Elsevier

Microelectronics Reliability

Volume 79, December 2017, Pages 270-275
Microelectronics Reliability

Nanomechanical properties of Ag solder bumps doped with Pd and Au

https://doi.org/10.1016/j.microrel.2017.05.037Get rights and content

Highlights

  • We evaluated the Pd and Au doped on Ag solder alloy using nanoindentation.

  • H enhancement as the RT and TTT and particular for Pd 2% and Au 3%

  • Changeable H and E are due to responses of bonding strength of doped Ag solder.

  • Stress exponent S2 (Pd:3%) > S3 (Pd:5%) > S4 (Pd:2%-Au:3%)and bond strength (S4 > S3 > S2) are inversely related from creep test.

Abstract

As an approach to increasing their reliability, we doped Ag solder bumps with Pd and Au alloys to enhance their creep resistance, which we characterized using a nanoindentation technique. The hardnesses of the pure Ag and the Ag solders doped with Pd:3%, Pd:5%, and Pd:2%-Au:3% were 0.6 ± 0.1, 1.8 ± 0.2, 2.8 ± 0.3, and 3.1 ± 0.5 GPa, respectively. Although the hardness of the Ag solder was enhanced significantly when it contained Pd:2%-Au:3%, it exhibited a lower creep exponent (n > 2). The mechanism of the thermal recovery of the sample appears to be associated with activation of dislocation sources—also the reason for the decrease in hardness. The compounds doped with Pd:2%-Au:3% displayed significantly greater bond strengths and creep exponents, whereas the Pd-only compounds exhibited poorer reliability.

Introduction

The fabrication of intermetallic compound (IMC) solders at an interface is crucial for good adhesion under bump metallurgy (UBM) [1], [2], [3]. For electronics assemblies, lead (Pb)-free solders are preferred over tin (Sn)-Pb solders for use in bump fabrication [4]. The Pb-free solders are acceptable as Sn-rich alloys when they incorporate elements such as silver (Ag), copper (Cu), indium (In), and antimony (Sb). Typically, high-Sn-content solders react rapidly with Cu UBM, creating a thick layer of Cusingle bondSn IMC as a result of interdiffusion, such that the solder joints of IMCs decay and, thus, form microcracks [5]. Silver metal is generally used as a solder in microelectronics packaging because it has a unique combination of a low melting point and good wettability, strength, and creep [6], [7], [8]. In particular, doped Ag alloys are believed to be promising materials for creating bumps with greater connection reliability [9]. It has been suggested that a rapid interaction tends to form a substantial amount of IMCs, which can be brittle—an important and challenging problem affecting solder reliability [10], [11], [12]. Shock impact, bending mechanical damage, and plastic strain deformation are common in integrated circuit packaging. The periodicity of plastic deformation stresses can lead to fatigue and failure of solder bump interconnects [13]. Thermal cycling, thermal and mechanical shock, vibration, and bending processes are generally used as accelerated reliability tests for solder bumps [14]. To address these problems, it will be necessary to increase the solder reliability that existing Pb-free solders can endure, or develop new solder composites. Nanoindentation technology can be used to evaluate the mechanical properties of IMCs [15], [16], [17] because it enables more precise measurements of mechanical performance on the nano- and microscale. The alloy content improves the resistance of Ag solder bumps to thermal shock. Accordingly, to meet the requirements for the conductive paste bonding process, an Ag system IMC must have good resistance to thermal fatigue, high ductility, and better solderability as a conductive solder bump [8]. In terms of reliability, doping with gold (Au) and/or palladium (Pd) can enhance the hardness (H) and Young's moduli (E) of Ag solder bumps. Variations in metal–metal bond strength have effects on stability. In bulk Ausingle bondAg alloys, the metal–metal bond strength follows the order Ausingle bondAu > Ausingle bondAg > Agsingle bondAg. Thus, increasing the number of Au atoms would generate more Ausingle bondAg bonds, which are stronger than Agsingle bondAg bonds, resulting in greater stability of Ausingle bondAg alloys [18].

Although nanomechanical behavior generally distributes thermal shock on a solder array, little is known from nanoindentation studies about the significant differences between Ag solder bumps and Ag solder bumps doped with Au and/or Pd in printed circuit boards (PCBs). The data suggests that the Pd and/or Au atoms in the Ag solder play significant roles in preserving the integrity of the solder joint. Preexisting grain boundaries can be used with the nanoindentation technique to determine the stress concentration [19], [20], [21]. In addition, the nanomechanical aspects of Au and/or Pd doping as a means of inducing changes in Ag solder bumps and the effects of thermal temperature treatment (TTT) have not yet to be reported in detail. Furthermore, although a mechanical test may well be useful for examining the well-grained microstructures developed in Ag solder bumps, it may not be ideal for determining creep resistance and even less so for measuring grain boundary sliding. An attractive method of enhancing the creep resistance of solder materials is to adopt a solder bump approach. In this study, we employed the nanoindentation technique to analyze the values of H and E of Ag solder bumps at RT and TTT. In addition, we adopted the creep method to track the change in the duration of grain boundary formation.

Section snippets

Experimental procedure

Bump grid array (BGA) packaging components were prepared using Ag solder bumps and solder paste on a printed wiring board (PWB)–solder pad interface (Fig. 1(a)). The bulk of the solder bumps typically contain small spheroidal particles. Four sample types were prepared: pure Ag (S1) and Ag alloys doped with 3% Pd (S2), 5% Pd (S3), and 2% Pd and 3% Au (S4). To study the intermetallic formations in the bulk solder, the interfaces of the solder bumps were examined on the PWB. The samples were

Results and discussion

Fig. 2 presents cross-sectional views of the SEM and respective BSE/EDS images of solder bumps of (a) pure Ag and Ag doped with (b) Pd:3%, (c) Pd:5%, and (d) Pd:2%-Au:3%. We could clearly observe the SEM and BSE morphologies of the IMC solders and the grain boundaries from the solder bumps; EDS analysis revealed that the Ag solder was a well-formed bump in the aluminum (Al) metal pad, and the Pd and Au elements are observed. The Pd atoms in the solder were evident near the interface in Fig. 2

Conclusions

We have employed a nanoindentation technique to measure the values of H and E of the metal bonds in Ag solder bumps. First, we compared the values of H for pure Ag, Agsingle bondPd, and Agsingle bondPdsingle bondAu determined by force loading to those found in the literature. A comparison of the values of H with respect to the degree of metal doping revealed that the introduction of Pd tended to change the compound formation. Although Au had a more important effect than Pd on bond strength, it slightly decreased the values of E

Acknowledgments

This study was supported by the National Science Council of Taiwan (contracts MOST 103-2119-M-009-002 and 104-2119-M-009-010-MY3). We thank the Center for Advanced Instrumentation (Department of Electrophysics, National Chiao Tung University, Hsinchu 300, Taiwan, R.O.C.) for assistance with the CL/SEM/EDS measurements (JEOL JSM-7001F field-emission scanning electron microscope) and Professor Y. R. Jeng for supporting the SPM equipment used in this study.

References (27)

  • Y.C. Chan et al.

    Prog. Mater. Sci.

    (2010)
  • J. Shen et al.

    Microelectron. Reliab.

    (2009)
  • M. Amagai et al.

    Microelectron. Reliab.

    (2002)
  • M. He et al.

    Acta Mater.

    (2004)
  • D.G. Kim et al.

    J. Alloys Compd.

    (2005)
  • D.G. Kim et al.

    J. Alloys Compd.

    (2008)
  • H. Chen et al.

    J. Alloys Compd.

    (2012)
  • B.S.S. Chandra Rao et al.

    Microelectron. Eng.

    (2010)
  • M.H. Lin et al.

    Appl. Surf. Sci.

    (2010)
  • H.C. Wen et al.

    J. Alloys Compd.

    (2015)
  • V.M.F. Marques et al.

    Acta Mater.

    (2013)
  • F. Guo

    J. Mater. Sci. Mater. Electron.

    (2007)
  • S. Arai et al.

    J. Electrochem. Soc.

    (2003)
  • Cited by (1)

    View full text