Microstructure and Properties of Laser Welds in C-Mn Steels

P L Moore
University of Cambridge and TWI Ltd

D S Howse
TWI Ltd

E R Wallach

University of Cambridge 

Paper presented at 6^th International Conference on Trends in Welding Research, 15 - 19 April 2002, Callaway Gardens Resort, Pine Mountain, Georgia.

Abstract

High-power Nd:YAG laser welding is finding increasing use for industrial applications as it can easily weld complex geometries in a single pass. For thicker steel section, satisfactory joint quality, properties and understanding of the weld microstructure is needed.

Autogenous welds were made in seven C-Mn steels using 5.2 kW of Nd:YAG laser power at the workpiece. This power was obtained by beam combination of a 3 and 4 kW Nd:YAG laser. The welding speed ranged from 0.2 m/min to 0.6 m/min with a focus depth of -3 mm. Welds made in each steel under the same welding conditions were compared to determine the effect of steel composition on microstructure and the incidence of weld flaws. Hardness and Charpy impact tests were performed on selected weld samples. Traditional pipeline grade steels were compared with a plate steel and also a laboratory cast patented steel designed for laser weldability.

The welds giving the best properties exhibited a weld metal microstructure of primarily acicular ferrite, plus grain-boundary ferrite. Acicular ferrite significantly reduced the effective weld grain size, giving good toughness by retardation of crack propagation. Solidification cracking, commonly a problem in laser welding, was present only in welds that did not contain acicular ferrite. The patent steel performed consistently well, showing a favourable acicular ferrite microstructure with no flaws and a correspondingly good toughness.

Introduction

The formation of deep and narrow welds with low distortion, and ease of automation makes laser welding increasingly attractive for industrial thick section applications. In the past, the relatively low power of the Nd:YAG laser, compared to the CO₂ laser, has limited its usefulness. Today, however, Nd:YAG lasers are commercially available up to powers of 6 kW. In this work, a 3 and 4 kW laser were combined to create 5.2 kW of power at the workpiece. Nd:YAG lasers also have the advantage that they can be used with fibre optic delivery, vastly increasing the range of weld geometries and applications for which they can be used, such as girth welding pipelines. In this instance, a single high power laser weld can replace many GMAW fill passes.

In order to obtain consistently high standards of joint quality and properties, it is important that the microstructures of the welds are understood. The microstructures of high power Nd:YAG laser welds in seven different pipeline and structural steels were investigated, and related to the resulting weld mechanical properties and defects. These microstructures and properties were then explained in terms of the compositional differences between the steels.

Experimental

Beam combination of 3 and 4 kW Trumpf/Haas Nd:YAG lasers was used to obtain a workpiece power of 5.2 kW. This was used to make autogenous bead-on-plate laser welds in seven carbon-manganese (C-Mn) steels. A range of welding speeds between 0.2 and 0.6 m/min was used to vary the heat input with good keyhole stability, using a focus depth of -3 mm. All of the steels were welded at 0.3 m/min and this allowed comparisons of the weld microstructures to be made under the same welding conditions. Hardness and Charpy impact test data were used to back up the microstructural analysis. Radiography was used to inspect for internal porosity and solidification cracking.

Table 1: Material form and thicknesses of the seven steels

Steel	A	B	C	D	E	F	G
Source	pipe	pipe	pipe	pipe	pipe	plate	plate
t/mm	15.9	14.2	9.8	14.3	13.4	12	12

Of the seven low-alloy C-Mn steels investigated, A-E are commercially available pipeline grade steels, F is a commercially available plate steel, and G is a laboratory cast of a patented steel, designed for laser weldability ^[1] (see Table 1). The individual steel compositions are shown in Table 2.

Weld microstructure is influenced not only by the steel composition, but also the cooling profile. This in turn depends on the laser heat input, itself a function of power, coupling efficiency and welding speed, and the weld depth/width ratio and plate thickness. Although the thicknesses of the steels vary, they are thick enough for all the welds to have a 3D cooling profile, which is independent of thickness. ^[2] For this reason it was possible to isolate the effect of composition alone on the weld microstructure.

Table 2: Compositions (in weight %) of the seven steels, including the calculated carbon-equivalent (PCM) ^[3]

Steel	A	B	C	D	E	F	G
C	0.11	0.04	0.06	0.07	0.08	0.1	0.09
S	0.004	0.003	0.005	0.003	<.002	0.002	<.002
P	0.021	0.01	0.005	0.007	0.013	0.006	0.008
Si	0.33	0.27	0.31	0.31	0.35	0.47	0.24
Mn	1.42	1.58	1.47	1.42	1.6	1.31	1.47
Ni	0.023	0.17	0.007	0.75	0.019	0.35	0.003
Cr	0.018	0.11	0.16	0.12	0.017	0.06	0.002
Mo	0.003	0.22	0.21	0.27	0.003	0.01	0.003
V	0.002	0.003	0.04	0.004	0.032	0.005	0.001
Cu	0.023	0.3	0.012	0.012	0.011	0.13	0.002
Nb	0.04	0.066	0.072	0.066	0.044	0.027	0.017
Ti	0.003	0.022	0.015	0.01	0.003	0.002	0.012
Al	0.045	0.027	0.038	0.034	0.03	0.032	0.004
Co	<.004	0.01	<.004	<.004	0.005	0.01	<.004
B	3E-04	0.0003	0.0004	3E-04	0.004	5E-04	0.001
Ca	1E-04	0.0031	<.003	5E-04	0.002	0.002	<.0003
O	8E-04	0.0021	0.0012	0.001	0.002	0.003	0.001
N	0.003	0.0096	0.0054	0.008	0.004	0.011	0.003
PCM	0.196	0.168	0.173	0.190	0.197	0.200	0.178

Results & discussion

Microstructures

The high power Nd:YAG laser welds in the steels investigated tended to fall into two categories of microstructure (these are illustrated in Figs.1 and 2):

• primarily acicular ferrite
• primarily ferrite with an aligned second phase

Fig. 1. Primarily acicular ferrite weld microstructure, from steel B

  

 

Fig. 2. Primarily ferrite with an aligned second phase weld microstructure, from steel A

Acicular ferrite consists of small, intra-granularly nucleated, randomly oriented ferrite particles, often lamellar in shape. It is desirable in a weld microstructure as it reduces the effective grain size, has random orientation that inhibits crack propagation, and generally gives high toughness in welds. Low transformation temperature phases such as martensite or bainite are unfavourable as they have high hardness due to the quenched-in carbon, which has not had time to diffuse out of the primary austenite before transformation has occurred, and hence also low toughness. ^[4]

Microstructures consisting primarily of acicular ferrite often had grain boundaries decorated with primary ferrite, and some Widmanstätten ferrite laths growing out from the grain boundaries. The interior of the grains contained acicular ferrite.

The primarily ferrite with an aligned second phase microstructure in Fig.2 encompasses a range of aligned ferrite type microstructures, including Widmanstätten ferrite and bainite. It was often difficult to distinguish between Widmanstätten ferrite, aligned ferrite and bainite.

Generally Widmanstätten ferrite nucleates from grain boundaries as coarse ferrite side-plates. The majority of the microstructures in this category consisted of stacks of coarse parallel ferrite lamellae occurring intra-granularly, formed by the diffusion of carbon during cooling to form a carbide phase between ferrite plates. Bainite is similar in structure to this, being a sheaf of ferrite laths with carbide between them, except that it is much finer and more difficult to resolve under the optical microscope (upper bainite). When carbon has insufficient time during cooling to diffuse to the lath boundaries, it can form carbides within the ferrite lath (lower bainite). Bainite generally nucleates from the grain boundaries.

The patent steel G had been designed to produce a primarily acicular ferrite microstructure, and indeed it contained both grain boundary ferrite and acicular ferrite. Only steel B also showed this type of microstructure.

The rest of the steels were largely ferrite with an aligned second phase, containing varying amounts of bainite and Widmanstätten ferrite. The remainder of the microstructure comprised other microphases, or grain boundary ferrite. Steels C and E had slightly coarser aligned ferrite-type microstructures, and the grain boundaries could often be discerned due to the presence of small amounts of grain boundary ferrite and Widmanstätten ferrite. There was even some acicular ferrite present in the welds made in these steels. No grain boundary phases could be seen in the welds in steels A, D and F. Steel F showed the aligned ferrite layers quite clearly, and contained less of the finer bainite. Steels A and D had quite fine microstructures and contained more bainite with aligned ferrite.

The formation of acicular ferrite instead of bainite has previously been linked to the effect of prior grain size in the parent steel. ^[4] Acicular ferrite is associated with large parent metal grain size, and bainite with small parent metal grain size. The parent metal grain size for each of the seven steels was measured; ^[5] all the steels had a very fine grain size (3-6 µm). When ranked in order of relative grain size the data did not show the expected effect of grain size on microstructure. The relationship between microstructure and parent metal grain size appeared to be random for these steels.

Mechanical properties

Fully martensitic samples were generated from these steels, by water quenching small steel samples; the hardness was measured to give a range between 320-380 Hv, depending on the carbon content. None of the weld hardnesses were this high, indicating that none of the welds contained significant amounts of martensite. Bainite would give a lower hardness than this, and ferrite phases lower still.

Table 3: Average hardness properties of the seven steels

Steel	A	B	C	D	E	F	G
Hardness parent (Hv) Hardness weld (Hv)	195 260	249 260	226 242	227 251	213 252	161 255	198 206
Overmatch ratio	1.33	1.04	1.07	1.11	1.18	1.58	1.04

Since the parent metal hardness varied amongst the steels, from 161 to 249 Hv, the hardness overmatch ratio between weld and parent was used to make comparisons between the steels ( Table 3 ). The hardness overmatch ratios reflected the weld microstructure:

• the acicular ferritic welds in steels B and G had the lowest hardness overmatch ratio of just 1.04
• steels C, D and E with an aligned ferrite microstructure had mid-range overmatch ratios
• the greatest hardness overmatching was for the more bainitic welds in steels A and F

Although the parent steels showed quite a range in hardness, the average weld metal hardnesses among all the commercially available steels ranged from only 240 Hv to 260 Hv. In contrast, the patent steel G had a significantly lower average weld hardness of only 206 Hv. There were large differences in the toughness results, though, as shown in Table 4.

Comparing the steels to a Charpy acceptance criterion of 35 J at -10°C, only steels B, C and G passed easily. Steel F was borderline and steels A, D and E failed to show acceptable toughness. However fracture path deviation, often a problem in CO₂ laser welds, was not present in any of the Charpy samples.

Table 4: Toughness and defect properties of the seven steels

Steel	A	B	C	D	E	F	G
Impact energy (J) at -10°C	10	70	75	10	30	40	110
Acceptance	Fail	Pass	Pass	Fail	Fail	Pass	Pass
% Porosity	0.6	3.0	0.4	0.2	1.0	0.7	0.2
Acceptance	Pass	Fail	Pass	Pass	Pass	Pass	Pass
Cracking (mm)	47	0	6	43	0	0	0
Acceptance	Fail	Pass	Fail	Fail	Pass	Pass	Pass

Defects

Porosity levels in the steels were generally well below the acceptance criterion of 2% maximum ^[6] - except steel B, which had 3% porosity. This steel contained a high calcium level, which is known to affect weld bead shape, and hence molten metal flow, in other welding processes. ^[7] These high levels of porosity were disappointing since this steel had shown exceptional microstructure and mechanical properties.

Solidification cracking is a problem influenced not just by steel composition, but also by welding speed and bead shape. Cracking is not acceptable in any welds in pipeline fabrication ^[6,8] and so steels A and D (that showed over 40 mm of cracking) would fail, along with steel C (showing only 6 mm). None of the other steels showed any solidification cracking. It should be noted that cracking was only present in welds that did not contain acicular ferrite.

Effect of composition:

a) on solidification cracking

Elements that promote the formation of solidification cracking include phosphorus and nickel. ^[9] Of the steels investigated, the highest Ni content is in steel D (0.75wt%), and the most P is in steel A (0.021wt%). These two steels did indeed show the greatest amount of cracking, over 40 mm in the total length of weld. Manganese will reduce the cracking tendency of a weld. ^[10] Steels B and E had the greatest amount of Mn, and neither of these steels showed any solidification cracking.

A cracking index based upon the steel composition can give some feel for the relative likelihood of a particular steel to suffer solidification cracking. The following two cracking indices are calculated using the weight percentages of the elements. The parameter L gives a relative indication of cracking, ^[11] and UCS stands for Units of Crack Susceptibility. ^[12]

L = -1123(C) + 3096(C)² + 816(S) + 90     [1]

UCS = 230(C) + 190(S) + 75(P) + 45(Nb) - 12.3(Si) - 5.4(Mn)     [2]

Equation 1 was an inaccurate predictor of the relative solidification cracking, predicting most cracking in steel B (which showed none) and least in steel A (which had most cracking). Equation 2 was a better relative predictor, giving the highest value for steel A, medium values for steels C, D and E, and predicted least cracking in B. Two steels that did not fit well were steels F and G; these were predicted high values for cracking susceptibility, when they actually had zero cracking. Equations 1 and 2 were developed for CO₂ laser welds and arc welds respectively, whereas the weld profile of an Nd:YAG laser weld falls somewhere between the two, hence the limited accuracy of these predictive equations for this application.

b) on porosity

The convection of molten steel in the weld pool is determined by the surface tension gradient across the weld pool, which depends on the amounts of dissolved oxygen and sulphur present. ^[7] These, in turn, are controlled by elements such as silicon, aluminium and calcium that readily combine with them. Reducing the amounts of dissolved S and O changes the flow pattern, from radially outward from the centre of the weld pool to radially in. The latter convection pattern is more likely to cause gas to become trapped in the solidifying weld as porosity. ^[7] It is therefore expected that greater porosity might be seen in the steels containing the most Ca, Si and Al.

Of all the steels only B, E and F had Ca contents above 5 ppm. Steel F had the highest Si content (0.47wt%), the rest of the steels varied only slightly from around 0.3wt%, except G, which had the lowest value of 0.24wt%. Steel G also had by far the lowest Al content (only 0.004wt%), steel A contained the most (0.045wt%). The greatest porosity was seen in steels B and E with levels over 1%, steels A and F showed about 0.7% and the other welds had under 0.5% porosity. These findings fit the hypothesis that greater weld porosity is associated with high Ca, Si and Al content.

c) on hardness

The harden ability of a steel is given by the carbon equivalent (the PCM value, although often used for cracking predictions, ^[3] was found to be more useful for hardness predictions in these low carbon steels the IIW CE formula ^[12] ). The largest hardness overmatches were in Steels A and F, which also had high PCM values (0.2 and 0.196 respectively). Medium range overmatching in steels D and E reflects the lower PCM values for these steels (0.19 and 0.197), whereas the lowest overmatch and lowest PCM were in steels B, C and G (all below 0.18). This work demonstrates the correlation between the PCM value, the hardness overmatch ratio and the corresponding weld microstructure.

d) on weld microstructure

Nucleation of acicular ferrite is encouraged by low Al and O contents, and also with additions of titanium, since this may precipitate oxide particles with enhanced nucleation efficiency. ^[13] Boron also lowers the grain boundary energy to promote ferrite nucleation in the grain interior. ^[1] Ni will suppress ferrite formation in a weld. ^[4]

Steel G had by far the lowest Al and O content, compared to Ti. Of the commercial steels, B has the lowest Al and highest Ti content, explaining why it also shows an acicular ferrite microstructure. The steels with the largest amount of Al and O, relative to Ti, were those that had no acicular ferrite in the weld (A, E and F). All of the steels contained under 5 ppm of boron, except G (11 ppm) and E (the highest value of 40 ppm). However, this boron level on its own in steel E did not promote much acicular ferrite.

Temperature measurement

The weld microstructure obtained from a particular welding process and set of conditions is caused partly by the composition of the steel and partly by the cooling rate locally in the molten pool. It has been shown that the effect of steel composition on weld microstructure and properties can be easily understood and evaluated. But by knowing the cooling rate of a weld under particular welding conditions, better microstructural predictions can be made.

However, determining the cooling conditions during welding is significantly less trivial compared to establishing a steel's composition. The cooling rate of a weld depends on the heat input, the weld profile (its depth/width ratio and cross sectional geometry) and also the thickness of the workpiece (determining a 2D or 3D cooling profile). It can be difficult to make accurate predictions of the cooling rate, given all the variables involved, whereas being able to measure the weld temperature experimentally during welding can provide this information. But this requires an accurate method of temperature measurement of the molten pool and then solid weld during laser welding.

This requirement has led to the development of a novel application of a method for recording the weld cooling profile, by directly harpooning a thermocouple into the molten pool. This allows the actual cooling rate of the weld to be measured, and related to the weld microstructure and properties obtained. Other methods of temperature recording, avoiding the effects of the highly damaging laser beam and surrounding plasma, record only the surface temperature, or the temperature of the heat affected zone.

During the fabrication of the welds described in this paper, cooling rate measurement was not successful, and so no data are presented here. However, a method is now successful and a full discussion of the effect of cooling rate on laser weld microstructures is to be published.

Evaluation

Although evaluating the relative merits of the different ferrite with an aligned second phase type microstructures in steels A, C, D, E and F was difficult, it is possible to rank the different steels, using the mechanical properties and radiography results.

• Steel A had a high carbon equivalent and showed high weld/parent hardness overmatching. This is due to the formation of hard metastable phases such as bainite in the weld microstructure, as seen in Fig.2. This is one of the worst of the seven steels investigated for Nd:YAG laser welding due to the high hardness, low toughness and unacceptably high levels of solidification cracking.
• Steel B showed a favourable acicular and grain-boundary ferrite weld microstructure, illustrated in Fig.1, which proved to have a correspondingly low hardness overmatch ratio and high toughness. Unfortunately, the Nd:YAG laser welds in this steel showed unacceptably high levels of porosity. However, it must still be considered among the best of the commercially available pipeline steels for high power Nd:YAG laser welding.
• Steel C performed averagely when Nd:YAG laser welded. Its mechanical properties were good, passing the toughness criterion, and it had low levels of cracking and porosity; however, any cracking is unacceptable. Its microstructure had some acicular ferrite in the primarily aligned ferrite matrix, which might account for its better mechanical properties than other steels.
• Steel D's microstructure is similar to steel A, but on a slightly finer scale and, like A, also had significant solidification cracking and failed the toughness criterion. This steel had the greatest nickel content of the seven, suppressing ferrite formation in the weld and promoting bainite. The Nd:YAG laser welds in this steel were not of an acceptable quality.
• Steel E had a ferrite with an aligned second phase microstructure with bainite, but little other ferrite phases in it. It showed a correspondingly high hardness and low toughness. Even though this steel had no cracking and little porosity, overall, it did not form acceptable Nd:YAG laser welds.
• Steel F, the plate steel, was one of the better commercially available steels for Nd:YAG laser welding. Even though the weld did not contain acicular ferrite, it had no cracking and little porosity and had borderline weld toughness despite a high hardness overmatch ratio. This steel might be acceptable for high power Nd:YAG laser welding.
• Steel G, the patented laboratory cast steel, was the only steel of the seven to easily pass all the toughness and defect acceptance criteria, making it an ideal material for high power Nd:YAG laser welding. It showed a good acicular ferrite type microstructure and illustrates the favourable weld properties that can be achieved with that microstructure. However, though it appears an excellent material for Nd:YAG laser welding, it is currently commercially unavailable, and is presently of only limited usefulness for laser welding applications.

Summary & conclusion

The best weld properties were exhibited by welds with a primarily acicular ferrite microstructure, with some grain boundary ferrite. Solidification cracking was present only in welds that did not contain any acicular ferrite. Several types of ferrite with an aligned second phase, such as Widmanstätten ferrite and upper bainite, were commonly present. These microstructures do not inhibit crack propagation, and their presence can be associated with low weld metal toughness. Lower bainite and martensite have high hardness and can also give low toughness welds, and are to be avoided.

To obtain a weld microstructure that will have the best mechanical properties and a reduced likelihood of cracking, the steel composition should have a low Al content (below 0.03wt%) and low O content (below 30 ppm) with Ti present, in order to promote acicular ferrite. To prevent solidification cracking, the Mn level should be above 1.4wt%, the Ni content below 0.5wt% with the P content below 0.01wt%. A high Ni content will also suppress the formation of ferrite and promote bainite, which is undesirable. Having a Ca content below 20 ppm and restricting Si to around 0.3wt% can prevent weld porosity.

Acknowledgements

This work forms part of a larger collaborative project at TWI. ^[14] The authors would like to acknowledge TWI for supplying the hardness and toughness data referred to in this paper, and thank CRC Evans Pipeline International, BP and Corus for providing the materials. 

References

1. D Senogles, UK Patent Application 2341613 A (1998).
2. J Degenkolbe, D. Uwer & H. G. Wegmann, BSi internal report 91/79918 (1991).
3. AWS D1.1/D1.1M:2002: Structural welding code-steel (2002).
4. R W K Honeycombe & H K D H Bhadeshia, Steels-Microstructure and Properties, Butterworth-Heinemann (1995).
5. ASTM E112-96: Standard test methods for determining average grain size (1996).
6. BS 4515-1: Specification for welding of steel pipelines on land and offshore-Part 1: Carbon and carbon manganese steel pipelines (2000).
7. K C Mills, R F Brooks & A A Shirali, NPL report DMM(D)59 (1991).
8. API 1104: Welding pipelines and related facilities, 19th edition (1999).
9. J D Russell, 'Laser weldability of C-Mn steels', Assessment of Power Beam Welds, European Symposium, Geesthacht, Germany (1999).
10. J Kristensen, L. Hansen, S. Nielsen & K. Borggreen, 'Laser welding of C-Mn steels and duplex stainless steels', Assessment of Power Beam Welds, European Symposium, Geesthacht, Germany (1999).
11. S M I Birch, TWI internal report 7394.01/99/1026-03 (1999).
12. BS 1011-2:2001 Welding-recommendations for welding of metallic materials (2001).
13. R C Cochrane & D J Senogles, 'The effect of titanium additions on the microstructure of autogenous laser welds', Titanium Technology in Microalloyed Steels, Sheffield, UK (1994).
14. D S Howse, A C Woloszyn, I Hadley, P H M Hart, G S Booth & P L Baines, TWI internal report 12940/24/01 (2001).