In this work, we aim to optimize the all-optical phase-controlled phase modulators that incorporate lasers, optical directional couplers, optical amplifiers, and phase shifters. Two types of the systems are studied: linear type and sinusoidal type in which the phase of the output light beam of the systems follows a linear and a sinusoidal dependence on the phase difference between the input light and the control light. Using the inverse design approach based on the 2-layer multilayer perceptron through the pseudoinverse of matrix, the coupling ratio of directional couplers and the phase delay of the phase shifters of the systems can be obtained. The standard errors of phase and power between the output modulated light and the target output light are less than those of our previous reports. The non-linear phase-controlled phase modulator could be applied in optical neural networks as an activation function.
Solar energy harvesting and radiative cooling through the atmospheric transparency window are complementary approaches for sustainable thermal management. However, their conflicting spectral requirements make it difficult for conventional static structures to simultaneously realize efficient cooling and heating. Here, we propose a thermally responsive selective emitter based on a Ge2Sb2Te5 (GST) phase-change metamaterial that enables reversible switching between radiative cooling and solar heating through GST phase modulation. In the crystalline state, the emitter achieves an average emissivity of 92.5% within the atmospheric window (8 – 13 μm), facilitating strong thermally activated radiative heat dissipation that significantly outweighs daytime solar heat gain. In the amorphous state, the atmospheric-window emissivity drops to 9% while solar absorptance reaches 85%, enabling effective photothermal heating. Unlike conventional passive emitters, the proposed device offers thermally reconfigurable spectral characteristics through the non-volatile GST phase transition, making it particularly attractive for elevated-temperature thermal management and thermal-protection applications. These results demonstrate the potential of GST phase-change metamaterials for dynamic spectral engineering and thermally adaptive photonic thermal management, offering a promising strategy for switchable solar heating and radiative heat dissipation in energy-efficient thermal-management systems.
Electroabsorption modulators (EAMs) are key components in contemporary photonics, enabling compact, low-power modulation in optical communications networks. Monolithic integration of EAMs with quantum well (QW) lasers, facilitated by quantum well intermixing (QWI), can potentially deliver high-performance electroabsorption-modulated lasers for deployment in high-speed photonic integrated circuits. Integrating lasers and EAMs in a single epitaxial structure via QWI presents a demanding high-dimensional design problem, requiring simultaneous selection of pre- and post-QWI band gaps, optimization of EAM performance via quantum-confined Stark effect engineering, and consideration of growth constraints. To enable systematic QW-EAM optimization, we establish an inverse design framework that combines predictive QW electroabsorption calculations with genetic algorithm maximization of theoretical EAM performance (quantified by the ratio of the extinction ratio and insertion loss at the target modulation wavelength). The physical model combines multi-band $\bf {k}\cdot \mathbf {p}$ electronic structure and excitonic absorption calculations with a quantitative multi-component diffusion theory of QWI. We undertake inverse design of standalone ( non-intermixed) strained Al$_{1-x-y}$In$_{y}$Ga$_{x}$As/InP QW-EAMs operating at 1550 nm, identify strain-compensated “intra-step” QWs as offering optimal performance, and demonstrate that intra-step QW-EAMs outperform conventional (“square”) and composition-graded (“triangular”) designs. Limiting the modulation electric field to target low-voltage operation yields optimized intra-step QWs exhibiting up to 61% figure-of-merit enhancement vs. square QWs. Extending to intermixed QWs identifies structures following the same intra-step motif that provide up to 12% figure-of-merit enhancement relative to non-intermixed structures, indicating that QWI can enhance high-field EAM performance. This establishes QWI as an EAM design degree of freedom and not solely a band gap tuning mechanism.
Freeform metasurfaces offer a vast design space to achieve performance beyond the conventional fixed-shape metasurfaces, but navigating this space efficiently remains a significant challenge. Reinforcement learning (RL), as one of the modern inverse design methods, holds the potential to handle the immensity of the design space without requiring prior training data. Yet, most demonstrations in freeform metasurface design have been limited to one-dimensional (1D) grating design problems and only a limited number of studies has been applied to two-dimensional (2D) metasurface. The exponential scaling of the design space and the expensive three-dimensional electromagnetic simulations make 2D metasurface inverse design considerably more challenging. To overcome this limitation, we propose Fourier level-set reinforcement learning (FLRL), which integrates the Fourier level-set representation and reinforcement learning into a unified optimization framework. The Fourier level-set representation effectively shrinks the design space by producing smooth, connected structures, thereby eliminating unnecessary designs. By leveraging RL's exploration capabilities within this reduced design space, FLRL successfully optimizes high-performance 2D metasurfaces, exceeding previous records achieved with grid-cell representation and achieving higher efficiency compared to gradient-based method on the same level-set representation. This work demonstrates the potential of level-set representation to enable the adoption of machine learning methods for highly complex 2D metasurface inverse design.
This paper presents frequency-domain-interleaved optical transceivers enabled by ultra-wideband analog integrated circuits (ICs) based on InP heterojunction bipolar transistor (HBT) technology. The rapid growth of artificial intelligence (AI) data centers is driving demand for future Ethernet systems operating beyond 400 Gb/s/lane, where symbol rates exceeding 200 GBaud are required. However, the limited bandwidth of digital-to-analog and analog-to-digital converters (DACs and ADCs) integrated into digital signal processors (DSPs) remains a major bottleneck. To address this issue, we present and validate a frequency-domain-interleaved transceiver architecture that extends the effective signal bandwidth using ultra-wideband up conversion and down-conversion circuits. The architecture relaxes the bandwidth requirements of baseband electrical interfaces by allocating the signal spectrum across multiple frequency bands. For validation, we developed mixer ICs and active combiner ICs with 200-GHz-class bandwidths using 250 nm InP-HBT technology with fT/fmax of 460/480 GHz. Ultra broadband test modules equipped with 0.8-mm coaxial connectors were also developed to support operation up to 150 GHz. Using these ICs and modules, we experimentally demonstrate intensity-modulation and direct-detection (IMDD) optical transmission over 11 km at a net bit rate of 651 Gb/s/lane using 248-GBaud probabilistically shaped (PS) twelve-level pulse-amplitude modulation (PAM-12) signals. These results highlight the effectiveness of frequency-domain interleaving in extending the signal bandwidth of future IMDD optical transceivers operating beyond 400 Gb/s/lane.
