Bondability of Copper Joints Formed Using a Mixed Pasteof Ag2O and CuO for Low-Temperature Sinter Bonding

Tomo Ogura+, Tomohiro Yagishita, Shinya Takata, Tomoyuki Fujimoto and Akio Hirose

Division of Materials and Manufacturing Science, Graduate School of Engineering, Osaka University, Suita 565-0871, Japan

The bondability of copper joints formed using a mixed paste of silver oxide (Ag2O) and copper oxide (CuO) that contained reducingsolvents was evaluated in order to achieve bonds that exhibited high migration tolerance and could serve as Pb-free alternatives to theconventional bonds formed using high-melting point solders in electronics packaging. The Ag2O particles reduced into silver nanoparticles at150°C, whereas the CuO reduced into copper nanoparticles about 300°C. The joints formed using the Ag2O/CuO mixed paste, when heated tothe appropriate levels, exhibited bondability superior to that of conventional Pb­5Sn joints. The oxide film formed on the copper substrate wasreduced by the combustion of polyethylene glycol 400, and bonding was achieved between the sintered layer and the copper substrate. A longerperiod resulted in the oxidisation of a few layers of sintered copper layers into Cu2O. The ion-migration tolerance of the Ag2O/CuO mixed pastewas approximately four times that of a layer of pure sintered silver. [doi:10.2320/matertrans.MD201202]

(Received November 30, 2012; Accepted February 18, 2013; Published April 5, 2013)

Keywords: silver oxide, copper oxide, ion migration, sintering, nanoparticles, copper joint

1. Introduction

It is desirable that electronic devices not only workefficiently at elevated temperatures but also consume lesserpower when packaged. However, bonding technology thatcan meet these requirements needs to be developed. Inparticular, there is a pressing demand for suitable alternativesto lead-rich high-melting-point solders, which contain chemi-cal substances that are hazardous to the environment. Thedifficulty in developing lead-free solders for use at hightemperatures arises from the requirement that the metallicbonds between mounted parts of semiconductor chips andvarious wiring connections exhibit long-term reliability.The mounted components of semiconductor chips (i.e.,die-bonded parts) generate considerable heat and thus needimproved thermal resistance. In addition, the dischargecharacteristics of certain parts of semiconductor chips mustbe maintained to enable the stable operation of the chips.1,2)

It is anticipated that the use of nanoparticles as bondingmaterials in die-attach technologies will provide feasiblealternatives to the current microsoldering techniques thatrequire the use of high-temperature solders such as Pb­10Snor Pb­5Sn. This is because particles smaller than 10 nm insize exhibit characteristics significantly different from thoseof the bulk state.3­6) This phenomenon is owing to the largesurface energy of the nanoparticles. If activated nanoparticlescould be made to affect the surface atoms of the bulk metalbeing bonded, metal-to-metal bonding using the nano-particles as filler material could be achieved at significantlylower temperatures than those in the case of conventionalfusion welding or diffusion bonding. This unique propertyof nanoparticles could be exploited in bonding materials inelectronic devices in order to achieve the improved thermalcharacteristics discussed above.

The use of nanoparticles as bonding materials for die-attach technologies is attracting significant research interest,and there have been numerous studies exploring nanoparticlepastes for use in low-temperature sintering.7­11) Researchers

have proposed a novel bonding process that involves silvernanoparticles.2,12­15) In addition, silver nanoparticles havebeen formed in-situ from silver oxide particles,1,16­20) andthese silver nanoparticles are being explored as an alternativeto the high-temperature solders currently used, such as Pb­10Sn or Pb­5Sn. Researchers have also studied the dynamicsof the initial stage of the sintering of nanoparticles on metalsubstrates using molecular dynamics (MD) simulationstudies.21) When bonding techniques involving the sinteringof nanoparticles are used in the assembly of fine pitchelectronic devices, attention must be paid to the ion-migrationtolerance of the bonded circuits, since a sintered layer of pureAg is formed on them.22) It has been found that the ion-migration tolerances of the sintered layers improve with theaddition of gold and palladium particles in the form of asecond metal to the Ag2O paste.19) However, these metals areexpensive, resulting in an increase in costs, whereas, from anindustrial point of view, it is important to devise sinteringmethods that are cheaper.

In this study, CuO was chosen as an alternative to gold andpalladium and used so as to add a second metal to the Ag2Opaste. This was done in the hope that CuO would be reducedby the solvent during bonding, and the formed coppernanoparticles would get distributed in the sintered silver,preventing the migration of the silver atoms. Yan et al. alsoreported the addition of cupper to silver improves bondabilityof joints using Cu­Ag mixed nanoparticles.23) The bond-ability of the joints formed using a reducing-solvents-containing mixed paste of silver oxide (Ag2O) and copperoxide (CuO) was investigated.

2. Experimental Procedure

Ag2O particles and CuO particles were prepared as shownin Fig. 1. The Ag2O particles and CuO particles were mixedin the ratio of 9 : 1 (in mass%) in the combined solvents ofpolyethylene glycol (PEG) 400 and a viscous solvent, used ina concentration of 180 µl/g, and processed into a paste forbonding. A copper substrate was used as the material to bebonded. The paste was applied to a lower sample that was+Corresponding author, E-mail: tomo.ogura@mapse.eng.osaka-u.ac.jp

Materials Transactions, Vol. 54, No. 6 (2013) pp. 860 to 865Special Issue on Nanojoining and Microjoining©2013 The Japan Institute of Metals and Materials

10mm in diameter and 25mm in length and then preheated to80°C. Next, an upper sample measuring 5mm in diameterand 15mm in length was placed on top of the lower sampleand bonded at temperatures ranging from 300 to 400°C. Thebonding was carried out in an infrared furnace for 5minunder pressure of 5MPa, applied using a dead weightmounted on top of the upper sample. The joint strength of theformed bond was evaluated using a tensile test performed at across-head speed of 1mm/min. The thermal characteristics ofthe combined paste were determined using a combination ofthermogravimetric (TG) and differential thermal (DT) analy-ses, performed using a heating rate of 10°C/min in ambientatmosphere. The cross section of the joint was observed usingtransmission electron microscopy (TEM). The ion-migrationtolerance of the sintered layer was evaluated using thepreviously employed water drop method.19) The bondingpastes were applied to a glass substrate. Then, the pastessintered at 300°C for 300 s under pressure after preheated. A10µl droplet of distilled water was placed on the gap betweenthe sintered samples, and a voltage of 3V was applied acrossthe samples. On the application of the voltage, silver ionsmigrated from the anode to the cathode, and silver dendritesgrew in the direction of the anode. The time taken by thesilver dendrites to reach the anode was measured.

3. Results and Discussion

3.1 Reduction behaviour of the Ag2O/CuO mixed pasteFigure 2 shows TG-DT analyses curves of the mixed

paste of Ag2O/CuO. Two exothermal peaks, at 150°C and at

300°C, were observed. A corresponding loss in the weightof the paste sample was also confirmed. Figure 3 shows theX-ray diffraction (XRD) patterns of the Ag2O/CuO mixedpaste samples that had been heated to 150 and 300°C. TheXRD pattern of the paste sample heated to 150°C showedpeaks attributable to silver and CuO, whereas the XRDpattern of the paste sample heated to 300°C showed peaksattributable to silver and copper. On the basis of the TG-DTanalyses curves and the XRD patterns of the paste samples,it was determined that Ag2O particles reduced into silvernanoparticles at 150°C. In addition, CuO also reduced intocopper nanoparticles at about 300°C. Residual amounts ofthe PEG solvent remained in the bonding layer until thebonding temperature reached 150°C, at which point the redoxreaction reached completion.20) Therefore, the reduction ofCuO was considered to be proceeded by the residual PEGsolvent because PEG solvent did not evaporate and combustuntil the temperature reached 300°C.20)

1μm1μm

(a) (b)

Fig. 1 SEM images of (a) Ag2O and (b) CuO particles.

-25

-20

-15

-10

-5

0

-200

0

200

400

600

800

1000

0 100 200 300 400

DTA

TG

DTA

(μV

)

TG

(m

ass%

)

Temperature, T /°C

Fig. 2 TG-DTA curves of the Ag2O/CuO mixed paste.

Non heated

423K

Ag2O Ag CuO Cu

573K

(a)

(b)

Inte

nsity

(a.

u.)

40°

Diffraction angle, 2θ /degree

30° 50° 60°In

tens

ity (

a.u.

)

2θ /degree

37°36°35°

Fig. 3 (a) X-ray diffraction patterns of the Ag2O/CuO mixed pastesamples after being heated to various temperatures and (b) magnifiedversion of the pattern of the paste sample heated to 150°C. The image isof the area corresponding to 2ª = 35.

Bondability of Copper Joints Formed Using a Mixed Paste of Ag2O and CuO for Low-Temperature Sinter Bonding 861

3.2 Tensile strength of the jointsFigure 4(a) shows tensile strength of the joints formed

using the Ag2O/CuO mixed paste heated to 400°C. Thestrength of the joints could be measured at 250°C. At 300°C,the strength increased, owing to the formation and sinteringof copper nanoparticles and the evaporation and combustionof the residual PEG solvent. In general, the strength increasedwith increases in the temperature and the holding time (seeFig. 4(b)). Figure 5 shows the tensile strengths of jointsheated to various temperatures and held at those temperaturesfor 300 s. Again, the strength increased with an increase inthe bonding temperature, and all specimens exhibited hashigher strengths than those of gold-to-gold joints formedusing Pb­5Sn (29MPa), which is the current industrialstandard for die bonding. This showed clearly that when the

joints formed using the Ag2O/CuO mixed paste were heatedto appropriately high temperatures, the bonding achieved wassuperior to that obtained using Pb­5Sn.

The fractured surface of a joint after the tensile test isshown in Fig. 6. It was confirmed that some silver particlessintered with the copper substrate at 200°C (Fig. 6(a)). Thenumber of sintered particles present on the fractured surfaceincreased as with temperature (Figs. 6(b) and 6(c)). The

Holding time, t /s

0

10

20

30

40

50

60

0 60 120 180 240 300

(a)

(b)

0

10

20

30

40

50

60

100 150 200 250 300 350 400

Temperature, T /°C

Tens

ile s

tren

gth,

σ/M

PaTe

nsile

str

engt

h, σ

/MPa

Fig. 4 Tensile strengths of the joints formed using the Ag2O/CuO mixedpaste (a) heated till up to 400°C and (b) held at 400°C for various periods,with the bonding pressure being 5MPa.

0

10

20

30

40

50

60

300 350 400

Temperature, T /°C

Tens

ile s

tren

gth,

σ/M

Pa

Fig. 5 Tensile strengths of the joints formed using the Ag2O/CuO mixedpaste and heated to various temperatures, while being held at thetemperatures for 300 s. The bonding pressure used was 5MPa.

1mm5μm

1mm

5μm

1mm5μm

1mm5μm

(a)

(b)

(c)

(d)

Fig. 6 Optical and SEM images of fractured surfaces of the joints formedusing the Ag2O/CuO mixed paste with the bonding temperature being (a)200°C, (b) 300°C, (c) 400°C and (d) 400°C held at this temperature for300 s. The bonding pressure used was 5MPa.

T. Ogura, T. Yagishita, S. Takata, T. Fujimoto and A. Hirose862

dimple-like fracture region also increased in area (Figs. 6(c)and 6(d)), indicating that the sintering of not only silvernanoparticles but also of copper nanoparticles occurred afterthe reduction of CuO. In addition, with an increase in theholding time, the circumferential regions of the fracturedsurfaces of the joints were increasingly seen to contain ablack region, which is typically observed on the surface ofcopper oxide,20) showing that the oxidation of both thecopper substrate and sintered copper nanoparticles startedduring the heating process.

3.3 Microstructures of the interfaces of the jointsFigure 7 shows the back-scattered electron (BSE) images

and the corresponding results of electron probe micro-analyzer (EPMA)-based element mapping of the crosssections of the joints bonded using the Ag2O/CuO mixedpaste. While gold and palladium are concentrated in distinctphases in the sintered silver layer in joints formed using agold- and palladium-supplemented Ag2O paste,19) it wasfound that copper was uniformly distributed in the sinteredsilver layer in the joints formed using the Ag2O/CuO mixedpaste (Fig. 7(a)). Holding the joint for 300 s increased theconcentration of oxygen around copper, as shown inFig. 7(b). This increase in the oxygen concentration wasprobably owing to the formation of copper oxide. A layer ofcopper oxide was distinctly observable using TEM. Figure 8shows bright field and energy-dispersive X-ray spectroscopy(EDS) mapping images of the joint formed using the Ag2O/CuO mixed paste and heated to and held at 400°C for 300 s.A film of the natural oxide of copper was not observed,showing the bond formed was achieved between the sinteredlayer and the copper substrate. The oxide film on the coppersubstrate was reduced by the combustion of PEG, andthe residual PEG suppressed the oxidation of the coppersubstrate during bonding.20) The results of the EDS mappingconfirmed the presence of silver and of a copper layer,with the oxygen being present within the copper layer.A magnified image of the sintered layer along with thecorresponding SAED pattern and the results of the EDSanalysis of the layer are shown in Fig. 9 and Table 1. TheSAED patterns indicated the presence of silver, copper andCu2O. However, CuO was not detected. The formation ofCu2O was probably owing to the oxidation of a few layers ofsintered copper during the heating process.24) The results ofthe EDS analysis also suggested that a few sintered layerswere in the form of a solid solution of the silver and coppernanoparticles (point 1 in Table 1), indicating that some of thecopper nanoparticles were alloyed with the silver during theformation and sintering of the copper nanoparticles.

3.4 Ion-migration toleranceIn the water-drop test, Ag atoms of the anode were ionised

and made to pass through water towards the cathode under anelectric field. On reaching the cathode, these Ag precipitated,forming Ag dendrites. The Ag dendrites grew in size andeventually reached the anode. The “arrival time” for thedendrites, which was defined as from the time periodextending from the application of the 3V potential to themoment the Ag dendrites reached the anode, was used as ameasure of the ion-migration tolerance. Figure 10 shows the

arrival time plotted against the volume fraction of CuO in thesintered layer. The arrival time for the mixed paste wasapproximately four times as large as that of a layer of puresintered silver. Although the quantitative evaluation such asmigration rate is not clarified yet, thus, the addition of CuOimproved the ion-migration tolerance of the sintered silverlayer. There are several possible reasons this improvement in

Cu

C

Ag

O

BSE

Cu

C

Ag

O

BSE

5μm

(a)

(b)

5μm

Fig. 7 BSE images and corresponding EPMA mappings of the crosssections of joints heated to (a) 400°C and (b) 400°C and held at thistemperature for 300 s with the bonding pressure being 5MPa.

Bondability of Copper Joints Formed Using a Mixed Paste of Ag2O and CuO for Low-Temperature Sinter Bonding 863

the ion-migration tolerance. The addition of a second metalreduces the area fraction of silver on the surface of the anode.This reduction can result in a reduction in the amount ofionised silver atoms at the anode because copper migrates toa lesser degree than silver. During the migration of silveratoms, the concentration of the second metal increases atthe anode surface, preventing the further migration of silveratoms from the anode.19) Moreover, the diffusion of Agatoms through the sintered Ag layer may be disrupted owingto the presence of the sintered layer of the second metal.This disruption may be an additional mechanism owing towhich the second metal prevents the migration of silveratoms. The limiting additive fraction of CuO is not revealedyet and the further research is needed for better understandingof this point.

4. Conclusions

(1) Ag2O particles reduced into silver nanoparticles at150°C. Then, CuO reduced into copper nanoparticlesat about 300°C. The reduction of CuO was consideredto be attributable to the presence of the residual PEGsolvent.

(2) The joints formed using the Ag2O/CuO mixed paste,when heated to the appropriate levels, exhibited bond-ability superior to that of a conventional Pb­5Sn joint.

(3) Bonding was achieved between the sintered layer andthe copper substrate. The oxide film on the coppersubstrate was reduced by the combustion of PEG. Alonger heating period resulted in the oxidisation of a

B.F. Ag

Cu O

500nm

Interface

Fig. 8 Bright field and EDS mapping images of the joint formed using theAg2O/CuO paste. The joint was heated to 400°C and held at thistemperature for 300 s with a bonding pressure of 5MPa.

1

2

100nmCu substrate

(a)

Interface

Cu{220}

Ag{220} Cu2O{220}(b)

Cu{111}

Fig. 9 (a) Bright field image of a joint and (b) diffraction pattern of theencircled region of the joint.

0

100

200

300

400

500

600

0 10

Arr

ival

tim

e, t

/s

Volume fraction of CuO (in mass%)

Fig. 10 Time taken by the silver dendrites in the sintered layer to growfrom the anode and reach the cathode during the water-drop test. Thisparameter was termed the “arrival time” of the dendrites.

Table 1 Results of a quantitative EDS analysis of the area of the jointshown in the TEM image in Fig. 9(a).

Composition (%)

Ag Cu O

1 25.6 65.4 9.0

2 0.5 80.4 19.1

T. Ogura, T. Yagishita, S. Takata, T. Fujimoto and A. Hirose864

few layers of sintered copper into Cu2O. A few of thesintered layers acted as a solid solution of the silvernanoparticles and the copper nanoparticles.

(4) The ion-migration arrival time of the Ag2O/CuO mixedpaste was approximately four times that of pure sinteredsilver.

Acknowledgements

The authors would like to thank Prof. M. Takahashi ofOsaka University and Prof. T. Shibayanagi of OsakaUniversity (now at Toyama University) for their valuablehelp with the TEM observations, and thank Mr. K. Ohmitsufor his help with the EPMA operations. A part of this studywas performed under the Cooperative Research Program ofInstitute for Joining and Welding Research Institute, OsakaUniversity. This work was supported by a Grant-in-Aid forScientific Research (B) No. 23360322, Japan, and by PriorityAssistance for the Formation of Worldwide RenownedCenters of Research®The Global COE Program (Project:Center of Excellence for Advanced Structural and FunctionalMaterials Design) from the Ministry of Education, Culture,Sports, Science and Technology (MEXT), Japan.

REFERENCES

1) T. Morita, Y. Yasuda, E. Ide, Y. Akada and A. Hirose: Mater. Trans. 49(2008) 2875­2880.

2) T. Morita, E. Ide, Y. Yasuda, A. Hirose and K. F. Kobayashi: Jpn. J.Appl. Phys. 47 (2008) 6615­6622.

3) A. N. Goldstein, C. M. Echer and A. P. Alivisatos: Science 256 (1992)1425­1427.

4) P. Pawlow: Z. Phys. Chem. 65 (1909) 545­548.5) M. Takagi: J. Phys. Soc. Japan 9 (1954) 359­363.

6) N. Gladkich, R. Niedermayer and K. Spiegel: Phys. Status Solidi 15(1966) 181­192.

7) J. G. Bai and G. Q. Lu: IEEE Trans. Device Mater. Reliab. 6 (2006)436­441.

8) A. Hu, J. Y. Guo, H. Alarifi, G. Patane, Y. Zhou, G. Compagnini andC. X. Xu: Appl. Phys. Lett. 97 (2010) 153117.

9) Y. Morisada, T. Nagaoka, M. Fukusumi, Y. Kashiwagi, M. Yamamoto,M. Nakamoto, H. Kakiuchi and Y. Yoshida: J. Electron. Mater. 40(2011) 2398­2402.

10) K. Suganuma, S. Sakamoto, N. Kagami, D. Wakuda, K. S. Kim and M.Nogi: Microelectron. Reliab. 52 (2012) 375­380.

11) Y. Kobayashi, T. Shirochi, Y. Yasuda and T. Morita: Int. J. Adhes.Adhes. 33 (2012) 50­55.

12) Y. Akada, H. Tatsumi, T. Yamaguchi, A. Hirose, T. Morita and E. Ide:Mater. Trans. 49 (2008) 1537­1545.

13) E. Ide, S. Angata, A. Hirose and K. F. Kobayashi: Acta Mater. 53(2005) 2385­2393.

14) E. Ide, A. Hirose and K. F. Kobayashi: Mater. Trans. 47 (2006) 211­217.

15) Y. Zhou (ed.): Microjoining and nanojoining, (Woodhead PublishingLimited, Cambridge, 2008) pp. 250­268.

16) A. Hirose, H. Tatsumi, N. Takeda, Y. Akada, T. Ogura, E. Ide and T.Morita: J. Phys. Conf. Ser. 165 (2009) 012074.

17) T. Morita, Y. Yasuda, E. Ide and A. Hirose: Mater. Trans. 50 (2009)226­228.

18) A. Hirose, N. Takeda, Y. Konaka, H. Tatsumi, Y. Akada, T. Ogura, E.Ide and T. Morita: Mater. Sci. Forum 706­709 (2012) 2962­2967.

19) T. Ito, T. Ogura and A. Hirose: J. Electron. Mater. 41 (2012) 2573­2579.

20) S. Takata, T. Ogura and A. Hirose: J. Electron. Mater. 42 (2013) 507­515.

21) T. Ogura, M. Nishimura, H. Tatsumi, W. Takahara and A. Hirose:Mater. Trans. 53 (2012) 2085­2090.

22) E. Ide, T. Morita and Y. Kawachi: Proc. 24th JIEP Ann. Meeting,(2010) pp. 210­211.

23) J. Yan, G. Zou, A. Wu, J. Ren, A. Hu and N. Zhou: J. Electron. Mater.41 (2012) 1886­1892.

24) A. Njeh, T. Wieder and H. Fuess: Surf. Interf. Anal. 33 (2002) 626­628.

Bondability of Copper Joints Formed Using a Mixed Paste of Ag2O and CuO for Low-Temperature Sinter Bonding 865