Liquation Cracks in Thick Section Welds in Al-Mg-Si Plate

Presented at INALCO 98, 7th International Conference on Joints in Aluminium, Abington, Cambridge, UK, 15-17 April 1998.

Abstract

Study has been carried out on the welding behaviour of heavy section aluminium alloy 6082-T651. Welds were produced using the metal inert gas (MIG) process in 80mm material, using Al-Mg and Al-Si filler metals. Metallographic examination was carried out, with assessment of heat affected zone (HAZ) toughness by R-curve testing. Butt welds, made using Al-Mg (5356) filler wire, contained HAZ liquation cracks but those made with Al-Si (4043A) filler wire did not. This is consistent with the trends and experience previously established with these filler metals.

In the weld containing HAZ cracks, crack propagation during toughness testing was in the HAZ and resulted in the linking of individual cracks by tearing to create a serrated fracture path. For the sound weld, made using Al-Si filler metal, crack propagation took place by tearing in weld metal close to the fusion boundary. The toughness data generated indicated that the liquation cracking had only a minor effect on toughness levels associated with initiation of tearing. Thus, although the thick plate material would appear to be more susceptible to the formation of HAZ liquation cracks than sheet material, the current results suggest that their geometry and distribution may result in marginal significance with respect to mechanical performance.

Introduction

Although it has been recognised that HAZ liquation cracking in 6xxx alloys can be detrimental to mechanical properties(1-8), no fracture toughness values have been generated from cracked HAZs and there is little information on the welding characteristics of thick material. The aim of this work was to determine the significance of HAZ cracks in thick material, welded by the MIG process, with particular consideration of toughness behaviour as assessed by R-curve testing.

Experimental procedure

Materials

Alloy 6082 plate, 80mm in thickness, in the T651* condition was used and the chemical analysis is shown in Table 1. The Table also shows the analyses of the two welding wires employed, 4043A (Al-5%Si) and 5356 (Al-5%Mg-Mn): the diameter was 2.4mm.

*T651 = solution heat treated, controlled stretched and artificially aged to peak strength.

Table 1 Chemical analyses (wt%) plate and filler wires.

Element	6082 plate	4043A wire	5356 wire
Al	Remainder	Remainder	Remainder
Mg	0.80	<0.02	4.47
Si	0.91	4.90	<0.01
Mn	0.55	<0.01	0.17
Cu	0.06	<0.01	<0.01
Fe	0.27	0.22	0.17
Zn	0.05	<0.01	<0.01
Cr	0.01	<0.01	0.08
Ti	0.04	<0.01	0.07

Welding

Two 500mm long (MIG) welds were made, using 4043A and 5356 filler wires, designated W1 and W2 respectively. A single 'J' weld preparation was employed, as shown in Fig.1, to produce a planar HAZ on one side of the weld normal to the plate surface, to facilitate location of the notch for subsequent mechanical testing. The plates were butted together and held in a heavy restraining rig.

Preweld and inter-run cleaning was achieved using a stainless steel wire brush. A preheat and minimum interpass temperature of 50°C was employed. The heating was carried out using a propane torch. After filling the preparation from one side, the plate was turned over, a 5mm groove was cut on the reverse side (the preparation had a 4mm root face) and the joint was completed with a single run. Both welds had approximately 90mm long 'run on' and 'run off' plates which were discarded on completion of welding. Welds W1 and W2 contained 54 and 55 runs respectively. Welding conditions are shown in Table 2.

Table 2 Welding conditions

* Run 1 + Runs 2-54 # Runs 2-55
Process	W1 (4043A wire)	W2 (5356 wire)
	MIG, dc electrode + ve
Shielding gas	75%He -25%Ar
Flow rate, 1/min	40
Nozzle diameter, mm	25
Torch angle	Forward 10°
Wire diameter, mm	2.4
Arc voltage, V	25.7* 29.1	24.2* 30 #
Current, A	258* 342 +	300* 336 #
Welding speed, mm/min	266* 380 +	530* 380 #
Linear arc energy, kJ/mm	0.9* 1.6 +	0.9* 1.6 #
Arc time, s	63* 94 +	27* 102 #

Metallography

In addition to full weld cross sections, metallographic sections were also prepared perpendicular to one of the fracture surfaces of each of the fracture specimens. This was to reveal the location of the fatigue pre-crack tip and the path of the crack produced during CTOD testing. The microstructures were examined by light microscopy. The second fracture face was used for fractography, including scanning electron microscopy.

Tensile testing

Hounsfield tensile specimens were machined from the weld metals with their axes perpendicular to the direction of welding and parallel to the plate surface.

R-curve testing

R-curve tests were carried out on specimens extracted from each of the welded plates. Single edge-notched bend specimens were machined, with a thickness (B) of 37.5mm, a width (W) of 75mm and a surface notch, as shown in Fig.2. Notch locations were:

- fusion line, notched in root to cap direction,

- fusion line, notched in cap to root direction,

- weld metal centreline, notched in root to cap direction.

The reason for notching the HAZ specimens from both the root and the cap was the location of HAZ cracks. Liquation cracking (in the Al-Mg weldment, W2) was concentrated in the half of the section thickness nearest the cap. Hence if notches had been machined from the root side, the crack would have propagated through a cracked HAZ. Conversely, notches made on the cap side of the specimen would have been presented with a sound, uncracked HAZ. The same procedure was followed for the Al-Si weldment (W1), even though it was free from liquation cracks. Triplicate tests were carried out on each of the fusion line specimens and a single test on each of the two weld metals.

R-curves were generated using the single-specimen unloading compliance technique, according to a procedure described by Gordon and Leggatt ⁽⁹⁾ . In this technique, the specimen is repeatedly loaded and unloaded, allowing the fatigue precrack to extend by ductile tearing. The relationship between clip gauge opening and applied load is recorded. The multiple loadings and unloadings are logged and, at each unloading, the specimen compliance is measured. From this, the extension of the crack ( _Δ a) is estimated and a complete R-curve can be generated from a single specimen. Accurate estimation of _Δ a requires the crack front to be as straight as possible, as the crack extends. In order to ensure development of straight crack fronts, some of the specimens tested were side grooved to a depth of approximately 3.7mm on each side.

Tests were carried out at room temperature using a crosshead displacement speed of 1mm/min. Both the Crack Tip Opening Displacement (CTOD, or _δ ) and J-integral were monitored as a function of crack extension. R-curves of the form y=m( _Δ a+b) ^c (i.e. an offset power law curve) were then fitted to the data. The relevant parameters which can be measured from the R-curve are as follows:

- initiation toughness, J _0.2. This parameter is a simple engineering estimate, analogous to the use of a 0.2% proof stress as an estimate of yield stress. It is determined by fitting a curve to the J vs _Δ a or _δ vs _Δ a data, and identifying the point at which 0.2mm of ductile tearing has occurred.

- J at maximum load, J _m. This is a simple, single-point value of CTOD or J corresponding to the value at the peak of the load-displacement trace produced during a standard test to BS7448: Part 1, i.e. without multiple loadings/unloadings. The parameter _{δ m} tends to be higher than _{δ 0.2}, being associated with greater levels of ductile tearing; however, it has been shown ⁽¹⁰⁾ that it can be used in defect assessment, where the mechanism of failure is by ductile tearing.

For the test weldments, initiation parameters were made in terms of J. All tests were therefore analysed in terms of J _0.2, using an Offset Power Law (OPL) fit over the domain 0.2mm< _Δ a<3.5mm.

Results

Metallography

A photomacrograph and micrographs taken from the two welds are shown in Figs.3-5. The HAZ of W1 (made with Al-Si filler) was crack-free, but the HAZ of W2 (made with Al-Mg filler) contained large 'open' cracks, Fig.5. The cracks were restricted to the top half of the HAZ, i.e. from the mid-section towards the cap of the weld.

On closer examination of the fusion boundary/HAZ regions of W1, grain boundary films were observed, Fig.4. These films ran from the weld metal up to 0.5mm into the HAZ. These features were much longer than the 'open' HAZ cracks associated with the HAZ of W2 (Fig.5).

Tensile data

The Al-Mg weld metal had higher proof stress, tensile strength and ductility than the Al-Si weld metal, as shown in Table 3.

Table 3 Results of all weld metal tensile tests on W1 and W2

GL = Gauge length
Weld identity (Filler wire) [Filler alloy]	0.2% proof stress (N/mm 2)	Tensile strength (N/mm 2)	Elongation on 23mm GL (%)
W1 (4043A) [Al-Si]	108	182	13
	87	181	14
	93	182	14
W2 (5356) [Al-Mg]	131	251	28
	130	253	31
	131	251	29

R-curve tests

Results of the R-curve tests are presented in Table 4 and Fig.6. From this data, the following features were apparent:

1. The toughness associated with initiation of tearing, (J _0.2) was higher in the Al-Mg weld metal (W02-03) than in the Al-Si weld metal (W01-03). In addition, the tearing resistance of the Al-Mg weld metal beyond the point of initiation was markedly superior to that of Al-Si.Furthermore, the W02-03 (Al-Mg) specimen was side-grooved, which would have the effect of making the R-curve more shallow relative to that for a plane-sided specimen such as W01-03. Had similar specimens been compared, the differencein the R-curves shown in Fig.6 would have been even more pronounced.
2. The direction of notching did not appear to influence the shape of the R-curves.

Table 4 Analysis of results in terms of Offset Power Law (OPL), using J

* All notches were positioned in the fusion boundary except W01-03 and W02-03 where the notch was placed in the weld metal. +Poor fit to OPL curve.
Specimen identification	Region sampled*	Predominant fracture path	J 0.2 (kJ/m 2) from OPL
W01-01 (plane-sided)	Cap-root	Weld metal	15.1
W01-02 (plane-sided)	Root-cap	Weld metal	11.0
W01-03 (plane-sided)	Weld metal	Weld metal	18.5
W01-04 (side-grooved)	Cap-root	Weld metal	12.1
W01-05 (side-grooved)	Cap-root	Weld metal	17.2
W01-06 (side-grooved)	Root-cap	Weld metal	16.4
W01-07 (side-grooved)	Root-cap	Weld metal	7.6
W02-01 (plane-sided)	Cap-root	HAZ	8.1
W02-02 (plane-sided)	Root-cap	HAZ	16.1
W02-03 (side-grooved)	Weld metal	Weld metal	28.9
W02-04 (side-grooved)	Cap-root	HAZ	205 +
W02-05 (side-grooved)	Cap-root	HAZ	4.0
W02-06 (side-grooved)	Root-cap	HAZ	6.1 +
W02-07 (side-grooved)	Root-cap	HAZ	13.3

Examination of R-curve samples

The metallographic sections showed that all the HAZ notches were sited correctly. In all of the tests on the W1 HAZ, the crack path moved away from the HAZ into the weld metal. The sections also showed that, in all the tests associated with W2, there were fine cracks approximately normal to the fracture faces. In specimens from W2, notched in the root to cap direction, interlinking of numerous small HAZ cracks occurred and the crack path remained within the HAZ. The 'serrated' fracture edge created in the HAZ is shown in Fig.7.

The fracture surfaces of the fracture mechanics specimens had different morphologies for the two welds. The surfaces from W1 were generally smoother than those from W2, which were characterised by a series of steps, Fig.8. These steps were present irrespective of testing direction (i.e. cap to root, or root to cap). Figure 9 shows that individual HAZ cracks were linked by tearing and Fig.10 illustrates the typical intergranular morphology of the liquation cracks.

Discussion

HAZ cracking behaviour

The work has confirmed the sensitivity of 6082 alloy to liquation cracking. From the metallographic examination carried out, HAZ liquation occurred with both Al-Mg and Al-Si fillers, but in the latter case there may have been successful back-filling by low melting point material from the weld metal. This difference is consistent with the behaviour observed by Gittos and Scott ⁽¹⁾ . However, their model predicted that dilution of the filler metal in excess of 50% would be required for the bulk weld metal melting point to be sufficiently high to cause cracking. Dilution in the present welds is likely to have been somewhat below this level. However, it is noted that the geometry of the cracks, perpendicular to the fusion boundary and parallel to the rolling plane, was different from the more serious liquation cracking, parallel to the fusion line, which can occur in sheet material ⁽¹⁾ . It is reasonable to suggest that the cracks in the plate may be related to bands of segregation through its thickness. Locally enhanced alloy content in the HAZ could explain crack formation at lower dilution levels than in thin material. 

Toughness testing

The weld metal results were similar to values measured in welds in thinner sections ^(1,9) . In all cases, samples failed by ductile tearing, and it may be remarked that the measurements of HAZ initiation toughness were fairly low, in fact below those obtained for the weld metal. However, all specimens displayed a degree of tearing resistance and the failure mode was microvoid coalescence.

When the toughness data from the two sets of specimens notched at the fusion boundary are compared, it is clear that W1 (Al-Si weld) is in general superior to W2 (Al-Mg weld). The appropriate R-curves are shown to the same scale in Fig.8. There does not appear to be any convincing evidence from Fig.8 as to whether the shape of the R-curve depends on the direction of notching. For example, samples W02-04, W02-01 and W02-05 (Al-Mg weld), were notched in the cap-root direction; the ligaments through which the cracks propagated were therefore essentially crack-free. However, they do not appear to be substantially better or worse than samples W02-06, W02-02 and W02-07, in which the crack propagated through a cracked HAZ.

Fracture faces associated with W1 were relatively flat, whereas, those for W2 showed more topography. The outline of the individual runs was seen more clearly than on W1 and steps or splits normal to the fracture corresponding to HAZ liquation cracks were also seen on W2. Moreover, cracks which were not apparent on the lower half of the macrosection of W2, could be seen on the fracture faces, and this notching orientation included the lowest value of toughness recorded. It is likely that these cracks opened during toughness testing, becoming more visible, and this may have resulted in the similar initiation behaviour of 'root to cap' and 'cap to root' samples in W2. The orientation of the HAZ cracks (perpendicular to the crack propagation plane) may have produced a 'crack blunting' effect which has helped to minimise their significance.

Conclusions

1. Liquation cracks were observed in the HAZ of the weld made with Al-Mg (5356) filler metal but not in that made with Al-Si (4043A) filler metal. This is consistent with the trends and experience previously established with thesefiller metals.
2. For the sound weld, made using Al-Si filler, crack propagation in the toughness tests took place by tearing in weld metal close to the fusion boundary.
3. For the weld containing HAZ cracks, made using Al-Mg filler, crack propagation occurred in the HAZ, resulting in the linking of the liquation cracks by tearing and the formation of a characteristically serrated fracture path.
4. The toughness data generated indicated that the liquation cracking had only a minor effect on toughness levels associated with initiation of tearing.
5. Although the thick plate material would appear to be more susceptible to the formation of HAZ liquation cracks than sheet material, the current results suggest that their geometry and distribution may result in marginal significance with respect to mechanical performance.

Acknowledgements

The work was carried out within the core research programme of TWI which was jointly funded by The Department of Trade and Industry, and Industrial Members of TWI. 

References

Copyright by TWI, 1999

N°	Author	Title
1	Gittos M F and Scott M H,	'Heat affected zone cracking of Al-Mg-Si alloys'. Welding Journal, 60, 6, June 1981, 95s-103s.
2	Gitter R, Maier J, Müller W and Schwellinger P,	'Formation and effect of grain boundary openings in Al-Mg-Si alloys caused by welding - physical fundamentals and basic considerations leading to the creation of a quantitative and qualitative experimental programme', 5th Inalco '92, International Conf on Aluminium Weldments, Munich, 27-29 April 1992. 4.1.
3	Borst G, Linden H,	'First results of test programme for the formation and effects of grain boundary openings on the welding of medium strength Al-Mg-Si alloys. Effects of the welding practice'. Ibid 4.2.
4	Tsujimoto K, Sukaguchi A, Kinoshia T, Tanaka K and Sasabe S,	'HAZ cracking of Al-Mg-Si alloys', IIW doc IX-1273-83, 1983.
5	Kerr H W and Katoh M,	'Investigation of heat affected zone cracking of GMA welds of Al-Mg-Si alloys using the varestraint test', Welding Journal, 66, 9, Sept 1987, p251s-259s.
6	Kerr H W and Katoh M,	'Investigation of heat affected zone cracking of GTA welds of Al-Mg-Si alloys using the varestraint test', Welding Journal, 66, 10, Oct 1987, p360s-368s.
7	Scott M H and Gittos M F,	'Tensile and toughness properties of arc welded 5083 and 6082 aluminium alloys', Welding Journal, 62, 9, 1983, p243s-252s.
8	Pirner M,	'Properties of gas-shielded arc welded joints in heat treatable Al-Mg-Si and Al-Zn-Mg alloys', FWP Journal, 22, 3, 1982m p19-20,22,24,26.
9	Gordon J R and Leggatt R H,	'A comparison of CTOD ductile fracture analyses of aluminium alloy wide plate tests', TWI Members' Report 350/1987, September 1987.
10	Towers O L and Garwood S J,	'The use of maximum load toughness for ductile fracture assessments', TWI Research Report 157/1981.