Direct electron beam writing of channel waveguides in nonlinear optical organic crystals

We report on optical channel waveguiding in an organic crystalline waveguide produced by direct electron beam pat terning. The refractive index profile as a function of the applied electro n fluence has been determined by a reflection scan method in the nonlinear optic al organic crystal 4-N, N-dimethylamino-4’-N’-methyl-stilbazoliu m tosylate (DAST). A maximal refractive index reduction of ∆n1 = −0.3 at a probing wavelength of 633 nm has been measured for an electron fluence of 2. 6 mC/cm2. Furthermore, a new concept of direct channel waveguide patt erning in bulk crystals is presented and waveguiding has been demonst rated in the produced structures by end-fire coupling. Mach-Zehnder modulators have been successfully realized and a first electro-optic mo dulation at a wavelength ofλ = 1.55μm has been demonstrated therein. © 2007 Optical Society of America OCIS codes: (230.7380) Waveguides, channeled; (160.2100) Electro-op tical materials; (130.4110) Integrated optics, modulators References and links 1. Y. Shi, C. Zhang, H. Zhang, J. H. Bechtel, L. R. Dalton, B. H. Robinson, and W. H. Steier, “Low (sub-1-Volt) halfwave voltage polymeric electro-optic modulators achiev ed by controlling chromophore shape,” Science 288 (5463), 119-122 (2000). 2. M. Lee, H. E. Katz, C. Erben, D. M. Gill, P. Gopalan, J. D. Heb er, D. J. McGee,“Broadband modulation of light by using an electro-optic polymer,” Science 298(5597), 1401-1403 (2002). 3. Y. Enami, C.T. Derose, D. Mathine, C. Loychik, C. Greenlee, R. A. Norwood, T.D. Kim, J. Luo, Y. Tian, A. K. Y. Jen and N. Peyghambarian, ”Hybrid polymer/so-gel waveguide mo dulators with exceptionally large electro-optic coefficients,” Nature Photonics 1 (3), 180-185 (2007). 4. S. R. Marder, J. W. Perry, and W. P. Schaefer, “Synthesis of organic salts with large second-order optical nonlinearities,” Science245(4918), 626-628 (1989). 5. F. Pan, G. Kn̈ opfle, Ch. Bosshard, S. Follonier, R. Spreiter, M. S. Wong, an d P. G̈unter, “Electro-optic properties of the organic salt 4-N, N-dimethylamino-4’-N’-methyl-stilb azolium tosylate,” Appl. Phys. Lett. 69 (1), 13-15 (1996). 6. R. Spreiter, Ch. Bosshard, F. Pan, and P. G ünter, “High-frequency response and acoustic phonon contr ibution of the linear electro-optic effect in DAST,” Opt. Lett. 22 (8), 564-566 (1997). 7. L. Mutter, A. Guarino, M. Jazbiň sek, M. Zgonik, P. G̈unter and M. D̈obeli,“Ion implanted optical waveguides in nonlinear optical organic crystal,” Opt. Express 15, 629-638 (2007). 8. L. Mutter, M. Jazbiňsek, M. Zgonik, U. Meier, Ch. Bosshard, and P. G ünter,“Photobleaching and optical properties of organic crystal 4-N, N-dimethylamino-4’-N’-methyl stilbazolium tosylate,” J. Appl. Phys. 94 (3), 13561361 (2003). 9. T. Kaino, B. Cai, and K. Takayama, “Fabrication of DAST chan nel optical waveguides,” Adv. Funct. Mater. 12 (9), 599-603 (2002). 10. Ph. Dittrich, R. Bartlome, G. Montemezzani, and P. G ünter, “Femtosecond laser ablation of DAST,” Appl. Surf. Sci.220(1-4), 88-95 (2003). #89777 $15.00 USD Received 15 Nov 2007; revised 29 Nov 2007; accepted 30 Nov 2007; published 4 Dec 2007 (C) 2007 OSA 10 December 2007 / Vol. 15, No. 25 / OPTICS EXPRESS 16828 11. W. Geis, R. Sinta, W. Mowers, S. J. Deneault, M. F. Marchan t, K. E. Krohn, S. J. Spector, D. R. Calawa, and T. M. Lyszczarz, “Fabrication of crystalline organic waveguide s with an exceptionally large electro-optic coefficient,” Appl. Phys. Lett.84 (19), 3729-3731 (2004). 12. S. Manetta, M. Ehrensperger, Ch. Bosshard, and P. G ünter, “Organic thin film crystal growth for nonlinear optic s: present methods and exploratory developments,” C. R. Physiqu e,3 (4), 449-462, (2002). 13. M. Thakur, J. Xu, A. Bhowmik, and L. Zhou, “Single-pass thi n-film electro-optic modulator based on an organic molecular salt,” Appl. Phys. Lett. 74 (5), 635-637 (1999). 14. A. J. Houghton and P. D. Townsend, “Optical waveguides fo rmed by low energy electron irradiation of silica,” Appl. Phys. Lett.29 (9), 565-566 (1976). 15. S. G. Blanco, A. Glidle, J.H. Davies, J.S. Aitchison, and J. M. Cooper, “Electron-beam-induced densification of Ge-doped flame hydrolysis silica for waveguide fabrication, ” Appl. Phys. Lett.79 (18), 2889-2891 (2001). 16. K. Komatsu, T. Abe, O. Sugihara, T. Kaino, “Waveguide patt ern formation in organic nonlinear optical crystal by electron beam irradiation,” Proc. of SPIE. 5724, 131-138 (2005). 17. F. Pan, M.S. Wong, Ch. Bosshard, and P. G ünter, “Crystal Growth and Characterization of the organic salt 4-N, N-dimethylamino-4’-N’-methyl-stilbazolium tosylate (DAST ),” Adv. Mater.8 (7), 592-595 (1996). 18. C. Solcia, D. Fluck, T. Pliska, P. G ünter, St. Bauer, M. Fleuster, L. Beckers, and Ch. Buchal, “T he refractive index distributionnc(z) of ion implanted KNbO3 waveguides,” Opt. Commun. 120(1-2), 39-46 (1995). 19. Ch. Bosshard, K. Sutter, Ph. Pr être, J. Hulliger, M. Fl̈ orsheimer, P. Kaatz, and P. G ünter,Organic nonlinar optical materials, (Gordon and Breach Science Publishers, 1995).


Introduction
Organic noncentrosymmetric materials have recently attracted much attention for applications in photonic devices because of their almost unlimited design possibilities to tailor the linear and nonlinear optical properties and the fact that the nonlinear active organic molecules, embedded either in a polymer matrix or orderly packed in a crystal, show an almost pure electronic origin of their nonlinearities [1,2,3].Therefore, organic nonlinear optical materials are very interesting and promising for high-speed electro-optic applications in future telecommunication networks.In contrast to the more widely studied electro-optic poled polymers, organic crystals offer the advantages of superior photochemical stability and potentially higher nonlinearities.Furthermore, in organic crystals no thermal relaxation of the chromophores is present since they are embedded in a crystalline structure, whereas poled polymers often tend to lose their electro-optic activity due to thermal relaxation of the oriented chromophores.Nevertheless, thin film growth and processing of organic crystals is still very challenging.
In this work we exploit the process of direct electron beam (e-beam) patterning of DAST.Ebeam direct structuring has already been used to create channel waveguides in silica, in which the refractive index was increased in the exposed area [14,15].On the contrary, for the organic nonlinear optical crystal AANP a decrease of the refractive index has been recently observed by e-beam irradiation [16], but no waveguiding configuration has been suggested yet.
We present our results on channel waveguide patterning in organic nonlinear optical bulk crystals by electron beam exposure, which is offering several advantages over previously reported structuring techniques.Using this method, channel waveguides can be directly patterned in a single process step and the waveguide dimensions can be precisely tailored to attain single mode waveguiding, which is a prerequisite for efficient electro-optic devices.Since no mechanical or chemical etching is required as compared to standard photolithography, the produced channel waveguides have smoother side walls.Furthermore, e-beam exposure offers the benefit of submicrometer resolution.
In the first part of this work we show that the refractive index of the DAST area irradiated with electrons is reduced and we introduce a simple model that relates the induced refractive index change with the deposited energy of the electrons.Then we propose a new concept of channel waveguide patterning in a single process step in bulk crystals, which is used to successfully realize channel waveguides and Mach-Zehnder modulator structures.Furthermore, a first demonstration of electro-optic modulation in these structures is described.

Electron beam experiments
DAST belongs to the monoclinic point group m [17].The crystallographic b axis and the dielectric x 2 axis are normal to the mirror plane, and the x 1 axis makes an angle of 5.4 • to the polar a axis in the symmetry plane [8].The noncentrosymmetric crystal packing is achieved by strong Coulomb interactions between the positively charged, nonlinear optical chromophore stilbazolium and the negatively charged counter ion tosylate, which are shown in Fig. 1 [4].The samples used in our experiments were grown from supersaturated methanol solution by the temperature-lowering method [17].The crystals were then cut almost perpendicular to the dielectric axes and the x 1 x 2 surface polished to λ /4 surface quality.The typical sample size was about 7 × 4 × 4 mm 3 with the longest dimension along x 1 .The samples were protected with a 150 nm thick polyvinylacetate (PVAc) layer and then covered with 20 nm Cr in order to prevent the sample from charging up during the electron beam exposure.For direct patterning the structures were exposed with a Raith electron beam system with an electron energy of 30 keV and a relatively low electron current of 0.3 nA to avoid thermal damage of the samples.After the exposure, the Cr layer was removed by Ar + sputtering and the PVAc layer, which protects DAST from Cr indiffusion during the sputtering, is dissolved in toluene.

Model for e-beam exposure
To describe the refractive index change induced by electron beam exposure, we have chosen a model analogous to the one used for H + implanted samples [7], in which the induced refractive index change was related to the deposited energy of the implanted ions in an organic crystal.
The energy deposition of electrons in DAST was calculated with the Monte Carlo simulation tool for electron trajectory in solids CASINO (www.gel.usherbrooke.ca/casino/).Figure 2 shows the deposited energy per electron G(z) = dE(z)/dz as a function of the electron penetration depth z for an electron energy of 30 keV and a homogeneous e-beam exposure.A maximal energy deposition occurs at a depth of about 7 µm.
The phenomenological relation between the induced refractive index change ∆n and the stored energy is given by [7]  where ∆n max is the saturation refractive index change, φ represents the electron fluence, G 0 is a normalization energy and γ an exponential factor.Note that for high electron fluences the expression tends to the saturation refractive index change ∆n max .

Reflection scan measurement method
The parameters ∆n max , G 0 , and γ in Eq. ( 1) are material depended and were determined with the following experiment.In a first step ten lines of 30 µm width were written with linearly increasing fluences from 0.26 mC/cm 2 up to 2.6 mC/cm 2 .The sample was then polished under a wedge of 1 • in order to increase the spatial resolution in z direction as shown in Fig. 3.The refractive index at the polished surface was determined by measuring the back reflected η z||x 3 light from the sample surface [18].The refractive index n is related to the intensity reflection coefficient R = I R /I 0 by the Fresnel formula, where I 0 is the incident light intensity and I R the reflected one.For normal incidence n is given by The laser light of a HeNe laser was focused on the sample surface to a beam diameter of less than 5 µm.It was linearly polarized along the dielectric x 1 axis, since the alteration of n 1 is of main interest in order to use the largest electro-optic coefficient r 11 of DAST.The lines were then scanned in η direction with a step size of 8 µm.Since the η and the z axis are related to each other by a simple projection, this measurement yields a depth resolution in z direction of less than 0.2 µm.The back reflected light from the sample surface was deflected by a beam splitter and detected with a photodiode.

Results: Refractive index profile
The measured refractive index profiles indicated by dots for five different fluences are shown in Fig. 4. The refractive index is decreased by electron beam exposure.For a fluence of φ = 2.6 mC/cm 2 , a maximal refractive index reduction of ∆n = −0.3 is achieved.The shape of the refractive index profile is similar to the one of the deposited energy, compare Fig.  factor γ is equal to 1 and that we are in a low fluence regime for the used fluences of up to 2.6 mC/cm 2 , which means that the argument of the exponential term is φ G(z)/G 0 << 1 and can be consequently linearized, which yields with c = −0.036± 0.001 cm 2 µm/(CeV) at a wavelength of λ = 633 nm.The refractive index change at a certain position is therefore simply proportional to the deposited energy φ G(z), and can be calculated for any exposure parameters in the low fluence regime by considering only one experimental constant c.

Dispersion of the refractive index change
Since we are mainly interested in waveguiding at telecommunication wavelengths, the dispersion of the proportionality constant c(λ ) is also important.A relation between the microscopic material modification and the resulting refractive index reduction will be derived with help of the Lorentz-Lorenz relation in order to finally calculate c at a wavelength of λ = 1.55 µm.
The refractive index in DAST is reduced after e-beam exposure due to changes of the molecular structure, which leads to a reduction of the molecular polarizability α.The refractive index n ω at a frequency ω is related to the molecular polarizability α(ω) by where N denotes the number of molecules per volume and f ω = (n 2 ω +2)/3 is the Lorentz local field factor.After exposure, the refractive index ñω (z) at depth z is altered due to the fact that some chromophores are modified.Therefore, the refractive index ñω is given by where p is the percentage of unmodified molecules after electron beam exposure and α represents the polarizability of the modified molecules.α can be expressed by where k ω denotes the percentage of remaining polarizability of the modified molecule.Combining Eqs.(4-6), the percentage of unmodified molecules p(z) at the depth z is given by Since the percentage of unmodified molecules p(z) is independent of the wavelength, the refractive index profile ñω ′ (z) at the frequency ω ′ can be calculated with help of the following equation Taking the reasonable assumption that k ω is constant with wavelength, i.e. k ω = k ω ′ , the refractive index profiles at 1.55 µm can be calculated from (8) and the data at 633 nm, which leads to c(λ = 1.55µm)= −0.019 ± 0.001 cm 2 µm/(CeV).In order to estimate the accuracy of this assumption, we calculated the lower limit for c at 1.55 µm by taking the strict condition that the polarizability of the modified molecule is dispersion free, i.e. α(λ = 633 nm) = α(λ = 1.55 µm).Lowering the polarizability of the modified molecules from 50% to 10% of their initial value at the zero frequency limit, the values of the corresponding c parameters were agreeing within 15% to the value obtained by considering k ω = k ω ′ .

Concept of channel waveguide patterning
The electrons of the writing beam are scattered in the material and therefore the beam is widened up.This circumstance can be exploited to directly write channel waveguides in a bulk crystal by exposing two lines separated by the waveguide width as depicted in Fig. 5. Between the two lines an unexposed region surrounded by an e-beam modified area with lowered refractive index remains and thus a channel waveguide is formed.The advantage of this configuration is that the waveguide core is mostly in the virgin material, in which the nonlinear and electro-optic properties are the same as in the bulk material.
With the CASINO simulation software the 2 dimensional energy distribution in the target material for an electron beam at a fixed position can be calculated.Subsequently the energy distribution for the configuration shown in Fig. 5 was calculated by overlapping the single beam distributions over the exposed area.Using the relation between the refractive index reduction and the electron fluence obtained in Sections 3 and 4, the two dimensional refractive index cross section was calculated.In Fig. 6    integrated optics software OlympIOs with the full vectorial complex bend 2D mode solver.The resulting intensity profiles of the first order and the third order mode are depicted in Fig. 6(a) and (b), respectively.Whereas the first mode is well confined and has simulated losses below 0.1 dB/cm, the third order mode is leaky with tunneling losses of over 20 dB/cm.

Waveguiding observation
We have produced several channel waveguide structures using fluences of 0.65, 1.3, and 2.6 mC/cm 2 with a line width of L = 4 µm and varying the waveguide core width W = 3, 6, 9 and 12 µm.The x 1 x 3 end-faces were subsequently polished and waveguiding experiments performed by standard end-fire coupling with the setup depicted in Fig. 7(a).The light was propagating along the x 2 direction and was polarized parallel to x 1 as shown in Fig. 7(b).Waveguiding was clearly observed for the fluences of 1.3 and 2.6 mC/cm 2 , whereas the structures exposed with the lowest fluence of 0.65 mC/cm 2 were having too large tunneling losses, as confirmed by simulations.In Fig. 7(c) a photograph taken from top is shown and in (d) photographs of the mode profiles taken at the output face of waveguides with core widths of 6, 9, and 12 µm and a fluence of 2.6 mC/cm 2 are depicted.We were not able to efficiently couple light into the waveguides with a core width of W = 3 µm because of the not perfect edge quality.
The waveguide losses were measured by detecting the scattered light from top with a CCD camera and are about 20 dB/cm, which is in contrast to the simulated losses of less than 0.1 dB/cm for the first order modes.We attribute this partly to the edge quality and therefore to a weak coupling efficiency to the first order mode, which is confined closest to the surface, (C) 2007 OSA whereas higher order modes are easier to excite, since they are less confined at the sample surface, but they have high tunneling losses, compare also Fig. 6.Therefore, from top mainly the decay of the higher order modes is observed since most of the energy is coupled into them.

Electro-optic modulation
Mach-Zehnder structures were successfully realized with similar parameters as the channel waveguides.The electron fluence was varied from 1.3 to 2.6 mC/cm 2 at core widths of W = 4, 6 and 8 µm.The line width L was chosen to be 5 µm in order to reduce the tunneling losses into the substrate at the Y-junctions of the structures.A schematic illustration of the Mach-Zehnder geometry is shown in Fig. 8. Electrodes were patterned subsequent to the electron beam exposure by the following procedure.First, the sample was covered with a SU-8 layer with a thickness of about 2 µm, which was acting as an optical buffer and at the same time as a protecting layer against different solvents used during the electrode structuring process.The electrodes were then patterned in an AZ-5214 photoresist by standard photolithography.After depositing 5 nm Cr and 100 nm Au, the AZ-resist was washed away with acetone.

Cladding Electrodes
The electro-optic modulation measurements were performed in the experimental configuration shown in Fig. 9(a).The light was propagating in x 2 direction and was polarized parallel to the dielectric x 1 axis in order to use the largest electro-optic coefficient of DAST i.e. r 11 = 47 pm/V at 1.55 µm.The phase shift between the two arms is given by [19] where L o is the arm length (overlap length between the modulation field the guided optical wave) and E 1 represents the magnitude of the modulation field along x 1 inside the DAST waveguides, which was calculated with Femlab (www.femlab.com).The modulation amplitude at the output of the Mach-Zehnder structure is proportional to cos 2 (φ /2).In of the modulation was about 20% of the output signal.At present, the half wave voltage is still higher than 10V, since the modulator dimensions and the electrode arrangement have not been optimized yet.The electrode spacing was with 20 µm relatively wide, and the effective interaction arm length, where the optical and electric fields were overlapping, was only L o = 0.85 mm long.

Towards single mode waveguides
Single mode characteristics is of importance for electro-optic applications.Some electron beam systems are offering now the possibility to irradiate the sample under an angle.This possibility can be used to produce single mode waveguides in DAST.In Fig. 10 one possible configuration

Conclusion
In conclusion, we have successfully applied direct electron beam patterning to organic crystalline materials.For the first time, optical channel waveguides and Mach-Zehnder structures in a bulk organic crystal have been produced in a single exposure step by applying a new concept, which is also offering the advantage that the waveguide core is not affected by the electron beam and thus the electro-optic and nonlinear optic properties are therein the same as in the bulk material.Waveguiding has been demonstrated in DAST channel waveguides and Mach-Zehnder modulators structured by using 30 keV e-beam irradiation.In addition, a first electro-optic modulation has been demonstrated in the produced Mach-Zehnder modulators.Therefore, this work gives a new perspective for the use of nonlinear optical organic materials for integrated optics.
The refractive index profile in DAST has been measured as a function of the electron range and the electron fluence in DAST.A reduction of the refractive index n 1 of ∆n 1 = −0.3 at a wavelength of 633 nm has been obtained using an electron fluence of 2.6 mC/cm 2 .Using this data and the 2 dimensional refractive index profile of the channel waveguide, the corresponding mode profiles and losses were determined.Furthermore, a configuration for single mode waveguides in DAST has been proposed which is of importance for future applications.

#Fig. 2 .
Fig. 2. Deposited energy in DAST as a function of the electron depth z for an electron energy of 30 keV and homogenous e-beam exposure as calculated by CASINO.

Fig. 3 .
Fig. 3. Experimental configuration for the determination of the refractive index profile by measuring the back reflected light from a wedged-polished sample surface.The electron beam patterned lines were scanned in η direction.With a projection the refractive index profile as a function of the depth z can be obtained.

Fig. 4 .
Fig. 4. Measured refractive index n 1 (dots) as a function of the depth z shown for five different fluences.The solid curves correspond best to the theoretical model of Eq. (1) or (3).

x 1 x 2 x 3 Fig. 5 .
Fig.5.Concept of channel waveguide patterning in DAST: Two lines spaced by the waveguide core width are exposed by e-beam (gray).Since the electron beam is widened up in the target material, an unexposed region surrounded by an exposed area with lowered refractive index is created and thus a waveguide formed (red).
the calculated refractive index profile of n 1 at the -12 -9 -6 -3

Fig. 6 .
Fig. 6.Calculated 2D profile of the refractive index n 1 at a wavelength of λ = 1.55 µm for an electron fluence of φ = 2.6 mC/cm 2 , a line width L = 4 µm and a waveguide core width of W = 6 µm.The corresponding first and third order modes are depicted in Figure (a) and (b), respectively.

Fig. 7 .
Fig. 7. (a) Experimental setup for the determination of the waveguiding characteristics and propagation losses: CCD: infrared camera.(b) Waveguiding configuration: the light polarized parallel to x 1 was propagating along the x 2 direction.(c) Photograph of the sample surface taken from top with the CCD camera and (d) from the end face for different waveguide widths W .The line width was L = 4 µm and the structure were exposed with an electron fluence of φ = 2.6 mC/cm 2 .

Fig. 8 .
Fig. 8. Mach-Zehnder modulator geometry in DAST with in-plane electrodes in order to use the largest electro-optic coefficient r 11 of DAST; L o : electrical and optical field overlap length, L : line width, and W : waveguide core width.The bend radii of the segments used to pattern the Y-junctions were 5 mm long.

Fig. 9 ((Fig. 9 .
Fig. 9. (a) Measurement setup for the electro-optic modulation experiment.The light at a wavelength of 1.55 µm is propagating along the x 2 direction and is polarized parallel to the x 1 axis.The modulated light intensity was detected with a photodiode.PD: photodiode, A: aperture and L: lens.(b) The applied modulation voltage at a frequency of 20 kHz (lower curve) and the detected signal with the photodiode (upper curve) are depicted.

Fig. 10 .
Fig. 10.A possible configuration for the production of a single mode waveguide obtained by tilting the electron beam by 30 • .The depicted refractive index profile at 1.55 µm corresponds to an electron fluence of 2.6 mC/cm 2 and an energy of 30 keV.
2. The full curves correspond best to the theoretical model described in Section 3.1 and were obtained by least-squares theoretical analysis with the following remarks.The analysis showed that the