Vertical-cavity surface-emitting phase shifter

In this work we theoretically propose a novel method towards obtaining a two-dimensional beam steering system which could be used for LIDAR applications. The method utilizes a hybrid plasmonic waveguide in a cylindrical topology creating a nanoscale vertical cavity surface emitting phaseshifter (VCSEP). This approach enables a sub-wavelength spacing between each phase shifter and therefore provides a larger field of view. Each VCSEP consists of a highly doped sub-micron silicon pillar coated with a thin layer of non-linear material then another layer of conductive metal. The input optical carrier is inserted from the polished back side of the silicon substrate, which is then modulated by the phase shifter and re-emitted in the vertical direction to form a beam in the far field. The hybrid plasmonic localization increases the effective interaction length of the light with the non-linear material within a silicon core resonant cavity. Increased interaction length due to plasmonic effects in the vertical architecture leads to an enhanced phase shift. A lower loss is possible while retaining high localization due to an overlap of photonic mode and plasmonic modes. The impedance mismatch between the vertical hybrid plasmonic waveguide and free space will create a low finesse, low-Q cavity resonator. The single VCSEP device can be considered as a low finesse Fabry-Perot resonator that can be used for phase modulation with less than 3dB amplitude modulation. We characterized the optical response of a single VCSEP, with an aspect ratio of 12.5 which has an FSR of 47.25±2.5nm and transmission variation of 3dB.


INTRODUCTION:
Current scanning of optical beams is performed using laser sources coupled to free-space optical systems comprised of mechanical components, moving parts, and bulk optics. Unfortunately, the application range of these legacy systems is limited by their size, weight, reliability and cost. Consequently, a substantial research effort has been directed towards the miniaturization and simplification of these systems. Recent work has focused on beam steering using phased arrays.
Although optical phased arrays are an elegant non-mechanical beam steering approach, the technical and environmental challenges compared to RF systems (10,000 times smaller wavelengths and tolerances) are daunting. The main challenge to construct optical phased array beam steering technology will rely on our ability to construct ultra-compact phase shifters with footprint < 1 µm 2 and power consumption < 5µW/2π. Moreover, these phase shifters should enable scalability to large size two dimensional (2D) arrays with period on the order of the wavelength of light to allow wide angular scanning range and at high angular resolution.
Currently, compact chip scale 2D phased array optics are implemented using 2D arrays of phase shifters based on waveguide coupled directional gratings [1]. These devices are based on a photonic architecture in which photonic waveguides feed a linear array of add/drop filters. The amount of coupling into each add/drop filter can be modulated thermally via metal heaters. From the add/drop, the light is passed to a grating nanoantennas which emits the light vertically. Compact 64 x 64 arrays of phase shifters with a pixel size of 9 μm × 9 μm or 89 μm 2 have been reported [1]. Such approaches are fundamentally limited in pixel size by the size of the waveguide couplers themselves, which due to the physical requirements of wavelength scaling, are necessarily longer than 1 µm, as seen in [1]. Likewise, the heaters themselves add a non-negligible footprint requirement both in terms of the physical size of the heaters, and the required thermal isolation via distance from nearby pixels. Additionally, the heaters have a limited response time. It may be possible to control coupling in such devices via direct current injection, but even then, that approach would still be limited by the physical size of the waveguide coupler.
The typical 2D beam steering scheme utilizes a phase difference between integrated optical phase shifter for each dimension. However, in order to realize a 2D beam steering the spacing between each component of each shifter needed to be very small to ensure an acceptable field of view; The beam steering angle (Φ) is related the array period (d) by the equation = arcsin( 2 ) . Additionally, most of the suggested 2D optical beam steering devices are only tuned in one dimension using a grating antenna, while the second dimension is tuned by shifting the operating wavelength in conjunction with a fixed diffraction grating (thereby changing the diffraction angle). This method greatly limits the field of view in the second dimensions. [1,2] To further progress the state of the art and achieve improved performance metrics of phase shifter arrays for beam forming applications, this work proposes an alternative approach to the design of a large 2D array with sub wavelength spacing by utilizing a hybrid plasmonic modulator.

APPROACH DESCRIPTION AND DESIGN:
Hybrid plasmonic modulators have recently been reported [3,4] combining the advantages of photonic and plasmonic structures. These hybrid plasmonic waveguides consist of a thin noble metal layer, typically on the order of 20 nm thick, sitting atop a thin layer of nonlinear dielectric, which in turn sits atop a traditional silicon photonic waveguide. The result is a structure where the photonic mode is "pulled" out of the Si waveguide by the noble metal and is located predominantly in the non-linear dielectric layer between the Si and the noble metal. Critically, pulling a mode out of a photonic waveguide via a metal layer, rather than using a traditional metal-dielectric-metal plasmonic guiding structure, reduces the interaction area between the mode and the lossy metal while still retaining most of the plasmonic localization. [3,4]. This reduces loss to be between 800 to 2000 dB/cm while still supporting localized higher k-vector plasmonic modes. As a further advantage, the resulting mode sits almost entirely in the non-linear material.
Our approach uses hybrid plasmonic resonators realized vertically with the input optical carrier inserted from an integrated bottom layer, which is modulated by the phase shifter and re-emitted in the vertical direction to beam form in the far field, as seen in Figure 1. When light passes through the vertical-cavity surface-emitting phase shifters (VCSEPs) it will be converted to a hybrid plasmonic mode where it will be localized and pass through a non-linear material. Through combined localization and resonant effects, the effective interaction length will be greatly increased. A voltage will be applied to the metal shell via a metal contact, and the doped Si core attached to a doped Si substrate will serve as the ground. This will induce an electric field within the non-linear electro optic material within the VCSEP, modulating the refractive index within an individual VCSEP device. The combination of greater effective interaction length and modulating refractive index within the cavity will achieve a phase shift. The light will then be re-emitted into free space; in this configuration the VCSEP operates in transmission mode. VCSEPs use hybrid plasmonic localization to increase effective interaction length of the light with the non-linear material within the resonant cavity, thus increasing the interaction length due to the plasmonic effect. Lower loss (~1000 dB /cm) is possible while retaining high localization due to overlap of photonic mode and plasmonic modes. Furthermore, the impedance mismatch between the vertical hybrid plasmonic waveguide and free space will create a low finesse, low-Q cavity resonator.
By utilizing a low-Q cavity the traditional limitations on bandwidth and spectral range found in a high Q photonic resonator-based modulator is reduced, while still gaining interaction length due to a cavity finesse of 10-20. This effect is then compounded with large k-vector plasmonic modes that are localized in space: hybrid plasmonic modes stimulated by a free space 1550nm laser will have an effective wavelength that is up to 20x shorter in the hybrid plasmonic structure [5,6]. Combined with an example cavity Q of 10 i.e. assuming a photon will, on average, make 10 round trips through the cavity, the VCSEP will provide a 400x increase in effective interaction length with the non-linear material. In other words, a 5 µm tall VCSEP may have an effective interaction length of 2000 µm or 2mm with the non-linear materials. The VCSEP couples photonic modes to plasmonic modes through a lower index dielectric (e.g. SiNx) [8,9] surrounding a high index core (e.g., Si) of the coaxial structure illustrated in Fig.1.Taken together in a vertical topology, this hybridizes the advantages of plasmonic localization with a low loss photonic mode, where metal also provides the electrical contact in an ultra-compact vertical topology, as shown in Fig. 2. Due to vertical architecture, the suggested VCSEP design can have a very large packing density, ≥1/µm 2 . Flexibility in hybrid plasmonic design allows us to optimize for low loss (<1000 dB/cm) or high localization (e.g. k mode>30 k0 were k0 is the wavevector in free space), depending on the respective radii of the Si core, the interstitial layer of non-linear dielectric, and the metal shell. With a thick Si core and metal cladding, the photonic mode is localized in the silicon core which reduces the propagation losses, as shown in Fig.2a. With a thin Si core, the field is well confined in the non-linear dielectric and strongly localized in a thin section between the metal and the Si core creating greater overall interaction with the non-linear dielectric while retaining some plasmonic localization of non-linear dielectric at the expense of increased loss, as shown in of Fig.2b. This would be the closest to a direct a b cylindrical analogue to the hybrid plasmonic waveguides found in literature, as shown in Fig.1.
Lumerical FDTD mode simulations of the vertical cylindrical hybrid plasmon cavity with a diameter of 400 nm are shown in Fig. 2. The simulated structure is estimated to have a loss of 2000 dB/cm, which is consistent with values seen in literature for hybrid plasmonic structures. The fabrication process begins with a spin coating of 500nm PMMA , and E-beam writing was performed using Vistec EBPG 5200 with dose of 1500 µC/cm 2 . The pattern consists of different circle diameters ranging from 400 to 300 nm. The separation distance between each VCSEP is varied from 1µm to 13µm, on different groups of structures. AL2O3 was sputtered on to the transferred pattern to act as a hard mask. In literature it has been demonstrated that aluminum oxide (Al2O3) has a selectivity of 70,000:1 to silicon [7], which is the best choice for difficult fabrication procedures. A 5nm layer of Al2O3 was RF sputtered using Denton Discovery 635 Sputter System. The chamber pressure was kept at 3.18 mTorr and the RF power was set to be 400 W, and a lift off process was done afterward to remove the excess resist. The single-step deep reactive ion etching (SDRIE) was done by the Oxford Plasmalab 100 using a highly controlled etching recipe that maintained smooth walls and has a high etching rate (200nm/min) [8]. The VCSEP pillars etching was executed by a combination of SF6/Ar plasma with the addition of C4F8 to protect the sidewalls of the pillars by the deposition of organic polymer. The flow rate of both the SF6 and C4F8 gases are kept at 28-sccma and 52-sccm, respectively. During the etching process, the chamber pressure was maintained to be 19mTorr. The RIE and ICP generated power is kept at 9W and 850W, respectively [8]. The height of each pillar was determined to be 4µm with an aspect ratio of 10. Afterward, the hard mask was removed by leaving the sample in a chemical etchant, Buffered Oxide Etch (BOE), for 10 second with etching rate of (1nm/sec). To ensure all the polymer is removed, the sample is cleaned with piranha mixture (H2SO4:H2O2 3:1) at a temperature of 80 º C. Next the Oxford PECVD was used to coat the VCSEP pillars with silicon-rich nitride (SRN). During the process, the chamber pressure was kept at 15 mTorr and RF power was kept at 50 W. The SRN was deposited using a combination of SiH4 and NH3. Lastly, the pillars were coated with a thin film of Au using the Denton 635. The chamber pressure was maintained at 2.88 mTorr during the sputtering process. The DC plasma power was kept at 200 W for 55 second to obtain a gold thickness of 40nm. To ensure uniform coating of the gold, the stage was rotated at 10 rpm. a b

RESULTS AND DISCUSSION:
The Characterization of the VECSEP is carried out using the experiment setup illustrated in Fig.5. The laser source (Agilent model 81980A) is fiber coupled and then collimated before the light is transmitted through the back end of the sample. The transmitted light is collected by a 0.5NA microscope objective and projected onto the camera by magnification setup which consist of lenses, acting as two sequential telescopes with a total magnification of 125. The camera has a 25µm-by-25µm pixel size. Figure 5: Experimental setup used to characterize the VCSEP which has a total magnification of 100x.
Due to the ~1um size of an individual VCSEP, locating the correct structure is challenging.
To simplify the characterization method, the sample is fabricated with large alignment marks, as seen in Fig.6, to identify the small area where a single VCSEP or pair of VCSEP are located. Fig.6 shows the SEM images of the fabricated VCSEP and image obtained from the IR camera.  An iris is used in the first image plane of the optical setup in order to isolate a single VCSEP structure. This spatial filter limits any unwanted noise that might affect the device a b a b measurements. Fig.7 Shows the optical image detected from the IR camera after using the iris to limit the illuminated area to be as close as possible to the VCSEP illuminated area.
The optical characterization of a single VCSEP device is performed using a tunable laser with a tuning range from 1460 nm to 1640 nm. The measured transmission results are summarized in Fig. 8. To further understand the analyzed structure each single VCSEP device can be considered as a sequence of Fabry-Perot resonators with three different resonators defined in the VCSEP cross section in Fig. 9. The first one consists of air-silicon wafer-VCSEP resonator, the second one consists of air-silicon wafer-air resonator and the third resonator is the one of interest: silicon wafer-VCSEP-air resonator. Identifying the resonance response of each resonator is necessary to understand the results obtained from characterizing the single VCSEP device, as seen Fig. 7, since the optical spot size is larger than the diameter of an individual VCSEP device and therefore, the measured result corresponds to the response of these three resonators. Further details on the response of these three resonators and their Free Spectral ranges (FSRs) are depicted in Fig. 10.  With this analysis in mind, the analyzed measurement response shown in Fig. 9 have provided the following observations: The transmission spectrum of the air-silicon waferair and the air-silicon wafer-VCSEP have a FSR of about 1 nm (see Fig. 8a). We also need to consider the power spectra of the source and the effect of spectra on coupling efficiency. These were taken into consideration to normalize the measured data and obtain Fig. 8b. Finally, we performed a Hilbert Transform to remove the fast oscillations due to wafer resonance and obtain a curve corresponding to transmission of a single VCSEP device shown in Fig. 8C. We estimate from the optical response of a single VCSEP device with an aspect ratio of 12.5 to have an FSR of 47.25±2.5nm and a transmission variation of 3dB. These results confirm our design performance.

CONCLUSIONS:
This manuscript introduces a novel approach for the realization of phase shifters utilizing hybrid resonance structures in a cylindrical topology. This new approach offers a compact architecture for integrated scanners useful for LIDAR applications. This manuscript also shows a novel design and simulation tools as well as nanofabrication procedures and demonstrated fabricated devices with aspect ratios as high as 15:1 with diameters less than 1 µm (footprint <1 µm 2 ) enabling dense integration of VCSEPs. Such a small footprint makes it possible to achieve the high density required for wide angular bandwidth scanning with high resolution across a large area. The constructed optical system was used to characterize the optical response of a single VCSEP device. The transmission spectrum is obtained by sweeping the laser source from 1460nm to 1640nm and it can be noticed that the transmission power is reduced as the wavelength increases. This is due to coupling efficiency and this effect can be eliminated by normalizing the transmission spectrum. The envelope of the spectrum can be obtained by applying a Hilbert transform function, thus we can estimate the response of a single VCSEP, with an aspect ratio of 12.5 which has an estimated FSR of 47.25±2.5nm and transmission variation of 3dB.