We report on the use of thin, i.e. 10µm-thick, single-crystal LiNbO3, in low-voltage electrooptic prism scanners. These devices are fabricated by electric-field poling of a series of electrooptic prisms in a bulk crystal followed by high-energy ion implantation and subsequent etching of the poled samples. Such a single-crystal thin-film scanner, while having the same scanning functionality as with a bulk device, has an order-of-magnitude reduction in its required voltage; for example, a series of two prisms, of 2mm in total length, yields a deflection angle of 0.7° at 100V compared to more than 1.7kV for the same device in standard 200µm-thick LiNbO3 wafers.
©2004 Optical Society of America
Devices that provide continuous optical-beam deflection are essential components for display and printing technologies, optical data-storage systems, and beam scanners. It was recognized early that the Pockel’s effect in linear electrooptic materials could be used to control electrically the optical-beam deflection angle in a cascaded prism-scanning device [1,2]. The application of an electric field to an electrooptic-crystal prism changes the propagation direction of an optical beam passing through the prism. The fact that the electrooptic coefficient is relatively small has been a major impediment in using this approach, since it requires very high voltage to achieve practical deflection in a single-prism device. Cascading several crystal prisms is thus commonly used to provide an appreciable total deflection at somewhat more acceptable voltages, with the operating voltage decreasing approximately linearly with the number of prisms. Scanners can be made more compact by integration, that is, by patterning prism- or prism-cascade-shaped poled regions or domains in a ferroelectric slab. This process has been applied to z-cut wafers of LiTaO3 , allowing devices of millimeter prism length. Further advances have resulted in larger deflection angles by optimizing the scanner geometry [4–6]. However, despite these improvements, electrooptic scanner devices still require voltages in the kilovolt range.
In order to fabricate beam-scanning devices which are useful in integrated-optics applications, it is essential to reduce the driving voltage. One possible solution would be to use a thin-film substrate material, since the required scanning voltage scales proportionally with the sample thickness. Such a thin-film device would allow the use of low-power drivers, achieve higher scanning frequency and also make heterointegration of the scanning device, onto a separate optical platform, feasible.
In this paper we demonstrate that thin-films of single-crystal LiNbO3, formed via crystalion-slicing (CIS)  can be used to fabricate low-voltage, thin-film electrooptic scanners. This technique involves the implantation of high-energy He+ ions into a bulk crystal substrate and subsequent wet etching of the sacrificial layer formed in the heavily implanted subsurface region. Using CIS on a bulk prism-scanning device yields a much thinner crystal slab than that used in previously reported electrooptic scanner devices [5, 6], while preserving the existing ferroelectric domain structure [8, 9]. The intent of this paper is to demonstrate the low-voltage operation of CIS thin-film beam deflectors and to show that the calculated performance per prism can be realized in our exfoliated patterned film.
where Δn is the index difference between the prisms and the surrounding medium. If the poled deflectors are made in z-cut LiNbO3, then the index difference, Δn, in a single-domain prism for TM polarized light is given by:
where ne is the extraordinary index of refraction of LiNbO3, r 33 is the corresponding electrooptic coefficient and V is the voltage applied across the sample thickness d. The choice of TM-polarized light enables use of r 33, the largest electrooptic coefficient of LiNbO3. Equations (1) and (2) show that the beam deflection is linearly proportional to the applied voltage and that the maximum scan angle is limited by the width, W, of the output prism. Our study was oriented toward the use of a sequence of prisms arranged in size and position so that they would form an optimized, horn-shaped scanner pattern . This design limits the width of each prism so that it can just accommodate the beam at maximum deflection and leads to an increase in prism widths as the beam propagates through the device.
Single-prism deflectors, as well as a simple series of two prisms, were investigated to demonstrate the fact that scaling in voltage and in prism number occurred as predicted from extrapolation from the behavior of the bulk device. For example, Fig. 1 shows a 2D beam propagation method (BPM) simulation result for a 2mm-length scanner composed of two prisms with widths ranging from 420µm to 490µm. The simulation was performed for TM polarized light with λ=632.8nm and the index change for an ±200V applied voltage across a LiNbO 3 film with d=10µm. For the given parameters, BPM calculations predict a ±1.6° external deflection angle. The difference between the fundamental TM mode index, i.e. the effective index, and material index was negligible due to our comparatively large LiNbO 3 slab thickness, and was ignored in these calculations. Note that the value of electrooptic coefficient r 33 used in the BPM simulation is the same as the one for bulk LiNbO3 (r 33=31pm/V). Experimental data, presented later, as well as earlier studies  demonstrate that CIS thin films retained the bulk optical properties of LiNbO3 parent wafer.
The devices were fabricated on a 0.5mm-thick z-cut LiNbO3 wafer by electric-field poling. The poling electrode was made by thermal evaporation of a 50nm Cr film onto the +z surface. A direct-laser-writing system was used to create a photomask with an optimized expanding or “horn-shaped” scanner pattern with a 10mm length. This pattern was then transferred to the poling electrode by optical lithography and wet etching. The poling electrode was covered with a thick insulating layer of photoresist, which had photolithographically defined openings for contacts. The bulk LiNbO3 sample prepared in this way was sandwiched between two rubber rings containing liquid electrolyte and connected to a high-voltage power supply. A controlled, high-voltage pulse (11kV) was applied to the crystal causing domain reversal in the area covered by the poling electrode and terminated after the poling current dropped to zero.
After poling, the insulating layer and poling electrode were removed and the samples were implanted with a 5×1016 cm-2 dose of 3.8MeV He+ ions. At this implantation energy, the calculated ion range using a transport of ions in matter code, TRIM, was ~10µm. The locally stressed, heavily implanted region, containing interstitial He, acts as a sacrificial layer during the etching step. Before liftoff, rapid thermal annealing (RTA) at 350 °C for 30s was performed to enhance the etch selectivity between the sacrificial layer and the rest of the crystal . Samples were then cut into smaller pieces containing one and two prisms, and the end facets were optically polished to enable low-loss light input and output coupling. Note that because of the smaller sample size for the one-prism device (~1mm), the end facet polishing was somewhat difficult; however, a satisfactory region was obtained in the center of the facet. The implanted surface was encapsulated with HF-resistant wax and then exfoliated by etching in a 5%-HF solution. After liftoff, thin-film devices were annealed at 550°C for 6h in an Ar-O2 atmosphere to restore the material bulk optical properties . Figure 2 shows samples with one and two electrooptic prisms prior to etching, and the same thin films obtained after crystal ion slicing. The triangular ferroelectric domains created by poling were examined using Nomarski microscopy. The prism length was 1mm in all cases, and the width was 540µm for the sample shown in Fig. 2(a) and Fig. 2(b) and 356µm and 415µm for the sample in Fig. 2(c) and Fig. 2(d). Note that, in these photographs, the surface quality of CIS films appears different than that of the parent samples due to different domain contrast in bulk vs. thin-film case. In particular, the greater contrast in the bulk sample results from intentionally over-etching the sample back-side in order to verify the complete domain reversal throughout the entire wafer thickness; the inverted domains were then more visible. On the other hand, in the thin-film case image contrast needed to be greatly increased to see the domains, and thus stray surface features such as scratches due to surface polishing, are more apparent in the micrograph.
Each of the thin-film scanners fabricated this way was then placed between two doped oxidized silicon wafer samples with the oxidized surfaces facing the film; these samples served as electrodes for the device. The 0.5µmthermally grown oxide provided optical buffer layers. This approach allowed preparation of a temporary structure, henceforth referred to as freestanding thin-film scanner, for rapid optical testing.
In order to facilitate handling of the CIS film, a permanent device structure was also devised. Another sample of approximately 2mm in length, and containing two electrooptic prisms, was prepared for testing using the following sequence. Prior to liftoff, a silicon-dioxide buffer layer was deposited onto the LiNbO3 implanted surface by RF sputtering, followed by thermal evaporation of a Cr/Au electrode. Sample-surface protection, slicing and post-liftoff annealing were then performed as described earlier. The sliced device was placed face down onto a metalized LiNbO3 support sample so that the electrode-covered film surface contacted the metalized substrate. This provided electrical contact for the bottom electrode. The film was then secured to the substrate by applying a nonconductive-epoxy layer around the film edges. This layer also provided insulation between top and bottom electrodes of the CIS scanning device. A silicon-dioxide buffer layer was then deposited onto the negative-z surface of the sliced film and packaging was completed by bonding another piece of bulk LiNbO 3 onto the CIS film/bottom substrate structure using silver epoxy, which also served as the top scanner electrode, see Fig.3. This “package” was found to provide sufficient mechanical and chemical protection for the CIS thin-film device during subsequent fabrication steps and device manipulation. Sample preparation was finalized by polishing the input and output facets to an optical finish.
3. Results and discussion
Freestanding and packaged scanner segments were tested using a He-Ne laser. Laser light was polarized along the z-axis (TM) using a linear polarizer, and coupled into the scanner segment using two cylindrical lenses. The beam was focused to a spot size of 100µm in the thin-film plane, and, thus, remained well collimated throughout the entire device length (~2mm). A second cylindrical lens provided a 8µm spot size in the transverse direction for coupling into the thin-film planar waveguide. Care was taken during vertical alignment to suppress coupling into higher-order TM modes. At the output of the scanner, light diverging in the transverse direction was collimated using another cylindrical lens and imaged with a camera. Laser-spot positions at the output plane were recorded at distances from 5cm to 15cm from the scanner output face and transformed into values of external deflection angles.
Figure 4(a) shows the results for two free-standing scanner segments containing one and two prisms. In the case of single-prism device, experimental results agree well with the theory for voltages up to ~70V, showing the scanning sensitivity of ~0.03 mrad/V. For higher voltages, the deflected beam impinged on the imperfect facet regions (see above) and thus resulted in less accurate determination of the external deflection angle. In the case of the two-prism deflector, scanning sensitivity of 0.088mrad/V was measured, which is within experimental error of the theoretically predicted value of 0.095mrad/V. Figure 4(b) shows the experimental results and BPM simulation data for the packaged two-prism scanner device, depicted schematically in Fig. 3. The fabricated thin film scanner had a length of 2mm, with prism-widths of 420µm and 490µm. Taking into account the total buffer-oxide thickness of around 0.2µm, the theoretical expectation for the device scanning sensitivity was 0.127mrad/V, a value in excellent agreement with the measured sensitivity of 0.125mrad/V.
Our measurements lead to further observations regarding the thin-film many-prism-element (typically ~10) scanning device. First, as in a bulk electrooptic scanner, the maximum scan voltage is limited by the value of electric field in the CIS film, which must be kept below the coercive field value (21kV/mm) in order to preserve the poled domains. For example, in our device a driving voltage of around 200V would lead to an internal field which is 20% below the coercive field, making the maximum deflection for this two prism geometry to be ~25mrad (1.4°); for the case of 10 optimized prisms  with the input prism width of W 0=150µm this would correspond to ~9°. Second, the number of resolvable spots, which is an important figure of merit for characterizing a scanner performance, remains unchanged in the thin-film device as compared with bulk devices of the same geometry. This value can be estimated as the ratio of maximum deflection angle and the beam divergence angle . For our device we calculate ~7 resolvable spots for the voltage range of ±100V. Finally, although the optical loss measurements were not conducted at this time, previous work  shows that CIS films of LiNbO 3 retain bulk linear-optical properties, which permits the fabrication of low-loss optical devices.
We have demonstrated that a low-voltage electrooptic scanner can be made in LiNbO 3 using thin ion-sliced material of 10µm thickness. The present study demonstrates voltages scaling with crystal thickness and with prism number. Our results enable an order-of-magnitude reduction in driving voltage compared to conventional bulk devices of the same material and geometry, and show promise for achieving low-power, high speed, heterointegrable electrooptic beam deflectors. For example, using the parameters shown here a device of 2mm in length would enable a scanning angle of ±0.7° with an applied voltage of ±100V. Further, other electrooptic devices using transverse electrode geometries could also benefit from the driving voltage reduction obtainable using crystal-ion-slicing technology in LiNbO 3 and related materials.
The authors would like to thank Dr. Antonije Radojevic and Dr. Tomoyuki Izuhara for many helpful discussions, and Dr. Antonio Sanchez for his especially valuable suggestions and continuing interest in this project. We acknowledge generous financial support from the Brown Univ. DARPA Optocenter (BROWNU-1119-24596) and from AFOSR (F49620-99-1-0038).
References and links
1. J. L. Lotspeich, “Electrooptic light-beam deflection,” IEEE Spectr. 5, 45–52 (1968). [CrossRef]
2. T. C. Lee and J. D. Zook, “Light beam deflection with electrooptic prisms,” IEEE J. Quantum Electron. QE-4, 442–454 (1968). [CrossRef]
3. Q. Chen, Y. Chiu, D. N. Lambeth, T. E. Schlesinger, and D. D. Stancil, “Guided-wave electro-optic beam deflector using domain reversal in LiTaO3,” J. Lightwave Technol. 12, 1401–1404 (1994). [CrossRef]
4. Y. Chiu, J. Zou, D. D. Stancil, and T. E. Schlesinger, “Shape-optimized electrooptic beam scanners: Analysis, design, and simulation,” J. Lightwave Technol. 17, 108–114 (1999). [CrossRef]
5. J. C. Fang, M. J. Kawas, J. Zou, V. Gopalan, T. E. Schlesinger, and D. D. Stancil, “Shape-optimized electrooptic beam scanners: Experiment,” IEEE Photon. Technol. Lett. 11, 66–68 (1999). [CrossRef]
6. D. A. Scrymegeour, A. Sharan, V. Gopalan, K. T. Gahagan, J. L. Casson, R. Sander, J. M. Robinson, F. Muhammad, P. Chandramani, and F. Kiamilev, “Cascaded electro-optic scanning of laser light over large angles using domain microengineered ferroelectrics,” Appl. Phys. Lett. 81, 3140–3142 (2002). [CrossRef]
7. M. Levy, R. M. Osgood Jr., R. Liu, L. E. Cross, G. S. Cargill, A. Kumar, and H. Bakhru, “Fabrication of single-crystal lithium niobate films by crystal ion slicing,” Appl. Phys. Lett. 73, 2293–2295 (1998). [CrossRef]
8. A. M. Radojevic, M. Levy, R. M. Osgood Jr., D. H. Jundt, and H. Bakhru, “Second-order optical nonlinearity of 10µm-thick periodically poled lithium niobate films,” Opt. Lett. 25, 1034–1036 (2000). [CrossRef]
9. D. A. Scrymegeour, V. Gopalan, and T. E. Haynes, “Crystal ion slicing of domain microengineered electro-optic devices on lithium niobate,” Integr. Ferroelectr. 41, 35–42 (2001). [CrossRef]
10. T. A. Ramadan, M. Levy, and R. M. Osgood Jr., “Electro-optic modulation in crystal-ion-sliced z-cut LiNbO3 thin films,” Appl. Phys. Lett. 76, 1407–1409 (2000). [CrossRef]
11. A. M. Radojevic, M. Levy, R. M. Osgood, A. Kumar, H. Bakhru, C. Tian, and C. Evans, “Large etch-selectivity enhancement in the epitaxial liftoff of single-crystal LiNbO3 films,” Appl. Phys. Lett. 74, 3197–3199 (1999). [CrossRef]