Integrated Circuits and Hardware AI
IME’s integrated circuits and hardware AI development team is supported by established research work in integrated circuits design and development of computing hardware, which has enabled an extensive range of IP building blocks including ultra-low-power sensor node processor (SNP), power management integrated circuit (PMIC), sensor readout IC, radio transceivers, hardware securities, as well as capabilities in system-on-chip design. Designed for supporting desired attributes of next-generation innovative devices and systems such as miniaturization, integrated power management, energy efficiency and wireless connectivity for AI-enabled Internet-of-Things (IoT), these capabilities serve well for the development of hardware AI targeting applications such as autonomous vehicles, machine health monitoring, environment sensing, surveillance, vitals monitoring, and data computing. IME’s integrated circuits team are focusing on the development of two AI technology platforms, specifically neuromorphic computing and deep learning. These platforms represents the two main memory-centric computing architectures that we are pursuing: (1) the deep learning-based digital approach which uses CNN or RNN and (2) neuro-based SNN analog approach. By employing such approaches with emerging NVM and TSVs, energy-efficient, high computational throughput and accuracy AI hardware can be demonstrated.
As device scaling powered by Moore’s law faces challenges in meeting industry requirements, advanced packaging and interconnect technologies have emerged as an alternative solution to achieve performance, power, form-factor, and cost requirements that are critical to key industry drivers such as mobile, 5G, data centres, automotive, and IoT. IME offers a wide range of advanced packaging technology solutions to address the above industry needs. Our platforms include 2.5D through-silicon interposer, 3D TSV, fan-out wafer level packaging, MEMS wafer level chip-scale packaging, fine-pitch Cu pillar, chip-on-wafer (direct Cu-to-Cu) bonding, electronic-photonic integration, as well as thermal solutions.
Flat optics have shown great potential to replace conventional optics components such as focal lens, color display, polarization bandpass filter and dot projector with its advantages of ultra-thin, flexible and nearly zero distortion value. The promising applications of flat optics are on 3D sensing, augmented reality (AR) / virtual reality (VR) devices, wearable devices and smart phones. Flat optics is ultra-thin and based on metasurface materials. Flat optics is able to control optical phase and polarization in sub-wavelength scale with very small distortion values, as compared to the conventional optics which is bulky, and is limited to control phase and polarization based on the finite design, where the distortion value is always an issue of quality. IME’s advanced optics program develops flat optics with ultra-fine light structures, works on design and material optimization, integration and device optimization, packaging and product development and application. IME has the full facility of advanced 12 inch semiconductor fabrication tools for flat optics R&D and mass production. IME provides total solutions from design, mask tapeout, fabrication processing (film deposition, patterning with immersion scanner and etching tool), packaging and final product testing.
Sensors, Actuators and Embedded Memory
Sensors development and storage solutions form the research focus of IME’s sensors, actuators and embedded memory team. IME’s sensors development leverages its comprehensive proprietary MEMS process technology platforms, especially for AlN MEMS capabilities, such as hybrid integrated thin film AlN, AlN on SOI platform and monolithic thin film piezoelectric platforms that address industry requirements in performance, cost and size to meet market needs for IIOT applications. IME’s digital and analog ReRAM technology storage solutions are suitable for edge computing, deep learning and neuromorphic computing applications. IME’s oxide-based ReRAM devices have been optimized to provide high speed, low power and good reliability. The device stack is highly scalable (for high memory density) and compatible with standard CMOS process flow.
IME’s medtech research development can be broadly classified into two areas, namely diagnostic platform technologies for molecular and cellular diagnosis, and sensor development platform for invasive/non-invasive medical devices. IME collaborates with renowned academic institutes, biologists, clinical partners/hospitals and leading biomedical companies to develop cutting edge technologies in the field. In additional, IME provides a range of specialty services to support and cater to biomedical companies’ R&D requirements. The bioelectronics team develops advanced biosensors and biochips for healthcare and life sciences applications. The R&D efforts have resulted in development of performance enhancing diagnostics platform technologies, which cover advanced detection methodologies coupled with microfluidic sample manipulation. The biomedical device team designs, fabricates micro-electro-mechanical system (MEMS) sensors and bio-packages with application-specific-integrated-circuits (ASIC), wireless data transmission module using flexible electronics. The highly integrated medical devices can measure various physiological signals from the human body for healthcare applications.