Comparison of thermal performance between glass and silicon interposers

Abstract

This paper compares thermal performance of glass and silicon interposers for mobile applications, using computational modeling. It is well known that while silicon is a good thermal conductor, glass is a poor conductor, potentially making it unsuitable for packaging applications. This study proposes to address this short coming of glass by comparing and contrasting with silicon. In this study, for more accurate thermal analysis, effective thermal conductivity of glass interposer substrates is measured by infrared microscopy. Subsequently, equivalent thermal conductivity of TPV (Through Package Via) is calculated through numerical analyses. For comparison of thermal performance of glass and silicon interposers, 2.5D interposer structures with logic and memory chips are considered. The comparison shows that by incorporating thermal vias, junction temperature for the glass interposer decreases by about 60%, while junction temperature for silicon interposer decreases 45%, making both acceptable. However, the glass interposer provides significantly better thermal isolation between logic and memory chips. Glass and silicon interposer structures placed in enclosures, representative of mobile applications, provide comparable performance in the presence of thermal vias. IntroductionThe 2.5D and 3D interposer technologies offer promising approaches for future microelectronic packaging, addressing miniaturization, cost and performance. These interposers connect multiple functional dies with a fine pitch microbumps and high density through-via interconnect technologies. In 2.5D technology, interposer, an intermediate substrate to which components are attached prior to PCB, is used to provide dense interconnections between multiple dies. Such a technology provides increased performance at reduced power consumption and yet enables thermal dissipation by enabling side-by-side, die to die interconnections, thereby improving the overall system performance.Silicon interposer technology is being developed to replace organic substrates whose performance is limited due to warpage, wiring density, number of I/Os, and acceptable cost. However, it faces a number of challenges, such as limited wafer size leading to high fabrication cost. Glass interposer has the potential to be the best packaging material except for its poor thermal conductivity. If this problem is addressed, It can replace organic substrates and silicon interposers due to a number of its advantages, including ultrahigh resistivity, low electrical loss, low dielectric constant, availability in thin and large sizes, and lower cost. This paper addresses the main problem with glass-its low thermal conductivity which can result in poor thermal dissipation. One

Communication—Exploration of Plasma Oxidized Copper Oxide as a Copper Passivation Layer

“…For example, the glass substrate dissipates heat poorly as compared to the Si wafer, i.e., 1 W m -1 •K −1 vs 150 W m -1 •K −1 . 36 Previously, it was reported that Cu lines on the Si substrate have longer lifetimes than Cu lines on the glass substrate, 15,16,22,27,35,37 which may be due to the substrate's heat dissipation capability. Also, the as-sputtered deposited Cu thin film has a larger resistivity than the electroplated Cu film has.…”

Multi-scale thermal modeling of glass interposer for mobile electronics application

“…However, using average die temperature is still a valid approach to evaluate and compare thermal characteristics of electronics packaging while a uniform heat generation is assumed. The approach was validated under uniform heat generation boundary condition by comparing the simplified model with detailed model, which showed only~1 percent difference between maximum temperatures predicted by the two models (Cho et al, 2013). More detailed validation of the approach under various geometric conditions is provided in the literature (Ma et al, 2014).…”

“…Heinig et al (2014) presented thermal analysis and optimization results for various 2.5D and 3D integrated processor configurations. These results indicated that maximum total power of the processor on 25 mm × 16 mm interposer can be increased up to 10 W when there is convective heat removal on bottom side with an effective heat transfer coefficient of 50 W/m 2 K. Nonetheless, most of the previous works focus on silicon-based integration technologies and thermal studies on glass-based integration technologies are currently lacking (Cho et al, 2013;Oprins and Beyne, 2014). This paper is organized into three parts.…”

Experimental demonstration of the effect of copper TPVs (Through package vias) on thermal performance of glass interposers

“…Its thermal conductivity (~1 W/m·K) is an order of magnitude higher than organics, but two orders of magnitude lower than silicon (~150 W/m·K). However, this limitation of glass interposers can potentially be overcome by implementation of large number of high thermal conductivity (~400 W/mK) copper through package vias (TPVs) [2]. If routing space allows, additional copper TPVs can be utilized as purely thermal vias, which can further improve thermal conduction in glass interposers.…”