Effect of Friction Stir Welding on Bulk Metallic Glasses

ZHANG Xiangyun, ZHANG Mi, CHEN Wenbin, YUAN Zizhou

(1. State Key Laboratory of Advanced Processing and Recycling of Nonferrous Metals, Lanzhou University of Technology, Lanzhou 730050,China; 2. Wenzhou Engineering Institute of Pump&Valve, Lanzhou University of Technology, Wenzhou 325105, China)

Abstract: In order to enlarge the size of bulk metallic glasses (BMGs), two Cu36Zr48Al8Ag8 BMGs plates were successfully welded by friction stir welding (FSW) without obvious crystallization. The effect of friction stir welding on microstructure and mechanical properties of the BMGs was investigated. X-ray diffraction (XRD)and scanning electron microscope (SEM) were used to evaluate the changes of the crystalline particles in the BMGs. Nanoindentation was applied to analyze the changes of the amorphous matrix. Micro-hardness of the stir zone before and after FSW was tested to explore changes of the mechanical properties. Results show that the original Al3Zr particles in the BMGs were refined and some cavities parallel to the rotating direction of the pin were observed in the amorphous matrix after FSW. Furthermore, micro-hardness of the stirred zone rises approximately 50 Hv compared with the as-cast sample due to structure relaxation of the amorphous matrix.

Key words: bulk metallic glasses; friction stir welding; micro-hardness; nanoindentation

1 Introduction

BMGs has great potential to be used as new structural material in engineering applications because of their outstanding mechanical properties such as superior strength, high hardness and excellent wear resistance. But their attainable maximum sizes is limited by the critical cooling rate[1]. FSW, a typical solid-state joining technique, is an optimal method to enlarge the size of BMGs. Materials shouldn’t be remelted and crystallization will be prevented to the maximum extent during the welding process.

In recent years, some BMGs have been successfully joined by FSW[2,3]. But for the research on FSW of BMGs at its beginning stage, there are so many problems need to be solved, among which,the microstructure and mechanical properties of the BMGs joint is the most important and urgent one. On the one hand, heat created by friction between the workpiece and rotating tool during the welding process will bring about hardening in BMGs due to structure relaxation of the amorphous structure[4]. On the other hand, intense plastic deformation of the workpiece will induce softening of the amorphous structure attributed to the increase in free volume[5]. Which one will finally decide the mechanical properties of the joint?The understanding of this problem is of the utmost importance to the application of FSW to the joining of BMGs.

In this paper, Cu36Zr48Al8Ag8BMGs was welded by friction stir welding in the supercooled liquid region.In addition, in order to study the effect of FSW on the BMGs, microstructure and hardness of the BMGs before and after FSW were observed.

2 Experimental

Plates of Cu36Zr48Al8Ag8BMGs with a size of 50 mm×20 mm×2.4 mm were prepared by copper mold suction casting method under an Ar atmosphere.Two plates of BMGs were successfully butt-jointed by a computer numerical controlled vertical milling machine. The rotating tool used in the weld process was made of tool steel and had a cylindrical shoulder with 12 mm in the diameter and a pin of 2 mm in length. The frustum cone-like pin is 3 mm and 2 mm in the diameter of upper and lower sides respectively.

In order to investigate the effect of FSW on the BMGs, contrasting experiments research between the as cast sample and stir zone of the welded specimen were conducted. The cross-sections of the joint perpendicular to the welding direction were cut off by wire cutting machine for testing. Microstructures of the samples were examined by SEM (QUANTA FEG 450)and XRD (Rigaku D/max-RB) with Cu-Kα radiation at a scanning rate of 2°/min and a detecting step of 0.02°.Differential scanning calorimetric (DSC, STA449C)was used to determine the glass transition temperature(Tg=714 K) and crystallization temperature (Tx=798 K) at a heating rate of 50 K/min under the protection of high purity Ar gas. Micro Vickers hardness test was carried out at an applied load of 500 gf and dwell time of 10 s. Nanoindentation creep tests were conducted at room temperature on a Nanotest600 nanoindenter utilizing a diamond Berkovich indenter. The indentation creep tests were loaded to 10 mN with different loading rates of 0.075, 0.2, and 1 mN/s, and then held for 10 s to evaluate their creep behaviors. Thermal drift rate was maintained below 0.05 nm/s during each test.

3 Results and discussion

3.1 Microstructure

A smooth nugget zone was achieved under the tool rotation speed of 400 rpm, welding speed of 20 mm/min, and the plunge depth of 0.1 mm. Fig.1 shows a typical cross-sectional macrograph of the joint. No unwelded parts were detected in the stir zone, but some pore defects were observed on the retreating side (RS)of the joint. As noted by many researchers[6-8], however,the advancing side (AS) is typically hotter than the retreating side. So the asymmetric pore distribution may be related to the different temperature distribution and the corresponding flow viscosity diversity of the material during the welding process.

Fig.1 Microphotograph of the cross-section of the joint

Fig.2 shows XRD patterns of the samples before and after FSW. The patterns display only broad halo peaks without appearance of sharp Bragg peaks,indicating that no obvious crystallization happened during the entire FSW process. But SEM images of the stir zone before and after FSW in Fig.3 indicate that changes happened during the FSW. Back-scattering electron image of the as-cast samples in Fig.3(a) shows that dendrites with the maximum size about 5 μm randomly distributed in the metallic glass matrix. These dendrites are indentified to Al3Zr by energy spectrum analysis (EDS) and our further study[9]. While the dendrites observed in the stir zone near the top surface of the sample, as shown in Fig.3(b), are much finer and more. Mukherjeeet al[10]also found that dendrites get fragmented during the friction stir process. The dendrites in the top surface of the welded sample in Fig.3(b) seems to be more than the as-cast sample,which may be resulting from nanocrystallization of a little part of the amorphous matrix. Furthermore, the dendrites size increases and the quantity decreases with increasing the depth from the top surface of the sample.The inhomogeneous dendrites distribution phenomenon may be also related to the inhomogeneous heat and viscosity flow behavior during FSW.

Fig.2 XRD patterns of the specimen before and after FSW

Fig.3 SEM images of the stir zone (a) before and (b) after FSW

3.2 Mechanical properties

Fig.4 shows micro-hardness distribution across the centre of the cross-section perpendicular to the welding direction. The average micro-hardness of the as-cast sample is 525 HV. However, the microhardness of the welded sample increases, and the average hardness of the stir zone reaches up to 579 Hv.To further understand the changes in micro-hardness,the plastic deformation regions around the indents after indentations were studied. Fig.5 shows back-scattered SEM images of the typical surface morphologies of indents for the specimens before and after FSW.Obvious circular patterns and pile-up representing the plastic and tough nature of materials can be observed around the indents edges for the as-cast specimen(Fig.5(a)). But in the case of the welded sample, no such distinct features were observed, indicating that the amorphous matrix is hard enough with no toughness and flow of material (Fig.5(b))[11]. Therefore, there must be some microstructure changes in the BMGs during the FSW process. Furthermore, a lot of cavities parallel to the rotating direction of the pin are presented in the amorphous matrix of the welded sample, which may be caused by the insufficient liquidity of the amorphous matrix during FSW.

Fig.4 Micro Vickers harness profile of the welded sample

Fig.5 Typical morphologies around the indents after indentation measurements: (a) as-cast specimen; (b) stir zone

All these elements, including the cavities,crystalline particles, and amorphous matrix, have impact on micro-hardness of the BMGs. Cavities always decrease micro-hardness of the BMGs[12]. But it’s difficult to test the effect of dendrite changes on hardness of the BMGs, while the amorphous matrix may be hardened or softened by the free volume annihilation and intense plastic deformation during the FSW process. Nanoindentation is an ideal technique to investigate the microscopic mechanical deformation of BMGs due to its high accuracy, including the plastic properties and the elastic properties[13,14]. In order to find out the main reason of micro-hardness change during FSW, the amorphous matrix as the majority of the material was further studied by nanoindention test.

All the indenters were located on the amorphous matrix of the samples. Fig.6 shows nanoindentation load-displacement (P-h) curves for the as-cast sample and stir zone of the welded sample. There are obvious indentation creep phenomena in all the curves during peak-load holding segment. Furthermore, the welded sample exhibits higher nanohardness (from approximately 5.5 GPa for the as-cast sample to approximately 5.9 GPa for the stir zone of the welded sample) at any given indentation loading rate than the as-cast samples. This observation is consistent with the above micro-hardness results. Therefore, the total hardness may be dominated by the amorphous matrix.

Fig.6 Load-displacement curves of the as-cast sample and stir zone of the welded sample

To further indentify changes of amorphous matrix during FSW, the time dependent creep behaviors of the samples were analyzed. Fig.7 shows the creep displacementversustime (h-t) data during the peakload holding period for the samples before and after FSW. Starting points of the creep displacement have been zeroed for comparison. It can be seen that the total creep displacement during the peak-load holding segment depends on the loading rate for all the samples. All the creep displacement increases with increasing the loading rate and all the creep process can be segmented into two stages: the transient creep (the creep displacement increases rapidly with holding time)and the steady-state creep[13]. Furthermore, even though the total creep displacement of the as-cast samples are just slightly larger than that of the welded specimen at any given loading rate, creep displacements of the welded samples reach their steady values after 5 seconds holding time, while the displacements of the as-cast specimens are continually increasing in the whole 10 seconds holding time. This phenomenon implies that severe structural relaxation occurred during the FSW process, which enhanced the creep strength and creep resistance[14].

Fig.7 Creep displacement versus time during the peak-load holding period with various loading rates for the (a) as-cast and (b) welded specimen

Considering that the maximum creep displacement during the peak-load holding stage can minimize the viscoelastic deformation in the loading stage, theh-tdata of the BMGs loaded at the loading rate of 10 mN/s was chosen to further study their creep behavior,as shown in Fig.8. It can be seen that both the creep displacement data of the BMGs before and after FSW can be accurately fitted by the generalized Kelvin model with the two exponential terms (Eq.(1))[15]. The fitting parameters are summarized in Table 1.

Fig.8 Creep displacement-time curve of the BMGs during the load holding stage with loading rate of 10 mN/s

whereherepresents the depth of the first spring pressure,hirepresents indentation depth of thei-th Kelvin element,τipresents the retardation time of thei-th Kelvin element. In amorphous polymers, the retardation times are related toαandβtransitions[16].μ0is a constant related to the viscosity coefficient of the last dashpot.

Creep complianceJ(t) and the delay spectrumL(t)are important mechanical physical quantity that reflects the properties of materials. Based on the generalized Kelvin model,J(t) andL(t) given by Eqs.(2) and (3)were induced from the fitted parameters in Table 1, as shown in Figs.9(a) and (b), respectively.

Table 1 Fitting parameters in Eq.(3) for the Cu36Zr48Ag8Al8 as cast and the stir zone

whereA0andP0are the contact area and the applied load corresponding to the virtual length, respectively.The virtual lengthhin, approximately equals the depth at the end of the loading.

Both compliance curves of the BMGs samples before and after FSW are initially constant and then increases with holding time due to structural relaxation during the peak-load holding stage[17,18], as shown in Fig.9(a). Both the retardation spectra are made up of two peaks, as shown in Fig.9(b), indicating that both the creep processes of the samples before and after FSW consist of two kinds of relaxation processes. Similar phenomena have been observed in Ti[17], La[18], Ce[15,18],Cu[19]and Mg[20]based BMGs. Compared with the welded sample, the as-cast specimen with relatively larger compliance value and sharper retardation peaks indicates its more relaxed state, which also reflects that severe structural relaxation had occurred during the FSW.

Fig.9 (a) Creep compliance and (b) delay spectrum of the as-cast and welded specimen during the peak-load holding period at the loading rate of 10 mN/s

Therefore, it is reasonable to suppose that the over all micro-hardness of the BMGs is dominated by the hardness of the amorphous matrix. Even though the deformation during the FSW process may produce a large number of free volumes, the overall free volume of the amorphous matrix decreased and the hardness increased due to severe structural relaxation caused by heat input.

4 Conclusions

A study on microstructure and mechanical properties of Cu36Zr48Al8Ag8BMGs before and after FSW was held to investigate the effect of FSW on BMGs. The BMGs were successfully welded with almost no crystallization in the stir zone. But the original particles in the BMGs were refined and some cavities parallel to the rotating direction of the pin are presented in the amorphous matrix during the FSW process. Furthermore, micro-hardness of the stir zone rises approximately 50 HV compared with the ascast sample due to severe structure relaxation of the amorphous matrix. This study may be useful for the optimization of the FSW process of the BMGs.