the workpiece (x, y) with respect to the moving heat source, forinstance, at x = -1cm and y = 4 cm shown in Figure 2.15. The temperatures atother locations along y = 4 cm can also be calculated, and the temperature distributionalong y = 4 cm can thus be determined. Table 2.1 lists the thermalproperties for several materials (27).B. Rosenthal’s Three-Dimensional Equation The analytical solutionderived by Rosenthal for three-dimensional heat flow in a semi-infinite workpieceduring welding, Figure 2.17, is as follows (25):(2.10)22p 0aT T kRQ( - ) V R x=È- ( - )Π͢˚ ˙exp22 200pa aT T kgQVxK( - ) Vr= Êˈ¯Êˈ¯expANALYSIS OF HEAT FLOW IN WELDING 49thickness, gthin sheets attemperature T0 before weldingheat source, Qweld poolfusion zoneT(x,y) shown at x = -1 cm and y = 4 cmy = 4 cmT(x,y)T2o-3 -2 -1 0 1 2 3 4 5yxweldingspeed, VFigure 2.15 Two-dimensional heat flow during welding of thin workpiece.where R is the radial distance from the origin, namely, (x2 + y2 + z2)1/2. For agiven material and a given welding condition, an isotherm T on a plane at agiven x has a radius of R. In other words, Equation (2.10) implies that on thetransverse cross section of the weld all isotherms, including the fusion boundaryand the outer boundary of the heat-affected zone, are semicircular inshape. Equation (2.10) can be used to calculate the steady-state temperatureT(x, y, z), with respect to the moving heat source, at any location in the workpiece(x, y, z), for instance, at x = 1cm, y = 4cm, and z = 0cm, as shown inFigure 2.17. The temperatures at other locations along y = 4 cm can also be50 HEAT FLOW IN WELDING2.82.42.01.61.20.80.400 1 2 3 4Modified Bessel functionof second kindzero orderK0 [Vr/(2 )]Vr/(2 α)αFigure 2.16 Modified Bessel function of second kind and zero order (26).TABLE 2.1 Thermal Properties for Several MaterialsVolumeThermal Thermal Thermal MeltingDiffusivity, Capacity, Conductivity, PointMaterial a (m2/s) rCs (J/m3K) k (J/m sK) (K)Aluminum 8.5 ¥ 10-5 2.7 ¥ 106 229.0 933Carbon steel 9.1 ¥ 10-6 4.5 ¥ 106 41.0 18009% Ni steel 1.1 ¥ 10-5 3.2 ¥ 106 35.2 1673Austenitic 5.3 ¥ 10-6 4.7 ¥ 106 24.9 1773stainless steelInconel 600 4.7 ¥ 10-6 3.9 ¥ 106 18.3 1673Ti alloy 9.0 ¥ 10-6 3.0 ¥ 106 27.0 1923Copper 9.6 ¥ 10-5 4.0 ¥ 106 384.0 1336Monel 400 8.0 ¥ 10-6 4.4 ¥ 106 35.2 1573Source: Gray et al. (27).calculated, and the temperature distribution along y = 4cm can thus bedetermined.C. Thermal Cycles and Temperature Distributions Equations (2.9) and(2.10) can be used to calculate the temperature distribution in the workpieceduring welding. The temperature distribution in the welding direction, forinstance, the T–x curves in Figures 2.15 and 2.17, are of particular interest.They can be readily converted into temperature–time plots, namely, thethermal cycles, by converting distance x into time t through t = (x - 0)/V.Figures 2.18 and 2.19 show calculated thermal cycles and temperature distributionsat the top surface (z = 0) of a thick 1018 steel at two different sets ofheat input and welding speed. The infinite peak temperature at the origin ofthe coordinate system is the result of the singularity problem in Rosenthal’ssolutions caused by the point heat source assumption.It should be mentioned, however, that Rosenthal’s analytical solutions,though based on many simplifying assumptions, are easy to use and have beengreatly appreciated by the welding industry.2.2.2 Adams’ EquationsAdams (28) derived the following equations for calculating the peak temperatureTp at the workpiece surface (z = 0) at a distance Y away from the fusionline (measured along the normal direction):(2.11)for two-dimensional heat flow and1 413 1T T0 0VYg Cp - Q T T= +-. rmANALYSIS OF HEAT FLOW IN WELDING 51semi-infinite platesat temperature T0 before weldingheat source, Qweld poolfusion zoneweldingspeed, VT(x,y,z)y = 4 cmTo10 2 4z-2T(x,y,z) shown at x = 1 cm, y = 4cm and z = 0 cmyxFigure 2.17 Three-dimensional heat flow during welding of semi-infinite workpiece.(2.12)for three-dimensional heat flow.Several other analytical solutions have also been derived for twodimensional(29–34) and three-dimensional (31, 35–37) welding heat flow.Because of the many simplifying assumptions used, analytical solutions havehad limited success.2.2.3 Computer SimulationMany computer models have been developed to study two-dimensional heatflow (e.g., refs. 38–43) and three-dimensional heat flow (e.g., refs. 44–55) duringwelding. Most assumptions of Rosenthal’s analytical solutions are no longerrequired. Figure 2.20 shows the calculated results of Kou and Le (24) for theGTAW of 3.2-mm-thick sheets of 6061 aluminum alloy. The agreement withobserved fusion boundaries and thermal cycles appears good.1 54422102T T 0kQVVYp - T T= +Êˈ¯ÈΠ͢˚ ˙+-. p aa m52 HEAT FLOW IN WELDINGFigure 2.18 Calculated results from Rosenthal’s three-dimensional heat flow equation:(a) thermal cycles; (b) isotherms. Welding speed: 2.4mm/s; heat input: 3200W;material: 1018 steel.2.3 EFFECT OF WELDING PARAMETERS2.3.1 Pool ShapeAs the heat input Q and welding speed V both increase, the weld pool becomesmore elongated, shifting from elliptical to teardrop shaped. Figure 2.21 showsthe weld pools traced from photos taken during autogenous GTAW of 304stainless steel sheets 1.6 mm thick (56). Since the pools were photographedfrom the side at an inclined angle (rather than vertically), the scale bar appliesonly to lengths in the welding direction. In each pool the cross indicates theposition of the electrode tip relative to the pool.The higher the welding speed,the greater the length–width ratio becomes and the more the geometric centerof the pool lags behind the electrode tip.Kou and Le (57) quenched the weld pool during autogenous GTAW of1.6-mm 309 stainless steel sheets and observed the sharp pool end shown in Figure2.22. The welding current was 85A, voltage 10V, and speed 4.2mm/s [10in./min(ipm)].The sharp end characteristic of a teardrop-shaped weld pool is evident.The effect of the welding parameters on the pool shape is more significantin stainless steel sheets than in aluminum sheets.The much lower thermal con-EFFECT OF WELDING PARAMETERS 53Figure 2.19 Calculated results similar to those in Figure 2.18 but with welding speedof 6.2 mm/s and heat input of 5000W.54 HEAT FLOW IN WELDINGFigure 2.20 Computer simulation of GTAW of 3.2-mm-thick 6061 aluminum, 110A,10V, and 4.23mm/s: (a) fusion boundaries and isotherms; (b) thermal cycles. From Kouand Le (24).2mm 35A, 7.4V0.42 mm/s70A, 8.4V2.5 mm/s100A, 9.0V4.2 mm/s304 stainless steel sheetsFigure 2.21 Weld pool shapes in GTAW of 304 stainless steel sheets. Courtesy ofSmolik (56).ductivity of stainless steels makes it more difficult for the weld pool to dissipateheat and solidify.2.3.2 Cooling Rate and Temperature GradientThe ratio EI/V represents the amount of heat input per unit length of weld.Lee et al. (58) measured the cooling rate in GTAW of 2024 aluminum by stickinga thermocouple into the weld pool. Figure 2.23 shows that increasing EI/VEFFECT OF WELDING PARAMETERS 55Figure 2.22 Sharp pool end in GTAW of 309 stainless steel preserved by ice quenchingduring welding. From Kou and Le (57).800(a)(b)(e)(d)(c)(f)(a) V:10 cm/min; EI/V:15.4 kJ/cm(b) V:15 cm/min; EI/V:9.79 kJ/cm(c) V:20 cm/min; EI/V:7.34 kJ/cm(d) V:30 cm/min; EI/V:4.76 kJ/cm(e) V:40 cm/min; EI/V:3.57 kJ/cm(f) V:60 cm/min; EI/V:2.38 kJ/cm0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4100200300400500600700900100011001200Time (sec)Temperature ( oC)Figure 2.23 Variation in cooling rates with heat input per unit length of weld (EI/V).Reprinted from Lee et al. (58).decreases the cooling rate (the slope). Kihara et al. (59) showed that thecooling rate decreases with increasing EI/V and preheating. Figure 2.24 showsthat the cooling rate in ESW, which is known to have a very high Q/V, is muchsmaller than that in arc welding (60). The effects of the heat input, weldingspeed, and preheat on the cooling rate and temperature gradient can be illustratedby considering the following example.Example: Bead-on-plate welding of a thick steel plate is carried out usingGTAW at 200A, 10V, and 2mm/s. Based on Rosenthal’s three-dimensionalequation, calculate the 500°C cooling rates along the x axis of the workpiecefor zero and 250°C preheating.The arc efficiency is 70% and the thermal conductivityis 35W/m°C.Along the x axis of the workpiece,y = z = 0 and R = x (2.13)Therefore, Equation (2.10) becomes(2.14)Therefore, the temperature gradient is(2.15)From the above equation and(2.16)∂∂xtVTÊˈ¯=∂∂ ppTxQx xkT TQ tÊˈ¯=-= -( - )212 202T TQkx- 0=2p56 HEAT FLOW IN WELDINGElectroslagweldArc weld0 2 4 6 8 10 12 14-1826054081510951370165019250500100015002000250030003500Time, minTemperature, oCTemperature, oFFigure 2.24 Thermal cycles of electroslag and arc welds. Reprinted from Liu et al.(60).the cooling rate is(2.17)Without preheating the workpiece before welding,(2.18)With 250°C preheating,(2.19)It is clear that the cooling rate is reduced significantly by preheating. Preheatingis a common practice in welding high-strength steels because it reducesthe risk of heat-affected zone cracking. In multiple-pass welding the interpasstemperature is equivalent to the preheat temperature T0 in single-passwelding.Equation (2.17) shows that the cooling rate decreases with increasing Q/V,and Equation (2.15) shows that the temperature gradient decreases withincreasing Q.2.3.3 Power Density DistributionFigure 2.25 shows the effect of the power density distribution of the heatsource on the weld shape (24). Under the same heat input and welding speed,weld penetration decreases with decreasing power density of the heat source.As an approximation, the power density distribution at the workpiece surfaceis often considered Gaussian, as shown by the following equation:(2.20)where q is the power density, Q the rate of heat transfer from the heat sourceto the workpiece, and a the effective radius of the heat source.2.3.4 Heat Sink Effect of WorkpieceKihara et al. (59) showed that the cooling rate increases with the thickness ofthe workpiece. This is because a thicker workpiece acts as a better heat sinkto cool the weld down. Inagaki and Sekiguchi (61) showed that, under theqQara=-ÈΠ͢˚ ˙32 32p 2exp∂∂pTt xÊˈ¯=(- ¥ ∞ ¥ ¥ )( ∞ - ∞)¥ ¥2 35 2 10- = ∞500 2500 7 200 103 202W m C m sC CA VC s.∂∂pTt xÊˈ¯=(- ¥ ∞ ¥ ¥ )( ∞ - ∞)¥ ¥2 35 2 10- = ∞500 250 7 200 103 712W m C m sC CA VC s.∂∂∂∂∂∂pTtTxxtkVT TQ x t TÊˈ¯=Êˈ¯Êˈ¯= -( - )2 02EFFECT OF WELDING PARAMETERS 57same heat input and plate thickness, the cooling time is shorter for filletwelding (a T-joint between two plates) than for bead-on-plate welding becauseof the greater heat sink effect in the former.2.4 WELD THERMAL SIMULATOR2.4.1 The EquipmentThe thermal cycles experienced by the workpiece during welding can be duplicatedin small specimens convenient for mechanical testing by using a weldthermal simulator called Gleeble, a registered trademark of Dynamic Systems.58 HEAT FLOW IN WELDINGFigure 2.25 Effect of power density distribution on weld shape in GTAW of 3.2-mm6061 aluminum with 880 W and 4.23mm/s. From Kou and Le (24).These simulators evolved from an original device developed by Nippes andSavage in 1949 (62). Figure 2.26 shows a specimen being resistance heated bythe electric current passing through the specimen and the water-cooled jawsholding it (63). A thermocouple spot welded to the middle of the specimen isconnected to a feedback control system that controls the amount of electriccurrent passing through the specimen such that a specific thermal cycle can beduplicated.2.4.2 ApplicationsThere are many applications for weld thermal simulators. For instance, aweld thermal simulator can be used in conjunction with a high-speeddilatometer to help construct continuous-cooling transformation diagramsuseful for studying phase transformations in welding and heat treating ofsteels.By performing high-speed tensile testing during weld thermal simulation,the elevated-temperature ductility and strength of metals can be evaluated.This is often called the hot-ductility test. Nippes and Savage (64, 65), forinstance, used this test to investigate the heat-affected zone fissuring inaustenitic stainless steels.Charpy impact test specimens can also be prepared from specimens (1 ¥1 cm in cross section) subjected to various thermal cycles. This syntheticspecimenor simulated-microstructure technique has been employed bynumerous investigators to study the heat-affected-zone toughness.WELD THERMAL SIMULATOR 59Figure 2.26 A weld simulator specimen held between two water-cooled jaws andresistance heated by electric current passing through it. Courtesy of Dynamic SystemsInc. (63).2.4.3 LimitationsWeld thermal simulators, though very useful, have some limitations and drawbacks.First, extremely high cooling rates during electron and laser beamwelding cannot be reproduced, due to the limited cooling capacity of the simulators.Second, because of the surface heat losses, the temperature at thesurface can be lower than that at the centerline of the specimen, especially ifthe peak temperature is high and the thermal conductivity of the specimen islow (66). Third, the temperature gradient is much lower in the specimenthan in the weld heat-affected zone, for instance, 10°C/mm, as opposed to300°C/mm near the fusion line of a stainless steel weld. This large differencein the temperature gradient tends to make the specimen microstructure differfrom the heat-affected-zone microstructure. For example, the grain size tendsto be significantly larger in the specimen than in the heat-affected zone, especiallyat high peak temperatures such as 1100°C and above.REFERENCES1. Kou, S., and Le,Y., Metall. Trans.A, 15A: 1165, 1984.2. Kou, S., and Lu, M. J., in Welding Metallurgy, 1st ed., S.Kou, 1987,Wiley, New York,p. 32; Lu, M. J., Ph.D. Thesis, Department of Materials Science and Engineering,University of Wisconsin, Madison, WI, 1988.3. Lu, M., and Kou, S., Weld. J., 67: 29s, 1988.4. Giedt,W. H., Tallerico, L. N., and Fuerschbach, P.W., Weld. J., 68: 28s, 1989.5. Fuerschbach, P.W., and Knorovsky, G. A., Weld. J., 70: 287s, 1991.6. Fuerschbach, P.W., Weld. J., 77: 76s, 1998.7. Dupont, J. N., and Marder, A. R., Weld. J., 74: 406s, 1995.8. Dupont, J. N., and Marder, A. R., Metall. Mater. Trans., 27B: 481s, 1996.9. Evans, D. M., Huang, D., McClure, J. C., and Nunes, A. C., Weld. J., 77: 53s, 1998.10. Lu, M., and Kou, S., Weld. J., 68: 452s, 1989.11. Lu, M., and Kou, S., Weld. J., 68: 382s, 1989.12. Essers,W. G., and Van Gompel, M. R. M., Weld. J., 63: 26s, 1984.13. Hasegawa, M., and Goto, H., IIW Document 212-212-71, International WeldingInstitute, London, 1971.14. Xie, J., and Kar, A., Weld. J., 78: 343s, 1999.15. Glickstein, S. S., Weld. J., 55: 222s, 1976.16. Savage,W. F., Strunk, S. S., and Ishikawa,Y., Weld. J., 44: 489s, 1965.17. Key, J. F., Weld. J., 59: 365s, 1980.18. Nestor, O. H., J. Appl. Phys., 33: 1638, 1962.19. Schoeck, P. A., in Modern Developments in Heat Transfer,Warren Ibele, Academic,New York, 1963, p. 353.20. Tsai, N., Ph.D. Thesis, Department of Materials Science and Engineering, MIT,Cambridge, MA, 1983.60 HEAT FLOW IN WELDING21. Kou, S., and Sun, D. K., Metall. Trans.A, 16A: 203, 1985.22. Oreper, G., Eagar, T.W., and Szekely, J., Weld. J., 62: 307s, 1983.23. Pavelic,V., Tan Bakuchi, L. R., Uyechara, O. A., and Myers, P. S.,Weld. J., 48: 295s,1969.24. Kou, S., and Le,Y., Metall. Trans. A, 14A: 2245, 1983.25. Rosenthal, D., Weld. J., 20: 220s, 1941.26. Abramowitz, M., and Stegun, I.A., Eds., Handbook of Mathematical Functions,National Bureau of Standards,Washington, DC, 1964.27. Gray, T. F. G., Spence, J., and North, T. H., Rational Welding Design, Newnes-Butterworth, London, 1975.28. Adams, C. M., Jr., Weld. J., 37: 210s, 1958.29. Grosh, R. J., Trabant, E. A., and Hawskins, G. A., Q. Appl. Math., 13: 161, 1955.30. Swift-Hook, D. T., and Gick, A. E. F., Weld. J., 52: 492s, 1973.31. Jhaveri, P., Moffatt,W. G., and Adams C. M., Jr., Weld. J., 41: 12s, 1962.32. Myers, P. S., Uyehara,O. A., and Borman,G. L.,Weld. Res. Council Bull., 123: 1967.33. Ghent, H.W., Hermance, C. E., Kerr, H.W., and Strong, A. B., in Proceedings of theConference on Arc Physics and Weld Pool Behavior,Welding Institute, ArbingtonHall, Cambridge, 1979, p. 389.34. Trivedi, R., and Shrinivason, S. R., J. Heat Transfer, 96: 427, 1974.35. Grosh, R. J., and Trabant, E. A., Weld. J., 35: 396s, 1956.36. Malmuth, N. D., Hall,W. F., Davis, B. I., and Rosen, C. D., Weld. J., 53: 388s, 1974.37. Malmuth, N. D., Int. J. Heat Mass Trans., 19: 349, 1976.38. Kou, S., Kanevsky, T., and Fyfitch, S., Weld. J., 62: 175s, 1982.39. Pavelic, V., and Tsao, K. C., in Proceedings of the Conference on Arc Physics andWeld Pool Behavior, Vol. 1, Welding Institute, Arbington Hall, Cambridge, 1980,p. 251.40. Kou, S., and Kanevsky, T., in Proceedings of the Conference on New Trends ofWelding Research in the United States, Ed. S. David, ASM International, MaterialsPark, OH, 1982, p. 77.41. Friedman, E., Trans. ASME; J. Pressure Vessel Techn., Series J, No. 3, 97: 206, 1965.42. Grill, A., Metall. Trans. B, 12B: 187, 1981.43. Grill, A., Metall. Trans. B, 12B: 667, 1981.44. Ushio, M., Ishmimura, T., Matsuda, F., and Arata, Y., Trans. Japan Weld. Res. Inst.,6: 1, 1977.45. Kou, S., in Proceedings of the Conference on Modeling of Casting and WeldingProcesses, Metall. Society of AIME,Warrendale, PA, 1980.46. Friedman, E., and Glickstein, S. S., Weld. J., 55: 408s, 1976.47. Friedman, E., in Numerical Modeling of Manufacturing Processes, Ed. R. F. Jones,Jr., American Society of Mechanical Engineers, New York, 1977, p. 35.48. Lewis, R.W., Morgan, K., and Gallagher, R. H., in Numerical Modeling of ManufacturingProcesses, Ed. R. F. Jones, Jr., American Society of Mechanical Engineers,New York, 1977, p. 67.49. Hsu, M. B., in Numerical Modeling of Manufacturing Processes, Ed. R. F. Jones, Jr.,American Society of Mechanical Engineers, New York, 1977, p. 97.REFERENCES 6150. Glickstein, S. S., and Friedman, E., Weld. J., 60: 110s, 1981.51. Krutz, G.W., and Segerlind, L. J., Weld. J., 57: 211s, 1978.52. Paley, Z., and Hibbert, P. D., Weld. J., 54: 385s, 1975.53. Hibbitt, H. D., and Marcal, P.V., Comput. Struct., 3: 1145, 1973.54. Mazumder, J., and Steen,W. M., J. Appl. Phys., 51: 941, 1980.55. Tsai, N. S., and Eagar, T.W., Weld. J., 62: 82s, 1983.56. Smolik, G., private communication, Idaho National Engineering Laboratories,Idaho Falls, Idaho, 1984.57. Kou, S., and Le,Y., Metall. Trans.A, 13A: 1141, 1982.58. Lee, J.Y., Park, J. M., Lee, C. H., and Yoon, E. P., in Synthesis/Processing of LightweightMetallic Materials II, Eds. C. M.Ward-Close, F. H. Froes, S. S. Cho, and D. J.Chellman,The Minerals, Metals and Materials Society,Warrendale, PA, 1996, p. 49.59. Kihara, H., Suzuki, H., and Tamura, H., Researches on Weldable High-StrengthSteels, 60th Anniversary Series,Vol. 1, Society of Naval Architects of Japan,Tokyo,1957.60. Liu, S., Brandi, S. D., and Thomas, R. D., in ASM Handbook, Vol. 6, ASM International,Materials Park, OH, 1993, p. 270.61. Inagaki, M., and Sekiguchi, H., Trans. Nat. Res. Inst. Metals, Tokyo, Japan, 2(2): 102(1960).62. Nippes, E. F., and Savage,W. F., Weld. J., 28: 534s, 1949.63. HAZ 1000, Duffers Scientific, Troy, NY, now Dynamic System, Inc.64. Nippes, E. F., Savage,W. F., Bastian, B. J., Mason, H. F., and Curran, R. M.,Weld. J.,34: 183s, 1955.65. Nippes, E. F., Savage,W. F., and Grotke, G. E.,Weld. Res. Council Bull., 33: February1957.66. Widgery, D. J., in Weld Thermal Simulators for Research and Problem Solving,Welding Institute, Cambridge, 1972, p. 14.FURTHER READING1. Rykalin,N.N., Calculation of Heat Flow in Welding, Trans. Z. Paley and C. M.Adams,Jr., Document 212-350-74, 1974, International Institute of Welding, London.2. Rosenthal, D., Weld. J., 20: 220s (1941).3. Meyers, P. S., Uyehara, O. A., and Borman, G. L., Weld. Res. Council Bull. 123:1967.4. Kou, S., Transport Phenomena and Materials Processing,Wiley, New York, 1996.PROBLEMS2.1 In one welding experiment, 50-mm-thick steel plates were joined usingelectroslag welding. The current and voltage were 480A and 34V, respectively.The heat losses to the water-cooled copper shoes and by62 HEAT FLOW IN WELDINGradiation from the surface of the slag pool were 1275 and 375 cal/s,respectively. Calculate the heat source efficiency.2.2 It has been reported that the heat source efficiency in electroslagwelding increases with increasing thickness of the workpiece. Explainwhy.2.3 (a) Consider the welding of 25.4-mm-thick steel plates. Do you preferto apply Rosenthal’s two- or three-dimensional heat flow equationfor full-penetration electron beam welds? What about bead-onplategas–tungsten arc welds?(b) Suppose you are interested in studying the solidification structureof the weld metal and you wish to calculate the temperature distributionin the weld pool. Do you expect Rosenthal’s equations toprovide reliable thermal information in the pool? Why or why not?(c) In multipass welding do you expect a higher or lower cooling ratein the first pass than in the subsequent passes? Why?2.4 Large aluminum sheets 1.6 mm thick are butt welded using GTAW withalternating current. The current, voltage, and welding speed are 100A,10V, and 2mm/s, respectively. Calculate the peak temperatures atdistance of 1.0 and 2.0 mm from the fusion boundary. Assume 50% arcefficiency.2.5 Bead-on-plate welding of a thick-section carbon steel is carried outusing 200A, 20V, and 2mm/s. The preheat temperature and arc efficiencyare 100°C and 60%, respectively. Calculate the cross-sectionalarea of the weld bead.2.6 (a) Do you expect to have difficulty in achieving steady-state heat flowduring girth (or circumferential) welding of tubes by keeping constantheat input and welding speed? Explain why.What is the consequenceof the difficulty? (b) Suggest two methods that help achieve steady-stateheat flow during girth welding.2.7 A cold-rolled AISI 1010 low-carbon steel sheet 0.6 mm thick was testedfor surface reflectivity in CO2 laser beam welding under the followingdifferent surface conditions: (a) as received; (b) oxidized in air furnaceat 1000°C for 20 s; (c) oxidized in air furnace at 1000°C for 40 s; (d)covered with steel powder. In which order does the reflectivity rank inthese surface conditions and why?2.8 It was observed in YAG laser beam welding of AISI 409 stainless steelthat under the same power the beam size affected the depth–width ratioof the resultant welds significantly. Describe and explain the effect.2.9 Calculate the thermal cycle at the top surface of a very thick carbonsteel plate at 5 mm away from the centerline of the weld surface. ThePROBLEMS 63power of the arc is 2kW, the arc efficiency 0.7, the travel speed 2mm/s,and the preheat temperature 100°C.2.10 Is the transverse cross section of the weld pool at a fixed value of xperfectly round according to Rosenthal’s three-dimensional heat flowequation? Explain why or why not based on the equation.What doesyour answer tell you about the shape of the transverse cross section ofa weld based on Rosenthal’s three-dimensional equation?64 HEAT FLOW IN WELDING3 Chemical Reactions in WeldingBasic chemical reaction during fusion welding will be described in this chapter,including gas–metal reactions and slag–metal reactions. The effect of thesechemical reactions on the weld metal composition and mechanical propertieswill be discussed.3.1 OVERVIEW3.1.1 Effect of Nitrogen, Oxygen, and HydrogenNitrogen, oxygen, and hydrogen gases can dissolve in the weld metal duringwelding. These elements usually come from air, the consumables such as theshielding gas and flux, or the workpiece such as the moist or dirt on its surface.Nitrogen, oxygen, and hydrogen can affect the soundness of the resultant weldsignificantly. Some examples of the effect of these gases are summarized inTable 3.1.3.1.2 Techniques for Protection from AirAs described in Chapter 1, various techniques can be used to protect the weldpool during fusion welding. These techniques are summarized in Table 3.2.Figure 3.1 shows the weld oxygen and nitrogen levels expected from severaldifferent arc welding processes (1, 2). As will be explained below, techniquesprovide different degrees of weld metal protection.A. GTAW and GMAW Gas–tungsten arc welding is the cleanest arc weldingprocess because of the use of inert shielding gases (Ar or He) and a short,stable arc. Gas shielding through the torch is sufficient for welding most materials.However, as shown in Figure 3.2, additional gas shielding to protect thesolidified but still hot weld is often provided both behind the torch and underthe weld of highly reactive metals such as titanium (3, 4).Welding can also beconducted inside a special gas-filled box. Although also very clean, GMAW isnot as clean as GTAW due to the less stable arc associated with the use of consumableelectrodes. Furthermore, the greater arc length in GMAW reducesthe protective effects of the shielding gas. Carbon dioxide is sometimesemployed as shielding gas in GMAW. Under the high temperature of the arc,65Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4decomposition into CO and O is favored, potentially increasing the weldoxygen level.B. SMAW The flow of gas in SMAW is not as well directed toward the weldpool as the flow of inert gas in GTAW or GMAW. Consequently, the protectionafforded the weld metal is less effective, resulting in higher weld oxygenand nitrogen levels. Carbon dioxide, produced by the decomposition of carbonateor cellulose in the electrode covering, can potentially increase the weldoxygen level.C. Self-Shielded Arc Welding Self-shielded arc welding uses strong nitrideformers such as Al, Ti, and Zr in the electrode wire alone to protect againstnitrogen. Since these nitride formers are also strong deoxidizers, the weld66 CHEMICAL REACTIONS IN WELDINGTABLE 3.1 Effect of Nitrogen, Oxygen, and Hydrogen on Weld SoundnessNitrogen Oxygen HydrogenSteels Increases strength but Reduces toughness Induces hydrogenreduces toughness but improves it if crackingacicular ferrite ispromotedAustenitic or Reduces ferrite andduplex promotes solidificationstainless steels crackingAluminum Forms oxide films Forms gas porositythat can be trapped and reduces bothas inclusions strength andductilityTitanium Increases strength but Increases strengthreduces ductility but reduces ductilityTABLE 3.2 Protection Techniques in CommonWelding ProcessesProtection Technique Fusion Welding ProcessGas Gas tungsten arc, gas metalarc, plasma arcSlag Submerged arc, electroslagGas and slag Shielded metal arc, fluxcoredarcVacuum Electron beamSelf-protection Self-shielded arcoxygen levels are slightly lower than in SMAW. Unfortunately, as can be seenin Figure 3.1, the weld nitrogen content is still rather high.D. SAW The weld oxygen level in submerged arc welds can vary significantly,depending on the composition of the flux; the very high oxygen levelsassociated with acidic fluxes containing large percentages of SiO2, accordingto Eagar (2), are the result of SiO2 decomposition. This is consistent with thelarge increase in the weld metal silicon content when acidic fluxes are used(2). If atmospheric contamination were the reason for the weld metal oxygencontent, the nitrogen content would also have been high.OVERVIEW 67Figure 3.1 Oxygen and nitrogen levels expected from several arc welding processes.From Rein (1); reproduced from Eagar (2).Backingshield(Argon)Trailingshield(Argon)WorkTorch(HP Argon)(Air-Argon)(CO2-Argon)Figure 3.2 Gas–tungsten arc welding of titanium with additional gas shielding.Reprinted from Harwig et al. (3). Courtesy of American Welding Society.3.2 GAS–METAL REACTIONSThe gas–metal reactions here refer to chemical reactions that take place at theinterface between the gas phase and the liquid metal. They include the dissolutionof nitrogen, oxygen, and hydrogen in liquid metal and the evolution ofcarbon monoxide.3.2.1 Thermodynamics of ReactionsIn steelmaking, exposure of molten steel to molecular nitrogen, N2, can resultin dissolution of nitrogen atoms in the molten steel, that is,(3.1)where the underlining bar denotes dissolution in molten metal. From thermodynamics(5), the equilibrium concentration of dissolved nitrogen, [N], atany given temperature T can be determined from the following relationship:(3.2)where KdNis the equilibrium constant for reaction (3.1) based on dissolutionfrom a diatomic gas N2, pN2 the partial pressure (in atmospheres) of N2 abovethe molten metal, DG° the standard free energy of formation (in caloriesper mole), and R the gas constant 1.987 cal/(K mol).Table 3.3 shows the valuesof DG° for several chemical reactions involving nitrogen, oxygen, and hydrogen(6–9). From Equation (3.2), the well-known Sievert law (10) for the dissolutionof a diatomic gas in molten metal can be written aslnK lnNpGRTdNN= [ ] Êˈ¯=- ∞2D12 N2(g)=N68 CHEMICAL REACTIONS IN WELDINGTABLE 3.3 Free Energy of Reactions Involving Nitrogen, Oxygen, and HydrogenFree Energy of Reaction,Gas Reaction DG° (cal/mol) ReferenceNitrogen 1/2N2(g) = N(g) 86596.0 - 15.659T (K) 6N(g) = N(wt % in steel) -85736.0 + 21.405T1/2N2(g) = N(wt % in steel) 860.0 + 5.71T 7Oxygen 1/2O2(g) = O(g) 60064 - 15.735T 6O(g) = O(wt % in steel) -88064 + 15.045 T1/2O2(g) = O(wt % in steel) -28000 - 0.69T 8Hydrogen 1/2H2(g) = H(g) 53500.0 - 14.40T 9H(g) = H(ppm in steel) -44780.0 + 3.38T 91/2H2(g) = H(ppm in steel) 8720.0 - 11.02T 8(3.3)In arc welding, however, a portion of the N2 molecules can dissociate (oreven ionize) under the high temperature of the arc plasma. The atomic N soproduced can dissolve in the molten metal as follows:N = N (3.4)(3.5)where KmN is the equilibrium constant for reaction (3.4) based on dissolutionfrom a monatomic gas N and pN the partial pressure (in atmospheres) of Nabove the molten metal.It is interesting to compare dissolution of nitrogen in molten steel frommolecular nitrogen to that from atomic nitrogen. Consider an arbitrary temperatureof 1600°C for molten steel for the purpose of discussion. Based onthe free energy of reaction DG° shown in Table 3.3, for molecular nitrogen apressure of pN2 = 1 atm is required in order to have [N] = 0.045wt %.For atomicnitrogen, however, only a pressure of pN = 2 ¥ 10-7 atm is required to dissolvethe same amount of nitrogen in molten steel.Similarly, for the dissolution of oxygen and hydrogen from O2(g) and H2(g),(3.6)(3.7)However, as in the case of nitrogen, a portion of the O2 and H2 molecules candissociate (or even ionize) under the high temperature of the arc plasma.Theatomic O and H so produced can dissolve in the molten metal as follows:O = O (3.8)H = H (3.9)DebRoy and David (11) showed, in Figure 3.3, the dissolution of monoatomic,rather than diatomic, nitrogen and hydrogen dominates in molten iron.As they pointed out, several investigators have concluded that the species concentrationin the weld metal can be significantly higher than those calculatedfrom dissolution of diatomic molecules. Dissociation of such molecules toneutral atoms and ions in the arc leads to enhanced dissolution in the moltenmetal.In the case of hydrogen the calculated results are consistent with the earlierones of Gedeon and Eagar (9) shown in Figure 3.4. The calculation is based12H2(g)=H12 O2(g)=OlnK lnNpGRT NmN= [ ] Êˈ¯=-D ∞[N]=KNd pN2GAS–METAL REACTIONS 69on a dissociation temperature of 2500°C, 0.01 atm hydrogen added to the argonshielding gas, and the pool surface temperature distribution measured byKrause (12). As shown, the majority of hydrogen absorption appears to takeplace around the outer edge of the weld pool, and monatomic hydrogenabsorption dominates the contribution to the hydrogen content. This contradictspredictions based on Sievert’s law that the maximum absorption occursnear the center of the pool surface where the temperature is highest. However,as they pointed out, the dissolution process alone does not determine thehydrogen content in the resultant weld metal. Rejection of the dissolvedhydrogen atoms by the solidification front and diffusion of the hydrogen atomsfrom the weld pool must also be considered. It is interesting to note that70 CHEMICAL REACTIONS IN WELDINGONH1,850 2,050 2,250 2,45000.020.040.060.080.100.12Temperature (K)Weight percent soluteONH1,850 2,050 2,250 2,45000.1Temperature (K)Weight percent solute0.20.3(a) (b)Figure 3.3 Equilibrium concentration of nitrogen, oxygen, and hydrogen in liquid ironas a function of temperature: (a) N2(g) with pN2 = 1atm,O2(g) with pO2 = 10-9 atm,H2(g)with pH2 = 1atm; (b) N(g) with pN = 10-6 atm, O(g) with pO = 10-8 atm, H(g) with pH =5 ¥ 10-2 atm. From DebRoy and David (11).-10 -5 0 5 10051015202530Weld pool width (mm)Hydrogen absorbed (ppm)Monatomic absorptionDiatomic absortionFigure 3.4 Equilibrium concentration of hydrogen as a function of weld pool location.Reprinted from Gedeon and Eagar (9). Courtesy of American Welding Society.Hooijmans and Den Ouden (13) suggested that a considerable amount ofhydrogen absorbed by the liquid metal during welding leaves the weld metalimmediately after the extinction of the arc.3.2.2 NitrogenFor metals that neither dissolve nor react with nitrogen, such as copper andnickel, nitrogen can be used as the shielding gas during welding. On the otherhand, for metals that either dissolve nitrogen or form nitrides (or both), suchas Fe, Ti, Mn, and Cr, the protection of the weld metal from nitrogen shouldbe considered.A. Sources of Nitrogen The presence of nitrogen in the welding zone isusually a result of improper protection against air. However, nitrogen is sometimesadded purposely to the inert shielding gas. Figure 3.5 shows the weldnitrogen content of a duplex stainless steel as a function of the nitrogen partialpressure in the Ar–N2 shielding gas (14). Nitrogen is an austenite stabilizer foraustenitic and duplex stainless steels. Increasing the weld metal nitrogencontent can decrease the ferrite content (Chapter 9) and increase the risk ofsolidification cracking (Chapter 11).B. Effect of Nitrogen The presence of nitrogen in the weld metal can significantlyaffect its mechanical properties. Figure 3.6 shows the needlelikestructure of iron nitride (Fe4N) in a ferrite matrix (15).The sharp ends of suchGAS–METAL REACTIONS 710 0.02 0.04 0.06 0.08 0.10.10.20.30.4Nitrogen partial pressure (MPa)Nitrogen content (mass%)Figure 3.5 Effect of nitrogen partial pressure in Ar–N2 shielding gas on nitrogencontent in welds of duplex stainless steel. Reprinted from Sato et al. (14).a brittle nitride act as ideal sites for crack initiation. As shown in Figure 3.7,the ductility and the impact toughness of the weld metal decrease with increasingweld metal nitrogen (15). Figure 3.8 shows that nitrogen can decrease theductility of Ti welds (4).C. Protection against Nitrogen In the self-shielded arc welding process,strong nitride formers (such as Ti, Al, Si, and Zr) are often added to the fillerwire (16). The nitrides formed enter the slag and nitrogen in the weld metalis thus reduced. As already shown in Figure 3.1, however, the nitrogen con-72 CHEMICAL REACTIONS IN WELDINGFigure 3.6 Iron nitride in a ferrite matrix (¥500). From Seferian (15).Figure 3.7 Effect of nitrogen on the room temperature mechanical properties of mildsteel welds. From Seferian (15).tents of self-shielded arc welds can still be rather high, and other arc weldingprocesses such as GTAW, GMAW, or SAW should be used if weld nitrogencontamination is to be minimized.3.2.3 OxygenA. Sources of Oxygen Oxygen in the weld metal can come from the air, theuse of excess oxygen in oxyfuel welding, and the use of oxygen- or CO2-containing shielding gases. It can also come from the decomposition of oxides(especially SiO2 and MnO and FeO) in the flux and from the slag–metal reactionsin the weld pool, which will be discussed subsequently.In GMAW of steels the addition of oxygen or carbon dioxide to argon (e.g.,Ar–2% O2) helps stabilize the arc, reduce spatter, and prevent the filler metalfrom drawing away from (or not flowing out to) the fusion line (17). Carbondioxide is widely used as a shielding gas in FCAW, the advantages being lowcost, high welding speed, and good weld penetration. Baune et al. (18) pointedout that CO2 can decompose under the high temperature of the welding arcas follows:(3.10)(3.11)B. Effect of Oxygen Oxygen can oxidize the carbon and other alloyingelements in the liquid metal, modifying their prevailing role, depressinghardenability, and producing inclusions. The oxidation of carbon is asfollows:C+O=CO(g) (3.12)CO(g)=C(s)+ 1O (g)2 2CO2g CO g O2g12( )= ( )+ ( )GAS–METAL REACTIONS 73.03Fe, 10 C/sec.03Fe, 28 C/sec.07Fe, 15 C/sec.07Fe, 10 C/sec.11Fe, 11 C/secOE = 2C+3.5N+O-0.14Fe0.15 0.2 0.25 0.3 0.35 0.4 0.4505101520253035% ElongationoooooFigure 3.8 Effect of oxygen equivalence (OE) on ductility of titanium welds.Reprinted from Harwig et al. (4). Courtesy of American Welding Society.The oxidation of other alloying elements, which will be discussed subsequentlyin slag–metal reactions, forms oxides that either go into the slag orremain in the liquid metal and become inclusion particles in the resultantweld metal.Table 3.4 shows the effect of gas composition in oxyacetylene welding ofmild steel on the weld metal composition and properties (15).When too muchoxygen is used, the weld metal has a high oxygen level but low carbon level.On the other hand, when too much acetylene is used, the weld metal has a lowoxygen level but high carbon level (the flame becomes carburizing). In eithercase, the weld mechanical properties are poor. When the oxygen–acetyleneratio is close to 1, both the impact toughness and strength (proportional tohardness) are reasonably good.If oxidation results in excessive inclusion formation in the weld metal orsignificant loss of alloying elements to the slag, the mechanical properties ofthe weld metal can deteriorate. Figure 3.9 shows that the strength, toughness,and ductility of mild steel welds can all decrease with increasing oxygen contamination(15). In some cases, however, fine inclusion particles can act asnucleation sites for acicular ferrite to form and improve weld metal toughness(Chapter 9). For aluminum and magnesium alloys, the formation of insolubleoxide films on the weld pool surface during welding can cause incompletefusion. Heavy oxide films prevent a keyhole from being established properlyin conventional PAW of aluminum, and more advanced DC variable-polarityPAW has to be used. In the latter, oxide films are cleaned during the electrodepositivepart of the current cycle.Bracarense and Liu (19) discovered in SMAW that the metal transferdroplet size can increase gradually during welding, resulting in increasing Mn74 CHEMICAL REACTIONS IN WELDINGTABLE 3.4 Effect of Oxygen–Acetylene Ratio on Weld Metal Composition andProperties of Mild Steela = O2/C2H2 > 1 a £ 1
Before a = 1.14 a = 1.33 a = 2 a = 2.37 a = 1 a = 0.82
C 0.155 0.054 0.054 0.058 0.048 0.15 1.56
Mn 0.56 0.38 0.265 0.29 0.18 0.29 0.375
Si 0.03
S 0.030
P 0.018
O — 0.04 0.07 0.09 — 0.02 0.01
N — 0.015 0.023 0.030 — 0.012 0.023
Impact value, — 5.5 1.40 1.50 1.30 6.9 2.3
kg/cm2
Hardness, HB — 130 132 115 100 140 320
Grain size — 6 5 4 4 4 5
Source: Seferian (15).
and Si transfer to the weld pool and hence increasing weld metal hardness
along the weld length, as shown in Figure 3.10. As the electrode is heated up
more and more during welding, the droplet size increases gradually (becomes
more globular).This reduces the surface area (per unit volume) for oxygen to
react with Mn and Si and hence improves the efficiency of Mn and Si transfer
to the weld pool.
3.2.4 Hydrogen
A. Steels The presence of hydrogen during the welding of high-strength
steels can cause hydrogen cracking (Chapter 17).
A.1. Sources of Hydrogen Hydrogen in the welding zone can come from
several different sources: the combustion products in oxyfuel welding; decomposition
products of cellulose-type electrode coverings in SMAW; moisture or
grease on the surface of the workpiece or electrode; and moisture in the flux,
electrode coverings, or shielding gas.
As mentioned previously in Chapter 1, in SMAW high-cellulose electrodes
contain much cellulose, (C6H10O5)x, in the electrode covering. The covering
decomposes upon heating during welding and produces a gaseous shield rich
in H2, for instance, 41% H2, 40% CO, 16% H2O, and 3% CO2 in the case of
E6010 electrodes (20). On the other hand, low-hydrogen electrodes contain
much CaCO3 in the electrode covering. The covering decomposes during
welding and produces a gaseous shield low in H2, for example, 77% CO, 19%
CO2, 2% H2, and 2% H2O in the case of E6015 electrodes. As such, to reduce
weld metal hydrogen, low-hydrogen electrodes should be used.
GAS–METAL REACTIONS 75
Figure 3.9 Effect of the oxygen content on the mechanical properties of mild steel
welds. From Seferian (15).
A.2. Measuring Hydrogen Content Various methods have been developed
for measuring the hydrogen content in the weld metal of steels. The mercury
method and the gas chromatography method are often used. Figure 3.11 shows
the mercury method (21). A small test specimen (13 ¥ 25 ¥ 127mm, or 1/2 ¥ 1
¥ 5 in.) is welded in a copper fixture. The welded test specimen is then
immersed in mercury contained in a eudiometer tube. As hydrogen diffuses
out of the welded test specimen, the mercury level in the eudiometer tube continues
to drop. From the final mercury level, H (in millimeters), the amount
of hydrogen that diffuses out of the specimen, that is, the so-called diffusible
hydrogen, can be measured. This method, however, can take days because of
the slow diffusion of hydrogen at room temperature. In the gas chromatography
method (22), the specimen is transferred to a leak-tight chamber after
76 CHEMICAL REACTIONS IN WELDING
430
420
410
400
390
380
100
80
60
40
20
0 20 40 60 80 100
weld length position, mm
120 140 160
[Mn] and [Si] x 10-2, wt %
manganese
silicon
[O], ppm
oxygen
(a)
40
0 20 40 60 80 100
weld length position, mm
120
Hardness, HRC
30
50
(b)
Figure 3.10 Variations of weld metal along the length of a weld made with an E7018
electrode: (a) composition; (b) hardness. Reprinted from Bracarense and Liu (19).
Courtesy of American Welding Society.
welding, which can be heated to accelerate the hydrogen evolution from the
specimen. After that, the chamber can be connected to a gas chromatograph
analyzer to measure the total amount of hydrogen present. The advantages
are that it can separate other gases present and measure only hydrogen and
it takes hours instead of days. One disadvantage is the relatively high cost of
the equipment.
Newer methods have also been developed. Albert et al. (23) developed a
new sensor for detecting hydrogen. The sensor is a conducting polymer film
coated with Pd on one side to be exposed to a hydrogen-containing gas and
the other side to air.The current going through the sensor is directionally proportional
to the hydrogen content in the gas. Figure 3.12 shows the hydrogen
contents measured by the new sensor as well as gas chromatography (GC).
These are gas–tungsten arc welds of a 0.5Cr–0.5Mo steel made with Ar–H2 as
the shielding gas. Smith et al. (24) developed a new hydrogen sensor that generates
results in less than 1 h and allows analysis to be done on the actual
welded structure. The sensor is a thin porous film of tungsten oxide, which
changes color upon reacting with hydrogen.
GAS–METAL REACTIONS 77
Figure 3.11 Mercury method for measuring diffusible hydrogen in welds. Reprinted
from Shutt and Fink (21). Courtesy of American Welding Society.
A.3. Hydrogen Reduction Methods The weld hydrogen content can be
reduced in several ways. First, avoid hydrogen-containing shielding gases,
including the use of hydrocarbon fuel gases, cellulose-type electrode coverings,
and hydrogen-containing inert gases. Second, dry the electrode covering
and flux to remove moisture and clean the filler wire and workpiece to remove
grease. Figure 3.13 shows the effect of the electrode baking temperature on
the weld metal hydrogen content (25). Third, adjust the composition of the
consumables if feasible. Figure 3.14 shows that CO2 in the shielding gas helps
reduce hydrogen in the weld metal (22), possibly because of reaction between
the two gases. Increasing the CaF2 content in the electrode covering or the flux
78 CHEMICAL REACTIONS IN WELDING
gas chromatography
sensor
Concentration of hydrogen in shielding
gas (%)
Hydrogen in weld metal (ml/100cc)
0.0 1.0 2.0 3.0 4.0 5.0 6.0
15.0
25.0
35.0
45.0
55.0
Figure 3.12 Hydrogen content in gas–tungsten arc welds of 0.5Cr–0.5Mo steel as a
function of volume percent of hydrogen in Ar–H2 shielding gas. Reprinted from Albert
et al. (23). Courtesy of American Welding Society.
As baked
Exposed at 80%
relative humidity at
28oC for 24 h
Baking temperature oC
Diffusible hydrogen
ml/100g deposited metal
0
0
100 200 300 400 500
5
10
15
20
25
30
Figure 3.13 Effect of electrode baking temperature on weld metal diffusible hydrogen
levels. Reprinted from Fazackerley and Gee (25).
has been reported to reduce the weld hydrogen content. This reduction in
hydrogen has been ascribed to the reaction between hydrogen and CaF2 (26).
Fourth, as shown in Figure 3.15, use postweld heating to help hydrogen diffuse
out of the weld (27).
GAS–METAL REACTIONS 79
(a)
(b)
Ar Ar+5%CO2
Ar+20%CO2CO2
Ar
Ar+5%CO2
Ar+20%CO2CO2
Shielding gas
Diffusible Hydrogen (mL/100g)
0
1
2
3
0
2
4
6
8
10
deposited metal
fused metal
deposited metal
fused metal
Figure 3.14 Effect of shielding gases on weld metal hydrogen content: (a) GMAW;
(b) FCAW. Reprinted from Mirza and Gee (22).
Figure 3.15 Effect of postweld heating on the weld metal hydrogen content of mild
steel. Reprinted from Flanigan (27).
B. Aluminum Hydrogen can cause porosity in aluminum welds (Chapter
10). The oxide films on the surface of the workpiece or electrode can absorb
moisture from the air and introduce hydrogen into molten aluminum during
welding. Grease on the surface of the workpiece or electrode and moisture in
shielding gas can also be the sources of hydrogen. Figure 3.16 shows the solubility
of hydrogen in aluminum (28). Since the solubility of hydrogen is much
higher in liquid aluminum than in solid aluminum, hydrogen is rejected into
the weld pool by the advancing solid–liquid interface. Consequently, hydrogen
porosity is often observed in aluminum welds. Devletian and wood (29) have
reviewed the factors affecting porosity in aluminum welds.
B.1. Effect of Hydrogen Porosity As shown in Figure 3.17, excessive hydrogen
porosity can severely reduce both the strength and ductility of aluminum
welds (30). It has also been reported to reduce the fatigue resistance of
aluminum welds (31).
B.2. Reducing Hydrogen Porosity To reduce hydrogen porosity, the surface
of Al–Li alloys has been scrapped, milled, or even thermovacuum degassed to
remove hydrogen present in the form of hydrides or hydrated oxides (32, 33).
Similarly, Freon (CCl2F2) has been added to the shielding gas to reduce hydrogen
in aluminum welds. The weld pool has been magnetically stirred to help
hydrogen bubbles escape and thus reduce hydrogen porosity (34). Keyhole
plasma arc welding, with variable-polarity direct current, has been used to
reduce hydrogen porosity in aluminum welds (35). The cleaning action of the
DCEP cycle helps remove hydrated oxides and hydrides. The keyhole, on the
other hand, helps eliminate entrapment of oxides and foreign materials in
the weld, by allowing contaminants to enter the arc stream instead of
being trapped in the weld. Consequently, the welds produced are practically
porosity free.
80 CHEMICAL REACTIONS IN WELDING
Figure 3.16 Solubility of hydrogen in aluminum. From Eastwood (28).
Consider the reduction of severe gas porosity in the welding of a highstrength,
lightweight Al–5Mg–2Li alloy (33, 36).The presence of Li in the alloy
promotes the formation of lithium hydride during heat treatment as well as
the hydration of the surface oxide at room temperature. Surface cleaning and
thermovacuum treatment before welding, which help reduce the hydrogen
content of the workpiece surface, have been reported to reduce the porosity
level of the weld metal. Reduction in porosity has also been achieved by using
an alternating magnetic field to stir the weld pool and by using variablepolarity
keyhole PAW (33).
Consider also the reduction of severe gas porosity in the welding of PM
(powder metallurgy) parts of Al–8.0Fe–1.7Ni alloy (33, 36). Oxidation and subsequent
hydration of the aluminum powder during and after powder production
by air atomization result in a high surface moisture content. When the
powder is consolidated into PM parts, the moisture is trapped inside the parts.
Because of the difficulty in removing the moisture from deep inside the workpiece,
thermovacuum treatments at temperatures as high as 595°C have been
found necessary. Unfortunately, the use of such a high temperature causes
unacceptable degradation in the base-metal strength. Therefore, atomization
and consolidation techniques that minimize powder oxidation and hydration
are required for producing porosity-free welds.
GAS–METAL REACTIONS 81
Figure 3.17 Effect of porosity on tensile properties of aluminum welds. From Shore
(30).
C. Copper Hydrogen can also cause problems in copper welding. It can react
with oxygen to form steam, thus causing porosity in the weld metal. It can also
diffuse to the HAZ and react with oxygen to form steam along the grain
boundaries.This can cause microfissuring in the HAZ.These problems can be
minimized if deoxidized copper is used for welding.
3.3 SLAG–METAL REACTIONS
3.3.1 Thermochemical Reactions
The thermochemical slag–metal reactions here refer to thermochemical reactions
that take place at the interface between the molten slag and the liquid
metal. Examples of such reactions are decomposition of metal oxides in the
flux, oxidation of alloying elements in the liquid metal by the oxygen dissolved
in the liquid metal, and desulfurization of the weld metal.
A. Decomposition of Flux In studying SAW, Chai and Eagar (37) suggested
that in the high-temperature environment near the welding plasma, all oxides
are susceptible to decomposition and produce oxygen. It was found that the
stability of metal oxides during welding decreases in the following order: (i)
CaO, (ii) K2O, (iii) Na2O and TiO2, (iv) Al2O3, (v) MgO, and (vi) SiO2 and MnO
(FeO was not included but can be expected to be rather unstable, too). For
instance, SiO2 and MnO can decompose as follows (2):
(3.13)
(3.14)
It was concluded that in fluxes of low FeO content (<10% FeO), SiO2 andMnO are the primary sources of oxygen contamination and the stability ofmetal oxides in welding is not directly related to their thermodynamic stability.It was also concluded that CaF2 reduces the oxidizing potential of weldingfluxes due to dilution of the reactive oxides by CaF2 rather than to reactivityof the CaF2 itself and significant losses of Mn may occur by evaporation fromthe weld pool due to the high vapor pressure of Mn.B. Oxidation by Oxygen in Metal(3.15)(3.16)(3.17)2Al+3O=(Al2O3) (3.18)Ti+2O=(TiO2)Si+2O=(SiO2)Mn+O=(MnO)(MnO)= Mn(g)+ 1O (g)2 2SiO2 SiO g O2g12( )= ( )+ ( )82 CHEMICAL REACTIONS IN WELDINGC. Desulfurization of Liquid Metal(3.19)3.3.2 Effect of Flux on Weld Metal CompositionBurck et al. (38) welded 4340 steel by SAW with manganese silicate fluxes,keeping SiO2 constant at 40wt % and adding CaF2, CaO, and FeO separatelyat the expense of MnO. Figure 3.18 shows the effect of such additions on theextent of oxygen transfer from the flux to the weld metal, expressed in D(weldmetal oxygen). A positive D quantity means transfer of an element (oxygen inthis case) from the flux to the weld metal, while a negative D quantity meansloss of the element from the weld metal to the flux. The FeO additions, at theexpense of MnO, increase the extent of oxygen transfer to the weld metal.Thisis because FeO is less stable than MnO and thus decomposes and producesoxygen in the arc more easily than MnO. The CaO additions at the expenseof MnO decrease the extent of oxygen transfer to the weld metal becauseCaO is more stable than MnO.The CaF2 additions at the expense of MnO alsodecrease the extent of oxygen transfer to the weld metal but more significantly.It is worth noting that Chai and Eagar (37) reported previously thatS +(CaO)=(CaS)+OSLAG–METAL REACTIONS 83FeOCaOCaF2-SiO2-MnO-CaF2 Flux-SiO2-MnO-CaO Flux AISI 4340-SiO2-MnO-FeO FluxSiO2 = 40 w/o (constant)0 10 20 30 40Flux addition (wt.%)60 50 40 30 20MnO in the flux (wt.%)0.00.020.040.060.08Delta weld metal oxygen (wt.%)Figure 3.18 Effect of flux additions to manganese silicate flux on extent of oxygentransfer to the weld metal in submerged arc welding of 4340 steel. Reprinted fromBurck et al. (38). Courtesy of American Welding Society.CaF2 reduces oxygen transfer by acting as a diluent rather than an activespecies.Figure 3.19 shows the effect of flux additions on the manganese changeof the weld metal (38). It is surprising that the CaO additions at the expenseof MnO do not decrease the extent of manganese transfer from the flux to theweld metal. From the steelmaking data shown in Figure 3.20, it appears thatthe CaO additions do not reduce the activity of MnO (39), as indicated by thedots along the constant MnO activity of about 0.30.The additions of FeO andCaF2 at the expense of MnO decrease the extent of manganese transferfrom the flux to the weld metal, as expected. Beyond 20% FeO, D(weldmetal manganese) becomes negative, namely, Mn is lost from the weld metalto the slag. This is likely to be caused by the oxidation of Mn by the oxygenintroduced into the liquid metal from FeO, namely, Mn + O = (MnO).The fluxadditions also affect the extents of loss of alloying elements such as Cr, Mo,and Ni.Therefore, the flux composition can affect the weld metal composition andhence mechanical properties rather significantly.The loss of alloying elementscan be made up by the addition of ferroalloy powder (e.g., Fe–50% Si andFe–80% Mn) to SAW fluxes or SMAW electrode coverings. In doing so, the84 CHEMICAL REACTIONS IN WELDINGFeOCaOCaF2-SiO2-MnO-CaF2 Flux-SiO2-MnO-CaO Flux AISI 4340-SiO2-MnO-FeO FluxSiO2 = 40 w/o (constant)0 10 20 30 40Flux addition (wt.%)60 50 40 30 20MnO in the flux (wt.%)-0.20.00.20.40.6Delta weld metal manganese (wt.%)Figure 3.19 Effect of flux additions to manganese silicate flux on extent of manganesetransfer to the weld metal in submerged arc welding of 4340 steel. Reprinted fromBurck et al. (38). Courtesy of American Welding Society.alloying element recovery, that is, the percentage of the element transferredacross the arc and into the weld metal, should be considered. The recoveryvaries significantly from element to element. In SMAW, for example, it can beabout 100% for Ni and Cr, 75% for Mn, 70% for Nb, 45% for Si, and 5% forTi (40).3.3.3 Types of Fluxes, Basicity Index, and Weld Metal PropertiesThe use of proper welding fluxes during fusion welding helps control the compositionof the weld metal as well as protect it from air.Welding fluxes can becategorized into the following three groups according to the types of mainconstituents (26):(a) Halide-type fluxes: for example, CaF2–NaF, CaF2–BaCl2–NaF,KC1–NaCl–Na3AlF6, and BaF2–MgF2–CaF2–LiF.(b) Halide–oxide-type fluxes: for example, CaF2–CaO–Al2O3,CaF2–CaO–SiO2, CaF2–CaO–Al2O3–SiO2, and CaF2–CaO–MgO–Al2O3.(c) Oxide-type fluxes: for example, MnO–SiO2, FeO–MnO–SiO2, andCaO–TiO2–SiO2.The halide-type fluxes are oxygen free and are used for welding titaniumand aluminum alloys (26, 41). The halide–oxide-type fluxes, which are slightlyoxidizing, are often used for welding high-alloy steels. The oxide-type fluxes,which are mostly oxidizing, are often used for welding low-carbon or low-alloysteels.When oxide-type fluxes are used for welding a reactive metal such astitanium, the weld metal can be contaminated with oxygen.The oxides in a welding flux can be roughly categorized into the followingthree groups (26):SLAG–METAL REACTIONS 85SiO2SiO2 saturationXSiO2XMnOXCaOMnO saturationlime and limesilicate saturation0.060.130.170.210.250.37 0.41 0.590.67CaO 0.2 0.4 0.6 MnO0.60.4 0.60.4Figure 3.20 Activity of MnO in CaO–MnO–SiO2 melts at 1500°C. Reprinted fromBurck et al. (38). Courtesy of American Welding Society.(a) Acidic oxides, in the order of decreasing acidity: SiO2, TiO2, P2O5,V2O5.(b) Basic oxides, in the order of decreasing basicity:K2O, Na2O, CaO,MgO,BaO, MnO, FeO, PbO, Cu2O, NiO.(c) Amphoteric oxides: Al2O3, Fe2O3, Cr2O3, V2O3, ZnO.Oxides that are donors of free oxide ions, O2-, are considered as basicoxides, CaO being the most well known example. Oxides that are acceptors ofO2- are considered as acidic oxides, SiO2 being the most well known example.Oxides that are neutral are considered as amphoteric oxides.3.3.4 Basicity IndexThe concept of the basicity index (BI) was adopted in steelmaking toexplain the ability of the slag to remove sulfur from the molten steel. Itwas later broadened to indicate the flux oxidation capability. The BI of a flux(especially an oxide-type one) can be defined in the following general form(42):(3.20)The concept of the BI was applied to welding. Tuliani et al. (43) used thefollowing well-known formula for the fluxes in SAW:(3.21)where components are in weight fractions. Using the above expression, theflux is regarded as acidic when BI < 1, as neutral when 1.0 < BI < 1.2, and asbasic when BI > 1.2. The formula correlates well with the oxygen content in
submerged arc welds.
Eagar and Chai (44, 45), however, modified Equation (3.21) by considering
CaF2 as neutral rather than basic and omitting the CaF2 term. As shown in
Figure 3.21, the formula correlates well with the oxygen content in submerged
arc welds (44, 45).The oxygen content decreases as the basicity index increases
up to about 1.25 and reaches a constant value around 250 ppm at larger basicity
values.
Baune et al. (18, 46) modified Equation (3.21) for FCAW electrodes by
using the composition of the solidified slag after welding rather than the composition
of the flux before welding. Furthermore, mole fraction was used rather
than weight fraction, and FeO was replaced by Fe2O3. The composition of the
solidified slag was thought to provide more information about the extent of
BI
CaF CaO MgO BaO SrO
Na O K O Li O MnO FeO
SiO Al O TiO ZrO
=
+ + + +
+ + + + ( + )
+ ( + + )
2
2 2 2
2 2 3 2 2
0 5
0 5
.
.
BI
basic oxides
nonbasic oxides
=
( )
( )
Â
Â
%
%
86 CHEMICAL REACTIONS IN WELDING
SLAG–METAL REACTIONS 87
Figure 3.21 Weld metal oxygen content in steel as a function of flux basicity in
submerged arc welding. From Chai and Eagar (45).
Basicity index based on
measured composition of solidified slag
515
510
505
500
495
490
485
480
475
470
465
1.9 1.95 2 2.05 2.1 2.15
Weld metal oxygen
content(ppm)
Wire #1
Bl=1.94
Wire #2
Bl=1.976
Wire #3
Bl=1.982
Wire #4
Bl=2.02
Wire #5
Bl=2.13
[o]=-199.69B.l.+895.37
R2 =0.9941
Figure 3.22 Weld metal oxygen content in steel as a function of flux basicity in fluxcore
arc welding. Reprinted from Baune et al. (46). Courtesy of American Welding
Society.
the slag–metal reactions during welding than the composition of the flux
before welding. The Fe2O3 was thought to be the iron oxide that forms in a
welding slag. As shown in Figure 3.22, the weld oxygen content correlates well
with the new basicity index (46).
Oxide inclusions in steel welds can affect the formation of acicular ferrite,
which improves the weld metal toughness (47–50). It has been reported that
acicular ferrite forms in the range of about 200–500 ppm oxygen (51–53).This
will be discussed later in Chapter 9.
For SAW of high-strength, low-alloy steels with the CaF2–CaO–SiO2 flux
system, Dallam et al. (47) used the following simple formula for the basicity
index:
(3.22)
Figure 3.23 shows the effect of the basicity index on sulfur transfer (47).
As the basicity index increases from 0 to 5, Dsulfur becomes increasingly
negative, namely, more undesirable sulfur is transferred from the weld metal
to the slag (desulfurization). This is because CaO is a strong desulfurizer,
as shown previously by Equation (3.19).
Excessive weld metal oxygen and hence oxide inclusions can deteriorate
weld metal mechanical properties. As shown in Figure 3.24, some inclusion
BI
CaO
SiO
=
2
88 CHEMICAL REACTIONS IN WELDING
-0.008
-0.012
-0.020
-0.016
0 1 2 3 4 5
CaO/SiO2
sulfur (wt%)
-0.004
∞
Δ
Figure 3.23 Desulfurization of high-strength, low-alloy steel welds as a function of
basicity index CaO/SiO2 of CaF2–CaO–SiO2 type flux. Reprinted from Dallam et al.
(47). Courtesy of American Welding Society.
Figure 3.24 Fracture initiation at an inclusion in flux-cored arc weld of high-strength,
low-alloy steel. Reprinted from Bose et al. (54).
particles can act as a fracture initiation site (54). Besides oxide inclusions,
oxygen in the weld pool can also react with carbon to form CO gas during
solidification. As shown in Figure 3.25, this can result in gas porosity in steel
welds (55). The addition of deoxidizers such as Al, Ti, Si, and Mn in the filler
metal helps reduce the amount of porosity. Figure 3.26 shows that the
toughness of the weld metal decreases with increasing oxygen content (56).
However, if the content of the acicular ferrite in the weld metal increases with
the weld oxygen content, the weld metal toughness may in fact increase
(Chapter 9).
Basic fluxes, however, can have some drawbacks. They are often found to
have a greater tendency to absorb moisture, which can result in hydrogen
embrittlement unless they are dried before welding.The slag detachability may
SLAG–METAL REACTIONS 89
Figure 3.25 Wormhole porosity in weld metal. From Jackson (55).
Figure 3.26 Relationship between the toughness at 20°C and the oxygen content of
steel welds. Reprinted from North et al. (56). Courtesy of American Welding Society.
not be very good in a fully basic flux. This makes slag removal more difficult,
especially in multiple-pass or narrow-groove welding. In the case of FCAW,
basic fluxes have also been observed to generate unstable arcs (57).
3.3.5 Electrochemical Reactions
Kim et al. (58, 59) studied the effect of electrochemical reactions on the weld
metal composition in SAW, and significant composition differences were
observed when the electrode polarity was varied in DC welding. The following
anodic oxidation reactions were proposed:
(3.23)
(3.24)
These reactions occur at the electrode tip–slag interface in the electrodepositive
polarity or the weld pool–slag interface in the electrode-negative
polarity.Therefore, oxidation losses of alloying elements and pickup of oxygen
are expected at the anode.
The following cathodic reduction reactions were also proposed:
(3.25)
(3.26)
(3.27)
The first two reactions are the reduction of metallic cations from the slag,
and the third reaction is the removal (refining) of oxygen from the metal.
These reactions occur at the electrode tip–slag interface in the electrodenegative
polarity or the weld pool–slag interface in the electrode-positive
polarity.The current density is much higher at the electrode tip–slag interface
than at the weld pool–slag interface. Therefore, reactions at the electrode tip
may exert a greater influence on the weld metal composition than those at the
weld pool.
A carbon steel containing 0.18% C, 1.25% Mn, and 0.05% Si was submerged
arc welded with a low-carbon steel wire of 0.06% C, 1.38% Mn, and
0.05% Si and a flux of 11.2% SiO2, 18.14% Al2O3, 33.2% MgO, 25.3% CaF2,
6.9% CaO, and 1.2% MnO. Figure 3.27 shows the oxygen contents of the
melted electrode tips and the detached droplets for both polarities, the 20ppm
oxygen content of the wire being included as a reference (58). A significant
oxygen pickup in the electrode tips is evident for both polarities, suggesting
that the excess oxygen came from decomposition of oxide components in the
flux and the surrounding atmosphere.The anodic electrode tip has about twice
as much oxygen as the cathodic electrode tip, suggesting the significant effect
of electrochemical reactions. The difference in the oxygen content is due to
O(metal)+2e-=O2-(slag)
Si4+(slag)+4e-=Si(metal)
M2+(slag)+2e-=M(metal)
O2-(slag)=O(metal)+ 2e-
M(metal)+nO2-(slag)= MOn(slag)+ 2ne-
90 CHEMICAL REACTIONS IN WELDING
oxygen pickup at the anode and oxygen removal (refining) at the cathode. In
either polarity the detached droplets contain more oxygen than the melted
electrode tip. This suggests that the electrochemical reactions cease after the
droplets separate from the electrode tips, but the droplets pick up more oxygen
from decomposed oxides while falling through the arc plasma. The higher
oxygen content in the droplets separated from the anodic electrode tip is due
to the higher oxygen content of the melted anodic electrode tip.
Figure 3.28 shows the silicon change in the weld pool for both polarities
(58). Since there is plenty of oxygen from the decomposition of oxides, loss
of silicon from the weld pool due to the thermochemical reaction Si + 2O =
(SiO2) is likely. But the silicon loss is recovered by the cathodic reduction
(Si4+) + 4e- = Si and worsened by the anodic oxidation Si + 2(O2-) = (SiO2) +
4e-. Consequently, the silicon loss is more significant at the anode than at the
cathode.
Figure 3.29 shows the manganese change in the weld pool for both polarities
(58). The manganese loss from the weld pool to the slag is significant in
both cases. This is because manganese is the richest alloying element in the
weld pool (judging from the compositions of the workpiece and the filler
metal) and MnO is among the poorest oxides in the flux. As such, the thermochemical
reaction Mn + O = (MnO) can shift to the right easily and cause
much manganese loss. Since manganese is known to have a high vapor pressure
(Chapter 4), evaporation from the liquid can be another reason for much
manganese loss. However, the manganese loss is partially recovered by the
cathodic reduction (Mn2+) + 2e- = Mn and worsened by the anodic oxidation
Mn + (O2-) = (MnO) + 2e-. Consequently, the manganese loss is more significant
at the anode than at the cathode.
SLAG–METAL REACTIONS 91
Oxygen Concentration (ppm)
0
200
400
600
800
1000
1200
Wire Electrode
Tip
Droplet
Anode
Wire
Cathode
Figure 3.27 Oxygen contents of the welding wire, melted electrode tips, and detached
droplets for both electrode-positive and electrode-negative polarities. Reprinted from
Kim et al. (58). Courtesy of American Welding Society.
REFERENCES
1. Rein, R. H., in Proceedings of a Workshop on Welding Research Opportunities, Ed.
B. A. McDonald, Office of Naval Research,Washington, DC, 1974, p. 92.
92 CHEMICAL REACTIONS IN WELDING
Anode
Cathode
Silicon change in weld pool
0.0 0.8 1.6 2.4 3.2
Travel speed (cm/s)
-0.15
-0.10
-0.05
0.00
0.05
0.10
0.15
silicon (wt.%) Δ
Figure 3.28 Gain or loss of weld metal silicon due to reactions in weld pool for
electrode-positive and electrode-negative polarities as a function of welding speed.
Reprinted from Kim et al. (58). Courtesy of American Welding Society.
Anode
Cathode
Manganese change in weld pool
0.0 0.8 1.6 2.4 3.2
Travel speed (cm/s)
-0.35
-0.30
-0.25
-0.20
-0.15
-0.10
-0.15
manganese (wt.%)
0.00
Δ
Figure 3.29 Loss of weld metal manganese due to reactions in weld pool for electrode-
positive and electrode-negative polarities as a function of welding speed.
Reprinted from Kim et al. (58). Courtesy of American Welding Society.
2. Eagar, T.W., in Weldments: Physical Metallurgy and Failure Phenomena, Proceedings
of the Fifth Bolton Landing Conference, General Electric Co., Schenectady,
NY, 1979, p. 31.
3. Harwig,D.D., Fountain,C., Ittiwattana,W., and Castner, H.,Weld. J., 79: 305s, 2000.
4. Harwig, D. D., Ittiwattana,W., and Castner, H., Weld. J., 80: 126s, 2001.
5. Darken, L. S., and Gurry, R.W., Physical Chemistry of Metals, McGraw-Hill, New
York, 1953.
6. Elliott, J. F., and Gleiser, M., Thermochemistry for Steelmaking, Vol. I, Addison
Wesley, Reading, MA, 1960.
7. Pehlke, R. D., and Elliott, J. F., Trans. Metall. Soc. AIME., 218: 1091, 1960.
8. Elliott, J. F., Gleiser, M., and Ramakrishna, V., Thermochemistry for Steelmaking,
Vol. II, Addison Wesley, Reading, MA, 1963.
9. Gedeon, S. A., and Eagar, T.W., Weld. J., 69: 264s, 1990.
10. Pehlke, R. D., Unit Processes in Extractive Metallurgy, Elsevier, New York, 1979.
11. DebRoy, T., and David, S. A., Rev. Modern Phys., 67: 85, 1995.
12. Krause, H. G., Weld. J., 66: 353s, 1987.
13. Hooijmans, J.W., and Den Ouden, G., Weld. J., 76: 264s, 1997.
14. Sato, Y. S., Kokawa, H., and Kuwana, T., in Trends in Welding Research, Eds. J. M.
Vitek, S. A. David, J. A. Johnson, H. B. Smartt, and T. DebRoy,ASM International,
Materials Park, OH, June 1998, p. 131.
15. Seferian, D., The Metallurgy of Welding, Wiley, New York, 1962.
16. Atarashi, N., Welding Metallurgy, Maruzen, Tokyo, 1972 (in Japanese).
17. Welding Handbook, Vol. 2, 7th ed., American Welding Society, Miami, FL, 1978,
p. 134.
18. Baune, E., Bonnet, C., and Liu, S., Weld. J., 79: 57s, 2000.
19. Bracarense, A. Q., and Liu, S., Weld. J., 72: 529s, 1993.
20. Christensen, N., Welding Metallurgy, Lecture Notes, Colorado School of Mines,
1979.
21. Shutt, R. C., and Fink, D. A., Weld. J., 64: 19, 1985.
22. Mirza, R. M., and Gee, R., Sci. Technol.Weld. Join., 4: 104, 1999.
23. Albert, S. K., Remash, C., Murugesan, N., Gill, T. P. S., Periaswami, G., and
Kulkarni, S. D., Weld. J., 76: 251s, 1997.
24. Smith, R. D., Benson, D. K., Maroef, I., Olson, D. L., and Wildeman, T. R.,Weld. J.,
80: 122s, 2001.
25. Fazackerley, W., and Gee, R., in Trends in Welding Research, Eds. H. B. Smartt,
J. A. Johnson, and S. A. David, ASM International, Materials Park, OH, June 1995,
p. 435.
26. Principles and Technology of the Fusion Welding of Metal, Vol. 1., Mechanical
Engineering Publishing Company, Peking, 1979 (in Chinese).
27. Flanigan, A. E., Weld. J., 26: 193s, 1947.
28. Eastwood, L. W., Gases in Non-ferrous Metals and Alloys, American Society for
Metals, Cleveland, OH, 1953.
29. Devletian, J. H., and Wood,W. E., Weld. Res. Council Bull., 290: December 1983.
30. Shore, R. J., M.S. Thesis, Ohio State University, Columbus, OH, 1968.
REFERENCES 93
31. Pense, A.W., and Stout, R. D., Weld. Res. Council Bull., 152: July 1970.
32. Fedoseev,V.A., Ryazantsev,V. I., Shiryaeva,N.V., and Arbuzov,Y.U. P., Svar. Proiz.,
6: 15, 1978.
33. Ishchenko, A. Y., and Chayun, A. G., paper presented at the Second International
Conference on Aluminum Weldments, Paper 110, Munich, F.R.G., May
1982.
34. Matsuda, F., Nakata, K., Miyanaga, Y., Kayano, T., and Tsukamoto, K., Trans. Jpn.
Weld. Res. Inst., 7: 181, 1978.
35. Tomsic, M., and Barhorst, S., Weld. J., 63: 25, 1984.
36. Baeslak, W. A. III, paper presented at the 1982 American Welding Society Convention,
Kansas City, MO, 1982.
37. Chai, C. S., and Eagar, T.W., Weld. J., 61: 229s, 1982.
38. Burck, P. A., Indacochea, J. E., and Olson, D. L., Weld. J., 69: 115s, 1990.
39. Abraham, K. P., Davies, M. W., and Richardson, F. D., J. Iron Steel Inst., 196: 82,
1960.
40. Linnert,G. E., in Welding Metallurgy,Vol. 1,American Welding Society, Miami, FL,
1965, Chapter 8, p. 367.
41. Gurevich, S. M., Metallurgy and Technology of Welding Titanium and Its Alloys,
Naukova Dumka Publishing House, Kiev, 1979 (in Russian).
42. Schenck, H., Physical Chemistry of Steelmaking, British Iron and Steel Research
Association, London, 1945.
43. Tuliani, S. S., Boniszewski, T., and Eaton, N. F., Weld. Metal Fab., 37: 327, 1969.
44. Eagar, T.W., Weld. J., 57: 76s, 1978.
45. Chai, C. S., and Eagar, T.W., Metall. Trans., 12B: 539, 1981.
46. Baune, E., Bonnet, C., and Liu, S., Weld. J., 79: 66s, 2000.
47. Dallam, C. B., Liu, S., and Olson, D. L., Weld. J., 64: 140s, 1985.
48. Lathabai, S., and Stout, R. D., Weld. J., 64: 303s, 1985.
49. Fleck, N. A., Grong,O., Edwards,G. R., and Matlock,D. K.,Weld. J., 65: 113s, 1986.
50. Liu, S., and Olson, D. L., Weld. J., 65: 139s, 1986.
51. Ito,Y., and Nakanishi, M., Sumitomo Search, 15: 42, 1976.
52. Abson, D. J., Dolby, R. E., and Hart, P. M. H., Trend in Steel and Consumables fore
Welding, pp. 75–101, 1978.
53. Cochrane, R.C., and Kirkwood, P. R.,Trend in Steel and Consumables fore Welding,
pp. 103–121, 1978.
54. Bose,W.W., Bowen, P., and Strangwood, M., in Trends in Welding Research, Eds.
H. B. Smartt, J. A. Johnson, and S. A. David, ASM International, Materials Park,
OH, June 1995, p. 603.
55. Jackson, C. E., in Weldability of Steels, 3rd. ed., Eds. R. D. Stout and W. D. Doty,
Welding Research Council, New York, 1978, p. 48.
56. North, T. H., Bell, H. B., Nowicki, A., and Craig, I., Weld. J., 57: 63s, 1978.
57. Siewert, T. A., and Ferree, S. E., Metal Prog., 119: 58, 1981.
58. Kim, J. H., Frost, R. H., Olson, D. L., and Blander, M., Weld. J., 69: 446s, 1990.
59. Kim, J. H., Frost, R. H., and Olson, D. L., Weld. J., 77: 488s, 1998.
94 CHEMICAL REACTIONS IN WELDING
FURTHER READING
1. Olson,D. L., Liu, S., Frost, R. H., Edwards,G. R., and Fleming,D. A., in ASM Handbook,
Vol. 6: Welding, Brazing and Soldering, ASM International, Materials Park,
OH, 1993, p. 55.
2. Gaskell, D. R., Introduction to Metallurgical Thermodynamics, 2nd ed., McGraw-
Hill, New York, 1983.
3. Richardson, F. D., Physical Chemistry of Melts in Metallurgy, Vols. 1 and 2.
Academic, London, 1974.
PROBLEMS
3.1 Lithium alloys are known to have severe hydrogen porosity problems
due to the hydration of oxides and formation of hydrides on the workpiece
surface. If you were to do GMAW of these alloys using Al–Li wires,
can you avoid porosity? If so, how?
3.2 Do you prefer using an oxidizing or reducing flame in gas welding of
high-carbon steels? Explain why or why not.
3.3 (a) Will decreasing welding speed help reduce weld porosity in gas–tungsten
arc welds of aluminum if the source of hydrogen is on the workpiece
surface?
(b) What about if the source of hydrogen is in the shielding gas? Explain
why or why not.
3.4 When welding rimmed steels or when doing GMAW of carbon steels
using CO2 as the shielding gas, Mn- or Si-containing electrodes are used
to prevent gas porosity. Explain why.
3.5 Austenitic stainless steels usually contain very low levels of carbon,
around or below 0.05wt %.When welding stainless steels using CO2 as
the shielding gas or covered electrodes containing abundant CaCO3, the
weld metal often tends to carburize. Explain why and indicate how to
avoid the problem.
3.6 It has been reported that gas–tungsten arc welds of aluminum made in
the overhead position tend to have a significantly higher porosity level
than those made in the flat position. Explain why.
3.7 The GTAW of pure iron with Ar–5% H2 as the shielding gas showed that
the weld metal hydrogen content increased with increasing heat input
per unit length of the weld. Explain why.
3.8 Electromagnetic stirring has been reported to reduce hydrogen porosity
in aluminum welds. Explain why.
PROBLEMS 95
3.9 A steel container was welded by SAW with a filler wire containing 1.38%
Mn and 0.05% Si and a flux containing 11.22% SiO2 and 1.15% MnO. Is
the electrode tip Mn content expected to be greater or smaller than the
wire Mn content and why? What about the Si content?
96 CHEMICAL REACTIONS IN WELDING
4 Fluid Flow and Metal
Evaporation in Welding
In this chapter fluid flow in both the welding arc and the weld pool will
be described. The effect of weld pool fluid flow on the geometry of the
resultant weld will be discussed. Evaporation of alloying elements from the
weld pool and its effect on the weld metal composition will be presented.
Finally, the effect of active fluxes on weld penetration in GTAW will be
discussed.
4.1 FLUID FLOW IN ARCS
Figure 4.1 shows a gas–tungsten arc in GTAW with DC electrode negative (1).
The shape of the electrode tip is characterized by the tip angle (also called
included or vertex angle) and the extent to which the sharp point is removed,
that is, the truncation.
4.1.1 Driving Force for Fluid Flow
The welding arc is an ionic gas, that is, a plasma, with an electric current passing
through.The driving force for fluid flow in the arc is the electromagnetic force
or Lorentz force.The buoyancy force is negligible. Mathematically, the Lorentz
force F = J ¥ B, where J is the current density vector and B is the magnetic
flux vector.The current density vector J is in the direction the electric current
flows. According to the right-hand rule for the magnetic field, if the thumb
points in the direction of the current, the magnetic flux vector B is in the direction
that the fingers curl around the path of the current. Vectors F, J, and B
are perpendicular to each other. According to the right-hand rule for the electromagnetic
force, F is in the direction out of and perpendicular to the palm
if the thumb points in the direction of J and the fingers stretch out and point
in the direction of B.
4.1.2 Effect of Electrode Tip Geometry
The welding arc is more or less bell shaped. The tip angle of a tungsten
electrode in GTAW is known to have a significant effect on the shape of
97
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
the arc (Chapter 2)—it tends to become more constricted as the electrode
tip changes from sharp to blunt. The change in the shape of the electrode tip
changes fluid flow in the arc plasma, which in turn changes the shape of the
arc.
A. Sharp Electrode Consider the case of GTAW with DC electrode negative.
The electric current converges from the larger workpiece to the smaller
electrode tip. It tends to be perpendicular to the electrode tip surface and the
workpiece surface, as illustrated in Figure 4.2a. The electric current induces
a magnetic field, and its direction is out of the plane of the paper (as indicated
by the front view of an arrow) on the left and into the paper (as indicated
by the rear view of an arrow) on the right. The magnetic field and the converging
electric current field together produce a downward and inward
force F to push the ionic gas along the conical surface of the electrode tip.
The downward momentum is strong enough to cause the high-temperature
ionic gas to impinge on the workpiece surface and turn outward along
the workpiece surface, thus producing a bell-shaped arc, as illustrated in
Figure 4.2b.
Fluid flow in welding arcs has been studied by computer simulation (1–6).
Tsai and Kou (1) investigated the effect of the electrode tip geometry on heat
and fluid flow in GTAW arcs. Figure 4.3 (left) shows the current density distribution
in a 2-mm-long, 200-A arc produced by a 3.2-mm-diameter electrode
98 FLUID FLOW AND METAL EVAPORATION IN WELDING
Figure 4.1 Gas–tungsten welding arc. Reprinted from Tsai and Kou (1). Copyright
1990 with permission from Elsevier Science.
with a 60° angle tip. The electric current near the electrode tip is essentially
perpendicular to the surface. The electromagnetic force, shown in Figure 4.3
(right), is downward and inward along the conical surface of the electrode tip.
As shown in Figure 4.4, this force produces a high-velocity jet of more than
200m/s maximum velocity and the jet is deflected radially outward along the
workpiece surface. This deflection of the high-temperature jet causes the
isotherms to push outward along the workpiece surface, thus resulting in a
bell-shaped arc.
FLUID FLOW IN ARCS 99
Tungsten
electrode (-)
Tungsten
electrode (-)
Shielding
gas
Arc
Workpiece (+) Workpiece (+)
(a) (b)
Electric
current
F F
Figure 4.2 Arc produced by a tungsten electrode with a sharp tip: (a) Lorentz force
(F); (b) fluid flow.
Tungsten
electrode (-)
0.5 mm
1.0 x 108
A/m2
5.0 x 106
N/m3
0.5 mm
current density Lorentz force
Figure 4.3 Current density field (left) and Lorentz force field (right) in an arc produced
by a tungsten electrode with a 60° tip angle. Modified from Tsai and Kou (1).
B. Flat-End Electrode With a flat-end electrode, on the other hand, there
is no longer an electrode tip to act as a fixed cathode spot, where the electric
current enters the electrode. Consequently, the cathode spot moves around
rather randomly and quickly in the flat electrode end. The time-averaged
diameter of the area covered by the moving spot can be considered as the
effective cathode spot size.Without a conical surface at the electrode end, the
resultant Lorentz force is still inward and downward but the downward component
is reduced, as illustrated in Figure 4.5a. Consequently, the resultant arc
can be expected to be more constricted, as shown in Figure 4.5b.
Figure 4.6 (left) shows the current density distribution in a 2-mm-long, 200-
A arc produced by a 3.2-mm-diameter electrode with a flat end and a 2.4-mmdiameter
cathode spot.The electromagnetic force, shown in Figure 4.6 (right),
is less downward pointing than that in the case of a 60° electrode (Figure 4.3).
As shown in Figure 4.7, the isotherms are not pushed outward as much and
the resultant arc is thus more constricted.
C. Power Density and Current Density Distributions These distributions at
the anode surface can be measured by the split-anode method (7–10). Figure
4.8 shows such distributions for a 100-A, 2.7-mm-long gas–tungsten arc measured
by Lu and Kou (10). These distributions are often approximated by the
following Gaussian distributions:
100 FLUID FLOW AND METAL EVAPORATION IN WELDING
Tungsten
electrode (-)
0.5 mm
200 m/sec
velocity field and isotherms
Figure 4.4 Velocity and temperature fields in an arc produced by a tungsten electrode
with a 60° tip angle. The isotherms from right to left are 11,000, 13,000, 15,000, 17,000,
19,000, and 21,000K. Modified from Tsai and Kou (1).
(4.1)
(4.2)
where q is the power density, Q the power transfer to the workpiece, a the
effective radius of the power density distribution, j the current density, I the
welding current, and b the effective radius of the current density distribution.
j
I
b
r
b
=
-
ÊË Á
ˆ¯ ˜
3
2 3
2
p 2
exp
q
Q
a
r
a
=
-
ÊË Á
ˆ¯ ˜
3
2 3
2
p 2
exp
FLUID FLOW IN ARCS 101
Tungsten
electrode (-)
Tungsten
electrode (-)
Workpiece (+)
F F
Workpiece (+)
Shielding
gas
(a) (b)
Electric
current
Figure 4.5 Arc produced by a tungsten electrode with a flat end: (a) Lorentz force
(F); (b) fluid flow.
Tungsten
electrode (-)
0.5 mm
5.0 x 107
A/m2
1.2 x 106
N/m3
current density Lorentz force
0.5 mm
Figure 4.6 Current density field (left) and Lorentz force field (right) in an arc produced
by a tungsten electrode with a flat end. Modified from Tsai and Kou (1).
The effective radius represents the location where q or j drops to 5% of its
maximum value. Equation (4.1) is identical to Equation (2.20).
Lee and Na (6) studied, by computer simulation, the effect of the arc length
and the electrode tip angle on gas–tungsten arcs. Figure 4.9 shows that both
the power and current density distributions at the anode (workpiece) flatten
and widen as the arc length increases.
102 FLUID FLOW AND METAL EVAPORATION IN WELDING
Tungsten
electrode (-)
0.5 mm
200 m/sec
velocity field and isotherms
Figure 4.7 Velocity and temperature fields in an arc produced by a tungsten electrode
with a flat end.The isotherms from right to left are 11,000, 13,000, 15,000, 17,000, 19,000,
and 21,000K. Modified from Tsai and Kou (1).
Experiment
Gaussian
approximation
Experiment
Gaussian
approximation
DC electrode negative
100 A, 12 V, Ar
2.7 mm arc length
2.4 mm electrode
50o tip angle
Radius, mm Radius, mm
(a) (b)
Current density, amp/mm2
Power density, W/mm2
0 2 4 6
0
10
20
30
0 2 4 6
0
20
40
60
80
q(r)=65.0e-0.27r2 j(r)=17.6e-0.54r2
Figure 4.8 Gas–tungsten welding arc: (a) power density distribution; (b) current
density distribution. Reprinted from Lu and Kou (10). Courtesy of American Welding
Society.
4.2 FLUID FLOW IN WELD POOLS
4.2.1 Driving Forces for Fluid Flow
The driving forces for fluid flow in the weld pool include the buoyancy force,
the Lorentz force, the shear stress induced by the surface tension gradient
at the weld pool surface, and the shear stress acting on the pool surface by the
arc plasma. The arc pressure is another force acting on the pool surface, but
its effect on fluid flow is small, especially below 200 A (11, 12), which is usually
the case for GTAW. The driving forces for fluid flow in the weld pool, shown
in Figure 4.10, are explained next.
A. Buoyancy Force The density of the liquid metal (r) decreases with
increasing temperature (T). Because the heat source is located above the
center of the pool surface, the liquid metal is warmer at point a and cooler at
point b. Point b is near the pool boundary, where the temperature is lowest at
the melting point. As shown in Figure 4.10a, gravity causes the heavier liquid
FLUID FLOW IN WELD POOLS 103
Calculated
Measured
3.2 mm
6.3 mm
12.7 mm arc length
3.2 mm
6.3 mm
I = 200 A
e = 30o
r-axis along base plate, mm
r-axis along base plate, mm
Heat flux, W/mm2 Current density, A/mm 2
0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
0
20
40
60
80
100
0
20
40
60
80
100
I = 200 A
e = 30o
Calculated
Measured
12.7 mm arc length
(a)
(b)
θ
θ
Figure 4.9 Effect of arc length on gas–tungsten welding arcs: (a) power density distributions;
(b) current density distributions. Modified from Lee and Na (6).
metal at point b to sink. Consequently, the liquid metal falls along the pool
boundary and rises along the pool axis, as shown in Figure 4.10b.
B. Lorentz Force Gas–tungsten arc welding with DC electrode negative is
used as an example for the purpose of discussion. The electric current in the
workpiece converges toward the tungsten electrode (not shown) and hence
near the center of the pool surface.This converging current field, together with
the magnetic field it induces, causes a downward and inward Lorentz force, as
shown in Figure 4.10c. As such, the liquid metal is pushed downward along the
pool axis and rises along the pool boundary, as shown in Figure 4.10d.The area
on the pool surface where the electric current goes through is called the anode
spot (pb2, where b is the effective radius of the current density distribution).
104 FLUID FLOW AND METAL EVAPORATION IN WELDING
Arc Shear Stress
(g) (h)
F F
Surface Tension Force
F F
b a b
(e) (f)
Lorentz Force
electric
current
F F
(c) (d)
b
Buoyancy Force
a
(a) T (b)
b a
T
b a
weld pool
base metal
b
F F
γ
ρ
Figure 4.10 Driving forces for weld pool convection: (a, b) buoyancy force; (c, d)
Lorentz force; (e, f ) shear stress caused by surface tension gradient; (g, h) shear stress
caused by arc plasma.
The smaller the anode spot, the more the current field converges from the
workpiece (through the weld pool) to the anode spot, and hence the greater
the Lorentz force becomes to push the liquid metal downward.
C. Shear Stress Induced by Surface Tension Gradient In the absence of a
surface-active agent, the surface tension (g) of the liquid metal decreases with
increasing temperature (T), namely, ∂g /∂T < 0. As shown in Figure 4.10e, thewarmer liquid metal with a lower surface tension at point a is pulled outwardby the cooler liquid metal with a higher surface tension at point b. In otherwords, an outward shear stress is induced at the pool surface by the surfacetension gradient along the pool surface. This causes the liquid metal to flowfrom the center of the pool surface to the edge and return below the poolsurface, as shown in Figure 4.10f. Surface-tension-driven convection is alsocalled thermocapillary convection or Marangoni convection.D. Shear Stress Induced by Plasma Jet The plasma moving outward at highspeeds along a pool surface (Figure 4.4) can exert an outward shear stress atthe pool surface, as shown in Figure 4.10g. This shear stress causes the liquidmetal to flow from the center of the pool surface to the pool edge and returnbelow the pool surface, as shown in Figure 4.10h.These driving forces are included either in the governing equations or asboundary conditions in the computer modeling of fluid flow in the weld pool(13). Oreper et al. (14) developed the first two-dimensional fluid flow modelfor stationary arc weld pools of known shapes. Kou and Sun (15) developed asimilar model but allowed the unknown pool shape to be calculated. Kou andWang (16–18) developed the first three-dimensional fluid flow model formoving arc and laser weld pools. Numerous computer models have beendeveloped subsequently for fluid flow in weld pools.4.2.2 Buoyancy ConvectionFigure 4.11 shows the buoyancy convection in a stationary weld pool of an aluminumalloy calculated by Tsai and Kou (19).The liquid metal rises along thepool axis and falls along the pool boundary. The maximum velocity is alongthe pool axis and is only about 2 cm/s. The pool surface is slightly above theworkpiece surface because of the expansion of the metal upon heating andmelting.4.2.3 Forced Convection Driven by Lorentz ForceA. Flow Field Figure 4.12 shows the calculated results of Tsai and Kou (20)for a stationary weld pool in an aluminum alloy. The liquid metal falls alongthe pool axis and rises along the pool boundary (Figure 4.12a). The electriccurrent converges from the workpiece to the center of the pool surface (Figure4.12b).The Lorentz force is inward and downward (Figure 4.12c), thus pushingFLUID FLOW IN WELD POOLS 105106 FLUID FLOW AND METAL EVAPORATION IN WELDINGFigure 4.11 Buoyancy convection in aluminum weld pool. From Tsai and Kou (19).(a)(b)fluidflowcurrentdensity5 x 105 N/m3(c)Lorentzforce0.4 m/sec5 x 107 A/m2 1 mmFigure 4.12 Convection in an aluminum weld pool caused by Lorentz force: (a)flow field; (b) current density field; (c) Lorentz force field. Modified from Tsai and Kou(20).the liquid downward along the pool axis.The maximum velocity, about 40 cm/s,is one order of magnitude greater than that in the case of buoyancy convection(Figure 4.11). The parameters used for the calculation include 150 A forthe welding current, 1800W for the power input, 2mm for the effective radiusof the power density distribution, and 4mm for the effective radius of thecurrent density distribution.B. Deep Penetration Caused by Lorentz Force The Lorentz force (Figure4.12) makes the weld pool much deeper, as compared to the buoyancy force(Figure 4.11). The liquid pushed downward by the Lorentz force carries heatfrom the heat source to the pool bottom and causes a deep penetration.C. Physical Simulation of Effect of Lorentz Force Kou and Sun (15) weldedWoods metal (a low-melting-point alloy) with a heated copper rod in contactwith it, as shown in Figure 4.13a. The weld became much deeper when a 75-A current was passed through to the weld pool (no arcing and negligible resistanceheating), as shown in Figure 4.13b. This confirms the effect of theLorentz force on weld penetration.4.2.4 Marangoni ConvectionA. Heiple’s Model Heiple et al. (21–25) proposed that, when a surfaceactiveagent is present in the liquid metal in a small but significant amount,∂g /∂T can be changed from negative to positive, thus reversing Marangoni convectionand making the weld pool much deeper. Examples of surface-activeFLUID FLOW IN WELD POOLS 107Weld(a)(b)Heatedrod75 AcurrentWeldHeatedrodWoods metal Woods metalFigure 4.13 Welds in Woods metal produced under the influence of (a) buoyancyforce and (b) Lorentz force. Modified from Kou and Sun (15).agents in steel and stainless steel are S, O, Se, and Te. Figure 4.14 shows thesurface tension data of two different heats of the stainless steel, one withapproximately 160 ppm more sulfur than the other (26). Figure 4.15 shows twoYAG laser welds made in 6.4-mm- (1/4-in.-) thick 304 stainless steel plates at3000W and 3.39mm/s (8ipm). The plate with the shallower weld containsabout 40 ppm sulfur and that with the deeper weld contains about 140ppmsulfur (27).Heiple’s model is explained in Figure 4.16. In the absence of a surface-activeagent (Figures 4.16a–c), the warmer liquid metal of lower surface tension nearthe center of the pool surface is pulled outward by the cooler liquid metal ofhigher surface tension at the pool edge. In the presence of a surface-activeagent (Figures 4.16d–f ), on the other hand, the cooler liquid metal of lowersurface tension at the edge of the pool surface is pulled inward by the warmerliquid metal of higher surface tension near the center of the pool surface.Theflow pattern in Figure 4.16e favors convective heat transfer from the heatsource to the pool bottom. In other words, the liquid metal carries heat from108 FLUID FLOW AND METAL EVAPORATION IN WELDINGSurface Te nsion (mN/m)160017001800190020001400 1600 1800 2000 2200Temperature ( oC)low sulfurstainless steelhigh sulfurstainless steelFigure 4.14 Surface tension data of two different heats of 316 stainless steel, one with160ppm more sulfur than the other. Modified from Heiple and Burgardt (26).Figure 4.15 YAG laser welds in two 304 stainless steels with (a) 40 ppm sulfur and(b) 140 ppm sulfur. From Limmaneevichitr and Kou (27).the heat source to the pool bottom more effectively, thus increasing the weldpenetration.B. Physical Simulation of Marangoni Convection Limmaneevichitr andKou (28) induced Marangoni convection in a transparent pool of NaNO3 witha defocused CO2 laser beam.The NaNO3 has a ∂g/∂T = -0.056dyn/cm/°C. Sinceits transmission range is from 0.35 to 3mm,NaNO3 is opaque to CO2 laser (10.6mm wavelength) just like a metal weld pool is opaque to an arc. Figure 4.17FLUID FLOW IN WELD POOLS 109(c)(b)(e) (f)weldingdirectiondistance, ltemperature,T(a)(d)121212121 212lsurface tension, surface tension,temperature,Tγγγ γFigure 4.16 Heiple’s model for Marangoni convection in a weld pool: (a, b, c) lowsulfursteel; (d, e, f) high-sulfur steel.Figure 4.17 Marangoni convection with an outward surface flow in a NaNO3 pool.Reprinted from Limmaneevichitr and Kou (28). Courtesy of American WeldingSociety.shows the flow pattern induced by a CO2 laser beam of 2.5W power and3.2mm diameter (28). The pool surface is just below the two arrows thatindicate the directions of flow at the pool surface. The two counterrotatingcells are, in fact, the two intersections between the donut-shaped flow patternand the meridian plane of the pool. The outward surface flow is much fasterthan the inward return flow, which is typical of Marangoni convection. Asthe beam diameter is reduced, convection grows stronger and penetratesdeeper.It is worth noting that in conduction-mode (no-keyholing) laser beamwelding, the pool surface can be concave due to Marangoni convection andsurface tension (29) and, in fact, this has been shown to be the case experimentally(30) and by computer simulation (31). The concave NaNO3 poolsurface, however, is just a coincidence; the melt wets the container wall andthe meniscus makes the pool surface concave.Limmaneevichitr and Kou (32) added C2H5COOK to the NaNO3 pool andreversed the direction of Marangoni convection.The C2H5COOK is a surfaceactiveagent of NaNO3 and, like S reduces the surface tension of liquid steel,it reduces the surface tension of NaNO3, ∂g /∂C being -22dyn/(cm/mol %) (33).The NaNO3 pool shown in Figure 4.18 contains 2mol % of C2H5COOK andits surface flow is inward. This flow reversion is because the surface tensionis now higher at the center of the pool surface than at the pool edge. At thecenter of the pool surface, C2H5COOK is lower in concentration because itdecomposes under the heating of the CO2 laser beam.Blocks of solid NaNO3, both pure and with C2H5COOK, were welded witha defocused CO2 laser beam. As shown in Figure 4.19, the weld in pure NaNO3is shallow and wide (32). This is because the thermal conductivity of NaNO3is low and heat transfer is dominated by the outward surface flow to the pool110 FLUID FLOW AND METAL EVAPORATION IN WELDINGFigure 4.18 Marangoni convection with an inward surface flow in a NaNO3 poolcontaining 2mol % C2H5COOK as a surface-active agent. Reprinted fromLimmaneevitchitr and Kou (32). Courtesy of American Welding Society.edge. The weld in NaNO3 with 2mol % C2H5COOK is deeper and slightlynarrower, which is consistent with the inward surface flow observed duringwelding.C. Thermodynamic Analysis of Surface Tension Sahoo et al. (34) andMcNallan and DebRoy (35) calculated the surface tension of liquid metalsbased on thermodynamic data. Figure 4.20 shows the surface tension of liquidiron as a function of temperature and the sulfur content (36). For pure Fe,∂g/∂T is negative at all temperatures. For sulfur-containing Fe, however, ∂g /∂Tcan be positive at lower temperatures, which is consistent with the surfacetension measurements by Sundell et al. (37).Based on the surface tension data in Figure 4.20, Pitscheneder et al. (36)calculated Marangoni convection in stationary steel weld pools. Figure 4.21shows the results for a laser power of 5200 W and an irradiation time of 5 s.FLUID FLOW IN WELD POOLS 111-3-2-10-7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 mmmmCO2 laser (12.4 W)basematerialweld: with surface-active agent; : withoutFigure 4.19 Laser welds in solid blocks of pure NaNO3 (open circles) and NaNO3 with2mol % C2H5COOK (solid squares). Reprinted from Limmaneevichitr and Kou (32).Courtesy of American Welding Society.Temperature Coefficient ofSurface Te n sion x 104(N/mK)Percent SulfurTe m perature (K)Surface Te n sion (N/m)0.02 0.01 0.00(a)2.21.71.22820242020201620 2020 2420 282001-135-3-5Temperature (K)(b)20 ppm150 ppm40 ppmFigure 4.20 Liquid iron with various levels of sulfur: (a) surface tension; (b) temperaturecoefficient of surface tension. Reprinted from Pitscheneder et al. (36).Courtesy of American Welding Society.For the steel containing only 20 ppm sulfur, the outward surface flow carriesheat from the heat source to the pool edge and results in a shallow and widepool (Figure 4.21a). For the steel containing 150 ppm sulfur, on the other hand,the inward surface flow turns downward to deliver heat to the pool bottomand results in a much deeper pool (Figure 4.21b). In the small area near thecenterline of the pool surface, the temperature is above 2000 K and the surfaceflow is outward because of negative ∂g /∂T (Figure 4.20). It is worth noting thatZacharia et al. (38) showed that computer simulations based on a positive∂g /∂T for liquid steel at all temperatures can overpredict the pool depth.4.2.5 Forced Convection Driven by Plasma JetMatsunawa and Shinichiro (39, 40) demonstrated that the plasma shear stressinduced by a long arc in GTAW can outweigh both the Lorentz force in the112 FLUID FLOW AND METAL EVAPORATION IN WELDINGFigure 4.21 Convection in stationary laser weld pools of steels with (a) 20 ppm sulfurand (b) 150 ppm sulfur. Reprinted from Pitscheneder et al. (36). Courtesy of AmericanWelding Society.weld pool and the surface tension gradients along the pool surface. Figure 4.22shows two series of stationary gas–tungsten arc welds in mild steel with 150,180, and 210 s of welding times, one with a 2-mm-long arc and the other witha 8-mm-long arc (39, 40). The 8-mm-arc welds are much wider and shallowerthan the 2-mm-arc welds. For a longer and thus wider arc, the Lorentz forcein the weld pool is smaller because of flatter and wider current density distributionat the pool surface (Figure 4.9b).The surface tension gradients are alsosmaller because of the flatter and wider power density distribution (Figure4.9a). However, location of the maximum shear stress shifts outward, thusallowing the shear stress to act on a greater portion of the pool surface.As shown in Figure 4.23, with a 2-mm-arc length the gas–tungsten arc weldis much deeper in the 304 stainless steel containing 77 ppm sulfur (39, 40).ThisFLUID FLOW IN WELD POOLS 113150 s180 s210 s5 0 5 mm2 mm arc length 8 mm arc length5 0 5mmFigure 4.22 Stationary gas–tungsten arc welds in a mild steel made with a 2-mm arc(left) and an 8-mm arc (right) for 150, 180, and 210 s. Modified from Matsunawa et al.(39).S: 18 ppm S: 77 ppm(a)(b)S: 18 ppm S: 77 ppm2 mm arc length5 mm8 mm arc length∂σ ∂Τ< 0 ∂σ ∂Τ = 0Figure 4.23 Stationary GTA welds in 304 stainless steels with 18 ppm sulfur and77ppm sulfur made with an arc length of (a) 2mm and (b) 8mm. Modified fromMatsunawa et al. (39).is expected because sulfur makes ∂g/∂T either less negative or even positive.With a longer arc of 8mm, however, the welds are shallow regardless of thesulfur level. This further demonstrates that forced convection driven by thearc plasma dominates in the welds made with the 8-mm-long arc.4.2.6 Effect of TurbulenceChoo and Szekely (41) first considered turbulence in gas–tungsten arc weldpools and showed that turbulence can affect the pool depth significantly.Hong et al. (42) demonstrated that a fluid flow model based on laminarcan over-predict the pool depth, as shown in Figure 4.24a for a GTA weldin a 304 stainless steel. When turbulence is considered, however, the effectiveviscosity increases (meff > m) and convection slows down. Furthermore,
the effective thermal conductivity (keff > k) increases and the effect of convection
on the pool shape thus decreases. Consequently, the calculated
pool depth decreases and agrees better with the observed one, as shown in
Figure 4.24b.
4.3 METAL EVAPORATION
4.3.1 Loss of Alloying Elements
Due to the intense heating of pool surface, evaporation from the weld pool
can be significant with some alloying elements. Evaporation-induced Mg losses
114 FLUID FLOW AND METAL EVAPORATION IN WELDING
Figure 4.24 Weld pool shapes and isotherms in a 304 stainless steel with 50 ppm sulfur
calculated based on (a) laminar flow and (b) turbulent flow. Reprinted from Hong
et al. (42).
(a)
(b)
from laser welds of aluminum alloys have been reported (43–45), and Figure
4.25 is an example (45). Since Al–Mg alloys are solution strengthened by alloying
with Mg, Mg loss can result in substantial reduction in the tensile strength
of the weld metal. Similarly, Figure 4.26 shows evaporation-induced Mn losses
in laser welds of stainless steels (46).
Figure 4.27 shows the vapor pressure of several metals (47). It is clear that
Mg has a much higher vapor pressure than Al at any temperature; that is, Mg
has a greater tendency to evaporate than Al.This explains Mg losses from laser
welds of aluminum alloys. It is also clear that Mn has a much higher vapor
pressure than Fe, which explains Mn losses from laser welds of stainless steels.
The Langmuir equation has been used to predict the evaporation rate of metal
from the weld pool surface (47). However, according to DebRoy et al. (48–51),
it can overpredict by a factor of 10 or more.
4.3.2 Explosion of Metal Droplets
Evaporation can also occur as metal droplets transfer from the filler wire to
the weld pool through the arc, considering the very high temperature of the
METAL EVAPORATION 115
Figure 4.25 Magnesium loss in a laser weld of an Al–Mg alloy. Reprinted from Pastor
et al. (45). Courtesy of American Welding Society.
arc. Unstable metal transfer has been reported in GMAW with Al–Mg
and Al–Mg–Zn filler wires (52). In fact, high-speed photography revealed
in-flight explosions of metal droplets, resulting in much spattering. Figure
4.27 shows that Zn has an even higher vapor pressure than Mg. Obviously,
the high vapor pressures of Mg and Zn are likely to have contributed to the
explosions.
4.4 ACTIVE FLUX GTAW
The use of fluxes in GTAW has been found to dramatically increase weld
penetration in steels and stainless steels (53–58). The flux usually consists of
oxides and halides, and it is mixed with acetone or the like to form a paste and
116 FLUID FLOW AND METAL EVAPORATION IN WELDING
AISI 202
AISI 201
USS
TENELON
BASE METAL WELD ZONE
Horizontal Distance (mm)
Weight % Mn
0 0.4 0.8 1.2 1.6
3
4
5
6
7
15
16
17
1.8
Figure 4.26 Manganese losses in laser welds of stainless steels. Reprinted from
DebRoy (46).
Zn
Mg
Li
Mn
Al
800 1000 1200 1400 1600 1800 2000
Temperature (K)
Vapor pressure (mm Hg)
1x102
1x100
1x10-2
1x10-4
1x10-6
1x10-8
Fe
Figure 4.27 Vapor pressure of several metals as a function of temperature.
painted as a thin coating over the area to be welded. Figure 4.28 shows gas–
tungsten arc welds made, without and with a flux, in a 6-mm-thick 316L stainless
steel containing low S (0.005 wt %) (56).
Howse and Lucus (56) observed that the arc becomes more constricted
when a flux is used. Consequently, they proposed that the deeper penetration
is caused by arc constriction and the vaporized flux constricts the arc by capturing
electrons in the cooler outer region of the arc. For the same welding
current, the more the arc is constricted, the smaller the area (pb2, where b is
the effective radius of the current density distribution) on the pool surface to
which the current field converges from the workpiece (through the weld pool),
and hence the greater the Lorentz force (F in Figure 4.10c) and deeper
penetration.
Tanaka et al. (57), however, proposed that the inward surface flow in the
presence of the oxygen from the oxide-containing flux causes the deeper
penetration. They gas–tungsten arc welded a 304 stainless steel containing
little (0.002wt %) S, with He for shielding and TiO2 as the flux (TiO2 is a main
ingredient in several commercial fluxes). They observed both an inward
surface flow and a significant decrease in the surface tension when the flux is
used. As mentioned previously, ∂g/∂T can become positive and cause inward
surface flow in the presence of a surface-active agent such as oxygen. They
observed a steep temperature gradient across the pool surface caused by the
inward surface flow. Spectroscopic analysis of the arc showed that the blue
luminous plasma appears to be mainly composed of metal vapor (Cr, Fe, etc.)
from the weld pool.The steeper temperature gradient across the smaller pool
surface suggests a more localized metal evaporation and hence a more constricted
arc, which also help increase the penetration.
REFERENCES
1. Tsai, M. C., and Kou, S., Int. J. Heat Mass Transfer, 33: 2089, 1990.
2. Ushio, M., and Matsuda, F., Trans. JWM, 11: 7, 1982.
REFERENCES 117
Figure 4.28 Gas–tungsten arc welds of 6-mm-thick 316L stainless steel: (a) without a
flux; (b) with a flux. Reprinted from Howse and Lucas (56).
3. Ushio, M., and Matsuda, F., J. Jpn.Weld. Soc., 6: 91, 1988.
4. McKelliget, J., and Szekely, J., Metall. Trans.A, 17A: 1139, 1986.
5. Choo, R. T. C., Szekely, J., and Westhoff, R. C., Weld. J., 69: 346, 1990.
6. Lee, S.-Y., and Na, S.-J., Weld. J., 75: 269, 1996.
7. Nestor, O. H., J. Appl. Phys., 33: 1638 (1962).
8. Schoeck, P. A., in Modern Developments in Heat Transfer, Ed.W. Ibele, Academic,
New York, 1963, p. 353.
9. Tsai, N., Ph.D. Thesis, MIT, Cambridge, MA, 1983.
10. Lu, M., and Kou, S., Weld. J., 67: 29s, 1988.
11. Lawson,W. H. S., and Kerr, H.W.,Welding Research International,Vol. 6, p. 1, 1976.
12. Lin, M. L., and Eagar, T.W., Weld. J., 64: 63s, 985.
13. Kou, S., Transport Phenomena in Materials Processing,Wiley, New York, 1996, pp.
3–115.
14. Oreper, G. M., Eagar, T.W., and Szekely, J., Weld. J., 62: 307s, 1983.
15. Kou, S., and Sun, D. K., Metall. Trans., 16A: 203, 1985.
16. Kou, S., and Wang,Y. H., Weld. J., 65: 63s, 1986.
17. Kou, S., and Wang,Y. H., Metall. Trans., 17A: 2265, 1986.
18. Kou, S., and Wang,Y. H., Metall. Trans., 17A: 2271, 1986.
19. Tsai, M. C., and Kou, S., Numerical Heat Transfer, 17A: 73, 1990.
20. Tsai, M. C., and Kou, S., Weld. J., 69: 241s, 1990.
21. Heiple, C. R., and Roper, J. R., Weld. J., 61: 97s, 1982.
22. Heiple, C. R., and Roper, J. R., in Trends in Welding Research in the United States,
Ed. S. A. David, American Society for Metals, Metals Park, OH, 1982, pp. 489–
520.
23. Heiple, C. R., Roper, J. R., Stagner, R. T., and Aden, R. J., Weld. J., 62: 72s, 1983.
24. Heiple, C. R., Burgardt, P., and Roper, J. R., in Modeling of Casting and Welding
Processes, Vol. 2, Eds. J. A. Dantzig and J. T. Berry, TMS-AIME,Warrendale, PA,
1984, pp. 193–205.
25. Heiple, C. R., and Burgardt, P., Weld. J., 64: 159s, 1985.
26. Keene, B. J., Mills, K. C., and Brooks, R. F., Mater. Sci. Technol., 1: 568–571, 1985;
Heiple, C. R., and Burgardt, P., ASM Handbook, Vol. 6: Welding, Brazing and
Soldering, ASM International, Materials Park, OH, 1993, p. 19.
27. Limmaneevichitr, C., and Kou, S., unpublished research, University of Wisconsin,
Madison, WI, 2000.
28. Limmaneevichitr, C., and Kou, S., Weld. J., 79: 126s, 2000.
29. Duley,W.W., Laser Welding,Wiley, New York, 1999, p. 76.
30. Mazumder, J., and Voekel, D., in Laser Advanced Materials Processing—Science
and Applications, Vol. 1, Eds. A. Matsunawa and S. Katayama, High Temperature
Society of Japan, Osaka, Japan, 1992, pp. 373–380.
31. Tsai, M. C., and Kou, S., Internal J. Numerical Methods Fluids, 9: 1503, 1989.
32. Limmaneevichitr, C., and Kou, S., Weld. J., 79: 324s, 2000.
33. Smechenko,V. K., and Shikobalova, L. P., Zh. Fiz. Khim., 21: 613, 1947.
118 FLUID FLOW AND METAL EVAPORATION IN WELDING
34. Sahoo, P., DebRoy, T., and McNallan, M. J., Metall. Trans. B., 19B: 483–491, 1988.
35. McNallan, M. J., and DebRoy, T., Metall. Trans. B, 22B: 557–560, 1991.
36. Pitscheneder,W., DebRoy,T., Mundra, K., and Ebner, R.,Weld. J., 75: 71s–80s, 1996.
37. Sundell, R. E., Correa, S. M., Harris, L. P., Solomon, H.D., and Wojcik, L. A., Report
86SRD013, General Electric Company, Schenectady, NY, December 1986.
38. Zacharia,T., David, S. A.,Vitek, J. M., and DebRoy,T.,Weld. J., 68: 510s–519s, 1989.
39. Matsunawa, A., and Shinichiro, Y., in Recent Trends in Welding Science and Technology,
Eds. S. A. David and J. M. Vitek, ASM International, Materials Park, OH,
1990, pp. 31–35.
40. Matsunawa, A., in International Trends in Welding Science and Technology, Eds.
S.A. David and J. M.Vitek, ASM International, Materials Park, OH, 1993, pp. 3–16.
41. Choo, R. T. C., and Szekely, J., Weld. J., 73: 25s, 1994.
42. Hong, K.,Weckman,D. C., Strong, A. B., and Zheng,W., Science and Technology of
Welding and Joining, in press;Weckman,D. C., in Trends in Welding Research, Eds.
J. M.Vitek, S. A. David, J. A. Johnson, H. S. Smartt, and T. DebRoy, ASM International,
Materials Park, OH, p. 3, 1999.
43. Moon, D.W., and Metzbower, E. A., Weld. J., 62: 53s, 1983.
44. Cieslak, M. J., and Fuerschbach, P.W., Metall. Trans., 19B: 319, 1988.
45. Pastor, M., Zhaq, H., Martukanitz, R. P., and Debroy, T., Weld. J, 78: 207s, 1999.
46. DebRoy, T., in International Trends in Welding Science and Technology, Eds. S. A.
David and J. M.Vitek, ASM International, Materials Park, OH, 1993, p. 18.
47. Block-Bolten, A., and Eagar, T.W., Metall. Trans., 15B: 461, 1984.
48. Mundra, K., and DebRoy, T., Weld. J., 72: 1s, 1991.
49. Mundra, K., and DebRoy, T., Metall. Trans., 24B: 146, 1993.
50. Yang, Z., and DebRoy, T., Metall. Trans., 30B: 483, 1999.
51. Zhao, H., and DebRoy, T., Metall. Trans., 32B: 163, 2001.
52. Woods, R. A., Weld. J., 59: 59s, 1980.
53. Gurevich, S. M., Zamkov,V. N., and Kushnirenko, N. A.,Avtomat Svarka, Automat.
Weld., 9: 1, 1965.
54. Gurevich, S. M., and Zamkov,V. N., Automat.Weld., 12: 13, 1966.
55. Paskell, T., Lundin, C., and Castner, H., Weld, J., 76: 57, 1997.
56. Howse, D. S., and Lucas,W., Sci. Technol.Weld. Join., 5: 189, 2000.
57. Tanaka, M., Shimizu,T.,Terasaki, H., Ushio, M., Koshi-ishi, F., and Yang, C.-L., Sci.
Technol.Weld. Join., 5: 397, 2000.
58. Kuo, M., Sun, Z., and Pan, D., Sci. Technol.Weld. Join., 6: 17, 2001.
FURTHER READING
1. Szekely, J., Fluid Flow Phenomena in Metals Processing, Academic, New York, 1979.
2. Kou, S., Transport Phenomena and Materials Processing,Wiley, New York, 1996.
3. Lancaster, J. F., The Physics of Welding, Pergamon, Oxford, 1984.
FURTHER READING 119
120 FLUID FLOW AND METAL EVAPORATION IN WELDING
PROBLEMS
4.1 Experimental results show that the depth–width ratio of stainless steel
welds increases with increasing electrode tip angle. How does the angle
affect the effective radius of the electric current at the pool surface (the
anode spot)? How does this radius in turn affect weld pool convection
and the weld depth–width ratio?
4.2 Experimental results show that the depth–width ratio of stainless steel
welds decreases with increasing arc length. Explain why.
4.3 It has been suggested that the weldability of stainless steels can be
improved by oxidizing the surface by subjecting it to an elevated temperature
in an oxidizing environment. From the penetration point of
view, do you agree or disagree, and why?
4.4 Two heats of a stainless steel with the same nominal composition but
significantly different sulfur contents are butt welded by autogenous
GTAW.Which side of the weld is deeper and why?
4.5 Two GTA welds of the same 304 stainless steel were made, one with a
shielding gas of Ar and the other with Ar plus 700ppm SO2 gas.Which
weld was deeper and why?
4.6 Consider Marangoni convection in a simulated weld pool of NaNO3
such as that shown in Figure 4.17. As the laser beam diameter is reduced
at the same power from 5.9 to 1.5mm, does Marangoni convection in
the pool become faster or slower and why?
4.7 In electron beam welding with the surface melting mode, which driving
force for flow is expected to dominate?
4.8 In conduction-mode laser beam welding of a 201 stainless steel sheet
7 mm thick, the welding speed was 3mm/s. The power was increased
from 400 to 600W. It was found that: (a) the Mn evaporation rate
increased, (b) the weld pool grew significantly larger in volume, and (c)
the Mn concentration decrease in the resultant weld metal decreased.
The Mn concentration was uniform in the resultant weld metal. Explain
(c) based on (a) and (b).
4.9 (a) Name the four different driving forces for weld pool convection.
(b) In GMAW with spray metal transfer, is there an additional driving
force for weld pool convection besides the four in (a)? If so,
explain what it is and sketch the flow pattern in the pool caused
by this force alone.
4.10 Paraffin has been used to study weld pool Marangoni convection. The
surface tension of molten paraffin decreases with increasing temperature,
and convection is dominated by the surface tension effect. A thin
slice of paraffin is sandwiched between two pieces of glass, and its top
surface is in contact with the tips of two hot soldering irons to produce
a weld pool that penetrates downward into the thin slice, as shown in
Figure P4.10. Sketch and explain the flow pattern and the shape of the
pool.
PROBLEMS 121
Figure P4.10
5 Residual Stresses, Distortion,
and Fatigue
In this chapter the causes of residual stresses, distortion, and fatigue failure in
weldments will be discussed, and the remedies will be described.
5.1 RESIDUAL STRESSES
Residual stresses are stresses that would exist in a body if all external loads
were removed. They are sometimes called internal stresses. Residual stresses
that exist in a body that has previously been subjected to nonuniform temperature
changes, such as those during welding, are often called thermal
stresses (1).
5.1.1 Development of Residual Stresses
A. Three-Bar Arrangement The development of residual stresses can be
explained by considering heating and cooling under constraint (2). Figure 5.1
shows three identical metal bars connected to two rigid blocks. All three bars
are initially at room temperature. The middle bar alone is heated up, but its
thermal expansion is restrained by the side bars (Figure 5.1a). Consequently,
compressive stresses are produced in the middle bar, and they increase with
increasing temperature until the yield stress in compression is reached. The
yield stress represents the upper limit of stresses in a material, at which plastic
deformation occurs.When heating stops and the middle bar is allowed to cool
off, its thermal contraction is restrained by the side bars (Figure 5.1b). Consequently,
the compressive stresses in the middle bar drop rapidly, change to
tensile stresses, and increase with decreasing temperature until the yield stress
in tension is reached. Therefore, a residual tensile stress equal to the yield
stress at room temperature is set up in the middle bar when it cools down to
room temperature. The residual stresses in the side bars are compressive
stresses and equal to one-half of the tensile stress in the middle bar.
B. Welding Roughly speaking, the weld metal and the adjacent base metal
are analogous to the middle bar, and the areas farther away from the weld
metal are analogous to the two side bars (Figure 5.1c). This is because the
122
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
expansion and contraction of the weld metal and the adjacent base metal are
restrained by the areas farther away from the weld metal. Consequently, after
cooling to the room temperature, residual tensile stresses exist in the weld
metal and the adjacent base metal, while residual compressive stresses exist
in the areas farther away from the weld metal. Further explanations are given
as follows.
Figure 5.2 is a schematic representation of the temperature change (DT)
and stress in the welding direction (sx) during welding (2). The crosshatched
area M–M¢ is the region where plastic deformation occurs. Section A–A is
ahead of the heat source and is not yet significantly affected by the heat input;
the temperature change due to welding, DT, is essentially zero. Along section
B–B intersecting the heat source, the temperature distribution is rather steep.
Along section C–C at some distance behind the heat source, the temperature
distribution becomes less steep and is eventually uniform along section D–D
far away behind the heat source.
Consider now the thermally induced stress along the longitudinal direction,
sx. Since section A–A is not affected by the heat input, sx is zero.Along section
B–B, sx is close to zero in the region underneath the heat source, since the
weld pool does not have any strength to support any loads. In the regions
somewhat away from the heat source, stresses are compressive (sx is negative)
because the expansion of these areas is restrained by the surrounding metal
of lower temperatures. Due to the low yield strength of the high-temperature
metal in these areas, sx reaches the yield strength of the base metal at
RESIDUAL STRESSES 123
middle bar: heated,
in compression
side bars in tension
after
heating
stops
middle bar: cooled, in tension
residual tensile stresses
side bars: residual
compressive stresses
weld metal and adjacent base metal:
residual tensile stresses
areas farther away from weld metal:
residual compressive stresses
(a) (b)
(c)
Figure 5.1 Thermally induced stresses: (a) during heating; (b) during cooling;
(c) residual stresses in weld.
corresponding temperatures. In the areas farther away from the weld sx is
tensile, and sx is balanced with compressive stresses in areas near the weld.
Along section C–C the weld metal and the adjacent base metal have cooled
and hence have a tendency to contract, thus producing tensile stresses (sx is
positive). In the nearby areas sx is compressive. Finally, along section D–D the
weld metal and the adjacent base metal have cooled and contracted further,
thus producing higher tensile stresses in regions near the weld and compressive
stresses in regions away from the weld. Since section D–D is well behind
the heat source, the stress distribution does not change significantly beyond it,
and this stress distribution is thus the residual stress distribution.
5.1.2 Analysis of Residual Stresses
Figure 5.3 shows typical distributions of residual stresses in a butt weld.
According to Masubuchi and Martin (3), the distribution of the longitudinal
residual stress sx can be approximated by the equation
sx y s (5.1)
y
b
y
b
( )= -ÊË
ˆ¯
È
Î Í
˘
˚ ˙
- ÊË
ˆ¯
È
Î Í
˘
˚ ˙
m 1
1
2
2 2
exp
124 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
Figure 5.2 Changes in temperature and stresses during welding. Reprinted from
Welding Handbook (2). Courtesy of American Welding Society.
where sm is the maximum residual stress, which is usually as high as the yield
strength of the weld metal. The parameter b is the width of the tension zone
of sx (Figure 5.3a).
The distribution of the transverse residual stress sy along the length of the
weld is shown in Figure 5.3b. As shown, tensile stresses of relatively low magnitude
are produced in the middle part of the weld, where thermal contraction
in the transverse direction is restrained by the much cooler base metal
farther away from the weld.The tensile stresses in the middle part of the weld
are balanced by compressive stresses at the ends of the weld. If the lateral contraction
of the joint is restrained by an external constraint (such as a fixture
holding down the two sides of the workpiece), approximately uniform tensile
stresses are added along the weld as the reaction stress (1).This external constraint,
however, has little effect on sx.
Figure 5.4 shows measured and calculated distributions of residual stresses
sx in a butt joint of two rectangular plates of 5083 aluminum (60 cm long, 27.5
cm wide, and 1cm thick) welded by GMAW (4). The calculated results are
based on the finite-element analysis (FEA), and the measured results are from
Satoh and Terasaki (5). The measurement and calculation of weld residual
stresses have been described in detail by Masubuchi (1) and will not be
repeated here. Residual stresses can cause problems such as hydrogen-induced
cracking (Chapter 17) and stress corrosion cracking (Chapter 18). Postweld
heat treatment is often used to reduce residual stresses. Figure 5.5 shows the
effect of temperature and time on stress relief in steel welds (2). Table 5.1 list
the temperature ranges used for postweld heat treatment of various types of
materials (2). Other techniques such as preheat, peening, and vibration have
also been used for stress relief.
RESIDUAL STRESSES 125
weld
in tension in compression
base
metal
(a)
weld
base
metal
in tension in compression
(b)
x
x
y
y
b
with external
constraint
σ
σ
Figure 5.3 Typical distributions of longitudinal (sx) and transverse (sy) residual
stresses in butt weld. Modified from Welding Handbook (2).
5.2 DISTORTION
5.2.1 Cause
Because of solidification shrinkage and thermal contraction of the weld metal
during welding, the workpiece has a tendency to distort. Figure 5.6 illustrates
several types of weld distortions (2). The welded workpiece can shrink in the
transverse direction (Figure 5.6a). It can also shrink in the longitudinal direction
along the weld (Figure 5.6b). Upward angular distortion usually occurs
when the weld is made from the top of the workpiece alone (Figure 5.6c).The
weld tends to be wider at the top than at the bottom, causing more solidification
shrinkage and thermal contraction at the top of the weld than at the
bottom. Consequently, the resultant angular distortion is upward. In electron
126 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
F.E.A.
Satoh et al. 1976
0 0.2 0.4 0.6 0.8 1
0.2
0.4
0.6
0.8
1
1.2
0
-0.2
-0.4
Distance along x-axis, x/W
Residual Stress, Sy/Sys
Figure 5.4 Measured and calculated distributions of residual stress in butt joint of
5083 aluminum. Reprinted from Tsai et al. (4). Courtesy of American Welding Society.
Figure 5.5 Effect of temperature and time on stress relief of steel welds. Reprinted
from Welding Handbook (2). Courtesy of American Welding Society.
beam welding with a deep narrow keyhole, the weld is very narrow both at
the top and the bottom, and there is little angular distortion.When fillet welds
between a flat sheet at the bottom and a vertical sheet on the top shrink, they
pull the flat sheet toward the vertical one and cause upward distortion in the
DISTORTION 127
TABLE 5.1 Typical Thermal Treatments for Stress
Relieving Weldments
Soaking
Material Temperature (°C)
Carbon steel 595–680
Carbon–1/2% Mo steel 595–720
1/2% Cr–1/2% Mo steel 595–720
1% Cr–1/2% Mo steel 620–730
11/4% Cr–1/2% Mo steel 705–760
2% Cr–1/2% Mo steel 705–760
21/4% Cr–1% Mo steel 705–770
5% Cr–1/2% Mo (Type 502) steel 705–770
7% Cr–1/2% Mo steel 705–760
9% Cr–1% Mo steel 705–760
12% Cr (Type 410) steel 760–815
16% Cr (Type 430) steel 760–815
11/4% Mn–1/2% Mo steel 605–680
Low-alloy Cr–Ni–Mo steels 595–680
2–5% Ni steels 595–650
9% Ni steels 550–585
Quenched and tempered steels 540–550
Source: Welding Handbook (2).
(a) Transverse shrinkage
in butt weld
(b) Longitudinal
shrinkage in butt weld
(c) Angular distortion
in butt weld
(d) Angular distortion
in fillet welds
Figure 5.6 Distortion in welded structures. Modified from Welding Handbook (2).
flat sheet (Figure 5.6d). Figure 5.7 shows angular distortions in butt welds
of 5083 aluminum of various thicknesses (6). As shown, angular distortion
increases with workpiece thickness because of increasing amount of the weld
metal and hence increasing solidification shrinkage and thermal contraction.
The quantitative analysis of weld distortion has also been described in detail
by Masubuchi (1) and hence will not be repeated here.
5.2.2 Remedies
Several techniques can be used to reduce weld distortion. Reducing the
volume of the weld metal can reduce the amount of angular distortion and
lateral shrinkage. Figure 5.8 shows that the joint preparation angle and the
root pass should be minimized (7). The use of electron or laser beam welding
can minimize angular distortion (Chapter 1). Balancing welding by using a
double-V joint in preference to a single-V joint can help reduce angular distortion.
Figure 5.9 shows that welding alternately on either side of the double-
V joint is preferred (7). Placing welds about the neutral axis also helps reduce
distortion. Figure 5.10 shows that the shrinkage forces of an individual weld
128 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
Figure 5.7 Distortion in butt welds of 5083 aluminum with thicknesses of
6.4–38mm. Reprinted from Gibbs (6). Courtesy of American Welding Society.
DISTORTION 129
larger angle
smaller angle
no angle
more distortion
less distortion
little distortion
Figure 5.8 Reducing angular distortion by reducing volume of weld metal and by
using single-pass deep-penetration welding. Modified from TWI (7).
one side
welded first
alternating
welding
more
distortion
less
distortion
before welding
Figure 5.9 Reducing angular distortion by using double-V joint and welding
alternately on either side of joint. Modified from TWI (7).
Neutral
axis
Neutral
axis
more
distortion
less
distortion
Figure 5.10 Reducing distortion by placing welds around neutral axis. Modified from
TWI (7).
(a) (b)
can be balanced by placing another weld on the opposite side of the neutral
axis (7). Figure 5.11 shows three other techniques to reduce weld distortion
(2, 8). Presetting (Figure 5.11a) is achieved by estimating the amount of
distortion likely to occur during welding and then assembling the job with
members preset to compensate for the distortion. Elastic prespringing (Figure
5.11b) can reduce angular changes after the removal of the restraint. Preheating
(Figure 5.11c), thermal management during welding, and postweld
heating can also reduce angular distortion (4).
130 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
Figure 5.11 Methods for controlling weld distortion: (a) presetting; (b) prespringing;
(c) preheating (8). (a), (b) Reprinted from Welding Handbook (2). Courtesy of
American Welding Society.
5.3 FATIGUE
5.3.1 Mechanism
Failure can occur in welds under repeated loading (9, 10).This type of failure,
called fatigue, has three phases: crack initiation, crack propagation, and fracture.
Figure 5.12 shows a simple type of fatigue stress cycling and how it can
result in the formation of intrusions and extrusions at the surface of a material
along the slip planes (10). A discontinuity point in the material (e.g., inclusions,
porosity) can serve as the source for a slip to initiate. Figure 5.13 shows
a series of intrusions and extrusions at the free surfaces due to the alternating
placement of metal along slip planes (10). Eventually, these intrusions and
extrusions become severe enough and initial cracks form along slip planes.The
FATIGUE 131
1 2 3 4
source
tensile
force
compressive
force
tensile
force
compressive
force
intrusion intrusion
extrusion
slip
plane
0 Time
max
min
Stress
1
2
3
4
cyclic loading
σ
σ
Figure 5.12 Fatigue stress cycling (top) and formation of intrusions and extrusions
(bottom).
Figure 5.13 Fatigue surface showing extrusions and intrusions. From Hull (10).
direction of crack propagation is along the slip plane at the beginning and then
becomes macroscopically normal to the maximum tensile stress (11).
5.3.2 Fractography
As pointed out by Colangelo and Heiser (11), the appearance of fatigue failures
is often described as brittle because of the little gross plastic deformation
and the fairly smooth fracture surfaces. Fatigue failures are usually easy to distinguish
from other brittle failures because they are progressive and they leave
characteristic marks. Macroscopically, they appear as “beach,” “clam-shell,” or
“conchoidal” marks, which represent delays in the fatigue loading cycle. Figure
5.14 shows a fatigue fracture surface, where the arrow indicates the origin of
fracture (12).
5.3.3 S–N Curves
Fatigue data are often presented in the form of S–N curves, where the applied
stress (S) is plotted against number of cycles to failure (N). As the applied
stress decreases, the number of cycles to failure increases. There are many
factors (13) that affect the fatigue behavior, such as material properties, joint
configuration, stress ratio, welding procedure, postweld treatment, loading
condition, residual stresses, and weld reinforcement geometry. Figures
5.15–5.17 show the effect of some of these factors observed by Sanders and
Day (14) in aluminum welds.
132 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
Figure 5.14 Fatigue fracture surface showing beach marks and origin of fracture.
Reprinted, with permission, from Wulpi (12).
FATIGUE 133
As welded transverse square butt joint
Number of kilocycles
Log maximum stress - Ksi
Log maximum stress - MPa
1 10 100 1,000 10,000 100,000
10
5
50
100
500
100
50
10
5456-H116
5052-H32
6061-T6
5056-O
1
Figure 5.15 Effect of alloy and material properties on fatigue of transverse butt joint.
Modified from Sanders and Day (14).
Log maximum stress - Ksi
Log maximum stress - MPa
Number of kilocycles
As welded
5083-O S.R. =0
Cruciform
Double transverse strap
Transverse attachment
Transverse
butt weld
Longitudinal butt weld
10 1,000 10,000 100,000
10
5
50
100
1
500
100
50
10
Cruciform
Double transverse strap
Transverse attachment
Transverse butt weld
Longitudinal butt weld
Figure 5.16 Effect of joint configurations on fatigue of 5083–O aluminum. Modified
from Sanders and Day (14).
5.3.4 Effect of Joint Geometry
As pointed out by Sanders and Day (14), in developing any fatigue behavior
criteria for welding, the severity of joint geometry is probably the most critical
factor. The more severe the geometry, the lower the fatigue strength, as
shown in Figure 5.16.The severity level of the longitudinal butt weld is lowest
because both the weld and the base metal carry the load.The severity level of
the cruciform, on the other hand, is highest since the welds alone carry the
load and the parts are joined perpendicular to each other.
5.3.5 Effect of Stress Raisers
It is well known that stress raisers tend to reduce fatigue life, namely, the socalled
notch effect. Stress raisers can be mechanical, such as toes with a high
reinforcement, lack of penetration, and deep undercuts.They can also be metallurgical,
such as microfissures (microcracks), porosity, inclusions, and brittle
and sharp intermetallic compounds. Figure 5.18 shows a fatigue crack originating
from the toe (Chapter 1) of a gas–metal arc weld of a carbon steel (15).
Figure 5.19 shows a fatigue failure originating from an undercut at the top of
an electron beam weld in a carbon steel and how undercutting can reduce
fatigue life (16).
5.3.6 Effect of Corrosion
As also shown in Figure 5.17, a corrosive environment (salt water in this case)
can often reduce fatigue life (14). This is called corrosion fatigue (17, 18). It
has been reported that the damage can be almost always greater than the sum
of the damage by corrosion and fatigue acting separately.
134 RESIDUAL STRESSES, DISTORTION, AND FATIGUE
Transverse butt weld 5456-H117
S.R. = -1.0
Number of kilocycles
Log maximum stress - Ksi
Log maximum stress - MPa
10 100 1,000 10,000
10
5
50
100
500
100
50
10
1
Reinforcement removed;
tested in salt water
As welded
Tes ted in salt water
Figure 5.17 Effect of reinforcement removal and saltwater environment on fatigue of
5456-H117 aluminum. Modified from Sanders and Day (14).
5.3.7 Remedies
A. Shot Peening Welding and postweld grinding can create tensile residual
stresses at the weld surface and promote fatigue initiation when under cyclic
loading. Shot and hammer peening, on the other hand, can introduce surface
compressive stresses to suppress the formation of intrusions and extrusions
and hence fatigue initiation (13). In shot peening, the metal surface is bombarded
with small spherical media called shot (19). Each piece of shot strik-
FATIGUE 135
Figure 5.18 Fatigue crack originating from weld toe of gas–metal arc weld of carbon
steel. Reprinted from Itoh (15). Courtesy of American Welding Society.
Figure 5.19 Effect of undercutting on fatigue in electron beam welds of carbon steel:
(a) photograph; (b) fatigue life. Reprinted from Elliott (16). Courtesy of American
Welding Society.
ing the surface acts as a tiny peening hammer, producing a tiny indentation or
dimple on the surface. Overlapping indentations or dimples develop a uniform
layer of residual compressive stresses. Figure 5.20 shows the residual stresses
as a function of depth, namely, the distance below the surface (19). It is clear
that tensile residual stresses (>0) in the as-welded condition can be reduced
by stress-relieving heat treatment and reversed to become highly compressive
residual stresses (<0) by shot peening.B. Reducing Stress Raisers Figure 5.21 shows stress raisers caused byimproper welding and how they can be reduced or eliminated. Figure 5.17shows that removing the reinforcement improves the fatigue life (14). This isconsistent with the results of Nordmark et al. (20) shown in Figure 5.22 for136 RESIDUAL STRESSES, DISTORTION, AND FATIGUEwelded conditionwelded & stress relieved (SR)welded, SR & SPwelded & shot peened (SP)Depth (inches)0 0.01 0.02 0.03-60-40-200.020Residual stress (ksi)Depth (mm)0 0.25 0.50 0.750-100-200-300-400100Residual stress (MPa)Figure 5.20 Effect of stress relieving and shot peening on residual stresses near themetal surface. Modified from Molzen and Hornbach (19).reinforcement stress concentrationundercutlack of penetrationHigher stressconcentrationweldsLower stressconcentrationweldsstressconcentrationFigure 5.21 Stress raisers in butt and T-welds and their corrections.transverse butt welds of 6.4-mm- (1/4-in.-) thick 5456-H116 aluminum.However, in the case of incomplete penetration, removing the reinforcementcan dramatically reduce the fatigue life.5.4 CASE STUDIES5.4.1 Failure of a Steel Pipe Assembly (21)Figure 5.23a is a sketch showing a steel pipe of 10 mm wall thickness bent toturn in 90° and welded at the ends (b) to connecting flanges. The pipe–flangeassembly was connected to a bottle-shaped container (a) and a heavy overhangingvalve weighing 95 kg (e). The working pressure was 425kg/cm2 andthe operating temperature was 105°C. The assembly was subjected to vibrationsfrom the compressors provided at intervals along the pipe system.Graphical recorders showed evidence of the vibrations. The assembly failedafter a service time of about 2500 h.Examination of the failed assembly showed two fractures: fracture A atposition c and fracture B at the weld at position d. Both fractures exhibitedradial lines and striations typical of a fatigue fracture. Fracture A (Figure5.23b) originated in small craters accidentally formed in the flange during arcwelding. Fracture B (Figure 5.23c) started at the weld toe, namely, the junctionbetween the weld reinforcement and the base metal surface.The fracturesurface (Figure 5.23d) shows the radial lines and striations of fatigue.The composition of the steel was unavailable, though metallographyshowed that it had a ferrite–pearlite structure, and the grains are fine andCASE STUDIES 137reinforcementintactreinforcementremovedreinforcementremoved10 4 10 5 10 6Cycles to failureincompletepenetrationreinforcementintactcompletepenetrationFigure 5.22 Effect of weld reinforcement and penetration on fatigue life of transversebutt welds of 5456 aluminum. Reprinted from Nordmark et al. (20). Courtesy ofAmerican Welding Society.equiaxed. Regarding fracture A, many cracks were observed in the fused metalin the craters and in the martensitic structure of the heat-affected zone. Underthe cyclic loading from the vibrations, these cracks served as the initiation sitesfor fatigue. As for fracture B, the notch effect due to reinforcement of the weldpromoted the fatigue failure.5.4.2 Failure of a Ball Mill (22)Figure 5.24 shows part of a ball mill used for ore crushing.The cylindrical shellof the ball mill was 10.3 m in length, 4.4 m in diameter, and 50 mm in thickness.138 RESIDUAL STRESSES, DISTORTION, AND FATIGUEFigure 5.23 Failure of a steel pipe assembly (21): (a) pipe system; (b) fracture Acaused by craters at position c; (c) fracture B at weld at position d; (d) surface offracture B.CASE STUDIES 139Figure 5.24 Failure of a ball mill: (a) design; (b) fatigue cracks (indicated by whitearrows); (c) origin of fatigue cracks. From Wallner (22).A flange with 100 mm wall thickness was welded to each end of the cylindricalshell (left half of Figure 5.24a). Both the cylindrical shell and the flangeswere made from a killed steel with the composition of the steel being 0.19%C, 0.25% Si, 0.65% Mn, 0.025% P, and 0.028% S.The entire weight of the millcharge was approximately 320 tons and the drum operated at a rotation speedof 14rpm.After about 3000 h of operation, long cracks appeared on the outsidesurface of the drum (Figure 5.24b).The failed drum was emptied, its inside inspected, and cracks ranging from100 to 1000 mm long were observed. It was found that these cracks had originatedfrom nearby tack welds, which had been made for holding insulationduring stress relieving of the drum (Figure 5.24c). Apparently, the high notcheffect has greatly reduced the fatigue life of the drum. It was subsequently suggestedthat the new joint design shown in the right half of Figure 5.24a be usedand a new stress relief method be employed in order to avoid the use of similartack welds.REFERENCES1. Masubuchi, K., Analysis of Welded Structures, Pergamon, Elmsford, NY, 1980.2. Welding Handbook, 7th ed.,Vol. 1, American Welding Society, Miami, FL, 1976.3. Masubuchi, K., and Martin, D. C., Weld. J., 40: 553s, 1961.4. Tsai, C. L., Park, S. C., and Cheng,W. T., Weld. J., 78: 156s, 1999.5. Satoh, K., and Terasaki, T., J. Jpn.Weld. Soc., 45: 42, 1976.6. Gibbs, F. E., Weld. J., 59: 23, 1980.7. TWI Job Knowledge for Welders, Part 34,Welding Institute, Cambridge, UK, March21, 1998.8. Watanabe, M., Satoh, K., Morii, H., and Ichikawa, I., J. Jpn.Weld. Soc., 26: 591, 1957.9. Hertzberg, R.W., Deformation and Fracture Mechanics of Engineering Materials,Wiley, New York, 1976, p. 422.10. Hull, D., J. Inst. Met., 86: 425, 1957.11. Colangelo, V. J., and Heiser, F. A., Analysis of Metallurgical Failures, Wiley, NewYork, 1974.12. Wulpi, D., Understanding How Components Fail, American Society for Metals,Metals Park, OH, 1985, p. 144.13. Sanders,W.W. Jr., Weld. Res. Council Bull., 171: 1972.14. Sanders, W. W. Jr., and Day, R. H., in Proceedings of the First InternationalAluminum Welding Conference, Welding Research Council, New York, 1982.15. Itoh,Y., Weld. J., 66: 50s, 1987.16. Elliott, S., Weld. J., 63: 8s, 1984.17. Uhlig, H. H., Corrosion and Corrosion Control, 2nd ed.,Wiley, New York, 1971.18. Fontana, M. G., and Greene, N. D., Corrosion Engineering, 2nd ed., McGraw-Hill,New York, 1978.19. Molzen, M. S., and Hornbach, D., Weld. J., 80: 38, 2001.140 RESIDUAL STRESSES, DISTORTION, AND FATIGUE20. Nordmark, G. E., Herbein, W. C., Dickerson, P. B., and Montemarano, T. W.,Weld. J., 66: 162s, 1987.21. Fatigue Fractures in Welded Constructions, Vol. II, International Institute ofWelding, London, 1979, p. 56.22. Wallner, F., in Cracking and Fracture in Welds, Japan Welding Society,Tokyo, 1972,P. IA3.1.FURTHER READING1. Masubuchi, K., Analysis of Welded Structures, Pergamon, Elmsford, NY, 1980.2. Fatigue Fractures in Welded Constructions,Vol. II, International Institute of Welding,London, 1979.3. Cracking and Fracture in Welds, Japan Welding Society, Tokyo, 1972.4. Gurney, T. R., Ed., Fatigue of Welded Structures, 2nd ed., Cambridge UniversityPress, Cambridge, 1979.PROBLEMS5.1 (a) Austenitic stainless steels have a high thermal expansion coefficientand a low thermal conductivity. Do you expect significant distortionin their welds?(b) Name a few welding processes that help reduce weld distortion.5.2 Aluminum alloys have a high thermal expansion coefficient, high thermalconductivity, and high solidification shrinkage. Do you expect significantdistortion in their welds?5.3 The lower the weld reinforcement usually means the less the stress concentrationat the weld toe. One quick way to reduce the height of thereinforcement is to grind its upper half flat without changing its contactangle with the workpiece surface. In this a proper treatment of the reinforcement?Explain why or why not.PROBLEMS 141PART IIThe Fusion ZoneWelding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-46 Basic Solidification ConceptsIn order to help explain the weld metal microstructure and chemical inhomogeneitiesin subsequent chapters, some basic solidification concepts will bepresented first in this chapter. These concepts include solute redistribution,solidification modes, constitutional supercooling, microsegregation and banding,the dendrite-arm or cell spacing, and the solidification path.6.1 SOLUTE REDISTRIBUTION DURING SOLIDIFICATIONWhen a liquid of uniform composition solidifies, the resultant solid is seldomuniform in composition.The solute atoms in the liquid are redistributed duringsolidification. The redistribution of the solute depends on both thermodynamics,that is, the phase diagram, and kinetics, that is, diffusion, undercooling,fluid flow, and so on.6.1.1 Phase DiagramFigure 6.1a is a portion of a phase diagram near the corner of a pure metal ofmelting point Tm, with S denoting the solid phase and L the liquid phase. Considerthe solidification of alloy C0, that is, with initial melt composition C0. Avertical line through C0 intersects the liquidus line at the liquidus temperature,TL, and the solidus line at the solidus temperature, TS. Assume that undercoolingis negligible so that the solid begins to form when the liquid cools toTL. Also assume that equilibrium between the solid and the liquid is maintainedat the solid–liquid (S/L) interface throughout solidification.This meansat any temperature T the composition of the solid at the interface, CS, and thecomposition of the liquid at the interface, CL, follow the solidus line and theliquidus line, respectively.A. Equilibrium Partition Ratio At any temperature T the equilibrium partitionratio, or the equilibrium segregation coefficient, k, is defined as(6.1)where CS and CL are the compositions of the solid and liquid at the S/L interface,respectively. The value of k depends on temperature T. For simplicity,kCC= SL145Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4k is assumed constant; that is, the solidus and liquidus lines are both assumedstraight lines.The first solid to form will have the composition kC0 according toEquation (6.1) and the phase diagram, that is, CL = C0 at T = TL. Considerthe case of k < 1 first. As the phase diagram in Figure 6.1a shows, the solidcannot accommodate as much solute as the liquid does, and the solid thusrejects the solute into the liquid during solidification. Consequently, the solutecontent of the liquid continues to rise during solidification. Since the solidgrows from the liquid, its solute content also continues to rise. As indicated bythe arrowheads on the solidus and liquidus lines in Figure 6.1a for k < 1, CSand CL both increase as temperature T of the S/L interface drops duringsolidification.Consider now the case of k > 1. As the phase diagram in Figure 6.1b shows,
the solid can accommodate more solute than the liquid does, and the solid thus
absorbs the solute from the liquid during solidification. Consequently, the
solute content of the liquid continues to drop during solidification. Since the
solid grows from the liquid, its solute content also continues to drop. As indicated
by the arrowheads on the solidus and liquidus lines in Figure 6.1b for
k > 1, CS and CL both decrease as temperature T of the S/L interface drops
during solidification.
146 BASIC SOLIDIFICATION CONCEPTS
solidus line liquidus line;
slope m L < 0Solute concentration, CTemperature, TEquilibriumsegregationcoefficient:k = CS/CL < 1melt of initialcomposition, COS S + LLkCOTmCLTTLCOTemperature, TCOSL S + LkCOTmCLCSTTLCOSolute concentration, Ck = CS/CL > 1
(a)
(b)
CT S S
TS
liquidus line;
slope m L > 0 solidus line
Figure 6.1 Portion of a binary phase diagram showing the equilibrium partition
ratio k.
B. Slope of Liquidus Line The slope of the liquidus line,mL, is less than zero
when k < 1 and vice versa, as shown in Figure 6.1. If the liquidus line is straight,a temperature T on the liquidus line can be expressed asT = Tm + mLCL (6.2)6.1.2 Complete Diffusion in Solid and LiquidTo help understand solute redistribution during solidification, consider theone-dimensional solidification experiment shown in Figure 6.2. A metal heldin a stationary container of an aluminum oxide tube is heated by a tubularfurnace at the top and cooled by a water cooler at the bottom. As thefurnace–cooler assembly rises steadily, the metal in the container solidifiesupward with a planar S/L interface. The growth rate R of the metal, that is,the travel speed of the S/L interface, can be adjusted by adjusting the risingspeed of the furnace–cooler assembly. The temperature gradient G in theliquid metal at the S/L interface can be adjusted by adjusting heating andcooling.This case of complete diffusion in both the solid and the liquid is shown inFigure 6.3, where a liquid metal of uniform initial composition C0 is allowedto solidify in one direction with a planar S/L interface, just like in the directionalsolidification experiment shown in Figure 6.2. This case is also calledequilibrium solidification because equilibrium exists between the entire solidand the entire liquid, not just at the interface. Diffusion is complete in the solidand the liquid, and the solid and liquid are thus uniform in composition.Uniform composition in the liquid requires either complete mixing by strongconvection or complete diffusion in the liquid. Complete diffusion of thesolute in the liquid requires DLt >> l2, where l is the initial length of the liquid,
DL the diffusion coefficient of the solute in the liquid, and t the time available
SOLUTE REDISTRIBUTION DURING SOLIDIFICATION 147
water
container
(stationary)
rising
tubular
furnace
(rising)
cooler
(rising)
distance, z
S/L (solid/liquid interface)
temperature, T
G: temperature gradient
dT/dz at S/L
R: growth rate
L: liquid metal
S: solid metal
L
R
s
Directional Solidification
Experiment
Figure 6.2 Directional solidification experiment.
for diffusion. This is because the square root of Dt is often considered as an
approximation for the diffusion distance, where D is the diffusion coefficient.
This requirement can be met when l is very small, when solidification is so slow
that the solute has enough time to diffuse across the liquid, or when the diffusion
coefficient is very high. Similarly, complete solute diffusion in the solid
requires DSt >> l2, where DS is the diffusion coefficient of the solute in the
solid. Since DS is much smaller than DL, complete diffusion is much more difficult
to achieve in solid than in liquid. Interstitial solutes (such as carbon in
steel) tend to have a much higher DS than substitutional solutes (such as Cr
in steel) and are more likely to approach complete diffusion in solid.
During solidification the composition of the entire solid follows the solidus
line and that of the entire liquid follows the liquidus line (Figure 6.3a). The
composition of the solid is a function of the fraction of the solid, that is, CS( fS).
Likewise the composition of the liquid is a function of the fraction of the solid,
148 BASIC SOLIDIFICATION CONCEPTS
C
0
Co
fL = 1 - fS fS
solid liquid
CL
CS
1
Concentration, C
Fraction of solid, fS
kCo
Co
C
0 1
Equilibrium
solidification
kCo Co/k
Co
Co/k
k < 1(b)CL(fS)CS(fS)(c)CSfS + CLfL = Cosolid liquid totalS/Lsolid liquidinitial meltcomposition,CoTemperature, TS S + LLTmTLTST CSCL (a)fStotal volume of materialFigure 6.3 Solute redistribution during solidification with complete diffusion in solidand liquid: (a) phase diagram; (b) CL( fs) and CS( fs); (c) composition profiles in solidand liquid.that is, CL(fS). At the onset of solidification at TL, the fraction of solid fs = 0,and the compositions of the solid and the liquid are kC0 and C0, respectively.As solidification continues and temperature drops from TL to TS, the compositionof the entire solid rises from kC0 to C0 and that of the entire liquid fromC0 to C0/k (Figure 6.3b). At any time during solidification, the compositions ofthe solid and liquid are uniform (as shown by the thick horizontal lines inFigure 6.3c). Solidification ends at the solidus temperature TS and the resultantsolid has a uniform composition of C0; that is, there is no solute segregationat all.In Figure 6.3c the hatched areas CS fS and CLfL represent the amountsof solute in the solid and liquid, respectively, where fS and fL are the fractionsof solid and liquid, respectively. The area C0( fS + fL) represents the amount ofsolute in the liquid before solidification. Based on the conservation of soluteand the fact that fS + fL = 1,(6.3)Substituting fS = 1 - fL into the above equation, the following equilibrium leverrule is obtained:(6.4)(6.5)and the composition of the liquid is(6.6)6.1.3 No Solid Diffusion and Complete Liquid DiffusionThis case is shown in Figure 6.4. Diffusion in the solid is assumed negligible,and the solid is thus not uniform in composition. This requires that DSt << l 2.On the other hand, diffusion in the liquid is assumed complete, and the liquidis thus uniform in composition. This requires that DLt >> l 2. If mixing caused
by strong convection is complete in the liquid, the liquid composition can also
be uniform. Equilibrium exists between the solid and the liquid only at the
interface.
Unlike the case of equilibrium solidification, the solute cannot back
diffuse into the solid and all the solute rejected by the growing solid has to
go into the liquid. Consequently, CL rises more rapidly during solidification
than in the case of equilibrium solidification. Since the solid grows from the
liquid, its composition at the S/L interface CS also rises more rapidly than in
C
C
f k f L
L L
=
+ (- )
0
1
f
C C
C C S
L
L S
=
-
-
0
f
C C
C C L
S
L S
=
-
-
0
CSfS+CLfL=C0(fS+fL)=C0
SOLUTE REDISTRIBUTION DURING SOLIDIFICATION 149
the case of equilibrium solidification. When CL rises beyond C0 /k, CS rises
beyond C0 (Figure 6.4b).
The two hatched areas in Figure 6.4c represent the amounts of solute in the
solid and liquid, and their sum equals the amount of solute in the liquid before
solidification, C0. Consider the conservation of solute shown in Figure 6.4d.
Since the total amount of solute in the system (total hatched area) is conserved,
the area under equals the area under. As such, hatched
area 1 equals hatched area 2 and, therefore,
abcd ab¢c¢d¢
150 BASIC SOLIDIFICATION CONCEPTS
(a)
(b)
Fraction of Solid, fs
kCo
Co
C
0 1
c
f
Co/k
Complete
Liquid Diffusion
CL(fS)
CS(fS)
kCo Co Co/k Concentration, C
k < 1Linitial meltcomposition,CoTemperature, TS S + LTmTLTSTCSCLC0 1solid liquidsolid liquidCoCLCS(c)Fraction of Solid, f SkCoCoConcentration, C0 1CLCSdfSdCL(CL - CS )dfS = (1 - fS)dCLarea 1 area 2area 1area 2fSa bcdc' d'b'S/LfL = 1 - f Ssolid liquidk < 1(d)S/Ltotal volume of materialfSFigure 6.4 Solute redistribution during solidification with complete diffusion in liquidand no diffusion in solid: (a) phase diagram; (b) CL( fs) and CS( fs); (c) composition profilesin solid and liquid; (d) conservation of solute.(6.7)Substituting CS = kCL into the above equation and integrating from CL = C0at fs = 0,(6.8)or(6.9)The above two equations are the well-known Scheil equation for solute segregation(1). Equation (6.9) can be written for the fraction liquid as follows:(6.10)The nonequilibrium lever rule, that is, the counterpart of Equation (6.5), canbe written as(6.11)From the Scheil equation the average composition of the solid, , can bedetermined as(6.12)As already mentioned, the solidus and liquidus lines are both assumedstraight.From Figure 6.1a, it can be shown that (TL - Tm)/C0 = mL. Furthermore,based on the proportional property between two similar right triangles, it canbe shown that (Tm - TL)/(Tm - T) = C0/CL. Based on these relationships andEquation (6.10), the Scheil equation can be rewritten as follows:(6.13)This equation can be used to determine the fraction of solid fs at a temperatureT below the liquidus temperature TL.f fT TT Tm CT TkokL Sm LmLm= - =--Êˈ¯=(- )-Êˈ¯- -11 1 1 1CkC f dfdfC fff kfkS0 S S0 SSSSS =(- )=- [-(- )] ÚÚ011 01 1CSfC CC CSLL S=--0fCCCCk kLLL=Êˈ¯=Êˈ¯( -) (- )01 101 1C CfkL= L-01C kC fk S= (-S)-01 1(CL-CS)dfS=(1-fS)dCLSOLUTE REDISTRIBUTION DURING SOLIDIFICATION 1516.1.4 No Solid Diffusion and Limited Liquid DiffusionThis case is shown in Figure 6.5. In the solid diffusion is assumed negligible.In the liquid diffusion is assumed limited, and convection is assumed negligible.Consequently, neither the solid nor the liquid is uniform in compositionduring solidification. Because of limited liquid diffusion, the solute rejected bythe solid piles up and forms a solute-rich boundary layer ahead of the growthfront, as shown in Figure 6.5c. Again, equilibrium exists between the solid andthe liquid only at the interface.152 BASIC SOLIDIFICATION CONCEPTSConcentration, CTemperature, Tinitial transientfinal transientS S + LL(a)steadystateCo/kinitial meltcomposition,CoinitialtransientfinaltransientsteadystateFraction Solid, f S(b)kCoCo0 1CTmTLTSkCo Co Co/kLimited LiquidDiffusionk < 10 1solute-richboundary layersolid liquid (c)CCL(fS)CS(fS)Co/kkCoCoS/Lsolid liquidfStotal volume of materialFigure 6.5 Solute redistribution during solidification with limited diffusion in liquidand no diffusion in solid: (a) phase diagram; (b) CL( fs) and CS( fs); (c) composition profilesin solid and liquid.Unlike the previous case of complete liquid diffusion and no solid diffusion,the solute rejected by the growing solid forms a solute-rich boundarylayer ahead of the growth front, rather than spreading uniformly in the entireliquid. Consequently, CL and hence CS rise more rapidly than those in the caseof complete liquid diffusion and no solid diffusion.The period of rising CL andCS is called the initial transient (Figure 6.5b).When CL and CS reach C0/k andC0, respectively, a steady-state period is reached, within which CL, CS, and theboundary layer (Figure 6.5c) remain unchanged. As the boundary layer movesforward, it takes in a liquid of composition C0 and it leaves behind a solid ofthe same composition C0. Since the input equals the output, the boundary layerremains unchanged. This steady-state condition continues until the boundarylayer touches the end of the liquid, that is, when the thickness of the remainingliquid equals that of the steady-state boundary layer. Here the volumeof the remaining liquid is already rather small, and any further decrease involume represents a very significant percentage drop in volume. As such,the remaining liquid quickly becomes much more concentrated in solute assolidification continues, and CL and hence CS rise sharply. This final period ofrapidly rising CL and CS is called the final transient, and its length equals thethickness of the steady-state boundary layer.The solute-rich boundary layer is further examined in Figure 6.6. It has beenshown mathematically (2) that the steady-state composition profile in theboundary layer isSOLUTE REDISTRIBUTION DURING SOLIDIFICATION 153solute-richboundarylayerConcentration, CC Co/k oTLTSSLTemperature, TCCodistance, xDL /R(a)(b)growth rate, RkCoCo/kTmS + LS/LkCoinitial meltcomposition,CoLimited LiquidDiffusionZk < 1solid liquidδ =Figure 6.6 Solidification with limited diffusion in liquid and no diffusion in solid: (a)phase diagram; (b) solute-rich boundary layer.(6.14)where R is the growth rate and Z the distance from the S/L interfacepointing into the liquid. At Z = 0, CL - C0 has a maximum value of C0 /k -C0 and at Z = DL/R it drops to 1/e (about one-third) of the maximumvalue. As such, the thickness of the boundary layer at the steady state can betaken as d DL/R. As such, the characteristic length of the final transient isDL/R.It also has been shown mathematically (3, 4) that, for k << 1, the compositionprofile in the initial transient can be expressed as(6.15)where x is the distance from the starting point of solidification. At x = 0, C0 -CS has a maximum value of C0 - kC0 and at x = DL/kR it drops to 1/e ofthe maximum value. As such, the characteristic length of the initial transientis DL/kR, which is much greater than that of the final transient because k issignificantly less than 1.Equation (6.15) can be rearranged as follows:(6.16)Upon taking the logarithm on both sides, the above equation becomes(6.17)The left-hand-side of Equation (6.17) can be plotted against the distancex. From the intercept the value of k can be checked, and from the slopethe length of the initial transient DL/kR can be found. From this thegrowth rate R can be found if the diffusion coefficient DL is known or viceversa.Figure 6.7 shows three different types of solute of redistributions thatcan occur when diffusion in the solid is negligible during solidification. Intype 1 either liquid diffusion or convection-induced mixing in the liquidis complete, and the resultant solute segregation is most severe (Figure 6.7d).In type 3, on the other hand, liquid diffusion is limited and there is noconvection-induced mixing in the liquid, and the resultant solute segregationis least severe. Type 2 is intermediate, so is the resultant solutesegregation.log loglog1 10- Êˈ¯= - ( )-Êˈ¯CCkeD kRx SL1 10-C =( - )-( )CS k e k R DL xC CC kC0 e k R DL x0 0--S = -( )C CC k CL - e R DL Z-0 = -( )0 0154 BASIC SOLIDIFICATION CONCEPTS6.2 SOLIDIFICATION MODES ANDCONSTITUTIONAL SUPERCOOLING6.2.1 Solidification ModesDuring the solidification of a pure metal the S/L interface is usually planar,unless severe thermal undercooling is imposed. During the solidification of analloy, however, the S/L interface and hence the mode of solidification can beplanar, cellular, or dendritic depending on the solidification condition and thematerial system involved. In order to directly observe the S/L interface duringsolidification, transparent organic materials that solidify like metals have beenused. Shown in Figure 6.8 are the four basic types of the S/L interfacemorphology observed during the solidification of such transparent materials:planar, cellular, columnar dendritic, and equiaxed dendritic (5, 6). TypicalSOLIDIFICATION MODES AND CONSTITUTIONAL SUPERCOOLING 155liquidCokCoCokCoCo/kCokCoSolute concentration, CCokCoType 1: complete liquiddiffusion or mixingType 3: limited liquid diffusion,no convectionFraction Solid, f s(a)(b)(c)Type 2 (d)Type 1 (mostsegregation)Type 3 (least segregation)Type 2: limited liquid diffusion,some convectionsolidsolidCliquidliquidsolidsolidsolid liquidFigure 6.7 Solute redistributions in the absence of solid diffusion: (a) type 1; (b) type2; (c) type 3; (d) resultant segregation profiles.microstructures resulting from the cellular, columnar dendritic, and equiaxeddendritic modes of solidification in alloys are shown in Figures 6.9a, b, and c,respectively (7–9). A three-dimensional view of dendrites is shown in Figure6.9d (10).6.2.2 Constitutional SupercoolingTwo major theories have been proposed to quantitatively describe the breakdownof a planar S/L interface during solidification: the constitutional supercoolingtheory by Chalmer and co-workers (11, 12) and the interface stabilitytheory by Mullins and Sekerka (13–15).The former theory considers only thethermodynamic aspect of the problem while the latter incorporates the interfacekinetic and heat transfer aspects. For simplicity, however, only the constitutionalsupercooling theory will be described here.Consider the solidification of alloy C0 at the steady state with a planar S/Linterface, as shown in Figure 6.10. As shown previously in Figure 6.6, the compositiondistribution in the solute-rich boundary layer is shown in Figure 6.10b.156 BASIC SOLIDIFICATION CONCEPTSFigure 6.8 Basic solidification modes (magnification 67¥): (a) planar solidification ofcarbon tetrabromide (5); (b) cellular solidification of carbon tetrabromide with a smallamount of impurity (5); (c) columnar dendritic solidification of carbon tetrabromidewith several percent impurity (5); (d) equiaxed dendritic solidification of cyclohexanolwith impurity (6). From Solidification (5), pp. 132–134, with permission.The liquidus temperature distribution corresponding to this composition distributioncan be constructed point by point from the liquidus line of the phasediagram and is shown in Figure 6.10c.A boundary layer consisting of the liquidphase alone is thermodynamically stable only if its temperature is above theliquidus temperature. If its temperature is below the liquidus temperature,solid and liquid should coexist.This means that the planar S/L interface shouldSOLIDIFICATION MODES AND CONSTITUTIONAL SUPERCOOLING 157Figure 6.9 Nonplanar solidification structure in alloys. (a) Transverse section of a cellularlysolidified Pb–Sn alloy from Journal of Crystal Growth (7) (magnification 48¥).(b) Columnar dendrites in a Ni alloy. From New Trends in Materials Processing (8),with permission. (c) Equiaxed dendrites of a Mg–Zn alloy from Journal of Inst. ofMetals (9) (magnification 55¥). (d) Three-dimensional view of dendrites in a Ni-basesuperalloy. Reprinted from International Trends in Welding Science and Technology(10).break down to a cellular or dendritic one so that solid cells or dendrites cancoexist with the intercellular or interdendritic liquid (Figures 6.8b–d). Theshaded area under the liquidus temperature distribution in Figure 6.10c indicatesthe region where the actual liquid temperature is below the liquidus temperature,that is, the region of constitutional supercooling. The shaded areabelow the liquidus line in Figure 6.10a, which depends on DL and R, indicatesthe corresponding region of constitutional supercooling in the phase diagram.Obviously, this area lies within the solid–liquid region of the phase diagram,that is, liquid alone is unstable and solid and liquid should coexist.The temperature difference across the boundary layer is the equilibriumfreezing range DT = TL - TS. As shown previously, the thickness of the boundarylayer at the steady state is DL/R. As such, the slope of the tangent to theliquidus temperature distribution at the S/L interface is DT/(DL/R) or RDT/DL.To ensure that a planar S/L interface is stable, the actual temperature gradientG at the S/L interface must be at least RDT/DL. Therefore, for a planar S/Linterface to be stable at the steady state, the following criterion must be met:(6.18)GRTD≥DL158 BASIC SOLIDIFICATION CONCEPTSdistance from S/Lsolute-richboundary layerregion ofconstitutionalsupercoolingC Concentration, C oC /k oTLTS SLTTemperature, TTTCCo/kCoLGDL Ror
Δ
Δ
Δ Δ Δ
Δ
δ =
Figure 6.10 Constitutional supercooling: (a) phase diagram; (b) composition profile
in liquid; (c) liquidus temperature profile in liquid.
This is the steady-state form of the criterion for planar growth. It says that for
an alloy to be able to grow with a planar solidification mode, the ratio G/R
must be no less than DT/DL. The constitutional supercooling theory has been
verified experimentally by many investigators (16–23). In general, this theory
predicts fairly closely the conditions required to initiate the breakdown of a
planar S/L interface in alloys with isotropic surface energy.
According to Equation (6.18), the higher the temperature gradient G and
the lower the growth rate, the easier for a planar S/L interface to be stable.
On the other hand, the higher the freezing range DT and the lower the diffusion
coefficient DL, the more difficult for a planar S/L interface to be stable.
For an Al–4% Cu alloy, for example, TL, TS and DL, are 650°C, 580°C, and 3 ¥
10-5cm2/s, respectively. If the temperature gradient G is 700°C/cm, the growth
rate R has to be less than or equal to 3 ¥ 10-4 cm/sec in order to have planar
solidification. If the growth rate is higher than this, the planar S/L interface
will break down and cellular or dendritic solidification will take place.
Figure 6.11 shows that the solidification mode changes from planar to cellular,
to columnar dendritic, and finally to equiaxed dendritic as the degree of
constitutional supercooling continues to increase.The region where dendrites
(columnar or equiaxed) and the liquid phase coexist is often called the mushy
zone (2). It is interesting to note that at a very high degree of constitutional
supercooling (Figure 6.11d) the mushy zone can become so wide that it is
easier for equiaxed dendrites to nucleate than for columnar dendrites to
stretch all the way across the mushy zone. Unfortunately, simple theories
SOLIDIFICATION MODES AND CONSTITUTIONAL SUPERCOOLING 159
Planar
Cellular
Columnar
dendritic
Equiaxed
dendritic
L
L
L
L
Equilibrium
Actual
M
M
S
S
S
S
Increasing constitutional supercooling
constitutional
supercooling
(a)
(b)
(c)
(d)
Figure 6.11 Effect of constitutional supercooling on solidification mode: (a) planar;
(b) cellular; (c) columnar dendritic; (d) equiaxed dendritic (S, L, and M denote solid,
liquid, and mushy zone, respectively).
similar to the constitutional supercooling theory are not available for predicting
the transitions from the cellular mode to the columnar dendritic mode
and from the columnar dendritic mode to the equiaxed dendritic mode.
Figure 6.12 is a series of photographs showing the breakdown of a planar
S/L interface into a cellular one during the solidification of a pivalic acid
alloyed with 0.32mol % ethanol (24). The temperature gradient G was
15°C/mm, and the growth rate R was suddenly raised to a higher level of
5.7mm/s, to suddenly lower the G/R ratio and trigger the breakdown by constitutional
supercooling.
6.3 MICROSEGREGATION AND BANDING
6.3.1 Microsegregation
Solute redistribution during solidification results in microsegregation across
cells or dendrite arms. The analysis of solute redistribution during the direc-
160 BASIC SOLIDIFICATION CONCEPTS
Figure 6.12 Breakdown of a planar S/L interface during the solidification of pivalic
acid–ethanol: (a) 0 s, (b) 120 s, (c) 210 s, (d) 270 s, (e) 294 s, ( f) 338 s, (g) 378 s, (h) 456 s,
(i) 576 s, and (j) 656 s. Reprinted from Liu and Kirkaldy (24). Copyright 1994 with permission
from Elsevier Science.
tional solidification of a liquid metal (Section 6.1) can be applied to solute
redistribution during the solidification of an intercellular or interdendritic
liquid during welding (or casting). The total volume of material in directional
solidification (Figures 6.3–6.5) is now a volume element in a cell or a dendrite
arm, as shown in Figure 6.13. Within the volume element the S/L interface
is still planar even though the overall structure is cellular or dendritic. The
volume element covers the region from the centerline of the cell or dendrite
arm to the boundary between cells or dendrite arms. Solidification begins
in the volume element when the tip of the cell or dendrite arm reaches the
volume element.
The case of the equilibrium partition ratio k < 1 is shown in Figure 6.14a.No segregation occurs when diffusion is complete in both the liquid and solid.MICROSEGREGATION AND BANDING 161LSSL(a) (b)volumeelementFigure 6.13 Volume elements for microsegregation analysis: (a) cellular solidification;(b) dendritic solidification.kCokCoCoCoboundary between cellsor dendrite armscomplete liquid diffusion,no solid diffusioncomplete liquid &solid diffusionlimited liquid diffusion,no solid diffusionCodistancecompositioncenterline of cellor dendrite armcell ordendrite armk < 1volumeelementkCokCoCoCocomplete liquid diffusion,no solid diffusioncomplete liquid &solid diffusionlimited liquid diffusion,no solid diffusionCodistancecompositionk > 1
cell or
dendrite arm
(a) (b)
Figure 6.14 Microsegregation profiles across cells or dendrite arms: (a) k < 1; (b) k > 1.
This requires that DLt >> l 2 and that DSt >> l 2, where l is now half the cell or
dendrite arm spacing (the length of the volume element). Segregation is worst
with complete diffusion in the liquid but no diffusion in the solid.This requires
that DLt >> l 2 and that DSt << l 2. Segregation is intermediate with limited diffusionin the liquid and no diffusion in the solid. When this occurs, there isa clear concentration minimum at the centerline of the cell or dendrite arm.Usually, there is some diffusion in the solid and the concentration minimummay not always be clear. The case of k > 1 is shown in Figure 6.14b. The segregation
profiles are opposite to those of k < 1.Consider the case of a eutectic phase diagram, which is common amongaluminum alloys. Assume complete liquid diffusion and no solid diffusion. Asshown in Figure 6.15, the solid composition changes from kC0 to CSM, themaximum possible solute content in the solid, when the eutectic temperatureTE is reached. The remaining liquid at this point has the eutectic composition162 BASIC SOLIDIFICATION CONCEPTSConcentration, CTemperature, TSS + LL(a)(b)Fraction of Solid, fskCoCoC0 1kCoCoinitial meltcomposition,CoTmTLTSCSMCompleteLiquid DiffusionCEeutecticCEeutecticCSMTEsolidfECL(f S)CS(f S)volumeelementFigure 6.15 Solute redistribution during solidification with complete diffusion inliquid and no diffusion in solid: (a) eutectic phase diagram; (b) composition profiles inliquid and solid.CE and, therefore, solidifies as solid eutectic at TE. The fraction of eutectic, fE,can be calculated from Equation (6.13) with fL = fE and T = TE, that is,(6.19)6.3.2 BandingIn addition to microsegregation, solute segregation can also occur as a resultof growth rate fluctuations caused by thermal fluctuations. This phenomenon,shown in Figure 6.16, is known as banding (2). As shown, steady-state solidificationoccurs at the growth rate R1. When the growth rate is suddenlyincreased from R1 to R2, an extra amount of solute is rejected into the liquidat the S/L interface, causing its solute content to rise. As a result, the materialsolidifies right after the increase in the growth rate has a higher solute concentrationthan that before the increase. The boundary layer at R2 is thinnerthan that at R1 because, as mentioned previously, the boundary layer thicknessis about DL/R. The solute concentration eventually resumes its steady value ifno further changes take place. If, however, the growth rate is then decreasedsuddenly from R2 back to R1, a smaller amount of solute is rejected into theliquid at the S/L interface, causing its solute content to drop. As a result, thematerial solidifies right after the decrease in the growth rate has a lower soluteconcentration than that just before the decrease. In practice, the growth ratecan vary as a result of thermal fluctuations during solidification caused byunstable fluid flow in the weld pool, and solute-rich and solute-depleted bandsform side by side along the solidification path.When the flow carrying heatfrom the heat source and impinging on the growth front suddenly speeds up,the growth rate is reduced and vice versa.6.4 EFFECT OF COOLING RATEIt has been observed that the higher the cooling rate, that is, the shorter thesolidification time, the finer the cellular or dendritic structure becomes (2).fT TT TkEm Lm E=--Êˈ¯1(1- )EFFECT OF COOLING RATE 163Solute concentration, CCoCo /kTime, tR = R1R = R1 < R2 R = R2 > R1
soluterich
band
solutedepleted
band
Figure 6.16 Formation of banding due to changes in the solidification rate R.
Figure 6.17 shows the dendrite arm spacing as a function of the cooling rate
or solidification time in three different materials (25–27).The relationship can
be expressed as (2)
(6.20)
where d is the secondary dendrite arm spacing, tf is the local solidification time,
e is the cooling rate, and a and b are proportional constants.
Figure 6.18 depicts the growth of a dendrite during solidification (2). The
window for viewing the growing dendrite remains stationary as the dendrite
tip advances. As can be seen through the window, large dendrite arms grow at
the expense of smaller ones as solidification proceeds. Since smaller dendrite
arms have more surface area per unit volume, the total surface energy of the
solidifying material in the window can be reduced if larger dendrite arms grow
at the expense of the smaller ones. The slower the cooling rate during solidid
at b f
n n = = ( )- e
164 BASIC SOLIDIFICATION CONCEPTS
10-4 10-2 1 102 104 106 108
1
102
104
Average local cooling rates, oC/s
Secondary dendrite
arm spacing, m
Al-4.5Cu
100 101 102 103 104 105 106
0.1
1
10
100 Brower et al.
Cooling rate, oC/s
Abdulgadar
201 stainless steel
Secondary dendrite
arm spacing, m
Solidification time, sec
.01 .02 .04.06 .10 .20
1
2
4
6 8
10
310 stainless steel
Secondary dendrite
arm spacing, m μ μ μ
Figure 6.17 Effect of cooling rate or solidification time on dendrite arm spacing: (a)
For Al–4.5Cu. From Munitz (25), (b) For 201 stainless steel. From Paul and DebRoy
(26). (c) For 310 stainless steel. From Kou and Le (27).
(a)
(b)
(c)
fication, the longer the time available for coarsening and the larger the dendrite
arm spacing (Figure 6.17).
Figure 6.19 shows how GR affects the cell spacing in Sn–Pb alloys solidified
in the cellular mode (2, 28).The product GR (°C/cm times cm/s) is, in fact,
the cooling rate, judging from its unit of °C/s. As shown, the higher the cooling
rate, that is, the lower (GR)-1, the finer the cells become.
The effect of the temperature gradient G and the growth rate R on the
solidification microstructure of alloys is summarized in Figure 6.20. Together,
G and R dominate the solidification microstructure.The ratio G/R determines
the mode of solidification while the product GR governs the size of the solidification
structure.
EFFECT OF COOLING RATE 165
Figure 6.18 Schematic showing dendritic growth of an alloy at a fixed position at
various stages of solidification. Note that several smaller arms disappear while larger
ones grow. Reprinted from Flemings (2).
10 20 30 50 70 100 200
0
20
40
60
80
100
(GR)-1 (oC/sec)-1
Cell spacing (microns)
Figure 6.19 Effect of GR on cell spacing in Sn–Pb alloys. Modified from Flemings (2)
based on data from Plaskett and Winegard (28).
6.5 SOLIDIFICATION PATH
Consider the eutectic phase diagram of a binary system A–B shown by the
broken lines in Figure 6.21a. To the left of the eutectic point, the primary
solidification phase is g, namely, g is the first thing to solidify. To the right, the
primary solidification phase is a. The solid line represents the solidification
path of alloy C0. The arrow in the solidification path indicates the direction
in which the liquid composition changes as temperature decreases. At step 1,
where the fraction of the liquid phase L is fL = 1, g begins to form from the
liquid phase, assuming negligible undercooling before solidification.The composition
of the liquid phase follows the solid line representing L Æ g. At step
2, solidification is complete and fL = 0.
In the case of a ternary system A–B–C, the phase diagram is three dimensional,
with the base plane showing the composition and the vertical direction
showing the temperature. The liquidus is a surface (curved) instead of a line
as in a binary phase diagram (Figure 6.21a).The intersection between two liquidus
surfaces is a line called the line of twofold saturation, instead of a point
intersection between two liquidus lines in a binary phase diagram. As shown
in Figure 6.21b, when projected vertically downward to the base plane, the
liquidus surfaces show areas of primary solidification phases with the lines of
twofold saturation separating them.
The solid line shows the solidification path of alloy C0. At step 1, where the
fraction of the liquid phase L is fL = 1, the solid phase g begins to form from
166 BASIC SOLIDIFICATION CONCEPTS
a
high
G/R
G/R determines morphology
of solidification structure
GxR determines size of
solidification structure
higher cooling rate (GxR);
finer structure
lower cooling rate (GxR);
coarser structure
Planar
Cellular
Columnar
Dendritic
Equiaxed
Dendritic Low
G/R b
Growth rate, R
Temperature gradient, G
S L
S L
Figure 6.20 Effect of temperature gradient G and growth rate R on the morphology
and size of solidification microstructure.
the liquid phase. As temperature decreases, the composition of the liquid
phase follows the solid line representing L Æ g until the line of twofold saturation
is reached. At step 2, the a phase begins to form from the liquid,
too. As temperature decreases further, the composition of the liquid phase
follows the solid line representing the ternary eutectic reaction L Æg + a until
the line of twofold saturation is reached. At step 3, the formation of a from
the liquid stops and is taken over by the formation of b. As temperature
decreases still further, the composition of the liquid phase follows the solid
line representing the ternary eutectic reaction L Æ g + b. This can go on until
the line representing binary A–B (the abscissa) is reached, if there is still liquid
available to get there. The fraction of liquid fL = 0 when solidification is
complete.
REFERENCES
1. Scheil, E., Z. Metallk., 34: 70, 1942.
2. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
3. Pohl, R. G., J. Appl. Phys., 25: 668, 1954.
3n
m3
REFERENCES 167
Temperature, T
wt % B
L
L
A B
A wt % B
wt % C
L
L L
f L = 1
f L = 1
f L = 0
f L = 0
1
2
2
1
3
4
(a)
(b)
Co
liquidus line
L
m
n
primary primary
primary
primary
primary
primary
L
to pure B
to pure C
solidification
path lines of twofold saturation:
m3, 3n and 34
solidification
path
γ α
α
+ α
γ
γ + γ
γ
γ
γ + α γ + β
α
β
γ
Figure 6.21 Solidification paths: (a) a binary A–B system; (b) a ternary A–B–C
system.
4. Kou, S., Transport Phenomena and Materials Processing,Wiley, New York, 1996.
5. Jackson, K. A., in Solidification, American Society for Metals, Metals Park, OH,
1971, p. 121.
6. Jackson, K. A., Hunt, J. D., Uhlmann, D. R., Sewand III, T. P., Trans. AIME, 236:
149, 1966.
7. Morris, L. R., and Winegard,W. C., J. Crystal Growth, 5: 361, 1969.
8. Giamei, A. F., Kraft, E. H., and Lernkey, F. D., in New Trends in Materials Processing,
American Society for Metals, Metals Park, OH, 1976, p. 48.
9. Kattamis, T. Z., Holmber, U. T., and Flemings, M. C., J. Inst. Metals, 55: 343, 1967.
10. David, S. A., and Vitek, J. M., in International Trends in Welding Science and Technology,
Eds. by S. A. David and J. M. Vitek, ASM International, Materials Park,
OH, 1993, p. 147.
11. Rutter, J.W., and Chalmer, B., Can. J. Physiol., 31: 15, 1953.
12. Tiller,W. A., Jackson, K. A., Rutter, J.W., and Chalmer, B., Acta Met., 1: 428, 1953.
13. Mullins,W.W., and Sekerka, R. F., J. Appl. Physiol., 34: 323, 1963; 35: 444, 1964.
14. Sekerka, R. F., J. Appl. Physiol., 36: 264, 1965.
15. Sekerka, R. F., J. Phys. Chem. Solids, 28: 983, 1967.
16. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
17. Jackson, K. A., and Hunt, J. S., Acta Met., 13: 1212, 1965.
18. Walton,D.,Tiller,W. A., Rutter, J.W., and Winegard,W. C., Trans.AIME, 203: 1023,
1955.
19. Cole, G. S., and Winegard,W. C., J. Inst. Metals, 92: 323, 1963.
20. Hunt, M. D., Spittle, J. A., and Smith, R.W., The Solidification of Metals, Publication
No. 110, Iron and Steel Institute, London, 1968, p. 57.
21. Plaskett, T. S., and Winegard,W. C., Can. J. Physiol., 37: 1555, 1959.
22. Bardsley,W., Callan, J. M., Chedzey, H. A., and Hurle, D. T. J., Solid-State Electron,
3: 142, 1961.
23. Coulthard, J. O, and Elliott, R., The Solidification of Metals, Publication No. 110,
Iron and Steel Institute, 1968, p. 61.
24. Liu, L. X., and Kirkaldy, J. S., J. Crystal Growth, 144: 335, 1994.
25. Munitz, A., Metall. Trans., 16B: 149, 1985.
26. Paul, A. J., and DebRoy, T., Rev. Modern Phys., 67: 85, 1995.
27. Kou, S., and Le,Y., Metall. Trans., 13A: 1141, 1982.
28. Plaskett, T. S., and Winegard,W. C., Can. J. Physiol., 38: 1077, 1960.
FURTHER READING
1. Chalmers, B., Principles of Solidification, Wiley, New York, 1964.
2. Davies, G. J., Solidification and Casting, Wiley, New York, 1973.
3. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20: 83, 1975.
4. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
5. Savage,W. F., Weld.World, 18: 89, 1980.
168 BASIC SOLIDIFICATION CONCEPTS
PROBLEMS
6.1 From the Al–Mg phase diagram, the equilibrium freezing range of 5052
aluminum (essentially Al–2.5Mg) is about 40°C. Suppose the welding
speed is 4 mm/s and the diffusion coefficient DL is 3 ¥ 10-5cm2/s. Calculate
the minimum temperature gradient required for planar solidification
at the weld centerline.What is the corresponding cooling rate? Can this
level of cooling rate be achieved in arc welding.
6.2 Let CE and CSM be respectively 35% and 15% Mg, and both the solidus
and liquidus lines are essentially straight in the Al–Mg system. The
melting point of pure Al is 660°C, and the eutectic temperature is 451°C.
What is the approximate volume fraction of the aluminum-rich dendrites
in the fusion zone of autogenous 5052 aluminum weld?
6.3 It has been observed that aluminum alloys welded with the electron
beam welding process show much finer secondary dendrite arm spacing
in the weld metal than those welded with GMAW. Explain why.
6.4 Which alloy has a greater tendency for planar solidification to break
down, Al-0.01Cu or Al-6.3Cu and why?
6.5 How would preheating of the workpiece affect the secondary dendrite
arm spacing in welds of aluminum alloys and why?
PROBLEMS 169
7 Weld Metal Solidification I:
Grain Structure
In this chapter we shall discuss the development of the grain structure in the
fusion zone and the effect of welding parameters on the grain structure.Then,
we shall discuss various mechanisms of and techniques for grain refining. The
grain structure of the fusion zone can significantly affect its susceptibility to
solidification cracking during welding and its mechanical properties after
welding.
7.1 EPITAXIAL GROWTH AT FUSION BOUNDARY
7.1.1 Nucleation Theory
Figure 7.1 shows the nucleation of a crystal from a liquid on a flat substrate
with which the liquid is in contact (1, 2). The parameters gLC, gLS, and gCS are
the surface energies of the liquid–crystal interface, liquid–substrate interface,
and crystal–substrate interface, respectively. According to Turnbull (2), the
energy barrier DG for the crystal to nucleate on the substrate is
(7.1)
where Tm is the equilibrium melting temperature, DHm the latent heat of
melting, DT the undercooling below Tm, and q the contact angle. If the liquid
wets the substrate completely, the contact angle q is zero and so is DG. This
means that the crystal can nucleate on the substrate without having to overcome
any energy barrier required for nucleation. The energy barrier can be
significant if no substrate is available or if the liquid does not wet the substrate
completely.
In fusion welding the existing base-metal grains at the fusion line act as the
substrate for nucleation. Since the liquid metal of the weld pool is in intimate
contact with these substrate grains and wets them completely (q = 0), crystals
nucleate from the liquid metal upon the substrate grains without difficulties.
When welding without a filler metal (autogenous welding), nucleation occurs
by arranging atoms from the liquid metal upon the substrate grains without
D
D D
G
T
H T
=
( )
4 (- + )
3
2 3 2
3 pg
q q LC
3
m
2
m
cos cos
170
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
altering their existing crystallographic orientations. Such a growth initiation
process, shown schematically in Figure 7.2, is called epitaxial growth or epitaxial
nucleation. The arrow in each grain indicates its <100> direction. For
materials with a face-centered-cubic (fcc) or body-centered-cubic (bcc) crystal
structure, the trunks of columnar dendrites (or cells) grow in the <100> direction.
As shown, each grain grows without changing its <100> direction.
EPITAXIAL GROWTH AT FUSION BOUNDARY 171
Liquid (L)
Substrate (S)
Crystal (C)
Radius (r)
LS CS
LC
γ
γ γ
θ
Figure 7.1 Spherical cap of a crystal nucleated on a planar substrate from a liquid.
Base Metal
(substrate)
Fusion
Line
Welding Direction
Weld Pool
(liquid) <100>
Epitaxial
Growth
Grain
(crystal)
Figure 7.2 Epitaxial growth of weld metal near fusion line.
7.1.2 Epitaxial Growth in Welding
Savage et al. (3–8) first discovered epitaxial growth in fusion welding. By using
the Laue x-ray back-reflection technique, they confirmed the continuity of
crystallographic orientation across the fusion boundary. Savage and Hrubec
(6) also studied epitaxial growth by using a transparent organic material
(camphene) as the workpiece, as shown in Figure 7.3. Three grains are visible
in the base metal at the fusion line. All dendrites growing from each grain
point in one direction, and this direction varies from one grain to another.
From the welds shown in Figure 7.4 epitaxial growth at the fusion line is
evident (9).
Epitaxial growth can also occur when the workpiece is a material of more
than one phase. Elmer et al. (10) observed epitaxial growth in electron beam
welding of an austenitic stainless steel consisting of both austenite and ferrite.
As shown in Figure 7.5, both austenite (A) and ferrite (F) grow epitaxially at
the fusion line (dotted line) from the base metal to the weld metal (resolidified
zone).
7.2 NONEPITAXIAL GROWTH AT FUSION BOUNDARY
When welding with a filler metal (or joining two different materials), the weld
metal composition is different from the base metal composition, and the weld
metal crystal structure can differ from the base metal crystal structure.When
172 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.3 Epitaxial growth during the “welding” of camphene in the area indicated
by the square (magnification 75¥). Modified from Savage (6).
this occurs, epitaxial growth is no longer possible and new grains will have to
nucleate at the fusion boundary.
Nelson et al. (11) welded a type 409 ferritic stainless steel of the bcc structure
with a Monel (70Ni–30Cu) filler metal of the fcc structure and produced
a fcc weld metal. Figure 7.6 shows the fusion boundary microstructure. They
proposed that, when the base metal and the weld metal exhibit two different
crystal structures at the solidification temperature, nucleation of solid weld
metal occurs on heterogeneous sites on the partially melted base metal at the
fusion boundary. The fusion boundary exhibits random misorientations
between base metal grains and weld metal grains as a result of heterogeneous
nucleation at the pool boundary.The weld metal grains may or may not follow
NONEPITAXIAL GROWTH AT FUSION BOUNDARY 173
100 μm
GB 2 FL
FL GB 1 GB 2
GB 1
base
(b) metal
weld
metal
Figure 7.4 Epitaxial growth. (a) Near the fusion boundary of electron beam weld of
C103 alloy (magnification 400¥). Reprinted from O’Brien (9). Courtesy of American
Welding Society. (b) Near the fusion boundary of as-cast Al–4.5Cu welded with 4043
filler (Al–5Si).
special orientation relationships with the base metal grains they are in contact
with, namely, orient themselves so that certain atomic planes are parallel to
specific planes and directions in the base-metal grains.
7.3 COMPETITIVE GROWTH IN BULK FUSION ZONE
As described in the previous section, the grain structure near the fusion line of
a weld is dominated either by epitaxial growth when the base metal and the weld
metal have the same crystal structure or by nucleation of new grains when they
have different crystal structures.Away from the fusion line, however, the grain
structure is dominated by a different mechanism known as competitive growth.
During weld metal solidification grains tend to grow in the direction perpendicular
to pool boundary because this is the direction of the maximum temperature
gradient and hence maximum heat extraction. However, columnar
dendrites or cells within each grain tend to grow in the easy-growth direction.
Table 7.1 shows the easy-growth directions in several materials (12) and, as
shown, it is <100> for both fcc and bcc materials. Therefore, during solidification
grains with their easy-growth direction essentially perpendicular to the
pool boundary will grow more easily and crowd out those less favorably
oriented grains, as shown schematically in Figure 7.7. This mechanism of
competitive growth dominates the grain structure of the bulk weld metal.
7.4 EFFECT OF WELDING PARAMETERS ON
GRAIN STRUCTURE
The weld pool becomes teardrop shaped at high welding speeds and elliptical
at low welding speeds (Chapter 2). Since the trailing pool boundary of a
174 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
resolidified zone
FA
20 μm
Figure 7.5 Epitaxial growth of austenite (A) and ferrite (F) from the fusion line of
an austenitic stainless steel containing both phases. From Elmer et al. (10).
Figure 7.6 Fusion boundary microstructure in 409 ferritic stainless steel (bcc) welded
with Monel filler wire (fcc): (a) optical micrograph; (b) scanning electron micrograph.
White arrows: fusion boundary; dark arrows: new grains nucleated along fusion boundary.
Reprinted from Nelson et al. (11). Courtesy of American Welding Society.
TABLE 7.1 Easy-Growth Directions
Crystal Structure Easy-Growth Direction Examples
Face-centered-cubic (fcc) <100> Aluminum alloys,
austenitic stainless steels
Body-centered-cubic (bcc) <100> Carbon steels, ferritic
stainless steels
Hexagonal-close-packed (hcp) <101¯0> Titanium, magnesium
Body-centered-tetragonal (bct) <110> Tin
Source: From Chalmers (12).
teardrop-shaped weld pool is essentially straight, the columnar grains are also
essentially straight in order to grow perpendicular to the pool boundary, as
shown schematically in Figure 7.8a. On the other hand, since the trailing
boundary of an elliptical weld pool is curved, the columnar grains are also
176 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Base
Metal
Fusion
Line
Welding Direction
Weld
Pool
<100>
Competitive
Growth
Epitaxial
Growth
Figure 7.7 Competitive growth in bulk fusion zone.
Figure 7.8 Effect of welding speed on columnar-grain structure in weld metal: (a, b)
regular structure; (c, d) with axial grains.
curved in order to grow perpendicular to the pool boundary, as shown in
Figure 7.8b. Figure 7.9 shows the gas–tungsten arc welds of high-purity
(99.96%) aluminum made by Arata et al. (13). At the welding speed of
1000mm/min straight columnar grains point toward the centerline, while at
250mm/min curved columnar grains point in the welding direction.
Axial grains can also exist in the fusion zone. Axial grains can initiate from
the fusion boundary at the starting point of the weld and continue along the
length of the weld, blocking the columnar grains growing inward from the
fusion lines. Like other columnar grains, these axial grains also tend to grow
perpendicular to the weld pool boundary.With a teardrop-shaped pool, only
a short section of the trailing pool boundary can be perpendicular to the axial
direction, and the region of axial grains is thus rather narrow, as shown in
Figure 7.8c.With an elliptical weld pool, however, a significantly longer section
of the trailing pool boundary can be perpendicular to the axial direction, and
the region of axial grains can thus be significantly wider, as shown in Figure
7.8d. Figure 7.10 shows aluminum welds with these two types of grain
EFFECT OF WELDING PARAMETERS ON GRAIN STRUCTURE 177
Figure 7.9 Gas–tungsten arc welds of 99.96% aluminum: (a) 1000mm/min welding
speed; (b) 250 mm/min welding speed. From Arata et al. (13).
structure. Axial grains have been reported in aluminum alloys (14–16),
austenitic stainless steels (17), and iridium alloys (18).
7.5 WELD METAL NUCLEATION MECHANISMS
The mechanisms of nucleation of grains in the weld metal will be discussed
in the present section. In order to help understand these mechanisms,
the microstructure of the material around the weld pool will be discussed
first.
Figure 7.11 shows a 2219 aluminum (essentially Al–6.3% Cu) weld
pool quenched with ice water during GTAW (19). The S + L region around
the weld pool consists of two parts (Figure 7.11a): the partially melted material
178 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.10 Axial grains in GTAW: (a) 1100 aluminum at 12.7 mm/s welding speed;
(b) 2014 aluminum at 3.6 mm/s welding speed.
(clear) associated with the leading portion of the pool boundary and the
mushy zone (shaded) associated with the trailing portion. Area 1 (Figure
7.11b) covers a small portion of the partially melted material. Area 2 (Figure
7.11c) covers a small portion of the mushy zone as well as the partially melted
material.
Based on Figure 7.11, the microstructure around the weld pool boundary
of an alloy is shown schematically in Figure 7.12, along with thermal cycles at
the weld centerline and at the fusion line and a phase diagram. The eutectictype
phase diagram is common among aluminum alloys. As shown, the mushy
WELD METAL NUCLEATION MECHANISMS 179
Figure 7.11 Weld pool of 2219 aluminum quenched during GTAW: (a) overall view;
(b) microstructure at position 1; (c) microstructure at position 2. Modified from Kou
and Le (19). Courtesy of American Welding Society.
zone behind the trailing portion of the pool boundary consists of solid dendrites
(S) and the interdendritic liquid (L). The partially melted material
around the leading portion of the pool boundary, on the other hand, consists
of solid grains (S) that are partially melted and the intergranular liquid (L).
In summary, there is a region of solid–liquid mixture surrounding the weld
pool of an alloy.
Figure 7.13a shows three possible mechanisms for new grains to nucleate
during welding: dendrite fragmentation, grain detachment, and heterogeneous
nucleation (19). Figure 7.13b shows the fourth nucleation mechanism, surface
nucleation. These mechanisms, which have been well documented in metal
casting, will be described briefly below. The techniques for producing new
grains in the weld metal by these mechanisms will be discussed in a subsequent
section.
7.5.1 Dendrite Fragmentation
Weld pool convection (Chapter 4) can in principle cause fragmentation of dendrite
tips in the mushy zone, as illustrated in Figure 7.13a.These dendrite fragments
are carried into the bulk weld pool and act as nuclei for new grains to
form if they survive the weld pool temperature. It is interesting to note that
this mechanism has been referred to frequently as the grain refining mechanism
for weld metals without proof.
180 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
pool (L)
distance, x
time, t
T
centerline
fusion line
centerline
fusion line
fusion line
base metal (S)
welding
direction
TE
partially melted
material (S+L)
fusion
zone (S)
TL
(a)
(b)
(c)
Concentration, C
A
Temperature, T
L
S S + L
TL
TE
Co
mushy zone (S+L)
partially melted zone
Figure 7.12 Microstructure around the weld pool boundary: (a) phase diagram; (b)
thermal cycles; (c) microstructure of solid plus liquid around weld pool.
7.5.2 Grain Detachment
Weld pool convection can also cause partially melted grains to detach themselves
from the solid–liquid mixture surrounding the weld pool, as shown in
Figure 7.13a. Like dendrite fragments, these partially melted grains, if they
survive in the weld pool, can act as nuclei for the formation of new grains in
the weld metal.
7.5.3 Heterogeneous Nucleation
Foreign particles present in the weld pool upon which atoms in the liquid
metal can be arranged in a crystalline form can act as heterogeneous nuclei.
Figure 7.14 depicts heterogeneous nucleation and the growth of new grains in
the weld metal. Figure 7.15a shows two (dark) heterogeneous nuclei at the
centers of two equiaxed grains in an autogenous gas–tungsten arc weld of a
6061 aluminum containing 0.043% titanium (20). Energy dispersive spectrometry
(EDS) analysis, shown in Figure 7.15b, indicates that these nuclei are
rich in titanium (boron is too light to be detected by EDS). A scanning electron
microscopy (SEM) image of a nucleus is shown in Figure 7.15c. The
WELD METAL NUCLEATION MECHANISMS 181
Figure 7.13 Nucleation mechanisms during welding: (a) top view; (b) side view.
Reprinted from Kou and Le (19). Courtesy of American Welding Society.
182 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
TE
TL
Pool
Welding
Direction
Nucleus
Equiaxed
dendrite
Figure 7.14 Heterogeneous nucleation and formation of equiaxed grains in weld
metal.
Figure 7.15 Heterogeneous nuclei in GTAW of 6061 aluminum: (a) optical micrograph;
(b) EDS analysis; (c) SEM image; (d) SEM image of TiB2 particles in a grain
refiner for aluminum casting. (a, b) From Kou and Le (20). (d) Courtesy of Granger
(21).
WELD METAL NUCLEATION MECHANISMS 183
(c)
Figure 7.15 Continued
morphology of the nucleus is similar to that of the agglomerated TiB2 particles,
shown in Figure 7.15d, in an Al–5% Ti–0.2% B grain refiner for ingot
casting of aluminum alloys (21). This suggests that the TiB2 particles in the
weld metal are likely to be from the Al–Ti–B grain refiner in aluminum ingot
casting. Figure 7.16 shows a heterogeneous nucleus of TiN at the center of an
equiaxed grain in a GTA weld of a ferritic stainless steel (22).
As mentioned previously, Nelson et al. (11) have observed nucleation of
solid weld metal on heterogeneous sites on the partially melted base metal at
the fusion boundary when the weld metal and the base metal differ in crystal
structure. Gutierrez and Lippold (23) have also studied the formation of the
nondendritic equiaxed zone in a narrow region of the weld metal adjacent to
the fusion boundary of 2195 aluminum (essentially Al–4Cu–1Li). Figure 7.17
shows an example of the equiaxed zone in a 2195 weld made with a 2319
(essentially Al–6.3Cu) filler metal. The width of the equiaxed zone was found
to increase with increasing Zr and Li contents in the alloy.They proposed that
near the pool boundary the cooler liquid is not mixed with the warmer bulk
weld pool. Consequently, near the pool boundary heterogeneous nuclei such
as Al3Zr and Al3(LixZr1-x), which are originally present as dispersoids in the
base metal, are able to survive and form the nondendritic equiaxed zone, as
illustrated in Figure 7.18. By using a Gleeble thermal simulator, Kostrivas and
Lippold (24) found that the equiaxed zone could be formed by heating in the
temperature range of approximately 630–640°C and at temperatures above
640°C the normal epitaxial growth occurred. This is consistent with the proposed
mechanism in the sense that the heterogeneous nuclei can only survive
near the cooler pool boundary and not in the warmer bulk weld pool.
In the metal casting process the metal is superheated during melting before
it is cast. Since the nuclei are unstable at higher temperatures, they dissolve in
the superheated liquid in the casting process. As such, there is no equiaxed
184 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.16 TiN particle as heterogeneous nucleus in GTAW of ferritic stainless steel.
Reprinted from Villafuerte and Kerr (22).
Figure 7.17 Nondendritic equiaxed zone in narrow region adjacent to fusion boundary
of 2195 Al–Cu–Li alloy. Reprinted from Gutierrez and Lippold (23). Courtesy of
American Welding Society.
zone in the weld of an as-cast alloy 2195. However, the equiaxed zone occurs
again if the as-cast alloy 2195 is solution heat treated first and then welded.
Gutierrez and Lippold (23) proposed that because of the relatively low solubility
of Zr in solid aluminum,Al3Zr and Al3(LixZr1-x) particles precipitate out
of the solid solution during heat treating.
7.5.4 Surface Nucleation
The weld pool surface can be undercooled thermally to induce surface nucleation
by exposure to a stream of cooling gas or by instantaneous reduction or
removal of the heat input.When this occurs, solid nuclei can form at the weld
pool surface, as illustrated in Figure 7.13b. These solid nuclei then grow into
new grains as they shower down from the weld pool surface due to their higher
density than the surrounding liquid metal.
7.5.5 Effect of Welding Parameters on Heterogeneous Nucleation
Before leaving this section, an important point should be made about the
effect of welding parameters on heterogeneous nucleation. Kato et al. (25),
Arata et al. (26), Ganaha et al. (16), and Kou and Le (20) observed in commercial
aluminum alloys that the formation of equaixed grains is enhanced by
higher heat inputs and welding speeds. As shown in Figure 7.19, equiaxed
grains can form a band along the centerline of the weld and block off columnar
grains as the heat input and welding speed are increased (20). Kou and Le
(27) showed in Figure 7.20 that, as the heat input and the welding speed are
increased, the temperature gradient (G) at the end of the weld pool is reduced.
Furthermore, as the welding speed is increased, the solidification rate of the
WELD METAL NUCLEATION MECHANISMS 185
welding
direction
zone of nondendritic
equiaxed grains formed
by hetrogeneous
nucleation
base metal
containing
nuclei
weak flow near
pool boundary
fusion
boundary
cooler unmixed
liquid containing
nuclei
nuclei unable
to survive in
warmer bulk
weld pool
normal dendritic
structure
Figure 7.18 Mechanism for formation of nondendritic equiaxed zone in Al–Cu–Li
weld according to Gutierrez and Lippold (23).
weld metal (R) is also increased. As illustrated in Figure 7.21, the ratio G/R
should be decreased and the constitutional supercooling (28) in front of the
advancing solid–liquid interface should, therefore, be increased. Kato et al.
(25) and Arata et al. (26) proposed that the transition to an equiaxed grain
186 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.19 Effect of welding parameters on grain structure in GTAW of 6061
aluminum: (a) 70A ¥ 11V heat input and 5.1 mm/s welding speed; (b) 120A ¥ 11V
heat input and 12.7 mm/s welding speed. From Kou and Le (20).
Pool Pool
175 A
15 V
20 mm/sec
0 50 100 0 50
Welding direction
1100 aluminum
500 500
1000 1000
Temperature, oC
Temperature, oC
60 A
10 V
1.7 mm/sec
Distance, mm
(a)
Distance, mm
(b)
Figure 7.20 Effect of welding parameters on temperature gradient at weld pool end
of 1100 aluminum: (a) higher welding speed and heat input; (b) lower welding speed
and heat input. From Kou and Le (27).
structure is due to the existence of a sufficiently long constitutionally undercooled
zone in the weld pool. Ganaha et al. (16), however, indicated that the
transition is not due to constitutional supercooling alone. In fact, it was
observed that significant amounts of equiaxed grains formed only in those
alloys containing around or more than 0.01wt% Ti or 0.10wt% Zr. Furthermore,
tiny second-phase particles rich in titanium and/or zirconium (possibly
TixZryC compounds) existed at the dendrite centers of the equiaxed grains.
Consequently, it was proposed that equiaxed grains in the fusion zone form
by heterogeneous nucleation aided by constitutional supercooling.The observation
of Ganaha et al. (16) was confirmed by Kou and Le (19, 20, 29). The
effect of welding parameters on grain refining in aluminum welds was discussed
by Kou and Le (20).
7.6 GRAIN STRUCTURE CONTROL
The weld metal grain structure can affect its mechanical properties significantly.
Arata et al. (13) tensile tested aluminum welds in the welding direction.
The weld metal ductility of 99.96% aluminum dropped greatly when the
columnar grains pointed to the weld centerline (Figure 7.9a), that is, when the
grains became nearly normal to the tensile axis. Also, the weld metal tensile
strength of 5052 aluminum increased as the amount of equiaxed grains
increased.
GRAIN STRUCTURE CONTROL 187
Figure 7.21 Effect of welding parameters on heterogeneous nucleation: (a) low
constitutional supercooling at low welding speed and heat input; (b) heterogeneous
nucleation aided by high constitutional supercooling at high welding speed and heat
input.
The formation of fine equiaxed grains in the fusion zone has two main
advantages. First, fine grains help reduce the susceptibility of the weld metal
to solidification cracking during welding (Chapter 11). Second, fine grains can
improve the mechanical properties of the weld, such as the ductility and fracture
toughness in the case of steels and stainless steels.Therefore, much effort
has been made to try to grain refine the weld fusion zone. This includes the
application of grain refining techniques that were originally developed for
casting. Described below are several techniques that have been used to control
the weld metal grain structure.
7.6.1 Inoculation
This technique has been used extensively in metal casting. It involves the addition
of nucleating agents or inoculants to the liquid metal to be solidified. As
a result of inoculation, heterogeneous nucleation is promoted and the liquid
metal solidifies with very fine equiaxed grains. In the work by Davies and
Garland (1), inoculant powders of titanium carbide and ferrotitanium–
titanium carbide mixtures were fed into the weld pool during the submerged
arc welding of a mild steel and very fine grains were obtained. Similarly,
Heintze and McPherson (30) grain refined submerged arc welds of C-Mn and
stainless steels with titanium. Figure 7.22 shows the effect of inoculation on
the grain structure of the weld fusion zone of the C-Mn steel. It is interesting
to note that Petersen (31) has grain refined Cr–Ni iron base alloys with aluminum
nitride and reported a significant increase in the ductility of the resultant
welds, as shown in Figure 7.23.
Pearce and Kerr (32), Matsuda et al. (33), Yunjia et al. (34), and
Sundaresan et al. (35) grain refined aluminum welds by using Ti and Zr as
inoculants. Yunjia et al. (34) showed the presence of TiAl3 particles at the
origins of equiaxed grains in Ti microalloyed 1100 aluminum welds. Figure 7.24
shows grain refining in a 2090 Al–Li–Cu alloy gas–tungsten arc welded with a
2319 Al–Cu filler inoculated with 0.38% Ti (35).
7.6.2 External Excitation
Different dynamic grain refining techniques, such as liquid pool stirring, mold
oscillation, and ultrasonic vibration of the liquid metal, have been employed
in metal casting, and recently similar techniques, including weld pool stirring,
arc oscillation, and arc pulsation, have been applied to fusion welding.
A. Weld Pool Stirring Weld pool stirring can be achieved by electromagnetic
stirring, as shown in Figure 7.25, by applying an alternating magnetic field
parallel to the welding electrode (36). Matsuda et al. (37, 38) and Pearce and
Kerr (32) increased the degree of grain refinement in aluminum alloys containing
small amounts of titanium by electromagnetic pool stirring. Pearce and
Kerr (32) suggested that the increased grain refinement was due to heteroge-
188 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
neous nucleation, rather than to dendrite fragmentation. This is because in
GTAW, unlike ingot casting, the liquid pool and the mushy zone are rather
small, and it is therefore difficult for the liquid metal in the pool to penetrate
and break away dendrites, which are so short and so densely packed together.
GRAIN STRUCTURE CONTROL 189
Figure 7.22 Effect of inoculation on grain structure in submerged arc welds of C–Mn
steel (magnification 6¥): (a) without inoculation; (b) inoculation with titanium.
Reprinted from Heintze and McPherson (30). Courtesy of American Welding Society.
0 0.04 0.08 0.12 0.16
100
0
200
300
400
Grain size, mm
Elongation, %
Figure 7.23 Effect of grain size on weld metal ductility of a Cr–Ni iron base alloy at
925°C. Modified from Petersen (31).
Stirring of the weld pool tends to lower the weld pool temperature, thus
helping heterogeneous nuclei survive. Figure 7.26 shows the increased grain
refinement by weld pool stirring in an Al–2.5wt% Mg alloy containing
0.11wt% Ti (32). As shown in Figure 7.27, Villafuerte and Kerr (39) grain
refined a weld of 409 ferritic stainless steel containing 0.32% Ti by the same
technique. The higher the Ti content, the more effective grain refining was.
Titanium-rich particles were found at the origin of equiaxed grains, suggesting
heterogeneous nucleation as the grain refining mechanism.
Pearce and Kerr (32) also found that by applying weld pool stirring, the
partially melted grains along the leading portion of the pool boundary, if they
are only loosely held together, could be swept by the liquid metal into the weld
190 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.24 Effect of inoculation on grain structure in GTAW of 2090 Al–Li–Cu alloy:
(a) 2319 Al–Cu filler metal; (b) 2319 Al–Cu filler metal inoculated with 0.38% Ti.
Reprinted from Sundaresan et al. (35).
magnetic field
current field
coil
weld pool
workpiece
tungsten
electrode
welding direction
Figure 7.25 Schematic sketch showing application of external magnetic field during
autogenous GTAW. Modified from Matsuda et al. (36).
GRAIN STRUCTURE CONTROL 191
Figure 7.26 Widening of equiaxed zone in GTAW of alloy Al–2.5Mg–0.011Ti by
magnetic stirring (at dotted line). Reprinted from Pearce and Kerr (32).
Figure 7.27 Effect of electromagnetic pool stirring on grain structure in GTAW of
409 ferritic stainless steel: (a) without stirring; (b) with stirring. Reprinted from Villafuerte
and Kerr (39). Courtesy of American Welding Society.
pool, where they served as nuclei. Significant amounts of equiaxed grains were
so produced in gas–tungsten arc welds of a 7004 aluminum alloy containing
very little Ti.
B. Arc Oscillation Arc oscillation, on the other hand, can be produced by
magnetically oscillating the arc column using a single- or multipole magnetic
probe or by mechanically vibrating the welding torch. Davies and Garland (1)
produced grain refining in gas–tungsten arc welds of Al–2.5wt% Mg alloy by
torch vibration. Resistance to weld solidification cracking was improved in
these welds. Figure 7.28 shows the effect of the vibration amplitude on the
grain size (1). Dendrite fragmentation was proposed as the grain refining
mechanism. It is, however, suspected that heterogeneous nucleation could
have been the real mechanism, judging from the fact that the Al–2.5wt% Mg
used actually contained about 0.15wt%Ti. Venkataraman et al. (40, 41)
obtained grain refining in the electroslag welds of steels by electrode vibration
and enhanced electromagnetic stirring of the weld pool and improved
their toughness and resistance to centerline cracking. Dendrite fragmentation
was also considered as the mechanism for grain refining.
Sharir et al. (42) obtained grain refinement in gas–tungsten arc welds of
pure tantalum sheets by arc oscillation. Due to the high melting point of pure
tantalum (about 3000°C), the surface heat loss due to radiation was rather significant.
As a result, the liquid metal was cooled down rapidly and was in fact
undercooled below its melting point when the heat source was deflected away
during oscillated arc welding. This caused surface nucleation and resulted in
grain refinement.
192 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
150A, 200 mm/min
Arc gap 2.4 mm
Arc vibration frequency 10 Hz
Arc vibration amplitude, mm
Grain size, mm
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
0.2
0.4
0.1
0.3
0.5
Parallel to welding direction
Perpendicular to welding
direction
Figure 7.28 Effect of arc vibration amplitude on grain size in Al–2.5Mg welds. Modified
from Davies and Garland (1).
Sundaresan and Janaki Ram (43) grain refined Ti alloys by magnetic arc
oscillation and improved the weld metal tensile ductility. No specific grain
refining mechanism was identified.
C. Arc Pulsation Arc pulsation can be obtained by pulsating the welding
current. Sharir et al. (42) also obtained grain refinement in gas–tungsten arc
welds of pure tantalum sheets by arc pulsation. The liquid metal was undercooled
when the heat input was suddenly reduced during the low-current cycle
of pulsed arc welding. This caused surface nucleation and resulted in grain
refinement.
Figure 7.29 shows grain refining in a pulsed arc weld of a 6061 aluminum
alloy containing 0.04wt% Ti (44). Heterogeneous nucleation, aided by thermal
undercooling resulting from the high cooling rate produced by the relatively
high welding speed used and arc pulsation, could have been responsible for
the grain refinement in the weld.
7.6.3 Stimulated Surface Nucleation
Stimulated surface nucleation was originally used by Southin (45) to obtain
grain refinement in ingot casting. A stream of cool argon gas was directed on
the free surface of molten metal to cause thermal undercooling and induce
surface nucleation. Small solidification nuclei formed at the free surface and
showered down into the bulk liquid metal.These nuclei then grew and became
small equiaxed grains. This technique was used to produce grain refining in
Al–2.5Mg welds by Davies and Garland (1) and in Ti alloys by Wells (46).
GRAIN STRUCTURE CONTROL 193
Figure 7.29 Equiaxed grains in pulsed arc weld of 6061 aluminum (magnification 9¥).
From Kou and Le (44).
7.6.4 Manipulation of Columnar Grains
Kou and Le (14, 15, 47) manipulated the orientation of columnar grains in
aluminum welds by low-frequency arc oscillation. Figure 7.30a shows a 2014
aluminum weld made with 1Hz transverse arc oscillation, that is, with arc
oscillating normal to the welding direction (14). Similarly, Figure 7.30b shows
a 5052 aluminum weld made with 1 Hz circular arc oscillation (15). In both
194 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
Figure 7.30 Grain structures in oscillated arc welds of aluminum alloys. (a) For alloy
2014 with transverse arc oscillation. From Kou and Le (14). (b) For alloy 5052 with
circular arc oscillation. From Kou and Le (15).
cases, columnar grains grew perpendicular to the trailing portion of the weld
pool, and the weld pool in turn followed the path of the moving oscillating
arc.
As will be discussed later in Chapter 11, periodic changes in grain orientation,
especially that produced by transverse arc oscillation at low frequencies,
can reduce solidification cracking and improve both the strength and ductility
of the weld metal (47).
7.6.5 Gravity
Aidun and Dean (48) gas–tungsten arc welded 2195 aluminum under the high
gravity produced by a centrifuge welding system and eliminated the narrow
band of nondendritic equiaxed grains along the fusion boundary. As shown in
Figure 7.31, the band disappeared when gravity was increased from 1 g to
10 g. It was suggested that buoyancy convection enhanced by high gravity
caused Al3Zr and Al3(LixZr1-x) nuclei near the pool boundary to be swept into
the bulk pool and completely dissolved, thus eliminating formation of
equiaxed grains by heterogeneous nucleation.
REFERENCES
1. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20(196): 83, 1975.
2. Turnbull, D., J. Chem. Physiol., 18: 198, 1950.
3. Savage,W. F., Lundin, C. D., and Aronson, A. H., Weld. J., 44: 175s, 1965.
4. Savage,W. F., and Aronson, A. H., Weld. J., 45: 85s, 1966.
5. Savage,W. F., Lundin, C. D., and Chase, T. F., Weld. J., 47: 522s, 1968.
6. Savage,W. F., and Hrubec, R. J., Weld. J., 51: 260s, 1972.
REFERENCES 195
Figure 7.31 Effect of gravity on grain structure in GTAW of 2090 Al–Li–Cu alloy: (a)
1 g; (b) 10g and with equiaxed zone (EQZ) near the fusion boundary eliminated.
Reprinted from Aidun and Dean (48). Courtesy of American Welding Society.
7. Savage,W. F., Weld.World, 18: 89, 1980.
8. Savage,W. F., in Weld Imperfections, Eds. A. R. Hunger and R. E. Lewis, Addison-
Wesley, Reading, MA, 1968, p. 13.
9. 18th ed., O’brien R. L., Ed., Jefferson’s Welding Encyclopedia, American Welding
Society, Miami, FL, 1997, p. 316.
10. Elmer, J.W., Allen, S. M., and Eagar, T.W., Metall. Trans.A, 20A: 2117, 1989.
11. Nelson, T.W., Lippold, J. C., and Mills, M. J., Weld. J., 78: 329s, 1999.
12. Chalmers, B., Principles of Solidification,Wiley, New York, 1964, p. 117.
13. Arata,Y., Matsuda, F., Mukae, S., and Katoh, M., Trans. JWRI, 2: 55, 1973.
14. Kou, S., and Le,Y., Metall. Trans., 16A: 1887, 1985.
15. Kou, S., and Le,Y., Metall. Trans., 16A: 1345, 1985.
16. Ganaha, T., Pearce, B. P., and Kerr, H.W., Metall. Trans., 11A: 1351, 1980.
17. Kou, S., and Le,Y., Metall. Trans., 13A: 1141, 1982.
18. David, S. A., and Liu, C. T., Metals Technol., 7: 102, 1980.
19. Kou, S., and Le,Y., Weld. J., 65: 305s, 1986.
20. Kou, S., and Le,Y., Metall. Trans., 19A: 1075, 1988.
21. Granger, D. A., Practical Aspects of Grain Refining Aluminum Alloy Melts, paper
presented at International Seminar on Refining and Alloying of Liquid Aluminum
and Ferro-Alloys, Trondheim, Norway,August 26–28, 1985.
22. Villafuerte, J. C., and Kerr, H.W., in International Trends in Welding Science and
Technology, Eds. S. A. David and J. M. Vitek, ASM International, Materials Park,
OH, March 1993, p. 189.
23. Gutierrez, A., and Lippold, J. C., Weld. J., 77: 123s, 1998.
24. Kostrivas, A., and Lippold, J. C., Weld. J., 79: 1s, 2000.
25. Kato, M., Matsuda, F., and Senda, T., Weld. Res. Abroad, 19: 26, 1973.
26. Arata,Y., Matsuda, F., and Matsui, A., Trans. Jpn.Weld. Res. Inst., 3: 89, 1974.
27. Kou, S., and Le,Y., unpublished research, University of Wisconsin, Madison, 1985.
28. Rutter, J.W., and Chalmer, B., Can. J. Physiol., 31: 15, 1953.
29. Le,Y., and Kou, S., in Advances in Welding Science and Technology, Ed. S.A. David,
ASM International, Metals Park, OH, 1986, p. 139.
30. Heintze, G. N., and McPherson, R., Weld. J., 65: 71s, 1986.
31. Petersen,W. A., Weld. J., 53: 74s, 1973.
32. Pearce, B. P., and Kerr, H.W., Metall. Trans., 12B: 479, 1981.
33. Matsuda, F., Nakata, K., Tsukamoto, K., and Arai, K., Trans. JWRI, 12: 93, 1983.
34. Yunjia, H., Frost, R. H., Olson, D. L., and Edwards, G. R., Weld. J., 68: 280s, 1983.
35. Sundaresan, S., Janaki Ram, G. D., Murugesan, R., and Viswanathan, N., Sci.
Technol.Weld. Join., 5: 257, 2000.
36. Matsuda, F., Ushio, M., Nakagawa, H., and Nakata, K., in Proceedings of the
Conference on Arc Physics and Weld Pool Behavior, Vol. 1, Welding Institute,
Cambridge, 1980, p. 337.
37. Matsuda, F., Nakagawa, H., Nakata, K., and Ayani, R., Trans. JWRI, 7: 111, 1978.
38. Matsuda, F., Nakata, K., Miyawaga,Y., Kayano,T., and Tsukarnoto, K.,Trans. JWRI,
7: 181, 1978.
39. Villafuerte, J. C., and Kerr, H.W., Weld. J., 69: 1s, 1990.
196 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
40. Venkataraman, S., Devietian, J. H., Wood, W. E., and Atteridge, D. G., in Grain
Refinement in Castings and Welds, Eds. G. J. Abbaschian and S. A. David,
Metallurgical Society of AIME,Warrendale, PA, 1983, p. 275.
41. Venkataraman, S.,Wood,W. E., Atteridge, D. G., and Devletian, J. H., in Trends in
Welding Research in the United States, Ed. S.A. David,American Society for Metals,
Metals Park, OH, 1982.
42. Sharir,Y., Pelleg, J., and Grill, A., Metals Technol., 5: 190, 1978.
43. Sundaresan, S., and Janaki Ram, G. D., Sci. Technol.Weld. Join., 4: 151, 1999.
44. Kou, S., and Le,Y., unpublished research, University of Wisconsin, Madison, 1986.
45. Southin, R. T., Trans. AIME, 239: 220, 1967.
46. Wells, M. E., and Lukens,W. E., Weld. J., 65: 314s, 1986.
47. Kou, S., and Le,Y., Weld. J., 64: 51, 1985.
48. Aidun, D. K., and Dean, J. P., Weld. J., 78: 349s, 1999.
FURTHER READING
1. Chalmers, B., Principles of Solidification, Wiley, New York, 1964.
2. Davies, G. J., Solidification and Casting,Wiley, New York, 1973.
3. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20(196): 83, 1975.
4. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
5. Savage,W. F., Weld.World, 18: 89, 1980.
6. Abbaschian, G. J., and David, S. A., Eds., Grain Refinement in Castings and Welds,
Metallurgical Society of AIME,Warrendale, PA, 1983.
PROBLEMS
7.1 In aluminum alloys such as 6061 and 5052, which often contain small
amounts of Ti (say about 0.02wt%), the Ti-rich particles in the workpiece
can be dissolved with a gas–tungsten arc by multipass melting. If the
preweld is a multipass weld intended to dissolve such particles and the
grain structure is shown in Figure P7.1, what is the grain refining mechanism
in the test weld and why?
7.2 Equiaxed grains can often be found in the crater of a weld that exhibits
an essentially purely columnar grain structure, as shown in Figure P7.2.
Explain why.
PROBLEMS 197
test weld preweld
Figure P7.1
7.3 Gutierrez and Lippold (23) made a preweld in aluminum alloy 2195 and
then a test weld perpendicular to it, as shown in Figure P7.3. (a) Do you
expect to see a nondendritic equiaxed zone near the fusion boundary of
the test weld in the overlap region and why or why not? (b) Same as (a)
but with the workpiece and the preweld solution heat treated before
making the test weld.
7.4 Part of a pure Ni ingot with large columnar grains is welded perpendicular
to the grains, as shown in Figure P7.4. Sketch the grain structure in
the weld.
7.5 A pulsed arc weld is shown in Figure P7.5. Sketch the grain structure in
the area produced by the last pulse.
198 WELD METAL SOLIDIFICATION I: GRAIN STRUCTURE
weld crater
Figure P7.2
preweld
test
weld
Figure P7.3
(a) Ni ingot (b) Ni ingot
Figure P7.4
welding
direction pulsed
arc weld
Figure P7.5
8 Weld Metal Solidification II:
Microstructure within Grains
In this chapter we shall discuss the microstructure within the grains in the
fusion zone, focusing on the solidification mode, dendrite spacing and cell
spacing, how they vary across the weld metal, and how they are affected by
welding parameters.The advantages of fine microstructure and techniques for
microstructural refining will also be discussed.
8.1 SOLIDIFICATION MODES
As constitutional supercooling increases, the solidification mode changes from
planar to cellular and from cellular to dendritic (Chapter 6). Figure 8.1 shows
schematically the effect of constitutional supercooling on the microstructure
within the grains in the weld metal. The solidification mode changes from
planar to cellular, columnar dendritic, and equiaxed dendritic as the degree of
constitutional supercooling at the pool boundary increases. Heterogeneous
nucleation aided by constitutional supercooling promotes the formation of
equiaxed grains in the weld metal (Chapter 7).
8.1.1 Temperature Gradient and Growth Rate
While the solidification mode can vary from one weld to another (Figure 8.1),
it can also vary within a single weld from the fusion line to the centerline.This
will be explained following the discussion on the growth rate R and the temperature
gradient G.
Figure 8.2 shows the relationship between the growth rate R and the
welding speed V. The distance a given point on the pool boundary travels in
the normal direction n during a very small time interval dt is
(8.1)
Dividing the above equation by dt cos(a - b) yields
R (8.2)
V
=
( - )
cos
cos
a
a b
Rndt =(Vdt)cosa=(Rdt)cos(a- b)
199
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
200 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
T T
T T
S L
Planar
(a)
x x
S L
S+L
actual
equilibrium
weld
pool
welding direction
<100>
<100>
fusion
line
base
metal
Cellular
(b)
S S + L L
x x
Equiaxed
dendritic
(d)
Columnar
dendritic
(c)
S + L
S L
constitutional
supercooling
magnified
welding
direction
weld
pool
weld
metal
Figure 8.1 Effect of constitutional supercooling on solidification mode during
welding: (a) planar; (b) cellular; (c) columnar dendritic; (d) equiaxed dendritic.
Constitutional supercooling increases from (a) through (d).
pool boundary
at time, t
pool n
R
V
centerline
fusion line
dendrite
n
Rdt
Vdt
Rndt
t+dt
t
R
V
pool boundary
at time, t+dt
welding
speed, V
dendrite
α
β
Figure 8.2 Relationship between growth rate R and welding speed V.
where a is the angle between the welding direction and the normal to the pool
boundary and b is the angle between the welding direction and the growth
direction of a dendrite at the point (<100> in fcc and bcc materials).This relationship
has been shown by Nakagawa et al. (1). If the difference between the
two angles is neglected (cos 0° = 1) as an approximation, Equation (8.2)
becomes
(8.3)
As shown in Figure 8.3, a = 0° and 90° at the weld centerline and the fusion
line, respectively. Therefore, the solidification rate at the centerline RCL = V
(maximum) while that at the fusion line RFL = 0 (minimum). As shown in
Figure 8.4, the distance between the maximum pool temperature (Tmax) and
the pool boundary (TL) is greater at the centerline than at the fusion line
R=Vcosa
SOLIDIFICATION MODES 201
centerline
fusion line
welding
speed, V
pool boundary
at time t
pool boundary
at time t + dt
pool
Vdt
Rdt
Vdt
Rdt
Vdt
at fusion line anywhere at centerline
Rdt 0
α =900
cos900 = 0
cos 0 FL R V α =
α
R Vcosα
Rdt = (Vdt) cosα
α = 00
cos00 =1
cos CL R V α = V
Figure 8.3 Variation in growth rate along pool boundary.
welding
speed, V
weld pool
centerline (CL)
fusion line (FL)
GFL
RCL=V
GCL
R FL 0
G
Tmax
TL ≈
Figure 8.4 Variations in temperature gradient G and growth rate R along pool
boundary.
because the weld pool is elongated. Consequently, the temperature gradient
normal to the pool boundary at the centerline, GCL, is less than that at the
fusion line, GFL. Since GCL < GFL and RCL >> RFL,
(8.4)
8.1.2 Variations in Growth Mode across Weld
According to Equation (8.4), the ratio G/R decreases from the fusion line
toward the centerline. This suggests that the solidification mode may change
from planar to cellular, columnar dendritic, and equiaxed dendritic across the
fusion zone, as depicted in Figure 8.5. Three grains are shown to grow epitaxially
from the fusion line. Consider the one on the right. It grows with the
planar mode along the easy-growth direction <100> of the base-metal grain.
A short distance away from the fusion line, solidification changes to the
cellular mode. Further away from the fusion line, solidification changes to
the columnar dendritic mode. Some of the cells evolve into dendrites and
their side arms block off the neighboring cells. Near the weld centerline
equiaxed dendrites nucleate and grow, blocking off the columnar dendrites.
The solidification-mode transitions have been observed in several different
materials (2–4).
Figure 8.6 shows the planar-to-cellular transition near the weld fusion line
of an autogenous gas–tungsten weld of Fe–49Ni (2). Figure 8.7 shows the
planar-to-cellular transition and the cellular-to-dendritic transition in 1100 aluminum
(essentially pure Al) welded with a 4047 (Al–12Si) filler metal. Figure
8.8 shows the transition from columnar to equiaxed dendrites in an electron
beam weld of a Fe–15Cr–15Ni single crystal containing some sulfur (4).These
columnar dendrites, with hardly visible side arms, follow the easy-growth
G
R
G
R
ÊË
ˆ¯
<<Êˈ¯CL FL202 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINScolumnardendriticplanarcellularequiaxeddendriticweldpoolweldingdirectionbasemetal<100>
<100> <100>
pool
boundary
fusion
line
centerline
Figure 8.5 Variation in solidification mode across the fusion zone.
direction <100> of the single crystal, which happens to be normal to the fusion
line in this case.
Before leaving the subject of the solidification mode, it is desirable to
further consider the weld metal microstructure of a workpiece with only one
and very large grain, that is, a single-crystal workpiece. Figure 8.9 shows the
columnar dendritic structure in a Fe–15Cr–15Ni single crystal of high purity
electron beam welded along a [110] direction on a (001) surface (5).The weld
is still a single crystal because of epitaxial growth. However, it can have
SOLIDIFICATION MODES 203
region of
planar growth
region of
cellular
growth
fusion
line
solidified
weld metal
100 weld pool
boundary
liquid weld
metal
Figure 8.6 Planar-to-cellular transition in an autogenous weld of Fe–49Ni. Modified
from Savage et al. (2).
Figure 8.7 Planar-to-cellular and cellular-to-dendritic transitions in 1100 Al welded
with 4047 filler.
204 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
Figure 8.8 Electron beam weld of single crystal of Fe–15Cr–15Ni with sulfur showing
transition from columnar to equiaxed dendrites. Reprinted from David and Vitek (4).
colonies of columnar dendrites of different orientations, as shown by the top
weld pass in Figure 8.9b (f is one of the angles characterizing the normal to
the pool boundary). This is because of competitive growth between columnar
dendrites along the three <100> easy-growth directions [100], [010], and [001].
Dendrites with an easy-growth direction closest to the heat flow direction
(normal to the pool boundary) compete better.
8.2 DENDRITE AND CELL SPACING
The spacing between dendrite arms or cells, just as the solidification mode, can
also vary across the fusion zone. As already mentioned in the previous section,
GCL < GFL and RCL >> RFL. Consequently,
(8.5)
where G ¥ R is the cooling rate, as explained previously in Chapter 6. According
to Equation (8.5), the cooling rate (G ¥ R) is higher at the weld centerline
and lower at the fusion line. This suggests that the dendrite arm spacing
decreases from the fusion line to the centerline because the dendrite arm
spacing decreases with increasing cooling rate (Chapter 6).
The variation in the dendrite arm spacing across the fusion zone can be
further explained with the help of thermal cycles (Chapter 2). Figure 8.10
(G¥R) >(G¥R) CL FL
shows a eutectic-type phase diagram and the thermal cycles at the weld centerline
and fusion line of alloy C0. As shown, the cooling time through the
solidification temperature range is shorter at the weld centerline (2¯4/V) and
longer at the fusion line (1¯3/V). As such, the cooling rate through the solidification
temperature range increases and the dendrite arm spacing decreases
from the fusion line to the centerline. As shown by the aluminum weld in
DENDRITE AND CELL SPACING 205
Figure 8.9 Electron beam weld of single crystal of pure Fe–15Cr–15Ni made in a [110]
direction on a (001) surface: (a) top cross section; (b) transverse cross section.
Reprinted from Rappaz et al. (5).
Figure 8.11, the solidification microstructure gets finer from the fusion line to
the centerline (6). The same trend was observed in other aluminum welds by
Kou et al. (6, 7) and Lanzafame and Kattamis (8).
8.3 EFFECT OF WELDING PARAMETERS
8.3.1 Solidification Mode
The heat input and the welding speed can affect the solidification mode of the
weld metal significantly. The solidification mode changes from planar to cel-
206 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
distance
time
T
1 2
3 4
a
b
3
1
2
4
Concentration, C
TL
TE
(a)
(b)
(c)
welding direction
shorter solidification
time, b, at centerline
fusion line
(coarser
microstructure)
A
Temperature, T
L
S
S + L
TL
Co
TE
centerline (finer
microstructure)
longer solidification
time, a, at fusion line
weld
pool
Figure 8.10 Variation in dendrite arm spacing across fusion zone: (a) phase diagram;
(b) thermal cycles; (c) top view of weld pool.
Figure 8.11 Transverse cross-section of gas–tungsten arc weld in 6061 aluminum:
(a) finer microstructure near centerline; (b) coarser microstructure near fusion line.
Magnification 115¥. Reprinted from Kou et al. (6). Courtesy of American Welding
Society.
lular and dendritic as the ratio G/R decreases (Chapter 6). Table 8.1 summarizes
the observations of Savage et al. (9) in HY-80 steel. At the welding speed
of 0.85mm/s (2ipm), the weld microstructure changes from cellular to dendritic
when the welding current increases from 150 to 450A. According to
Equation (2.15), the higher the heat input (Q) under the same welding speed
(V), the lower the temperature gradient G and hence the lower the ratio G/R.
Therefore, at higher heat inputs G/R is lower and dendritic solidification prevails,
while at lower heat inputs G/R is higher and cellular solidification prevails.
Although analytical equations such as Equations (2.15) and (2.17) are
oversimplified, they can still qualitatively tell the effect of welding parameters.
8.3.2 Dendrite and Cell Spacing
The heat input and the welding speed can also affect the spacing between dendrite
arms and cells. The dendrite arm spacing or cell spacing decreases with
increasing cooling rate (Chapter 6). As compared to arc welding, the cooling
rate in laser or electron beam welding is higher and the weld metal microstructure
is finer. The 6061 aluminum welds in Figure 8.12 confirm that this is the
case (10).
As shown in Table 8.1, at the welding current of 150A, the cells become
finer as the welding speed increases. From Equation (2.17), under the same
heat input (Q), the cooling rate increases with increasing welding speed V.
Therefore, at higher welding speeds the cooling rate is higher and the cells are
finer, while at lower welding speeds the cooling rate is lower and the cells are
coarser. Elmer et al. (11) also observed this trend in EBW of two austenitic
stainless steels of similar compositions, as shown in Figure 8.13.The difference
in the cell spacing can be seen even though the magnifications of the micrographs
are different.
According to Equation (2.17), the cooling rate increases with decreasing
heat input–welding speed ratio Q/V. This ratio also represents the amount
EFFECT OF WELDING PARAMETERS 207
TABLE 8.1 Effect of Welding Parameters on Weld Metal Microstructure
Travel speed 150A 300A 450A
0.85mm/s Cellular Cellular dendritic Coarse cellular
(2ipm) dendritic
1.69mm/s Cellular Fine cellular Coarse cellular
(4ipm) dendritic dendritic
3.39mm/s Fine cellular Cellular, slight Severe undercutting
(8ipm) undercutting
6.77mm/s Very fine cellular Cellular, Severe undercutting
(16ipm) undercutting
Source: From Savage et al. (9).
208 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
Figure 8.12 Autogenous welds of 6061 aluminum: (a) coarser solidification structure
in gas–tungsten arc weld: (b) finer solidification structure in electron beam weld.
Reprinted from Metals Handbook (10).
Figure 8.13 Effect of welding speed on cell spacing in EBW of austenitic stainless
steels: (a) 100mm/s; (b) 25mm/s. From Elmer et al. (11).
of heat per unit length of the weld (J/cm or cal/cm). Therefore, the dendrite
arm spacing or cell spacing can be expected to increase with increasing
Q/V or amount of heat per unit length of the weld. This has been observed
in several aluminum alloys (7, 8, 12–14), including those shown in Figure
8.14.
8.4 REFINING MICROSTRUCTURE WITHIN GRAINS
It has been shown in aluminum alloys that the finer the dendrite arm spacing,
the higher the ductility (15) and yield strength (8, 15) of the weld metal and
the more effective the postweld heat treatment (8, 12, 15), due to the finer
distribution of interdendritic eutectics.
REFINING MICROSTRUCTURE WITHIN GRAINS 209
10 20 30 40
10
20
(Heat input/welding speed, W/mm)1/2
Dendrite arm spacing, m
30 60 90
(Heat input/welding speed, J/mm)1/2
(b)
(a)
just above toe of weld
level with surface of plate
14
12
10
8
6
4
2
0 10 20 30 40 50 60 70 80
Dendrite spacing, m
Heat input cal/cm length
0
0
μ μ
Figure 8.14 Effect of heat input per unit length of weld on dendrite arm spacing.
(a) For Al–Mg–Mn alloy. From Jordan and Coleman (13). (b) For 2014 Al–Cu alloy.
Modified from Lanzafame and Kattamis (14). Courtesy of American Welding Society.
8.4.1 Arc Oscillation
Kou and Le (16) studied the microstructure in oscillated arc welds of 2014 aluminum
alloy. It was observed that the dendrite arm spacing was reduced significantly
by transverse arc oscillation at low frequencies, as shown in Figure
8.15. This reduction in the dendrite arm spacing has contributed to the significant
improvement in both the strength and ductility of the weld, as shown in
Figure 8.16.
As illustrated in Figure 8.17, when arc oscillation is applied, the weld pool
gains a lateral velocity, v, in addition to its original velocity u in the welding
direction (16). The magnitude of v can be comparable with that of u depending
on the amplitude and frequency of arc oscillation.
As shown, the resultant velocity of the weld pool, w, is greater than that of
the unoscillated weld pool, u. Furthermore, the temperature gradient ahead
of the solid–liquid interface, G, could also be increased due to the smaller
distance between the heat source and the pool boundary. Consequently, the
product GR or the cooling rate is increased significantly by the action of arc
oscillation. This explains why the microstructure is finer in the oscillated arc
weld.
Example: Suppose the welding speed of a regular weld is 4.2mm/s (10ipm).
Calculate the increase in the velocity of the weld pool if the arc is oscillated
210 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
NO
OSCILLATION
TRANSVERSE
OSCILLATION
(a) (b)
Figure 8.15 Microstructures near fusion line of gas-tungsten arc welds of 2014 aluminum:
(a) coarser dendrites in weld made without arc oscillation; (b) finer dendrites
in weld made with transverse arc oscillation. Magnification 200¥. Reprinted from Kou
and Le (16). Courtesy of American Welding Society.
transversely at a frequency of 1 Hz and an amplitude (the maximum deflection
of the arc from the weld centerline) of 1.9mm. Do you expect the dendrite
arm spacing to keep on decreasing if the frequency keeps on increasing
to, say, 100Hz?
REFINING MICROSTRUCTURE WITHIN GRAINS 211
Figure 8.16 Tensile testing of two gas–tungsten arc welds of 2014 aluminum made
without arc oscillation and with transverse arc oscillation. From Kou and Le (16).
No
oscillation
u
w v
u
(a)
(b)
Transverse
oscillation
Figure 8.17 Increase in weld pool travel speed due to transverse arc oscillation. Modified
from Kou and Le (16). Courtesy of American Welding Society.
Since the arc travels a distance equal to four oscillation amplitudes
per second, v = 4 ¥ 1.9mm/s = 7.6mm/s. Therefore, the resultant velocity w =
(u2 + v2)1/2 = (4.22 + 7.62)1/2 = 8.7mm/s. The increase in the weld pool velocity
is 8.7 - 4.2 = 4.5mm/s. The increase in the weld pool velocity and hence the
decrease in the dendrite arm spacing cannot keep on going with increasing
oscillation frequency. This is because when the arc oscillates too fast, say at
100 Hz, the weld pool cannot catch up with it because there is not enough time
for melting and solidification to occur.
It is interesting to point out that Kou and Le (16) also observed that the
microstructure in oscillated arc welds is much more uniform than that in welds
without oscillation. In oscillated arc welds the weld centerline is no longer a
location where the cooling rate (or GR) is clearly at its maximum.
Tseng and Savage (17) studied the microstructure in gas–tungsten arc
welds of HY-80 steel made with transverse and longitudinal arc oscillation.
Refining of the grain structure was not obtained, but refining of the dendritic
structure within grains (subgrain structure) was observed and solidification
cracking was reduced, as shown in Figure 8.18. It is possible that solidification
cracking was reduced because the crack-causing constituents were diluted
to a greater extent by the larger interdendritic area in the welds with a finer
dendritic structure.
212 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
Half-amplitude =
1.65mm (0.065in)
0 0.23 0.42 1.20
0
10
20
30
Frequency of oscillation-Hz
Subgrain diameter
(microns)
Oscillation frequency
Amplitude =1.65mm
(0.065in)
No
oscillation
0.42
Hz
1.19
Hz
Mean maximum
crack length (mm)
0
0.25
0.50
0.75
Mean maximum
crack length (mils)
0
10
20
30
(a)
(b)
Figure 8.18 Effect of magnetic arc oscillation on gas–tungsten arc welds of HY-80
steel: (a) refining of microstructure within grains; (b) solidification cracking. Modified
from Tseng and Savage (17).
8.4.2 Arc Pulsation
Becker and Adams (18) studied the microstructure in pulsed arc welds of
titanium alloys. It was observed that the cell spacing varied periodically
along the weld, larger where solidification took place during the high-current
portion of the cycle. Obviously, the cooling rate was significantly lower during
the high-current portion of the cycle. This is because, from Equation
(2.17), the cooling rate can be expected to decrease with increasing heat
input, though, strictly speaking, this equation is for steady-state conditions
only.
REFERENCES
1. Nakagawa, H., et al., J. Jpn.Weld. Soc., 39: 94, 1970.
2. Savage,W. F., Nippes, E. F., and Erickson, J. S., Weld. J., 55: 213s, 1976.
3. Kou, S., and Le,Y., unpublished research, University of Wisconsin, Madison, 1983.
4. David, S. A., and Vitek, J. M., Int. Mater. Rev., 34: 213, 1989.
5. Rappaz, M., David, S. A., Vitek, J. M., and Boatner, L. A., in Recent Trends in
Welding Science and Technology, Eds. S. A. David and J. M. Vitek, ASM International,
Materials Park, OH, May 1989, p. 147.
6. Kou, S., Kanevsky, T., and Fyfitch, S., Weld. J., 61: 175s, 1982.
7. Kou, S., and Le,Y., Metall. Trans. A, 14A: 2245, 1983.
8. Lanzafame, J. N., and Kattamis, T. Z., Weld. J., 52: 226s, 1973.
9. Savage,W. F., Lundin, C. D., and Hrubec, R. J., Weld. J., 47: 420s, 1968.
10. Metals Handbook, Vol. 7, 8th ed., American Society for Metals, Metals Park, OH,
1972, pp. 266, 269.
11. Elmer, J.W., Allen, S. M., and Eagar, T.W., Metall. Trans., 20A: 2117, 1989.
12. Brown, P. E., and Adams, C. M. Jr., Weld. J., 39: 520s, 1960.
13. Jordan, M. F., and Coleman, M. C., Br.Weld. J., 15: 552, 1968.
14. Lanzafame, J. N., and Kattamis, T. Z., Weld. J., 52: 226s, 1973.
15. Fukui, T., and Namba, K., Trans. Jpn.Weld. Soc., 4: 49, 1973.
16. Kou, S., and Le,Y., Weld. J., 64: 51, 1985.
17. Tseng, C., and Savage,W. F., Weld. J., 50: 777, 1971.
18. Becker, D.W., and Adams, C. M. Jr., Weld. J., 58: 143s, 1979.
19. Savage, W. F., in Weldments: Physical Metallurgy and Failure Phenomena, Eds.
R. J. Christoffel, E. F. Nippes, and H. D. Solomon, General Electric Co., Schenectady,
NY, 1979, p. 1.
FURTHER READING
1. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20: 83, 1975.
2. Savage,W. F., Weld.World, 18: 89, 1980.
FURTHER READING 213
3. David, S. A., and Vitek, J. M., Int. Mater. Rev., 34: 213, 1989.
4. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
PROBLEMS
8.1 It has been suggested that the secondary dendrite arm spacing d along
the weld centerline can be related quantitatively to the heat input per
unit length of weld, Q/V. Based on the data of the dendrite arm spacing
d as a function of cooling rate e, similar to those shown in Figure 6.17a,
it can be shown that d = ae-1/b, where a and b are constant with b being
in the range of 2–3. (a) Express the dendrite arm spacing in terms of Q/V
for bead-on-plate welds in thick-section aluminum alloys. (b) How do
the preheat temperature and thermal conductivity affect the dendrite
arm spacing? (c) Do you expect the relationship obtained to be very
accurate?
8.2 The size of the mushy zone is often an interesting piece of information
for studying weld metal solidification. Let d = ae-1/b, where d is the dendrite
arm spacing and e the cooling rate. Consider how measurements of
the dendrite arm spacing across the weld metal can help determine
the size of the mushy zone. Express the width of the mushy zone in the
welding direction Dx, as shown in Figure P8.2, in terms of the dendrite
arm spacing d, the welding speed V, and the freezing temperature range
DT (= TL - TE).
8.3 It has been observed that the greater the heat input per unit length of
weld (Q/V), the longer it takes to homogenize the microsegregation in
the weld metal of aluminum alloys for improving its mechanical properties.
Let d = ae-1/b, where d is the dendrite arm spacing and e the cooling
rate. Express the time required for homogenization (t) in terms of Q/V.
8.4 An Al–1% Cu alloy is welded autogenously by GTAW, and an Al–5%
Cu alloy is welded under identical condition.Which alloy is expected to
develop more constitutional supercooling and why? Which alloy is likely
to have more equiaxed dendrites in the weld metal and why?
8.5 An Al–5% Cu alloy is welded autogenously by GTAW and by EBW
under the same welding speed but different heat inputs (much less in the
214 WELD METAL SOLIDIFICATION II: MICROSTRUCTURE WITHIN GRAINS
V
mushy zone
TE
weld pool
TL Δx
Figure P8.2
case of EBW).Which weld is expected to experience more constitutional
supercooling and why? Which weld is likely to have more equiaxed dendrites
and why?
8.6 In autogenous GTAW of aluminum alloys, how do you expect the amount
of equiaxed grains in the weld metal to be affected by preheating and
why?
8.7 In autogenous GTAW of aluminum alloys, how do you expect the dendrite
arm spacing of the weld metal to be affected by preheating and
why?
8.8 Figure P8.8 is a micrograph near the fusion line of an autogenous
gas–tungsten arc weld in a Fe–49% Ni alloy sheet (19). Explain the solidification
microstructure, which is to the right of the fusion line (dark vertical
line).
PROBLEMS 215
Figure P8.8
9 Post-Solidification Phase
Transformations
Post-solidification phase transformations, when they occur, can change the
solidification microstructure and properties of the weld metal. It is, therefore,
essential that post-solidification phase transformations be understood
in order to understand the weld metal microstructure and properties. In
this chapter two major types of post-solidification phase transformations in
the weld metal will be discussed. The first involves the ferrite-to-austenite
transformation in welds of austenitic stainless steels, and the second involves
the austenite-to-ferrite transformation in welds of low-carbon, low-alloy
steels.
9.1 FERRITE-TO-AUSTENITE TRANSFORMATION IN
AUSTENITIC STAINLESS STEEL WELDS
9.1.1 Primary Solidification Modes
The welds of austenitic stainless steels normally have an austenite (fcc) matrix
with varying amounts of d-ferrite (bcc) (1–7). A proper amount of d-ferrite in
austenitic stainless steel welds is essential—too much d-ferrite (10 vol %)
tends to reduce the ductility, toughness, and corrosion resistance, while too
little d-ferrite (5 vol %) can result in solidification cracking.
A. Phase Diagram Figure 9.1 shows the ternary phase diagram of the
Fe–Cr–Ni system (8). The heavy curved line in Figure 9.1a represents the
trough on the liquidus surface, which is called the line of twofold saturation.
The line declines from the binary Fe–Ni peritectic reaction temperature
to the ternary eutectic point at 49Cr–43Ni–8Fe. Alloys with a composition on
the Cr-rich (upper) side of this line have d-ferrite as the primary solidification
phase, that is, the first solid phase to form from the liquid. On the other hand,
alloys with a composition on the Ni-rich (lower) side have austenite as the
primary solidification phase. The heavy curved 1ines on the solidus surface in
Figure 9.1b more or less follow the trend of the liquidus trough and converge
at the ternary eutectic temperature.
The development of weld metal microstructure in austenitic stainless steels
is explained in Figure 9.2.The weld metal ferrite can have three different types
216
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
Figure 9.1 The Fe–Cr–Ni ternary system: (
a) liquidus surface; (
b) solidus surface. Reprinted from Metals Handbook (8).
217
218 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
of morphology: interdendritic (Figure 9.2a), vermicular (Figure 9.2b), and
lathy (Figure 9.2c). Figure 9.2d shows a schematic vertical (isoplethal) section
of the ternary phase diagram in Figure 9.1, for instance, at 70 wt % Fe and
above 1200°C. This has also been called a pseudo-binary phase diagram. The
apex (point 1) of the three-phase eutectic triangle (L + g + d) corresponds to
the intersection between the vertical section and the heavy curved line in
Figure 9.1a.The two lower corners (points 2 and 3) of the triangle, on the other
hand, correspond to the intersections between the vertical section and the two
heavy curved lines in Figure 9.1b.
B. Primary Austenite For an alloy on the Ni-rich (left-hand) side of the apex
of the three-phase eutectic triangle, austenite (g) is the primary solidification
phase. The light dendrites shown in Figure 9.2a are austenite, while the dark
particles between the primary dendrite arms are the d-ferrite that forms when
interdendritic
ferrite
(a)
vermicular
ferrite
(b)
lathy
ferrite
(c)
liquid liquid liquid
L
Primary austenite
solidification
increasing Ni
increasing Cr
(d)
austenite, ferrite,
L L
Temperature
liquid, L
1
2 3
4
or
Primary ferrite
solidification
γ
γ γ
δ δ
γ
+ γ
+ γ + δ
γ
+ δ
δ + γ δ
Figure 9.2 Schematics showing solidification and postsolidification transformation in
Fe–Cr–Ni welds: (a) interdendritic ferrite; (b) vermicular ferrite; (c) lathy ferrite; (d)
vertical section of ternary-phase diagram at approximately 70% Fe.
the three-phase triangle is reached during the terminal stage of solidification.
These are called the interdendritic ferrite. For dendrites with long secondary
arms, interdendritic ferrite particles can also form between secondary dendrite
arms.
C. Primary Ferrite For an alloy on the Cr-rich (right-hand) side of the apex
of the three-phase eutectic triangle, d-ferrite is the primary solidification phase.
The dark dendrites shown in Figure 9.2b are d-ferrite.The core of the d-ferrite
dendrites, which forms at the beginning of solidification, is richer in Cr (point
4), while the outer portions, which form as temperature decreases, have lower
chromium contents. Upon cooling into the (d + g) two-phase region, the outer
portions of the dendrites having less Cr transform to austenite, thus leaving
behind Cr-rich “skeletons” of d-ferrite at the dendrite cores. This skeletal
ferrite is called vermicular ferrite. In addition to vermicular ferrite, primary
d-ferrite dendrites can also transform to lathy or lacy ferrite upon cooling
into the (d + g) two-phase region, as shown in Figure 9.2c.
D. Weld Microstructure Figure 9.3a shows the solidification structure at the
centerline of an autogenous gas–tungsten arc weld of a 310 stainless steel
sheet, which contains approximately 25% Cr, 20% Ni, and 55% Fe by weight
(9).The composition is on the Ni-rich (left) side of the apex of the three-phase
eutectic triangle, as shown in Figure 9.4a, and solidification occurs as primary
austenite. The microstructure consists of austenite dendrites (light etching;
mixed-acids etchant) and interdendritic d-ferrite (dark etching; mixed-acids
etchant) between the primary and secondary dendrite arms, similar to those
shown in Figure 9.2a.
Figure 9.3b, on the other hand, shows the solidification structure at the
centerline of an autogenous gas–tungsten arc weld of a 309 stainless steel
sheet, which contains approximately 23 wt% Cr, 14 wt% Ni, and 63 wt % Fe.
The composition lies just to the Cr-rich side of the apex of the three-phase
eutectic triangle, as shown in Figure 9.4b, and solidifies as primary d-ferrite.
The microstructure consists of vermicular ferrite (dark etching; mixedacids
etchant) in an austenite matrix (light etching; mixed-acids etchant)
similar to those shown in Figure 9.2b. In both welds columnar dendrites grow
essentially perpendicular to the teardrop-shaped pool boundary as revealed
by the columnar dendrites.
Kou and Le (9) quenched welds during welding in order to preserve
the as-solidified microstructure, that is, the microstructure before postsolidification
phase transformations. For stainless steels liquid-tin quenching
is more effective than water quenching because steam and bubbles reduce heat
transfer.With the help of quenching, the evolution of microstructure during
welding can be better studied. Figure 9.5 shows the d-ferrite dendrites (light
etching; mixed-chloride etchant) near the weld pool of an autogenous gas–
tungsten arc weld of 309 stainless steel, quenched in during welding with liquid
tin before the d Æ g transformation changed it to vermicular ferrite like that
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 219
220 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
Figure 9.3 Solidification structure at the weld centerline: (a) 310 stainless steel; (b)
309 stainless steel. Magnification 190¥. Reprinted from Kou and Le (9).
L L L
55%Fe 63%Fe 73%Fe
1400
1000
600
15
30
25
20
35
10
10
27
20
17
30
7
5
22
15
12
25
2
wt%Cr
wt%Ni
Temperature, oC
(a) (b) (c)
L + γ + δ
γ
γ γ
δ
δ
δ
δ + γ
δ
+γδ
+
γ
Figure 9.4 The Fe–Cr–Ni pseudo-binary phase diagrams: (a) at 55 wt % Fe; (b) at 63
wt % Fe; (c) at 73 wt % Fe. Reprinted from Kou and Le (9).
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 221
shown in Figure 9.3b. Liquid-tin quenching was subsequently used by other
investigators to study stainless steel welds (10, 11).
9.1.2 Mechanisms of Ferrite Formation
Inoue et al. (11) studied vermicular and lathy ferrite in autogenous GTAW of
austenitic stainless steels of 70% Fe with three different Cr–Ni ratios. It was
found that, as the Cr–Ni ratio increases, the ratio of lathy ferrite to total ferrite
does not change significantly even though both increase. A schematic of the
proposed formation mechanism of vermicular and lathy ferrite is shown in
Figure 9.6.Austenite first grows epitaxially from the unmelted austenite grains
at the fusion boundary, and d-ferrite soon nucleates at the solidification front.
The crystallographic orientation relationship between the d-ferrite and the
austenite determines the ferrite morphology after the postsolidification transformation.
If the closed-packed planes of the d-ferrite are parallel to those of
the austenite, the d Æ g transformation occurs with a planar d/g interface,
resulting in vermicular ferrite. However, if the so-called Kurdjumov–Sachs orientation
relationships, namely, (1¯10)d //(1¯11)g and [1¯1¯1]d //[1¯1¯0]g, exist between
the d-ferrite and the austenite, the transformation occurs along the austenite
habit plane into the d-ferrite dendrites. The resultant ferrite morphology is
lathy, as shown in Figure 9.7. For the lathy ferrite to continue to grow, the
Figure 9.5 Liquid-tin quenched solidification structure near the pool of an autogenous
gas–tungsten arc weld of 309 stainless steel. Magnification 70. Mixed-chloride
etchant. Reprinted from Kou and Le (9).
222 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
Figure 9.6 Mechanism for the formation of vermicular and lathy ferrite. Reprinted
from Inoue et al. (11).
Figure 9.7 Lathy ferrite in an autogenous gas–tungsten arc weld of Fe–18.8Cr–11.2Ni.
Reprinted from Inoue et al. (11).
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 223
preferred growth direction <100> of both d-ferrite and austenite must be
aligned with the heat flow direction.
9.1.3 Prediction of Ferrite Content
Schaeffler (12) first proposed the quantitative relationship between the composition
and ferrite content of the weld metal. As shown by the constitution
diagram in Figure 9.8, the chromium equivalent of a given alloy is determined
from the concentrations of ferrite formers Cr, Mo, Si, and Cb, and the austenite
equivalent is determined from the concentrations of austenite formers Ni,
C, and Mn. DeLong (13) refined Schaeffler’s diagram to include nitrogen, a
strong austenite former, as shown in Figure 9.9. Also, the ferrite content is
expressed in terms of the ferrite number, which is more reproducible than the
ferrite percentage and can be determined nondestructively by magnetic
means. Figure 9.10 shows that nitrogen, introduced into the weld metal by
adding various amounts of N2 to the Ar shielding gas, can reduce the weld
ferrite content significantly (14). Cieslak et al. (6), Okagawa et al. (7), and
Lundin et al. (15) have reported similar results previously.
The WRC-1992 diagram of Kotecki and Siewert (16), shown in Figure 9.11,
was from the Welding Research Council in 1992. It was modified from the
WRC-1988 diagram of McCowan et al. (17) by adding to the nickel equivalent
the coefficient for copper (18) and showing how the axes could be
extended to make Schaeffler-like calculations for dissimilar metal joining.
20
30
10
0
10 20 30 40
Chromium equivalent =
%Cr + %Mo + 1.5 X %Si + 0.5 X %Cb
Nickel equivalent =
%Ni + 30 X %C + 0.5 X %Mn
Austenite
0% Ferrite
5%
20%
40%
100%
Ferrite
Martensite
M + F
A + M
0
28
26
24
22
12
14
16
18
8
64
2
2 4 6 8 12141618 2224 26 28 32 34 36 38
F + M
A + M + F
80%
10%
A + F
Figure 9.8 Schaeffler diagram for predicting weld ferrite content and solidification
mode. From Schaeffler (12).
224 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
Kotecki (19, 20) added the martensite boundaries to the WRC-1992 diagram,
as shown in Figure 9.12. More recent investigations of Kotecki (21, 22) have
revealed that the boundaries hold up well with Mo and N variation but
not as well with C variation. Balmforth and Lippold (23) proposed the
ferritic–martensitic constitution diagram shown in Figure 9.13.Vitek et al. (24,
0
4
2
WRC ferrite
number
Austenite
Austenite
plus ferrite
Schaeffler
A + M line
16 17 18 19 21 22 23 24 25 26 27
10
11
12
13
14
15
16
17
18
19
20
21
Chromium equivalent =
%Cr + %Mo + 1.5 X %Si + 0.5 X %Cb
Nickel equivalent =
%Ni + 30 X %C + 30 X %N + 0.5 X %Mn
12
14
16
18
10
6
8
0
2
4
6
7.6
9.2
10.7
12.3
13.820
Prior magnetic ferrite %
Figure 9.9 DeLong diagram for predicting weld ferrite content and solidification
mode. Reprinted from DeLong (13). Courtesy of American Welding Society.
0.1 0.2 0.3 0.4
0
20
40
60
80
Nitrogen content (mass%)
Ferrite content (%)
Figure 9.10 Effect of nitrogen on ferrite content in gas–tungsten arc welds of duplex
stainless steel. Reprinted from Sato et al. (14).
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 225
Figure 9.11 WRC-1992 diagram for predicting weld ferrite content and solidification
mode. Reprinted from Kotecki and Siewert (16). Courtesy of American Welding
Society.
Figure 9.12 WRC-1992 diagram with martensite boundaries for 1, 4, and 10% Mn.
Reprinted from Kotecki (20). Courtesy of American Welding Society.
226 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
25) developed a model FNN-1999 using artificial neural networks to improve
ferrite number prediction, as shown in Figure 9.14. Twelve alloying elements
besides Fe were considered: C, Cr, Ni, Mo,N, Mn, Si, Cu,Ti, Cb,V, and Co.The
model is not in a simple pictorial form, such as the WRC-1992 diagram,
because it allows nonlinear effects and element interactions.
9.1.4 Effect of Cooling Rate
A. Changes in Solidification Mode The prediction of the weld metal ferrite
content based on the aforementioned constitution diagrams can be inaccurate
10 30
20 40
50
60 70 80 90
A+M+
F
Cr + 2Mo + 10(Al + Ti)
Ni + 35C + 20N
8 10 12 14 16 18 20 22
0
1
2
3
4
5
6
M + F
Martensite
Ferrite
Figure 9.13 Ferritic–martensitic stainless steel constitution diagram containing a
boundary for austenite formation and with iso-ferrite lines in volume percent of ferrite.
Reprinted from Balmforth and Lippold (23).
Figure 9.14 Experimentally measured ferrite number (FN) versus predicted FN: (a)
FNN-1999; (b) WRC-1992. Reprinted from Vitek et al. (25). Courtesy of American
Welding Society.
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 227
when the cooling rate is high, especially in laser and electron beam welding
(3, 26–37). Katayama and Matsunawa (28, 29), David et al. (31), and Brooks
and Thompson (37) have compared microstructures that form in slow-coolingrate
arc welds with those that form in high-cooling-rate, high-energy-beam
welds.Their studies show two interesting trends. For low Cr–Ni ratio alloys the
ferrite content decreases with increasing cooling rate, and for high Cr–Ni ratio
alloys the ferrite content increases with increasing cooling rate. Elmer et al.
(33) pointed out that in general low Cr–Ni ratio alloys solidify with austenite
as the primary phase, and their ferrite content decreases with increasing
cooling rate because solute redistribution during solidification is reduced at
high cooling rates. On the other hand, high Cr–Ni ratio alloys solidify with
ferrite as the primary phase, and their ferrite content increases with increasing
cooling rate because the d Æ g transformation has less time to occur at
high cooling rates.
Elmer et al. (33, 34) studied a series of Fe–Ni–Cr alloys with 59% Fe and
the Cr–Ni ratio ranging from 1.15 to 2.18, as shown in Figure 9.15. The apex
of the three-phase triangle is at about Fe–25Cr–16Ni. Figure 9.16 summarizes
the microstructural morphologies of small welds made by scanning an electron
beam over a wide range of travel speeds and hence cooling rates (33). At
low travel speeds such as 0.1–1mm/s, the cooling rates are low and the alloys
with a low Cr–Ni ratio (especially alloys 1 and 2) solidify as primary austenite.
The solidification mode is either single-phase austenite (A), that is, no
ferrite between austenite dendrites or cells (cellular–dendritic A), or primary
austenite with second-phase ferrite (AF), that is, only a small amount of ferrite
between austenite dendrites (interdendritic F). The alloys with a high Cr–Ni
Cr 21 23 25 27 29 31
A + F
L + A L + F
A F
Composition, wt%
Ni 20 18 16 14 12 10
1 2 3 4 5 6 7
1200
1250
1300
1350
1400
1450
1500
1550
Temperature, oC
L 59% Fe
Figure 9.15 Vertical section of Fe–Ni–Cr phase diagram at 59% Fe showing seven
alloys with Cr–Ni ratio ranging from 1.15 to 2.18. Modified from Elmer et al. (33).
228 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
ratio (especially alloys 5–7), on the other hand, solidify as primary ferrite.The
solidification mode is primary ferrite with second-phase austenite (FA), that
is, vermicular ferrite, lacy ferrite, small blocks of austenite in a ferrite matrix
(blocky A), or Widmanstatten austenite platelets originating from ferrite grain
boundaries (Widmanstatten A).
At very high welding speeds such as 2000mm/s, however, the cooling rates
are high and the alloys solidify in only the single-phase austenite mode (A) or
the single-phase ferrite mode (F). An example of the former is the alloy 3
(about Fe–24.75Cr–16.25Ni) shown in Figure 9.17a. At the travel speed of
25mm/s (2 ¥ 103 °C/s cooling rate) the substrate solidifies as primary austenite
in the AF mode, with austenite cells and intercellular ferrite. At the much
higher travel speed of 2000 mm/s (1.5 ¥ 106 °C/s cooling rate) the weld at the
top solidifies as primary austenite in the A mode, with much smaller austenite
cells and no intercellular ferrite (cellular A). An example of the latter is
alloy 6 (about Fe–27.5Cr–13.5Ni) shown in Figure 9.17b. At 25 mm/s the substrate
solidifies as primary ferrite in the FA mode, with blocky austenite in a
ferrite matrix. At 2000 mm/s the weld at the top solidifies as primary ferrite in
the F mode, with ferrite cells alone and no austenite (cellular F).
Figure 9.16 also demonstrates that under high cooling rates an alloy that
solidifies as primary ferrite at low cooling rates can change to primary austenite
solidification. For instance, alloy 4 (about Fe–25.5Cr–15.5Ni) can solidify
as primary ferrite at low cooling rates (vermicular F) but solidifies as primary
10-1
Widmanstatten A
blocky A
lacy F
vermicular F
intercellular A
interdendritic F intercellular F
cellular-dendritic A
cellular F
massive A
cellular A
100 101 102 103 104
Electron-beam travel speed, mm/s
Composition, wt%
1
2 3
4
5
6
7
Cr
30
28
24
22
20
Ni
11
13
15
17
19
21
26
alloys
Figure 9.16 Electron beam travel speed (cooling rate) versus composition map of
microstructural morphologies of the seven alloys in Figure 9.15 (A and F denote
austenite and ferrite, respectively). The solid lines indicate the regions of the four
primary solidification modes, while the dashed lines represent the different morphologies
resulting from postsolidification transformation from ferrite to austenite. Modified
from Elmer et al. (33).
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 229
austenite at higher cooling rates (intercellular F or cellular A). Another interesting
point seen in the same figure is that at high cooling rates alloy 5 can
solidify in the fully ferritic mode and undergoes a massive (diffusionless) transformation
after solidification to austenite (massive A). Under very high
cooling rates there is no time for diffusion to occur.
B. Dendrite Tip Undercooling Vitek et al. (27) attributed the change solidification
mode, from primary ferrite to primary austenite, at high cooling rates
to dendrite tip undercooling. Brooks and Thompson (37) explained this undercooling
effect based on Figure 9.18. Alloy C0 solidifies in the primary ferrite
mode at low cooling rates. Under rapid cooling in laser or electron beam
welding, however, the melt can undercool below the extended austenite
liquidus (CLg), and it becomes thermodynamically possible for the melt to
solidify as primary austenite. The closer C0 is to the apex of the three-phase
triangle, the easier sufficient undercooling can occur to switch the solidification
mode from primary ferrite to primary austenite.
Figure 9.17 Microstructure of the low-cooling-rate substrate (2 ¥ 103 °C/s) and the
high-cooling-rate electron beam weld at the top: (a) alloy 3 in Figure 9.15; (b) alloy 6.
Reprinted from Elmer et al. (34).
230 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
Kou and Le (9) made autogenous gas–tungsten arc welds in 309 stainless
steel, which has a composition close to the apex of the three-phase triangle,
as shown in Figure 9.4b. At 2 mm/s (5 ipm) welding speed, primary ferrite was
observed across the entire weld (similar to that shown in Figure 9.3b). At a
higher welding speed of 5mm/s (12ipm), however, primary austenite was
observed along the centerline, as shown in Figure 9.19. Electron probe microanalysis
(EPMA) revealed no apparent segregation of either Cr or Ni near
the weld centerline to cause the change in the primary solidification phase.
From Equation (8.3) the growth rate R = Vcosa, where a is the angle between
the welding direction and the normal to the pool boundary. Because of the
teardrop shape of the weld pool during welding (Figure 2.22), a drops to zero
Ni
Cr
Temperature
Composition
+ L + L
+
Co
CS CS
CL
CL
δ γ
γ
δ
δ γ
γ
δ
δ γ
Figure 9.18 Vertical section of Fe–Cr–Ni phase diagram showing change in solidification
from ferrite to austenite due to dendrite tip undercooling. Reprinted from
Brooks and Thompson (37).
Figure 9.19 Weld centerline austenite in an autogenous gas–tungsten arc weld of 309
stainless steel solidified as primary ferrite. From Kou and Le (9).
TRANSFORMATION IN AUSTENITIC STAINLESS STEEL WELDS 231
and R increases abruptly at the weld centerline. As such, the cooling rate (GR)
increases abruptly at the weld centerline, as pointed out subsequently by
Lippold (38).
Elmer et al. (34) calculated the dendrite tip undercooling for the alloys in
Figure 9.16 under various electron beam travel speeds. An undercooling of
45.8°C was calculated at the travel speed of 175mm/s, which is sufficient to
depress the dendrite tip temperature below the solidus temperature (Figure
9.15). This helps explain why alloy 4 can change from primary ferrite solidification
at low travel speeds to primary austenite solidification at much higher
travel speeds.
9.1.5 Ferrite Dissolution upon Reheating
Lundin and Chou (39) observed ferrite dissolution in multiple-pass or repair
austenitic stainless steel welds. This region exists in the weld metal of a previous
deposited weld bead, adjacent to but not contiguous with the fusion zone
of the deposited bead under consideration. Both the ferrite number and ductility
are lowered in this region, making it susceptible to fissuring under strain.
This is because of the dissolution of d-ferrite in the region of the weld metal
that is reheated to below the g-solvus temperature. Chen and Chou (40)
reported, in Figure 9.20, a significant ferrite loss in a 316 stainless steel weld
Figure 9.20 Effect of thermal cycles on ferrite content in 316 stainless steel weld: (a)
as welded; (b) subjected to thermal cycle of 1250°C peak temperature three times after
welding. Reprinted from Chen and Chou (40).
232 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
subjected to three postweld thermal cycles with a 1250°C peak temperature,
which is just below the g + d two-phase region of about 1280–1425°C.
9.2 AUSTENITE-TO-FERRITE TRANSFORMATION IN
LOW-CARBON, LOW-ALLOY STEEL WELDS
9.2.1 Microstructure Development
The dendrites or cells in the weld metal are not always discernible. First, significant
solute partitioning does not occur during solidification if the partition
ratio k is too close to 1. The miscrosegregation, especially solute segregation
to the interdendritic or intercellular regions, in the resultant weld metal can
be too little to bring out the dendritic or cellular structure in the grain interior
even though the grain structure itself can still be very clear. Second, if
solid-state diffusion occurs rapidly, microsegregation either is small or is
homogenized quickly, and the dendrites or cells in the resultant weld metal
can be unclear. Third, post-solidification phase transformations, if they occur,
can produce new microstructures in the grain interior and/or along grain
boundaries and the subgrain structure in the resultant weld metal can be
overshadowed.
Several continuous-cooling transformation (CCT) diagrams have been
sketched schematically to explain the development of the weld metal
microstructure of low-carbon, low-alloy steels (41–45). The one shown in
Figure 9.21 is based on that of Onsoien et al. (45).The hexagons represent the
transverse cross sections of columnar austenite grains in the weld metal. As
austenite (g) is cooled down from high temperature, ferrite (a) nucleates at
the grain boundary and grows inward.The grain boundary ferrite is also called
“allotriomorphic” ferrite, meaning that it is a ferrite without a regular faceted
grain boundary ferrite
sideplate ferrite
accicular ferrite
bainite
austenite
grain
log Time
Temperature
inclusion particle
grain boundary
martensite
cooling
curve
austenite
Figure 9.21 Continuous-cooling transformation diagram for weld metal of lowcarbon
steel.
TRANSFORMATION IN LOW-CARBON, LOW-ALLOY STEEL WELDS 233
shape reflecting its internal crystalline structure. At lower temperatures the
mobility of the planar growth front of the grain boundary ferrite decreases
and Widmanstatten ferrite, also called side-plate ferrite, forms instead. These
side plates can grow faster because carbon, instead of piling up at the planar
growth front, is pushed to the sides of the growing tips. Substitutional atoms
do not diffuse during the growth of Widmanstatten ferrite. At even lower temperatures
it is too slow for Widmanstatten ferrite to grow to the grain interior
and it is faster if new ferrite nucleates ahead of the growing ferrite. This new
ferrite, that is, acicular ferrite, nucleates at inclusion particles and has randomly
oriented short ferrite needles with a basket weave feature.
Figure 9.22 shows the microstructure of the weld metal of a low-carbon,
low-alloy steel (46). It includes in Figure 9.22a grain boundary ferrite (A),
Figure 9.22 Micrographs showing typical weld metal microstructures in low-carbon
steels: A, grain boundary ferrite; B, polygonal ferrite; C, Widmanstatten ferrite; D,
acicular ferrite; E, upper bainite; F, lower bainite. Reprinted from Grong and Matlock
(46).
(a)
(b)
234 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
Widmanstatten ferrite (C), and acicular ferrite (D) and in Figure 9.22b
upper bainite (E) and lower bainite (F). A polygonal ferrite (B) is also found.
Examination with transmission electron microscopy (TEM) is usually needed
to identify the upper and lower bainite. The microstructure of a low-carbon
steel weld containing predominately acicular ferrite is shown in Figure 9.23
and at a higher magnification in Figure 9.24 (47). The dark particles are
inclusions.
9.2.2 Factors Affecting Microstructure
Bhadeshia and Suensson (48) showed in Figure 9.25 the effect of several
factors on the development of microstructure of the weld metal: the weld
metal composition, the cooling time from 800 to 500°C (Dt8–5), the weld metal
oxygen content, and the austenite grain size. The vertical arrows indicate the
directions in which these factors increase in strength. This will be explained
with the help of CCT curves.
A. Cooling Time Consider the left CCT curves (broken lines) in Figure 9.26.
As cooling slows down (Dt8–5 increases) from curve 1 to curve 2 and curve 3,
and the transformation product can change from predominately bainite
(Figure 9.25c), to predominately acicular ferrite (Figure 9.25b) to predominately
grain boundary and Widmanstatten ferrite (Figure 9.25a).
Figure 9.23 Predominately acicular ferrite microstructure of a low-carbon, low-alloy
steel weld. Reprinted from Babu et al. (47).
TRANSFORMATION IN LOW-CARBON, LOW-ALLOY STEEL WELDS 235
Figure 9.24 Acicular ferrite and inclusion particles in a low-carbon, low-alloy steel
weld. Reprinted from Babu et al. (47).
Figure 9.25 Schematic showing effect of alloy additions, cooling time from 800 to
500°C, weld oxygen content, and austenite grain size. Reprinted from Bhadeshia and
Svensson (48).
B. Alloying Additions An increase in alloying additions (higher hardenability)
will shift the CCT curves toward longer times and lower temperatures.
Consider now cooling curve 3 in Figure 9.26. The transformation product can
change from predominately grain boundary and Widmanstatten ferrite (left
CCT curves) to predominately acicular ferrite (middle CCT curves) to predominately
bainite (right CCT curves). This is like what Figure 9.25 shows.
236 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
C. Grain Size Similar to the effect of alloying additions, an increase in the
austenite grain size (less grain boundary area for ferrite nucleation) will also
shift the CCT curves toward longer times and lower temperatures.
D. Weld Metal Oxygen Content The effect of the weld metal oxygen
content on the weld metal microstructure is explained as follows. First, as
shown in Figure 9.27, Fleck et al. (49) observed in submerged arc welds that
the austenite grain size before transformation decreases with increasing
weld metal oxygen content. Liu and Olson (50) observed that increasing the
weld metal oxygen content increased the inclusion volume fraction and
decreased the average inclusion size. In fact, a large number of smaller size
inclusions of diameters less than 0.1mm was found. Since fine second-phase
particles are known to increasingly inhibit grain growth by pinning the grain
boundaries as the particles get smaller and more abundant (51), increasing the
weld metal oxygen content should decrease the prior austenite grain size.
1 2 3
grain boundary ferrite
sideplate ferrite
acicular ferrite
bainite
austenite
log Time
Temperature
increasing alloying additions
increasing grain size
decreasing oxygen
Figure 9.26 Effect of alloying elements, grain size, and oxygen on CCT diagrams for
weld metal of low-carbon steel.
Q&T C-Mn-Mo-Nb
plate 3.0 MJ/m welds
Weld metal oxygen content, wt %
0.01 0.02 0.03 0.04 0.05
70
80
90
100
110
Prior austenite grain
diameter, m μ
Figure 9.27 Prior austenite grain diameter as a function of weld metal oxygen content
in submerged arc welds. Reprinted from Fleck et al. (49). Courtesy of American
Welding Society.
TRANSFORMATION IN LOW-CARBON, LOW-ALLOY STEEL WELDS 237
Therefore, the effect of decreasing the weld metal oxygen content is similar to
that of increasing the prior austenite grain size. This is just like what Figure
9.25 shows.
Second, larger inclusions, which are favored by lower weld metal oxygen
contents, can act as favorable nucleation sites for acicular ferrite (50). Appropriate
inclusions appear to be in the size range 0.2–2.0mm, and the mean size
of about 0.4mm has been suggested to be the optimum value (49, 51–53). Fox
et al. (54) suggested in submerged arc welds of HY-100 steel that insufficient
inclusion numbers are generated for the nucleation of acicular ferrite if the
oxygen content is too low (<200ppm). On the other hand, many small oxideinclusions (<0.2mm) can be generated if the oxygen content is too high(>300ppm). These inclusions, though too small to be effective nuclei for acicular
ferrite, reduce the grain size and thus provide much grain boundary area
for nucleation of grain boundary ferrite. As such, an optimum oxygen content
can be expected for acicular ferrite to form.This is just like what Figure 9.25b
shows.
The existence of an optimum oxygen content for acicular ferrite to form
has also been reported by Onsoien et al. (45) in GMAW with oxygen or carbon
dioxide added to argon, as shown clearly in Figure 9.28. With Ar–O2 as the
shielding gas, the shielding gas oxygen equivalent is the volume percentage of
O2 in the shielding gas. With Ar–CO2 as the shielding gas, it becomes the
volume percentage of CO2 in the shielding gas that will produce the same
oxygen content in the weld metal. As expected, the experimental results show
that the higher the shielding gas oxygen equivalent, the more hardenability
elements such as Mn and Si from the filler wire are oxidized. Consider again
cooling curve 3 in Figure 9.26. As the shielding gas oxygen equivalent is
reduced, the CCT curves can shift from left (broken lines) to middle (solid
0 5 10 15 20
0
20
40
60
80
100
Shielding gas oxygen
equivalent (vol. pct.)
Acicular ferrite content (vol. pct.)
ER 70S-3, 1.8 MJ/m
Oxygen in shielding gas
Carbon dioxide in
shielding gas
Figure 9.28 Acicular ferrite content as a function of shielding gas oxygen equivalent
for gas–metal arc welds. Reprinted from Onsoien et al. (45). Courtesy of American
Welding Society.
238 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
lines) and a predominately acicular microstructure is produced. However, as
the shielding gas oxygen equivalent is reduced further, the CCT curves can
shift from middle (solid lines) to right (dotted lines) and acicular ferrite no
longer predominates.
Other factors have also been reported to affect amount of acicular ferrite
in the weld metal. For example, it has been reported that acicular ferrite
increases with increasing basicity index of the flux for submerged arc welding
(54), Ti (55, 56), and Mn and Ni (57).
9.2.3 Weld Metal Toughness
Acicular ferrite is desirable because it improves the toughness of the weld
metal (55, 56). As shown in Figure 9.29, Dallam et al. (57) observed that the
Volume percent acicular ferrite
Energy absorbed (ft-lbs)
Joules
0 30 40 50 60 70 80
10
20
40
30
30
10
20
40
50
Q&T C-Mn-Mo-Nb plate
3.0 MJ/m welds
test temp. -40oC
7.6 X 7.6 mm izod tests
Figure 9.29 Subsize Charpy V-notch toughness values as a function of volume
fraction of acicular ferrite in submerged arc welds. Reprinted from Fleck et al. (49).
Courtesy of American Welding Society.
0 2 4 6 8 10
-60
-50
-40
-30
-20
-10
0
Shielding gas oxygen
equivalent (vol. pct.)
Transition temperature at
35J (deg. C)
Figure 9.30 Weld metal Charpy V-notch toughness expressed as transition temperature
as a function of shielding gas oxygen equivalent. Reprinted from Onsoien et al.
(45). Courtesy of American Welding Society.
REFERENCES 239
weld metal Charpy V-notch toughness in submerged arc welds increases with
increasing volume fraction of acicular ferrite in the weld metal. The interlocking
nature of acicular ferrite, together with its fine grain size, provides the
maximum resistance to crack propagation by cleavage.The formation of grain
boundary ferrite, ferrite side plates, or upper bainite is detrimental to weld
metal toughness, since these microstructures provide easy crack propagation
paths.
Onsoien et al. (45) tested the Charpy V-notch toughness of GMA weld
metal using an energy absorption of 35 J as the criterion for measuring the
transition temperature for ductile-to-brittle fracture. Figure 9.30 showed that
the maximum toughness (minimum transition temperature)occurs at a shielding
gas oxygen equivalent of about 2 vol %. This, as can be seen from Figure
9.28, essentially corresponds to the maximum amount of acicular ferrite in the
weld metal, thus clearly demonstrating the beneficial effect of acicular ferrite
on weld metal toughness. Ahlblom (58) has shown earlier a clear minimum in
the plots of Charpy V-notch transition temperature versus weld metal oxygen
content.
REFERENCES
1. David, S. A., Goodwin, G. M., and Braski, D. N., Weld. J., 58: 330s, 1979.
2. David, S. A., Weld. J., 60: 63s, 1981.
3. Lippold, J. C., and Savage,W. F., Weld. J., 58: 362s, 1974.
4. Lippold, J. C., and Savage,W. F., Weld. J., 59: 48s, 1980.
5. Cieslak, M. J., and Savage,W. F., Weld. J., 59: 136s, 1980.
6. Cieslak, M. J., Ritter, A. M., and Savage,W. F., Weld. J., 62: 1s, 1982.
7. Okagawa, R. K., Dixon, R. D., and Olson, D. L., Weld. J., 62: 204s, 1983.
8. Metals Handbook, Vol. 8, 8th ed., American Society for Metals, Metals Park, OH,
1973.
9. Kou, S., and Le,Y., Metall. Trans., 13A: 1141, 1982.
10. Brooks, J. A., and Garrison, Jr.,W. M., Weld. J., 78: 280s, 1999.
11. Inoue, H., Koseki,T., Ohkita, S., and Fuji, M., Sci.Technol.Weld. Join., 5: 385, 2000.
12. Schaeffler, A. L., Metal Prog., 56: 680, 1949.
13. Delong,W. T., Weld. J., 53: 273s, 1974.
14. Sato,Y. S., Kokawa, H., and Kuwana, T., Sci. Technol.Weld. Join., 4: 41, 1999.
15. Lundin, C. D., Chou, C. P. D., and Sullivan, C. J., Weld. J., 59: 226s, 1980.
16. Kotecki, D. J., and Siewert, T. A., Weld. J., 71: 171s, 1992.
17. McCowan, C. N., Siewert, T. A., and Olson, D. L., WRC Bull., 342: 1–36, 1989.
18. Lake, F. B., Expansion of the WRC-1988 Ferrite Diagram and Nitrogen Prediction,
Abstracts of Papers, 1988 AWS Convention, Detroit, MI, pp. 214–215.
19. Kotecki, D. J., Weld. J., 78: 180s, 1999.
20. Kotecki, D. J., Weld. J., 79: 346s, 2000.
240 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
21. Kotecki, D. J., Weld Dilution and Martensite Appearance in Dissimilar Metal
Welding, IIW Document II-C-195-00, 2000.
22. Kotecki, D. J., private communications, Lincoln Electric Company, Cleveland, OH,
2001.
23. Balmforth, M. C., and Lippold, J. C., Weld. J., 79: 339s, 2000.
24. Vitek, J. M., Iskander,Y. S., and Oblow, E. M., Weld. J., 79: 33s, 2000.
25. Vitek, J. M., Iskander,Y. S., and Oblow, E. M., Weld. J., 79: 41s, 2000.
26. Vitek, J. M., and David, S. A., in Trends in Welding Research in the United States,
Ed. S. A. David, American Society for Metals, Metals Park, OH, 1982.
27. Vitek, J. M., DasGupta, A., and David, S. A., Metall. Trans., 14A: 1833, 1983.
28. Katayama, S., and Matsunawa, A., Proc. International Congress on Applications of
Laser and Electro-Optics 84, 44: 60–67, 1984.
29. Katayama, S., and Matsunawa, A., in Proc. International Congress on Applications
of Laser and Electro-Optics 85, San Francisco, 1985, IFS Ltd., Kempston, Bedford,
UK, 1985, p. 19.
30. Olson, D. L., Weld. J., 64: 281s, 1985.
31. David, S. A.,Vitek, J. M., and Hebble, T. L., Weld. J., 66: 289s, 1987.
32. Bobadilla, M., Lacaze, J., and Lesoult, G., J. Crystal Growth, 89: 531, 1988.
33. Elmer, J.W., Allen, S. M., and Eagar, T.W., Metall. Trans., 20A: 2117, 1989.
34. Elmer, J.W., Eagar, T.W., and Allen, S. M., in Weldability of Materials, Eds. R. A.
Patterson and K. W. Mahin, ASM International, Materials Park, OH, 1990,
pp. 143–150.
35. Lippold, J. C., Weld. J., 73: 129s, 1994.
36. Koseki, T., and Flemings, M. C., Metall. Mater. Trans., 28A: 2385, 1997.
37. Brooks, J. A., and Thompson, A.W., Int. Mater. Rev., 36: 16, 1991.
38. Lippold, J. C., Weld. J., 64: 127s, 1985.
39. Lundin, C. D., and Chou, C. P. D., Weld. J., 64: 113s, 1985.
40. Chen, M. H., and Chou, C. P., Sci. Technol.Weld. Join., 4: 58, 1999.
41. Abson, D. J., and Dolby, R. E., Weld. Inst. Res. Bull., 202: July 1978.
42. Dolby, R. E., Metals Technol. 10: 349, 1983.
43. Classification of Microstructure in Low Carbon Low Alloy Weld Metal, IIW Doc.
IX-1282-83, 1983, International Institute of Welding, London, UK.
44. Vishnu, P. R., in ASM Handbook, Vol. 6: Welding, Brazing and Soldering, ASM
International, Materials Park, OH, 1993, pp. 70–87.
45. Onsoien, M. I., Liu, S., and Olson, D. L., Weld. J., 75: 216s, 1996.
46. Grong, O., and Matlock, D. K., Int. Metals Rev., 31: 27, 1986.
47. Babu, S. S., Reidenbach, F., David, S. A., Bollinghaus,Th., and Hoffmeister, H., Sci.
Technol.Weld. Join., 4: 63, 1999.
48. Bhadeshia, H. K. D. H., and Svensson, L. E., in Mathematical Modelling of Weld
Phenomena, Eds. H. Cerjak and K. Easterling, Institute of Materials, 1993.
49. Fleck, N. A., Grong,O., Edwards,G. R., and Matlock,D. K.,Weld. J., 65: 113s, 1986.
50. Liu, S., and Olson, D. L., Weld. J., 65: 139s, 1986.
51. Ashby, M. F., and Easterling, K. E., Acta Metall., 30: 1969, 1982.
52. Bhadesia, H. K. D. H., in Bainite in Steels, Institute of Materials, London, 1992,
Chapter 10.
53. Edwards, G. R., and Liu, S., in Proceedings of the first US-Japan Symposium
on Advanced Welding Metallurgy, AWA/JWS/JWES, San Francisco, CA, and
Yokohama, Japan, 1990, pp. 213–292.
54. Fox, A. G., Eakes, M.W., and Franke, G. L., Weld. J., 75: 330s, 1996.
55. Dolby, R. E., Research Report No. 14/1976/M,Welding Institute, Cambridge 1976.
56. Glover, A.G., McGrath, J.T., and Eaton, N. F., in S Toughness Characterization and
Specifications for HSLA and Structural Steels. ed. P. L. Manganon, Metallurgical
Society of AIME, NY, pp. 143–160.
57. Dallam, C. B., Liu, S., and Olson, D. L., Weld. J., 64: 140s, 1985.
58. Ahlblom, B., Document No. IX-1322-84, International Institute of Welding,
London, 1984.
FURTHER READING
1. Grong, O., and Matlock, D. K., Int. Meter. Rev., 31: 27, 1986.
2. Brooks, J. A., and Thompson, A.W., Int. Mater. Rev., 36: 16, 1991.
3. Vishnu, P. R., in ASM Handbook, Vol. 6: Welding, Brazing and Soldering, ASM
International, Materials Park, OH, 1993, pp. 70–87.
4. Brooks, J. A., and Lippold, J. C., in ASM Handbook, Vol. 6: Welding, Brazing and
Soldering, ASM International, Materials Park, OH, 1993, pp. 456–470.
PROBLEMS
9.1 (a) Construct pseudobinary phase diagrams for 55% and 74% Fe. Mark
on the diagrams the approximate compositions of 310 (essentially
Fe–25 Cr–20 Ni) and 304 (essentially Fe–18Cr–8Ni) stainless steels.
(b) From the diagrams and the approximate compositions, indicate the
primary solidification phases.
9.2 A 308 stainless-steel filler (essentially Fe–20Cr–10Ni) is used to weld 310
stainless steel.What is the primary solidification phase if the dilution ratio
is about 60%?
9.3 A 304 stainless-steel sheet with a composition given below is welded
autogenously with the GTAW process.The shielding gas is Ar-2% N2, and
the nitrogen content of the weld metal is about 0.13%. The contents of
other alloying elements are essentially the same as those in the base
metal.
(a) Calculate the ferrite numbers for the base metal and the weld metal.
(b) The weld metal exhibits the primary solidification phase of austenite,
and the ferrite content measurements indicate essentially zero
PROBLEMS 241
ferrite number. Is the calculated ferrite number for the weld metal
consistent with the observed one? (Composition: 18.10Cr, 8.49Ni,
0.060C, 0.66Si, 1.76Mn, 0.36Mo, 0.012S, 0.036P, and 0.066N.)
9.4 A significant amount of ferrite is lost in a 316 stainless steel weld after
being subjected to three postweld thermal cycles with a 1250°C peak temperature,
which is just below the g + d two-phase region of about 1280 to
1425°C. Sketch a curve of ferrite number vs. temperature from 900 to
1400°C and explain it.
9.5 Kou and Le (9) quenched 309 stainless steel during autogenous GTAW.
The weld metal side of the quenched pool boundary showed dendrites
of d-ferrite but the weld pool side showed dendrites of primary austenite.
Explain why.
9.6 It has been observed in welding austenitic stainless steel with a teardropshaped
weld pool that the weld metal solidifies with primary ferrite
except near the centerline, where it solidifies as primary austenite. Sketch
a curve of the growth rate R versus the distance y away from the weld
centerline. How does your result explain the ferrite content change near
the centerline?
242 POST-SOLIDIFICATION PHASE TRANSFORMATIONS
10 Weld Metal Chemical
Inhomogeneities
In this chapter we shall discuss chemical inhomogeneities in the weld metal,
including solute segregation, banding, inclusions, and gas porosity. Solute segregation
can be either microsegregation or macrosegregation. Microsegregation
refers to composition variations across structures of microscopic sizes, for
instance, dendrite arms or cells (Chapter 6). Macrosegregation, on the other
hand, refers to variations in the local average composition (composition averaged
over many dendrites) across structures of macroscopic sizes, for instance,
the weld. Macrosegregation has been determined by removing samples across
the weld metal with a small drill for wet chemical analysis (1). Microsegregation,
on the other hand, has been determined by electron probe microanalysis
(EPMA) (2–8) or scanning transmission electron microscopy (STEM) (9–14).
The spacial resolution is lower in the former (e.g., about 1mm) and higher in
the latter (e.g., about 0.1mm).
10.1 MICROSEGREGATION
Alloying elements with an equilibrium segregation coefficient k < 1 tend tosegregate toward the boundary between cells or dendrite arms, and those withk > 1 tend to segregate toward the core of cells or dendrite arms (Chapter 6).
Microsegregation can have a significant effect on the solidification cracking
susceptibility of the weld metal (Chapter 11).
10.1.1 Effect of Solid-State Diffusion
Microsegregation can be reduced significantly by solid-state diffusion during
and after solidification. Consequently, microsegregation measured after
welding may not represent the true microsegregation during welding, which is
more relevant to solidification cracking (Chapter 11).
Kou and Le (15) quenched stainless steels with liquid tin during autogenous
GTAW in order to preserve the high-temperature microstructure around
the pool boundary. In 430 ferritic stainless steel the dendrites were clear near
the quenched pool boundary but became increasingly blurred away from it,
suggesting homogenization of microsegregation by the solid-state diffusion.
243
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
Ferrite has a body-centered-cubic (bcc) structure, which is relatively open
and thus easy for diffusion. In 310 austenitic stainless steel, however, the
dendrites were clear even away from the quenched pool boundary, suggesting
much less homogenization due to solid-state diffusion. Austenite has a facecentered-
cubic (fcc) structure, which is more close packed than bcc and thus
more difficult for diffusion.
Brooks and Garrison (8) quenched a precipitation-strengthened martensitic
stainless steel with liquid tin during GTAW. They measured microsegregation
across the columnar dendrites, as shown in Figure 10.1a, where the
quenched weld pool is in the upper right corner of the micrograph. The weld
metal solidified as a single-phase ferrite. As shown in Figure 10.1b, near the
244 WELD METAL CHEMICAL INHOMOGENEITIES
Figure 10.1 Microsegregation across columnar dendrites near quenched weld pool in
a martensitic stainless steel: (a, b) near growth front; (c) 400mm behind. Reprinted from
Brooks and Garrison (8). Courtesy of American Welding Society.
solidification front segregation of Ni, Cu, and Nb toward the boundaries
between dendrites is clear, the measured equilibrium segregation coefficients
being 0.85, 0.9, and 0.36, respectively. The concentration peaks correspond to
the dendrite boundaries. The microsegregation profiles at 400mm behind the
solidification front are shown in Figure 10.1c. As shown, microsegregation is
reduced significantly by solid-state diffusion at elevated temperatures.
Instead of using the Scheil equation (Chapter 6), Brooks and Baskes
(16–19) calculated weld metal microsegregation considering solid-state diffusion
and Lee et al. (20) further incorporated the kinetics of dendrite coarsening.
Figure 10.2a shows the calculated microsegregation in a Fe–23Cr–12Ni
stainless steel that solidifies as ferrite (17, 18), where r is the radius of the
growing cell and R is the final cell radius. Since k > 1 for Cr, the cell core (r =
0) is rich in Cr. The cell grows to about 60% of its full radius in 0.015 s after
solidification starts and 100% in 0.15 s, with Cr diffusing away from the cell
MICROSEGREGATION 245
0 0.2 0.4 0.6 0.8 1
10
20
30
40
0
Normalized Distance (r/R)
Cr Concentration ( wt % )
t=.18 sec
t= 3.2 sec
Cr
Ni
0 0.2 0.4 0.6 0.8 1
15
20
25
30
Normalized Distance (r/R)
Cr Concentration ( wt % )
t=.015 sec
t=.075 sec
t=.25 sec
t=.75 sec
t=.15 sec
(a)
(b)
Figure 10.2 Calculated microsegregation: (a) Fe–23Cr–12Ni (solid-state diffusion significant);
(b) Fe–21Cr–14Ni. Reprinted from Brooks (18).
core as solidification proceeds. Diffusion is fast in ferrite because of its
relatively loosely packed bcc structure. Chromium diffusion continues after
solidification is over (0.25 s) and the resultant cell is nearly completely
homogenized in 0.75 s after solidification starts. Figure 10.2b, shows the calculated
microsegregation in a Fe–21Cr–14Ni stainless steel that solidifies as
austenite at the time of final solidification (0.18 s) and 3 s later. Both Cr and
Ni are highly segregated to the cell boundary. Little solid-state diffusion occurs
during cooling except near the cell boundary, where some diffusion occurs
because of the steep concentration gradients. Diffusion is slow in austenite
because of its more densely packed fcc structure.
Figure 10.3 shows the STEM microsegregation profiles across a dendrite
arm in the weld metal of 308 austenitic stainless steel (about Fe–20Cr–10Ni)
measured by David et al. (10). The primary solidification phase is d-ferrite
246 WELD METAL CHEMICAL INHOMOGENEITIES
Figure 10.3 Microsegregation in 308 stainless steel weld: (a) phase diagram; (b) TEM
micrograph of a dendrite arm; (c) microsegregation across the arm. Reprinted from
David et al. (10). Courtesy of American Welding Society.
(Figure 10.3a), but the dendrite arm has transformed to g except for the
remaining d-ferrite core (Figure 10.3b). Nickel (k < 1) segregates toward thedendrite boundary, while Cr (k > 1) segregates toward the dendrite core
(Figure 10.3c). Similar microsegregation profiles have been measured in welds
of 304L (11, 12) and 309 (13) stainless steels, and it has been confirmed that
diffusion occurs during the d Æ g transformation (6, 7, 10, 14).
10.1.2 Effect of Dendrite Tip Undercooling
In addition to solid-state diffusion, microsegregation can also be affected by
the extent of dendrite tip undercooling. The difference between the equilibrium
liquidus temperature TL and the dendrite tip temperature Tt is the total
undercooling DT, which can be divided into four parts:
DT = DTC + DTR + DTT + DTK (10.1)
where DTC = concentration-induced undercooling
DTR = curvature-induced undercooling
DTT = thermal undercooling
DTK = kinetic undercooling
The solute rejected by the dendrite tip into the liquid can pile up and cause
undercooling at the dendrite tip DTC, similar to constitutional supercooling at
a planar growth front (Chapter 6). The equilibrium liquidus temperature in a
phase diagram is for a flat solid–liquid interface, and it is suppressed if the
interface has a radius of curvature like a dendrite tip. Thermal undercooling
is present where there is a significant nucleation barrier for the liquid to
transform to solid. The kinetic undercooling, which is usually negligible, is
associated with the driving force for the liquid atoms to become attached to
the solid. It has been observed that the higher the velocity of the dendrite tip,
Vt, the smaller the radius of the dendrite tip,Rt, and the larger the undercooling
at the dendrite tip, DT (21).This is shown schematically in Figure 10.4. Models
have been proposed for dendrite tip undercooling, including those by Burden
and Hunt (22) and Kurz et al. (23).
Brooks et al. (19) calculated weld metal microsegregation in Fe–Nb alloys
considering dendrite tip undercooling as well as solid-state diffusion. Composition
analyses of tin-quenched gas–tungsten arc welds indicated a dendrite tip
undercooling of 3.07°C in Fe–1.8Nb and 5.63°C in Fe–3.3Nb.Figure 10.5a show
the calculated results for a Fe–3.3Nb alloy, based on the model of Kurz et al.
(23) with k = 0.28 for Nb and DL = 6.8cm2/s. Undercooling results in a higher
core Nb concentration during the initial stages of solidification (0.08 s). But as
solidification proceeds to completion (1.2 s) and the weld cools (5.6 s), the concentration
profiles with and without undercooling start to converge due to
the significant effect of solid-state diffusion. It was, therefore, concluded that
MICROSEGREGATION 247
248 WELD METAL CHEMICAL INHOMOGENEITIES
TL
Tt1
Vt1
Vt2
Tt2
Rt1
Rt2
lower tip velocity,
larger tip radius,
smaller tip
undercooling
T1 T2
higher tip velocity,
smaller tip radius,
larger tip
undercooling
Liquid
Temperature, T
Distance, x
Dendrite Tips:
Δ
Δ
Figure 10.4 Dendrite tip undercooling.
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
0
1
2
3
4
5
With tip undercooling
Without tip undercooling
t = 5.6s
t = 1.2s
t = 0.08s
Fraction solidified ( r/R)
Nb concentration (wt%)
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
1
2
3
4
5
With tip undercooling
Without tip undercooling
t = 0.008s
Fraction solidified (r/R)
Nb concentration (wt%)
t = 0.08s
(a)
(b)
Figure 10.5 Calculated microsegregation in a Fe–3.3Nb weld: (a) slower cooling rate;
(b) higher cooling rate. Reprinted from Brooks et al. (19).
although tip undercooling can result in a higher Nb concentration during
initial solidification, its effect on the final microsegregation is small because of
the overwhelming effect of solid-state diffusion.With a very high cooling rate
of 104°C/s such as in high-energy beam welding (less time for solid-state diffusion),
Figure 10.5b shows that the effect of undercooling can be important
throughout solidification and subsequent cooling. Final microsegregation is
significantly less with dendrite tip undercooling than without.
10.2 BANDING
10.2.1 Compositional and Microstructural Fluctuations
In addition to microsegregation across dendrites, microsegregation can
also exist in the weld metal as a result of banding during weld pool
solidification (Chapter 6). Banding in welds can cause perturbations in the
solidification structure as well as the solute concentration (24). Figure 10.6
shows banding and rippling near the centerline of the as-welded top surface
of a YAG laser weld (conduction mode) in a 304 stainless steel, with alternating
bands of dendritic and planarlike structures. Figure 10.7 shows alternating
bands of austenite (light) and martensite (dark) in an A36 steel
welded with a E309LSi filler by GTAW (25). The composition of A36 is
Fe–0.71Mn–0.29Cu–0.18Si–0.17C–0.13Ni–0.09Cr–0.05Mo and that of E309LSi
is Fe–23.16Cr–13.77Ni–1.75Mn–0.79Si–0.20Cu–0.16Mo–0.11N–0.02C. The
high hardness (smaller indentation marks) in the darker regions reflects the
presence of martensite. The hydrogen crack in the martensite near but within
BANDING 249
Figure 10.6 Banding and rippling near centerline of as-welded top surface of a 304
stainless steel YAG laser welded in conduction mode.
the fusion boundary was promoted by the use of Ar–6% H2 as the shielding
gas.
10.2.2 Causes
Banding in the weld metal can occur due to a number of reasons. Fluctuations
in the welding speed during manual welding or arc pulsing during pulsed arc
welding can cause banding. However, even under steady-state welding conditions,
banding can still occur, as evidenced by the surface rippling of the weld.
Beside fluctuations in the welding speed and the power input, the following
mechanisms also have been proposed: Solidification halts due to the rapid evolution
of latent heat caused by high solidification rates during welding (26–28),
oscillations of weld pool metal due to uncontrollable variations in arc stability
and the downward stream of the shielding gas (29), and fluctuations in weld
pool turbulence due to electromagnetic effects (30, 31).
10.3 INCLUSIONS AND GAS POROSITY
Inclusions and gas porosity tend to deteriorate the mechanical properties of the
weld metal. Gas–metal and slag–metal reactions can produce gas porosity and
inclusions in the weld metal and affect the weld metal properties (Chapter 3).
Inclusions can also result from incomplete slag removal during multiple-pass
welding (32), the large dark-etching particle near position D in Figure 10.8
being one example (33). Figure 10.9 shows trapped surface oxides as inclusions
in the weld metal and their elimination by modifying the joint design (34).
250 WELD METAL CHEMICAL INHOMOGENEITIES
Figure 10.7 Banding near fusion boundary of a carbon steel welded with an austenitic
stainless steel filler metal. Reprinted from Rowe et al. (25). Courtesy of American
Welding Society.
Various ways of reducing gas porosity in the weld metal have already been
described in Chapter 3. Gas pores can be round or interdendritic, as shown in
Figure 10.10, which shows a gas–metal arc weld in 7075 aluminum made with
a 4043 filler metal. Similar gas pores have been reported in other aluminum
welds (35–37). Although round gas pores can be randomly distributed in the
weld metal, they can also line up and form bands of porosity when banding
is severe during weld metal solidification (35, 36). It is often difficult to tell
whether interdendritic pores are due to gas formation or due to solidification
INCLUSIONS AND GAS POROSITY 251
Figure 10.8 Multipass weld with slag inclusions (D) and other defects, including lack
of fusion (A), lamellar tearing (B), poor profile (C), and undercut (E). Reprinted from
Lochhead and Rodgers (33). Courtesy of American Welding Society.
Figure 10.9 Joint designs and trapping of surface oxides in aluminum welding: (a)
oxide trapped; (b) oxide not trapped (34).
shrinkage (21). However, if they are due to gas formation, they must have
formed during the latter stages of solidification, where the dendritic structure
has essentially been established.
10.4 INHOMOGENEITIES NEAR FUSION BOUNDARY
Dissimilar metal welding is often encountered in welding, where a filler metal
different in composition from the base metal is used or where two base metals
different in composition are welded together. In dissimilar metal welding, the
region near the fusion boundary often differ significantly from the bulk weld
metal in composition and sometimes even microstructure and properties. The
region, first discovered by Savage et al. (38, 39), has been called the unmixed
zone (38, 39), filler-metal-depleted area (40), partially mixed zone (41), intermediate
mixed zone (42), and hard zone (43). It has been observed in various
welds, including stainless steels, alloy steels, aluminum alloys, and superalloys
(38–47).
10.4.1 Composition Profiles
Ornath et al. (46) determined the composition profiles across the fusion
boundary of a low-alloy steel welded with a stainless steel filler of
Fe–18Cr–8Ni–7Mn, as shown in Figure 10.11. Plotting the concentration
against the distance from the fusion boundary according to Equation (6.17)
(for solute segregation during the initial transient of solidification), they
obtained a straight line. As such, they proposed that segregation rather than
diffusion is responsible for the observed composition profiles. In other words,
the composition profiles are caused by the rejection of Cr, Ni, and Mn into the
melt by the solid weld metal during the initial stage of solidification.
252 WELD METAL CHEMICAL INHOMOGENEITIES
Figure 10.10 Porosity in aluminum weld showing both spherical and interdendritic
gas pores. One interdendritic pore is enlarged for clarity.
Baeslack et al. (45) determined the composition profiles across the fusion
boundary of a 304L stainless steel welded with a 310 stainless steel filler, as
shown in Figure 10.12. Unlike the composition profiles shown in Figure 10.11,
the weld metal next to the fusion line, labeled as the unmixed zone by the
authors, has essentially the same composition as the base metal, suggesting
stagnant melted base metal unmixed with the filler metal. Figure 10.13 is a
schematic sketch for an unmixed zone.The arrows show the directions of local
INHOMOGENEITIES NEAR FUSION BOUNDARY 253
Figure 10.11 Fusion boundary of a low-alloy steel welded with an austenitic stainless
steel electrode: (a) microstructure (magnification 55¥); (b) segregation. Reprinted from
Ornath et al. (46). Courtesy of American Welding Society.
Figure 10.12 Unmixed zone in a 304L stainless steel welded with a 310 filler metal:
(a) microstructure; (b) composition profiles. Reprinted from Baeslack et al. (45). Courtesy
of American Welding Society.
heat-affected
zone
partially
melted zone
unmixed zone
base metal
fusion
boundary
bulk weld metal
Figure 10.13 Zone of unmixed melted base metal along the fusion boundary.
254 WELD METAL CHEMICAL INHOMOGENEITIES
fluid flow during welding, which is not strong enough for thorough mixing but
strong enough to move parts of the melted base metal.Also shown in the figure
are the partially melted zone (Chapters 12 and 13) and the heat-affected zone
(Chapters 14–18).
10.4.2 Effect of Inhomogeneities
Inhomogeneities in the region along the fusion boundary have been reported
to cause problems, including hydrogen cracking (48, 49), corrosion (50), and
stress corrosion cracking (45). Martensite often exists in the region in carbon
or alloy steels welded with austenitic stainless steel fillers. This is because the
weld metal composition here can be within the martensite region of the constitutional
diagrams (Chapter 9). This composition transition shown in Figure
10.11 covers compositions in the martensite range of the Schaeffler diagram,
thus explaining the formation of martensite (46). Linnert (48) observed
martensite and hydrogen cracking in weld metal along the fusion boundary of
a Cr–Ni–Mo steel welded with a 20Cr–10Ni stainless steel filler. Savage et al.
(49) reported similar hydrogen cracking in HY-80 welds. Rowe et al. (25)
showed hydrogen cracking along the weld metal side of the fusion boundary
in a A36 steel gas–tungsten arc welded with a ER308 stainless steel filler metal
and an Ar-6% H2 shielding gas.
Omar (43) welded carbon steels to austenitic stainless steels by SMAW.
Figure 10.14 shows that the hard martensite layer in the weld metal along the
carbon steel side of the fusion boundary can be eliminated by using a Ni-base
filler metal plus preheating and controlling the interpass temperature. The
austenitic stainless steel electrode E309, which has much less Ni and more Cr,
did not work as well.
10.5 MACROSEGREGATION IN BULK WELD METAL
Weld pool convection (Chapter 4) can usually mix the weld pool well to
minimize macrosegregation across the resultant weld metal. Houldcroft (1) found
no appreciable macrosegregation in pure aluminum plates single-pass welded
with Al–5% Cu filler. Similar results were observed in Al–l.0Si–l.0Mg plates singlepass
welded using either Al–4.9% Si or Al–1.4% Si as the filler metal (1).
However, in single-pass dissimilar-metal welding, macrosegregation can still occur
if weld pool mixing is incomplete. In multiple-pass dissimilar-metal welding,
macrosegregation can still occur even if weld pool mixing is complete in each pass.
10.5.1 Single-Pass Welds
Macrosegregation can occur in a dissimilar weld between two different base
metals because of insufficient mixing in the weld pool. Matsuda et al. (51)
showed Cu macrosegregation across an autogenous gas–tungsten arc
weld between thin sheets of 1100 aluminum (essentially pure Al) and 2024
MACROSEGREGATION IN BULK WELD METAL 255
256 WELD METAL CHEMICAL INHOMOGENEITIES
(a)
(b)
Figure 10.14 Carbon steel side of weld metal in a weld between a carbon steel and
an austenitic stainless steel made with a Ni-based filler metal: (a) martensite along
fusion boundary; (b) martensite avoided by preheating and controlling interpass temperature.
Reprinted from Omar (43). Courtesy of American Welding Society.
(a) (b)
Figure 10.15 Macrosegregation in a laser beam weld between Ti–6Al–4V and
Ti–3Al–8V–6Cr–4Mo–4Zr (bC): (a) transverse cross section; (b) composition profiles.
Reprinted from Liu et al. (53). Courtesy of American Welding Society.
aluminum (essentially Al–4.4Cu). Macrosegregation was reduced through
enhanced mixing by magnetic weld pool stirring. In laser or electron
beam welding, the welding speed can be too high to give the weld metal
sufficient time to mix well before solidification (52, 53). Figures 10.15
shows macrosegregation in a laser weld between Ti–6Al–4V (left) and
Ti–3Al–8V–6Cr–4Mo–4Zr (right) (53).
Macrosegregation due to lack of weld pool mixing has also been observed
in GTAW of some powder metallurgy alloys. Such alloys are made by consolidation
of rapidly solidified powder having some special properties, for
instance, extended solubility of alloying elements. Figure 10.16 shows the lack
MACROSEGREGATION IN BULK WELD METAL 257
(a)
(b)
Figure 10.16 Powder metallurgy Al–10Fe–5Ce alloy gas–tungsten arc welded with
Al–5Si filler metal: (a) AC; (b) DCEN. Reprinted from Metzger (54). Courtesy of
American Welding Society.
Filler D
Base
metal A
Base
d metal B
a b
c
Backing C
(2) % element E in weld bead =
[a (% E in a) + b (% E in b) + c (% E in c)
+ d (% E in d)] / (a + b + c + d)
(1) % dilution =
×
a + b + c
100
a + b + c + d
Figure 10.17 Filler metal dilution and composition in dissimilar-metal welding.
Reprinted from Estes and Turner (55). Courtesy of American Welding Society.
258 WELD METAL CHEMICAL INHOMOGENEITIES
Figure 10.18 Macrosegregation in a multiple-pass weld between 4130 steel and 304
stainless steel. Reprinted from Estes and Turner (55). Courtesy of American Welding
Society.
of mixing between an Al–10Fe–5Ce powder metallurgy base metal and a 4043
(Al–5Si) filler metal in AC GTAW and improved mixing and reduced gas
porosity with DCEN GTAW (54). This may be because in GTAW weld penetration
is higher with DCEN than with AC (Chapter 1).
10.5.2 Multiple-Pass Welds
Figure 10.17 shows the filler metal dilution and composition of the first bead
(root pass) in a dissimilar-metal weld (55). The composition of the bead
depends not only on the compositions of the base and filler metals but also on
the extent of dilution. Apparently, the second bead to be deposited on top of
the first bead will have a different composition from the first bead regardless
of the extent of weld pool mixing.
Figure 10.18 shows the composition profiles across a multiple-pass weld
between 4130 alloy steel and 304 stainless steel with a 312 stainless steel as
the filler metal (55). Composition differences between beads are evident. For
instance, the Cr content varies from 18% in the first bead to 25% in the third.
As a result of such macrosegregation, the ferrite content (which affects the
resistance to solidification cracking and corrosion) varies from one bead to
another, as shown in Figure 10.19.
MACROSEGREGATION IN BULK WELD METAL 259
Figure 10.19 Variations in microstructure in a multiple-pass weld between 4130 steel
and 304 stainless steel. Reprinted from Estes and Turner (55). Courtesy of American
Welding Society.
REFERENCES
1. Houldcroft, R. T., Br.Weld. J., 1: 468, 1954.
2. Lippold, J. C., and Savage,W. F., Weld. J., 58: 362s, 1979.
3. Takalo, T., Suutala, N., and Moisio, T., Metall. Trans., 10A: 1173, 1979.
4. Suutala, N., Takalo, T., and Moisio, T., Metall. Trans., 10A: 1183, 1979.
5. Ciestak, M. J., and Savage,W. F., Weld. J., 59: 136s, 1980.
6. Suutala, N., Takalo, T., and Moisio, T., Weld. J., 60: 92s, 1981.
7. Leone, G. L., and Kerr, H.W., Weld. J., 61: 13s, 1982.
8. Brooks, J. A., and Garrison,W. M. Jr., Weld. J., 78: 280s, 1999.
9. Gould, J. C., Ph.D. Thesis, Carnegie-Mellon University, Pittsburgh, PA, 1983.
10. David, S. A., Goodwin, G. M., and Braski, D. N., Weld. J., 58: 330s, 1979.
11. Lyman, C. E., Manning, P. E., Duquette,D. J., and Hall, E., Scan. Electron. Microsc.,
1: 213, 1978.
12. Lyman, C. E., Weld. J., 58: 189s, 1979.
13. Brooks, J. A., Ph.D. Thesis, Carnegie-Mellon University, Pittsburgh, PA, 1982.
14. Cieslak, M. J., Ritter, A. M., and Savage,W., Weld. J., 61: 1s, 1982.
15. Kou, S., and Le,Y., Metall. Trans.A, 13A: 1141, 1982.
16. Brooks, J. A., and Baskes, M. I., in Advances in Welding Science and Technology,
Ed. S. A. David, ASM International, Materials Park, OH, March 1986, p. 93.
17. Brooks, J. A., and Baskes, M. I., in Recent Trends in Welding Science and Technology,
Eds. S. A. David and J. M. Vitek, ASM International, Materials Park, OH,
March 1990, p. 153.
18. Brooks, J. A., in Weldability of Materials, Eds. R. A. Patterson and K. W. Mahin,
ASM International, Materials Park, OH, March 1990, p. 41.
19. Brooks, J. A., Li, M., Baskes, M. I., and Yang, N. C. Y., Sci. Technol.Weld. Join., 2:
160, 1997.
20. Lee, J.Y., Park, J. M., Lee, C. H., and Yoon, E. P., in Synthesis/Processing of Lightweight
Metallic Materials II, Eds. C. M.Ward-Close, F. H. Froes, S. S. Cho, and D. J.
Chellman,The Minerals, Metals and Materials Society,Warrendale, PA 1996, p. 49.
21. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
22. Burden, M. H., and Hunt, J. D., J. Crystal Growth, 22: 109, 1974.
23. Kurz,W., Giovanola, B., and Trivedi, R., Acta Metall., 34: 823, 1986.
24. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20: 83, 1975.
25. Rowe, M. D., Nelson, T.W., and Lippold, J. C., Weld. J., 78: 31s, 1999.
26. Gurev, H. S., and Stout, R. D., Weld. J., 42: 298s, 1963.
27. Cheever, D. L., and Howden, D. G., Weld. J., 48: 179s, 1969.
28. Morchan, B. A., and Abitdnar, A., Automat.Weld., 21: 4, 1968.
29. Ishizaki, K., J. Jpn.Weld. Soc., 38: 1963.
30. Jordan, M. F., and Coleman, M. C., Br.Weld. J., 15: 552, 1968.
31. Waring, J., Aust.Weld. J., 11: 15, 1967.
32. Gurney, T. R., Fatigue of Welded Structures, Cambridge University Press,
Cambridge, 1968, p. 156.
260 WELD METAL CHEMICAL INHOMOGENEITIES
33. Lochhead, J. C., and Rodgers, K. J., Weld. J., 78: 49, 1999.
34. Inert Gas Welding of Aluminum Alloys, Society of the Fusion Welding of Light
Metals, Tokyo, Japan, 1971 (in Japanese).
35. D’Annessa, A. T., Weld. J., 45: 569s, 1966.
36. D’Annessa, A. T., Weld. J., 49: 41s, 1970.
37. D’Annessa, A. T., Weld. J., 46: 491s, 1967.
38. Savage,W. F., and Szekeres, E. S., Weld. J., 46: 94s, 1967.
39. Savage,W. F., Nippes, E. F., and Szekeres, E. S., Weld. J., 55: 260s, 1976.
40. Duvall, D. S., and Owczarski,W. A., Weld. J., 47: 115s, 1968.
41. Kadalainen, L. P., Z. Metallkde., 70: 686, 1979.
42. Doody, T., Weld. J., 61: 55, 1992.
43. Omar, A. A., Weld. J., 67: 86s, 1998.
44. Lippold, J. C., and Savage,W. F., Weld. J., 59: 48s, 1980.
45. Baeslack,W. A. III, Lippold, J. C., and Savage,W. F., Weld. J., 58: 168s, 1979.
46. Ornath, F., Soudry, J.,Weiss, B. Z., and Minkoff, I., Weld. J., 60: 227s, 1991.
47. Albert, S. K., Gills,T.P. S.,Tyagi,A. K., Mannan, S. L.,Kulkarni, S.D., and Rodriguez,
P., Weld. J., 66: 135s, 1997.
48. Linnert, G. E., Welding Metallurgy, Vol. 2. American Welding Society, Miami, FL,
1967.
49. Savage,W. F., Nippes, E. F., and Szekeres, E. S., Weld. J., 55: 276s, 1976.
50. Takalo, T., and Moisio, T., IIW Annual Assembly, Tel Aviv, 1975.
51. Matsuda, F., Ushio, M., Nakagawa, H., and Nakata, K., in Proceedings of the
Conference on Arc Physics and Weld Pool Behavior, Vol. 1, Welding Institute,
Arbington Hall, Cambridge, 1980, p. 337.
52. Seretsky, J., and Ryba, E. R., Weld. J., 55, 208s, 1976.
53. Liu, P. S., Baeslack III,W. A., and Hurley, J., Weld. J., 73: 175s, 1994.
54. Metzger, G. E., Weld. J., 71: 297s, 1992.
55. Estes, C. L., and Turner, P.W., Weld. J., 43: 541s, 1964.
FURTHER READING
1. Davies, G. J., Solidification and Casting,Wiley, New York, 1973.
2. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20: 83, 1975.
3. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.
4. Savage,W. F., Weld.World, 18: 89, 1980.
PROBLEMS
10.1 With the help of Schaeffler’s diagram, show that martensite can form in
the fusion zone at 70mm from the fusion boundary of the weld shown
in Figure 10.11.
PROBLEMS 261
10.2 Butt welding of 5052 aluminum (Al–2.5Mg) with a single-V joint is
carried out with 5556 filler (Al–5.1Mg). The dilution ratio of the first
pass is 80%. In the second pass 40% of the material comes from the
filler wire, 40% from the base metal, and 20% from the first pass.
Calculate the compositions of the two passes, assuming uniform mixing
in both.
10.3 Suppose that in the previous problem the workpiece composition is
Fe–25Cr–20Ni and the filler composition is Fe–20Cr–10Ni.What is the
difference in the ferrite content between the two passes based on
Schaeffler’s diagram.
10.4 Consider the pseudo-binary-phase diagram shown in Figure 10.3a.
Sketch the Ni and Cr concentration profiles across a dendrite arm for
an alloy that has a composition just to the left of point b.
10.5 Consider welding Ni to Ti. Can macrosegregation occur in LBW? Why
or why not? Is the chance of macrosegregation higher or lower in
GTAW than in LBW?
10.6 Explain why gas porosity can be severe in the GTAW of powder metallurgy
alloy Al–10Fe–5Ce (Figure 10.16a). Explain why gas porosity
can be significantly less with DCEN than with AC.
10.7 Consider banding in the YAG laser weld of 304 stainless steel (Figure
10.6).What could have caused banding in this weld? Is the growth rate
higher during dendritic or planarlike solidification and why?
262 WELD METAL CHEMICAL INHOMOGENEITIES
11 Weld Metal
Solidification Cracking
Solidification cracking in the weld metal will be described, the metallurgical
and mechanical factors affecting the crack susceptibility will be discussed, and
the methods for reducing cracking will be presented.
11.1 CHARACTERISTICS, CAUSE, AND TESTING
11.1.1 Intergranular Cracking
Solidification cracking, which is observed frequently in castings and ingots, can
also occur in fusion welding, as shown in Figure 11.1. Such cracking, as shown
in Figure 11.2, is intergranular, that is, along the grain boundaries of the weld
metal (1). It occurs during the terminal stage of solidification, when the tensile
stresses developed across the adjacent grains exceed the strength of the almost
completely solidified weld metal (2–4). The solidifying weld metal tends to
contract because of both solidification shrinkage and thermal contraction.The
surrounding base metal also tends to contract, but not as much, because it is
neither melted nor heated as much on the average.Therefore, the contraction
of the solidifying metal can be hindered by the base metal, especially if the
workpiece is constrained and cannot contract freely. Consequently, tensile
stresses develop in the solidifying weld metal. The severity of such tensile
stresses increases with both the degree of constraint and the thickness of the
workpiece.
The various theories of solidification cracking (4–7) are effectively identical
and embody the concept of the formation of a coherent interlocking solid
network that is separated by essentially continuous thin liquid films and thus
ruptured by the tensile stresses (2).The fracture surface often reveals the dendritic
morphology of the solidifying weld metal, as shown in Figure 11.3 for
308 stainless steel fractured under augmented strain during GTAW (8). If a
sufficient amount of liquid metal is present near the cracks, it can “backfill”
and “heal” the incipient cracks.
The terminal stage of solidification mentioned above refers to a fraction of
solid, fS, close to 1, and not necessarily a temperature near the lower limit of
the solidification temperature range. Depending on how it varies with temperature
in an alloy, fS can be close to 1 and an essentially coherent inter-
263
Welding Metallurgy, Second Edition. Sindo Kou
Copyright �� 2003 John Wiley & Sons, Inc.
ISBN: 0-471-43491-4
264 WELD METAL SOLIDIFICATION CRACKING
Figure 11.1 Solidification cracking in a gas–metal arc weld of 6061 aluminum.
Figure 11.2 Solidification cracking in an autogenous bead-on-plate weld of 7075
aluminum (magnification 140¥). From Kou and Kanevsky (1).
locking solid network can form even though the temperature is only slightly
below the liquidus temperature. In fact, Matsuda et al. (9, 10) have reported
that solidification cracking occurs in some carbon and stainless steels at temperatures
slightly below their liquidus temperatures.
11.1.2 Susceptibility Testing
A. Houldcroft Test This test, also called the fishbone test, is shown in
Figure 11.4 (11, 12). It is often used for evaluating the solidification cracking
CHARACTERISTICS, CAUSE, AND TESTING 265
Figure 11.3 SEM fracture surface of a 308 stainless steel weld fractured under
augmented strain during GTAW. Reprinted from Li and Messeler (8). Courtesy of
American Welding Society.
Figure 11.4 Houldcroft test: (a) design (11); (b) cracking in an aluminum alloy
specimen. Reprinted from Liptax and Baysinger (12). Courtesy of American Welding
Society.
susceptibility of sheet gage materials. The crack length from the starting
edge of the test specimen is used to indicate the susceptibility to cracking
(11–14).
The specimen is free from constraints and a progression of slots of varying
depth allows the dissipation of stresses within it. In such a test, solidification
cracking initiates from the starting edge of the test specimen and
propagates along its centerline. As the heat source moves inward from the
starting edge of the test specimen, solidification begins and the solidifying
structure is torn apart because the starting edge continues to expand as a result
of continued heat input to the specimen. The reason for using the slots is
explained with the help of specimens cracking under tension, as shown
in Figure 11.5. The crack may run all the way through the entire length of
the specimen (Figure 11.5a). Reducing the width of the specimen can reduce
the amount of stress along the length and bring the crack to a stop (Figure
11.5b). In order not to dramatically change the heat flow condition along
the length of the weld, however, the material next to the slots is not cut off
(Figure 11.5c).
B. Varestraint Test This test was developed by Savage and Lundin (15). As
shown in Figure 11.6a, an augmented strain is applied to the test specimen
(usually 12.7 mm thick) by bending it to a controlled radius at an appropriate
moment during welding (15). Both the amount of the applied strain and the
crack length (either the total length of all cracks or the maximum crack length)
serve as an index of cracking sensitivity.The specimen can also be bent transverse
to the welding direction, that is, the transverse Varestraint test, as shown
in Figure 11.6b (16). This may promote cracking inside the weld metal more
than that outside. Figure 11.6c shows the cracking pattern observed in a Varestraint
test specimen of a 444 Nb-stabilized ferritic stainless (17). Figure 11.7
compares the solidification cracking susceptibility of several materials by Varestraint
testing (18).
Other methods have also been used for testing the solidification cracking
susceptibility of weld metals (19, 20), including the circular patch test (21–23)
and Sigamajig test (24).
266 WELD METAL SOLIDIFICATION CRACKING
(a) (b) (c)
Figure 11.5 Three types of specimens cracking under tension: (a) uniform width along
crack; (b) decreasing width along crack; (c) uniform width along crack but slotted.
Figure 11.6 Solidification cracking tests: (a) Varestraint (15); (b) transverse
Varestraint (16); (c) cracking in a ferritic stainless steel Varestraint specimen (17).
(c) Reprinted from Krysiak et al. (17), courtesy of ASM International.
HR-160
Maximum crack length, mm
0 1 2 3 4 5
0
0.5
1.0
1.5
2.0
2.5
3.0
IN718
310SS
304SS
Applied strain, %
Figure 11.7 Varestraint test results showing solidification cracking susceptibility of
several different materials. Reprinted from DuPont et al. (18). Courtesy of American
Welding Society.
(c)
11.2 METALLURGICAL FACTORS
Metallurgical factors that have been known to affect the solidification cracking
susceptibility of weld metals include (i) the solidification temperature
range, (ii) the amount and distribution of liquid at the terminal stage of solidification,
(iii) the primary solidification phase, (iv) the surface tension of the
grain boundary liquid, and (v) the grain structure. All these factors are directly
or indirectly affected by the weld metal composition. The first two factors are
affected by microsegregation during solidification. Microsegregation in turn
can be affected by the cooling rate during solidification. In fact, in austenitic
stainless steels the primary solidification phase can also be affected by the
cooling rate. These metallurgical factors will be discussed in what follows.
11.2.1 Solidification Temperature Range
Generally speaking, the wider the solidification (freezing) temperature range,
the larger the (S + L) region in the weld metal or the mushy zone and thus
the larger the area that is weak and susceptible to solidification cracking. The
solidification temperature range of an alloy increases as a result of either the
presence of undesirable impurities such as sulfur (S) and phosphorus (P) in
steels and nickel-base alloys or intentionally added alloying elements.
A. Effect of S and P Impurities such as S and P are known to cause severe
solidification cracking in carbon and low-alloy steels even at relatively low
concentrations. They have a strong tendency to segregate at grain boundaries
(25) and form low-melting-point compounds (FeS in the case of S), thus widening
the solidification temperature range. In addition, S and P can cause severe
solidification cracking in nickel-base alloys (26–28) and ferritic stainless steels
(29). In the case of austenitic stainless steels, their detrimental effect on solidification
cracking can be significantly affected by the primary solidification
phase, as will be discussed subsequently.
Figure 11.8 shows the effect of various elements, including S and P, on the
solidification temperature range of carbon and low-alloy steels (30).As shown,
S and P tend to widen the solidification temperature range of steels tremendously.
The wider the solidification temperature range, the more grain boundary
area in the weld metal remains liquid during welding and hence susceptible
to solidification cracking, as depicted in Figure 11.9.
B. Effect of Reactions during Terminal Solidification Reactions, especially
eutectic reactions, can often occur during the terminal stage of solidification
and extend the solidification temperature range, for instance, in aluminum
alloys and superalloys.
Dupont et al. (31–35) studied the effect of eutectic reactions on the
solidification temperature range and solidification cracking. The materials
studied include some superalloys and stainless steels containing niobium (Nb).
268 WELD METAL SOLIDIFICATION CRACKING
Figure 11.10 shows the weld metal microstructure of a Nb-bearing superalloy
(34). Two eutectic-type constituents are present in the microstructure, g-NbC
and g-Laves. The liquid solidifies as the primary g and two eutectic reactions
occur during the terminal stage of solidification, first L Æ g + NbC and then
L Æ g + Laves. The second eutectic reaction, when it occurs, takes place at a
considerably lower temperature than the first, thus extending the solidification
METALLURGICAL FACTORS 269
S
B
P
C
Si
0 5
0
100
200
300
Weight %
Freezing range, C
10
o
Figure 11.8 Effect of alloying elements on the solidification temperature range
of carbon and low-alloy steels. Modified from Principles and Technology of the Fusion
Welding of Metals (30).
Pool (L) Weld (S)
no grain boundary
(GB) liquid
GB
GB
(a)
(b)
(d)
(c)
pure Fe: zero freezing
temperature range;
not susceptible
Fe with silicon: narrow
freezing temperature
range; somewhat
susceptible
Fe with sulfur: wide
freezing temperature
range; highly susceptible
GB
GB
GB
GB
grain boundary
liquid
L
L
L
S
S
S
grain boundary
liquid
Figure 11.9 Effect of impurities on grain boundary liquid of weld metal: (a) weld
metal near pool; (b) no liquid in pure Fe; (c) some liquid with a small amount of Si;
(d) much more liquid with a small amount of sulfur.
temperature range significantly. For the alloy shown in Figure 11.10, for
instance, solidification starts at 1385.6°C (liquidus temperature), L Æ g + NbC
takes place at 1355.2°C and L Æ g + Laves at a considerably lower temperature,
1248.2°C (31).The solidification temperature range is 137.4°C. As shown
in Figure 11.11, the maximum crack length in Varestraint testing increases with
increasing solidification temperature range (17).
Before leaving the subject of eutectic reactions, two stainless steel welds
will be used here to illustrate the two eutectic reactions along the solidification
path (35).The first weld, containing 0.48 wt % Nb and 0.010wt % C, is an
270 WELD METAL SOLIDIFICATION CRACKING
Figure 11.10 Scanning electron micrographs showing weld metal microstructure of a
Nb-bearing superalloy. Reprinted from DuPont et al. (34).
Experimental superalloys
HR-160
IN718
Maximum crack length, mm
0.5
1.0
1.5
2.0
2.5
3.0
0
Solidification temperature range, C
0 50 100 150 200250 300
o
Figure 11.11 Maximum crack length as a function of solidification temperature range
for Nb-bearing superalloys and two other materials. Reprinted from DuPont et al. (18).
Courtesy of American Welding Society.
autogenous gas–tungsten arc weld of a Nb-stabilized austenitic stainless steel
20Cb-3. To increase the Nb content, a relatively large composite weld was
made in alloy 20Cb-3 with an INCO 112 filler wire containing 3.81wt % Nb
and then machined flat at the top surface for further welding.The second weld,
containing 2.20 wt % Nb and 0.014wt % C, is an autogenous gas–tungsten arc
weld within the 20Cb-3–INCO 112 composite weld.
The solidification paths of the two welds are shown in Figures 11.12. As
mentioned previously in Chapter 6, the arrows in the solidification path indicate
the directions in which the liquid composition changes as temperature
decreases. Here, solidification of the 20Cb-3 weld initiates by a primary L Æ
g reaction, as shown in Figure 11.12a. Because of the relatively high C–Nb ratio
of the alloy, the interdendritic liquid becomes enriched in C until the g/NbC
twofold saturation line is reached. Here, the L Æ g + NbC reaction takes over.
However, since the fraction of liquid is already very small (fL = 0.005), solidification
is soon over (fL = 0) at about 1300°C before the L Æ g + Laves reaction
takes place. In contrast, the lower C–Nb ratio of the 20Cb-3–INCO 112
composite weld caused the interdendritic liquid to become more highly
enriched in Nb as shown in Figure 11.12b.The solidification path barely intersects
the g/NbC twofold saturation line (fL = 0.01) before it reaches the g/Laves
twofold saturation line (fL = 0.008). The L Æ g + NbC reaction takes only
briefly, and the L Æ g + Laves reaction takes over until solidification is complete
(fL = 0) at about 1223°C, which is significantly lower than 1300°C.
11.2.2 Amount and Distribution of Liquid during Terminal Solidification
A. Amount of Liquid Figures 11.13a–d show the effect of composition on
the solidification cracking sensitivity of several aluminum alloys (36–41).
Figure 11.13e shows the crack sensitivity in pulsed laser welding of Al–Cu
METALLURGICAL FACTORS 271
(b)
NbC
fL=0.005
0 5 10 15 20 25 30
0.25
0.20
0.15
0.10
0.05
0
Liquid composition, wt% C
Liquid composition, wt% Nb
20Cb-3
fL=0
fL=1 Laves
NbC
fL=0.01
0 5 10 15 20 25 30
0.25
0.20
0.15
0.10
0.05
0
20Cb-3/
INCO 112
fL=0 Laves
fL=1
fL=0.008
(a)
γ
γ
Figure 11.12 Solidification paths (solid lines) of Nb-stabilized austenitic stainless
steels: (a) 20Cb-3; (b) 20Cb-3–INCO 112 composite. Reprinted from DuPont (35).
Courtesy of American Welding Society.
alloys (41). Figure 11.14a shows an aluminum weld with little Cu (alloy 1100
gas–metal arc welded with filler 1100), and there is no evidence of cracking.
Figure 11.14b shows a crack in an aluminum weld with about 4% Cu (alloy
2219 gas–metal arc welded with filler 1100). Figure 11.14c shows a crack healed
by the eutectic liquid in an aluminum weld with about 8% Cu (alloy 2219
gas–metal arc welded with filler 2319 plus extra Cu).
272 WELD METAL SOLIDIFICATION CRACKING
0
0
1 2 3 4 5 6 7 8
0
0
0
Composition of weld, % alloying element
(a) Al-Si (Singer
et al. 1947)
(b) Al-Cu
(Pumphrey et
al. 1948
(c) Al-Mg
(Dowd, 1952)
(d) Al-Mg2Si
(Jennings et al. 1948)
Relative crack sensitivity
0 1 2 3 4 5 6 7 8
Copper content (wt%)
0
10
20
30
Total crack length (mm)
(e) Al-Cu
(Michaud et al. 1995)
)
Figure 11.13 Effect of composition on crack sensitivity of some aluminum alloys.
(a–d) From Dudas and Collins (40). (e) Reprinted from Michaud et al. (41).
Figure 11.14 Aluminum welds with three different levels of Cu: (a) almost no Cu; (b)
4% Cu; (c) 8% Cu.
As shown in Figure 11.13, the maximum crack sensitivity occurs somewhere
between pure aluminum and highly alloyed aluminum (say no less than 6wt
% solute).The presence of a maximum in the crack susceptibility composition
curve (Figure 11.13) is explained qualitatively with the help of Figure 11.15.
Pure aluminum is not susceptible to solidification cracking because there is no
low-melting-point eutectic present at the grain boundary to cause solidification
cracking. In highly alloyed aluminum, on the other hand, the eutectic
liquid between grains can be abundant enough to “heal” incipient cracks (3).
Somewhere in between these two composition levels, however, the amount of
liquid between grains can be just large enough to form a thin, continuous grain
boundary film to make the materials rather susceptible to solidification cracking
but without extra liquid for healing cracks. A fine equiaxed dendritic structure
with abundant liquid between grains (Figure 11.15f) can deform more
easily under stresses than a coarse columnar dendritic structure (Figure
11.15e) and thus has a lower susceptibility to cracking.
B. Calculation of Fraction Liquid Dupont et al. (31–35) calculated the
fraction of the liquid (fL) as a function of distance (x) within the mushy zone
in Nb-bearing superalloys and austenitic stainless steels. The fL–x curve provides
the quantitative information for the amount and distribution of interdendritic
liquid in the mushy zone. The two stainless steel welds discussed
previously will be considered here, namely, the weld in alloy 20Cb-3 and the
METALLURGICAL FACTORS 273
I. pure metal: no grain
boundary (GB) liquid; not
susceptible to cracking
GB
(b)
(e)
(d)
L L
S
(f)
II: some solute: just enough
liquid to form a continuous
GB film; most susceptible
Pool (L) Weld (S)
(a)
S
III. more solute: more GB
liquid to heal cracks; less
susceptible to cracking
IV. much more solute: much
GB liquid to heal cracks and
less rigid dendritic structure;
least susceptible to cracking
L L
I
II
III
IV
crack
susceptibility
solute content
(c)
Figure 11.15 Effect of composition on crack susceptibility: (a) weld; (b) crack susceptibility
curve; (c) pure metal; (d) low solute; (e) more solute; ( f) much more solute.
weld in the 20Cb-3–INCO 112 composite weld. According to DuPont et al.
(35), Nb, Si, and C are treated as solutes in these alloys and, as an approximation,
the remaining elements in the solid solution with g are treated as the
“g-solvent.”
Assume that the slopes of the liquidus surface with respect to Nb, Si, and
C, that is, mL,Nb, mL,Si, and mL,C, respectively, are constant. Equation (6.2) can
be extended to find the temperature of a liquid on the liquidus surface T as
follows:
T = Tm + mL,NbCL,Nb + mL,SiCL,Si + mL,CCL,C (11.1)
Assuming negligible solid diffusion but complete liquid diffusion for Nb
and Si, the following equations can be written based on the Scheil equation
[Equation (6.9)]:
(11.2)
(11.3)
Since diffusion in solid and liquid is fast for a small interstitial solute such as
C, the following equation can be written based on the equilibrium lever rule
[Equation (6.6)]:
(11.4)
Note that the equilibrium partition ratios kNb, kSi, ans kC have been assumed
constant in Equations (11.2)–(11.4) for simplicity. Inserting these equations
into Equation (11.1) yields the following relationship between temperature
and fraction liquid:
(11.5)
The cooling rate (e or GR) can be determined from the secondary dendrite
spacing (d). According to Equation (6.20),
(11.6)
For instance, from the dendrite arm spacing of the weld metal the cooling rate
GR is about 250°C/s. The growth rate at the weld centerline R is the welding
speed 3mm/s. As such, the temperature gradient G is about 83°C/mm.
Assuming that G is constant in the mushy zone and taking x = 0 at the
liquidus temperature of the alloy TL,
d at b bGR f
n n n = = ( ) = ( ) - - e
T T m C f m C f m
C
f k f
= + k + k +
+ (- )
ÊË
ˆ¯
( -) ( -)
m L,Nb 0,Nb L L,Si 0,Si L L,C
C
L C L
Nb1 Si1 0
1
,
C
C
f k f L,C
0,C
L C L
=
+ (1- )
C C fk
L,Si 0,Si L
= (Si -1)
C C fk
L,Nb 0,Nb L
= (Nb-1)
274 WELD METAL SOLIDIFICATION CRACKING
(11.7)
which can be used to find the temperature T at any distance x.
Equations (11.5) and (11.7) can be combined to determine how the liquid
fraction fL varies with distance x within the mushy zone, and the results are
shown in Figure 11.16.The liquid fraction drops rapidly with distance near the
pool boundary but slowly further into the mushy zone. Also, the mushy zone
is significantly wider in the 20Cb-3–INCO 112 composite weld than in the
20Cb-3 weld due to the significantly larger solidification temperature range of
the former. This explains why the former is more susceptible to solidification
cracking.
C. Liquid Distribution As shown previously in Figure 11.11 for Nb-bearing
superalloys, the alloys with a narrower solidification temperature range are
less susceptible to solidification cracking. In fact, some of these alloys have a
wide solidification temperature range just like the more susceptible alloys. As
in the more crack-susceptible alloys, the L Æ g + Laves reaction follows the L
Æ g + NbC reaction during the terminal stage of solidification. However, as
shown in Figure 11.17 for one of these less susceptible alloys, the amount of
terminal liquid undergoing the L Æ g + Laves reaction in these alloys is small
and remains isolated. This type of morphology, unlike the continuous grain
boundary liquid undergoing the L Æ g + Laves reaction in the more cracksusceptible
alloys, should be more resistant to crack propagation throughout
the mushy zone. Since an isolated L Æ g + Laves reaction does not really
contribute to solidification cracking, it should not have to be included in the
solidification temperature range, and the lower bound of the effective solidification
temperature range should more accurately be represented by the L
Æ g + NbC reaction (31, 34).
x
T T
G
=
L -
METALLURGICAL FACTORS 275
0 1.0
0
Fraction liquid
Distance along centerline, mm
20Cb-3/INCO 112
(a)
20Cb-3
1.0
0.8
0.6
0.4
0.2
2.0 3.0 4.0 0 1.0
0
20Cb-3/INCO 112
(b)
20Cb-3
0.10
0.08
0.06
0.04
0.02
2.0 3.0 4.0
L
L γ
γ
Figure 11.16 Fraction liquid as a function of distance within the mushy zone of
20Cb-3 and 20Cb-3–INCO 112 composite (a) and enlarged (b). Reprinted from
DuPont (35). Courtesy of American Welding Society.
11.2.3 Ductility of Solidifying Weld Metal
The less ductile a solidifying weld metal is, the more likely it will crack during
solidification. Nakata and Matsuda (16) used the transverse Varestraint test
(Figure 11.6b) to determine the so-called ductility curve, as illustrated in Figure
11.18. At any given strain the ductility curve ranges from the liquidus temperature
TL to the temperature at the tip of the longest crack.To construct the
curve, a strain of e1 is applied during welding, and the maximum crack length
is examined after welding (Figure 11.18a). From the temperature distribution
276 WELD METAL SOLIDIFICATION CRACKING
Figure 11.17 Scanning electron micrographs showing morphology of g-NbC and g-
Laves constituents in solidification cracks of a Nb-bearing superalloy. Reprinted from
DuPont et al. (34).
METALLURGICAL FACTORS 277
T 1
Strain,
min
T L
brittle temperature
range (BTR)
slope = critical strain rate for
temperature drop (CST)
Distance
Temperature
tip of longest
crack induced by
applying strain
weld
pool
T L
(a)
(b)
direction
100
50
0
1 2
Relative cracking ratio, %
CST (x 10-oC)
0
A7N01
A2017 A2219
A5052 A1100(3.75)
A5083
GTAW crater
weld test (d)
region of low ductility
region of low
ductility
TL2
Strain,
min1
TL1
(c)
BTR1
BTR CST
min2 CST
CST < CST : alloy 2 moresusceptible to cracking than alloy 1Temperatureductility curvewelding-4/oTT2 CST1CST22 1Temperatureε1ε1εεεε εFigure 11.18 Ductility of solidifying weld metal: (a) temperature distribution (temperatureincreases from right to left); (b) ductility curve; (c) ductility curves with differentcrack susceptibility; (d) effect of CST on cracking. (d) From Nakata and Matsuda(16).along the weld centerline measured with a thermocouple during welding, thetemperature T1 at the tip of the longest crack and hence the point (T1, e1) canbe determined (Figure 11.18b). By repeating the experiment for variousapplied strains and finding the corresponding crack tip temperatures, the ductilitycurve can be determined.The maximum crack length and hence the temperaturerange of the ductility curve first increase with increasing appliedstrain but then level off as the applied strain increases further.The widest temperaturerange covered by the ductility curve is called the brittle temperaturerange (BTR).The weld metal is “brittle” in the sense that it is much less ductilein this temperature range than either the weld pool or the completely solidifiedweld metal. The minimum strain required to cause cracking is called emin.The slope of the tangent to the ductility curve is called the critical strain ratefor temperature drop (CST), that is, the critical rate at which the strain varieswith temperature drop.In general, the lower emin, the greater BTR or the smaller CST is, the greaterthe susceptibility to solidification cracking (Figure 11.18c). According toNakata and Matsuda (16), CST best correlates with the cracking susceptibilityof the weld metal (Figure 11.18d). It should be pointed out, however, thatthe values of CST in Figure 11.18d were not determined exactly as describedabove. Because the emin values were too small to be determined, a slow-bendingtransverse Varestraint test had to be used to modify the ductility curve.For a given material cracking can be avoided if the strain rate for temperaturedrop can be kept below the critical value, that is, if -de/dT < CTS. Mathematically,-de/dT = (∂e/∂t)/(-∂T/∂t), where t is time. The strain rate (∂e/∂t)consists of both the self-induced tensile strain in the solidifying weld metalwhose contraction is hindered by the adjacent base metal and the augmentedstrain if it is applied. -∂T/∂t is the cooling rate, for instance, 150°C/s. InVarestraint testing an augmented strain is applied essentially instantaneously,and the high ∂e/∂t and hence -de/dT easily cause cracking. If the augmentedstrain were applied very slowly on a solidifying weld metal whose self-inducedstrain rate during solidification is low, ∂e/∂t can be small enough to keep -de/dT below the CST, and the weld metal can solidify without cracking. Crackingcan also be avoided if the cooling rate -∂T/∂t is increased dramatically toreduce -de/dT to below the CST.Yang et al. (42) avoided solidification crackingin GTAW of 2024 aluminum sheets by directing liquid nitrogen behind theweld pool to increase the cooling rate. They also showed, with finite-elementmodeling of heat flow and thermal stresses, that the condition -de/dT < CTScan be achieved by cooling the area right behind the weld pool.Before leaving the subject of the ductility curve, the relationship betweenthe BTR and the solidification temperature range needs to be discussed.Nakata and Matsuda (16) observed in aluminum alloys such as alloys 2017,2024, and 2219 that the cooling curve showed a clear solidification temperaturerange. A distinct point of arrest (a short flat region) in the curve correspondingto the formation of eutectics was observed as well as a discontinuityat the liquidus temperature corresponding to the formation of the aluminum-278 WELD METAL SOLIDIFICATION CRACKINGrich solid. For these alloys the BTR was found to be the same as the solidificationtemperature range, that is, TL to TE. In other words, the region of lowductility behind the weld pool corresponds to the mushy zone discussed in previouschapters. On the other hand, for aluminum alloys such as alloys 5052,5083, and 6061, the cooling curve did not show a distinct point of arrest, andthe inflection point in the curve had to be taken as the “nominal” solidus temperatureTS. For these alloys the BTR was found significantly (about 20%)larger than the nominal solidification temperature range of TL to TS and wasthus a more realistic representation of the nonequilibrium solidification temperaturerange of TL to TE. Presumably, eutectics still formed below TS but notin quantities enough to show a distinct point of arrest in the cooling curve. Itis not clear if differential thermal analysis was tried, which is known to detecteutectic temperatures well.In the case of steels and stainless steels, the solidification temperature rangeis relatively narrow, and impurities such as P and S are found to enlarge theBTR, presumably by forming eutectics with a low melting point (16).11.2.4 Primary Solidification PhaseFor austenitic stainless steels the susceptibility to solidification cracking ismuch lower when the primary solidification phase is d-ferrite rather thanaustenite (43–45). As the ratio of the Cr equivalent to the Ni equivalentincreases, the primary solidification phase changes from austenite to d-ferrite,and cracking is reduced. As shown by Takalo et al. (44) in Figure 11.19a forarc welding, this change occurs at Creq/Nieq = 1.5. Similarly, as shown by Lienert(45) in Figure 11.19b for pulsed laser welding, this change occurs in theCreq/Nieq range of 1.6–1.7. In both cases Creq = Cr + 1.37Mo + 1.5Si + 2Nb +3Ti and Nieq = Ni + 0.3Mn + 22C + 14.2N + Cu. As discussed previously(Chapter 9), under high cooling rates in laser or electron beam welding, a weldmetal that normally solidifies as primary ferrite can solidify as primary austenitebecause of undercooling, and this is consistent with the change occurringat a higher Creq–Nieq ratio in pulsed laser welding.In general, austenitic stainless steels containing 5–10% d-ferrite are significantlymore resistant to solidification cracking than fully austenitic stainlesssteels (46). As mentioned previously in Chapter 9, ferrite contents significantlygreater than 10% are usually not recommended, for the corrosion resistancewill be too low (47–50). Furthermore, upon exposure to elevated temperatures(600–850°C), d-ferrite can transform to brittle s-ferrite and impair the mechanicalproperties of austenitic stainless steels, unless the ferrite contentis kept low (51).It is generally believed that, since harmful impurities such as sulfur andphosphorus are more soluble in d-ferrite than in austenite (see Table 11.1), theconcentration of these impurities at the austenite grain boundaries, and thustheir damaging effect on solidification cracking, can be reduced if d-ferrite ispresent in significant amounts (43, 52, 53). In addition, it is also believed thatMETALLURGICAL FACTORS 279280 WELD METAL SOLIDIFICATION CRACKINGcracking no crackingCreq/NieqTotal S+P+B (wt %)1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.00.000.010.020.030.040.050.060.070.08pulsedNd:YAGlaser welds(b)1.0 1.2 1.4 1.6 1.8 2.00.040.080.120.160.200Creq/Nieqsusceptiblesomewhatsusceptiblenot susceptiblecrackingno crackingTotal S+P (wt %)(a)primaryaustenitemixedprimaryferriteprimary austenite primary ferriteFigure 11.19 Solidification crack susceptibility of austenitic stainless steels:(a) arc welds; (b) laser welds. (a) From Takalo et al. (44). (b) Reprinted fromLienert (45).TABLE 11.1 Solubility of Sulfur and Phosphorus inFerrite and Austenite (wt%)In d-Ferrite In AusteniteSulfure 0.18 0.05Phosphorus 2.8 0.25Source: Borland and Younger (52).when d-ferrite is the primary solidification phase, the substantial boundaryarea between d-ferrite and austenite acts as a sink for sulfur and phosphorus.This decreases the concentration of such impurities at the austenite grainboundaries and, therefore, reduces solidification cracking (43, 54–57). Hull(46) suggests that the propagation of solidification cracks in cast austeniticstainless steels is halted by d-ferrite due to the fact that the ferrite–austeniteinterface energy is lower than the austenite–austenite interface energy (58,59). Solidification cracks stopped by d-ferrite have also been observed in 309stainless steel welds by Brooks et al. (60).In addition to austenitic stainless steels, the primary solidification phase canalso affect the solidification cracking susceptibility of carbon steels. Accordingto the iron–carbon phase diagram shown in Figure 11.20, when the carboncontent is greater than 0.53, austenite becomes the primary solidification phaseand solidification cracking becomes more likely. In fact, the wider solidificationtemperature range at a higher carbon content further increases the potentialfor solidification cracking.11.2.5 Surface Tension of Grain Boundary LiquidThe effect of the amount and distribution of the grain boundary liquid on thesolidification cracking of weld metals has been described earlier in this section.The higher the surface tension of the grain boundary liquid, the larger its dihedralangle is. Figures 11.21a and b show the dihedral angle and distribution ofthe grain boundary liquid (61). Figure 11.21c shows the effect of the dihedralangle on the susceptibility to solidification cracking in several aluminum alloysevaluated in autogenous spot GTAW (16). As shown, except for alloy 1100,which is essentially pure aluminum and thus not susceptible to cracking, theMETALLURGICAL FACTORS 281Figure 11.20 Iron–carbon phase diagram.susceptibility decreases with increasing dihedral angle of the grain boundaryliquid.The effect of the surface tension of the grain boundary liquid on solidificationcracking is further depicted in Figure 11.22. If the surface tension betweenthe solid grains and the grain boundary liquid is very low, a liquid film will282 WELD METAL SOLIDIFICATION CRACKING10050020 40 60 80Relative cracking ratio, %Dihedral angle , degree0A7N01A2017A2219A5052 A5083A1100GTAW craterweld test= 120 o(a)(b)= 60 o = 0 oGB liquid(c)grainboundaryGB (GB)liquidθ θ θθθFigure 11.21 Grain boundary liquid: (a) dihedral angle; (b) distribution of liquid atgrain boundary; (c) effect of dihedral angle on solidification cracking. (c) From Nakataand Matsuda (16).(b)GBGBL Sgrain boundaryliquid(c)GBGBL Sgrain boundaryliquidcontinuous grainboundary liquid;highly susceptibleisolated grainboundary liquid;less susceptiblePool (L) Weld (S) (a)Figure 11.22 Effect of grain boundary liquid morphology on crack susceptibility: (a)weld; (b) continuous; (c) isolated.form between the grains and the solidification cracking susceptibility is high.On the other hand, if the surface tension is high, the liquid phase will be globularand will not wet the grain boundaries. Such discontinuous liquid globulesdo not significantly reduce the strength of the solid network and, therefore,are not as harmful. For example, FeS forms films at the grain boundaries ofsteels while MnS forms globules. Due to its globular morphology and highermelting point, MnS has been known to be far less harmful than FeS.11.2.6 Grain Structure of Weld MetalFine equiaxed grains are often less susceptible to solidification cracking thancoarse columnar grains (16, 62–64), as shown in Figure 11.23a (16) for severalMETALLURGICAL FACTORS 283Figure 11.23 Effect of grain structure on solidification cracking: (a) aluminum alloys;(b) centerline cracking in a coarse-grain 310 stainless steel weld. (a) From Nakata andMatsuda (16). (b) From Kou and Le (65).(a)(b)aluminum alloys. Alloy A1070 is not susceptible to solidification crackingbecause it is essentially pure aluminum. Fine equiaxed grains can deform toaccommodate contraction strains more easily, that is, it is more ductile, thancolumnar grains. Liquid feeding and healing of incipient cracks can also bemore effective in fine-grained material. In addition, the grain boundary areais much greater in fine-grained material and, therefore, harmful low-meltingpointsegregates are less concentrated at the grain boundary.It is interesting to note that, due to the steep angle of abutment betweencolumnar grains growing from opposite sides of the weld pool, welds madewith a teardrop-shaped weld pool tend to be more susceptible to centerlinesolidification cracking than welds made with an elliptical-shaped weld pool.Asteep angle seems to favor the head-on impingement of columnar grainsgrowing from opposite sides of the weld pool and the formation of the continuousliquid film of low-melting-point segregates at the weld centerline. Asa result, centerline solidification cracking occurs under the influence of transversecontraction stresses. Centerline cracking is often observed in welding.Figure 11.23b is an example in an autogenous gas–tungsten arc weld of a 310stainless steel made with a teardrop-shaped weld pool (65).11.3 MECHANICAL FACTORS11.3.1 Contraction StressesSo far, the metallurgical factors of weld solidification cracking have beendescribed. But without the presence of stresses acting on adjacent grainsduring solidification, no cracking can occur. Such stresses, as already mentioned,can be due to thermal contraction or solidification shrinkage or both.Austenitic stainless steels have relatively high thermal expansion coefficients(as compared with mild steels) and, therefore, are often prone to solidificationcracking.Aluminum alloys have both high thermal expansion coefficients and highsolidification shrinkage (5). As a result, solidification cracking can be ratherserious in some aluminum alloys, especially those with wide solidification temperatureranges.11.3.2 Degree of RestraintThe degree of restraint of the workpiece is another mechanical factor ofsolidification cracking. For the same joint design and material, the greaterthe restraint of the workpiece, the more likely solidification cracking willoccur.Figure 11.24 illustrates the effect of workpiece restraint on solidificationcracking (66). As shown, solidification cracking occurred in the second(left-hand-side) weld of the inverse “T” joint due to the fact that the degree284 WELD METAL SOLIDIFICATION CRACKINGof restraint increased significantly after the first (right-hand-side) weld wasmade.11.4 REDUCING SOLIDIFICATION CRACKING11.4.1 Control of Weld Metal CompositionWeld metals of crack-susceptible compositions should be avoided. In autogenouswelding no filler metal is used, and the weld metal composition is determinedby the base-metal composition. To avoid or reduce solidificationcracking, base metals of susceptible compositions should be avoided.When abase metal of a crack-susceptible composition has to be welded, however, afiller metal of a proper composition can be selected to adjust the weld metalcomposition to a less susceptible level.A. Aluminum Alloys According to Figure 11.13, an Al–3% Cu alloy can berather crack susceptible during welding. If the Cu content of the base metal israised to above 6%, solidification cracking can be significantly reduced. In fact,2219 aluminum, one of the most weldable Al–Cu alloys, contains 6.3% Cu.When a filler metal is used, the weld metal composition is determined bythe composition of the base metal, the composition of the filler metal, and thedilution ratio. The dilution ratio, as mentioned previously, is the ratio of theamount of the base metal melted to the total amount of the weld metal.Againusing Al–3% Cu as an example, the weld metal Cu content can be increasedby using 2319 filler metal, which is essentially an Al–6.3% Cu alloy. If the jointdesign and heat input are such that the dilution ratio is low, the weld metalcopper content can be kept sufficiently high to avoid solidification cracking.Figure 11.25 shows the approximate dilution in three typical joint designs (40).REDUCING SOLIDIFICATION CRACKING 285Figure 11.24 Solidification cracking in steel weld. Reprinted from Linnert (66). Courtesyof American Welding Society.Table 11.2 is a guide to choice of filler metals for minimizing solidificationcracking in welds of high-strength aluminum alloys (40). Experimental datasuch as those in Figure 11.26 from solidification cracking testing by ring castingcan also be useful (67, 68).286 WELD METAL SOLIDIFICATION CRACKINGFigure 11.25 Approximate dilution of three weld joints by base metal in aluminumwelding. Reprinted from Dudas and Collins (40). Courtesy of American WeldingSociety.TABLE 11.2 Guid to Choice of Filler Metals for Minimizing Solidification Cracking inWelds of High-Strength Aluminum Alloys7000 7000 6000 5000 2000Base Metals (Al–Zn–Mg–Cu) (Al–Zn–Mg) (Al–Mg–Si) (Al–Mg) (Al–Cu)2000 (Al–Cu) NRa NR NR NR 4043414523195000 (Al–Mg) 5356 5356 5356 5356 —b5556 5556 55565183 5183 51836000 (Al–Mg–Si) 5356 5356 4043 — —5556 46435183 53567000 (Al–Zn–Mg) 5356 5356 — — —55567000 (Al–Zn–Mg–Cu) 5356 — — — —5556a NR, not recommend.b Charts that recommend filler choice for many applications are available from filler metal suppliers.Source: Dudas and Collins (40).REDUCING SOLIDIFICATION CRACKING 287Figure 11.26 Solidification cracking susceptibility of aluminum alloys: (a) Al–Mg–Si;(b) Al–Cu–Si (c) Al–Mg–Cu. (a, b) Modified from Jennings et al. (67). (c) Modifiedfrom Pumphrey and Moore (68).Minor alloying elements have also been found to affect the solidificationcracking susceptibility of aluminum alloys. For example, the Fe–Si ratio hasbeen found to significantly affect the solidification cracking susceptibility of3004 and other Al–Mg alloys (69, 70); therefore, proper control of the contentof minor alloying elements can be important in some materials.B. Carbon and Low-Alloy Steels The weld metal manganese content cansignificantly affect solidification cracking. It is often kept high enough toensure the formation of MnS rather than FeS. This, as described previously, isbecause the high melting point and the globular morphology of MnS tend torender sulfur less detrimental. Figure 11.27 shows the effect of the Mn–S ratioon the solidification cracking tendency of carbon steels (71). At relatively lowcarbon levels the solidification cracking tendency can be reduced by increasingthe Mn–S ratio. However, at higher carbon levels (i.e., 0.2–0.3% C),increasing the Mn–S ratio is no longer effective (72). In such cases loweringthe weld metal carbon content, if permissible, is more effective.One way of lowering the weld metal carbon content is to use low-carbonelectrodes. In fact, in welding high-carbon steels one is often required to makethe first bead (i.e., the root bead) with a low-carbon electrode.This is because,as shown in Figure 11.28, the first bead tends to have a higher dilution ratioand a higher carbon content than subsequent beads (73). A high carboncontent is undesirable because it promotes not only the solidification crackingof the weld metal but also the formation of brittle martensite and, hence,the postsolidification cracking of the weld metal. Therefore, in welding steelsof very high carbon contents (e.g., greater than 1.0% C), extra steps should betaken to avoid introducing excessive amounts of carbon from the base metalinto the weld metal. As shown in Figure 11.29, one way to achieve this is to288 WELD METAL SOLIDIFICATION CRACKINGCarbon content, %000.10 0.12 0.14 0.161020504030Ratio of manganese to sulfurIntermediate ZoneNo CrackingCrackingFigure 11.27 Effect of Mn–S ratio and carbon content on solidification cracking susceptibilityof carbon steel weld metal. Reprinted from Smith (71).“butter” the groove faces of the base metal with austenitic stainless steel (suchas 25–20 stainless) electrodes before welding (73). In welding cast irons, purenickel electrodes have also been used for buttering. In any case, the surfacelayers remain in the ductile austenitic state, and the weld can then be completedeither with stainless electrodes or with other cheaper electrodes.C. Nb-Bearing Stainless Steels and Superalloys The C–Nb ratio of the weldmetal can affect the susceptibility to solidification cracking significantly (17,31–35). A high C–Nb ratio can reduce solidification cracking by avoiding thelow-temperature L Æ g + Laves reaction, which can widen the solidificationtemperature range.D. Austenitic Stainless Steels As mentioned previously, it is desirable tomaintain the weld ferrite content at a level of 5–10% in order to obtain soundwelds. The quantitative relationship between the weld ferrite content and theweld metal composition in austenitic stainless steels has been determined bySchaeffler (74), DeLong (75), Kotecki (76, 77), Balmforth and Lippold (78),and Vitek et al. (79, 80). These constitution diagrams have been shown previouslyin Chapter 9. Alloying elements are grouped into ferrite formers (Cr,Mo, Si, and Cb) and austenite formers (Ni, C, and Mn) in order to determinethe corresponding chromium and nickel equivalents for a given alloy.Example: Consider the welding of a 1010 steel to a 304 stainless steel. Forconvenience, let us assume the dilution ratio is 30%, half from 304 stainlesssteel and half from 1010 steel, as shown in Figure 11.30. Estimate the ferritecontent and the solidification cracking susceptibility of a weld made with atype 310 electrode that contains 0.12% carbon and a weld made with a type309 ELC (extra low carbon) electrode that contains 0.03% carbon.For the weld made with the 310 stainless steel electrode, the weld metalcomposition can be calculated as follows:REDUCING SOLIDIFICATION CRACKING 289Figure 11.28 Schematic sketch of multipass welding. Note that the root pass has thehighest dilution ratio. From Jefferson and Woods (73).Figure 11.29 Buttering the groove faces of very high carbon steel with a 310 stainlesswire steel before welding. From Jefferson and Woods (73).Element Electrode ¥70% Type 304 ¥15% 1010 Steel ¥15% Weld MetalCr 26.0 18.2 18.0 2.7 0 0 20.9Ni 21.0 14.7 8.0 1.2 0 0 15.9C 0.12 0.084 0.05 0.0075 0.10 0.015 0.1065Mn 1.75 1.23 2.0 0.30 0.4 0.06 1.59Si 0.4 0.28 — — 0.2 0.03 0.31According to the WRC-1992 diagram shown in Figure 9.11, the chromium andnickel equivalents of the weld metal are as follows:Chromium equivalent = 20.9Nickel equivalent = 15.9 + 35 ¥ 0.1065 = 19.6Based on the diagram, the weld metal is fully austenitic and, therefore, is susceptibleto solidification cracking. If the 309 ELC electrode is used, the weldmetal composition can be calculated as follows:Element Electrode ¥70% Type 304 ¥15% 1010 Steel ¥15% Weld MetalCr 24 16.8 18.0 2.7 0 0 19.5Ni 13 9.1 8.0 1.2 0 0 10.3C 0.03 0.021 0.05 0.0075 0.10 0.015 0.0435Mn 1.98 1.39 2.0 0.30 0.4 0.06 1.75Si 0.4 0.28 0 0 0.2 0.03 0.31Therefore, the chromium and nickel equivalents of the weld metal areChromium equivalent = 19.5Nickel equivalent = 10.3 + 35 ¥ 0.0435 = 11.8According to the WRC-1992 diagram, the weld metal now is austenitic with aferrite number of about 8 and, therefore, should be much more resistant tosolidification cracking.290 WELD METAL SOLIDIFICATION CRACKINGFigure 11.30 Welding 304 stainless steel to 1010 carbon steel.It should be emphasized that neither the constitution diagrams nor themagnetic measurements of the weld metal ferrite content (such as thosedetermined by Magne–Gage readings) reveal anything about the weld metalsolidification. In fact, the primary solidification phase and the quantity of dferriteat high temperatures (i.e., during solidification) are more importantthan the amount of ferrite retained in the room temperature microstructurein determining the sensitivity to solidification cracking (81). Also, the amountof harmful impurities such as sulfur and phosphorus should be consideredin determining the weldability of a material; a material with a higher ferritecontent can be more susceptible to solidification cracking than another materialwith a lower ferrite content if the impurity level of the former is alsohigher. The cooling rate during solidification is another factor that the constitutiondiagrams fail to recognize (Chapter 9).11.4.2 Control of Solidification StructureA. Grain Refining As mentioned previously, welds with coarse columnargrains are often more susceptible to solidification cracking than those with fineequiaxed grains. It is, therefore, desirable to grain refine the weld metal. Infact, both 2219 aluminum and 2319 filler metal are designed in such a way thatthey not only have a non-crack-sensitive copper content but also have smallamounts of grain refining agents such as Ti and Zr to minimize solidificationcracking. Dudas and Collins (40) produced grain refining and eliminated solidificationcracking in a weld made with an Al–Zn–Mg filler metal by addingsmall amounts of Zr to the filler metal. Garland (82) has grain refined weldsof aluminum–magnesium alloys by vibrating the arc during welding, therebyreducing solidification cracking.B. Magnetic Arc Oscillation Magnetic arc oscillation has been reportedto reduce solidification cracking in aluminum alloys, HY-80 steel, and iridiumalloys (83–88). Kou and Le (86–88) studied the effect of magnetic arc oscillationon the grain structure and solidification cracking of aluminum alloy welds.As already shown in Figure 7.30, transverse arc oscillation at low frequenciescan produce alternating columnar grains. Figure 11.31 demonstrates that thistype of grain structure can be effective in reducing solidification cracking (86).As illustrated in Figure 11.32, columnar grains that reverse their orientationat regular intervals force the crack to change its direction periodically, thusmaking crack propagation difficult (87). Figure 11.33 shows the effect of theoscillation frequency on the crack susceptibility of 2014 aluminum welds (86).As shown, a minimum crack susceptibility exists at a rather low frequency,where alternating grain orientation is most pronounced. This frequency canvary with the welding speed.As shown in Figure 11.34, arc oscillation at much higher frequencies than1Hz, though ineffective in 2014 aluminum, is effective in 5052 aluminum (87).REDUCING SOLIDIFICATION CRACKING 291As shown in Figure 11.35, this is because grain refining occurs in alloy 5052welds at high oscillation frequencies (88). Heterogeneous nucleation isbelieved to be mainly responsible for the grain refining, since a 0.043wt% Tiwas found in the 5052 aluminum used.292 WELD METAL SOLIDIFICATION CRACKINGFigure 11.31 Effect of transverse arc oscillation (1 Hz) on solidification cracking ingas–tungsten arc welds of 2014 aluminum. Reprinted from Kou and Le (86).Figure 11.32 Schematic sketches showing effect of arc oscillation on solidificationcracking. From Kou and Le (87).REDUCING SOLIDIFICATION CRACKING 2932014 aluminumtransverse oscillationFrequency, HzCrack length, mm0 5 10 35150100500Figure 11.33 Effect of oscillation frequency on solidification cracking in gas–tungstenarc welds of 2014 aluminum. From Kou and Le (86).5052 aluminumtransverse oscillationFrequency, HzCrack length, mm0 10 20150100500amplitude 1.1 mm1.9 mmFigure 11.34 Effect of arc oscillation frequency on the solidification cracking ingas–tungsten arc welds of 5052 aluminum. From Kou and Le (87).Figure 11.35 Grain structure of gas–tungsten arc welds of 5052 aluminum: (a) no arcoscillation; (b) 20 Hz transverse arc oscillation. Reprinted from Kou and Le (87).11.4.3 Use of Favorable Welding ConditionsA. Reducing Strains As mentioned previously in Chapter 1, the use ofhigh-intensity heat sources (electron or laser beams) significantly reduces thedistortion of the workpiece and hence the thermally induced strains. Lessrestraint and proper preheating of the workpiece can also help reduce strains.Dyatlov and Sidoruk (89) and Nikov (90) found that preheating the workpiecedecreased the magnitude of strains induced by welding. Sekiguchi andMiyake (91) reduced solidification cracking in steel plates by preheating.Hernandez and North (92) positioned additional torches behind and along theside of the welding head and inhibited solidification cracking in aluminumalloy sheets. It was suggested that the local heating decreased the amount ofplastic straining resulting from the welding operation and produced a lessstressful situation behind the weld pool.B. Improving Weld Geometry The weld bead shape can also affect solidificationcracking (93).When a concave single-pass fillet weld cools and shrinks,the outer surface is stressed in tension, as show in Figure 11.36a. The outersurface can be considered as being pulled toward the toes and the root.However, by making the outer surface convex, as shown in Figure 11.36b,pulling toward the root actually compresses the outer surface and offsets thetension caused by pulling toward the toes. Consequently, the tensile stressesalong the outer surface are reduced, and the tendency for solidification crackingto initiate from the outer surface is lowered. It should be pointed out,however, that excessive convexity can produce stress concentrations andinduce fatigue cracking (Section 5.3) or hydrogen cracking (Section 17.4) atthe toes. In multiple-pass welding, as illustrated in Figure 11.37, solidificationcracking can also initiate from the weld surface if the weld passes are too wideand concave (93).The weld width-to-depth ratio can also affect solidification cracking. Asdepicted in Figure 11.38, deep narrow welds with a low width-to-depth ratiocan be susceptible to weld centerline cracking (93) This is because of the steepangle of the abutment between columnar grains growing from opposite sides294 WELD METAL SOLIDIFICATION CRACKINGsurface intensionConcave fillet weld Convex fillet weld(a)surface lessin tension(b)root toetoeFigure 11.36 Effect of weld bead shape on state of stress at center of outer surface:(a) concave fillet weld; (b) convex fillet weld. Modified from Blodgett (93).of the weld pool. This type of cracking is often observed in deep and narrowwelds produced by EBW or SAW.11.5 CASE STUDY: FAILURE OF A LARGE EXHAUST FANFigure 11.39a shows a schematic sketch of a large six-bladed exhaust fan fordrawing chemical mist from a chemical processing chamber (94). The hub ofthe fan had a diameter of about 178mm (7in.) and a washer-shaped verticalfin. Two flat-bar spokes were welded to the fin at each 60° interval and providedthe support for the fan blades. The fan acted only as the drawing forceto pull the mist from the process area; it did not come into direct contact withthe mist. A “scrubber” was positioned between the fan and the process unit toremove the chemical from the exhausted air mass.The spokes and blades werefabricated from type 316 austenitic stainless steel bars and plates, and the electrodesused for welding were type E316 covered electrodes.The fan measuredabout 1.5m (5ft) between blade tips, weighed about 320kg (700 lb), androtated at 1200 rpm in service. It failed after 12 weeks service time.Figure 11.39b shows the fractured fillet weld joining the underside ofthe blade to the supporting spoke. The weld metal was nonmagnetic, thatis, containing little d-ferrite, and was therefore susceptible to solidificationcracking. Note the jagged nature of the microfissures following the columnargrain structure above the arrowhead, as compared with the smooth cracktypical of fatigue failure below the arrowhead. In fact, the fracture surface ofCASE STUDY: FAILURE OF A LARGE EXHAUST FAN 295Crack Crack NoCracktoo wide and concave(also poor slagremoval)washed up toohigh andconcaveflat or slightly convexnot quite full width(also good slag removal)(a) (b) (c)Figure 11.37 Effect of weld bead shape on solidification in multipass weld: (a)concave; (b) concave; (c) convex. Modified from Blodgett (93).No Crack(b)Crack(a)Figure 11.38 Effect of weld depth–width ratio on centerline cracking: (a) ratio toohigh; (b) ratio correct. Modified from Blodgett (93).the failed spoke member exhibited the “clam shell” markings typical of fatiguefailure.In summary, the fully austenitic weld metal produced by type E316 electrodessuffered from solidification cracking, which in turn initiated fatiguecracking and led to final failure. Type 309 or 308 electrodes, which containmore d-ferrite to resist solidification cracking, could have been used to avoidsolidification cracking in the weld metal (94).REFERENCES1. Kou, S., and Kanevsky, Y., unpublished research, Carnegie-Mellon University,Pittsburgh, PA, 1980.2. Davies, G. J., and Garland, J. G., Int. Metal Rev., 20: 83, 1975.3. Lees, D. C. G., J. Inst. Metals, 72: 343, 1946.4. Singer, A. R. E., and Jennings, P. H., J. Inst. Metals, 73: 273, 1947.5. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.6. Bishop, H. F., Ackerlind, C. G., and Pellini,W. S., Trans. AFS, 65: 247, 1957.7. Borland, J. C., Br.Weld. J., 7: 508, 1960.8. Li, L., and Messler, Jr., R.W., Weld. J., 78: 387s, 1999.9. Matsuda, F., Nakagawa, H., and Sorada, K., Trans. JWS1, 11: 67, 1982.10. Matsuda, F., Nakagawa, H., Kohomoto, H., Honda, Y., and Matsubara, Y., Trans.JWSI, 12: 73, 1983.11. Houldcroft, P. T., Br.Weld. J., 2: 471, 1955.12. Liptax, J. A., and Baysinger, F. R., Weld. J., 47: 173s, 1968.296 WELD METAL SOLIDIFICATION CRACKINGFigure 11.39 Failure of a large welded exhaust fan (94): (a) location of failure; (b)failed component.13. Garland, J. G., and Davies, G. J., Metals Const. Br.Weld. J., vol. 1, p. 565, December1969.14. Rogerson, J. H., Cotterell, B., and Borland, J. C., Weld. J., 42: 264s, 1963.15. Savage,W. F., and Lundin, C. D., Weld. J., 44: 433s, 1965.16. Nakata, K., and Matsuda, F., Trans. JWRI, 24: 83, 1995.17. Krysiak, K. F., Grubb, J. F., Pollard, B., and Campbell, R. D., in Welding, Brazing,and Soldering, Vol. 6, ASM International, Materials Park, OH, December 1993,p. 443.18. DuPont, J. N., Michael, J. R., and Newbury, B. D., Weld. J., 78: 408s, 1999.19. Stout, R. D., in Weldability of Steels, 3rd ed., Eds. R. D. Stout and W. D. Doty,Welding Research Council, New York, 1978, p. 252.20. Messler, R.W., Jr., Principles of Welding, Processes, Physics, Chemistry and Metallurgy,Wiley, New York, 1999, pp. 557–589.21. Cieslak, M. J., Weld. J., 66: 57s, 1987.22. David, S. A., and Woodhouse, J. J., Weld. J., 66: 129s, 1987.23. Nelson, T.W., Lippold, J. C., Lin,W., and Baeslack III, Weld. J., 76: 110s, 1997.24. Goodwin, G. M., Weld. J., 66: 33s, 1987.25. Hondros, E. D., and Seah, M. P., Int. Metals Rev., 222: 12, 1977.26. Pease, G. R., Weld. J., 36: 330s, 1957.27. Canonico, D. A., Savage,W. F.,Werner,W. J., and Goodwin, G. M., paper presentedat Effects of Minor Elements on the Weldability of High-Nickel Alloys, WeldingResearch Council, NY, 1969, p. 68.28. Savage,W. F., Nippes, E. F., and Goodwin, G. M., Weld. J., 56: 245s, 1977.29. Kah, D. H., and Dickinson, D.W., Weld. J., 60: 135s, 1981.30. Principles and Technology of the Fusion Welding of Metals,Vol. 1, Mechanical EngineeringPublishing Co., Peking, China, 1979 (in Chinese).31. DuPont, J. N., Robino, C.V., and Marder, A. R., Weld. J., 77: 417s, 1998.32. DuPont, J. N., Robino, C.V., Marder, A. R., Notis, M. R., and Michael, J. R., Metall.Mater. Trans. A., 29A: 2785, 1998.33. DuPont, J. N., Robino, C.V., and Marder, A. R., Acta Mater., 46: 4781, 1998.34. DuPont, J. N., Robino, C.V., and Marder, A. R., Sci.Technol.Weld. Join., 4: 1, 1999.35. DuPont, J. N., Weld. J., 78: 253s, 1999.36. Singer, A. R. E., and Jennings, P. H., J. Inst. Metals, 73: 197, 1947.37. Pumphrey,W. I., and Lyons, J.V., J. Inst. Metals, 74: 439, 1948.38. Dowd, J. D., Weld. J., 31: 448s, 1952.39. Jennings, P. H., Singer, A. R. E., and Pumphrey,W. L., J. Inst. Metals, 74: 227, 1948.40. Dudas, J. H., and Collins, F. R., Weld. J., 45: 241s, 1966.41. Michaud, E. J., Kerr, H.W., and Weckman, D. C., in Trends in Welding Research,Eds. H. B. Smartt, J. A. Johnson, and S. A. David, ASM International, MaterialsPark OH, June 1995, p. 154.42. Yang,Y. P., Dong, P., Zhang, J., and Tian, X., Weld. J., 79: 9s, 2000.43. Gueussier, A., and Castro, R., Rev. Metall., 57: 117, 1960.44. Takalo, T., Suutala, N., and Moisio, T., Metall. Trans., 10A: 1173, 1979.REFERENCES 29745. Lienert, T. J., in Trends in Welding Research, Eds. J. M. Vitek, S. A. David, J. A.Johnson, H. B. Smartt, and T. DebRoy, ASM International, Materials Park, OH,June 1998, p. 726.46. Hull, F. C., Weld. J., 46: 399s, 1967.47. Baeslack,W. A., Duquette, D. J., and Savage,W. F., Corrosion, 35: 45, 1979.48. Baeslack,W. A., Duquette, D. J., and Savage,W. F., Weld. J., 57: 175s, 1978.49. Baeslack,W. A., Duquette, D. J., and Savage,W. F., Weld. J., 58: 83s, 1979.50. Manning, P. G., Duquette, D. J., and Savage,W. F., Weld. J., 59: 260s, 1980.51. Thomas, R. G., Weld. J., 57: 81s, 1978.52. Borland, J. C., and Younger, R. N., Br.Weld. J., 7: 22, 1960.53. Thomas, R. D. Jr., Metals Prog., 70: 73, 1956.54. Medovar, B. I., Avtomatischeskaya Svarka, 6: 3, 1953.55. Curran, R. M., and Rankin, A.W., Weld. J., 34: 205, 1955.56. Borland, J. C., Br.Weld. J., 7: 558, 1960.57. Rollason, E. C., and Bystram, M. C. T., J. Iron Steel Inst., 169: 347, 1951.58. Smith, C. S., Interphase Imperfections in Nearly Perfect Crystals,Wiley, New York,1952, pp. 377–401.59. Taylor, J.W., J. Inst. Metals, 86: 456, 1958.60. Brooks, J. A.,Thompson, A.W., and Williams, J. C., in Physical Metallurgy of MetalJoining, Eds. R. Kossowsky and M. E. Glickstein, Metallurgical Society of AIME,Warrendale, PA, 1980, p. 117.61. Smith, C. R., Trans. AIME, 175: 15, 1948.62. Matsuda, F., Nakata, K., Tsukamoto, K., and Arai, K., Trans. JWRI, 12: 93, 1983.63. Matsuda,F., Nakata, K.,Tsukamoto, K., and Uchiyama,T.,Trans.JWRI, 13: 57, 1984.64. Dvornak, M. J., Frost, R. H., and Olson, D. L., Weld. J., 68: 327s, 1989.65. Kou, S., and Le,Y., unpublished research, Carnegie-Mellon University, Pittsburgh,PA, 1981.66. Linnert,G. E.,Welding Metallurgy, Vol. 2, 3rd ed., American Welding Society, NewYork, 1967.67. Jennings, P. H., Singer, A. R. E., and Pumphrey,W. I., J. Inst. Metals, 74: 227, 1948.68. Pumphrey,W. I., and Moore, D. C., J. Inst. Metals, 73: 425, 1948.69. Savage,W. F., Nippes, E. F., and Varsik, J. D., Weld. J., 58: 45s, 1979.70. Evancho, J. W., and Baker, C. L., Weld Crack Susceptibility of Al-Mg Alloys,ALCOA Report,ALCOA Technology Center,ALCOA Center, PA, March 1980.71. Smith, R. B., in Welding, Brazing, and Soldering, Vol. 6, ASM International,Materials Park, OH, December 1993, p. 642.72. Borland, J. C., Br.Weld. J., 8: 526, 1961.73. Jefferson, T. B., and Woods, G., Metals and How to Weld Them, James Lincoln ArcWelding Foundation, Cleveland, OH, 1961.74. Schaeffler, A. L., Metal. Prog., 56: 680, 1949.75. DeLong,W. T., Weld. J., 53: 273s, 1974.76. Kotecki, D. J., Weld. J., 78: 180s, 1999.77. Kotecki, D. J., Weld. J., 79: 346s, 2000.298 WELD METAL SOLIDIFICATION CRACKING78. Balmforth, M. C., and Lippold, J. C., Weld. J., 79: 339s, 2000.79. Vitek, J. M., Iskander,Y. S., and Oblow, E. M., Weld. J., 79: 33s, 2000.80. Vitek, J. M., Iskander,Y. S., and Oblow, E. M., Weld. J., 79: 41s, 2000.81. Cieslak, M. J., and Savage,W. F., Weld. J., 60: 131s, 1981.82. Garland, J. G., Metal Const. Br.Weld. J., 21: 121, 1974.83. Tseng, C., and Savage,W. F., Weld. J., 50: 777, 1971.84. David, S. A., and Liu, C. T., in Grain Refinement in Castings and Welds, Eds. G. J.Abbaschian and S.A. David, Metals Society of AIME,Warrendale, PA, 1983, p. 249.85. Scarbrough, J. D., and Burgan, C. E., Weld. J., 63: 54, 1984.86. Kou, S., and Le,Y., Metall. Trans., 16A: 1887, 1985.87. Kou, S., and Le,Y., Metall. Trans., 16A: 1345, 1985.88. Kou, S., and Le,Y., Weld. J., 64: 51, 1985.89. Dyatlov,V. I., and Sidoruk,V. S., Autom.Weld., 3: 21, 1966.90. Nikov, N.Y., Weld. Production, 4: 25, 1975.91. Sekiguchi, H., and Miyake, H., J. Jpn.Weld. Soc., 6(1): 59, 1975.92. Hernandez, I. E., and North, T. H., Weld. J., 63: 84s, 1984.93. Blodgett, O.W., Weld. Innovation Q., 2(3): 4, 1985.94. Fatigue Fractures in Welded Constructions, Vol. 11, International Institute ofWelding, London, 1979.FURTHER READING1. Davies, G. J., and Garland, J. G., Int. Metall. Rev., 20: 83, 1975.2. Flemings, M. C., Solidification Processing, McGraw-Hill, New York, 1974.PROBLEMS11.1 Compare the solidification temperature range and fraction eutectic ofAl–3.0Cu with those of Al–6.0Cu. For simplicity, assume Scheil’s equationis a valid approximation and both the solidus and liquidus linesare straight in the Al–Cu system. CSM = 5.65, CE = 33, TE = 548°C, andTm = 660°C (pure Al).11.2 Centerline cracking is often observed in deep-penetration electron orlaser beam welds. Explain why.11.3 Fillet welds of 5052 Al (essentially Al–2.5Mg) are made with 5556 filler(essentially Al–5.1Mg). What are the approximate dilution ratio andweld metal composition? Is the weld metal susceptible to solidificationcracking?11.4 Solidification cracking in 2014 aluminum sheets can be reducedsignificantly by using low-frequency transverse arc oscillation. Low-PROBLEMS 299frequency circular and longitudinal arc oscillations, however, are lesseffective. Explain why.11.5 Low-frequency arc pulsation during autogenous GTAW of aluminumalloys, such as 6061 and 2014, is often found detrimental rather thanbeneficial in controlling solidification cracking. Explain why.11.6 A structural steel has a nominal composition of 0.16 C, 1.4 Mn, 0.4 Si,0.022 S, and 0.016 P. Because of macrosegregation of carbon duringingot casting, some of the steel plates produced contained as much as0.245% C. Severe solidification cracking was reported in welds of thesesteel plates. Explain why. The problem was solved by using a differentfiller wire. Comment on the carbon content of the new filler wire.11.7 Consider welding 1018 steel to 304 stainless steel by GMAW (dilutionnormally between 30 and 40%). Assume approximately equal contributionto weld metal dilution from each side of the joint.Will the weldmetal be susceptible to solidification cracking if ER308L Si is used asthe electrode? Will the weld metal be dangerously close to the martensiteboundary? 1018 steel: 0.18C–0.02Cr–0.03Ni–0.01Mo–0.07Cu–0.01N; 304 stainless steel: 0.05C–18.30Cr–8.80Ni–0.05Mo–0.08Cu–0.04N; ER308L Si: 0.03C–19.90Cr–10.20Ni–0.21Mo–0.19Cu–0.06N.11.8 Repeat the previous problem but with ER309L Si as the electrode.ER309L Si: 0.02C–24.10Cr–12.70Ni–0.13Mo–0.16Cu–0.05N.11.9 Stainless steels normally considered resistant to solidification crackingin arc welding based on the constitution diagram (such as Schaeffleror WRC 1992), for instance, 304L, 316L, and 321Mo, can become rathersusceptible in laser or electron beam welding. Explain why.11.10 It has been observed in 1100 aluminum alloy that the Ti or Zr contentof the alloy, up to about 0.5wt%, can significantly affect its susceptibilityto solidification cracking. Explain how Ti or Zr affects weld metalsolidification cracking of the alloy and why.11.11 It has been reported that autogenous gas–tungsten arc welds ofInvar (Fe–36wt% Ni) are rather susceptible to solidification cracking.Explain why based on a constitution diagram. The addition of Ti(e.g., 0.5–1.0%) has been found to change the Mn sulfide films alongthe grain boundaries to tiny Ti sulfide particles entrapped betweendendrite arms. What is the effect of the Ti addition on solidificationcracking?300 WELD METAL SOLIDIFICATION CRACKINGPART IIIThe Partially Melted ZoneWelding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-412 Formation of the PartiallyMelted ZoneSevere liquation can occur in the partially melted zone during welding. Severalfundamental liquation-related phenomena are discussed in this chapter,including liquation mechanisms, solidification of the grain boundary (GB)liquid, and the resultant GB segregation.12.1 EVIDENCE OF LIQUATIONThe partially melted zone (PMZ) is the area immediately outside the weldmetal where liquation can occur during welding. Figure 12.1a shows a portionof the PMZ in a gas–metal arc weld of a 6061 aluminum made with a 4145filler metal. The presence of dark-etching GBs along the fusion boundary inthis micrograph is an indication of GB liquation.The microstructure inside thewhite rectangle is enlarged in Figure 12.1b.The liquated and resolidified materialalong the GB consists of a dark-etching eutectic GB and a lighter etchinga (Al-rich) band along the GB.Figure 12.2 shows the PMZ microstructure in alloy 2219, which is essentiallyAl–6.3Cu (1). The liquated and resolidified material along the GB consistsof a dark-etching eutectic GB and a light-etching a band along the GB.The a bands here appear lighter than those in Figure 12.1b because of the useof a different etching solution. As shown, the large dark-etching eutectic particleswithin grains are surrounded by a light-etching a phase, thus indicatingthat liquation also occurs within grains.The formation of the PMZ in 2219 aluminum is explained in Figure 12.3.As shown in the phase diagram (Figure 12.3a), the composition of alloy 2219is C0 = 6.3% Cu. As shown by the thermal cycles (Figure 12.3b), the materialat position b is heated up to between the eutectic temperature TE and the liquidustemperature TL during welding.Therefore, the material becomes a solidplus-liquid mixture (a + L), that is, it is partially melted. The material atposition a is completely melted while that at position c is not melted at all.Asimilar explanation for the formation of the PMZ, in fact, has been given previouslyin Figure 7.12.303Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-412.2 LIQUATION MECHANISMSHuang et al. (1–5) have conducted a series of studies on liquation and solidificationin the PMZ of aluminum welds. Figure 12.4 shows five different PMZliquation mechanisms. The phase diagram shown in Figure 12.4a is similar tothe Al-rich side of the Al–Cu phase diagram. Here, AxBy is an intermetalliccompound, such as Al2Cu in the case of Al–Cu alloys. Alloy C1 is within thesolubility limit of the a phase, CSM, and alloy C2 is beyond it. In the as-castcondition both alloys C1 and C2 usually consist of an a matrix and the eutectica + AxBy along GBs and in between dendrite arms.12.2.1 Mechanism 1: AxBy Reacting with MatrixThis mechanism is shown in Figure 12.4b. Here alloy C2, for instance, alloy2219, consists of an a matrix and AxBy particles at any temperature up to the304 FORMATION OF THE PARTIALLY MELTED ZONEFigure 12.1 PMZ microstructure of gas–metal arc weld in 6061 aluminum made with4145 aluminum filler wire. Rectangular area in (a) enlarged in (b).eutectic temperature TE, regardless of the heating rate to TE during welding.At TE liquation is initiated by the eutectic reaction AxBy + a Æ L.The liquation mechanism for alloy 2219 can be explained with the helpof Figure 12.5. The base metal contains large and small q (Al2Cu) particlesboth within grains and along GBs (Figure 12.5a), as shown by the SEMimage at the top. At the border line between the PMZ and the basemetal (Figure 12.5b), the material is heated to the eutectic temperature TE.Regardless of the heating rate q (Al2Cu) particles are always present whenTE is reached. Consequently, liquation occurs by the eutectic reaction a +q Æ LE, where LE is the liquid of the eutectic composition CE. Upon cooling,the eutectic liquid solidifies into the eutectic solid without compositionchanges and results in eutectic particles and some GB eutectic (Figure12.5e).Above TE, that is, inside the PMZ, liquation intensifies. The a matrix surroundingthe eutectic liquid dissolves and the liquid increases in volumeLIQUATION MECHANISMS 305Figure 12.2 PMZ microstructure of gas–metal arc weld in 2219 aluminum made with2319 aluminum filler wire: (a) left of weld; (b) right of weld. Reprinted from Huangand Kou (1). Courtesy of American Welding Society.(Figure 12.5c). This causes the liquid composition to change from eutectic tohypoeutectic (
C1 < CSMPMZ (5)TL2TEWeldMetalTEWeldMetalsolvus3. Residual AxBy reacting with matrix:(constitutional liquation)AxBy+ L at TEif AxBy still present at TE5. Melting of matrix:L at TS1if no AxBy or eutectic present at TE1. AxBy reacting with matrix:AxBy+ L at TEAxBy always present at TEregardless of heating rateBMTV4. Melting of residual eutectic:eutetic(S) eutectic(L) at TEif eutectic still present at TE2. Melting of eutectic:eutectic(S) eutectic(L) at TEeutectic always presentat TE regardless of heating rate+ AxByααααααFigure 12.4 Five mechanisms for liquation in PMZ of aluminum alloys: (a) phasediagram; (b) two mechanisms for an alloy beyond the solid solubility limit (CSM); (c)three mechanisms for an alloy within the solid solubility limit.308 FORMATION OF THE PARTIALLY MELTED ZONEFigure 12.5 Microstructure evolution in PMZ of 2219 aluminum with SEM image ofthe base metal on the top and optical micrograph of the PMZ at the bottom. Reprintedfrom Huang and Kou (1). Courtesy of American Welding Society.interface. Further heating to above TE would allow additional time for furtherdissolution of AxBy and further formation of the liquid phase. Therefore, it isclear that localized melting should be possible with rapid heating rates at temperaturessignificantly below the equilibrium solidus temperature TS1.Figure 12.6 shows the microstructure of the PMZ of a resistance spot weldof 18-Ni maraging steel (6).This micrograph reveals the four stages leading toGB liquation.The first stage, visible at point A, shows a rodlike titanium sulfideinclusion beginning to form a thin liquid film surrounding the inclusion. Thesecond stage is visible at point B, which is closer to the fusion boundary andhence experiences a higher peak temperature than at point A.As shown, liquationis more extensive and an elliptical liquid pool surrounds the remainingsmall gray inclusion. Still closer to the fusion boundary, at point C, no moresolid inclusion remains in the liquid pool, and penetration of GBs by the liquidphase is evident along the GBs intersecting with the liquid phase. Finally, atposition D, GB penetration by the liquid phase is so extensive that the GBsare liquated.Constitutional liquation has also been observed in several nickel-basealloys (8–11), such as Udimet 700,Waspaloy, Hastelloy X, and Inconel 718, andin 347 stainless steel (12). Constitutional liquation can be initiated by the interactionbetween the matrix and particles of carbide or other intermetallic compounds.Examples include M6C in Hastelloy X, MC carbide in Udimet 700,Waspaloy and Inconel 718, and Ni2Nb Laves phase in Inconel 718. Figure 12.7shows the PMZ of an Inconel 718 weld (13). Constitutional liquation occursby the eutectic reaction between the Ni2Nb Laves phase and the nickel matrix.LIQUATION MECHANISMS 309Figure 12.6 PMZ of an electric resistance spot weld of 18-Ni maraging steel showingconstitutional liquation. The fusion zone is at the top. Magnification 385¥. Reprintedfrom Pepe and Savage (6). Courtesy of American Welding Society.Constitutional liquation alone, however, is not enough to cause the liquidto penetrate most GBs in the PMZ. For this to occur in Ni-base alloys, GBmigration is also needed. Figure 12.8 shows schematically the formation of GBfilms in the PMZ due to the simultaneous occurrence of constitutional liquationand GB migration (6).The microstructure at location d0 is representativeof the as-received plate. At location d1, significant grain growth occurs abovethe effective grain coarsening temperature. Meanwhile, some of the movingGBs intersect with the solute-rich pools formed by constitutional liquation,thus allowing the solute-rich liquid to penetrate these GBs.At location d2 moregrain growth occurs and sufficient solute-rich liquid penetrates the GBs andforms GB films. These GBs are pinned due to the wetting action of the films.No further grain growth is expected until either the solute-rich liquid phase isdissipated by homogenization or the local temperature decreases to below theeffective solidus of the solute-rich liquid to cause it to solidify. If insufficienttime is available to dissipate the solute-rich liquid GB films before the localtemperature decreases to below the effective solidus of the liquid, the liquidGB films will solidify as a solute-rich GB network.A “ghost” GB network willthus remain fixed when grain growth resumes, as shown by the dashed linesat location d3. Grain growth will continue until either an equilibrium GBnetwork is formed or the temperature decreases to below the effective graincoarsening temperature. Figure 12.9 shows the ghost GB network near thefusion boundaries of gas–tungsten arc welds of 18-Ni maraging steel (6) anda Ni-base superalloy 690 (14).310 FORMATION OF THE PARTIALLY MELTED ZONEFigure 12.7 PMZ of Inconel 718 weld showing constitutional liquation due to Laveseutectic reaction. Reprinted from Kelly (13).Figure 12.8 Schematic representation of constitutional liquation and formation ofghost GB network. Reprinted from Pepe and Savage (6). Courtesy of AmericanWelding Society.Figure 12.9 Ghost GBs near fusion boundary of gas–tungsten arc welds: (a) 18-Nimaraging steel; magnification 125¥. Reprinted from Pepe and Savage (6). Courtesy ofAmerican Welding Society. (b) Ni-base superalloy 690. Reprinted from Lee and Kuo(14).(a)(b)12.2.4 Mechanism 4: Melting of Residual EutecticThis mechanism is also shown in Figure 12.4c. Here alloy C1, for instance, anas-cast Al–4.5Cu alloy, still contains the residual eutectic a + AxBy along GBsand in between dendrite arms when the eutectic temperature TE is reached.If the alloy is heated very slowly to above the solvus temperature TV, the eutecticcan dissolve completely in the a matrix by solid-state diffusion. However,if it is heated rapidly to above TV, as in welding, the eutectic does not haveenough time to dissolve completely in the a matrix because solid-state diffusiontakes time. Consequently, upon further heating to the eutectic temperatureTE, the residual eutectic melts and becomes liquid eutectic. Above TE thesurrounding a phase dissolves in the liquid and the liquid becomes hypoeutectic.Upon cooling, the hypoeutectic liquid solidifies first as solute-depleteda and last as solute-rich eutectic when its composition increases to CE. Figure12.10 shows the PMZ of a gas–metal arc weld of an as-cast Al–4.5% Cu alloy(4). Liquation is evident at the prior eutectic sites along the GB and in betweendendrite arms. A light-etching a band is present along the eutectic GB. Likewise,light-etching a rings surround the eutectic particles near the fusionboundary.12.2.5 Mechanism 5: Melting of MatrixThis mechanism is also shown in Figure 12.4c. Here alloy C1 contains neitherAxBy particles nor the a + AxBy eutectic when the eutectic temperature TE isreached. An alloy Al–4.5Cu solution heat treated before welding is anexample. Slow heating to TE can also cause complete dissolution of AxBy oreutectic in the a matrix before TE, but this usually is not likely in welding.The312 FORMATION OF THE PARTIALLY MELTED ZONEFigure 12.10 PMZ microstructure of gas–metal arc weld in as-cast Al–4.5Cu.Reprinted from Huang et al. (4).PMZ ranges from the solidus temperature (TS1) to the liquidus temperature(TL1), instead of from the eutectic temperature to the liquidus temperature,as in all previous cases. Figure 12.11 shows the PMZ microstructure in anAl–4.5% Cu alloy solution heat treated and quenched before GMAW (4).Eutectic is present both along GBs and within grains in the PMZ even thoughthe base metal is free of eutectic. Another example is a 6061-T6 aluminumcontaining submicrometer-size Mg2Si precipitate, which is reverted in the amatrix before reaching TE during GMAW (5).12.2.6 Mechanism 6: Segregation-Induced LiquationLippold et al. (15) proposed a segregation-induced liquation mechanism foraustenitic and duplex stainless steels. In such a mechanism the alloy and/orimpurity elements that depress the melting point segregate to GBs, lowerthe melting point, and cause GB liquation. In other words, GB segregationtakes place first and GB liquation next.This is opposite to all the other mechanismsdescribed in this chapter, where liquation takes place first and segregationoccurs during the solidification of the liquated material, as will bediscussed subsequently. They suggested that such GB segregation can beLIQUATION MECHANISMS 313Figure 12.11 PMZ microstructure of gas–metal arc weld in cast Al–4.5Cu homogenizedbefore welding (a) and magnified (b). Reprinted from Huang et al. (4).caused by (i) equilibrium diffusion of atoms of the elements to GBs, (ii) GBsweeping of such atoms into migrating GBs during grain growth, and (iii)“pipeline” diffusion of such atoms along GBs in the fusion zone that are continuousacross the fusion boundary into the PMZ. More details are availableelsewhere (15, 16).12.3 DIRECTIONAL SOLIDIFICATION OF LIQUATED MATERIALHuang et al. (1, 4) observed that the GB liquid has a tendency to solidify essentiallyupward and toward the weld regardless of its location with respect to theweld, as shown schematically in Figure 12.12. This directional solidification iscaused by the high-temperature gradients toward the weld during welding. Ithas been generally accepted that the GB liquid between two neighboringgrains solidifies from both grains to the middle between them. However, themicrographs in Figure 12.2 show that it solidifies from one grain to the other—in the direction upward and toward the weld.This, in fact, suggests GB migrationin the same direction.However, if the grains in the PMZ are very thin or very long, there may notbe much GB area facing the weld. Consequently, solidification of the GB liquidis still directional but just upward (5).12.4 GRAIN BOUNDARY SEGREGATIONAs the GB liquid solidifies, solute atoms are rejected by the solid into theliquid if the equilibrium partition coefficient k < 1, such as in the phase diagram314 FORMATION OF THE PARTIALLY MELTED ZONEPartially MeltedZoneBase MetalWeldMetal: direction ofsolidificationRollingdirectioneutecticsolutedepletedαFigure 12.12 Directional solidification of GB liquid in PMZ.shown in Figure 12.3a. The GB liquid solidifies first as a solute-depleted a butfinally as eutectic when the liquid composition reaches CE.Figure 12.13 depicts the GB segregation that develops during solidificationof the GB liquid in the PMZ (5).Take alloy 2219 as an example. Let C0 be theconcentration of Cu in the base metal (6.3%).Theoretically, the measured concentrationof the GB eutectic, Ce, is the eutectic composition CE (33%) if theGB eutectic is normal and q or Al2Cu (55%) if it is divorced. A normal eutecticrefers to a eutectic with a composite-like structure of a + q. A divorcedeutectic, on the other hand, refers to a eutectic that nucleates upon an existinga matrix and thus looks like q alone. In practice, with EPMA (electronprobe microanalysis) the value of Ce can be less than CE if the GB is thinnerthan the volume of material excited by the electrons, that is, the surroundinga of low-Cu is included in the composition analysis. This volume depends onthe voltage used in EPMA and the material being analyzed.In the absence of back diffusion, the concentration of the element at thestarting edge of the a strip should be kC0, where k is the equilibrium partitionratio of the element. The dashed line shows the resultant GB segregation ofthe element. However, if back diffusion of the solute from the growth frontinto the solute-depleted a strip is significant, the concentration of the elementat the starting edge of the a strip will be greater than kC0, as the solid lineindicates.Severe GB segregation of alloying elements has been observed in the PMZof gas–metal arc welds of alloys 2219, 2024, 6061, and 7075 (2, 5). Figure 12.14shows that in alloy 2219 the composition varies from about 2% Cu at the startingedge of the a strip to 30% Cu at the GB eutectic (2).The 2–3% Cu contentof the a strip is significantly lower than that of the base metal (C0 = 6.3% Cu),GRAIN BOUNDARY SEGREGATION 315Partially MeltedZoneBase MetalWeldMetal(a)(b)GB eutecticabDistance, zSoluteConcentration, C(c)a bk < 1backdiffusionno backdiffusionCokCoCesolute-depletedααFigure 12.13 Grain boundary segregation in PMZ. From Huang and Kou (5).thus confirming that the a strip is Cu depleted.The 2% Cu content at the startingedge of the a strip is higher than kC0 (1.07% = 0.17 ¥ 6.3%), thus implyingback diffusion of Cu into the a strip during GB solidification.The 30% Cu concentrationat the GB is close to the 33% Cu composition of a normal eutectic.12.5 GRAIN BOUNDARY SOLIDIFICATION MODESAs shown in Figures 12.1, 12.2, and 12.11, the a band along the eutectic GB isplanar, namely, without cells or dendrites. This suggests that the solidificationmode of the GB liquid is planar. The vertical temperature gradient G and the316 FORMATION OF THE PARTIALLY MELTED ZONEFigure 12.14 Grain boundary segregation in PMZ of 2219 aluminum weld: (a) electronmicrograph; (b) composition profile. Reprinted from Huang and Kou (2). Courtesyof American Welding Society.Partially MeltedZoneBase MetalWeldMetal(a)(b) (c)eutectic eutecticGrainCellularSolidificationPlanarSolidificationthinner thickera bplanar solutedepletedcellular soluteαα depleted α αFigure 12.15 Solidification modes of GB liquid in PMZ.GRAIN BOUNDARY SOLIDIFICATION MODES 317Figure 12.16 Solidification modes of GB liquid in PMZ of 2219 aluminum weld: (a)planar; (b) cellular. Reprinted from Huang and Kou (3). Courtesy of American WeldingSociety.vertical growth rate R were determined in the PMZ of alloy 2219 and theupward solidification of the GB liquid was analyzed (3). The ratio G/R forplanar GB solidification was found to be on the order of 105°Cs/cm2, which isclose to that required for planar solidification of Al–6.3% Cu.Although planar solidification of the GB liquid predominates in the PMZ,cellular solidification can also occur (3, 5). These cellular a bands share twocommon characteristics. First, they are often located near the weld bottom.Second, on average, they appear significantly thicker than the planar a bandsnearby. These characteristics, depicted in Figure 12.15, can be because of thelower vertical temperature gradient G in the area or backfilling of liquid fromthe weld pool (3, 5). Since a thicker GB liquid has to solidify faster, the verticallyupward solidification rate R is higher.The lower G/R in the area suggestsa greater chance for constitutional supercooling and hence cellular insteadof planar solidification. Figure 12.16 shows the PMZ microstructure in agas–metal arc weld of alloy 2219 (3). However, it should be pointed out thatplanar solidification changes to cellular solidification gradually, and a thinnerGB liquid may not have enough room for the transition to take place. Therefore,planar GB solidification may not necessarily mean that G/R is highenough to avoid cellular solidification.12.6 PARTIALLY MELTED ZONE IN CAST IRONSFigure 12.17 shows the PMZ in a cast iron weld, where g, a, and C representaustenite, ferrite, and graphite, respectively. This area tends to freeze as whiteiron due to the high cooling rates and becomes very hard (17).REFERENCES1. Huang, C., and Kou, S., Weld. J., 79: 113s, 2000.2. Huang, C., and Kou, S., Weld. J., 80: 9s, 2001.318 FORMATION OF THE PARTIALLY MELTED ZONEFe 1 2Carbon, wt%Temperature, oC16001200800L+ C+ CL3 4(a)(b)Cast ironpartially melted zoneheat-affected zonebase metalfusion zone+ γγγαFigure 12.17 PMZ in a cast iron.3. Huang, C., and Kou, S., Weld. J., 80: 46s, 2001.4. Huang, C., Kou, S., and Purins, J. R., in Proceedings of Merton C. Flemings Symposiumon Solidification and Materials Processing, Eds. R. Abbaschian, H. Brody,and A. Mortensen, Minerals, Metals and Materials Society,Warrendale, PA, 2001,p. 229.5. Huang, C., and Kou, S., Weld. J., in press.6. Pepe, J. J., and Savage,W. F., Weld. J., 46: 411s, 1967.7. Pepe, J. J., and Savage,W. F., Weld. J., 49: 545s, 1970.8. Owczarski,W. A., Duvall, D. S., and Sullivan, C. P., Weld. J., 45: 145s, 1966.9. Duvall, D. S., and Owczarski,W. A., Weld. J., 46: 423s, 1967.10. Savage,W. F., and Krantz, B. M., Weld. J., 45: 13s, 1966.11. Thompson, R. G., and Genculu, S., Weld. J., 62: 337s, 1983.12. Dudley, R., Ph.D. Thesis, Rensselaer Polytechnic Institute, Troy, NY, 1962.13. Kelly,T. J., in Weldability of Materials, Eds. R. A. Patterson and K.W. Mahin,ASMInternational, Materials Park, OH, 1990, p. 151.14. Lee, H. T., and Kuo, T.Y., Sci. Technol.Weld. Join., 4: 94, 1999.15. Lippold, J. C., Baselack III,W. A., and Varol, I., Weld. J., 71: 1s, 1992.16. Lippold, J. C., in Technology and Advancements and New Industrial Applicationsin Welding, Proceedings of the Taiwan International Welding Conference ’98, Eds.C. Tsai and H. Tsai, Tjing Ling Industrial Research Institute, National University,Taipei, Taiwan, 1998.17. Bushey, R. A., in ASM Handbook, Vol. 6: Welding, Brazing and Soldering, ASMInternational, Materials Park, OH, 1993, p. 708.PROBLEMS12.1 Sheets of alloy 2219 (Al–6.3Cu) 1.6mm thick are welded with theGTAW process using the following welding conditions: I = 60A, E = 10V, U = 3mm/s. Suppose the arc efficiency is 70%. Estimate the PMZwidth using the Adams two-dimensional equation (Chapter 2). AssumeTL = 645°C and TE = 548°C.12.2 The heating rate in resistance spot welding can be much faster than thatin GTAW. Is constitutional liquation expected to be more severe in thePMZ of 18-Ni-250 maraging steel in resistance spot welding or GTAW?Explain why.12.3 It has been pointed out that Fe3C tends to dissociate upon heatingmore easily than most other alloy carbides. Also, the diffusion rate ofthe interstitial solute, carbon, is much faster than substitutional solutes,for example, sulfur in the case of titanium sulfide inclusion. Do youexpect a plain carbon eutectoid steel containing fine Fe3C particles tobe susceptible to constitutional liquation when heated to the eutectictemperature under normal heating rates during welding (say less than500°C/s)? Why or why not?PROBLEMS 31912.4 It has been reported that by replacing Cb with Ta in Ni-base superalloy718, the Laves eutectic reaction temperature increases from 1185 to1225°C. Does the width of the PMZ induced by constitutional liquation(and liquation-induced cracking) increase or decrease in the alloy andwhy?12.5 It has been suggested that in welding cast irons reducing the peak temperaturesand the duration at the high temperatures is the most effectiveway to reduce PMZ problems. Is the use of a low-melting-point fillermetal desirable in this respect? Is the use of high preheat temperature(to prevent the formation of martensite) desirable in this respect?12.6 Based on the PMZ microstructure of the as-cast alloy Al–4.5Cu shownin Figure 12.10, what can be said about the solidification direction andmode of the grain boundary liquid and why?12.7 Based on the PMZ microstructure of the homogenized alloy Al–4.5Cushown in Figure 12.11, what can be said about the solidification directionand mode of the grain boundary liquid and why?320 FORMATION OF THE PARTIALLY MELTED ZONE13 Difficulties Associated with thePartially Melted ZoneThe partially melted zone (PMZ) can suffer from liquation cracking, loss ofductility, and hydrogen cracking. Liquation cracking, that is, cracking inducedby grain boundary liquation in the PMZ during welding, is also called PMZcracking or hot cracking. The causes of these problems and the remedies,especially for liquation cracking, will be discussed in this chapter. Liquationcracking and ductility loss are particularly severe in aluminum alloys. Forconvenience of discussion, the nominal compositions of several commercialaluminum alloys and filler wires are listed in Table 13.1 (1).13.1 LIQUATION CRACKINGFigure 13.1 shows the longitudinal cross section at the bottom of a gas–metalarc weld of 2219 aluminum made with a filler wire of 1100 aluminum. Therolling direction of the workpiece is perpendicular to the plane of the micrograph.Liquation cracking in the PMZ is intergranular (2–6). Liquation crackingcan also occur along the fusion boundary (3). The presence of a liquidphase at the intergranular fracture surface can be either evident (4) or unclear(5, 6).13.1.1 Crack Susceptibility TestsThe susceptibility of the PMZ to liquation cracking can be evaluated usingseveral different methods, such as Varestraint testing, circular-patch testingand hot-ductility testing, etc.A. Varestraint Testing This is usually used for partially penetrating welds inplates (Chapter 11). In brief, the workpiece is subjected to augmented strainsduring welding, and the extent of cracking in the PMZ is used as the index forthe susceptibility to liquation cracking (7–12).B. Circular-Patch Testing This is usually used for fully penetrating welds inthin sheets. A relatively high restraint is imposed on the weld zone transverseto the weld (5).The fixture design shown in Figure 13.2 was used for liquation321Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4cracking test of aluminum welds (13). The specimen is sandwiched betweentwo copper plates (the upper one having a big round opening for welding) andtightened by tightening the bolts against the stainless steel base plate.A similardesign was used by Nelson et al. (14) for assessing solidification cracking insteel welds. Cracking is at the outer edge of the weld, not the inner. This isbecause contraction of the weld during cooling is hindered by the restraint,thus rendering the outer edge in tension and the inner edge in compression.Figure 13.3a shows a circular weld made in 6061 aluminum with a 1100 aluminumfiller.The three cracks in the photo all initiate from the PMZ near theouter edge of the weld and propagate into the fusion zone along the welding322 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONETABLE 13.1 Nominal Compositions of Some Commercial Aluminum AlloysSi Cu Mn Mg Cr Ni Zn Ti Zr Fe1100 — 0.122014 0.8 4.4 0.8 0.52024 — 4.4 0.6 1.52219 — 6.3 0.3 — — — — 0.06 0.182319 — 6.3 0.3 — — — — 0.15 0.184043 5.25083 — — 0.7 4.4 0.155356 — — 0.12 5.0 0.12 — — 0.136061 0.6 0.28 — 1.0 0.26063 0.4 — — 0.76082 0.9 — 0.5 0.7 — — — — — 0.37002 — 0.75 — 2.5 — — 3.57075 — 1.6 — 2.5 0.23 — 5.6Source: Aluminum Association (1).Figure 13.1 PMZ cracking in 2219 aluminum welded with filler metal 1100.direction (clockwise). Figure 13.3b shows a circular weld between 2219 aluminumand 1100 aluminum made with a 1100 aluminum filler (13). A longPMZ crack runs along the outer edge of the weld.C. Hot Ductility Testing This has been used extensively for evaluating thehot-cracking susceptibility of nickel-base alloys (15, 16). It is most often performedon a Gleeble weld simulator (Chapter 2), which is also a tensile testinginstrument. The specimen is resistance heated according to a predeterminedthermal cycle resembling that in the PMZ. It is tensile tested, for instance, ata stroke rate of 5 cm/s, at predetermined temperatures along the thermal cycle,either during heating to the peak temperature of the thermal cycle or duringcooling from it. Several different criteria have been used for interpreting hotductility curves (17). For instance, one of the criteria is based on the ability ofthe material to reestablish ductility, that is, how fast ductility recovers duringcooling from the peak temperature. If ductility recovers right below the peaktemperature, the alloy is considered crack resistant, such as that shown inFigure 13.4a for a low-B Cabot alloy 214 (6). On the other hand, if ductilityrecovers well below the peak temperature, the alloy is considered crack sensitive,such as that shown in Figure 13.4b for a high-B Cabot 214. The mechanismthat boron affects liquation cracking in Ni-base superalloys is not clear(18).LIQUATION CRACKING 323stainless steel base platecopperplatespecimenthreadswasher2.5cm10 cmspecimenweld10 cmweldholebolt copperplate witha holeFigure 13.2 Schematic sketch of a circular-patch test.Figure 13.3 Cracking in circular-patch welds: (a) 6061 aluminum made with a 1100filler wire; (b) 2219 aluminum (outside) welded to 1100 aluminum (inside) with a 1100filler wire. From Huang and Kou (13).OCPeaktemperature(1345 oC)900 1000 1100 1200 1300 14000204060801001650 1830 2010 2370 2550Temperature, oCTemperature, oFReduction of area, %2019OHOCPeaktemperature(1345 oC)900 1000 1100 1200 1300 14000204060801001650 1830 2010 2370 2550Temperature, oCTemperature, oFReduction of area, %2019OH(a) (b)Figure 13.4 Hot ductility response of Cabot alloy 214 with two different boron levels:(a) 0.0002 wt% B; (b) 0.003 wt% B. OH, testing done on heating to 1345°C; OC, oncooling from 1345°C. From Cieslak (6). Reprinted from ASM Handbook, vol. 6,ASMInternational.324 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONE13.1.2 Mechanisms of Liquation CrackingFigure 13.5a is a schematic showing the formation of liquation cracking in thePMZ of a full-penetration aluminum weld (13). Since the PMZ is weakenedby grain boundary liquation, it cracks when the solidifying weld metal contractsand pulls it. Most aluminum alloys are susceptible to liquation cracking.This is because of their wide PMZ (due to wide freezing temperature rangeand high thermal conductivity), large solidification shrinkage (solid densitysignificantly greater than liquid density), and large thermal contraction (largethermal expansion coefficient). The solidification shrinkage of aluminum is ashigh as 6.6%, and the thermal expansion coefficient of aluminum is roughlytwice that of iron base alloys. Figure 13.5b shows liquation cracking in an alloy6061 circular weld (Figure 13.3a). The light etching a band along the grainboundary is a clear evidence of the grain boundary liquid that weakened thePMZ during welding.LIQUATION CRACKING 325weldpoolsolidifying weldmetal pulling PMZpullingcrackmagnifiedgrainfusion boundaryweldpool base metalPMZ weakened by grainboundary (GB) liquationPMZFigure 13.5 Formation of PMZ cracking in a full-penetration aluminum weld: (a)schematic; (b) PMZ cracking in 6061 aluminum. From Huang and Kou (13).(a)Figure 13.6 shows the effect of the weld metal composition on liquationcracking in 2219 aluminum, which is essentially Al-6.3Cu (13). The circularpatchweld on the right is identical to the PMZ (or the base metal) in composition,that is, Al-6.3Cu. No liquation cracking occurs. The circular-patchweld on the left (same as that in Figure 13.3b), however, has a significantlylower Cu content than the PMZ, and liquation cracking was severe.This effectof the weld metal composition will be explained as follows.Since the cooling rate during welding is too high for equilibrium solidification,it is inappropriate to discuss liquation cracking based on the solidustemperature from an equilibrium phase diagram. From Equation (6.13) fornonequilibrium solidification, the fraction of liquid fL at any given temperatureT can be expressed as follows:(13.1)where mL (<0) is the slope of the liquidus line in the phase diagram, Co thesolute content of the alloy, Tm the melting point of pure aluminum, and kthe equilibrium partition ratio. Therefore, at any temperature T the lower Co,the smaller fL is, that is, the stronger the solid/liquid mixture becomes.Consider the circular-patch weld on the left in Figure 13.6.The weld metal hasfm CT TkLL om=(- )-Êˈ¯1 1-326 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEAl wt % CuL10 20 30400500600700Temperature, oC+TE = 548 oC620solid solubility limit at 5.65liquationcrackingsolidificationcrackingdecreasing liquation cracking inalloy 2219 (essentially Al-6.3Cu)+ L2 cm2219weldingdirectioncircular-patch welds in workpiece of alloy 2219weld weldααα θFigure 13.6 Effect of weld metal composition on PMZ cracking in 2219 aluminum.From Huang and Kou (13).a significantly lower Co (Al-0.95Cu) than the PMZ (Al-6.3Cu). This suggeststhat at any temperature T the weld metal is significantly stronger than thePMZ, thus causing liquation cracking. As for the circular-patch on the right,the weld metal and the PMZ have the same Co and hence the similar strengthlevel. As such, no liquation cracking occurs.Figure 13.7a is a schematic showing the formation of liquation cracking inthe PMZ of a partial-penetration GMA weld of an aluminum alloy. The papillary(nipple) type penetration pattern shown in the figure is common inGMAW of aluminum alloys with Ar shielding, where spray transfer is themode of filler metal transfer through the arc (13).The welding direction is per-LIQUATION CRACKING 327(a)pullingforceweld poolpartially meltedzonesolidifying and contracting weld metalgrainboundaryrollingbase directionmetalcracks form if weld metal develops sufficientstrength to pull away while GBs are still liquatedgrain boundary(GB) liquidfusion boundaryFigure 13.7 Weld metal pulling and tearing PMZ: (a) schematic sketch; (b) 7075 aluminumwelded with filler 1100. From Huang and Kou (13).pendicular to the rolling direction. The weld metal in the papillary penetration,as indicated by its very fine cell spacing of the solidification microstructure,solidifies rapidly. This suggests that the rapidly solidifying and thuscontracting weld metal in the papillary penetration pulls the PMZ thatis weakened by grain boundary liquation. Figure 13.7b shows the transversecross section near the bottom of a GMA weld of 7075 aluminum made witha filler wire of 1100 aluminum. As shown, the weld metal pulls and tears thePMZ near the tip of the papillary penetration (13).13.2 LOSS OF STRENGTH AND DUCTILITYAs mentioned in the previous chapter, Huang and Kou (19–21) studied liquationin the PMZ of 2219 aluminum gas–metal arc welds and found both a Cudepleteda band next to the Cu-rich GB eutectic and a Cu-depleted a ringsurrounding each large Cu-rich eutectic particle in the grain interior. Results ofmicrohardness testing showed that the Cu-depleted a was much softer than theCu-rich eutectic.This suggests that the liquated material solidifies with severesegregation and results in a weak PMZ microstructure with a soft ductile a anda hard brittle eutectic right next to each other. Under tensile loading, the ayields without much resistance while the eutectic fractures badly.Figure 13.8 shows the tensile testing results of a weld made perpendicularto the rolling direction (20).The maximum load and elongation before failureare both much lower in the weld specimen than in the base-metal specimen,as shown in Figure 13.8a. Fracture of eutectic is evident both along the GBand at large eutectic particles in the grain interior, as shown in Figures 13.8band c. The fluctuations in the tensile load in Figure 13.8a are likely to be associatedwith the fracture of the eutectic.13.3 HYDROGEN CRACKINGSavage et al. (22) studied hydrogen-induced cracking in HY-80 steel. Theyobserved intergranular cracking in the PMZ and the adjacent region in thefusion zone where mixing between the filler and the weld metal is incomplete,as shown in Figure 13.9.It was pointed out that the creation of liquated films on the GBs in the PMZprovides preferential paths along which hydrogen from the weld metal candiffuse across the fusion boundary. This, according to Savage et al. (22), isbecause liquid iron can dissolve approximately three to four times morenascent hydrogen than the solid, thus making the liquated GBs serve as“pipelines” along which hydrogen from the weld metal can readily diffuseacross the fusion boundary.When these segregated films resolidify, they notonly are left supersaturated with hydrogen but also exhibit a higher hardenabilitydue to solute segregation. Consequently, they serve as preferred nucleationsites for hydrogen-induced cracking.328 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEFigure 13.8 Results of tensile testing of a gas–metal arc weld of 2219 aluminum madeperpendicular to the rolling direction. Reprinted from Huang and Kou (20). Courtesyof American Welding Society.Figure 13.9 Hydrogen-induced cracking in the PMZ of HY-80 steel. Reprinted fromSavage et al. (22). Courtesy of American Welding Society.HYDROGEN CRACKING 32913.4 REMEDIESRemedies for the problems associated with the PMZ can be grouped into fourcategories: filler metal, heat source, degree of restraint, and base metal. Thesewill be discussed below.13.4.1 Filler MetalLiquation cracking can be reduced by selecting the proper filler metal.Metzger (23) reported the significant effect of the weld metal composition onliquation cracking in aluminum alloys. Liquation cracking occurred in 6061aluminum welds produced with Al–Mg fillers at high dilution ratios but not inwelds made with Al–Si fillers at any dilution ratios. Metzger’s study has beenconfirmed by subsequent studies on alloys 6061, 6063, and 6082 (5, 10–12,24–26).Gittos and Scott (5) studied liquation cracking in alloy 6082 welded with5356 and 4043 fillers using the circular-patch test. Like Metzger (23), Gittosand Scott (5) observed liquation cracking welds made with the 5356 filler athigh dilution ratios (about 80%) but not in welds made with the 4043 filler atany dilution ratios.When it occurred, liquation cracking was along the outeredge of the weld and no cracking was observed along the inner edge.Gittos and Scott (5) proposed the criterion of TWS > TBS for liquation cracking
to occur, where TWS and TBS are the solidus temperatures of the weld metal
and the base metal, respectively. They assumed that if the weld metal composition
is such that TWS > TBS, then the PMZ will solidify before the weld metal
and thus resist tensile strains arising from weld metal solidification. The weld
metal solidus temperature TWS and the base-metal solidus temperature TBS
were taken from Figure 13.10, which shows the solidus temperatures in the Alrich
corner of the ternary Al–Mg–Si system (27).They found the variations in
TWS and TBS with the dilution ratio shown in Figure 13.11a to be consistent
with the results of their circular-patch testing.
Katoh and Kerr (10, 11) and Miyazaki et al. (12) studied liquation cracking
in 6000 alloys, including 6061, using Varestraint testing. Longitudinal liquation
cracking occurred when alloy 6061 was welded with a 5356 filler but not with
a 4043 filler. They measured the solidus temperatures of the base metals and
filler metals by differential thermal analysis. The solidus temperature TBS of
alloy 6061 was 597°C. Contrary to the TWS > TBS cracking criterion proposed
by Gittos and Scott (5), Miyazaki et al. (12) found TWS < TBS whether the fillermetal was 5356 or 4043 (12). This is shown in Figure 13.11b. It was proposedthat the base metal of 6061 aluminum probably liquated at 559°C by constitutionalliquation induced by the Al–Mg2Si–Si ternary eutectic.It should be noted that when attempting to avoid liquation cracking in thePMZ by choosing a proper filler metal, the solidification cracking susceptibilityof the fusion zone still needs to be checked. Solidification cracking–composition diagrams (Figure 11.26) can be useful for this purpose (28).330 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEREMEDIES 331560570560000.5 1.0 1.5 2.0 2.5 3.0 3.5 4.00.51.01.52.02.53.03.54.05.04.55.56.04.5 5.0 5.5 6.0 6.5 7.0Al - Mg2Si - Si555 oCQuasi-BinaryMagnesium, wt%Silicon, wt%570580 590580570590595580550560Solidus Temperatures forTernary Al-Mg-Si System590600610620570Al6082 workpiece6061 workpiece4043 filler5356 fillerFigure 13.10 Ternary Al–Mg–Si phase diagram showing the solidus temperature.Modified from Phillips (27).0 20 40 60 80 100550560570580590600base metal53564043Dilution by base metal, D(%)Solidus temperature, Ts(oC)(a)0 20 40 60 80 10050052054056058060053564043Dilution by base metal, D(%)Solidus temperature, Ts(oC)6061(b)Figure 13.11 Variation of weld metal solidus temperature with dilution: (a) in 6082aluminum (5); (b) in 6061 aluminum (12). (a) from Gittos et al. (5) and (b) fromMiyazaki et al. (12). Reprinted from Welding Journal, Courtesy of American WeldingSociety.13.4.2 Heat SourceThe size of the PMZ and hence the extent of PMZ liquation can be reducedby reducing the heat input, as illustrated in Figure 13.12. Figure 13.13 showsthe effect of the heat input on liquation cracking in Varestraint testing ofgas–metal arc welds of alloy 6061 made with a 5356 filler metal (12). To minimizethe difficulties associated with the PMZ, the heat input can be kept lowby using multipass welding or low-heat-input welding processes (such as EBWand GTAW) when possible.Kou and Le (29) reduced GB liquation and thus liquation cracking in thePMZ of 2014 aluminum alloy by using transverse arc oscillation (1 Hz) duringGTAW. The extent of GB melting is significantly smaller with arc oscillation.With the same welding speed, the resultant speed of the heat source isincreased by transverse arc oscillation (Chapter 8). This results in a smallerweld pool and a narrower PMZ.13.4.3 Degree of RestraintLiquation cracking and hydrogen-induced cracking in the PMZ are bothcaused by the combination of a susceptible microstructure and the presenceof tensile stresses.The sensitivity of the PMZ to both types of cracking can be332 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEfusion zonelower Q/V, narrower PMZhigher Q/V, wider PMZQ/V: (heat input)/(welding speed) orheat input per unit length of weldpool pool fusion zoneFigure 13.12 Effect of heat input on width of PMZ.Q/V=9.2kJ/cmQ/V=5.6kJ/cmQ/V=3.4kJ/cmLongitudinal crack0.0 1.0 2.0 3.0 4.0 5.0 6.00481216Augmented strain, (%)Crack length, L(mm)εFigure 13.13 Effect of heat input on liquation cracking in Varestraint testing of 6061aluminum welded with 5356 filler metal. Modified from Miyazaki et al. (12). Courtesyof American Welding Society.reduced by decreasing the degree of restraint and hence the level of tensilestresses.13.4.4 Base MetalLiquation cracking can be reduced by selecting the proper base metal forwelding if it is feasible. The base-metal composition, grain structure, andmicrosegregation can affect the susceptibility of the PMZ to liquation crackingsignificantly.A. Impurities When impurities such as sulfur and phosphorus are present,the freezing temperature range can be widened rather significantly (Chapter11). The widening of the freezing temperature range is due to the lowering ofthe incipient melting temperature, which is effectively the same as the liquationtemperature in the sense of liquation cracking. The detrimental effect ofsulfur and phosphorus on the liquation cracking of nickel-base alloys has beenrecognized (15, 16).The effect of minor alloying elements on the liquation temperatureof 347 stainless steel is shown in Figure 13.14 (30).B. Grain Size The coarser the grains are, the less ductile the PMZ becomes.Furthermore, the coarser the grains are, the less the GB area is and hence themore concentrated the impurities or low-melting-point segregates are at theGB, as shown in Figure 13.15. Consequently, a base metal with coarser grainsis expected to be more susceptible to liquation cracking in the PMZ, as shownin Figure 13.16 by Varestraint testing the gas–tungsten arc welds of 6061 aluminum(12).Thompson et al. (31) showed the effect of the grain size on liquationcracking in Inconel 718 caused by constitutional liquation. Guo et al. (32)showed in Figure 13.17 the effect of both the grain size and the boron contenton the total crack length of electron beam welded Inconel 718 specimens.Figure 13.18 shows liquation cracking in gas–metal arc welds of two Al–4.5%Cu alloys of different grain sizes (33). Cracking is much more severe withcoarse grains.REMEDIES 3330 0.16 0.24 0.32 0.402380240024202440246024801/2%Cb / 30 (%C) + 50 (%N)Liquation temperature, oF130513151325133513451355Liquation temperature, oC0.08Figure 13.14 Effect of minor alloying elements on liquation temperature of 347 stainlesssteel. From Cullen and Freeman (30).C. Grain Orientation Lippold et al. (34) studied liquation cracking inthe PMZ of 5083 aluminum alloy and found that PMZ cracking wasmore severe in welds made transverse to the rolling direction than thosemade parallel to the rolling direction. It was suggested that in the latterthe elongated grains produced by the action of rolling were parallel to theweld and, therefore, it was more difficult for cracks to propagate into the basemetal.D. Microsegregation In the welding of as-cast materials, the PMZ is particularlysusceptible to liquation cracking because of the presence of334 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEfiner grainsmore grainboundary arealower concentration ofliquation-inducing materialat grain boundary(a)coarser grainsless grainboundary areahigher concentration ofliquation-inducing materialat grain boundary(b)Figure 13.15 Effect of grain size on concentration of liquation-causing material atgrain boundaries.0 2 4 6 8 10012345= 3.0%Average grain size, d(mm)Maximum crack length, Lmax(mm)εFigure 13.16 Effect of grain size on liquation cracking in Varestraint testing of 6061aluminum gas–tungsten arc welds. Reprinted from Miyazaki et al. (12). Courtesy ofAmerican Welding Society.REMEDIES 335Low B alloyHigh B alloy0 100 2000400800Grain size, mTotal crack length, mμμFigure 13.17 Effect of grain size and boron content on liquation cracking in PMZ ofInconel 718 electron beam welds. Reprinted from Guo et al. (32).Figure 13.18 Liquation cracking in two Al–4.5% Cu alloys: (a) small grains; (b) coarsegrains. Reprinted from Huang et al. (33).low-melting-point GB segregates. Upon heating during welding, excessive GBliquation occurs in the PMZ, making it highly susceptible to liquation cracking.Figure 13.19 shows liquation cracking in a cast 304 stainless steel (35) anda cast corrosion-resistant austenitic stainless steel (36). It initiates from thePMZ and propagates into the fusion zone.REFERENCES1. Aluminum Association, Aluminum Standards and Data, Aluminum Association,Washington, DC, 1982, p. 15.2. Kreischer, C. H., Weld. J., 42: 49s, 1963.3. Dudas, J. H., and Collins, F. R., Weld. J., 45: 241s, 1966.4. Thompson, R. G., in ASM Handbook, Vol. 6, ASM International, Materials Park,OH, 1993, p. 566.336 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONEFigure 13.19 Liquation cracking originated from PMZ extending into the fusion zone.(a) Cast 304 stainless steel. Reprinted from Apblett (35). Courtesy of AmericanWelding Society. (b) Cast corrosion-resistant austenitic stainless steel. Reprinted fromCieslak (36).5. Gittos, N. F., and Scott, M. H., Weld. J., 60: 95s, 1981.6. Cieslak, M. J., in ASM Handbook, Vol. 6: Welding, Brazing and Soldering, ASMInternational, Materials Park, OH, 1993, p. 88.7. Savage,W. F., and Dickinson, D.W., Weld. J., 51: 555s, 1972.8. Savage,W. F., and Lundin, C. D., Weld. J., 44: 433s, 1965.9. Lippold, J. C., Nippes, E. F., and Savage,W. F., Weld. J., 56: 171s, 1977.10. Katoh, M., and Kerr, H.W., Weld. J., 66: 360s, 1987.11. Kerr, H.W., and Katoh, M., Weld. J., 66: 251s, 1987.12. Miyazaki, M., Nishio, K., Katoh, M., Mukae, S., and Kerr, H.W., Weld. J., 69: 362s,1990.13. Huang, C., and Kou, S., Weld. J., submitted for publication.14. Nelson,T.W., Lippold, J. C., Lin,W., and Baselack III,W. A.,Weld. J., 76: 110s, 1997.15. Effects of Minor Elements on the Weldability of High-Nickel Alloys, WeldingResearch Council, 1969.16. Methods of High-Alloy Weldability Evaluation,Welding Research Council, 1970.17. Yeniscavich,W., in Methods of High-Alloy Weldability Evaluation, p. 1.18. Kelly,T. J., in Weldability of Materials, Eds. R. A. Patterson and K.W. Mahin,ASMInternational, Materials Park, OH, 1990, p. 151.19. Huang, C., and Kou, S., Weld. J., 79: 113s, 2000.20. Huang, C., and Kou, S., Weld. J., 80: 9s, 2001.21. Huang, C., and Kou, S., Weld. J., 80: 46s, 2001.22. Savage,W. F., Nippes, E. F., and Szekeres, E. S., Weld. J., 55: 276s, 1976.23. Metzger, G. E., Weld. J., 46: 457s, 1967.24. Gitter, R., Maier, J., Muller, W., and Schwellinger, P., in Proceedings of the FifthInternational Conference on Aluminum Weldments, Eds. D. Kosteas, R. Ondra, andF. Ostermann, Technische Universita Munchen, Munchen, 1992, pp. 4.1.1–4.1.13.25. Powell, G. L. F., Baughn, K., Ahmed, N., Dalton J. W., and Robinson, P., in Proceedingsof International Conference on Materials in Welding and Joining, Instituteof Metals and Materials Australasia, Parkville,Victoria,Australia, 1995.26. Ellis, M. B. D., Gittos, M. F., and Hadley, I., Weld. Inst. J., 6: 213, 1997.27. Philips, H. W. L., Annotated Equilibrium Diagrams of Some Aluminum AlloySystems, Institute of Metals, London, 1959, p. 67.28. Jennings, P. H., Singer, A. R. E., and Pumphrey,W. I., J. Inst. Metals, 74: 227, 1948.29. Kou, S., and Le,Y., Weld. J., 64: 51, 1985.30. Cullen, T. M., and Freeman, J.W., J. Eng. Power, 85: 151, 1963.31. Thompson, R. G., Cassimus, J. J., Mayo, D. E., and Dobbs, J. R., Weld. J., 64: 91s,1985.32. Guo, H., Chaturvedi, M. C., and Richards, N. L., Sci. Technol. Weld. Join., 4: 257,1999.33. Huang, C., Kou, S., and Purins, J. R., in Proceedings of Merton C. Flemings Symposiumon Solidification and Materials Processing, Eds. R. Abbaschian, H. Brody,and A. Mortensen, Minerals, Metals and Materials Society,Warrendale, PA, 2001,p. 229.REFERENCES 33734. Lippold, J. C., Nippes, E. F., and Savage,W. F., Weld. J., 56: 171s, 1977.35. Apblett,W. R., and Pellini,W. S., Weld. J., 33: 83s, 1954.36. Cieslak, M. J., in ASM Handbook, Vol. 6: Welding, Brazing and Soldering, ASMInternational, Materials Park, OH, 1993, p. 495.PROBLEMS13.1 Hot-ductility testing was performed on an 18% Ni maraging steel followinga thermal cycle with a peak temperature of 1400°C. The onheatingpart of the testing showed that the ductility dropped to zero at1380°C (called the nil ductility temperature), and the on-cooling partshowed that the ductility recovered from zero to about 7% at 1360°C.Is this maraging steel very susceptible to liquation cracking? Explainwhy or why not. Do you expect the specimen tensile tested on heatingat 1380°C to exhibit brittle intergranular fracture of ductile transgranulardimple fracture? Why? What do you think caused PMZ liquationin this maraging steel?13.2 (a) The effect of the carbon content and the Mn–S ratio on weld metalsolidification cracking in steels has been described in Chapter 11. It hasbeen reported that a similar effect also exists in the liquation crackingof the PMZ of steels. Explain why. (b) Because of the higher strengthof HY-130 than HY-80, its chemical composition should be more strictlycontrolled if liquation cracking is to be avoided. Assume the followingcontents: HY-80: £0.18 C; 0.1–0.4 Mn; £0.025 S; £0.025 P; HY-130: £0.12C; 0.6–0.9 Mn; £0.010 S; £0.010 P. Do these contents suggest a more strictcomposition control in HY-130?13.3 Sulfur can form a liquid with nickel that has a eutectic temperature of635°C. Do you expect high-strength alloy steels containing Ni (say morethan 2.5%) to be rather susceptible to liquation cracking due to sulfur?Explain why or why not.13.4 Low-transverse-frequency arc oscillation (Figure 8.17) has beenreported to reduce PMZ liquation. Sketch both the weld and the PMZbehind the weld pool and show how this can be true.13.5 Consider the circular-patch weld in Figure 13.3b.Will liquation crackingoccur if the outer piece is alloy 1100 (essentially pure aluminum)and the inner piece (the circular patch) is alloy 2219 (Al-6.3Cu)?Explain why or why not.13.6 Like aluminum alloy 7075, alloy 2024 is very susceptible to liquationcracking. In GMAW of alloy 2024 do you expect liquation cracking tobe much more severe with filler metal 4043 or 1100? Why?338 DIFFICULTIES ASSOCIATED WITH THE PARTIALLY MELTED ZONE13.7 In a circular-patch test alloy 2219 (Al-6.3Cu) is welded with alloy 2319(Al-6.3Cu) plus extra Cu as the filler metal. The resultant compositionof the weld metal is about Al-8.5Cu. Do you expect liquation crackingto occur? Explain why or why not.13.8 Do you expect liquation cracking to occur in autogenous GTAW of7075? Why or why not?PROBLEMS 339PART IVThe Heat-Affected ZoneWelding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-414 Work-Hardened MaterialsMetals can be strengthened in several ways, including solution hardening, workhardening, precipitation hardening, and transformation hardening. The effectivenessof the last three methods can be reduced significantly by heatingduring welding in the area called the heat-affected zone (HAZ), where thepeak temperatures are too low to cause melting but high enough to causethe microstructure and properties of the materials to change significantly.Solution-hardening materials are usually less affected unless they have beenwork hardened and thus will not be discussed separately. This chapter shallfocus on recrystallization and grain growth in the HAZ of work-hardenedmaterials, which can make the HAZ much weaker than the base metal.14.1 BACKGROUNDWhen a metal is cold worked and plastically deformed, for instance, coldrolled or extruded, numerous dislocations are generated. These dislocationscan interact with each other and form dislocations tangles. Such dislocationtangles hinder the movement of newly generated dislocations and, hence,further plastic deformation of the metal. In this way, a metal is strengthenedor hardened by cold working. This strengthening mechanism is called workhardening.14.1.1 RecrystallizationMost of the energy expended in work hardening appears in the form of heatbut, as shown in Figure 14.1, a finite fraction is stored in the material as strainenergy (1).When a work-hardened material is annealed, the deformed grainsin the material tend to recrystallize by forming fresh, strain-free grains thatare soft, just like grains that have not been deformed.The stored strain energyis the driving force for recrystallization of a work-hardened material (2), andthis energy is released as fresh, strain-free grains form. Figure 14.2 shows thevarious stages of recrystallization in a work-hardened brass (3). Slip bands,which have formed during severe work hardening, serve as the nucleation sitesfor new grains.The extent of recrystallization increases with increasing annealing temperatureand time (4). Therefore, it can be expected that the strength or hardness343Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-40 10 20 30 40051015Fraction of energystoredStored energyPercent elongationStored energy, cal/moleFraction of stored energy, %Figure 14.1 Stored energy and fraction of stored energy as a function of tensile elongationduring cold working of high-purity copper. From Gordon (l).Figure 14.2 Deformation and recrystallization structures of a-brass: (a) 33% coldreduction; (b) short anneal at 580°C; (c) longer anneal; (d) completely recrystallized;(e) grain growth. From Burke (3).344 WORK-HARDENED MATERIALSof a work-hardened material tends to decrease with increasing annealingtemperature and time. Figure 14.3 shows the hardness of a work-hardenedcartridge brass as a function of annealing temperature (5).Table 14.1 summarizes the recrystallization temperature for various metals.For most metals the recrystallization temperature is around 40–50% of theirmelting point in degrees Kelvin (6). It should be pointed out that the recrystallizationtemperature of a metal can be affected by the degree of work hardeningand the purity level (2).In fact, before recrystallization takes place, there exists a period of timeduring which certain properties of the work-hardened material, for instance,the electrical resistivity, tend to recover without causing any microstructuralchanges. This phenomenon is called recovery. However, since the mechanicalproperties of the material, such as strength or hardness, do not change significantlyduring recovery, recovery is not important in welding.BACKGROUND 34520406080100120400 800 1200 1600Drawn 200 400 600 80063%Tensilestrength150300450600750Annealing temperature, oCAnnealing temperature, oFStrength, ksiStrength, MPaFigure 14.3 Strength of cartridge brass (Cu–35Zn) cold rolled to an 80% reductionin area and then annealed at various temperatures for 60 min. Reprinted from MetalsHandbook (5).TABLE 14.1 Recrystallization Temperatures forVarious MetalsMinimumRecrystallization MeltingMetal Temperature (°C) Temperature (°C)Aluminum 150 660Magnesium 200 659Copper 200 1083Iron 450 1530Nickel 600 1452Molybdenum 900 2617Tantalum 1000 3000Source: Brick et al. (6).14.1.2 Grain GrowthUpon completion of recrystallization, grains begin to grow. The driving forcefor grain growth is the surface energy. The total grain boundary area and thusthe total surface energy of the system can be reduced if fewer and coarsergrains are present. This can be illustrated by the growth of soap cells in a flatcontainer (7), as shown in Figure 14.4. It should be pointed out that since thedriving force for grain growth is the surface energy rather than the storedstrain energy, grain growth is not limited to work-hardened materials.Like recrystallization, the extent of grain growth also increases with increasingannealing temperature and time. Figure 14.5 shows grain growth in coldrolledbrass as a function of temperature and time (5).346 WORK-HARDENED MATERIALSFigure 14.4 Growth of soap cells in a flat container.The numbers indicate growth timein minutes. From Smith (7).0.080.060.040.030.021 10 1004006008001000120014001600300400500600700800900Grain size,mmDuration of annealing, minAnnealing temperature, oFAnnealing temperature, oCFigure 14.5 Grain growth of Cu–35Zn brass cold rolled to 63% reduction in area.Reprinted from Metals Handbook (5).It is worth noting that carbide and nitride particles can inhibit graingrowth in steels by hindering the movement of grain boundaries (2). Theseparticles, if not dissolved during welding, tend to inhibit grain growth in theHAZ.14.2 RECRYSTALLIZATION AND GRAIN GROWTH IN WELDINGThe effect of work hardening is completely gone in the fusion zone becauseof melting and is partially lost in the HAZ because of recrystallization andgrain growth. These strength losses should be taken into account in structuraldesigns involving welding.14.2.1 MicrostructureFigure 14.6 shows the weld microstructure of a work-hardened 304 stainlesssteel (8). The microstructure of the same material before work hardening isalso included for comparison (Figure 14.6a). Recrystallization (Figure 14.6d)and grain growth (Figure 14.6e) are evident in the HAZ. Figure 14.7 showsgrain growth in the HAZ of a molybdenum weld (9). Severe HAZ graingrowth can result in coarse grains in the fusion zone because of epitaxialgrowth (Chapter 7). Fracture toughness is usually poor with coarse grains inthe HAZ and the fusion zone.RECRYSTALLIZATION AND GRAIN GROWTH IN WELDING 347Figure 14.6 Microstructure across the weld of a work-hardened 304 stainless steel: (a)before work hardening; (b) base metal; (c) carbide precipitation at grain boundaries;(d) recrystallization; (e) grain growth next to fusion boundary; ( f) fusion zone. Magnification137¥. Reprinted from Metals Handbook (8).14.2.2 Thermal CyclesThe loss of strength in the HAZ can be explained with the help of thermalcycles, as shown in Figure 14.8. The closer to the fusion boundary, the higherthe peak temperature becomes and the longer the material stays above348 WORK-HARDENED MATERIALSFigure 14.7 Grain growth in electron beam weld of molybdenum, arrows indicatingfusion boundary. Reprinted from Wadsworth et al. (9). Copyright 1983 with permissionfrom Elsevier Science.Tx3211 2 3123(a)Temperature, TStrength orhardnessDistanceWeld(b)HAZTLWorkhardenedbase metalloss ofstrengthTime, tFigure 14.8 Softening of work-hardened material caused by welding: (a) thermalcycles; (b) strength or hardness profile.the effective recrystallization temperature, Tx. Under rapid heating duringwelding, the recrystallization temperature may increase because recrystallizationrequires diffusion and diffusion takes time. Since the strength of a workhardenedmaterial decreases with increasing annealing temperature andtime, the strength or hardness of the HAZ decreases as the fusion boundaryis approached. Figure 14.9 shows the HAZ strength profiles of two workhardened5083 aluminum plates (10). It appears that the harder the base metal,the greater the strength loss is.Grain growth in the HAZ can also be explained with the help of thermalcycles, as shown in Figure 14.10. The closer to the fusion boundary, the higherthe peak temperature becomes and the longer the material stays at high temperatures.Since grain growth increases with increasing annealing temperatureand time (Figure 14.5), the grain size in the HAZ increases as the fusionboundary is approached.14.3 EFFECT OF WELDING PARAMETERS AND PROCESSThe effect of welding parameters on the HAZ strength is explained in Figure14.11. Both the size of the HAZ and the retention time above the effectiverecrystallization temperature Tx increase with increasing heat input per unitlength of the weld, that is, the ratio of heat input to welding speed. Consequently,the loss of strength in the HAZ becomes more severe as the heatinput per unit length of the weld is increased. Figure 14.12 shows the effect ofEFFECT OF WELDING PARAMETERS AND PROCESS 349Distance from weldcenterline, cmDistance from weldcenterline, inYield strength, KSI0 2.5 50 1 2H32H11310203040Yield strength, MPa100200150250Figure 14.9 Yield strength profiles across welds of two work-hardened 5083 aluminumplates. Reprinted from Cook et al. (10). Courtesy of American Welding Society.welding parameters on the HAZ strength of a work-hardened 5356-H321 aluminumalloy (11).Finally, Figure 14.13 shows the effect of the welding process on the HAZmicrostructure of a work-hardened 2219 aluminum (12). Because of the lowheat input and the high cooling rate in EBW, very little recrystallization isobserved in the HAZ of the work-hardened material. On the other hand,350 WORK-HARDENED MATERIALSWeight %, C Time, tLSTxCo 3211 2 312 3(b)(a)Temperature, TGrain sizeDistanceWeld(c)HAZTmTLFigure 14.10 Grain growth in HAZ: (a) phase diagram; (b) thermal cycles; (c) grainsize variations.123(a)(b)(c)1 2 3Distance fromweld centerline0Strength orhardnessIncreasing heatinput per unitlength of weldTL 1 2 3TxTimeFigure 14.11 Effect of heat input per unit length of weld on: (a) width of HAZ(shaded), (b) thermal cycles near fusion boundary, and (c) strength or hardnessprofiles.because of the higher heat input and lower cooling rate in GTAW, recrystallizationand even some grain growth are observed in the HAZ.REFERENCES1. Gordon, P., Trans. AIME, 203: 1043, 1955.2. Reed-Hill, R. E., Physical Metallurgy Principles, 2d ed.,Van Nostrand, New York,1973.3. Burke, J. E., in Grain Control in Industrial Metallurgy,American Society for Metals,Cleveland, OH, 1949.4. Decker, B. F., and Harker, D., Trans. AIME, 188: 887, 1950.REFERENCES 3513,940 J/cm (10,000 J/in)5,905 J/cm(15,000 J/in)11,810 J/cm (30,000 J/in)Distance from weld centerline, cm0 2.5Distance from weld centerline, in50 1 2Hardness, RB1020304050Figure 14.12 Effect of heat input per unit length of weld on HAZ hardness in a workhardened5356 aluminum. Reprinted from White et al. (11). Courtesy of AmericanWelding Society.Figure 14.13 Microstructure near fusion boundary of a work-hardened 2219-T37aluminum: (a) electron beam weld; (b) gas–tungsten arc weld. Magnification 80¥.Reprinted from Metals Handbook (12).5. Metals Handbook, 8th ed., Vol. 2, American Society for Metals, Metals Park, OH,1972, p. 285.6. Brick, R. M., Pense, A. W., and Gordon, R. B., Structure and Properties of EngineeringMaterials, 4th ed., McGraw-Hill, New York, 1977, p. 81.7. Smith, C. S., ASM Seminar, Metal Interfaces, ASM, Metals Park, OH, 1952, p. 65.8. Metals Handbook, 8th ed., Vol. 7, American Society for Metals, Metals Park, OH,1972, p. 135.9. Wadsworth, J., Morse, G. R., and Chewey, P. M., Mater. Sci. Eng., 59: 257 (1983).10. Cook, L. A., Channon, S. L., and Hard, A. R., Weld. J., 34: 112, 1955.11. White, S. S., Manchester. R. E., Moffatt,W. G., and Adams, C. M., Weld. J., 39: 10s,1960.12. Metals Handbook, 8th ed., Vol. 7, American Society for Metals, Metals Park, OH,1972, p. 268.FURTHER READING1. Reed-Hill, R. E., Physical Metallurgy Principles, 2nd ed.,Van Nostrand, New York,1973.2. Brick, R. M., Pense,A.W., and Gordon, R.B., Structure and Properties of EngineeringMaterials, 4th ed., McGraw-Hill, New York, 1977.PROBLEMS14.1 A 301 stainless steel sheet work-hardened to about 480 Knoop hardnesswas welded, and in the HAZ the hardness droped to a minimumof about 240. Explain the loss of strength in the HAZ. The weld reinforcementwas machined off and the whole sheet including the weld wascold rolled.What was the purpose of cold rolling?14.2 It is known that bcc is less close-packed than fcc and thus has a higherdiffusion coefficient. Are ferritic stainless steels (bcc at high temperatures)more or less susceptible to HAZ grain growth than austeniticstainless steels (fcc at high temperatures)? Explain why.14.3 Do you expect grain growth during the welding of tantalum (Tm =2996°C) to be more severe than during the welding of Al? Explain why.352 WORK-HARDENED MATERIALS15 Precipitation-HardeningMaterials I: Aluminum AlloysAluminum alloys are more frequently welded than any other types ofnonferrous alloys because of their widespread applications and fairly goodweldability. In general, higher strength aluminum alloys are more susceptibleto (i) hot cracking in the fusion zone and the PMZ and (ii) losses of strength/ductility in the HAZ. Aluminum–lithium alloys and PM (powder metallurgy)aluminum alloys can be rather susceptible to porosity in the fusion zone.Table15.1 summarizes typical problems in aluminum welding and recommendedsolutions. The problems associated with the fusion zone and the PMZ havebeen discussed previously. In this chapter, we shall focus on the HAZ phenomenain heat-treatable aluminum alloys, which are strengthened throughprecipitation hardening.Table 15.2 shows the designation for aluminum alloys.As shown, the 2000, 6000, and 7000 series are heat treatable, while the rest arenon–heat treatable.15.1 BACKGROUNDAluminum–copper alloys are a typical example of precipitation-hardeningmaterials. As shown in the Al-rich side of the Al–Cu phase diagram inFigure 15.1, the solubility of Cu in the a phase increases with increasingtemperature—a necessary criterion for precipitation hardening. Consider theprecipitation hardening of Al–4% Cu as an example. Step 1, solution heattreating, is to heat treat the alloy in the a-phase temperature range until itbecomes a solid solution. Step 2, quenching, is to rapidly cool the solid solutionto room temperature to make it supersaturated in Cu. Step 3, aging, is toallow the strengthening phase to precipitate from the supersaturated solidsolution. Aging by heating (e.g., at 190°C) is called artificial aging and agingwithout heating is called natural aging. In the heat-treating terminology, T6and T4 refer to a heat-treatable aluminum alloy in the artificially aged conditionand the naturally aged condition, respectively.Five sequential structures can be identified during the artificial aging ofAl–Cu alloys:Supersaturated solid solutionÆGPÆq¢¢Æq¢Æq(Al2Cu) (15.1)353Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4where q (Al2Cu) is the equilibrium phase with a body-centered-tetragonal(bct) structure. The GP zones (Guinier–Preston, sometimes called GP1), theq≤ phase (sometimes called GP2), and the q¢ phase are metastable phases.Figure 15.2 shows the solvus curves of these metastable phases, which representthe highest temperatures these phases can exist (1, 2).The GP zones are coherent with the crystal lattice of the a solid solution.They consist of disks a few atoms thick (4–6 Å) and about 80–100 Å in diameter,formed on the {100} planes of the solid solutions (3). Since a Cu atom isabout 11% smaller than an Al atom in diameter and the GP zones are richerin Cu than the solid solution, the crystal lattice is strained around the GPzones.The strain fields associated with the GP zones allow them to be detectedin the electron microscope.The q≤ phase is also coherent with the crystal lattice354 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSTABLE 15.1 Typical Welding Problems in Aluminum AlloysSectionsTypical Problems Alloy Type Solutions in bookPorosity Al-Li alloys (severe) Surface scraping or milling 3.2Thermovacuum treatment 10.3Variable-polarity keyholePAWPowder-metallurgy Thermovacuum treatment 3.2alloys (severe) Minimize powder oxidationand hydration duringatomization andconsolidationOther types (less Clean workpiece and wire 3.2severe) surfaceVariable-polarity keyholePAWSolidification Higher-strength alloys Use proper filler wires and 11.4cracking in FZ (e.g., 2014, 6061, dilution7075) In autogenous GTA welding, 11.4use arc oscillation or less 7.6susceptible alloys (2219)Hot cracking and Higher-strength alloys Use low heat inputa 13.1low ductility in Use proper filler wires 13.2PMZ Low-frequency arc oscillationSoftening in HAZ Work-hardened Use low-heat input 14.2materials 14.3Heat-treatable alloys Use low-heat input 15.2Postweld heat treating 15.3a Low heat input processes (e.g., EBW, GTAW) or multiple-pass welding with low-heat input ineach pass and low interpass temperature.355TABLE 15.2 Designation of Wrought Aluminum AlloysNot Heat Heat Not Heat Not Heat Not Heat HeatTreatable Treatable Treatable Treatable Treatable Heat Treatable TreatableSeries 1000 2000 3000 4000 5000 6000 7000Major None Cu Mn Si Mg Mg/Si ZnalloyingelementsAdvantages Electrical/ Strength Formability Filler Strength Strength, Strengththermal wires after extrudabilityconductivity weldingExample 1100 2219 3003 4043 5052 6061 7075of the solid solution, its size ranging from 10 to 40 Å in thickness and 100 to1000Å in diameter. The q¢ phase, on the other hand, is semicoherent with thelattice of the solid solution. It is not related to the GP zones or the q≤ phase;it nucleates heterogeneously, especially on dislocations.The size of the q¢ phaseranges from 100 to 150 Å in thickness and 100 to 6000 Å or more in diameterdepending on the time and temperature of aging (4). Figure 15.3 shows a transmissionelectron micrograph of the q¢ phase in 2219 aluminum (Al–6Cu) (5).Finally, the q phase, which can either form from q¢ or directly from the solidsolution, is incoherent with the lattice of the solid solution.Figure 15.4 shows the correlation of these structures with the hardnessof Al–4Cu (6).The maximum hardness (or strength) occurs when the amountof q≤ (or GP2) is at a maximum, although some contribution may also beprovided by q¢ (6). As q¢ grows in size and increases in amount, the coherentstrains decrease and the alloy becomes overaged. As aging continues evenfurther, the incoherent q phase forms and the alloy is softened far beyondits maximum-strength condition. As shown schematically in Figure 15.5,the lattice strains are much more severe around a coherent precipitate356 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYS2004006002 4 6 8 10Wt % CuAl231Temperature, oCLiquid5.6αθFigure 15.1 Aluminum-rich side of Al–Cu phase diagram showing the three steps ofprecipitation hardening.Al 2 4 6 8Wt % Cu200400600Liquid5.6'''GPTemperature, oCαθθθFigure 15.2 Metastable solvus curves for GP, q≤, and q¢ in Al–Cu phase diagram. FromHornbogeni (1) and Beton and Rollason (2).(Figure 15.5b) than around an incoherent one (Figure 15.5c).The severe latticestrains associated with the coherent precipitate make the movement of dislocationsmore difficult and, therefore, strengthen the material to a greaterextent.Similar to Al–Cu alloys, the precipitation structure sequence may be representedas follows for other alloy systems (8):BACKGROUND 357Figure 15.3 Transmission electron micrograph of a 2219 aluminum heat treated tocontain q¢ phase. From Dumolt et al. (5).0.01 0.1 1.0 10 100 1000406080100120140Aging time, daysVickers hardness number'GP[1] 190 oC (374F)130 oC (266F)GP[2]θθFigure 15.4 Correlation of structures and hardness of Al–4Cu at two aging temperatures.From Silcock et al. (6).(15.2)(15.3)(15.4)where SS denotes the supersaturated solid solution.Table 15.3 shows the compositionsof the commercial aluminum alloys mentioned above (10). It shouldbe pointed out, however, that coherency strains are not observed in the GPzones or b¢ transition stages of precipitation in Al–Mg–Si alloys such as 6061.Al–Zn–Mg e.g., 7005 SS GP Zn2Mg Zn2Mg ( ): Æ Æh¢( )Æh( )Al–Mg–Si e.g., 6061 SS GP Mg2Si Mg2Si ( ): Æ Æb¢( )Æb( )Al–Cu–Mg e.g., 2024 SS GP S Al2CuMg S Al2CuMg ( ): Æ Æ ¢( )Æ ( )358 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSFigure 15.5 Three types of structure in Al–Cu precipitation hardening: (a) supersaturatedsolid solution; (b) coherent metastable phase; (c) incoherent equilibriumphase. From Guy (7).TABLE 15.3 Compositions of Some Heat-TreatableAluminum AlloysAlloy Si Cu Mn Mg Cr Ni Zn Ti2014 0.8 4.4 0.8 0.5 — — — —2024 — 4.4 0.6 1.5 — — — —2219 — 6.3 0.3 — — — — 0.066061 0.6 0.3 — 1.0 0.2 — — —7005 — — 0.4 1.4 0.1 — 4.5 0.047039 0.3 0.1 0.2 2.8 0.2 — 4.0 0.17146 0.2 — — 1.3 — — 7.1 0.06Source: Aluminum Standards and Data (10).Therefore, it has been suggested that precipitation hardening in suchaluminum alloys is due to the increased energy required for the dislocationsto break the Mg–Si bonds as they pass through the precipitate, rather thandue to coherency strains (4).Figure 15.6 shows the precipitation-hardening curves of 6061 aluminum (8).The alloy has been naturally aged at room temperature (T4) before heat treating.The initial strength decrease is due to reversion (dissolution) of the GPzones formed in natural aging.As shown, the higher the temperature, the fasteroveraging occurs and the strength decreases.15.2 Al–Cu–Mg AND Al–Mg–Si ALLOYS15.2.1 Welding in Artificially Aged ConditionThe 2000-series (Al–Cu–Mg) and 6000-series (Al–Mg–Si) heat-treatablealloys are known to have a tendency to overage during welding, especiallywhen welded in the fully aged condition (T6).Dumolt et al. (5) studied the HAZ microstructure of 2219 aluminum, abinary Al–6.3Cu alloy. Figure 15.7 shows the transmission electron micrographsof a 2219 aluminum plate artificially aged to contain only onemetastable phase, q¢, before welding and preserved in liquid nitrogen afterwelding to inhibit natural aging (5). Since the composition of alloy 2219 isbeyond the maximum solid solubility, large q particles are still present afterheat treating, but the matrix is still a containing fine q¢ precipitate. As shownin the TEM images, the volume fraction of q¢ decreases from the base metalto the fusion boundary because of the reversion of q¢ during welding. Thereversion of q¢ is accompanied by coarsening; that is, a few larger q¢ particlesAl–Cu–Mg AND Al–Mg–Si ALLOYS 359120oC(250oF)150oC(300oF)170oC(340oF)205oC(400oF)230oC(450oF)260oC(500oF)172.5207241.5310.52763452530354045500 0.01 0.1 1 10 100 103 104 1056061 aluminumDuration of precipitation heat treatment, hrTensile strength, MPaTensile strength, 1000 psiFigure 15.6 Aging characteristics for 6061-T4 aluminum (9). Modified from MetalsHandbook, vol. 2, 8th edition, American Society for Metals, 1964, p. 276.grow at the expense of many small ones (middle TEM image). The presenceof such coarse precipitates suggests overaging and hence inability to recoverstrength by postweld artificial aging, as will be discussed later.The microstructure in Figure 15.7 can be explained with the help of Figure15.8.The base metal is heat treated to contain the q¢ phase. Position 4 is heatedto a peak temperature below the q¢ solvus and thus unaffected by welding.Positions 2 and 3 are heated to above the q¢ solvus and partial reversion occurs.Position 1 is heated to an even higher temperature and q¢ is fully reverted.Thecooling rate here is too high for reprecipitation of q¢ to occur during coolingto room temperature. The q¢ reversion causes the hardness to decrease in theHAZ, which is evident in the as-welded condition (AW). During postweldnatural aging (PWNA), the GP zones form in the solutionized area near position1, causing its hardness to increase and leaving behind a hardness minimumnear position 2. During postweld artificial aging (PWAA), q≤ and some q¢ precipitatenear position 1 and cause its hardness to increases significantly. However,near position 2, where overaging has occurred during welding due toq¢ coarsening, the hardness recovery is not as much.A somewhat similar situation is welding a workpiece that has been heattreated to the T6 condition. Figure 15.9 shows the hardness profiles in a 3.2-mm-thick 6061 aluminum autogenous gas–tungsten arc welded in the T6condition at 10V, 110A, and 4.2mm/s (10ipm) (11). A hardness minimum isevident after PWNA and especially after PWAA. Malin (12) welded 6061-T6aluminum by pulsed GMAW with a filler metal of 4043 aluminum (essentiallyAl–5Si) and measured the HAZ hardness distribution after postweld natural360 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSFigure 15.7 Transmission electron micrographs of a 2219 aluminum artificially agedto contain q¢ before welding. From Dumolt et al. (5).aging. He observed a hardness minimum similar to the PWNA hardness profilein Figure 15.9, where the peak temperature was 380°C (716°F) during weldingand where failure occurred in tensile testing after welding. He pointed outthat the precipitation range for the most effective strengthening phase b¢¢Al–Cu–Mg AND Al–Mg–Si ALLOYS 3614AWPWNAPWAAHardnessDistancefrom weldTemperatureConcentration Time Time(a) (b) (c)(d)(e)WeldHAZ edge1 213'' reversion4' '3'2α θ θ θθθαFigure 15.8 Al–Cu alloy heat treated to contain q ¢ before welding: (a) phase diagram;(b) thermal cycles; (c) reversion of q ¢; (d) microstructure; (e) hardness distribution. qin base metal not shown.Postweld artificial aging(155 C, 18 hours)Postweld natural aging(7 days)Right after weldingDistance from fusion line, mm0 15 3050100Knoop hardness (500g)oFigure 15.9 HAZ hardness profiles in a 6061 aluminum welded in T6 condition. FromKou and Le (11).is 160–240°C (320–464°F) and that for the less effective strengthening phaseb¢ is 240–380°C (464–716°F) (13, 14). He proposed that the losses of hardnessand strength is a result of overaging due to b≤ coarsening and b¢ formation.Malin also observed a sharp hardness decrease immediately outside the fusionboundary (PMZ) and speculated that this was caused by Mg migration intothe Mg-poor weld.Rading et al. (15) determined the fully naturally aged HAZ hardness profilesin a 9.5-mm-thick 2095 aluminum (essentially Al–4.3Cu–1.3Li) welded inthe peak aged T8 condition with a 2319 filler metal (Al–6.3Cu), as shown inFigure 15.10.The T8 condition stands for solution heat treating, cold working,followed by artificially aging.The principal strengthening precipitate in the T8condition is T1 (Al2CuLi) while in the naturally aged (T4) condition strengtheningis provided mainly by d¢ (Al3Li) (16). The hardness profiles in Figure15.10 are similar to the PWNA hardness profile in Figure 15.9 except for thesharp decrease near the fusion line (FL). According to TEM micrographs, thehardness minimum at 5 mm from the fusion line is due to overaging caused byT1 coarsening during welding, while the hardness peak at 2 mm from the fusionline is caused by d¢ precipitation during natural aging. It was proposed that thesharp hardness decrease near the fusion boundary is caused by diffusion of Liinto the Li-poor weld, which reduces the propensity for d¢ precipitation.15.2.2 Welding in Naturally Aged ConditionFigure 15.11 shows TEM micrographs of a 2219 aluminum heat treated tocontain GP zones alone before welding and preserved in liquid nitrogen after362 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSWM HAZ BMLine1Line2Line3Line4FLI II III IV-5 0 5 10 15 20 2550100150250200Hardness (HV1)Distance from fusion line (mm)Figure 15.10 HAZ hardness profiles in a 2095 aluminum welded in T8 conditionand measured after natural aging. Reprinted from Rading et al. (15). Courtesy ofAmerican Welding Society.welding to inhibit natural aging (5). The GP zones in the HAZ are easilyreverted during welding because of their small size. Precipitation of q¢ occursin the middle of the HAZ. Figure 15.12 shows the optical micrograph of a 2024aluminum (Al–4.4Cu–1.5Mg) plate that was welded in the T4 (naturally aged)condition (17).The precipitation region in the HAZ is visible as a dark-etchingband.The microstructure in Figure 15.11 can be explained with the help of Figure15.13. Positions 1–3 are heated to above the solvus of the GP zones, and GPzones are thus reverted. Since position 2 is heated to a maximum temperaturewithin the precipitation temperature of q¢, q¢ precipitates and causes a smallhardness peak right after welding (AW), as shown in Figure 15.13e. Precipitationof q ≤, however, is not expected since the time typical of a welding cycleis not sufficient for its formation (18). During PWNA, the hardness increasesslightly in the solutionized area at position 1 because of the formation of theGP zones. During PWAA, the hardness increases significantly in both this areaand the base metal because of q≤ and q¢ precipitation. The hardness recovery,however, is not as good near position 2, where some overaging has occurredduring welding due to q¢ precipitation.Figure 15.14 shows the results of hardness measurements in a 3.2-mm-thick6061 aluminum gas–tungsten arc welded in the T4 condition at 10V, 110A, and4.2 mm/s (10 ipm) (11).A small peak appears in the as-welded condition, whichis still visible after PWNA here but may be less clear in other cases. Similarresults have been reported by Burch (19).Al–Cu–Mg AND Al–Mg–Si ALLOYS 363Figure 15.11 Transmission electron micrographs of a 2219 aluminum aged to containGP zones before welding. From Dumolt et al. (5).Figure 15.12 Precipitation zone in HAZ of a 2024 aluminum welded in naturally agedcondition (weld metal at upper right corner). Reprinted from Arthur (17). Courtesy ofAmerican Welding Society.GP1241 2 4AWPWNAPWAAHardnessDistance from weldGP' precipitationGP reversionTemperatureConcentration Time Time(a) (b) (c)(d)(e) HAZ edgeWeld33"''' precipitationC-curveα θθαθθθαFigure 15.13 Al–Cu alloy heat treated to contain GP zones before welding: (a) phasediagram; (b) thermal cycles; (c) precipitation C curves; (d) microstructure; (e) hardnessdistribution. q in base metal not shown.364 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSFigure 15.15 shows the results of hardness measurements in a 6061aluminum (20). It suggests that welding a heat-treatable 6000- or 2000-seriesalloy in the T6 (artificially aged) condition can result in severe loss of strength(hardness) due to overaging. For this reason, welding in the T4 condition isoften preferred to welding in the T6 condition (19, 20).15.2.3 Effect of Welding Processes and ParametersThe loss of strength in the HAZ can be significantly affected by the weldingprocess and by the heat input and welding speed. Figure 15.16 shows that asAl–Cu–Mg AND Al–Mg–Si ALLOYS 365Distance from fusion line, mm0 15 3050100Knoop hardness (500g)Right after weldingPostweld artificialaging (155 C, 18hours)Postweld naturalaging (7 days)oFigure 15.14 HAZ hardness profiles in a 6061 aluminum gas–tungsten arc welded inT4 condition. From Kou and Le (11).Distance from fusion line, cm0 0.5 1.0 1.5 2.0 2.50.2 0.4 0.6 0.8 1.0Distance from fusion line, in0Hardness, DPH (500g)60708090100110T4, Postweld AAT6, Postweld AAT6, Postweld NAT4, Postweld NAFigure 15.15 HAZ hardness profiles in 6061 aluminum welded in T4 or T6 andpostweld naturally or artificially aged. Reprinted from Metzger (20). Courtesy ofAmerican Welding Society.compared to variable-polarity PAW, LBW results in significantly less strengthloss in the HAZ of a 2195-T8 aluminum (Al–Cu–Li) (21). Again, T8 stands forsolution heat treating, cold working, followed by artificially aging. As shownin Figure 15.17, the higher the heat input per unit length of the weld (the higherthe ratio of power input to welding speed), the wider the HAZ and the moresevere the loss of strength (19). Therefore, the heat input should be limitedwhen welding heat-treatable aluminum alloys.Apparently, full strength can be recovered if the entire workpiece is solutionized,quenched, and artificially aged after welding. The preweld condition366 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSBase metal mean = 614FZVPPAWLBWDistance from interface (mm)Yield strength (MPa)-5 0 5 10 15 20 25200250300350400450500550600650Figure 15.16 HAZ strength distributions in 2195-T8 aluminum made by LBW andPAW. Reprinted from Martukanitz and Howell (21).Distance, cm21 mm/s; 200 J/mm8.5 mm/s; 315 J/mm1.7 mm/s; 900 J/mm00Distance, inches1 2 350 2.5 5.0 7.560708090Hardness, Rockwell FFigure 15.17 HAZ hardness profiles in 6061-T4 aluminum after postweld artificialaging. Reprinted from Burch (19). Courtesy of American Welding Society.can be either solution heat treated or fully annealed, although the latter issomewhat inferior due to its poor machinability (18). However, for largewelded structures heat-treating furnaces may not be available. In addition,distortion of the welded structure developed during postweld solution heattreating and quenching may be unacceptable.15.3 Al–Zn–Mg ALLOYSThe age hardening of Al–Zn–Mg alloys is quite different from that of either6000- or 2000-series aluminum alloys. As shown in Figure 15.18, alloy 7005(Al–4.5Zn–1.2Mg) ages much more slowly than alloy 2014 (Al–4.5Cu–0.6Mg)(22). In fact, Al–Zn–Mg alloys, such as 7005, 7039, and 7146, age much moreslowly than either 6000- or 2000-series aluminum alloys. As a result,Al–Zn–Mg alloys have a much smaller tendency to overage during weldingthan the other alloys. Furthermore, unlike 6000- or 2000-series aluminumalloys, Al–Zn–Mg alloys recover strength slowly but rather significantly bynatural aging. For these reasons Al–Zn–Mg alloys are attractive when postweldheat treatment is not practical. Figure 15.19 shows the hardness profilesin alloy 7005 after welding (23). Welding such an alloy in its naturally agedcondition is most ideal since the strength in the HAZ can be recovered almostAl–Zn–Mg ALLOYS 36720oC150oC 100oC200oC100oC20oC120oC130oC15 30 60 120 18030343842607080901000 1 10 100 1000Time, hrTime, hrTensile strengthb x 9.8, N/mm 2Hardness, Rockwell F(b)(a)σFigure 15.18 Aging characteristics of heat-treatable aluminum alloys: (a)Al–4.5Cu–0.6Mg quenched from 500°C; (b) Al–4.5Zn–1.2Mg quenched from 450°C(22).368 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSFigure 15.19 HAZ hardness profiles in Al–4.5Zn–1.2Mg alloy: (a) naturally agedbefore welding (1 : 3 hour, 2: 4 days, 3: 30 days, 4: 90 days); (b) artificially aged at 130°Cfor 1 h before welding. From Mizuno (23).Postweld natural aging (8 days)4 hours after welding0 10 20 30Distance from fusion line, mmKnoop hardness (500g)75100125Figure 15.20 HAZ hardness profiles in 7146 aluminum naturally aged before welding.From Kou and Le (11).completely by postweld natural aging. Figure 15.20 shows similar results inalloy 7146 (Al–7.1Zn–1.3Mg) (11).Similar to the welding of 6000- or 2000-series alloys, excessive heat inputsshould be avoided in welding Al–Zn–Mg alloys in the artificially aged condition.Figure 15.21 shows the hardness profiles in the HAZ of alloy 7039 artificiallyaged before welding (24).As shown, the loss of strength can be reducedsignificantly by increasing the number of passes (thus decreasing the heat inputin each pass) and maintaining a low interpass temperature.Welding a heat-treatable aluminum alloy in the annealed condition isalmost the opposite of welding it in the aged condition. Figure 15.22 shows thehardness profiles in an annealed heat-treatable aluminum alloy after welding(11). The base metal is relatively soft because of annealing. The HAZ is solutionizedduring welding and thus gets stronger by solution strengthening,as evident from the hardness increase in the HAZ after welding. The furtherAl–Zn–Mg ALLOYS 369Figure 15.21 HAZ hardness profiles in 3-cm-thick artificially aged 7039 aluminum:(a) 4 passes, continuous welding; (b) 16 passes, 150°C interpass temperature. Reprintedfrom Kelsey (24). Courtesy of American Welding Society.hardness increases in the HAZ after natural aging are consistent with theremarkable ability of Al–Zn–Mg alloys to gain strength by natural aging.15.4 FRICTION STIR WELDING OF ALUMINUM ALLOYSFriction stir welding is a solid-state joining process developed at the WeldingInstitute (25). As shown in Figure 15.23a, a rotating cylindrical tool with aprobe is plunged into a rigidly clamped workpiece and traversed along thejoint to be welded.Welding is achieved by plastic flow of frictionally heatedmaterial from ahead of the probe to behind it. For welding aluminum alloys,the tool is usually made of tool steel. As shown in Figure 15.23b, the resultantweld consists of three zones: thermally affected zone, thermomechanicallyaffected zone, and dynamically recrystallized zone. In the thermally affectedzone the grain structure is not affected by welding. In the thermomechanicallyaffected zone, however, the grains are severely twisted. In the dynamicallyrecrystallized zone, which is also called the weld nugget, all old grains370 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYSDistance from fusion line, mmKnoop hardness (500g)0 10 20 30751001254 hours after weldingPostweld naturalaging (8 days)Postweld natural aging(30 days)Figure 15.22 HAZ hardness profiles in alloy 7146 welded in annealed condition. FromKou and Le (11).friction stir welddynamicallyrecrystallized zonethermomechanicallyaffected zonethermally affected zone(b)workpieceproberotatingtoolweldingdirectionjoint(a)toolshoulderFigure 15.23 Friction stir welding: (a) process; (b) transverse cross section ofresultant weld.disappear and numerous small new grains recrystallize. As in the HAZ infusion welds of precipitation-hardened aluminum alloys, heating duringwelding can cause considerable loss of strength. Reversion of precipitate,overaging, and solutionizing occur during welding, from near the base metalto the weld centerline. Figure 15.24 shows a typical hardness distribution inprecipitation-hardened aluminum alloys (26).REFERENCES1. Hornbogen, E., Aluminum, 43(part 11): 9, 1967.2. Beton, R. H., and Rollason, E. C., J. Inst. Metals, 86: 77, 1957–58.3. Nutting, J., and Baker, R. G., The Microstructure of Metals, Institute of Metals.London, 1965, pp. 65, 67.4. Smith, W. F., Structure and Properties of Engineering Alloys, McGraw-Hill, NewYork, 1981.5. Dumolt, S. D., Laughlin, D. E., and Williams, J. C., in Proceedings of the First InternationalAluminum Welding Conference, Welding Research Council, New York,p. 115.6. Silcock, J. M., Heal, J. J., and Hardy, H. K., J. Inst. Metals, 82: 239, 1953.7. Guy, A. G., Elements of Physical Metallurgy, Addison-Wesley, Reading, MA, 1959.8. Hundicker, H.Y., in Aluminum, Vol. 1, American Society for Metals, Metals Park,OH, 1967, Chapter 5, p. 109.9. Metals Handbook, vol. 2, 8th edition, American Society for Metals, Metals Park,OH, 1964, p. 276.10. Aluminum Standards and Data, Aluminum Association, New York, 1976, p. 15.11. Kou, S., and Le.Y., unpublished research, Carnegie-Mellon University, Pittsburgh,PA, 1982.12. Malin,V., Weld. J., 74: 305s, 1995.13. Panseri, C., and Federighi, T., J. Inst. Metals, 94: 94, 1966.14. Miyauchi, T., Fujikawa, S., and Hirano, K., J. Jpn. Inst. Light Metals, 21: 595, 1971.15. Rading, G. O., Shamsuzzoha, M., and Berry, J. T., Weld. J., 77: 411s, 1998.REFERENCES 371Distance from weld centerline, mm160140120100-15 -10 -5 0 5 10 15Vickers microhardness(VHN)2024 AlFigure 15.24 Hardness profile across friction stir weld of an artificially aged alloy 2024(26). The arrows indicate the weld nugget.16. Langan, T. J., and Pickens, J. R., in Aluminum-Lithium Alloys, Vol. II, Eds. T. H.Sanders, Jr. and E. A. Starke, Jr., Materials and Component Engineering Publications,Birmingham, UK, p. 691.17. Arthur, J. B., Weld. J., 34: 558s, 1955.18. Introductory Welding Metallurgy, American Welding Society, Miami, FL, 1968,p. 65.19. Burch,W. L., Weld. J., 37: 361s, 1958.20. Metzger, G. E., Weld. J., 46: 457s, 1967.21. Martukanitz, R. P., and Howell, P. R., in Trends in Welding Research, Eds. H. B.Smartt, J. A. Johnson, and S. A. David, ASM International, Materials Park, OH,1996, p. 553.22. Principles and Technology of the Fusion Welding of Metals, Vol. 2, MechanicalEngineering Publishing Co., Peking, China, 1981 (in Chinese).23. Mizuno, M., Takada, T., and Katoh, S., J. Japanese Welding Society, vol. 36, 1967,pp. 74–81.24. Kelsey, R. A., Weld. J., 50: 507s, 1971.25. Dawes, C. J., Friction Stir Welding of Aluminum, IIW-DOC XII-1437-96, 1996,pp. 49–57.26. Murr, L. E., Li, Y., Trillo, E. A., Nowak, B. M., and McClure, J. C., Alumin. Trans.,1(1), 141–154, 1999.27. A. Umgeher and H. Cerjak, in Recent Trends in Welding Science and Technology,Eds. S. A. David and J. M. Vitek, ASM International, Materials Park, OH, 1990,p. 279.FURTHER READING1. Aluminum, edited by J. E. Hatch, American Society for Metals, Metals Park, OH,1984.2. Mondolfo, L. F., Aluminum Alloys: Structure and Properties, Butterworths, London,1976.3. Polmear, I. J., Light Alloys, Edward Arnold, London, 1981.4. Smith, W. F., Structure and Properties of Engineering Alloys. McGraw-Hill, NewYork, 1981.PROBLEMS15.1 A reciprocal relationship between the tensile strength of 2219 aluminum(essentially Al–6.3Cu) weldments and the heat input per unitlength of weld per unit thickness has been observed. Explain why.15.2 (a) Based on the results of mechanical testing for 2219 aluminumgiven in Table P15.2, comment on the effect of postweld heat treatmenton the tensile strength.372 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYS(b) The bend angle (i.e., the maximum angle the specimen can be bentprior to failure) is an indication of the ductility of the material.Does the weld ductility seem to recover as effectively as the tensilestrength? Explain why or why not.15.3 A 12.7-mm-thick plate of 6061-T6 aluminum (TL = 652°C) was gas–tungsten arc welded with DC electrode negative. The welding parameterswere I = 222A, E = 10.4V, and V = 5.1mm/s. Microhardnessmeasurements after welding indicated that softening due to overagingstarts about 5.3 mm from the fusion line and gradually increases as thefusion line is approached. Thermal measurements during weldingrevealed a peak temperature of about 300°C at the position where softeningstarted. Calorimetric measurements showed that the arc efficiencywas around 80%. How does the width of the HAZ compare with thatpredicted from Adams’s equation (Chapter 2)?15.4 Figure P15.4 show the hardness distributions measured by Umgeher andCerjak (27) in 7075 aluminum after natural aging three months at roomtemperature (pwna), after artificial aging (pwaa), and after full postweldPROBLEMS 373TABLE P15.2 2219 AluminumTensile Strength Bend AngleTest Specimen Procedure (N/mm2) (deg)Base metal SS 320 180SS + AA 412 121Weldment SS + welding 254 64SS + AA + welding 287 54SS + welding + AA 300 44SS + welding + SS + AA 373 84AN + welding + SS + AA 403 93Abbreviations: SS, solid solution; AA, artificial aging;AN, annealing.0 10 20 30 4080120160distance from fusion line, mmpeak temperature, oChardness, HV 56004002000pwnapwaapwshpeak tempFigure P15.4heat treatment of solutionizing, quenching, and then artificial aging(pwsh). Indicate the temperature ranges in which overaging and solutionizingoccur during welding. Explain how these hardness distributionscompare with each other.15.5 Al–Li–Cu alloy 2095 was welded by LBW, GTAW, and GMAW and theHAZ hardness profiles of the resultant welds were measured. Rank thewelds in the order of increasing hardness in the HAZ.15.6 Al–Li–Cu alloy 2090 was welded with various filler metals such as 2319,2090, 4047, and 4145. Joint efficiencies up to 65% of base-metal strengthwere obtained in the as-welded condition. After postweld solution heattreatment and artificial aging, joint efficiencies up to 98% wereobtained. Explain why.15.7 Sketch the hardness profiles in the following aluminum alloys afterfriction stir welding: (a) artificially aged 2219; (b) work-hardened 5083.374 PRECIPITATION-HARDENING MATERIALS I: ALUMINUM ALLOYS16 Precipitation-HardeningMaterials II: Nickel-Base AlloysBecause of their high strength and good corrosion resistance at hightemperatures, Ni-base alloys have become the most extensively used hightemperaturealloys.Table 16.1 summarizes typical welding problems in Ni-basealloys and recommended solutions. The problems associated with the fusionzone and the PMZ have been discussed previously. In this chapter, we shallfocus on weakening the HAZ and postweld heat treatment cracking in heattreatableNi-base alloys.16.1 BACKGROUNDTable 16.2 shows the chemical compositions of several representativeheat-treatable Ni-base alloys (1, 2). Aluminum and Ti are the two major precipitation-hardening constituents in heat-treatable Ni-base alloys, equivalentto Cu in heat-treatable Al–Cu alloys. As can be seen in Figure 16.1, the solubilityof Ti or Al in the g phase increases significantly with increasing temperature—a necessary criterion for precipitation hardening as in Al–Cu alloys (3).Like heat-treatable Al alloys, the precipitation hardening of heattreatableNi-base alloys can be obtained by solutionizing at temperaturesabove the solvus, followed first by water quenching and then by artificial agingin the precipitation temperature range. In mill practice, however, the alloys areusually air cooled from the solutionizing temperature (usually in the range1040–1180°C, or 1900–2150°F) to an intermediate aging temperature and heldthere for a number of hours before being further air cooled to a final agingtemperature of about 760°C (1400°F). After aging at this final temperature forabout 16 h, the alloys can be air cooled to room temperature. For best resultssome alloys are aged at two, rather than one, intermediate temperatures. Forexample, both Udimet 700 and Astroloy are sometimes aged first at 980°C(1975°F) for 4 h, then at 815°C (1500°F) for 24 h before the 16-h final agingat 760°C (1400°F). For applications at low temperatures, the aging operationcan be carried out solely at 760°C (1400°F) to avoid grain boundary carbideprecipitation (2, 4).The precipitation reaction to form the strengthening phase g ¢ can be writtenas follows (2):375Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4376 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSTABLE 16.1 Typical Problems in Welding Nickel-Base AlloysSections inTypical Problems Alloy Types Solutions BookLow strength in Heat-treatable Resolution and artificial aging 16.2HAZ alloys after weldingReheat cracking Heat-treatable Use less susceptible gradealloys (Inconel 718)Heat treat in vacuum or inertatmosphereWelding in overaged condition 16.3(good for Udimet 500)Rapid heating through criticaltemperature range, if possibleHot cracking in PMZ All types Reduce restraint 13.1Avoid coarse-grain structure 13.2and Laves phaseTABLE 16.2 Composition of Heat-Treatable Nickel-Base SuperalloysAlloy C Cr Co W Mo Al Ti OthersInconel X-750 0.04 16 — — — 0.6 2.5 7Fe, 1CbWaspaloy 0.07 19 14 — 3 1.3 3.0 0.1ZrUdimet 700 0.10 15 19 — 5.2 4.3 3.5 0.02BInconel 718 0.05 18 — — 3 0.6 0.9 18Fe, 5CbNimonic 80A 0.05 20 <2 — — 1.2 2.4 <5FeMar-M200 0.15 9 10 12 — 5.0 2.0 1Cb, 2HfRene 41 0.1 20 10 — 10 1.5 3.0 0.01BSource: Owczarski (1) and Sims (2).(16.1)where g is a fcc matrix while g ¢, the precipitate, is an ordered fcc intermetalliccompound. The Ni3(Al,Ti) is only the abbreviation of g ¢; the exact compositionof g ¢ can be much more complicated. For example, the compositions ofg ¢ in Inconel 713C and IN-731 have been determined to be (5)(Ni0.980Cr0.004Mo0.004)3(Al0.714Cb0.099Ti0.048Mo0.038Cr0.103)g¢ in Inconel 713Cand(Ni0.884Co0.070Cr0.032Mo0.008V0.003)3(Al0.632Ti0.347V0.013Cr0.006Mo0.002)g¢ in IN-731Ni, Cr, Co, Mo, Al, Ti Ni Al, Ti Cr, Co, Momatrix components3 ( )Æ ( )¢+ ( )g g14444244443 14243 1442443The precipitate g ¢ can assume several different shapes, such as spherical,cubical, and elongated. Figure 16.2 shows examples of cubical and spherical g ¢precipitates in two different Ni-base alloys (6). It is interesting to note that,after the general precipitation of the dominant g ¢ particles, very fine g ¢ particlescan further precipitate during cooling to room temperature. Such g ¢, called“cooling g ¢,” often generates roughening of the g matrix. It should be pointedout here that the precipitation of g ¢ depletes the surrounding g matrix of Aland Ti and results in a decrease in the lattice parameter of the matrix. Thisdecrease creates the so-called aging contraction, which has been reported tobe of the order of 0.1% (0.001 in./in.) in Rene 41 and 0.05% (0.0005 in./in.) inInconel X-750 (7). As will be mentioned later in this chapter, the contractionBACKGROUND 377Figure 16.1 Effect of alloying elements on the solvus temperature of g ¢: (a) Ti; (b) Al.From Betteridge (3).Figure 16.2 g ¢ in Ni-base alloys: (a) cubical g ¢ in IN-100 (magnification 13,625¥);(b) spherical and cooling g ¢ in U500 (magnification 5450¥). From Decker and Sims (6).strains so created hinder the relaxation of residual stresses in the HAZ and,therefore, promote the chance of postweld heat treatment cracking.In most Ni-base alloys the high-temperature carbide MC can react with theg matrix and form lower carbides, such as M23C6 and M6C, according to the followingreactions (6):(16.2)and(16.3)In alloys such as Udimet 700 the M23C6 carbide formed by reaction (16.2)appears as blocky carbide lining the grain boundaries. The g ¢ phase, on theother hand, envelops the M23C6 carbide along the grain boundaries, as shownschematically in Figure 16.3. If the M23C6 carbide develops in a brittle, cellularform rather than a hard, blocky form, the ductility and rupture life of thealloys are reduced. Since alloys that generate profuse g ¢ at grain boundariesappear to be resistant to cellular M23C6, grain boundary g ¢ formed by reaction(16.2) may play an important role in blocking its growth. The hard, blockyM23C6 carbide may initially strengthen the grain boundary beneficially. Ultimately,however, such M23C6 particles are the sites of the initiation of rupturefracture (2). In alloys such as Nimonic 80A and Inconel-X, the grain boundaryg ¢ has not been noted as a product of reaction (16.2). In fact, as shown inFigure 16.3, the areas adjacent to the grain boundary are depleted of g ¢. Thiscan be caused by diffusion of Cr to form grain boundary carbides. Since theseareas are depleted in Cr, their solubility for Ni and Al increases, thus causingthe disappearance of g ¢ (2).Ti, Mo CMCNi, Co, Al, Ti Mo Ni, Co CM C3 Ni Al, Ti63 ( ) +( )Æ ( ) + ( )¢14243 1442443 1442443 14243g g3Ti, Mo CMCNi, Cr, Al, Ti Cr Mo CM C21 2 Ni Al, Ti23 63 ( ) +( )Æ + ( )¢14243 1442443 1442443 14243g g6378 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSFigure 16.3 Schematic sketch of microstructure observed in some Ni-base superalloys.From Decker and Sims (6).From the above discussion it is clear that the high strength of heattreatableNi-base alloys is due primarily to the precipitation hardening of g ¢and the resistance to grain boundary sliding provided by carbides. However,Inconel 718 is an exception. It utilizes niobium (Nb) as its primary strengtheningalloying element, and g ≤ (an ordered bct intermetallic compound ofcomposition Ni3Nb) rather than g ¢ is responsible for precipitation hardeningduring aging.In addition to g ¢, g ≤, and carbides, a group of phases called topologicallyclose packed phases can also be present in certain Ni-base alloys where compositioncontrol has not been carefully watched (6). Such phases, for instance,s and m, often appear as hard, thin plates and thus promote lowered rupturestrength and ductility. However, in most Ni-base alloys such undesired phasesdo not usually appear, unless significant alteration of the matrix compositionhas occurred as a result of extensive exposure in the aging temperature range.Therefore, they are not of great concern during welding (4). Nevertheless, itshould be pointed out that the presence of the Laves phase, which is also atopologically close packed phase, has been reported to promote hot crackingin Inconel 718 and A-286 due to its lower melting point (8–10).Like heat-treatable Al alloys, heat-treatable Ni-base alloys can also overage.As seen in Figure 16.4a, the optimum aging temperature for Inconel X-750 isaround 760°C (1400°F), above which it tends to overage. The aging characteristicsof several heat-treatable Ni-base alloys at this temperature are shownin Figure 16.4b. Inconel 718, which is precipitation hardened by g ≤, ages muchmore slowly than other alloys. As will be discussed later in this chapter, theslow aging characteristic of Inconel 718 makes it more resistant to crackingduring postweld heat treatment.16.2 REVERSION OF PRECIPITATE AND LOSS OF STRENGTHConsider welding a heat-treatable Ni-base alloy in the aged condition, asshown in Figure 16.5. The area adjacent to the weld is heated above the precipitationtemperature range of g ¢. Reversion of g ¢ ranges from partial reversionnear the edge of the HAZ (point 2) to full reversion near the fusionboundary (point 1).This g ¢ reversion causes loss of hardness or strength in theHAZ.16.2.1 MicrostructureOwczarski and Sullivan (13) studied the reversion of the strengthening precipitatesin the HAZ of Udimet 700 during welding. This material was in thefull aged condition produced by the following heat treatment before welding:1165°C/4 h + air cool (solution)1080°C/4 h + air cool (primary age)REVERSION OF PRECIPITATE AND LOSS OF STRENGTH 379845°C/4 h + air cool (intermediate age)760°C/16 h + air cool (final age)The resultant HAZ microstructure is shown in Figure 16.6. The unaffectedbase metal consists of coarse angular g ¢ and fine spherical g ¢ between thecoarse g ¢ (Figure 16.6a). The initial stage of reversion just inside the HAZ ischaracterized by the disappearing of the fine g ¢ and the rounding of the coarseangular g ¢ (Figure 16.6b). Further reversion of coarse g ¢ is evident in themiddle of the HAZ (Figure 16.6c). Since this material is highly alloyed with380 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSFigure 16.4 Aging characteristics of Ni-base alloys. (a) Inconel X-750. FromEiselstein (11). (b) Some other Ni-base alloys. Reprinted from Wilson and Burchfield(12). Courtesy of American Welding Society.Ti and Al, the areas near the reverted coarse g ¢ become so supersaturated withTi and Al that finer g ¢ reprecipitates during cooling. This localized supersaturationis due to the fact that the retention time at high temperatures is tooshort to allow homogenization to occur in this region. Reversion continuestoward completion as the fusion boundary is approached (Figure 16.6d). Theprior sites of coarse g ¢ particles are marked by a periodic pattern of very fineg ¢ reprecipitated during cooling. The ultimate solution and distribution of g ¢occur in the weld metal itself (Figure 16.6e).The weld metal contains a uniformdistribution of fine g ¢ precipitate.Figure 16.7 shows the HAZ microstructure after 16 h postweld heat treatmentat 760°C (1400°F). Reprecipitation of very fine g ¢ occurs in the regionwhere coarse g ¢ has begun to dissolve during welding (Figure 16.6b). This iscaused by the precipitation of the elements that were taken into solutionduring welding.16.2.2 Hardness ProfilesLucas and Jackson (14) and Hirose et al. (15) measured hardness profilesacross welds of Inconel 718. Figure 16.8 shows the hardness distributions ofREVERSION OF PRECIPITATE AND LOSS OF STRENGTH 381ConcentrationTemperaturematrix123123 4Distance from weldweldTime Time' precipitation(a) (b) (c)(d)(e)12344HAZ edge' reversionpartial reversion of 'full reversion of '' precipitate'HardnessprecipitationC-curvesolvusγγγγγγ + γγγFigure 16.5 Reversion of g ¢ in HAZ: (a) phase diagram; (b) thermal cycles; (c) precipitationC curve; (d) microstructure; (e) hardness distribution.382 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSFigure 16.6 Microstructure of Udimet 700 weld: (a) as-received material (magnification10,000¥); (b) initial solution of fine g ¢; (c) further solution of coarse g ¢; (d)advanced stage of solution of coarse g ¢; (e) weld metal containing fine g ¢. (b–d). Magnification15,000¥. Reprinted from Owczarski and Sullivan (13). Courtesy of AmericanWelding Society. Reduced to 84% in reproduction.Hirose et al. (15) in Inconel 718 laser and gas–tungsten arc welded in the asweldedcondition. The solutionized and laser-welded (SL) workpiece was notmuch affected by welding.The workpiece solutionized, aged, and laser welded(AL), however, became much softer in the HAZ. This is because of precipitatereversion in the HAZ and the fusion zone. The HAZ is much narrowerin the laser weld (AL) than in the gas–tungsten arc weld (AT) because of thelower heat input used in the former. Aging after welding helped achieve themaximum hardness, either aged after welding or solutionized and then agedafter welding, as shown in Figure 16.9.REVERSION OF PRECIPITATE AND LOSS OF STRENGTH 383Figure 16.7 HAZ of Udimet 700 showing reprecipitation of fine g ¢ in region wherecoarse g ¢ begins to dissolve. Magnification 16,000¥. Reprinted from Owczarski andSullivan (13). Courtesy of American Welding Society.Weld metalSLALVickers hardness,Hv (load: 1.96N)0 2 4 6 8 10200250300350400450500(a)HAZWeldmetalSTAT0 2 4 6 8 10(b)HAZDistance from weld centerline, mmVickers hardness,Hv (load: 1.96N)200250300350400450500Distance from weld centerline, mmFigure 16.8 Hardness profiles in Inconel 718 welds in as-welded condition: (a) laserwelds; (b) gas–tungsten arc welds. S: solutionized; A: aged after solutionization; L: laserwelded; T: gas–tungsten arc welded. Broken line indicates fusion line. From Hiroseet al. (15).16.3 POSTWELD HEAT TREATMENT CRACKINGCracking can occur during the postweld heat treatment of heat-treatable Nibasealloys. Hot cracking in both the fusion zone and the partially melted zoneof heat-treatable Ni-base alloys (8, 14, 16–28) are similar to those in othermaterials (Chapters 11–13) and will not be discussed separately here.16.3.1 Reasons for Postweld Heat TreatmentHeat-treatable Ni-base alloys are often postweld heat treated for two reasons:(i) to relieve stress, and (ii) to develop the maximum strength. To develop itsmaximum strength, the weldment is first solutionized and then aged. Duringsolutionization the residual stresses in the weldment are also relieved. Theproblem is that aging may occur in the weldment while it is being heated upto the solutionization temperature because the aging temperature range isbelow the solutionization temperature. Since this aging action occurs beforethe residual stresses are relieved, it can cause cracking during postweld heattreatment. Such postweld heat treatment cracking is also called strain-age384 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSWeldmetalSLAALADistance from weld centerline (mm)Vickers hardness, Hv(load: 1.96N)0 2 4 6 8 10200250300350400450500(a)ALSAWeldmetalSTAATA0 2 4 6 8 10(b)ATSADistance from weld centerline (mm)Vickers hardness, Hv(load: 1.96N)200250300350400450500Figure 16.9 Hardness profiles in Inconel 718 welds after postweld heat treating: (a)laser welds; (b) gas–tungsten arc welds. S: solutionized; A: aged after solutionization;L: laser welded; T: gas–tungsten arc welded. Broken line indicates fusion line. FromHirose et al. (15).cracking or simply reheat cracking. The term strain-age cracking arises fromthe fact that cracking occurs in highly restrained weldments, as they are heatedthrough the temperature range in which aging occurs.16.3.2 Development of CrackingFigure 16.10 shows the development of postweld heat treatment cracking.Theprecipitation temperature range is from T1 to T2 (Figure 16.10a). To relievethe residual stresses after welding, the workpiece is brought up to the solutionizationtemperature (Figure 16.10b). It passes through the precipitationtemperature range. Unless the heating rate is high enough to avoid intersectingthe precipitation C curve, precipitation and hence cracking will occur(Figure 16.6c). The microstructural changes in the HAZ are illustrated inFigures 16.10d and e.Postweld heat treatment cracks usually, though not always, initiate in theHAZ. However, as shown in the weld circle-patch test in Figure 16.11a, it canpropagate into regions unaffected by the welding heat (4). Such a test, whichis often used for evaluating the strain-age cracking tendency of a material, isachieved by welding the circle-patch specimen to a stiffener (strong back) sothat the combination can be heat treated without relaxation (stress relief) dueto mechanical factors. As shown in Figure 16.11b, the cracks in the HAZ areintergranular (29).POSTWELD HEAT TREATMENT CRACKING 385matrixresidual stresses + agingcrackingsolutionizingagingprecipitationtemperatureresidualstressesTime, t Time, tweldingFZT T(a) (b) (c)Concentration, CCot1 t2 t3234 21 1T1T2(d) a b'Temperature, Tprecipitate(e)t1 t2 t4 t3heatingt4coolingheatingprecipitationC-curvegg + grangeFigure 16.10 Postweld heat treatment cracking: (a) phase diagram; (b) thermal cyclesduring welding and heat treating; (c) precipitation C curve; (d) weld cross-section;(e) changes in microstructure.16.3.3 Effect of CompositionFigure 16.12 shows the effect of Al and Ti contents on the postweld heat treatmentcracking tendency in Ni-base alloys (30). Such a plot was first proposedby Prager and Shira (4). As can be seen, the g ¢-strengthened Ni-base alloyswith high Al and Ti contents are particularly difficult to weld because of high386 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYSFigure 16.11 Postweld heat treatment cracking. (a) Circular-patch specimen of Rene41. From Prager and Shira (4). (b) Scanning electron micrograph showing intergranularcracking. Reprinted from McKeown (29). Courtesy of American Welding Society.Inconel 909Inconel 718Rene` 62 Inconel X-750M 252Rene` 41Inconel XWaspaloyIN939Unitemp 1753Udiment500Inconel 700GMR 235AstroloyUdimet 700Udimet 600AF2-1DAMar-M-200IN 100B1900R`108713C0 1 2 3 4 5 60123456Titanium, Wt %Aluminum, Wt %Easy to weldDifficult to weldFigure 16.12 Effect of Al and Ti contents on postweld heat treatment cracking. Modifiedfrom Kelly (30).susceptibility to cracking.This is because such alloys tend to age harden ratherrapidly and because their ductility is low.16.3.4 Proposed MechanismsPostweld heat treatment cracking in Ni-base alloys is a result of low ductilityand high strains in the HAZ (31, 32). So far several mechanisms have beenproposed for the causes of low ductility in the HAZ, including, for example,embrittlement of the grain boundary due to liquidation or solid-state reactionsduring welding (33–36), embrittlement of the grain boundary by oxygen duringheat treatment (37–39), and a change in deformation mode from transgranularslip to grain boundary sliding (14, 19, 22). The causes of high strains in theHAZ, on the other hand, can be the welding stresses and the thermal expansionand contraction of the material. In heat-treatable Ni-base alloys, the precipitationof strengthening phases results in contraction during aging. Thisaging contraction, in fact, has been proposed by several investigators (7, 22,33, 40) to be a contributing factor to postweld heat treatment cracking in heattreatableNi-base alloys.16.3.5 RemediesSeveral different methods of avoiding postweld heat treatment crackinghave been recommended. Most of these methods are based on experimentallyobserved crack susceptibility C curves. A crack susceptibility C curveis a curve indicating the onset of postweld heat-treatment cracking in atemperature–time plot. It is usually obtained by isothermal heat treating ofwelded circle patches at different temperatures for different periods of timeand checking for cracking. Such a curve usually resembles the shape of a “C”and, therefore, is called a crack susceptibility C curve. Since the aging rate atthe lower end of the aging temperature range is relatively slow, the occurrenceof cracking approaches asymptotically some lower temperature limit. Likewise,since residual stresses are relaxed at the higher temperature end of theaging temperature range and precipitates formed at lower temperature aredissolved, the occurrence of cracking approaches asymptotically some highertemperature limit. At any temperature between the upper and lower asymptoticlimits, there exists a minimum time, prior to which no cracking is possibleand beyond which cracking is certain to occur (37).Figure 16.13 shows the crack susceptibility C curves of Waspaloy andInconel 718 (1). Inconel 718 ages much more sluggishly than g ¢-strengthenedNi-base alloys (Figure 16.4b). As a result, the C curve of Inconel 718 is far tothe right of the C curve of Waspaloy, suggesting that the former is much moreresistant to postweld heat treatment cracking. In fact, Inconel 718 is a materialdesigned specifically to minimize postweld heat treatment cracking andshould be considered when postweld heat treatment cracking is of majorconcern. However, other alloys may still have to be used because of specificPOSTWELD HEAT TREATMENT CRACKING 387requirements involved, and the following approaches to reduce the crackingsusceptibility are worth considering.Preweld overaging, either by multistep-type overaging or simply by coolingslowly from the solutionizing temperature, appears to significantly reducepostweld heat treatment cracking in Rene 41 (36, 37, 41), Figure 16.14 beingan example (36).The base metal of an overaged workpiece is more ductile anddoes not age contract during postweld heat treatment, and this helps preventsevere residual stresses in the HAZ. But, as Franklin and Savage (32) pointedout, the overaged precipitate is dissolved in the HAZ during welding, and consequentlythe HAZ may still harden and age contract during postweld heattreatment.The effect of preweld solution annealing appears to be controversial.Gottlieb (42) reported that solution-annealed Rene 41 (solutionized at 1080°C,or 1975°F, for half an hour followed by water quenching) is in fact more susceptibleto postweld heat treatment cracking than fully age hardened Rene 41388 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYS1200140016001800No cracksaging rate andcrack reducedWaspaloy Alloy 718Time (minutes)Temperature, oFTemperature, oC70080090010001 101 102 103 104 105 106Figure 16.13 Crack susceptibility C curves for Waspaloy and Inconel 718 welds.Reprinted, with permission, from Owczarski (l).uncracked1.0 10.0 100.0 1000.0110012001300140015001600170018001900Aging temperature, oC6001000700800900Aging temperature, oFFigure 16.14 Crack susceptibility C curve for a Rene 41 solution annealed beforewelding and crack test data () from a Rene 41 overaged before welding. Reprintedfrom Berry and Hughes (36). Courtesy of American Welding Society.[solutionized at 1080°C for half an hour, water quenched, aged at 760°C(1400°F) for 4–16h and then air cooled]. Berry and Hughes (36), on the otherhand, reported that fully age hardened Rene 41 behaves essentially the sameas solution-annealed Rene 41 during postweld heat treatment.It is true that the base metal of a solution-annealed workpiece can age contractas well as harden during postweld heat treatment, and effective stressrelief can be difficult. However, if the same weldment is heated rapidly duringpostweld heat treatment, effective stress relief can be achieved before it has achance to harden and age contract, thus avoiding cracking (36, 37). As shownin Figure 16.15, no cracking occurs if the weldment is heated rapidly to avoidintersecting the crack susceptibility C curve (36). This approach is feasiblewhen the welded structure can be heated rapidly in a furnace and when distortionsdue to nonuniform heating are not excessive.POSTWELD HEAT TREATMENT CRACKING 389C-Curve0.1 1.0 10.0 100.08001000Time, min.Temperature, oCNotcracked CrackedFigure 16.15 Effect of heating rate on postweld heat treatment cracking of a Rene41 solution annealed before welding. Reprinted from Berry and Hughes (36). Courtesyof American Welding Society.Heat 95285Heat 8179Filled point = melt-through0 10,000 30,000 50,000 70,000048121620Weld energy, joules/inchTotal crack length, inches4,000 12,000 20,000Weld energy, joules/cmTotal crack length, cm10203040500Figure 16.16 Effect of welding heat input on postweld heat treatment cracking ofRene 41. Reprinted from Thompson et al. (37). Courtesy of American Welding Society.Another way of avoiding postweld heat treatment cracking is to use vacuumor inert atmospheres for heat treating (4, 36–38). It has been postulated thatthis technique is effective since there is no oxygen present to embrittlethe grain boundary during postweld heat treatment (37–39, 43). Other recommendedapproaches include using low welding heat inputs (Figure 16.16),using small grain-size materials (36, 39), and controlling the composition(Figure 16.17). Of course, using low-restraint joint designs is helpful.Finally, reheat cracking has also been reported in other alloys including1/2Cr–1/2Mo–1/4V steel (a creep-resistant ferritic steel), A517F (T1) steel (a structuralsteel), and 18Cr–12Ni–1Nb steel (a Nb-stabilized stainless steel) (44).Thereheat cracking of creep-resistant ferritic steels will be discussed in Chapter17.REFERENCES1. Owczarski,W. A., in Physical Metallurgy of Metal Joining, Eds. R. Kossowsky andM. E. Glicksman, Metallurgical Society of AIME, New York, 1980, p. 166.2. Sims, C. T., J. Metals, 18: 1119, October 1966.3. Betteridge,W., The Nimonic Alloys, Arnold, London, 1959.4. Prager, M., and Shira, C. S., Weld. Res. Council Bull., 128: 1968.5. Mihalisin, J. R., and Pasquzne,D. L., Structural Stability in Superalloys, vol. 1, JointASTM-ASME, 1968, pp. 134–170.6. Decker, R. F., and Sims, C. T., in The Superalloys, Eds. C. T. Sims and W. C. Hagel,Wiley, New York, 1972, p. 33.7. Blum, B. S., Shaw, P., and Wickesser, A., Technical Report ASD-TRD-63-601,Republic Aviation Corp., Farmingdale, NY, June 1963.8. Thompson, E. G., Weld. J., 48: 70s, 1969.9. Thompson, E.G., unpublished research, University of Alabama, Birmingham, 1981.10. Brooks, J. A., Weld. J., 53: 517s, 1974.390 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYS1 10 100 100012001400160018002000High Fe, Si,Mn and S Low Fe, Si,Mn and SAging temperature, oF6007008001000900Aging temperature, oC1000Figure 16.17 Effect of composition on postweld heat treatment cracking of Rene 41.Reprinted from Berry and Hughes (36). Courtesy of American Welding Society.11. Eiselstein, H. L., Adv. Technol. Stainless Steels, Special Technical Publication No.369.12. Wilson, R. M. Jr., and Burchfield, L.W. G., Weld. J., 35: 32s, 1956.13. Owczarski,W. A., and Sullivan, C. P., Weld. J., 43: 393s, 1964.14. Lucas, M. J. Jr., and Jackson, C. E., Weld. J., 49: 46s, 1970.15. Hirose, A., Sakata, K., and Kobayahi, K. F., in Solidification Processing 1997, Eds.J. Beech and H. Jones, Department of Engineering Materials, University ofSheffield, Sheffield, United Kingdom, 1997, p. 675.16. Savage,W. F., Nippes, E. F., and Goodwin, G. M., Weld. J., 56: 245s, 1977.17. Savage,W. F., and Krantz, B. M., Weld. J., 50: 29s, 1971.18. Savage,W. F., and Krantz, B. M., Weld. J., 45: 13s, 1966.19. Owczarski,W. A., Duvall, D. S., and Sullivan, C. P., Weld. J., 45: 145s, 1966.20. Duvall, D. S., and Owczarski,W. A., Weld. J., 46: 423s, 1967.21. Yeniscavich,W., Weld. J., 45: 344s, 1966.22. Wu, K. C., and Herfert, R. E., Weld. J., 46: 32s, 1967.23. Yeniscavich, W., in Proceedings of the Conference on Methods of High-AlloyWeldability Evaluation, Welding Research Council, New York, 1970, p. 2.24. Gordine, J., in Proceedings of the Conference on Methods of High-Alloy WeldabilityEvaluation,Welding Research Council, New York, 1970, p. 28.25. Owczarski,W. A., in Proceedings of the Conference on Effects of Minor Elementson the Weldability of High-Nickel Alloys, Welding Research Council, New York,1967, p. 6.26. Canonico, D. A., Savage,W. F.,Werner,W. J., and Goodwin, G. M., in Proceedingsof the Conference on Effects of Minor Elements on The Weldability of High-NickelAlloys,Welding Research Council, New York, 1967, p. 68.27. Valdez, P. J., and Steinman, J. B., in Proceedings of the Conference on Effects ofMinor Elements on The Weldability of High-Nickel Alloys, Welding ResearchCouncil, New York, 1967, p. 93.28. Grotke, G. E., in Proceedings of the Conference on Effects of Minor Elements onThe Weldability of High-Nickel Alloys,Welding Research Council, New York, 1967,p. 138.29. McKeown, D., Weld. J., 50: 201s, 1971.30. Kelly,T. J., in Weldability of Materials, Eds. R. A. Patterson and K.W. Mahin,ASMInternational, Materials Park, OH, 1990, p. 151.31. Baker, R. G., and Newman, R. P., Metal Construction Br. Weld. J., 1: 4, 1969.32. Franklin, J. G., and Savage,W. F., Weld. J., 53: 380s, 1974.33. Chang,W. H., Report DM58302 (58AD-16), General Electric, October 1958.34. Hughes,W. P., and Berry, T. F., Weld. J., 46: 361s, 1967.35. Morris, R. J., Metal Prog. 76: 67, 1959.36. Berry, T. F., and Hughes,W. P., Weld. J., 46: 505s, 1969.37. Thompson, E. G., Nunez, S., and Prager, M., Weld. J., 47: 299s, 1968.38. Carlton, J. B., and Prager, M., Weld. Res. Council Bull., 150: 13, 1970.39. Prager, M., and Sines, G., Weld. Res. Council Bull., 150: 24, 1970.40. Schwenk,W., and Trabold, A. F., Weld. J., 42: 460s, 1963.REFERENCES 39141. Fawley, R.W., and Prager, M., Weld. Res. Council Bull., 150: 1, 1970.42. Gottlieb, T.: unpublished Rocketdyne data.43. Prager, M.A., and Thompson, E.G., Report R-71 11, Rocketdyne, September 1967.44. Nichols, R.W., Weld. World, 7(4): 1969, p. 245.FURTHER READING1. Sims, C. T., and Hagel,W. C., Eds., The Superalloys,Wiley, New York, 1972.2. Thamburaj, R.,Wallace,W., and Goldak, J. A., Int. Metals Rev., 28: 1, 1983.PROBLEMS16.1 Explain why the susceptibility of Rene 41 welded in the solution annealcondition to postweld heat treatment cracking increases with increasingwelding heat input.16.2 It has been observed that the temperature of solution heat treatmentbefore welding can significantly affect the susceptibility of Rene 41 topostweld heat treatment cracking. For instance, specimens subjectedto 2150°F preweld solution heat treatment have been found to be moresusceptible than those subjected to a 1975°F treatment. Explain why.16.3 It has been reported that, in developing strain-age cracking C curves forRene 41, water quenching following isothermal heat treatment of thewelded circle patches often results in cracking. (a) Do you expect theC curves so developed to be reliable? (b) It has been suggested that atthe end of isothermal heat treatment the furnace temperature be raisedto 1975°F and kept there for 30 min and that the welded circle patchesthen be allowed to furnace cool at a rate of about 3–8°F/min (1.7–4.4°C/min). Cracking during cooling has been eliminated this way.Explain why. Do you expect the cracking C curves so obtained to bemore reliable than those mentioned earlier?16.4 Two rules are often quoted in postweld heat treatment of nickel-basealloys. First, never directly age weldments of heat-treatable nickel-basealloys. Second, the aging temperatures should exceed the service temperaturesof the weldments. Explain why.392 PRECIPITATION-HARDENING MATERIALS II: NICKEL-BASE ALLOYS17 Transformation-HardeningMaterials: Carbon and Alloy SteelsCarbon and alloy steels are more frequently welded than any other materialsbecause of their widespread applications and good weldability. In general,carbon and alloy steels with higher strength levels are more difficult to weldbecause of the risk of hydrogen cracking. Table 17.1 summarizes some typicalwelding problems in carbon and alloy steels and their solutions.The problemsassociated with the fusion zone and the partially melted zone have been discussedin previous chapters.This chapter deals with basic HAZ phenomena inselected carbon and low-alloy steels.17.1 PHASE DIAGRAM AND CCT DIAGRAMSThe HAZ in a carbon steel can be related to the Fe–C phase diagram, as shownin Figure 17.1, if the kinetic effect of rapid heating during welding on phasetransformations is neglected. The HAZ can be considered to correspond tothe area in the workpiece that is heated to between the lower criticaltemperature A1 (the eutectoid temperature) and the peritectic temperature.Similarly, the PMZ can be considered to correspond to the areas betweenthe peritectic temperature and the liquidus temperature, and the fusion zoneto the areas above the liquidus temperature.The Fe–C phase diagram and the continuous-cooling transformation (CCT)diagrams for heat treating carbon steels can be useful for welding as well, butsome fundamental differences between welding and heat treating should berecognized. The thermal processes during the welding and heat treating of acarbon steel differ from each other significantly, as shown in Figure 17.2. First,in welding the peak temperature in the HAZ can approach 1500°C. In heattreating, however, the maximum temperature is around 900°C, which is notmuch above the upper critical temperature A3 required for austenite (g) toform. Second, the heating rate is high and the retention time above A3 is shortduring most welding processes (electroslag welding being a notable exception).In heat treating, on the other hand, the heating rate is much slower andthe retention time aboveA3 is much longer.TheA1 andA3 temperatures duringheating (chauffage) are often referred to as the Ac1 and Ac3 temperatures,respectively.393Welding Metallurgy, Second Edition. Sindo KouCopyright �� 2003 John Wiley & Sons, Inc.ISBN: 0-471-43491-4For kinetic reasons the Ac1 and Ac3 temperatures tend to be higher thanthe equilibriumA1 andA3 temperatures, respectively, and they tend to increasewith increasing heating rate during welding (1, 2). Kinetically, phase transformationsrequire diffusion (the transformation to martensite is a well-knownexception) and diffusion takes time. Consequently, upon rapid heating duringwelding, phase transformations may not occur at the equilibrium A1 and A3temperatures but at higher temperatures Ac1 and Ac3. For steels containinggreater amounts of carbide-forming elements (such as V,W, Cr, Ti, and Mo),394 TRANSFORMATION-HARDENING MATERIALSTABLE 17.1 Typical Welding Problems and Practical Solution in Carbon andAlloy Steels, and Their Locations in the TextTypical Problems Alloy Types Solutions LocationsPorosity Carbon and low- Add deoxidizers (Al, Ti, 3.2alloy steels Mn) in filler metal 3.3Hydrogen cracking Steels with high Use low-hydrogen or 3.2carbon equivalent austenitic stainless 17.4steel electrodesPreheat and postheatLamellar tearing Carbon and low- Use joint designs that 17.6alloy steels minimize transverserestrainButter with a softerlayerReheat cracking Corrosion and heat- Use low heat inputa to 17.5resisting steels avoid grain growthMinimize restraint andstress concentrationsHeat rapidly throughcritical temperaturerange, if possibleSolidification cracking Carbon and low- Keep proper Mn/S 11.4alloy steels rationLow HAZ toughness Carbon and low- Use carbide and nitride 17.2due to grain growth alloy steels formers to suppress 17.3grain growthUse low heat inputaLow fusion-zone Carbon and low- Grain refining 7.6toughness due to alloy steels Use multipass welding 17.2coarse columnar to refine grainsgrainsa Low heat input processes (GMAW and SMAW vs. SAW and ESW) or multipass welding withlow heat input in each pass.the effect of the heating rate becomes more pronounced. This is because thediffusion rate of such elements is orders of magnitude lower than that ofcarbon and also because they hinder the diffusion of carbon. As a result, phasetransformations are delayed to a greater extent.The combination of high heating rates and short retention time above Ac3in welding can result in the formation of inhomogeneous austenite duringheating. This is because there is not enough time for carbon atoms in austeniteto diffuse from the prior pearlite colonies of high carbon contents to priorferrite colonies of low carbon contents. Upon rapid cooling, the former canPHASE DIAGRAM AND CCT DIAGRAMS 395Fe 1 2Carbon, wt%Temperature, oC16001200800L ++ Fe3C+ Fe3CL3 4(a)(b)Carbon steelpartially melted zoneheat-affected zonebase metalfusion zoneA1A3γγγαFigure 17.1 Carbon steel weld: (a) HAZ; (b) phase diagram.A3TemperatureTimeHeat TreatingWelding (much highermaximum temperature andshorter time above Atemperature) 3TL(a) (b)Fe 1 2Carbon, wt%16001200800400L++Fe 3C+Fe 3CLiquid, LA Austenite, 33 4 5AFe rrite, 1Ferrite,T, oCγγαγδαFigure 17.2 Comparison between welding and heat treating of steel: (a) thermalprocesses; (b) Fe–C phase diagram.transform into high-carbon martensite colonies while the latter into lowcarbonferrite colonies. Consequently, the microhardness in the HAZ canscatter over a wide range in welds made with high heating rates.As a result of high peak temperatures during welding, grain growth can takeplace near the fusion boundary. The slower the heating rate, the longer theretention time above Ac3 is and hence the more severe grain growth becomes.In the heat treating, however, the maximum temperature employed is onlyabout 900°C in order to avoid grain growth.The CCT diagrams (Chapter 9) for welding can be obtained by using a weldthermal simulator (Chapter 2) and a high-speed dilatometer that detects thevolume changes caused by phase transformations (3–6). However, since CCTdiagrams for welding are often unavailable, those for heat treating have beenused. These two types of CCT diagrams can differ from each other becauseof kinetic reasons. For instance, grain growth in welding can shift the CCTdiagram to longer times favoring transformation to martensite.This is becausegrain growth reduces the grain boundary area available for ferrite and pearliteto nucleate during cooling. However, rapid heating in welding can shift theCCT diagram to shorter times, discouraging transformation to martensite.Carbide-forming elements (such as Cr, Mo, Ti, V, and Nb), when they are dissolvedin austenite, tend to increase the hardenability of the steel. Because ofthe sufficient time available in heat treating, such carbides dissolve morecompletely and thus enhance the hardenability of the steel.This is usually notpossible in welding because of the high heating rate and the shorthigh-temperature retention time encountered in the HAZ (7).17.2 CARBON STEELSAccording to the American Iron and Steel Institute (AISI), carbon steels maycontain up to 1.65wt% Mn, 0.60wt% Si, and 0.60wt% Cu in addition tomuch smaller amounts of other elements. This definition includes the Fe–Csteels of the 10XX grades (up to about 0.9% Mn) and the Fe–C–Mn steels ofthe 15XX grades (up to about 1.7% Mn). The last two digits in the alloy designationnumber denote the nominal carbon content in weight percent, forinstance, about 0.20% C in a 1020 and about 0.41% C in a 1541 steel. Manganeseis an inexpensive alloying element that can be added to carbon steelsto help increase hardenability.17.2.1 Low-Carbon SteelsThese steels, in fact, include both carbon steels with up to 0.15% carbon, calledlow-carbon steels, and those with 0.15–0.30% carbon, called mild steels (8).For the purpose of discussion 1018 steel, which has a nominal carbon contentof 0.18%, is used as an example. Figure 17.3 shows the micrographs of agas–tungsten arc weld of 1018 steel. The base metal consists of a light-etching396 TRANSFORMATION-HARDENING MATERIALSCARBON STEELS 397A CB DFigure 17.3 HAZ microstructure of a gas–tungsten arc weld of 1018 steel (magnification200¥).ferrite and a dark-etching pearlite (position A). The HAZ microstructure canbe divided into essentially three regions: partial grain-refining, grain-refining,and grain-coarsening regions (positions B–D).The peak temperatures at thesepositions are indicated in the phase diagram.The partial grain-refining region (position B) is subjected to a peak temperaturejust above the effective lower critical temperature,Ac1. As explainedin Figure 17.4, the prior pearlite (P) colonies transform to austenite (g) andexpand slightly into the prior ferrite (F) colonies upon heating to above Ac1and then decompose into extremely fine grains of pearlite and ferrite duringcooling. The prior ferrite colonies are essentially unaffected. The grain-refiningregion (position C) is subjected to a peak temperature just above the effectiveupper critical temperature Ac3, thus allowing austenite grains to nucleate.Such austenite grains decompose into small pearlite and ferrite grains duringsubsequent cooling. The distribution of pearlite and ferrite is not exactlyuniform because the diffusion time for carbon is limited under the high heatingrate during welding and the resultant austenite is not homogeneous.The graincoarseningregion (position D) is subjected to a peak temperature well aboveAc3, thus allowing austenite grains to grow. The high cooling rate and largegrain size encourage the ferrite to form side plates from the grain boundaries,called the Widmanstatten ferrite (9).Grain coarsening near the fusion boundary results in coarse columnargrains in the fusion zone that are significantly larger than the HAZ grains onthe average. As shown in Figure 17.5, in multiple-pass welding of steels thefusion zone of a weld pass can be replaced by the HAZs of its subsequentpasses (10). This grain refining of the coarse-grained fusion zone by multiplepasswelding has been reported to improve the weld metal toughness.Although martensite is normally not observed in the HAZ of a low-carbonsteel, high-carbon martensite can form when both the heating and the cooling398 TRANSFORMATION-HARDENING MATERIALSFe 1 2Carbon, wt%Temperature, oC16001200800400L+ Fe3C+ Fe3CLiquid, LAustenite,A33 4 5A1132Ferrite,Ferrite,PP PPPearlite, P1 2 3δα+ γγαα αααα αααα ααααγγ γγFigure 17.4 Mechanism of partial grain refining in a carbon steel.rates are very high, as in the case of some laser and electron beam welding.Figure 17.6 shows the HAZ microstructure in a 1018 steel produced by a highpowerCO2 laser beam (11). At the bottom of the HAZ (position B) highcarbonmartensite (and, perhaps, a small amount of retained austenite) formedin the prior-pearlite colonies. The high-carbon austenite formed in thesecolonies during heating did not have time to allow carbon to diffuse out, andit transformed into hard and brittle high-carbon martensite during subsequentrapid cooling. Hard, brittle martensite embedded in a much softer matrix offerrite can significantly degrade the HAZ mechanical properties. Further upinto the HAZ (positions C and D), both the peak temperature and the diffusiontime increased. As a result, the prior-pearlite colonies expanded whiletransforming into austenite and formed martensite colonies of lower carboncontents during subsequent cooling.High-carbon martensite can also form in the HAZ of an as-cast low-carbonsteel, where microsegregation during casting causes high carbon contents inthe interdendritic areas. Aidun and Savage (12) have studied the repairing ofCARBON STEELS 399(b)Figure 17.5 Grain refining in multiple-pass welding: (a) single-pass weld; (b)microstructure of multiple-pass weld. Reprinted from Evans (10). Courtesy ofAmerican Welding Society.400 TRANSFORMATION-HARDENING MATERIALSFigure 17.6 HAZ microstructure of 1018 steel produced by a high-power CO2 laser.Magnification of (A)–(D) 415¥ and of (E) 65¥. From Kou et al. (11).cast steels used in the railroad industry. A series of materials with 0.21–0.31%C, 0.74–1.57% Mn, 0.50% Si, and up to about 0.20% Cr and Mo were spotwelded using covered electrodes E7018. As a result of the microsegregationof carbon and alloying elements during casting, continuous networks of interdendriticpearlite nodules with carbon contents ranging from about 0.5 to0.8% were present in the as-cast materials, as shown in Figure 17.7a.The resultantHAZ microstructure is shown in Figure 17.8. During welding of the as-castmaterials, the continuous networks of pearlite nodules formed continuous networksof high-carbon austenite upon heating, which in turn transformed tocontinuous networks of high-carbon martensite upon cooling (region E). Theislands scattered in the networks are untransformed ferrite. These networksof interdendritic pearlite nodules can be eliminated by homogenizing at 954°Cfor 2 h, as shown in Figure 17.7b.17.2.2 Higher Carbon SteelsThese steels include carbon steels with 0.30–0.50% carbon, called mediumcarbonsteels, and those with 0.50–1.00% carbon, called high-carbon steels(8). Welding of higher carbon steels is more difficult than welding lowerCARBON STEELS 401Figure 17.7 Microstructure of a carbon steel: (a) as-cast condition; (b) after homogenization.Reprinted from Aidun and Savage (12). Courtesy of American WeldingSociety.carbon steels because of the greater tendency of martensite formationin the HAZ and hence hydrogen cracking. For the purpose of discussion,1040 steel, which has a nominal carbon content of 0.40%, is used as anexample.Figure 17.9 shows the micrographs of a gas-tungsten arc weld of 1040 steel.The base metal of the 1040 steel weld consists of a light-etching ferrite and adark-etching pearlite (position A), as in the 1018 steel weld discussed previously.However, the volume fraction of pearlite (C rich with 0.77 wt % C) issignificantly higher in the case of the 1040 steel because of its higher carboncontent. As in the case of the 1018 steel, the HAZ microstructure of the 1040steel weld can be divided essentially into three regions: the partial grainrefining,grain-refining, and grain-coarsening regions (positions B–D). TheCCT diagram for the heat treating of 1040 steel shown in Figure 17.10 can beused to explain qualitatively the HAZ microstructure (13). In the grain coarseningregion (position D), both the high cooling rate and the large grain sizepromote the formation of martensite.The microstructure is essentially martensite,with some dark-etching bainite (side plates) and pearlite (nodules). In thegrain-refining region (position C), on the other hand, both the lower coolingrate and the smaller grain size encourage the formation of pearlite and ferrite.The microstructure is still mostly martensitic but has much smaller grainsand more pearlite. Some ferrite and bainite may also be present at grainboundaries.Because of the formation of martensite, preheating and control of interpasstemperature are often required when welding higher carbon steels. For1035 steel, for example, the recommended preheat and interpass temperaturesare about 40°C for 25-mm (1-in.) plates, 90°C for 50-mm (2-in.) plates,and 150°C for 75-mm (3-in.) plates (assuming using low-hydrogen electrodes).For 1040 steel, they are about 90°C for 25-mm (1-in.) plates, 150°C for50-mm (2-in.) plates, and 200°C for 75-mm (3-in.) plates (14). The reason for402 TRANSFORMATION-HARDENING MATERIALSFigure 17.8 HAZ microstructure of a stationary repair weld of an as-cast carbon steel.Fusion zone: A, B; HAZ: C–F; base metal: G. Reprinted from Aidun and Savage (12).Courtesy of American Welding Society.CARBON STEELS 403ABCDFigure 17.9 HAZ microstructure of a gas–tungsten arc weld of 1040 steel (magnification400¥).more preheating for thicker plates is because for a given heat input thecooling rate is higher in a thicker plate (Chapter 2). In addition to thehigher cooling rate, a thicker plate often has a slightly higher carbon contentin order to ensure proper hardening during the heat-treating step of the steelmakingprocess.The hardness profile of the HAZ of the 1040 steel weld is shown in Figure17.11a. When welded with preheating, the size of the HAZ increases but themaximum hardness decreases, as shown in Figure 17.11b. Examination of theHAZ microstructure near the fusion boundary of the preheated weld revealsmore pearlite and ferrite but less martensite. This is because the cooling ratedecreases significantly with preheating (Chapter 2).17.3 LOW-ALLOY STEELSThree major types of low-alloy steels will be considered here: high-strength,low-alloy steels; quenched-and-tempered low-alloy steels; and heat-treatablelow-alloy steels.404 TRANSFORMATION-HARDENING MATERIALSAc3FPBM1 10 102 103 104 1052004006008001,6001,4001,2001,000800600400200Temperature, oFTemperature, oCCooling time, sec10%Bainite50% Ferrite50% PearliteAISI 1040 steel:0.39C, 0.72Mn, 0.23SiF: ferrite; P: pearliteB: bainite; M: martensiteFigure 17.10 Continuous cooling transformation diagram for 1040 steel. Modifiedfrom Atlas of Isothermal Transformation and Cooling Transformation Diagrams (13).10 0 10 10 0 10400800FZa HAZ bc da bDistance, mm Distance, mmknoop hardness (1000 g)(a) (b)c dFZHAZ250 oCpreheatingNopreheatingFigure 17.11 Hardness profiles across HAZ of a 1040 steel; (a) without preheating;(b) with 250°C preheating.17.3.1 High-Strength, Low-Alloy SteelsHigh-strength, low-alloy (HSLA) steels are designed to provided higherstrengths than those of carbon steels, generally with minimum yield strengthsof 275–550MPa (40–80ksi). Besides manganese (up to about 1.5%) andsilicon (up to about 0.7%), as in carbon steels, HSLA steels often contain verysmall amounts of niobium (up to about 0.05%), vanadium (up to about0.1%), and titanium (up to about 0.07%) to ensure both grain refinementand precipitation hardening. As such, they are also called microalloyedsteels. Typically the maximum carbon content is less than 0.2% and the totalalloy content is less than 2%. Alloys A242, A441, A572, A588, A633, and A710are examples of HSLA steels, and their compositions are available elsewhere(15).Niobium (Nb), vanadium (V), and titanium (Ti) are strong carbide andnitride formers. Fine carbide or nitride particles of these metals tend to hinderthe movement of grain boundaries, thus reducing the grain size by makinggrain growth more difficult. The reduction in grain size in HSLA steelsincreases their strength and toughness at the same time. This is interestingbecause normally the toughness of steels decreases as their strength increases.Among the carbides and nitrides of Nb, V, and Ti, titanium nitride (TiN) ismost stable; that is, it has the smallest tendency to decompose and dissolve athigh temperatures. This makes it most effective in limiting the extent of graingrowth in welding.The higher the heat input during welding, the more likely the carbide andnitride particles will dissolve and lose their effectiveness as grain growthinhibitors.The low toughness of the coarse-grain regions of the HAZ is undesirable.It has been reported that steels containing titanium oxide (Ti2O3) tendto have better toughness (16, 17). The Ti2O3 is more stable than TiN and doesnot dissolve even at high heat inputs. The undissolved Ti2O3 particles do notactually stop grain growth but act as effective nucleation sites for acicularferrite. Consequently, acicular ferrite forms within the coarse austenite grainsand improves the HAZ toughness (16).The HSLA steels are usually welded in the as-rolled or the normalized condition,and the weldability of most HSLA steels is similar to that of mild steel.Since strength is often the predominant factor in the applications of HSLAsteels, the filler metal is often selected on the basis of matching the strengthof the base metal (15). Any common welding processes can be used, but lowhydrogenconsumables are preferred.The preheat and interpass temperatures required are relatively low. Formost alloys they are around 10°C for 25-mm (1-in.) plates, 50°C for 50-mm (2-in.) plates, and 100°C for 75-mm (3-in.) plates. For alloy A572 (grades 60 and65) and alloy A633 (grade E), they are about 50°C higher (15). The amountof preheating required increases with increasing carbon and alloy content andwith increasing steel thickness.LOW-ALLOY STEELS 40517.3.2 Quenched-and-Tempered Low-Alloy SteelsThe quenched-and-tempered low-alloy (QTLA) steels, usually containing lessthan 0.25% carbon and less than 5% alloy, are strengthened primarily byquenching and tempering to produce microstructures containing martensiteand bainite.The yield strength ranges from approximately 345 to 895MPa (50to 130 ksi), depending on the composition and heat treatment. Alloys A514,A517, A543, HY-80, HY-100, and HY-130 are some examples of QTLA steels,and their compositions are available elsewhere (15).Low carbon content is desired in such alloys for the following two reasons:(i) to minimize the hardness of the martensite and (ii) to raise the Ms (martensitestart) temperature so that any martensite formed can be tempered automaticallyduring cooling. Due to the formation of low-carbon auto-temperedmartensite, both high strength and good toughness can be obtained. Alloyingwith Mn, Cr, Ni, and Mo ensures the hardenability of such alloys. The use ofNi also significantly increases the toughness and lowers the ductile–brittletransition temperature in these alloys.Any common welding processes can be used to join QTLA steels, but theweld metal hydrogen must be maintained at very low levels. Preheating is oftenrequired in order to prevent hydrogen cracking. The preheat and interpasstemperatures required are higher than those required for HSLA steels but stillnot considered high. For HY130, for example, the preheat and interpass temperaturesare about 50°C for 13-mm (0.5-in.) plates, 100°C for 25-mm (1-in.)plates, and 150°C for 38-mm (1.5-in.) plates (15). However, too high a preheator interpass temperature is undesirable. It can decrease the cooling rate of theweld metal and HAZ and cause austenite to transform to either ferrite orcoarse bainite, both of which lack high strength and good toughness. Postweldheat treatment is usually not required.Excessive heat input can also decrease the cooling rate and produce unfavorablemicrostructures and properties. High heat input processes, such asESW or multiple-wire SAW, should be avoided. Figure 17.12 shows the CCTcurves of T1 steel, that is, A514 and A517 grade F QTLA steel (18). Curves p,f, and z represent the critical cooling rates for the formation of pearlite, ferrite,and bainite, respectively. The hatched area represents the region of optimumcooling rates. If the cooling rate during welding is too low, for instance,between curve p and the hatched area indicated, a substantial amount offerrite forms.This can, in fact, be harmful since the ferrite phase tends to rejectcarbon atoms and turn its surrounding areas into high-carbon austenite. Suchhigh-carbon austenite can in turn transform to high-carbon martensite andbainite during cooling, thus resulting in a brittle HAZ. Therefore, the heatinput and the preheating of the workpiece should be limited when weldingquenched-and-tempered alloy steels.On the other hand, if the cooling rate during welding is too high, to theleft of curve z in Figure 17.12, insufficient time is available for the autotemperingof martensite. This can result in hydrogen cracking if hydrogen is406 TRANSFORMATION-HARDENING MATERIALSpresent.Therefore, low-hydrogen electrodes or welding processes and a smallamount of preheating are recommended. The hatched area in the figurerepresents the region of best cooling rates for welding this steel.To meet the requirements of both limited heat inputs and proper preheating,multiple-pass welding is often used in welding thick sections of QTLAsteels. In so doing, the interpass temperature is maintained at the same levelas the preheat temperature. Multiple-pass welding with many small stringerbeads improves the weld toughness as a result of the grain-refining and temperingeffect of successive weld passes. The martensite in the HAZ of a weldpass is tempered by the heat resulting from deposition in subsequent passes.As a result, the overall toughness of the weld metal is enhanced. Figure 17.13shows the effect of bead tempering (19). The HAZ of bead E is temperedby bead F and is, therefore, softer than the HAZ of bead D, which is nottempered by bead F of any other beads.17.3.3 Heat-Treatable Low-Alloy SteelsThe heat-treatable low-alloy (HTLA) steels refer to medium-carbonquenched-and-tempered low-alloy steels, which typically contain up to 5% oftotal alloy content and 0.25–0.50% carbon and are strengthened by quenchingto form martensite and tempering it to the desired strength level (15).Thehigher carbon content promotes higher hardness levels and lower toughnessand hence a greater susceptibility to hydrogen cracking than the quenchedand-tempered low-alloy steels discussed in the previous section. Alloys 4130,4140, and 4340 are examples of HTLA steels.The HTLA steels are normally welded in the annealed or overtemperedcondition except for weld repairs, where it is usually not feasible to anneal orovertemper the base metal before welding. Immediately after welding, theLOW-ALLOY STEELS 407Figure 17.12 CCT curves for A514 steel. From Inagaki et al. (18).entire weldment is heat treated, that is, reaustenized and then quenched andtempered to the desired strength level, or at least stress relieved or temperedto avoid hydrogen cracking.Any of the common welding processes can be usedto join HTLA steels. To avoid hydrogen cracking, however, the weld metalhydrogen must be maintained at very low levels, proper preheat and interpasstemperatures must be used, and preheat must be maintained after welding iscompleted until the commencement of postweld heat treatment.In applications where the weld metal is required to respond to the samepostweld heat treatment as the base metal in order to match the base metalin strength, a filler metal similar to the base metal in composition is used. Inrepair welding where it is possible to use the steel in the annealed or overtemperedcondition, the filler metal does not have to be similar to the base metalin composition. The weldment is stress relieved or tempered after welding.In some repair welding where neither annealing or overtempering the steelbefore welding nor stress relieving the weld upon completion is feasible, electrodesof austenitic stainless steels or nickel alloys can be used. The resultantweld metal has lower strength and greater ductility than the quenched-andtemperedbase metal, and high shrinkage stresses during welding can result inplastic deformation of the weld metal rather than cracking of the HAZ.High preheat and interpass temperatures are often required for weldingHTLA steels. For alloy 4130, for instance, they are around 200°C for 13-mm408 TRANSFORMATION-HARDENING MATERIALS0102030405060ACBWeldmetalDFEHardness traverseof upper welded faceLimit of temperingby beads C and FLimit of hardeningby beads C and FHardness traverseof lower welded faceZone hardened bybeads B and EZone hardenedby beads A and DRockwell C hardnessFigure 17.13 Tempering bead technique for multiple-pass welding of a butt joint in aquenched-and-tempered alloy steel. Reprinted form Linnert (19). Courtesy of AmericanWelding Society.(0.5-in.) plates, 250°C for 25-mm (1-in.) plates, and almost 300°C for 50-mm(2-in.) plates. For alloys 4140 and 4340, they are even higher (15).In addition to the use of preheating, the weldment of heat-treatable alloysteels is often immediately heated for stress-relief heat treatment beforecooling to room temperature. During the stress-relief heat treatment, martensiteis tempered and, therefore, the weldment can be cooled to room temperaturewithout danger of cracking. After this, the weldment can be postweldheat treated to develop the strength and toughness the steel is capable ofattaining.A sketch of the thermal history during welding and postweld stress relievingis shown in Figure 17.14a. The preheat temperature and the temperatureimmediately after welding are both maintained slightly below the martensitefinishtemperature Mf. Stress relieving begins immediately after welding, at atemperature below the A1 temperature.This can be further explained using a 25-mm (1-in.) 4130 steel as anexample. Based on the isothermal transformation diagram for 4130 steel (13)shown in Figure 17.15, the ideal welding procedure for welding 4130 steel isto use a filler metal of the same composition, preheat to 250°C (about 500°F),weld while maintaining a 250°C interpass temperature using low-hydrogenelectrodes, and stress-relief heat treat at about 650°C (1200°F) immediatelyupon completion of welding (19). The postweld heat treatment of the entireweldment can be done in the following sequence: austenitizing at about 850°C(about 1600°F), quenching, and then tempering in the temperature range400–600°C (about 800–1100°F).As shown in Figure 17.14a, the HAZ should be cooled to a temperatureslightly below the martensite-finish temperature Mf before being heated forstress relieving. If this temperature is above Mf, as shown in Figure 17.14b,there can be untransformed austenite left in the HAZ and it can decomposeinto ferrite and pearlite during stress relieving or transform to untemperedmartensite upon cooling to room temperature after stress relieving.In the event that stress-relief heat treatment cannot be carried out immediatelyupon completion of welding, the temperature of the completed weldmentcan be raised to approximately 400°C (750°F), which is the vicinity ofthe bainite “knee” for 4130 and most other heat-treatable alloy steels. ByLOW-ALLOY STEELS 409WeldingStress relievingPreheatingTimeTemperatureTimeTemperature(a) (b)A1 A1Mf MfFigure 17.14 Thermal history during welding and postweld stress relieving of a heattreatablealloy steel: (a) desired; (b) undesired.holding at this temperature for about 1 h or less, the remaining austenite cantransform to bainite, which is more ductile than martensite. Therefore, whenthe weldment is subsequently cooled to room temperature, no cracking shouldbe encountered. Further heat treatment can be carried out later in order tooptimize the microstructure and properties of the weldment.In cases where heat-treatable low alloy steels cannot be postweld heattreated and must be welded in the quenched-and-tempered condition, HAZsoftening as well as hydrogen cracking can be a problem.To minimize softening,a lower heat input per unit length of weld (i.e., a lower ratio of heat inputto welding speed) should be employed. In addition, the preheat, interpass, andstress-relief temperatures should be at least 50°C below the tempering temperatureof the base metal before welding. Since postweld heat treating of theweldment is not involved, the composition of the filler metal can be substantiallydifferent from that of the base metal, depending on the strength level ofthe weld metal required.17.4 HYDROGEN CRACKINGSeveral aspects of hydrogen cracking have been described previously (Chapter3), including the various sources of hydrogen during welding, the hydrogen410 TRANSFORMATION-HARDENING MATERIALSFigure 17.15 Isothermal transformation curves for 4130 steel. Reprinted from Atlasof Isothermal Transformation and Cooling Transformation Diagrams (13).levels in welds made with various welding processes, the solubility of hydrogenin steels, the methods for measuring the weld hydrogen content, and thetechniques for reducing the weld hydrogen content have been described previouslyin Chapter 3.17.4.1 CauseHydrogen cracking occurs when the following four factors are present simultaneously:hydrogen in the weld metal, high stresses, susceptible microstructure(martensite), and relatively low temperature (between -100 and 200°C).High stresses can be induced during cooling by solidification shrinkage andthermal contraction under constraints (Chapter 5). Martensite, especially hardand brittle high-carbon martensite, is susceptible to hydrogen cracking. Sincethe martensite formation temperature Ms is relatively low, hydrogen crackingtends to occur at relatively low temperatures. For this reason, it is often called“cold cracking.” It is also called “delayed cracking,” due to the incubation timerequired for crack development in some cases.Figure 17.16 depicts the diffusion of hydrogen from the weld metal to theHAZ during welding (20). The terms TF and TB are the austenite/(ferrite +pearlite) and austenite/martensite transformation temperatures, respectively.As the weld metal transforms from austenite (g) into ferrite and pearlite (a +Fe3C), hydrogen is rejected by the former to the latter because of the lowersolubility of hydrogen in ferrite than in austenite. The weld metal is usuallylower in carbon content than the base metal because the filler metal usuallyhas a lower carbon content than the base metal. As such, it is likely that theweld metal transforms from austenite into ferrite and pearlite before the HAZtransforms from austenite into martensite (M). The build-up of hydrogen inthe weld metal ferrite causes it to diffuse into the adjacent HAZ austenitenear the fusion boundary, as indicated by the short arrows in the figure. Asshown in Figure 17.17, the diffusion coefficient of hydrogen is much higher inferritic materials than austenitic materials (21). The high diffusion coefficientHYDROGEN CRACKING 411+Fe3CB H+H+TFTBAmartensiteH+H+H+H+ H+H+crackaustenite ( )base metalwelding directionweldpoolarcelectrodeαγγFigure 17.16 Diffusion of hydrogen from weld metal to HAZ during welding. Modifiedfrom Granjon (20).of hydrogen in ferrite favors this diffusion process. On the contrary, the muchlower diffusion coefficient of hydrogen in austenite discourages hydrogen diffusionfrom the HAZ to the base metal before the HAZ austenite transformsto martensite.This combination of hydrogen and martensite in the HAZ promoteshydrogen cracking.The mechanism for hydrogen cracking is still not clearly understood, thoughnumerous theories have been proposed. It is not intended here to discuss thesetheories since this is more a subject of physical metallurgy than welding metallurgy.For practical purposes, however, it suffices to recognize that Troiano(22) proposed that hydrogen promotes crack growth by reducing the cohesivelattice strength of the material. Petch (23) proposed that hydrogen promotescrack growth by reducing the surface energy of the crack. Beachem (24) proposedthat hydrogen assists microscopic deformation ahead of the crack tip.Savage et al. (25) explained weld hydrogen cracking based on Troiano’s theory.Gedeon and Eagar (26) reported that their results substantiated and extendedBeachem’s theory.17.4.2 AppearanceFigure 17.18 is a typical form of hydrogen crack called “underbead crack” (27).The crack is essentially parallel to the fusion boundary. Hydrogen cracking can412 TRANSFORMATION-HARDENING MATERIALS60050040030010 100 20010-310-410-510-610-710-81000/T, where T is temperature oK4 3 2 1FerriticmaterialsAusteniticmaterials10-9Overall diffusion coefficient, cm 2sec-1Temperature, oCFigure 17.17 Diffusion coefficient of hydrogen in ferritic and austenitic materials asa function of temperature. Modified from Coe (21).be accentuated by stress concentrations. Figure 17.19 shows cracking at thejunctions between the weld metal surface and the workpiece surface of a filletweld of 1040 steel (28).This type of crack is called “toe crack.”The same figurealso shows cracking at the root of the weld, where lack of fusion is evident.This type of crack is called “root cracks.”HYDROGEN CRACKING 413Figure 17.18 Underbead crack in a low-alloy steel HAZ (magnification 8¥).Reprinted from Bailey (27).Figure 17.19 Hydrogen cracking in a fillet weld of 1040 steel (magnification 4.5¥).Courtesy of Buehler U.K., Ltd., Coventry, United Kingdom.17.4.3 Susceptibility TestsThere are various methods for testing the hydrogen cracking susceptibility ofsteels, such as the implant test (20, 29), the Lehigh restraint test (30), the RPIaugmented strain cracking test (31), the controlled thermal severity test (32),and the Lehigh slot weldability test (33). Due to the limitation in space, onlythe first two tests will be described here.Figure 17.20 is a schematic of the implant test. In this test, a cylindrical specimenis notched and inserted in a hole in a plate made from a similar material.A weld run is made over the specimen, which is located in such a way thatits top becomes part of the fusion zone and its notch lies in the HAZ. Afterwelding and before the weld is cold, a load is applied to the specimen and thetime to failure is determined. As an assessment of hydrogen cracking susceptibility,the stress applied is plotted against the time to failure, as shown inFigure 17.21 for a high-strength, low-alloy pipeline steel (33). In this caseloading was applied to the specimen when the weld cooled down to 125°C.As414 TRANSFORMATION-HARDENING MATERIALSweldHAZloadbase metaltestspecimenarcFigure 17.20 Implant test for hydrogen cracking.80070060050040030020010 100 1000 10000 100000Time to Failure (sec)Stress to Failure (MPa)(Ar + 2% O2 )E7018E7010Involved in AE StudyNo FractureFigure 17.21 Implant test results for a HSLA pipe line steel. Reprinted fromVasudevan et al. (33). Courtesy of American Welding Society.shown, the welds made with low-hydrogen electrodes (E7018, basic limestonetypecovering) have a higher threshold stress below which no cracking occursand a longer time to cracking than the welds made with high-hydrogen electrodes(E7010, cellulose-type covering). In other words, the former is less susceptibleto hydrogen cracking than the latter. The gas–metal arc welds madewith Ar + 2% O2 as the shielding gas are least susceptible. Obviously, no electrodecovering is present to introduce hydrogen into these welds.Figure 17.22 shows the Lehigh restraint specimen (30). The specimen isdesigned with slots cut in the sides (and ends). The longer the slots, the lowerthe degree of plate restraint on the weld is. A weld run is made in the root ofthe joint, and the length of the slots required to prevent hydrogen cracking isdetermined. Cracking is detected visually or by examining transverse crosssections taken from the midpoint of the weld.17.4.4 RemediesA. Control of Welding ParametersA.1. Preheating As already described in the previous section, the use of theproper preheat and interpass temperatures can help reduce hydrogen cracking.Figure 17.23 shows such an example (26). Two general approaches haveHYDROGEN CRACKING 415groovefor welding20 o1.6 mm(1/16") gap38 mm (1 1/2") 25 mm (1")x305 mm (12")203 mm(8")89 mm (3 1/2") for plate < 25 mm (1")140 mm (5 1/2") for plate 25 mm (1")13 mm (1/2")x
.jpg)
bookmarked!!, I really like your blog!
BalasHapusHave a look at my web page ... Www.Candecarpetsandvents.com
If you are going for best contents like myself,
BalasHapussimply go to see this site everyday because it gives feature contents, thanks
My blog : Office Furniture Toronto
I am really enjoying the theme/design of your
BalasHapuswebsite. Do you ever run into any web browser compatibility problems?
A handful of my blog readers have complained about my blog not working correctly
in Explorer but looks great in Safari. Do you have any advice to help fix this problem?
Feel free to surf my web-site :: kids fishing clothes
If you would like to take much from this paragraph then you have
BalasHapusto apply such methods to your won weblog.
Feel free to surf my blog post - Contactez-Nous
If you wish for to grow your knowledge simply keep visiting
BalasHapusthis website and be updated with the most recent news update posted here.
Also visit my webpage - Kauai Weddings
Excellent website you have here but I was wanting to know
BalasHapusif you knew of any community forums that cover the same topics discussed
here? I'd really like to be a part of online community where I can get comments from other knowledgeable people that share the same interest. If you have any suggestions, please let me know. Appreciate it!
Also see my page > www.rainbowcottoncandy.com
It's not my first time to pay a quick visit this web site, i am browsing this web site dailly and get fastidious facts from here every day.
BalasHapusFeel free to surf my web page - blu ray player
Magnificent goods from you, man. I have understand your stuff previous
BalasHapusto and you are just extremely magnificent.
I really like what you've acquired here, really like what you're
stating and the way in which you say it. You make it entertaining and you
still care for to keep it wise. I can not wait to read far
more from you. This is really a tremendous site.
Have a look at my web site; public golf courses in kissimmee fl
Good post. I'm experiencing many of these issues as well..
BalasHapusAlso visit my blog post: residential led lighting
I wanted to thank you for this good read!! I definitely loved every bit of it.
BalasHapusI have got you bookmarked to check out new
things you post…
Feel free to visit my weblog easy link building
This is my first time visit at here and i am genuinely pleassant
BalasHapusto read all at one place.
my web page - Get Laid
I used to be able to find good advice from your blog articles.
BalasHapusmy web site :: wsmisports.com
Hello it's me, I am also visiting this site regularly, this web site is really nice and the users are really sharing nice thoughts.
BalasHapusStop by my web site ... Abercrombie Et Fitch
Hi just wanted to give you a brief heads up and let you know a few of the images aren't loading properly. I'm not sure why
BalasHapusbut I think its a linking issue. I've tried it in two different web browsers and both show the same results.
Here is my web blog: Mario Lemieux Authentic Jersey
Hi colleagues, its enormous paragraph about tutoringand entirely
BalasHapusexplained, keep it up all the time.
Feel free to visit my website: Louis Vuitton Outlet
Thanks for finally talking about > "WeLdinG mEtaLLurGi" < Loved it!
BalasHapusHere is my site ... Gafas De Sol Oakley
Hurrah! Finally I got a weblog from where I know how to genuinely
BalasHapusobtain useful data concerning my study and knowledge.
My web blog ... Get More Information
Greetings! I know this is kind of off topic but I was wondering which blog platform are you
BalasHapususing for this website? I'm getting fed up of Wordpress because I've had problems with hackers and
I'm looking at alternatives for another platform. I would be awesome if you could point me in the direction of a good platform.
my weblog; Michael Kors Handbags
For newest news you have to visit the web and on world-wide-web
BalasHapusI found this web page as a finest web site for hottest updates.
My web blog; Air Max
I have fun with, cause I discovered just what I was taking a look for.
BalasHapusYou've ended my four day lengthy hunt! God Bless you man. Have a great day. Bye
Also visit my web-site - Air Jordan Femme
Hi mates, how is everything, and what you want to say regarding this article, in
BalasHapusmy view its really awesome designed for me.
Feel free to surf to my blog :: Louis Vuitton Handbags Outlet
Hello, constantly i used to check webpage posts here in the early hours in
BalasHapusthe morning, for the reason that i like to find out more and more.
My blog ... Sidney Crosby Black Jersey
Thankfulness to my father who stated to me concerning this website, this
BalasHapusweb site is actually awesome.
my blog: wealthwayonline.com
Hi to every one, it's really a fastidious for me to go to see this site, it includes valuable Information.
BalasHapusAlso visit my site :: www.explorethecapabilities.com
What's up i am kavin, its my first occasion to commenting anyplace, when i read this piece of writing i thought i could also create comment due to this brilliant paragraph.
BalasHapusCheck out my page; Cheap Jerseys
Your style is unique compared to other people I have read stuff from.
BalasHapusMany thanks for posting when you've got the opportunity, Guess I will just bookmark this blog.
Here is my blog post ... Abercrombie
Wonderful blog! I found it while searching on Yahoo
BalasHapusNews. Do you have any tips on how to get listed in Yahoo News?
I've been trying for a while but I never seem to get there! Thank you
Here is my blog kitchen design ideas for small kitchens in india
Awesome article.
BalasHapusFeel free to surf to my site: Cheap Oakley Sunglasses
Excellent post but I was wondering if you could write a litte more on this
BalasHapustopic? I'd be very thankful if you could elaborate a little bit further. Appreciate it!
Also visit my web-site; Air Max Femme
wah banyak spamm neh
BalasHapus