Open Access
Add to BibSonomyAbstract
We describe an immersion grating (IG) with both low polarization-dependent loss (PDL) and high diffraction efficiency. Our immersion grating consists of a silicon (Si) prism and a Si grating coated with dielectric film. We analyze the effect of the refractive index of dielectric film and clarify the refractive index ratio between a Si grating and dielectric film. Selecting an adequate material for the dielectric film, we design the diffraction efficiency of TM polarization without changing that of TE polarization. The simulation results for the optimized IG show a PDL of 0.2 dB and diffraction efficiency of −0.4 dB. A prototype with high dispersion power provides a low PDL of 0.6 dB, while maintaining a high diffraction efficiency of −0.4 dB.
©2013 Optical Society of America
1. Introduction
In optical devices for wavelength-division-multiplexing (WDM) telecommunications networks, such as optical channel monitors and wavelength selective switches, the dispersion element is a key component for multiplexing and demultiplexing the signal for each channel.
A high-dispersion element is necessary for downsizing the devices, and an arrayed waveguide grating based on the planar devices structure [1] is a solid solution. However, a grating is used in free-space optics so that the requirement for multiple ports or functional integration in some telecommunications devices can be met [2]. An immersion grating (IG) is attracting much attention as a high-dispersion element [3–8] for free-space optics. An IG is composed of a reflection grating and a prism [Fig. 1]. A high dispersion power can be obtained because light is dispersed both by the grating and prism, which amplifies the dispersion angle at the boundary of the prism and air. Therefore, the higher the refractive index of the prism is, the higher the dispersion power the IG has. Silicon (Si) is a good choice for this kind of grating because it has a high refractive index (n = 3.5) in the C + L band for optical communications [6] and Si fabrication technology is very mature. However, in a conventional Si IG, there is a trade-off between diffraction efficiency and polarization-dependent loss (PDL). The reflection grating of the IG is commonly coated with metal to reflect light. However, the absorption of the metal reduces the efficiency. For example, an IG coated with aluminum has diffraction efficiency of about −1 dB and low PDL of less than 0.3 dB [6]. In contrast, an IG based on total internal reflection (TIR) [7] has high diffraction efficiency of more than −1 dB because of the absence of absorption in metal layers, but produces very high PDL of more than 20 dB due to the great difference in propagation in the grating layer between TE and TM polarizations.
In this paper, we propose a new IG structure based on TIR to achieve both low PDL and high diffraction efficiency. The target values are a dispersion power of over 0.3 degree per nanometer, diffraction efficiency of more than −1 dB, and PDL of less than 1 dB. The paper is organized as follows. In section 2, we cover the principles of a low PDL IG with high diffraction efficiency. In section 3, we describe the design and fabrication of IGs. In section 4, we present the experimental results and discuss the characteristics of IG prototypes, and the paper is concluded in section 5.
2. Low PDL IG structure
Figures 2(a) and 2(b) show the electrical field intensity distribution in and above the cross-section of a conventional Si IG structure at a wavelength of 1.550 μm. The grating shape is rectangular because it is easy to design and fabricate. Parameters of the Si grating profile are pitch, depth, and width. A pitch of 0.29 μm and the Littrow configuration result in a dispersion power of 0.34 deg./nm in first-order diffraction. The width is 0.15 μm because of the easiness of fabrication for width and depth. The depth is designed so as to attain the highest diffraction efficiency for TE-polarized light and 0.26 μm, which is the shallowest. The intensity of TE polarization is high in the Si grating structures [Fig. 2(a)], whereas the intensity of TM polarization is high between them [Fig. 2(b)]. The large difference in the distribution between TE and TM polarizations makes it difficult to lower the PDL. To reduce the PDL of an IG with high diffraction efficiency, we propose a TIR-based IG coated with dielectric film whose refractive index (n2) is close to that of the base material (n1) [Fig. 3]. Figures 2(c) and 2(d) show the electrical field intensity distribution in and above the cross-section of a Si IG coated with dielectric film (n2 = 2.2). The high-intensity region of a Si IG [Fig. 2(a)] and a Si IG coated with dielectric film [Fig. 2(c)] for TE polarization is the same. In contrast, the distributions of electrical field intensity in Figs. 2(b) and 2(d) are largely different for TM polarization. By changing the refractive index (n2), the distributions of electrical field intensity and the diffraction efficiency for TM polarization can be designed without changing those for TE polarization. We calculated the PDL of the Si IG coated with various dielectric films by DiffractMOD [9], which uses a rigorous coupled-wave analysis (RCWA). The domain for calculating the diffraction efficiency and PDL was determined by using the same scheme as in Fig. 2, because of the absence of absorption in bulk Si in the C band. The parameters are pitch, depth, and width, the same as for the Si IG in Fig. 2.
Fig. 2 Electrical field intensity distribution in and above the cross-section of the Si IG for (a) TE and (b) TM and of the Si IG coated with dielectric film for (c) TE and (d) TM. The white line is the boundary between Si grating and air.
Figure 4 shows the relationship between PDL and the refractive index ratio (n2/n1). The green line is the boundary of the TIR condition. The higher the ratio is, the lower the PDL becomes. PDL of less than 1 dB can be achieved when the ratio of the refractive index ranges from 0.54 to 0.74. For a Si IG (n1 = 3.5), the refractive index of dielectric film should be within the range of 1.9 to 2.6. Figures 5(a) and 5(b) show the diffraction efficiency maps as the function of the Si IG profile, depth and width, for TE and TM polarizations, respectively. It is necessary to choose the Si IG’s profile that makes TE and TM polarizations diffraction efficiency the same in order to minimize PDL. The regions of a diffraction efficiency higher than −1 dB in both TE and TM polarizations is over 0.5 μm in depth [10], which make the Si grating very difficult to fabricate. Figures 5(c) and 5(d) show the diffraction efficiency maps of the Si IG coated with dielectric film (n2 = 2.2). The form of the map of TM polarization [Fig. 5(d)] changes to become close to TE’s one. The regions of high diffraction efficiency of more than −1 dB in both TE and TM polarizations appear in a shallow depth area, which makes it easy to fabricate the Si grating. Therefore, a TIR-based Si IG coated with dielectric film has both high diffraction efficiency and low PDL with a high diffraction power.
Fig. 4 PDL vs. refractive index ratio (orange line). The green line is the boundary of the TIR condition.
Fig. 5 Diffraction efficiency maps of the Si IG for (a) TE and (b) TM and the Si IG coated with dielectric film for (c) TE and (d) TM.
3. Grating structure design and fabrication
We use Si as the base material due to its high refractive index and silicon nitride (SiN, n = 2.2) as the coating material to provide 0.63 as the suitable ratio of the refractive index (SiN/Si). We designed three samples (A, B, and C), which have dispersion powers of 0.34, 0.42, and 0.48 deg./nm, respectively. The IGs consist of a triangular prism, the size of which was determined by the size of grating plane of 20 x 10 mm2 (vertical and parallel axes for grating lines). The relationships among the pitch of the grating and angle of the prism between the grating plane and the launch plane were determined from the dispersion power of the first-order diffraction angle. The grating shapes are rectangular. The width and depth that would achieve both the PDL < 1 dB and diffraction efficiency > −1 dB in the C band were determined as follows. Figures 6(a) and 6(b) show the region of more than −1-dB diffraction efficiency in sample A for TE and TM polarizations. The red and blue are regions for wavelengths of 1.530 and 1.565 μm, respectively. Because the regions are continuous for the wavelengths range, the overlap region of both the shortest and longest wavelengths covers the full C-band wavelength range. Figure 6(c) shows the overlap region. By selecting the depth and width in the overlap region, diffraction efficiency of −1 dB and PDL of less than 1 dB both in the C band are achieved. Table 1 shows the designed values of the Si grating profile for the three samples. Table 2 shows the expected diffraction efficiency and PDL of each sample. The fabrication tolerances for diffraction efficiency of −1 dB are −20 to + 10 μm in depth and between ± 0.3 μm in width. Both tolerances were fairly similar to the order of the fabrication repeatability.
Fig. 6 The region of less than −1 dB diffraction efficiency at wavelengths of 1.530 and 1.565 μm for (a) TE and (b) TM polarizations. (c) The overlap region of (a) and (b).
Table 1. Designed Grating Profiles
Table 2. Expected Diffraction Efficiency (TE) and PDL of the Designed Gratings
To verify the effect of the dielectric coating, we fabricated three additional IGs having the same profile without dielectric film (samples D, E, and F). Our designed TIR-based Si IGs can be fabricated by common fabrication processes. The Si grating was fabricated by electron beam lithography and dry etching of double-side polished wafers. For samples A, B, and C, the SiN layer was deposited by low-pressure chemical vapor deposition after dicing into chips. Figure 7(a) shows a photograph of the grating structure of sample A taken by transmission electron microscopy (TEM). The grating is trapezoidal and shallower than the designed profile. Figure 7(b) shows a model of the trapezoidal grating that has parameters of upper width (width1), lower width (width2), and depth. Table 3 shows the values for the trapezoidal grating profile in samples A, B, and C. The profiles of samples D, E, and F are the same as those of samples A, B, and C because they were fabricated by the same process, respectively. The grating and prism were directly bonded to each other [8, 11]. The prism surfaces for an incident and launch plane were covered with AR coating for the C band. Figure 8 shows the fabricated IG.
Fig. 7 (a) TEM image of the Si grating coated with SiN (sample A) and (b) the model of trapezoidal grating.
Table 3. Fabricated Grating Profiles
4. Experimental results and discussion
Figure 9 shows the output profile of the optical intensity for sample A. The main lobe profiles are the same as the reference and there is no dependence of wavelength and polarization in the profile. It is considered that the side peaks are caused by the grating aperture and could be reduced by decreasing the beam size. Figure 10 shows results of the launch angle for wavelength in the C band. For samples A, B, and C, the dispersion powers, which are equal to the gradients of the launch angle were calculated from the results to be 0.34, 0.42, 0.48 deg./nm. They agree with the designed values, which is attributed to the absence of error in pitch. The dispersion powers of samples D, E and F were also the same as those of samples A, B, and C, respectively. Table 4 shows the diffraction efficiency of TE polarization and the PDL of all samples. The diffraction efficiency of sample A is −0.36 dB and the PDL is 0.61 dB. In contrast, sample D, which has the same profile as sample A without the dielectric film, has a diffraction efficiency of −0.36 dB and a PDL of 22.8 dB which is much worse than sample A’s. These results achieve our target value. The results for other samples also demonstrate the advantage of the dielectric coating.
Fig. 9 Output profiles of optical intensity of three wavelengths for (a) TE and (b) TM polarizations (sample A). The reference (black) shows the output profile without samples and contains both TE and TM.
Fig. 10 Launch angle vs. wavelength. The circles are experimental results and lines are designed values. Black, dark gray, gray show sample A, B, and C, respectively.
Table 4. Diffraction Efficiency and PDL of the Fabricated Gratings
The measured PDL of samples A, B and C is larger than the expected one shown in Table 2. These differences are due to errors in the grating profile. Figure 11 shows the diffraction efficiency of TE (green) and TM (blue) polarizations in the C band. The circles are the experimental results, and the dashed and solid lines are the simulation results using the designed profile [Table 1] and the actual profile [Table 3], respectively. For the six samples,the measured diffraction efficiency for TE and TM polarizations agree with simulation results using the actual profile (solid line). Figure 12 compares simulation results using the grating shape of the trapezoidal model (solid line) and rectangular model (dashed line). The diffraction efficiency for sample A and the experimental results are also plotted as circles. The trapezoidal model uses the actual shape shown in Table 3. The rectangular model uses the mean value of width1 and width2 as the width. The results for the two models are almost the same. As shown in Fig. 6(c), the overlap region of less than −1-dB diffraction efficiency has a wide tolerance in width. Therefore, it is considered that the profile error in depth mainly causes the PDL degradation. By optimizing the etching condition to minimize depth error, the PDL can be reduced regardless of the dispersion power.
Fig. 11 Diffraction efficiency vs. wavelength in the C band. (a), (c) and (e) data for samples A, B and C, respectively. (b), (d) and (f) data for samples D, E, and F. The circles are the measurement results, and the dashed and solid lines are the simulation results for the designed grating model and the actual grating model, respectively. Green and blue show TE and TM polarizations, respectively.
Fig. 12 Comparison between rectangular and trapezoidal structure in diffraction efficiency of sample A. The dashed-dotted and solid lines are the simulation results for the rectangular grating model and trapezoidal grating model, respectively. Green and blue show TE and TM polarizations, respectively.
5. Conclusion
We developed a new type of silicon immersion grating structure that is coated with dielectric film. By changing the refractive index of dielectric film, the diffraction efficiency of TM polarization can be controlled without changing that of TE polarization. This provides both high diffraction efficiency and low PDL. When the refractive-index ratio ranges from 0.54 to 0.74, PDL is expected to be less than 1 dB. We found silicon nitride to be a suitable material for obtaining a low PDL when the base material is silicon. We designed three IGs with different dispersion powers of 0.34 to 0.48 deg./nm. Simulation results show the characteristics of one of the three IGs are diffraction efficiency of −0.4 dB and PDL of 0.2 dB. A prototype immersion grating achieved high dispersion power of 0.34 degrees per nanometer, high diffraction efficiency of −0.4 dB, and low PDL of 0.6 dB. The PDL differences between the measured and the designed values are due to error in the depth of the grating. By optimizing the etching condition, the PDL can be reduced. We expect the new IG structure to be useful in optical telecommunications devices for WDM networks.
References and links
1. K. Suzuki, T. Mizuno, M. Oguma, T. Shibata, H. Takahashi, Y. Hibino, and A. Himeno, “Low loss fully reconfigurable wavelength-selective optical 1×N switch based on transversal filter configuration using silica-based planar lightwave circuit,” IEEE Photon. Technol. Lett. 15, 138–140 (2003).
2. Y. Ishii, K. Hadama, J. Yamaguchi, Y. Kawajiri, E. Hashimoto, T. Matsuura, and F. Shimokawa, “MEMS-based 1x43 Wavelength-Selective Switch with Flat Passband,” in Proceedings of 35th European Conference on Optical Communication, (Vienna, Austria, 2009), paper PD 1.9.
3. G. Wiedemann and D. E. Jennings, “Immersion grating for infrared astronomy,” Appl. Opt. 32(7), 1176–1178 (1993). [CrossRef] [PubMed]
4. P. J. Kuzmenko, D. R. Ciarlo, and C. G. Stevens, “Fabrication and testing of a silicon immersion grating for infrared spectroscopy,” Proc. SPIE 2266, 566–577 (1994). [CrossRef]
5. J. P. Marsh, D. J. Mar, and D. T. Jaffe, “Production and evaluation of silicon immersion gratings for infrared astronomy,” Appl. Opt. 46(17), 3400–3416 (2007). [CrossRef] [PubMed]
6. E. Popov, J. Hoose, B. Frankel, C. Keast, M. Fritze, T. Y. Fan, D. Yost, and S. Rabe, “Low polarization dependent diffraction grating for wavelength demultimlexing,” Opt. Express 12(2), 269–275 (2004). [CrossRef] [PubMed]
7. J. R. Marciante and D. H. Raguin, “High-efficiency, high-dispersion diffraction gratings based on total internal reflection,” Opt. Lett. 29(6), 542–544 (2004). [CrossRef] [PubMed]
8. Y. Ikeda, N. Kobayashi, H. Terada, A. Shibayama, A. Ozawa, C. Yasui, S. Kondo, T. S. Pyo, and H. Kawakita, “High-efficiency silicon immersion grating by electron-beam lithography,” Proc. SPIE 7014, 701469, 701469-12 (2008). [CrossRef]
9. Rsoft Design Group, “DiffractMOD”, http://www.rsoftdesign.com.
10. S. Wang, C. Zhou, Y. Zhang, and H. Ru, “Deep-etched high-density fused-silica transmission gratings with high efficiency at a wavelength of 1550 nm,” Appl. Opt. 45(12), 2567–2571 (2006). [CrossRef] [PubMed]
11. M. Shimbo, K. Furukawa, K. Fukuda, and K. Tanzawa, “Silicon-to-silicon direct bonding method,” J. Appl. Phys. 60(8), 2987–2989 (1986). [CrossRef]
