• 沒有找到結果。

Thin-film reactions during diffusion soldering of Cu/Ti/Si and Au/Cu/Al2O3 with Sn interlayers

N/A
N/A
Protected

Academic year: 2021

Share "Thin-film reactions during diffusion soldering of Cu/Ti/Si and Au/Cu/Al2O3 with Sn interlayers"

Copied!
5
0
0

加載中.... (立即查看全文)

全文

(1)

952

INTRODUCTION

Diffusion soldering (also known as solid-liquid interdiffusion bonding) is a novel joining technique based on the principle of isothermal solidification. A low-melting, metallic thin-film interlayer is em-ployed in the process, which melts at low tempera-tures and reacts rapidly with both high-melting (HT1 and HT2) layers or with substrates to form

intermetallic compounds (IMCs). Because the IMCs formed at the interfaces possess much higher melt-ing points than the original low-meltmelt-ing interlayer, a special feature of bonding at lower temperatures and usage at higher temperatures can be achieved.1

Such a superior characteristic enables diffusion soldering to broaden its application potentials in the electronics industry.2–4

For the manufacturing of a ceramic multichip modulus, Si dice are attached to multilayer ceramic substrates.5 The bonding temperature for this die

attachment process must be lower than 400°C to avoid any damage to integrated circuit chips. Poly-mer adhesives, glass bonding, metallic soldering, and Au-Si eutectic bonding are the traditional meth-ods applied for this purpose. However, this causes strength degradation in the joining interfaces when the integrated circuit chips are functioning at elevated

temperatures. Given the advantage of the diffusion soldering technique in pairing high operation tem-perature with low bonding temtem-perature, its applica-bility in die attachment for high-density ceramic packages is, hence, examined.

According to the underlying principle of the diffu-sion soldering process, it is obvious that interfacial reactions play a key role in the joining efficiency of this technique. The effort of this study is thus con-cerned with the IMCs formed at the interfaces and their growth kinetics during the diffusion soldering of the multilayer thin-film systems bonded onto Si wafers and Al2O3substrates. In addition, the tensile

strengths of the diffusion-soldered specimens are evaluated.

EXPERIMENTAL

For the diffusion soldering of Si with Al2O3

sub-strates, Ti (20 nm), Cu (6m), and Sn (4 m) were sputter-deposited sequentially on a Si wafer. A Cu layer (4m) and an Au layer (6 m) were also de-posited on an Al2O3 substrate by sputtering. The

specimens with dimensions of 4 mm× 4 mm were cut with a diamond saw. The surfaces of the specimens were stripped with a deoxidized agent prior to diffu-sion soldering to remove any oxide film. The multi-layer thin-film specimens were then sandwiched, as shown in Fig. 1, and heated at various temperatures

Thin-Film Reactions during Diffusion Soldering

of Cu/Ti/Si and Au/Cu/Al

2

O

3

with Sn Interlayers

M.W. LIANG,1T.E. HSIEH,1S.Y. CHANG,2and T.H. CHUANG2,3

1.—Department of Materials Science and Engineering, National Chiao-Tung University, 300 Hsinchu, Taiwan, Republic of China. 2.—Institute of Materials Science and Engineering, National Taiwan University, 106 Taipei, Taiwan. 3.—E-mail: [email protected]

The multilayer thin-film systems of Cu/Ti/Si and Au/Cu/Al2O3were

diffusion-soldered at temperatures between 250°C and 400°C by inserting a Sn thin-film interlayer. Experimental results showed that a double layer of intermetallic compounds (IMCs) -(Cu0.99Au0.01)6Sn5/-(Au0.87Cu0.13)Sn was formed at the

interface. Kinetics analyses revealed that the growth of intermetallics was diffusion-controlled. The activation energies as calculated from Arrhenius plots of the growth rate constants for (Cu0.99Au0.01)6Sn5 and (Au0.87Cu0.13)Sn are

16.9 kJ/mol and 53.7 kJ/mol, respectively. Finally, a satisfactory tensile strength of 132 kg/cm2could be attained under the bonding condition of 300°C for 20 min. Key words: Diffusion soldering, die attachment, intermetallic compounds,

kinetics analysis, bonding strength

(2)

reactions,-Cu6Sn5has been found to react further

with Cu to form the ε-Cu3Sn IMC, which might be too

thin to be observed in this diffusion-soldering study. The chemical composition of the IMC adjacent to the Au layer is Au:Cu:Sn 44.4:6.9:48.7, i.e., (Au0.87Cu0.13)Sn, corresponding to the -AuSn phase

on the Au-Sn phase diagram. The relatively high Cu content of this -AuSn intermetallic phase is attrib-uted to the rapid diffusion of Cu into the Au layer in Au/Cu/Al2O3, which simultaneously participates

in the interfacial reaction between Au and the Sn interlayer during the diffusion soldering process. For the soldering reaction between liquid Sn and the Au substrate, the interfacial IMC most commonly formed is the AuSn4phase.7The appearance of the

-AuSn phase can be attributed to a further reaction of AuSn4with the Au layer following the exhaustion

of the Sn interlayer.

The average thickness (x) of the intermetallic layers formed at the interface was measured and listed in Table I. The data are plotted against the square root of reaction time (t) and shown in Figs. 4 and 5 for the (Cu0.99Au0.01)6Sn5and (Au0.87Cu0.13)Sn

phases, respectively. In both cases, the growths of IMCs follow a parabolic rate law, implying that their reactions are diffusion-controlled. The growth rate constants (k x/t1/2) as calculated from Figs. 4 and 5 ranging from 250°C to 400°C in a vacuum furnace of

5.3× 10−4 Pa. After diffusion soldering, the speci-mens were cross-sectioned, ground with SiC paper, and polished with 1-m and 0.3-m Al2O3powders.

Morphology observations and growth rate measure-ments of the IMCs were mostly conducted via a scan-ning electron microscope. Chemical compositions of the IMCs were analyzed using an electron probe microanalyzer (EPMA). For the evaluation of bond-ing strengths of the diffusion-soldered specimens, tensile tests were conducted using a microforce tester at a crosshead speed of 0.01 mm/s.

RESULTS AND DISCUSSION

The typical morphology of the diffusion-soldered joints for the bonding of Cu/Ti/Si and Au/Cu/Al2O3

with Sn interlayers is shown in Fig. 2. The EPMA line profiles for Au, Sn, and Cu elements across the multi-layers of the diffusion-soldered specimen (Fig. 2) are plotted in Fig. 3. The Sn interlayer after diffusion soldering is eliminated and replaced by bilayered IMCs between Cu/Ti/Si and Au/Cu/Al2O3. The IMCs

adjacent to the Cu layer have a chemical composition (at.%) of Cu:Au:Sn 54.8:0.2:45, i.e., (Cu0.99Au0.01)6

Sn5, which corresponds to the-Cu6Sn5phase on the

Cu-Sn phase diagram. The -Cu6Sn5 intermetallic

phase has often been reported on in studies of Cu/Sn interfacial reactions.6 During the Cu/Sn soldering

Fig. 1. The scheme of diffusion soldering for Cu/Ti/Si and Au/Cu/ Al2O3with Sn interlayers.

Fig. 2. The morphology of IMCs formed after diffusion soldering between Cu/Ti/Si and Au/Cu/Al2O3 at 300°C for 20 min with Sn

interlayers.

Fig. 3. The Au, Sn, and Cu concentrations across the multilayers of the diffusion-soldered specimen (Fig. 2).

(3)

are given in Table II. From the Arrhenius diagram of ln k versus 1/T, as shown in Fig. 6, the activation energies (Q) for the growths of (Cu0.99Au0.01)6Sn5

and (Au0.87Cu0.13)Sn IMCs can be determined as

16.9 kJ/mol and 53.7 kJ/mol, respectively. The for-mer value (16.9 kJ/mol) is quite consistent with the activation energy for the diffusion of Cu in liquid Sn (19.5 kJ/mol), as reported by Ma and Swalin.8It

implies that the rate-limiting step in the growth of the (Cu0.99Au0.01)6Sn5 intermetallic is the diffusion

of Cu dissolved near the intermetallic reaction front into the surrounding liquid-Sn thin film.

Tu and Thompson9reported that the growth of the Cu6Sn5IMC during the solid-state reaction between

Cu and Sn thin films at room temperature exhibited a linear rate. According to their discussion, the rate-limiting step should be the release of Cu atoms from

Table I. Thicknesses of Intermetallic Compounds Formed during Diffusion Soldering Between

Cu/Ti/Si and Au/Cu/Al2O3with Sn Interlayers

Temper- Time (Cu0.99Au0.01)6Sn5 (Au0.87Cu0.13)Sn

ature (°C) (min) (m) (m) 250 10 0.51 0.89 20 0.79 1.18 30 1.05 1.38 40 1.22 1.59 300 10 1.28 1.18 20 1.63 1.46 30 1.85 2.38 40 2.08 2.50 350 10 1.78 2.17 20 2.11 3.04 30 2.46 3.68 40 2.67 4.24 400 10 2.39 2.64 20 2.84 3.47 30 3.26 4.89

Fig. 4. The average thickness (x) of (Cu0.99Au0.01)6Sn5IMCs formed

during diffusion soldering between Cu/Ti/Si and Au/Cu/Al2O3with Sn

interlayers.

Fig. 5. The average thickness (x) of (Au0.87Cu0.13)Sn IMCs formed

during diffusion soldering between Cu/Ti/Si and Au/Cu/Al2O3with Sn

interlayers.

Table II. Growth Rate Constants of Intermetallic Compounds Formed during Diffusion Soldering

between Cu/Ti/Si and Au/Cu/Al2O3 with Sn Interlayers

Temperature (Cu0.99Au0.01)6Sn5 (Au0.87Cu0.13)Sn

(°C) (m/min1/2) (m/min1/2)

250 0.228 0.219

300 0.250 0.462

350 0.287 0.653

400 0.374 0.955

the Cu film, rather than the diffusion across Cu6Sn5.

In contrast to the thin-film reaction discussed by Tu and Thompson, Vianco et al.10studied the solid-state interfacial reaction between Cu and hot-dipped Sn at temperatures ranging from 70°C to 205°C. Parabolic Fig. 6. The Arrhenius plots of growth rate constants (k) for (Cu0.99Au0.01)6Sn5and (Au0.87Cu0.13)Sn IMCs formed during diffusion

(4)

found that an AuSn4IMC was formed under diffusion

control during the reaction. The activation energy was 40 kJ/mol, similar to the value of our measure-ment for the solid /liquid thin-film reaction of the Au/Sn system (53.7 kJ/mol). Their results sustain the view held by this present study that the growth of (Au0.87Cu0.13)Sn is controlled by a solid/solid

inter-facial reaction as discussed previously.

Tensile strengths of the specimens after diffusion soldering at various temperatures for 20 min are shown in Fig. 8. A maximum value of 132 kg/cm2is

attained at the bonding temperature of 300°C. As Table I indicates, such an optimized bonding condi-tion (300°C, 20 min) is conducive to the growth of (Cu0.99Au0.01)6Sn5and (Au0.87Cu0.13)Sn intermetallic

layers to the thicknesses of 1.63 m and 1.46 m, respectively.

growth kinetics for Cu6Sn5IMCs was reported. The

deviation of the reaction mechanism in Tu and Thompson’s study9from the research of Vianco et al.10

should be attributed to the ultra-thin-film specimens (Sn thickness under 0.5m) and the quite low reac-tion temperature (room temperature) adopted by the former study. For the solid-liquid interdiffusion be-tween Cu and Sn thin films at 240°C and 300°C, Bader et al. showed that the growth of the Cu6Sn5

IMC did not follow the parabolic growth law.1As was

explained by them, the deviation was attributed to the reduction of transport grooves resulting from the growth of the scallop-shaped Cu6Sn5IMCs. Because

the kinetic measurements obtained by Bader et al. were focused on the initial stage of reaction (reaction time: shorter than 2 min; intermetallic thickness: thinner than 2 m), it is believed that once all the grooves along the interface have “closed” after a longer reaction time as similar for our study (reaction time: 10–40 min), the reaction-controlling step in the growth of Cu6Sn5should turn out to be the diffusion

through the continuous Cu6Sn5 intermetallic layer.

Hayashi et al.11studied the soldering reactions

be-tween Cu and liquid Sn saturated with Cu and found that the growth of the Cu6Sn5 IMC was

diffusion-controlled with an activation energy of 29 kJ/mol, a value quite near our own kinetic measurement (16.9 kJ/mol). In summary of the preceding results, it can be implied that the solid-liquid interfacial reac-tions in the Cu/Sn thin film case are carried out simi-larly to a “normal” soldering reaction.

The calculated activation energy for the growth of the (Au0.87Cu0.13)Sn intermetallic (53.7 kJ/mol) is

close to the activation energy for the lattice diffu-sion of Au in Sn (C: 46.1 kJ/mol, C: 74.1 kJ/mol), as reported by Dyson.12 The growth of the

(Au0.87Cu0.13)Sn intermetallic is, therefore, believed

to be controlled by the lattice diffusion of Au through the solid IMC. The discrepancy in growth rate-controlling mechanisms for (Cu0.99Au0.01)6Sn5 and

(Au0.87Cu0.13)Sn intermetallics may be attributed to

the much quicker liquid/solid reaction for Sn(l)/Au(s)

than that for Sn(l)/Cu(s). This explanation can be

vali-dated by the quite different wettability of liquid Sn on Au and Cu substrates. Figure 7 shows that the con-tact angle of liquid Sn on an Au substrate decreases rapidly and vanishes at complete wetting (∼0°), while the contact angle in the case of a Cu substrate re-mains at about 30°, implying that it is much easier for liquid Sn to react with Au than with Cu. In other words, Au0.87Cu0.13 reacts rapidly during diffusion

soldering with Sn to form (Au0.87Cu0.13)Sn4. After

the thin-film Sn is exhausted, a further solid /solid interfacial reaction takes place: (Au0.87Cu0.13)Sn4+

3(Au0.87Cu0.13)→ 4(Au0.87Cu0.13)Sn. Because the

reac-tion efficiency is governed by the slow mode of the latter reaction, the rate-controlling mechanism for the growth of (Au0.87Cu0.13)Sn is, therefore, the

diffu-sion of Au in this solid intermetallic phase. The solid/solid interfacial reaction in the Au/Pb-Sn solder system has been studied by Hannech and Hall.13They

Fig. 7. The contact angles () of liquid Sn on the surfaces of Au and Cu substrates at 350°C. Sn(l)/Au(s)reacts much more quickly than

Sn(l)/Cu(s).

Fig. 8. The tensile strengths () of Cu/Ti/Si wafer diffusion-soldered with Au/Cu/Al2O3substrates at various temperatures for 20 min using

(5)

at the bonding temperature of 300°C, corresponding to the growth thicknesses of 1.63m and 1.46 m for (Cu0.99Au0.01)6Sn5and (Au0.87Cu0.13)Sn intermetallic

layers.

REFERENCES

1. S. Bader, W. Gust, and H. Hieber, Acta Metall. Mater. 43, 329 (1995).

2. L. Barnstein and H. Bartholomew, Trans. TMS-AIME 236, 405 (1966).

3. D.M. Jacobson and S.P.S. Sangha, IEEE Trans. Comp.,

Packag. Manuf. Technol. 21, 515 (1998).

4. C.C. Lee, C.Y. Wang, and G. Matijasevic, J. Electron.

Packag. 115, 201 (1993).

5. J.J. Licari, Multichip Module Design, Fabrication and

Testing (New York, McGraw-Hill, 1995), pp. 137–203.

6. K.N. Tu, Mater. Chem. Phys. 46, 217 (1996). 7. D.H. Daebler, Surface Mount. Technol. 5, 43 (1991). 8. C.A. Ma and R.A. Swalin, Acta Metall. 8, 388 (1960). 9. K.N. Tu and R.D. Thompson, Acta Metall. 30, 947 (1982). 10. P.T. Vianco, P.F. Hlava, and A.C. Kilgo, J. Electron. Mater.

23, 583 (1994).

11. A. Hayashi, C.R. Kao, and Y.A. Chang, Scripta Mater. 37, 393 (1997).

12. B.F. Dyson, J. Appl. Phys. 37, 2375 (1966).

13. E.-B. Hannech and C.R. Hall, Mater. Sci. Technol. 8, 817 (1992).

CONCLUSIONS

To evaluate the applicability of the diffusion-soldering technology in the die attachment process for ceramic packages, the multilayer thin-film sys-tems of Cu/Ti/Si and Au/Cu/Al2O3were bonded with

Sn interlayers at various temperatures ranging from 250°C to 400°C. The IMCs formed after diffusion sol-dering as analyzed by EPMA are-(Cu0.99Au0.01)6Sn5

and -(Au0.87Cu0.13)Sn on the respective sides of Si

and Al2O3. The growth of both IMCs is

diffusion-controlled but with different rate-limiting steps. The activation energy for the (Cu0.99Au0.01)6Sn5

inter-metallic is 16.9 kJ/mol, which is near that for the dif-fusion of Cu in liquid Sn (19.5 kJ/mol). The growth of the (Cu0.99Au0.01)6Sn5 intermetallic is controlled by

the diffusion of dissolved Cu into the liquid-Sn thin film. The growth of (Au0.87Cu0.13)Sn presents an

acti-vation energy of 53.7 kJ/mol, a result in agreement with that for the lattice diffusion of Au in Sn (C: 46.1 kJ/mol, C: 74.1 kJ/mol), thus revealing that the rate-limiting step in the growth of (Au0.87Cu0.13)Sn is

the diffusion of Au through the IMC. Tensile tests for the specimens diffusion-soldered at various temper-atures for 20 min give a maximal value of 132 kg/cm2

數據

Fig. 3. The Au, Sn, and Cu concentrations across the multilayers of the diffusion-soldered specimen (Fig
Table I. Thicknesses of Intermetallic Compounds Formed during Diffusion Soldering Between
Fig. 7. The contact angles ( ) of liquid Sn on the surfaces of Au and Cu substrates at 350°C

參考文獻

相關文件

(Another example of close harmony is the four-bar unaccompanied vocal introduction to “Paperback Writer”, a somewhat later Beatles song.) Overall, Lennon’s and McCartney’s

The short film “My Shoes” has been chosen to illustrate and highlight different areas of cinematography (e.g. the use of music, camera shots, angles and movements, editing

1.9 Chapters 3 to 7 cover the concerns and suggestions received and elaborate on our support measures covering the five proposed actions, including enhancing schools’

Wang, Solving pseudomonotone variational inequalities and pseudocon- vex optimization problems using the projection neural network, IEEE Transactions on Neural Networks 17

volume suppressed mass: (TeV) 2 /M P ∼ 10 −4 eV → mm range can be experimentally tested for any number of extra dimensions - Light U(1) gauge bosons: no derivative couplings. =>

Define instead the imaginary.. potential, magnetic field, lattice…) Dirac-BdG Hamiltonian:. with small, and matrix

incapable to extract any quantities from QCD, nor to tackle the most interesting physics, namely, the spontaneously chiral symmetry breaking and the color confinement.. 

• Formation of massive primordial stars as origin of objects in the early universe. • Supernova explosions might be visible to the most