<p>Neuromorphic photonics promises sub-nanosecond latency, ultrawide bandwidth, and high parallelism, but practical scalability is constrained by fabrication tolerances, spectral alignment, and tuning energy. Here, we present a large-scale, compact, and reconfigurable photonic neuron in which each microring performs modulation and weighting simultaneously. By exploiting both carrier and thermal tuning within a single device, this architecture reduces footprint, relaxes spectral alignment requirements to just two optical components, and yields a steep transfer response that lowers tuning energy. The proposed neuron supports multiple operating configurations, allowing its dynamical behavior to be adapted to different computational tasks. In particular, a short electrical feedback path enables recurrent operation, providing tunable short- and long-term memory for temporal processing. Using a 10-microring resonator array, we demonstrate both spatial and temporal computing, including a 3<InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\times$</EquationSource> <EquationSource Format="MATHML"><math> <mo>×</mo> </math></EquationSource> </InlineEquation>3 convolution for image processing with an error of <5% and high-frequency financial time-series prediction. Each modulation-weighting element occupies 80<InlineEquation ID="IEq2"> <EquationSource Format="TEX">$\times$</EquationSource> <EquationSource Format="MATHML"><math> <mo>×</mo> </math></EquationSource> </InlineEquation>45  <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\upmu$</EquationSource> <EquationSource Format="MATHML"><math> <mi mathvariant="normal">μ</mi> </math></EquationSource> </InlineEquation>m<InlineEquation ID="IEq4"> <EquationSource Format="TEX">$^2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mn>2</mn> </mmultiscripts> </math></EquationSource> </InlineEquation> and consumes an average of 0.186 mW, corresponding to a compute density of 4.67 TOPS/s/mm<InlineEquation ID="IEq5"> <EquationSource Format="TEX">$^2$</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mn>2</mn> </mmultiscripts> </math></EquationSource> </InlineEquation>. Excluding electronic power, the on-chip tuning efficiency reaches approximately 105 TOPs/W, which is comparable to state-of-the-art implementations. These results indicate that modulation-and-weighting microring resonator banks provide a scalable building block for large-scale neuromorphic photonic systems, offering a favorable combination of compact footprint, low power consumption, and functional flexibility.</p>

Compact, reconfigurable, and scalable photonic neurons by modulation-and-weighting microring resonators

期刊论文
出版模式：

开放获取
发表日期：

2026年2月9日

查看全文

Weipeng Zhang,
Yuxin Wang,
Joshua C. Lederman,
Bhavin J. Shastri,
Paul R. Prucnal

摘要

Neuromorphic photonics promises sub-nanosecond latency, ultrawide bandwidth, and high parallelism, but practical scalability is constrained by fabrication tolerances, spectral alignment, and tuning energy. Here, we present a large-scale, compact, and reconfigurable photonic neuron in which each microring performs modulation and weighting simultaneously. By exploiting both carrier and thermal tuning within a single device, this architecture reduces footprint, relaxes spectral alignment requirements to just two optical components, and yields a steep transfer response that lowers tuning energy. The proposed neuron supports multiple operating configurations, allowing its dynamical behavior to be adapted to different computational tasks. In particular, a short electrical feedback path enables recurrent operation, providing tunable short- and long-term memory for temporal processing. Using a 10-microring resonator array, we demonstrate both spatial and temporal computing, including a 3 $\times$ $\times$ 3 convolution for image processing with an error of <5% and high-frequency financial time-series prediction. Each modulation-weighting element occupies 80 $\times$ $\times$ 45 $\upmu$ $μ$ m $^2$ $^{2}$ and consumes an average of 0.186 mW, corresponding to a compute density of 4.67 TOPS/s/mm $^2$ $^{2}$ . Excluding electronic power, the on-chip tuning efficiency reaches approximately 105 TOPs/W, which is comparable to state-of-the-art implementations. These results indicate that modulation-and-weighting microring resonator banks provide a scalable building block for large-scale neuromorphic photonic systems, offering a favorable combination of compact footprint, low power consumption, and functional flexibility.