Yeyu Tong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yeyu Tong has a total of 148 co-authors. Here are the top co-authors by number of shared publications:
- Hon Ki Tsang
The Chinese University of Hong Kong
28 shared publications Hao Chen
Microelectronics Thrust of The Hong Kong University of Science and Technology (Guangzhou)
14 shared publicationsWu Zhou
Microelectronic Thrust
12 shared publications- Hao Chen
The Hong Kong University of Science and Technology (Guangzhou)
11 shared publications - Kaihang Lu
The Hong Kong University of Science and Technology (Guangzhou)
11 shared publications - Wu Zhou
The Hong Kong University of Science and Technology (Guangzhou)
11 shared publications
Latest Publications
- 9 October 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (20) , 5798-5801
Low-parasitic-resistance and fabrication-tolerant silicon microring modulators operating beyond 100 GHz
Silicon microring modulators (MRMs) are promising for compact, energy-efficient optical interconnects, yet their electro-optic bandwidth is constrained by the RC time constant, and their resonance wavelength is highly susceptible to fabrication variations. In this work, we present a carrier-depletion MRM that mitigates these issues by integrating an adiabatic transition with a widened low-parasitic-resistance region. The design extends the low-resistance cross-section over 61% of the cavity circumference, reducing the junction series resistance to 36.9 ohms, which is 35.9% lower than conventional designs. Across eight dies, the resonance wavelength spread is reduced from 3.58 to 1.07 nm. The modulator exhibits a measured 3-dB electro-optic bandwidth exceeding 100 GHz over a bias range of −1 to −5 V. It supports 144 Gbps NRZ and 224 Gbps PAM-4, with the latter demonstrating a bit error rate below the 7% hard-decision forward-error-correction threshold. Moreover, the device can operate in a self-bias mode at 0 V bias, achieving 128 Gbps NRZ with an estimated energy consumption of 0.861 fJ/bit, while depletion-mode operation consumes 8.878 fJ/bit.
- 9 October 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (20) , 5798-5801
Low-parasitic-resistance and fabrication-tolerant silicon microring modulators operating beyond 100 GHz
Silicon microring modulators (MRMs) are promising for compact, energy-efficient optical interconnects, yet their electro-optic bandwidth is constrained by the RC time constant, and their resonance wavelength is highly susceptible to fabrication variations. In this work, we present a carrier-depletion MRM that mitigates these issues by integrating an adiabatic transition with a widened low-parasitic-resistance region. The design extends the low-resistance cross-section over 61% of the cavity circumference, reducing the junction series resistance to 36.9 ohms, which is 35.9% lower than conventional designs. Across eight dies, the resonance wavelength spread is reduced from 3.58 to 1.07 nm. The modulator exhibits a measured 3-dB electro-optic bandwidth exceeding 100 GHz over a bias range of −1 to −5 V. It supports 144 Gbps NRZ and 224 Gbps PAM-4, with the latter demonstrating a bit error rate below the 7% hard-decision forward-error-correction threshold. Moreover, the device can operate in a self-bias mode at 0 V bias, achieving 128 Gbps NRZ with an estimated energy consumption of 0.861 fJ/bit, while depletion-mode operation consumes 8.878 fJ/bit.
- 30 September 2026
- Published by Optica Publishing Group in Photonics Research
- Vol. 14 (10) , 4500
Monolithic silicon photonic multiplexer, transceiver, and processor for 4 × 196 Gbps/ λ multidimensional inter-chip communication
The rapid advancement of artificial intelligence (AI) necessitates significant improvements in optical interconnects that offer higher capacity, lower power consumption, and minimal latency. In this study, we present a multidimensional inter-chip optical interconnect that employs polarization and mode division multiplexing techniques. We achieve per-wavelength data rates of up to 4×126 Gbps with NRZ-OOK and 4×196 Gbps with PAM-4 through the monolithic integration of high-speed GeSi electro-absorption modulators, grating-based (de)multiplexers, and a reconfigurable optical MIMO signal processor. By optically compensating for polarization rotation and mode mixing within a two-mode few-mode fiber connecting two photonic chips, we obtain inter-channel crosstalk levels below −20 dB . The optical MIMO processor exhibits an estimated thermal tuning energy of 0.14 pJ/bit and an estimated optical propagation latency of 0.14 ns. Furthermore, our system demonstrates transparency to modulation formats and symbol rates, effectively supporting both high-speed NRZ-OOK and PAM-4 transmission. These findings provide a promising pathway for the development of high-capacity, energy-efficient, and low-latency optical interconnects, essential for the next generation of AI and data center infrastructures.
- 30 September 2026
- Published by Optica Publishing Group in Photonics Research
- Vol. 14 (10) , 4500
Monolithic silicon photonic multiplexer, transceiver, and processor for 4 × 196 Gbps/ λ multidimensional inter-chip communication
The rapid advancement of artificial intelligence (AI) necessitates significant improvements in optical interconnects that offer higher capacity, lower power consumption, and minimal latency. In this study, we present a multidimensional inter-chip optical interconnect that employs polarization and mode division multiplexing techniques. We achieve per-wavelength data rates of up to 4×126 Gbps with NRZ-OOK and 4×196 Gbps with PAM-4 through the monolithic integration of high-speed GeSi electro-absorption modulators, grating-based (de)multiplexers, and a reconfigurable optical MIMO signal processor. By optically compensating for polarization rotation and mode mixing within a two-mode few-mode fiber connecting two photonic chips, we obtain inter-channel crosstalk levels below −20 dB . The optical MIMO processor exhibits an estimated thermal tuning energy of 0.14 pJ/bit and an estimated optical propagation latency of 0.14 ns. Furthermore, our system demonstrates transparency to modulation formats and symbol rates, effectively supporting both high-speed NRZ-OOK and PAM-4 transmission. These findings provide a promising pathway for the development of high-capacity, energy-efficient, and low-latency optical interconnects, essential for the next generation of AI and data center infrastructures.
- 18 September 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (19) , 5345-5348
High-responsivity germanium photodetector enabled by dual corner reflectors
Germanium waveguide photodetectors are a fundamental building block for high-speed silicon photonic transceivers. While the responsivity is normally proportional to the area of the absorbing germanium layer, too large an active region increases the dark current and sacrifices the bandwidth. It is thus essential to improve the responsivity per unit area to achieve an overall optimal performance. Here, we demonstrate a vertical germanium photodetector with Si/Ge dual corner reflectors that recycle unabsorbed light in both the Si and Ge layers. The active region occupies a compact chip area of only 4.8 μm × 3.7 μm. It achieves a high responsivity of 0.93 A/W at 1550 nm and a low dark current of 1.7 nA at –2 V. This corresponds to a significantly high responsivity per unit area. Owing to the small active region, the measured 3 dB bandwidth reaches 77 GHz, supporting reception of 112 Gbps non-return-to-zero and 224 Gbps 4-level pulse-amplitude signals.
- 18 September 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (19) , 5345-5348
High-responsivity germanium photodetector enabled by dual corner reflectors
Germanium waveguide photodetectors are a fundamental building block for high-speed silicon photonic transceivers. While the responsivity is normally proportional to the area of the absorbing germanium layer, too large an active region increases the dark current and sacrifices the bandwidth. It is thus essential to improve the responsivity per unit area to achieve an overall optimal performance. Here, we demonstrate a vertical germanium photodetector with Si/Ge dual corner reflectors that recycle unabsorbed light in both the Si and Ge layers. The active region occupies a compact chip area of only 4.8 μm × 3.7 μm. It achieves a high responsivity of 0.93 A/W at 1550 nm and a low dark current of 1.7 nA at –2 V. This corresponds to a significantly high responsivity per unit area. Owing to the small active region, the measured 3 dB bandwidth reaches 77 GHz, supporting reception of 112 Gbps non-return-to-zero and 224 Gbps 4-level pulse-amplitude signals.
- 9 September 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (18) , 5121
High-speed and high-efficiency silicon slow-light modulator based on 1D band-edge waveguides
Driven by the increasing demand for high-speed and energy-efficient optical interconnects, silicon modulators have emerged as key building blocks due to their compact footprint and CMOS compatibility. Here, we demonstrate a high-speed and high-efficiency silicon MZM based on one-dimensional super-Gaussian apodized slow-light grating waveguides. By operating in a moderate slow-light regime, the device enhances light–matter interaction while preserving bandwidth, achieving a modulation efficiency of 0.34 V·cm within a 124-μm phase shifter. The apodized grating design suppresses mode mismatch and structural reflections, resulting in a low insertion loss of 2.6 dB. A 3-dB electro-optic bandwidth exceeding 70 GHz is obtained, enabling high-speed data transmission up to 144 Gbaud with NRZ signaling. These results indicate that properly engineered slow-light structures can simultaneously achieve high-efficiency and high-speed performance, providing a promising solution for high-density and energy-efficient silicon photonic interconnects.
- 9 September 2026
- Published by Optica Publishing Group in Optics Letters
- Vol. 51 (18) , 5121
High-speed and high-efficiency silicon slow-light modulator based on 1D band-edge waveguides
Driven by the increasing demand for high-speed and energy-efficient optical interconnects, silicon modulators have emerged as key building blocks due to their compact footprint and CMOS compatibility. Here, we demonstrate a high-speed and high-efficiency silicon MZM based on one-dimensional super-Gaussian apodized slow-light grating waveguides. By operating in a moderate slow-light regime, the device enhances light–matter interaction while preserving bandwidth, achieving a modulation efficiency of 0.34 V·cm within a 124-μm phase shifter. The apodized grating design suppresses mode mismatch and structural reflections, resulting in a low insertion loss of 2.6 dB. A 3-dB electro-optic bandwidth exceeding 70 GHz is obtained, enabling high-speed data transmission up to 144 Gbaud with NRZ signaling. These results indicate that properly engineered slow-light structures can simultaneously achieve high-efficiency and high-speed performance, providing a promising solution for high-density and energy-efficient silicon photonic interconnects.
- 3 September 2026
- Published by Optica Publishing Group in Optica
- Vol. 13 (9) , 1769
Self-configuring universal multichannel and multidimensional integrated photonic processing engine
Arbitrary manipulation of light across multiple physical dimensions is crucial for leveraging its parallelism in fundamental research and advanced applications, including interconnects, computing, imaging, sensing, and quantum networks. However, creating a universal device for the arbitrary parallel manipulation of structured light has been challenging. In this study, we experimentally demonstrate a self-configuring integrated photonic processor designed for arbitrary linear and parallel manipulation of multiple structured light, including cylindrical vector beams and orbital angular momentum beams. Utilizing an optical singular-value decomposition engine, the processor effectively sorts orthogonal input beams and implements precise linear transformations, despite the random nature of the input speckle. Notably, the photonic processor can be self-programmed in situ , enabling versatile functionalities such as beam shaping, optical switching, and reconfigurable optical add-drop multiplexing. Our findings advance the manipulation of multidimensional optical beams through a CMOS-compatible photonic integration approach, paving the way for fully exploiting the parallelism of light in various applications.
- 3 September 2026
- Published by Optica Publishing Group in Optica
- Vol. 13 (9) , 1769
Self-configuring universal multichannel and multidimensional integrated photonic processing engine
Arbitrary manipulation of light across multiple physical dimensions is crucial for leveraging its parallelism in fundamental research and advanced applications, including interconnects, computing, imaging, sensing, and quantum networks. However, creating a universal device for the arbitrary parallel manipulation of structured light has been challenging. In this study, we experimentally demonstrate a self-configuring integrated photonic processor designed for arbitrary linear and parallel manipulation of multiple structured light, including cylindrical vector beams and orbital angular momentum beams. Utilizing an optical singular-value decomposition engine, the processor effectively sorts orthogonal input beams and implements precise linear transformations, despite the random nature of the input speckle. Notably, the photonic processor can be self-programmed in situ , enabling versatile functionalities such as beam shaping, optical switching, and reconfigurable optical add-drop multiplexing. Our findings advance the manipulation of multidimensional optical beams through a CMOS-compatible photonic integration approach, paving the way for fully exploiting the parallelism of light in various applications.
- 1 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (12) , 5116-5122
High-Bandwidth and Efficient Silicon Microring Modulator Enabling Low-Complexity DMT Transmission
High-speed and energy-efficient optical interconnects are critical for next-generation data center and high-performance computing systems. Silicon microring modulators offer an attractive solution owing to their compact footprint and low power consumption; however, simultaneously achieving large electro-optic bandwidth and high modulation efficiency remains challenging, particularly for low-complexity intensity-modulation and direct-detection links. In this work, we experimentally demonstrate a high-performance silicon microring modulator with an electro-optic modulation bandwidth exceeding 110 GHz and modulation efficiency up to 50 pm/V, enabling high-speed transmission with simplified transmitter and digital signal processing. By employing discrete multi-tone modulation with one-tap frequency-domain equalization, direct-drive transmission from an arbitrary waveform generator is realized without RF amplifiers or bias tees at a data rate of 262.5 Gbit/s below the soft-decision forward error correction threshold using only 0.8 Vpp driving voltage. With electrical amplification, a back-to-back data rate of 300 Gbit/s is further achieved. A comprehensive comparison with single-carrier PAM-4 transmission shows that PAM-4 achieves 260 Gbit/s in a back-to-back configuration using linear equalization, but only 200 Gbit/s over 5 km of standard single-mode fiber with nonlinear equalization due to its sensitivity to chromatic dispersion. In contrast, the DMT-based system maintains robust operation over fiber links up to 5 km with low DSP complexity, achieving 225 Gbit/s. These results represent the highest reported data rate for microring modulator based DMT transmission using linear equalization and demonstrate a scalable, low-power solution for short-reach optical interconnects in future high-capacity silicon photonic systems.
- 1 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Lightwave Technology
- Vol. 44 (12) , 5116-5122
High-Bandwidth and Efficient Silicon Microring Modulator Enabling Low-Complexity DMT Transmission
High-speed and energy-efficient optical interconnects are critical for next-generation data center and high-performance computing systems. Silicon microring modulators offer an attractive solution owing to their compact footprint and low power consumption; however, simultaneously achieving large electro-optic bandwidth and high modulation efficiency remains challenging, particularly for low-complexity intensity-modulation and direct-detection links. In this work, we experimentally demonstrate a high-performance silicon microring modulator with an electro-optic modulation bandwidth exceeding 110 GHz and modulation efficiency up to 50 pm/V, enabling high-speed transmission with simplified transmitter and digital signal processing. By employing discrete multi-tone modulation with one-tap frequency-domain equalization, direct-drive transmission from an arbitrary waveform generator is realized without RF amplifiers or bias tees at a data rate of 262.5 Gbit/s below the soft-decision forward error correction threshold using only 0.8 Vpp driving voltage. With electrical amplification, a back-to-back data rate of 300 Gbit/s is further achieved. A comprehensive comparison with single-carrier PAM-4 transmission shows that PAM-4 achieves 260 Gbit/s in a back-to-back configuration using linear equalization, but only 200 Gbit/s over 5 km of standard single-mode fiber with nonlinear equalization due to its sensitivity to chromatic dispersion. In contrast, the DMT-based system maintains robust operation over fiber links up to 5 km with low DSP complexity, achieving 225 Gbit/s. These results represent the highest reported data rate for microring modulator based DMT transmission using linear equalization and demonstrate a scalable, low-power solution for short-reach optical interconnects in future high-capacity silicon photonic systems.
- Open Access
- 20 February 2026
- Published by American Chemical Society (ACS) in ACS Photonics
- Vol. 13 (5) , 1284-1291
200 Gb/s All-Silicon Photodetection Enabled by Response-Engineered Tandem Microrings
- 20 February 2026
- Published by American Chemical Society (ACS) in ACS Photonics
- Vol. 13 (5) , 1284-1291
Open Access200 Gb/s All-Silicon Photodetection Enabled by Response-Engineered Tandem Microrings
- Top Cited PapersOpen Access
- 18 February 2026
- Published by Springer Nature in Nature
- Vol. 651 (8105) , 348-355
Integrated photonics enabling ultra-wideband fibre–wireless communication
Telecommunication systems are evolving towards ultrawide bandwidth and low latency, supporting wired and wireless links and their non-blocking interconnection 1 . However, a long-standing bandwidth mismatch between fibre communication and its wireless counterpart arises from fundamental disparities in signal architectures and hardware constraints 2,3 , which prevent high-speed and compatible transmission across the two domains. This challenge further complicates unified system design and hinders the realization of high-throughput-density, congestion-free fibre–wireless links under wideband-access scenarios 4 . Here we present an ultra-wideband (UWB) integrated photonics scheme that facilitates fibre–wireless communication over a shared-bandwidth infrastructure. Built on electro–optic (EO) and optic–electro (OE) conversions featuring 3-dB operational bandwidths exceeding 250 GHz and cross-architecture adaptability, our system demonstrates unprecedented data transmission capabilities in both wired and wireless links. Using the same set of devices and powered by the proposed complex bidirectional gated recurrent unit (complex-biGRU) algorithm, ultrahigh single-lane data rates of 512 Gbps for short-reach fibre and, for the first time to the authors’ knowledge, 400-Gbps high-speed wireless transmission have been achieved. Furthermore, high-density access is enabled by an all-optically assisted ultra-broadband wireless scheme. Real-time multichannel 8K video transmission is successfully demonstrated across 86 channels, seamlessly using a spectral range from 138 to 223 GHz. These findings in unified telecommunication development show the potential for the development of high-speed, densified and low-latency communication networks.
- 18 February 2026
- Published by Springer Nature in Nature
- Vol. 651 (8105) , 348-355
Top Cited PapersOpen AccessIntegrated photonics enabling ultra-wideband fibre–wireless communication
Telecommunication systems are evolving towards ultrawide bandwidth and low latency, supporting wired and wireless links and their non-blocking interconnection 1 . However, a long-standing bandwidth mismatch between fibre communication and its wireless counterpart arises from fundamental disparities in signal architectures and hardware constraints 2,3 , which prevent high-speed and compatible transmission across the two domains. This challenge further complicates unified system design and hinders the realization of high-throughput-density, congestion-free fibre–wireless links under wideband-access scenarios 4 . Here we present an ultra-wideband (UWB) integrated photonics scheme that facilitates fibre–wireless communication over a shared-bandwidth infrastructure. Built on electro–optic (EO) and optic–electro (OE) conversions featuring 3-dB operational bandwidths exceeding 250 GHz and cross-architecture adaptability, our system demonstrates unprecedented data transmission capabilities in both wired and wireless links. Using the same set of devices and powered by the proposed complex bidirectional gated recurrent unit (complex-biGRU) algorithm, ultrahigh single-lane data rates of 512 Gbps for short-reach fibre and, for the first time to the authors’ knowledge, 400-Gbps high-speed wireless transmission have been achieved. Furthermore, high-density access is enabled by an all-optically assisted ultra-broadband wireless scheme. Real-time multichannel 8K video transmission is successfully demonstrated across 86 channels, seamlessly using a spectral range from 138 to 223 GHz. These findings in unified telecommunication development show the potential for the development of high-speed, densified and low-latency communication networks.