Cracking in the stir zones of Mg-alloy friction stir spot welds

Abstract

Liquid penetration induced (LPI) cracking is investigated during friction stir spot weld of AZ91, AZ31 and AM60 magnesium alloys. A combination of stir zone temperature measurement and detailed metallography has revealed differences in the cracking tendencies of different magnesium alloys when the dwell time during spot welding is varied. LPI cracking in AZ91 spot welds involves the following sequence of events: the formation of $\alpha-\hbox{Mg}\,+\,\hbox{Mg}_{17}\hbox{Al}_{12}$ eutectic films in the thermo-mechanically affected zone (TMAZ) region immediately adjacent to the stir zone extremity, engulfment of melted eutectic films as the stir zone width increases during the dwell period, penetration of α−Mg grain boundaries and crack propagation when torque is applied by the rotating tool. Cracking occurs early in the dwell period during AZ91 spot welding and almost the entire stir zone is removed when the rotating tool is withdrawn. However, crack-free AZ31 and AM60 spot welds are produced when a dwell time of 4 s is used since the stir zone temperatures are much higher than the α-Mg + Mg₁₇Al₁₂ eutectic temperature (437 °C) and melted eutectic films dissolve rapidly following their engulfment by the growing stir zone. In contrast, the temperature during the dwell period in AZ91 spot welding is close to 437 °C and melted eutectic films are not completely dissolved so that spot welds produced using a dwell time of 4 s exhibit LPI cracking.

Appendix

Dissolution of a Melted Eutectic Films during AZ91, AM60 and AZ31 Spot Welding

Figure A.1 shows part of the binary Mg–Al equilibrium phase diagram:

Fig. A.1

Binary equilibrium phase diagram for Mg–Al

The driving force for diffusion and dissolution of liquid droplets is determined by the relation:

$$ k=2\left[ \frac{C_{T_2 }^{\alpha /liq-C_{T_1 } ^{\alpha /\beta }}}{{C_{T_2 } ^{liq/\alpha }-C_{T_2 } ^{\alpha /liq}}} \right] $$

For AZ91; k = 0.4, $C_{T_1 } ^{\alpha /\beta }=9\% \,,\,C_{T_2 } ^{\alpha /liq}=13\% \,,\,C_{T_2 } ^{liq/\alpha }=33\% \,,\,T_2 =710\,\hbox{K}$

For AM60; k = 0.7, $C_{T_1 } ^{\alpha /\beta }=6\% \,,\,C_{T_2 } ^{\alpha /liq}=13\% \,,\,C_{T_2 } ^{liq/\alpha }=33\% \,,\,T_2 =710\,\hbox{K}$

For AZ31; k = 1.0, $C_{T_1 } ^{\alpha /\beta }=3\% \,,\,C_{T_2 } ^{\alpha /liq}=13\% \,,\,C_{T_2 } ^{liq/\alpha }=33\% \,,\,T_2 =710\,\hbox{K}$

The diffusion coefficient depends on the relation:

$$ D=D_o \exp \left[ {-\frac{Q}{RT}} \right] $$

where$R=8.314\times 10^{-3}$ kJ/(mol · K) and the diffusion rate of Al in Mg is [20]:

$$ D_o =1.53\times 10^{7}\,\mu \hbox{m}^{2}/\hbox{s},\quad Q=125 \hbox{kJ/mol}$$

For AZ91;

$$ D_{\rm Al\,in\,Mg\,at\,710\,K} =9.73\times 10^{-3}\,\mu {\rm m}^{2}/{\rm s}\quad \hbox{at 710\,K} (437\,{\deg}\hbox{C}) $$

$$ D_{\rm Al\,in\,Mg\,at\,733\,K} =1.89\times 10^{-2}\,\mu \hbox{m}^{2}/\hbox{s}\quad \hbox{at 733\,K} (460\,{\deg}\hbox{C}) $$

For AM60;

$$ D_{\rm Al\,in\,Mg\,at\,773\,K} =5.47\times 10^{-2}\,\mu \hbox{m}^{2}/\hbox{s} \quad \hbox{at 773\,K} (500\,{\deg}\hbox{C})$$

For AZ31;

$$ D_{\rm Al\,in\,Mg\,at\,787\,K} =7.73\times 10^{-2}\,\mu \hbox{m}^{2}/\hbox{s} \quad \hbox{at 787\,K} (514\,{\deg}\hbox{C}) $$

$$ D_{\rm Al\,in\,Mg\,at\,823\,K} =1.78\times 10^{-1}\,\mu \hbox{m}^{2}/\hbox{s at 823\,K} (550\,{\deg}\hbox{C}) $$

The relation between the half-thickness of a plate-shaped liquid droplet and the time available for dissolution at 710, 733, 773, 787 and 823 K is determined by the relation:

$$ B_0 =\frac{k}{\sqrt{\pi }}\sqrt{D_{\rm Al\,in\,Mg\,at\,(710,733,773,787,823)K} \cdot t} $$

It is worth noting that this equation is a first order approximation and is derived by assuming no overlapping of the concentration fields from neighboring films.