Laser & Optoelectronics Progress, Volume. 62, Issue 17, 1739011(2025)

Research Progress of Photonic Spiking Neural Networks (Invited)

Yahui Zhang1,2, Shuiying Xiang1,2、*, Xingxing Guo1,2, Yanan Han1,2, Changjian Xie1, Tao Wang1, and Yue Hao2
Author Affiliations
  • 1State Key Laboratory of Integrated Service Networks, School of Telecommunications Engineering, Xidian University, Xi'an 710071, Shaanxi , China
  • 2State Key Discipline Laboratory of Wide Bandgap Semiconductor Technology, School of Microelectronics, Xidian University, Xi'an 710071, Shaanxi , China
  • show less
    Figures & Tables(13)
    Circuit diagram of LIF neuron
    Schematic diagrams of encoding method[23]
    Photonic spiking neurons with different material systems and device structures[26-34]
    Simulation results of neuron-like properties of two-section semiconductor lasers. (a) Neuron-like properties of the classical two-section semiconductor laser model[29]; (b) neuron-like properties of the time-dependent traveling wave model[38]; (c) intrinsic plasticity and activation-weight integration properties of the time-dependent traveling wave model[39]
    Simulation results of neuron-like properties of VCSEL-based photonic spiking neuron based on spin inversion model. (a) Spike properties[40]; (b) simulation of tonic and phasic spikes[39]
    Implementation schemes of partial optical spiking neurons based on commercial devices. (a) VCSEL photonic spiking neuron[46]; (b) DFB photonic spiking neuron[41]; (c) FP photonic spiking neuron[54]; (d) dual-IQ modulator photonic spiking neuron[55]
    Microscope images of partial optical spiking neuron chips. (a) MRR[34]; (b) FP-SA[32]; (c) DFB-SA[36]
    Models and architectures of partial photonic spiking neural networks. (a) In-memory computing architecture based on phase-change materials[72]; (b) spiking neural network computing architecture based on all-VCSEL[75]; (c) connection between two-layer FP-SA photonic spiking neurons[79]; (d) hybrid integrated photonic spiking neural network based on 4×4 MZI array and VCSELs-SA[77]
    Integrated chips of partial photonic spiking neural networks. (a) Spiking neural network chip based on PCM[64]; (b) photonic spiking convolutional neural network array chip[37]
    • Table 1. Partial parameters of VCSEL-SA

      View table

      Table 1. Partial parameters of VCSEL-SA

      ParametersValues in gain sectionValues in absorber section
      Cavity volume V2.4×10-18 m32.4×10-18 m3
      Confinement factor Γ0.060.05
      Carrier lifetime τ1 ns100 ps
      Differential gain/loss g2.9×10-12m3s-114.5×10-12m3s-1
      Transparency carrier density n01.1×1024m-30.89×1024m-3
      Lasing wavelength λ850 nm
      Speed of light c3×108ms-1
      Bimolecular recombination Br10×10-16m3s-1
      Spontaneous emission coupling factor β1×10-4
      Output power coupling coefficient ηc0.4
      Photon lifetime τph4.8 ps
      Planck constant h6.63×10-34 Js
    • Table 2. Partial parameters of time-dependent traveling wave model

      View table

      Table 2. Partial parameters of time-dependent traveling wave model

      ParameterValueParameterValue
      Grating coupling coefficient κ1000 m-1Confinement factor Γ1
      Linear recombination coefficient A1×108 s-1Internal loss αs5000 m-1
      Bimolecular recombination coefficient B1×10-16 m3/sLength of the laser cavity L300 μm
      Auger recombination coefficient C3.5×10-41 m3/sLength of gain section LG280 μm
      Effective refractive index neff3.2Length of SA section LSA20 μm
      Linewidth enhancement factor αm1.5×1024Gain suppression coefficientε6×10-23 m3
      spontaneous emission factor nsp5×10-5Transparent carrier density NT3.6
    • Table 3. Dynamical properties of photonic spiking neurons for partial commercial devices

      View table

      Table 3. Dynamical properties of photonic spiking neurons for partial commercial devices

      DeviceThresholdTechnical routeFunction
      VCSEL43-501‒3 mAIntensity perturbation,frequency switching,electronic perturbationSpike coding, edge detection, firing threshold, temporal integration, refractory period
      DFB51-5310‒50 mAIntensity perturbationSpike coding, radio frequency signal processing, firing threshold, temporal integration, refractory period
      Dual-IQ modulation54Intensity, phase-encodingLinear photonic neuron
      FP Laser55~10 mAIntensity perturbationUltra-fast obstacle avoidance
      Quantum dot laser56-58~200 mAIntensity perturbation, electronic perturbationSpike coding, firing threshold
    • Table 4. Integrated chips of photonic spiking neurons

      View table

      Table 4. Integrated chips of photonic spiking neurons

      ChipMaterial/wavelengthSpike rate /GHzEnergy spikeArea
      BPD+MRR59Si/1550 nm8 μm×50 μm, 11.5 μm(diameter)
      MRR+PCM30Si/1550 nm0.0250.7 pJ100 μm2
      MRR34Si/1550 nm0.253.528 pJ~1575 μm2
      Micropillar laser28InP/980 nm0.08250‒700 fJ4 μm(diameter)
      DFB + BPD31InP/1550 nm210.4 pJ~0.055 mm2
      FP-SA32InP/1550 nm3.37.239 fJ300 μm×1500 μm
      DFB-SA36InP/1550 nm519.99 fJ250 μm×300 μm
      Opto-electronic neuron34CMOS/1310 nm11.18 pJ~4 cm2
    Tools

    Get Citation

    Copy Citation Text

    Yahui Zhang, Shuiying Xiang, Xingxing Guo, Yanan Han, Changjian Xie, Tao Wang, Yue Hao. Research Progress of Photonic Spiking Neural Networks (Invited)[J]. Laser & Optoelectronics Progress, 2025, 62(17): 1739011

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: AI for Optics

    Received: Apr. 16, 2025

    Accepted: Jun. 19, 2025

    Published Online: Sep. 12, 2025

    The Author Email: Shuiying Xiang (syxiang@xidian.edu.cn)

    DOI:10.3788/LOP251019

    CSTR:32186.14.LOP251019

    Topics