Interconnect

Generative A.I.’s (GAI) popularity has made photonics based computation an attractive approach for its potential to meet the demands for higher energy efficiency performance (EEP). However, previous optical solutions for multiply-accumulate (MAC) operations focused on either analog architecture [1-7] which is limited by its precision and data conversion, or free-space optical architecture with limited scalability [8]. Here, a world’s first on-chip large-scale Digital Optical Computing System (DOC) for GAI training is reported. DOC employs a novel wafer-based system integration technology featuring multilayer low-loss photonic interconnect fan-out (PIFO) and EIC/PIC stack architecture leveraging TSMC SoICR. It reduces the data movement and memory hierarchy leading to improvement in critical path latency and system energy efficiency (EE). Compared to conventional electrical designs, DOC can be scaled to larger coherent networks and operate at higher speeds with lower energies per MAC operation. A low energy consumption of <0.08 pJ/MAC at 8-bit operation with a >20x improvement in EEP compared to the state-of-the-art GPU [10] is achieved for a 512 x 512 MAC large scale operation. The EEP further improves at higher precisions due to the relative minimal fan-out energy. This architecture has full potential for continuous EEP scaling in future generations.

TSMC has developed an advanced silicon photonics foundry platform tailored to meet the increasing demands of next-generation data communication applications. This paper presents an overview of the platform and the performance of key photonic devices.

One of the prominent challenges for widespread adoption of Si photonics (SiPh) technology is the availability of an integration platform that can simultaneously meet a wide range of power, performance, and cost criteria in different applications. As a result, there is a diversity of Si photonics integrated solutions proposed or demonstrated, but none is considered as a common solution. In this paper, we will first survey industry proposed photonic engine (PE) structures in monolithic, 2D, 2.5D, and 3D on their strengths and weaknesses. We will then propose a compact and universal PE structure- COUPE (COmpact Universal Photonic Engine) that could consolidate different requirements onto the same integration platform. COUPE has the EIC-PIC integration with the electrical interface designed to minimize the EIC-PIC coupling loss. Compared with industry proposed PE technology, COUPE can provide low insertion loss for both grating coupler (GC) and edge coupler (EC). For either GC or EC, the COUPE is a solid structure without cavities or mechanically weak parts, thus enabling low insertion loss without contamination or mechanical concerns. COUPE also has the flexibility to be integrated easily with host ASIC to form a co-package structure. The COUPE integration scheme can meet the most demanding system requirements and pave the way for silicon photonics based wafer level system integration (WLSI) for high performance computing (HPC) applications.

An array-based test-vehicle for tracking bit-error-rate (BER) degradation of signal interconnects subject to DC electromigration (EM) stress was implemented in a 16nm FinFET process. A unit interconnect path comprises five identical interconnect stages where each wire is driven by inverter based buffers. Accelerated EM stress testing is achieved entirely on-chip using metal heaters located directly above the devices-under-test (DUTs) and separate stress circuits driving both ends of the wire. BER measurement results from four individual interconnect paths are presented and analyzed.

A direct silicon water cooling solution using fusion bonded silicon lid is proposed. It is successfully demonstrated as an effective cooling solution with total power >2600 W on a single SoC, equivalent to power density of 4.8 W/mm2. Low temperature logic chip to silicon lid fusion bonding, with trench/grid cooling structure cutting into silicon lid enables minimal thermal resistance between active device and cooling water and best cooling efficiency. Direct water cooling on logic chip silicon backside has also been demonstrated with power density better than 7 W/mm2.

The RC delay, electro migration (EM) and TDDB performance become more challenges to meet device requirement as continuous geometry shrink on BEOL dual damascene interconnects. To overcome these challenges from interconnect patterning point of view, we proposed Cu subtractive RIE as a potential solution for next generation Cu/Low-k interconnects.

As dimension shrinks the volume percent occupied by conventional barrier and liner increases and line resistance (Rs) and via resistance (Rc) increases dramatically. An ultrathin ALD MnN barrier is being evaluated as a single layer barrier for resistance reduction in small structures. >20% and >80% Rs and Rc reduction was demonstrated, while 4× better mean time to failure (MTTF) on the time dependent dielectric breakdown (TDDB) was achieved comparing to conventional barrier/liner. ALD MnN is a potential barrier candidate for future interconnects technology.

In this work, a low-resistance and low-cost PVD-TiZrN barrier is evaluated for BEOL interconnect. Comparing to conventional PVD barrier, comparable Cu barrier and Cu wetting properties are obtained. Moreover, up to 55% of via resistance reduction is achieved, with comparable voltage breakdown performance comparing to conventional one.

High stresses generated from chip-package interactions (CPI), especially when large die is flip mounted on organic substrate using Pb-free C4 bumps, can easily cause low-k delamination. A novel scheme by applying an elastic material can effectively reduce the transmitted stresses and, thus, resolve the interfacial delamination issue. Along with an optimized chip-package integration solution, a reliable interconnect structure with good electrical performance, has been successfully demonstrated.

This study evaluated plasma treatment processes on 193i and EUV photoresist to improve the line width roughness (LWR) performance in porous low-k/ultra-thin barrier Cu interconnect. We successfully demonstrated 20% LWR reduction for 193i PR and 11% for EUV PR. Furthermore, the influence of LWR on reliability was evaluated on 45nm line-width test vehicle. A boost of 10 times Time Dependent Dielectric Breakdown (TDDB) and 2 times Eelectrical Migration (EM) was demonstrated.