We report on microelectromechanical systems (MEMS)-actuated 32 × 32 optical phased arrays (OPAs) with high fill-factors and microsecond response time. To reduce the mirror weight and temperature-dependent curvature, we use high-contrast-grating (HCG) mirrors comprising a single layer of sub-wavelength polysilicon gratings with 400 nm thickness, 1250 nm pitch, and 570 nm grating bar width. The mirror has a broad reflection band and a peak reflectivity of 99.9% at 1550 nm wavelength. With 20 × 20 μm2 pixels and 2 μm, the OPA has a total aperture of 702 × 702 μm2 and a fill factor of 85%. The OPA is electrostatically controlled by voltage and has a total field of view of ± 2°, an instantaneous field of view (beam width) of 0.14°, and a response time of 3.8 μs. The latter agrees well with the mechanical resonance frequency of the HCG mirror (0.42 MHz).
© 2014 Optical Society of America
Optical phased arrays (OPAs) have emerged as a powerful technology for agile, high-resolution, random-access pointing/tracking with multiple simultaneous beams . Applications of OPAs range from 3D display and printing, optical data-storage, telecommunication to military and other industrial applications. Several technologies have been developed for OPAs. The most mature OPAs are based on liquid crystals [2,3]. They are low cost, and can be readily integrated on electronic integrated circuit drivers. The so-called liquid crystal on silicon (LCoS) technology has been used for both microdisplays as well as OPAs . However, the liquid crystal-based OPAs have some drawbacks, including slow response time, fringe field effect, and low steering efficiency at large angles . Liquid prism with electrowetting actuation has been proposed as an alternative approach for beamsteering. It is capable of polarization independent operation and high steering efficiency, but suffers from a slow response time (~milliseconds), limiting its applications . Recently, compact OPAs have been reported using silicon photonic waveguides on a silicon-on-insulator (SOI) with either wavelength tuning  or thermo-optic phase modulation . Large arrays have been made using silicon integrated circuit foundries . For wavelength-tuning approach, the beam direction is dependent on wavelength, and it is not possible to achieve monochromatic beamsteering . Phase tuning using thermo-optic modulators overcome this limit, however, it suffers from high power consumption, particularly for large arrays . In addition, the maximum optical powers in silicon photonic OPAs are limited by the power handling capability of sub-micron-sized waveguides.
Here, we describe a 32 × 32 OPA with high-contrast grating (HCG) reflectors and micro-electromechanical systems (MEMS) actuators. Although MEMS-based OPAs have been reported before, earlier devices employed metal-coated silicon mirrors  which suffer from residual stress-induced curvature and coefficient of thermal expansion (CTE) mismatch between metal and silicon. High-reflectivity HCG have been employed in a tunable vertical cavity surface emitting laser (VCSEL) with a record fast mechanical tuning speed and a high resonance frequency >27 MHz [11,12]. The HCG comprises a thin layer of sub-wavelength gratings with high dielectric constants. The reflectivity and the reflection spectrum can be tailored by varying the grating thickness, pitch, and grating bar width. HCG mirrors with high reflectivity (>99%) and reflection bandwidth larger than distributed Bragg reflectors (DBR) have been reported . Such lightweight mirrors result in fast tuning response. Moreover, the HCG mirrors are made of a single material (e.g. silicon) and free from stress- or thermally-induced curvature. Previously, we have reported a MEMS-actuated 8 × 8 OPA with single crystalline silicon HCG mirrors . Two-dimensional beamsteering was demonstrated. However, the fill factor was limited to 36% due to the large anchor structures in single-etch MEMS process on SOI. In this paper, we use a multilayer polysilicon process with sub-micron polysilicon/silicon nitride posts as anchors. This enables us to achieve high fill-factor (85%). The array size is also increased from 8x8 to 32x32, resulting in narrower beams. Thanks to the light mirror weight, we have achieved a response time of less than 4 μs.
The proposed MEMS OPA comprises a 32 × 32 array of HCG mirrors. Figure 1(a) shows the top view of the OPA layout. Each HCG mirror is tethered to four mechanical springs, one at each corner. The mirrors are electrostatically actuated by applying a voltage between the mirror and the substrate. The mirrors in the same column are electrically connected via their springs and move together when a voltage is applied. As a result, the current OPA can perform one-dimensional beamforming. Potentially, an individually addressable electrode array can be integrated underneath the HCG mirrors for two-dimensional beamforming. Figure 1(b) shows an optical image of the released array. Here, every other row of the OPA has been pulled downwards by electrostatic actuation, creating a periodic phase shift pattern.
For high performance OPA, it is desirable to reduce the array pitch and increase the fill-factor. The pitch determines the total field of view (TFOV), the largest angle one can steer to. High fill-factor is associated with better diffraction efficiency. Since we used the same polysilicon layer for both the HCG mirrors and the springs, the fill factor is limited by the width of the springs and the clearance between the structures. For a given spacing between HCG mirrors, the fill factor reduces with mirror size. Moreover, small MEMS mirrors require higher actuation voltage. To address these issues, the mirror area is chosen to be 20 × 20 μm2. We use a deep ultraviolet (DUV) stepper (ASML PAS 5500/300) with a resolution of 250 nm to pattern the HCG and MEMS structures. Figure 1(c) shows the calculated fill factor and TFOV versus the mirror spacing for an OPA with 20 × 20 μm2 mirrors. The spacing between HCG mirrors is designed to be 2 μm, resulting in a fill factor of 85% and a TFOV of ± 2° at 1550 nm wavelength. It should be mentioned that the reduction of fill factor with mirror pitch is not fundamental. High fill factor can be achieved for small mirror pitch with “hidden springs” by adding a structural layer between the HCG mirror and the substrate.
The schematic and the scanning electron micrograph (SEM) of the HCG mirror are shown in Fig. 2. The HCGs are of interest for a wide range of integrated optoelectronic device applications, such as lasers, filters, splitters, and couplers. In order to design the HCG used in our optical phased array, it is necessary to determine the grating period (Λ) for the desired operating wavelength (λ). For example, the diffraction regime, where the grating period (Λ) is greater or much less than the wavelength (λ) doesn’t result in high reflectivity. In contrast, in the near-wavelength regime, where the period (Λ) is between λ/nr and λ/na, where nr is the refractive index of the grating bars and na is the refractive index of air, extraordinary features such as high reflectivity (~99.9%) and high quality-factor resonance (Q>107) have been realized. It has been demonstrated that the reflectivity of HCGs can exceed 99% over an extraordinarily broad wavelength range of ∆λ/λ~30% [13,15]. For optical MEMS devices, the fact that the HCG is composed of only a single polysilicon layer eliminates the problems of residual stress and differential thermal expansion (e.g. bimorph effects) commonly observed in multi-layer reflectors.
Because the HCG’s grating bar width is on the order of 500 nm, significant (~20%) variation in bar width may result from optical lithography and dry etching. Thus, it is important to design HCGs with large fabrication tolerance. The HCG’s reflectivity is polarization dependent and simulations show that the transverse electric (TE) mode offers a larger fabrication tolerance than the transverse magnetic (TM) mode. The HCG was designed using finite-difference time-domain (FDTD) simulation to compute the reflectivity as a function of bar width, period, and thickness. Figure 3(a) shows the contour map of HCG mirror reflectivity versus the thickness of the HCG and the wavelength for HCG period of 1250 nm in TE mode (defined as optical polarization parallel to the grating bars). To achieve low actuation voltage, thinner devices (and therefore thinner springs) are preferred so the region inside the black square is selected. Figure 3(b) depicts the enlarged reflectivity map. Based on this map, the device thickness of 400 nm is selected to achieve high reflectivity at 1550 nm as well as large fabrication tolerance. Figure 3(c) shows the reflectivity contours versus the bar width and grating period. The dotted line corresponds to the resolution limit (250 nm) of our ASML300 stepper. The HCG design in this paper has a bar width of 570 nm and a period of 1250 nm. The fabrication tolerance for 99% reflectivity is ± 35 nm, well within the capability of our fabrication process. Figure 3(d) depicts the calculated reflection spectrum of the HCG mirror. It has a peak reflectivity of 99.9% at 1550 nm and a broad bandwidth.
3. Fabrication process
A polysilicon surface micromachining process, shown in Fig. 4, is used to fabricate our MEMS OPA. The fabrication begins with depositing a sacrificial low temperature oxide (LTO) film of 3.4 μm thickness on a silicon wafer [Fig. 4(a)]. Then 0.3-μm-diameter holes are patterned and etched through the LTO using DUV lithography and anisotropic dry etching in Fig. 4(b). In Fig. 4(c), 0.1-μm-thick stoichiometric silicon nitride is deposited to coat the holes in LTO for electrical isolation. The silicon nitride over the planar part of LTO is removed by chemical mechanical polishing (CMP), leaving silicon nitride only on the sidewalls of the LTO holes. The slurry used in the CMP process features very high selectivity to silicon dioxide and polysilicon so the silicon nitride filled in the holes remains while the silicon nitride on top of the LTO is removed. Polysilicon deposition is performed as shown in Fig. 4(d). The polysilicon layer is in situ doped to a resistivity of 8 × 10−4 Ω-cm in a low pressure chemical vapor deposition (LPCVD) furnace to achieve low electrical resistance for HCG mirrors and electrical lines. The grain sizes and therefore the surface roughness of polysilicon films deposited by LPCVD are on the order of tens of nanometers. Thus, CMP of the polysilicon surface is required to improve feature definition (i.e. fabrication of gratings with sub-wavelength width) and reduce optical scattering loss. The polysilicon plugs filling the silicon nitride-coated holes serve as the anchors of the HCG mirrors. Repeating the DUV stepper lithography and anisotropic dry etching, the HCG array is patterned in Fig. 4(e). Experimentally, the sidewall angle of HCG is measured to be 80 degrees. This is well within the tolerance by theoretical simulation in . Electrical fan-outs and wire bonding pads are also formed in the same process step. The release process, Fig. 4(f), is performed at the die level. The etch rate of LTO in diluted HF (49% concentrated HF:deionized water = 1:1) is measured to be ~1.1 μm/min so that 2 min dipping is sufficient to remove the LTO sacrificial layer. The device is then rinsed and dried in a critical point dryer (CPD).
Figure 5(a) shows the SEM image of a released 32 × 32 MEMS HCG optical phased array. Figure 5(b) shows an enlarged image of the HCG mirrors. Each HCG mirror is supported by four polysilicon mechanical springs anchored by silicon nitride-coated polysilicon plugs. The mirrors are electrically insulated from the substrate, thanks to the silicon nitride liner. To achieve high fill-factor, the HCG mirrors are electrically connected along the same row via the anchors, as marked by the yellow dotted lines in the Fig. 5(b). Thirty-two gold wire bonding pads are patterned at the edges of the die.
CMP is generally able to achieve a thickness uniformity of ± 10% over a 6-inch wafer, and it is important to determine the remaining thickness of polysilicon after the short CMP process. It is measured using an Alpha-step IQ surface profiler (KLA-Tencor Inc.) after the dry etching of polysilicon in Fig. 4(e). The targeted thickness of polysilicon HCG layer is 0.4 μm. The measured thickness of the HCG layer varies across the wafer from 0.34 μm to 0.41 μm. The red curve in Fig. 6 shows the reflectivity of HCG mirrors versus their thickness calculated using rigorous coupled wave analysis (RCWA). A reflectivity of 99.9% is expected over the thickness range of 316 nm to 402 nm at the wavelength of 1550 nm. The blue curve is the calculated resonance frequency versus HCG thickness. Since we use the same polysilicon layer for both the HCG and the spring, the mirror mass increases linearly with the HCG thickness and the spring constant increases with the cubic power of the spring (and HCG) thickness. As a result, the resonance frequency increases linearly with the HCG thickness. Based on the measured thickness, the mechanical resonance frequency can be expected to be ~0.4 MHz from the blue curve in Fig. 6.
4. Experimental results
The experimental setup for characterizing the OPA is shown in Fig. 7. It includes an imaging interferometer to measure, and subsequently control, the phase shift of each pixel.
Light from a laser source is collimated and directed to the MEMS OPA. Part of the light reflected from the OPA is tapped to a measurement path where it is interfered with a reference beam to form an interferogram on a camera (Xenics Bobcat-1.7-320) . The interferometer allows static and dynamic measurements of phase shift at pixel level. Figure 8(a) shows the measured temporal response of an HCG mirror in response to a 25 kHz square wave excitation at the peak-to-peak voltage of 20V. The phase response is stable to within ± 10% of the final value (0.36π) within 3.8 μs following each transition. The temporal response is fitted well by a damped-harmonic oscillator model with a damping ratio of 0.2 and a damped natural frequency of 0.46 MHz.
Our imaging interferometer is capable of measuring the phase shifts of all mirrors within the field of view. It is a powerful tool to characterize the uniformity of the mirrors in the OPA. We have measured the temporal responses of 180 mirrors for 25 kHz square wave actuation. For each mirror, we obtained a temporal response similar to that shown in Fig. 8(a). The aggregated temporal responses of all 180 mirrors are condensed in Fig. 8(b). The vertical axis is the time base, and the horizontal axis is the mirror number. The phase is represented by pseudo color whose scale is shown on the right. The 180 HCG mirrors exhibit uniform phase shift and similar transient response at each phase transition, with only 2 HCG mirrors, #29 and #45, showing different responses. The mean step height is 0.38 π, the rms. variation in step height is 0.08π. The uniformity of the oscillation period is shown by the histogram in Fig. 8(c). The average oscillation period is calculated to be 2.64 μs, corresponding to an average resonance frequency of 0.42 MHz. The measurement is in good agreement with the calculation based on the thickness measurement in Fig. 6. The resonance frequency of a few representative mirrors was also measured using a laser Doppler vibrometer (LDV), as shown in Fig. 8(d). The resonance frequency agrees well with the value derived from the time response. The resonance frequency of the HCG mirror is higher than conventional MEMS mirrors thanks to the lightweight HCG mirrors. The switching time can be further decreased by employing stiffer springs, allowing for even faster beamsteering.
Figure 9 shows the measured and fitted reflection power spectra of a fabricated polysilicon HCG and a gold mirror in the wavelength range of interest. The dots represent the measured data. The solid lines are polynomial fits of the measured data. The reflectance of the gold mirror is expected to be 98.5% at the 1550 nm wavelength. The measured data shows the reflectivity of the HCG mirror is comparable to the Au mirror. Precise measurement of the exact reflectance of the fabricated polysilicon HCG mirror is difficult due to the small mirror size, 20 × 20 μm2. The measured results indicate that OPAs with single-layer polysilicon HCG mirrors have low optical insertion loss.
Binary phase patterns, with the phase shift of each column of mirrors set to either 0 or π, were used for one-dimensional beamsteering experiments, as shown in Fig. 10(a). Figure 10(b) shows the measured far-field steering results corresponding to the mirror states in Fig. 10(a). State 0 is the case when no voltage is applied to the OPA and the beam is not deflected. From State 1 to State 3, the 0th order beam is suppressed and the light is directed into symmetric diffraction orders by the binary phase pattern. The diffracted beam angle is given by θ = arcsin(λ/2NΛ) where λ = 1550 nm is wavelength of the laser, Λ = 22 μm is the mirror pitch, and 2N is the number of mirrors per phase period. The maximum deflection angle measured at State 3 is ± 2° with the full width at half maximum (FWHM) beam width of 0.14°.
Figure 11(a) illustrates the measured intensity profile for 2N = 0 (State 0), 2 (State 3), 4 (State 2), 8 (State 1), 12, 16, and 20. The calculated beam location matches the measured beamsteering angles in the graph. The intensity of 0th order beam (black curve) appears to be well-suppressed with beamsteering as shown in the colored curves. The diffracted intensity was studied as a function of phase shift φ applied to the mirrors at the maximum steering angle (2N = 2). The intensity in the 1st order diffraction lobe at the maximum angle is given by I = Imax sin2(φ/2) , where the maximum intensity Imax occurs at φ = 1π. The measured intensity versus phase shift agrees very well with the theoretical calculation, as shown in Fig. 11(b). Figure 11(c) shows asymmetrical beamsteering results using a linear phase ramp modulo 2π. The experiments for asymmetrical beamsteering were performed using no more than 1.2π phase shift to avoid possible unwanted pull-in during HCG actuation. Thirty-three different stair-step phase structures using the OPA implement individual beam positioning to the left and right from 0th order beam location (see Media 1).
A novel microelectromechanical systems (MEMS) optical phased array with high fill-factor has been successfully developed for fast beamsteering. The 32 × 32 array is comprised of sub-wavelength high-contrast-grating (HCG) reflectors made of polysilicon. The size of each HCG mirror is 20 × 20 μm2. The spacing between HCG mirrors is 2 μm, resulting in 85% fill-factor. The thickness, width and period of the grating are 400 nm, 570 nm and 1250 nm, respectively. The theoretically calculated reflectivity of the polysilicon HCG mirror is ~99.9%. Each HCG is electrostatically actuated to create a voltage-controlled phase shift. The measured maximum beamsteering angle is ± 2°. The average resonance frequency of the array is 0.42 MHz, allowing for fast beamsteering. Our experimental results show that MEMS-actuated HCG mirrors are promising candidate for (~MHz) optical phased array (OPA) with high resolution.
The authors would like to acknowledge support from the DARPA SWEEPER program (No. HR0011-10-2-0002) and National Science Foundation Center for Integrated Access Network (CIAN) under grant #EEC-0812072.
References and links
1. P. F. McManamon, T. A. Dorschner, D. L. Corkum, L. J. Friedman, D. S. Hobbs, M. Holz, S. Liberman, H. Q. Nguyen, D. P. Resler, R. C. Sharp, and E. A. Watson, “Optical phased array technology,” Proc. IEEE 84(2), 268–298 (1996). [CrossRef]
2. D. P. Resler, D. S. Hobbs, R. C. Sharp, L. J. Friedman, and T. A. Dorschner, “High-efficiency liquid-crystal optical phased-array beam steering,” Opt. Lett. 21(9), 689–691 (1996). [CrossRef] [PubMed]
3. B. Wang, G. Zhang, A. Glushchenko, J. L. West, P. J. Bos, and P. F. McManamon, “Stressed liquid-crystal optical phased array for fast tip-tilt wavefront correction,” Appl. Opt. 44(36), 7754–7759 (2005). [CrossRef] [PubMed]
4. S. Serati and J. Stockley, “Advanced liquid crystal on silicon optical phased arrays,” in Aerosp. Conf. Proc.3, 1395–1402 (2002). [CrossRef]
5. P. F. McManamon, P. J. Bos, M. J. Escuti, J. Heikenfeld, S. Serati, H. Xie, and E. A. Watson, “A review of phased array steering for narrow-band electrooptical systems,” Proc. IEEE 97(6), 1078–1096 (2009). [CrossRef]
6. R. D. Niederriter, A. M. Watson, R. N. Zahreddine, C. J. Cogswell, R. H. Cormack, V. M. Bright, and J. T. Gopinath, “Electrowetting lenses for compensating phase and curvature distortion in arrayed laser systems,” Appl. Opt. 52(14), 3172–3177 (2013). [CrossRef] [PubMed]
8. 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]
10. I. W. Jung, U. Krishnamoorthy, and O. Solgaard, “High fill-factor two-axis gimbaled tip-tilt-piston micromirror array actuated by self-aligned vertical electrostatic combdrives,” J. Microelectromech. Syst. 15(3), 563–571 (2006). [CrossRef]
11. M. C. Y. Huang, Y. Zhou, and C. J. Chang-Hasnain, “A surface-emitting laser incorporating a high-index-contrast subwavelength grating,” Nat. Photon. 1(2), 119–122 (2007). [CrossRef]
13. C. F. R. Mateus, M. C. Y. Huang, Y. Deng, A. R. Neureuther, and C. J. Chang-Hasnain, “Ultrabroadband mirror using low-index cladded subwavelength grating,” IEEE Photon. Technol. Lett. 16(2), 518–520 (2004). [CrossRef]
14. B. W. Yoo, M. Megens, T. Chan, T. Sun, W. Yang, C. J. Chang-Hasnain, D. A. Horsley, and M. C. Wu, “Optical phased array using high contrast gratings for two dimensional beamforming and beamsteering,” Opt. Express 21(10), 12238–12248 (2013). [CrossRef] [PubMed]
15. C. J. Chang-Hasnain and W. Yang, “High-contrast gratings for integrated optoelectronics,” Adv. Opt. Photon. 4(3), 379–440 (2012). [CrossRef]
16. M. C. Y. Huang, K. Lee, D. Parekh, and C. J. Chang-Hasnain, “Large fabrication tolerance of ultra broadband sub-wavelength grating reflctor,” in IEEE LEOS Ann. Mtg.1, 364–365 (2004). [CrossRef]
17. T. K. Chan, M. Megens, B. W. Yoo, J. Wyras, C. J. Chang-Hasnain, M. C. Wu, and D. A. Horsley, “Optical beamsteering using an 8 × 8 MEMS phased array with closed-loop interferometric phase control,” Opt. Express 21(3), 2807–2815 (2013). [CrossRef] [PubMed]
18. D. T. Amm and R. W. Corrigan, “Grating Light ValveTM technology: update and novel applications,” in SID Int. Symp. Dig. Tec.29, 29–32 (1998).