The Effects of Minor Elements on the Welding Characteristics of Stainless Steel

The Effects of Minor Elements on the Welding Characteristics of Stainless Steel

Sulfur, oxygen and deoxidant contents determine penetration, while the silicon-to-aluminum ratio controls slagging characteristics

BY B. POLLARD

ABSTRACT. The effects of systematic inclusions in the steel because it tended who supply stainless steel strip to welded variations in the aluminum, sulfur, silicon, to approach an equilibrium correspond­ tubing manufacturers. Welded stainless manganese, phosphorus, molybdenum ing to temperatures near the surface of steel tubing is made by continuously roll- and copper contents on the penetration the weld pool, which were much higher forming a strip of steel into a tube and and slagging characteristics of laboratory than those encountered in steelmaking. welding the butt joint of the strip togeth­ heats of Type 304 stainless steel have Provided that sufficient amounts of the er. For wall thicknesses up to 0.125 in. been investigated. The effects of alumi­ appropriate deoxidant was present, (3.18 mm), the welding process most num and sulfur on penetration in tung­ these high temperatures favored the for­ commonly used is gas tungsten arc sten inert gas welding can be satisfactorily mation of slags that were richer in the (GTA). Since no filler wire is used, welding explained in terms of their effects on the high-stability oxides, such as those of characteristics of the steel are deter­ temperature coefficient of surface ten­ calcium, aluminum, titanium and silicon, mined completely by its composition. sion and the theory of surface tension and poorer in the low-stability oxides, The two welding characteristics of partic­ driven fluid flow. However, the results such as those of manganese and chromi­ ular concern to tubemakers are penetra­ for silicon and manganese are more com­ um, than the inclusions. In general, the tion and slagging. Poor penetration char­ plex and, while not inconsistent with the slagging characteristics of stainless steel acteristics limit welding speed and, there­ surface tension driven fluid flow model, were found to be determined primarily fore, have an adverse effect on the indicate that other factors besides the by the silicon-to-aluminum ratio of the economics of the tubemaking operation. temperature coefficient of surface ten­ steel. Patch type slags, which caused an In extreme cases, they can result in mate­ sion must be taken into account. In the uneven weld bead and irregular penetra­ rial that is unweldable because there is case of silicon, its influence on the viscos­ tion, were formed when the silicon-to- essentially no difference between the ity of the molten metal appears to have a aluminum ratio was less than 50 and the current required for full penetration and significant effect on penetration. As less troublesome globular slags formed that which causes excessive melt expected, phosphorus, molybdenum and when it was greater than 50. through. The second problem, slag for­ copper had no effect on penetration mation, is covered in Part II. Compositional limits for stainless steel since they are not surface active ele­ with welding characteristics suitable for ments, or likely to interact with surface The problem of heat-to-heat variations the production of welded tubing are active elements and, therefore, have no in the penetration characteristics of stain­ given. effect on the temperature coefficient of less steel was first reported in the litera­ surface tension. ture by Oyler, et al. (Ref. 1), in 1967. Part I—Penetration Characteristics However, it was not until 1975 that it Penetration values for production became a major problem for steelmak­ heats of Types 304 and 316 stainless Introduction ers. At that time, steelmakers noticed a steels showed a strong dependence on The effects of minor elements on the significant increase in the number of com­ sulfur content, but were lower for the welding characteristics of stainless steel plaints from tube-making customers laboratory heats, which was attributed to are of particular interest to steelmakers regarding the poor penetration charac­ their lower oxygen contents. teristics of 300 Series stainless steels. Slag formation during the GTA welding There were two reasons for these com­ of stainless steels was found to be caused plaints: first, many tube mills had installed primarily by the oxide inclusions in the new high-speed welding lines, which steel. However, EDS analyses indicated KEY WORDS required steel with improved penetration that the composition of the slag was not characteristics in order to exploit the always equivalent to a mixture of all the GTA Welding higher welding speeds which the equip­ Stainless Steels ment was capable of; and second, the Minor Element Effect argon-oxygen decarburization (AOD) Weld Characteristics process was rapidly replacing the electric B. POLLARD is with the Research Center, Weld Penetration Armco, Inc., Middletown, Ohio. The paper furnace (EF) for refining stainless steel. was prepared while the author was with I TV Weld Slagging The most obvious difference between Steel Co., Independence, Ohio. Welded Tubing AOD and EF steel is the oxygen content, Si-to-AI Ratio which typically averages about 60 ppm Paper presented at a conference on "Progress Sulfur-02 Deoxidants for AOD steel versus 120 ppm for EF in Resolving Undesirable Trace Element Effects Weld Bead Geometry steel. Therefore, poor penetration was on the Weldability of Steels," held November originally attributed to the low oxygen 21, 1985, in Charlotte, N.C.

202-s | SEPTEMBER 1988 content of AOD steel (Ref. 2). Evidence low-sulfur heats, which were solved by going to lower sulfur contents if this for the beneficial effect of oxygen was increasing the sulfur content to 0.015%. could be done without sacrificing weld­ provided by the results of other investi­ The adverse effect of aluminum and the ability. gators who showed that oxygen present beneficial effect of sulfur on weld pene­ In order to determine if it were possi­ as an oxide film improved penetration tration have been confirmed by more ble to produce low-sulfur steel with good (Refs. 3, 4). However, laboratory welding recent studies (Refs. 7-9). Unfortunately, penetration characteristics, a better tests and customer experience revealed although high-sulfur contents result in understanding of the interaction be­ that, although on average EF steel exhib­ good penetration characteristics, they tween the surface active elements oxy­ ited slightly better penetration than AOD have an adverse effect on the hot work­ gen and sulfur, and the deoxidants alumi­ steel, there were wide variations in pene­ ability, cold formability and pitting corro­ num, silicon and manganese was tration for steel produced by either pro­ sion resistance of stainless steel. There­ required. Optimum manganese and sili­ cess. Low penetration was found to be fore, the present practice at LTV Steel con contents were suggested by Oyler, associated with the presence of strong Specialty Products Co.1 for common et al. (Ref. 1), but no data were provided deoxidants, specifically aluminum, mag­ grades of stainless steels, such as Types to support their recommendations. A nesium, cerium and calcium (Refs. 5, 6). If 304 and 316 and their low-carbon equiv­ systematic study was, therefore, made of oxygen increases penetration, then it is to alents, is to produce both a regular quali­ the effects of aluminum, sulfur, silicon be expected that strong deoxidants will ty grade, which may be low-sulfur if and manganese on the welding charac­ reduce penetration. Of the foremen- desired, and a tubing quality grade in teristics of Type 304 stainless steel using tioned deoxidants, only aluminum has which maximum aluminum and minimum laboratory heats in which one element at been widely used in stainless steel. It is sulfur contents are specified so as to a time was varied. The effects of phos­ sometimes added to the desulfurization ensure good welding characteristics. phorus, molybdenum and copper addi­ mix to facilitate the removal of sulfur by Recently, questions about the effect of tions in amounts typical of those found in the reaction composition on the welding characteris­ production heats were also investigated.

3 (CaO)s 4- 2 [AI]M + 3 [5]*,— tics of stainless steel have again been The results of this study were then com­ 3 (CaS)s + (Al203)s (1) raised. The reasons for this resurgence in pared with penetration data for produc­ Low penetration was found to be interest are primarily economic. First, a tion heats of Types 304 and 316 stainless associated with high-aluminum residuals, significant reduction in inventory cost steel. i.e., aluminum contents up to 0.030%. To could be achieved by combining regular prevent high-aluminum levels in steel for and tubing grades; and second, an tubing applications, aluminum additions improvement in the hot workability of Experimental Procedure were prohibited and penetration was difficult-to-roll grades such as Type 316, Material generally found to be satisfactory when which would reduce surface defects and the aluminum content did not exceed increase yields, could be obtained by

tion problems were still e xperienced with 7. Now l&L Specialty Products Co ingots were made with the aluminum,

Table 1--Compositions and Penetration Ratios for Laboratory Heats

'ercent Heat . No. C Mn Si S P Cr Ni Mo Cu Al N O, ppm D/W

V154 0.040 0.050 0.50 0.014 0.016 18.79 8.50 0.01 0.05 0.001 0.024 194 0.472 1329 0.034 0.12 0.47 0.009 0.012 18.92 8.39 0.01 0.04 0.005 0.018 178 0.493 1330 0.044 0.50 0.49 0.013 0.014 18.93 8.57 0.02 0.05 0.002 0.018 159 0.551 1331 0.038 0.96 0.49 0.013 0.014 19.09 8.60 0.03 0.05 0.001 0.024 195 0.542 1332 0.037 1.38 0.47 0.014 0.013 18.95 8.51 0.03 0.05 0.001 0.024 152 0.549 1333 0.033 2.02 0.50 0.013 0.014 17.28 8.79 0.04 0.05 0.001 0.015 158 0.508 V155 0.038 1.41 0.01 0.014 0.015 18.78 8.37 0.01 0.05 0.001 0.024 197 0.423 1334 0.015 1.46 0.05 0.012 0.013 19.09 8.59 0.04 0.05 0.001 0.034 160 0.418 V156 0.035 1.47 0.22 0.015 0.017 18.75 8.35 0.01 0.05 0.001 0.025 206 0.475 1335 0.026 1.45 0.26 0.014 0.014 19.04 8.82 0.04 0.05 0.005 0.033 183 0.440 1336 0.046 1.48 0.73 0.014 0.015 19.19 8.74 0.04 0.05 0.001 0.019 151 0.525 1337 0.042 1.42 0.92 0.012 0.013 19.08 8.52 0.03 0.05 0.001 0.018 149 0.477 1338 0.035 1.54 0.49 0.004 0.014 17.66 9.26 0.03 0.05 0.002 0.014 156 0.546 1339 0.045 1.43 0.51 0.007 0.013 19.10 8.55 0.03 0.05 0.004 0.024 194 0.528 1340 0.044 1.44 0.50 0.010 0.014 18.95 8.60 0.03 0.05 0.009 0.023 132 0.432 1341 0.040 1.46 0.49 0.017 0.014 18.88 8.53 0.03 0.05 0.002 0.055 166 0.533 1342 0.043 1.46 0.49 0.026 0.014 19.00 8.55 0.03 0.05 0.010 0.052 121 0.491 1343 0.037 1.49 0.50 0.014 0.005 18.98 8.58 0.02 0.05 0.004 0.029 160 0.533 1344 0.034 1.44 0.50 0.014 0.005 18.80 8.40 0.03 0.05 0.001 0.029 125 0.535 1345 0.040 1.48 0.48 0.014 0.019 18.85 8.65 0.03 0.05 0.002 0.014 164 0.537 1346 0.041 1.53 0.50 0.014 0.014 18.96 8.68 0.03 0.05 0.010 0.049 174 0.473 1347 0.037 1.47 0.49 0.014 0.014 18.90 8.60 0.03 0.05 0.015 0.015 156 0.460 1348 0.035 1.55 0.53 0.013 0.014 16.95 8.75 0.03 0.05 0.031 0.033 74 0.455 1349 0.035 1.56 0.52 0.016 0.016 18.80 8.74 0.44 0.05 0.001 0.029 174 0.533 1350 0.045 1.52 0.52 0.015 0.013 19.00 8.43 0.03 0.45 0.002 0.010 197 0.543 1509 0.072 1.42 0.46 0.005 0.023 18.52 8.44 0.02 0.05 0.019 0.065 68 0.383 1510 0.065 1.41 0.44 0.007 0.023 18.75 8.47 0.03 0.05 0.029 0.072 88 0.384 1511 0.060 1.45 0.46 0.014 0.024 18.51 8.49 0.03 0.05 0.040 0.075 73 0.415 1512 0.065 1.47 0.47 0.028 0.024 18.56 8.46 0.03 0.05 0.033 0.73 52 0.528 1513 0.060 1.44 0.45 0.052 0.022 18.48 8.46 0.03 0.05 0.021 0.076 75 0.510

WELDING RESEARCH SUPPLEMENT 1203-s Table 2- Compositions and Penetration Ratios for Types 304 and 304L Production Heats

Thickness Percent D/W Heat No Grade in. mm C Mn Si S P Cr Ni Mo Cu Al Ti N O, ppm Measured Corrected

70024 304L 0.108 2.74 0.023 1.66 0.54 0.023 0.028 18.37 8.97 0.35 0.41 0.001 ND 0.090 ND 0.457 0.501 71261 304L 0.064 1.63 0.026 1.63 0.56 0.010 0.029 18.40 8.88 0.23 0.35 0.001 ND 0.079 64 0.367 0.371 93504 304L 0.061 1.55 0.022 1.72 0.69 0.021 0.032 18.19 8.97 0.25 0.38 0.001 0.006 0.091 57 0.480 0.481 70002 304 0.063 1.60 0.051 1.60 0.52 0.017 0.030 18.32 8.14 0.18 0.19 0.003 ND 0.090 ND 0.408 0.411 70030 304 0.071 1.80 0.070 1.66 0.55 0.004 0.025 18.38 8.92 0.16 0.19 0.003 ND 0.048 ND 0.401 0.411 70037 304 0.032 0.81 0.063 1.67 0.57 0.003 0.026 18.35 9.08 0.16 0.25 0.001 ND 0.039 ND 0.380 0.354 70069 304 0.052 1.32 0.055 1.67 0.43 0.018 0.027 18.39 8.60 0.16 0.22 0.001 ND 0.050 ND 0.450 0.443 70097 304 0.050 1.27 0.055 1.68 0.49 0.017 0.030 18.26 8.18 0.15 0.23 0.003 ND 0.078 ND 0.530 0.521 70099 304 0.108 2.74 0.054 1.75 0.52 0.014 0.027 18.31 8.15 0.20 0.21 0.004 ND 0.084 ND 0.415 0.459 70919 304 0.027 0.69 0.044 1.64 0.68 0.002 0.024 18.22 8.92 0.20 0.36 0.004 ND 0.037 ND 0.403 0.373 71028 304 0.060 1.52 0.049 1.70 0.66 0.022 0.026 18.34 8.17 0.18 0.30 0.003 ND 0.079 ND 0.456 0.456 71030 304 0.060 1.52 0.036 1.67 0.61 0.015 0.025 18.32 8.14 0.15 0.19 0.003 ND 0.078 ND 0.433 0.433 803175 304 0.046 1.17 0.050 1.87 0.57 0.025 0.034 18.21 8.24 0.35 0.37 0.001 0.007 0.061 36 0.472 0.459 894104 304 0.046 1.17 0.034 1.84 0.57 0.019 0.034 18.42 8.28 0.30 0.33 0.001 0.007 0.072 37 0.473 0.460 91776 304 0.058 1.47 0.036 1.61 0.61 0.019 0.035 18.13 8.28 0.25 0.32 0.003 ND 0.080 92 0.496 0.494

sulfur, silicon, manganese and phospho­ Penetration Measurements and the penetration ratio, D/W, was rus contents varied within the following defined as the ratio of the strip thickness ranges: aluminum 0 to 0.03%, sulfur 0 to Welding tests were performed on to the bead width. 0.05%, silicon 0 to 1.0%, manganese 0 to strips of metal approximately 13 in. The ratio D/W varies with welding 2.0% and phosphorus 0 to 0.020%. long X 1.5 in. wide (33 mm X 3.8 mm), conditions as well as steel composition, Also, one heat was made with an addi­ which were washed with mineral spirits, and the choice of welding conditions tion of 0.5% molybdenum and one with followed by acetone, before testing. The under which D/W measurements are an addition of 0.5% copper, which are test specimens were held in a Jetline made is somewhat arbitrary. D/W typical maximum values for these ele­ longitudinal positioner with a copper increases with welding current to a maxi­ ments in Type 304 stainless steel. The backing bar containing a 0.50-in. mum value when the weld just pene­ compositions of the laboratory heats are wide X 0.25-in. deep (12.7-mm X 6.4- trates the underside of the strip. The presented in Table 1. All material was mm) groove, and weld penetration was D/W measurements were made at this made into 0.060-in. (1.52-mm) thick measured using a variable current tech­ current level, because measurements annealed and pickled strip using standard nique and the welding conditions listed in made under these conditions should processing procedures for Type 304 Table 4. After striking the arc at a current relate more closely to the situation in an stainless steel. level well below that required to pene­ actual welding operation than those Production material for weldability trate the strip, the current was continu­ made at lower current levels. As noted studies was obtained in the annealed and ously increased at a rate of 6 A/s to a previously, the D/W measurements on pickled condition in thicknesses of 0.020 value above that required for full pene­ laboratory process material were made to 0.110 in. (0.51 to 2.79 mm), depending tration before the arc was extinguished. at a fixed strip thickness. on availability. The compositions of the The bead width was then measured at Tests on various thicknesses of strip production heats are presented in Tables the point where the weld bead just made from the same heat of Type 304 2 and 3. penetrated the underside of the material stainless steel showed that D/W

Table 3- Compositions and Penetration Ratios for Types 316 and 316L Production Heats

Thickness Percent D/W Heat No Grade in. mm C Mn Si S P Cr Ni Mo Cu Al Ti N O, ppm Measured Corrected

70008 316L 0.077 1.96 0.022 1.75 0.45 0.017 0.021 17.00 10.37 2.08 0.16 0.001 ND 0.084 ND 0.396 0.412 70143 316L 0.075 1.91 0.015 1.78 0.62 0.016 0.027 16.74 10.24 2.14 0.37 0.005 0.007 0.068 ND 0.376 0.390 70411 316L 0.102 2.59 0.018 1.68 0.67 0.016 0.024 17.55 12.58 2.66 0.31 0.001 ND 0.061 ND 0.390 0.429 90587 316L 0.036 0.91 0.015 1.82 0.59 0.010 0.026 16.34 11.02 2.12 0.30 0.004 0.016 0.017 ND 0.447 0.425 91711 316L 0.062 1.57 0.020 1.72 0.50 0.005 0.027 17.33 11.24 2.32 0.30 0.001 0.009 0.072 78 0.346 0.348 91716 316L 0.063 1.60 0.018 1.68 0.60 0.007 0.027 16.69 11.05 2.14 0.32 0.001 0.009 0.025 ND 0.364 0.367 91804 316L 0.062 1.57 0.015 1.67 0.54 0.007 0.034 16.75 11.07 2.03 0.31 0.001 0.010 0.040 ND 0.362 0.364 91958 316L 0.060 1.52 0.017 1.70 0.46 0.006 0.031 16.67 11.05 2.06 0.27 0.001 0.008 0.026 ND 0.360 0.360 92131 316L 0.060 1.52 0.022 1.66 0.60 0.007 0.029 16.48 11.16 2.10 0.37 0.001 0.008 0.037 ND 0.358 0.358 70248 316 0.105 2.67 0.063 1.46 0.55 0.025 0.028 16.77 10.22 2.07 0.56 0.001 0.005 0.045 ND 0.442 0.483 90291 316 0.036 0.91 0.048 1.54 0.61 0.023 0.025 16.99 10.36 2.10 0.24 0.001 0.006 0.036 ND 0.455 0.433 90448 316 0.112 2.84 0.058 1.57 0.63 0.017 0.024 16.79 10.35 2.10 0.29 0.001 0.018 0.037 ND 0.451 0.499 90532 316 0.060 1.52 0.053 1.56 0.70 0.015 0.024 16.77 10.90 2.12 0.65 0.001 0.012 0.014 ND 0.422 0.422 90591 316 0.103 2.62 0.037 1.60 0.62 0.018 0.026 16.28 10.92 2.05 0.33 0.002 0.008 0.014 ND 0.413 0.453 90690 316 0.106 2.69 0.042 1.59 0.69 0.004 0.026 16.33 10.95 2.11 0.30 0.001 0.014 0.020 ND 0.360 0.400 91712 316 0.062 1.57 0.057 1.62 0.52 0.003 0.025 16.79 10.79 2.03 0.30 0.001 0.010 0.030 64 0.437 0.439 91715 316 0.062 1.57 0.049 1.68 0.69 0.004 0.028 16.98 11.11 2.09 0.31 0.001 0.010 0.029 ND 0.377 0.379 91717 316 0.061 1.55 0.049 1.67 0.67 0.010 0.028 16.70 11.06 2.10 0.32 0.001 0.010 0.035 ND 0.447 0.448 91992 316 0.060 1.52 0.046 1.71 0.54 0.008 0.042 16.84 10.90 2.06 0.49 0.001 0.008 0.043 ND 0.434 0.434 92132 316 0.060 1.52 0.059 1.72 0.68 0.001 0.031 16.78 11.07 2.17 0.24 0.001 0.012 0.029 ND 0.391 0.391 SS 316 0.117 2.97 0.065 1.10 0.61 0.002 0.028 16.77 10.26 1.99 0.28 0.001 0.008 0.017 ND 0.381 0.433

204-s I SEPTEMBER 1988 0.70 r— Table 4—Welding Conditions

Process Gas tungsten arc HEAT 36I96 Nozzle diameter 0.50 in. (12.7 mm) 0.60 Shielding gas 25 cfh (11.8 L/min) argon Backing gas 25 cfh (11.8 L/min) argon Electrode Tungsten —2% thoria, 0.125 in. (3.18 mm) ** 11_ diameter, 60-deg point 0.50 Arc length 0.10 in. (2.54 mm) Current 90 to 140 A for full penetration Welding speed 20 ipm (0.85 cm/s)

0.40 -co l - o- A decreased with increasing strip thick­ SLOPE =-0.92 ness—Fig. 1. The slope of the curve for the welding conditions used was —0.92 0.30 in.-1 (—2.34 cm-1), and this value was 0 .020 .040 .060 .080 .100 .120 used to correct the D/W values for specimens of production material to a THICKNESS, INCHES common thickness of 0.060 in. (1.52 Fig. 1 —Effect of strip thickness on the penetration ratio, D/W mm).

Oxygen Content ganese, 0.5% silicon and 0.014% sulfur nese and 0.5% silicon was found to Although oxygen content was not decreased 14% for an increase in alumi­ depend on the aluminum content —Fig. varied systematically, measurements of num content from 0.001 to 0.015%, but 4. Sulfur content had no effect on pene­ oxygen content were required to explain only an additional 1% for a further tration for heats containing no more than the difference in penetration characteris­ increase to 0.031 % — Fig. 2. It is significant 0.004% aluminum. In contrast, penetra­ tics between laboratory and production that most of the change in the penetra­ tion ratios for heats containing 0.02 to heats. tion ratio occurred at aluminum contents 0.04% aluminum increased with an The oxygen contents of the strip were of less than 0.010%, which had little increase in the sulfur content up to measured instead of the oxygen contents effect on the measured oxygen content 0.028%, beyond which a further increase of the welds because prior experience (Fig. 3); and conversely, aluminum con­ to 0.052% had no further effect on pen­ had indicated little difference between tents of 0.010 to 0.031%, which had little etration. At the 0.005% sulfur level, base metal and weld oxygen contents. effect on penetration, significantly 0.02% aluminum reduced penetration by Moreover, weld oxygen contents are reduced the measured oxygen content. 31.5%. Two heats which contained 0.009 more difficult to measure accurately These results indicate that it is the amount and 0.010% aluminum had similar D/W because of inhomogeneous distribution of oxygen in solution in the molten steel values as the heats containing 0.02 to of slag particles within the welds. rather than the total oxygen content that 0.04% aluminum, which again empha­ Since some oxygen present in the steel affects penetration. sizes the fact that to obtain maximum as inclusion is lost as slag that floats to the The effect of sulfur content on pene­ penetration the aluminum content must weld surface, base metal content may, in tration for heats containing 1.5% manga- be tightly controlled, preferably to a level fact, be a more appropriate value to use than weld metal oxygen content. The latter is characteristic of the weld after it has solidified, while the penetration char­ 0.60 acteristics of the weld are determined by the oxygen content of the molten metal during weld pool formation. It seems evident that significant oxida­ tion of the weld pool by the air during the formation of the weld did not occur, because the compositions of the slags were consistent with their formation V 0.50 from inclusions in the steel. For example, Q a slag with a high-calcium content could not have been formed by air reacting with the weld pool, because essentially all of the very low calcium content of the steel was already present as oxide.

Results Laboratory Heats 0.40 .035 Penetration results for the laboratory 0 .005 .010 .015 .020 .025 .030 heats are presented in Table 1. Penetra­ WT % ALUMINUM tion ratios for heats containing 1.5% man- Fig. 2—Effect of aluminum content on penetration

WELDINC RESEARCH SUPPLEMENT I 205-s *t(JU for Types 316 and 316L in Table 3. No 1 1 1 difference in penetration was found 300 between production heats of Types 304 and 304L stainless steels containing less than 0.004% aluminum, which both had Z 200 k~ -o- —'•—«^^ lower penetration ratios than the best a. c —-c 3 — a. _ 13 — —* laboratory heats —Fig. 8. Penetration m increased with increasing sulfur content, \ 5 IOO but was not quite equal to the laboratory v\~ «> 80 ^w c s heats even at the 0.020% sulfur level. Q >o The Types 316 and 316L heats had X 60 o \ "^ > similar penetration ratios to the Types O 304 and 304L heats. However, on aver­ V s age the penetration ratio was about 14% 40 higher for the Type 316 than for the Type 30 316L heats —Fig. 9. The penetration ratios for the Types 316 and 316L production on heats also increased with increasing sulfur content, but, like the 304 and 304L heats, .001 .002 .004.006 .010 .020 .040.060 .100 none had penetration ratios equal to the (10) (100) (1000) best laboratory heats. WT% ALUMINUM.(PPM) Fig. 3 - Effect of aluminum content on the oxygen content of the laboratory heats Discussion Penetration Mechanism

of not more than 0.005%. beyond which further manganese addi­ Various mechanisms have been pro­ The effects of silicon on penetration tions reduced penetration slightly —Fig 6. posed to explain the effect of small for heats containing 1.5% manganese and The increase in penetration associated variations in composition on the penetra­ 0.014% sulfur appear to be more com­ with a 0.80% addition to a manganese- tion characteristics of stainless steels and plex than those of aluminum and sulfur, free heat was about 20%. other metals. In general, these mecha­ since penetration first increased with sili­ Phosphorus contents in the range nisms can be divided into three groups: con additions up to about 0.50% and 0.005 to 0.019% had no effect on pene­ those that involve changes in arc charac­ then decreased with further additions up tration for heats containing 1.5% manga­ teristics (Refs. 5, 7, 10-12), such as the to 0.92% —Fig. 5. The increase in pene­ nese, 0.5% silicon and 0.014% sulfur - Fig. size of the anode spot (Ref. 10) or the tration associated with a 0.50% silicon 7. Likewise, additions of 0.5% molybde­ anode drop voltage (Ref. 12); those that addition to a silicon-free heat was about num or 0.5% copper had negligible effect involve changes in the liquid-gas or liquid- 30%. on penetration —Table 1. solid interfacial tensions (Refs. 13-18); The effect of manganese on penetra­ and those that depend on changes in the tion for heats containing 0.50% silicon pattern of fluid flow in the weld pool Production Heats and 0.014% sulfur was similar to that of (Refs. 9, 19). silicon. Penetration increased with the Penetration test results for production The circulation of molten metal within addition of manganese to a maximum heats of Types 304 and 304L stainless the weld pool may be a result of magne- value at about 0.80% manganese, steel are presented in Table 2, and those tohydrodynamic flow (Refs. 20, 21) or surface tension driven fluid flow (Ref. 9), otherwise known as Marangoni convec­ tion, or a combination of both processes. 0.60 At the current levels typical of most GTA welding, Marangoni convection appears to be the dominant driving force, and the 0.55 o o surface tension driven fluid flow model o o D proposed by Heiple, ef al. (Ref. 9), best explains the effects of very small amounts D 0.50 of elements such as sulfur and aluminum /• on penetration. The basis of the phenom­ enon of surface tension driven fluid flow is that at the surface of the weld pool molten metal is pulled from a region of 1 low surface tension to one of high sur­ WT % &l face tension, and molten metal flows O 0.004 °>'.MA X upwards from the depths of the pool to maintain continuity —Fig. 10. If the tem­ • 0.009- 0.010 / perature coefficient of surface tension is 0.35 positive, then the hot metal immediately beneath the arc is drawn down into the depths of the pool, producing deep pen­ 0.30 i 1 1 1 1 etration. Conversely, if the temperature .010 .020 .030 .040 .050 .060 coefficient of surface tension is negative, then the hot metal immediately beneath WT % SULPHUR the arc flows outward to the edges of the Fig. 4 —Effect of sulfur content on penetration

206-s I SEPTEMBER 1988 pool, thereby widening the weld pool 0.60 and producing shallow penetration. The temperature coefficients of surface ten­ sion for pure metals are negative (Ref. 22) and the beneficial effects of surface active elements such as sulfur and oxygen on weld penetration are believed to be due to the fact that they produce positive temperature coefficients of surface ten­ sion (Refs. 9, 23). A mathematical model 0.50 for convection in weld pools developed by Oreper, et al. (Ref. 24), is in good agreement with the Heiple, et al. model, particularly in regard to the effect of the temperature coefficient of surface ten­ U?f--y sion on the direction of fluid flow within the weld pool.

Effects of Sulfur and Aluminum 0.40 The results obtained in this investiga­ 0.2 0.4 0.6 0.8 I.O tion for the effects of sulfur and alumi­ num on weld penetration are in complete WT% SILICON agreement with the Heiple, et al. theory that surface tension driven fluid flow is the dominant driving force for metal Fig. 5 —Effect of silicon content on penetration circulation within the weld pool. As explained previously, the positive effect of sulfur on penetration is due to the fact that it produces a positive temperature 0.60 coefficient of surface tension, while the negative effect of aluminum is due to it combining with the oxygen, thereby reducing the soluble oxygen content of the molten metal and producing a nega­ tive temperature coefficient of surface tension. 0.50

Effects of Silicon and Manganese The results for silicon and manganese additions to Type 304 stainless steel both p showed penetration increasing up to a maximum value beyond which further additions of either element caused a reduction in penetration. These results are not inconsistent with the surface 0.40 tension driven fluid flow model, but they 0.50 I.O 1.5 2.0 2.5 do indicate that other factors besides the temperature coefficient of surface ten­ WT % MANGANESE sion must be taken into account. Silicon is Fig. 6 — Effect of manganese content on penetration a moderately strong deoxidant, and its addition to a steel that does not contain any stronger deoxidants must lower the where the soluble oxygen content weld pool. The patch-type slag formed soluble oxygen content of the weld pool. became the dominant factor, the temper­ on the manganese-free heat produced a However, the laboratory heats contained ature coefficient of surface tension was puckered surface on the weld bead, 0.014% sulfur, which was apparently suf­ reduced sufficiently to more than offset which was not observed with any other ficient to produce a positive temperature any further decrease in viscosity by the heat. This puckered surface suggests a coefficient of surface tension and a silicon and penetration subsequently high slag-metal interfacial tension, which, downward movement of the hot metal decreased. although not itself detrimental to penetra­ at the center of the weld pool. The The same explanation cannot be used tion, may also be associated with a reduc­ beneficial effect of the initial silicon addi­ for the initial increase in penetration tion in the temperature coefficient of tion was most likely due to it reducing the observed with manganese additions to surface tension. viscosity of the molten steel (Ref. 25). If a laboratory heats containing 0.5% silicon downward flow pattern already existed and 0.014% sulfur, since manganese addi­ because of the 0.014% sulfur content, tions up to 1.0% have little effect on the Effects of Phosphorus, Molybdenum and Copper then a reduction in the viscosity of the viscosity of iron (Ref. 26). A more likely molten steel would increase the flow explanation for the beneficial effect of a Additions of 0.005 to 0.019% phos­ velocity and hence increase penetration. 0.5% manganese addition on penetration phorus, 0.5% molybdenum or 0.5% cop­ However, with increasing silicon addi­ is that it drastically changes the nature of per had no effect on penetration. This is tions, a point was eventually reached the slag formed on the surface of the not surprising since there is no evidence

WELDING RESEARCH SUPPLEMENT 1207-s to indicate that these elements are sur­ the laboratory heats, not only at low- 28) to the shielding gas. However, the face active, interact in any way with sulfur contents, but also to a lesser extent amount required is 500 to 1000 ppm, surface active elements, or have a signifi­ at high-sulfur levels —Figs. 8 and 9. The which is sufficient to cause accelerated cant effect on viscosity at the levels most likely reason for the difference in electrode wear in production welding of added. penetration characteristics between the tubing, where uninterrupted runs of one laboratory and production heats was hour or more are normal practice. Also, oxygen content, which was only 40 to 80 sulfur dioxide is toxic and its use would Production Heats ppm for the production heats versus 150 create potential health and environmen­ The results of tests on the laboratory to 200 ppm for the laboratory heats. tal problems. Therefore, oxygen or sulfur heats suggested that good penetration Increasing penetration by increasing dioxide additions to the shielding gas are characteristics could be obtained with the oxygen content of the steel is not a not likely to alleviate the need for ade­ low-sulfur steel by simply limiting the practical means of compensating for a quate sulfur levels in stainless steel for aluminum content. However, tests on low-sulfur content because it would welded tubing applications. production material with aluminum con­ result in low chromium recoveries, poor The optimum sulfur content is a trade­ tents in the range of 0.001 to 0.003% surface quality and increased slagging. off between weld penetration and other showed penetration ratios that were sig­ Penetration can be increased by adding properties such as the hot workability of nificantly lower than those obtained with oxygen (Refs. 2, 27) or sulfur dioxide (Ref. the steel and the cold formability of the finished tubing. For good penetration, not less than 0.010% and preferably 0.015% sulfur is recommended. A range 0.6 of 0.005 to 0.010% sulfur will give improved hot workability and cold form- ability at the cost of some decrease in penetration, while less than 0.005% sulfur o gives a significant improvement in the hot — workability of difficult-to-work grades, cT such as Type 316, but penetration is generally only poor to fair. At the other end of the sulfur range, sulfur contents of V. 0.5 up to 0.025% have been satisfactory for Q thin-walled tubing (e.g., 0.060 in./1.52 mm thick) not subject to severe forming operations. For heavy-wall tubing, too — high a sulfur content can reduce the surface tension to the point where it cannot support the weld pool and exces­ sive melt through occurs. For this type of application, sulfur content is usually lim­ 0.4 ited to a maximum value of 0.015%. .004 .008 .012 .016 .020 With a sulfur content of 0.014%, the effect of aluminum on weld penetration WT % PHOSPHORUS was found to be greatest at aluminum contents below 0.010%-Fig. 2. With Fig. 7—Effect of phosphorus content on penetration appropriate steelmaking practice, most heats contain less than 0.005% aluminum, but aluminum contents of 0.005 to .60 0.010% sometimes occur. A maximum aluminum content of 0.010% is, there­ LAB HEATS 150 TO 200 PPM OXYGEN fore, specified. This limit guarantees only slightly better penetration than would be .50 obtained at higher aluminum contents, but prevents the formation of patch-type slags and uneven penetration, which will be discussed further in Part II of this paper. .40 Maximum penetration for the labora­ tory heats occurred with a silicon content of 0.50%, which is the same value that has been found in practice to provide .30 good weldability with production materi­ al. With this silicon content, maximum O 304L 40 TO 80 PPM OXYGEN penetration occurred at a manganese • 304 0.004% MAX Al content of 0.80%. However, since man­ ganese is an austenite former, it is used as .20 a low-cost substitute for nickel, and the .005 .010 .015 .020 .025 .030 manganese content of production heats of Type 304 stainless steel is usually in the WT% SULPHUR range 1.50 to 2.00%, the maximum value allowed by the AISI specification for this Fig. 8 — Effect of sulfur content on penetration for production heats of Types 304 and 304L stainless steels containing 0.004% maximum aluminum grade. Fortunately, the decrease in the

208-s I SEPTEMBER 1988 penetration ratio, D/W, attributable to 60 the higher manganese content is only 4 to LAB HEATS 9%. I50 TO 200 PPM OXYGEN

Conclusions .50 1) The results of this investigation con­ firm that the effects of small amounts of sulfur, oxygen and aluminum on weld penetration in the GTA welding of stain­ less steels are due to their effects, either .40 directly or indirectly, on the temperature coefficient of surface tension and the resultant pattern of surface tension driv­ en fluid flow within the weld pool. .30 2) The effects of silicon and manga­ 03I6L 40 TO 80 PPM OXYGEN nese on weld penetration are not incon­ 0.004% MAX Al sistent with the surface tension driven • 3I6 fluid flow model for penetration, but in the case of silicon, its effect on viscosity .20 must also be taken into account. .005 .010 .015 .020 .025 .030 3) Good penetration can be obtained with sulfur contents of less than 0.005% WT % SULPHUR provided that sufficient oxygen is present. However, with oxygen contents Fig. 9 — Effect of sulfur content on penetration for production heats of Types 316 and 316L stainless typical of AOD steelmaking, there is a steels containing 0.004% maximum aluminum steady decrease in penetration with decreasing sulfur content when the latter is less than about 0.025%. 4) The optimum sulfur content is a trade-off between hot workability, weld­ ability and formability requirements. For Type 304 stainless steel tubing, a sulfur content of 0.010 to 0.015% is a good compromise for most applications. 5) The effect of aluminum on penetra­ tion is greatest in the range 0 to 0.010%, so the aluminum content should prefera­ bly be limited to 0.005%, but steelmaking limitations dictate a more realistic limit of 0.010% maximum. 6) Phosphorus, molybdenum and cop­ per, in amounts typical of those normally found in Type 304 stainless steel, have no effect on penetration.

Part II—Slagging Characteristics Introduction By comparison with the penetration problem, slag formation during the GTA welding of stainless steel has received much less attention. Nevertheless, exces­ sive slag formation is particularly trouble­ some in automatic welding operations, Fig. 10 —Effect of Marangoni convection on penetration such as making an autogenous weld in tubing, where it has been reported to cause arc instability, an irregular weld but silicon, chromium, manganese and effects of minor elements on the slagging bead and reduced penetration (Refs. 6, 7, iron have also been found in welding characteristics of both laboratory and 29). Also, it can cause weld discontinuities slags (Refs. 10, 29). production heats of 300 Series stainless during postweld rolling and drawing Although it is widely accepted that steels were investigated. operations (Ref. 29). oxide inclusions in the steel are the source of the slag, a correlation between Welding tests in an inert gas chamber Experimental Procedure have proved that the source of the slag is the compositions of the steel, the inclu­ the steel and not oxidation due to incom­ sions and the welding slag has not been Specimens for slagging studies were plete shielding (Ref. 6). Slagging has been established. The effect of slag character­ made using the same specimen size, attributed primarily to high melting-point istics on weldability in tubemaking also cleaning procedure and welding condi­ oxides, particularly the oxides of alumi­ has not been adequately explained. tions as used in the penetration studies, num, calcium, magnesium, titanium, zir­ Therefore, in order to determine how to except for the current level, which was conium and cerium (Refs. 6, 7, 10, 29); minimize the effect of slagging, the constant and about 50 A above the

WELDING RESEARCH SUPPLEMENT 1209-s Results Globular slags were formed when the %Si/%AI ratio was greater than about 50 Laboratory Heats and patch-type slags when it was less As expected, the types of inclusions than 50. The only exceptions were heats present in the laboratory heats and the that were very low in manganese (Heat resultant weld slag compositions were 1329) or silicon (Heat 1334), which found to depend upon the manganese, formed chromium-rich, patch-type slags silicon and aluminum contents of the even though they had %Si/%Al ratios of steel-Table 5. With a %Mn/%Si ratio of 94 and 50, respectively. 0.26 (Heat 1329), the inclusions were pure silica, but the slag contained primar­ Production Heats ily chromium oxide and only a small The compositions of the weld slags amount of silica. For heats with %Mn/ formed with production heats are in %Si ratios of 1.0, 2.0 to 2.9 and 4.0 to Table 6. They varied more widely than 5.6, the inclusions identified were 100% those formed with laboratory heats, even manganese chromium silicates, a mixture though their range of manganese, silicon of manganese chromium silicates and and aluminum contents was more chrome galaxites (Cr203-MnO), and restricted. These slags contained as few 100% chrome galaxites, respectively, but as two or as many as six different oxides. all formed manganese chromium silicate Their wide range of compositions slags. However, Heat 1334, which con­ undoubtedly reflects differences in steel­ tained only 0.05% silicon, and hence Fig. 11 — Typical globular slag particle on a making practices since the samples were weld crater. 10X chrome galaxite inclusions, formed a slag from many different sources. Another of essentially the same composition. notable difference in regard to slag com­ The majority of the inclusions in heats positions between laboratory and pro­ current level required to just penetrate containing 0.010% or more aluminum duction heats was the presence of signif­ the strip. Slag particles formed on the top were alumina, but manganese chromium icant amounts of calcium, from the lime surface of the weld crater, sometimes on aluminum oxides and manganese chromi­ used in steelmaking, in the latter. the underside of the weld crater, and on um silicates were also present. With The slags formed with production the top and bottom surfaces of the weld these heats, alumina patch-type slags heats were classified on the basis of their bead. These slag particles were examined were formed on the top and bottom appearance in the same manner as the visually and classified according to their faces of the weld crater and the weld slags formed with the laboratory heats. appearance. They were basically two bead, but several small manganese chro­ The high-silica slags generally formed dis­ distinct types: more or less hemispherical- mium silicate globules containing varying tinct globules, whereas the high-alumina ly shaped globules, which were often amounts of aluminum were also slags formed thin patches. However, glassy in appearance (Fig. 11), and thin observed on the rear half of the weld there were a few exceptions, most patches-Fig. 12. In Fig. 12A, the patch- crater top surf ace — Table 5. The patch- noticeably the manganese chromium type slag is clearly visible as the comet- type slag present on the top surface of oxide and calcium titanium oxide slags, shaped gray area covering the weld the weld bead was extremely variable in which were the patch type, and the crater and forming a tail downstream. composition. It contained chromium, alu­ calcium titanium aluminum oxide and cal­ The weld discontinuity in Fig. 12B is a minum and silicon, varying amounts of cium chromium aluminum oxide slags, hump on the underside of the weld bead manganese and, with some heats, a small which were the globular type —Table 6. caused by the slag patch. Specimens with amount of calcium. The variable compo­ As with the laboratory heats, globular attached slag particles were cut from the sition of this slag was due to the fact that slags were formed when the silicon-to- welds. Qualitative energy dispersive it consisted of a mixture of the alumina aluminum ratio was greater than 50 and spectroscopic (EDS) analyses were then and manganese chromium aluminum sili­ patch-type slags when less than 50. performed on the slag particles and on cate slags in various proportions. In addi­ inclusions in polished longitudinal metallo­ tion, the manganese content of the slag graphic sections of the base metal. decreased with an increase in the alumi­ Effect of Slag on Weld Penetration num content of the steel. The small The effect of slag on weld penetration Globular slags had no significant effect amount of calcium present, which was was examined by measuring penetration, on weld penetration. Yet, when a patch- not added intentionally and not identified defined in this case as the ratio of the type slag was present, although width of in the inclusions, was probably from the weld root width to the weld face width, the weld face, Wp, was fairly constant, furnace lining. at intervals along a weld. local increases in root width, WR, were observed — Fig. 13. This variation in pene­ tration was more clearly shown by a plot of WR/WF versus distance along the weld. Penetration peaks were found to coincide with the locations of slag patches, which also produced humps on the surface of the weld bead.

Discussion Slagging Mechanism The results of this investigation confirm that the source of the slag formed in the Fig. 12 — Typical patch-type slags on: A — the weld crater; B — the underside of the weld bead.GT A welding of stainless steels is primarily 2X the oxide inclusions in the steel. A slag is

210-s I SEPTEMBER 1988 generated by the motion of molten metal small. This result is a consequence of the the inclusion compositions for the slags to in the weld pool transferring inclusions high melting point of the alumina inclu­ be considered simply as mixtures of all from the interior to the surface of the sions (2020°C/3668°F) (Ref. 30), which the inclusions in the steel. There are a pool. Not all slag particles, however, were solid in all regions of the weld pool number of possible explanations for dif­ remain on the surface of the weld pool. except the hot zone immediately ferences between the slag and inclusion Some are subsequently swept back into beneath the arc and, therefore, did not compositions. In most cases, the smallest the depths of the pool by the same coalesce and grow to a size where the inclusions analyzed were those visible at forces that brought them to the surface. buoyancy force was significant. In con­ 500X magnification in an optical micro­ Nevertheless, in the process of making a trast, silicate inclusions were found pri­ scope, which was used to locate areas long continuous weld, as in tubemaking, marily on the top surface of the weld for subsequent EDS analysis in a scanning there is a net transfer of inclusions to the pool. Since most silicates have melting electron microscope. It is quite possible surface of the weld pool, and the slag points in the range 1291° to 1723°C that smaller inclusions of different com­ particle formed increases in size until it is (2350° to 3133°F) (Ref. 30), they were positions could have been present in caught on the surface of the advancing molten in the weld pool and were able to some specimens. However, although weld bead. A new slag particle then coalesce and grow by collision with other these inclusions might modify the slag begins to form, and the whole process is inclusions until they attained a size such composition to some extent, it is unlikely that they would be the major constitu­ repeated. that the buoyancy force in combination ents of the slag when many larger inclu­ The motion of inclusions within the with Marangoni convection propelled sions were present. weld pool is a combination of an upward them to the surface of the weld pool. motion resulting from buoyancy forces A possible source of higher chromium and the pattern of fluid flow established contents in some slags, as compared to Slag Composition by Marangoni convection. Alumina slags the inclusions from which they were were formed on both the top and bot­ A comparison of inclusion and slag formed, is the passive film present on the tom surfaces of the weld pool, which compositions for the laboratory heats surface of stainless steel. However, this indicates that the effect of buoyancy (Table 5) showed that not all the heats film is usually very thin and would only be forces on the alumina inclusions was had slag compositions similar enough to significant in cases where the volume

Table 5—Inclusio n and Slag Analyses for Laboratory Heats Heat Percent Elements in Inclusions Elements in Slag No. Mn Si Al Si/AI Mn/Si Major Minor Trace Slag Location Slag Type Major Minor Trace

1329 0.12 0.47 0.005 94 0.26 Si Weld bead Patch Cr Si Ca, Al top surface 1330 0.50 0.49 0.002 245 1.02 Mn, Cr, Si Ti Crater top surface Globular Mn Cr, Si 1331 0.96 0.49 0.001 490 1.96 1)Mn, Cr Ti, Al, Si Crater top surface Globular Mn Cr, Si Ca 2)Mn, Cr, Si 1332 1.38 0.47 0.001 470 2.94 1)Mn, Cr Ti, Al, Si Crater top surface Globular Mn Cr, Si 2) Mn, Cr, Si 1333 2.02 0.50 0.001 500 4.04 Mn, Cr Ti, Al, Si Crater top surface Globular Mn Cr, Si Ca, Al 1334 1.46 0.05 0.001 50 29.20 Mn, Cr Ti, Al, Si Crater top surface Patch Mn Cr Ca, Al, Ti, Si 1335 1.45 0.26 0.005 52 5.58 Mn, Cr Ti, Al. Si Crater top surface Globular Mn Cr, Si 1336 1.48 0.73 0.001 730 2.03 Mn, Cr, Si Ti, W Crater top surface Globular Mn Cr, Si 1337 1.42 0.92 0.001 920 1.54 Mn, Cr, Si Ti, Al Crater top surface Globular Si Mn Cr Al 1346 1.53 0.50 0.010 50 3.06 1) Al Crater top Patch Al Si, Cr surface, front 2) Ma Cr, Al Ti Crater top Globular Mn Cr, Al, Si Ti surface, rear Weld bead top Patch 1)Cr, Al Si surface 2)Mn Cr, Al, Si Ca Crater bottom Patch Al Cr surface Weld bead Patch Al Cr bottom surface 1347 1.47 0.49 0.015 32.7 3.00 1) Al Cr Crater top Patch Al Cr Ca surface, front 2) Al Si Cr Crater top Globular Mn Cr, Al, Si surface, rear 3)Mn, Cr, Si Weld bead Patch 2) Al, Mn Cr, Ca top surface 2)Cr, Al, Si Crater bottom Patch Al Cr surface Weld bead Patch Al Cr bottom surface 1348 1.55 0.53 0.031 17.1 2.92 1) Al Crater top Patch Al Ca surface, front 2)Mn, Cr, Si Crater top Globular Mn Cr, Al, Si surface, rear Weld bead Patch Cr, Al, Si top surface Crater bottom Patch Al Cr surface Weld bead Patch Al bottom surface

WELDING RESEARCH SUPPLEMENT!211-s 0.8 er, silicon, reduced part of the manga­ HEAf 1348 (0.031 %AI) nese and chromium oxides in the manga­ ^ WELDING SPEED 30"/MIN nese chromium aluminum oxide inclu­ o sions to form a manganese chromium SLAG PATCHES aluminum silicate slag. \ 0.6 Slag Physical Characteristics

The melting point and surface tension CO of the molten slag at temperatures just UJ 0.4 X above the melting point of steel are u clearly the two properties which most strongly influence the behavior of weld­ ing slags. High-silica slags typically have low melting points and high surface ten­ £ 0.2 sions at temperatures close to the melting ft point of steel, so that they solidify as tx hemispherically shaped globules on the weld surface. In contrast, pure alumina slags have high melting points (2020°C/ 4 6 8 3668 CF), and on the bottom of the weld and the outer regions of the weld pool DISTANCE, INCHES surface, they form patch-type slags, Fig. 13 — Effect of slag patches on weld penetration which are actually rafts of solid alumina inclusions. Nevertheless, when the slag patch on the surface of the weld pool grows to a size such that it covers most fraction of oxide inclusions was very low complex slags formed in the same man­ of the weld pool surface, the portion or where very few of them become a ner will be richer in calcium, aluminum closest to the center of the weld pool will part of the slag. Heat 1329 is a good and titanium, and poorer in manganese example of the latter situation. and chromium than the inclusions in the be molten. However, since alumina slags apparently have lower surface tensions More frequently, the difference in steel. The results for the aluminum-con­ than high-silica slags, the solid and liquid composition between the slag and inclu­ taining laboratory heats show this trend. portions of the slag form a continuous sions consists of the slag being richer in Heats 1347 and 1348, which contained the more stable oxides, such as those of alumina and manganese chromium sili­ thin layer over the weld pool surface. calcium, aluminum, titanium and silicon, cate inclusions, formed a mixture of alu­ High-alumina slags containing other than the inclusions. The most likely rea­ mina and manganese chromium alumi­ oxides have significantly lower melting son for this difference is the much higher num silicate slags, with aluminum replac­ points than pure alumina, but form patch- temperatures that exist near the surface ing part of the manganese and chromi­ type slags, presumably because they also of a weld pool, as compared to those um. Heat 1346, which contained only have low surface tension. encountered in steelmaking. Howden 0.010% aluminum, contained alumina and and Milner estimated the effective reac­ manganese chromium aluminum oxide Effect of the Slag Type on the Weld Bead Geometry tion temperature of the hot zone at the inclusions, but formed a mixture of alumi­ na and manganese chromium aluminum surface of an iron weld pool under an Other investigators have stated that silicate slags. In this case, essentially all the argon arc to be 2100°C (3812°F) (Ref. weld slagging results in poor penetration aluminum was probably in the form of 31). At this temperature, a different equi­ (Refs. 7,12, 29). In this study, it was found oxide so that the next strongest deoxidiz­ librium exists between metal and slag that patch-type slags produced irregular than that which existed at steelmaking temperatures of about 1600°C (2912°F). This high-temperature equilibrium favors the formation of the more stable oxides. Table 6—Summary of Major Elements Found in Weld Slags Formed on Production Heats of Therefore, it is postulated that, provided Type 304 Stainless Steel they are present in sufficient quantities, elements such as calcium, aluminum, tita­ Patch Type Globular Type nium and silicon will tend to reduce the Mn Cr Mn, Si Ca, Si Al, Si less stable oxides, such as those of man­ Ca, Al ganese and chromium, during slag forma­ Ca, Ti Ca, Ti, Al tion. The simplest examples of weld slags Ti, M formed in this manner were the laborato­ Mn Cr, Al Mn, Cr, Si Ca, Mn, Si Ca, Al, Si Ca, Cr, Al ry Heats 1333 and 1335, which both M contained approximately 0.5% silicon and Mn Cr, Al Si Mn, Cr, Al, Si Ca, Mn, Cr, Si Ca, Mn, Al, Si formed manganese chromium silicate slags although they contained only Si chrome galaxite inclusions — Table 5. In Al contrast, laboratory Heat 1334, which Ca. Cr, Al, Si •— Ca, Cr, Al, Si Ca, Cr, Ti, Si also contained chrome galaxite inclusions, Si formed a manganese chromium oxide Al slag of similar composition because it Ca, Mn, Cr, Al, Si -<— Ca, Mn, Cr, AI, Si contained only 0.05% silicon, which was clearly insufficient to reduce the chrome galaxites. It is to be expected that more Ca, Mn, Cr, Ti, Al, Si Ca, Mn, Cr, Mg, Al, Si

212-s | SEPTEMBER 1988 penetration with localized regions of high making. Provided that they are present in Welding lournal 56(4): 126-s to 132-s. 13. Hazlett, T. H., and Parker, E. R. 1956. penetration occurring beneath slag sufficient quantities, elements such as cal­ Effect of individual coating ingredients on sur­ patches. The apparent disagreement with cium, aluminum, titanium and silicon tend face tension of iron electrode. Welding journal other investigators is due to the fact that to reduce the less stable oxides, such as 35(3):113-S to 114-s. elements that cause poor penetration, those of manganese and chromium, dur­ 14. Bradstreet, B. |. 1968. Effect of surface e.g., aluminum and magnesium, are also ing slag formation. tension and metal flow on weld bead forma­ major causes of slagging. A possible 3) Slags formed with production mate­ tion. Welding journal 47(7):314-s to 322-s. explanation for the effect of a patch-type rial contain from two to six oxides of 15. Ishizaki, K. 1966. Interfacial tension the­ slag on weld penetration is that it con­ calcium, aluminum, titanium, magnesium, ory of arc welding phenomena — formation of stricts the arc root so that magnetohydro- silicon, manganese and chromium. How­ welding bead. Welding Research Abroad 12(3):68-86. dynamic flow within the weld pool ever, in general, the slagging characteris­ 16. Ishizaki, K. 1962. Interfacial tension the­ increases. When a slag patch is caught on tics of stainless steel were found to be ory of the phenomena of arc welding — the surface of the advancing weld bead, determined primarily by the ratio of sili­ mechanism of penetration. Physics of Welding weld penetration abruptly decreases. con to aluminum in the steel. Patch-type Arc Symposium, The Welding Institute, Cam­ However, a new slag patch soon starts to slags were formed when % silicon/% bridge, U.K., pp. 195-209. form, penetration increases and the aluminum was less than 50 and globular 17. Roper, j. R., and Olson, D. L. 1978. whole cycle is repeated. slags when it was greater than 50. Capillarity effects in the GTA weld penetration Patch-type slags also cause the forma­ 4) To prevent the formation of patch- of 21-6-9 stainless steel. Welding journal 57(47): 104-s to 107-s. tion of humps on the weld bead surface. type slags, which cause an uneven weld 18. Friedman, E. 1978. Analysis of weld These humps are due to the fact that, at bead and irregular penetration, the alumi­ puddle distortion and its effect on penetration. temperatures just above the freezing num content must not exceed 0.010% for Welding lournal 57'(6): 161-s to 166-s. point of the steel, the metal-slag interfa­ a silicon content of 0.50%. 19. Mills, G. S. 1977. Analysis of high man­ cial tension is higher than the metal-gas ganese steel weldability problem. Welding interfacial tension. A ckno wledgments lournal 56(6): 186-s to 188-s. 20. Woods, R. A., and Milner, D. R. 1971. Practical Significance of the Slag Type The author wishes to thank D. ). Motion in the weld pool in arc welding. in Tubemaking Modrak for performing the welding and Welding lournal 50(4): 163-s to 173-s. P. I. Henderson for the EDS analysis. 21. Willgoss, R. A. 1980. Electromagnetic The results of this investigation have control of fluid motion in TIG weld pools. Arc confirmed tubemill experience that Physics and Weld Pool Behavior, The Welding References patch-type slags are to be avoided Institute, Cambridge, U.K., Vol. 1, pp. 361- because they produce an uneven weld 1. Oyler, G. E., Matuszesk, R. A., and Carr, 373. bead and irregular penetration. It was C. R. 1967. Why some heats of stainless steel 22. Allen, B. C. 1972. The surface tension of also shown that in most cases globular may not weld. Welding journal 46(12)4006- liquid metals. Liquid Metals— Chemistry and instead of patch-type slags will be formed 1011. Physics. Ed. S. Z. Beer, M. Dekker, Inc., New if the %Si/%AI ratio is greater than 50, 2. Willgoss, R. H., and Yeo, R. B. G. 1975. York, N.Y., pp. 161-212. i.e., if for a silicon content of 0.50% the Union Carbide Corp., Linde Division, Research 23. Keene, B. )., Mills, K. C, Bryant, I. W., report. aluminum content is less than 0.010%. and Hondros, E. D. 1982. Effects of interaction 3. Majetic, |. C, and Yeo, R. B. G. lune 8, between surface active elements on the sur­ Globular slags are generally less harmful 1971. Method of Welding Stainless Steel. U.S. face tension of iron. Canadian Metallurgical than patch-type slags, except for critical Patent 3,584,187. Quarterly 21(4):393-403. applications such as dairy tubing. Some 4. Bukarov, V. A., and Ishchenko, Yu. S. 24. Oreper, G. M., Eagar, T. W., and Szeke­ welding lines utilize rollers downstream 1974. Melting of metal and seam formation in ly, ]. 1983. Convection in arc weld pools. from the arc to contour the weld while it the welding of a steel of Type 18-8 with an Welding lournal b2(-}-\)-307-s to 312-s. is still hot. Slag globules are rolled into the oxidized surface. Welding Production 21(12): 25. Romanov, A. A., and Kochegarov, V. G. weld surface and then etched out during 22-25. 1964. Viscosity of binary iron-silicon melts in pickling forming pits, which are unaccept­ 5. Bennett, W. S., and Mills, G. S. 1974. GTA the low concentration range of the second able on dairy tubing since they are poten­ weldability studies on high manganese stainless component. Physics of Metals and Metallogra­ tial sites for the growth of bacteria. steel. Welding lournal 53(12):548-s to 553-s. phy 17(2):136-138. Although weld slagging cannot be com­ 6. Franklin, I. E. June 25, 1975. Effects of 26. Romanov, A. A., and Kochegarov, V. G. Addition Alloys on the Weldability of 304 1964. Viscosity of Fe-Mn, Fe-P, Fe-Cr, and Fe-V pletely eliminated, for this critical applica­ Stainless Steel. INCO Research report 600.5. melts in the initial concentration range of the tion it must be minimized by using clean 7. Metcalf, ]. C, and Quigley, M. B. C. second component. Physics of Metals and steel and welding under conditions that 1977. Arc and pool instability in GTA welding. Metallography 18(6):67-73. produce a narrow weld bead, so that the Welding lournal 56(5): 133-s to 139-s. 27. Heiple, C. R„ Burgardt, P., and Roper, volume of metal melted and hence the 8. Heiple, C. R„ Cluley, R. ]., and Dixon, |. R. 1984. The effect of trace elements on slag volume are held to a minimum. R. D. 1980. Effect of aluminum on CTA geo­ GTA weld penetration. Modelling of Casting metries in stainless steel. Physical Metallurgy of and Welding Processes II. Eds. |. A. Danzig and Metal joining. Eds. R. Kossowsky and M. E. |. T. Berry. TMS-AIME, Warrendale, Pa., pp. Conclusions Glicksman. The Metallurgical Society —AIME, 193-205. Warrendale, Pa., pp. 160-165. 28. Heiple, C. R., and Burgardt, P. 1985. 1) Slag is formed in the CTA welding 9. Heiple, C. R., and Roper, ]. R. 1982. Effects of SO2 shielding gas additions on GTA of stainless steel when oxide inclusions Mechanism for minor element effect on GTA weld shape. Welding journal 64(6): 159-s to are transferred from the interior to the fusion zone geometry. Welding Journal 61(4): 162-s. surface of the weld pool by a combina­ 97-s to 102-s. 29. Linnert, G. F. 1967. Weldability of aus­ tion of buoyancy forces and Marangoni 10. Ludwig, H. C. 1968. Current density and tenitic stainless steels as affected by residual convection. anode spot size in the gas tungsten arc. elements. Effects of Residual Elements on 2) The composition of the slag is not Welding journal 47{5):234-s to 240-s. Properties of Austenitic Stainless Steels, ASTM Symposium lune 1966. ASTM Pub. No. 418. always equivalent to a mixture of all the 11. Glickstein, S. S., and Yeniscavich, W. May 1977. A Review of Minor Element Effects 30 CRC Press, Inc. CRC Handbook of inclusions present in the steel because it on the Welding Arc and Weld Penetration. Chemistry and Physics, 60th Edition. Boca tends to approach an equilibrium corre­ Welding Research Council Bulletin 226. Raton, Fla. 1979-80. sponding to temperatures near the sur­ 12. Savage, W. F., Nippes, E. F., and Good­ 31. Howden. D. G., and Milner, D. R. 1963. face of the weld pool, which are much win, G. M. 1977. Effect of minor elements on Hydrogen absorption in arc melting. British higher than those encountered in steel­ fusion zone dimensions of Inconel® 600. Welding lournal 42(6):304-316.

WELDING RESEARCH SUPPLEMENT I 213-s WRC Bulletin 328 November 1987 This bulletin contains two reports covering related studies conducted at The University of Kansas Center for Research, Inc., on the CTOD testing of A36 steel. Specimen Thickness Effects for Elastic-Plastic CTOD Toughness of an A36 Steel By G. W. Wellman, W. A. Sorem, R. H. Dodds, Jr., and S. T. Rolfe

This paper describes the results of an experimental and analytical study of the effect of specimen size on the fracture-toughness behavior of A36 steel. An Analytical and Experimental Comparison of Rectangular and Square CTOD Fracture Specimens of an A36 Steel By W. A. Sorem, R. H. Dodds, Jr., and S. T. Rolfe

The objective of this study was to compare the CTOD fracture toughness results of square specimens with those of rectangular specimens, using equivalent crack depth ratios.

Publication of these reports was sponsored by the Subcommittee on Failure Modes in Pressure Vessel Materials of the Pressure Vessel Research Committee of the Welding Research Council. The price of WRC Bulletin 328 is $20.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, Suite 1301, 345 E. 47th St., New York, NY 10017.

WRC Bulletin 330 January 1988 This Bulletin contains two reports covering the properties of several constructional-steel weldments prepared with different welding procedures.

The Fracture Behavior of A588 Grade A and A572 Grade 50 Weldments By C. V. Robino, R. Varughese, A. W. Pense and R. C. Dias

An experimental study was conducted on ASTM A588 Grade A and ASTM A572 Grade 50 microalloyed steels submerged arc welded with Linde 40B weld metal to determine the fracture properties of base plates, weld metal and heat-affected zones. The effects of plate orientation, heat treatment, heat input, and postweld heat treatments on heat-affected zone toughness were included in the investigation.

Effects of Long-Time Postweld Heat Treatment on the Properties of Constructional-Steel Weldments By P. J. Konkol

To aid steel users in the selection of steel grades and fabrication procedures for structures subject to PWHT, seven representative carbon and high-strength low-alloy plate steels were welded by shielded metal arc welding and by submerged arc welding. The weldments were PWHT for various times up to 100 h at 1100°F (593°C) and 1200°F (649°C). The mechanical properties of the weldments were determined by means of base-metal tension tests, transverse-weld tension tests, HAZ hardness tests, and Charpy V-notch (CVN) impact tests of the base metal, HAZ and weld metal.

Publication of these reports was sponsored by the Subcommittee on Thermal and Mechanical Effects on Materials of the Welding Research Council. The price of WRC Bulletin 330 is $20.00 per copy, plus $5.00 for postage and handling. Orders should be sent with payment to the Welding Research Council, 345 E. 47th St., Suite 1301, New York, NY 10017.

214-s I SEPTEMBER 1988