Heterogeneous silicon photonics is uniquely positioned to address the photonic sensing needs of upcoming autonomous cars and provide the necessary cost reduction for widespread deployment. This is because it allows for wafer-scale active/passive integration, including optical sources. We present our recent research and the development of interferometric optical gyroscopes and LiDAR sensors. More specifically, we show a fully integrated gyroscope front-end occupying an area of only 4.5 mm2. We also show the first dense pitch optical phased array using heterogeneous phase shifters. The 4 µm pitch heterogeneous phase shifters provide very low V2π of only 0.35-1.4 V across 200 nm, low residual amplitude modulation of only 0.1-0.15 dB for 2π phase shift, extremely low static power consumption (<3 nW), and high speed (> 1 GHz). All of these factors make them ideal for next-generation LiDAR systems that employ optical phased arrays.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
19 March 2019: A typographical correction was made to the title.
There has been an explosive push for autonomous driving as it has the potential to disrupt our lives by increasing safety and reducing the cost of ownership by sharing vehicles as a fleet, resulting in maximized utilization. It is generally believed that robust autonomous driving requires state-of-the-art neural networks processing input from a variety of complementary sensors to provide robustness and fail-safe operation in the complex environments where cars are operated.
Heterogeneous silicon photonics is uniquely positioned to address the photonic sensing needs of upcoming autonomous cars by providing a number of key benefits . The most important are the integration of optical sources and superior photonic devices based on III-V materials with silicon photonics combined with CMOS economies of scale and wafer scale testing capabilities. By utilizing state-of-the-art 300 mm CMOS facilities, there is a path for improved yield, uniformity and to providing a way to substantially reduce per unit costs to levels that will allow widespread deployment. The cost has to be decreased by hopefully more than an order of magnitude compared to current systems, and we believe that photonic integration is the most likely route to achieving that. Integration can also significantly reduce size and weight, allowing such sensors to be unobtrusively integrated in bumpers or around the car, similarly like today’s parking sensors.
Heterogeneous silicon photonics addresses a key limitation of silicon photonics by bringing electrically pumped lasers to the platform. The approach, pioneered by University of California, Santa Barbara and Intel in 2006 , has been adopted by many and is currently commercialized in Datacom applications by Intel and Juniper Networks and are being produced on 300 mm diameter wafers at the level of millions of devices per annum . The approach utilizes molecular bonding between III-V materials and silicon and is CMOS scalable. Initially the solution addressed light generation, but it was also shown to allow improved performance of other optical components, such as detectors and phase shifters/modulators, the latter being extremely important for sensing applications .
In this overview paper, we first briefly introduce the heterogeneous silicon photonics platform (Section 2) and then show our recent results in developing photonic sensors based on said platform, including a fully integrated fiber-optic gyroscope driver chip (Section 3) and an optical phased-array for chip-scale light detection and ranging (LiDAR) system (Section 4). Finally, we give conclusions and visions for the future in Section 5.
2. Heterogeneous III-V/Si integration
The heterogeneous III-V/Si integration [4–6] has widely been used for the development of active photonic integrated circuits (PICs) for silicon photonics during the past decade, since the first laser demonstration using this heterogeneous platform . A variety of heterogeneous III-V/Si building blocks has been realized, including mode locked lasers [7,8], tunable narrow-linewidth lasers [9,10], amplifiers [11,12], high speed modulators [13–15], and photodetectors [16,17]. Moreover, high-level integrated heterogeneous PIC systems on silicon photonics have also been demonstrated such as fully integrated silicon photonic network on chip , fiber-optic gyroscope driver chip , and free-space beam scanner . The above mentioned heterogeneous III-V/Si devices used relatively low integration density (with individual devices spaced at ~100 µm or more) and all details of fabrication and process can be found in those works. In this work, we extend the heterogeneous integration technology to even high-density integrated III-V/Si PICs especially for our optical phase arrays (OPAs) as discussed in Section 4.
In general, on-chip OPAs require micron pitch dimension (ideally sub-wavelength scale) between elements and the resulting high aspect ratio and large surface topology (>2 µm) in heterogeneous process call for significant process development, in particular, a novel metal lift-off fabrication process allowing for patterning dense metal lines. Figure 1 schematically illustrates the fabrication flow for our III-V/Si dense phase shifter array structure presented in more detail in Section 4. Similarly, to previous demonstrations, the overall process flow consists of passive Si waveguide/grating patterning on SOI wafer, III-V die bonding, post processing of III-V structures, and final metallization for contacts. For processing the silicon, we pattern the Si ridge waveguide layer by dry etching 231 nm on a 500 nm silicon-on-insulator (SOI) substrate with a 1.0 μm buried oxide and the lithography is carried on a 100 mm wafer using a DUV stepper. The ridge waveguide geometry is chosen for the benefits of low passive waveguide loss and high coupling efficiency between Si and III-V/Si regions. During said SOI processing all passive elements, including waveguides, splitters, and grating antennas, are defined; and other materials such as silicon nitride (SiN) can also be incorporated for instance to realize weak grating emitters as discussed in Section 4. After the passive patterning, the III-V multiple quantum-well (MQW) dies are selectively bonded onto Si waveguide using a direct bonding technique . After bonding, III-V substrate is removed and mesa and MQW are defined and etched by using reactive-ion etching (RIE) (CH4/H2/Ar) and selective wet etching (H3PO4/H2O2/H2O), respectively. The III-V is then passivated by a SiO2 deposition and n/p contact metallization follows. Due to the small feature size and large aspect ratio of the dense structure, we have developed a triple-layer process for metal lift-off in which a photoresist first defines the pattern on a polymer/SiO2 bi-layer and an undercut profile is then created by dry etching (O2) the polymer with SiO2 as a mask. The developed process allows us to define dense metal lines with a pitch down to 2 µm or even less and a thickness of up to 1 µm. We use Pd/Ge/Pd/Au and Pd/Ti/Pd/Au metal stacks as n- and p- type contacts, respectively. After n/p metallization, a second SiO2 passivation and DVS-BCB polymer planarization are performed. Finally, the vias for n/p contact are opened and probe metal is deposited. In Fig. 2, we show the scanning electron microscope (SEM) images of the cross sections of the fabricated III-V/Si structure, demonstrating high quality of the process.
3. Integrated fiber-optic gyroscope
The interferometric fiber-optic gyroscope (IFOG) is a fascinating sensing apparatus, capable of measuring miniscule phase shifts on the order of 10−15; which is made possible by careful optimization of individual components and utilizing reciprocity . The performance of such units is state-of-the-art, but modern IFOGs still suffer from size, weight, power and cost (SWaP-C) limitations due to being based on discrete optical components. Each discrete component is normally packaged with a fiber pigtail to form the Sagnac reciprocal interferometer optical circuit leading to larger size and increased cost. The SWaP-C limitation is the reason why many low-cost applications utilize micro-electro-mechanical systems (MEMS) based gyroscopes, despite inferior performance and limitations. One of key limitations of MEMS based gyroscopes is the increased sensitivity to vibrations and shock. Optical gyroscopes, on the other hand, are almost completely insensitive to vibration and shock, and also to electromagnetic interference, which is very important in today’s complex environments packed with electronic devices. The robustness would only increase with further integration. For that motivation, miniaturization of optical gyroscopes has been an active research topic for many years on various waveguide platforms [19,22–26].
An IFOG consists of two main parts: front-end and sensing coil. The front-end commonly consists of an optical source, one or more detectors, two splitters and two phase shifters. Currently only the last segment comprising of one splitter and two phase shifters is integrated in a LiNbO3 platform, while source and detectors are separate components. The sensing coil is usually optimized for gyroscope performance (small size, high reciprocity) by optimizing the cladding thickness and advanced fiber winding techniques to cancel out non-reciprocal noise and drift . Depending on the specific implementation, the coil can be single-mode (SM) fiber or polarization-maintaining (PM) single-mode fiber. PM fiber generally has better performance, but at an increased cost.
Integration of an IFOG would bring a huge SWaP-C improvement and would allow deployment of superior rotation sensing instrumentations in many areas where it was currently unfeasible [Fig. 3]. Autonomous driving is one such area. For this, and similar applications, we have developed a first prototype of a fully integrated interferometric optical driver chip (IOD) that comprises a light source, three photodiodes, two phase modulators and two 3-dB couplers within an area of 4.5 mm2 providing a path for significant reduction in SWaP-C of optical gyroscopes . We have also demonstrated chips with the IOD and sensing coil integrated on silicon.
The chip utilizes multiple bonding technology inherent to heterogeneous silicon photonics, where different III-V epitaxial structures are bonded to different regions of the integrated photonic chip to provide optimized performance . In the above demonstration, we have utilized one III-V structure optimized for providing optical gain for lasers, and a second III-V structure optimized for phase shifter operation. Such functionality would require re-growths in native substrate technology, while suffering from higher propagation losses typically encountered in III-V waveguides. Compared to a pure silicon photonics approach, our demonstration includes fully-integrated optical sources, but also superior performing phase shifters. Detailed characterization of heterogeneous III-V/Si based phase shifters is given in Section 3.3. Photodetectors use the same material as lasers, but are reverse biased. This is possible as gyroscope operation does not require very high-bandwidth photodetectors.
The IOD is driven by an optimized Fabry-Perot laser with broad spectra, which is beneficial for gyroscope operation to reduce the coherence length and remove the effects of unwanted reflections . Photodetectors have low-leakage current (<5 nA @ −3 V), responsivity of around 0.7 A/W and bandwidth in the order of 6 GHz. Phase shifters have low insertion loss (< 1 dB over whole C-band) and have ~2 GHz electrical bandwidth that is more than sufficient for gyroscope operations. For splitters, we utilized optimized adiabatic 3dB couplers that provide very broadband operation. The initial characterization of the IOD demonstrated rotation measurement, while more in-depth characterization requires permanent packaging with a sensing coil.
Initial characterization was made by combining the IOD with a fiber based coil for sensing, but further reduction in size and increase in robustness can be made by using an integrated waveguide coil as one demonstrated in  where 3 m large-area coil made in SiN was used to measure rotation. By combining both the IOD and SiN coil, a chip-scale optical gyroscope can be made, paving the way for mass deployment in future autonomous cars and other applications where low SWaP-C is crucial. Such processes are still in development and ideally they would be done in heterogeneous way so the fabrication process is wafers scale for very high volume fabrication, but hybrid approaches can also be envisioned, at least in nearer time frame.
4. Integrated LiDAR
LiDAR is one of key sensors for autonomous driving, along with radars, cameras and ultrasound sensors, but at the same time it is also the most expensive one . For the implementation of LiDAR in commercial vehicles, this will require at least an order of magnitude cost reduction, which we believe is possible only by photonic integration and volume manufacturing at 300 mm semiconductor foundries with wafer-scale testing capabilities. A LiDAR contains two key building blocks: the transceiver and the beam steerer. There are many ways to realize both, and arguably the transceiver is the easier part as it can leverage many of the techniques developed for coherent telecom transceivers. Beam steerers have historically been mechanical or MEMS based, but they generally are slow and need precise assembly with the transceiver. Recently there has been intense interest in development of monolithic beam steerers realized as OPAs [20,29–41]. By leveraging existing PIC platform on silicon photonics and the mature CMOS fabrication technique, it becomes feasible to realize compact and scalable OPAs in massive production. Thus, integrated OPAs provide a great opportunity for building up low-cost and energy-efficient practical systems on a chip, compared with traditional beam steering architectures mostly using complex mechanical apparatus that are usually expensive and bulky. Moreover, the advantage of heterogeneous III-V/Si platform allows for integrating high-performance active components such as laser sources, amplifiers, and photodetectors in a single chip. Eventually, a fully-integrated OPA-based system like a LiDAR becomes commercially viable for a range of applications.
Exact requirements on OPAs depend on application which in turn defines range, resolution, field-of-view, but it is generally agreed that >1 mm scale apertures are needed for distances of interest for autonomous cars. A key issue in all OPAs, regardless of their size, is to integrate high-performance phase shifters with low drive voltage and low power consumption, low residual amplitude modulation (RAM), and high bandwidth, in order to achieve large-scale and energy-efficient beam steerers for practical applications. Various OPAs on SOI platform have been demonstrated by using Si-based thermo-optic (TO) [31,32,34–37,39] or electro-optic (EO) phase shifters [20,33,40,41]. Ideally TO phase shifters can have negligible optical loss but a low bandwidth of a few KHz which limits beam scanning speed and final system performance. In addition, they suffer from high power consumption commonly on the order of tens to hundreds of mW for 2π phase shift, which is problematic in a practical OPA easily consisting of hundreds to thousands of elements. The thermal crosstalk is also a drawback limiting the integration density of the TO phase shifters. On the other hand, EO-type Si phase shifters are capable of high speed operation (tens of GHz) and low power consumption [40,41]. Furthermore, using state-of-the-art CMOS technology, they allow for dense integration with an even wavelength-scale pitch for phase shifter arrays . However, a remaining limitation in Si EO phase shifters is their high RAM that accompanies the phase shift and negatively impacts OPA performance. This limitation stems from the plasma dispersion effect in Si where considerable carrier absorption is unavoidable, due to the coupled real and imaginary parts of the refractive index via Kramers–Kronig relations . Moreover, the modulation efficiency in Si EO modulators is generally low (i.e. a high voltage–length product VπL, typically 1-3 V⋅cm), which results in high operation voltage or excessive lengths , consequently imposing a challenge on driving electronics or sacrificing chip footprint. In order to overcome the trade-offs among those performances in a phase shifter, an alternative choice is to implement a heterogeneous III-V/Si platform for OPA phase shifters to improve device performance comprehensively, since the III-V can leverage additional modulation effects such as Franz-Keldysh or quantum-confined Stark effect, band-filling, band shrinkage and Pockels effects among others to increase total EO effect [13–15,44–46]. In addition, compared with Si the higher electron mobility in III-V  contributes to a reduced carrier absorption loss and lower resistance–capacitance time constant (i.e., a higher operation speed). The demonstrated III/Si heterogeneous modulators have revealed a high modulation efficiency (VπL ~0.05-0.2 V⋅cm) and a low optical loss (<1 dB for a 2π phase shift) in a single device.
In the scope of this work, we focus on the OPA as we see it as the biggest challenge, mostly due to dense level of integration. The density requirements on OPAs are at least an order of magnitude higher than in typical telecom/datacom products. For that reason, in this section we address two key challenges: (a) the efficiency of the grating emitters and (b) the efficiency of phase shifters while minimizing insertion loss and RAM. Finally, we demonstrate record density heterogeneous silicon photonics OPA.
4.1 Grating emitters
Surface normal emission is generally preferred for OPAs as it allows utilizing theoretically the whole chip area (reducing the beam width) and allows steering in both dimensions, either by two-dimensional (2D) phased arrays or by steering with phase in one direction and utilizing wavelength to steer in the other direction in one-dimensional (1D) OPAs. We utilize the latter approach. As shown in Fig. 4, the emission angle θ in waveguide axis (longitudinal direction) is controlled by optical source wavelength and is determined by , where neff is the effective index of the waveguide, Λ is the grating period, and λ is the free-space wavelength of light. Steering in the other direction (ψ) is realized by controlling the phases in individual waveguide gratings. For an OPA, there are three critical performance factors for emitters − aperture size, directivity, and radiation efficiency. A centimeter-scale aperture is desired in a practical OPA as it directly reduces the beam-width of the main lobe. Meanwhile, it’s necessary to constrain the power in the main beam as much as possible. This is done by reducing the pitch between individual emitters and suppressing the grating lobes. Finally, the emission strength should be finely controlled to form Gaussian-like beams in the far field. For centimeter long emitters, to satisfy these requirements, an essential issue is to obtain weak and controllable emission rate. Due to the high refractive index contrast in typical silicon waveguide formed by Si (n = 3.48) an SiO2 (n = 1.44), very shallow etched Si surfaces (e.g., 5-15 nm) are needed. A CMOS compatible alternative is to introduce lower-index dielectric materials like SiN (n ≈2.0) to form the gratings. We investigate both approaches.
We first study the SiN/Si waveguide grating, and focus on two types of surface gratings: strip-line grating and fishbone grating; as schematically shown in Figs. 5(a) and 5(b) respectively. In order to control the emission strength and control the beam quality, we can chirp the duty cycle of strip-line grating or the width of fishbone grating [47–49]. With extensive simulations, we narrow down the design to have SiN thickness of 60 nm deposited above the 800 nm wide Si waveguide. Said SiN is patterned with a period of 560 nm and has the same width as the waveguide. The waveguide is formed by etching 231 nm on a 500 nm SOI substrate with a 1.0 μm buried oxide. Note that the same SOI is also used throughout the paper hereafter and all passive components are designed for fundamental transverse-electric (TE) mode. We first calculated the emission strength of the two gratings as a function of duty cycle or fishbone width and plot the results in Figs. 6(a) and 6(b). The wider emission strength benefits both effective aperture size and radiation efficiency, but it requires smaller feature size and can become limited by used fabrication technology. By optimization algorithm, we synthesized the optimal emission strength profile for both types of grating with a length of 10 mm as shown in Fig. 6(c), based on the constraint that our fabrication doesn’t allow features smaller than 120 nm, and have some non-ideal distortion when features pattern is smaller than 200 nm. Since both approaches to modulate emission rate introduce perturbation of neff, the period should be slightly chirped simultaneously to keep the emission direction unchanged. In Fig. 6(d) we show the far field beam profiles for the two gratings with optimized emission strength and the simulated full width at half maximum (FWHM) beam width is ~0.0065° at 1550 nm for both gratings corresponding to an equivalent Gaussian beam divergence of 7.5 mm. Note that for fishbone grating the element factor profile is slightly changed during the chirping of fishbone width, which could be ameliorated in design, for example, using a narrower waveguide as well as fishbone width. Such analysis will be in our separated work in future.
We studied both types of SiN/Si gratings. The fabrication started with the deposition of 60 nm SiN on a SOI wafer using plasma-enhanced chemical vapor deposition (PECVD). The SiN grating was then patterned and dry etched using inductively coupled plasma (ICP) RIE. Next, the Si waveguide was defined aligning to the grating and formed by a dry etching. The whole process was done on a 100 mm SOI wafer using 248 nm DUV lithography with an overlay accuracy of better than 30nm. Figures 7(a) and 7(b) show the SEM images of the fabricated strip-line and fishbone waveguide gratings. It can also be seen that the alignment between grating and waveguide is almost perfect while the corners of the grating are rounded due to image distortion in lithography. Note that the fabricated grating width is smaller than the waveguide width. This is because in the design we intentionally reduce grating width to compensate any possible alignment error and guarantee the grating defined within the waveguide window thus to avoid any masking effects by grating patterns during waveguide etching. In fact, such smaller shrinkage of the grating width has negligible effect on the radiation properties since the mode perturbation by grating predominantly occurs at the center of waveguide. Nevertheless, grating antennas perfectly aligned to waveguides can also be achieved in future by simply using advanced lithography tools or optimizing fabrication procedure (e.g., first patterning waveguide and then defining grating to eliminate said masking effects).
The fabricated grating was characterized with a Fourier imaging setup, as shown in Fig. 8. The setup is equipped with two different Fourier optical systems that we refer to as Optics-1 and Optics-2. Optics-1 has 25 mm tube-lens optics with a fixed optical axis normal to the chip surface and a high numerical aperture (NA) of NA = 0.83 with a far field resolution of 0.3°. Optics-2 is rotatable with respect to the chip surface and uses 50 mm optics with a NA = 0.23 and a theoretical resolution of 0.0043°. Through rotating the optical axis, the rotational optics is theoretically capable of imaging the beam of large aperture (~50 mm) over the entire upper 180° range in longitudinal direction. Optics-2 is also used to calibrate the emission angle. An infrared (IR) camera with 320 × 256 pixels and 25 μm pitch size is used for capturing images. Thus a maximum measurable field of view in longitudinal and lateral directions is ~96° × 76° and ~1.4° × 1.2° for Optics-1 and Optics-2, respectively. The setup is also used to characterize our OPAs later.
The measured far-field beams of the grating for the wavelength of 1520 nm to 1580 nm are shown in Fig. 9(a). In lateral direction, the beam is broad for both gratings due to narrow aperture of the grating in this direction (comparable to the waveguide width). In contrast, the beam is very narrow in waveguide direction, indicating a large effective aperture. The relative emission angle Δθ is almost linearly changed by tuning the wavelength as shown in Fig. 9(b), with a tuning efficiency of 0.133°/nm for both strip-line and fishbone gratings as expected. In order to measure the accurate beam width, we used the high-resolution optics (Optics-2) and plot the longitudinal beam profiles in Fig. 9(c). The FWHM beam width is measured to be 0.008° at 1550 nm by fitting a Gaussian function for the strip-line grating, corresponding to an effective grating aperture size of ~6.1 mm which is very close to the designed value of 7.5 mm. The discrepancy between design and measurement could be attributed to lithography limitations and distortions. For the fishbone grating, we obtained a bit larger FWHM beam width of 0.01°. The performance of both gratings can be further improved by employing optimizations to account for lithography related distortions. We also measured the transmission of the gratings as shown in Fig. 9(d); and by comparing with a straight reference waveguide a ~10 dB loss mainly due to grating emission is obtained over the measured wavelength range for both type gratings.
Apart from SiN/Si gratings, we also investigated the shallow etched Si strip-line grating. In our demonstration, the grating, aligned on the top of 800 nm wide Si waveguide, has a period of 560 nm with a uniform duty cycle of 0.5 and an etching depth of ~14 nm corresponding to ~2.5 mm effective aperture size in simulation. Compared to SiN/Si hybrid gratings, the Si gratings can simplify fabrication process by removing one deposition step, while making the etch depth a critical step of fabrication. We developed a dry etching process with a very low etch rate of ~12nm/min for Si and were able to control the etched depth within an accuracy of ± 1.0 nm. Figure 10 presents SEM and atomic force microscope (AFM) images of the fabricated shallow-etched Si grating, showing a very good control for the etching depth of the grating. Note that we used the same SiN grating mask for this investigation and thus the grating shows a non-ideal etched strip with a reduced width. A full strip grating can be achieved by simply using a wider grating design. The characterized results for far-field beams are presented in Fig. 11. Note that the emission angle was calibrated with our rotational optics. The wavelength tuning efficiency in longitudinal emission angle is 0.138 ± 0.002°/nm, which is in excellent agreement with a theoretical value of 0.139°/nm; and a total steering angle in θ direction is ~28° over a 200 nm wavelength tuning range. The beam width was measured to be 0.02° at 1550 nm, corresponding to a ~2.5 mm effective aperture size. Compared with the results of SiN grating, the etched Si grating has higher emission strength (~28 dB/cm for 14 nm etching), but the aperture can be increased by optimizing the emission strength as well as its profile along the length of the waveguide. Nonetheless, for the sake of simplicity, we utilized Si grating in the heterogeneous OPA demonstration described below.
4.2 Star couplers
Many-channel OPAs are required for small pitch, large area designs. Depending on the dimensions of the array, the input beam has to be split to multiple hundreds or even thousands of channels. Such splitting can commonly be done via multimode interferometer (MMI) couplers or via a star coupler. MMI coupler tree structures are a robust way to achieve splitting and also are capable of achieving wavelength-scale emitter pitch . The power distribution of the MMI tree is uniform, which can be a benefit or a disadvantage – depending on complete design. In the case of a single OPA, the uniform power distribution results with high side-lobes as the response of rectangular aperture is a sinc function. A potential benefit of MMI tree would be the uniform phase delay (equal path length delay), but fabrication tolerance, performance variation across phase shifters and thermal gradients might still require look-up tables for beam-steering.
We opted for star couplers  as they provide the ability for high fan-out count at small pitch and the potential for sub-wavelength pitch OPAs . Moreover, the star couplers are broadband, have low loss and provide power profile shaping for reduced sidelobes in the phase steered direction. A potential drawback for star couplers is that they are not path-length-matched and a possible solution is to modulate the widths of output waveguides to match the phases between them. We have designed 32, 240 and 480 channel star couplers with excellent performance. There is a tradeoff in power taper and total insertion loss, but as ~10-13 dB taper actually helps with sidelobe performance, extremely low insertion loss star coupler with large channel count can be designed. Here, we choose the input waveguide width as 800 nm and with this aperture ~95% of total power can be coupled into waveguide array. Note that our star coupler is more compact (only 1 mm × 1 mm footprint for a 1 × 240 coupler) compared with MMI couplers used in some other OPA demonstrations [29–33]. The fabrication is the same as Si waveguide process and we show the fabricated device in Fig. 12(a). We coupled the TE light to the input waveguide and collected the light at output waveguide facets. Figures 12(b) and 12(c) show the IR images of the star splitter region and the output facets. It can be seen that there is small scattering loss at the interfaces between waveguide/array and the free propagation region. In order to estimate the insertion loss, we measured the transmitted powers every 10th channel and then fitted the transmission profile with a Gaussian function as shown in Fig. 12(d). The total power is obtained by summing up the transmission for all channels using the fitted curve, and by comparing with a reference waveguide the insertion loss of the star coupler is estimated to be less than 1 dB, limited by coupling loss variations. Such a low-loss star coupler is suitable for building compact and scalable OPAs.
4.3 Phase shifters
On-chip OPAs require densely integrated high-performance phase shifter arrays that are capable of low operating voltage and low power consumption, low optical loss and large bandwidth, thus allowing for energy-efficient and scalable OPA systems for practical applications. In Si-based phase shifters, it remains hard to overcome the trade-offs among those device performances, due to an inherent low plasma-dispersion effect and considerable carrier absorption in Si. In our phase shifter design , we integrate III-V PN diodes on Si waveguides and the diodes provide phase tuning for the optical mode in the hybrid III-V/Si waveguides. The III-V stack consists of a top p contact, p-InP cladding layer, InAlGaAs III-V MQW between two separate confinement heterostructure (SCH) layers, and a n-InP layer for bottom n contact. The details of the III-V epitaxial layer design can be found in .
In order to study the characteristics of the individual III-V/Si phase shifter, we implement an unbalanced Mach–Zehnder interferometer (MZI) modulator with one arm integrated with a III-V phase shifter, as schematically shown in Fig. 13(a). The MZI consists of two 2 × 2 MMI couplers with ports 1, 2 and ports 3, 4 functioning as optical input and output, respectively. The designed power splitting ratio of the MMI is nearly 1:1 at 1550 nm. The coupling of the optical mode between Si and III-V is accomplished by a III-V taper which is carefully designed to minimize the coupling loss. The III-V/Si phase shifter is designed in a 3-channel array configuration with two side channels as dummy devices as shown in Fig. 13(b), thus to resemble the fabrication condition in our OPAs guaranteeing a similar device performance between the test phase shifter and those in the OPAs. We choose the phase shifter length of 5 mm and the III-V mesa width of 2 µm in the 4 µm pitch array, and the Si waveguide width of 600 nm with a confinement factor in MQWs of 27% for fundamental TE mode in the hybrid waveguide section as shown in Fig. 13(c). The devices were fabricated on 100 mm SOI wafer using 248 nm DUV lithography and a III-V/Si heterogeneous integration technique as described in Section 2. In Fig. 14, we show the fabricated MZI modulator with a 5mm long III-V phase shifter on one arm for phase modulation.
We first characterized the transmission spectrum of the MZI without biasing the III-V diode and present the spectra from port 1 to 3 and port 1 to 4 in Fig. 15(a). The extinction ratio is around 30 dB for the tested device and between 23 and 30 dB for all our working devices. Using our previously developed method , the extra loss induced by III-V was estimated to be 0.3-0.6 dB from transmission spectrum, which is mainly attributed to taper coupling loss. The power splitting ratio is also extracted to be ~50%. The I-V curve of III-V diode is shown in Fig. 15(b) and a dark current at −2 V bias is only 3 nA, implying a static power consumption at a level of nanowatts for the phase shifter. The excellent optical and electrical performance of the MZI also prove the ability of the III-V/Si process for large-dimension heterogeneous devices.
When the III-V diode is reversely biased, the optical index as well as the absorption changes. In order to extract the absorption due to bias, we measured the voltage-dependent transmission of a 5 mm long III-V/Si waveguide and show the result in Fig. 16(a). It can be seen that for the wavelength below 1500 nm, the transmission decreases rapidly when increasing the reverse bias, since the wavelength is close to the absorption edge of QW even at zero bias. For longer wavelengths (e.g., 1550 nm), the transmission decay is very slow with the bias from 0 to −2 V but fast with larger reverse biases due to the redshift of absorption edge by QCSE. Note that for long wavelengths (>1600 nm) there is a bias region in which the transmission slight increases from zero to a certain reverse bias, which could be attributed to reduced absorption by depleting carriers at reverse bias. The phase modulation properties were studied by measuring voltage-dependent transmission of the MZI. Figure 16(b) shows the transmission from port 1 to 3 and port 1 to 4 respectively, when changing the bias voltage applied on the III-V diode. We obtained a V2π of 0.9 V for a 2π phase shift at 1550 nm and an extinction ratio larger than 23 dB from first two transmission peaks. When further increasing the reverse bias, the V2π almost doesn’t change while the transmission peak decreases due to increasing absorption loss in III-V as discussed in Fig. 16(a). In Fig. 16(c), we present the V2π as a function of wavelength and the operating voltage is in a range of 0.35 V to 1.4 V over a wavelength range of 1450 nm to 1650 nm. From the result in Fig. 16(a), the absorption loss (i.e., RAM) within V2π (in lowest absorption window) operation is also extracted and shown in Fig. 16(c). It can be seen that the loss penalty is only 0.1-0.15 dB for the wavelength from 1530 nm to 1650 nm. For shorter wavelengths, the loss increases as explained previously, which could be alleviated by shifting the absorption edge of III-V QWs to shorter wavelengths in future design. The V2π can be further reduced via increasing mode confinement in III-V. Finally, we measured the frequency response of the phase shifter as shown in Fig. 16(d) and obtain a 3dB electrical bandwidth of ~1.65 GHz limited by capacitance. Such operation speed is adequate for OPA applications. The achieved high-performance III-V phase shifter enables low optical loss (~0.15 dB for C-band), low static power (< 3 nW), low operating voltage (about −1 V for C-band), and high speed, which are ideal for OPAs.
4.4 Optical phased arrays
An on-chip OPA needs a densely integrated phase shifter array in order to maximize effective aperture size relative to the chip size in the OPA dimension and to pursue even diffraction-limited and grating-lobe-free far-field beams. The demonstrated phase shifters allow us to make efficient OPAs and we then employ them in our 1D OPA demonstration by integrating those phase shifters in a dense configuration. Figure 17 shows a schematic design of the OPA which consists of a 1 × N (N = 32 or 240 in our design) star coupler, a 5mm III-V/Si phase-shifter array, and a 10 mm antenna array made of ~14 nm shallow etched Si surface grating as studied above. The cross section of the phase shifters is the same as that in Fig. 13(b) and the pitch of the OPA is 4 µm. We also designed 2 µm pitch OPA where the III-V mesa width is 1 µm. The grating array has two designs with a pitch of 4 µm and 2 µm, respectively. Especially, the probe pads for phase shifter are designed in a cascaded layout along the phase shifter so that all channels can be probed and biased separately. They can also be driven by flip chipping the chip onto a CMOS driver array. In order to study the far field beam of the OPA, we couple the light into the input waveguide and distribute it into the III-V phase shifter array through the star coupler. After the phase shift, the light is coupled back into Si waveguide layer and radiated coherently by the grating array. The emission direction of the beam in free space, as denoted by the coordinates of θ and ψ as shown in Fig. 17, are defined as the emission angles in waveguide dimension (i.e., longitudinal axis) and OPA dimension (i.e., lateral axis), respectively. In a 1D OPA, 2D beam steering can be accomplished in θ and ψ axes by wavelength tuning and OPA phase tuning, respectively.
The fabrication of the OPAs is similar to our phase shifters as described above. In Figs. 18(a) and 18(b), we show the fabricated 240-channel and 32-channel OPAs with a pitch of 4 µm. For the specific 240-channel OPA in Fig. 18(a), there are a few phase shifters failed due to local peeling of III-V layer during fabrication and the yield of working diodes is ~90%, which can readily be improved by optimizing the wafer-bonding process. For 2 µm phase-shifter pitch OPA, the yield of III-V diodes is low, mainly limited by lithography which could be addressed by using more advanced tools.
Since an electrical driver and interposer are needed to drive the 240-channel OPA for beam steering, under laboratory conditions we only focus on the 32-channel OPA for the demonstration of 2D beam steering simply using a probe card to drive all phase shifters. The 32-channel OPA with a 4 µm pitch grating array was characterized on the Fourier imaging setup as we described in Fig. 8. The phase shifters were reverse biased using a computer-controlled 32-channel voltage source as shown in Fig. 19(a). The TE polarized light from a broadband tunable laser was coupled into the input waveguide via a lens fiber and the IR camera was used to acquire the far field image and to provide feedback for real-time phase optimization. We first characterized the I-V property of the phase shifters and show the result in Fig. 19(b). It can be seen that the electrical performance of III-V diode is quite uniform and the dark current is in a range of 1-3 nA at −1 V bias, suggesting the power consumption on the order of nanowatts for all phase shifters at static operation. The uniformity of the performance to some extent proves the reliability of the heterogeneous III-V/Si platform for larger-scale devices once processes are transferred to more advanced fabrication facilities.
Next, we studied the beam properties and first need to align the phases of all channels to focus the beam at a fixed wavelength, since the initial phases of the light from the grating array is not aligned due to the lack of path-length-matching strategy and accumulation of phase errors by fabrication imperfections. Here, we employ the gradient descent method to optimize the bias voltages of the 32 phase shifters and define an objective function as , where P(θ,ψ) is the beam power at a target angle (θ,ψ) and P0 is the total power in a first grating-lobe region. The initial voltages are set to – V(i = 1→32) = −0.5 V for all channels and thus the voltage change between zero and −1 V is able to cover 2π phase shift at 1550 nm. Figure 20(a) shows the evolution of the beam at different iterations. We found that the objective function shows a very fast convergence; after only five iterations the voltages converge and the target function reaches 98% of the final optimized value, which can also be seen from the beam profiles in Fig. 20(b). We obtain a side-mode suppression ratio (SMSR) of 16 dB after focus (here the SMSR is defined in a single grating-lobe region). In practice, a look-up table of optimized voltages for phase shifters can be generated to steer the beam arbitrarily without further need for real-time optimization. From Fig. 20(b) it is also evident that for the 4 µm pitch OPA the maximum beam steering angle in ψ axis is ψmax ≈22° in which the grating lobes are not present, matching with the theoretical value given by with λ = 1550 nm and Λ = 4 µm. The steering range is ≈51° in the OPA with a 2µm pitch grating array.
We demonstrated the beam steering in ψ axis by tuning the phases of the OPA at a fixed wavelength and show the results in Fig. 21. For the OPA with a 4µm pitch for both phase-shifter array and grating array, the beam steering in ψ axis covers the entire free tuning range of 22° as shown in Fig. 21(a), and at 1550 nm the FWHM beam width was measured to be δψ = 0.78° which can be easily reduced by increasing the number of elements in OPAs (, e.g., 0.1° for a 240-channel OPA). The steering range in ψ axis can be further increased via shrinkage of the pitch of grating array. By tapering the pitch from 4 µm in phase shifters and to 2 µm in grating array, we further demonstrated 51° steering range in ψ axis, as shown in Fig. 21(b) with the FWHM beam width of δψ = 1.75°. However, a more useful way to obtain a large steering range or even to pursue diffraction-limited and grating-lobe-free beams, is pushing the OPA pitch down to even subwavelength. Moreover, this also increases the usage efficiency of the reticle size, which is especially important for large-scale OPAs with centimeter-scale apertures. For the III-V/Si OPAs, a pitch down to 2 µm or smaller should be achievable by optimization and use of more advanced tools.
By tuning the wavelength, we further demonstrated steering the beam in θ axis. Figure 22(a) presents IR images of the far-field beams (only main lobes) with the wavelength tuning from 1450 nm to 1650 nm, which also verifies the broad band operation in the phase shifters. Figure 22(b) shows the beam profiles in θ axis with a total steering range of ~28° over a 200 nm wavelength tuning and a noise floor less than −20 dB (limited by the IR camera used). Thus the tuning efficiency by wavelength is 0.138°/nm, agreeing with a calculated value of 0.139°/nm. The FWHM beam width was measured to be 0.02° at 1550 nm meaning a ~2.5 mm effective aperture size in θ axis. Note that this aperture can be further increased by engineering the emission rate of surface gratings in our developed platform as we pointed out previously. To illustrate the 2D beam steering, we also present the 3D plots for the beams (only main lobes) obtained by tuning both phases of OPA and wavelength (from 1520 nm to 1580 nm) in Figs. 23(a) and 23(b) for the OPAs with 4 µm and 2 µm grating pitch, respectively.
The demonstrated heterogeneous III-V/Si OPA, with its dense phase shifters, paves the way for fully-integrated LiDAR systems addressing one of key concerns of silicon photonics OPAs whose phase shifters are either too power hungry and slow (thermal tuning) or suffer from large residual amplitude modulation (carrier based tuning). Furthermore, it can be mass produced utilizing same tools and techniques employed at 300 mm wafer semiconductor facilities as demonstrated by heterogeneous datacom products currently being shipped in volume .
In summary we have demonstrated a path for two key photonics based sensors for autonomous cars based on heterogeneous silicon photonics – gyroscope and LiDAR. Heterogeneous silicon photonics, not only addresses a key limitation of silicon photonics by bringing electrically pumped lasers to the platform, but also brings superior phase shifters which are crucial for best performance of both types of sensors. Furthermore, it can be mass produced using same tools and techniques, allowing for significant SWaP-C reduction needed for mass deployment.
In the case of the gyroscope, the whole demonstrated front-end is only 4.5 mm2 in size providing drastic size reduction and also increased robustness to vibration, shock and electromagnetic interference. LiDAR is an extremely active area of research, with the main push for reducing the cost while providing sufficient levels of performance needed for autonomous cars. One of the key components of LiDARs is the beam steerer, and there is great promise for cost and size reduction in developing monolithic beam steerers that can be integrated with transceiver and mass fabricated. Most current demonstrations utilize silicon photonics and OPAs, but standard phase shifters available in silicon photonics generally don’t meet the requirements either due to power consumption, speed or residual amplitude modulation. To address that shortcoming, we have designed, fabricated and characterized a dense III-V/Si OPA system with a pitch of 4 µm. Our phase shifters have low residual amplitude modulation (~0.15 dB in C-band), low static power consumption (few nW), low operating voltage (about −1 V in C-band), wide optical bandwidth (200 + nm) and high operating speed (> 1 GHz), paving the way to fully-integrated LiDAR chips comprising of both the transceiver and OPA manufactured on 300 mm lines using state-of-the-art tools.
Defense Advanced Research Projects Agency (DARPA) MTO (MOABB HR0011-16-C-0106 and iWOG HR0011-14-C-0111).
The authors thank Larry Coldren and Jonathan Klamkin from University of California, Santa Barbara and Paul Suni and James R. Colosimo from Lockheed Martin for useful discussions, and MJ Kennedy for wafer-bonding help. This research was funded by the Defense Advanced Research Projects Agency (DARPA). The views, opinions and/or findings expressed are those of the author and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.
1. T. Komljenovic, D. Huang, P. Pintus, M. A. Tran, M. L. Davenport, and J. E. Bowers, “Photonic integrated circuits using heterogeneous integration on silicon,” in Proc. of the IEEE, doi: [CrossRef]
2. A. W. Fang, H. Park, O. Cohen, R. Jones, M. J. Paniccia, and J. E. Bowers, “Electrically pumped hybrid AlGaInAs-silicon evanescent laser,” Opt. Express 14(20), 9203–9210 (2006). [CrossRef] [PubMed]
3. R. Jones, P. Doussiere, J. B. Driscoll, W. Lin, H. Yu, Y. Akulova, T. Komljenovic, and J. E. Bowers, “Heterogeneously integrated photonics,” IEEE Nanotechnol. Mag. (In Press).
4. G. Roelkens, L. Liu, D. Liang, R. Jones, A. Fang, B. Koch, and J. Bowers, “III-V/silicon photonics for on-chip and inter-chip optical interconnects,” Laser Photonics Rev. 4(6), 751–779 (2010). [CrossRef]
5. D. Liang, G. Roelkens, R. Baets, and J. E. Bowers, “Hybrid integrated platforms for silicon photonics,” Materials (Basel) 3(3), 1782–1802 (2010). [CrossRef]
6. M. J. R. Heck, J. F. Bauters, M. L. Davenport, J. K. Doylend, S. Jain, G. Kurczveil, S. Srinivasan, Y. B. Tang, and J. E. Bowers, “Hybrid silicon photonic integrated circuit technology,” IEEE J. Sel. Top. Quantum Electron. 19(4), 6100117 (2013). [CrossRef]
8. A. W. Fang, B. R. Koch, K. G. Gan, H. Park, R. Jones, O. Cohen, M. J. Paniccia, D. J. Blumenthal, and J. E. Bowers, “A racetrack mode-locked silicon evanescent laser,” Opt. Express 16(2), 1393–1398 (2008). [CrossRef] [PubMed]
9. T. Komljenovic, S. Srinivasan, E. Norberg, M. Davenport, G. Fish, and J. E. Bowers, “Widely tunable narrow-linewidth monolithically integrated external-cavity semiconductor Lasers,” IEEE J. Sel. Top. Quantum Electron. 21(6), 214–222 (2015). [CrossRef]
10. M. A. Tran, D. Huang, T. Komljenovic, J. Peters, and J. E. Bowers, “A 2.5 kHz linewidth widely tunable laser with booster SOA integrated on silicon,” 2018 IEEE International Semiconductor Laser Conference (ISLC), 1 (2018). [CrossRef]
11. H. Park, A. W. Fang, O. Cohen, R. Jones, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs-silicon evanescent amplifier,” IEEE Photonics Technol. Lett. 19(4), 230–232 (2007). [CrossRef]
12. M. L. Davenport, S. Skendzic, N. Volet, J. C. Hulme, M. J. R. Heck, and J. E. Bowers, “Heterogeneous silicon/III-V semiconductor optical amplifiers,” IEEE J. Sel. Top. Quantum Electron. 22(6), 78–88 (2016). [CrossRef]
13. H. W. Chen, Y. H. Kuo, and J. E. Bowers, “A hybrid silicon-AlGaInAs phase modulator,” IEEE Photonics Technol. Lett. 20(23), 1920–1922 (2008). [CrossRef]
16. H. Park, A. W. Fang, R. Jones, O. Cohen, O. Raday, M. N. Sysak, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs-silicon evanescent waveguide photodetector,” Opt. Express 15(10), 6044–6052 (2007). [CrossRef] [PubMed]
17. H. Park, Y. H. Kuo, A. W. Fang, R. Jones, O. Cohen, M. J. Paniccia, and J. E. Bowers, “A hybrid AlGaInAs-silicon evanescent preamplifier and photodetector,” Opt. Express 15(21), 13539–13546 (2007). [CrossRef] [PubMed]
18. C. Zhang, S. J. Zhang, J. D. Peters, and J. E. Bowers, “8 x 8 x 40 Gbps fully integrated silicon photonic network on chip,” Optica 3(7), 785–786 (2016). [CrossRef]
19. M. A. Tran, T. Komljenovic, J. C. Hulme, M. J. Kennedy, D. J. Blumenthal, and J. E. Bowers, “Integrated optical driver for interferometric optical gyroscopes,” Opt. Express 25(4), 3826–3840 (2017). [CrossRef] [PubMed]
20. J. C. Hulme, J. K. Doylend, M. J. R. Heck, J. D. Peters, M. L. Davenport, J. T. Bovington, L. A. Coldren, and J. E. Bowers, “Fully integrated hybrid silicon two dimensional beam scanner,” Opt. Express 23(5), 5861–5874 (2015). [CrossRef] [PubMed]
21. H. C. Lefevre, The Fiber-Optic Gyroscope, second edition (Artech House, 2014).
22. S. Gundavarapu, M. Belt, T. A. Huffman, M. A. Tran, T. Komljenovic, J. E. Bowers, and D. J. Blumenthal, “Interferometric optical gyroscope based on an integrated Si3N4 low-loss waveguide coil,” J. Lightwave Technol. 36(4), 1185–1191 (2018). [CrossRef]
23. F. Dell’Olio, T. Tatoli, C. Ciminelli, and M. N. Armenise, “Recent advances in miniaturized optical gyroscopes,” J. Eur. Opt. Soc-Rapid. 91 (2014).
24. J. Li, M. G. Suh, and K. Vahala, “Microresonator Brillouin gyroscope,” Optica 4(3), 346–348 (2017). [CrossRef]
25. W. Liang, V. S. Ilchenko, A. A. Savchenkov, E. Dale, D. Eliyahu, A. B. Matsko, and L. Maleki, “Resonant microphotonic gyroscope,” Optica 4(1), 114–117 (2017). [CrossRef]
26. P. P. Khial, A. D. White, and A. Hajimiri, “Nanophotonic optical gyroscope with reciprocal sensitivity enhancement,” Nat. Photonics 12(11), 671–675 (2018). [CrossRef]
27. J. Sawyer, P. B. Ruffin, and C. C. Sung, “Investigation of the effects of temporal thermal gradients in fiber optic gyroscope sensing coils,” Opt. Eng. 36(1), 29–34 (1997). [CrossRef]
29. K. Van Acoleyen, W. Bogaerts, J. Jágerská, N. Le Thomas, R. Houdré, and R. Baets, “Off-chip beam steering with a one-dimensional optical phased array on silicon-on-insulator,” Opt. Lett. 34(9), 1477–1479 (2009). [CrossRef] [PubMed]
31. K. Van Acoleyen, K. Komorowska, W. Bogaerts, and R. Baets, “One-dimensional off-chip beam steering and shaping using optical phased arrays on silicon-on-insulator,” J. Lightwave Technol. 29(23), 3500–3505 (2011). [CrossRef]
32. J. K. Doylend, M. J. R. Heck, J. T. Bovington, J. D. Peters, L. A. Coldren, and J. E. Bowers, “Two-dimensional free-space beam steering with an optical phased array on silicon-on-insulator,” Opt. Express 19(22), 21595–21604 (2011). [CrossRef] [PubMed]
33. J. K. Doylend, M. J. R. Heck, J. T. Bovington, J. D. Peters, M. L. Davenport, L. A. Coldren, and J. E. Bowers, “Hybrid III/V silicon photonic source with integrated 1D free-space beam steering,” Opt. Lett. 37(20), 4257–4259 (2012). [CrossRef] [PubMed]
35. D. Kwong, A. Hosseini, J. Covey, Y. Zhang, X. Xu, H. Subbaraman, and R. T. Chen, “On-chip silicon optical phased array for two-dimensional beam steering,” Opt. Lett. 39(4), 941–944 (2014). [CrossRef] [PubMed]
37. D. N. Hutchison, J. Sun, J. K. Doylend, R. Kumar, J. Heck, W. Kim, C. T. Phare, A. Feshali, and H. S. Rong, “High-resolution aliasing-free optical beam steering,” Optica 3(8), 887–890 (2016). [CrossRef]
39. C. V. Poulton, A. Yaacobi, D. B. Cole, M. J. Byrd, M. Raval, D. Vermeulen, and M. R. Watts, “Coherent solid-state LIDAR with silicon photonic optical phased arrays,” Opt. Lett. 42(20), 4091–4094 (2017). [CrossRef] [PubMed]
40. C. V. Poulton, P. Russo, E. Timurdogan, M. Whitson, M. J. Byrd, E. Hosseini, B. Moss, Z. Su, D. Vermeulen, and M. R. Watts, “High-performance integrated optical phased arrays for chip-scale beam steering and LiDAR,” in Conference on Lasers and Electro-Optics. p. ATu3R.2 (2018). [CrossRef]
41. C. V. Poulton, M. J. Byrd, E. Timurdogan, P. Russo, D. Vermeulen, and M. R. Watts, “Optical phased arrays for integrated beam steering,” 15th International Conference on Group IV Photonics (GFP) (2018). [CrossRef]
42. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thomson, “Silicon optical modulators,” Nat. Photonics 4(8), 518–526 (2010). [CrossRef]
44. Q. S. Huang, Y. C. Wu, K. Q. Ma, J. H. Zhang, W. Q. Xie, X. Fu, Y. C. Shi, K. X. Chen, J. J. He, D. Van Thourhout, G. Roelkens, L. Liu, and S. L. He, “Low driving voltage band-filling-based III-V-on-silicon electroabsorption modulator,” Appl. Phys. Lett. 108(14), 141104 (2016). [CrossRef]
45. T. Hiraki, T. Aihara, K. Hasebe, K. Takeda, T. Fujii, T. Kakitsuka, T. Tsuchizawa, H. Fukuda, and S. Matsuo, “Heterogeneously integrated III-V/Si MOS capacitor Mach-Zehnder modulator,” Nat. Photonics 11(8), 482–485 (2017). [CrossRef]
46. J. H. Han, F. Boeuf, J. Fujikata, S. Takahashi, S. Takagi, and M. Takenaka, “Efficient low-loss InGaAsP/Si hybrid MOS optical modulator,” Nat. Photonics 11(8), 486–490 (2017). [CrossRef]
48. M. Zadka, Y. C. Chang, A. Mohanty, C. T. Phare, S. P. Roberts, and M. Lipson, “On-chip platform for a phased array with minimal beam divergence and wide field-of-view,” Opt. Express 26(3), 2528–2534 (2018). [CrossRef] [PubMed]
49. W. Xie, J. Huang, T. Komljenovic, L. Coldren, and J. E. Bowers, “Diffraction limited centimeter scale radiator: metasurface grating antenna for phased array LiDAR,” arXiv:1810.00109 (2018).
50. E. J. Stanton, N. Volet, T. Komljenovic, and J. E. Bowers, “Star coupler for high-etendue LIDAR,” in Conference on Lasers and Electro-Optics. STh1M.4 (2017). [CrossRef]
51. M. R. Kossey, C. Rizk, and A. C. Foster, “End-fire silicon optical phased array with half-wavelength spacing,” APL Photonics 3(1), 011301 (2018). [CrossRef]
52. W. Xie, T. Komljenovic, J. Huang, M. L. Davenport, and J. E. Bowers, “Dense III-V/Si phase-shifter based optical phased array,” arXiv (2019).
53. M. A. Tran, T. Komljenovic, J. C. Hulme, M. L. Davenport, and J. E. Bowers, “A robust method for characterization of optical waveguides and couplers,” IEEE Photonics Technol. Lett. 28(14), 1517–1520 (2016). [CrossRef]