Acicular ferrite in high oxygen weld metals

TWI Bulletin, May 1987

Norman Bailey, BMet, CEng, FIM, FWeldI, is a Principal Research Metallurgist in the Materials Department.

A study of weld metals containing more than 0.045% oxygen has shown that acicular ferrite microstructures of potentially good toughness can be developed by addition of a small amount of titanium to the weld metal, provided some aluminium is also present.

The toughness of weld metals is clearly dependent on their microstructure and for many ferritic steel weld metals it is advantageous to achieve a fine acicular ferrite microstructure. Although the practical requirements for this are by now reasonably well known, virtually all the research has concentrated on low oxygen weld deposits, i.e. below about 0.045% oxygen. The present article briefly reviews the situation regarding such weld metals, and then considers the conditions under which acicular ferrite formation can be obtained at higher oxygen levels.

Low oxygen weld metals

The development of intergranular acicular ferrite in weld metal is dependent on achieving a suitable balance between oxygen content and the presence of deoxidant elements, especially Ti and Al but also Mn and Si. When the weld oxygen content is low, it is possible for sufficient aluminium to enter the weld metal via dilution with parent metal for the formation of acicular ferrite to be prevented so that the toughness falls dramatically. ^[1-5] It is generally accepted that the formation of acicular ferrite is promoted by some sort of nucleant - probably an oxide of titanium assuming (as is the case in this article) that the degree of alloying by Mn and other elements is sufficient to reduce the amount of grain boundary ferrite below about 50%, but is not so high as to give a bainitic or martensitic microstructure. If this is the case, it is reasonable to suppose on thermodynamic grounds that aluminium ^[4] is oxidised preferentially to titanium, so that an excess of aluminium would react with all the oxygen, leaving none to form titanium oxide; this view is borne out by a rough rule of thumb that the weld Al content should not exceed 9/8 the oxygen content ( i.e. the ratio of Al and O in Al ₂O ₃) by more than a few thousandths of one per cent. ^[6] Some aluminium, however, appears to be helpful to acicular ferrite formation and optimum contents in the parent metal (increasing from 0.012%Al at 0.020% oxygen in the weld to nearly 0.1%Al in the plate with 0.065% oxygen) have been proposed. ^[8]

Evidence for the importance of titanium in these low oxygen weld metals is obtained from microanalysis of the somewhat complex non-metallic inclusions in the weld. With an excess of aluminium present, a large proportion of inclusions is found to contain only aluminium (light elements such as oxygen being undetectable by the techniques used), whereas if oxygen is not in excess most inclusions are more complex and frequently contain titanium, albeit as a minor constituent. ^[3,5,7-9]

Evidence for the importance of titanium from the bulk analysis of low oxygen weld metals is, however, scarce. Such weld metals often contain a few thousandths of one per cent, but acicular ferrite microstructures have been documented in submerged-arc weld metal where the weld titanium content is below the analytical detection limit, typically 0.002%, ^[1,2,10] although in the last instance a small amount (0.03%) of vanadium in the weld metal ^[10] may have assisted acicular ferrite formation.

In a series of carefully controlled tests on MIG welds, ^[11] deposits containing <0.002%Ti (and probably appreciably less than 0.001%) failed to show acicular ferrite microstructures in weld metal containing 0.004-0.005%Al and 0.02-0.05% oxygen. The presence of ~0.02%Ti at similar oxygen and aluminium levels gave predominantly acicular ferrite microstructures, although an excess of aluminium, as with submerged-arc weld metals, prevented acicular ferrite formation.

High oxygen weld metals

In so called normal weld metals, i.e. without deliberate additions of aluminium, acicular ferrite microstructures fail to appear with oxygen contents above ~0.045%. This has been observed ^[12] in manual metal arc (MMA) welding at about 0.045%O in welds made with cellulosic electrodes and containing appreciable quantities of titanium ( i.e. 0.01-0.03%) but <0.005% Al; although in these welds the acicular ferrite was coarse because of their low Mn content. In submerged-arc welds a similar change was seen at about 0.040%O. ^[13] In this investigation, however, it appeared that if ~0.02%Al was present in the weld metal ( i.e. from the use of alumina fluxes), microstructures containing acicular ferrite were found with oxygen contents as high as 0.055%. In these submerged arc welds the analytical limit for detecting titanium (0.01%) was too large to know whether significant amounts were present or not. ^[13] However, some later tests ^[14] with calcium silicate flux welds, in which small amounts of titanium (0.003%) were present together with 0.02-0.03%Al diluted from the parent plate, showed essentially acicular ferrite microstructures with oxygen contents between 0.05-0.06% and with arc energies just exceeding 7 kJ/mm.

Other work on cellulosic MMA welds ^[15] involved deposits made with two different electrodes, but only one of these (giving weld oxygen contents of 0.047-0.050%) give microstructures containing coarse acicular ferrite. The other (giving 0.045-0.077% oxygen) contained FS(A) (ferrite with second phase (aligned)). Both weld metals contained some titanium (0.008-0.02%) but little (

The previously described MIG investigation ^[11] also included results on high oxygen welds. In a critical series of tests on weld metals containing 0.05-0.06% oxygen and 0.007-0.048%Al, predominantly acicular ferrite microstructures were obtained with titanium contents down to an estimated 0.0013 ( i.e. below the 'normal' detection limit of 0.002%). Acicular ferrite was not, however, developed in a similar weld containing appreciably below 0.0005%Ti.

In other work on submerged-arc weld metals, it was found that an acicular ferrite microstructure could be developed in bead-on-plate welds at 3 kJ/mm using a high oxygen manganese silicate flux doped with ferro-titanium, but which gave a weld containing as much as 0.12% oxygen. ^[16] Further tests on 30mm thickness plate with two pass welds at 5.6 kJ/mm confirmed this effect ( Fig.1) and showed a substantial improvement in Charpy toughness, the 35J transition temperature being reduced by over 50°C compared with a weld made with flux to which no ferro-titanium had been added. ^[17] Other tests in this series were made with a lower heat input of 3 kJ/mm. In the first series the flux was doped with rutile (TiO ₂) and a partial conversion to an acicular ferrite structure was obtained ( Fig.2). ^[17] In multipass welds, the fill passes with flux doped with ferro-titanium again only achieved a partial change and the toughness was unaltered( Fig.3 ). ^[17]

The results were complicated because of unintentional differences in oxygen, titanium and aluminium contents between some of the welds. In all the welds, some aluminium had been picked up from aluminium treated parent plate, and was only absent (i.e. <0.003%) from the low dilution runs of the multipass welds. ^[17] A further test was therefore carried out using a parent plate essentially free from aluminium to see whether aluminium was necessary for acicular ferrite formation.

The available silicon-killed steel was a lean alloyed steel to BS 1501:271 whose composition is given in Table 1. The aluminium content was <0.003% and the only alloying element known t0 influence acicular ferrite formation (other than by the degree of alloying) was 0.10% vanadium - an element known to promote acicular ferrite to a small extent. ^[18] The S1 (0.5%Mn) wire used was also free from aluminium ( Table 1); this and the flux (manganese silicate flux Lincoln Arcmaker No. 1) were the same as in some of the previous work. ^[16,17]

Table 1 Analyses of plate, wire and weld metal

The flux was ground to a fine powder and mixed dry with a 4% addition of ferro-titanium and agglomerated after mixing in a small quantity of mixed sodium and potassium silicate solution as a binder. The flux was dried for 3 hours at 250°C before use.

A single bead was deposited in a groove 13.5mm deep in the 25mm thickness plate with an included angle of 45°. Welding conditions of 900A, 35V DC +, 325 mm/min welding speed and 35mm electrode extension gave an arc energy of 5.6 kJ/mm. The analysis of the weld, also included in Table 1, showed 0.003%Ti with only 0.004%Al and an oxygen content of 0.067%.

The microstructure ( Fig.4) consisted of a little grain boundary ferrite with mainly fine acicular ferrite in the intragranular regions. The acicular ferrite tended to be of high aspect ratio, however, and some ferrite with aligned second phase was present at the edges of the grain boundary ferrite. The intragranular microstructure, although fine, was intermediate between the acicular ferrite shown in Fig.1b and that in Fig.2b.

Discussion

To clarify the relationships between aluminium and titanium content and microstructure of high oxygen weld metals, Table 2 has been prepared for submerged-arc, MIG and MMA results available from recent Welding Institute data. [11,12,14,16,17] One reason for this restriction is that the spectrographic analysis for small amounts of elements requires careful standardisation and understanding of inter-element effects. Although it is not claimed that The Welding Institute's analyses are necessarily better than those of anybody else, they have been consistent over a number of years and thus enable a clear picture to be obtained.

Considering the results in Table 2 it can be seen that the weld metals with coarse ferrite with aligned second phase structures ( Fig.1a and 3a) contain less than 0.005%Al and/or less than 0.002%Ti, except in the case of weld 161C, which contains an estimated 0.0013%Ti with 0.007%Al. Even as much as 0.024%Ti is insufficient without the presence of Al (weld 7702-A12). Mixed microstructures ( Fig.2, 3b) were obtained from weld metals containing 0.003-0.004%Ti with no more than 0.005%Al, although one weld (7702W-101) did contain as much as 0.030%Al with a microstructure described as 'mainly acicular ferrite'. On the other hand, weld W108 with 0.003% each of Ti and Al gave a microstructure mainly of acicular ferrite. To achieve an essentially acicular ferrite microstructure at least 0.003%Ti with at least 0.006%Al should preferably be present in the weld metal.

Table 2 Summary of composition and microstructure of high oxygen (>0.045%) weld metals

Weld 9425-W100 was unusual in several respects. The high degree of alloying from the parent plate gave an usually fine intragranular microstructure, the oxygen content of 0.067% is unusually low for the manganese silicate flux used and the presence of 0.07% vanadium in the weld metal would be expected to favour nucleation of acicular ferrite. All these factors appear to have promoted acicular ferrite in a weld having a marginally low aluminium content of 0.004%.

The results generally support the view that a small amount of aluminium (at least 0.006%) is required to enable titanium oxide to nucleate acicular ferrite in weld metals of high oxygen content ( i.e. above 0.045%), but that a degree of alloying, including vanadium, may reduce the amount necessary if oxygen content is not too high (say 0.07% or less).

Summary

Tests on high and low dilution weld metals of high oxygen content, i.e. above 0.045%, have shown that tough acicular ferrite microstructures can be developed by small additions of titanium and aluminium, provided the degree of alloying with Mn and other elements is such as to avoid high proportions of grain boundary ferrite, bainite and martensite. Such acicular ferrite formation appeared to be easier with some vanadium present.

Acknowledgements

Thanks are due to colleagues, particularly Mrs W Martin, Miss S Stevens, Mrs V M Shaw and K M Filby for their assistance.

References