We show that silicon microrings with adiabatically widened bends are more tolerant to dimensional variations than conventional microring designs with uniform waveguide widths. Through wafer-scale measurements of test structures fabricated in the IMEC Standard Passives process (193 nm DUV lithography, 200 mm SOI wafer), improvements in the intra-die and wafer-scale variation of the resonance wavelength are demonstrated. A 2.1× reduction in the standard deviation of the resonance wavelength across the wafer was observed.
© 2014 Optical Society of America
Microring resonators have emerged as a unique and important class of devices in high index contrast integrated photonics platforms, such as silicon-on-insulator (SOI). The high index contrast in SOI enables waveguides that have sub-micron dimensions to be bent with micron-scale radii of curvature, amenable to microring resonators. SOI microrings are attractive candidates for ultra-compact optical filters, muxes/demuxes, switches, and modulators [1–6]. However, a critical challenge with microrings is that their spectral properties are extraordinarily sensitive to fabrication variations. This is fundamentally due to the sensitivity of the effective index of high confinement single-mode strip waveguides to nanometer-scale dimensional variations . Such high confinement strip waveguides with fully etched sidewalls are needed to realize tight bend radii in microrings. For a microring formed from standard 500 nm wide and 220 nm tall SOI strip waveguides, we calculate that the resonance wavelength shifts by about 0.6 nm per 1 nm change in the waveguide width. In a 248 nm or 193 nm photolithography fabrication process, dimensional variations are about tens of nanometers, which can hinder device yield and necessitates tuning for spectral alignment [4,8–10]. Indeed, with today’s foundry processes, the thermal tuning required to align a resonance to a desired wavelength is the dominant source of power consumption in microring devices [4, 6, 11, 12].
In this work, we show that by incorporating wide, multi-mode waveguide bends into the microrings, the spectral characteristics of the microring become less sensitive to fabrication variations. To excite only the fundamental mode of the wide waveguide, the bend is adiabatically widened from the coupler region. Adiabatically widened microrings have been studied for electrical contacts and efficient thermal tuning of microrings [13–15], but their impact on variation tolerance is only recently being explored [16, 17]. Here, we show reductions in both intra-die and wafer-scale variations in adiabatically widened microrings compared to microrings formed from standard strip waveguides using a IMEC Standard Passives multi-project-wafer (MPW) shuttle which uses 193 nm DUV lithography over 200 mm SOI wafers.
2. Device design
Figure 1(a) illustrates the adiabatically widened microring design. The top half of the microring, which contains the coupler region, uses single-mode waveguides. The bottom half consists of a 180 degree waveguide bend in which the single-mode input/output waveguides are widened, adiabatically transforming the optical mode to the lowest order whispering gallery mode (WGM) at the apex of the bend. The effective index of a WGM is less sensitive to variations in width, since it depends only on the radius of curvature, R, defined by the outer wall. The phase accumulation in the bend, ϕbend, isEq. 1, we parameterize c with respect to polar angle, θ. We use an eigenmode solver to numerically compute neff (θ), which depends on both the local width, W(θ), and the local radius of curvature, R(θ). The calculations are done for λ = 1550 nm and assuming material refractive indices for silicon and silica of 3.48 and 1.45 respectively. The bend structure is tailored to ensure that the transitions to WGM modes are short while remaining adiabatic (low-loss). We use bends with elliptically shaped outer walls and a parabolically widening local width of the form
Figure 1(b) shows that the sensitivity of ϕbend to the width offset, α, decreases as Wmax increases. For a semi-circular bend with R = 15 μm and Wmax = 2.0 μm, is 8× smaller than that of a standard, non-widened bend. Figure 1(c) shows the electric field intensity computed using 3D finite difference time domain (FDTD) simulations of the semi-circular bend and the adiabatic transformations of the strip waveguide mode to the WGM. The computed insertion loss is <0.02 dB. Therefore, the incorporation of adiabatically widened bends within microrings can significantly improve the tolerance to dimensional variations compared to designs formed solely from single-mode strip waveguides.
3. Intra-die variation measurements
We first investigate the effect of the adiabatically widened bends on intra-die variations, which tend to be smaller than the variation across the wafer. Figure 2(a) shows our test structure for measuring intra-die variation, which was a series of nominally identical rings coupled to a common bus waveguide. The transmission through this structure is the product of the transmission spectra of the individual rings. As the number of rings increases, the composite transmission spectrum approaches that of a notch filter with a deep extinction ratio (ER) and a full-width-half-maximum linewidth (FWHM) that characterizes the intra-die resonance wavelength variability . This effect has also been used to design a balanced SCISSOR device .
The microring design tested consisted of a ridge waveguide coupler section integrated with a waveguide bend in the shape of a half-ellipse, with major and minor axes of 56 μm and 33.6 μm, respectively. We used an elliptically shaped bend to increase the FSR while accommodating the long length of the coupling region. The ridge waveguide coupler was used to further improve the fabrication tolerance. It had a partially-etched silicon slab thickness of 150 nm, waveguide widths of 400 nm, a coupling gap of 400 nm, and a coupling length of 21 μm. We have previously shown that the coupling coefficient of such ridge couplers are highly tolerant to variations in width, coupling gap, SOI thickness and partial etch depth . Outside of the coupler region, the ridge waveguide mode was converted to the more tightly confined mode of a 500 nm wide strip waveguide using 3 μm long ridge-to-strip tapers. From a separate series of cutback structures, we measured the loss to be about 0.04 dB per taper. The strip waveguides enable tight bends and compact microring sizes. Lastly, for a low-loss transition from the 500 nm wide strip waveguide mode to a WGM, the local waveguide width was parabolically tapered around the bend circumference with the vertex located at the apex.
We compare two sets of test structures formed from microrings that were identical except for the waveguide bends. One set used adiabatically widened waveguides, where the width changed from 500 nm to Wmax = 2.5 μm, and the other set used waveguides with a constant width of 500 nm. Pairs of test structures containing widened and standard microrings were laid out as close to each other as possible to compare their intra-die variations as shown in Fig. 2(b). Figure 2(c) shows the spectra of 1, 2, 5, and 10 bus-coupled widened rings. As the spectra of the microrings start to overlap, the ER of the notch increases. Generally, as shown in Fig. 2(d), for the lineshape to become smooth, about 20 bus-coupled adiabatically widened microrings were needed, while more than 20 standard microrings would be required, in agreement with , which shows transmission ripples even after 56 microrings.
Table 1 summarizes the spectral features of the single-microring and 20-microring bus-coupled test structures on one representative die. From the FWHM of the 20-microring transmission spectra, the widened microrings exhibited a reduction in the resonance wavelength spread of about 15% compared to the standard microrings. The widening bend did not degrade the Q factors, which were about 8.2 × 103, and it slightly increased the FSR from 4.45 nm to 4.8 nm. The latter effect was due to lower group indices of WGMs, which were less dispersive than strip waveguide modes.
4. Wafer-scale measurements
To characterize the wafer-scale variation tolerance, we measured equivalent microring devices in 16 dies across the 200 mm SOI wafer. For the microrings discussed in the previous section, since the wafer-scale variation of the resonance wavelength was larger than the FSR, we were unable to resolve resonances of different orders. Therefore, in this section, we focus on two microrings designs that are smaller in size and do not suffer from this problem. The first is highly compact, and the FSR was significantly larger than the wafer-scale resonance variation. The larger second design contains a ridge waveguide coupler for a greater variation tolerance.
The highly compact microring is shown in Fig. 3(a). It had a strip waveguide coupler section and an elliptical waveguide bend with major and minor axes of 11.2 μm and 9.25 μm, respectively. In the adiabatically widened versions, Wmax = 1.2 μm. Figures 3(b) and (c) show the wafer maps of the measured resonance wavelengths of the widened and standard microrings, and the first two rows in Table 2 summarize the measurement results. Incorporating the adiabatically widened waveguide bend reduced the standard deviation of the resonance wavelengths, σλres, by about 27% compared to the standard microrings. The improvement was limited by the relatively long sections of 500 nm strip waveguides near the coupler region, which accounted for the majority of the round-trip phase accumulation.
The microring design using the ridge waveguide coupling sections is shown in Fig. 4(a). Since these microrings incorporated low-curvature ridge waveguide bends and ridge-to-strip waveguide tapers, they were larger than microring designs that used strip waveguides only. The microring had a coupling section with 500 nm wide ridge waveguides and a 500 nm gap connected to an elliptical strip waveguide bend with major and minor axes of 33.6 μm and 20.2 μm, respectively. For the widened microring design, Wmax = 2.0 μm. Since the wafer-scale resonance variation was similar to the FSR, we used the results of Figures 3(b) and (c) to inform the construction of the wafer maps and ensure that the same resonance order is tracked across the wafer. Figures 4(b) and (c) are the resultant wafer maps, which are similar in form to those for the compact microrings in Fig. 3(b) and (c). The measurement results are summarized in the last two rows of Table 2. The adiabatically widened microring exhibited a 2.1× improvement in σλres compared to the standard microring. The observed improvement was slightly lower than the expected value of 2.5× calculated from an analysis of the phase response to width variations. The discrepancy was most likely due to the additional variations in the SOI thickness and partial etch depth. The range of the measured ER was smaller compared to the microrings with strip waveguide couplers, indicating that the ridge waveguide couplers were also more variation tolerant .
In summary, using the IMEC Standard Passives process, we have demonstrated that microrings with adiabatically widened bends are more variation tolerant than standard designs that use strip waveguides with a uniform width. The improved intra-die and wafer-scale variation tolerance can potentially increase the device yield and reduce the amount of power required for post-fabrication tuning. The variation tolerance can be further improved with refinements to the design of the couplers and the adiabatic transitions.
Access to the IMEC Standard Passives MPW was supported by CMC Microsystems. We thank Dan Deptuck and Jessica Zhang of CMC Microsystems for their help. The financial support of the Natural Science and Engineering Research Council of Canada is gratefully acknowledged.
References and links
1. F. Xia, M. Rooks, L. Sekaric, and Y. Vlasov, “Ultra-compact high order ring resonator filters using submicron silicon photonic wires for on-chip optical interconnects,” Opt. Express 15, 11934–11941 (2007). [CrossRef] [PubMed]
2. J. R. Ong, R. Kumar, and S. Mookherjea, “Ultra-high-contrast and tunable-bandwidth filter using cascaded high-order silicon microring filters,” IEEE Photon. Technol. Lett. 25, 1543–1546 (2013). [CrossRef]
3. P. Dong, S. Liao, D. Feng, H. Liang, D. Zheng, R. Shafiiha, C.-C. Kung, W. Qian, G. Li, X. Zheng, A. V. Krishnamoorthy, and M. Asghari, “Low Vpp, ultralow-energy, compact, high-speed silicon electro-optic modulator,” Opt. Express 17, 22484–22490 (2009). [CrossRef]
4. A. Krishnamoorthy, X. Zheng, G. Li, J. Yao, T. Pinguet, A. Mekis, H. Thacker, I. Shubin, Y. Luo, K. Raj, and J. Cunningham, “Exploiting CMOS manufacturing to reduce tuning requirements for resonant optical devices,” IEEE Photonics Journal 3, 567–579 (2011). [CrossRef]
5. W. Bogaerts, P. De Heyn, T. Van Vaerenbergh, K. De Vos, S. Kumar Selvaraja, T. Claes, P. Dumon, P. Bienstman, D. Van Thourhout, and R. Baets, “Silicon microring resonators,” Laser and Photonics Reviews 6, 47–73 (2012). [CrossRef]
6. G. Li, A. Krishnamoorthy, I. Shubin, J. Yao, Y. Luo, H. Thacker, X. Zheng, K. Raj, and J. Cunningham, “Ring resonator modulators in silicon for interchip photonic links,” IEEE J. Sel. Top. Quant. Elect. 19, 3401819 (2013). [CrossRef]
7. M. A. Popovic, T. Barwicz, E. P. Ippen, and F. X. Kartner, “Global design rules for silicon microphotonic waveguides: sensitivity, polarization and resonance tunability,” “Conference on Lasers and Electro-Optics/International Quantum Electronics Conference,” (Optical Society of America, 2006), p. CTuCC.
8. S. Selvaraja, W. Bogaerts, P. Dumon, D. Van Thourhout, and R. Baets, “Subnanometer linewidth uniformity in silicon nanophotonic waveguide devices using cmos fabrication technology,” IEEE J. Sel. Top. Quant. Elect. 16, 316–324 (2010). [CrossRef]
9. T. Baehr-Jones, T. Pinguet, G.-Q. Lo, S. Danziger, D. Prather, and M. Hochberg, “Myths and rumours of silicon photonics,” Nature Photonics 6, 206–208 (2012). [CrossRef]
11. P. Dong, W. Qian, H. Liang, R. Shafiiha, N.-N. Feng, D. Feng, X. Zheng, A. V. Krishnamoorthy, and M. Asghari, “Low power and compact reconfigurable multiplexing devices based on silicon microring resonators,” Opt. Express 18, 9852–9858 (2010). [CrossRef] [PubMed]
14. M. R. Watts, W. A. Zortman, D. C. Trotter, G. N. Nielson, D. L. Luck, and R. W. Young, “Adiabatic resonant microrings (ARMs) with directly integrated thermal microphotonics,” “Conference on Lasers and Electro-Optics/International Quantum Electronics Conference,” (Optical Society of America, 2009), p. CPDB10.
15. A. Biberman, E. Timurdogan, W. A. Zortman, D. C. Trotter, and M. R. Watts, “Adiabatic microring modulators,” Opt. Express 20, 29223–29236 (2012). [CrossRef]
16. J. C. Mikkelsen, W. D. Sacher, and J. K. S. Poon, “Adiabatically widened silicon microring resonators with improved tolerance to wafer-scale variations,” “Optical Fiber Communication Conference,” (Optical Society of America, 2014), p. Th2A.12.
17. Z. Su, E. S. Hosseini, E. Timurdogan, J. Sun, G. Leake, D. D. Coolbaugh, and M. R. Watts, “Reduced wafer-scale frequency variation in adiabatic microring resonators,” “Optical Fiber Communication Conference,” (Optical Society of America, 2014), p. Th2A.55.
18. F. Xia, L. Sekaric, and Y. Vlasov, “Ultracompact optical buffers on a silicon chip,” Nat. Photon. 1, 65–71 (2007). [CrossRef]
19. J. Cardenas, P. A. Morton, J. B. Khurgin, C. B. Poitras, and M. Lipson, “Super-ring resonators: Taking advantage of resonance variability,” “Conference on Lasers and Electro-Optics 2012,” (Optical Society of America, 2012), p. CTu3I.4.