A novel high-efficiency silicon-chip-to-fiber grating coupler is investigated here. By introducing a dual layer grating structure with an inter-layer lateral shift to mimic 45° tilted mirror behavior, perfectly vertical coupling is successfully demonstrated. Our numerical results show that a peak silicon-chip-to-fiber coupling efficiency about 70% is possible near 1550 nm. Meanwhile, for the entire telecom C-band, i.e. wavelengths from 1530 nm to 1565 nm, the coupling efficiency is > 50% and the back reflection is less than < 1%. Our proposed high-performance silicon perfectly vertical coupling structure is suitable for interfacing with multi-core fiber platform, which may play an important role in the future CMOS photonic integration technology.
© 2015 Optical Society of America
In recent years, Silicon-on-Insulator (SOI) or CMOS photonic platform has been considered as one of the most promising platforms for future dense photonic integrated circuits (PICs)  because of its high-index contrast and compatibility with CMOS fabrication processes. The SOI optical interconnection technology enables tremendous optical communication bandwidth for on-chip applications. However, utilizing such bandwidth for off-chip applications requires high-performance and scalable chip-to-fiber coupling technology. A possible candidate for such technology will involve the use of grating coupler in conjunction with the latest multi-core fiber platform [2–4]. In the past, much effort has already been devoted to the design and optimization of grating couplers [5–12] for coupling into and out-of tilted single mode fiber. The slight tilt (usually ~10°) of the fiber axis with respect to the normal direction of the SOI chip is a usual practice in order to minimize back-reflection into the SOI bus waveguides caused by the grating coupler’s second-order Bragg reflection. Under such tilted configuration, multiple strategies have been developed to break out-of-plane scattering symmetry for the upward and downward directions in order to increase the coupling directionality. These include the use of a silicon or poly-silicon overlay layer [10,13], a metal reflecting substrate, a bottom distributed Bragg reflection grating mirror [7,11], or slanted gratings . In addition, there are also attempts to apodize the grating structures to design scattering patterns [8,9,15] in order to match the mode profile of the single mode fiber for better coupling efficiency.
However, so far little effort has been devoted to develop perfectly vertical grating coupling technology, which could have great potentials in large-scale high-density-integration CMOS photonic technology. For example, future CMOS photonic chips may require a large number of fiber ports to communicate with off-chip components, systems and networks . A traditional fiber bundle with multiple closely packed single-mode fibers mounted on V-grooves with a typical fiber-to-fiber spacing ~125μm (cladding diameter of a typical single-mode fiber) can in principle serve as fiber I/O ports as demonstrated by Luxtera . However, the emergence of multi-core fibers may provide a much compact solution by having much less space per physical channel or core [2–4]. The fiber industry can easily produce a fiber with a cladding diameter of 130μm and seven single-mode cores arranged in a hexagon pattern with core-to-core spacing only about 38μm . Such kind of fiber may not work well with tilted grating couplers, because even if the tilted angle is as small as 10° (a typical value found in many literatures), the vertical position difference among all the cores with respect to the surface of the chip can be as large as 12μm, which suggests that an optimized grating coupler structure working best for one core may not work well for the rest. Considering this factor, an efficient perfectly vertical grating coupler with low back-reflection is highly desired.
Using perfectly vertical coupler is also very advantageous for packaging. For example, Fig. 1(a) shows our conceptual potential packaging scheme. Compared with the case of tilted grating coupler which requires a tilted optical pathway with precise tilt angle control, a perfectly vertical optical via through an oxide layer can be easily fabricated by standard CMOS process, through which a cleaved multi-core fiber with a flat facet can then be inserted vertically and fixed with UV-curable adhesives after alignment. In this regard, the use of multi-core fiber and vertical coupling mechanism can become a natural choice in future SOI photonic packaging technology. In order to achieve perfectly vertical coupling, Roelkens et al. introduced a grating structure with an extra slit in the SOI bus waveguide to block the second-order back reflection wave . However, the wavelength of operation with low back reflection is limited. Covey and Chen proposed a grating structure for coupling between a multi-slotted waveguide and a vertical fiber . In this configuration, a distributed Bragg reflection grating (DBR) is needed to break the in-plane symmetry in order to achieve high directionality and thus coupling efficiency. The use of DBR may increase design complexity and device footprint. Here, we propose an efficient perfectly vertical silicon-chip-to-fiber coupling structure, which consists of two layers of silicon gratings with an inter-layer lateral shift to mimic 45° tilted mirror reflection behavior. Our analysis shows that >78% of the power in silicon waveguides can be directed upwards for the entire wavelength range from 1500 nm to 1600 nm in our simulations. A peak coupling efficiency ~70% from a silicon waveguide into a perfectly vertical fiber at a wavelength of ~1550 nm can be achieved. The results we demonstrate here may find great applications in future photonic integration technology.
2. Device model and principle
Figure 1(b) shows our proposed perfectly vertical silicon chip-to-fiber coupling structure. Compared with the typical mono-layer grating structure , our proposed structure consists of two vertically stacked layers of fully etched gratings. Each of the grating layers has a height of h and an etched width of d. The two layers of gratings have an inter-layer lateral shift as shown in Fig. 1(b). This lateral shift can be characterized by an effective tilted angle defined as . In order to have a perfectly vertical coupling into upward direction, i.e., to redirect the input optical wave in the silicon bus waveguide from 0° to 90°, the effective tilt angle is set to be 45° to mimic a mirror reflection behavior as shown in Fig. 2(a). Note that a traditional blazed grating can also serve the same purpose as indicated in several theoretical reports [20,21]. Our silicon grating coupler structure is, however, different and has several advantages over a traditional blazed grating, which could include preservation of surface planarity and easy effective tilt angle tuning by changing only the lateral inter-layer shift. Our grating structure has broken symmetries for both in-plane and vertical directions, which give rise to a high coupling directionality (defined as the ratio between the optical power coupled upwards and the total input optical power in the bus waveguide) and a low back reflection into the bus waveguide. As shown in Fig. 2 (a-c), scattered optical fields in the upward, downward and backward directions can be described as the coherent sum of two individual fields originated from the two different scattering channels marked by red and blue arrows. If the horizontal and vertical phase shifts are defined as and respectively, the optical fields in the upward, downward and backward direction are proportional to, and respectively. Here and are the corresponding amplitudes of the scattered optical field from the two scattering channels. Clearly when and are designed to have identical value of , the scattered light will interfere constructively in the upward direction but destructively in the downward as well as in the backward direction, leading to a very high coupling directionality. Next, in order to achieve a high chip-to-fiber coupling efficiency, the wave-front of the scattered field is matched to the flat horizontal wave-front of the optical mode in the vertical fiber by carefully choosing the grating period so that the phase difference between two successive scattering paths as shown in Fig. 2(d) equals to .
3. Device performance and simulation analysis
Our proposed coupling structure shown in Fig. 1 is investigated in a commercial 3D Finite-difference Time-domain (FDTD) solver (FDTD Solutions from Lumerical Solutions, Inc). In all the simulations, the fiber is vertically placed 250 nm above the top oxide layer. The distance between its center and the left edge of the first grating tooth on the bottom layer is 3.5 μm. The fiber is modeled with a core diameter of 9 μm, core index of 1.44 and NA of 0.13. The thickness of top and buried oxide layer is chosen to be 250 nm and 2000 nm, respectively. A 500 nm thick silicon handle wafer is also included in the simulations. The thickness of each silicon grating layer is h = 150 nm, which makes the height of silicon bus waveguide 2h = 300 nm. Note that the thickness of 300 nm used here is primarily to illustrate the feasibility of our concept, which in principle can be equally applied to a different silicon layer thickness, e.g., 220 nm or 250 nm, a typical silicon layer thickness used by many groups. Since the effective tilt angle is set to 45°, the lateral shift determined by as indicated in Fig. 1(b) is the same as h and equals to 150 nm. There are 22 grating periods for each grating layer in the simulated structure. Since the grating period and duty cycle determine the phase shift ,and , their values are varied to optimize the chip-to-fiber coupling efficiency. Fundamental quasi-TE mode of the silicon waveguide (with electric field perpendicular to the structures shown in Fig. 1(b)) is used as an input. Notice that due to the reciprocity of this coupling structure, the same performance is expected for the fiber-to-chip coupling.
Figure 3(a) shows our obtained coupling efficiency contour at a wavelength of 1550 nm when the grating period is varied from 500 nm to 650 nm and the duty cycle (defined as the ration between the width of unetched section and the grating period) is varied from 65% to 85%. Clearly for most of the grating period and duty cycle values, the coupling efficiency is very low. However, when the grating structure has a period ~565 nm and a duty cycle ~77%, the coupling efficiency can reach to a maximum value nearly 70%. In addition, as shown in Fig. 3(a) high coupling efficiency occurs for grating periods primarily around 565 nm. This observation is in fact consistent with our discussion above in Fig. 2(c), because this grating period gives rise to for an effective index for the silicon slab. Figure 3(b) shows the wavelength-dependent coupling efficiency for coupling into air and into buried oxide for an optimized grating structure with a grating period of 565 nm and a duty cycle of 77% determined from Fig. 3(a). Clearly this unique dual-layer grating structure has a very high coupling directionality, > 78% into the air and < 10% in the buried oxide layer over the entire simulated wavelength range from 1.5 μm to 1.6 μm. It also has a peak coupling efficiency of ~70% at a wavelength about 1550 nm into the fundamental mode of the perfectly vertical fiber as indicated in Fig. 3(c). For the entire telecom C-band from 1530 nm to 1565 nm, our proposed grating structure has a coupling efficiency greater than 50%. Although this grating structure is optimized for coupling efficiency between the silicon waveguide and the perfectly vertical fiber, this structure enables a low reflection into the back-propagating mode in the bus waveguide, less than 1% as suggested by the blue curve in Fig. 3(c) for the entire wavelength window in our simulations. Such high coupling efficiency and low back reflection is in fact consistent with the broken symmetry along the in-plane and vertical direction, as outlined above. Note that our grating structure can in principle also be optimized to achieve very low back-reflection, but a very high coupling efficiency cannot be guarantee simultaneously. In fact there are several studies [22,23] focusing primarily on removing the back-reflection. Extremely low back-reflection < 0.1% is shown to be possible but with a large reduction of the coupling efficiency down to < 30%. A study on reflectionless gratings capable of high coupling efficiency yet extremely low back-reflection is beyond the scope of this investigation.
This dual-layer grating structure’s high directionality and coupling efficiency can be explained by examining its field distributions, which are shown in Fig. 4 for three different wavelengths, 1500 nm, 1550 nm, and 1600 nm with an input light along the x-axis. The scattered field profiles into the upward (air) and downward (buried oxide) directions for these wavelengths shown in Fig. 4(a)-(c) are very similar, which are located mostly in the air, demonstrating a high directionality. In addition, the scattering fields decay as the input light transmits through the grating structures. Such decaying intensity distribution does not match with the Gaussian profile of the fiber’s fundamental mode. Nevertheless, a high coupling efficiency is still possible as indicated in Fig. 3, and the coupling efficiency may be further improved by apodizing the grating profile to optimize the scattered field pattern [8,9,15]. The wavelength dependence of the coupling efficiency here is primarily caused by the changing wave-front as shown in Fig. 4(d)-(f). The best coupling efficiency occurs when the wave-front of scattered field matches with the flat horizontal wave-front for the optical mode in a perfectly vertical fiber, which corresponds to a wavelength of operation at 1550 nm shown in Fig. 4(e). In this case, the propagation phase difference between two successive scattering paths along x-axis as illustrated in Fig. 2(c) is near 2π. However, when the wavelength deviates from 1550 nm, the phase difference moves away from 2π, giving rise to a tilted wave-front and thus the declining coupling efficiency as illustrated by two examples in Fig. 4(d) & (f) for wavelengths of 1500 nm and 1600 nm.
A series of simulations were conducted to study our grating structure’s performance degradation due to the grating parameter (grating period, duty cycle, and lateral shift) deviation possibly existing in the fabrication process. In our simulations, the wavelength is fixed at 1550 nm. Only one grating parameter is varied at a time while others are kept at their optimized values determined from our previous study shown in Fig. 3. When any of the three grating parameters deviate from their optimum values as shown in Fig. 5, the coupling efficiency rolls off from its peak value. Since the grating pattern is transferred from the photomask used in an advanced optical deep UV optical lithography, the grating period deviation is primarily determined by the quality of the photomask, which can easily have sub-10 nm accuracy when produced by e-beam lithography. The deviation of the grating period is thus unlikely to be large. If a 10 nm deviation is considered, according to Fig. 5(a) the coupling efficiency decreases only by about 5%. Similar analysis can be applied to study the impact of the duty cycle variation on the coupling efficiency. It is found that a 10nm width change of the etched section, corresponding to a duty cycle change of 0.018, has negligible impact on the grating coupling efficiency as shown in Fig. 5(b). In addition, if a layer-to-layer alignment accuracy of < 6.5 nm in advanced deep UV lithography  has to be included in the lateral shift, the coupling efficiency drops no more than 10% as seen from Fig. 5(c). All of the above analysis indicates that our device should have good tolerance to fabrication errors caused by geometric patterning and alignment.
It is worth noting that implementing our device in practice may also encounter filling artifacts when the fully etched section in the silicon layer, having a width on the order of 100 nm, needs to be filled with SiO2 in our current design. However, if our concept is applied towards mid-IR silicon photonic applications in the 3-8 μm wavelength range, the filling artifacts may not pose a major fabrication challenge since the width of the etched section can be much large and the filling edge effect diminishes. The filling artifacts can also be completely avoided in an alternative design conceptual similar to our current design, where index changes and thus gratings in the silicon layer are formed by ion implantation [25,26] instead of etching and filling. Besides, concerns may arise from the fact that the top silicon layer needs to be deposited in an amorphous or polycrystalline form, which may introduce additional optical loss. However, several studies have shown that low-loss silicon waveguides are possible in these non-single-crystalline silicon platforms [27–31] and integrated silicon devices may perform better even in some cases . In short, although there are potential fabrication challenges associated with our current design, none of them are totally preventing a possible implementation of conceptually similar devices in future.
A novel grating coupler for light coupling between a SOI waveguide and a perfectly vertical fiber is proposed. It consists of two vertically stacked grating layers with an interlayer lateral shift to mimic a 45° mirror reflection. Both in-plane and vertical symmetry is broken for such grating structure, giving rise to high coupling directionality and low back reflection. Our simulation results demonstrate that this novel coupling structure enables a coupling efficiency > 50% for the entire telecom C-band with a peak coupling efficiency as high as 70% at 1550 nm when the wave-front of the scattering field aligns well with that of the mode of the perfectly vertical fiber. The fabrication complication associated with our grating structure is also investigated and no fundamental limiting issues are found. The coupling structure we demonstrate here may play an important role in future dense photonic IC packaging technology.
We would like to thank the National Science Foundation of China (NSFC) for providing financial support with Grant No. 61378009. And we also thank the financial support from National Basic Research of China with Grant No. 2012CB921503 and No. 2013CB632702.
References and links
1. M. Hochberg and T. Baehr-Jones, “Towards fabless silicon photonics,” Nat. Photonics 4(8), 492–494 (2010). [CrossRef]
2. C. Doerr and T. Taunay, “Silicon Photonics Core-, Wavelength-, and Polarization-Diversity Receiver,” IEEE Photon. Technol. Lett. 23(9), 597–599 (2011). [CrossRef]
3. Y. Ding, F. Ye, C. Peucheret, H. Ou, Y. Miyamoto, and T. Morioka, “On-chip grating coupler array on the SOI platform for fan-in/fan-out of multi-core bers with low insertion lossand crosstalk, ” in ECOC2014, paper We.1.1.3.
4. N. Fontaine, C. Doerr, M. Mestre, R. Ryf, P. Winzer, L. Buhl, Y. Sun, X. Jiang, and R. Lingle, Jr., “Space-division multiplexing and all-optical MIMO demultiplexing using a photonic integrated circuit,” in OFC/NFOEC2012, paper PDP5B.1.
5. Z. Cheng and H. K. Tsang, “Experimental demonstration of polarization-insensitive air-cladding grating couplers for silicon-on-insulator waveguides,” Opt. Lett. 39(7), 2206–2209 (2014). [CrossRef] [PubMed]
6. A. Mekis, S. Gloeckner, G. Masini, A. Narasimha, T. Pinguet, S. Sahni, and P. De Dobbelaere, “A Grating-Coupler-Enabled CMOS Photonics Platform,” IEEE J. Sel. Top. Quantum Electron. 17(3), 597–608 (2011). [CrossRef]
8. M. Antelius, K. B. Gylfason, and H. Sohlström, “An apodized SOI waveguide-to-fiber surface grating coupler for single lithography silicon photonics,” Opt. Express 19(4), 3592–3598 (2011). [CrossRef] [PubMed]
9. X. Chen, C. Li, C. K. Fung, S. M. Lo, and H. K. Tsang, “Apodized waveguide grating couplers for efficient coupling to optical fibers,” IEEE Photon. Technol. Lett. 22(15), 1156–1158 (2010). [CrossRef]
10. D. Vermeulen, S. Selvaraja, P. Verheyen, G. Lepage, W. Bogaerts, P. Absil, D. Van Thourhout, and G. Roelkens, “High-efficiency fiber-to-chip grating couplers realized using an advanced CMOS-compatible Silicon-On-Insulator platform,” Opt. Express 18(17), 18278–18283 (2010). [CrossRef] [PubMed]
11. F. Van Laere, G. Roelkens, M. Ayre, J. Schrauwen, D. Taillaert, D. Van Thourhout, T. F. Krauss, and R. Baets, “Compact and highly efficient grating couplers between optical fiber and nanophotonic waveguides,” J. Lightwave Technol. 25(1), 151–156 (2007). [CrossRef]
12. D. Taillaert, F. Van Laere, M. Ayre, W. Bogaerts, D. Van Thourhout, P. Bienstman, and R. Baets, “Grating couplers for coupling between optical fibers and nanophotonic waveguides,” Jpn. J. Appl. Phys. 45(8A), 6071–6077 (2006). [CrossRef]
14. B. Wang, J. H. Jiang, and G. P. Nordin, “Embedded, slanted grating for vertical coupling between fibers and silicon-on-insulator planar waveguides,” IEEE Photon. Technol. Lett. 17(9), 1884–1886 (2005). [CrossRef]
15. R. Halir, P. Cheben, J. H. Schmid, R. Ma, D. Bedard, S. Janz, D. X. Xu, A. Densmore, J. Lapointe, and I. Molina-Fernández, “Continuously apodized fiber-to-chip surface grating coupler with refractive index engineered subwavelength structure,” Opt. Lett. 35(19), 3243–3245 (2010). [CrossRef] [PubMed]
16. M. Asghari and A. V. Krishnamoorthy, “Silicon photonics: Energy-efficient communication,” Nat. Photonics 5(5), 268–270 (2011). [CrossRef]
17. B. Zhu, T. F. Taunay, M. F. Yan, J. M. Fini, M. Fishteyn, E. M. Monberg, and F. V. Dimarcello, “Seven-core multicore fiber transmissions for passive optical network,” Opt. Express 18(11), 11117–11122 (2010). [CrossRef] [PubMed]
18. G. Roelkens, D. Van Thourhout, and R. Baets, “High efficiency grating coupler between silicon-on-insulator waveguides and perfectly vertical optical fibers,” Opt. Lett. 32(11), 1495–1497 (2007). [CrossRef] [PubMed]
20. K. Li, G. Li, F. Xiao, F. Lu, Z. Wang, and A. Xu, “Unidirectionally optical coupling from free space into silicon waveguide with wide flat-top angular efficiency,” Opt. Express 20(17), 18545–18554 (2012). [CrossRef] [PubMed]
21. K. Li, G. Li, F. Lu, and A. Xu, “Compact, Broadband, and Wide-Angle Optical Coupling for Silicon Waveguide,” IEEE Photon. J. 4(6), 2116–2125 (2012). [CrossRef]
22. Y. Li, L. Li, B. Tian, G. Roelkens, and G. Baets, “Reflectionless Tilted Grating Couplers With Improved Coupling Efficiency Based on a Silicon Overlay,” IEEE Photon. Technol. Lett. 25(13), 1195–1198 (2013). [CrossRef]
23. D. Vermeulen, Y. De Koninck, Y. Li, E. Lambert, W. Bogaerts, R. Baets, and G. Roelkens, “Reflectionless grating couplers for Silicon-on-Insulator photonic integrated circuits,” Opt. Express 20(20), 22278–22283 (2012). [CrossRef] [PubMed]
25. R. Loiacono, G. T. Reed, G. Z. Mashanovich, R. Gwilliam, S. J. Henley, Y. Hu, R. Feldesh, and R. Jones, “Laser erasable implanted gratings for integrated silicon photonics,” Opt. Express 19(11), 10728–10734 (2011). [CrossRef] [PubMed]
26. R. Topley, L. O’Faolain, D. J. Thomson, F. Y. Gardes, G. Z. Mashanovich, and G. T. Reed, “Planar surface implanted diffractive grating couplers in SOI,” Opt. Express 22(1), 1077–1084 (2014). [PubMed]
27. K. Selvaraja, E. Sleeckx, M. Schaekers, W. Bogaerts, D. V. Thourhout, P. Dumon, and R. Baets, “Low-loss amorphous silicon-on-insulator technology for photonic integrated circuitry,” Opt. Commun. 282(9), 1767–1770 (2009). [CrossRef]
28. J. S. Orcutt, S. D. Tang, S. Kramer, K. Mehta, H. Li, V. Stojanović, and R. J. Ram, “Low-loss polysilicon waveguides fabricated in an emulated high-volume electronics process,” Opt. Express 20(7), 7243–7254 (2012). [CrossRef] [PubMed]
29. D. Kwong, J. Covey, A. Hosseini, Y. Zhang, X. Xu, and R. T. Chen, “Ultralow-loss polycrystalline silicon waveguides and high uniformity 1x12 MMI fanout for 3D photonic integration,” Opt. Express 20(19), 21722–21728 (2012). [CrossRef] [PubMed]
30. S. Zhu, G. Q. Lo, and D. L. Kwong, “Low-loss amorphous silicon wire waveguide for integrated photonics: effect of fabrication process and the thermal stability,” Opt. Express 18(24), 25283–25291 (2010). [CrossRef] [PubMed]
31. R. Meade, O. Tehar-Zahav, Z. Sternberg, E. Megged, G. Sandhu, J. S. Orcutt, R. Ram, V. Stojanovic, M. R. Watts, E. Timurdogan, J. Shainline, and M. Popovic, “Integration of silicon photonics in a bulk CMOS memory flow,” in IEEE Optical Interconnects Conference2013, paper WC1. [CrossRef]
32. A. Shastri, M. Webster, G. Jeans, P. Metz, S. Sunder, B. Chattin, B. Dama, and K. Shatri, “Experimental Demonstration of Ultra-Low-Power Single Polarization 56 Gb/s QAM-16 Generation without DAC using CMOS Photonics,” in ECOC2014, paper PD.2.5.