Experimental investigation of silicon and silicon nitride platforms for phase-change photonic in-memory computing

Abstract

Advances in artificial intelligence have greatly increased demand for data-intensive computing. Integrated photonics is a promising approach to meet this demand in big-data processing due to its potential for wide bandwidth, high speed, low latency, and low-energy computing. Photonic computing using phase-change materials combines the benefits of integrated photonics and co-located data storage, which of late has evolved rapidly as an emerging area of interest. In spite of rapid advances of demonstrations in this field on both silicon and silicon nitride platforms, a clear pathway towards choosing between the two has been lacking. In this paper, we systematically evaluate and compare computation performance of phase-change photonics on a silicon platform and a silicon nitride platform. Our experimental results show that while silicon platforms are superior to silicon nitride in terms of potential for integration, modulation speed, and device footprint, they require trade-offs in terms of energy efficiency. We then successfully demonstrate single-pulse modulation using phase-change optical memory on silicon photonic waveguides and demonstrate efficient programming, memory retention, and readout of > 4 bits of data per cell. Our approach paves the way for in-memory computing on the silicon photonic platform. Published by The Optical Society under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.

Programmable phase-change metasurfaces on waveguides for multimode photonic convolutional neural network

“…For example, assuming a moderate datarate of 10 Gbits/sec and 4 WDM wavelengths in parallel per channel, the computing density will reach an upperbound value of 25 TOPS/mm 2 (Tera-operations per second per mm 2 ), which is significantly higher than that of digital electronic accelerators such as GPUs and tensor processing units (TPUs) 60 , 61 . Using silicon instead of silicon nitride can further reduce the device footprint to increase the computing density 62 . Besides MAC operation, the equally important computing processes of applying nonlinear functions and pooling can also be achieved optically by using elements such as nonlinear optical resonators, modulators, and amplifiers 27 , 28 , 63 .…”

Overview of Phase-Change Materials Based Photonic Devices

“…Their work was originally addressed to the problem of SiN-based PICs having a fairly large footprint and offering relatively low contrast. It is worth noting that Li et al [89] actively compared the relative merits of Si and SiN platforms ( Figure. 9(b)) and found that Si is better in relation to its potential for integration with PCMs, its modulation speed and its footprint ( Figure. [87,88] was to introduce GST-clad microring resonators and directional couplers that could be tuned according to the state of the GST ( Figure. 9(e)). They also experimented with using two low loss PCMs, Sb 2 S 3 (SbS) and GeSe, with the electrical actuation of SbS being accomplished on-chip by using an external indium tin oxide (ITO) heater.…”

“…As mentioned above, basic tuning of the condition of the platform was accomplished by on the one hand achieving amorphization of the GST by means of a single short laser pulse (Figure 9(f)), on the other hand heating the GST to above its crystallization temperature by using a repeated sequence of short laser pulses or a single long pulse (Figure 9(g)). The heat dissipates more quickly in Si than for SiN when applying the same amount of optical power, so short, higher amplitude pulses tend to be more effective for PCMs [89]. By proceeding in this way, Fang et al [87,88] enabled highly controllable multi-level switching of the microrings by being able to shift the GST to different and intermediate states between its fully amorphous and fully crystalline states, each with different refractive properties ( Figure 10(a)).…”

“…By optimizing the structure of these two waveguides it was possible to control the phase-matching condition for the GST by switching the evanescent coupling of the input TEpolarized light. Others have also noted the potential to dope Si waveguides with PCMs to act as effective heaters because of their attrative thermal conductivity (Figures 10(c) and (d)) [89,90]. Fang et al's decision to use ordinary Si waveguides and coat them with GST is important because Si and SiN have very different properties regarding thermal conductivity and heat capacity, with Si having higher thermal conductivity and lower evanescent coupling [89].…”

“…Others have also noted the potential to dope Si waveguides with PCMs to act as effective heaters because of their attrative thermal conductivity (Figures 10(c) and (d)) [89,90]. Fang et al's decision to use ordinary Si waveguides and coat them with GST is important because Si and SiN have very different properties regarding thermal conductivity and heat capacity, with Si having higher thermal conductivity and lower evanescent coupling [89]. Fang et al [87,88] experimented with using both 1 x 2 DC and a 2 x 2 DC switches.…”