WELDING RESEARCH SUPPLEMENT TO THE WELDING JOURNAL. MARCH 1989 Sponsored by the American Welding Societv and the Welding Research Council All papers published in the Welding Journal's Wekfalg Research Supplement undergo Peer Review before publication for: 1) originality of the contribution; 2) technical value to the welding community; 3) prior publication of the material being reviewed; 4) proper credit to others working in the same area; and 5) justification of the conclusions, based on the work performed. The names of the more than 170 individuals serving on the AWS Peer Review Panel are published periodically. All are experts in specific technical areas, and all are volunteers in the program. Submerged Arc Welding: Evidence for Electrochemical Effects on the Weld Pool It is hypothesized that weld metal chemistry is significantly affected by an electrochemical reaction at the weld pool-flux interface BY J. E. INDACOCHEA, M. BLANDER AND S. SHAH ABSTRACT. Compositional changes in toughness essentially equivalent to the occurring during heat treatment of steel. weld metal from welds made by sub­ base metal. However, because of the large noniso- merged arc welding of steel using CaF2- It is well established in submerged arc thermal behavior encountered in the CaO-Si02 fluxes are consistent with an welding (SAW), as in steel making, that welding processes, the circumstances electrochemical mechanism in which the low oxygen levels usually lead to better affecting the phase transformations in a filler wire is anodically oxidized to form toughness (Refs. 3-5). There seems to be weld are significantly different from those oxides and fluorides, and metals are general agreement that a microstructure occurring in steel production. For in­ cathodically deposited at the weld pool- normally consisting of acicular ferrite (Fig. stance, the nonmetallic inclusion volume flux interface. This speculative mecha­ 1) (Ref. 6) will yield the best weld metal fraction in low-carbon steel welds is nism, if correct, could make it simpler to mechanical properties in terms of much larger than that of normal steel predict flux compositions, which will strength and toughness, by virtue of its products, due to the very short time improve the quality of welds and control small grain size (typically 1 to 3 jum) and available for growth and flotation of such weld metal microstructures. high-angle grain boundaries (Ref. 7). particles (Ref. 8). For a given welding flux, The microstructural changes taking the final weld metal oxygen content has Introduction place in the weld metal on cooling been observed to be directly related to through the critical transformation tem­ the inclusion population (Ref. 9). On the During the past two decades, the steel perature are, in principle, similar to those other hand, oxide inclusions are known industry has experienced tremendous to strongly influence the transformation technological progress leading to many from austenite to ferrite, both by restrict­ alloys possessing excellent mechanical ing the austenite grain growth (Fig. 2), as and corrosion-resistant properties. Re­ well as by becoming nucleation sites for KEY WORDS cent developments in, for example, HSLA different ferrite morphologies. steel plate manufacturing technology Submerged Arc Welding High-temperature ferrite products, (Refs. 1, 2) have called for new formula­ Electrochemical Effects such as grain boundary ferrite (GF) and tions of welding consumables to produce Weld Pool Chemistry polygonal or blocky ferrite (PF) are pre­ weld metal deposits with strengths and CaF2-CaO-Si02 Fluxes dominant in welds with high inclusion Anodic Oxidation densities; while welds with low inclusion Weld Metal Composition densities have microstructures consisting /. £ INDACOCHEA and S. SHAH are with the Composition Changes primarily of lower temperature transfor­ University of Illinois at Chicago, Metallurgical Oxidized Welding Wire mation products, acicular ferrite and the Department, Chicago, III. M. BLANDER is with Cathodic Deposits Argonne National Laboratory, Chemical Tech­ aligned martensite-austenite-carbide (AC) Flux Predictions nology Division/Materials Science and Tech­ structure. These microstructures are nology Program, Argonne, III. shown in Fig. 3 (Ref. 10). WELDING RESEARCH SUPPLEMENT 177-s TM ft v %J iV ItV .»» • -j X > • t - >X« I % V ' i 5* i 1 ' /vg. 1 —Acicular ferrite morphology in a low-carbon-steel weld microstructure. A—A low-heat-input weld, 50.3 kj/in. (198 kj/mm); B — for a high-heat-input weld, 97.4 kj/in. (3.83 kj/mm). 25 fim In view of the significant role of oxy­ weld metal. In this paper, we consider the measurements have indicated that the gen on the austenite-to-ferrite transfor­ possible importance of electrochemical higher the basicity indexes, the lower the mation of low-carbon steel welds, and, effects on the chemistry of weld metal oxygen content of the weld metal. The Bl consequently, its effect on notch tough­ produced in submerged arc welding is, however, a poor measure of the ness, an understanding of the mecha­ (SAW) with CaF2-CaO-Si02 fluxes. chemistry of slags. In addition, it is proba­ nisms that control the oxygen, in particu­ Thermodynamic models have been ble that equilibrium is not attained and lar, and the chemistry of the weld metal, proposed to predict the final composition that kinetic factors are important. For in general, is important for producing of submerged arc welds (Refs. 11-17). In example, Thier and Dring (Ref. 12) pro­ high-quality welds. view of the very high temperatures posed a diffusional model to predict the Much prior work on weld chemistry involved and small molten volumes, final content of elements in the weld has focused primarily on kinetically limit­ some investigators (Ref. 11) assume that metal. Thier and Dring concluded that ed thermodynamically driven reaction equilibrium is attained. Davis and Bailey the slag composition where an element mechanisms. This pyrometallurgical ap­ (Ref. 11) propose that the transfer of was not transferred to the metal, the proach has proven to be useful in analyz­ elements between the slag and the weld neutral point, is affected only by the ing some data on weld metal chemistry. metal depends on the oxide activities in current, but not by the voltage changes However, such analyses have not yet led the slag, which are directly connected to and that the neutral point is characteristic to methods of predicting the chemistry of the basicity index (Bl) of the slags. Limited for a specific type of flux. Ekstrom and Olson (Ref. 14) reported that the change in Si in weld metal is influenced by the basicity of the slag and that this influence is higher when the o D & o LOW HEAT INPUT basicity index is less than two. Dallam, ef I30 al. (Ref. 15), found that while the Si level • • A •HIGH HEAT INPUT in the weld metal was correlated with the Ui basicity index, the Mn content of the weld depends on the amount of MnO in 55 I30 the slag. Indacochea, et al. (Ref. 16), z qualitatively showed a similar correlation < between flux and weld metal composi­ o I 10 tion. Despite qualitative agreement among researchers regarding the flux type and direction of elemental transfer, there is \- 90 no precise determination of a "neutral </> z> •0° A point," the composition where a given < element is not transferred, even though rr 70 very similar types of fluxes were used in o several of these studies. This failing may rr % o°-^_ o be attributed to the possibility that the rx mechanism of elemental transfer is not so O — — 50 simple, to the influence of different weld­ ing parameters used, and to differences I00 300 400 700 900 IIOO I300 I300 in the wire and flux compositions. It is clear that essentially all prior studies WELD METAL OXYGEN CONTENT (PPM) have been relatively narrowly defined. A Fig. 2 —Prior austenite grain size as a function of oxygen content in the weld metal for two heat more comprehensive approach is inputsfRef. 6). The 'prior austenite" is the parent phase for the lower temperature microstructural needed in which the diverse kinetic and products thermodynamic factors involved in SAW 78-s|MARCH 1989 fe3?' 2%, tJHr*f " " * • " Sfcflf? t:. <lMkMp GF - . r. / .:. ,J Ps l^fc*'^B 5%: /:' # . ? : 'V *c ,, SfiH - ,. ^Vi* *^ - ^ , fig. i — Typical ferrite morphologies present in low-carbon-steel weld microstructures. CE- Grain boundary ferrite; PE-polygonal or blocky ferrite; AC—aligned M-A-C (martensite-austenite-carbide); and acicular ferrite in the unlabeled micrograph (Ref. 6) (Refs. 16, 17) are examined and which have not been considered before. All In this paper, we examine the possibili­ also explores the possibility of mechanis­ welding fluxes when molten are, at least ty that the changes in chemistries of weld tic factors that have not been considered. in part, ionic and the number of cou­ metal with changes in the compositions Such an approach is made difficult by the lombs passed per mole of metal is very of CaF2-containing fluxes might be relat­ complexity of the processes in SAW. The large. If even a small fraction of the total ed to an electrochemical mechanism. DC voltages and currents are very high current (<0.01) is involved in a Faradayic (e.g., 30 V, 400 A) and lead to a system process, electrochemical effects at metal- containing the four principal phases, a flux interfaces could be a major factor Experimental Work weld wire, a molten flux, a plasma arc controlling the chemistry of weld metal Materials and a weld pool with five interfaces (Ref. 18). In addition, because the very among them. Most of the current is hot plasma is essentially electrically neu­ Low-carbon steel coupons of dimen­ transported by electrons in the plasma tral and in contact with the relatively sions 63.5 X 254.0 X 12.7 mm (2.5 X 10 from the generally cathodic weld pool to volatile fluxes, there have to be at least as X 0.5 in.) and a 2.38-mm (0.094-in.) diam­ the generally anodic weld wire (or filler many positive ions as there are electrons eter filler welding wire (AWS Type ER70S- wire).