Journal of Materials Science & Technology  2019 , 35 (7): 1388-1392 https://doi.org/10.1016/j.jmst.2019.03.007

Orginal Article

In-situ study on hydrogen bubble evolution in the liquid Al/solid Ni interconnection by synchrotron radiation X-ray radiography

Zongye Dinga, Qiaodan Hua*, Wenquan Lua, Xuan Gea, Sheng Caob, Siyu Suna, Tianxing Yanga, Mingxu Xiaa, Jianguo Lia

aShanghai Key Laboratory of Materials Laser Processing Modification, School of Materials Science and Engineering, Shanghai Jiao Tong University, 200240, Shanghai, China
bMonash Centre for Additive Manufacturing (MCAM), Monash University, Clayton, VIC, 3800, Australia

Corresponding authors:   *Corresponding author.E-mail address: qdhu@sjtu.edu.cn (Q. Hu).

Received: 2019-01-5

Revised:  2019-02-1

Accepted:  2019-02-14

Online:  2019-07-20

Copyright:  2019 Editorial board of Journal of Materials Science & Technology Copyright reserved, Editorial board of Journal of Materials Science & Technology

More

Abstract

Synchrotron X-ray radiography was used to carry out an in-situ observation of the hydrogen bubble evolution in the liquid Al/solid Ni interconnection. The individual bubble mainly grows in a stochastic way during heating. The size distribution for groups of bubbles follows a Gaussian distribution in the early stage and Lifshitz-Slyozov-Wagner (LSW) diffusion controlled distribution in the final stage. The intermetallic compounds (IMCs) first form during solidification, following by the hydrogen bubbles. The bubbles between two adjacent Al3Ni grains grow unidirectionally along the liquid channel, with the bottom being impeded by the Al3Ni phase and the radius of the growth front being smaller. For the bubbles at triple junctions, they grow along the liquid channel and the crack with morphology transition.

Keywords: Synchrotron radiation ; Liquid Al/solid Ni interconnection ; Hydrogen bubble ; Intermetallic compounds ; Growth behavior

0

PDF (2608KB) Metadata Metrics Related articles

Cite this article Export EndNote Ris Bibtex

Zongye Ding, Qiaodan Hu, Wenquan Lu, Xuan Ge, Sheng Cao, Siyu Sun, Tianxing Yang, Mingxu Xia, Jianguo Li. In-situ study on hydrogen bubble evolution in the liquid Al/solid Ni interconnection by synchrotron radiation X-ray radiography[J]. Journal of Materials Science & Technology, 2019, 35(7): 1388-1392 https://doi.org/10.1016/j.jmst.2019.03.007

1. Introduction

Due to considerable practical importance in Transient Liquid Phase Bonding (TLPB), aluminide coating of nickel, soldering and brazing, it is necessary to understand the diffusion reaction processes in the liquid Al/solid Ni interconnection [[1], [2], [3]]. The interaction included dissolution of solid Ni into liquid Al, migration of Al atoms into the solid Ni lattice, and formation of intermetallic compounds (IMCs) [4]. There were IMCs and porosities observed at the interface, which were essential to the mechanical properties for the interconnection. It was well documented that two kinds of IMCs, Al3Ni2 and Al3Ni phases, formed at the interface with the remnant liquid Al after solidification [5,6]. These previous studies mainly focused on the IMCs formation and growth at the liquid Al/solid Ni interface, and there is still a lack of understanding in porosity formation and growth during diffusion.

The insoluble gas evolution during solidification always resulted in the formation of porosities in the liquid/solid interconnection, which was detrimental to mechanical properties [7]. Therefore, it is vital to understand the formation and growth behavior of the bubbles in order to achieve porosity-free products [8]. While to the authors’ knowledge, the gas porosity evolution at the liquid Al/solid Ni interface has been not reported.

Hydrogen is the only gas that is considerably soluble in molten aluminum. The dissolution of atomic hydrogen in Al melt involved dissociation of gaseous hydrogen molecule into atomic form at the bubble edge, dissolution of atomic hydrogen through the melt boundary layer, and diffusion into the melt [9]. Those formed hydrogen bubbles are important source for porosity after solidification. For the liquid Al/solid Ni interconnection, the bubble growth behavior and the interaction between IMCs and bubbles are still unclear, which are key factors to further understand and control the interfacial microstructure [10].

Several studies recently show that the synchrotron radiation X-ray quantification is feasible for in situ observation of the dendritic and hydrogen bubble evolution in light alloys [[11], [12], [13]], and growth behavior of IMCs at liquid/solid interfaces [14]. Comparing to the conventional methods, it provides the direct and dynamic information on microstructural evolution. In this paper, the hydrogen bubble evolution in the liquid Al/solid Ni interconnection was investigated by using synchrotron radiation real-time imaging technology. The growth behavior of hydrogen bubbles during heating and solidification were discussed.

2. Experimental

Pure Al (99.999%) and pure Ni (99.99%) sheets with the dimensions of 10 mm × 10 mm × 0.4 mm were pre-polished for the Al/Ni interconnection. Two pieces of Al2O3 ceramic sheets with a thickness of 0.5 mm were used to fix the interconnection. The synchrotron radiation experiment was carried out in the BL13W1 beam line at Shanghai Synchrotron Radiation Facility, China. The samples were placed in a designed furnace. Two windows vertically aligned with the X-ray beam with energy of 26 keV, as shown in Fig. 1(a). The sample was slowly heated to 800 °C for 1 h in order to achieve homogenization during holding period, followed by rapid cooling in the furnace. Two thermal couples were used to measure the temperatures of the Al and Ni plates, and the average value was considered as the sample temperature. A high speed CCD camera with a resolution of 3.25 μm/pixel and an exposure time of 1 s was used to record the images. The working distance between the detector and sample was 85 cm. For the image analysis, the raw images should be processed through noise reduction, filtering, and background correction. The size of every bubble was measured for three times to obtain the average value. For the growth behavior of groups of bubbles, all of the bubbles in the imaging field were measured. In addition to the in-situ synchrotron characterization, scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) was used to examine the morphology and chemical composition of the interfacial microstructures after solidification.

Fig. 1.   (a) A schematic diagram of experimental set-up; (b)-(i) bubble evolution during heating in liquid Al/solid Ni system.

3. Results and discussion

3.1. Bubble evolution during heating

The hydrogen bubble evolution during heating in the liquid Al/solid Ni interconnection is shown in Fig. 1(b-i). The upper gray region is liquid Al, and the black zone at bottom is Ni substrate. There are not hydrogen bubbles observed at beginning, as shown in Fig. 1(b). After 12.5 s, a large number of spherical bubbles formed in the liquid Al at 687.1 °C, the interface was smooth, and the liquid Al was bonded well with the Ni substrate (Fig. 1(c)). With further increased temperature, the number of bubbles increased rapidly followed by a gradual decreasing, and finally disappeared at 117.5 s, as shown in Fig. 1(d)-(i). Three kinds of bubble behaviors can be distinguished: (1) most of bubbles grew to a maximum, then shrunk and disappeared, such as bubbles 1, 2 and 3 in Fig. 1(d); (2) some bubbles (bubble 4) in Fig. 1(f) grew first, and then suddenly burst; (3) some bubbles, such as bubble 5 in Fig. 1(g), grew and burst along with the shape changing from sphere to spheroid.

The solubility of hydrogen in liquid Al is determined by log10S=-A/T+B, where S is the solubility in cm3 of H2(g) at standard pressure per 100 g of Al, A and B are the parameters determined by the Ni concentration. The hydrogen solubility in the Al melt increased with temperature increasing. The interfacial concentration of dissolved hydrogen gas in the liquid Al can be obtained from Seivert’s law Cint=qPg, where q is the Seivert’s constant and Pg the gas pressure in the bubble [15]. The pressure of hydrogen inside the bubble (Pg) can be expressed by [16]:

Pg=Pm+2σ/r (1)

where Pm is the pressure at the bubble boundary, σ is the gas-liquid interfacial tension, and r is the bubble radius. The higher pressure in the bubble resulted in a higher hydrogen concentration compared with the concentration in the liquid Al. The concentration difference generated a driving force, leading to the dissolution of hydrogen bubble. On the contrary, the concentration of hydrogen in the liquid Al increased with temperature, leading to bubble growth. Therefore, the bubble behavior was determined by the competition between dissolution and growth of bubbles [17].

Fig. 2 (a) shows the plots of bubble radius versus time for the three different bubbles numbered 1, 2 and 3 in Fig. 1(d). The bubble size was well described by a Gaussian function, which indicated that individual bubbles grew in a stochastic way.

r=a+[(b/t0(π/2)1/2)exp-(2(t-tc)2/t0)] (2)

Fig. 2.   (a) Variations of bubble radius with time for the three distinguishable bubbles labeled in Fig. 1(d); (b) A schematic diagram of bubble evolution.

where a and b are constants, t0 and tc are deviation and mean of Gaussian time distribution function, respectively. The similar growth behavior of hydrogen bubbles was observed in Al-Ca alloy [17] and Al-Bi alloy [18]. The growth characteristics of a single hydrogen bubble can be described by a comprehensive mathematical model by the process of rectified diffusion [19]. A schematic diagram of bubble evolution was presented in Fig. 2(b). The individual hydrogen bubble growth can be determined by the equation relating to the radial concentration gradient [20].

(CH-CB)dr/dt=-DH(∂C/∂r)+PT (3)

where CH and CB are the hydrogen concentration at the surface of the shell and bubble, the first term on the right is the hydrogen diffusion flux into the bubble, r is the bubble radius and ∂C/∂r is the concentration gradient, PT is a temperature dependent hydrogen term representing dissolution into the melt. For the bubbles 1, 2 and 3, the growth behaviors were mainly ascribed to increased concentration (CH > CB), which resulted in the diffusion of hydrogen from liquid Al into these bubbles. The shell became thinner and the concentration gradient increased, accompanied by an increased bubble growth rate. Adversely, the shrinkage behavior mainly attributed to dissolution of bubble into liquid Al. The shell became thicker and the concentration gradient decreased, accompanied by a decreased bubble growth rate. As the temperature was higher than 750 °C, the γ-Al2O3 at the surface of the melt transformed into denser α-Al2O3, preventing any further increases of hydrogen concentration in the melt [21]. Subsequently, these hydrogen bubbles dissolved into the melt and disappeared. For the bubble 4, the abrupt increasing temperature caused to a rapid bubble growth. The growth rate was higher than the movement of gas-liquid interface, leading to a bubble breakage. For the bubble 5, the drag force, developing by the disturbance of local liquid phase, resulted in bubble shape changing from spherical to ellipsoid, because the bubble expanded against liquid force [22].

For groups of bubbles in superheated Al-Ca and Al-Bi alloy melt, the size distribution at different stages was found to be dominated by Gaussian or/and Lifshitz-Slyozov-Wagner (LSW) diffusion controlled distribution [17,23]. Fig. 3(a) and (b) show the scaled size distribution histograms of experimental bubble sizes at 44.5 s and 65 s, comparing with Gaussian distribution and LSW-diffusion controlled distribution fitted to the histograms. The abscissa represents scaled length ρ(=r/ $\bar{r}$), and the ordinate represents scaled frequency h(ρ)(=f/$\bar{f}$). The experimental data at 44.5 s was found to be well-fitted by the Gaussian distribution, as shown in Fig. 3(a). This attributed to the stochastic characteristics of bubble nucleation and growth in the early stage. However, the LSW-diffusion controlled distribution fitting, rather than Gaussian distribution, appeared to be satisfactory to the experimental data at 65 s (Fig. 3(b)). This was ascribed to the growth of discretely distributed bubbles controlled by Ostwald coarsening at the expense of smaller bubbles in the final stage.

Fig. 3.   (a, b) Scaled frequency of bubble size distribution at t + 44.5 s and t + 65 s. The comparison between experimental data fitted by Gaussian distribution and LSW-diffusion controlled distribution was also shown; (c) Dependence of cubic radius difference on the negative time difference for group bubbles at the final stage.

To further confirm the growth of bubbles following the LSW law, the relationship between bubble radius and time can be expressed by $\bar{r}$3-$\bar{r}^{3}_{0}$=K1t, where r0 is the average radius at t = 0, and K1 is the rate constant [24]. Based on this, the cubic radius difference with the negative time difference during a duration time of 7 s in the final stage was calculated. From Fig. 3(c), the linear relationship between $\bar{r}$3-$\bar{r}^{3}_{0}$ and time t had a good correlation with the experimental measurement, which demonstrated that the bubble evolution was dominated by LSW law during the final stage. This result was similar to the Al-Ca alloy instead of Al-Bi alloy, which attributed to the trap of Bi atoms at the surface of hydrogen bubbles [23].

3.2. Bubble evolution during solidification

The hydrogen bubble evolution in the liquid Al/solid Ni system during cooling is shown in Fig. 4. During the early stage of cooling, the IMCs first formed at the interface and grew into the melt, as shown in Fig. 4(a)-(d). The IMCs were identified as Al3Ni2 and Al3Ni phases through the microstructural characterization after solidification (Fig. 5(a), (b)). The Al3Ni phase grew until the intersection, where the Ni concentration of solidification front decreased, but the H concentration increased adversely. After 2060s, there were ellipsoidal hydrogen bubbles 1 and 2 observed at the intersection between two adjacent Al3Ni grains (Fig. 4(e)). The bubbles were entrapped by the growing Al3Ni front. The upper radius (r1) and bottom radius (r2) of the bubbles 1 and 2 varied with the increasing solidification time, as shown in Fig. 5(c). The upper ends grew along the liquid channel with smaller radius, with the bottom impinging on the Al3Ni phases (Fig. 4(f)). After a prolonged cooling, two spherical bubbles 3 and 4 formed at the triple junctions (intersection between liquid channel and crack), and grew along the liquid channel and crack, as shown in Fig. 4(g), (h). The morphology of the bubbles changed from sphere to be irregular. The radius of the growth front of the bubble trended to be smaller, which attributed to the widths of the liquid channel and crack.

Fig. 4.   Microstructural evolution during solidification in liquid Al/solid Ni system.

Fig. 5.   (a, b) Morphologies of the Al/Ni interconnection after solidification: the subimage shows the porosity in the liquid channel; (c) the varied upper (r1) and bottom radius (r2) of the bubbles 1 and 2 with increasing solidification time; (d), (e) Schematic model of the different hydrogen bubbles.

The porosities developing from the hydrogen bubbles were illustrated in Fig. 5(a)-(b). During solidification, the atomic hydrogen was rejected from the precipitated Al3Ni phase to the remaining liquid phase. As the hydrogen concentration in the melt was supersaturated, hydrogen bubbles formed and grew in the melt. The bubble growth during solidification can be described by the continuum governing equation [25]:

RH=∂[(Cllfl+kρsfs))]/∂t+∇j (4)

where RH is the consumption of hydrogen bubble per unit volume, the first term on the right is the hydrogen mass in the liquid and solid mixture, j is the flux of hydrogen. The hydrogen concentration in the remaining melt increased with temperature decreasing, the RH term promoted the bubble growth. However, the growth was constrained by the formed Al3Ni phase. A schematic diagram of bubble growth was shown in Fig. 5(d), (e), where ΔF indicated the obstruction from the Al3Ni phase, the black solid arrows and dotted lines represented bubble growth direction and growth front shape respectively. For the bubbles at the liquid channel, they were impinged on the formed Al3Ni phase under three directions, and their shapes changed under the increased pressure inside bubbles. The increased pressure and unbalanced force (ΔF) from three directions could result in the motion of the bubble [26]. However, the bubble had no motion, indicating an additional force (F) existing from the upper side.

F=∂P/∂x≈2σ(1/r1-1/r2)/l (5)

where r1 is the radius on the upper side, r2 on the bottom, l is the bubble length. To balance the bubble, r1 should be smaller than r2, indicating the bubble growth front with smaller radius, as shown in Fig. 5(d). For the bubbles at the triple junctions, the directions of the unbalanced forces were different from the ones at the liquid channel, resulting in the growth in three directions along the cracks and liquid channel, as shown in Fig. 5(e). The radius of the growth front of bubbles were limited by the liquid channels and cracks. Consequently, the bubbles had two different growth characteristics during solidification.

4. Conclusion

In this work, the hydrogen bubble evolution in the liquid Al/solid Ni interconnection during heating and solidification were studied. During heating, the individual bubble behaved in three different ways. The size distribution for groups of bubbles was dominated by Gaussian controlled distribution in the early stage, but Lifshitz-Slyozov-Wagner (LSW) diffusion at the final stage. During solidification, the IMCs first formed at the interface, followed by the formation of hydrogen bubbles. For the bubbles at the liquid channel, the bottom of bubbles were impeded by the Al3Ni phase with the upper radius being smaller during growth. The bubbles at the triple junctions grew along the liquid channel and the cracks, accompanied by a morphological transition.

Acknowledgements

This work is supported by the National Key Research and Development Program (2017YFA0403800), the National Natural Science Foundation of China (51374144, 51727802), the Shanghai Municipal Natural Science Foundation (13ZR1420600) and Shanghai Rising-Star Program (14QA1402300). The support of synchrotron radiation phase-contrast imaging by the BL13W1 beam line of Shanghai Synchrotron Radiation Facility (SSRF), China, is gratefully acknowledged.

The authors have declared that no competing interests exist.


/