Local brittle zones in C-Mn steel multipass welds
TWI Bulletin, September/October 1990
Adrienne Barnes joined TWI in August 1988 after obtaining a degree in metallurgy and materials science from Imperial College, London.
She is working in one of the ferrous sections within the Materials Department, principally on a number of failure investigations and projects relating to the toughness of heat affected zones.
Adrienne is also studying for an external PhD with Imperial College, looking in detail at the intercritically reheated portion of the HAZ - specifically the role played by a number of variables, including composition, on the development of microphases within this region - and their ultimate effect on the toughness and fracture mechanism.
Within the HAZ of multipass welds made in C-Mn microalloyed steels, regions of impaired toughness occur, commonly known as local brittle zones (LBZS). Many research projects, at TWI and elsewhere, are in hand to gain a better understanding of factors controlling their formation. Adrienne Barnes outlines the effects of various elements on HAZ microstructure and LBZ development.
The HAZ in a multipass weld, see Figure 1, is considerably more complex than that in a single-pass weld. The underlying material is continually reheated by subsequent passes and a wide range of microstructures results, depending on the distance from the fusion boundary, the peak temperatures reached, and the subsequent cooling rates during the multiple heating.
Fig.1. The complexity of the HAZ in a multipass weld: GC - grain coarsened, GR - grain refined, IC - intercritical, SC - subcritical.
Following this microstructural change, areas with limited cleavage resistance can develop, generally termed local brittle zones (LBZS). Such sites of preferential cleavage initiation can form at various positions in a multipass HAZ, depending on the specific microstructure formed.
Historically, lowest toughness was expected in the coarse-grained as-welded HAZ associated with the last pass of a weld, where coarse and/or hardened structures were most likely. In recent years, and with more modern steels of lower carbon equivalent and associated hardenability, it has become evident that minimum toughness may arise in reheated grain-coarsened HAZ regions, [1,2] whether intercritical and leading to partial re-austenitisation and grain-boundary decoration, or subcritical, or even occasionally beyond the visible HAZ. [3]
The interest in these regions has increased because of the higher demands on materials performance, particularly in the offshore sector as structures move into regions of deeper water and lower air temperature. Many new specifications now stipulate stringent testing of the HAZ, and demand performance close to that of the parent material.
LBZS are difficult to test because they are small. It is known that CTOD (crack tip opening displacement) values, used to assess a material's resistance to fracture initiation, are influenced by a number of variables such as weld type, specimen geometry and notch profile. However, testing the HAZ becomes further complicated by the inhomogeneity of the microstructure [4] and the considerable difficulty of locating the highly-constrained fatigue crack tip exactly at one of these LBZS during the toughness test. There may be wide scatter within the test data, making correlation difficult. Thermal simulation techniques somewhat reduce the inherent variability and allow close matching of the thermal conditions and microstructure at various locations in a multipass HAZ; this in turn allows easier and more accurate testing and examination.
The mechanical properties of both the parent material and the HAZ are strongly influenced by chemical composition, and fracture toughness is no exception. A number of researchers [5,6] have demonstrated a correlation between toughness levels within the HAZ and the volume fraction of martensite-austenite (M-A) microphases present.
The transformation to M-A is dependent on the hardenability of the carbon-enriched austenite formed either during the final stages of the austenite-ferrite reaction or as a result of partial re-austenitisation on intercritical reheating. The formation of M-A - or indeed of pearlite or other ferrite/carbide structures - thus depends on the composition of this region and therefore on the bulk composition and possibly the prior austenite grain size.
The sections following discuss briefly the role of some of the principal alloying elements used in modern steels, in terms of both the original transformed HAZ structure and the formation of LBZS, especially because of M-A constituent. Note that the effects are often interactive, the presence of one element influencing the effect of another.
Alloying element effects
Carbon and carbon equivalent
Fig.2. Effect of C equivalent on HAZ CTOD toughness (after Terasaki et al [8] )
There has, over the last decade or so, been a general trend within the industry to reduce the C content and carbon equivalent (CE) of steels. Although this has had an overall beneficial effect on HAZ toughness, too low a C level or CE may promote coarse sideplate structures [7] with an associated deterioration in toughness [8] as shown in Figure 2.
A decrease in C content or CE induces a decrease in hardenability, and may reduce the tendency for M-A formation with an increased tendency to autotempering because of the increase in M S, the martensite start temperature. It is reported [9] that CTOD levels increase for C levels up to 0.06% but decrease when this level is exceeded, although a value of 0.12wt% is proposed by Amano et al. [10] At higher C levels M-A formation is promoted and the overall microstructure is refined. [9] Since these have opposing effects on toughness, there must be an optimum C level.
Generally, HAZ toughness decreases with increased heat input because of a change in transformation product - from, for example, tough autotempered low-C martensite to more brittle upper bainite/Widmanstätten ferrite - and an increase in grain size. However, the low-CE steels will not generally produce the autotempered martensite transformation product, other than at very low heat input. Their toughness at low heat input can therefore be lower than that of higher CE steels, but is not necessarily unsatisfactory. Their performance is often better at higher heat input where less hard bainitic microstructures can develop. [11]
Manganese
Since the 1970s, there has been a tendency to increase the level of Mn used, to counter the effect of lower C content on base-plate strength. An increase in Mn, principally through solid solution hardening, but with an effect of refinement, leads to an increase in the yield strength, impact transition temperature and hardness of the steel.
[12] The extent to which Mn can be added, however, is limited, since it also affects the austenite hardenability and the risk of forming martensite. Work carried out at TWI
[13] has indicated that Mn may have a strong influence on the development of microphases within the intercritically reheated coarse-grained HAZ. The increased level of Mn raises the hardenability of the austenite and promotes formation of M-A constituent as opposed to pearlitic microphases, with an associated adverse effect on toughness.
A further restriction is imposed on the Mn content, particularly in normalised and hot-rolled plates, to minimise segregation and microstructural banding, which results in anisotropy of properties [7] and can give rise to varied microstructure and hence toughness in the HAZ on welding.
Silicon
Fig.3. Effect of Si on the toughness of: a) The intercritically reheated grain-coarsened HAZ (ICGCHAZ) b) The ICGCHAZ heat treated at 450°C (after Amano et al [10] )
There is no clear trend in Si content, but there is some evidence that levels have increased (although this is not universal) to maintain adequate plate strength and hardness by virtue of solid solution strengthening.
However, Si is ineffective as a grain-refining element and consequently the increased strength and hardness are generally accompanied by a reduction in toughness. [12] It is reported [14] that plate strength and toughness are not affected by Si levels less than 0.35%.
Si does appear to have a detrimental effect on HAZ toughness, particularly in the intercritically reheated grain-coarsened region, because of the resulting increase in the volume fraction of M-A constituent. [2,9] The Si promotes partial transformation of austenite to proeutectoid ferrite and the remaining austenite becomes enriched with C. This enhanced hardenability encourages transformation to M-A constituent. [15] For Si levels below ~0.1-0.2% the M-A constituent readily decomposes to cementite by subsequent heat treatment [10,14] and, as the simulation data presented in Figure 3 show, the adverse effect on toughness is therefore reduced. Such decomposition could occur during a post-weld heat treatment (PWHT) operation at, say, 600°C, or possibly during multipass welding.
Niobium
Nb has an effect on all regions of the HAZ [6,17] although its effect is strongly dependent on heat input. At medium to high heat input, and quite apart from any precipitation hardening effect via Nb(C,N), Nb has a detrimental influence on the microstructure and hence the toughness - of the grain-coarsened HAZ. [17,18] The Nb reduces the amount of grain-boundary ferrite and promotes formation of a coarse structure of ferrite with aligned M-A-C (martensite-austenite-carbide) resulting in increased hardness. The increase in hardness and the associated increase in M-A constituent are believed to be responsible for the monotonic decrease in toughness with increased Nb content. [5,14]
There appears, however, to be a degree of confusion in the literature as to what level of Nb can be tolerated within a material, as this depends strongly on the C content of the parent material. Work carried out by Hulka and Heisterkamp [19] suggests that, for reduced-carbon grades (~0.09%), Nb contents of 0.18% can be tolerated. It also suggests that, even at high cooling rates, martensite formation is prevented because the transformation is shifted to shorter times, and despite the higher strength, resistance to brittle failure is even improved.
This level is considerably higher than generally reported, levels of ~0.03wt% are more widely accepted as the limit above which a deterioration in toughness is observed. As with many of the other microalloying elements, Nb is added to enhance the strength of the base material, and it is reported [7] that, for low-C steels, small additions of Nb can actually enhance the toughness at low heat input because of a reduction in the width of side plates and the size of carbides through pinning effects.
The toughness of Nb-containing steels can be improved by reducing the CE [20,21] which in turn reduces the tendency for M-A formation. PWHT leads to decomposition of the M-A constituent, but also enhances precipitation of Nb(C,N); hence, provided that the Nb content is low enough for the beneficial decomposition of M-A to exceed precipitation effects, PWHT will improve the fracture toughness.
Vanadium
As with Nb, V is added to the base plate as a grain refiner and precipitation strengthener, although its interaction with C and N to produce V(C,N) is weaker than that of Nb and thus the temperature of precipitation is lower. [7] The effect of V on the coarse-grained HAZ microstructure is quite different from that of Nb. V tends to induce a beneficial reduction in the colony size and promote intragranular nucleation of ferrite during the initial weld thermal cycle. But at the same time it has a detrimental effect on the subcritical and subcritically reheated grain-coarsened regions, through increased secondary hardening. [17] It is also reported [22] that V additions can reduce the toughness of the intercritical HAZ because of a change in the second-phase microstructure, promoting the formation of the potentially brittle M-A constituent.
The influence of V on toughness depends heavily on the base metal C level. The effect is generally less in low-C steels but, under normal welding conditions, moderate alloying with V would appear not to cause significant deterioration in HAZ toughness [23,24] The reason for this behaviour (good toughness despite high strength and hardness) is not clear, but Levine and Hill put forward a possible explanation from their work on weld metals [25] in terms of V affecting grain-boundary angles and hence increasing the resistance to crack initiation. It is reported [12] that the free nitrogen level in the matrix may also be reduced by small additions of V, and this may improve toughness by reducing solid solution hardening, even if some V-containing precipitates are present.
The level of V permissible before the toughness begins to decrease is dependent on the heat input used; for high welding energies HAZ embrittlement occurs for V contents over 0.lwt%, [26] whereas up to 0.25wt% can be present at low heat input with no detrimental effect on toughness ( Figure 4). It should be noted that these are data from simulated single-pass welds and the effect of multiple runs is not taken into account. The loss in ductility can be attributed to precipitation hardening by V(C,N) which is enhanced at high heat input. The V level at which toughness deterioration occurs may also be dependent on the C level, a lower C level being more tolerant to microalloying.
Fig.4. The effect of V on the toughness of the coarse-grained HAZ produced by single-cycle thermal simulation (after Hannerz and Jonsson-Holmquist [26] )
The effect of V can alter markedly in the presence of other alloy additions. For example, addition of Nb and V can yield low toughness in the coarse-grained HAZ after PWHT. [27,28]
Titanium
Attempts have been made to enhance the toughness of the grain-coarsened HAZ by various alloying additions to restrict the austenite grain growth in this region. The effectiveness of, for example, Al and Nb is limited to below 1100°C by the reduced pinning effect of Al, N and Nb(C,N) above this temperature. So Ti, which ties up free N to produce nitrides of high thermal stability, is widely used. [7] If the Ti is present in excess of stoichiometry, a significant reduction in toughness may occur, possibly because of coarsening of the TiN particles which encourage the formation of coarse ferrite side plates in intragranular regions and austenite grain growth. [30]
Fig.5. Effect of Ti on HAZ CTOD values for various steels (after Gray and Pontremoli [29] )
Use of high heat input during welding (> 10 kJ/mm) enhances the formation of upper bainite within the coarse-grained HAZ with an increased volume fraction of M-A constituent. [31] Microalloying with Ti (~0.015%), either alone or in combination with elements such as B, reduces the grain growth and the volume fraction of M-A constituent, thereby improving the toughness.
Figure 5 shows the detrimental effect of increased levels of Ti for various steel compositions. [29] This deterioration in toughness may be a result of the precipitation of Ti carbides.
Boron
B markedly increases steel hardenability and for this reason is incorporated into certain quenched and tempered grades to maintain a suitable level of plate strength. B can lead to embrittlement in the HAZ because of precipitation of borocarbides
[7] . However, with the introduction of low-C Steel grades, B is being more widely used, although high levels can encourage the formation of M-A constituent.
[5] To achieve an adequate HAZ microstructure, and hence toughness, there is an optimum B content, determined by austenite grain size. The B addition restricts austenite grain growth for heat input >10 kJ/mm and the fine precipitates encourage ferrite nucleation. This reduces the quantity of upper bainite formed and thereby improves toughness.
[31] It is reported by Grong and Akselsen [7] that the primary effect of B is to raise the energy barrier for ferrite nucleation, and that 11ppm is sufficient, even at high heat input, to suppress the formation of proeutectoid ferrite in the coarse-grained region. A rapid transformation occurs, however, in the grain-refined region as a result of the large grain boundary area. If the B level is further increased, the hardenability of these two regions can be balanced, but the risk of a high degree of precipitation causing embrittlement is increased [32] so the B level is generally limited to ~10-15ppm.
The effect of B microalloying, generally considered as detrimental, can vary significantly depending on the presence of other alloying additions; for example its effect on hardenability can be reduced if it is not shielded from the effects of N and O by treating with such elements as Al and Ti. It has also been suggested that, when added to steels treated with rare earth metals (REMS), intragranular nucleation of ferrite is promoted by the precipitation of nitrides of B or oxysulphides of REM which can give rise to improved toughness at high heat input. [33]
Nitrogen
N is especially damaging to the toughness of the HAZ, particularly the coarse-grained region, when in the 'free' or uncombined State. [3] The aim is therefore to maintain as low a level of N as possible within the coarse-grained HAZ. [14,17,34] An increase in N content significantly raises the steel hardenability, and stability of the austenite, and hence promotes the development of M-A constituent. [5,9] In Al-treated steels in which the N is 'combined' in the form of nitrides, the high temperatures within the coarse-grained HAZ lead to dissolution of these nitrides and 'free' N can be produced which is unlikely to recombine due to the subsequent high cooling rate. [17,35] This free N, if present in sufficient quantity, can interact with the strain field of dislocations at temperatures between 100 and 200°C and introduce a propensity for strain-ageing embrittlement. [4] At the same time, it is believed that there is an optimum, non-zero, N level, even in Ti- and Al-treated steels. This is because, at very low concentrations, there may be a decrease in austenite grain growth resistance in the coarse-grained HAZ, caused by a reduction in the number of critical size nitrides, resulting in decreased toughness.
Phosphorus
The detrimental effect of P on HAZ toughness is most pronounced in the mid-thickness regions of continuously cast steels following PWHT. [11,36] Toughness also deteriorates in the subcritical and intercritical HAZ and the subcritical coarse-grained HAZ following PWHT. In the latter case the fracture mode becomes intergranular, resulting from segregation of P to the prior austenite grain boundaries leading to grain boundary embrittlement. [1] The embrittlement is enhanced by high base material Si and Mn levels which promote the segregation of P. Grong and Akselsen [7] suggest, however, that the effect can be reduced by increasing the Mo content, which has a scavenging effect on the P.
Summary
The effect of some (but by no means all) alloying elements on HAZ toughness and the development of LBZS has been briefly discussed. The discussion has also indicated the strong influence of factors such as heat input and PWHT. The topic of HAZ toughness and LBZS is extensive and complex owing to the interactive effect of the many variables, and it is beyond the scope of this article to cover all aspects in detail.
Steels are continually being modified through varied alloying and microalloying additions and processing routes, in order to enhance HAZ toughness and to restrict - or, if possible, eliminate - the formation of the detrimental LBZS. Clearly, this review shows that there is a fair level of understanding of elemental effects, and this is being exploited in steel development. At the same time, the conditions under which LBZS form in various regions of a weld HAZ are still not fully understood and it is not possible to predict quantitatively the effects of individual elements, steel processing route or total weld thermal cycle. A Group Sponsored Project is in progress at TWI to clarify the role played by individual elements; further information can be obtained from the author.
References
| N° | Author | Title |
|
| 1 | Haze T and Aihara S: | IIW doc IX-1423-86. |
|
| 2 | Harrison P L and Hart P H M: | 2nd int conf on 'Trends in welding research', ASM International, Gatlinburg, Tennessee, May 1989. | Return to text |
| 3 | Dolby R E: | Welding Journal 1979 58(8) 225-238. |
|
| 4 | Pisarski H G and Pargeter R J: | Metal Construction 1984 16(7) 412-417. |
|
| 5 | Haze T and Aihara S: | 7th int conf on 'Offshore mechanics and arctic engineering', Houston, Texas, February 1988. |
|
| 6 | Fairchild D P: | TMS-AIME Annual meeting, Denver, Colorado, February 1987. |
|
| 7 | Grong O and Akselsen O M: | Metal Construction 1986 18(9) 557-562. |
|
| 8 | Terasaki F et al: | Ibid |
|
| 9 | Nakanishi M, Komizo Y and Fukada Y: | The Sumitomo Search 1986 (33, November) 22-34. |
|
| 10 | Amano K, Kudo J, Itakura N and Nakano Y: | 8th int conf on 'Mechanics and arctic engineering', The Hague, Netherlands, March 1989. |
|
| 11 | Hart P H M and Harrison P L: | To be presented at 'The metallurgy, welding and qualification of microalloyed (HSLA) steel weldments', Houston, Texas, November 1990 AWS/MA1. |
|
| 12 | Otterberg R, Sandstrom R and Sandberg A: | Metals Technology 1987 7(10) 397-408. |
|
| 13 | Barnes A M: | TWI Members' Report 402/1989. | Return to text |
| 14 | Nakano Y, Kudo J, Kobayashi E and Amano K: | 8th int conf on 'Offshore mechanics and arctic engineering', The Hague, Netherlands, March 1989. |
|
| 15 | Koo J Y and Bangaru N V: | Private communication. | Return to text |
| 16 | Kirkwood P R: | Proc AIME int conf Niobium 1981. |
|
| 17 | Hart P H M: | Proc one-day conf, Copenhagen, Denmark, September 1984. |
|
| 18 | Wang G R, North T H and Leewis K G: | Welding Journal 1990 69(1)14-22. |
|
| 19 | Hulka K and Heisterkamp F: | 'HSLA steels technology and applications', Proc int conf, Philadelphia, October 1983. | Return to text |
| 20 | Sakaguchi S, Hirai F and Nishiyama N: | Trans ISIJ 1987 27. | Return to text |
| 21 | Endo S, Tsukamoto H, Suga M, Matsumoto K and Ishikawa S: | Trans ISIJ 1986 27(l). | Return to text |
| 22 | Uchino K and Ohno Y: | Proc 6th int 'Offshore mechanics and arctic engineering symposium', Houston, Texas, March 1987. | Return to text |
| 23 | Levine E and Hill D C: | Metal Construction 1977 9(8) 346-353. | Return to text |
| 24 | Sawhill J M, Boussel P and Morrow J W: | Int conf 'Welding of HSLA structural steels', Rome, November 1976, published ASM 235-257. |
|
| 25 | Levine E and Hill D C: | Met Trans 1977 8A 1453. | Return to text |
| 26 | Hannerz N E and Jonsson Holmquist B M: | Metal Science 1974 8(7) 228-234. |
|
| 27 | Rothwell A B et al: | Canmet report 79-6. | Return to text |
| 28 | Signes E G and Baker J G: | Welding Journal 1979 58 179s-187s. |
|
| 29 | Gray J M and Pontremoli M: | 'Pipe Technology', Rome, November 19 1987. |
|
| 30 | Akselsen O M: | SINTEF report STF34 A84065 1984. | Return to text |
| 31 | Kosazu I: | Proc 'HSLA steels, technology and application', ASM, Philadelphia, October 1983. |
|
| 32 | Akselsen O M: | SINTEF report STF34 A84067 1986. | Return to text |
| 33 | Funakoshi T et al: | Trans ISIJ 1977 17 419-427. | Return to text |
| 34 | Thaulow C, Paauw A J and Guttormsen K: | Welding Research Supplement 266-279. |
|
| 35 | Suzuki M, Tsukada K and Watanbe I: | Int conf on 'Offshore welded structures', TWI, London, November 1982. |
|
| 36 | Metz R, Coleman S and Denny A: | Proc 7th int conf on 'Offshore mechanics and arctic engineering', Houston, Texas, February 1988. |
|
A complete list of references including article titles, etc, is available from the Editor of the Bulletin.
