Search

Haptics in virtual reality has become increasingly important for improving realism and immersiveness. However, matching the complexity of real-world tactile interactions is a challenging problem due to the complexity of our sense of touch. Motivated by the recognition that multi-modal haptic feedback is needed to fully recreate human touch, we created PATCH, a Pump-Actuated Thermal Compression Haptic device for presenting simultaneous thermal and compression cues. The system uses water of varying temperature to provide compression cues and transfer heat to or from the user's skin. The wearable component of the system is constructed solely of soft, flexible components. When compared to an established, unimodal, pressure-based haptic device, the PATCH system was found to perform similarly in terms of recognition and saliency, but it was rated more favorably in terms of wearability. The PATCH device can display temperatures ranging from 17° C to 42° C, fitting the desired temperatures with an R2 = 0.75, and can display pulsed cues at a rate of 0.22 Hz. The success of our PATCH system can serve to inform the development of the next generation of multi-modal haptic devices.

Driven by massive cost down demands fan-out packaging is facing now a transition from wafer to panel level. To address the market needs a panel level fan-out bonder has to provide high productivity exceeding 7000 UPH as a key requirement. On the other hand embedded multi-die interconnect bridge packaging (EMIB) requires very high placement accuracy with a roadmap pointing to sub-micron level. This paper presents a panel level die-attach machine suitable for substrates up to 730 mm x 920 mm (GEN 4.5). The bonder provides global placement accuracy of 2 µm @ 3σ and local placement accuracy of 1 µm @ 3σ at 6000 UPH, the bonder's productivity can be scaled up to 7000 UPH. The machine concept is based on a building block approach and therefore can be configured for various fan-out pick and place processes, for silicon bridge embedding, thermal compression, and other processes.

Grain size and orientation of microbump play very important roles on thermal-cycling test and electromigration failure. In this study, Cu/Sn-2.5Ag/Cu and Ni/Sn-2.5Ag/Ni microbumps assembled by thermal-compression bonding are investigated grain size and orientation by electron-backscattering diffraction inverse pole figure. Cu/Sn-2.5Ag/Cu microbumps appear much more single crystal-like structures than Ni/Sn-2.5Ag/Ni. By the distribution of grain size, a large quantity of the small Sn grains appears in Ni/Sn-2.5Ag/Ni microbump due to the very low solubility of Ni in Sn. Finally, Cu/Sn-2.5Ag/Cu microbumps possess more grain areas with the lower angles between c-axis and normal dirction.

We demonstrated surface tension-driven self-assembly of chips with Cu/Sn-Ag microbumps in order to satisfy requirements for both high throughput and high alignment accuracy toward 3D system integration. The chips were singulated with different dicing methods: standard single-cut, precise single-cut, and modified step-cut. The alignment accuracies were compared among the three methods. The chips obtained by modified step-cut were precisely aligned within approximately 2 µm and comparable to that obtained by precise single-cut. By optimizing liquid volumes, the step-cut chips having Cu/Sn-Ag microbumps were accurately self-assembled irrespective of microbump densities. The self-assembled chips were successfully bonded at 280°C by thermal compression. The Cu/Sn-Ag daisy chains indicated good electrical characteristics with a resistance of 35 mΩ/joint.

We have proposed a new multichip-to-wafer 3D stacking method with high throughput and high yield based on a capillary self-assembly method using liquid droplets. In this paper, we optimized conditions in self-assembly and microbump bonding using non-conductive film (NCF)-covered known good dies (KGDs). Self-assembly of the NCF-covered KGDs provided high chip alignment accuracy within approximately 1 µm. After the self-assembly and a subsequent thermal compression, resultant microbump chains composed of over 7,000 microbump joints exhibited good electrical properties of 32 mΩ/joint without bridge short and open failures. The microbump joint resistance varied within 5% of the initial values after thermal cycle test (TCT) of 1,000 cycles. In addition, we demonstrated a multi-layer 3D stacking by the self-assembly method with the NCF-covered KGDs.

In this study, Cu-Ag composite pastes with the mixture of Cu@Ag core-shell particles and Ag nanoparticles prepared by one-step spray pyrolysis are developed for microelectronic bonding applications. Cu to Cu bonding can be successfully achieved by thermal compression with the Cu/Ag pastes at 250°C. The joint strength reaches 28.2 MPa and 32.4MPa respectively when bonding pressure holds at 5 MPa and 10MPa at 275°C. It has been demonstrated the joints thus formed remain robust at elevated temperatures up to 250°C and excellent reliability subjected to thermal cycling ranged from −65 °C∼150 °C.

This paper summarizes key learnings on 20/16nm CPI (Chip-Package-Interaction) challenges at 100um pitch and below to support ever increasing performance/cost/form factor demands for high performance mobile SoCs. CPI solutions for two types of Cu pillar interconnects using mass reflow and thermal compression type assembly process respectively are studied in technology development/production, and separate bump cell structures are proposed.

High through put thermal compression NCF bonding was studied and the new process consisting of dividing pre and main bonding, and the multi die gang main bonding has been developed. The dividing could change the process from serial to parallel and enabled to use the constant heated bonder head, which eliminated the time consuming head cooling process of the conventional serial thermal compression bonding. The die of 7.3 × 7.3 × 0.1 mm size with bumps of 38 × 38 μm 2 square Cu pillar covered by Sn-Ag cap, which had the pitches of 80 μm at peripheral and 300 μm at corer area, and the organic laminated substrate with Cu/OSP trace were used as the test vehicle in this study. Firstly, the dividing of pre and main bo ndi ng process in the case of si ngle die was investigated. The prebonding was the die placement to the NCF on the substrate, which was carried out at 150°C for 0.5 second. The substrate was kept at 80°C during the process. After the pre bonding the test vehicle was removed out from the equipment and cooled down to room temperature. And then it was mounted back to the equipment again and main bonding was carried out at 240°C for 20 seconds. The same substrate temperature as the pre bonding process was kept. Solder joint formation and NCF curing was made at the process. The assembled test vehicle was evaluated. The cross sectional observation results showed that the bump solder wetted the Cu trace on the substrate and no void was detected in the NCF by C-SAM observation. Secondly, the multi die main gang bonding was studied. The equipment was newly designed and built. 15 dies were pre bonded on the substrate with the same condition as that of the single die experiment. After the pre bonding was finished, the substrate was moved to the main gang bonder. During the transportation the substrate was cooled down to room temperature. The 15 dies were bonded at one time at 240°C for 10 seconds. The substrate was heated at 240°C during the process. The evaluation of the assembled dies revealed that the solder wettability of the joints and void detection in the NCF was almost the same as those of the single die pre and main divided bonding. This main bonding process time corresponded to 2700 UPH.

Ni/Sn solid-state reaction might induce the void formation with low-volume solder under a severe space confinement in 3D IC packaging, and Ag addition could effectively eliminate void formation. In the present study, the different Ag concentration in solder was carried out to figure out the lower limit of Ag concentration to avoid void formation. The solid-state diffusion couple, Ni/Sn-xAg/Ni (x = 0 wt.%, 1.0 wt.%, 2.4 wt.%, 3.5 wt.%, 4.5 wt.%, 8.0 wt.%), was prepared by thermal compression. The isothermal aging process was carried out at 200°C. There were only two IMCs being observed during the entire aging process: Ag 3 Sn and Ni 3 Sn 4 . The Ag 3 Sn initially formed as fine and eutectic network in the Sn matrix after cooling. The Ag 3 Sn particles were found eventually located at the middle of sandwich structure. After Sn was totally consumed, there was very different microstructure in the sandwich structure in different Ag concentration.

In order to overcome problems associated with thermal compression flip chip bonding with post-applied liquid underfill, we have developed a novel process with pre-applied Non Conductive Film (NCF) underfill. This paper describes failure modes in post-applied liquid underfill and the development of the novel NCF process which has several advantages for IC-package assembly. Along with Au bump, this process is applicable to Sn/Ag capped Cu-pillar with ideal electrical connections and high reliability performance.

In flat panel displays, the assembly format of driving ICs, due to the rectangular chip shape, is usually with a difference of conventional commercial ICs constructed with a square configuration. Because of this concern, the electroplating gold bump technology in package was desirably developed. Additionally, in the characteristics of hardness, melting point, and stability, gold material is generally better than tin material. Although tin in cost is lower than gold, the previous just can be applied to the soldering joint, but not to the thermal compression process. Therefore, in this study, the electroplating gold bump in assembly was probed in detail to analyze the possibility of mass-production.

In training neural networks, it is important to reduce input variables for saving memory, reducing network size, and achieving fast training. The paper proposes two kinds of selecting methods for useful input variables. One of them is to use information of connection weights after training. If a sum of absolute value of the connection weights related to the input node is large, then this input variable is selected. In some cases, only positive connection weights are taken into account. The other method is based on correlation coefficients among the input variables. If a time series of the input variable can be obtained by amplifying and shifting that of another input variable, then the former can be absorbed in the latter. These analysis methods are applied to predicting cutting error caused by thermal expansion and compression in machine tools. The input variables are reduced from 32 points to 16 points, while maintaining good prediction within 6/spl mu/m, which can be applicable to real machine tools.