Sessions Index

Bio:

2008: Researcher, Inst. Industrial Science, U.Tokyo
2011: Assistant Professor, Dept. Materials Engineering, U.Tokyo
2016: Lecturer, Dept. Materials Engineering, U.Tokyo
2023: Associate Professor, Research and Education Institute for Semiconductors and Informatics, Kumamoto University
2024: Professor
2025: Special Assistant to the President

Abstract:

This paper introduces SCFD (Supercritical Fluid Deposition), a film formation technology that utilizes chemical reactions in supercritical fluids. This technology can form uniform thin films even on deep structures with aspect ratios exceeding 100 and can achieve film formation rate of 10 nm/min, making it promising for application in Cu formation technology for TSVs.

 

Bio:
Professor Oliver Chyan directs the Interfacial Electrochemistry and Materials Research Laboratory at the University of North Texas. With over 30 years of R&D experience, he specializes in resolving critical interfacial materials problems related to high-volume microelectronic fabrication and IC packaging. His current research interests focus on developing novel copper interconnects to enhance performance and reliability.

Abstract:
The exponential growth in computational demands driven by AI, Machine Learning, and data-heavy applications has accelerated the transition toward 3D integration in microelectronics, now seen as a successor to Moore’s Law. As transistor scaling approaches physical limits, 3D integration enables vertical stacking for increased bandwidth, lower latency, and higher density. To meet these demands, high-performance communication between and within processors especially between GPUs and HBM requires robust, ultra-dense interconnects. Cu-solder micro-bumps and Cu-to-Cu direct bonding are key candidates, offering compact form factors, high-speed signaling, and thermal efficiency.
However, Cu’s high chemical reactivity presents a major challenge. Oxidation in ambient and elevated temperature environments forms CuO and Cu2O, which impede solder wetting and atomic diffusion, degrading interconnect performance. Conventional mitigation methods like no-clean flux, formic acid vapor clean, and metal coatings face critical drawbacks: no-clean flux still leaves corrosive residues, formic acid demands complex engineering due to its hazardous nature, and metal coatings add resistivity and processing complexity. These issues limit fine-pitch scalability, yield, and process efficiency, highlighting the pressing need for a cleaner, low-cost, scalable oxidation control method for next-generation 3D packaging.
To overcome this persistent challenge of Cu oxidation, we developed an ultra-thin (~2-5 nm) Cu-selective passivation coating, applied using industry-compatible deposition methods such as CVD or LPD. Notably, the coating maintained its integrity even after two months of ambient storage, proving its long-term stability. It effectively suppressed Cu oxide formation by up to 63% at ~240°C, ensuring a clean, bond-ready surface during high-temperature processes like TCB and solder reflow. In Cu-solder flip-chip assemblies bonded in ambient conditions, the passivation coating enabled robust Cu–Sn intermixing with well-formed Cu₆Sn₅ and Cu₃Sn IMCs, free from voids or delamination (Figure 1(a)). Mechanical reliability was exceptional, with joints achieving 5.16 MPa shear strength, exceeding MIL-STD-883G requirements (~1.5 MPa). Electrical performance was equally strong, with an initial contact resistance of 6.4 × 10⁻⁶ Ω·cm² and negligible drift after 1000 hours at 150°C and 1000 thermal cycles (–40°C to +125°C), along with only a slight increase in IMC growth demonstrating excellent environmental resilience and bond stability.
For Cu-to-Cu direct bonding, the same passivation coating enabled significant oxidation resistance, achieving approximately 56% oxide suppression after 20 minutes at 300°C in ambient air. When bonded under optimized TCB conditions, CMP-treated Cu substrates with the passivation layer yielded average shear strengths of 40.7 ± 4.2 kgf, surpassing the MIL-STD-883 Method 2019.9 requirement (~31.2 kgf for a 0.25 cm² die area). Cross-sectional SEM and EDX analyses confirmed clean Cu to Cu bonding interface with no residual oxides or interference from the passivation coating (Figure 1(b)).
This comprehensive strategy offers a cost-effective and practical solution for the microelectronics packaging industry, with the Cu-selective passivation coating enabling robust, fluxless Cu-solder and Cu-to-Cu interconnects. By effectively mitigating oxidation without sacrificing bond integrity or scalability, it presents a promising integration-ready approach for next-generation, high-density, AI-driven 3D integration platforms.

 

Bio:
Chen-Ning Li is a Ph.D. candidate in the Department of Materials Science and Engineering at National Yang Ming Chiao Tung University in Taiwan, working under the supervision of Professor Chih Chen. His research focuses on low-temperature nanocrystalline Cu bonding for 3D-IC applications.

Abstract:
Cu-to-Cu bonding has emerged as a promising approach for achieving fine-pitch hybrid bonding in heterogeneous integration applications. As bump diameters continue to scale down to the sub-micron level, the use of nanocrystalline Cu (NC-Cu) becomes essential to meet via-filling requirements. NC-Cu is not only the most suitable Cu microstructure for fine-pitch via filling, but it also exhibits exceptional properties that facilitate its application in low-temperature Cu bonding. Leveraging its high grain boundary density, grain boundary diffusion plays a critical role in surface atom rearrangement and interfacial void closure. Moreover, the internal energy stored within grain boundaries can be released during the thermo-compression bonding (TCB), providing a driving force for grain growth across the bonding interface. However, surface oxidation remains a critical challenge, as it hinders atomic diffusion and inhibits interfacial void closure. Therefore, various methods have been investigated to remove native Cu oxide prior to the bonding process, including formic acid vapor treatment, citric acid wet cleaning, and plasma surface cleaning. Although these pre-treatments can effectively remove the native oxide, the subsequent room temperature storage between the cleaning and bonding processes may lead to re-oxidation of the Cu surface. Thus, the processing window related to post-treatment storage time emerges as an important yet underexplored factor that can significantly influence the final bonding quality.
In this work, NC-Cu films were electroplated on a Ti/Cu substrate. The samples were then diced into 1 × 1 cm2 chips and subjected to a chemical mechanical planarization (CMP) process to reduce film thickness and improve surface planarity. Two types of surface pre-treatments, citric acid cleaning and plasma cleaning, were applied to evaluate the post-treatment durability of oxide removal. After surface pre-treatment, the samples were stored at room temperature for queue times of 1 h, 6 h, and 12 h to evaluate the effect of ambient exposure on surface oxidation. The bonding process was then performed at 280 °C under an external pressure of 22 MPa for 1 h. Cross-sectional focused ion beam (FIB) imaging was utilized to evaluate bonding quality by quantifying void size and distribution, while shear tests were conducted to assess the mechanical strength of the bonding interface. By correlating the FIB imaging results with the shear test data, we gained insights into the threshold queue time for different pre-treatment conditions. Additionally, the bonding fraction and void fraction, obtained from the cross-sectional images, showed a clear correlation with the measured shear strengths, suggesting a strong connection between the observed interfacial microstructure and mechanical performance. These findings highlight the importance of controlling post-treatment storage time to ensure reliable bonding performance in low-temperature NC-Cu bonding.

 

Bio:
I am a Chinese-Malaysian currently pursiung a Master's Degree at National Yang Ming Chiao Tung University, Hsinchu. It is my fifth year studying abroad in Taiwan, and had the wonderful opportunity to have Professor Chih Chen as my advisor. My main field of research is on improving the reliability of fine-pitch RDL structures, and aim to develop a thin and conformal passivation coating to inhibit electromigration induced failure. The material of choice here is tin for it's unique capability of forming stable intermetallic compounds (IMC) with copper, which can act as a blockade for atomic diffusion paths along the surface of copper lines.

Abstract:
As the pitch of Cu redistribution layers (RDLs) continues to shrink, challenges such as increased susceptibility to oxidation, corrosion, and electromigration become more pronounced due to the higher surface-area-to-volume ratio and elevated current densities. These stressors can significantly degrade the reliability of the interconnects in advanced packaging technologies. To mitigate these issues, the introduction of passivation or capping layers over Cu RDLs has become increasingly important. Such layers serve as physical and chemical barriers, protecting the copper from environmental and electrochemical degradation, thereby enhancing the overall thermal, chemical, and electrical stability of the RDL structures. Among potential passivation materials, tin (Sn) has emerged as a promising candidate due to its favorable interaction with copper. Upon annealing, Sn reacts with Cu to form stable Cu6Sn5 and Cu₃Sn intermetallic compounds (IMC), which not only prevents further oxidation but also helps restrict surface diffusion pathways of Cu atoms, effectively reducing the driving forces for electromigration. This makes Sn a particularly attractive option for next-generation Cu RDL passivation schemes. However, achieving uniform and conformal Sn coverage, especially on high-aspect-ratio structures, remains a considerable challenge. Conventional deposition methods such as electrodeposition and thermal evaporation often fail to provide adequate coverage on the vertical sidewalls and foot of the RDL lines. In our work, we deposited an 800 μm long, 5 μm thick, 10 μm wide nanotwinned copper (NT-Cu) line on to a RDL die consisting of Ti barrier layer and Cu seed layer with well-defined lithography patterning. Following electroplating, the surrounding Ti, Cu, and photoresist layers were removed using acetone and appropriate etchants to expose the NT-Cu line. To achieve a more conformal passivation layer, a thin Sn film was subsequently deposited onto the NT-Cu line using an electroless plating method. This approach allows for uniform deposition along the top surface, sidewalls, and bottom corners of the Cu line. The NT-Cu RDL was electroplated using a periodic reverse (PR) method, with forward and reverse current densities of 4 A/dm² and -1 A/dm² respectively, using a 40 ms forward and 4 ms reverse cycle. The Sn layer was deposited at various immersion times in a commercial plating solution, maintained at 68 °C and stirred with a magnetic stirrer on a hot plate to ensure uniformity. Cross-sectional microstructure analysis was carried out using focused ion beam (FIB) and scanning electron microscopy (SEM), while surface elemental composition was examined with energy-dispersive X-ray spectroscopy (EDS). Preliminary results show that longer immersion times yield thicker Sn coatings, while at extremely short durations may result in weaker coverage, particularly at the foot of the Cu RDL sidewalls. This suggests that coating conformality improves with immersion time and that total coverage of the structure is possible. Future work will focus on optimizing annealing conditions to promote the controlled formation of Cu₃Sn IMCs and to quantify their thickness across the RDL structure.

 

Bio:
Department of Materials Science and Engineering, National Chung Hsing University, Taichung 402, Taiwan

Abstract:
With the continuous trend toward thinner, lighter, and more compact electronic devices, the development of reliable, low-temperature metal bonding techniques has become increasingly important for advanced packaging. Ultrasonic bonding stands out due to its short processing time, room-temperature operation, and elimination of filler materials, making it a highly promising approach for direct metal-to-metal interconnections. This study focuses on the effects of vacuum ultraviolet (VUV) surface treatment under an oxalic acid atmosphere on Al to Al ultrasonic bonding. The results demonstrated a significant improvement in bond strength for specimens can be achieved when treated with VUV in oxalic acid for a proper duration. This enhancement is attributed to the formation of stable bidentate bonds between the carboxyl functional groups and aluminum surface atoms, which promote stronger interfacial adhesion during ultrasonic bonding. Additionally, the increased surface energy resulting from oxalic acid-VUV treatment contributes to improved wettability and atomic-level contact between the bonding interfaces. It can also be observed that excessive treatment time or high vapor concentration led to the presence of organic residues that were not fully decomposed under the given irradiation conditions. These residues adversely affected the bonding interface by creating contamination layers or voids, thereby reducing the mechanical strength of the joints. Oxalic acid assisted by VUV surface modification offers a promising route to enhance Al/Al ultrasonic bonding by engineering surface chemistry at the molecular level. These insights provide valuable guidance for developing low-temperature, high-reliability bonding processes in future electronic packaging applications.
Keywords: Aluminum direct bonding, Ultrasonic bonding, Oxalic acid, VUV surface treatment

 

Bio:
Dong Jun Kim is currently pursuing a Ph.D. in Mechanical Engineering at the Korea Advanced Institute of Science and Technology (KAIST), South Korea. He received his B.S. degree in Mechanical Engineering from Hanyang University and his M.S. degree in the same field from KAIST. He was recognized with the Student Grant Award for an outstanding presentation and paper at the IEEE Electronic Components and Technology Conference (ECTC). His research interests include the mechanical reliability of thin films used in semiconductor devices, particularly materials such as SiO₂ and SiN. His work focuses on tensile properties, interfacial adhesion, and thermal expansion behavior of these films. CV page : https://jkim111w.wixsite.com/mysite

Abstract:
As semiconductor devices continue to scale down, the mechanical reliability of back-end-of-line (BEOL) interconnect structures becomes increasingly critical for device performance and manufacturing yield. This study presents a comprehensive investigation of interfacial adhesion energies in BEOL structures using the integrated measurement method, double cantilever beam (DCB) fracture mechanics testing to identify the weakest interface and optimize material selection and process parameters. We performed mapping and comparing the adhesion energy results and strain results in the BEOL structure.
We systematically characterized three critical interfaces: Cu–Cap, Cu–IMD (inter-metal dielectric), and IMD–Cap using unified measurement methods. For Cu–Cap interfaces, two different capping materials (Cap1 and Cap2) were evaluated under various surface treatment conditions. NH₃ plasma treatment and consecutive H₂/NH₃ plasma treatments were applied, with treatment effects analyzed through chemical bonding characterization. Results showed that the enhanced surface treatment increased Cu–Si bonding ratios at the Cu–Cap1 interface through the removal of copper oxide layers, thereby improving interfacial adhesion energy. For the Cu–Cap2 interface, the SiH₄ gas treatment between H₂ plasma and NH₃ plasma treatments resulted in dramatic adhesion energy enhancement through selective Si atom incorporation, achieving approximately 260% improvement compared to the baseline treatment. For Cu–IMD interfaces, we compared two IMD materials with different dielectric constants (κ values). IMD1 with lower κ exhibited significantly lower adhesion energy (0.72 ± 0.03, normalized) compared to IMD2 with higher κ (1.47 ± 0.30), attributed to increased porosity and defect densities in low-κ materials that introduce nonuniform bonding characteristics and stress concentration points. The IMD–Cap interface demonstrated remarkable sensitivity to deposition sequence. When Cap2 was deposited first (Cap2–IMD1), adhesion energy was approximately six times higher than the reverse sequence (IMD1–Cap2), with crack propagation shifting from the target interface to the IMD–epoxy interface. Surface energy analysis revealed that the higher surface energy of Cap2 (49.01 mJ/m²) compared to IMD1 (32.54 mJ/m²) enhanced physical adsorption of CVD precursors, explaining the deposition sequence dependency.
Finite element method (FEM) simulation analyzed thermal stress distributions in BEOL structures, identifying geometrically discontinuous regions as vulnerable failure points. Combining experimental adhesion measurements with simulation results, we determined that the Cu–IMD (low-κ) interface represents the weakest link in BEOL structures, with particularly high delamination potential at Cu-filled via sidewalls.
These findings provide critical insights for enhancing BEOL reliability through optimized material selection, surface treatment protocols, and control of deposition sequence. The quantitative adhesion energy database and mechanical analysis will contribute to the development of more robust and reliable interconnect structures for advanced semiconductor packaging applications.
Keywords: BEOL, interfacial adhesion, semiconductor packaging, dielectric materials, surface treatment, adhesion mapping

 

Bio:
Ke-Wei Hsieh is a first-year M.S. student in the Department of Materials Science and Engineering at National Yang Ming Chiao Tung University. He is currently conducting research under the supervision of Prof. Chih Chen, focusing on nanocrystalline copper and low-temperature Cu–Cu bonding for advanced packaging applications.

Abstract:
Copper-to-copper (Cu–Cu) direct bonding has emerged as a highly promising alternative to conventional solder-based interconnects, which often suffer from limited high-temperature reliability, electromigration, and the formation of brittle intermetallic compounds. However, achieving strong Cu–Cu bonding at reduced temperatures remains a significant challenge due to the limited atomic mobility of conventional Cu surfaces.
To address this, we developed (111)-oriented dominant nanocrystalline Cu films that enable low-temperature Cu–Cu bonding. The (111) surface provides high surface diffusivity, low oxidation rate, and strong atomic packing, which together enhance bonding efficiency and promote interfacial elimination. Meanwhile, the nanocrystalline structure offers high grain boundary density and stored energy, which facilitate grain boundary diffusion, grain growth, and intimate contact. This combination enables effective bonding at reduced temperatures, making it well-suited for advanced electronic packaging.
Nanocrystalline Cu films with a strong (111) surface texture were fabricated using direct current electrodeposition with three specific additives—DP115, DP116, and DP117—provided by Chemleader, Inc. Chronopotentiometric (constant-current) electrochemical analysis showed that the incorporation of these additives significantly increased the overpotential, indicating their ability to promote fine-grained microstructures. According to electron backscattered diffraction (EBSD) analysis, the average grain size of the as-deposited Cu films was 132.8 nm, with a (111) orientation ratio of 57.4%, confirming the successful formation of a nanocrystalline, texture-dominant structure.
To evaluate bottom-up filling potential, a rotation speed-dependent electrochemical experiment was conducted to simulate via-filling conditions—where high rotation speed represents the via opening and low rotation speed the via bottom. The potential difference between the two conditions reached 21.5 mV, suggesting the feasibility of superfilling behavior enabled by the additive chemistry.
The microstructure and thermal stability of the as-deposited nanocrystalline Cu films were systematically evaluated. After annealing at 150 °C for 2 hours, the average grain size increased only slightly from 132.8 nm to 145.2 nm, demonstrating excellent thermal stability and resistance to spontaneous grain growth at room temperature. Morphology was further characterized using atomic force microscopy (AFM), while mechanical and electrical properties were assessed via hardness testing and sheet resistance measurements.
Cu–Cu bonding was performed at 200 °C under 22 MPa for 1 hour. Post-bonding annealing at 280 °C for 6 hours was carried out to evaluate interfacial grain growth. The bonding interfaces were analyzed by focused ion beam (FIB) microscopy to examine grain coalescence and migration behavior. Preliminary results revealed significant grain coalescence across the bonding interface after thermal treatment, indicating successful grain-boundary-mediated bonding and the potential for enhanced joint integrity.
To further validate the mechanical reliability of the bonded joints, shear strength testing is planned. Based on the observed microstructural evolution and the high-density grain boundaries in the nanocrystalline Cu, strong interfacial adhesion is anticipated. These findings support the feasibility of employing (111)-oriented nanocrystalline Cu films for low-temperature Cu–Cu bonding in advanced electronic packaging applications.
Overall, this study demonstrates a practical approach for fabricating thermally stable nanocrystalline Cu with a strong (111) surface orientation to meet industrial demands for low-temperature bonding. The results contribute to the development of more robust Cu interconnects suitable for next-generation electronic packaging technologies.

 

Bio:
Lee Kuan Fang origin from Malaysia, is a Principal Engineer of X-Fab Sarawak Sdn. Bhd. (an analog/mixed-signal semiconductor applications foundry). He is responsible for the quality and reliability assessment of wafer products. Bond pad quality and wire bonding process is the area he focused on, as well as electromigration test of the wafer devices. He is also experienced in the investigation of chip assembly failure and failure analysis of wafer level reliability failure analysis. He is involved actively in field quality feedback and investigation of various assembly packaging issues.

Abstract:
Abstract — Long term microelectronic device reliability is important for crucial applications. There is a high percentage of field reliability failure related to wire bond defects. This pointed out that device reliability and system reliability are significantly affected by wire bond reliability. In the scenario of gold wire bonding on aluminum metallization, Excessive intermetallic compound (IMC) growth is a major reliability concern. IMC growth is contributed by gold and aluminum diffusion processes. This is an ageing process that will eventually cause high resistance and open circuits of the product. In this study, aluminum bond pads with various characteristics were carefully evaluated to understand IMC growth behavior. Physical factors such as aluminum layer thickness and grains size are the focus of the study. Gold wire material was fixed during wire bonding for all samples, as well as wire bonding parameter settings. Subsequently after wire bonding, accelerated ageing conditioning was applied to the wire bonded samples. High Temperature Storage Life (HTSL) and Temperature Cycle (TC) are two common accelerated ageing methods for device reliability assessment. HTSL treatment exposed wire bonded samples with continuous high temperatures for a long duration. Meanwhile, TC treatment conditioned the sample between cold and hot environments. Both ageing methods will accelerate IMC growth of Au-Al bonding. Ball bond cross-sectioning was conducted at various phases of HTSL and TC. SEM images were obtained from cross-sectioned samples for close examination of IMC growth patterns. This is followed by a wire bond shear test to evaluate the integrity of Au-Al bonding. The level of IMC growth influenced by aluminum layer thickness and grain size are highly interested in this study. These two characteristics are determined by deposition rate and temperature used in wafer fabrication process. In view of this, data obtained from this study is valuable information for bond pad structure design and improvement activity in wafer fabrication process.
Keywords— Intermetallic growth, aluminium thickness, aluminium grain size, HTSL, TC.

I. INTRODUCTION

Good reliability of microelectronic devices is highly important to ensure long term functionality for crucial applications (such as automotive, healthcare, and telecommunication etc.). It has been reported that more than 30% to 40% of the field reliability failures are due to wire bond failure. Intermetallic compound (IMC) formed during wire bonding process plays a significant role in determine device reliability. [1]
IMC or purple plague is an alloy formed at the interface of gold wire and aluminium bond pad during thermosonic wire bonding. The type of intermetallic compound depends on the temperature of exposure and length of time exposure. Usually, Al5Au2 is the phase that is formed. Nevertheless, many different phases (such as AuAl2, Au5Al2, and Au4Al) are formed in the subsequent processes that involved heat treatment to the wire bonded sample.[2] IMC showing different characteristics in term of electrical resistance, thermal expansion co-efficient and hardness properties as compared to its original form of gold and aluminium. It is necessary for good interfacial adhesion of gold wire and aluminium bond pad. However, it also affects device reliability in two different ways, namely excessive IMC growth and brittle fracture of IMC. At the interface of ball bond, diffusion of gold and aluminium happen in opposite direction into each other to form a compound. However, when one of the atomic species diffuses faster than the other, it leaves vacancies behind. These vacancies may cluster together and form a void. Typically, the voids formed around bond periphery. This may cause high electrical resistance and lead to open circuit failure. The bonding interface will become mechanically weak with the presence of severe voids (such as Kirkendall voids) (Figure 1). Gold-aluminium intermetallic compounds are stronger than the pure metals. Nevertheless, they are also more brittle. When a micro defect exists in the intermetallic and subject to thermal cycling conditions, the crack will propagate at a high speed and causes brittle fracture of the IMC.[3], [5]-[8]

As a source of aluminium, this experiment was focus on bond pad characteristics that might be key factors of IMC growth. Aluminium layer thickness and grain size were included in the study. This information will also serve as important reference for bond pad structure design and aluminium deposition process finalization.