sign in sign up
<<
Home , Partial pressure

Materials Transactions, Vol. 45, No. 3 (2004) pp. 625 to 629 Special Issue on Lead-Free Soldering in Electronics #2004 The Japan Institute of Metals

Effect of on of Ag-Sn Alloys

Joonho Lee1, Toshihiro Tanaka1, Masaya Yamamoto2 and Shigeta Hara1

1Department of Materials Science and Processing, Osaka University, Suita 565-0871, Japan 2Formerly Graduate Student, Department of Materials Science and Processing, Osaka University, Suita 565-0871, Japan

The effect of oxygen on the surface tension of liquid Ag-Sn alloys used as one of the main components in lead-free soldering alloys was measured by the sessile drop method. The surface tension of liquid Ag-Sn alloys decreased with increasing tin content. The effect of oxygen was investigated for oxygen partial ranging from about 1017 to 1012 atm. The effect of oxygen on the surface tension of liquid Ag-Sn alloys is closer to pure tin rather than pure silver. Thermodynamic simulation using Butler’s equation showed that the surface of liquid Ag-Sn alloys is enriched with tin. Accordingly, the effect of oxygen adsorption on the surface tension of liquid Ag-Sn alloys was considered to be closer to that of tin than silver.

(Received September 22, 2003; Accepted December 2, 2003) Keywords: surface tension; liquid silver-tin alloys; sessile drop method; lead-free soldering

1. Introduction In the present work, the surface tension of liquid Ag-Sn alloys, one of the main components in lead-free solders, was Since the surface tension of liquid solder alloy determines measured by the sessile drop method with varying oxygen its wetting behavior, many researchers have investigated the partial pressure in atmosphere. In order to investigate the surface tension of the lead-free soldering alloys (e.g. Ag-Sn surface tension in the whole composition range, the exper- after Moser et al.1)). Although oxygen in atmosphere is imental was determined as 1253 K. Then, the known to decrease the surface tension of liquid metals and oxygen adsorption behavior on the surface of liquid Ag-Sn alloys,2–10) studies on the effect of oxygen for lead-free alloys was interpreted by the surface calculated solders have been only restricted to pure metals.8–10) using thermodynamic database. Generally, Gibbs adsorption isotherm (1) may express the 11) oxygen adsorption (O) on liquid metals and alloys. 2. Experimental 1 d 2 d O ¼ ¼ ð1Þ The experimental apparatus used for the measurements of RT d ln a RT d ln p O O2 the surface tension of liquid Ag-Sn alloys consisted of an where R: universal constant, T: temperature, : surface image capturing system with a He-Ne laser and a SiC heating tensions of liquid metal or alloy, aO: oxygen activity in the element furnace. Figure 1 shows a schematic diagram of the metal or alloy and pO2 : oxygen partial pressure, respectively. experimental setup. A quartz reaction tube (5.5 cm O.D., Hence, the information of the surface tension dependence on 5.0 cm I.D., 20.0 cm high) with investigation windows from a the oxygen partial pressure reflects the oxygen adsorption horizontal direction was used. The tube was sealed by means behavior in the surface of liquid metals and alloys. For of O-rings fitted onto copper water-cooled jackets at the ends example, if the surface of liquid metals and alloys is saturated of the tube. High purity samples (Ag: 99.99+% and Sn: with oxygen, d=d ln pO2 have a constant slope, yielding the 99.999+%) installed in an alumina crucible with a hole surface excess concentration of oxygen (oxygen adsorption) at saturation. The oxygen adsorption for pure metals has been Gas inlet investigated by many researchers. Table 1 summarizes some Alumina rod of the reported oxygen adsorption at saturation for pure 9) Cooling water inlet metals after Bernard and Lupis. However, there is no research on the oxygen adsorption for lead-free solder alloys siliconit as far as the authors concern. Alumina crucible

Table 1 Saturation parameters of oxygen at the surface of pure liquid metals.9)

Maximum surface excess Area per adsorbed , He-Ne laser CCD camera Temp. System concentration, S/10 2 nm2 Sample alloy Alumina substrate T/K 6 2 O.sat/10 molm experiment calculation Copper jacket Thermocouple Ag 1253 5 33 23/38þ Gas outlet Connected to Cu 1423 9.3 17.8 18/30þ a vacuum pump þ Fe 1823 18:123:47:19:2 8 Fig. 1 Schematic diagram of experimental apparatus for the sessile drop (100) face and þ(111) face of the compound measurements of surface tension. 626 J. Lee, T. Tanaka, M. Yamamoto and S. Hara

fitted with an optical zoom lens and filters, where a pixel Alumina tube corresponds to 18.7166 mm with the CCD camera. The threshold shape of the drop was determined using image Alumina analysis software. Then, this drop profile was used to Glass determine the surface tension by comparing the observed Alumina particles shape with that of the to the Laplace equation using a computer program developed by Krylov et al.13) Reference electrode (Metal/Metal oxide, p ref . ) O2 3. Results and Discussion Solid electrolyte tube (ZrO –8mol%Y O ) 2 2 3 3.1 Surface tension of liquid Ag-Sn alloys Platinum wire Initially the method was applied to liquid Ag-Sn alloys in

Gas ( p O ) 2 Ar-10%H2 atmosphere. In Fig. 3, the experimental results are plotted as a function of concentration of tin in the bulk phase. Fig. 2 Schematic diagram of zirconia oxygen sensor. For comparison, the surface tension of pure silver and tin measured by Bernard and Lupis9) (1253 K), and Taimatsu and Sangiorgi10) (1273 K) are also plotted. In addition, the (1.4 mm) at the bottom side, was set about 15 mm above an experimental results for alloys were compared with the alumina (sapphire) substrate. The alumina substrate was set calculated results based on Butler’s eq. (6) using thermody- in the uniform-temperature area of the furnace, and accu- namic database. rately leveled by adjusting three poles sustaining the furnace. ! S The temperature was measured with a Pt/Pt-13%Rh thermo- RT 1 NAg ¼ Sn þ ln B couple, which was set directly under the substrate. After the SSn 1 NAg sample was placed in the furnace, the quartz tube was 1 Ex,S S Ex,B B evacuated with a mechanical pump producing a vacuum of þ GSn ðT; NAgÞGSn ðT; NAgÞ SSn 1 Pa and filled with high-purity Ar-10%H2 gas . This ! procedure was repeated for three times to eliminate residual RT NS ¼ þ ln Ag oxygen in the reaction tube. Then, the furnace was Ag B SAg NAg programmed for the heating cycle up to 1253 K in 3 h. When 1 temperature in the furnace was stabilized, the sample alloys Ex,S S Ex,B B þ GAg ðT; NAgÞGAg ðT; NAgÞ ð6Þ in the alumina crucible were dropped onto the alumina SAg substrate using an alumina rod. During experiments, the where R: universal gas constant, T: temperature, , : oxygen partial pressure was controlled with introducing a Sn Ag surface tensions of pure liquid tin and silver, respectively, suitable mixture of purified Ar-10%H and CO . In 2 2 S , S : molar surface areas of pure liquid tin and silver, order to confirm the oxygen potential in the gas, zirconia Sn Ag S S ¼ S oxygen sensor equipped with reference electrodes (Fe/FeO, NAg: fraction of silver in the surface (NSn 1 NAg), B B B Ni/NiO and Cu/Cu2O couples) was placed in the furnace. A NAg: of silver in the bulk (NSn ¼ 1 NAg). schematic illustration of the oxygen sensor is shown in Fig. 2. Here, the ‘‘surface’’ is defined as outermost monolayer. Ex,S S When the effect of the electronic conduction is negligible, Gi ðT; Ni Þ: partial excess Gibbs energy of i in the surface S Ex,B B from the electromotive (E) expressed by eq. (2) one can as a function of T and Ni , Gi ðT; Ni Þ: partial excess Gibbs obtain the oxygen potential in the atmosphere (p ). ! O2

RT pO2 E ¼ ln ð2Þ 900 4F pref. Calculation O2 Experiment where F: Faraday’s constant, pref.: the oxygen potential in the -1 Bernarad et al. O2 800 reference electrode. Here, the oxygen potential in the Taimatsu et al.

reference electrodes was determined using reported values /mN m σ below.12) (pref. ¼ expðG=RT)) O2 700 Fe þ 1/2O2 ¼ FeO G ¼263:4 þ 64:81T ðkJÞð3Þ Ni þ 1/2O2 ¼ NiO G ¼233:6 þ 83:94 ðkJÞð4Þ 600 2Cu þ 1/2O2 ¼ Cu2O G ¼166:9 þ 71:13 ðkJÞð5Þ

Preliminary experiments showed that the equilibrium oxygen Surface Tension, 500 potential was obtained in 60 min in the present experimental condition. Surface tension measurements, therefore, were 0.0 0.2 0.4 0.6 0.8 1.0 conducted after 150 min to confirm the oxygen partial Concentration in the Bulk, N B pressure. In order to capture a clear drop shadow image, a Sn He-Ne laser ( ¼ 632:8 nm) was used. The drop shadow Fig. 3 Change in surface tension of liquid Ag-Sn alloys as a function of tin profile was taken using a CCD camera (512 512 pixels) content at 1253 K. Effect of Oxygen on Surface Tension of Liquid Ag-Sn Alloys 627

Table 2 Thermo-physical data and partial excess Gibbs energies for the calculation of the surface tension of Sn-Ag alloys.14,17,18)

¼ 1 Sn 601 00886T i/mN m Ag ¼ 1207 0:2279T ¼ 6f þ 4ð Þg 3 1 VSn 17:0 10 1 0:87 10 T 504:99 Vi/m mol 6 4 VAg ¼ 11:6 10 f1 þ 0:98 10 ðT 1233:65Þg Ex G / NAgNSnfð3902:15 4:96927TÞþð16974:05 þ 7:424515TÞðNAg NSnÞ 1 2 3 Jmol þð14299:05 þ 10:67712TÞðNAg NSnÞ þð5979:25 þ 6:497125TÞðNAg NSnÞ g

B Table 3 The maximum limit of oxygen partial pressure forming SnO . energy of i in the bulk as a function of T and Ni . The molar 2 15,16) surface area of component i can be obtained by eq. (7). 28) 8 Composition, at% aSn pO2 /10 Pa 23 1=3 2=3 Ag 0 — Si ¼ 1:091 ð6:02 10 Þ Vi ð7Þ Ex,B B Ag-10%Sn 0.087 1.78 where Vi is molar of the element i. Gi ðT; Ni Þ can be obtained directly from thermodynamic database, and Ag-20%Sn 0.132 3.18 Ex,S S Ag-30%Sn 0.183 2.29 Gi ðT; Ni Þ is obtained using eq. (8) assuming that the partial excess Gibbs energy in the bulk and the surface have Ag-40%Sn 0.253 1.66 the same concentration dependence. Sn 1 0.419 GEx,SðT; NSÞ¼0:83 GEx,BðT; NSÞ15Þ ð8Þ i i i i the calculations. Thermo-physical data and excess Gibbs energy used in the present calculation are listed in Table 2.14,17,18) This method 3.2 Surface tension of liquid Ag-Sn-O alloys has already been applied to various alloy and slag systems by 3.2.1 Thermodynamics of Ag-Sn-O system the authors with succeed.15,16,19–27) As shown in Fig. 3, the In order to understand oxygen behavior on the surface of agreement between calculations (solid line) and measure- Ag-Sn alloys, it is necessary to calculate the thermodynamic ments (solid circles) in the present work is also reasonable. limits for experiments (no oxides formation). In liquid Ag-Sn Accordingly, the surface tension of liquid Ag-Sn alloys at alloys, SnO2 is preferentially formed by the reaction (9). different could be obtained by the above Sn(l) þ O (g) ¼ SnO (s) ð9Þ calculation method. 2 2 The thermodynamic calculations using eq. (6) also suggest where the equilibrium constant of eq. (9) is expressed as eq. us information on the surface concentration of tin as shown in (10). Fig. 4. Even though the actual surface concentration of tin 1 would not be exactly the same plotted in Fig. 4 due to the K ¼ ¼ 2:415 1013 ðat 1253 KÞ28Þ ð10Þ a p simple assumption of the surface by a monolayer, it may be Sn O2 acceptable that tin preferentially adsorbs on the surface of Using the reported thermodynamic data for the activity of tin liquid Ag-Sn alloys. In Fig. 4, the surface concentration at in the alloys,29) the maximum limit of oxygen partial pressure 1) 493 K (the eutectic temperature of Sn-3.8%Ag solder alloy ) to prohibit the formation of SnO2 is calculated. In Table 3, is also plotted for comparison. The surface concentration of the calculated results of thermodynamic limits for experi- tin dramatically increased with decreasing temperature from ments are shown. 3.2.2 Effect of oxygen on surface tension of liquid Ag-Sn 1.0 alloys Figure 5 shows the measured surface tensions for pure

S 493K silver and tin and Ag-5, 10, 20, 30, 40%Sn alloys as a Sn 0.8 N 1253K function of logarithm of the oxygen partial pressure. The dependence of the surface tension for liquid Ag-Sn alloys on

0.6 the oxygen potential (the slope of d=d ln pO2 in Fig. 5) was found different from those for pure tin or silver. Nevertheless, it is noteworthy that the oxygen potential dependence on the 0.4 surface tension of liquid Ag-Sn alloys was much closer to pure tin even for the samples containing 10% tin. As described previously, when d=d ln pO has a constant 0.2 2 slope, we can determine the oxygen adsorption at saturation in the surface. Table 4 shows the maximum surface excess Concentraion in the Surface, 0.0 concentration of oxygen and the minimum area occupied by 0.0 0.2 0.4 0.6 0.8 1.0 single oxygen atom at saturation on the liquid metals and B alloys. Here, the minimum area (S) occupied by an oxygen Concentration in the Bulk, NSn atom was calculated from eq. (11). Fig. 4 Relationship between the of tin in the bulk and the S ¼ 1=ðN Þð11Þ surface at 493 and 1253 K in liquid Ag-Sn alloys. A O,sat. 628 J. Lee, T. Tanaka, M. Yamamoto and S. Hara

tin and Ag-Sn alloys are much larger than that of pure silver. As shown in Table 1, the minimum area occupied by single 900 oxygen atom at saturation on the surface of metal M is close -1 to those for oxide compound MnO at about the melting 800 temperature of the metals. Taimatsu et al.10) and Yuan et al.8) Bernard et al.

/mN m reported that the dependence of oxygen partial pressure on σ 700 the surface tension of pure tin increased with lowering Ag Ag-5% Sn temperature. Hence, we may consider that on the surface of 600 Ag-10% Sn liquid Ag-Sn alloys segregated with tin, oxygen adsorption Ag-20% Sn would decrease with increasing temperature. Probably, the Ag-30% Sn reason for the increase in the minimum area occupied by 500 Ag-40% Sn

Surface Tension, single oxygen atom in the surface of pure tin and Ag-Sn Sn Taimatsu et al. alloys is that the experimental temperature of the present 400 work is much higher than the melting temperature of pure tin. -15 -10 -5 0 In addition, we may expect that the oxygen adsorption on the log pO / atm 2 surface of liquid Ag-Sn alloys at much lower temperature (for example, eutectic temperature of Sn-3.8%Ag alloy, 493 K as Fig. 5 Surface tension of liquid Ag-Sn alloys as a function of logarithm of shown in Fig. 4) would be close to that of pure tin rather than oxygen partial pressure at 1253 K. pure silver, because the surface segregation of tin increases with decreasing temperature as shown in Fig. 4. Conse- Table 4 The maximum surface excess concentration of oxygen and the quently, the effect of the oxygen on the surface tension of Sn- minimum area occupied by single oxygen atom at saturation on liquid Ag- Ag based soldering alloy, of which silver content is not so Sn alloys. much, is estimated to be almost the same as that for pure tin Maximum surface excess reported by Yuan et al.8) near the liquidus temperature. Area per adsorbed atom, Composition, at% concentration, O,sat/ S/102 nm2 107 molm2 4. Conclusions 36.0 Ag 46.7 329) In the present work, the effect of oxygen on the surface Ag-10%Sn 9.24 180 tension of liquid Ag-Sn alloys at 1253 K was investigated using the sessile drop method. The followings are the major Ag-20%Sn 8.49 196 conclusions of the present work. Ag-30%Sn 7.01 237 (1) The surface tension of liquid Ag-Sn alloys decreased Ag-40%Sn 6.37 261 with increasing the concentration of tin, and the 460 Sn 3.61 experimental results were reasonably accord with the 69010) calculated results using Butler’s equations. (2) The dependence of the surface tension for liquid Ag-Sn alloys on the oxygen potential was found different from where NA and O,sat denote Avogadro number and the those for pure tin or silver. The oxygen potential maximum surface excess concentration of oxygen at satu- dependence for liquid Ag-Sn alloys was much closer to ration, respectively. Even though the present results for pure pure tin even for the samples containing 10% tin. metals slightly differ from those after previous research- ers,9,10) the tendency of the variation with metals is accept- REFERENCES able, i.e., oxygen adsorption for pure silver was relatively higher than those for pure tin and Ag-Sn alloys. Thereafter, 1) Z. Moser, W. Gasior and J. Pstrus: J. Phase Equil. 22 (2001) 254–258. the minimum area occupied by single oxygen atom at 2) T. Tanaka and S. Hara: Steel Res. 72 (2001) 439–445. 2 3) I. Jimbo and A. W. Cramb: ISIJ Int. 32 (1992) 26–35. saturation for pure silver (0.36 nm ) has much smaller value 4) Z. Jun and K. Mukai: ISIJ Int. 38 (1998) 1039–1044. 2 than those for pure tin (4.60 nm ) and Ag-Sn alloys 5) A. Sharan and A. W. Cramb: Metall. Mater. Trans. B. 28B (1997) 465– (1:802:61 nm2). Since tin preferentially adsorbs on the 472. surface of liquid Ag-Sn alloys as predicted in Fig. 4, the 6) K. Mukai, Z. Li and M. Zeze: Mater. Trans. 43 (2002) 1724–1731. oxygen adsorption on liquid Ag-Sn alloys is considered 7) B. Gallios and C. H. P. Lupis: Metall. Trans. B. 12B (1981) 549–557. 8) Z. Yuan, K. Mukai, K. Takagi and M. Ohtaka: J. Japan Inst. Metals 65 closer to that of tin rather than silver. Many researchers have (2001) 21–28. shown that the minimum area occupied by single oxygen 9) G. Bernard and C. H. P. Lupis: Metall. Trans. 2 (1971) 2991–2998. atom at saturation on the surface of metal M is close to those 10) H. Taimatsu and R. Sangiorgi: Surf. Sci. 261 (1992) 375–381. 2,9,26,27,30,31) for oxide compound MnO. If we assume Ag2O 11) G. R. Belton: Can. Metall. Q. 21 (1982) 137–143. structure is formed on the surface of liquid silver as suggested 12) R. A. Rapp: Physicochemical Measurements in Metals Research, (New 9) York, Interscience Publishers, 1970) Part. 2. by Bernard and Lupis, the minimum areas occupied by 13) A. S. Krylov, A. V. Vvedensky, A. M. Katsunelson and A. E. single oxygen atom on (100) and (111) faces, respectively, Tugovigov: J. Non-Crystal 156–158 (1993) 845–848. equal to 0.32 and 0.38 nm2, which are very close to the 14) T. Iida and R. I. L. Guthrie: The Physical Properties of Liquid Metals, present result (0.36 nm2). On the other hand, the minimum (Clarendon press, Oxford, 1993). area occupied by single oxygen atom in the surface of pure 15) T. Tanaka and T. Iida: Steel Res. 65 (1994) 21–28. Effect of Oxygen on Surface Tension of Liquid Ag-Sn Alloys 629

16) T. Tanaka, K. Hack, T. Iida and S. Hara: Z. Metallkd. 87 (1996) 380– 25) T. Tanaka, M. Matsuda, K. Nakao, Y. Katayama, D. Kaneko, S. Hara, 389. X. Xing and Z. Qiao: Z. Metallkd. 92 (2001) 1242–1246. 17) P.-Y. Chevalier: Thermochim. Acta 136 (1988) 45–54. 26) J. Lee and K. Morita: ISIJ Int. 42 (2002) 588–594. 18) W. Shimoda, J. Lee and T. Tanaka: Collected Abstracts of the 2003 27) J. Lee and K. Morita: Steel Res. 73 (2002) 367–372. Autumn Meeting of The Japan Institute of Metals, 373. 28) I. Barin: Thermochemical Data of Pure Substances Part II (1989) 19) T. Tanaka, S. Hara, M. Ogawa and T. Ueda: Z. Metallkd. 89 (1998) p 1404. 368–374. 29) R. Hultgen, P. D. Desai, D. T. Hawkins, M. Gleiser and K. K. Kelley: 20) T. Tanaka, K. Hack and S. Hara: MRS Bulletin 24 (1999) 45–50. Selected Values of the Thermodynamic Properties of Binary Alloys, 21) T. Tanaka and S. Hara: Electrochemistry 67 (1999) 573–580. (ASM, Ohio, 1973) 110. 22) T. Tanaka and S. Hara: Z. Metallkd. 90 (1999) 348–354. 30) K. Monma and H. Suto: J. Japan Inst. Metals 24 (1960) 377–379. 23) T. Ueda, T. Tanaka and S. Hara: Z. Metallkd. 90 (1999) 342–347. 31) J. P. Hajra, M. G. Frohberg and H.-K. Lee: Z. Metallkd. 82 (1991) 718– 24) T. Tanaka, K. Hack and S. Hara: Calphad 24 (2001) 463–474. 720.