Open Access
Add to BibSonomy
Abstract
We report on the design, fabrication, and evaluation of grating devices, specifically (1) low back-reflection grating couplers and (2) focused Bragg grating mirrors, fabricated on silicon nitride technology platforms for applications in the visible wavelength range. Experiments show that the designed grating couplers exhibit 8 dB-reduced back reflections and a coupling penalty of less than 0.8 dB compared to a conventional grating coupler design. Taking advantage of the 193 nm immersion lithography, focused Bragg gratings are fabricated for the visible wavelengths of 638 nm and 532 nm and over 80% reflectivity is achieved.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Photonic integrated circuits (PICs) based on silicon nitride (SiN) waveguides have received a great interest from both industry and academia in applications ranging from telecom, data-com, to biomedical, and image sensing to name a few [1,2]. PICs based on SiN platform exhibit exquisite features such as compatibility with complementary metal-oxide semiconductor (CMOS) lines, wide transparency from the visible to the mid-IR wavelength range. In fact, propagation loss less than 1 dB/cm in plasma-enhanced chemical vapor deposition (PECVD) at wavelengths up to 900 nm [3], and as low as 0.1 dB/m in low-pressure chemical vapor deposition (LPCVD) waveguide operating at a wavelength of 1580 nm [4] have been demonstrated. Furthermore, SiN has nonlinear and thermo-optic coefficients of 0.28×10−18 m2/W and (2.45 ± 0.09) ×10−5 RIU/°C, respectively, that are lower compared to silicon coefficients, thus enabling high optical power feeding, which is required, for example, in optical phased arrays, and higher tolerance against temperature drifts. Among the photonic integrated components developed and experimentally demonstrated in SiN CMOS-compatible platform, high-performance grating couplers are of crucial importance for providing outstanding interconnectivity between large index contrast waveguides and optical fibers. Therefore, efforts in optimizing grating performance, such as high coupling efficiency, polarization diversity and independency, and reduced back-reflections are needed. SiN grating couplers patterned by using 193 nm optical lithography and partially etched (70–140 nm) by inductive coupled plasma-reactive ion-etch process, have been demonstrated at wavelengths of 532, 780, and 900 nm with a typical efficiency of ∼7 dB/coupler [3].
In this paper, we report for the first time to the best of our knowledge, on our progress on efficient grating devices designed and fabricated at IMEC. Taking advantage of the 193 nm immersion lithography [5], record low propagation losses on single mode PECVD SiN waveguides and 85nm-small critical dimensions for Bragg periods have been achieved compared to prior art [3]. As a result, grating couplers with back-reflections of ∼8 dB less compared to those achieved in standard grating devices and with less than 0.8 dB penalties in coupling efficiencies have been demonstrated at visible wavelengths, specifically at 532 nm and 638 nm. Finally, more than 80% of reflectivity is measured in focused Bragg gratings at wavelengths of 532 nm and 638 nm. At these wavelengths, life science applications such as ultra-sensitive chemical and biochemical sensing based on fluorescence and refractive index principles as well as next generation on-chip holography can be enabled, thus representing a very vibrant research and industrial topics nowadays.
2. Platform fabrication
SiN waveguides were fabricated on 300 mm wafers. A stack of 70 nm SiN and 2550 nm silicon oxide were deposited on silicon using PECVD process. The 70 nm SiN layer operates as an antireflection coating (ARC). A ‘touch up’ chemical mechanical polishing (CMP) was done to smoothen the oxide surface before depositing the waveguide layer. The roughness of the oxide surface has changed from 4 nm to 0.15 nm after CMP. For a wavelength of 532 nm, 180 nm PECVD SiN film was deposited on top of the smoothened oxide layer. Waveguides were patterned using 193 nm immersion lithography tool and subsequently dry etched using fluorine based etch chemistry to remove 180nm-SiN. Similarly, the 220 nm of waveguide deposition was implemented for a wavelength of 638 nm. Waveguides were patterned and dry etched using the same method. After resist strip and wet clean, a thin layer (∼100 nm) of high aspect ratio patterning (HARP) oxide was used to ensure void-free gap fill. This was followed by a thick oxide (∼2.5μm) deposition using PECVD. The top oxide layer was planarized using CMP to leave ∼2 μm thick oxide on top of the waveguide. All PECVD films were deposited at 400°C, while the HARP film was deposited at 430°C. Figure 1 shows the cross section of the waveguide platform.
Fig. 1. SiN waveguide platform. The PECVD SiN waveguide thicknesses are 220 nm and 180 nm for wavelengths of 638 nm and 532 nm, respectively.
When using gratings for in and out coupling of light in photonic circuits and/or as functional filtering devices on the platform, the 70 nm SiN ARC layer operates in order to minimize power efficiency fluctuations due to a non-uniform thickness of the under cladding layer (2.55 µm oxide). Although this layer generates a small penalty in grating coupler efficiency, it provides stable in and output power coupling across the whole wafer because the effect on the interference of reflected light from the silicon substrate is mitigated. Figure 2(a) shows that even thickness variations up to 50 nm reduce the coupling efficiency of grating couplers dramatically if the ARC is not applied. Conversely, the coupling efficiency is relatively stable against the thickness variations when the ARC layer is applied. The fabrication tolerance for the under cladding layer, which is fabricated by deposition and CMP, can be relaxed as shown in Fig. 2(b). In summary, the thickness of under cladding layer is one of the most sensitive parameters for the performance of grating couplers.
Fig. 2. Simulations showing the effect of a 70nm-ARC layer in the SiN platform: (a) without ARC layer (b) with ARC layer when varying the thickness of the under-cladding. Here, hundercladding: the height of a under cladding.
According to simulation results, it is worth specifying that the coupling efficiency of 638nm-grating couplers can be increased from 30% (using 70 nm ARC) to 45% when replacing the ARC layer with a 70nm-thick Titanium Nitride (TiN) mirror. However, in the case of 532nm-grating couplers, the efficiency is not increased when using a 70nm-TiN mirror at the same location. In conclusion, proper design of the distance between the waveguide layer and the TiN metal mirror layer is required to achieve higher coupling efficiency at both wavelengths.
The immersion lithography is a resolution enhancement technique for fabricating integrated circuits [5]. The air gap between a final lens in a lithography tool and a wafer surface is replaced by a liquid medium which has a higher refractive index than air. The resolution is increased by a factor equal to the refractive index of the liquid. Therefore, this technique enables the patterning of critical dimension (CD) of 85 nm (pitch = 170 nm) which is used for this experiment. Additionally, line edge roughness on the waveguides (side wall roughness) is reduced and the propagation loss which is directly affected by the roughness is decreased. Overall, the combination of small waveguide patterning and reduced surface roughness makes the immersion lithography an exquisite tool to enable design and fabrication of low-loss photonic devices operating, for the first time, at visible wavelengths.
The propagation test structures were designed to be characterized by five different waveguide lengths. Consequently, losses are plotted versus the different lengths at wavelengths of 638 nm and 532 nm. Propagation losses as low as 0.9 dB and 1.6 dB are achieved in single mode waveguides at 638 nm and 532 nm, respectively. The widths of single mode waveguides operating at 638 nm and 532 nm are 340 nm and 350 nm, respectively. The refractive indices of SiN and silicon oxide are 1.89 and 1.45 at 638 nm, respectively. Moreover, the refractive indices of SiN and silicon oxide are 1.91 and 1.46 at 532 nm, respectively.
The platform’s propagation losses are summarized in Table 1. Moreover, propagation loss performance comparison between waveguides fabricated using the standard and immersion lithography is reported demonstrating an improvement of more than 40% when using the latter.
Table 1. Propagation losses in the SiN waveguide platforms
3. Grating devices for visible light
3.1 Grating couplers for 638 nm
Three types of grating couplers operating at 638 nm were designed and evaluated. The low-back reflection grating couplers [7,8] are mainly used for in and out coupling of light in photonic integrated circuits. Most of the photonics circuits in the SiN platform are designed for sensing applications using visible light. Therefore, back-reflections from a grating coupler are undesirable. Back-reflections are due to Fresnel reflections occurring at the interface between an input waveguide and a grating trench region. They represent a detrimental effect in photonic circuits as they increase the optical instability in interferometric devices and generate ripples and/or oscillations in resonant spectra when two grating couplers are cascaded to generate a Fabry-Perot cavity. Furthermore, optoelectronic devices such as lasers and photodiodes are typically sensitive to back-reflections because they contribute to noise generation, especially in optical circuits where optical feedback schemes are implemented. For these reasons, we designed grating couplers for visible light applications with low back-reflections. For the design purpose, the platform can be assumed to be substrateless due to the 70-nm ARC layer. Figure 3 shows simulation results and their comparison between a normal grating coupler with and without ARC and silicon substrate. Due to the ARC layer, the assumption is acceptable. The difference of coupling efficiency is only less than 0.6 dB and the offsets of the two peak-wavelengths are reasonably similar. For the low back-reflection grating design, a 3-dimensional finite difference time domain (3D FDTD) simulation is required because of its asymmetric shape. However, under that assumption, the computational time can be significantly reduced.
Fig. 3. Comparison of simulations between normal grating couplers with and without ARC and substrate at wavelengths of (a) 532 nm and (b) 638 nm.
Figure 4(a) shows the schematics of the grating coupler and its design parameter for low back- reflection. The angle parameter φ defines the tilt angle of the grating coupler. Here we used φ equal to 3 degree and it results in a 20 degree tilting of the out coupling as shown in the inset of Fig. 4(a). Therefore, the grating has been intentionally 20-degree tilted for measurement and packaging when drawing a mask. We also designed and fabricated a normal grating coupler as shown in Fig. 4(b) for comparison. Additionally, an apodized grating coupler is designed and fabricated. Taking advantage of the immersion lithography, less than 100 nm trench can be fabricated. Grating periods are 460 nm and fully etched with a width of 225 nm. The apodization technique was implemented by changing the duty cycle of the front periods linearly as shown in Fig. 4(c). The first line trench of 92 nm is the narrowest. Figure 4(d) shows the test structure for back reflection measurements.
Fig. 4. (a) Low back-reflection, (b) normal and (c) apodized grating couplers. (d) Test structure for back reflection measurement. The inset of (a) shows the out coupling far field angle of the low back-reflection grating coupler.
The coupling efficiencies and back reflection characteristics of grating couplers are shown in Fig. 5. We also compare simulations and experimental data. A single mode fiber is assumed in simulations and practically used in experiments. In particular, the mode field diameter (MFD) of the fiber is calculated to be ∼4.5μm using a numerical aperture of ∼0.12 at 630 nm, which is from the specification of the fiber we used. The incident angle of the fiber is set at 10degree in the longitudinal direction (θx). For the low back reflection grating coupler, the fiber angles are tilted at 10 and 20 degrees in θx and θz, respectively. The simulated coupling efficiency of low back reflection grating coupler is −6.6 dB at the peak corresponding to a wavelength of 640 nm. The measurement result is −7.5 dB at 644 nm. For comprison, the normal grating coupler was measured, and the efficiency was −6.8 dB at a wavelength of 644 nm. Moreover, the coupling penalty was 0.7 dB. This penalty is likely to derive from the shape and size mismatch between the fiber mode and the mode from the grating coupler. Simulations and measurements of back-reflection characteristics are in good agreement. Furthermore, the reflection of the low back-reflection grating coupler is 8 dB less than that of normal grating coupler as shown in Fig. 5(a) and 5(b). The back reflections of the normal and designed grating couplers were −23 dB and −31 dB at Full width half maximum (FWHM), respectively. In the case of grating apodization as shown in Fig. 5(c), the efficiency increase is not observed in simulations and experiments due to the small size of fiber and grating modes. Measured bandwidths of all grating couplers, which are determined by the effective refractive index of the grating area and operating wavelength [9], are ∼30 nm at FWHM. The fluctuation of back reflections is reduced in the simulation and experiments even though the reflection is still higher than the normal grating coupler. It would most probably come from the first trench which is a nearly half of Bragg period.
Fig. 5. Coupling efficiency and back-reflection of (a) low back-reflection, (b) normal, and (c) apodized grating couplers. Solid lines are simulations, while circles and triangles are experiments.
The discrepancy between simulated and measured coupling efficiencies is mainly due to the longitudinal misalignment between the grating coupler and the fiber. In particular, cleaved fibers were used and a minimum gap of 11μm between the fiber and grating coupler physically occurred thus generating the longitudinal misalignment because of 10 degree tilted angle of the fiber. For a conventional grating coupler and fiber coupling in the C-band, a ∼11μm-longitudinal misalignment can induce negligible loss. In fact, the theoretical longitudinal misalignment loss between 10μm-MFDs for C-band is less than 0.2 dB up to a 50μm-gap. Conversely, misalignment loss induced by a 11μm-gap are not negligible at visible wavelengths because the mode spot sizes are much smaller compared to the C-band. For example, when the coupling between two 4.5μm-modes occurs at the visible light, the estimated 11μm-longitudinal misalignment loss is ∼0.22 dB [10]. However, a longer gap between the fiber and the grating coupler was used in experiments in order to avoid breaking of the fiber tip. In the case of using the gap of∼25μm, the estimated misalignment loss is 1 dB.
For product packaging purposes, the proposed low-back reflection grating couplers can be pigtailed using commercial fiber blocks with typical fiber-pigtailing processes [8,11]. In particular, the periods of the grating couplers must be properly designed for index matching when using an adhesive medium for fiber attachment. In fact, the grating peak wavelength will shift to shorter wavelengths because of the presence of the index matching medium. For example, when using the adhesive with the index of 1.5, the peak wavelength will be shifted to near 600 nm. This results in different performance and specifications compared to the nominal design requirements. Consequently, the actual period of the grating couplers reported here and designed for chip-based applications (no packaging) must be longer in order to compensate for the wavelength shift aforementioned and make the grating couplers operating within specifications.
3.2 Bragg gratings for mirror applications
Multilayer based Bragg mirrors for visible light have been reported [12–14]. However, a waveguide-base Bragg grating mirror for visible light is rarely reported experimentally because Bragg periods below 100 nm are required for the visible light. We designed a focused Bragg grating as shown in Fig. 6 [15,16]. The focused Bragg mirror has advantages compared to conventional designs (see. inset of Fig. 7(c)). This focused Bragg mirror exhibits lower loss and higher reflectivity than the conventional Bragg mirror when both are characterized by the same length (N = 10) as shown in Fig. 7. The Bragg period (Λ) is 200 nm and the width of trench is 100 nm for the Bragg wavelength of 638 nm. The design parameters are described in Fig. 6. Figure 8 shows the length optimization for the focused Bragg mirror. Here N is the number of trenches. When N is 15, the focused grating mirror exhibit nearly 90% of reflectivity.
Fig. 7. Comparison of Bragg gratings when N = 10: focused (solid lines) vs. conventional (dotted lines).
Fig. 8. Focused Bragg grating simulations when having different lengths: (a) reflection, and (b) transmission.
We fabricated Bragg mirrors with different N values, specifically N equal to 15 and 30. Using the immersion lithography, a 200 nm pitch can be fabricated which is corresponding to 100 nm trench (50% duty cycle). Figure 9 shows the results of fabricated Bragg gratings. The reflection measurement was implemented with the same test structures in Fig. 4(d). The Bragg grating is integrated on the output arm (see Fig. 4(d), ‘Output for T’). The measured reflections are −1.1 dB and −0.7 dB when N = 15 and N = 30, respectively. FWHMs are approximately 50 nm and 58 nm when N = 15 and N = 30, respectively. The Bragg grating with N = 30 filters adjacent light strongly and the Bragg wavelength of 645 nm can be transmitted below −35 dB.
Fig. 9. Characteristics of Bragg gratings when (a) N = 15 and (b) N = 30 (solid lines: simulations, circles and squares: experiments).
3.3 Grating devices for 532 nm
We also designed and fabricated grating couplers operating at a wavelength of 532 nm. At this wavelength propagation losses are higher than that achieved at 638 nm as we described in Section 2. Also, smaller sizes are required for these devices considering the shorter operating wavelength. The low back reflection grating couplers are characterized by a period and φ of 383 nm and 3 degree, respectively. The theoretical fiber coupling efficiency with ∼4.5μm-mode field diameter and back-reflection of a normal grating couplers are −6.0 dB and −22.3 dB at the peak wavelength of 532 nm, respectively. Theoretical coupling efficiency and back-reflection of the low back reflection grating coupler are −6.8 dB and −35.4 dB at the peak wavelength of 532 nm, respectively. Figure 10(b) shows the comparative simulation results of the coupling efficiencies and back reflections of normal and low back-reflection grating couplers. The penalty of coupling efficiency between the two grating typologies is ∼0.8 dB. Simulation results also show that the back-reflection of the grating couplers is ∼13 dB lower than that of normal grating couplers. The measured coupling efficiency and back-reflections are −9.1 dB and −27 dB, respectively. The design parameter of φ which is resulted in the output of 20 degree tilting is shown in Fig. 10(c). The inset of Fig. 10(c) shows the scanning electron microscope (SEM) photo of the fabricated grating coupler for 532 nm. The discrepancy between theory and experiment in the coupling efficiency is higher than that of case of 638 nm. Grating couplers for shorter wavelengths are more sensitive to process variations of the under cladding than those for longer wavelengths. Even though experimental results show that grating couplers operating at 532 nm can be more sensitive to fabrication tolerances than those working at 638 nm, we believe that the ARC layer can mitigate this sensitivity.
Fig. 10. Experimental characteristics of low back-reflection grating coupler at λ of 532nm: (a)transmission and reflection, (b) simulation comprison between normal and low back-reflection grating couplers, and (c) far field angle of designed grating coupler and SEM photo.
532 nm-Bragg gratings are also fabricated and characterized. Figure 11 shows SEM pictures of the fabricated Bragg grating with 30 Bragg periods. The Bragg period is 170 nm corresponding to an 85 nm trench width. Figure 12 shows the results of the focused Bragg mirrors operating at 532 nm. It shows that reflections of −0.7 dB and −0.62 dB have been achieved at peak wavelengths of ∼526 nm when N = 15 and N = 30, respectively. We also compared simulations and measurements demonstrating a good agreement as shown in Fig. 12. We detected strong sidelobes in reflection-measurements plotted in Fig. 12. These might be originated from unexpected reflections occurring in the test structures. Furthermore, it is worth specifying that we did not apply any apodization technique to reduce sidelobes in these Bragg gratings.
Fig. 12. Bragg grating characteristics with (a) N = 15 and (b) N = 30 (solid lines are simulations, while circles and squares are experiments).
4. Conclusion
Low back-reflection grating couplers and focused Bragg grating mirrors were designed, fabricated and tested for applications in the visible wavelength range. These grating devices were fabricated on SiN technology platforms using a 193 nm immersion lithography tool within optimal process flow and layer stack developed at IMEC. As a result, the platform is characterized by record low propagation loss in single mode waveguides, specifically ∼0.9 and ∼1.7 dB/cm at operating wavelengths of 638 nm and 532 nm, respectively. Furthermore, Bragg gratings can be fabricated with critical dimensions of 85 nm, thus enabling, for the first time to the best of our knowledge, new spectral functionalities for visible light applications. In particular, grating couplers exhibited ∼8 dB lower back reflections and less than 0.8 dB penalties in coupling efficiencies compared to normal gratings. More than 80% of reflectivity was measured in 638nm- and 532nm-focused Bragg gratings. Finally, the developed SiN technology platform offers uniform and stable grating performance across a full processed wafer.
References
1. P. Munoz, G. Micó, L. A. Bru, D. Pastor, D. Pérez, J. D. Doménech, J. Fernández, R. Banos, B. Gargallo, R. Alemany, A. M. Sanchez, J. M. Cirera, R. Mas, and C. Dominguez, “Silicon nitride photonic integration platforms for visible, near-infrared and mid-Infrared applications,” Sensors 17(9), 2088 (2017). [CrossRef]
2. S. Romero-García, F. Merget, F. Zhong, H. Finkelstein, and J. Witzens, “Silicon nitride CMOS-compatible platform for integrated photonics applications at visible wavelengths,” Opt. Express 21(12), 14036–14046 (2013). [CrossRef]
3. A. Z. Subramanian, P. Neutens, A. Dhakal, R. Jansen, T. Claes, X. Rottenberg, F. Peyskens, S. Selvaraja, P. Helin, B. Du Bois, K. Leyssens, S. Severi, P. Deshpande, R. Baets, and P. Van Dorpe, “Low-loss singlemode PECVD silicon nitride photonic wire waveguides for 532–900 nm wavelength window fabricated within a CMOS pilot line,” IEEE Photon. J. 5(6), 2202809 (2013). [CrossRef]
4. J. F. Bauters, M. J. R. Heck, D. D. John, J. S. Barton, C. M. Bruinink, A. Leinse, R. G. Heideman, D. J. Blumenthal, and J. E. Bowers, “Planar waveguides with less than 0.1 dB/m propagation loss fabricated with wafer bonding,” Opt. Express 19(24), 24090–24101 (2011). [CrossRef]
5. S. K. Selvaraja, G. Winroth, S. Locorotondo, G. Murdoch, A. Milenin, C. Delvaux, P. Ong, S. Pathak, W. Xie, G. Sterckx, G. Lepage, D. V. Thourhout, W. Bogaerts, J. V. Campenhout, and P. Absil, “193 nm immersion lithography for high performance silicon photonic circuits,” Proc. SPIE 9052, 90520F (2014). [CrossRef]
6. K. Lodewijks, S. Jayachandran, T. D. Kongnyuy, S. Lenci, S. Das, P. V. Dorpe, A. Humbert, R. Jansen, S. Severi, and X. Rottenberg, “Low loss high refractive index niobium oxide waveguide platform for visible light applications,” in Advanced Photonics 2018 (BGPP, IPR, NP, NOMA, Sensors, Networks, SPPCom, SOF), OSA Technical Digest (online) (Optical Society of America, 2018), paper ITh2B.3.
7. J. H. Song and X. Rottenberg, “Low-back-reflection grating couplers using asymmetric grating trenches,” IEEE Photon. Technol. Lett. 29(4), 389–392 (2017). [CrossRef]
8. J. H. Song, B. Snyder, K. Lodewijks, R. Jansen, and X. Rottenberg, “Grating coupler design for reduced back-reflections,” IEEE Photon. Technol. Lett. 30(2), 217–220 (2018). [CrossRef]
9. C. R. Doerr, L. Chen, Y.-K. Chen, and L. L. Buhl, “Wide bandwidth silicon nitride grating coupler,” IEEE Photon. Technol. Lett. 22(19), 1461–1463 (2010). [CrossRef]
10. L. A. Wang and C. D. Su, “Tolerance analysis of aligning an astigmatic laser diode with a single-mode optical fiber,” J. Lightwave Technol. 14(12), 2757–2762 (1996). [CrossRef]
11. B. Snyder and P. O’Brien, “Packaging process for grating-coupled silicon photonic waveguides using angle-polished fibers,” IEEE Trans. Compon., Packag., Manuf. Technol. 3(6), 954–959 (2013). [CrossRef]
12. T.-H. Chang, T.-E. Lee, N.-K. Hsueh, C. H. Lin, and C.-F. Yang, “Investigation of TiO2–Al2O3 bi-layer films as Bragg reflector of blue light by using electron beam evaporation,” Microsyst. Technol. 24(10), 3941–3948 (2018). [CrossRef]
13. J. M. Fastenau and G. Y. Robinson, “Low-resistance visible wavelength distributed Bragg reflectors using small energy band offset heterojunctions,” Appl. Phys. Lett. 74(25), 3758–3760 (1999). [CrossRef]
14. J. W. Leem, X.-Y. Guan, and J. S. Yu, “Tunable distributed Bragg reflectors with wide-angle and broadband high-reflectivity using nanoporous/dense titanium dioxide film stacks for visible wavelength applications,” Opt. Express 22(15), 18519–18526 (2014). [CrossRef]
15. C.-J. Chae, E. Skafidas, and D.-Y. Choi, “Compact Bragg grating reflectors in silicon waveguides and their application to resonator filters,” Th2A.20, OFC 2014.
16. C. J. Chae and E. Skafidas “Circular Bragg grating reflector suitable for integration with Narrow silicon nanowires,” Optical Society of America, In Integrated Photonics Research, Silicon and Nanophotonics, IT2A-3 (2013).
