An electro-optic device applied as an optical beam deflector and switch at different wavelengths has been built and tested. The electro-optic device is based on domain-engineered lithium niobate (LiNbO3). In this paper, for the first time, its operation has been visualized by an imaging camera. The device has been characterized both at the visible wavelength (632.8 nm) and at a typical telecom wavelength (1532 nm). Furthermore, the device has been tested as an amplitude modulator in the mid-infrared region as well, at a wavelength of ~4.3 µm, where no Pockels cells are available. A detailed description of this device is given, and the experimental results are discussed.
© 2003 Optical Society of America
In the past ten years, the growing need for optical switches, laser scanning, and high-bandwidth signal processing systems—which require fast modulation, low cost, and low power consumption—has resulted in the construction of many novel high-speed photonic devices. Electro-optical beam deflectors are good potential candidates for these applications, and in particular domain-engineered samples of ferroelectric materials, such as lithium niobate (LN), seem to satisfy all the aforementioned specifications [1–6]. In this paper, an electro-optically addressable device in a sample of LiNbO3, which can be used as either an optical deflector or an optical switch, is investigated. A similar device was originally developed by Eason and co-workers at the Optoelectronics Research Center in Southampton (UK) [5, 6].
The device presents a sharp boundary between two oppositely oriented domain regions obtained by means of an electric-field poling process. When an external electric field is applied to this boundary, equal magnitude refractive-index changes of opposite sign will occur between the adjacent domain regions. In this way, the normal electro-optic-induced refractive-index change is doubled. Thus, an optical beam crossing the interface is subjected to both reflection and refraction.
A great improvement in the angular deflection sensitivity can be obtained by use of a socalled grazing angle geometry, which is when the angle of the beam impinging on the interface approaches the total internal reflection (TIR) limit angle. In this configuration it is possible to achieve relatively large deflection angles by use of a control signal of a few hundred volts. Moreover, when the incident angle exceeds the limit angle, i.e., the TIR phenomenon onsets, the device operation is not continuous. In fact, it exhibits an abrupt deflection of the light beam. In other words, without application of voltage the light passes through the boundary practically unchanged, and only a residual poling-induced strain occurring at the boundary can slightly affect the propagation of the beam. With application of external voltage, an opposite refractive-index change will be induced to the adjacent domain regions so that if the beam light arrives from the region whose refractive index has increased and impinges on the interface at angles larger than the limit, it will be reflected with a theoretical efficiency of 100%.
The angle necessary for TIR, at an electro-optically controlled interface, is given by the usual expression:
The main equation relating the induced refractive-index change Δn, under application of a voltage V, to the electric field amplitude, is given by:
where r33 is the largest electro-optic coefficient accessed by extraordinary s-polarized light, ne is the wavelength dependent value of extraordinary refractive index, and d is the thickness of the device. The induced refractive-index change related to p-polarized light can be derived from Eq. (2) by replacing the electro-optic coefficient r33 with r13 and the extraordinary refractive index (ne ) with the ordinary refractive index (no ). Anyway, this induced refractive-index change is not particularly useful because its value is only about one-third than that obtained using r33 and ne in the above equation.
The aforesaid device provides several advantages, including an easy fabrication, with the possibility of achieving high contrast ratios, relatively low electric drive power, and finally a wavelength dependence that is much less critical than in other electro-optic devices, such as Pockels cells. In fact, a Pockels cell is characterized by the half wave voltage, i.e., Vπ=λ/(2··r33), that is directly proportional to the wavelength. In the domain-engineered device, the switching voltage, i.e. V=(Δn·d)/(·r33), is effectively wavelength independent, because intrinsic wavelength dispersion characteristics of both the electro-optic coefficient (r33 ) and refractive index (ne ) are almost negligible in a relatively small wavelength range. This feature is particular interesting for the spectral ranges where no Pockels cells are available and switch performances faster than mechanical chopping are demanded.
The presented device has been tested at visible wavelength (632.8 nm), at typical telecom wavelength (1532 nm) and, finally, in the mid-infrared region at a wavelength of 4.3 micron, where no Pockels cells are available and such electro-optic device could be of great interest.
In this paper a detailed description of the design, fabrication and testing of the electro-optic device are reported. In particular, technological steps for room-temperature electric-field poling are given in Section 2. Performance of the device working like either an optical deflector or optical switch is detailed in Section 3. Finally, further developments and cocnclusions are discussed in Section 4.
2. Technological processes
In Fig. 1 the external electrical circuit employed to reverse ferroelectric domains in LN samples is reported [7–12]. The local reversing of the sample spontaneous polarization is achieved by applying a high positive voltage slightly exceeding the LN coercive field (21.0 kV/mm). A conventional Voltage Generator (VG) drives an High Voltage Amplifier (HVA-2000x), provided by Trek, Inc., with a series current limiting resistor, RS =50Mω, in order to produce a positive voltage of 12kV. A diode rectifier was connected to the output of the HVA to prevent backswitch current flowing in the circuit.
The polarization reversal has been carried out on 500-µm-thick lithium niobate single domain crystal samples (Crystal Technology Inc). The samples were z-cut 3-inch diameter wafers with both sides polished. After solvent cleaning, the z+ face was spin coated with a 1.3-µm-thick photoresist layer (Shipley S1813-J2) and then exposed through a standard mask. The positive high voltage was applied over the z+ patterned crystal face by using liquid electrolyte (LiCl in deionized water). The liquid electrode fixture consists of two-electrolyte containing chambers which squeeze the LiNbO3 sample between O rings.
The LN conductivity at room temperature is low enough that the poling current Ipol flowing in the circuit can be readily monitored. This current was measured by recording the voltage drop across the resistance Rm=10 kω, while the poling voltage Vpol was measured using a conventional High Voltage Probe (HVP). Both the current and voltage waveforms were visualized on the oscilloscope during the poling process.
Figure 2 illustrates an optical micrograph of the interface between the two reversed ferroelectric domains. The interface has been revealed by an etching process of one hour in a HF:HNO3=1:2 acid mixture at room temperature. This acid mixture etches the z- face much faster than the z+ face. On a zoomed scale, the interface shows evident lack of homogeneities, but locally it seems to be smooth enough for the construction of our devices.
The electro-optic device was fabricated by cutting a 20 mm×15 mm large sample from the obtained domain-engineered LN wafer and by polishing its edges to have optical quality facets for light coupling. Finally, gold electrodes were sputtered on both z- and z+ faces across the domain interface region. Thus, applying an electrical field Ez over these electrodes, a refractive-index change, equal in magnitude and opposite in sign, can be achieved between the adjacent anti-parallel domain regions.
3. Device characterizations
The aforementioned technological steps were used for fabricating a device which is able to work as both beam deflector and Total Internal Reflection-based switch depending on the set-up employed. In order to characterize both the operation modes, the device has been mounted on an insulating goniometric support, which allows a fine control of the angle of incidence for the light beam (Fig. 3).
For the characterization in the visible range, experiments have been performed using a He-Ne polarized laser source at λ=632.8 nm, with an output optical power of about 1.0 mW. A half-wave plane has inserted at the output of the laser to select the required linear light polarization. Then, the laser beam has been focused onto the LN sample edge, and the output beam has been collected by means of a 560×800 pixels silicon CCD array with a pixel dimension of 10 µm. Finally, the electrical CCD output was acquired by a standard PCI frame grabber card. For the visible spectral range, the electro-optic coefficient (r33 ) and the extraordinary refractive index (ne ) are assumed to be equal to 33·10-12 m/V and 2.14, respectively .
The device has been characterized also in the infrared range. A first experiment has been performed at a telecommunication wavelength by employing a He-Ne laser source emitting at λ=1532 nm with an output power of about 5.0 mW, and replacing the CCD array with an infrared Vidicon camera. Moreover, a Glan-Thompson filter has been used instead of the half-wave plane. Additionally, also radiation at a wavelength of about 4.3 µm, generated by a nonlinear difference-frequency process in a periodically poled lithium niobate crystal , has been used to characterize the device. For this experiment an InSb liquid-N2-cooled photodiode and CaF2 lenses have been employed.
The experimental results, obtained for both operation modes, are illustrated in the following paragraphs.
3.1. Beam deflector
In Fig. 4 is shown the simple geometry for a beam deflection at the interface between the two regions for a domain engineered LN scanner, in the case of incident angle (θi) at interface smaller than the limit angle (θTIR).
We should note the values of the transmitted angle (θt) as function of the applied voltage and for different incident angles at the visible wavelength of 632.8 nm (Fig. 5).
It can be noted that, unlike the case for different kind of electro-optic scanner geometry, as the prism based one, the grazing incidence geometry is extremely sensitive to small changes of the local refractive index. Moreover, it is evident that approaching the limit angle substantial deflection can be obtained even for low control voltage amplitude. Although for small values of the incident angle (θi), as 88.0 and 88.4 degrees, the relationship between the transmitted angle and the applied voltage is approximately linear, whereas when the incident angle approaches the value of 89.0 degrees, the TIR phenomenon occurs and the relationship is non-linear. Thus, it is very advantageous to use the beam deflector with an angle of incidence approaching, but not equal to the limit angle, in order to obtain maximum deflection for a few hundreds of applied voltage. Obviously, the advantage from applying across the interface a voltage from negative to positive values is to get a doubled total deflection angle. In Fig. 6 both the simulated and measured shift of the transmitted optical beam, in the visible range, are reported, for three values of the input angle (θinp).
The sign of the displacement is related to the initial position (i.e., without any applied voltage) of the light beam; thus the minus (plus) sign reflects a separation from the initial position towards right (left). Simulated values are obtained considering the following relationship:
where d=20cm is the distance between the sample and the CCD array (see Fig. 3), and θout is evaluated taking in account the change of the refracted beam direction at the electro-optically addressed interface and refractions due both to input and output interfaces between LiNbO3 and air (see Fig. 5). So, applying the Snell’s law to the three interfaces, and bearing in mind some easy trigonometrical relationships, the output angle is given by:
where Δn is defined in Eq. (2). As is possible to observe, there is a good agreement between simulated and theoretical values of the displacement. The slight discrepancy mainly arises in the non-ideal profile of the domains interface region and in the uncertainty measure both of the absolute value of the input angle (θinp) and of distance between the device and the CCD array (d). Moreover, for the input angle of 2.0 degrees and for applied voltage larger than 300V, the output angle has not been measured, because the deflection was so large to leave the sensitive area of the CCD array. From the theoretical prediction relating to the value of the input angle of 2.0°, can be observed that for applied voltage larger than 400 V the Total Internal Reflection effect arises and the deflected beam is not more present, so the shift cannot be defined.
In Fig. 7 are shown the optical field distribution taken from the CCD camera at a distance of 20 cm far from the sample, setting an input angle of 3.3 degrees, without applied voltage and for the two extreme values of the applied voltage V=±650 V.
The resulting lateral shift is about 3.5mm for positive value of the applied voltage and 3.0 mm for the negative one. The difference can be justified observing that, when a positive voltage is applied, the deflector approaches the TIR condition, moving on the non linear region of the curves in Fig. 5, while when a reversed voltage is applied, it goes away from the more efficient non-linear region, moving towards the less efficient linear part of the characteristic. In Fig. 8 is reported a movie, obtained by successive acquisitions of the deflected beam applying a voltage from -650 V to +650 V. The shown behavior points up the possibility to employ the device as an optical scanner.
The same device has been characterized in near infra-red range with the set-up described at beginning of the paragraph, and in Fig. 9 the shift obtained placing the Vidicon camera at a distance of 27.5 cm from the device, is reported. These data, obtained with a large input angle (θinp), confirm the linear relationship between the applied voltage and the beam shift.
3.2. TIR-based switch
The same device illustrated in the previous paragraph can be used as a digital optical switch exploiting the Total Internal Reflection effect in order to obtain an abrupt deflection of the light beam from the transmitted path to the reflected one. This effect occurs when the angle of incidence (θi) is larger than the limit angle (θTIR). This angle can be evaluated from the Eq. (1) for both crystal polarizations, and its value becomes handy only for applied voltage larger than 300 V.
The schematic of the geometry of the interaction between the beam and the interface is reported Fig. 10. The schematic characterization of the switch has performed at an angle of incidence of about 89.0 degrees, determined with an error of ±0.1°.
For the visible range (λ=632.8 nm), optical reflected power for s- and p-polarization are illustrated in Fig. 11(a), as function of the applied voltage. As expected, considering the values of the respective electro-optic coefficient for each polarization orientation, the switching of s-polarized light occurs for smaller applied voltage than of p-polarized. For this polarization, the observed extinction ratio between the TIR-on and the TIR-off state is greater then 20 dB. Moreover, in Fig. 11(b) is reported the optical transmitted power for s-polarization (the more efficient configuration). Even in this case the contrast reachable in greater than 20 dB. From Fig. 11(a) and Fig. 11(b) can be noted that when TIR is approached there is not a sharp variation both of reflected and transmitted power. In fact, unless the beam is very well collimated, each portion of the beam experiences a different angle of incidence and, as consequence, the voltage required for TIR is slightly different for each portions of the beam. For this reason near TIR, a smooth transition to the reflected beam has been achieved.
In Fig. 12 is reported a movie, obtained by successive acquisitions of both the deflected and reflected beam for an s-polarized light at the visible wavelength, applying a voltage from 0 to +650 V. Though the image of the optical beam is not well defined owing to a no perfect optical alignment, it’s evident that as long as the TIR condition is not satisfied, increasing the applied voltage, both the transmitted angle and optical reflected power increase; when the TIR condition is reached, the optical transmitted power becomes null, whereas the reflected power gets to the maximum.
The lithium niobate TIR switch has been tested with a 4.3 µm optical radiation too. This radiation, generated in a periodically-poled lithium niobate crystal, is useful for spectroscopy applications . A positive sinusoidal voltage, of 500 V peak-to peak, has been applied between the upper and lower golden electrodes of the device. A knife edge stopped the beam transmitted by the internal interface between the two domains. A maximum modulation depth of 70% on the reflected beam has been measured, up to a 10 kHz frequency, limited by the bandwidth of employed high-voltage amplifier. This experiment has demonstrated the possibility to perform amplitude modulation faster than mechanical chopping, in a wavelength region where no Pockels cells are available.
4. Further developments and conclusions
The deflection and switching characteristics mentioned above could be significantly improved in an optimized version of the poled region, where a lateral deep wet etch can be used to produce a local narrowing  of the device in correspondence with the interface between the inverted domains, thus reducing the needed applied voltage. This approach is currently under development in our laboratory. Moreover, the device could be integrated in a guided-wave structure, which could be fabricated by three different methods of proton exchange, ion-beam implantation, and titanium indiffusion . These optical waveguides can be designed and realized to illuminate the active domain-engineered region of the device and to collect refracted and/or reflected light from it. This kind of design could make the device efficient and easy to couple to external optical components and circuits.
In conclusion, an electrooptic device able control a light beam spatially by means of induced deflection and/or reflection has been presented. When used as a deflector, it permits a wide angular scan range. When used as a TIR switch, the device exhibits digital behavior, with abrupt switching of light from direct to reflected path. In the visible range it has shown a good contrast ratio, greater than 20 dB. Experiments for visualizing the two operations have been conducted, and results have been discussed. Moreover, the possibility of performing amplitude modulation has also been demonstrated in the mid-infrared range. In particular, an experiment at λ=4.3 µm, where no Pockels cells are available, has been carried out.
The authors thank M. Varasi of Alenia Marconi System and A. Zani of IMM-CNR for their help in performing this research.
References and links
1. Y. Chiu, V. Gopalan, M. J. Kawas, T. E. Schlesinger, D. D. Stancil, and W. P. Risk, “Integrated optical device with second-harmonic generator, electrooptic lens, and electrooptic scanner in LiTaO3,” J. Lightwave Technol. 17, 462–465 (1999). [CrossRef]
2. K. T. Gahagan, V. Gopalan, J. M. Robinson, Q. X. Jia, T. E. Mitchell, M. J. Kawas, T. E. Schlesinger, and D. D. Stancil, “Integrated electro-optic lens/scanner in a LiTaO3 single crystal,” Appl. Opt. 38, 1186–1190 (1999). [CrossRef]
3. M. Yamada and M. Saitoh, “Electric-field induced cylindrical lens, switching and deflection devices composed of the inverted domains in LiNbO3 crystal,” Appl. Phys. Lett. 69, 3659–3661 (1996). [CrossRef]
4. H. Gnewuch, C. N. Pannell, G. W. Ross, P. G. R. Smith, and H. Geiger, “Nanosecond response of Bragg deflectors in periodically poled LiNbO3,” Phot. Technol. Lett. 10, 1730–1732 (1998). [CrossRef]
5. A. J. Boyland, S. Mailis, J. M. Hendricks, P. G. R. Smith, and R. W. Eason, “Electro-optically controlled beam switching via total internal reflection at a domain-engineered interface in LiNbO3,” Opt. Commun. 197, 193–200 (2001). [CrossRef]
6. R. W. Eason, A. J. Boyland, S. Mailis, and P. G. R. Smith, “Electro-optically controlled beam deflection for grazing incidence geometry on a domain-engineered interface in LiNb O3,” Opt. Commun. 197, 201–207 (2001). [CrossRef]
7. M. Yamada, N. Nada, M. Saitoh, and K. Watanabe, “First-order quasi-phase matched LiNbO3 waveguide periodically poled by applying an external field for efficient blue second-harmonic generation,” Appl. Phys. Lett. 62, 435–436 (1993). [CrossRef]
8. J. Webjörn, V. Pruneri, P. St. J. Russell, J. R. M. Barr, and D. C. Hanna, “Quasi-phase-matched blue light generation in bulk lithium niobate, electrically poled via periodic liquid electrodes,” Electron. Lett. 30, 894–895 (1994). [CrossRef]
9. V. Pruneri, J. Webjörn, P. St. J. Russell, J. R. M. Barr, and D. C. Hanna, “Intracavity second harmonic generation of 0.532µm in bulk periodically poled lithium niobate,” Opt. Commun. 116, 159–162 (1995). [CrossRef]
10. G. D. Miller, R. G. Batchko, M. M. Fejer, and R. L. Byer, “Visible quasi-phase-matched harmonic generation by electric-field-poled lithium niobate,” SPIE 2700, 34–36 (1996). [CrossRef]
11. S. Grilli, S. De Nicola, P. Ferraro, A. Finizio, P. De Natale, M. Iodice, and G. Pierattini, “Investigation on overpoled lithium niobate patterned crystal,” in ICO XIX, 19th Congress of the International Commission for Optics, Technical Digest, Italy, 25–31 August 2002, Part 2, pp. 735–736.
12. S. Grilli, S. De Nicola, P. Ferraro, A. Finizio, P. De Natale, G. Pierattini, and M. Chiarini, “Investigation on poling of lithium niobate patterned by interference lithography,” in Integrated Optical Devices: Fabrication and Testing, Proc. SPIE4944 (2002).
13. A. M. Prokhorov and Y. S. Kuzminov, Physics and Chemistry of Crystalline Lithium Niobate (Hilger, Bristol, UK, 1990).
14. D. Mazzotti, P. De Natale, G. Giusfredi, C. Fort, J. A. Mitchell, and L. W. Hollberg, “Difference-frequency generation in PPLN at 4.25 µm: an analysis of sensitivity limits for DFG spectrometers,” Appl. Phys. B 70, 747–750 (2000). [CrossRef]
15. D. Mazzotti, S. Borri, P. Cancio, G. Giusfredi, and P. De Natale, “Low-power Lamb-dip spectroscopy of very weak CO2 transitions near 4.25 µm,” Opt. Lett. 27, 1256–1258 (2002). [CrossRef]
16. M. Reich, F. Korte, C. Fallnich, H. Welling, and A. Tunnermann “Electrode geometries for periodic poling of ferroelectric materials,” Opt. Lett. 23, 1817–1819 (1998). [CrossRef]
17. I. E. Barry, G. W. Ross, P. G. R. Smith, and R. W. Eason, “Ridge waveguides in lithium niobate fabricated by differential etching following spatially selective domain inversion,” Appl. Phys. Lett. 74, 1487–1488 (1999). [CrossRef]