1. Introduction

Focused on the bottleneck problem of the thermal expansion mismatch in the development of flip-chip interconnection using conventional bonding methods at high temperatures, surface activated bonding (SAB) method is concerned with reducing the bonding temperature. Since using activated surfaces is the basic concept of SAB, high vacuum is regarded as a necessary condition to achieve and maintain cleaned surfaces. With this common wisdom, no work has previously been attempted to check the possibility of treating surfaces under low vacuum and performing the bonding in ambient air except with Au. Bonding in ambient air has the advantages of low cost and short fabrication time. If it is possible, low-temperature bonding in ambient air would be highly prospected.

However, the residue gases under low vacuum might influence the cleanliness of Ar-plasma pretreatment. Moreover, typically the electrodes of environmentally-friendly Sn-based alloys are easily oxidized in ambient air. Therefore, the key problem of bonding in ambient air is surface contamination on the activated surfaces. However, less information is available regarding the characterization of surface contamination before bonding and their influence on bonding of lead-free micro-bumps. Furthermore, if the surface contamination could be dispersed by chemical reaction and mechanical contact deformation, the bonding might be achieved in some bonding boundaries, which can be controlled by the process parameters of bonding pressure, temperature and time and influenced by material properties and geometrical factors. However, little studies have addressed these issues, and their influence on the bonding of micro-bumps is not well understood. Numerous experimental data and a theory model are necessary to establish the relationship of such factors to find suitable process windows enabling bonding of micro-bumps in ambient air.

The purpose of this research is to realize the bonding of micro-bumps below 200C in ambient air by the SAB method by approaching the study of the characterization of surface contamination and the influencing factors to achieve suitable process parameters. The achieved bonding boundaries and the determined relationships of the influencing factors are prospected to apply in the bonding of micro-bumps between different substrates.

2. Experimental Models and Methods

Typical electrodes, the hardly-oxidized Au, and easily-oxidized Sn and Sn-Ag or Sn-Ag-Cu are selected is this the study. X-ray photoelectron spectrometer (XPS) was selected for analysis the surface contamination of Au, Sn, and Sn-2.0Ag (wt%) alloy films. A submicron flip-chip bonder and various Au, Au/Sn and Sn-Ag-Cu micro-bumps were used for investigating the functions of the influencing factors. Some evaluation methods were carried out for understanding the surface conditions, electrical characteristics, and mechanical characteristics of the interconnections, including atomic force microscopy (AFM), contact angle measuring, surface profiling, electrical resistance test, die shear test, tensile test, scanning electron microscopy (SEM), and electron probe micro-analysis (EPMA).

3. Characterization of Surface Contamination and Its Influence on Bonding

Few effects of the annealing and flattening processes and the addition of 2.0wt% Ag composition were detected on the initial thickness of surface contamination. The initial thickness of carbon contaminants is around 2 nm on Au, Sn, or Sn-2.0Ag surfaces, and the initial thickness of oxides of Sn or Sn-2.0Ag is around 5-11 nm. The growth of contamination of Sn follows a logarithmic rate law.

The influence of the vacuum background is not sensitive to Au. Carbon contaminants are removed from the surfaces regardless of whether the background is high or low vacuum. In the case of Sn, carbon contaminants are removed from the surfaces, whereas Ar-etched Sn is oxidized at the same time due to the residual air in the low vacuum conditions, and the contamination ratio is seldom reduced. Compared with that pretreated under the high-vacuum background, the water droplet contact angles proved the cleanliness in terms of carbon contaminants of activated surfaces pretreated under the low-vacuum background. Regardless of the vacuum background, Ar-plasma pretreatment improves the bondability of Sn-Ag-Cu micro-bumps at low temperatures in ambient air. The status of carbon contaminants is a critical factor in the low-temperature bonding of ambient air. The low-vacuum background of Ar-plasma pretreatment is available for the low-temperature bonding of Sn-based micro-bumps in ambient air, even though the thickness of the oxides is not obviously reduced.

In the low-temperature bonding of micro-bumps in ambient air, the critical values of the Ar-plasma pretreatment time and the air exposure time are related to the thickness of carbon contaminants, material properties, and bonding parameters. To achieve a high bond quality, 2-nm-thick carbon contaminants should be removed by Ar-plasma pretreatment. Deformation and diffusion is effective on the bonding of micro-bumps. The critical time of Ar-plasma pretreatment and air exposure is changed depending on the material property and process parameters of bonding pressure and temperature. The shear strength on the bonding of Au micro-bumps at room temperature under 300 MPa increased more than 8 times by Ar-plasma pretreatment, whereas the shear strength on the bonding of Sn-Ag-Cu micro-bumps at 100°C under the bonding pressure of 250 MPa reached more than 20 MPa without Ar-plasma pretreatment, and there is no critical time of Ar-plasma pretreatment and air exposure in this case. The Ar-plasma pretreatment time and air exposure time should be controlled according to the material properties and bonding parameters and was determined as follows to study other influencing facotrs:

Ar-plasma pretreatment time: 30 s for Au and 120 s for Sn or Sn-Ag-Cu under a low vacuum background of 5-7 Pa;

Air exposure time: 3-30 min.

4. Influencing Factors on Micro-Bump Bonding

Quick diffusion between bonding couples may accelerate the surface contact at low-temperature bonding in ambient air. The diffusion accelerates the shrinking process of gaps between the contact surfaces. With a large diffusion rate of Au and Sn, the required bonding pressure in the bonding of Au and Sn is smaller than that of Au and Au at 150°C. The oxides on the bond interface of Au and Sn were found to be dispersed quickly with the help of diffusion. The intermetallic compounds formed in the bonded interfaces is assumed to be AuSn on the Au pads side, and AuSn2 and AuSn4 on the Au/Sn bump side at 100 and 150°C. One phase, AuSn5, emerges between the AuSn phase and Au at 200°C. The diffusion rate becomes increases with the temperature increase.

In the bonding of Au/Sn bumps to Au pads, surface roughness is not critical, whereas the key point is for a certain thick Sn layer to remain on top of the Au/Sn bumps for deformation and diffusion. In the bonding of Au micro-bumps, a smooth surface increases the actual contact area, and therefore increases the bonding strength and bond yield, and reduces the required bonding pressure and temperature. With a lower hardness of Sn than that of Au, a sufficient contact area of Au-Sn was achieved more easily than that of Au-Au, even though the Au/Sn bumps had rougher surfaces than the Au bumps. Through annealing to lower the hardness of Au, the required bonding pressure of Au micro-bumps was reduced more than 30% at room temperature. Since the self-diffusion rates are low, the contact deformation is the dominant function of the bonding of Au and Sn-Ag-Cu micro-bumps at low temperatures. A high bump profile (h/w) lowers the required bonding pressure.

The bonding parallelism between chip and substrate influences the bond yield. When the bump size and pattern were determined, a smaller bonding pressure was required to achieve a high bond yield and strength due to a smaller planarity angle. The planarity angle used in this study was controlled at around 0.005.

The relationship between bonding strength and bonding time of Au and Sn-Ag-Cu micro-bumps follows a logarithmic law. Bonding time is assumed to have a minor influence in these cases, whereas it has a significant influence on the bonding of Au-Sn due to the contribution of diffusion. The actual contact area of micro-bumps was strongly dependent on the bonding pressure and temperature. A suitable bonding boundary was selected depending on the material properties and geometric factors by controlling the bonding pressure and temperature. A model of the relationship with bonding pressure, temperature and bump profile was given.

5. Applications

The SAB method was successfully applied to connect various substrates in bonding chip-on-board, with non-conductive film, and surface acoustic wave components at 25-150C using bonding boundaries and the relationship of the bump profile, bonding temperature and pressure achieved. The critical problem of thermal mismatch in conventional methods due to the large difference of the coefficient of thermal expansion of each substrate is overcome.

6. Conclusions

In this research, the bonding of micro-bumps of Au and Sn as well as Sn-Ag and Sn-Ag-Cu is realized in the temperature range below 200°C by the SAB method using Ar-plasma activation under low vacuum background and contact in ambient air. Suitable process parameters were determined for low temperature micro-bonding in ambient air, such as pretreatment time and air exposure time for surface activation process, and parallelism, bonding temperature, bonding pressure, and bonding time for bonding process, depending on the materials, hardness, surface roughness, and bump profile. It demonstrates that the successful application of SAB to connect various substrates at low temperatures, and therefore overcome the critical problem of the thermal mismatch in conventional methods. It contributes to the microelectronic industry a systemically fundamental and applicable guide of the low-temperature flip-chip bonding technology.

本論文の目的は、表面活性化手法を用いたマイクロバンプの大気圧中、低温接合を実現することである。

高密度実装においては、20ミクロンレベルを切る微細マイクロバンプの接合が必須となっており、その実現のためには接合の低温化が必要とされている。しかし、従来の接合手法では接合温度が高く、そのため表面活性化を利用した低温接合に期待が大きい。しかし、表面活性化手法は従来、高真空中での接合が前提となっており実用には遠いものであった。本研究では、これを低真空中でのプラズマ表面活性化と大気中での接合を試み、その可能性を明らかにしたものである。

本研究では、まず、接合のプロセスを表面活性化のプロセスと接合のプロセスにわけ、それぞれについて、プロセスパラメータの影響因子とその最適化を図った。

表面活性化プロセスについては、表面活性化時間および大気中への露出時間、接合プロセスについては、接合時間、接合圧力、接合温度をパラメータとし、これらのパラメータと接合のメカニズムの関係を明らかにするとともに、これらプロセスパラメータの最適化とその境界を明確にすることで、さまざまな応用が可能であることを示した。また対象としては、接続における電気的特性の要求から、典型的な接続材料である、金、金スズ、スズ-銀?銅系のはんだの3系統を選定した。

これらの内容について、審査では下記のような議論があった。

最初に、多数のデータが網羅的に提示されていることが原因で、本研究の目的・成果が分かりにくいという点が指摘された。これについては、表面分析による表面活性化プロセスに関するデータを、表面活性化時間および大気中への露出時間という2つのプロセスパラメータで整理することにより、接合性におよぼす表面活性化プロセスの影響を明らかにした。また、接合プロセスについては、接合圧力と接合温度との相関について、あらたな理論的な関係式を導入し、そのパラメトリックな解析により、実験値の整理と接合性の関係を明示的に示した。これにより、従来の研究では明確にされていなかった接合圧力を接合温度の関係が初めて理論的に明快に示されることとなった。

応用例の提示についても網羅的な印象が強いという指摘がなされた。これに対しては、接合の可能なプロセスウィンドゥを明示し、応用例がどの範囲にあるかということを示すことで、今後の新たな応用の可能性が示される結果となった。

議論の過程から、金とスズ系の両者には、表面エネルギーに起因する接合性の本質的な違いに加え、硬さ、変形能の違い、さらに酸化性の違いがあることから、金については、バンプ表面の平坦性が極めて大きな影響を示すこと、一方で、スズ系については、変形の影響が大きく、圧力に対する依存性が大きく、そのため、表面に残留している酸化層の存在に関わらず、大気中での接合が可能であることが結論された。また、従来知られていなかった、バンプの縦横比が接触圧力に大きな影響を及ぼし、結果的に接合プロセスパラメータの限界値に大きな影響を与えていることが定量的に示された。

以上の結果から、本研究では、従来の手法では不可能であった常温・低温での大気中の接合を実現したのみならず、そのメカニズムを膨大な実験データおよび解析により明らかにしたものであり、高く評価された。以上のように、本研究で得られた工学的知見は極めて大きく、また、工学の発展に寄与するところは多大である。

よって本論文は博士(工学)の学位請求論文として合格と認められる。