Optical waveguide demultiplexer was designed and fabricated by integration of two types of gratings, namely, guided-mode-selective focusing grating couplers (GMS-FGCs) and different-guided-mode-coupling distributed Bragg reflectors (DGM-DBRs) in a slab waveguide for constructing a wavelength-division-multiplexing chip-to-chip optical interconnecting board. In the waveguide demultiplexer, guided signal waves were separated wavelength-selectively by DGM-DBRs, and coupled out by GMS-FGCs to focused free space waves. Two-channel demultiplexing with 5-nm-wavelength spacing was demonstrated at around 850-nm wavelength.
©2004 Optical Society of America
Clock frequencies in LSI chips have been increasing steadily, while data rates in electrical interconnects between boards and chips are approaching the limits of their performance. The increase of the data rates in the interconnects is one of the most serious problems in constructing future ultrahigh-speed signal-processing units, which is called the pin-bottleneck problem. Optical interconnects are paid attention as strong candidates for solving the pin-bottleneck problem, and have been investigated in board-to-board, chip-to-chip, and intra-chip communications. It has been expected that intra-board chip-to-chip optical interconnects would be needed in several years . Most techniques proposed for the chip-to-chip optical interconnects so far [2–6] have utilized free space or multimode waveguides as optical signal paths. These techniques may provide high connecting efficiency with large fabrication tolerance, but two-dimensional (2-D) massive transmission of optical signals needs bulky components such as micro-mirrors, micro-prisms, and micro-lenses, resulting in little compatibility with current heat radiation architectures as well as mass-production technologies. On the other hand, we have proposed and investigated an optical waveguide add-drop multiplexing device in order to realize wavelength-division-multiplexing (WDM) ultra-broadband chip-to-chip optical interconnects [7–9]. The configuration utilizing WDM technique enables a massive signal transmission from a 2-D vertical-cavity surface-emitting laser (VCSEL) array to a 2-D photodiode (PD) array through a thin film waveguide with integrated-optic components suitable for planar fabrication processes. For example, 512 transmission channels will be provided within 10-mm width by combination of 64 waveguide channels and 8-channel WDM . Figure 1 illustrates a schematic view of a part of the proposed interconnects. Electronic chips with optoelectronic interposer integrating 2-D VCSEL array and 2-D PD array are surface-mounted on an optical waveguide integrated in a board. The proposed waveguide device consists of two types of gratings, namely, guided-mode-selective focusing grating couplers (GMS-FGCs)  and different-guided-mode-coupling distributed Bragg reflectors (DGM-DBRs), and TE0 and TE1 modes are utilized as a signal propagation mode and an input/output mode, respectively. A free space wave diverging from VCSEL is coupled to TE1 mode in the waveguide by GMS-FGC, and TE1 mode is contra-directionally coupled to TE0 mode. A propagated TE0 mode is wavelength-selectively coupled to TE1 mode by another DGM-DBR, and TE1 mode is coupled out to a free space wave focusing to PD by another GMS-FGC. In this letter, we present first fabrication of an optical waveguide demultiplexer from guided waves to free space waves, and demonstrate two-channel demultiplexing with 5-nm spacing at around 850-nm wavelength.
2. Device configuration
A schematic view of the designed two-channel demultiplexer at around 850-nm wavelength is illustrated in Fig. 2. The demultiplexer consists of two pairs of GMS-FGCs and DGM-DBRs. Propagated TE0 mode (signal propagation mode) at wavelength of λ1 is not coupled out but passes through GMS-FGC1, and is contra-directionally coupled by DGM-DBR1 to TE1 mode (output mode). The TE1 mode is coupled out by GMS-FGC1 to a focused free space wave. Propagated TE0 mode at wavelength of λ2 passes through GMS-FGC1, DGM-DBR1, and GMS-FGC2, and is contra-directionally coupled by DGM-DBR2 to TE1 mode. The TE1 mode is coupled out by GMS-FGC2 to another focused free space wave. A cross-sectional structure and a refractive index profile of the designed waveguide device are illustrated in Fig. 3. The waveguide consists of a Ge-SiO2 main-guiding core (refractive index of 1.53, thickness of 0.7 µm), a SiO2 sub-guiding core (refractive index of 1.46, thickness of 1.26 µm), and a Au high-reflection layer (thickness of 0.1 µm) on a glass substrate. A Si-N grating layer (refractive index of 2.01, thickness of 0.04 µm) for GMS-FGCs and an electron-beam (EB) resist grating layer (refractive index of 1.55, thickness of 0.1 µm) for DGM-DBRs are formed in the middle of the SiO2 sub-guiding core and on the Ge-SiO2 main-guiding core, respectively.
The device characteristics were predicted theoretically by using the conventional coupled-mode analysis. The theoretical estimation of the device performances is similar to that of our previous study [7, 11]. Calculated electric-field profiles of TE0 and TE1 modes are also shown in Fig. 3. TE0 mode is mainly confined in the main-guiding core by total internal reflections at the upper boundary of the EB-resist grating layer and at the lower boundary of the main-guiding core. TE1 mode is confined in the whole of the waveguide by total internal reflection at the upper boundary of the EB-resist grating layer and by high reflection at the lower boundary of the sub-guiding core. The propagation losses due to substrate radiation of TE0 and TE1 modes were calculated to be 1.6 dB/cm and 8.8 dB/cm, respectively. Two GMS-FGCs are formed in the Si-N grating layer. The distance between GMS-FGC1 and GMS-FGC2 was 2 mm. Grating line patterns of GMS-FGCs were curved with curvature varied from 1.30 mm-1 to 1.85 mm-1 and chirped in period from 0.53 µm to 0.63 µm. An aperture, a focal length, and an output angle of the GMS-FGCs were 0.1 mm×0.3 mm, 0.4 mm, and 1.7 deg, respectively. Radiation decay factors of the GMS-FGC for TE0 and TE1 modes were calculated to be 1.2 mm-1 and 17 mm-1, respectively. Such a large difference in the radiation decay factor results from the difference in the electric field at the Si-N grating layer between TE0 and TE1 modes. The Au high-reflection layer contributes to suppress the substrate-radiation mode and enhance the radiation decay factor for TE1 mode. The output-coupling efficiencies of the GMS-FGCs were calculated to be 22 % and 97 % for TE0 and TE1 modes, respectively. Grating patterns of DGM-DBRs were formed in the EB-resist grating layer with the gap of 0.03 mm from the GMS-FGC. The effective refractive indices of TE0 and TE1 modes at the DGM-DBRs were calculated to be 1.494 and 1.469, respectively. Grating period of the DGM-DBR for 850-nm wavelength was calculated to be 286.9 nm. A coupling factor of the DGM-DBR between TE0 and TE1 modes was calculated to be 9.5 mm-1. The coupling efficiency and full-width at half maximum (FWHM) of wavelength-selectivity were calculated to be 99.97 % and 2 nm, respectively, with the grating length of 0.5 mm. The output efficiencies of the first and second pairs of GMS-FGCs and DGM-DBRs were calculated to be 66 % and 48 %, respectively, at the corresponding wavelength by using the coupling efficiencies of GMS-FGCs and DGM-DBRs and the propagation losses of TE0 and TE1 modes. The FWHM of the focus spot size was calculated to be 3 µm×1 µm for each output wave.
A Cr contact layer and a Au high reflection layer were deposited sequentially by thermal evaporation on a glass substrate (AN100 provided by Asahi Glass Co., Ltd.). A SiO2 first sub-guiding core was deposited by plasma enhanced chemical vapor deposition (PECVD) using Si(OC2H5)4 as a precursor. A Si-N grating layer was deposited by reactive DC sputtering with Si targets and N2 gas. An EB resist (Nippon Zeon, ZEP-520) of 0.4-µm thickness was spin-coated on the Si-N grating layer. Two GMS-FGC patterns with 2-mm space were written directly by an EB scanning. After developing, the grating patterns were transferred to the Si-N layer by reactive ion etching using C3F8 gas. A SiO2 second sub-guiding core and a Ge-SiO2 main-guiding core were sequentially deposited by the PECVD. The surface of the SiO2 second sub-guiding core on the GMS-FGC patterns should be flattened in order to suppress excess loss for the propagating TE0 mode. Therefore, the SiO2 second sub-guiding core was deposited with 1/3 operating pressure of the normal deposition condition, though a deposition rate was reduced to 1/10 of the normal deposition . In the deposition of the Ge-SiO2 main-guiding layer, Ge(OCH3)4 was used as a precursor for GeO2. The flow ratio of Si(OC2H5)4 and Ge(OCH3)4 was 1:1. An EB-resist grating layer was spin-coated on the Ge-SiO2 main-guiding core. DGM-DBR patterns of 0.5 mm×0.5 mm aperture with 287.6-nm and 286.0-nm periods were written directly by an EB scanning. Figure 4 shows an optical microscopic photograph and scanning electron microscope (SEM) photographs of the fabricated GMS-FGC and DGM-DBR. The SEM photographs of the GMS-FGC show the grating pattern before the deposition of the SiO2 second sub-guiding core. The GMS-FGC and the DGM-DBR could be integrated at the designed position. The gratings with line/space ratio of 1:1 and rectangular cross-section could be obtained for both GMS-FGC and DGM-DBR patterns. A GC of 1.45-µm period and 0.5 mm×0.3 mm aperture was also fabricated in the EB-resist grating layer by an EB direct writing for exciting TE0 mode into the waveguide. The distance between the GC and GMS-FGC1 was 10 mm, as shown in Fig. 2.
4. Characterization of optical properties
Optical experiments were carried out by using a wavelength-tunable laser diode. TE0 and TE1 modes were separately excited with different incidence angle by the GC. Effective refractive indices of TE0 and TE1 modes were measured to be 1.496 and 1.470, respectively, from the incidence angle. These values agree well with the theoretical values within our measurement accuracy. The coupling wavelengths for DGM-DBR1 of 287.6-nm period and DGM-DBR2 of 286.0-nm period were estimated to be 853.0 nm and 848.3 nm, respectively. A propagation loss for TE0 mode was measured to be 7 dB/cm from an intensity decay of scattering light from the waveguide. One of the reasons why the propagation loss was larger than the predicted value of 1.6 dB/cm would be a scattering at the boundary between the main-guiding and the sub-guiding cores as well as a scattering and an absorption in the Si-N grating layer. Figure 5(a) and Fig. 5(b) show photographs of the top views of the fabricated device excited with TE0 mode at 854-nm (λ1) and 849-nm (λ2) wavelengths, respectively. The guided wave propagated from left-hand side to right-hand side. A bright spot seen at the right-hand side in Fig. 5(a) was a focusing output wave coupled out by combination of GMS-FGC1 and DGM-DBR1. A guided wave passing through GMS-FGC1 and DGM-DBR1 could not be observed in Fig. 5(a). In the case of Fig. 5(b), on the other hand, most guided wave passed through GMS-FGC1 and DGM-DBR1 and was coupled out by combination of GMS-FGC2 and DGM-DBR2 to be a focusing wave, while a part of the guided wave was directly radiated by GMS-FGC1 to be a diverging wave. Figure 6 shows the wavelength dependence of the output efficiencies for two pairs of GMS-FGCs and DGM-DBRs. The output efficiency was estimated as the output power against the propagated power in just front of GMS-FGC1. In the estimation, we assumed that there were no excess losses such as scattering loss or mode-conversion loss. The maximum output efficiencies for the first and the second pairs of GMS-FGCs and DGM-DBRs were estimated to be about 40 % at 853.2-nm wavelength and about 27 % at 848.3-nm wavelength, respectively. The FWHM of wavelength-selective widths were estimated to be 4 nm and 3 nm for the first and the second grating pairs, respectively.
Although the experimentally obtained efficiencies were smaller than the theoretically predicted values, coupling wavelengths agree well with the predicted values for both the grating pairs. The discrepancy between the theoretical and experimental values in output efficiency and wavelength selectivity would be caused primarily by our insufficient fabrication processes. More precise clarification of the causes for the discrepancy is currently under study. The FWHM of the output spot size was measured to be 3 µm×2 µm each, which was close to the diffraction-limited value within our measurement accuracy.
We have fabricated an optical waveguide demultiplexer for two-wavelength channels from guided waves to free space waves. The waveguide demultiplexer consisted of two types of gratings, i.e. GMS-FGCs and DGM-DBRs. The wavelength demultiplexing with 5-nm spacing was experimentally demonstrated. The focus spot sizes close to the diffraction-limited value were obtained for both output waves. Experimental work is being continued to improve device performances such as output efficiency, wavelength selectivity and propagation loss. An add-drop multiplexing device is also under study.
This research work was done as a part of a collaboration research project “Development of Innovative Designing/Manufacturing Process and Creation of New Opto-electronic System with Ultra High Performance and Integration” in Japan. This work was financially supported in part by International Communications Foundation in Japan, and in part by a Grant-in-Aid for Scientific Research (A) No. 15206008 of Japan Society for the Promotion of Science.
References and links
1. N. Savage, “Linking with light,” IEEE Spectrum 39, 8, 32–36 (2002). [CrossRef]
2. S. Voigt, S. Kufner, M. Kufner, and I. Frese, “A refractive free-space microoptical 4×4 interconnect on chip level with optical fan-out fabricated by the LIGA technique,” Photon. Technol. Lett. 14, 1484–1486 (2002). [CrossRef]
3. D. V. Plant, M. B. Venditti, E. Laprise, J. Faucher, K. Razavi, M. Châteauneuf, A. G. Kirk, and J.S. Ahearn, “256-channel bi-directional optical interconnect using VCSELs and photodiodes on CMOS,” J. Lightwave Technol. 19, 1093–1103 (2001). [CrossRef]
4. H. Sasaki, K. Kotani, H. Wada, T. Takamori, and T. Ushikubo, “Scalability analysis of diffractive optical element-based free-space photonic circuits for interoptoelectronic chip interconnections,” Appl. Opt. 40, 1843–1855 (2001). [CrossRef]
5. H. Takahara, “Optoelectronic multichip module packaging technologies and optical input/output interface chip-level packages for the next generation of hardware systems,” IEEE J. Sel. Top. Quantum Electron. 9, 443–451 (2003). [CrossRef]
6. C. Debaes, M. Vervaeke, V. Baukens, H. Ottevaere, P. Vynck, P. tuteleers, B. Volckaerts, W. Meeus, M. Brunfaut, J. V. Campenhout, A. Hermanne, and H. Thienpont, “Low-cost microoptical modules for MCM level optical interconnections,” IEEE J. Sel. Top. Quantum Electron. 9, 518–530 (2003). [CrossRef]
7. S. Ura, “Selective guided mode coupling via bridging mode by integrated gratings for intraboard optical interconnects,” in Optoelectronic Interconnects, Integrated Circuits and Packaging, L.A. Eldada, R. A. Heyler, and J. R. Rowlette, eds., Proc. SPIE4652, 86–96 (2002).
8. S. Ura, R. Nishida, and H. Nishihara, “Waveguide gratings for optical add-drop multiplexing of free space waves and guided waves,” in Tech. Dig. Integrated Photonics Research, Vol. 58 of OSA Proceedings Series (Optical Society of America, Washington, D.C., 2001), pp. ITuG3-1–ITuG3-3.
9. S. Ura, K. Kintaka, R. Satoh, and H. Nishihara, “Integrated-optic wavelength-division-add/drop device for broad-band chip-to-chip optical interconnects,” in Proc. 16th Ann. Meet. IEEE Lasers & Electro-Optics Society, (Institute of Electrical and Electronics Engineers, New York, 2003), pp. 177–178. [CrossRef]
10. K. Kintaka, J. Nishii, Y. Imaoka, J. Ohmori, S. Ura, R. Satoh, and H. Nishihara, “A guided-mode-selective focusing grating coupler,” IEEE Photon. Technol. Lett. 16, 512–514 (2004). [CrossRef]
11. Y. Imaoka, J. Ohmori, M. Hamada, S. Ura, K. Kintaka, R. Satoh, and H. Nishihara, “Guided-mode-selective focusing grating coupler for optical add/drop multiplexing of free-space wave with guided waves,” in Tech. Dig. 9th Microoptics Conf., (The Japan Society of Applied Physics, Tokyo, 2003), pp. 376–379.