Effects of SO2 Shielding Gas Additions on GTA Weld Shape Small additions of SO2 in argon shielding gas for GTA welding both suppresses variability and improves joint penetration in 21-6-9 and 304 stainless steels BY C R. HEIPLE AND P. BURGARDT ABSTRACT. Substantial increases in GTA joint penetration, can vary substantially For pure metals and many alloys, the weld depth/width ratio resulted from between different heats (and even within surface tension decreases as temperature small additions of sulfur dioxide (S02) to the same heat) of material with the same increases (Ref. 2). For weld pools in such the torch shielding gas when welding two nominal composition. materials, the surface tension will be stainless steels. The improvement was Heiple and Roper (Ref. 1) have recently greatest on the coolest part of the pool demonstrated on both Types 304 and proposed a mechanism for the origin of surface at the edge, and lowest on the 21-6-9 austenitic stainless steels, but variable GTAW joint penetration by hottest part under the arc near the center would be expected for iron-base alloys which weld pool shape can be substan­ of the pool. Such a surface tension gradi­ generally. The weld pool shape achieved tially altered by variations in the concen­ ent produces outward surface fluid flow was essentially independent of variations tration of certain trace elements in the (as shown schematically in Fig. 1A) and in both SO2 content of the torch gas and weld pool. They proposed that the major generates a relatively wide and shallow base metal composition when SO2 in the factor determining weld shape is general­ weld. shielding gas was in the range of 500 to ly fluid flow in the weld pool and that, Surface-active trace elements, when 1400 ppm. With 700 ppm S02 in the under many common GTA welding con­ present in sufficient quantity in the weld torch gas, less than 30 ppm sulfur was ditions, the major factor driving fluid flow pool, create a positive surface tension added to an autogenous weld bead. For is a surface tension gradient. Trace ele­ temperature dependence (Ref. 2). Under alloys where this additional sulfur can be ments affect weld pool shape by altering these conditions, the surface tension will tolerated and appropriate measures can surface tension gradients, thereby chang­ be highest near the center of the weld be taken to handle the toxic SO2, this ing the magnitude and/or direction of pool and an inward surface fluid flow technique offers a promising way to fluid flow in the weld pool. results (as shown schematically in Fig. 1B). improve GTA weld joint penetration while suppressing variable penetration. Introduction The shape of GTA welds has long been of concern, in part because the GTA welding process tends to be used in applications where high quality, high pre­ cision welds are required. Furthermore, the welds are often made by automatic equipment, which is presently unable to compensate for variability in weld pool geometry once the welding variables have been set. For these reasons, consid­ WELDING erable interest has been generated by the DIRECTION observations that weld shape, especially Paper presented at the AWS 66th Annual Convention, held April 28-May 3, 1985, in Las Vegas, Nev. C. R. HEIPLE is in Physical Metallurgy R&D and P. BURGARDT is in Nuclear joining R&D at B Rockwell International, Rocky Flats Plant, North Fig. 1 — Proposed (Ref. 1) surface ana subsurface fluid flow in the weld pool. A —surface tension American Space Operations, Golden, Colo. temperature coefficient negative; B — surface tension temperature coefficient positive WELDING RESEARCH SUPPLEMENT 1159-s This fluid flow pattern produces relatively The second experiment was designed welding conditions for both the experi­ deep, narrow weld beads. to determine if variable joint penetration ment with varying S02 shielding gas con­ In iron-base alloys, sulfur and oxygen could be suppressed with S02 shielding tent and the experiment with doped base are the surface-active trace elements gas additions. Strips in a 17 mm (0.67 in.) metal were: current— 150 A DCEN; elec­ most commonly present. Historically, sul­ thick plate of 304L stainless steel were trode to work distance —2 mm (0.080 fur and oxygen levels have been high doped with either sulfur or selenium, and in.); travel speed— 150 mm/min (6 ipm); enough so that complete joint penetra­ GTA welds were subsequently passed electrode tip angle —45 degrees; and tion was observed consistently, i.e., weld across the doped zones and undisturbed shielding gas flow rate —0.57 m3/hr (20 depth/width (d/w) ratios of about 0.5. base metal. The doping technique cfh). Finally, both autogenous welds and However, improvements in steelmaking involved electron beam melting foils of welds with cold filler metal addition were practice have made available steels, par­ 303 stainless steel (for sulfur doping) and made on several joint configurations to ticularly stainless steels, which have 303 Se stainless steel (for selenium dop­ confirm that the changes observed in extremely low sulfur and oxygen con­ ing) into the base plate. The addition of bead-on-plate welds would also be seen tents (less than 10 ppm each). Such steels sulfur (Ref. 3) and selenium (Ref. 4) to in full joint penetration welds. exhibit poor joint penetration in GTA stainless steel is known to drastically welding and can have weld d/w ratios as increase GTA weld d/w ratio when the Results low as 0.15. These wide, shallow weld steel has an initially low d/w ratio. The pools create a number of significant doping procedure has been described in SO2 in the shielding gas had a substan­ welding problems. detail previously (Ref. 5). The 304L plate tial effect on weld pool shape. Weld d/w used in this experiment had an exception­ One approach to improving GTAW ratio is plotted versus torch gas SO2 ally low sulfur and oxygen content and joint penetration and reducing variable content in Fig. 2. Cross sections of welds exhibited a d/w ratio of about 0.17. The joint penetration is to add small concen­ made using pure argon and 750 ppm S02 composition is given in Table 2. It was trations of surface active elements to the are shown in Fig. 3. One of the encour­ chosen to give the largest possible differ­ weld pool. This can be accomplished in a aging aspects of the results presented in ence in weld shape between the doped number of ways, including the addition of Fig. 2 is that there is a substantial range zones and undoped base metal. The a doped filler metal wire to the weld (approximately 500-1400 ppm) of S02 pool, coating the joint prior to welding, contents for which the d/w ratio is inde­ and doping the shielding gas delivered pendent of SO2 concentration. This can through the torch. The additions need to be small, since most elements known to Table 2—Analysis of 304L Plate, wt-% be surface active in iron (S, Se, and Te) promote hot cracking. We report here Cr 19.0 the effects of small SO2 shielding gas Ni 10.3 Table 3 —Sulfur Contents of Welds on additions on GTA welds on 21-6-9 and Mn 1.9 21-6-9, ppm 304 austenitic stainless steels. C 0.034 Al 0.002 Argon Argon + Argon + o 0.001 2 Shield 750 ppm 2000 ppm S02 Experimental S 0.002 so2 Fe balance 40 67 150 The initial experiment consisted of pro­ ducing GTA bead-on-plate welds on a 25 mm (1 in.) thick plate of 21-6-9 stainless steel. Since the intent of the experiment was to determine if S02 additions to the torch gas would improve weld d/w ratio, a plate with known-poor d/w ratio was used. The composition of the plate is given in Table 1. All welds were made under identical welding conditions except that the torch gas composition was va­ ried between pure argon and argon con­ taining 2000 ppm S02. The welding was performed in a large chamber filled with argon. The SO2 content in the shielding gas was varied by mixing pure argon with gas from a second cylinder containing argon with 2000 ppm SO2. Table I—Analysis of 21-6-9 Plate, wt-% UJ Cr 20.1 >UJ Ni 6.8 Q Mn 8.7 X N2 0.27 O C 0.025 200 400 600 800 1000 1200 1400 1600 1800 2000 tr Al 0.0058 TORCH GAS S02 < 02 0.0058 CONTENT, ppm cUoJ S 0.0024 UJ Fe balance Fig. 2 - Effect of sulfur dioxide in the shielding gas on GTA weld depth/width ratio. 21-6-9 stainless steel 160-slJUNE 1985 0.5 S ft X ' o \ 0.4 - / O \ i l I \ I \ t i a 0.3 D -i A UJ £ \ \ N 0.2 o 0.1 a 100 AMPS O 150 AMPS Fig. 3-Cross sections of GTA welds made A 200 AMPS with (top) pure argon shielding gas and (bot­ tom) argon plus 750 ppm 502 shielding gas I J_ 500 1000 1500 2000 TORCH GAS OXYGEN be contrasted with the results obtained in CONTENT(PPM) a similar experiment when oxygen was Fig. 4-Effect of oxygen in shielding gas on GTA weld depth/width ratio (Ref. 6) added to the shielding gas, where a relatively sharp maximum in d/w ratio is obtained with increasing oxygen concen­ tration—Fig. 4. Welds made with shield­ was added to the weld; the increase was gas additions, are shown in Fig.