In the last decades, dual-band optical filters (DBOFs) have attracted great attentions as they play important roles in various optical applications, such as optical modulation [1], sensing [2], fast/slow light [3], etc. Multiple microring resonators have been utilizing in various methods to realize the dual-band filtering spectrum [4-6]. However, the multiple-microring-resonators structure is not suitable for high volume integration due to its large form factor. In this work, we firstly developed a novel design based on single-microracetrack-resonator structure embedded with Bragg gratings, which is more compact and feasible for large-scale integration. Besides, through changing the dimensional sizes of the Bragg gratings, the reflectivity of the Bragg grating can be tuned and the transmission spectrum can be engineered as well.

The schematic of the microracetrack resonator embedded with Bragg gratings is shown in Fig. 1. Two embedded Bragg gratings act as partial reflective elements to form a Fabry-Perot resonator with a part of the microracetrack. The coupling between the F-P resonator and the microracetrack gives rise to the dual-band filtering transmission spectrum. In the simulation, we regard the Bragg grating as a periodic structure consisting of wide waveguide segments, narrow segments and reflective interfaces [7]. As shown in Fig. 2, the transmission spectrum is calculated with a specially designed numerical model based on the transfer matrix method. In this simulation, the parameters are chosen as: width of the waveguides = 500 nm, dw = 20 nm, pitch = 324 nm, the number of pitches = 20, radius of the microracetrack = 10.2 μm. The effective indices of the waveguides under different wavelengths around 1.55 µm are calculated with the BeamPROP module of Rsoft software by taking the dispersion into account.

As shown in Fig. 3, the microracetrack resonator embedded with Bragg is fabricated on a SOI platform with a 220-nm top silicon slayer and a 2um buried oxide (BOX) layer. The grating layer and the waveguide layer both are patterned with electron beam lithography. The grating layer is etched with RIE to the depth of 70 nm. The waveguide layer is etched down to the BOX layer. Then the sample is coated with a 1 um SiO2 cladding layer. The characterizations are carried out at room temperature. The experimental results are plotted in Fig. 4. As it can be seen, the extinction ratios (ERs) of the dual-band filtering spectrum are ~ 4.5 dB. The separation between two notches is ~ 0.5 nm. The full-width-at-half-maximums (FWHMs) are 0.05 nm and 0.045 nm, which corresponds to the Q factors of 30900 and 34400. The insertion loss is < 0.5 dB. The uneven transmission may be caused by the F-P resonator formed by the two end facets of the bus waveguide. By reducing the roughness of the waveguide sidewalls, the ER can be further increased.

This compact dual-band filter is expected to have good potential for applications in optical modulation, optical signal processing, and biological/chemical sensing.

Integrating nanoscale electromechanical transducers and nanophotonic devices potentially can enable new acousto-optic devices to reach unprecedented high frequencies and modulation efficiency. We demonstrate acousto-optic modulation of a photonic crystal nanocavity using acoustic waves with frequency up to 19 GHz, reaching the microwave K band. Both the acoustic and photonic devices are fabricated in piezoelectric aluminum nitride thin films. Excitation of acoustic waves is achieved with interdigital transducers with periods as small as 300 nm. Confining both acoustic wave and optical wave within the thickness of the membrane leads to improved acousto-optic modulation efficiency in the new devices than that obtained in previous surface acoustic wave devices. Our system demonstrates a novel scalable optomechanical platform where strong acousto-optic coupling between cavity-confined photons and high frequency traveling phonons can be explored.

Exploitation of light-sound interactions in various types of media has led to a plethora of important optical technologies ranging from acousto-optic devices for optical communication to photo-acoustic imaging in biomedicine. With advances of integrated photonics and nanofabrication technology, it is now more feasible to miniaturize and integrate acousto-optic devices to augment their speed and performance so they can assume indispensable roles in integrated photonic systems for chip-scale optical communication. Moreover, in addition to electromechanical excitation, acoustic waves or localized mechanical vibrations can also be optically stimulated through optomechanical forces including radiation pressure, gradient force and electrostriction. Such optomechanical effects recently has been extensively investigated in various optomechanical systems with dimensional scales ranging from meters to nanometers. Therefore, with these recent developments, acousto-optics is entering a new era with plenty of research opportunities.

To combine acoustic and photonic devices, previously we and other groups have used piezoelectric aluminum nitride (AlN) film deposited on silicon wafers with a layer of silicon dioxide (SiO₂). Since AlN has a relatively high refractive index of about 2.1, photonic waveguides and cavities can be fabricated in AlN with the SiO₂ layer as the cladding. At the same time, acoustic waves can be exited in the AlN layer. On this platform, we have demonstrated SAW wave with frequency up to 12 GHz and its acousto-optic modulation of optical ring resonators and photonic crystal nanocavities. However, an important drawback of above devices based on SAW is that the SiO₂ layer has a lower sound velocity than both the top AlN layer and the bottom silicon substrate. As a result, the excited acoustic waves in the AlN layer tend to leak into and be guided in the SiO₂ layer whereas the optical modes are highly confined in the top AlN layer because of its high refractive index. Consequently, the modal overlap between the optical and acoustic modes is low, leading to relatively weak acousto-optic coupling efficiency.

To circumvent this problem, we further implemented integrated acousto-optic devices on a suspended AlN membrane. When the membrane thickness is less than or comparable to the acoustic wave, the generated acoustic wave will propagate in Lamb wave mode with very high acoustic velocity. The removal of the substrate enforces the optical mode and the acoustic wave to maximally overlap within the thickness of the membrane. With this approach, we demonstrate acousto-optic modulation of a photonic crystal nanocavity at frequency up to 19 GHz with significantly improved modulation efficiency. The system overcomes the limitation of the unsuspended AlN platform where the acoustic wave leaks into the substrate layer without contributing to the acousto-optic interaction. The platform is promising for studying interaction between cavity confined photons and propagating phonons of microwave frequency. In additional, the strong and high frequency acoustic waves realized in this platform can provide spatially-coherent time-domain modulation to induce non-reciprocity and break time-reversal symmetry in photonic systems. At this high acoustic frequency, this system can also be applied for microwave photonics technology where optical and microwave channels of communication can be linked and interchanged.

Refractometer is an important device to measure the concentration of biochemical such as biotoxins in drinking water based on refractive index change [1, 2]. Silicon photonics device is a promising platform for refractometer development with the advantages of high sensitivity, fast response, capable of real-time measurement. Moreover, silicon photonics device fabrication is CMOS compactable, which can significantly reduce device footprint and manufacturing cost. In this paper, a Mach-Zehnder interferometer based refractometer is proposed to achieve high sensitivity and wide measurement range.

A Mach-Zehnder interferometer is used for refractive index measurement. The schematic of the device is shown in Figure 1(a). Light is coupled into the waveguide from the left terminal and then split in the Y-branch into the upper sensing arm and the bottom reference arm. The waveguide used has thickness of 340 nm and width of 500 nm. The sensing and reference arms have a total length of 5 mm and 4.9 mm, respectively. The ultra-low loss Y-branch has a loss less than 0.28 dB [3], which is shown in Figure 1(b). The top cladding layer of the sensing arm is removed to expose the silicon waveguide. When the sensing arm is immersed into different media (refractive index), the refractive index change in the sensing area will cause the change in the optical path difference of the two arm. Thus, the interference pattern measured on the right terminal will change and such change can be quantified to measure the concentration of chemical and biochemical molecules in drinking water.

In our sensor design, the 340-nm thickness and 500-nm width waveguide will support both TE- and TM- mode. Since TE- and TM- mode has different group index, there present two different periods corresponded to TE- and TM- in the device spectrum. And it can be distinguished by doing fast Fourier transform to the spectrum. Additionally, the two modes have different sensitivities, which can be used to complement each other.

The refractometry response is measured by adding ethanol (with refractive index of 1.36) on top of chip surface. The refractive index change will be 0.36 in this case. Figure 2 (a) shows the spectrum measured when cladding is ethanol and Figure 2 (b) shows the spectrum measured in air. And after using fast Fourier transform to those spectrums and using chirp Z-transform to refine the result, the power changing frequency of each condition is shown in Figure 3. There are mainly 2 peak represented the TE and TM mode. Further improvement can be made to add polarization controller to filter out only one mode and detect the two modes separately.

As calculated from Figure 3, the period sensitivity ST is 2.9 nm/RIU for TE mode and 4.21 nm/RIU for TM mode. This period sensitivity can be used to determine the large shift of cladding refractive index. And the very high wavelength sensitivity of 600 nm/RIU can be used to calculate the small change of cladding refractive index. By combine both the TE and TM mode result, and both the period shift and wavelength data, high sensitivity and large measurement range can be achieved.

An optical mode converter is proposed and designed. It consist of Y-branch splitters, combiners, and phase shifter array. The input fundamental mode can be converted to TE₁, TE₂, TE₃ mode by controlling the phase shifters. Furthermore, the proposed mode converter support high order optical modes conversion by the expansion of the structure. It is designed for the wavelength of mid infra-red (MIR). This device shows potential applications for the mode division multiplexing (MDM) system at MIR.

.

In recent years, on-chip mode division multiplexing system has attracted much more attention since its pave a way to increase the capacity in single wavelength. Plenty of building blocks for the MDM system have been reported and show excellent performance, such as the mode-division switch ^[1-2], mode multiplexer^[3], mode converter^[4], etc. Among these devices, one of the important device is the optical mode converter, which manage the optical mode in the MDM network. This work is focus on the demonstration of CMOS-compatible integrated silicon optical mode converter, which support high order modes conversion, with scalable structure. We also forward the working wavelength of the device from the NIR to MIR, since MIR photonics provide great potential application in the sensing, communication, thermal imaging ^[5-7], etc.

The proposed silicon mode converter contains three parts, including splitter, phase shifter array and combiners, as shown in Fig. 1. The input light TE₀ mode divided into four beams. After propagation through the phase shifters array, the light will be modulated with different phase. The four beams will be combined by two stage Y branch. We can choose the proper phase modulation to get the different optical mode. The working manner is listed in Table 1. Taking the conversion from TE₀ mode and TE₂ mode for example, in order to generate the TE₂ order optical mode, two in-phase TE₁ optical mode is required to superposition at the second-stage Y-branch combiner. So, phase shifter A and B works to generate two in-phase optical beam and superposition at the first-stage combiner to generate two in-phase TE₁ order optical mode, as shown in Fig. 2(c). The thermo/electric-optic effect could be employed to generate π shift. The simulation is performed by the FD-BPM method.

In conclusion, the silicon optical mode converter can support the conversion between TE ₀ mode and 0-3^th mode, which provide much more flexibility for mode management in the optical interconnects. This structure could be updated to higher modes conversion by extension the structures.

Opto-mechanical systems offer one of the most sensitive methods for detecting mechanical motion using shifts in the optical resonance frequency of the opto-mechanical resonator. Presently, these systems are used for measuring mechanical thermal noise displacement or mechanical motion actuated by optical forces. Meanwhile, electrostrictive and piezoelectric actuation and detection are the main transduction schemes used in RF MEMS resonators.

In this talk, I will introduce a method for actuating an opto-mechanical resonator using MEMS transducers and sensing of mechanical motion by using the optical intensity modulation at the output of an opto-mechanical resonator. I will discuss classical applications enabled by this hybrid platform such as multi-GHz Acousto-Optic Modulators (AOM) and Opto-Acoustic Oscillators.  I will conclude my talk by providing a glimpse of how we are leveraging our mastery of micromachining and MEMS to achieve coherent transduction between spin-defects, phonons and photons.

We develop single-photon sources that simultaneously combines high purity, efficiency, and indistinguishability. We demonstrate entanglement among ten single photons. We construct high-performance multi-photon boson sampling machines to race against classical computers to reach the goal of quantum computational supremacy.

Boson sampling is considered as a strong candidate to demonstrate the “quantum supremacy” over classical computers. However, previous proof-of-principle experiments suffered from small photon number and low sampling rates owing to the inefficiencies of the single-photon sources and multi-port optical interferometers. In this talk, I will report two routes towards building Boson Sampling machines with many photons.

In the first path, we developed SPDC two-photon source with simultaneously a collection efficiency of ~70% and an indistinguishability of ~91% between independent photons. With this, we demonstrate genuine entanglement of ten photons [1]. Such a platform will provide enabling technologies for teleportation of multiple properties of photons [2] and efficient scattershot Boson Sampling.

In the second path, using a QD-micropillar, we produced single photons with high purity (>99%), near-unity indistinguishability for >1000 photons [3], and high extraction efficiency [4]—all combined in a single device compatibly and simultaneously. We build 3-, 4-, and 5-bosonsampling machines which runs >24,000 times faster than all the previous experiments, and for the first time reaches a complexity about 100 times faster than the first electronic computer (ENIAC) and transistorized computer (TRADIC) in the human history [5,6]. Our architecture is feasible to be scaled up, and might provide experimental evidence against the Extended Church-Turing Thesis.

We study the trapping dynamics of a nanoparticle near a silicon slot waveguide. It is shown for the first time that a nanoparticle can go along different paths before it becomes trapped. The optical forces originating from different parts of an optical mode compete with each other during particle transport, forming a complicated particle trajectory and particularly a critical area where particle transport becomes unstable. Brownian motion is simulated for 50-nm and 100-nm nanoparticles, and the induced instability is analyzed.

Infrared (IR) spectroscopy probes the phonon vibrational states of molecules by measuring wavelength-dependent optical absorption in the mid-infrared regime (2.5-25 µm wavelength or 400-4,000 cm^-1 in wave number).  It is widely recognized as the gold standard for chemical analysis given its superior specificity. Traditional IR spectroscopy relies on fragile bench-top instruments located in dedicated laboratories, and is thus not suitable for emerging field-deployed applications such as in-line industrial process control, environmental monitoring, and point-of-care diagnosis. Recent strides in photonic integration technologies provide a promising route towards enabling miniaturized, rugged platforms for IR spectroscopic analysis. It is therefore attempting to simply replace the bulky discrete optical elements used in conventional IR spectroscopy with their on-chip counterparts. This size down-scaling approach, however, cripples the system performance as both the sensitivity of spectroscopic sensors and spectral resolution of spectrometers scale with optical path length.

To address this challenge, we present the development of two novel photonic device designs uniquely capable of reaping performance benefits from microphotonic scaling [1].  Firstly, we leverage strong optical and thermal confinement in judiciously designed microcavities to circumvent the thermal diffusion and optical diffraction limits in conventional photothermal sensors and achieve parts- per-billion (ppb) level gas molecule limit of detection. Photothermal spectroscopy has been predicted as a technique 10⁴ times more sensitive than conventional microcavity absorption spectroscopy [2, 3]. We propose a new technique that monitors the shift in cavity resonance as a result of heating.  Resonant absorption of cavity light by the gas molecules thermally heats the cavity, producing a large resonant shift in a suitably designed microdisk cavity, as schematically illustrated in Fig. 1. 

In the second example, a novel spectrometer technology based on on-chip digital Fourier Transform InfraRed (dFTIR) is proposed to overcome the aforementioned spectral resolution limit facing current on-chip spectrometers. As schematically depicted in Fig. 2, the dFTIR spectrometer consists of an interferometer with arms comprising a series of cascaded optical switches connected by waveguides of varying lengths. Each permutation of the switches modifies the physical waveguide path lengths and thus spectrometer configurations. Such a dFTIR approach is far more effective for varying the optical path length than index modulation and hence enables perform metrics comparable or even superior than conventional benchtop instruments.  The spectrometer design fulfills Fellgett’s advantage offering dramatically-improved SNR and enables sub-nm spectral resolution (as projected in Fig, 3) on a millimeter-sized, fully-packaged photonic integrated chip without mechanical moving parts. 

In addition to the advantages of miniaturization, low cost, high integration and large array, MEMS technologies also endow the traditional optical elements with totally new functionalities, showing great potentials for such applications as imaging, optical communication, sensing, projection, and so on. In this paper, two kinds of MEMS-based spatial light modulators (SLMs) will be concerned, one named as the micro programmable blazed grating developed in our group, and another one, the digital micromirror device (DMD) commercialized by TI. In addition, the prototypes of multispectral and hyperspectral imaging systems utilizing them will also be demonstrated. The preliminary experimental results for imaging artificial targets seem encouraging.

.

We  propose  and  experimentally  demonstrate  broadband,  low-crosstalk,  and  low-loss  polarization splitter-rotators  (PSRs)  with  optimized  taper structures  at  1550-nm  and  1310-nm  wavelengths, respectively.  The  PSRs  consist  of  particle  swarm  optimization  (PSO)  based  bi-level  tapers  and
shortcuts to adiabaticity (STA) based ridge-waveguide couplers. Ridge waveguides are introduced to increase  the  coupling  coefficient  of  the  STA  based  coupler  and  to  reduce  the  crosstalk  from  TM0 mode. The measured polarization conversion losses (PCLs) and crosstalk (CT) are less than 0.6 dB and  –20 dB,  respectively,  from  1500 nm  to  1600 nm  wavelength  for  the  1550-nm  PSR.  The measured  PCLs  and  CT  are  less  than  1 dB  and  –22 dB, respectively,  from  1260 nm  to  1340 nm wavelength for the 1310-nm PSR.
 
High-performance  PSRs  are  preferred to  realize  polarization  diversity  for  silicon  photonic  circuits. Various  structures  of  PSR  have  been  proposed, including  asymmetrical  directional  coupler  (ADC) [1,2],  bi-level  taper  plus  ADC  [3],  bi-level  taper  plus  multimode  interference  coupler  [4],  bi-level
taper plus adiabatic coupler [5], and bi-level taper plus asymmetric Y-junction [6]. However, none of them  can  have  a  PCL  value  less  than  1 dB  and  a  CT  value  less  than  –20 dB  within  a  80-nm-wide wavelength range.
 
The scanning electron microscope (SEM) pictures of the 1550-nm and 1310-nm PSRs are shown in Fig.  1  and  Fig.  2,  respectively.  The  PSR  consists  of  a  TM0-TE1  bi-level  taper  and  a  TE0-TE1 demultiplexer.  The  bi-level  tapers  are  optimized  based  on  the  PSO  method.  The  tapers,  which  are 20-μm  long,  are  divided  into  4  and  10  segments  with  equal  length  for  the  1550-nm  and  1310-nm PSR  respectively.  The  maximum  values  of  the  average  TM0–TE1 conversion efficiencies over  the wavelengths from 1500 nm to 1600 nm and from 1260 nm to 1360 nm are set to the Figure of Merit for the 1550-nm  and  1310-nm  PSRs,  respectively.  The  TE0-TE1  demultiplexer is  optimized  based on  the  STA  method  [7].  The  coupling  length  is  70 μm.  The  minimum  edge-gaps  are  200 nm  and 160 nm for the 1550-nm and 1310-nm PSRs, respectively. The total PSR lengths are about 120 μm.
 
Fig. 3 shows the measured results of the 1550-nm PSR. The PCLs and CT are less than 0.6 dB and –20 dB,  respectively,  from  1500 nm  to  1600 nm  wavelength.  Fig.  4  shows  the  measured  results  of the 1310-nm PSR. The PCLs and CT are less than 1 dB and –22 dB, respectively, from 1260 nm to
1340 nm  wavelength.  The  measured  broadband  operation  is  limited  by  the  available  bandwidth  of grating couplers.
 
In  conclusion,  we  have  proposed  and  demonstrated  practical  PSRs  at  1550-nm  and  1310-nm wavelengths, respectively. The PSRs show lower PCL less than 1 dB and lower polarization CT less than –20 dB within a 80-nm-wide wavelength range.