Solid-state optical beam steering devices desire a large field of view (FOV), good beam quality, and reconfigurable beamforming of complex patterns, which are not available in a single system yet. Having not been demonstrated, an active beamformer using an optical phased array (OPA) could potentially fulfill these requirements simultaneously, because it can control both the wavefront and beam pattern. Half-wavelength-pitch OPAs theoretically can achieve the three requirements concurrently, but suffer from crosstalk. Most previous efforts focus on mitigating/avoiding crosstalk. Instead, here we appreciate its existence and propose/demonstrate a programmable architecture to compensate for it. Using a tree of composite variable splitters with a full splitting-ratio range, we achieve arbitrary amplitude/phase modulation to pre-correct scrambled phase/amplitude by crosstalk. With comprehensive stray-light minimization strategies, the sidelobe suppression ratio (SLSR) is significantly improved. Our design achieves a 180∘ FOV, a peak SLSR of 24 dB, and complex-pattern beamforming simultaneously in a half-wavelength-pitch 64-waveguide array. Within the ±60∘ range, a SLSR of >20dB is achieved. Our OPA demonstrates Bayliss difference, pulse-shaped, and asymmetric three-beam patterns with high SLSRs of >20dB, ∼10dB, and >18dB, respectively. These performance metrics are important for various applications in light detection and ranging, imaging, and communication.
Photonic neural networks are emerging as promising computing platforms for artificial intelligence (AI). Particularly, integrated photonic unitary neural networks (IPUNNs) are capable of mitigating gradient vanishing/explosion problems when deeper neural networks are constructed. Furthermore, their optical implementations are also much simpler compared to non-unitary counterparts. Meanwhile, real-valued datasets still dominate AI research and the encoding strategy is critical for IPUNNs' performances. However, there are few studies to compare different encoding strategies of IPUNNs to represent these real-valued datasets and their impacts on IPUNNs' performances. Here, in the scope of encoding strategies for real-valued features, we first compare different schemes, such as phase, amplitude and hybrid encoding using numerical simulations, with benchmarks of decision boundary and image recognition tasks. These encoding strategies of IPUNNs are also compared to non-unitary real-valued neural networks (RVNNs) with trainable biases for the same benchmarks. The results suggest that phase encoding outperforms amplitude and hybrid encoding, and exhibits comparable performances to non-unitary RVNNs. To verify the numerical results, a 10×10 IPUNN chip is designed and fabricated. The phase encoding is chosen to be implemented because of its superior performances in numerical studies. We reconfigure the IPUNN chip to perform decision boundary and image recognition tasks by on-chip in-situ training. The experimental results match the simulations well. Our work provides insights for implementing reconfigurable IPUNNs in AI computing.
An on-chip, high-extinction-ratio transverse electric (TE) pass polarizer utilizing a silicon oxynitride (SiON) slab has been proposed and experimentally verified. The power confinement ratio of the mode field is manipulated by using a SiON slab, where most of the power of the transverse magnetic (TM) mode is transferred to the upper SiON slab and then attenuated through radiation, while the TE mode passes through with relatively low propagation loss. Experimental results show that our proposed device can achieve an extinction ratio that varies from 20.5 to 32.7 dB in the wavelength range of 790 to 870 nm, with an insertion loss of 0.6 to 1.7 dB. Potentially, this design has lower material refractive index contrast, larger minimum etching size, smaller lengths, and less stray light crosstalk, which is beneficial for systems applications such as gyroscopes.
This study introduces a novel distributed temperature sensing (DTS) technique on silicon-on-insulator (SOI) chips by the optical frequency domain reflectometry (OFDR) technology. In contrast to traditional on-chip silicon photonics temperature sensors which rely on transmission spectrum detection, this method is based on Rayleigh backscatter induced by the sidewall roughness of as-fabricated waveguides, eliminating the need for a specially designed structure. On-chip DTS results with a remarkable sensing spatial resolution of 200 μm and a high temperature sensitivity of 88 pm/K are demonstrated within the temperature range from 22.8 ∘C to 200 ∘C. Furthermore, the technology is employed to measure non-uniform temperature distributions along an SOI waveguide generated by integrated heaters. Importantly, this approach offers a straightforward sensing structure, opening new possibilities for investigating temperature profiles and thermal crosstalk across the chip.
Spectral scanning, which utilizes the dispersive effect of light, is a simple and robust method for solid-state beam steering in light detection and ranging (LiDAR) applications. Powered by a tunable laser source, optical frequency-domain reflectometry (OFDR) is a high-precision measurement scheme that is inherently compatible with spectral scanning. Here, we propose a spectral-scanning LiDAR based on OFDR technology and demonstrate that, by connecting the measured spectral reflectivity and group delay of the targets with the dispersion equation, their cloud point data can be obtained. Moreover, compared to the spectral-scanning LiDAR based on the frequency-modulated continuous-wave (FMCW) ranging method, our proposed LiDAR scheme offers a more than tenfold improvement in range resolution with a large number of angular pixels. This enhancement enables high-resolution 3D imaging along both the angular and range axes.