A spatial light modulator (SLM) based on a Fabry-Perot interferometer configuration has been fabricated and tested. The Fabry-Perot spacer layer is a thin film of the SEO100 electro-optic polymer which serves as the nonlinear medium. Measurement results demonstrate the modulation of multiple pixels operating simultaneously at frequencies ranging from 300 kHz to 800 kHz which is significantly faster than SLMs based on liquid crystal and digital micromirror device technology. An average modulation contrast of 50% for all pixels is achieved with a drive voltage of 70 Vrms at 100 kHz. Microwave speeds and CMOS compatibility are feasible with improved transmission line and cavity design.
©2011 Optical Society of America
Spatial light modulators (SLMs) have been used extensively for various applications such as data encryption , digital holography , interferometry , and wavefront correction . Some of the most popular SLM devices on the market today are based on liquid crystal (LC) , digital micromirror device (DMD) , and quantum well (QW) [7,8] technology. While these devices have a proven track record of high performance and reliability they suffer from intrinsic material or device structure properties that make them unsuitable for high speed applications that require a wide optical bandwidth and wavelength stability. Although LC SLMs have the advantage of having low operating voltage requirements, the response time of the LCs is confined to the millisecond to sub-millisecond regime . DMD SLMs also exhibit low voltage operation but the operating speeds are generally limited to a few kHz due to the slow mechanical response and settling time of each individual mirror . Operating speeds in excess of 10 MHz with reasonably low drive voltages have been demonstrated for multiple QW SLMs which in theory would make these devices prime candidates for high speed applications . However, these devices suffer from a narrow optical bandwidth since the operating wavelength must be very close to the semiconductor absorption band. This design constraint is undesirable because small temperature fluctuations can have a significant effect on pixel modulation depth. QW SLMs using a stepped QW structure have demonstrated much wider optical bandwidth operation as compared to multiple QW SLMs but they still are very sensitive to temperature change making wavelength stability a concern .
The use of electro-optic (EO) polymers as the active medium in a SLM is attractive because of their very wide optical bandwidth, room temperature stability, potential for high speed performance due to picosecond response times, and the ease of incorporation into standard microfabrication processes. Previous work has been done demonstrating the operation of single pixel EO polymer and sol-gel based Fabry-Perot interferometers [10–12]. In this paper we demonstrate a potential, next generation EO SLM based on a similar design with operating speeds much faster than current LC and DMD SLM technology.
2. Device design
The general design for a thin film Fabry-Perot interferometer has been described previously in literature [10–13]. The fabricated device consists of an 8x8 array of individually addressable pixels which were designed to operate at a resonant wavelength of 1550 nm. This array is divided into four quadrants each containing 16 identical pixels with pixel sizes of 800 μm, 692 μm, 565 μm, and 400 μm. The quadrant with 565 μm pixels was chosen for analysis due to this region having the least amount of surface roughness in the EO polymer layer. The structural layout of this region is illustrated in Fig. 1 . A high reflectance distributed Bragg relfector (DBR) mirror consisting of 10 periods of alternating SiO2 and Ta2O5 quarter-wave layers serves as one mirror in the Fabry-Perot cavity. The other mirror is a thin film of Au that also serves as the top electrode. The Au is coated with a quarter-wave layer of SiO2 which serves as both an anti-reflection coating and encapsulaton layer to protect the Au surface from damage during the poling process. The pixels are defined by etched 2% Al-doped ZnO (AZO) which also serve as the bottom electrodes. A layer of Ge is patterned in between the pixels for the purpose of drastically reducing the amount of light transmitted due to the high absorption of Ge at 1550 nm coupled with the multiple reflections inside the cavity. This has the effect of increasing the image contrast of the SLM significantly. The spacer layer consists of the SEO100 EO polymer from Soluxra. Simulation of this multi-layer thin film design was performed using Essential Macleod software.
3. Device fabrication
A step by step overview of the fabrication process is illustrated in Fig. 2 . Initially a high reflectance DBR mirror was sputtered onto a clean 2 inch diameter, 2 mm thick fused silica substrate. Afterwards, a 50 nm layer of AZO was deposited onto the surface by pulsed laser deposition (PLD) then annealed at 400 °C.
Next, a bi-layer lift-off process to deposit contact electrodes with connecting traces consisting of layers of Ti/Au was performed as follows: A 400 nm layer of MicroChem LOR 3A lift-off resist was spin-cast onto the substrate then softbaked on a hotplate for 5 minutes at 175 °C. A 1.3 μm layer of Shipley 1813 positive photo-resist was spin-cast on top of the lift-off resist then softbaked for an additional 2 minutes at 115 °C. The sample was then patterned by UV lithography using a dark field mask and the positive resist was developed using Shipley CD-26 metal ion free developer. The sample was baked on a hotplate for an additional 5 minutes at 125 °C to harden the positive resist in prepartion for the undercutting step. The sample was then submerged in CD-26 developer long enough to undercut the lift-off resist sufficiently to ensure good lift-off of the evaporated material and photo-resist. A 30 nm layer of Ti, which serves as an adhesion layer between the Au and substrate and a 250 nm layer of Au were then deposited by electron-beam evaporation. The masked regions of the sample were removed by placing the sample in a bath of Microposit Remover PG that was heated to 80 °C and placed inside a ultrasonicator to ensure complete removal of material. The detailed lift-off procedure is illustrated in Fig. 3(a) .
Next, an etching procedure to pattern the AZO was performed as follows: A 1.3 μm layer of Shipley 1813 positive photo-resist was spin-cast onto the substrate then softbaked on a hotplate for 2 minutes at 115 °C. The sample was patterned by UV lithography using a clear field mask, then developed using Shipley 352 developer. Etching of the unmasked regions of AZO was done using a dilute etchant consisting of HCl and H2O in a ratio of 1:800 which gives an etch rate of approximately 5 nm/s. The resist was then stripped to reveal patterned pixels with connecting traces that are aligned directly underneath the Ti/Au traces created in the previous lift-off step. The detailed etching procedure is illustrated in Fig. 3(b).
Next, the Ge contrast enhancment layer is deposited using the previously described lift-off procedure. The higher the cavity finesse is, the lower the thickness of Ge required to block essentially all light in between pixels. A thickness of 50 nm was sufficient for this application. The EO polymer used for this device is the Soluxra produced guest-host polymer SEO100 which is a blended polymer and chromophore mixture that is dissolved in dibromomethane to create a polymer solution that is 6 wt%. The EO polymer was spin-cast to create a 1625 nm thick film and dried under vacuum at 80 °C overnight. A 45 nm layer of Au and a 270 nm layer of SiO2 were deposited on top of the polymer by electron-beam evaporation. Kynar 30-gauge wires were soldered to the bottom Au electrodes. Since the top Au electrode/mirror is very thin it was not possible to solder a wire directly to it. Instead a two-part high operating temperature silver conductive epoxy was used to attach a 30-gauge wire. All wires were covered in two-part insulating epoxy to provide additional mechanical stability. The device was poled using a field of 75 V/μm on a hot plate that was initially at room temperture while increasing the temperature as fast as possible until it reached 133 °C. The poled device was then cooled to ambient room temperature and the applied field was removed.
4. Device characterization
The detailed setup created for characterizing SLM device performance is illustrated in Fig. 4 . The setup is structured such that the upper portion is used for characterizing the throughput of the SLM while the lower portion is used to characterize the performance of each pixel individually. A fiber coupled C-band tunable laser and a C-band ASE source are combined further using a 2-1 coupler with the output connected to a fiber collimator. This output is combined with the output from a spatially filtered free-space C-band tunable laser using a beam splitter. The light that is traveling towards the upper portion of the setup then passes through a variable ND Filter and is folded using a plane mirror oriented at 45 degrees. The light then goes through an adjustable polarizer and is expanded using a Galilean beam expander to create a collimated beam approximately 30 mm in diameter that is normally incident on the sample. The sample is attached to a rigid kinematic mount that provides two-axis translation. A magnification system consisting of two plano-convex lenses is used after the sample to completely fill the CCD array of the camera (Goodrich SU320M) with an iris being used to remove any stray background light. The reason for coupling the two C-band tunable lasers is to help correct the non-uniformity in the device throughput caused by thickness variations in the spin-cast EO polymer films. The divided light that travels towards the lower portion of the setup passes through a reversed Keplerian beam expander that decreases the size of the beam to a collimated spot size of approximately 500 μm which is normally incident on the sample. The sample is translated in the transverse directions to couple light through each pixel. An objective is used to focus the light that passes through the pixel onto a high speed detector (InGaAs) or a multimode fiber that is attached to a precision 5-axis stage and connected to an optical spectrum analyzer (OSA).
4.1 DC performance
After poling, pixel-by-pixel characterization was performed on the SLM. Using a C-band ASE source the optical spectrum of each pixel is measured by coupling light through the multimode fiber that is connected to the OSA. By taking into account the phase dispersion at the active medium interfaces of the SLM the average finesse is estimated to be 52. To measure the spectrum shift due to an applied electric field, a DC voltage is applied and varied from −200 V to 200 V with the results shown for an arbitrary pixel in Fig. 5 (a) . The noticeable change in peak intensity as the applied voltage is varied is due to the unbalanced mirror reflectance of the Fabry-Perot cavity. The average tunability is calculated to be 20 pm/V and the average resonant wavelength shift is illustrated in Fig. 5(b). The ASE source is then replaced with a C-band tunable laser and tuned to the resonant wavelength of each pixel. The average insertion loss is 22 dB and determined by measuring the ratio of the power throughput of each pixel over the power after the sample is removed. The applied DC voltage is varied from −200 V to 200 V while the pixel output is monitored using a detector. From this data the average normalized intensity and the average isolation ratio, which is the average normalized intensity in dB, are calculated with the results being displayed in Fig. 5(c).
4.2 Modulation performance
To measure the modulation performance of the SLM a sinusoidal voltage is applied simultaneously to all pixels while performing optical measurements on each pixel individually. The reason for applying voltages to all pixels at the same time was to ensure that high speed performance could be observed with the device in full operation. A C-band tunable laser is tuned to the resonant wavelength and the applied voltage is varied from 10 Vrms to 100 Vrms at a frequency of 100 kHz to measure the average modulation contrast for all pixels with the results illustrated in Fig. 6(a) . Since the transmitted light is modulated regardless of the sign of the applied voltage, the signal has a response frequency that is twice the modulation frequency. An oscilloscope screenshot showing a pixel modulation contrast of 28% and a drive signal of 50 Vrms at 500 kHz is shown in Fig. 6(b). The Pockel’s coefficient r13 is estimated to be 22.3 pm/V based on the device performance, which is in close agreement to the measurement results obtained from the reflection ellipsometry  and Mach-Zehnder interferometry  methods. To determine the broadband performance of each pixel the drive frequency is varied and the modulation contrast is recorded. Due to the structure of the SLM each pixel can be approximately modeled as a first order RC circuit. The normalized modulation contrast frequency response for select pixels is shown in Fig. 6(c).
4.3 Device throughput
Next the device throughput of the SLM was examined. The C-band ASE source was used to show the uniformity of the SLM throughput over a broad range of wavelengths with the resulting image shown in Fig. 7(a) . To demonstrate the spatial operation of the SLM array a single wavelength source is needed while a DC voltage is applied. Due to the surface of the polymer being rough we chose to use two coupled C-band lasers tuned to wavelengths of 1545 nm and 1555 nm to increase image uniformity. A 200 VDC signal was applied to selected pixels to demonstrate the contrast between the on and off states. The contrast of the camera was adjusted so that a clear distinction between the on and off states could be observed. Images of the device throughput as a voltage is applied to selected pixels are shown in Fig. 7 (b)-(l). A video file demonstrating the modulation of the top two rows of the 4x4 array with a drive signal 150 Vrms at 2 Hz is shown in Fig. 8 .
5. Future design considerations
Due to the difficulty in controlling both the amplitude and phase of the light transmitted through the SLM, this device is best suited for applications that require operation in binary mode. Dynamic holography  is a current application that needs both wavelength stability and high speed modulation which current SLM technology is not able to provide. Even though our current SLM design shows potential to be suitable for holography applications there are a few design improvements that need to be made in our next generation of devices to make this a reality.
In order to provide the high resolution and pixel count that holography requires a set of photomasks with much smaller pixel sizes and an electronic driver with a much larger number of channels would need to be designed. Driving an SLM using a large drive voltage at high frequencies in large densely packed pixel arrays may prove to be difficult therefore the device voltage requirements need to be addressed. The modulation voltage can be dramatically reduced by increasing the finesse of the device or using alternative poling methods such as coplanar electrode poling to utilize the larger Pockel’s coefficient, r33. The major source of loss in the SLM is caused by the high absorption of the Au electrode/mirror and the absorption of the AZO layer due to the multiple reflections inside the cavity. The insertion loss can be reduced dramatically by replacing the Au layer with the combination of a DBR mirror and an AZO or other transparent conductive oxide (TCO) layer to serve as the top electrode. If both bottom and top electrode TCO layers are placed outside of the cavity the insertion loss can be further reduced to less than 1 dB but the tradeoff will be an increase in both modulation voltage and RC time constant due to the voltage drop across the DBR mirrors. The speed is currently limited because of the high resistance of the traces which consist of an AZO region overlapped with a Ti/Au region. The variation in speed for each pixel is due to the different AZO trace length for each pixel. Speeds in the microwave regime can be achieved simply by making the traces entirely out of Ti/Au and using microwave engineering design software to properly design the transmission lines and impedance match the device to the drive electronics.
In conclusion we have demonstrated a potentially high speed SLM performance with the modulation of individual pixels at speeds ranging from 300 kHz to 800 kHz which is significantly faster than current LC and DMD SLM technology. An average modulation contrast of 50% is achieved with a drive voltage of 70 Vrms at 100 kHz. Microwave speeds can be achieved simply by improving transmission line design circuitry. By improving the cavity finesse, increasing the poling efficiency, and adopting alternative poling geometries it will be possible to reduce the drive voltage significantly, making CMOS compatibility possible.
The authors acknowledge support from the National Science Foundation MDITR Science and Technology Center under Grant # 0120967 and the National Science Foundation through CIAN NSF ERC under grant # EEC-0812072.
References and links
1. D. S. Monaghan, U. Gopinathan, B. M. Hennelly, D. P. Kelly, T. J. Naughton, and J. T. Sheridan, “Applications of spatial light modulators in optical signal processing systems,” Proc. SPIE 5827, 358–368 (2005). [CrossRef]
4. L. Hu, L. Xuan, Y. Liu, Z. Cao, D. Li, and Q. Mu, “Phase-only liquid crystal spatial light modulator for wavefront correction with high precision,” Opt. Express 12(26), 6403–6409 (2004). [CrossRef] [PubMed]
5. K. M. Johnson, D. J. McKnight, and I. Underwood, “Smart spatial light modulators using liquid crystals on silicon,” IEEE J. Quantum Electron. 29(2), 699–714 (1993). [CrossRef]
6. D. Dudley, W. M. Duncan, and J. Slaughter, “Emerging digital micromirror device (DMD) applications,” Proc. SPIE 4985, 14–25 (2003). [CrossRef]
7. B. Noharet, Q. Wang, S. Junique, D. Agren, and S. Almqvist, “Multiple quantum well spatial light modulators for optical signal processing,” Proc. SPIE 5618, 146–155 (2004). [CrossRef]
8. H. Mohseni, W. K. Chan, H. An, A. Ulmer, and D. Capewell, “Tunable surface-normal modulators operating near 1550 nm with a high-extinction ratio at high temperatures,” IEEE Photon. Technol. Lett. 18(1), 214–216 (2006). [CrossRef]
9. K. A. Bauchert, S. A. Serati, and A. Furman, “Advances in liquid crystal spatial light modulators,” Proc. SPIE 4734, 35–43 (2002). [CrossRef]
10. H. Gan, H. Zhang, C. T. DeRose, R. A. Norwood, N. Peyghambarian, M. Fallahi, J. Luo, B. Chen, and A. K.-Y. Jen, “Low drive voltage Fabry-Perot etalon device tunable filters using poled hybrid sol-gel materials,” Appl. Phys. Lett. 89(4), 041127 (2006). [CrossRef]
11. H. Gan, H. Zhang, C. T. DeRose, R. A. Norwood, M. Fallahi, J. Luo, A. K.-Y. Jen, B. Liu, S.-T. Ho, and N. Peyghambarian, “Hybrid Fabry-Perot etalon using an electro-optic polymer for optical modulation,” Appl. Phys. Lett. 89(14), 141113 (2006). [CrossRef]
12. H. Gan, C. Greenlee, C. Sheng, R. A. Norwood, M. Fallahi, S. Wang, W. Lin, M. Yamamoto, K. Mohanalingam, and N. Peyghambarian, “Near-resonance electro-optic activity enhancement and improved modulation performance for polymer based Fabry-Perot interferometers,” Appl. Phys. Lett. 92(20), 203302 (2008). [CrossRef]
14. C. C. Teng and H. T. Man, “Simple reflection technique for measuring the electro-optic coefficient of poled polymers,” Appl. Phys. Lett. 56(18), 1734–1736 (1990). [CrossRef]
15. C. Greenlee, A. Guilmo, A. Opadeyi, R. Himmelhuber, R. A. Norwood, M. Fallahi, J. Luo, S. Huang, X.-H. Zhou, A. K.-Y. Jen, and N. Peyghambarian, “Mach-Zehnder interferometry method for decoupling electro-optic and piezoelectric effects in poled polymer films,” Appl. Phys. Lett. 97(4), 041109 (2010). [CrossRef]
16. P.-A. Blanche, A. Bablumian, R. Voorakaranam, C. Christenson, W. Lin, T. Gu, D. Flores, P. Wang, W.-Y. Hsieh, M. Kathaperumal, B. Rachwal, O. Siddiqui, J. Thomas, R. A. Norwood, M. Yamamoto, and N. Peyghambarian, “Holographic three-dimensional telepresence using large-area photorefractive polymer,” Nature 468(7320), 80–83 (2010). [CrossRef] [PubMed]