Search

A coherent receiver that can demodulate high-speed in-phase and quadrature signals of light is an essential component for optical communication, interconnects, imaging, and computing. Conventional waveguide-based coherent receivers, however, exhibit large footprints, difficulty in coupling a large number of spatial channels efficiently, and limited operating bandwidth imposed by the waveguide-based optical hybrid. Here, we present a surface-normal coherent receiver with nanometallic-grating-based polarizers integrated directly on top of photodetectors without the need for an optical hybrid circuit. Using a fabricated device with the active section occupying a 70-{\mu}m-square footprint, we demonstrate demodulation of high-speed (up to 64 Gbaud) coherent signals in various formats. Moreover, ultra-broadband operation from 1260 nm to 1630 nm is demonstrated, thanks to the wavelength-insensitive nanometallic polarizers. To our knowledge, this is the first demonstration of a surface-normal homodyne optical receiver, which can easily be scaled to a compact two-dimensional arrayed device to receive highly parallelized coherent signals., Comment: 23 pages, 4 figures (main manuscript) + 4 pages, 2 figures (supporting info)

Stokes-vector receiver with compact metasurface and two-dimensional photodetector array is developed to demonstrate 15-Gbaud 16QAM self-coherent transmission over 25-km single-mode fiber. Surface-normal configuration enables direct coupling to multicore fibers in the future highly parallelized systems.

We demonstrate 108-ch (12-core×9-WDM) frequency comb-based self-homodyne multi-core transmission with a net bit rate of 8.64-Tb/s. This demonstration shows the feasibility of high-capacity short-reach SDM/WDM systems using only a single laser.

We experimentally demonstrate error-tolerant optical unitary processors with multiport directional couplers. Thanks to the inherent redundancy of the multi-plane light conversion scheme, equivalent performance is obtained in the presence of large fabrication errors.

High-speed and robust optical beam-steering device will be the key component for various applications, such as LiDAR (light detection and ranging) and free-space optical communication. Optical phased arrays (OPAs) integrated on semiconductor chips have recently received increasing attention due to the high-speed operation, compactness, and low cost. In particular, indium phosphide (InP)-based OPAs are advantageous for high-output-power applications at 1.55- $\mu \text{m}$ eye-safe wavelength, owing to their capability of monolithically integrating active components, such as high-power lasers and optical amplifiers. In this letter, we design and fabricate, to the best of our knowledge, the largest-scale InP-based OPA and experimentally demonstrate its beam-steering operations. The fabricated OPA consists of 100 waveguides with carrier-injection-based phase shifters, densely integrated on 7 mm $\times5$ mm footprint. A focused beam width of 0.11° is steered across the free spectral range of 8.88°, corresponding to more than 80 resolvable points. This is the largest number of resolvable points obtained by a monolithic InP OPA. The response time of the OPA is confirmed to be less than 16 ns, which is limited by the driver circuit. This work paves the way for realizing compact beam-steering modules for high-speed and high-output-power imaging applications.

Recent studies have shown that the self-coherent systems, including the bi-directional, the Stokes-vector-modulation direct-detection (SVM-DD), and the Kramers-Kronig (KK) schemes, have the potential to reduce the cost of short-reach networks. One of the most attractive features of the self-coherent systems is that they may eliminate the necessity of a narrow-linewidth laser, which is mandatory in the conventional full coherent systems employing high-order quadrature-amplitude-modulation (QAM) formats. On the other hand, it is also recognized that the laser phase noise may have a significant impact on the system performance if there exists a large length mismatch between the signal and the CW tone in the transmission paths. In this paper, we analyze the impact of laser phase noise on the self-coherent systems numerically and experimentally by considering the bi-directional system as an example case. As a result, we show that the performance of the self-coherent system can be described efficiently by using a limited number of normalized parameters. We then reveal the existence of a distinctive threshold on the path length mismatch that influences the bit-error-rate (BER) performance; the penalty increases in a stepwise manner as the length mismatch exceeds the threshold and converges to a constant value. Based on the results, we suggest the feasibility of employing high-order QAM formats to transmit >800 Gb/s signals per polarization using a 10-MHz-linewidth laser. Finally, proof-of-concept experimental results are presented to verify the theoretical and numerical analyses.

The optical-phased array (OPA) has gained special interest in recent years as a high-speed and compact imaging device. While large-scale OPAs have been demonstrated in single-pixel imaging, the complexity of the driver circuit is becoming a crucial problem as the number of phase shifters increases. Here, we investigate the phase shift requirement of OPA for single-pixel imaging and demonstrate, for the first time, that full 2π phase shifts are not mandatory to generate a set of illumination patterns with a sufficient degree of randomness required to reconstruct the image. Using a silicon photonic OPA chip with 128 carrier-depletion-based phase shifters, we experimentally confirm this property by successfully retrieving an image under a maximum phase shift of only ∼1.5π without affecting the quality of the image. Consequently, the input voltage can be reduced significantly. Since the carrier-depletion phase shifters generally require high driving voltages, this finding paves the way to high-speed OPA-based imaging with a minimum device requirement.

Due to the unique nature of offering minimal invasiveness and high spatial resolution simultaneously, multimode fibers (MMFs) are receiving significant attention in bio-imaging applications. While a spatial light modulator is typically used for controlling the wavefront of the light emitted from the MMF, it makes the system slow, bulky, and expensive. To solve this problem, in this work, we demonstrate the use of a chip-scale integrated optical phased array (OPA) for imaging through an MMF. A silicon OPA with 128 independent phase shifters is fabricated and combined with a 3D waveguide interface to generate speckle patterns at the MMF output. Using the generated speckles and the optical power detected by a bucket detector, we experimentally obtain fine 2D images of the target. From the point spread function analysis of the system, the number of resolvable points is derived to be 1007 points, which is much larger than the number of phase shifters.

An optical unitary processor (OUP) is a programmable photonic circuit to achieve arbitrary unitary operation for various applications, including optical communication, deep learning, and quantum computing. Conventionally, OUPs are implemented by cascading 2 × 2 reconfigurable interferometers, but this scheme cannot easily be extended to multiple wavelength and polarization channels due to the strict requirement to employ 50:50 beam splitters. Here, we demonstrate that an OUP using multiport directional couplers (DCs) can realize independent unitary conversion of multiple wavelength and polarization channels simultaneously. This OUP is based on the multi-plane light conversion (MPLC) principle, which does not require a specific transformation at each layer, unlike the conventional scheme. Thanks to this unique robustness of the MPLC method and strong wavelength/polarization dependence of multiport DCs, we numerically show that independent unitary transformations can be applied to up to 16 channels (2 polarization × 4 wavelengths) using a single device.

We demonstrate integrated optical neural network processor with excellent error tolerance using multiport directional couplers. Thanks to the robust multi-plane light-conversion mechanism, high data-classifying accuracy over 95% is confirmed, insensitive to the exact coupling ratio.

Recent progress of developing universal optical unitary processors (OUPs) based on the concept of multi-plane light conversion (MPLC) is reviewed. The inherent redundancy of MPLC provides unique scalability and excellent robustness against fabrication imperfectness, enabling large-scale OUPs integrated on silicon and InP platforms.

We demonstrate normal-incident Stokes vector receiver with superimposed metasurfaces to realize sorting and focusing in three different polarization bases. By using the fabricated metasurface with Si nanopost array, we successfully retrieve arbitrary state of polarization.

Highly efficient and fabrication-error-resistant InP/ InGaAsP polarization rotator based on the half-ridge waveguide is experimentally demonstrated. A thin InAlAs layer is embedded inside the InGaAsP core layer to provide 20-fold etching selectivity with respect to InGaAsP, enabling precise control of the dry-etched core layer thickness. The designed device is fabricated to show a high polarization conversion efficiency of more than 95% with a large etching time tolerance of ±16%. Demonstrated scheme offers a practical path towards realizing compact and low-cost InP-based monolithic dual-polarization transceivers with on-chip polarization (de)multiplexers.

Direct-detection (DD) Stokes vector receiver (SVR) has rapidly gained interest for the high-speed datacenter and short-reach links, owing to its capability of recovering 3D optical signals without using expensive coherent detection. In order to reduce the size and cost of SVR, we have recently proposed and demonstrated a simple and compact device with single-ended photodetectors (S-PDs), which can be monolithically integrated on InP. In this paper, we provide comprehensive analysis on the receiver sensitivity for different SVR configurations. From rigorous theoretical investigations, we derive optimal design of SVR to maximize the sensitivity as well as explicit analytical expression of the bit-error-rate assuming an unamplified thermal-noise-limited case. From the obtained results, we find that the optimized 4-port S-PD-based SVR exhibits a moderate penalty of 2.1 dB with respect to the conventional more complicated SVR configuration based on 3-port balanced PDs. Since the proposed S-PD-based SVR can reduce the number of PDs, and requires neither a polarization-beam splitter nor precisely tuned optical interferometer, it may be an attractive candidate in realizing a low-cost and compact fully integrated DD-based receiver for the future >Tb/s short-reach systems, where the power budget is not the primary concern.

An integrated optical unitary converter (OUC), which can realize arbitrary N × N unitary transformation on chip, is promising for widespread applications in various areas, such as optical communication, quantum information processing, and optical neural networks. Most of the integrated OUCs demonstrated to date comprise cascaded 2 × 2 Mach-Zehnder interferometers (MZIs) based on Reck's scheme or its variation, or cascaded multimode interference (MMI) couplers based on multi-plane light conversion (MPLC). However, these schemes are sensitive to fabrication errors, making it difficult to realize high-performance large-scale OUCs. In this work, we experimentally demonstrate a silicon photonic 4 × 4 OUC that is inherently robust to fabrication errors, based on a novel structure. The proposed structure consists of cascaded multiport directional couplers (DCs) and phase shifter arrays. In this scheme, the multiport DC assures unitary transformation with inherent robustness against fabrication errors. By optimizing all the phase shifters using a simulated annealing algorithm, we experimentally demonstrate reconfigurable 4-mode demultiplexing and error-free demultiplexing of 40-Gb/s non-return-to-zero (NRZ) signals.

Gain-coupled 4 × 25 Gb/s electroabsorption modulated laser (EML) array is implemented by an identical epitaxial layer integration scheme. Gain-coupling is incorporated via absorptive gratings to improve the single mode yield of the device. The EML array exhibits excellent single-mode performance, high extinction ratio, and a modulation bandwidth over 28 GHz. Clear eye opening under 28-Gb/s nonreturn-to-zero modulation has been demonstrated for all four channels. Transmission performance of the EML array over 20- and 40-km single-mode fiber is tested, confirming its potential for 100 GbE applications.

We fabricate and demonstrate a monolithic InP-based optical unitary converter (OUC) with nonuniform multimode interference (MMI) couplers. Reconfigurable 4-mode sorting is experimentally demonstrated with a significantly reduced footprint compared with the conventional OUCs using uniform MMI couplers.

An optical phased array (OPA) is a compact high-speed wavefront modulation device that is promising for next-generation optical sensing systems. In particular, speckle-based single-pixel imaging (SSPI) using OPA is an attractive scheme since precise tuning of optical phases is unnecessary. In this work, we present a comprehensive analysis of SSPI using an OPA withNphase shifters by comparing two classes of OPAs: uniformly spaced OPA (UOPA) and non-redundant OPA (NROPA). Through singular value decomposition analysis of the illumination patterns generated from the OPA, we clarify the theoretical limit of the imaging resolution for each case. As a result, the number of resolvable points can be as large asN2−N+1for the case of NROPA. This is in clear contrast to the case of UOPA, where the number of resolvable points can only be as large as2N−1. Finally, imaging results of a test target are compared to study the impact of the array layout in OPA-based SSPI. Our work provides theoretical understanding of OPA-based SSPI and reveals the effectiveness of SSPI using NROPA.

Optical unitary converters (OUC) that enable reconfigurable unitary transformation between the input and output optical bases has gained increased interest for various applications. We compare different architectures of integrated OUC in terms of fabrication tolerance and demonstrate that all-to-all-coupled interferometer significantly enhances the robustness.

We review our recent activities on developing surface-normal high-speed modulators with electro-optical polymer embedded inside plasmonic and silicon metasurface structures. By judiciously designing the grating structures, we obtain sharp optical resonances, which can be used to achieve efficient modulation of the reflected and transmitted light.

Optical phased arrays (OPAs) are promising beam-steering devices for various applications such as light detection and ranging, optical projection, free-space optical communication, and optical switching. However, previously reported OPAs suffer from either an insufficient number of resolvable points, or complicated control requirements due to an extremely large number of phase shifters. To solve this issue, we introduce the non-redundant array (NRA) concept to the OPA devices. Based on this design, we can realize high-resolution OPAs whose number of resolvable points scales quadratically with the number of antennas N . In contrast, that of traditional OPAs scales only linearly with N . Thus, a significant reduction in the number of required phase shifters can be attained without sacrificing the number of resolvable points. We first investigate the impact of employing the NRA theoretically by considering the autocorrelation function of the array layout. We then develop a Costas-array-based silicon OPA and experimentally demonstrate 2D beam steering with ∼ 19 , 000 resolvable points using only 127 phase shifters. To the best of our knowledge, this corresponds to the largest number of resolvable points achieved by an OPA without sweeping the wavelength.

We report a 10-port optical unitary converter integrated on a silicon photonic chip. Reconfigurable unitary operations are demonstrated for both the transverse-electric (TE) and the transverse-magnetic (TM) polarization modes.

We design and fabricate the largest-scale InP-based optical phased array (OPA) with 100 waveguides. Beam steering with more than 80 resolvable points and the response time of less than 16 ns are demonstrated experimentally.

This erratum corrects the error in Fig. 5 of our earlier paper [Optica 8, 1350 (2021)OPTIC82334-253610.1364/OPTICA.437453].

Symmetric polarization rotator-splitter (PRS) is proposed and experimentally demonstrated on InP for the first time. Instead of integrating a mode-selective splitter, we employ a symmetric multimode-interference (MMI) splitter at the output of an adiabatic taper section to extract the linear superpositions of the transverse-electric (TE) and the transverse-magnetic (TM) components of the input signal. As a result, the entire device functions as a PRS with its basis on the S2-S3 plane of the Poincaré sphere, whereas we can fully eliminate complicated asymmetric structures that are challenging to fabricate on InP. Moreover, the adiabatic taper, which operates as a mode-evolution-based polarization converter, is designed judiciously to minimize the overall length. The designed InP PRS with a total length of 750 µm is fabricated by a simple single-etching process. A polarization extinction ratio of more than 16.3 dB and a polarization-dependent loss of 0.67 dB are demonstrated experimentally at a 1550-nm wavelength.

Integrated reconfigurable optical unitary converters (OUCs) are crucial in realizing all-optical spatial mode demultiplexing for mode-division-multiplexed transmission systems and programmable photonic processing for optical neural networks. In this work, we present the first experimental demonstration of 4×4 OUC monolithically integrated on InP. To avoid the difficulty of integrating a large number of Mach-Zehnder interferometer couplers on the InP platform, we apply the concept of multi-plane light conversion and use cascaded stages of 4-port multimode interference couplers, which are more scalable and easier to fabricate on InP. By optimizing the phase shifters, we demonstrate reconfigurable 4-mode sorting as well as error-free switching of 40-Gbit/s signal.

Stokes-vector modulation (SVM) formats have recently drawn attention for realizing low-cost and high-capacity optical transceivers. However, SVM transmitters reported so far are still complicated because they are based on optical components employed mainly in coherent systems. In order to reduce the device size, cost and power consumption, a more compact transmitter configuration is desired. In this invited paper, we first experimentally demonstrate transmission of 30-Gb/s 8-ary SVM signal over 50 km using the proposed configuration. Then we numerically investigate system performance of our recently-proposed compact transmitter for the 8-ary SVM format.

We demonstrate that single-pixel imaging resolution of optical phased array can be enhanced by transmitting through a multimode fiber. Using only 128 phase shifters, >1000 points are resolved, determined by the number of fiber modes.

Carrier-injection-based polarization controller with strained MQW layer is demonstrated. Based on novel design concept, both polarization-rotating and phase-shifting sections are integrated monolithically on regrowth-free InGaAsP/InP platform to achieve efficient conversion over the entire Poincaré sphere.

An integrated Stokes vector receiver (SVR) that can retrieve state of polarization of light in the three-dimensional (3D) Stokes space has widespread applications, such as short-reach communication links, polarization-sensitive imaging, and sensing. While various approaches have been demonstrated to date, monolithic integration of polarization components on InP has been a challenging issue. In this paper, we develop a novel 4-port SVR circuit integrated on a compact InP chip to retrieve complete Stokes parameters of incoming light with various intensity and degree-of-polarization. By judiciously designing the lengths and positions of asymmetric waveguide sections, we demonstrate that the SV of signal can be projected onto four vertices of a regular tetrahedron inscribed in the Poincare sphere. Additionally, we employ this device in decoding 10-Gbaud 4-ary and 8-ary Stokes-vector-modulated signals in the 3D Stokes space.

Stokes vector modulation direct-detection (SVM-DD) formats have attained special attention in the short-reach communication links to transmit multilevel signals without using costly coherent receivers. In this paper, we fabricate and experimentally demonstrate a potentially low-cost DD-based SV receiver with on-chip photodetectors (PD) on a compact InP chip. Compressively strained multiple-quantum-well PD array is monolithically integrated with the SV projection circuit, which consists of multimode interference splitter and SV rotators based on self-aligned half-ridge structure. From the photocurrent signals at three on-chip PDs, the SV of input light is retrieved successfully over the entire Poincare sphere. With potentially broad operating bandwidth of InP-based PDs, the demonstrated device may be an attractive candidate to realize low-cost and high-speed receivers for the future short-reach SVM-DD systems.