The size effect on intermetallic microstructure ...

Soldering & Surface Mount Technology The size effect on intermetallic microstructure evolution of critical solder joints for flip chip assemblies Ye Tian Justin Chow Xi Liu Suresh K. Sitaraman

Article information: To cite this document: Ye Tian Justin Chow Xi Liu Suresh K. Sitaraman , (2015),"The size effect on intermetallic microstructure evolution of critical solder joints for flip chip assemblies", Soldering & Surface Mount Technology, Vol. 27 Iss 4 pp. 178 - 184 Permanent link to this document: http://dx.doi.org/10.1108/SSMT-10-2014-0020

Downloaded by Georgia Institute of Technology At 21:10 09 September 2015 (PT)

Downloaded on: 09 September 2015, At: 21:10 (PT) References: this document contains references to 25 other documents. To copy this document: [email protected] The fulltext of this document has been downloaded 16 times since 2015*

Users who downloaded this article also downloaded: Hong Gao, Jianhua Ma, Lilan Gao, Dunji Yu, Jinsheng Sun, (2015),"Thermal cycling aging effects on the ratcheting behavior of anisotropic conductive film", Soldering & Surface Mount Technology, Vol. 27 Iss 4 pp. 185-194 http://dx.doi.org/10.1108/ SSMT-04-2015-0016 Vadimas Verdingovas, Morten Stendahl Jellesen, Rajan Ambat, (2015),"Relative effect of solder flux chemistry on the humidity related failures in electronics", Soldering & Surface Mount Technology, Vol. 27 Iss 4 pp. 146-156 http://dx.doi.org/10.1108/ SSMT-11-2014-0022

Access to this document was granted through an Emerald subscription provided by emerald-srm:168551 []

For Authors If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service information about how to choose which publication to write for and submission guidelines are available for all. Please visit www.emeraldinsight.com/authors for more information.

About Emerald www.emeraldinsight.com Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online products and additional customer resources and services. Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Related content and download information correct at time of download.

The size effect on intermetallic microstructure evolution of critical solder joints for flip chip assemblies Ye Tian The School of Mechanical and Electrical Engineering, Henan University of Technology, Zhengzhou, China, and The George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA, and

Justin Chow, Xi Liu and Suresh K. Sitaraman


The George W. Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, Georgia, USA Abstract Purpose – The purpose of this paper is to study the intermetallic compound (IMC) thickness, composition and morphology in 100-␮m pitch and 200-␮m pitch Sn–Ag–Cu (SAC305) flip-chip assemblies after bump reflow and assembly reflow. In particular, emphasis is placed on the effect of solder joint size on the interfacial IMCs between metal pads and solder matrix. Design/methodology/approach – This work uses 100-␮m pitch and 200-␮m pitch silicon flip chips with nickel (Ni) pads and stand-off height of approximately 45 and 90 ␮m, respectively, assembled on substrates with copper (Cu) pads. The IMCs evolution in solder joints was investigated during reflow by using 100- and 200-␮m pitch flip-chip assemblies. Findings – After bump reflow, the joints size controls the IMC composition and dominant IMC type as well as IMC thickness and also influences the dominant IMC morphology. After assembly reflow, the cross-reaction of the pad metallurgies promotes the dominant IMC transformation and shape coarsened on the Ni pad interface for smaller joints and promotes a great number of new dominate IMC growth on the Ni pad interface in larger joints. On the Cu pad interface, many small voids formed in the IMC in larger joints, but were not observed in smaller joints, combined with the drawing of the IMC growth process. Originality/value – With continued advances in microelectronics, it is anticipated that next-generation microelectronic assemblies will require a reduction of the flip-chip solder bump pitch to 100 ␮m or less from the current industrial practice of 130 to150 ␮m. This work shows that as the packaging size reduced with the solder joint interconnection, the solder size becomes an important factor in the intermetallic composition as well as morphology and thickness after reflow. Keywords Flip-chips, Pb-free, Microstructure, Interconnections, Intermetallic compounds, Reflow Paper type Research paper

Generally, IMCs are desirable to form a good bond between the solder matrix and a pad’s metallization. However, due to their brittle nature, excessive IMC formation can potentially weaken the solder joint strength. It has been reported that the strength of a solder joint decreases with increasing thickness of the IMC at the interface. Therefore, the interface with the IMC is believed to be the initiation site for micro-cracks (Bang et al., 2008; Peng et al., 2006;Tian et al., 2011; Yang et al., 2011) and, thus, shorter thermal cycling fatigue life (Che and Pang, 2012). Also, the increment of the ratio of the IMC in smaller-sized solder joints can change the failure mode from ductile to a mixture of ductile and brittle (Li et al., 2011; Yang et al., 2012). Apart from the effect of excessive IMC on joint reliability, the IMC composition and morphology also have a

1. Introduction Solder joints work as interconnections in most electronic packages, and they are often the weakest link controlling the overall life times of electronic products(Shnawah et al., 2012; Tian et al., 2012). As there is a continual requirement for multi-functional capabilities in miniaturized electronic products, a strong driving force has accelerated the development of high-density electronic packages in recent years (Lu et al., 2014). Based on this trend, solder joint pitches have been reduced, and joint sizes have been scaled down correspondingly, bringing a great challenge in terms of solder joint reliability (Ouyang and Jhu, 2013). Intermetallic compounds (IMCs) form at the interface between the solder matrix and the metal pads (Cu, Ni and Au) used for the interconnection of substrate and component.

The authors would like to acknowledge the funding from the Chinese Scholarship Council that provided support for Mr Ye Tian’s research at Georgia Institute of Technology. Support from the National Science Foundation (ECCS-0901679) is also gratefully acknowledged. The authors also acknowledge the valuable support from the Packaging Research Center at Georgia Institute of Technology for substrate fabrication and flip-chip assembly.

The current issue and full text archive of this journal is available on Emerald Insight at: www.emeraldinsight.com/0954-0911.htm

Soldering & Surface Mount Technology 27/4 (2015) 178 –184 © Emerald Group Publishing Limited [ISSN 0954-0911] [DOI 10.1108/SSMT-10-2014-0020]

Received 28 October 2014 Revised 4 April 2015 Accepted 14 July 2015

178


Evolution of critical solder joints

Soldering & Surface Mount Technology

Ye Tian, Justin Chow, Xi Liu and Suresh K. Sitaraman

Volume 27 · Number 4 · 2015 · 178 –184

Figure 1 Flip-chip configuration

great influence. For example, Cu6Sn5 has good structural integrity under mechanical shock-loading conditions, but Ni additives in Cu6Sn5 could weaken its mechanical properties, resulting in crack formation (Mattila and Kivilahti, 2006). Also, the stress concentrations are inclined to be caused by the protruding region of the non-uniform IMC morphology when the solder joints are subjected to loading. Therefore, the strength of the joints with prism-type IMC grains at the interface was higher than those with the scallop-type IMC grains, because of a lower shear height (Xu et al., 2005; Lee et al., 2003). Furthermore, if multi-layer IMCs are formed at the pad interface, they may substantially weaken the mechanical properties. This is because the cracks are more likely to form at the interface between them (Zeng and Tu, 2002; Ho et al., 2007). As the solder joint size reduces, the IMC thickness and morphology as well as the IMC composition could experience a great change (Anderson et al., 2012). Therefore, it is necessary to investigate the size effect on the IMC evolution and realize the evolution mechanism, which could provide a theoretical basis and data for evaluating and improving the reliability of smaller-sized joints. Although the existing literature has explored the size effect on the IMC microstructure evolution in the solder joints, only larger solder joints with diameters of more than 150 ␮m have been studied, and some only considered one single interface between the pad metallization and the solder bumps (Wong et al., 2008; Ho et al., 2005). In addition, limited studies have been carried out on the IMC microstructure evolution in solder joints by using pure Ni and pure Cu as chip-side pad and substrate-side pad, respectively, for flip-chip assembly. Studies on the IMC growth are extremely important in emerging microelectronic applications, such as compliant interconnects and microbumps for three-dimensional (3D) stacked dies, where the solder stand-off height is in the range of 10-20 ␮m (Ostrowicki et al., 2012; Liu et al., 2012). The objective of this research was to study the IMC thickness, composition and morphology in 100- and 200-␮m pitch Sn–Ag–Cu (SAC305) flip-chip assemblies after bump reflow and assembly reflow. In particular, emphasis was placed on the effect of the solder joint size on the interfacial IMCs between the metal pads and the solder matrix. For the proposed study, this work used 100- and 200-␮m pitch silicon flip chips with Ni pads and stand-off heights of approximately 45 and 90 ␮m, respectively, assembled on substrates with Cu pads.

reflow, the assembly was underfilled with an epoxy-based underfill and cured at 160°C for 7 minutes. For cross-sectional scanning by scanning electron microscopy (SEM) imaging, the samples were cold-mounted in an acrylic polymer and polished down to a 0.05 ␮m finish. The samples were then etched for 5 seconds using a 5 per cent HCl–95 per cent C2H5OH solution to reveal the microstructure. To observe the 3D morphology of the IMC, the entire solder matrix had to be etched away by using a 5 per cent HCl–95 per cent C2H5OH solution for 1 minute, followed by etching with a 10 per cent HNO3–90 per cent C2H5OH solution for 8 minutes with ultrasonic agitation. The corner solder joint was imaged to consider the effect of the thermal mismatch-induced stress on the microstructure. SEM was used to analyze the microstructural morphology of the IMCs. Energy-dispersive X-ray spectrometry (EDS) was used to characterize the composition and elemental distribution of the IMCs.

3. Results and discussion 3.1 The IMC evolution of the 100- and 200-␮m pitch solder joints after bump reflow 3.1.1 The IMC formation and growth of the 100- and 200-␮m pitch solder bumps Figure 2 shows the cross-sectional back-scattered electron (BSE) image of a Ni/SnAgCu solder joint for a 20-␮m pitch solder bump. The BSE image mode of the SEM was used to distinguish different types of IMCs. Depending on the BSE contrast, three types of interfacial IMCs were observed on the Ni pad interface, as seen in Figure 2(b). Results of the EDX analysis for the IMC identity can be seen in Table I. The black Figure 2 Cross-section BSE images of the interfacial IMC layers after bump reflow for 200-␮m pitch

2. Experimental procedure

Ni pad

As shown in Figure 1, the samples used in this study were 10 ⫻ 10 mm silicon flip chips with a peripheral row of 376 and 186 solder bumps at 100- and 200-␮m pitch separately. The composition of the solder balls was Sn-3.0Ag-0.5Cu (weight per cent) (SAC305). The solder ball diameters were 60 and 120 ␮m. The flip chips had Ni pads, while the substrate pads were composed only of Cu. The solder balls were bumped on the silicon die first, and then flip chips were assembled onto the BT substrate. The reflow used standard lead-free thermal profiles with a peak temperature of approximately 240°C, and a time above liquidus (220°C) of about 35 and 60s. After

Ni N 3P

Ni Pad

(Cu,Ni)6Sn5

(Ni,Cu)3Sn4

Solder bu ump

(a)

(b)

Notes: (a) The entire solder bump; (b) the local magnified image 179




Volume 27 · Number 4 · 2015 · 178 –184

Figure 4 The top view of the IMC on the Ni pad interface for 100-␮m pitch solder joint and 200-␮m pitch solder bumps after bump reflow

Table I EDS compositional analysis of phases formed at the SAC305/Ni interface for 200 ␮m pitch Analysis sites


Die Die Die Die

side side side side

Composition (at %) Cu Sn P

Ni 22.16 30.91 73.57

29.42 9.9

48.42 59.19 4.62 24.37

Ag

21.81 75.63

Phase (Cu,Ni)6Sn5 (Ni,Cu)3Sn Ni3P Ag3Sn

(Ni,Cu u)3Sn4

(Ag3Sn3

layer-shaped IMC next to the Ni pad interface was Ni3P, the blocky-shaped IMC near the solder matrix was (Cu,Ni)6Sn5 and the medium light layer-shaped was (Ni,Cu)3Sn4, the dominant IMC was (Cu,Ni)6Sn5. Figure 3 shows the cross-sectional BSE image of a Ni/SnAgCu solder bump for the 100-␮m pitch. As seen in Figure 3(b), two interfacial IMCs formed on the Ni pad interface. According to the EDX compositional analysis, as seen in Table II, the needle-shaped IMC was (Ni,Cu)3Sn4 and the black layer-shaped IMC next to the Ni pad interface was Ni3P and the (Ni,Cu)3Sn4 was the dominant IMC. The protruding region of the needle-shaped (Ni,Cu)3Sn4 in the smaller-sized joints could be inclined to cause stress concentrations and, thus, have a negative influence on reliability. Figure 4 shows the top-view image of the IMC on the Ni pad interface of 100- and 200-␮m pitch bumps. As seen in Figure 4(b) for the 200-␮m pitch, the numerous bean-shaped (Ni,Cu)3Sn4 particles were close together on the Ni pad interface. The dominant (Cu,Ni)6Sn5 exhibited a diamond-shape and were scattered on the (Ni,Cu)3Sn4 particles. Additionally, some small Ag3Sn particles were observed to be attached on the (Cu,Ni)6Sn5 interface. As seen in Figure 4(a) for the 100-␮m pitch, the (Ni,Cu)3Sn4 showed long needle shapes and grew toward the solder matrix in a random orientation away from the Ni pad interface.

Ni pad

3.1.2 The size effect on the IMC evolution during the solder bumps preparation Based on the experimental results above, the discrepancy of the IMC types and dominant IMC in two types of solder bumps could be due to the difference in the bump size. Previous literature has reported that IMCs formed on the solder/Ni interface were strongly dependent on the Cu concentration in the solder bump. For a low concentration of Cu below 0.4 weight per cent, only (Ni,Cu)3Sn4 was formed at the interfaces. If the Cu concentration rose up to 0.6 weight per cent, only (Cu,Ni)6Sn5 was observed at the interfaces. When the Cu concentration was in the range of 0.4-0.6 weight per cent, (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4 coexisted at the interface(Mattila and Kivilahti, 2006; Vuorinen et al., 2008). In this study, although both bumps contained 0.5 weight per cent of Cu originally, the total amount of Cu was much greater in the larger bumps, meaning that the Cu concentration in the larger bumps could still be above 0.4 weight per cent, while in the smaller bumps, it could be less than 0.4 weight per cent after bump reflow. Therefore, the dual IMC structure was only observed on the Ni pad interface in larger bumps. To prove this assumption, the mass balance of Cu in the solder matrix can be used to calculate the Cu concentration after reflow, which is the amount of Cu in the solder before reflow is equal to that of the Cu inside the solder and the IMCs after the reflow. The relationship can be obtained as following equation:

Ni3P

冉冊冉冊兺冋冉冊

4 djoint 3 4 djoint 3 0 ␲ ␳solder · wCu ⫽ ␲ ␳solder · wCu 3 2 3 2 2 dpad ⫹ IMC ␲ hIMC␳IMC · wCu in IMC 2

Solder bump

(b)

Table II EDS compositional analysis of phases formed at the SAC305/Ni interface for 100 ␮m pitch Composition (at %) Cu Sn

Ni

Die side Die side

28.78 78.43

18.65

52.57 5.77

P 15.8

册

0 where wCu and wCu represent the Cu concentration in the solder matrix before and after the reflow, respectively. The djoint and dpad represent the diameter of the solder ball and Ni pad. ␳solder and ␳IMC are the density of the solder and the interfacial IMC. hIMC is the average thickness of the interfacial IMC. In this study, (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4 were the Cu-bearing IMCs at the Sn/Ni interface for the 100- and 200-␮m pitch bumps, respectively. Although no density value can be obtained from the existing literature for (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4, Cu and Ni have similar atomic weights and radii, as well as lattice constants, compared with Cu6Sn5 and

Notes: (a) The entire solder ball; (b) the local magnified image

Analysis sites

(b)

Notes: (a) 200-µm pitch; (b) 100-µm pitch

(Ni,Cu)3Sn4

(a)

(Cu,Ni)6Sn5

(a)

Figure 3 Cross-section BSE images of the interfacial IMC layers after bump reflow for 100-␮m pitch Ni pad

(Ni,Cu))3Sn4

Phase (Ni,Cu)3Sn4 Ni3P

180





Volume 27 · Number 4 · 2015 · 178 –184

Figure 5 Cross-section images of the interfacial IMC layers after assembly reflow for 200-␮m pitch

Ni3Sn4. It means their densities are close. Therefore, it is feasible to approximate the densities of (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4 with those of Cu6Sn5 (8.28 g/cm3) and Ni3Sn4 0 (8.65 g/cm3) (Wu et al., 2009). wCu in the solder matrix is 0.5 weight per cent. The wCu in the interfacial IMC was determined by EDX for both size bumps as seen in Table I. For smaller bumps, hIMC was calculated to be 1.603 ␮m, and the error was approximately 0.212 ␮m. The wCu can be calculated to be 0.15 weight per cent, which is much less than 0.4 weight per cent and consistent with the presumption above. For larger joints, hIMC was measured and calculated to be 1.18 ␮m, and the error was approximately 0.212 ␮m. The wCu was calculated to be 0.31 weight per cent, which is still less than 0.4 weight per cent, and it exhibits a discrepancy with our presumption. The discrepancy is attributed to two aspect;, first, the presumption based on the Ni–Sn–Cu isothermal ternary phase diagrams at 240°C from Zeng and Tu, 2002; Ho et al. (2007). However, Vuorinen et al. (2008) reported that the effect of Cu concentration on IMC type depended on the temperature; a reflow temperature reduction can lower the Cu concentration for IMC types such that only (Cu,Ni)6Sn5 formation takes place when the Cu concentration is above 0.3 weight per cent at 217°C. Second, they also reported that the Ag atoms in the solder matrix could lower the Cu concentration for the IMC types control. In this paper, although the maximum reflow temperature was approximately 240°C, the reflow temperature was in the range from approximately 240 to 217°C. Moreover, the solder matrix contained 3 weight per cent of Ag atoms. These are the reasons for the difference between the results and the presumptions stated above. To understand the effect of the joint size on the IMC evolution, the mechanism of IMC transformation in the solder bumps will be discussed. An IMC’s transformation in the solder bumps mainly depends on the phase growth kinetic and thermodynamic phase diagram (Vuorinen et al., 2008). According to the thermodynamic phase diagram, the (Cu,Ni)6Sn5 forms prior to the (Ni,Cu)3Sn4 on the Ni pad interface. Due to the unsaturated solder liquid for Cu in the SAC305 solder matrix, the (Cu,Ni)6Sn5 formation and dissolution takes place at the same time. When the Cu concentration is above approximately 0.4 weight per cent, the (Cu,Ni)6Sn5 formation rate is higher compared to (Cu,Ni)6Sn5 dissolution, and thus, interfacial reaction results show (Cu,Ni)6Sn5 growth. However, if the Cu concentration drops to 0.4 weight per cent, no new (Cu,Ni)6Sn5 forms and only (Cu,Ni)6Sn5 is dissolved until all (Cu,Ni)6Sn5 disappears. In this study, the total amount of Cu in the smaller bumps was lower. During the reflow, the Cu concentration was rapidly consumed down to 0.4 weight per cent; thus, no (Cu,Ni)6Sn5appeared on the Ni pad interface. However, in the larger bumps, the Cu concentration in the solder matrix was consistently maintained at more than approximately 0.4 weight per cent, and thus, the (Cu,Ni)6Sn5 coexisted with the (Ni,Cu)3Sn4.

Ni Pad

Ni pad d

(Ni,Cu)3Sn4

(Cu,Ni)6SSn5

Voids

(Cu,Ni)6Sn5

Cu pad d

(Cu,Ni)6Sn5

(a)

Cu Pad

(b)

(c)

Notes: (a) Entire solder ball; (b) ni pad interface; (c) cu pad interface Cu solder joints for the 200-␮m pitch. As seen in Figure 5(a), the solder joints show a barrel shape with a width of 125 ␮m and a stand-off height of 90 ␮m. On the Ni pad interface, as seen in Figure 5(b), EDS was used to determine the IMC composition. Comparing the IMC types after assembly with those before assembly reflow, no change was observed. However, the thickness and morphology of the (Cu,Ni)6Sn5 changed. The IMC thickness clearly increased, and the morphology was transformed to being a continuous wave-shape from the discontinuous block-shape. On the Cu side, as seen in Figure 5(c), the needle-shaped IMC with a layer was formed. EDX compositional analysis in Table III indicated that it was (Cu,Ni)6Sn5. Furthermore, some scattered voids were observed inside the (Cu,Ni)6Sn5; the reason for this is discussed below. After assembly reflow, some voids had formed inside the IMCs in the larger joints, while none existed in the smaller joints, which would have a positive effect on the smaller joint reliability. Figure 6 shows a top view of the interfacial IMC on both pad interfaces. On the Ni side interface, as seen in Figure 6(a), the (Cu,Ni)6Sn5 still exhibited a diamond-like shape. A large amount of new (Cu,Ni)6Sn5 formed and was close to array on the Ni pad interface. The (Cu,Ni)6Sn5 size had almost no change. In conclusion, the interface of the particle-shaped (Ni,Cu)3Sn4 provided the preferred lower-energy location to promote (Cu,Ni)6Sn5 nucleation and growth during assembly reflow. Table III EDS compositional analysis of phases formed at the SAC305/Ni interface for 200 ␮m pitch Composition (at %) Cu Sn

Analysis sites

Ni

Die side Die side Substrate side

22.75 31.91 6.03

29.90 9.90 46.34

47.35 58.19 47.62

Phase (Cu,Ni)6Sn5 (Ni,Cu)3Sn Ni3P

Table IV EDS compositional analysis of phases formed at the SAC305/Ni interface for 100 ␮m pitch

3.2 IMC evolution on the pad interface for both sizes of joints after the assembly reflow 3.2.1 The formation and growth of the IMC on the pad interface Figure 5 shows the cross-section BSE image of Ni/SnAgCu/ 181

Composition (at %) Cu Sn

Analysis sites

Ni

Die side Die side Substrate side

34.92 23.31 5.48

12.67 31.70 49.60

52.41 44.99 44.92

Phase (Ni,Cu)3Sn (Cu,Ni)6Sn5 Ni3P




Volume 27 · Number 4 · 2015 · 178 –184


Figure 6 Top view of the interfacial IMCs at the Ni and Cu pad interfaces after assembly reflow for 200-␮m pitch

will transform to the (Cu,Ni)6Sn5 compound (Wang and Liu, 2003). As a result, the (Ni,Cu)3Sn4 transformation and new (Cu,Ni)6Sn5 formation made (Cu,Ni)6Sn5 the dominant IMC, instead of (Ni,Cu)3Sn4 after assembly reflow. At the Cu pad interface, a thin layer of needle-shaped IMC was observed. As identified by EDS in Table II, the IMC was determined to be (Cu,Ni)6Sn5. Figure 8 shows the top-view images for the interfacial microstructure on the chip side and substrate side, respectively. As seen in Figure 8(a), on the Ni pad interface, the needle-shaped (Ni,Cu)3Sn4 prior to the assembly transformed to the coarse rod-shaped (Cu,Ni)6Sn5 IMC. On the Cu pad interface, as seen in Figure 8(b), the short-needle (Cu,Ni)6Sn5 formed. Although, after assembly reflow, the needle-shaped (Ni,Cu)3Sn4 in the smaller-sized joint was transformed to the coarse rod-shaped (Cu,Ni)6Sn5, the morphology could still weaken the reliability of the solder joint, and the phase transformation stress was inclined to cause crack formation. Also, the volume ratio in the smaller-sized joint was still higher.

On the Cu side interface, as seen in Figure 5(b), a great number of (Cu,Ni)6Sn5 needles were formed on the interface. However, their lengths varied with different growth orientation. In addition, the (Cu,Ni)6Sn5 morphology on the Ni interface was diamond-shaped, while it was needle-shaped on the Cu pad interface. Based on the differences in their composition and growth conditions, the Ni concentration and their nucleation interface play a critical role. Figure 7 shows a cross-section image of Ni/SnAgCu/Cu solder joints after assembly for the 100-␮m pitch. As seen in Figure 7(a), the solder joint shows a similar shape to the 200-␮m pitch example. Figures 7(b and c) are locally magnified micrographs of the interfaces on the Ni pad interface and the Cu pad interface. At the Ni pad interface, (Cu,Ni)6Sn5, which was not present before assembly, was the dominant IMC. EDS was used for the compositional analysis of the phases formed at the Cu/ SAC305/Ni interface. The original (Cu,Ni)3Sn4 layer had almost entirely converted to (Cu,Ni)6Sn5 after assembly reflow. The diffusion coefficient of Cu in the liquid Sn-based solder is high enough so that the Cu atoms from the Cu pad can diffuse across the entire joint to join in the interfacial reaction on the Ni pad (Rizvi et al., 2009). According to the Sn–Cu–Ni ternary phase diagram at 240°C (Zeng and Tu, 2002; Ho et al., 2007), the presence of a small amount of Ni (0.15 weight per cent) in the liquid Sn phase could reduce the solubility of Cu from 1.1 to 0.4 weight per cent. Because Ni lowers the solubility of Cu on the Ni pad interface, the concentration gradient between the two interfaces induced a large driving force, which could force the Cu atoms to diffuse toward the Ni pad interface. The Cu from the Cu pad interface diffused into the (Ni,Cu)3Sn4. Once the Cu content exceeds 8.5 atom per cent, the (Ni,Cu)3Sn4 compound phase

3.3 The size effect on the IMC evolution during assembly reflow Although the IMC types in both sizes of joints were the same, several differences still existed between them after assembly reflow. On the Ni pad interface, the (Cu,Ni)6Sn5 morphology was coarse rod-shaped for the smaller joints, while it was diamond-shaped for the larger joints. This discrepancy was closely related to the original IMC morphology before assembly reflow. On the Cu pad interface, the (Cu,Ni)6Sn5 thickness was obviously thicker for the larger joints, and some voids appeared inside the (Cu,Ni)6Sn5, but were not observed in the (Cu,Ni)6Sn5 of the smaller joints. The interfacial IMC growth process of both joint sizes explains the difference found in the experimental results, as seen in Figures 9 and 10. As seen in Figure 9(a), when the interfacial reaction started, the needle-shaped (Cu,Ni)6Sn5 formed on the Cu pad interface. As the reflow time increased, a large amount of needle-shaped (Cu,Ni)6Sn5 grew and intersected with some liquid solder trapped in them, as seen in Figure 9(b). In the subsequent reflow time, some liquid solder trapped in the IMC transformed to (Cu,Ni)6Sn5 with the Cu reaction, and thus, some voids formed as a result of the volume reduction from the liquid–solid transformation. Furthermore, the solder matrix could not provide a sufficient amount of liquid solder Figure 8 The top view of the IMC on both side pad interfaces for 100-␮m pitch solder

Figure 7 Cross-section images of the interfacial IMC layers after assembly reflow for 100-␮m pitch Ni pad (Ni,Cu)3Sn4 (C Cu,Ni)6Sn5

Cu pad

(a)

(Cu,Ni))6Sn5 Cu pad

(b)

Ag3Sn

(c)

Notes: (a) Entire solder ball; (b) ni pad interface; (c) cu pad interface 182




Volume 27 · Number 4 · 2015 · 178 –184

Figure 9 The IMC growth process after assembly reflow for 200-␮m pitch


Figure 10 The IMC growth process after assembly reflow for 100-␮m pitch

inside the IMC in the larger joint, while they did not exist in the smaller joints, which would have a positive effect for its reliability.

into all the voids with the Cu reaction by diffusion through the IMC during the assembly time. As a result, some voids were left in the IMC after assembly reflow however, as seen in Figure 9(c) for the smaller joints. Due to the low Cu concentration and high Cu concentration gradient in the smaller joints before reflow, the new (Cu,Ni)6Sn5 formed on the Cu pad interface had to dissolve into the solder matrix quickly, and thus, the (Cu,Ni)6Sn5 length was shorter, and its thickness was less during assembly reflow. Because of the short length of the (Cu,Ni)6Sn5, only a few voids formed inside the (Cu,Ni)6Sn5, as seen in Figure 10(b). Due to the thinner IMC and a few voids, enough solder in the solder matrix could diffuse to voids with the Cu reaction, and the (Cu,Ni)6Sn5 formed could be full of all voids as seen in Figure 10(c).

References Anderson, I.E., Boesenberg, A., Harringa, J., Riegner, D., Steinmetz, A. and Hillman, D. (2012), “Comparison of extensive thermal cycling effects on microstructure development in micro-alloyed Sn-Ag-Cu solder joints”, Journal of Electronic Materials, Vol. 41 No. 2, pp. 390-397. Bang, W., Moon, M.W., Kim, C.U., Kang, S.H., Jung, J.P. and Oh, K.H. (2008), “Study of fracture mechanics in testing interfacial fracture of solder joints”, Journal of Electronic Materials, Vol. 37 No. 4, pp. 417-428. Che, F. and Pang, J.H. (2012), “Characterization of IMC layer and its effect on thermomechanical fatigue life of Sn-3.8Ag-0.7Cu solder joints”, Journal of Alloys and Compounds, Vol. 541, pp. 6-13. Ho, C., Yang, S. and Kao, C. (2007), “Interfacial reaction issues for lead-free electronic solders”, Journal of Materials Science: Materials in Electronics, Vol. 18 Nos 1/3, pp. 155-174. Ho, C.E., Lin, Y.W., Yang, S.C. and Kao, C.R. (2005), “Volume effect on the soldering reaction between SnAgCu solders and Ni”, Proceedings International Symposium on Advanced Packaging Materials: Processes, Properties and Interfaces, pp. 39-44. Lee, H.T., Chen, M.H., Jao, H.-M. and Liao, T.-L. (2003), “Influence of interfacial intermetallic compound on fracture behavior of solder joints”, Materials Science and Engineering: A, Vol. 358 Nos 1/2, pp. 134-141. Li, X., Xia, J., Zhou, M., Ma, X. and Zhang, X.P. (2011), “Solder volume effects on the microstructure evolution and shear fracture behavior of ball grid array structure

4. Conclusions The results of this work have shown that solder joint size is an important factor affecting the IMC type, morphology and thickness. After reflow, in the smaller joints, the dominant IMC was needle-shaped (Ni,Cu)3Sn4, while dual-layer IMCs with (Cu,Ni)6Sn5 and (Ni,Cu)3Sn4 formed in larger joints, with the dominant IMC being blocky-shaped (Cu,Ni)6Sn5. After assembly reflow, the cross-reaction of the pad metallurgies promoted the dominant IMC transformation and the shape coarsened on the Ni Pad interface for the smaller joints due to the joint size effect. The dominant IMC was transformed as coarse rod-shaped (Cu,Ni)6Sn5, and no voids formed inside the IMC in the smaller-sized joints. In addition, although the IMC thickness on the two pad surfaces was smaller than that in the larger joints, due to the joint size effect, the volume ratio of IMCs in the joint was still higher. Compared with the larger joints, the IMC morphology, type and thickness could weaken the solder joint and impact the reliability. However, after assembly reflow, some voids formed 183





Volume 27 · Number 4 · 2015 · 178 –184

Sn-3.0Ag-0.5Cu solder interconnects”, Journal of Electronic Materials, Vol. 40 No. 12, pp. 2425-2435. Liu, X., Chen, Q., Sundaram, V., Wachtler, K.P., Tummala, R.R. and Sitaraman, S.K. (2012), “Reliability assessment of through-silicon vias in multi-die stack packages”, IEEE Transactions on Device and Materials Reliability, Vol. 12 No. 2, pp. 263-271. Lu, H., Takagi, Y., Suzuki, Y., Sawyer, B., Taylor, R., Sundaram, V. and Tummala, R. (2014), “Demonstration of 3–5 ␮m RDL line lithography on panel-based glass interposers”, Electronic Components and Technology Conference (ECTC), 2014 IEEE 64th,, Orlando, FL, pp. 1416-1420. Mattila, T. and Kivilahti, J. (2006), “Reliability of lead-free interconnections under consecutive thermal and mechanical loadings”, Journal of Electronic Materials, Vol. 35 No. 2, pp. 250-256. Peng, C.T., Chiang, K.-N., Ku, T. and Chang, K. (2006), “Experimental characterization and mechanical behavior analysis of intermetallic compounds of Sn–3.5 Ag lead-free solder bump with Ti/Cu/Ni UBM on copper chip”, Microelectronics Reliability, Vol. 46, pp. 523-534. Shnawah, D.A., Sabri, M.F. and Badruddin, I.A. (2012), “A review on thermal cycling and drop impact reliability of SAC solder joint in portable electronic products”, Microelectronics Reliability, Vol. 52 No. 1, pp. 90-99. Ostrowicki, G.T., Fritz, N.T. and Sitaraman, S.K. (2012), “Domed and released thin-film construct-an approach for material characterization and compliant interconnects”, IEEE Transactions on Device and Materials Reliability, Vol. 12 No. 1, pp. 15-23. Ouyang, F.Y. and Jhu, W.C. (2013), “Comparison of thermomigration behaviors between Pb-free flip chip solder joints and microbumps in three dimensional integrated circuits: bump height effect”, Journal of Applied Physics, Vol. 113, pp. 043711-043711-8. Rizvi, M., Lu, H. and Bailey, C. (2009), “Modeling the diffusion of solid copper into liquid solder alloys”, Thin Solid Films, Vol. 517 No. 5, pp. 1686-1689. Tian, Y., Hang, C., Wanga, C., Yang, S. and Lin, P. (2011), “Effects of bump size on deformation and fracture behavior of Sn3.0Ag0. 5Cu/Cu solder joints during shear testing”, Materials Science and Engineering: A, Vol. 529, pp. 468-478.

Tian, Y., Chow, J., Liu, X., Ping, Y., Suresh, Wu. and Sitaraman, K. (2012), “Study of intermetallic growth and kinetics in fine-pitch lead-free solder bumps for next-generation flip-chip assemblies”, Journal of Electronic Materials, Vol. 42 No. 2, pp. 230-239. Vuorinen, V., Yu, H. and Kivilahti, J. (2008), “Formation of Intermetallic Compounds Between Liquid Sn and Various CuNi x Metallizations”, Journal of Electronic Materials, Vol. 37 No. 6, pp. 792-805. Wang, S. and Liu, C. (2003), “Study of interaction between Cu-Sn and Ni-Sn interfacial reactions by Ni-Sn3.5Ag-Cu sandwich structure”, Journal of Electronic Materials, Vol. 32 No. 11, pp. 1303-1309. Wong, C., Pang, J.H., Tewa, J.W., Loka, B.K., Lua, H.J., Nga, F.L. and Sunb, Y.F. (2008), “The influence of solder volume and pad area on Sn–3.8 Ag– 0.7 Cu and Ni UBM reaction in reflow soldering and isothermal aging”, Microelectronics Reliability, Vol. 48 No. 4, pp. 611-621. Wu, W.H., Chung, H.L. and Ho, C.E. (2009), “The influence of current direction on the Cu-Ni Cross-Interaction in Cu/Sn/Ni diffusion couples”, Journal of Electronic Materials, Vol. 38 No. 12, pp. 2563-2572. Xu, L., Pang, J.H. and Che, F.X. (2005), “Intermetallic growth and failure study for Sn-Ag-Cu/ENIG PBGA solder joints subject to thermal cycling”, Proceedings 55th Electronic Components and Technology Conference, Lake Buena Vista, FL, pp. 682-686. Yang, M., Li, M., Wang, L., Fu, Y., Kim, J. and Weng, L. (2011), “Cu6Sn5 morphology transition and its effect on mechanical properties of eutectic Sn-Ag solder joints”, Journal of Electronic Materials, Vol. 40 No. 2, pp. 176-188. Yang, L., Zhang, Q. and Zhang, Z.F. (2012), “Effects of solder dimension on the interfacial shear strength and fracture behaviors of Cu/Sn-3Cu/Cu joints”, Scripta Materialia, Vol. 67 Nos 7/8, pp. 637-640. Zeng, K. and Tu, K. (2002), “Six cases of reliability study of Pb-free solder joints in electronic packaging technology”, Materials Science and Engineering: R: Reports, Vol. 38 No. 2, pp. 55-105.

Corresponding author Ye Tian can be contacted at: [email protected]

For instructions on how to order reprints of this article, please visit our website: www.emeraldgrouppublishing.com/licensing/reprints.htm Or contact us for further details: [email protected]

184