Fig. 1. Silicon-based photonics-electronics convergence

Fig. 2. Research progress in silicon high-speed electro-optic modulators. (a) McGill University^[20]; (b) Huazhong University of Science and Technology^[21]; (c)‒(d) Laval University^[22-23]; (e) Peking University^[24]; (f) Yokohama National University^[25]; (g) National Institute of Information and Communication Technology^[26]; (h) Zhejiang University^[27]; (i) Advanced Micro Foundry^[28]; (j) Intel Corporation^[41]; (k) Ghent University^[42]; (l) National Information Optoelectronics Innovation Center^[45]; (m) Intel Corporation^[46]; (n) Advanced Micro Devices^[48]; (o) Ghent University^[47]; (p) Hewlett Packard Laboratories^[38]

Fig. 3. Research progress in Si/Si-Ge (avalanche) PDs. (a) National University of Singapore^[49]; (b) Huazhong University of Science and Technology^[50]; (c) Leibniz Institute for High-Performance Microelectronics^[51]; (d) Massachusetts Institute of Technology^[52]; (e)‒(i) Huazhong University of Science and Technology^[53-56,63]; (j) Fujitsu^[64]; (k) Hewlett-Packard Laboratories^[60]; (l) Huazhong University of Science and Technology^[61]; (m)‒(n) Hewlett-Packard Laboratories^[61,65]

Fig. 4. Schematic cross-section of Global Foundries 45 nm monolithic integration process^[69]

Fig. 5. Applications of silicon photonics in optical communications. (a) Global internet user growth^[73]; (b) worldwide IDC global DataSphere forecast^[77]; (c) continued growth of power consumption in communication technologies^[79]; (d) evolution of integrated device count across different material photonic integration platform^[82]

Fig. 6. Research progress in silicon photonic multi-channel optical transmitter and receiver chips. (a) Low-power 16-channel parallel silicon MZM transmitter^[84]; (b) linearly driven high-density 16-channel parallel silicon-based MZM transmitter^[85]; (c) 3D-integrated low-power and high-bandwidth 80-channel WDM silicon-based MRM transmitter^[86]; (d) grating-based 8-channel Lan-WDM silicon-based photonic transceiver^[87]; (e) monolithic 8-channel WDM transmitter based on silicon-based MRM^[88]; (f) 4-channel polarization-insensitive WDM silicon-based receiver based on dual-ring filters^[89]; (g) Kerr comb-driven 32-channel WDM silicon-based MRM transceiver^[90]

Fig. 7. Interconnection within large-scale data centers. (a) Typical hyperscale data center network architecture^[91]; (b) optical module supply chain in 2024^[95]

Fig. 8. Advances in CPO. (a) Bandwidth evolution trends in data center switches^[83]; (b) progression of optical interconnect technologies^[82]; (c) Broadcom CPO switches^[96]; (c) NVIDIA CPO switches^[4]

Fig. 9. Next-generation PAM/coherent communications experimental systems. (a) Aloe Semiconductor’s single-wave 425 Gbit/s dual-polarization PAM-4 silicon transmitters^[97]; (b) University of British Columbia's single-wave 528 Gbit/s dual-polarization coherent receiver^[98]

Fig. 10. Optical interconnect development roadmap ^[99]

Fig. 11. Five types of mainstream OPAs. (a) 1D dispersion-assisted OPA^[103]; (b) 1D phase-sweep OPA^[104]; (c) combined 1D dispersion-assisted and 1D phase-sweep OPA^[105]; (d) 2D phase-sweep OPA^[106]; (e) 2D dispersion-assisted OPA^[107]

Fig. 12. FMCW-OPA ranging system architecture

Fig. 13. Core components of FMCW-OPA scheme on silicon platforms. (a) External cavity laser^[108]; (b) on-chip 2D sparse matrix-based beam steering scheme^[109-112]; (c) on-chip coherent reception scheme^[101]; (d) 3D CPO OPA chip^[113]

Fig. 14. Aperture-beam angular resolution relationship. (a) Main beam accuracy of subwavelength-spaced arrays with different apertures; (b) survey of notable works on aperture scaling since 2009^{［113-117］}

Fig. 15. Recent typical research progress of silicon FMCW-OPA lidar^{［113，117，128-140］}

Fig. 16. Typical application scenarios of lidar. (a) Environment sensing and obstacle detection in autonomous driving^[141]; (b) industrial inspection, security monitoring, and robot navigation^[142]; (c) depth perception and interaction enhancement in AR/VR scenarios^[143]

Fig. 17. ANN equivalent mathematical model and silicon-based photonic computing accelerator architectures. (a) ANN equivalent mathematical model comprising linear matrix-vector multiplication computation and nonlinear activation functions; silicon-based photonic computing accelerator architectures including: (b) MZI mesh-based architecture; (c) MRR-based architecture; (d) intensity modulation array-based architecture; (e) metasurface diffraction-based architecture

Fig. 18. Development timeline of silicon-based on-chip optical computing accelerators. (a) Based on MZI meshes^{[152, 158-160]}; (b) based on MRR weight banks^{[161,163,165-166]}; (c) based on intensity-modulator arrays^{[154, 167-168]}; (d) based on on-chip diffractions^{[155, 170-172]}; (e) based on non-volatile phase change materials^[173-176]; (f) other materials^[177-180]

Fig. 19. Application scenarios of integrated optical computing chips

Fig. 20. Schematic diagram of silicon optical computing architecture ^[184]