Thermal Behavior of Silver Nanoparticles for Low-Temperature Interconnect Applications
Moon, Kyoung-Sik
Low-temperature sintering behavior of Ag nanoparticles was investigated. The nano Ag particles used (~20 nm) exhibited obvious sintering behavior at significantly lower temperatures (~150°C) than the T^sub m^ (960°C) of silver. Coalescence of the nano Ag particles was observed by sintering the particles at 150°C, 200°C, and 250°C. The thermal profile of the nanoparticles was examined by a differential scanning calorimeter (DSC) and a thermogravimetric analyzer (TGA). Shrinkage of the Ag-nanoparticle compacts during the sintering process was observed by thermomechanical analysis (TMA). Sintering of the nanoparticle pellet led to a significant increase in density and electrical conductivity. The size of the sintered particles and the crystallite size of the particles increased with increasing sintering temperature.
Key words: Silver nanoparticles, thermal behavior, sintering, melting point depression, low-temperature interconnect
INTRODUCTION
Considerable attention has been paid to polymermetal composites as an interconnect material in the microelectronic packaging area because of their light weight, low processing temperature, and environmentally benign properties. Nonetheless, the polymer-metal composites have been applied only in low-end products because of their limited performance. The contact resistance of the interconnect joints made by the polymer-metal composites is high and less stable in harsh environments compared with those of eutectic metal-solder joints, such as SnPb, SnAg, and SnAgCu.
Figure la and b illustrate that by using fine particles, more interfaces between the fillers can be generated, which leads to an increasing contact resistance of the interconnect joints. The R^sub fillers^ term depends on the intrinsic property of the fillers, which is already low compared to R^sub btw fillers^- Thus, in order to decrease the R^sub total^, we can see that the R^sub btw^ ^sub fillers^ and R^sub filler to bond pad^ terms must be reduced.
At lower filler concentrations, the insulation layer (e.g., polymer resin) is more prevalent between the fillers. Direct contact between the fillers occurs for higher filler loadings, which reduces R^sub btw fillers^ and enables electrons to flow through the fillers, resulting in more electrically conductive materials. Although a direct contact between the fillers can be accomplished, a contact resistance at the interface is still present unless the interface is eliminated by fusion, as shown in Fig. Ic. This approach minimizes the total contact resistance because fusion of the fillers with each other reduces total interface area. The approach to fuse metal fillers does not appear to be feasible with previous materials because the typical, organic printed-circuit board on which the metal-filled polymer is applied cannot withstand the necessary high temperature for fusion.
Interestingly, studies have shown that decreasing the size of materials can dramatically reduce their melting point (T^sub m^). Melting, freezing, and diffusion behaviors of finite systems have been of theoretical and experimental interest for many years.4-6 It has been reported that surface premelting and sintering processes are a primary mechanism of T^sub m^ depression of fine particles. From kinetics and dynamics of the phenomena, it is believed that high surface-area nanopowders result in an increased proportion of defective interfacial atoms that also cause a depression of the sintering temperature (T^sub s^).
If interfaces between metal fillers can be eliminated, the use of fine metal particles would be promising for fabricating metal-polymer composites with high electrical conductivity.
A system containing 3-7-nm Ag/Au nanoparticles and a polymer binder for fabricating fine-pitch electrical line patterns on circuit boards was reported.7 However, the report lacks information about thermal behavior of the nanoparticles and their crystal structure change. This novel approach was applied by Dr. Wong’s group in electronic and optopackaging areas to replace high-temperature soldering materials, such as Sn-based alloys, with nanomaterials.8’9
In this paper, thermal behavior of Ag nanoparticles with respect to the sintering reaction is discussed. Surface changes of the particles during sintering and crystal structure variation are addressed as well.
EXPERIMENTAL
Materials
For synthesizing Ag nanoparticles, AgNO^sub 3^ was used as a precursor in the combustion chemicalvapor condensation (CCVC) method.10 In CCVC production of nanopowders, the NanoSpray Process (a platform technology that relies on the proprietary Nanomiser device to produce aerosols with controllable droplet size distribution (nGimat, Atlanta, GA)) was used to convert a starting liquid solution containing chemical precursors into an ultrafine aerosol that is efficiently combusted to produce nanopowders. The average particle size of Ag nanoparticles was 20 nm as shown in Fig. 2.
Measurements
A differential scanning calorimeter (DSC, TA Instruments, New Castle, DE) was used for thermal profile measurement. The weight loss during heating and thermal shrinkage was monitored by a thermogravimetric analyzer (TGA, TA Instru-ments) and by a thermomechanical analyzer (TMA, TA Instruments), respectively. For TMA measurement, the nanoparticles were cold pressed in a mold and made into a tablet (13 mm in diameter). X-ray photoelectron spectroscopy (XPS, Surface Science Model SSX-IOO) and Fourier transformation infrared (FTIR, Nicolet Co., San Jose) spectroscopy were used to characterize the surface of the nanoparticles at different temperatures. Scanning electron microscopy (SEM, Hitachi S-800, Tokyo) was used to observe the morphology of the particles.
Nanopowders were compacted with ~50 MPa pressure. Density of the compacts before and after sintering at different temperatures was determined by the Archimedes’ method, in which the three weight method, measuring dry, immersed, and wet weights of each sample, was used. Electrical conductivity of cold-pressed pellets was measured by the four-point probe method.
RESULTS AND DISCUSSION
Microstructure during Annealing
Bulk Ag has a high T^sub m^, 960°C, and heat of fusion (ΔH^sub f^) of 284.9 kJ/mol. However, if the Ag particle size is reduced to nanoscale, the T^sub s^ can be significantly lowered. In particular, this study found interesting behavior of nanosized Ag particles at low temperature.
Figure 3 shows SEM photographs of the Ag pellet prepared from nanoparticles as pressed and then as sintered at different temperatures. As-received particles have a very small particle size. Particles sintered at 100°C showed no significant difference in size compared with as-received particles. On the other hand, particles sintered at 150°C show a dramatic increase in size, with many 0.1-0.3-µm particles. Furthermore, particle size gradually increased at 20O°C and 25O°C. In Fig. 3d and e, 0.4-0.5-µm and 1-1.5-µm-sized particles were observed, respectively.
The particles shown in Fig. 3c-e are sintered between each particle, and there is necking rather than complete melting. These particles are connected through their surfaces, and dumbbell-type particles can be found. This is very similar to the morphology of an initial stage in the typical sintering process. Therefore, the sintering process obviously occurred in the particles at 150°C and proceeded with increasing temperature.
Thermal Profile and Surface Characterization
In order to investigate why the nanoparticles used showed a strong sintering behavior, thermal properties of the particles were studied. The weight loss during heating was monitored in TGA, as shown in Fig. 4. The as-received particles showed weight loss at two regions, one around 100°C and the other around 20O°C. Weight loss at 10O°C likely arose from moisture adsorbed on the particle surface. A line with solid squares in the figure indicates particles that were cleaned in ethanol by sonication and dried at 50°C. There was no significant difference in weight loss at 200°C between the annealed particles at 100°C, 150°C, and the cleaned particles. Thus, the substance that caused weight loss at 200°C may not be simply washed by ethanol, while heat treatment above 200°C could eliminate the weight loss at 200°C. Interestingly, the loss at 200°C disappeared by 200°C and 250°C annealing. It is postulated that oxygen or other material was bonded or adsorbed to the surface of the Ag particles and debonded at about 200°C.
Figure 5 shows DSC profiles of the particles treated at different temperatures. The as-received particles showed a sharp exothermic peak around 220°C. The annealed particles at 100°C and 150°C also showed a strong exothermic peak, but their peak temperatures were shifted down by ~20°C. The area of these peaks decreased with increasing annealing temperature. The area of the as-received particles was 86.4 J/g and the area of the heattreated particles at 100°C and 150°C were 52 J/g and 25.5 J/g, respectively. These peaks disappeared from the 200°C and 250°C annealed specimens. Evaporation or debonding of organic compounds shows typically an endothermic peak. Some surfactants on Ag flakes can show exothermic peaks in DSC when they are acids and the test is run under air because of an oxidation process. Under N^sub 2^, even acidic surfactants on Ag particles have neither endothermic nor exothermic peaks. Accordingly, the exothermic peak may not be due to the evaporation or debonding of organic compounds.
The XPS was used to examine the presence of any organic compounds on the particle surface. Figure 6 shows survey scan results for the as-received particles and the nanoparticles annealed at 150°C and 250°C. From this XPS test, binding energy peaks of Ag 3d^sub 2/3^ (397.9 eV) were found. High-resolution scanning confirmed that the Ag 3d^sub 2/3^ peaks for all the samples were identical. In addition, small amounts of oxygen and carbon peaks were observed. Their atomic compositions are shown in Table I, in which there was no significant difference in carbon content between the as-received particles and the ones annealed at 150°C, while even higher carbon content was recorded from the particles annealed at 250°C. This may arise from sample contamination during heat treatment at 250°C in the thermal oven under the ambient environment. If there were substantial amounts of organic materials on the particles, the carbon contents should have shown lesser amounts in the 150°C- or 250°C-treated particles. Thus, the source of the weight loss and the exothermic peak from TGA and DSC is most likely not only from organic materials.
The FTIR was used to study the surface of the nanoparticles. The FTIR spectra for nanoparticles during heating were obtained and are shown in Fig. 7. This figure shows FTIR spectra of a series of heat treatments ranging from room temperature to 300°C. The hydroxyl group related with moisture on the particles was found at around 3,000 cm^sup -1^, and the peak disappeared by heat treatment. Any specific peak in FTIR that could be suspected from organic compounds was not observed in the as-received particles. During heating, there was no marked difference in characteristics peaks. Therefore, no significant amount of organic materials was believed to be on the nanoparticles.
In general, exothermic peaks of metal particles can also be found in DSC because of the crystallization of amorphous metal particles or recrystallization of strained metal particles by heating. In addition, the exothermic heat can be generated by diffusion between unstable atoms on the nanoparticles during the surface sintering reaction. During this reaction, stabilization of the crystal structure, reformation, and reallocation of grain boundaries at the particle/ particle interface can occur. The atoms on nanoparticles are generally very unstable from higher surface energy because of their extremely large surface area and defective structure. For this reason, surface sintering or melting of nanoparticles occurs at a lower temperature than that of the bulk materials. Therefore, surface sintering could cause this exothermic reaction that was found in DSC. Creation of gas phases from the nanoparticle could be another source of the exothermic peak. This could explain the weight loss at the temperature in TGA. However, further investigation is needed to clearly explain this phenomenon.
Shrinkage during Heat Treatment
In general, during the sintering process of powder compacts, the material experiences significant densification together with shrinkage. This can be monitored by changes in thermal diffusivity, electrical conductivity, density, or dimension. In this study, a dimension change was measured by TMA.
Figure 8 shows TMA results of the compacts sintered at different temperatures in which linear dimensional change of the compacted samples was monitored. The as-received particles show a continuous decrease in dimension with increasing temperature. This decrease continued up to 40O°C, and a different profile was observed at above 400°C. Note that a small abnormal region in the TMA curve was found at around 230°C, where a very slight increase in dimension with increasing temperature and an abrupt decrease in dimension occurred. This was observed for both the as-received particles and the particles sintered at 150°C, while the particles sintered at 250°C and at 500°C did not show this behavior. This may be related to weight loss in TGA and exothermic reaction in DSC that were found at similar temperatures.
At the initial stage, three of the samples (the asreceived and sintered at 150°C and 250°C) showed an increase in dimensions. This may be from the thermal expansion of the Ag compacts. At a certain temperature, dimension began to decrease up to 400°C and then showed a plateau or an increasing region. Their overall dimension-change behavior as a function of temperature appeared similar, except for the sample sintered at 500°C. Interestingly, for the as-received particles and the particles sintered at 150°C and 250°C, their dimensions began to decrease at ~60°C, 150°C, and 250°C, respectively. Sintering started to occur for these nanopowder compacts once the temperature exceeded the sintered temperatures. The dimension decrease of the compacts up to 400°C is thought to be caused by sintering shrinkage.
While typical sintering shrinkage of ceramics is about 10-20% (linear dimension change), a dimension decrease of the nanoparticles used was as small as -2.5 % (up to 400°C). The reason the amount of shrinkage of the nano Ag particles was much smaller than typical ceramics may be that shrinkage of the nanoparticles was generated only by the surface premelting and surface sintering process; the temperature range in this study was not high enough to examine the entire sintering behavior before melting.
Figure 9 shows the electrical conductivity of the compact specimens with different sintering temperatures. The electrical conductivity increased almost linearly with increasing sintering temperature. The increase in the conductivity could be due to densification of the specimens by the sintering process. The density of the pressed silver-nanoparticle specimens before and after sintering and their morphology are shown in Fig. 10. The density of the compact specimens increased with increasing sintering temperature. Therefore, it was confirmed that the sintering process led to particle densification, resulting in increasing electrical conductivity.
Crystal Structure during Annealing
Crystal structure of the nanoparticles was investigated using XRD. Figure 11 shows XRD spectra of the nano Ag particles annealed at different temperatures. Overall, the nanoparticles showed a typical, polycrystalline-Ag crystal structure regardless of annealing. However, the as-received particles showed slightly broadened peaks at each corresponding orientation angle. This is due to the smaller particle size compared to the annealed powder.
A polycrystalline material that does not contain lattice strain and consists of particle sizes larger than 500 nm shows sharp lines in a powder diffractogram. Imperfections in the structure of the crystallites constituting a sample cause broadening of the diffraction line profiles. Large crystallites give rise to sharp peaks; as the crystallite size decreases, the peak width increases and the intensity decreases. This is due to the periodicity of the individual crystallite domains, in phase, reinforcing the diffraction of the x-ray beam, resulting in a tall narrow peak. If the crystals are defect free and periodically arranged, the x-ray beam is diffracted to the same angle even through multiple layers of the specimen. If the crystals have a lower degree of periodicity, the result is a broader peak. Peak broadening and shifting can also originate from variations in lattice spacings caused by lattice strain.
Figure 12 shows the crystallite size of the Ag nanoparticles as a function of the annealing temperature. The crystallite size increased with increasing annealing temperature. Thus, during the surface premelting and the sintering process, not only particle size increased but also the crystal grains in the particles grew. With annealing, the crystal structure of the nanoparticles reach higher degrees of periodicity, develop a more ordered pattern, and become more stabilized in terms of the free energy by the sintering process.
CONCLUSIONS
Sintering behavior of Ag nanoparticles, which were synthesized through the CCVC process, was investigated. The nano Ag particles used (-20 nm) exhibited sintering behavior at significantly lower temperatures than larger grained Ag.
Coalescence of the nano Ag particles was observed by sintering the particles at 150°C, 200°C, and 250°C. From the FTIR and XPS results, no significant amount of organic compounds was found on the nanoparticle surfaces.
The sharp exothermic peak of the nanoparticles found in DSC may be related to a chemical reaction of oxygen, or gas phase creation on the particle surface, to the sintering reaction, by lattice diffusion, stabilization of the crystal structure, recrystallization, reallocation, and recreation of grain boundaries at the particle/particle interface. The fact that there is no substantial amount of organic surfactant on the Ag nanoparticles could cause more active sintering at low temperature compared with surfactant-coated nano Ag particles.
Coalescence of the Ag nanoparticles during the sintering process was observed. Annealing the nanoparticles led the density and the electrical conductivity of the compact specimens to dramatically increase. The size of the coalesced particles and the crystallite size of the particles increased with increasing annealing temperature.
During the surface premelting and the sintering process, not only the particle size increased, but also the crystal grains in the particles grew. The crystal structure of the nanoparticles showed higher degrees of periodicity, developed a more ordered pattern, and became more stabilized in terms of free energy by the sintering process. This is driven by the fact that nanoparticles possess high surface energy and have substantial lattice strain.
ACKNOWLEDGEMENTS
The authors thank the National Science Foundation for partial financial support for this research (NSF Grant No. DMI-0217910). The authors thank Professor Robert Speyer, Mr. NamTae Cho, and Mr. Runrun, Georgia Institute of Technology, for valuable discussions. The authors thank Dr. Rosario A. Gerhardt and Ms. Yangyang Sun, Georgia Institute of Technology, for electrical conductivity and FTIR measurements, respectively.
REFERENCES
1. J.C. Jagt, P.J.M. Beris, and G.F.C.M. Lijten, IEEE Trans. CPMT 18, 292 (1995).
2. D. Lu, Q.K. Tong, and C.P. Wong, IEEE Trans. CPMT, Part C 22, 223 (1999).
3. K.S. Moon, C. Rockett, C. Kretz, W.F. Burgoyne, and C.P. Wong,J.Adhes. Sd. Technol. 17, 1785 (2003).
4. L. Allen, R.A. Bayles, W.W. GiIe, and W.A. Jesser, Thin Solid Films 144, 297 (1986).
5. M.S. Daw and M.I. Baskes, Phys. Rev. B 29, 6443 (1984).
6. M.S. Daw and M.I. Baskes, Phys. Rev. Lett. 50, 1285(1983).
7. M. Oda, T. Abe, T. Suzuki, N. Saito, H. Iwashige, and G. Kutluk, Mater. Res. Soc. Symp. Proc. 704, 3 (2002).
8. C.P. Wong, K.S. Moon, and Y. Li, Georgia Tech. Corp. Invention Disclosure 2003, U.S. patent pending.
9. H. Dong, K.S. Moon, and O.P. Wong, J. Electron. Mater. 33, 1326 (2004).
10. A.T. Hunt, W.B. Carter, and J.K. Cochran, Jr., Appl. Phys. Lett. 63, 266 (1993).
KYOUNG-SIK MOON,1 HAI DONG, RADENKA MARIO,2 SURESH POTHUKUCHI,1 ANDREW HUNT,2 YI LI,1 and O.P. WONG1,3
1.-School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, GA 30332. 2.-nGimat Corporation, Atlanta, GA 30341. 3.-E-mail: cp-wong@mse.gatech.edu
Copyright Minerals, Metals & Materials Society Feb 2005
Provided by ProQuest Information and Learning Company. All rights Reserved
