Effect of Deoxidation Sequence on Carbon Manganese Steel Weld Metal Microstructures

Deoxidation sequence has a strong influence on nonmetallic inclusions formation and the subsequent weld refinement

BY F-C. LIAO AND S. LIU

ABSTRACT. During investigation of indi­ boundary ferrite and Widmanstatten side­ the molten metal for the purpose of vidual and combined effects of aluminum plate ferrite are often grouped together as deoxidation. Some common deoxidizers and titanium on low-carbon low-alloy primary ferrite (PF). As the weld tempera­ used in steel making are aluminum, silicon, steel weld metal microstructures, experi­ ture continues to drop, fine acicular ferrite manganese, and titanium. In arc welding, mental welds were made on 12.7-mm (AF) laths begin to nucleate intragranularly. these same elements are also used for (0.5-in.) thick A516 C70 pressure vessel Finally, the remaining austenite transforms weld pool deoxidation and alloying. They steel plates using ER70S-3 welding wire to a variety of microstructural features, enter the weld pool from the base metal, and a low-oxygen potential commercial which include bainite (B), martensite (M) electrode, or fluxes. In ladle refining of flux. In the aluminum or titanium (individ­ and pearlite (P). Together with the re­ steels, the deoxidation reactions occur at ual) addition series welds, the results indi­ tained austenite, martensite and carbides near isothermal and equilibrium condi­ cated that the final weld metal microstruc­ form the microconstituents known as tions quite different from those found in tures are related to the inclusion size dis­ MAC. The relative proportions of the dif­ most welding conditions. The nonisother- tribution and the amount of aluminum or ferent transformation products are mal, nonequilibrium nature of arc welding titanium in solid solution. In the aluminum- strongly influenced by the nonmetallic in­ makes it very difficult to identify clearly titanium or titanium-aluminum (combined) clusions in the weld metal. the effect of deoxidation sequence in the addition welds, deoxidation sequence Several models have been proposed to weld pool on weld metal microstructure. plays an important role in the formation of explain the effects of nonmetallic inclu­ In submerged arc welding, oxygen specific types of nonmetallic inclusions sions such as TiO and MnO • AI2O3 on the comes mainly from the flux, which may and in the determination of solid solution formation of acicular ferrite (Refs. 1-5). contain easily reduced oxides such as iron elements content, which are fundamental However, the influence of deoxidizers oxide, manganese oxide and silica (Ref. 6), in microstructural refinement. (individually or combined) on inclusion which often exist in a dissociated state, formation and inclusion size distribution is that is, metal cations and oxygen anions, in the weld pool. During solidification, the Introduction still not fully understood. This research fo­ cuses on the weld pool deoxidation se­ weld pool oxygen concentration estab­ In order to satisfy many critical engi­ quence to better understand the relation­ lished at high temperatures will readjust as neering applications, steel weld metals ship between weld metal microstructures a result of decreasing oxygen solubility with high strength and high toughness and nonmetallic inclusions. and the combination of oxygen with deox­ have been widely investigated in recent idizers that exist in the weld metal. Kluken years to determine the factors that control and Grong (Ref. 7) divided the weld pool Weld Pool Deoxidation Practice weld metal microstructure. During the into two reaction zones. One is the "hot" austenite-to-ferrite decomposition, the In steel making, elements with higher reaction zone, immediately beneath the first transformation product is grain affinity for oxygen than iron are added to arc, where the deoxidation products are boundary ferrite (CBF), which forms along continually separated by highly turbulent prior austenite grain boundaries. Follow­ flows that sweep those products to the ing, Widmanstatten sideplate ferrite (SP) trailing edge of the weld pool. The other nucleates and grows from grain boundary KEY WORDS is the "cold" reaction zone where most of ferrite as long needle-like laths that pro­ the precipitated products are entrapped trude into the austenite grains. Grain Deoxidation Sequence in the weld metal as finely dispersed inclu­ Weld Metal Microstructure sion particles. A516 Chemical Composition Aluminum Additions Inclusions Formation F-C. LIAO and S. LIU are with the Center for Titanium Additions Welding and joining Research, Colorado School Inclusion Composition In fusion welding, the level of residual of Mines, Golden, Colo. Weld Sequence oxygen in liquid iron is strongly dependent Transformation Paper presented at the 71st Annual AWS on the amount and kind of deoxidizers Meeting, held April 22-27, 1990, in Anaheim, Hardenability added. If only one deoxidizer is used in the Calif. welding system, the deoxidation sequence

94-s I MARCH 1992 and products are much easier to predict. When the residual oxygen content is less nominal heat input for the first pass weld­ In reality, however, multiple deoxidizers than 0.06 wt-%, CX-AI2O3 is the product, as ing was approximately 3.0 kj/mm (76 kj/ are used and complex inclusions arc described in Equation 5 (Ref. 11). in.). The welding parameters are shown in formed. The amounts of alloying elements Table 2. After the first passes were made, 2AI + 4FeO ^ and deoxidizers strongly affect the final (l) (l) V-grooves were machined from the cen­ . AI 0 -Fe0 + 3Fe i) (4) inclusion's composition, size, and shape. 2 3 {s) ( ter of the beads. Gas tungsten arc (GTA)

2AI(i, + 3FeO(l) — Al2Ol(i) + 3Fe(l) (5) welds with eight levels of aluminum addi­ Manganese as Deoxidizer tions, from 0.009 to 0.228 wt-%, were In low-carbon steel weld pools, where made with argon shielding with a nominal When a steel weld pool is deoxidized oxygen content is reasonably low, it is heat input of approximately 2.9 kj/mm by manganese alone, the final deoxidation common to observe clusters of AI2O3 (74 kj/in.). GTA welding was chosen to product will be rich in MnO with a small particles. ensure that the variations in oxygen con­ amount of FeO (Ref. 8). The deoxidation tent observed in the second passes were reaction in the weld pool is (Refs. 9,10) Titanium as Deoxidizer caused mainly by aluminum addition since all other elements were maintained con­ Mri(i) + FeO, • MnO(i) + Fe,•m (1) Using titanium as the only deoxidizer, a 10' stant. Due to the nonuniform bead shape number of oxides are formed. At increas­ with of some of the second passes, a third au­ ing oxygen content, the following se­ togenous GTA weld pass at 2.9 kj/mm AG°= -29,469+ 13.57 (2) quence of phases is expected: TiO, Ti20 , 3 was applied in the transverse direction of Ti 0 , Ti0 , FeO-Ti0 , and Fe Ti0 (Ref. In Equation 2, T is the weld pool tem­ 3 5 2 2 2 4 all welds to homogenize the weld metal. 11). The final form of the inclusions, how­ perature. The inclusion type and shape The welding sequence is shown schemat­ ever, may be rich in TiO (because of its are strongly dependent on the ratio of ically in Fig. 2. thermodynamic stability)—Fig. 1. MnO/FeO. With high manganese level However, in real arc welding processes, The second part of the experiment fol­ MnO, (high ), many inclusions will exhibit more than one deoxidizer is added to the lowed similar procedures as the first part' FeO weld pool and the deoxidation sequence with the exception of the sequence of dendritic morphology (Ref. 11). At a low- is complicated and most of the inclusions deoxidizers addition (aluminum in the first manganese level (low _ _), however, occur in a combined form, with multiple pass and titanium in the second pass). oxides. Some sulfides and oxysulfides may the inclusions are always spherical and also be present. Analyses of the Experimental single phase (Ref. 11). Welds Materials and Welding Procedure Silicon as Deoxidizer For reasons indicated previously, only the first and third passes were studied. If silicon alone is used as the deoxidizer The base metal used in this research was an ASTM A516 G70 pressure vessel They were examined using a light micro­ in the welding system, the final deoxida­ steel. A 3.2-mm (Vs-in.) diameter ER70S-3 scope to evaluate quantitatively the vol­ tion product in the weld metal is solid sil­ electrode and a high MgO-CaF low Si0 ume fractions of the various microstruc­ ica (Refs. 8,11) and the deoxidation reac­ 2/ 2 commercial flux were used in the experi­ tures. Systematic two-dimensional point tion can be written as ments. The compositions of the base counting was performed on each of the % + 2FeO(l) — Si02(s) + 2Fe(i) (3) metal and the welding consumables are weld specimens. Among all quantitative given in Table 1. methods available in phase proportions At the presence of other deoxidizers, A two-part experiment concerning the determination, the technique of system­ however, a liquid silicate product is ex­ sequence of weld pool deoxidation was atic two-dimensional point counting re­ pected to form. designed. The first part examined the tita­ sults in the smallest relative error, approx­ nium-aluminum addition sequence and imately ten percent. Aluminum as Deoxidizer the second part, the aluminum-titanium The chemical composition of the weld ln the case of deoxidation by aluminum, addition sequence. In the case of the tita­ passes were analyzed using an optical the final products can be AI2O3 or a com­ nium-aluminum addition sequence welds, emission spectrometer and reported in pound of AI2O3 and FeO. If the oxygen the first pass of each weld was made by Table 3. The carbon, sulfur, oxygen, and content is greater than 0.06 wt-%, that is, bead-in-groove submerged arc welding nitrogen content of the weld metals were low aluminum content in the molten metal, (SAW) to contain eight different levels of determined using interstitial analyzers. In­ Al203 • FeO is formed, Equation 4 (Ref. 11). titanium, from 0.007 to 0.355 wt-%. The clusions from the different passes were

Table 1—Chemical Compositions of the Base Plate, the ER70S-3 Electrode, and the Commercial Flux Used in this Research

C S Mn P Si Cr Mo Ti Cu Al (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%)

Base metal A516 0.257 0.012 1.20 0.008 0.174 0.023 0.002 0.001 0.019 0.028 G70 PVQ V2 in. ER70S-3 Electrode 0.124 — 1.152 0.021 0.56 — — — 0.070 0.010 Si02 AI2O3 MgO CaO MnO Ti02 CaF2 Na20 Fe203 C (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) Commercial flux 10.7 17.3 31.7 6.6 1.1 0.86 24.1 0.78 1.9 0.35

WELDING RESEARCH SUPPLEMENT I 95-s - 50 First Pass

151-100 u -V-*£-"' __£ c j£>^ .g Second Pass E Ui O fc -150 €>^<^ ^£^$§5^ c ^/sp^ PJ ££» u '—\^jy^ ^^oi-- <0' ^

Third Pass

-250 i i i 500 1000 1500 2000 Temperature (CC)

F/£. 7 —Free energy of formation of some oxides (Ref. 17). Fig. 2—Schematic drawing of the welding sequence.

isolated using carbon extraction replica Table 2—Summary of Welding Parameters for the Experimental Welds technique and analyzed with a scanning electron microscope. A minimum of 500 Welding Welding Travel Nominal particles were examined from each weld Welding Voltage Current Speed Heat Input and the diameter of the particles mea­ Process (V) (A) (mm/s) (KJ/mm) sured directly on the microscope. The First pass SAW 31 340 3.5 3.0 chemical compositions of these inclusions Second pass CTAW 16 250 1.4 2.9 were determined using SEM/EDS (energy Third pass GTAW 16 250 1.4 2.9 dispersive spectroscopy).

Results and Discussion Individual Effects Relationship between Weld Metal Chemical Table 3-- Chemical Composition of the Experimental Welds Discussed Composition and Microstructures

C Mn Si Ti Al O N To correlate weld metal microstruc­ Weld (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) tures with chemical composition, the hard­ enability of the weld metals according to t1 0.143 0.84 0.19 0.007 0.005 218 50 the IIW carbon equivalent (CE) equation t2 0.151 0.88 0.20 0.014 0.007 221 49 (Ref. 12) was first considered. Only the t3 0.149 0.95 0.22 0.036 0.009 283 49 welds with similar CE values (0.30 wt-%) Tl 0.127 0.90 0.20 0.046 0.006 217 58 were evaluated to determine the alumi­ Tl 0.141 0.85 0.28 0.201 0.012 274 56 a1 0.132 0.88 0.18 0.002 0.009 258 54 num and titanium effects and the results a2 0.132 0.88 0.22 0.003 0.013 246 54 shown in this paper. a3 0.107 0.88 0.20 0.003 0.026 302 il In the first pass weldments, Figure 3 A1 0.156 0.96 0.21 0.003 0.039 365 55 shows that there is a threshold concen­ A2 0.138 0.87 0.25 0.003 0.054 418 59 tration of titanium (approximately 0.05 a1t1 0.139 0.72 0.14 0.010 0.004 191 68 wt-%) above which bainite increased at a1t2 0.118 0.72 0.14 0.043 0.004 179 87 the expense of acicular ferrite and grain a1t3 0.119 0.67 0.12 0.083 0.004 146 66 boundary ferrite. The increase in bainite at t3a1 0.109 0.58 0.12 0.005 5^ 0.013 93 higher titanium concentrations seems to t3a2 0.148 0.74 0.17 0.022 0.024 188 55 suggest that titanium in excess of 0.05 t3a3 0.122 0.64 0.15 0.019 0.033 178 53 wt-% may exist in the form of solid solu-

96-s I MARCH 1992 100 • AF% 90 o GBF% 80 A B% 70 3 60 - A O _3- 50 • O ^*" A w 40 "•&• n "^""i s ^\~~———^o o B 30 c 0) a> o \- 0 u 20 .O w a. a. 10

I.I. u.00 0.05 0.10 0.15 0.20 0.25 u.00 0.02 0.04 0.06 0.08 0.10 0.12 Weld Metal Ti Content (wt.%) Weld Metal Al Content (wt.%)

Fig. 3 — Variation in weld metal microstructure as a function of weld Fig. 4 — Variation in weld metal microstructure as a function of weld metal titanium concentration. metal aluminum concentration. tion and promote the formation of bain­ Note that the large amount of fine intra­ be observed to decrease with increasing ite. At low-titanium concentrations, both granular features shown in Fig. 5C is not titanium content. Figure 9, however, acicular ferrite and grain boundary ferrite acicular ferrite. More careful examination shows that prior austenite grain size re­ increased at the expense of bainite with showed that those are, in fact, bainite, as mained approximately constant with alu­ titanium addition. indicated by the arrows in Fig. 7. minum additions. This seems to indicate The variation of weld metal microstruc­ that the welds with titanium additions may Relationship between Prior Austenite Grain tures with only aluminum addition is shown also have inclusion populations that pin Size and Chemical Composition in Fig. 4. Below approximately 0.055 wt-% the austenite grain boundaries more ef­ of aluminum, the volume fraction of acic­ Another factor to be considered in fectively and restrict austenite growth. ular ferrite and bainite increased with alu­ weld metal phase transformation is the The aluminum bearing welds did not ex­ minum, while grain boundary ferrite de­ prior austenite grain size. For a similar heat hibit the same behavior. In the case of creased. Above this point, grain boundary input (and thus, cooling rate), the different weld specimens with titanium addition, ferrite and bainite increased at the ex­ alloying elements and their levels of addi­ the austenite grain size observation also pense of acicular ferrite. This result agrees tions showed different effects on prior agreed with the weld metal microstruc­ with the data reported by Brownlee (Ref. austenite grain size. Figures 8 and 9 show ture. Large austenite grain size also exhib­ 13). Figures 5 and 6 are light micrographs the variation of prior austenite grain size as ited smaller amount of grain boundary taken from specimens with different lev­ a function of titanium and aluminum addi­ ferrite. The effect of inclusions on weld els of titanium and aluminum addition. tion. In Fig. 8, prior austenite grain size can metal phase transformations will be dis­ cussed in the following paragraphs.

•$*L*mv

Fig. 5 — Weld metal microstructure showing mixture of grain bound­ ary ferrite, PF, SP, acicular ferrite, and bainite for different levels of titanium addition. A -0.007 wt-% (40% AF,21% PF, 34% B); B -0.014 wt-% (44% AF, 27% PF, 29% B);C—0.201 wt-% (20% AF, 34% PF, 46% wm B).

(b)

WELDING RESEARCH SUPPLEMENT I 97-s k'J--

«5 ^i ?icw^ r%. J#£^A-^ __* **~ m\i:J

k t fflt&

F;g. 7—High magnification micrograph showing that the fine intra­ granular features shown in Fig. 5C are not acicular ferrite, but rather, bainitic in nature (with carbide between parallel ferrite laths).

•60

Fig. 6 — Weld metal microstructure showing mixture of grain bound­ ary ferrite, PF, SP, acicular ferrite, and bainite for different levels of aluminum addition. A —0.009 wt-%;B—0.054 wt-%; C—0. 118 wt-%.

u.OO 0.05 0.10 0.15 0.20 0.25

Weld Metal Tl Content (wt.%) Fig. 8 — Variation of prior austenite grain size as a function of weld metal titanium addition.

uo £ • a. t • • • 120 • = 5 100 e « 30 u

9 60

40 o a 20

u.OO 0.02 0.04 0.06 0.08 0.10 0.12

Weld Metal Al Content (wt.%) Fig. 9 — Variation of prior austenite grain size as a function of weld metal aluminum addition.

98-s I MARCH 1992 50 50 t3 t2

40 • _ 40

30 30 c 3 3 a- 20 5) a- 20 -

10 • 10

1 1 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95 1 05 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95 105 Inclusion Diameter (pm) Inclusion Diameter (pm) Fig. 10—Simple size distribution of the inclusions extracted from: A —Specimen t30.03 (Ti 6 wt-%; Al = 0.009 wt-%; B-Specimen t2 Ti = 0.014 wt-% Al = 0.007 wt-%) with only titanium addition.

Influence of Inclusion Size Distribution on shown in Fig. 11B, which further supports Figures 12 and 13 show the variation of Weld Metal Transformation the discussion that along with inclusion oxide types in inclusions as a function of weld metal aluminum and titanium con­ From Figs. 10 and 11, it is clear that the size, the size distribution of the inclusions tent. To simplify the discussion, it was as­ aluminum addition series welds exhibited is also important in defining the final weld sumed that only simple oxides such as inclusion populations with larger particles metal microstructures. Al 0 , TiO, Si0 , and MnO resulted dur­ than the titanium addition welds (for ex­ 2 3 2 ing deoxidation and that these oxides ample, the number of inclusions with Influence of Weld Metal combine to form the final inclusions. Fig­ diameter larger than 0.7 p.m). This obser­ Composition on the Inclusion ure 12 shows that the amount of AI2O3 in vation explains in part the small variation Oxide Types the inclusions increased with aluminum of austenite grain size with weld metal addition, while other oxides such as Si02, aluminum content, since large inclusions With the addition of different levels of MnO, and TiO decreased. Figure 13shows are ineffective in pinning grain boundaries. titanium or aluminum, the chemical con­ that with titanium additions TiO is the main Figures 10A and 11A show the inclusion stitution of the inclusions was also quite component. Small amounts of AI2O3, size distribution in the optimal titanium different. In C-Mn-Si steels weld metals, MnO, and Si02 were present but de­ and aluminum addition welds, 0.050 and the inclusions contain mainly SiOi and creased with increasing titanium content. 0.055 wt-%, respectively. No small parti­ MnO. As aluminum is added, the compo­ cles (with diameter <0.2 pm) were ob­ sition of the inclusions were observed to served and the particle size mode of both change, with increasing AI2O3 content. Combined Effects welds was 0.45 pm. These welds also ex­ Cochrane, ef al. (Ref. 16), observed simi­ hibited higher volume fraction of acicular lar behavior in his investigation. Increasing Due to the large number of welds made ferrite than the other welds, which seems weld metal titanium concentration led to in both the aluminum-titanium (A-T) se­ to agree with the findings of Barbaro, etal. an enrichment of titanium in the inclusions. quence and titanium-aluminum (T-A) se­ (Ref. 14), and Jang, etal. (Ref. 15), that in­ TiO is assumed to be the predominant quence, a typical set was chosen from clusions of diameters within the range of phase in the inclusions because of the each group to illustrate the effect of 0.4 to 0.6 pm are more efficient in acicu­ many reports in the literature (Refs. 17-19) deoxidation sequence on weld metal lar ferrite formation. Cochrane, etal. (Ref. and the greater thermodynamic stability transformation behavior. 16), also reported similar particle size (AGf) of TiO compared to the other tita­ range being effective in acicular ferrite nium oxides —Fig. 1. Liu and Olson (Refs. Weld Metal Composition Effect in nucleation. 20-21), and Bhatti, et al. (Ref. 22), have Aluminum-Titanium Sequence Welds At higher aluminum additions, the weld also observed similar results in their inves­ The A-T sequence welds (a1t1, a1t2, metals exhibited even larger particles, as tigations. and a1t3) showed that titanium additions

50 A2 A3

40 " 40

30 - 30

20 20 -

10 •

J I, I I, I I, I II I 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0 35 0.93 105 0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.95 105 Inclusion Diameter (um) Inclusion Diameter (um) Fig. 11 —Inclusion size distribution as a function of weld metal aluminum content. A —Al = 540 ppm and Ti = 30 ppm; B —Al = 1180 ppm and Ti = 30 ppm.

WELDING RESEARCH SUPPLEMENT I 99-s 100 100

0.009 0.013 0.039 0.054 0.118 0.112 0.007 0.014 0.036 0.046 0.201 0.206

Weld Metal Al Content (wt.%) Weld Metal Ti Content (wt.%) Fig. 12 —Effect of weld metal aluminum content on inclusion compo­ Fig. 13 —Effect of weld metal titanium content on inclusion compo­ sition. sition. in the second pass decreased the oxygen quent titanium addition during the second 20 ppm of titanium. Only 171 ppm of ox­ content residual from Weld al. Note that pass. On the other hand, MnO and Si02 ygen remained to combine with other el­ Weld a1 was deoxidized with aluminum formed initially are assumed to be reduc­ ements such as manganese and silicon re­ alone. With the addition of titanium in the ible to manganese and silicon by titanium sulting in MnO and Si02- Indeed, the second pass, the oxygen content of Weld addition. Finally, it is assumed that all tita­ presence of manganese oxide and silica a1 was reduced to 191,179, and 146 ppm nium atoms in the weld metal are in the was confirmed by inclusion analyses in Welds a1t1,a1t2, and a1t3, respectively form of TiO —Fig. 1. shown in Fig. 12. (indicated as closed circles in Fig. 14). To Chemical analysis showed that the first During the second pass welding (a1t1), explain the microstructural changes ob­ pass (weld a1 with only aluminum addi­ the oxygen content dropped from 258 to served in these welds as a function of ox­ tion) has the following composition: 191 ppm—Fig. 14. The loss could be ygen and titanium content, it is assumed Al = 90ppm,Ti = 20 ppm, O = 258 ppm, attributed to the coalescence of inclu­ that with sufficient oxygen in the system, and N = 54 ppm. Of the 258 ppm of ox­ sions, which were later eliminated by flo­ all aluminum in the weld metal will form ygen, approximately 80 ppm of oxygen tation or convective flow in the weld the stable AI2O3 and that these alumina combined with the 90 ppm of aluminum pool. This is verified by the reduction of particles will not be reduced by subse­ and 7 ppm of oxygen combined with the aluminum from 90 ppm to 40 ppm after

0.030

al [O] = 2S8 ppm 100 0.025 • AF % 90 O GBF % 80 . a tl - 0.020 A B% , a 1.2 70 o Al.Oj o - Al.Oj "*"*' "•^^altS 5C «c 0.015 01 50 Al-O, MnO 40 0.010 • SiO, others TiO 30 2 TiO 3 20 3 0.005 •

" •c TiO 0.000 U.OO 0.02 0.04 0.06 0.08 0.10 0.02 0.04 0.06 0.08 0.10

Weld Metal Tl Content (wt.%) Weld Metal Ti Content (wt.%) Fig. 14 — Variation of weld metal oxygen content as a function of ti­ Fig. 15 — Variation of weld metal microstructure as a function ot tita­ tanium additions in the aluminum-titanium group welds. nium additions in the aluminum-titanium group welds.

100-s I MARCH 1992 the titanium addition. In this weld, 36 ppm Table 4—Chemical Composition of Selected Experimental Welds Showing the Total and Acid- of oxygen were consumed to form AI2O3 Soluble Contents of Aluminum and Titanium in Low-Carbon Low-Alloy Steel Weld Metal with the 40 ppm of aluminum. To react with the 100 ppm of titanium, 33 ppm of Al Al Ti Ti oxygen were needed. The remaining ox­ c O N total soluble total soluble ygen in the second pass combined with (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) (wt-%) manganese and silicon to form MnO and Si02, which explains the inclusion analysis 0.09 0.037 0.005 0.02 0.002 0.025 0.018 reported earlier. The three vertical lines in 0.09 0.039 0.005 0.037 0.004 0.022 0.018 Fig. 14 were drawn to illustrate the relative 0.09 0.040 0.006 0.044 0.006 0.058 0.046 0.10 0.031 0.005 0.062 0.013 0.032 0.029 amounts of Al 0 , TiO, MnO, and S.O2 2 3 0.09 0.031 0.006 0.053 0.008 0.053 0.052 that would result in each weld after the addition of the second deoxidizer. The relative proportions of these oxides in the inclusions were also confirmed experi­ that despite the high affinity of titanium for Weld Metal Composition Effect in mentally in this work. carbon and nitrogen, a significant amount Titanium-Aluminum Sequence Welds Increasing titanium addition in the sec­ of titanium atoms remain in solution, de­ ond weld, the amount of residual oxygen termined as acid soluble titanium—Table The T-A welds (t3a1, t3a2, and t3a3) decreased further. Weld a1t2 contained 4 (Ref. 7). showed that aluminum increased the sec­ 40 ppm of aluminum, 430 ppm of tita­ From the calculations above, it is easy to ond pass weld metal oxygen content to a nium, and 179 ppm of oxygen. Following explain the variation of microstructures in maximum point followed by a slight drop, the procedure described in the previous Fig. 15. Figure 15 shows that maximum as shown in Fig. 16. From Fig. 17, it can be paragraph, it was determined that 143 acicular ferrite was observed at approxi­ seen that above 0.024 wt-% of aluminum ppm of oxygen was necessary to combine mately 0.042 wt-% of titanium. Above this addition, acicular ferrite is replaced by with 430 ppm of titanium and no oxygen point, acicular ferrite was replaced by bainite. would be available for other elements. bainite. Below this point, acicular ferrite Below 0.024 wt-% of aluminum, acicu­ This confirms the observation of numer­ increased at the expense of grain bound­ lar ferrite increased at the expense of ous titanium-bearing particles in the inclu­ ary ferrite. From Fig. 14, TiO was ob­ bainite. Grain boundary ferrite did not sion population. The increase in acicular served to increase with titanium additions seem to be affected by aluminum addi­ ferrite in the weld is then attributed to the at the composition range below 0.04 tion. larger amount of TiO inclusions in the wt-%, which led to the conclusion that the In the case of Weld t3 (with titanium weld. Since a small amount of MnO and volume fraction of acicular ferrite in­ addition in the first pass), the "residual" Si02 was detected in the inclusions, some creased with increasing TiO. Above the composition is Al = 90 ppm, Ti = 360 titanium must have remained in the weld optimal value, uncombined titanium at­ ppm, and O = 283 ppm. Since approxi­ metal in the form of nitride or solid solu­ oms remain in the weld pool and behave mately 80 ppm of oxygen are required to tion. as hardenability agent, promoting bainite combine with the 90 ppm of aluminum to Further increasing the titanium content formation. In all A-T group welds, high-ti­ form aluminum oxide, only 203 ppm of to 830 ppm, Weld a1t3 contained 40 ppm tanium addition led to high-volume frac­ oxygen remained to combine with tita­ of aluminum and 146 of oxygen. In this tions of bainite, sometimes close to 100% nium, manganese, silicon, and other minor weld, 36 ppm of oxygen combined with (Ref. 25). elements. Stoichiometric calculations 40 ppm of aluminum to form AI2O3. Assuming that all remaining oxygen atoms reacted with titanium to form TiO, then the 110 ppm of oxygen would have con­ 0.030 sumed only 330 ppm of titanium. This cal­ t3 [0] = 283ppm culation shows that at least 500 ppm of ti­ tanium must have remained in the weld metal, in the form of carbide, nitride, and «_; 0.025 - solid solution. During solidification, only a fraction of c the titanium atoms will form nitride or £ 0.020 c carbide (as individual particles or on the o surface of already existing inclusions) in o t3a2 the liquid steel. Taking into account the c t3a3 solubility product of TiN and TiC in aus­ o 0.015 - / D) tenite (Ref. 23) and that the nucleation and >. growth of TiN and TiC are controlled by x O titanium diffusion (Ref. 24), only part of 75 0.010 the titanium atoms will form TiN and TiC 4_- on the outer layer of existing inclusions 0) t3al (oxide particles) (Refs. 7,24). The remain­ ing titanium atoms will be dispersed in the 2 0.005 weld metal as solid solution atoms and hardenability agent. Experimental evi­ 5 dences of Kluken and Grong (Ref. 7) and 1 . 1 Es-Souni and Beaven (Ref. 24) seem to 0.000 u.OO 0.01 0.02 0.03 0.04 support the above statement. These au­ thors investigated the differences be­ Weld Metal Al Content (Wt.%) tween total and acid soluble contents of Fig. 16 — Variation of weld metal oxygen content as a function of aluminum addi­ aluminum and titanium, and determined tions in the titanium-aluminum group welds.

WELDING RESEARCH SUPPLEMENT I 101-s u.OO 0.01 0.02 0.03

Weld Metal Al Content (wt.%) Weld Metal Al Content (wt.%) Fig. 17 — Variation of weld metal microstructure as a function of alu­ Fig- 18 —Variation of inclusion volume fraction (Vv) as a function of minum additions in the titanium-aluminum group welds. aluminum additions in the titanium-aluminum group welds. showed that only 120 ppm of oxygen are of oxygen is in the form of MnO and Si02- and silicon atoms remained in the form of needed to combine with titanium to form This is also confirmed by inclusion analysis. solid solution and promoted bainite for­ TiO. Therefore, MnO and Si02 are formed Weld t3a2 has 240 ppm of aluminum, mation. and is confirmed by the inclusion analysis. 220 ppm of titanium, and 188 ppm of ox­ Additionally, the optimal aluminum con­ During the second pass welding, Weld ygen, assuming that all oxygen available centration and acicular ferrite volume frac­ t3a1, the oxygen content dropped from has reacted with aluminum to form AI2O3. tion could also be explained as the result 283 to 93 ppm. The decrease was prob­ The 188 ppm of oxygen will combine with of an increase in inclusion volume frac­ ably due to inclusion loss by convective 212 ppm of aluminum, which means that tion-Fig. 18. Above 0.023 wt-% of alu­ flow in the weld pool, and coalescence the weld metal does not have enough ox­ minum, considerable amounts of alumi­ and flotation, because AI2O3 inclusions ygen to react with all aluminum. There­ num, titanium, manganese, and silicon are are known to coalesce quite readily (Ref. fore, some aluminum atoms, together in solution which increased the harden­ 7). After adding aluminum in the second with most of the titanium, manganese and ability of the alloy and promoted bainite pass, Weld t3a 1 contained 50 ppm of alu­ silicon atoms, will remain in the weld metal formation at the expense of acicular fer­ minum and 130 ppm of titanium. The de­ as solid solution atoms and become hard­ rite. The decrease in inclusion density ob­ creasing aluminum observed was an indi­ enability agents. In this case, the increasing served in weld specimens where alumi­ cation that indeed inclusions were lost by amount of titanium bearing inclusions was num concentrations were below 0.023 flotation or convective flow in the weld responsible for the increase of acicular wt-% confirms the coalescence and floata­ pool. Assuming that all 50 ppm of alumi­ ferrite in the weld metal. tion of aluminum oxide particles (Ref. 7). num are in the form of oxide, then 44 ppm Following the similar procedure, the Figure 19 illustrates the importance of of oxygen are needed. If all titanium are in 178 ppm of oxygen in Weld t3a3 were deoxidation sequence by showing the mi­ the form of titanium oxide (requiring 43 calculated to combine with 200 of the 330 crostructures of two welds with the same ppm of oxygen), mass balance showed ppm of aluminum to form AI2O3, which weld pool aluminum and titanium addi­ that only 6 ppm of oxygen would remain, means that at least 130 ppm of aluminum, tions, but at different addition sequence. which means that only very little amount most of titanium (190 ppm), manganese These micrographs show that low-carbon

Fig. 19 —Optical micrographs illustrating the importance of deoxidation sequence on weld metal microstructure. A—T1A1, predominantly grain bound­ ary ferrite, SP and acicular ferrite; B—A1T1, predominantly bainite.

102-s I MARCH 1992 steel weld metal microstructure is indeed 4) Weld metal with a larger amount of Treating of Steel, ed., H. E. McGannon. United a function of deoxidation sequence. The acicular ferrite are found to exhibit coarser States Steel Corp. titanium-aluminum sequence weld (Fig. prior austenite grains with a large number 10. Falconi Campos, V. 1974. Theory of de­ oxidation. Deoxidation and Solidification of 19A), showed a mixture of grain boundary of nonmetallic inclusions of diameter Steels. Associacao Brasileira de Metais. ferrite, Widmanstatten sideplate ferrite, around 0.45 pm. 11. Kiessling, R. 1978. Non-metallic Inclu­ and acicular ferrite, while the aluminum- sions in Steels. Part l-IV. The Metal Society, titanium weld (Fig. 19B) exhibited almost Combined Effect London, England. exclusively bainite. In the case of the latter 12. Ito, Y., and Bessyo, K. 1968. Weldability weld, the small amount of oxygen remain­ 1) Deoxidation sequence is important formula of high strength steels related to heat- ing after deoxidation with aluminum was in weld pool deoxidation and microstruc­ affected zone cracking. IIW DOC IX-576-68. insufficient to react with titanium to form tural refinement. Aluminum and titanium 13. Brownlee, |. K. 1985. Effects of alumi­ titanium oxide. The lack of nucleation sites (in solid solution) increase the weld metal num and titanium on the microstructure and for acicular ferrite and higher acid-soluble hardenability which results in increasing properties of microalloyed steel weld metal. titanium content contributed to the bai­ bainite. Colorado School of Mines, M. Sc. Thesis No. T-3064, Golden, Colo. nitic microstructure. 2) In the aluminum-titanium (A-T) group 14. Barbaro, F.)., Krauklis, P., and Easterling, welds, higher titanium additions resulted To utilize the beneficial effects of deox­ K. E. 1989. Formation of acicular ferrite at oxide in higher titanium in solution, which led to idation sequence as discussed previously, particles in steels. Materials Science and Tech­ different types of electrodes must be used higher volume fraction of bainite. nology 5(11):1058-1068. on multipass welding. As an example, in 3) In the titanium-aluminum (T-A) group 15. )ang, J. W., Shah, S., and Indacochea,). thick-plate fabrication using tandem weld­ welds, no matter how high the aluminum E. 1987. Influence of SAW fluxes on low-carbon ing, the chemical composition of the elec­ addition, the weld metal microstructure steel weld microstructure. Journal of Materials trodes can be customized such that the always show a mixture of bainite, grain for Energy Systems 8(4):391-401. sequence of addition of deoxidizers into boundary ferrite and acicular ferrite. 16. Cochrane, R. C, Ward, J. L, and Keville, B. R. 1983. The influence of deoxidation and/or the weld pool can be controlled to max­ 4) The concept of welding systems us­ desulphurisation practice on the weld metal imize acicular ferrite in the weld metal. ing multiple wires (in tandem) that contain toughness of high dilution welds. The Effect of From the present work, it is known that different deoxidizers can be used to bet­ Residual, Impurity and Micro-alloying Elements the titanium-aluminum sequence is pre­ ter control and refine the weld metal mi­ on Weldability and Weld Properties. Paper 16, ferred. Then, the combination of leading crostructure. The Welding Institute, U.K. electrodes that contain adequate amounts 17. Kohno, R., Takami, T., Mori, N., and Na­ of titanium followed by trailing electrodes Acknowledgment gano, K. 1982. New fluxes of improved weld with optimal aluminum content will result metal toughness for HSLA steels. Welding Jour­ The authors appreciate and acknowl­ in improved microstructure and mechan­ nal 61(12):373-s to 380-s. edge the research support of the Office of ical properties. The sequenced deoxida­ 18. Mori, N., Homma, H„ Ohkita, S., and Naval Research. tion will also result in weld metal oxygen Wakabayashi, M. 1981. Mechanisms of notch levels much lower than those achieved by toughness improvement in Ti-B-bearing weld metals. IIW DOC IX-1196-81. individual or simultaneous additions of References 19. Mori, N., Homma, H., Wakabayashi, M., deoxidizers. 1. Abson, D. ). 1978. The Role of Inclusions and Ohkita, S. 1982. Characteristics of mechan­ in controlling Weld Metal Microstructures in ical properties of Ti-B-bearing weld metals. IIW C-Mn Steels. The Welding Institute, Cambridge, DOC IX-980-82. Conclusions England. 20. Liu, S. 1984. The role of nonmetallic in­ 2. Ferrante, M., and Farrar, R. A. 1982. The- The results of this investigation on alu­ clusions in controlling weld metal microstruc­ role of oxygen-rich inclusions in determining tures in niobium microalloyed steels. Colorado minum and titanium deoxidation sequence the microstructure of weld metal deposits. School of Mines, Ph.D. Thesis T-2923, Golden, can be summarized in the following: lournal of Materials Science 17:3293. Colo. 3. Bhadeshia, H. K. D. H, and Svensson, L.-E. 21. Liu, S., and Olson, D. L. 1987. The influ­ 1991. Modelling the evolution of microstruc­ Individual Effect ence of inclusion chemical composition on ture in steel weld metal. IIW DOC ll-A-846-91. weld metal microstructure. Journal of Materials 1) Increasing weld metal titanium con­ 4. Ricks, R. A., Howell, P. R., and Barrite, Engineering 9(3):237-251. tent up to approximately 0.05 wt-% in­ C. S. 1982. The nature of acicular ferrite in HSLA 22. Bhatti, A. R., Saggese, M. E., Hawkins, steel weld metals. Journal of Materials Science creased the amount of acicular ferrite and D. N., Whiteman, J. A., and Golding, M. S. 1984. 17:732. grain boundary ferrite at the expense of Analysis of inclusions in submerged arc weld in 5. Liu, S., and Olson, D. L. 1986. The role of bainite. At titanium contents above 0.05 microalloyed steels. Welding journal 63(7): inclusions in controlling HSLA steel weld micro- 224-s to 230-s. wt-%>, both acicular ferrite and grain structures. Welding journal 65(6):139-s to 23. Suzuki, S., Weatherly, G. C, and Hough- boundary ferrite are replaced by bainite. 150-s. ton, D. C. 1987. The response of carbo-nitride 2) Increasing weld metal aluminum con­ 6. Choi, C. S., and Eagar, T. W. 1981. Slag- particles in HSLA steels to weld thermal cycles. tent up to approximately 0.055 wt-% metal equilibrium during submerged arc weld­ Acta. Met. 35(2):341-352. ing. Met. Trans. A 12B:539-547. increased the amount of acicular ferrite 24. Es-Souni, M., and Beaven, P. A. 1989. and bainite at the expense of grain bound­ 7. Kluken, A. O, and Grong, O. 1989. Microanalysis of inclusion/matrix interfaces in ary ferrite. High aluminum content, that is, Mechanisms of inclusion formation in Al-Ti- weld metals. Application of Surface and Inter­ Si-Mn deoxidized steel weld metals. Met. Trans. above 0.055 wt-%, resulted in less acicu­ face Analysis-FCASIA 89. Antibes, France. A 20(8): 1335-1349. lar ferrite. 25. Liao, F. C. 1990. The effect of deoxida­ 8. Twidwell, L. G. Physical Chemistry of Iron 3) Titanium-containing inclusions have tion sequence on carbon manganese steel weld and Steelmaking. Montana College of Mineral metal microstructures. Colorado School of better austenite grain boundary pinning Science and Technology. Mines, M. Sc. Thesis No. T-3928, Golden, Colo. effect than aluminum-bearing inclusions. 9. USS. 1971. The Making, Shaping and

WELDING RESEARCH SUPPLEMENT I 103-s