made available by this study were finally examined, as a last attempt to link fracture characteristics, microstructure, and com- position. Figure 12A is a low-magnifica- tion secondary electron image of a typical weld interfacial fracture in DP20. In welds with sizeable voids, like in Fig. 12A, the possibility that defects such as voids influ- enced fracture was not dismissed, knowing that voids notoriously decrease spot weld properties (Refs. 15, 31). Figure 12B is a high-magnification view of the large round void of Fig. 12A. The fine dendrites at its surface present additional strong ev- idence that this void resulted from solidi- fication shrinkage. Also, the cracks at its perimeter indicate that shrinkage caused sufficient tension to split the fine inter- locked dendrites apart; i.e., induce a cracking that this time could not be linked to a local change in chemical composition, as was found in Fig. 11. For the interme- diate region between the void and the edge of the weld, Fig. 12C shows that a substantial part of the weld exhibits the characteristic microvoids of a ductile frac- ture, in spite of the high hardness values (Fig. 9) associated with the martensite mi- crostructures (Fig. 10E). Figure 12D shows that brittle fracture by cleavage also occurred, particularly on the side of the weld (i.e., HAZ) where the microstructure was the hardest — Fig. 9.
As for the zinc expulsion, the weld but- ton diameter, and the type of weld frac- ture, the effects of current, force, and time on the shrinkage voids were all investi- gated for both steels. Quantitative results to compare voids in DP18 and DP20 are provided first in Fig. 13, which like Fig. 14, was constructed after estimating pro- jected areas of all sizeable voids seen on the interfacial fracture surfaces. Figure 13 clearly shows that the percentage of the fused area covered by the voids was greater in DP20 than in DP18, a feature that is also in line with the greater suscep- tibility of DP20 welds toward interfacial fracture.
Figure 13 shows that increasing the weld time, thus the weld diameter, re- duces the percentage area made by voids in the fusion zone, and thus any contribu- tion voids might have to fracture. Figure 14 confirms these results by showing that an increase in weld diameter, as achieved by raising the current, also decreased the voids rapidly and eliminated weld interfa- cial fracture (beyond 12 kA). Of the two process variables in Fig. 14, weld force af- fected most distinctively the relative im- portance of voids. Voids were reduced dramatically at 1200 lb (5.3 kN) and 1700 lb (7.6 kN), where they were nearly nonexistent. This contrasted with the 20% of the fusion zone they covered at 600 lb (2.7 kN).
Despite evidence that voids could be controlled by process parameters, and in particular reduced by high forces, we found no indication that voids directly caused interfacial fracture. However, we proved that voids and sheet thickness were related by showing that more voids were present in the thicker steel — Fig. 13.
Among all fracture surfaces, we dis- covered that those associated to abnor- mally fast weld schedules (e.g., 24 kA and 5-cycle weld time) provided a new expla- nation for spot weld fracture in DP600 steels. Unlike the welds examined previ- ously, the chisel testing of welds made with high currents and extremely short current pulses produced buttons with un- common characteristics. Figure 15A–F detail one of those welds. First, note the presence of several large eccentric voids, including two on the right of Fig. 15A that appear to be superimposed. Figure 15B, for a part of the fusion zone periphery, shows an abrupt separation between a ductile region with microvoids, and a den- dritic solidification structure. Typical characteristics of ductile fracture are also visible in Fig. 15C, as are those of a cleav- age brittle fracture. Also, carbon-rich in- clusions (likely carbides) were frequently detected in the ductile regions, where they were found deep inside microvoids. Their concentrations were not quantified, since they would have improbably ex- plained DP20’s greater weld interfacial fracture susceptibility.
The microstructure shown in Fig. 15D revealed another characteristic feature of these welds made with abnormally fast schedules. Figure 15D is a high-magnifi- cation view of the crack that was seen in Fig. 15B. In Fig. 15D, a heavy coating of zinc, confirmed by EDS, is seen on den- drites. Zinc was also encountered for the microconstituent of glassy appearance of Fig. 15E. EDS measurements revealed that its composition matched that of an iron-zinc spinel; i.e., FeZn2O4, a phase we however did not try to validate. In Fig. 15F, where part of the circular void of Fig. 15A is depicted, zinc was also found quite ho- mogeneously distributed over its surface. The fact that zinc was found at dendrites and voids simply confirms that solidifica- tion cracking occurred. This cracking, best revealed in Fig. 15D, can be well under- stood from the binary phase diagram with iron (Ref. 30). The Fe-Zn phase diagram indicates that zinc rejection from the iron solid solution stabilizes a zinc-rich liquid at temperatures as low as the zinc melting temperature; i.e., a perfect condition for cracking to occur, even under normal shrinkage conditions. Although this crack-
ing is clearly detrimental to the welds, we found that chisel-testing zinc-infiltrated welds had consistently produced buttons; an example that demonstrated that chisel testing, and other mechanical tests alone would be inappropriate for these particu- lar welds.
1) The two DP600 steels were success- fully resistance spot welded to produce welds resisting interfacial fracture during chisel testing. Low currents (<10 kA), ex- tended weld times (>25 cycles), and high weld forces (>900 lb, or 4.0 kN) promoted weld button formation by producing large fusion zones and occasionally deep (but acceptable) electrode indentations. The greater susceptibility of the DP20 steel to produce weld interfacial fracture was con- sidered to be the results of its thicker gauge (2.0 mm) compared to the DP18 steel (1.8 mm).
2) As explained by carbon equivalents, weld fusion zones in DP600 steels contained mainly martensite, and weld microstruc- tures in the DP18 steel were harder than in the DP20 steel. The dominant effect of sheet thickness was demonstrated by the fact that 1.8-mm-thick DP180 steel showed lower susceptibility to weld interfacial frac- ture than the thicker DP20 steel, despite its higher carbon equivalent number.
3) Galvanized coatings generally have insignificant effects on weld fracture. How- ever, decreasing weld time and increasing weld current (not typical in production) re- sulted in zinc ingestion into the fusion zone, which caused solidification cracks. In this investigation, ingestion of zinc was observed at the same time type of fracture changed from one type to the other.
4) Shrinkage voids, recognizable by their dendritic surface morphology, were ob- served in many welds. High current, long weld time, and high weld force all helped re- duce shrinkage voids. Shrinkage voids were less pronounced in the thinner DP18 steel, which also had lower susceptibility for weld interfacial fracture.
The authors would like to acknowledge Chris Chen, Alexander Turley, and David Sigler from General Motors for providing frequent valuable discussions. Technical support from Bob Cubic for the scanning electron microscopy and Rich Valdo for the microprobe analysis are also deeply appre-
1. ULSAB-AVC Consortium, Technical Transfer Dispatch #6 (Body Structure Materi- als), May 26, 2001.
WELDING JOURNAL 181 -s