Appl Phys A (2009) 95: 329 334 DOI 10.1007/s00339-009-5076-7 Fabrication of three-dimensional (3D) woodpile structure photonic crystal with layer by layer e-beam lithography Li Wang Sasa Zhang Qingpu Wang Jiaqi Chen Wei Jiang Ray T. Chen Received: 28 August 2008 / Accepted: 15 December 2008 / Published online: 21 January 2009 Springer-Verlag 2009 Abstract Photonic crystal based superprism offers a way to design new optical components for beam steering and DWDM (Dense Wavelength Division Multiplexing) application. Three-dimensional (3D) photonic crystals are especially attractive as they could offer more control of the light beam. A FCT (Face-Centered-Tetragonal) woodpile structure has been fabricated using layer by layer stacking techniques with E-Beam lithography. Special planarizations and processes have been introduced to ensure the survivability and good alignment of the fabricated nanostructures. Scanning electron microscopy results proved the structure uniformity. With the proper design, the structure exhibits superprism effects around 1550 nm, and such effects have been observed in the experiments. PACS 42.70.Qs 42.82.Cr 81.05.-t 1 Introduction The prediction and the confirmation that artificial periodic dielectric structures can be used to manipulate electromag- L. Wang S. Zhang ( ) J. Chen R.T. Chen Department of Electrical & Computer Engineering, The University of Texas at Austin, Austin, TX 78731, USA e-mail: sasazhang@sdu.edu.cn R.T. Chen ( ) e-mail: raychen@uts.cc.utexas.edu S. Zhang Q. Wang School of Information Science & Engineering, Shandong University, Ji nan, Shandong 250100, P.R. China W. Jiang Electrical & Computer Engineering Department, Rutgers, The State University of New Jersey, New York 08854-8058, USA netic wave propagation affect significantly the development of the micro- and nano-optoelectronics [1 5]. One of the many interesting phenomena is the superprism effect in the photonic crystal. It is the anomalous refraction of light at an interface between a photonic crystal and a homogeneous medium. The refraction angle is found to be very sensitive to the change of incident angle and wavelength under proper conditions. Such an effect arises from the anisotropy of the bands in the photonic crystal and such dispersion effects could be hundreds of times stronger than the conventional prism. And that is where it gains the name superprism. A number of groups have previously designed and fabricated superprism devices since its introduction by Kosaka and coworkers [6]. The superprismeffect intwo-dimensionalperiodic systems was investigated by Baba et al. [7] and Chung et al. [8] afterward. The first experimental demonstration of the two-dimensional (2D) superprism effect was reported by Wu et al. [9] who employed an asymmetric GaAs AlGaAs heterostructure to provide light confinement in the third dimension via total internal reflection. On the other hand, in the three-dimensional (3D) photonic crystal case by careful design you can get a better control of light in all directions, and you can achieve a more versatile design. Also the 3D photonic crystal can offer you a real 3D superprism effect instead of the in-plane superprism effect. In the literature, there is a great diversity in the fabrication approaches to make two-dimensional and three-dimensional photonic crystals. Therefore, improving the quality in terms of being more feature-size flexible, materials flexible and less time consuming is needed. Despite the remarkable progress in the fabrication of two-dimensional photonic crystals [10], there remain significant challenges for the fabrication of 3D photonic crystals, especially for producing sub-micron periodicity for near-ir applications. Many 3D fabrication approaches have been studied on a number of material plat-
330 L. Wang et al. forms. Among them, holographic fabrication [11], microassembly of planar semiconductor layers [12] and multiphoton absorption [13] have been investigated to create certain microstructures. However, for the aforementioned approaches the final structure is not that versatile [14 16]. E-beam lithography is a common approach to make nano-size structures. It can produce sub-10 nm structures easily, and with the advance of technologies people can build 3D structures with an alignment accuracy of less than 50 nm. Here we based on it and used an Electron Beam Lithography (EBL) machine (Jeol JBX6000) to fabricate 3D photonic crystal structures. The structure we have chosen is the woodpile Face Centered Tetragonal (FCT) structure. Basically it possesses one-dimensional (1D) periodic structure on each layer, and is produced by controlling the period of both horizontal/vertical directions and the aspect ratio of the periodic structures. We can achieve superprism effects around 1550 nm, which fits into the conventional optical communication transmission window. 2 Fabrication process Most of the semiconductors have atoms arranged in a diamond lattice structure, and energy bandgaps for electrons occur in this particular geometry. Similarly, the widest photonic bandgap takes place in the diamond lattice structure. Moreover, the diamond lattice structure is the most preferable configuration as the structure exhibits a photonic bandgap with the lowest contrast of the refractive index, i.e., the value of 2 [17]. However, its complicated arrangement of lattice points hinders the practical fabrication for optical wavelengths. Thus, in order to cope with the difficulties of making diamond lattice structures, a structure with a diamond-like properties and with a practical geometry for fabrication is required [18 22]. A sketch of a woodpile structure that can make the superprism effect and periodic arrays of dielectric rods placed on another array of rods perpendicular to one another forming a photonic crystal is shown in Fig. 1. For its appearance the structure is called the woodpile structure. The stacking sequence is such that a unit cell consists of every four layers. The distance between four adjacent layers is denoted by c. Within each layer, it has the layers of one-dimensional rods in-plane rod spacing d with a stacking sequence that repeats itself every four layers, the distance between in-plane rods is d; w and h are the width and height of each in-plane rod, respectively. The adjacent layers are rotated by 90. Between every other layer, the rods are shifted relative to each other by a half of a period (d/2). Generally, the resulting structure has a face-centeredtetragonal (FCT) lattice symmetry. For the special case of c/d = 2, the structure has a face-centered-cubic (FCC) Fig. 1 A sketch of a woodpile structure, where c is the distance between four adjacent layers, d is the distance between in-plane rods, w and h are the width and height of each in-plane rod, respectively symmetry. By the above procedures, the 3D photonic crystal structures were fabricated by the layer-by-layer stacking method. Our 3D polymer photonic crystals are fabricated using the layer-by-layer stacking method. The structure consists of layers one-dimensional rods, stacking according to certain crystal symmetry to form a lattice structure. The fabrication procedure can be summarized as shown in Fig. 2.We first write the alignment marks on the silicon substrate. And each layer s pattern is written at a correct position referencing to the alignment marks fabricated in the first step. After developing the E-beam resist, the pattern is transferred to the SiO 2 layer by Reactive Ion Etch (RIE). The following step is the planarization of each layer. The polymer DUV30J is spin-coated to planarize the wafer surface. Then RIE is used to expose the SiO 2 surface by etching back at the same rate of polymer and SiO 2. Digital Instrument AFM is used to measure the surface roughness and planarization results. By repeating the process, we have fabricated four layers of a woodpile structure. Here we will fabricate 3D polymer/ SiO 2 photonic crystal superprism with low refractive index contrast. It has low loss at 1550 nm wavelength and thermal stability. The polymer also needs to have good planarizing property. So DUV30J (Brewer Science) is chosen as the polymer material for superprism. The first layer e-beam resist ZEP 520A spun at 4000 rpm for 60 seconds was patterned by the e-beam lithography. A RIE step was executed afterwards to create the desired pattern in the underlying SiO 2 layers. To create the second layer, a key issue is the planarization. We coated the 1st layer with DUV30J (Brewer Science) and then carefully controlled the etch back time in Oxford RIE, 30 mtorr total pressure, 20 sccm CHF 3, 20 sccm Ar, SiO 2 etch rate 35 nm/min. We firstly only use O 2 to remove the polymer DUV30J to expose SiO 2 layer. It is difficult to control the etching stop time. If the etching time is over 1 2 min compared with the threshold etch time, the max peak-to-valley
Fabrication of three-dimensional (3D) woodpile structure photonic crystal with layer by layer e-beam 331 Fig. 2 Process flow of a 3D woodpile structure using layer-by-layer stacking method Fig. 3 a Top view of four layers of the woodpile structure fabricated using layer-by-layer stacking method. b Side view of four layers of the woodpile structure distance can reach 20 40 nm. This etch bias will transfer to the following layer and it is more difficult to keep global wafer flatness. Then we improved the etch-back process by simultaneously etching-back both the polymer DUV30J and SiO 2 at the same rate until the SiO 2 layer is exposed on the planarized surface. The chemistry of the etch-back process is based on CHF 3 and O 2 gas. High planarization level and low surface roughness can be obtained by adjusting CHF 3 and O 2 gas flow and RF power [23 25]. The etch rate of both the polymer and SiO 2 can be controlled at about 26 nm/min with a power 200 W and a chamber pressure 30 mtorr. The gas flow rates are 20 sccm CHF 3 and 3.8 sccm O 2. Then we can over-etch 1 2 min to expose SiO 2 surface. The AFM image and profile of the first layer etched back by RIE show the max peak-to-valley distance is less than 10 nm after etchback process to create a flat basis for the second layer SiO 2 deposition. The further steps are just a repetition of the first 8 steps, and eventually we got a layer-by-layer SiO 2 /polymer woodpile structures. 3 Fabrication results of 3D woodpile photonic crystal structure Figure 3 shows the SEM images of the fabricated 3D polymer photonic crystal structures. The four layers of woodpile structure with in-plane rod spacing of d = 1.387 µm and rod width of w = 0.369 µm. The height of each layer
332 L. Wang et al. Fig. 4 a The photonic band structure of the 3D woodpile structure. b Reciprocal lattice (first Brillouin zone) of the FCT woodpile structure is 0.423 µm. The cross-section pictures show the alignment error between the first and the third layer below 50 nm. 4 In-plane superprism effects simulation and demonstration When fabricating, according to the symmetry in FCT and FCT approximate calculation, we used the FCT unit vectors as the unit vectors for the bulk simulation. With the structure data extracted from the previous simulation and experiment we calculated the y z plane, or (100) plane, in-plane superprism effects. The 3 unit vectors in the real space we used in the simulation were a 1 = 0.864x 0.430z, a 3 = 0.693y 0.430z, a 2 = 0.864x 0.693y, where all the units were in µm. And the corresponding unit vectors for the reciprocal lattice were b 1 = 3.74x + 4.52y 7.36z, b 2 = 3.74x 4.52y + 7.36z, b 3 = 3.74x 4.52y + 7.36z, where all the units were in µm 1. Fig. 5 a The 4th band dispersion surface in (100) plane.b The wavelength sensitive superprism effect in (100) plane The band structure is calculated using BANDSOLVE software package which utilizes the plane wave expansion method. Figure 4(a) shows the photonic band structure calculated for FCT lattice woodpile with w/d = 0.266 and h/d = 0.305 in the polymer/sio 2 medium and with n = 0.17. There is no complete bandgap in such a woodpile due to low refractive index contrast. This band structure only displays the energy along lines connecting the high symmetry points on the Brillouin zone surface (shown in the inset of Fig. 4(b)). To calculate the dispersion surface, we need to calculate the entire band structure throughout the first Brillouin zone. Now we are looking at the in-plane optical properties. We have drawn the dispersion surface of the 4th band in (100) plane as shown in Fig. 5(a). At low normalized frequency (ωa/2πc < 0.7), far from the partial bandgap, the band structure is isotropic, and the dispersion surface is circle-
Fabrication of three-dimensional (3D) woodpile structure photonic crystal with layer by layer e-beam 333 Fig. 6 a The optical setup to observe the superprism effect. b Beam propagation without the superprism effect at 1573 nm. c The superprism effect at 1581 nm like with a radius given by the magnitude of wave vector. At high normalized frequency (0.7 <ωa/2πc <1.1), near the photonic bandage, the band structure becomes anisotropic. As a result, the shape of dispersion surface deviates from circle. Here we chose a from 1088 to 1092 nm, the results could be in good agreement with the following simulation data. The propagation direction is obtained by computing the normal to the dispersion surface at the end point of the propagation wave vector based on the momentum conservation rule. And in this case with an input angle of 11, we can achieve a maximum beam steering from 14 to more than 48 within 3 nm wavelength tuning. The simulation results are shown in Fig. 5(b). Due to the fabricating error and the limitation in experimental techniques, much bigger error with more multilayers, we only used the light in plane direction for limiting propagation. Thus, 4 layers of rods are enough to demonstrate the superprism effect in two-dimensional plane of the fabricated samples. To demonstrate the superprism effect in the fabricated samples we have conducted transmission experiments as shown in Fig. 6(a). The 3D sample was mounted on the optical stage, and a lensed fiber with an output beam diameter of 2 µm was used to couple the laser light into the fabricated sample. The beam direction inside the photonic crystal was monitored by a CCD camera mounted right above the sample. By tuning the tunable laser wavelength we have observed that the beam inside the photonic crystal region changed 37 when the wavelength was changed from 1573 to 1581 nm (shown in the inset of Figure 6(b), (c)). Because the input angle is very sensitive for observing superprism effect and the laser light is very difficult to couple into the fabricated 3D sample, we only discussed two wavelengths 1573 and 1581 nm. Such an experiment result is very similar to the simulation and is a direct confirmation of the feasibility of using woodpile structure for beam steering application. Compared with the superprism effect around 1580 nm in the experiments and the simulation results at 1668 nm, this difference between the experiment and the simulation can be attributed to the rods that are not exactly rectangular solid, etch bias and the defects in the fabricated samples.
334 L. Wang et al. 5 Conclusions In conclusion, we have demonstrated the fabrication of a nano 3D woodpile photonic crystal structure with layer-bylayer stacking techniques combined with e-beam lithography. Such a structure provided strong, both wavelength sensitive superprism effects around 1550 nm, which is suitable for the telecommunication application. Acknowledgements This work was supported by the Air Force Research Laboratory and the travel fund for Sasa Zhang was from China Scholarship Council (Grant No. 2005B47011) and Shandong University. References 1. E. Yablonovitch, Phys. Rev. Lett. 58, 2059 (1987) 2. S. John, Phys. Rev. Lett. 58, 2486 (1987) 3. J.D. Joannopoulos, R.D. Meade, J.N. Winn, Photonic Crystals (Princeton, New York, 1995) 4. W. Jiang, R.T. Chen, Phys. Rev. Lett. 91, 213901 (2003) 5. Y. Jiang, W. Jiang, L. Gu, X. Chen, R.T. Chen, Appl. Phys. Lett. 87, 221105 (2005) 6. H. Kosaka, T. Kawashima, A. Tomita, M. Notomi, T. Tamamura, T.Sato,S.Kawakami,Phys.Rev.B.58, R10096 (1998) 7. T. Baba, M. Nakamura, J. Quantum Electron. 38, 909 (2002) 8. K.B. Chung, S.W. Hong, Appl. Phys. Lett. 81, 1549 (2002) 9. L. Wu, Y. Zhong, C.T. Chan, K.S. Wong, G.P. Wang, Appl. Phys. Lett. 86, 241102 (2005) 10. H. Benisty, J. Lourtioz, A. Chelnokov, S. Combrie, X. Checoury, Proc. IEEE 94, 997 (2006) 11. L. Wang, W. Jiang, X. Chen, L. Gu, J. Chen, R.T. Chen, J. Appl. Phys. 101, 114316 (2007) 12. K. Aoki, H.T. Miyazaki, H. Hirayama, K. Inoshita, T. Baba, Nat. Mater. 2, 117 (2003) 13. M. Deubel, G.V. Freymann, M. Wegner, S. Pereira, K. Busch, C.M. Soukoulis, Nat. Mater. 3, 444 (2004) 14. G. Subramania, S.Y. Lin, Appl. Phys. Lett. 85, 5037 (2004) 15. M. Qi, E. Lidorikis, P.T. Rakich, S.G. Johnson, J.D. Joannopoulos, E.P. Ippen, H.I. Smith, Nature 429, 538 (2004) 16. G. Subramania, Y.J. Lee, B.A. Hernandez-Sanchez, A.J. Fischer, T.S.Luk, I.Brener,P.G.Clem,T.J.Boyle, Opt.Express 15, 13049 (2007) 17. K.M. Ho, C.T. Chan, C.M. Soukoulis, Phys. Rev. Lett. 65, 3152 (1990) 18. K.M. Ho, C.T. Chan, C.M. Soukoulis, R. Biswas, M. Sigalas, Solid State Commun. 89, 413 (1994) 19. S.Y. Lin, J.G. Fleming, D.L. Hetherington, B.K. Smith, R. Biswas, K.M. Ho, M.M. Sigalas, W. Zubrzycki, S.R. Kurtz, J. Bur, Nature 394, 251 (1998) 20. Serbin, M. Gu, Opt. Express 14, 3563 (2006) 21. Chutinan, S. Noda, Phys. Rev. B 57, R2006 R2008 (1998) 22. Feigel, M. Veinger, B. Sfez, A. Arsh, M. Klebanov, V. Lyubin, Appl. Phys. Lett. 83, 4480 4482 (2003) 23. R. Hsiao, J. Carr, Mat. Sci. Eng. B 52, 63 (1998) 24. R. Hsiao, IBM J. Res. Dev. 43, 89 (1999) 25. L. Chen, G.S. Mathad, European Patent EP0151948 (1985)