Journal Information
Vol. 8. Issue 5.
Pages 4141-4150 (September - October 2019)
Share
Share
Download PDF
More article options
Visits
3
Vol. 8. Issue 5.
Pages 4141-4150 (September - October 2019)
Original Article
DOI: 10.1016/j.jmrt.2019.07.023
Open Access
The Zn accumulation behavior, phase evolution and void formation in Sn-xZn/Cu systems by considering trace Zn: a combined experimental and theoretical study
Visits
3
Jieshi Chena,b,d,
Corresponding author
cjshbb@sjtu.edu.cn

Corresponding authors.
, Hongkui Zhanga, Peilei Zhanga,b, Zhishui Yua,b, Yongzhi Zhangc, Chun Yud, Hao Lud,
Corresponding author
shweld@sjtu.edu.cn

Corresponding authors.
a School of Materials Engineering, Shanghai University of Engineering Science, Shanghai 201602, PR China
b Shanghai Collaborative Innovation Center of Laser Advanced Manufacturing Technology, Shanghai, 201620, PR China
c AECC Commercial Aircraft Engine manufacturing CO., TLD, Shanghai 200241, PR China
d School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai, 200240, PR China
This item has received
3
Visits

Under a Creative Commons license
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Figures (6)
Show moreShow less
Tables (3)
Table 1. EDS results of IMCs phase at the interface of Sn-xZn/EP Cu (x = 0.2, 0.5 and 0.8 wt.%) joints following different thermal aging times.
Table 2. IMCs evolution in following different thermal aging times.
Table 3. Structural properties obtained by GGA calculations, in comparison with other theoretical and experimental works. Lattice parameters a (Å), b(Å), c (Å); formation enthalpy Hf (kJ · mol−1) and cohesive energy (ΔEC). Information in square brackets shows the final setting values for k-point meshes and their corresponding k-point numbers.
Show moreShow less
Abstract

The strong effects of Zn addition on the reaction in Sn/Cu systems are widely known. Nevertheless, the micro-mechanism for Zn accumulation behavior, phase evolution and void formation in Sn-xZn/Cu systems remain unclear. In this work, the effect of Zn addition on the interface behavior is investigated. Structures of Sn-xZn/Cu are prepared by the reaction between Sn-xZn solders (x = 0, 0.2, 0.5 and 0.8 wt.%) and the electroplated Cu (EP Cu) joints. During the thermal aging of Sn/EP Cu joints, many voids formed inside the Cu3Sn phase, especially close to the Cu3Sn/EP Cu interface. Whereas the voids were greatly suppressed with Zn addition. During the thermal aging, Zn gradually accumulated near the interface of IMCs/EP Cu. As thermal aging time to168 h and Zn content to 0.8 wt.%, Zn participated in the interface reaction, three layers with different contrasts, (Cu, Zn)6Sn5, Cu6(Sn, Zn)5 and (Cu,Sn)3Zn were formed. Theoretical analysis shows that Zn atom accumulated near the interface of IMCs/EP Cu due to lower diffusion energy barriers (Edf) of Zn than Cu atom. The preferential occupation of Zn atoms in Cu6(Zn,Sn)5 and (Cu,Zn)6Sn5 are Sn3 (4e) and Cu3 (4a) site because of the lowest lattice strain. The diffusion path interstice size for Cu atom via the IMCs layer gradually shrink with the increment of charge transfer in Zn-rich layers, corresponding the Edf of Cu atom gradually rise. As a result, the fluxes of Cu and Sn via the IMCs layer were balanced out, which reduced the Kirkendall voids formation.

Keywords:
Sn/Cu solder joint
Zinc
Accumulation behavior
Phase evolution
Kirkendall void
Full Text
1Introduction

With the trends of microelectronic products from 2-dimensional integration of circuits (2D IC) to 3-dimensional integration of circuits (3D IC), the size of solder joints for interconnections becomes smaller and smaller. The interfacial microstructure plays a more and more important role in the reliability of joints [1]. During the thermal aging of Sn-based solders/Cu joints, the Cu-Sn intermetallic compounds (IMCs) are usually formed at the interface [2,3]. The large IMCs layer can degrade the reliability of solder joints because of the inherent brittleness of IMCs [4–6]. Meanwhile, a large number of Kirkendall voids (KVs) are often observed at Cu3Sn/Cu interface, which has detrimental effects on the reliability of joints [7,8]. Therefore, interfacial behaviors of IMCs growth, phase evolution and void formation in Sn/Cu systems have attracted lots of attentions [1–8].

In Sn-Cu system, when the molten Sn-based solder alloy wets the Cu, Cu6Sn5 and Cu3Sn IMCs form at the solder/Cu interface. KVs are frequently observed near the Cu3Sn/Cu interface and/or in Cu3Sn [9,10]. Essentially, the mechanism of KVs formation in the Cu/Sn interface is basically caused by unbalanced diffusion fluxes (Cu flux > Sn flux), that is to say, the Kirkendall effect [11]. Since KVs are formed due to unbalanced diffusion fluxes, tiny amounts of additive elements will affect KVs formation and the growth of IMCs. It was reported that adding minor alloying elements into the solder could suppress the formation of voids and retard the growth of Cu3Sn layer [3,8,12,13]. Among the numerous Sn-based alloys proposed to replace the traditional Sn-Pb solder, Zn is one of the most promising alloy elements [3,4,14]. Microalloying with Zn is found to effectively reduce the growth of Cu3Sn layer and inhibit the formation of KVs [15,16]. Two assumptions emphasizing that Zn level controls the Kirkendall voiding level were proposed. Fukumoto et al. reported that the Zn segregation not only inhibited diffusion of Cu due to the effect of solute drag, which delayed growth of the Cu3Sn layer, but also balanced out the fluxes of Cu and Sn, which reduced the KVs formation [7]. Kotadia et al. confirmed that Zn atoms can dissolve into the IMCs layers and substitute part of component atoms in the IMCs crystals [15,17,18]. The partial substitution of Zn could increase the lattice strain in the IMCs layers, reducing the diffusion rate of vacancy and accordingly suppressing the formation of voids [11,19]. However, the exact micro-mechanism of how Zn addition affects the fluxes of Cu and Sn atom diffusion in solder/Cu interface is still not clear. Especially, how the partial substitution of Zn could increase the lattice strain in the IMC layers, has not been reported. Moreover, When Zn was added into the solder, Zn was found to accumulate near the interface of Cu3Sn /Cu after the aging test [7]. The reason needs more detailed explanations.

Thus, in order to better understand the role of Zn in the interface behaviors in Sn-based solders/Cu joints. A systematical and comparative study was performed to characterize the Zn accumulation behavior, phase evolution and void formation between Sn-xZn solders (x = 0, 0.2, 0.5 and 0.8 wt.%) and the electroplated Cu (EP Cu) joints. Meanwhile, the IMCs layer models considering Zn distribution were built. The formation enthalpy (ΔHf), cohesive energy ( ΔEC) and lattice volume (V) varying with the substitutional site of Zn atom in Cu6Sn5-based structure, the minimum-energy path (MEP) and diffusion activation energy of Cu and Zn fluxes were calculated. Also, relative electronic structures (electron density difference, charge transfer and lattice strain, and so on) of the IMCs layer models were analyzed using density functional theory. Based on the experimental and theoretical results, the effects of Zn addition on interfacial behaviors in Sn/Cu systems were revealed. The study is organized as follows: in section 2, we present the experimental and theoretical methods. In section 3, we provide the experimental results, the underlying mechanism that Zn accumulation behavior, phase evolution and void formation in Sn-based solders/Cu joints.

2Experimental and theoretical methods2.1Experimental procedures

The solders used in this work were Sn-xZn (x = 0, 0.2, 0.5 and 0.8 wt.%), which were fabricated from pure Sn (99.99%) and Zn (99.99%). The substrate was high purity Cu foil with a layer of Cu film (10 μm in thickness) electroplated on the surface. The electroplating solution contained CuSO4, H2SO4, HCl and PEG (polyethylene glycol). The current density used during the electroplating process was 1.7 mA/cm2[20]. The solder joints were prepared by melting Sn on the foils at 260 ℃ for 1 min. To investigate the interfacial microstructure in the solder joints, the isothermal aging for the as-reflowed samples was performed at 180 ℃ with different aging time. Then these samples were mounted in epoxy and metallurgically polished. The cross-sectional microstructure at the interface was observed by SEM (Sirion200). The compositions of IMC layers were determined by energy dispersive spectroscopy (EDS). The average thickness of IMCs layer was measured with the software Image J. The average thickness of the IMC layer was calculated by dividing the IMC layer area by the length of the interface.

2.2Calculation details

The calculations were carried out by using the CASTEP plane-wave code [21] in the scheme of generalized gradient approximation (GGA-PBE) [22]. The Vanderbilt ultrasoft pseudopotentials [23] were employed to treat the valence electrons for Cu (3d104s1), Sn (5s25p2) and Zn (3d104s2). Brillouin-zone integrations were performed using Monkhorst and Pack k-point meshes [24]. The settings for these calculations are shown in Table 3. The energy cutoff was 400 eV. The convergence tolerance was selected: minimum energy less than 1.0 × 10−5 eV/atom, maximum force less than 0.03 eV · Å−1, maximum stress less than 0.05 GPa, and maximum displacement less than 1 × 10−3 Å.

The transition state (TS) was obtained through calculation of the minimum-energy path (MEP) using the nudged elastic band (NEB) method in Dmol3 [25]. The interface phase models considering Zn distribution were constructed for atom diffusion. Also, the spin polarization was considered in the calculations. The charge distributions and electronic density differences of the systems were evaluated by Mulliken charge analysis and electron density deformation analysis, which were performed using a linear combination of atomic orbitals (LCAO) basis [26,27].

3Results and discussion3.1Interfacial microstructures evolution of Sn-xZn/EP Cu joints

The interfacial microstructure of Sn-xZn/EP Cu (x = 0, 0.2, 0.5 or 0.8 wt.%) joints with different aging time are shown in Figs. 1 and 2. In all of the images, the upper, bottom and middle materials are solder alloy, copper and interfacial IMC layer, respectively. The chemical compositions of the IMCs layers have been characterized by EDS analysis and the results are listed in Table 1.

Fig. 1.

Cross-sectional BSE micrographs of Sn/EP Cu joints after aging at 180 °C. (a) 0 h; (b) 24 h and (c) 72 h.

(0.26MB).
Fig. 2.

Microstructure evolution of Sn-xZn/EP Cu (x = 0.2, 0.5 and 0.8 wt.%) joints following different thermal aging times, (a) 0 h; (b) 72 h; (c)168 h.

(1.28MB).
Table 1.

EDS results of IMCs phase at the interface of Sn-xZn/EP Cu (x = 0.2, 0.5 and 0.8 wt.%) joints following different thermal aging times.

Layer  Cu (at. %)  Sn (at. %)  Zn (at. %)  Phase  Layer  Cu (at. %)  Sn (at. %)  Zn (at. %)  Phase 
A#  59.82  40.18  Cu6Sn5  I#  73.28  26.62  Cu3Sn 
B#  59.91  40.09  Cu6Sn5  J#  57.34  42.56  Cu6Sn5 
C#  58.33  41.67  Cu6Sn5  K#  75.18  24.82  Cu3Sn 
D#  58.87  41.13  Cu6Sn5  L#  57.76  42.24  Cu6Sn5 
E#  74.50  25.50  Cu3Sn  M#  73.24  25.44  1.32  Cu3Sn 
F#  57.92  42.08  Cu6Sn5  N#  33.15  46.58  20.27  (Cu, Zn)6Sn5 
G#  75.23  24.77  Cu3Sn  O#  54.62  40.12  5.26  Cu6(Sn, Zn)5 
H#  59.11  40.89  Cu6Sn5  P#  61.70  12.98  25.32  (Cu,Sn)3Zn 

Fig. 1a shows a cross-sectional back scattered electron (BSE) images of the Sn/EP Cu joints. At this scale, only a layer of scallop-like Cu6Sn5 formed at the interface after reflow soldering, and no void at Sn/EP Cu interface was observed. After a 24 h aging at 180 °C, a Cu3Sn layer was observed between the Cu6Sn5 and Cu layers, and several voids were found at the Cu3Sn/EP Cu interface, as shown in Fig. 1b. When the aging time was extended to 72 h, as seen in Fig. 1c, the IMCs layer thickness gradually increased in Sn/EP Cu joint, and Cu3Sn layer grew much quicker than the Cu6Sn5 layer. The morphology of IMCs became flattened, because Cu diffusion through the scallop valleys is faster than that through the scallop hills [28]. In addition, a great number of voids with considerable size can be observed at the Cu3Sn/ EP Cu interface.

Fig. 2a1–a3 show the interfacial microstructures of as-reflowed Sn-xZn/EP Cu (x = 0.2, 0.5 and 0.8 wt.%) joints. Similar to the as-reflowed Sn/EP Cu joint, a thin layer of Cu6Sn5 was formed at the interface, and no void was observed in all samples.

After a 72 h aging at 180 °C, as shown in Fig. 2b1–b3, the IMC layer thickness increased obviously. A thin Cu3Sn layer emerged between Cu6Sn5 and EP Cu. The thickness of Cu3Sn layer was found to be sensitive to the Zn content. With Zn content was increased from 0.2 to 0.8 wt.%, the thickness of Cu3Sn layer became thinner. In addition, a few voids can be found at Cu3Sn/EP Cu interface in Sn-0.2Zn/EP Cu joint, while, as the Zn content in solder was increased up to 0.5 wt.% or 0.8 wt.%, the voids disappeared from the interface. Moreover, comparing Figs. 1 and 2, it was easily observed that the Zn addition could suppress the formation of voids and the growth of Cu3Sn layer.

When the aging time was extended to 168 h, as shown in Fig. 2c1–c3. The total IMC thickness in the Sn-0.2Zn/EP Cu interface did almost not increase, but the Cu3Sn layer became discontinuous and thinner. For Sn-0.5Zn/EP Cu joint, similar to Sn-0.2Zn/EP Cu, the total IMC thickness had no significant increase, and no void was observed in Sn-0.5Zn/EP Cu interface. For the Sn-0.8Zn/EP Cu joint, IMC thickness was very different from that in Sn-0.2Zn/EP Cu and Sn-0.5Zn/EP Cu, it was significantly increased. Moreover, three layers with different contrasts were observed in the Sn-0.8Zn/EP Cu interface, a new dark-gray layer close to the Sn-0.8Zn alloy was found, it had a unique composition of Cu-20.27Zn-46.58Sn (in at.%), and it should be the (Cu, Zn)6Sn5 phase with Zn solubility of 20.27 at.%. Compositions of the other two layers from the Cu side to Sn-0.8Zn alloy were Cu-12.98Sn-25.32Zn (in at. %) and Cu-40.12Sn-5.26Zn (in at. %), corresponding they should be (Cu, Sn)3Zn and Cu6(Sn, Zn)5 phase, respectively.

3.2IMCs evolution of Sn-xZn/EP Cu joints

The evolution of IMCs at Sn-xZn/EP Cu (x = 0, 0.2, 0.5 and 0.8 wt.%) joints by considering Zn content and aging condition was summarized in Table 2. In the Sn-xZn/EP Cu (x = 0, 0.2 and 0.5 wt.%) joints, the Cu6Sn5 and Cu3Sn phases were formed at the interface. As the aging time was extended to 168 h, the (Cu, Zn)6Sn5, Cu6(Sn, Zn)5 and (Cu, Sn)3Zn phases were formed at the Sn-0.8Zn/EP Cu joint interface (Fig. 2c3). These results revealed that, with the Zn content and thermal ageing time increased, the Zn atoms were incorporated into the Cu and Sn sublattice of the Cu-Sn IMCs.

Table 2.

IMCs evolution in following different thermal aging times.

Sn–Zn alloyAs reflowing at 260 ℃Aging time at 180 ℃
24 h  72 h  168 h 
SnCu6Sn5Cu6Sn5  Cu6Sn5  Cu6Sn5 
Cu3Sn  Cu3Sn  Cu3Sn 
Sn-0.2ZnCu6Sn5Cu6Sn5  Cu6Sn5  Cu6Sn5 
Cu3Sn  Cu3Sn  Cu3Sn 
Sn-0.5ZnCu6Sn5Cu6Sn5  Cu6Sn5  Cu6Sn5 
Cu3Sn  Cu3Sn  Cu3Sn 
Sn-0.8ZnCu6Sn5Cu6Sn5  Cu6Sn5  (Cu, Zn)6Sn5 
Cu3Sn  Cu3Sn  Cu6(Sn, Zn)5 
    (Cu, Sn)3Zn 

To understand the phase evolution of η′-Cu6Sn5-based structures with Zn substitution. The optimized lattice parameters, lattice volumes (V), formation enthalpy (ΔHf) and cohesive energy( ΔEC) for η′-Cu6Sn5-based structures with Zn substitution were calculated, and the results are listed in Table 3. It can be seen that the obtained lattice parameters and cell volumes for η′-Cu6Sn5 are in good agreement with the available experimental [29] and theoretical values [29,30], where the deviations are less than 5%. For pure Cu, the face-centered cubic (fcc), and the space group (FM-3 M) were chose, and the obtained lattice parameter and cell volume also agree well with experimental results [31]. With respect to Zn has a hexagonal structure, and Sn has a tetragonal structure (β-Sn). The calculated bond length are 2.912 Å and 2.320 Å, respectively, which are close to the experimental results [32,33]. Therefore, the above results verified the reliability of this calculation method. Moreover, to gain deep insight into the phase stability of η′-Cu6Sn5-based structures with Zn substitution. The formation enthalpy (ΔHf) and cohesive energy ( ΔEC) of a compound were calculated. The ΔHf and ΔEC of CumZnnSnl (n, m, and l are the number of atoms) structure can be expressed as follows:

where ET is the total energy of CumZnnSnl, and ECusolid, EZnsolid and ESnsolid are the total energy per atom of Cu, Zn, and Sn crystals, ECuatom, EZnatom and ESnatom are the energies of a single Cu, Zn and Sn atom, respectively. The calculated results are summarized in Table 3 and Fig. 3(a). The ΔHf of CumZnnSnl are almost negative values, indicating that the Cu-Zn-Sn ternary structures are thermodynamic stable and can be synthesized. The substitutional sites of Zn atom at Cu4 (4e), Cu1 (8f) + Cu2 (8f), Cu3 (4a) + Cu4 (4e), Cu3 (4a) + Cu1 (8f) and Cu3 (4a) + Cu2 (8f) are less stability than that at Cu3 (4a), this indicates that the substitutional site of a Zn atom at Cu3 (4a) site is the most stable structures (e.g. Cu5ZnSn5, −6.35 kJ/mol). Among the Cu6(Sn, Zn)5 structures, the substitutional sites of Zn atom at Sn3 (4e) is preferential occupation (e.g. Cu6ZnSn4, −5.57 kJ/mol). Above theoretical results well agree with the experiment phenomenon. Lattice volume (V) varying with the substitutional site of Zn atom in Cu6Sn5-based structure are shown in Fig. 3(b), the results indicate that the substitutional site of Zn atom at Cu site, the lattice volume expansion of Cu6Sn5 with the increase of the Zn content. Whereas, the substitutional site of Zn atom at Sn site, the lattice volume shrinkage of Cu6Sn5 with the increase of the Zn content. Moreover, the results also show that the preferential occupation of Zn in (Cu,Zn)6Sn5 and Cu6(Zn,Sn)5 structure are Cu3 (4a) and Sn3 (4e) site, because of the preferential occupation leads to the influence on lattice strain is least, i.e. the lowest lattice strain energy.

Table 3.

Structural properties obtained by GGA calculations, in comparison with other theoretical and experimental works. Lattice parameters a (Å), b(Å), c (Å); formation enthalpy Hf (kJ · mol−1) and cohesive energy (ΔEC). Information in square brackets shows the final setting values for k-point meshes and their corresponding k-point numbers.

Phase  Substitutional site  Method  Lattice parameters  V(Å3ΔHf (kJ/mol)  ΔEC (eV/atom) 
η′-Cu6Sn5  –  GGA[6 × 6 × 6]  a = 11.054 Å, b = 7.435 Å  808.87  −6.65  0.167 
    Exp.by others  c = 9.965 Å, β = 99.04º  779.37a  −7.35a 
      a = 11.022 Å, b = 7.282Åa    −7.13b   
      c = 9.827 Å, β = 98.84º       
Cu4Zn2Sn5  Cu1 (8fGGA[6 × 6 × 6]  a = 11.159 Å, b = 7.540 Å  856.42  −5.08  0.147 
      c = 10.254 Å, β= 96.97°    –   
Cu4Zn2Sn5  Cu2 (8fGGA[6 × 6 × 6]  a =11.124 Å, b =7.521 Å  854.87  −4.97  0.145 
      c = 10.29 Å, β= 96.87°       
Cu5ZnSn5  Cu3 (4aGGA[6 × 6×6]  a = 11.104 Å, b = 7.490 Å  829.34  −6.35  0.157 
      c = 10.074 Å, β= 98.22°Å       
Cu5ZnSn5  Cu4 (4eGGA[6 × 6 × 6]  a = 11.107 Å, b = 7.483 Å  830.51  −4.36  0.143 
      c = 10.112 Å, β= 98.85°       
Cu2Zn4Sn5  Cu1 (8f)  + Cu2 (8fGGA[6 × 6×6]  a = 11.450 Å, b = 7.708 Å  902.04  −3.38  0.127 
      c = 10.315 Å, β= 97.81°       
Cu4Zn2Sn5  Cu3 (4a)  + Cu4 (4eGGA[6 × 6 × 6]  a = 11.157 Å, b = 7.538 Å  852.47  −4.81  0.146 
      c = 10.221 Å, β= 98.37°       
Cu3Zn3Sn5  Cu3 (4a)  + Cu1 (8fGGA[6 × 6 × 6]  a = 11.278 Å, b = 7.645 Å  876.23  −4.92  0.146 
      c = 10.226 Å, β= 96.41°       
Cu3Zn3Sn5  Cu3 (4a)  + Cu2 (8fGGA[6 × 6 × 6]  a = 11.345 Å, b = 7.610 Å  878.89  −4.30  0.143 
      c = 10.267 Å, β= 97.47°       
Cu6Zn2Sn3  Sn1 (8fGGA[6 × 6 × 6]  a = 11.031 Å, b = 7.078 Å  743.58  −3.62  0.134 
      c = 9.667 Å, β= 99.93°       
Cu6Zn2Sn3  Sn2 (8fGGA[6 × 6 × 6]  a = 10.985 Å, b = 7.089 Å  742.89  −1.48  0.105 
      c = 9.66 Å, β= 99.29°       
Cu6ZnSn4  Sn3 (4eGGA[6 × 6 × 6]  a = 10.979 Å, b = 7.234 Å  773.01  −5.57  0.156 
      c = 9.856 Å, β= 99.12°       
Cu6Zn4Sn1  Sn1(8f)  + Sn2 (8fGGA[6 × 6 × 6]  a = 10.756 Å, b = 7.206 Å  681.18  1.46  0.083 
      c = 8.917 Å, β= 99.77°       
Cu6Zn3Sn2  Sn1 (8f)  + Sn3 (4eGGA[6 × 6 × 6]  a = 10.842 Å, b = 7.061 Å  714.55  −2.09  0.123 
      c = 9.497 Å, β= 100.62°       
Cu6Zn3Sn2  Sn2 (8f)  + Sn3 (4eGGA[6 × 6 × 6]  a = 10.867 Å, b = 7.007 Å  711.16  −0.37  0.109 
      c = 9.446 Å, β= 98.69°       
Cu  –  GGA[8 × 8 × 8]  a = = b =c = 3.630 Å  47.85  Bond length   
    Exp.by others  a = b =c = 3.615 Åc  47.24c     
Zn  –  GGA[11 × 11 × 6]  a = b = 2.709 Å, c = 4.795 Å  30.48  2.912   
    Exp.by others  a = b = 2.665 Å, c = 4.945 Åd  30.40d  2.863d   
β-Sn  –  GGA[6 × 6 × 5]  a = b = 5.742 Å, c = 3.201 Å  105.53  2.320   
    Exp.by others  a = b = 5.832 Å, c = 3.182 Åe  108.22e  2.211e   
a

Ref. [29] ab initio calculations and experimental results.

b

Ref. [30] first-principles DFT calculations.

c

Ref. [31].

d

Ref. [32] experimental data at 40 K to 500k.

e

Ref. [33].

Fig. 3.

(a) Formation enthalpy (ΔHf), cohesive energy (ΔEC) and (b) lattice volume (V), lattice strain varying with the substitutional site of Zn atom in Cu6Sn5-based structure.

(0.62MB).
3.3Zn accumulation behavior of Sn-xZn/EP Cu joints

The composition distributions at the interface in different kinds of joints were analyzed by using the EDS technique (line scan mode), as shown in Fig. 4. The date reveal that the distribution of Zn is relatively homogeneous at the Sn-0.2Zn/EP Cu interface, except for a few Zn accumulate at Cu3Sn/EP Cu interface. In the Sn-0.5Zn/EP Cu joint, the accumulation of Zn at Cu3Sn/EP Cu interface significantly increased, and it further increased at the Cu3Sn/EP Cu interface of Sn-0.8Zn/EP Cu joint. In the meantime, there was a sharp rise of Zn content in the solder/Cu6(Sn,Zn)5 interface, i.e. the (Cu, Zn)6Sn5 layer. Fukumoto et al. investigated the effects of Zinc addition on void formation in solid–liquid interdiffusion bonding of copper. They also found that the Zn was segregated near the interface of Cu/Cu3Sn and the grain boundaries of Cu3Sn [7]. In addition, Kao et al. studied the effects of Zn additions on Cu-Sn microjoints. They found that there is a peak Zn concentration at the (Cu,Zn)6Sn5/Cu interface. The maximum Zn concentration is 4.5 at.%. After aging at 150 °C for 500 h, this Zn concentration peak remains and even becomes higher, reaching nearly 6 at.% [14]. In our work, the Zn atoms in the solder first inclined to accumulate in Cu3Sn/EP Cu interface. With increasing aging time and Zn content, Zn participated in the interface reaction, the Zn atoms were incorporated into the Cu and Sn sublattice of the Cu-Sn IMCs, corresponding new interface layers (i.e. (Cu, Zn)6Sn5, Cu6(Sn, Zn)5 and (Cu, Sn)3Zn) were formed in Sn-0.8Zn/EP Cu joint.

Fig. 4.

Concentration profile of Sn-xZn/EP Cu joints after aging at 180 °C for168 h, (a) Sn-0.2Zn/EP Cu, (b) Sn-0.5Zn/EP Cu, (c) Sn-0.8Zn/EP Cu, (d) Zn concentration profile of Sn-xZn/EP Cu (x = 0.2, 0.5, 0.8 wt.%) joints. Precise locations of the line scans are indicated by the red arrows marked in the micrographs.

(0.87MB).

The optimal energy path (MEP) for the Zn or Cu atom diffusion along the Cu6Sn5 layer are searched based on the NEB method, and the population analysis, bond length and the energy barriers (Edf) are shown in Fig. 5. It is seen that the population analysis are reduced from Zn (0.133e) to Cu (−0.091e), which indicates that the charge transfer between M (=Zn or Cu) atom and neighboring Sn and Cu atoms decreases from Zn to Cu. It is well known that the binding strengths of the Cu–Sn bond around the diffusion interstice are decided by the number of electrons, which becomes weak with the electron loss. It is also found that the Sn and Cu atoms around the diffusion interstice lose less electrons from Zn to Cu, which illustrates that the Cu–Sn bonds around the interstice will be more difficult to deform from Zn to Cu (see change of bond length of Cu–Sn bonds). As a result, there will be increasing difficulty for M atoms to penetrate from Zn to Cu (the Edf are 0.842 eV for Cu and 0.487 eV for Zn). Therefore, when Zn was added into the solder, Zn atom was the preferential diffusion species in the Sn-xZn/Cu interface than that of Cu atom, the Zn atom accumulated near the interface of IMCs/EP Cu.

Fig. 5.

The optimal energy path (MEP), population analysis, bond length and the energy barriers (Edf) for the Zn or Cu atom diffusion along the Cu6Sn5 layer.

(0.26MB).
3.4The micro-mechanism for retardation of void growth

The probable retardation mechanism of Zn addition on thermal aging induced void formation is proposed as follows. It is generally believed that the Kirkendall voiding is induced by an unbalanced diffusion [11]. For Sn/Cu system, the unbalanced diffusion at the interface can be expressed as:

where JCu and JSn are the diffusion fluxes of Cu and Sn, respectively. JV is the diffusion flux of vacancies. When the joints with Zn addition, Zn atoms can serve as a diffusing species and participate in the interfacial reaction. The interdiffusion at the Sn-xZn/EP Cu interface is given by

From 3.3 section analysis, Zn atoms in solder can pass across the IMCs layer and fill up vacancy sites before they grow into voids. In addition, the diffusion rate of vacancy (Jv) also can be reduced by the lattice deformation in the IMCs layer due to the Zn addition in solder. Schematic diagram of the micro-mechanism for retardation of void growth is shown in Fig. 6. The IMCs models considering Zn distribution are built as shown in Fig. 6(a). The charge transfer of IMCs models considering Zn distribution are shown in Fig. 6(b). It is easy to found that the electron density difference distribution curve of IMCs layer after Zn doping greater than that of Cu6Sn5 layer, that means more charge transfer of IMCs layer when Zn atoms are incorporated into the Cu and Sn sublattice of the Cu-Sn IMCs. The results can also be visualized in the images of electron density difference (Fig. 6(b)). It is seen that the yellow regions are increased from Cu6Sn5 layer to Zn-rich layer, which indicates that the charge transfer between Zn atom and neighboring Cu atoms increases gradually from Cu6Sn5 layer to Zn-rich layer. In other words, there exists a stronger affinity between Cu and Zn in the case of Zn addition into the Cu–Sn system. Feng et al. also pointed out that the orbital interaction of Cu–Zn is much stronger than those of Cu–Sn, Cu–Cu, and Sn–Zn [34]. The lattice volume evolution of IMCs layers from Cu6Sn5 layer to Zn-rich layer are shown in Fig. 6(c). The lattice volume are 808.87 Å3 for Cu6Sn5, 773.01 Å3 for Cu6(Sn,Zn)5 and 714.55 Å3 for Zn-rich layer, respectively. It is easy to obtain the fact that the lattice volume shrinkage of IMCs layers because of stronger affinity between Cu and Zn from Cu6Sn5 layer to Zn-rich layer. The optimal diffusion path interstice size for Cu atom diffusion along the IMCs layers are analyzed in Fig. 6(d). For the Cu6Sn5 layer, the bond length of Cu1-Sn1, Sn1-Cu2, Cu2-Sn2 and Sn2-Cu1 are 2.724 Å, 2.793 Å, 2.909 Å and 2.754 Å, respectively. For Cu6(Sn,Zn)5 layer, the bond length of Cu1-Sn1, Sn1-Cu2, Cu2-Zn1 and Zn1-Cu1 are 2.663 Å, 2.733 Å, 2.865 Å and 2.614 Å, respectively. For Zn-rich layer, the bond length of Cu1-Sn1, Sn1-Cu2, Cu2-Zn1 and Zn1-Cu1 are 2.534 Å, 2.688 Å, 2.766 Å and 2.563 Å, respectively. The results of the data show that the diffusion path interstice size for Cu atom in IMCs layers will shrink from Cu6Sn5 layer to Zn-rich layer. The optimal diffusion activation energy (Edf) fitting curve for the Cu atom diffusion along IMCs layers with different Zn content are shown in Fig. 6(e). The Edf increases dramatically when the Zn content increases in IMCs layers. That is, the JCu will decrease with the increase in Zn content. According to Eq. (4), the fluxes of Cu and Sn via the IMCs layers were balanced out. The formation of voids was suppressed.

Fig. 6.

Schematic diagram of the micro-mechanism for retardation of void growth. (a) Interface model of Cu6Sn5, Cu6(Sn,Zn)5 and Zn-rich layer. (b) Electron density difference of Cu6Sn5, Cu6(Sn,Zn)5 and Zn-rich layer. The yellow and green regions show regions of electron accumulation and loss. (c) Lattice volume of Cu6Sn5, Cu6(Sn,Zn)5 and Zn-rich layer. (d) Diffusion path interstice size of Cu6Sn5, Cu6(Sn,Zn)5 and Zn-rich layer. (e) Relation between diffusion activation energy for Cu atom along CumZnnSnl layers vs. Zn content x ( x=n/(m+n+l)).

(0.72MB).

The micro-mechanism for Zn addition inhibiting the diffusion of Cu atom (Edf from 0.842 to 1.678 eV) via the interface was due to more charge transfer (exists a stronger affinity between Cu and Zn in the case of Zn addition into the Cu–Sn system) of IMC layer after Zn addition, resulting in that the interstice path for Cu atom diffusion would be shrinked. Thus, Zn is supposed as an effective alloying element in Pb-free solders to retard the Kirkendall voids formation.

4Conclusions

In this study, The Zn accumulation behavior, phase evolution and void formation in Sn-xZn/Cu systems by considering trace Zn have been investigated with experimental and theoretical methods. All the results are summarized as follows:

  • (1)

    For Sn/EP Cu joints, many voids formed in the Cu3Sn phase, especially close to the Cu3Sn/EP Cu interface. Whereas the voids were greatly suppressed with trace Zn addition.

  • (2)

    Zn accumulated near the interface of IMCs/ EP Cu due to lower Edf of Zn atom via Cu6Sn5 layer than Cu atom.

  • (3)

    With the thermal aging to168 h and Zn content to 0.8 wt.%, Zn participated in the interface reaction, three layers with different contrasts, (Cu, Zn)6Sn5, Cu6(Sn, Zn)5 and (Cu, Zn)3Sn were formed in interface joints. The preferential occupation of Zn in Cu6(Zn,Sn)5 and (Cu,Zn)6Sn5 were Sn3 (4e) and Cu3 (4a) sites, respectively.

  • (4)

    As Zn substituting Cu atoms, the lattice volume expansion of Cu6Sn5 increased with the increase of the Zn content. Whereas, As Zn substituting Sn atoms, the lattice volume of Cu6Sn5 shrinked with the increase of the Zn content.

  • (5)

    The unbalanced diffusion of Cu and Sn fluxes could be greatly reduced by the addition of Zn into the solder. The formation of voids was suppressed. The micro-mechanism for Zn addition inhibiting the diffusion of Cu atom (Edf from 0.842 to 1.678 eV) via the interface was due to more charge transfer (exists a stronger affinity between Cu and Zn in the case of Zn addition into the Cu–Sn system) of IMC layer after Zn addition, resulting in that the interstice path for Cu atom diffusion would be shrinked.

Acknowledgement

This project is supported by National Natural Science Foundation of China (Grant no. 51805316), China Postdoctoral Science Foundation (No. 2019M651491) and Shanghai Science Technology Development Funds (Nos. 18FY1424900, ZZGCD18014 and 17JC1400601).

References
[1]
C. Chen, H.Y. Hsiao, Y.W. Chang, F. Ouyang, K.N. Tu.
Thermomigration in solder joints.
Mater Sci Eng R, 73 (2012), pp. 85-100
[2]
D.K. Mu, S.D. Mcdonald, J. Read, H. Huang, K. Nogita.
Critical properties of Cu6 Sn5 in electronic devices: recent progress and a review.
Curr Opin Solid State Mater Sci, 20 (2015), pp. 55-76
[3]
G. Zeng, S.D. Mcdonald, Q. Gu, Y. Terada, K. Uesugi, H. Yasuda, et al.
The influence of Ni and Zn additions on microstructure and phase transformations in Sn–0.7Cu/Cu solder joints.
Acta Mater, 83 (2015), pp. 357-371
[4]
Y.W. Yen, C.Y. Lin, G.N. Hermana, P.Y. Chen, Y.P. Wu.
Interfacial reactions in the Au/Sn- x Zn/Cu sandwich couples.
J Alloys Compd, 710 (2017), pp. 479-490
[5]
X. Li, T. Ivas, A.B. Spierings, K. Wegener, C. Leinenbach.
Phase and microstructure formation in rapidly solidified Cu-Sn and Cu-Sn-Ti alloys.
J Alloys Compd, 735 (2018), pp. 1374-1382
[6]
J.Y. Park, W. Seo, S. Yoo, Y.H. Kim.
Effect of Cu electroplating parameters on microvoid formation and high-speed shear strength in Sn-3.0Ag-0.5Cu/Cu joints.
J Alloys Compd, 724 (2017), pp. 492-500
[7]
S.Y. Zhang, X.Y. Xu, T.S. Lin, P. He.
Recent advances in nano-materials for packaging of electronic devices.
J Mater Sci Mater Electron, (2019),
[8]
F.J. Wang, H. Chen, Y. Huang, L. Liu, Z. Zhang.
Recent progress on the development of Sn–bi based low-temperature pb-free solders.
J Mater Sci Mater Electron, 30 (2019), pp. 3222-3243
[9]
C. Yu, J. Chen, Z. Cheng, Y. Huang, J. Chen, J. Xu, et al.
Fine grained Cu film promoting Kirkendall voiding at Cu3Sn/Cu interface.
J Alloys Compd, 660 (2016), pp. 80-84
[10]
C. Yu, Y. Yang, J. Chen, J. Xu, J. Chen, H. Lu.
Effect of deposit thickness during electroplating on Kirkendall voiding at Sn/Cu joints.
Mater Lett, 128 (2014), pp. 9-11
[11]
M.Y. Tsai, S.C. Yang, Y.W. Wang, C.R. Kao.
Grain growth sequence of Cu3Sn in the Cu/Sn and Cu/Sn–Zn systems.
J Alloys Compd, 494 (2010), pp. 123-127
[12]
Y.W. Wang, Y.W. Lin, C.R. Kao.
Kirkendall voids formation in the reaction between Ni-doped SnAg lead-free solders and different Cu substrates.
Microelectron Reliab, 49 (2009), pp. 248-252
[13]
Y.W. Wang, C.C. Chang, C.R. Kao.
Minimum effective Ni addition to SnAgCu solders for retarding Cu3Sn growth.
J Alloys Compd, 478 (2009), pp. L1-L4
[14]
Y.W. Wang, T.L. Yang, J.Y. Wu, C.R. Kao.
Pronounced effects of Zn additions on Cu-Sn microjoints for chip-stacking applications.
J Alloys Compd, 750 (2018), pp. 570-576
[15]
T. Laurila, V. Vuorinen, M. Paulasto-Kröckel.
Impurity and alloying effects on interfacial reaction layers in Pb-free soldering.
Mater Sci Eng R, 68 (2010), pp. 1-38
[16]
M.G. Cho, S.K. Kang, D.Y. Shih, H.M. Lee.
Effects of minor additions of Zn on interfacial reactions of Sn-Ag-Cu and Sn-Cu solders with various Cu substrates during thermal aging.
J Electron Mater, 36 (2007), pp. 1501-1509
[17]
S.C. Yang, C.E. Ho, C.W. Chang, C.R. Kao.
Strong Zn concentration effect on the soldering reactions between Sn-based solders and Cu.
J Mater Res, 21 (2010), pp. 2436-2439
[18]
H.R. Kotadia, O. Mokhtari, M.P. Clode, M.A. Green, S.H. Mannan.
Intermetallic compound growth suppression at high temperature in SAC solders with Zn addition on Cu and Ni–P substrates.
J Alloys Compd, 511 (2012), pp. 176-188
[19]
I.E. Anderson, J.L. Harringa.
Suppression of void coalescence in thermal aging of tin-silver-copper-X solder joints.
J Electron Mater, 35 (2006), pp. 94-106
[20]
C. Yu, D. Wang, J. Chen, J. Xu, J. Chen, H. Lu.
Study of Cu6Sn5 and Cu3Sn growth behaviors by considering trace Zn.
Mater Lett, 121 (2014), pp. 166-169
[21]
S.J. Clark, M.D. Segall, C.J. Pickard, P.J. Hasnip, M.I.J. Probert, K. Refson, et al.
First principles methods using CASTEP.
Z Kristallogr Cryst Mater, 220 (2005), pp. 567-570
[22]
J.P. Perdew, K. Burke, M. Ernzerhof.
Generalized gradient approximation made simple [phys. Rev. Lett. 77, 3865 (1996)].
Phys Rev Lett, 78 (1997), pp. 3865
[23]
D. Vanderbilt.
Soft self-consistent pseudopotentials in a generalized eigenvalue formalism.
Phys Rev B, 41 (1990), pp. 7892
[24]
H.J. Monkhorst.
Special points for Brillouin-zone integrations.
Phys Rev B Condens Matter, 16 (1976), pp. 1748-1749
[25]
R.A. Olsen, G.J. Kroes, G. Henkelman, A. Arnaldsson, H. Jónsson.
Comparison of methods for finding saddle points without knowledge of the final states.
J Chem Phys, 121 (2004), pp. 9776
[26]
M.D. Segall, C.J. Pickard, R. Shah, M.C. Payne.
Population analysis in plane wave electronic structure calculations.
Mol Phys, 89 (1996), pp. 571-577
[27]
M.D. Segall, R. Shah, C.J. Pickard, M.C. Payne.
Population analysis of plane-wave electronic structure calculations of bulk materials.
Phys Rev B Condens Matter, 54 (1996), pp. 16317
[28]
H.F. Zou, H.J. Yang, Z.F. Zhang.
Morphologies, orientation relationships and evolution of Cu6Sn5 grains formed between molten Sn and Cu single crystals.
Acta Mater, 56 (2008), pp. 2649-2662
[29]
G.G.M. Asta.
Phase stability, phase transformations, and elastic properties of Cu6Sn5: ab initio calculations and experimental results.
J Mater Res, 20 (2005), pp. 3102-3117
[30]
Y. Yang, Y. Li, H. Lu, C. Yu, J. Chen.
First-principles calculations of Zn substitutions in Cu6Sn5.
Comput Mater Sci, 65 (2012), pp. 490-493
[31]
H.M. Otte.
Lattice parameter determinations with an X-Ray spectrogoniometer by the debye-scherrer method and the effect of specimen condition.
J Appl Phys, 32 (1961), pp. 1536-1546
[32]
J. Nuss, U. Wedig, A. Kirfel, M.J. Dr.
The structural anomaly of zinc: evolution of lattice constants and parameters of thermal motion in the temperature range of 40 to 500K.
Z Fãƒâ¼r Anorg Und Allg Chemie, 636 (2010), pp. 309-313
[33]
V.T. Deshpe, D.B. Sirdeshmukh.
Thermal expansion of tin in the β–γ transition region.
Acta Crystallogr, 15 (2010), pp. 294-295
[34]
W.F. Feng, C.Q. Wang, M. Morinaga.
Electronic structure mechanism for the wettability of Sn-based solder alloys.
J Korean Inst Electr Electron Mater Eng, 31 (2002), pp. 185-190
Copyright © 2019. The Authors
Journal of Materials Research and Technology

Subscribe to our newsletter

Article options
Tools
Cookies policy
To improve our services and products, we use cookies (own or third parties authorized) to show advertising related to client preferences through the analyses of navigation customer behavior. Continuing navigation will be considered as acceptance of this use. You can change the settings or obtain more information by clicking here.