Analysis of Inclusions in Submerged Arc Welds in Microalloyed Steels

The level of weld bead acicular ferrite is found to be less in Ca-treated vs. Ca-free base metal of the same composition

BY A. R. BHATTI, M. E. SAGGESE, D. N. HAWKINS, J. A. WHITEMAN, AND M. S. GOLDING

Introduction ered the importance of inclusion crystal­ In this study, a range of weld beads lography. Others (Refs. 8, 9) do not share made on API 5LX65 base material have Toughness requirements for arctic this view and have instead concentrated been investigated to simulate the high grade line pipe have become increasingly on the indirect effect that inclusion chem­ dilution situation in practical pipe manu­ more stringent with the advent of tech­ istry may have on hardenability. ln their facture. Two base metals were selected nological demanding applications. This view, acicular ferrite is apparently devel­ with similar composition and low sulphur has led to developments in steel technol­ oped when the threshold value of 1.1 content, with one plate having under­ ogy giving increased through-thickness wt-% Mn is exceeded in the weld metal gone inclusion modification by calcium properties, with a move to low sulphur matrix. treatment. Three wires were steels and inclusion control by the use of selected; they were Oerlikon Welding rare earth treatments (e.g., Ca). Industries' S2, SD3Mo and Tibor 22 (con­ It has been shown (Ref. 1) that these taining microalloyed additions of Ti and B, Table 1—Base Metal Composition, wt-% types of steels provide excellent resis­ and identified as no. 22 elsewhere in this tance to hydrogen-induced cracking paper). The submerged arc fluxes cov- Ca-free Ca-treated (HIC) in sour gas environments. However, steel processing variables such as calcium C 0.052 0.060 treatment can have an indirect effect on Mn 1.37 1.45 the properties obtained in submerged arc Si 0.224 0.142 Table 2—Base Metal Microstructural Features weld metal in terms of generating a S 0.005 0.006 microstructural phase — acicular ferrite — P 0.018 0.024 which is known to optimize toughness Ni 0.003 0.003 Microstructural Base metal Cr 0.032 0.032 properties in unrefined weld metal. It is feature Ca-free Ca-treated Mo 0.012 0.007 known that the formation of acicular Nb 0.036 0.032 Ferrite, % 96 91 ferrite is favored in a particular range of Cu 0.34 0.32 Pearlite, % 4 9 oxygen content; however, there is some V 0.069 0.070 Ferrite grain 6.15 6.55 disagreement regarding the factors favor­ Sn 0.007 0.006 size, Mm ing the formation of this phase. Al(sol) 0.054 0.025 Pearlite colony 3.57 6.88 Many authors (Refs. 2-6) think that N 0.003 0.006 size, /am inclusions play an important role in nucle­ ation. In this case, the controlling factors, in determining the effectiveness of inclu­ sions as nucleants, are volume fraction Table 3—Welding Wire Composition, Wt-% and size distribution. More recently, a some workers (Ref. 7) have also consid- Element S2 SD3Mo No. 22< > C 0.10 0.10 0.10 Mn 1.06 1.60 1.19 Si 0.22 0.13 0.04 Based on a paper presented at the 64th Annual s 0.012 0.012 0.010 AWS Convention in Philadelphia, Pennsylva­ p 0.013 0.007 0.008 nia, on April 29, 1983. Ni 0.07 0.08 Cr 0.03 0.10 A. R. BHATTI is a Research Fellow and D. N. Mo 0.47 0.30 HAWKINS and J. A. WHITEMAN are Senior Cu 0.11 Lecturers, Sheffield University, England; M. E. Ti 0.043 SAGGESE is with CNEA Argentina, and M. S. B 0.004 GOLDING is Technical Services Manager, Oer­ likon Welding (Ireland) Limited, Dublin. (a) Marketed by Oerlikon Welding Industries Ltd.. Houston

224-s I JULY 1984 ered a range of basicities (Ref. 10) from Base Metal Effect acid to fully basic; these were flux A with Figure 2 shows the percentage of acic­ a basicity index (Bl) = 3.1, flux B with ular ferrite (AF) plotted against the basici­ Bl = 1.8, and flux C with Bl = 0.6. ty index (Bl) of the flux with both welding wire and base metal composition as inde­ pendent variables. Weld metal micro- Experimental structure in Ca-free base metal always had a larger proportion of acicular ferrite All the beads were deposited under than that in the Ca-treated base metal; identical conditions: process — sub­ however, the decrease in acicular ferrite merged , current —600 A, level is influenced by the flux-welding m voltage —32 V, speed —460 mm/min wire combination. With the microalloyed (18.1 ipm), and heat input (HI)-2.5 kj/ no. 22 welding wire, the reduction is mm (62.5 kj/in.). relatively small, regardless of the flux The base metals were 16 mm (0.63 in.) basicity index. For the other two welding thick, and details of their chemistry and wires, flux B with the basicity of 1.8 microstructure are given in Tables 1 and produces the greatest decrease. This flux, 2. The compositions of the filler metals when used with S2 welding wire, gives a and fluxes used are presented in Tables 3 low acicular ferrite content even when and 4, respectively. used on a Ca-free base metal. The acicular ferrite level in each weld metal was quantified by point counting Wire Effect using the schemes proposed by Abson and Dolby (Ref. 11) and Pargeter (Ref. For the whole range of welds exam­ Fig. 1 — Typical microstructural features ob­ 12). The types of inclusions were exam­ ined, use of the no. 22 welding wire always gave improved acicular ferrite served in the weld metal: A — Ca-free base ined by means of a Philips 400 electron metal; B — Ca-treated metal microscope, fitted with EDAX and STEM levels with little change resulting from changes in the flux type. This behavior is facilities, using carbon extraction repli­ large for an interstitial atom, boron readi­ in agreement with previously published cas. ly segregates to atomically strained literature (Refs. 13, 14, 15) on Ti- and The replica technique was best suited regions; very small quantities (typically B-containing consumables. for this work. This is because mean inclu­ <50 ppm in the deposit) will reduce the sion size was less than 0.5 nm, and any Additions of boron and titanium are energy of these sites and hence inhibit other technique using bulk specimens known to promote acicular ferrite forma­ grain boundary ferrite nucleation. would have been severely affected by tion. It is presumed that boron carries out the role of minimizing grain boundary interaction with the matrix. The inclusions Flux Basicity were further line-scanned along their ferrite nucleation, while titanium protects diameters to determine their internal vari­ boron from oxygen and nitrogen due to Referring to Fig. 2, it can be seen that ation in chemical composition. This tech­ its strong deoxidizing potential. In addi­ there is no relationship between the AF nique was also employed to analyze the tion, titanium (as TiO) is proposed to level and the basicity index of the flux. extraction replicas taken from the base enhance acicular ferrite nucleation (Refs. Although the best results are those given metals. Diffraction work was also under­ 16, 17). by fully basic flux A, the worst ones are taken to study any crystallographic The beneficial effect of boron results not given by the acid flux but by the nature of the inclusions. from its small atomic size. It is present as semi-basic flux B. These results are con­ an interstitial solute and is, therefore, trary to those reported in the literature relatively mobile. Since it is comparatively (Refs. 8, 16-17, 18); however, they would

Results and Discussion Metallography Table 4-—Results of Flux Analyses The typical microstructural features (Fig. 1) observed in these welds were: Ra<;iritv Composition, wt-% 3 1. Acicular ferrite (AF). Flux index' ' Si02 + Ti02 CaO + MgO Al203 + MnO CaF2 2. Grain boundary ferrite (CF). 3. Polygonal ferrite (PF). A 3.1 15 35 20 25 4. Ferrite with aligned , aus­ B 1.8 22 30 25 21 C 0.60 27 1 51 14 tenite or carbide (MAC). The percentage of acicular ferrite for - MgO + BaO + Na 0 + K 0 + ViMnO B, = Ca°^ 2 2 the welds studied are shown in Table 5. Si02 + Vi (Al203 + Ti02 + Zr02)

Table 5—Percentages of Acicular Ferrite (AF) in Weld Microstructures

FlNvnn»nd No. 22 welding wire SD3Mo welding , wire S2 welding wire basicity Ca-free Ca-treated Ca-free Ca-treated Ca-free Ca-treated index (Bl) base metal base metal base metal base metal base metal base metal

1 (Bl = 3.1) A-85% AF B-76% AF C-69% AF H-64% AF M-70% AF N-68% AF 2 (Bl = 1.8) C-70% AF D-65% AF 1-61% AF 1-17% AF 0-35% AF P-16% AF 3 (Bl = 0.6) E-78% AF F-65% AF K-73% AF L-61% AF

WELDING RESEARCH SUPPLEMENT 1225-s seem to support the findings of Terashi­ ma and Hart (Ref. 19) where high levels of acicular ferrite were generated in a higher oxygen content. It is well known that there is a direct so correlation between basicity index of the flux and weld metal oxygen content. This relationship holds for the present series of welds examined as can be seen in Fig. 3; here the levels of oxygen content in the weld metal were taken from the weld chemical analyses — Table 6. However, no direct relationship was found between the oxygen content of the weld metal and the amount of acicular ferrite present in the welds —Fig. 4. This result is contrary to that referred to in the litera­ ture (Refs. 14, 20, 21) where it has been suggested that the range 0.025-0.06 wt-% oxygen provides the optimum dis­ persion of inclusions for the nucleation of acicular ferrite —for example, at 0.044 wt-% oxygen in the welds, the levels of 40 acicular ferrite ranged from 16 to 70%.

Electron Microscopy and 30 X-Ray Microanalysis Electron microscope study of the car­ bon replicas showed that the inclusions

20 are generally globular in shape. The glob­ ular inclusions were, however, found to be supplemented by distinctly geometri­ cal inclusions in the beads made from the high basicity flux A; this is consistent with the work of Cochrane and Keville (Ref. 7) —Fig. 5. The geometrically shaped _)_ inclusions could be triangular, rectangu­ 3 I lar, pentagonal and hexagonal. Flux C Flux B Flux A Figures 6A, 7A, and 8A show the Basicity index results of the energy dispersive spectros­ Fig. 2 —Percentage of acicular ferrite in the weld metal microstructure vs. flux basicity index. copy on the typical inclusions of the weld Independent variables were welding wire and base metal composition metal. The spectra are very similar in terms of the elements present; however, there were differences in the actual con­ centration of Al, Si and Mn. In addition to the random spot analysis, typical inclusions were line-scanned. This revealed that a change in inclusion com-

.09

.08

.07

.06

.05 • • .04

.03 )-

.02 Ca free 0 .01 Ca treated %

Flux basicity J I L _l I 1_ Fig. 3 — Relationship between flux basicity and oxygen content in the weld metal. The oxygen values of 0.07 and 0.08 wt-% correspond to a flux with a basicity of 0.63 Fig. 4 — Percentage of acicular ferrite vs. weld metal O2 content

226-s | JULY 1984 Table 6—Weld Metal Composition, % I- z LU Weld Total S no. C Mn Cr Mo Ni V Ti Cu Sn Nb B Al N o a. 2 o (M) .04 .24 1.19 .014 .027 <.005 .032 .036 .005 .25 .007 .005 .029 <.001 .032 .0048 .022 -I UJ (N) .05 .17 1.23 .016 .027 <.005 .029 .036 <.005 .25 .006 <.005 .022 <.001 .018 .007 .03 > (O) .04 .25 1.24 .014 .028 <.005 .032 .035 .005 .25 .007 .005 .025 <001 .025 .0052 .044 LU (P) .04 .20 1.29 .017 .029 <.005 .031 .037 <.005 .26 .007 .005 .022 <.001 .014 .0068 .044 a (A) .04 .19 1.21 .014 .045 .089 .043 .036 .006 .23 .006 .005 .028 .001 .032 .0045 .027 ^. (B) .05 .13 1.26 .018 .044 .083 .041 .038 .006 .23 .005 .005 .023 .0012 .020 .0068 .028 x .03 .22 1.26 .015 .044 .086 .042 .036 .006 .23 .006 .005 .027 .0012 .028 .0054 .044 o (Q IE (D) .04 .16 1.30 .018 .045 .089 .041 .038 .006 .24 .006 .005 .021 .001 .018 .0075 .044 < .028 (C) .05 .20 1.29 .014 .024 .18 .038 .034 <.005 .26 .007 .007 .027 <.001 .0048 .021 LU .17 .035 .037 .006 .025 •C001 .016 (H) .05 .14 1.36 .017 .027 .005 .28 .006 .007 .025 to .23 1.37 .014 .024 .15 .036 .038 .006 .030 <.001 .026 (I) .04 <.005 .28 .009 .0047 .034 .17 1.43 .017 .024 .14 .033 .039 .28 .006 .027 <.001 .014 (I) .04 .005 .006 .0065 .035 .23 1.25 .014 .023 .17 .038 .037 .26 .008 .007 .029 <.001 .036 (K) .06 .013 .0054 .049 .18 1.31 .017 .025 .16 .035 .039 .28 .007 .006 .027 -C001 .024 (L) .06 .010 .0068 .040 .05 .22 1.21 .015 .043 .09 .044 .037 .23 .007 .006 .030 .0012 .034 a. (E) .018 .0058 .051 o .07 .16 1.23 .018 .045 .098 .043 .039 .2i .007 .005 .026 .0012 .029 (F) .020 .0075 .051

X position does occur from one flux-weld­ ~-*Zru ' -% s O ing wire combination to another. The IE < three predominant types of inclusions LU found were: Al-rich, Mn-rich and a type l/> LU categorized as "others"; these latter LE were generally complex inclusions. ^ The line-scans corresponding to the y Z .'if ui three types are shown in Figs. 6B, 7B and .-."•s* £ 8B. a. Figure 9A shows a typical example of o X7K the spectra taken from geometrically shaped inclusions. As can be seen, the geometrically shaped inclusions are very rich in Al. This is also seen in the line-scan of Fig. 9B. Electron diffraction studies of o these geometrically shaped particles CE show that they are crystalline <5-AI 03 • \ < 2 V UJ (F.c.c. a0 = 7.859 A). af B Vi ^3 UJ Table 7 shows the results found for the CE --^ welds deposited in Ca-free base metal. I- The percentage of acicular ferrite present srr Vf Z in the microstructure is listed together UJ with the percentage of Al-rich, Mn-rich S a. and "others" types of inclusions. The O range of Mn/A[ ratios and the average Fig. 5-Carbon extraction replica micrograph showing globular and geometrically shaped inclu­ _i UJ Mn/AI ratio, Mn/AI, were calculated sions > UJ a x o Al CE < UJ tf) UJ CE

UJ S a. ® O _i UJ >

6 .18 20 37 _?a .26 SS .30 ,3? ,3a .35 .38 .ao .*2 DISTANCE, )SB X o cr 4 6 < [NERGY, KeV UJ tf) Fig. 6 — Spectography and line-scan results for Al-rich inclusions: A • UJ typical x-ray spectra; B (right) — STEM, EDAX line scan CE

WELDING RESEARCH SUPPLEMENT 1227-s DISTANCE, Jim

Fig. 7 — Spectography and line-scan results for Mn-rich inclusions: A (left) — typical x-ray spectra; B (above)—STEM, EDAX line scan

from the line-scans. Table 7 also shows ratios. In these inclusions it can be seen Microprobe analysis showed that there the range of inclusion sizes. (Table 7 from Fig. 11 that low AF levels are associ­ were no noticeable changes in Mn levels X includes a similar tabulation of the results ated with a high Mn/AI ratio. It was in the matrix in relation to variation in o for the welds deposited on Ca-treated observed for Mn/AI ratios below 0.24 percentage of Mn-rich inclusions. Conse­

Mn Al O Al

Q. ® o @ > LU Mn Q I Cu Ti Cu ^^ Ti X A. . y\ L /c_ 4 6 o J CE TNERGY, KeV < ENERGY , KeV UJ V) UJ IE /^ ^"x ® 0. O / Mn \ / ^ \ > LU Q X /^ — --^A o

228-s|JULY 1984 can also be seen that the highest levels of AF are those where the presence of Table 7—Results of Analyses for Acicular Ferrite and Inclusions Mn-rich inclusions is minimized. z Again, no systematic relationship was Mn UJ Al found between the levels of oxygen in £ range the welds and the type of inclusions A Inclusion oQ. present. For example, if four of the ana­ Al-rich Mn-rich Mn Sizes, Flux Welding _J a UJ lyzed welds with identical oxygen con­ Code %AF, inclusions °o inclusions % Others, % Al rim (Bl)< > wire tents are examined and their percentage > Ca-free base metal: of acicular ferrite and the types of inclu­ OUJ .22 Q sions present are recorded, the results .33 CE 32 A < (A) 85 100 0 0 to No. 22 appear as shown in Table 8. This rein­ A UJ (3.1) W) forces the idea that the chemistry of the .27 1.46 inclusions rather than their volume frac­ .22 tion is of prime importance in the devel­ 33 ,32 A opment of acicular ferrite. (M) 70 100 0 0 to S2 A (3.1) UJ .28 .85 Analyses of the inclusions present in £ the base metals were also performed. .18 .53 Results revealed that the Ca-free base .34 A Q. (C) 69 62 5 33 to SD3Mo metal had predominantly Mn inclusions A (3.1) o .23 1.45 with an average size of 0.42 ^m, accom­ _l LU panied by some Si-Cu (average size = .14 > 0.30 jim), Nb-S (average size = 0.30 ^m) 81 .25 B LU (C) 70 100 0 0 to No. 22 and S-Cu (average size = 0.15 ^m) parti­ A (1.8) c .44 1.64 cles. Ca-treated base metal showed a X wider variety of particles, although no .44 30 Mn particles were detected. The pres­ .75 C o (K) 67 45 38 17 to SD3M0 ence of large round Al-Ca particles A (0.6) LE .55 1.10 < approximately 75 /um in diameter was UJ most noticeable. Other particles were .53 tf) 25 UJ similar to those found in the Ca-free base 76 B (I) 61 30 40 30 to SD3Mo CE metal except for the presence of Si-S A (1.8) Q. .65 1.55 clusters. o .83 Although the differences in chemistry 24 96 B of the inclusions found in the two base (O) 35 25 75 0 to S2 metals are significant, they do not pro­ A (1.8) .91 1.84 vide any straightforward explanation of x Ca-treated base metal: o the differences in the corresponding (E .55 weld metal inclusions. 30 < .81 A UJ (H) 64 58 16 26 to SD3Mo V) A (3.1) UJ .67 1.35 CE Conclusion .12 .20 The results do not give an insight into .34 B (D) 65 100 0 0 to No. 22 the basic mechanisms in terms of base A (1.8) 2.96 metal chemistry, flux and welding wire .23 £ combinations that have contributed to .31 Q. .30 o weld metal microstructure variations. .66 C —I (L) 61 57 5 38 to SD3Mo UJ However, the observed effects are a A (0.6) .85 foundation for future work, and it is .44 > UJ hoped that investigation will be extended 1.72 .30 Q to examine these factors. In the mean­ 3.03 B 100 0 to SD3Mo X (I) 17 0 A (1.8) while the following findings are to be 1.20 noted: 2.50 o 1.51 CE 1. For similar welding conditions the .23 3 03 B < level of acicular ferrite in the weld bead is (P) 16 0 100 0 to S2 UJ lower in Ca-treated than in Ca-free base A (1.8) tf) 2.13 .66 UJ metal of similar chemical composition. UCJE 2. The use of the microalloyed no. 22 (a) Bl = basicity index. £ welding wire generated high levels of a. o acicular ferrite on both calcium-treated —I and calcium-free base metal. >UJ 3. Variations in flux basicity do not produce the expected changes in weld bead microstructure. This is because acid X and fully basic fluxes produce higher o CE levels of acicular ferrite than that pro­ < duced when a semi-basic flux is used. This UJ is true for both base metal compositions tf) LU and all three welding wires used. CE

WELDING RESEARCH SUPPLEMENT 1229-s • Ca Free Plate Table 8—Variations in Acicular Ferrite and Inclusion Types at 0.44 Wt-% Constant • Ca Treated Plate Oxygen Content

** too - Weld no. AF, % Inclusion type

C 70 100% Al-rich D 65 100% Al-rich o O 35 25% Al-rich LL + 75% Mn-rich 50 - P 16 100% Mn-rich

submerged arc weld metal inclusions. The Welding Institute members report 151/1981. 7. Cochrane, R. C, and Keville, B. R. 1981. I Paper 5 in Proceedings of conference on steel 50 100 for line pipe and pipeline fittings, Crosvenor House, London. 8. Farrar, R. A., and Watson, M. N. 1979 Al rich inclusions % (lune). Effects of oxygen and manganese on submerged arc weld metal microstructures. Fig. 10 —AF content vs. Al-rich inclusions Metal Construction: 285-286. 9. Harrison, P. L, and Farrar, R. A. 1981. Influence of oxygen rich inclusion on the 5 to a Ca Free Plate phase transformation in HSLA steel weld met­ als. Journal of Material Science: 2218-2226. Ca Treated Plate 10. Tuliani, S. S., Boniszewski, T., and Eaton, N. F. 1979 (August). Notch toughness of com­ 100- mercial submerged arc weld metal. Welding and : 327-329. 11. Abson, D. ]., and Dolby, R. E. 1980 (April). A scheme for the quantitative descrip­ tion of ferritic weld metal microstructure. The m We/ding Institute Research Bulletin: 100-103. 50- 12. Pargeter, R. ). 1980. Quantification of ferritic weld metal microstructures — results of an interim exercise. IIW doc. IXJ-37-80. 13. Barrite, C. S„ et al. 1981. Quantitative microanalysis with high spatial resolution, pp. 112-118. London: The Metals Society. ~> I n—I—| 1 1 r —i i i i i ; 14. Still,). R., and Rogerson, ]. H. 1978 ()uly). The effect of Ti and B additions to multipass 0.5 1.0 1.5 2.0 2.5 submerged arc welds in 500 plate. Metal Construction: 339-342. 15. Homma, H., et al. 1978. Effects of Ti and Mn Nb additions on the mechanical properties of submerged arc weld metals. IIW doc. IX 1072- Al 78. Fig. 11 —AF content vs. average MrVAl ratios 16. Mori, N. et al. 1981. Mechanism of notch toughness improvements in Ti-B bearing weld metals. IIW doc. IX 1196.81. 17. Mori, N. et al. 1982. Characteristics of 4. From inclusion analyses it appears References mechanical properties of Ti-B bearing weld that the formation of acicular ferrite is 1. Taira et al. 1982. Resistance of pipeline metals. IIW doc. H-980-82, IX-1229-82. favored by Al-rich rather than Mn-rich steels to wet sour gas: 173-180. Metals Park, 18. Koukabi, A. H., et al. 1979 (December). inclusions. The chemical composition and Ohio: American Society for Metals. Properties of submerged arc deposits — effects structure of the inclusions seem to be the 2. Dolby, R. E. 1976. Factors controlling of zirconium, vanadium and titanium/boron. primary factors controlling their efficiency weld toughness — the present position, pt. II- Metal Construction: 639-642. as nudeants of acicular ferrite forma­ weld metals. The Welding Institute members 19. Terashima, H., and Hart, P. H. M. 1983. tion. report 14/1976/M(1976). Effects of aluminium on C-Mn steel submerged 3. Kirkwood, P. R. 1978 (May). Microstruc­ arc weld metal properties, Part II. Presented at tural and toughness control in low carbon the Welding institute conference on the weld metals. Metal Construction: 260-264. effects of residual, impurity and micro-alloyed 4. Abson, D. |. et al. 1978. Investigation into elements on and mechanical prop­ A ckno wledgments the role of nonmetallic inclusions on ferrite erties, November, 1983. The authors would like to thank Shef­ nucleation in carbon steel weld metals. The 20. Watanbe, I. et al. 1980. Applicability of field University, Sheffield, U.K. and Oerli­ Welding Institute members report 67/1978/ large current MIG arc welding technique. Pre­ M(1978). kon Welding Limited, Hayes, Middlesex, sented at conference on welding research in 5. Garland, |. C, and Kirkwood, P. R. 1976 the 1980's, Osaka, Japan, 1980 (43). U.K. for permission to publish this paper. (April). A reappraisal of the relationship 21. Cochrane, R. C, and Kirkwood, P. R. It should be noted that the opinions between flux basicity and mechanical proper­ 1975. The effect of oxygen on weld metal expressed in this paper are those of the ties in . Welding and microstructure. The Welding Institute confer­ authors and are not necessarily those of Metal Fabrication: 44-49. ence on trends in steels and consumables, the organizations they represent. 6. Pargeter, R. ). 1981. Investigation into London.

230-s | JULY 1984