Open Access
Add to BibSonomyAbstract
We propose a novel electric field poling technique for the fabrication of nonlinear photonic crystals in congruent LiNbO3 substrates, based on a hybrid bi-dimensional mask, which combines periodic proton-exchange and electrode patterns. With it we demonstrate rectangular bulk lattices with a periodicity of 8 µm x 6.78 µm in 500 µm-thick substrates.
©2011 Optical Society of America
1. Introduction
The reversible polarization of ferroelectric materials is at the heart of their widespread use in electronics [1] and photonics [2], for devices ranging from random access memories and high density storage media [3], to nonlinear optical frequency converters. The field of nonlinear optics has particularly benefited over the past years from the development of reliable technologies to engineer ferroelectric gratings by electric field poling techniques [4], providing effective means to implement the idea of Quasi-Phase-Matching (QPM), originally proposed by Armstrong et al. in 1962 [5], in optical materials such as LiNbO3 [6], LiTaO3 [7] and KTP [8].
In more recent years, the extension of electric field poling techniques to two-dimensional lattices [9] has enabled the demonstration of purely nonlinear photonic crystals (NPC) [10] and quasi-crystals [11]. Furthermore, the new degrees of freedom affordable through domain engineering in 2D have led to variety of novel nonlinear optical devices, such as multiple-beam frequency converters [12], tunable soliton switches [13] and Airy beam generators [14].
Most of such devices have been implemented in periodically poled LiNbO3 (PPLN) and LiTaO3 (PPLT). PPLN is particularly appealing for its high nonlinear coefficients, the proven scalability of its poling process to wafer sizes and the maturity of the waveguide technology developed for congruent substrates, already exploited for integrated NPCs [15]. Yet several challenges still remain to be faced in the fabrication of advanced NPC structures [16], involving stringent control over domain sizes and complex two-dimensional (2D) topologies.
The conventional approach to fabricating NPCs consists in a direct generalization of the standard 1D electric field poling (EFP) technique based on photoresist (insulator) patterning [6], as illustrated in Fig. 1 . The domain topology in the x-y plane of a z-cut CLN substrate is defined by patterning in 1D (Fig. 1a) or 2D (Fig. 1b) a photoresist layer on one of the z-faces. An electric field (Ez) exceeding the coercive value (Ec~21kV/mm) is then selectively applied between electrical contacts made in the openings of the photoresist on one face of the crystal and an uniform electrode on the other. The inhomogeneous (x-y) field distribution generated by the patterned electrodes close to the CLN surface induces the polarization switching in the areas where Ez(x,y) > Ec. In Figs. 1c and 1d, we illustrate the electrostatic distribution of Ez(x,y) for the case of 1D and 2D electrodes, respectively, calculated for typical photoresist on CLN.
Fig. 1 Conventional electric field poling of z-cut CLN crystals with photoresist insulator patterns. Insulating mask geometries for the: (a) 1D and (b) 2D case. Calculated in-plane (x-y) distributions of the polar component (Ez) of the electrostatic field close to the patterned surface (z = 500nm) for: (c) 1D and (d) 2D patterns with a period Λ = 10 μm. Simulations done with a commercial solver of the Poisson equation (Comsol Multiphysics@), for an external field of 21 kV/mm applied to 0.5 mm-thick CLN (εLN = 34), with a 1.8 μm-thick photoresist layer (εpr = 3).
In analogy to the 1D case, the main technological difficulties encountered for domain engineering in 2D concern avoiding domain merging at short-periods. In standard EFP configurations, part of the problem arises from the fringing fields at the edges of the photoresist [17], apparent in the field plots of Figs. 1c and 1d. In order to overcome such limitations, novel EFP techniques employing controlled domain back-switching [18] or substrate chemical patterning [19,20] have recently been devised for short-period (<10μm) poling of 0.5mm-thick CLN substrates.
Additional constraints affecting the poling in 2D geometries (Fig. 1b) stem from the crystal symmetry, which naturally favors hexagonal lattice topologies, making it significantly more complicated to fabricate e.g., rectangular lattices with comparable periods in the two orthogonal crystal directions (x-y). Specifically, due to the faster growth of CLN domains along the y crystallographic axis with respect to the x-axis [6], it proves more challenging to reduce the poling periods in the former than in the latter direction. The finest-pitch 2D bulk domain structures in CLN to date have been demonstrated by Peng et al. [21]. With a chemical patterning technique, they obtained rectangular domain arrays with periodicities of 6.6 µm and 13.6 µm along the x and y directions, respectively (implying a period along y which is still twice the one along x).
Here we present a novel technique suitable for the fabrication of 2D bulk PPLN structures, which relies on a hybrid 2D poling mask, obtained as the combination of 1D periodic chemical patterning of the substrate (via proton-exchange) and 1D periodic electrodes deposited on its surface (gel contacts through photoresist openings, as in Fig. 1a). With this technique we successfully fabricated 2D ferroelectric rectangular lattices with periodicities of 8 x 6.78 µm2 (along x and y, respectively) in 0.5 mm-thick CLN, representing, to the best of our knowledge, the densest 2D PPLN bulk structures achieved to date.
2. The hybrid mask
We performed our electric-field poling experiments on commercially available, 0.5 mm-thick z-cut congruent LiNbO3 substrates (Castech Inc.). The poling masks consisted of rectangular 2D lattices, with periods of Λx = 8 μm and Λy = 6.78 μm along the x and y crystallographic directions, as depicted in Fig. 2 . The rectangular 2D mask patterns were a hybrid combination of two orthogonal 1D gratings, made by periodic PE and periodic surface electrodes, respectively. As illustrated in Fig. 2a, the PE grating lines were aligned with the y axis, while the electrodes were parallel to x.
Fig. 2 EFP of CLN with a 2D hybrid mask. (a) Sketch of the mask geometry in 3D (blue stripes = PE regions, red stripes = photoresist). (b) top view of the mask, highlighting its elementary cell. (c) calculated in-plane (x-y) distributions of the polar component (Ez) of the electrostatic field at a depth z = 2.3 μm beneath the patterned surface. Electrostatic simulations under the same conditions as for Fig. 1 Eext = 21 kV/mm, CLN (εLN = 34) and insulator (εpr = 3) thicknesses of 500 μm and 1.8 μm, respectively.
The hybrid 2D mask was fabricated in two steps. First, we selectively proton-exchanged the substrates through the openings of periodic ~100nm-thick Titanium stripes (patterned by standard photolithography and reactive ion etching). The 1D Ti gratings had a periodicity Λx = 8 μm and a duty cycle (stripe width over grating period) of 70%. A uniform thin layer of Ti was additionally evaporated on the opposite (unpatterned) side of the crystals to prevent PE. The samples were then exchanged for 24 hours at 200 °C in pure benzoic acid. This resulted in PE surface gratings extending to a (measured) depth d PE~2.3 μm, with a duty cycle of 50% (exceeding the 20% Ti-mask openings) due to the lateral diffusion of protons along x, underneath the Ti stripes [22]. After PE, the Ti mask layers were removed by wet-etching, leaving a surface chemical pattern in the crystals as illustrated in Fig. 2a (blue stripes = PE regions).
The second patterning step consisted in depositing periodical electrodes on the substrate, orthogonally to the chemical grating. This was done by patterning 1.8 µm-thick photoresist (insulating) stripes with a period of Λy = 6.78 μm, a duty cycle of 50% (at the top) and a trapezoidal (~80° wall slope) cross-section (red stripes in Fig. 2a). As in conventional poling (Fig. 1), the openings of the photoresist were then filled with an electrolyte to achieve a periodic electrical contact at the sample surface.
In the unitary cell of the final 2D hybrid pattern, four different areas can be identified, corresponding to: bare CLN (grey), PE-CLN (blue), photoresist-covered CLN (red) and photoresist-covered PE-CLN (violet) regions, respectively, as highlighted in Fig. 2b.
In Fig. 2c we also plot electrostatic calculations of the spatial distribution in the x-y plane of the polar field (Ez) in the crystal. The latter results from the superposition of the internal fields associated to the periodic proton-exchange [19] and of the external field applied via the patterned electrodes. It is worth noticing how the field patterning due to PE mitigates the edge effects of the external electrodes in comparison to the case of Fig. 1d, yielding a smoother 2D field profile in the crystal. This significantly limits lateral domain broadening during the poling, as previously demonstrated for the 1D case [19].
In order to evaluate also the effect of the substrate polarity on the poling, the hybrid patterning of Fig. 2 was fabricated on multiple samples, either on the + z or on the –z face.
3. The poling experiments
Samples patterned with the 2D hybrid mask described above were poled with a standard EFP technique, using gel electrodes to contact the crystals. We employed high voltage pulses of the type of ref [6], with poling plateaux of durations Δτpol = 50 ÷ 100 ms, applied fields E ext ~22 kV/mm and voltage ramp-down times greater than 100 ms.
After the poling, the samples were etched in a solution of 40% hydrofluoric acid and water for 60 min. The differential etching rates of domains of opposite polarity and of PE versus non-PE regions [23], allowed the simultaneous visualisation of the chemical mask and of the final domain distributions.
Substantially different results were obtained for patterning on –z and + z, as illustrated by Fig. 3 and Fig. 5 , respectively. In what follows, regardless of the original substrate polarity, we will simply refer to the patterned face of the crystals as the ‘top’ side and to the unpatterned face as the ‘bottom’ side.
Fig. 3 Results of EFP with a hybrid mask on -z. Top views of the patterns, revealed after the poling by a wet-etch in an HF:H2O solution: (a) top (patterned) surface, originally -z and (b) bottom (upatterned) surface, originally + z.
Fig. 5 Results of EFP with a hybrid mask on + z. Top views of the patterns, revealed after the poling by a wet-etch in an HF:H2O solution: (a) top (patterned) surface, originally + z and (b) bottom (upatterned) surface, originally -z.
When the masks were made on −z, the double patterning due to periodic PE in one direction and periodic contacts through the photoresist lines in the other resulted in 2D bulk domain lattices with a rectangular topology, as shown in Fig. 3, a result which intuitively agrees with the expected field distributions based on the simple electrostatic model discussed in the previous section (Fig. 2c). Figures 3a and 3b provides more detailed views of the structures observed after the etching on the top and bottom faces of the samples, respectively. For a comparison, in Fig. 3 we have also sketched the original mask geometry, highlighting in blue the (vertical) PE regions and in red the (horizontal) photoresist stripes.
The PE regions can also be clearly recognised on the top face, as the darker areas in Fig. 3a. On the other hand, as discussed in [19], the actual ferroelectric domain patterns are best identified by the images taken on the bottom face, where no PE layer is present. From Fig. 3b it is apparent how the hybrid mask on –z results in regular 2D domain arrays, which, even on the backside, follow well the periodicities of the hybrid mask created on the top. A comparison between Figs. 3a and 3b illustrates also how the individual domain shapes evolve from rectangles on the top surface to hexagons after propagation through the bulk, well reflecting the symmetry of the CLN crystal. Additional sub-micrometric structures, preferentially aligned along the crystallographic y axis, can also be distinguished within the switched hexagons on the bottom face (Fig. 3b). We are currently further investigating their nature. Due to their resemblance with structures reported elsewhere [24], we suspect these features to be surface nanodomains, possibly originating from back-switching preferentially occurring at the + y corners of the poled hexagons.
The dimension of the 2D poled array is ~4 mm x 1 mm. The hexagons on the bottom face are ~4 µm x 4.6 µm (along x and y), corresponding to aspect ratios (domain width / depth) of 125 and 250, respectively. The inverted area related to each hexagonal domain is ~13.84 µm2. Along y (where short pitch poling with a photoresist mask is normally extremely challenging, because of the higher domain propagation speed) the inverted domains lie ~2 µm apart without merging. To the best of our knowledge, this represents the shortest period achieved along y in 2D bulk PPLN.
Second harmonic generation (SHG) measurements, made on these samples at higher order QPM with a tuneable continuous-wave Ti:sapphire laser source, confirmed the microscopic investigations. The 2D lattice results in multiple in-plane SHG resonances, as illustrated by the SHG image of Fig. 4a , showing a picture of the blue output from the PPLN in the far field, recorded at λp = 820.97 nm. The three blue spots in Fig. 4a correspond to SH beams emerging at angles of ± 3.46° and 0°, generated by QPM via the reciprocal lattice vectors G1, ± 1 and G01 (collinear) of the 2D lattice, respectively. The spectral and angular positions of the SHG resonances agree well with theoretical predictions based on Sellmeier equations for LiNbO3 [25]. In Fig. 4b we show also the calculated SHG tuning curve (magenta line) of an ideal (4mm-long) grating for 5th order QPM via G01 and compare it with the corresponding experimental data (blue dots), measured on the central lobe of Fig. 4a at temperature of 178°C.
Fig. 4 Optical characterization of the 2D PPLN sample by means of SHG. a) SH beams emerging at ± 3.46° and 0° ; b) the ideal tuning curve (magenta line) calculated for the central SHG peak (5th order QPM with G01) in a 4mm-long grating and the measured ones: blue dots = SHG tuning curve in the middle of the sample – black stars = SHG close to the patterned surface
The experimental full-width at half maximum is Δλ = 2 nm, somewhat larger than the theoretical one (Δλ = 1.4 nm), but this could also be attributed to the limited resolution we could achieve in tuning the pump wavelength. The SHG measurements indicated a good quality of the 2D PPLN pattern throughout the crystal thickness, with the only exception of a shallow layer close to the patterned face, where we recorded a ~53% reduction of the peak conversion efficiency (cf dark curve with black markers in Fig. 4b), presumably due to scattering effects and surface perturbations of the domain pattern induced by the periodic PE, similarly to what seen in 1D PE: PPLN [19].
Substantially different results were obtained in the poling experiments performed, under the same conditions, on samples patterned with the 2D hybrid mask on + z, as illustrated by the images in Fig. 5 (next page). Etching of the bottom faces (−z, Fig. 5b), revealed regular 1D PPLN domain structures, which followed the chemical but not the electrode patterns. Optical SHG measurements confirmed that also in this case the structures seen on the bottom face corresponded to the actual bulk domain distribution, extending through the sample thickness. The results obtained for patterning on + z indicate that the sample polarity plays also a major role in determining the balance between the actions of the PE and of the external electrodes used in the poling.
It is this balance/imbalance which ultimately determines whether the final bulk domain distribution will result in a 2D or 1D pattern. In light of previous investigations on the effect of PE on the coercive field of CLN [26], the poling selectivity associated to PE should be stronger on + z than on -z and this would be consistent with our experimental observations. Furthermore, our studies on 1D poling with photoresist masks of the same type (Fig. 1a) and gel electrodes (as the ones used in the 2D experiments) gave better results for patterning on -z. In light of this, it might see reasonable that for the case on Fig. 5 (mask on + z), the enhanced action of the chemical patterning proves to be predominant over that of the photoresist, yielding a complete switching of the non-PE regions in the bulk, regardless of whether covered or not with insulator. A closer investigations of the domain patterns on the top surfaces of our samples (Fig. 5a) reveals also the presence of fine structures between the PE stripes (dark horizontal lines in Fig. 5a), which most likely are residual unswitched surface nano-domains located under the resist.
4. Conclusions
We have presented a novel poling technique suitable for the fabrication of 2D NPCs in CLN, combining two 1D masks, made by combination of periodic proton exchange and photoresist patterning. With those we demonstrated rectangular domain lattices with periodicities of 8 µm x 6.78 µm in 0.5 mm-thick z-cut CLN substrates, which represent to the best of our knowledge the densest 2D NPCs made in bulk CLN.
The proposed hybrid poling technology overcomes some of the constraints imposed by the LN crystallographic structure in standard electric field poling and can in principle allow even shorter-period bulk domain patterning, currently under investigation. Furthermore, the experimental results provide a proof-of-principle for enhanced possibilities in tailoring the 3D distributions of the electric field at the sample surfaces, by suitably weighting the contributions arising from the chemical patterning inside the crystal with those created externally by non conventional in-plane electrode geometries. The additional degrees of freedom associated to the independent engineering of the internal and external poling fields holds promise for enabling higher-resolution sophisticated 2D domain engineering suitable for the implementation of a variety of novel nonlinear photonic crystals and quasi-crystals.
Further improvements of this technology would involve a 2D PE mask. Numerical simulations of the field distributions suggest that this configuration might be the most promising to attain even denser patterning in congruent lithium niobate by further weakening 2D later domain broadening.
Acknowledgments
This work was supported by the Swedish Scientific Research Council (Vetenskapsrådet, VR 621-2008-3601) and the Linné centre for Advanced Optics and Photonics (ADOPT). Katia Gallo gratefully acknowledges support from the EU and Vetenskapsrådet through Marie Curie (PIEF-GA-2009-234798) and Rådforskare (622-2010-526) fellowships.
References and links
1. J. F. Scott, Ferroelectric Memories (Springer, 2000).
2. P. Ferraro, S. Grilli, and P. De Natale, eds., Ferroelectric Crystals for Photonic Applications, Vol. 91 of Springer Material Science Series (Springer, 2008), pp. 229–250.
3. K. Tanaka and Y. Cho, “Actual information storage with a recording density of 4 Tbit/in2 in a ferroelectric recording medium,” Appl. Phys. Lett. 97(9), 092901 (2010). [CrossRef] [PubMed]
4. 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(5), 435–436 (1993). [CrossRef]
5. J. A. Armstrong, N. Bloembergen, J. Ducuing, and P. S. Pershan, “Interactions between light waves in a nonlinear dielectric,” Phys. Rev. 127(6), 1918–1939 (1962). [CrossRef]
6. L. E. Myers, R. C. Eckardt, M. M. Fejer, R. L. Byer, W. R. Bosenberg, and J. W. Pierce, “Quasi-phase-matched optical parametric oscillator in periodically poled LiNbO3,” J. Opt. Soc. Am. B 12(11), 2102–2116 (1995). [CrossRef]
7. H. Ishizuki and T. Taira, “High energy quasi-phase matched optical parametric oscillation using Mg-doped congruent LiTaO3 crystals,” Opt. Express 18(1), 253–258 (2010). [CrossRef]
8. H. Karlsson and F. Laurell, “Electric field poling of flux grown KTiOPO4,” Appl. Phys. Lett. 71(24), 3474–3476 (1997). [CrossRef]
9. N. G. R. Broderick, G. W. Ross, H. L. Offerhaus, D. J. Richardson, and D. C. Hanna, “Hexagonally poled lithium niobate: a two-dimensional nonlinear photonic crystal,” Phys. Rev. Lett. 84(19), 4345–4348 (2000). [CrossRef] [PubMed]
10. V. Berger, “Nonlinear photonic crystals,” Phys. Rev. Lett. 81(19), 4136–4139 (1998). [CrossRef]
11. R. Lifshitz, A. Arie, and A. Bahabad, “Photonic quasicrystals for nonlinear optical frequency conversion,” Phys. Rev. Lett. 95(13), 133901 (2005). [CrossRef] [PubMed]
12. P. Xu, S. H. Ji, S. N. Zhu, X. Q. Yu, J. Sun, H. T. Wang, J. L. He, Y. Y. Zhu, and N. B. Ming, “Conical second harmonic generation in a two-dimensional χ(2) photonic crystal: a hexagonally poled LiTaO3 crystal,” Phys. Rev. Lett. 93(13), 133904 (2004). [CrossRef] [PubMed]
13. K. Gallo, A. Pasquazi, S. Stivala, and G. Assanto, “Parametric solitons in two-dimensional lattices of purely nonlinear origin,” Phys. Rev. Lett. 100(5), 053901 (2008). [CrossRef] [PubMed]
14. T. Ellenbogen, N. Voloch-Bloch, A. Ganany-Padowicz, and A. Arie, “Nonlinear generation and manipulation of Airy beams,” Nat. Photonics 3(7), 395–398 (2009). [CrossRef]
15. K. Gallo, C. Codemard, C. B. Gawith, J. Nilsson, P. G. R. Smith, N. G. R. Broderick, and D. J. Richardson, “Guided-wave second-harmonic generation in a LiNbO3 nonlinear photonic crystal,” Opt. Lett. 31(9), 1232–1234 (2006). [CrossRef] [PubMed]
16. A. Arie and N. Voloch, “Periodic, quasi-periodic, and random quadratic nonlinear photonic crystals,” Laser Photonics Rev. 4(3), 355–373 (2010). [CrossRef]
17. G. Rosenman, Kh. Garb, A. Skliar, M. Oron, D. Eger, and M. Katz, “Domain broadening in quasi-phase-matched nonlinear optical devices,” Appl. Phys. Lett. 73(7), 865–867 (1998). [CrossRef]
18. R. G. Batchko, M. M. Fejer, R. L. Byer, D. Woll, R. Wallenstein, V. Y. Shur, and L. Ermann, “CW quasi-phase-matched generation of 60 mW at 465 nm by single-pass frequency doubling of a laser diode in backswitch poled lithium niobate,” Appl. Phys. Lett. 24, 1293–1295 (1999).
19. M. Manzo, F. Laurell, V. Pasiskevicius, and K. Gallo, “Electrostatic control of the domain switching dynamics in congruent LiNbO3 via periodic proton-exchange,” Appl. Phys. Lett. 98(12), 122910 (2011). [CrossRef]
20. L.-H. Peng, Y.-C. Zhang, and Y.-C. Lin, “Zinc oxide doping effects in polarization switching of lithium niobate,” Appl. Phys. Lett. 78(1), 4–6 (2001). [CrossRef]
21. L.-H. Peng, C.-C. Hsu, and Y.-C. Shih, “Second harmonic green generation from two-dimensional χ(2) nonlinear photonic crystal with orthorhombic lattice structure,” Appl. Phys. Lett. 83(17), 3447–3449 (2003). [CrossRef]
22. D. F. Clark, A. C. G. Nutt, K. K. Wong, P. J. R. Laybourn, and R. M. De La Rue, “Characterization of proton exchange slab optical waveguides in z cut LiNbO3,” J. Appl. Phys. 54(11), 6218–6220 (1983). [CrossRef]
23. F. Laurell, J. Webjorn, G. Arvidsson, and J. Holmberg, “Wet etching of proton-exchanged lithium niobate-a novel processing technique,” J. Lightwave Technol. 10(11), 1606–1609 (1992). [CrossRef]
24. C. E. Valdivia, C. L. Sones, J. G. Scott, S. Mailis, R. W. Eason, D. A. Scrymgeour, V. Gopalan, T. Jungk, E. Soergel, and I. Clark, “Nanoscale surface domain formation on the +z face of lithium niobate by pulsed ultraviolet laser illumination,” Appl. Phys. Lett. 86(2), 022906 (2005). [CrossRef]
25. D. E. Zelmon, D. L. Small, and D. Jundt, “Infrared corrected Sellmeier coeffcients for congruently grown lithium niobate and 5 mol. % magnesium oxide-doped lithium niobate,” J. Opt. Soc. Am. B 14(12), 3319–3322 (1997). [CrossRef]
26. W. H. Li, R. Tavlykaev, R. V. Ramaswamy, and S. Samson, “On the fabrication of annealed proton exchanged waveguides with electric field poled domain reversals in Z‐cut LiNbO3,” Appl. Phys. Lett. 68(11), 1470–1472 (1996). [CrossRef]
