Synthesis with Glucose Reduction Method and Low Temperature Sinteringof Ag-Cu Alloy Nanoparticle Pastes for Electronic Packaging

Dongyue Zhang1, Guisheng Zou1,+, Lei Liu1, Yingchuan Zhang1,Chen Yu1, Hailin Bai1 and Norman Zhou1,2

1Department of Mechanical Engineering, Tsinghua University, Beijing 100084, P. R. China2Department of Mechanical and Mechatronics Engineering, University of Waterloo,Waterloo, ON, N2L 3G1, Canada

The metallic nanoparticle paste is receiving great interests recently because it is a potential interconnect material which can perform joiningat low temperature and serves at high temperature. The nano-Ag paste and nano-Cu paste have been the hot areas of research, whereas the highcost and low resistance of electrochemical migration of the former and the relatively low anti-oxidation property of the latter limit theirapplications. In this study, Ag-Cu alloy nanoparticles with the size of 20­50 nm were synthesized with glucose as the reducing agent and NaOHas accelerator. The Ag-Cu nanoparticle paste showed no oxidation after sintering up to 350°C in the air, indicating that the antioxidant capacitywas superior to that of the mechanically mixed Ag nanoparticles and Cu nanoparticles. In addition, the electrochemical migration resistance ofthe sintered Ag-Cu alloy pastes was better than that of the Ag nanoparticle paste. This paste can be used to effectively bond silver-plated copperbulks with maximum shear strength of 35MPa. [doi:10.2320/matertrans.MI201406]

(Received December 8, 2014; Accepted May 14, 2015; Published June 26, 2015)

Keywords: electronic packaging, nanoparticle pastes, nanoalloy, electrochemical migration

1. Introduction

The rapid development of microelectronics has beenplacing increasingly greater emphasis on electronic productsand systems that are more powerful, integrated, reliable andcost-effective.1) High temperature serving is one of theparticularly crucial requirements, which requests the packag-ing material having high service temperature.2) As is known,nano-scale metal particle has much higher surface energy dueto its particularly high ratio of surface atoms comparing withthe bulk metal.3­5) Ag nanoparticles can be sintered at 250°C,which is much lower than the bulk silver melting point(960°C).6­9) Ag nanoparticles sintered at low temperatureform micro-scaled silver layer, where the joints can providestable connection and fine electric as well as thermalconduction.10­12) As a result, low temperature sintering andthen high temperature serving can be realized. Studies in thisfield are mostly focused on Ag nanoparticle paste and Cunanoparticle paste.13,14) However, both nano-Ag and nano-Cupastes have their disadvantages. Silver is a relatively weakcandidate against electrochemical migration, which may leadto short circuit during service. Furthermore, silver is toocostly to find wide use in industry. Copper nanoparticles tendto oxidize in the air, and the strength of the copper joints islower than that of silver.15­17) In response to the abovedisadvantages, this study proposed Ag-Cu alloy nanoparticlepaste as one of the improved options. Firstly, the alloynanoparticle paste was prepared by a novel process ofglucose chemical reduction method and characterized.Secondly, two Ag/Ni-coated copper discs were bonded withthe nano-alloy paste. The joint properties were compre-hensively evaluated through combining the joint strength andthe electrochemical migration resistance of the sinteredcounterelectrodes using nano-particle pastes followed indetails.

2. Experimental Procedure

Ag-Cu alloy nanoparticles were synthesized by glucose(C6H12O6) from AgNO3 and Cu(NO3)2 simultaneously, whileNaOH was added as accelerating agent and PVP was addedas dispersing agent. The solution was heated in an oil bath to95°C. Acetone extraction, centrifuge and stewing methodswere employed to gain the resultant alloy nanoparticle paste.For comparison, Ag nanoparticle paste and Cu nanoparticlepaste were prepared by polyol chemical reduction process.Oxidative stability comparison between Ag-Cu alloy nano-particle paste and mechanically mixed Ag+Cu nanoparticlepaste (mole ratio 1 : 1) was conducted by heating the pastesat 80°C, 180°C, 200°C, 250°C, 300°C, 350°C, for 10min,respectively, in ambient atmosphere and then examining thechange of the pastes using XRD.

Counter electrodes with a gap of 1mm were formed on aglass substrate coated with the Ag and nano Ag-Cu alloypastes. The glass plates with the coated counter electrodeswere sintered at 350°C for 30min in air. The electrochem-ical migration resistance of the sintered electrodes wasevaluated by the water drop test (Fig. 1). The time for shortcircuit of the counter electrodes with distilled water underan applied voltage of 5V was measured. In order tocompare the electrochemical migration resistance of Ag andnano Ag-Cu alloy pastes, the short circuit current was set at1mA.

In consideration of simulating the actual joints inelectronic packaging, two Ag/Ni plated copper discs withdiameter 6mm and 10mm, respectively, with a height of5mm, were used as base material. The layer of Ag-Cu alloynanoparticle pastes was sandwiched between the two kinds ofcopper cylinders/discs. These discs were bonded at variousprocessing temperatures for different periods of time underdifferent pressures in air (called “Sintering-bonding”) toevaluate the influence of the bonding parameters on sinteringAg-Cu alloy nanoparticle paste.+Corresponding author, E-mail: zougsh@tsinghua.edu.cn

Materials Transactions, Vol. 56, No. 8 (2015) pp. 1252 to 1256©2015 The Japan Institute of Metals and Materials

The microstructures and compositions of the synthesizednanoparticle pastes were observed and analyzed by X-raydiffraction (XRD), UV-vis spectrum, scanning electronicmicroscope (SEM) and transmission electronic microscope(TEM). The fracture surface was also observed by SEM. Theaverage shearing strength of three disc joints under the samesintering-bonding condition were evaluated through shearingtests using the Thermal-Mechanical Simulator Gleeble1500D with a displacement speed of 5mm/min at roomtemperature. In addition, the short circuit current and timewere monitored by semi-digital multimeter keithley 2000.

3. Results and Discussions

3.1 Microstructures of the Ag-Cu alloy nanoparticlepaste

Figure 2 shows the UV-vis spectrum of Ag-Cu alloynanoparticles. As is shown, Ag-Cu alloy nanoparticles hadspecific absorb peak at 434 nm, which was in the middle ofthe characteristic Ag nanoparticle absorb peak (408 nm) andCu nanoparticle absorb peak (574 nm). Instead of an evidentabsorb peak, a gentle slope was observed in mechanicallymixed Ag+Cu nanoparticles’ spectrum. This result suggeststhat the Ag and Cu elements in the obtained paste coexist in ananoparticle and form an alloy structure rather than a core-shell structure, because two distinct UV absorption peakscorresponding to each element can be observed in core-shellstructure.12­15)

Figure 3 shows the XRD result of vacuum stored Ag-Cualloy nanoparticle paste. Compared to standard PDF cards,peak No. 1, No. 3, No. 5 and No. 6 slightly deviated fromstandard Ag peaks, and the peaks shits were 38.1145 to38.202, 44.299 to 44.348, 64.443 to 64.624 and 77.397 to77.631, respectively. Peak No. 2 and No. 4 slightly shiftedfrom standard Cu peaks, 43.473 to 43.419 and 50.375 to50.159, respectively. The different shit directions of Ag andCu peaks indicate the existence of Ag-Cu alloy, which wasalso observed by Ref. 11, 13).

TEM pictures of Ag-Cu alloy nanoparticles are shown inFig. 4. As can be seen from the pictures, the diameter of thenanoparticles spread from 20 nm to 50 nm. An organic shellwas found to surround the nanoparticle and protect it fromagglomerating with others (Fig. 4(b)). The EDS analysis ofthe Ag-Cu alloy nanoparticle was done by focusing the TEMelectron beam to an isolated nanoparticle. The TEM grid was

pure nickel. Therefore the analyzed nanoparticle containedboth silver and copper elements. Another isolated Ag-Cunanoparticle was also analyzed and the Cu:Ag atomic ratiowas 18 : 82. It indicates that the nanoparticles were either Curich or Ag rich. The Ag-Cu phase diagram shows limitedmutual solid solubility between Cu and Ag. Althoughreducing material dimension could enhance the solidsolubility, the solute contains in the present study stillshowed a remarkable higher value. It is believed that in onesingle nanoparticle, the phase boundary between Cu and Agstill existed. The Ag-Cu alloy nanoparticle consisted of Cu-rich grains and Ag-rich grains.

3.2 Oxidative stability of the Ag-Cu alloy nanoparticlepaste

Figure 5(a) shows the XRD results of Ag-Cu alloynanoparticles after being sintered at different temperatures.

Fig. 2 UV-vis spectrums of Ag, Cu, Ag-Cu alloy nanoparticle pastes.

Fig. 3 XRD result of Ag-Cu alloy nanoparticle paste.

Fig. 1 Schematic illustration of the water-drop test.

Synthesis with Glucose Reduction Method and Low Temperature Sintering of Ag-Cu Alloy Nanoparticle Pastes for Electronic Packaging 1253

As can be seen, the diffraction peaks of CuO or Cu2O werenot observable in the Ag-Cu paste sample subjected tosintering at temperatures from 180°C to 350°C, meaning thatthe Cu in Ag-Cu alloy nanoparticle paste was protected frombeing oxidized. On the other hand, the appearance of CuOpeak when sintered at 80°C might be accounted for by theincomplete decomposition of organic shell leading to areduced sintered density and the oxidation of copper as aresult of copper reacting with the organic residue. Figure 5(b)shows a comparison of the resistance to oxidation betweenAg-Cu alloy nanoparticles and mechanically mixed Ag+Cunanoparticles. As is shown in the figure, Cu2O diffractionpeak was found in the XRD result of mechanical mixedAg+Cu nanoparticles after sintering at 350°C for 10min,whereas no similar phenomenon was observed in the Ag-Cualloy nanoparticles, which supports the conclusion that Ag-Cu alloy nanoparticles appeared to be more resistant againstoxidation than mechanically mixed Ag+Cu nanoparticles. Itis believed that the exposure area of Cu to air accounted forthis difference. For the mechanically mixed Ag and Cunanoparticles, all of the Cu exposed to oxygen duringsintering because the Cu and Ag were separated by air gap orPVP organic shell. For the Ag-Cu alloy nanoparticle, thecopper at the nanoparticle surface still exposed to oxygen butthe rest of Cu was insulated and protected by Ag-Cu phase

boundaries in the nanoparticle. Therefore, the Ag-Cu alloynanoparticle can also oxide but the oxides volume and extentwere much lower that the mechanically mixed Ag and Cunanoparticles. Another reason was recently proposed byKim et al. that electron transfer from Cu to Ag within thesebimetallic nanoparticles allows far better resistance tooxidation than monometallic Cu nanoparticles.18)

3.3 Electrochemical migration resistance of the Ag-Cualloy nanoparticle paste

Figure 6 shows the results of the water-drop test. The timefor short circuit (current reaching 1mA) of Ag counter-electrodes with distilled water under 5V was less than 60 s,while the current of Ag-Cu alloy counterelectrodes keptdecreasing and never reached the short circuit current duringthe experiment. These results indicate that the electrochem-ical migration resistance of the sintered nano Ag-Cu alloypastes is better than that of the Ag nanoparticle paste.

This different electrochemical migration resistance mightresult from the distinct electrode potentials of Ag and Cu.As is known, during the process of the electrochemicalmigration, Ag and Cu react with water under certain electricpotential gradient, producing Ag2O and CuO, respectively.The chemical reaction equations are as follows, equationsA1-A6. The silver oxide was electric conductive, making the

Fig. 4 TEM pictures of Ag-Cu alloy nanoparticles, (a), (b) TEM images, (c) chemical composition.

(a) (b)

Fig. 5 Oxidation resistance at different sintering temperature for 10min bonding time, (a) Ag-Cu alloy nanoparticles paste, (b)Mechanically mixed Ag+Cu nanoparticles paste.

D. Zhang et al.1254

electrochemical progress continue until short circuit occurs.Cu had a lower standard electrode potential. It is believed thatthe Cu element in the Ag-Cu alloy nanoparticle paste notonly produced non-conducting CuO but also inhibited Agfrom losing electrons. Consequently, the electrochemicalmigration resistance of Ag-Cu alloy nanoparticle paste wassuperior to that of Ag nanoparticle paste. Detail study of theelectrochemical behavior of the Ag-Cu alloy nanoparticlepaste will be conducted in the future.

3.4 Effects of the sintering-bonding parameters on thejoints using Ag-Cu alloy nanoparticle paste

The curves of the shearing strength of the joints vs thesintering-bonding temperature using nano Ag-Cu alloy pasteare shown in Fig. 7. The joints were bonded at 180°C,200°C, 250°C, 300°C and 350°C, under 2MPa pressure for10min. As can be seen, the maximum shearing strength ofthe joint is acquired when sintering­bonding temperature is300°C. Figure 8 shows the fracture surface of samplessintered at different temperature. Organic shells can be foundin the fracture surfaces in samples sintered at 180°C and200°C. The joint strength peaked at 300°C bonding temper-

ature. Most of the organic shells and reacting residue wereevaporated and decomposed at 300°C. The nanoparticlessintered layer became dense, leading to effective joints withthe average joint shear strength of 33MPa and maximumstrength of 45MPa.

Figure 9 shows the effect of bonding pressure on the jointstrength. The bonding temperature and time kept constantat 250°C and 10min, while the bonding pressure variedbetween 0 and 10MPa. When sintered under 0MPa, no jointstrength can be achieved. The maximum shearing strengthpeaked under applied pressure 2MPa. When the pressureexceeded 5MPa, the joint strength dropped because thenanoparticle paste was squeezed to the edge of the joints.Figure 10 shows the effect of bonding time on the joint

Fig. 6 Comparison of the electrochemical migration resistance between Agand Ag-Cu alloy nanoparticle pastes.

Fig. 7 Effect of bonding temperature on the joint strength when thebonding pressure and time was 2MPa and 10min.

Fig. 8 Fracture surface structures of joints at different bonding temper-ature, (a) 180°C, (b) 200°C, (c) 250°C, (d) 300°C, (e) 350°C.

Fig. 9 Effect of bonding pressure on joint strength when sintered at 250°Cand 10min.

Synthesis with Glucose Reduction Method and Low Temperature Sintering of Ag-Cu Alloy Nanoparticle Pastes for Electronic Packaging 1255

strength when sintered at 250°C under 5MPa. Increasedholding time resulted in more complete evaporation anddecomposition of organic shells and organic residue, higherdensity of the sintering layer as well as shearing strength.

4. Conclusion

(1) Ag-Cu alloy nanoparticle paste was successfullysynthesized by reducing AgNO3 and Cu(NO3)2 withglucose as reducing agent and NaOH as accelerateagent. The diameter of the nanoparticles ranged from20 nm to 50 nm.

(2) Ag-Cu alloy nanoparticle paste acquired by our methodshowed no oxidization when heating at 350°C for10min. Its oxidation resistance was superior to that ofmechanically mixed Ag+Cu nanoparticle paste.

(3) The electrochemical migration resistance of the sinterednano Ag-Cu alloy pastes had an advantage over that ofthe Ag nanoparticle paste.

(4) The maximum shear strength was obtained under thefollowing parameters: 300°C, 10min, 2MPa. Prolong-ing heating time generated stronger joints.

Acknowledgments

This research was supported by National Natural ScienceFoundation of China (Grant No. 51375261), by State KeyLaboratory of Automotive Safety and Energy, Tsinghua

University (Grant No. 2013XC-B-02), by Tsinghua Univer-sity Initiative Scientific Research Program (Grant No.2010THZ 02-1) and by State Key Lab of Advanced Weldingand Joining, Harbin Institute of Technology (Grant No. AWJ-M14-05).

REFERENCES

1) L. Ma, M. Li and H. Xian: Weld. Technol. 38 (2009) 6­10 (in Chinese).2) G. Dong: Master Thesis., Tianjin University, (2009) (in Chinese).3) J. Yan, G. Zou and J. Li: Mater. Eng. 10 (2010) 5­8 (in Chinese).4) J. Yan, G. Zou, A. Hu and Y. Norman Zhou: J. Mater. Chem. 21 (2011)

15981­15986.5) J. Yan, G. Zou, A. Wu, J. Ren, J. Yan, A. Hu and Y. Zhou: Scr. Mater.

66 (2012) 582­585.6) J. Yan, G. Zou, A. Wu, J. Ren, A. Hu and Y. Norman Zhou: J. Electron.

Mater. 41 (2012) 1924­1930.7) J. Yan, G. Zou, A. Wu, J. Ren, A. Hu and Y. Norman Zhou: J. Electron.

Mater. 41 (2012) 1886­1892.8) J. Yan, G. Zou, A. Wu, J. Ren, J. Yan, A. Hu and Y. Norman Zhou:

J. Phys. Conf. Ser. 379 (2012) 012024.9) J. Yan, G. Zou, Y. Zhang, J. Li, L. Liu, A. Wu and Y. Norman Zhou:

Mater. Trans. 54 (2013) 879­883.10) Y. Akada, H. Tatsumi, T. Yamaguchi, A. Hirose, T. Morita and E. Ide:

Mater. Trans. 49 (2008) 1537­1545.11) H. Jiang, K. Moon and C. P. Wong: 2005 10th Int. Symp. on Advanced

Packaging Materials, Processes, Properties and Interfaces, (2005).12) D. Wakuda, C.-J. Kim, K.-S. Kim and K. Suganuma: 2nd Electronics

System Integration Technology Conference, (2008) pp. 909­914.13) Y. Li: Master Thesis., National Tsinghua University, (2006).14) M. Maruyama, R. Matsubayashi, H. Iwakuro, S. Isoda and T. Komatsu:

Appl. Phys. A 93 (2008) 467­470.15) L.-ur Rahman, R. Qureshi, M. M. Yasinzai and A. Shah: C. R. Chim.

15 (2012) 533­538.16) G. Zou, J. Yan, F. Mu, A. Wu, H. Bai, J. Ren, A. Hu and Y. Norman

Zhou: The Open Surface Journal (2011) 70­75.17) H. Nishikawa, T. Hirano, T. Takemoto and N. Terada: The Open

Surface Science Journal (2011) 60­64.18) N. R. Kim, K. Shin, I. Jung, M. Shim and H. Mo Lee: J. Phys. Chem. C

118 (2014) 26324­26331.

Appendix

Ag� e� ! Agþ

Agþ þ OH� ! AgOH

2AgOH ! Ag2Oþ H2O

Cu� 2e� ! Cu2þ

Cu2þ þ 2OH� ! CuðOHÞ2CuðOHÞ2 ! CuOþ H2O

Fig. 10 Effect of bonding time on joint strength when sintered at 250°C,5MPa.

D. Zhang et al.1256