Open Access
Add to BibSonomy
Abstract
We propose and demonstrate a simple all-optic control for Si photonics using a photo-thermal heater. The control light is absorbed in a heavily doped control waveguide and the signal light phase is tuned through thermal diffusion in a signal waveguide adjacent to but not optically coupled with the control waveguide. We designed and fabricated Mach–Zehnder- and microring-type devices requiring 17 (π-phase shift) and 4 (switching between resonance and non-resonance with 6 dB extinction) mW of control power, respectively. We confirmed that the heating efficiency of all-optic control exceeded that of an electrical heater placed above the signal waveguide.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
In Si photonics, the guided light is often controlled by the thermo-optic effect [1]. Common optical devices such as Mach–Zehnder interferometers (MZI) and microring resonators (MRR) are partially heated by an electrical heater to adjust their interference condition [2], resonant wavelength [3–7], and coupling efficiency [7–9]. The heater is constructed from process-compatible metals such as TiN [5,7], Ti [6], W [3,4], and NiCr [8] adjacent to Si waveguides with sufficient spacing to avoid optical absorption. Direct current injection through Si waveguides [10–13] improves the heating efficiency but causes absorption losses due to doping and complicates the process.
In this paper, we discuss a simple on-chip all-optic control method using the photo-thermal effect in Si photonics, which is supposed to be used remotely from base equipment. In special environments such as at cryogenic temperatures or in the presence of explosives, electrical components might need to be fully replaced with all-optic control. Unlike ultrahigh-speed all-optic signal processing [14,15], we consider replacing the electrical heaters with photo-thermal all-optic control working at relatively low speeds. Some related studies using cavities and metallic structures [16–24] have focused only on local heating and are less efficient or incompatible with the standard Si photonics process, which is provided by complementary-metal-oxide semiconductor-like foundry services.
Our control method employs a photo-thermal heater (here called an optical heater) composed of a heavily doped Si waveguide that absorbs control light via free carrier absorption, generates heat, and diffuses it to an adjacent signal waveguide (Fig. 1) [20,24]. Such a structure is fully compatible with the standard process and the absorption loss of signal light is suppressed by an appropriate design. As a proof of concept, we fabricated MZI and MRR devices and demonstrated that the optical heater can be more efficient than an electrical heater. In the following contents, we first present the modeling and simulation of the device, and then show the experimental results.
Fig. 1. (a) 3D structure of the optical heater, which is located at the central part of (b) describing a 2D planar layout including curved input and output waveguides. (c) equivalent modal index of signal waveguide calculated as a function of gap g.
2. Modeling
The optical heater consists of two parallel Si rib waveguides embedded on a common thin Si slab and cladded by SiO2. The control waveguide is heavily n+ -doped to absorb the control light, whereas the signal waveguide simply transmits the signal light. The light absorbed in the control waveguide heats both the control and signal waveguides. The doped region is triangular-shaped so that light is absorbed gradually and the heating is moderately expanded along the waveguide. The Si slab assists thermal diffusion from the control waveguide to the signal waveguide. Unnecessary parts of the Si slab are removed as much as possible to reduce unwanted thermal diffusion. The gap g between the two waveguides is a critical parameter because an excessively large g degrades the heating efficiency, whereas an excessively small g results in optical coupling and absorption of the signal light.
The absorbed light power Pabs(z) at distance z in the control waveguide is given by
where Pctrl is the incident power of the control light and α(z) is the absorption coefficient in the doped region. The temperature distribution at the position r around the signal waveguide, T(r), depends on Pabs(z) and three-dimensional heat diffusion. The phase of the signal light is changed by the change in the equivalent modal index along the signal waveguide, Δneq(z), as a function of T(r). Here, the material indices relate with T(r) as nSi(r) = 0.00018T(r) + 3.50 and nSiO2(r) = 0.000012T(r) + 1.44 in the range that the linearity is maintained. We calculated T(r) from Pabs(z) using Lumerical HEAT and assuming a perfect heat sink sufficiently far below the Si slab. We also calculated neq(z) from T(r) using Lumerical FDE. The total phase change is given as3. Simulations
The light propagation and heating characteristics in the model of Fig. 1(a) and (b) were simulated using the charge and heat solvers in Lumerical FDTD and their material parameter library. The thickness and width of the Si rib waveguides were 220 nm and 450 nm, respectively, and the thickness of the Si thin slab was 110 nm. These values allow the single modal transmission of signal light at λ ∼1550 nm. The control region was 30 µm long. In the heavily n+ -doped region, we assumed an experimental doping concentration of 2.273 × 1020 cm−3. Figure 1(c) plots neq of the signal waveguide as a function of g calculated with the Lumerical FDE solver, where g represents the separation between the slab-connected control waveguide (orange part in Fig. 1(c) inset, not heated here) and the signal waveguide. The two waveguides were surrounded by SiO2 cladding. When g exceeded 0.5 µm, it negligibly affected the signal waveguide mode since the adjacent waveguide. Therefore, we set g = 0.5, 1.0, and 1.5 µm in the following calculations. The width of the slab beside the signal waveguide on the opposite side of the control waveguide was set at 0.5 µm. In this configuration, the signal mode was almost symmetric and suppressed the excess heat diffusion.
Figure 2(a) and (b) shows the propagation of the signal and control light when g= 0.5 µm. Optical coupling can occur in the control section like a directional coupler. The maximum coupling length was calculated for g = 0.5 µm using Lumerical FDE solver as $L = \mathrm{\lambda}/2({n_{\textrm{even}}} - {n_{\textrm{odd}}}) = 10.1\ \mathrm{\mu}\textrm{m}$, where neven = 2.570 and nodd = 2.493 are the effective indices of the two coupled modes at λ = 1550 nm. Therefore, the optical heater length of 30 µm is approximately 3 times the coupling length and almost satisfies the maximum coupling condition. However, due to the different cross-sectional shapes of the two parallel waveguides, the maximum coupled power from the signal waveguide was only 9%. For the signal light passing through the control section, the insertion loss caused by the coupling and absorption was calculated to be 0.63 dB as shown in Fig. 2(a). For g = 1.0 µm, the calculated loss was reduced to 0.34 dB. Optical coupling is suppressed and insertion loss is reduced further by employing different waveguide widths in the control section. As shown in Fig. 2(b), the control light was gradually attenuated and almost disappeared after passing through the 30-µm long control region. The optical absorption was distributed over the control waveguide with a maximum at the center of the waveguide (Fig. 2(c)).
Fig. 2. Simulation of (a) signal light and (b) control light propagation, and (c) absorption of control light when g = 0.5 µm.
We compared the temperature distribution in the optical heater with that in an electrical heater consisting of 50 nm thick TiN located 2.1 µm above the signal waveguide. Panels (a)‒(c) of Fig. 3 show the planer and cross-sectional temperature distributions when the input power to both heaters was 40 mW and 300 K was the reference temperature without heating. In the optical heater case, the temperature at the signal waveguide with g = 0.5 µm reached a maximum of 643 K owing to direct thermal diffusion through the thermally conductive Si slab. In the electrical heater case, the maximum temperature at the signal waveguide was 512 K although the TiN was heated to 666 K.
Fig. 3. Planar (left) and cross-sectional (right) temperature distributions through the optical heater with (a) g = 0.5 µm and (b) 1.5 µm, and (c) through the electrical heater.
The changes in neq and total phase [Eq. (2)] were also calculated using Lumerical FDE solver and are plotted in Fig. 4. The all-optic heating in the triangular doped region resulted in a wide bell-shaped distribution of neq. The distribution under electrical heating was flatter because the TiN heater uniformly heated the waveguide and the heat diffusion through the SiO2 cladding was limited. The total phase change increased with heating power. The π shift powers Pπ were 19.2, 20.8, and 22.4 mW for the optical heater with g = 0.5, 1.0, and 1.5 µm, respectively, and 22.5 mW for the electrical heater. The optical heater exhibited higher efficiency than the electrical heater within the parameters of this calculation.
Fig. 4. Equivalent modal index neq and total phase shift of guided mode in the signal waveguide with heating: (a), (b) distribution of neq with optical and electrical heating, respectively. For the optical heating, g = 0.5 µm is assumed. (c) Phase shift versus control power Pctrl.
4. Fabrication
We designed and fabricated MZI- and MRR-type optical switches (Fig. 5) to demonstrate the proposed all-optic control using 300-mm-diameter SOI wafer process of Advanced Industrial Science and Technology (AIST) commercial multi-project wafer (MPW) service. Although our devices require a doping process, all-optic configuration allows them to be fabricated without electrical components such as electrical heaters, contacts, and wirings. We multiplexed signal light and control light at different wavelengths in off-chip optical fibers and launched them onto the Si waveguide. The light signals were demultiplexed by an on-chip directional coupler and then incident on the control section. Setting the length of the directional coupler at 680 µm, the wavelengths of the control and signal light could be easily chosen in the C band. Although a wavelength-insensitive multiplexer in operating broadband might be better in further applications, we employed here a DC for a simple proof-of-concept demonstration. In practical applications, it will be recommended to use two fibers, one for signal input and the other for control input. Rib waveguides of 30 µm were inserted into both arms of the MZI device, one for the control section and the other for maintaining symmetry between the two arms. In the control waveguide, the doped region of Si was triangular as described in Fig. 1. The bending radius was 5 µm for the wire waveguides and 10 µm for the rib waveguides. The ring of the MRR device was a rib waveguide with the thin slab inside the ring removed to suppress unwanted thermal diffusion. The ring was shaped into a rounded square with 7.14-µm long straight parts to obtain a 30-µm long control section. The doped region was gradually widened in the straight part, while the rounded part was maintained at one half-width of the waveguide. We also fabricated similar devices with a 50 nm thick TiN heater placed 2.1 µm above the signal waveguide for comparison.
Fig. 5. Fabricated devices: (a-b) Mach–Zehnder interferometers (MZI)-type device without (a) and with (b) a TiN heater. (c-d) Microring resonators (MRR)-type device without (c) and with (d) a TiN heater. The black section in the optical heater is the n+ doped region.
5. Measurements
Figure 6(a) shows the measurement setup. Lasers 1 and 2 were benchtop tunable lasers for the control and signal lights, respectively. Device under test (DUT) was MZI- or MRR-type device. The remaining components were an erbium-doped fiber amplifier (EDFA, 30 dBm maximum output), polarization-maintaining fibers with a lens module for light input and output from an on-chip spot-size converter (SSC), a bandpass filter (BPF, 1 nm bandwidth) and an optical power meter (PM). Transverse-electric-polarized light was launched on the chip with a loss of 3.7 dB, including 1.2 dB in the lens module and 2.5 dB at the SSC. Figure 6(b) shows the transmission spectrum of the directional coupler. The signal wavelength at a peak of the through port and the control light wavelength at a peak of the tap port were both set near 1550 nm. The coupling loss at the directional coupler was evaluated as 0.9 dB.
Fig. 6. (a) Experimental setup for optical control; (b) transmission spectrum of the on-chip directional coupler for wavelength demultiplexing.
Panels (a), (b) and (c) of Fig. 7 show the transmissions of the MZI devices with g = 0.5, 1.0 and 1.5 µm, respectively, as functions of Pctrl of the optical heater. Panel (d) shows the corresponding results for the electrical heater. Here Pctrl denotes the estimated power input to the control section, which was varied by controlling the EDFA gain. All responses were sinusoidal, although the initial phase of the MZI differed between the two devices owing to nonuniform fabrication. The closer control waveguide improved the heating efficiency; Pπ was 17, 19, and 24 mW for the optical heaters with g = 0.5, 1.0, and 1.5 µm, respectively, and 22.5 mW for the electrical heater (larger than that of the optical heater with smaller g). The output destabilized and was no longer reproducible when Pctrl exceeded 50 mW. This deterioration was attributed to damage at the Si SSC caused by the two-photon absorption, and might be ameliorated by employing a SiN SSC suitable for high-power input [24]. Therefore, we limited Pctrl to be less than 50 mW, and we could not evaluate Pπ from repeated sinusoidal response in this experiment. Figure 7(d) summarizes the measured Pπ. The experimental results of the optical heater roughly agreed with the simulation results (solid line). The theoretical value of Pπ for the electrical heater was 22.4 mW and deviated from the experimental result by only 0.1 mW.
Fig. 7. Measured operations of MZI devices controlled (a), (b), (c) optically with g = 0.5 µm, 1.0 µm and 1.5 µm, respectively, and (d) electrically. (e) Evaluation of Pπ versus g.
The resonance tuning of the MRR device is shown in Fig. 8. The resonance dip in the transmission spectrum with slowly varying Pout was attributed to the directional coupler. However, for the optical heater with g = 0.5 µm, no resonance was observed. This is because in the MRR device, the coupling between signal waveguide and control waveguide is larger than the straight waveguides due to the increased field penetration to outside at bends. For g = 0.5 µm, light coupled into the MRR from the bus waveguide propagates counterclockwise and particularly decreases at the upper left bend and is partly transferred to the control waveguide, as seen in Fig. 9(a), which severely dampens the resonance. For g = 1.0 µm, such light coupling is suppressed, as seen in Fig. 9(b), and the resonance can occur. Therefore, the optical heaters with g = 1.0 and 1.5 µm produced clear resonances that red-shifted with increasing Pctrl. As this optical transmission spectrum was measured by sweeping the signal wavelength, the control light was efficiently demultiplexed by the directional coupler only at the peak of the slow variation. Therefore, the effective control power was lower than Pctrl except at the peak, and its ratio changed with Pctrl. For this reason, the resonance shift was not always proportional to Pctrl. For the optical heater with g = 1.0 µm and Pctrl <12 mW, the tuning efficiency (0.18 nm/mW) was comparable to that of the electrical heater. As the spectral dip of the resonance was roughly 1 nm width, Pctrl = 4 mW was sufficient to switch the resonance on and off with an extinction ratio of 6 dB. At higher Pctrl and g = 1.5 µm, the efficiency was apparently as low as 0.08 nm/mW for the above-stated reason. Meanwhile, the electrical heater provided a clear and stable resonance shift that increased linearly with Pctrl.
Fig. 8. Resonance shifts of MRR devices with optical heaters: (a) g = 0.5 µm, (b) 1.0 µm and (c) 1.5 µm. Each graph is centered at λ = 1552.5 nm. The value in each curve indicates Pctrl at the maximum of the curve. (d) That with the electrical heater. (e) Evaluation of tuning efficiency.
Fig. 9. Simulated signal light propagation in MRR device with (a) g = 0.5 µm, (b) g = 1.0 µm at resonant wavelength.
We also evaluated the temporal response of the MZI device with the optical heater. The control light was modulated using a LiNbO3 modulator with a 1 kHz square-wave voltage. The output waveform of the signal light was observed on an oscilloscope (see Fig. 10). For g = 0.5, 1.0, and 1.5 µm, the 10%‒90% rise times were 14, 13, and 13 µs, respectively, and the 90%‒10% fall times were 13, 12, and 12 µs, respectively. The response time of such thermo-optic devices is determined by its heat capacity, and those of electrical heaters reported ranges from 1 µs [13] to 30 µs [25]. The response of our device can be further reduced by optimizing the waveguide length.
Fig. 10. Temporal response of the optical heater: (a) waveform of the 1 kHz output signal, (b) rising edge, and (c) falling edge.
6. Conclusion
We proposed and demonstrated a simple all-optic control in Si photonics using an optical heater. Control light is absorbed in the heavily doped control waveguide of the optical heater and the adjacent signal waveguide is controlled via heat diffusion. As these waveguides are closely placed and connected through a thermally conductive Si thin slab, this optical heater achieves higher heating efficiency than an electrical heater formed in the SiO2 cladding above the Si waveguides. All structures are fabricated by the standard wafer process of Si photonics; therefore, the device can be used when a photonic device must be remotely operated without electrical wirings.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. R. L. Espinola, M. C. Tsai, J. T. Yardley, and R. M. Osgood, “Fast and low-power thermooptic switch on thin silicon-on-insulator,” IEEE Photonics Technol. Lett. 15(10), 1366–1368 (2003). [CrossRef]
2. Y. Shoji, K. Kintaka, S. Suda, H. Kawashima, T. Hasama, and H. Ishikawa, “Low-crosstalk 2×2 thermo-optic switch with silicon wire waveguides,” Opt. Express 18(9), 9071–9075 (2010). [CrossRef]
3. H. Yu, D. Ying, M. Pantouvaki, J. Van Campenhout, P. Absil, Y. Hao, J. Yang, and X. Jiang, “Trade-off between optical modulation amplitude and modulation bandwidth of silicon micro-ring modulators,” Opt. Express 22(12), 15178–15189 (2014). [CrossRef]
4. M. Pantouvaki, S. A. Srinivasan, Y. Ban, P. De Heyn, P. Verheyen, G. Lepage, H. Chen, J. De Coster, N. Golshani, S. Balakrishnan, and P. Absil, “Active components for 50 Gb/s NRZ-OOK optical interconnects in a silicon photonics platform,” J. Lightwave Technol. 35(4), 631–638 (2017). [CrossRef]
5. F. Shinobu, N. Ishikura, Y. Arita, T. Tamanuki, and T. Baba, “Continuously tunable slow-light device consisting of heater-controlled silicon microring array,” Opt. Express 19(14), 13557–13564 (2011). [CrossRef]
6. P. Dong, R. Shafiiha, S. Liao, H. Liang, N. N. Feng, D. Feng, G. Li, X. Zheng, A. V. Krishnamoorthy, and M. Asghari, “Wavelength-tunable silicon microring modulator,” Opt. Express 18(11), 10941–10946 (2010). [CrossRef]
7. H. Shoman, H. Jayatilleka, A. H. Park, A. Mistry, N. A. Jaeger, S. Shekhar, and L. Chrostowski, “Compact wavelength-and bandwidth-tunable microring modulator,” Opt. Express 27(19), 26661–26675 (2019). [CrossRef]
8. M. J. Strain, C. Lacava, L. Meriggi, I. Cristiani, and M. Sorel, “Tunable Q-factor silicon microring resonators for ultra-low power parametric processes,” Opt. Lett. 40(7), 1274–1277 (2015). [CrossRef]
9. L. Chen, N. Sherwood-Droz, and M. Lipson, “Compact bandwidth-tunable microring resonators,” Opt. Lett. 32(22), 3361–3363 (2007). [CrossRef]
10. J. Sun, R. Kumar, M. Sakib, J. B. Driscoll, H. Jayatilleka, and H. Rong, “A 128 Gb/s PAM4 silicon microring modulator with integrated thermo-optic resonance tuning,” J. Lightwave Technol. 37(1), 110–115 (2019). [CrossRef]
11. M. T. Wade, J.M. Shainline, J. S. Orcutt, C. Sun, R. Kumar, B. Moss, M. Georgas, R. J. Ram, V. Stojanović, and M. A. Popović, “Energy-efficient active photonics in a zero-change, state-of-the-art CMOS process,” In Optical Fiber Communication Conference, Optica Publishing Group, Tu2E-7 (2014).
12. G. Li, X. Zheng, J. Yao, H. Thacker, I. Shubin, Y. Luo, K. Raj, J. E. Cunningham, and A. V. Krishnamoorthy, “25Gb/s 1V-driving CMOS ring modulator with integrated thermal tuning,” Opt. Express 19(21), 20435–20443 (2011). [CrossRef]
13. H. Jayatilleka, H. Shoman, L. Chrostowski, and S. Shekhar, “Photoconductive heaters enable control of large-scale silicon photonic ring resonator circuits,” Optica 6(1), 84–91 (2019). [CrossRef]
14. A. D. Neira, G. A. Wurtz, P. Ginzburg, and A. V. Zayats, “Ultrafast all-optical modulation with hyperbolic metamaterial integrated in Si photonic circuitry,” Opt. Express 22(9), 10987–10994 (2014). [CrossRef]
15. T. Tanabe, M. Notomi, S. Mitsugi, A. Shinya, and E. Kuramochi, “All-optical switches on a silicon chip realized using photonic crystal nanocavities,” Appl. Phys. Lett. 87(15), 151112 (2005). [CrossRef]
16. Y. Feng, Y. Zheng, F. Zhang, J. Yang, T. Qin, and W. Wan, “Passive fine-tuning of microcavity whispering gallery mode for nonlinear optics by thermo-optical effect,” Appl. Phys. Lett. 114(10), 101103 (2019). [CrossRef]
17. K. D. Heylman and R. H. Goldsmith, “Photothermal mapping and free-space laser tuning of toroidal optical microcavities,” Appl. Phys. Lett. 103(21), 211116 (2013). [CrossRef]
18. H. C. Tapalian, J. P. Laine, and P. A. Lane, “Thermooptical switches using coated microsphere resonators,” IEEE Photonics Technol. Lett. 14(8), 1118–1120 (2002). [CrossRef]
19. R. Wang, A. Vasiliev, M. Muneeb, A. Malik, S. Sprengel, G. Boehm, M. C. Amann, I. Šimonytė, A. Vizbaras, K. Vizbaras, and R. Baets, “III–V-on-silicon photonic integrated circuits for spectroscopic sensing in the 2–4 µm wavelength range,” Sensors 17(8), 1788 (2017). [CrossRef]
20. J. Zhu, Ş Kaya Özdemir, L. He, and L. Yang, “Optothermal spectroscopy of whispering gallery microresonators,” Appl. Phys. Lett. 99(17), 171101 (2011). [CrossRef]
21. T. Murai, Y. Shoji, and T. Mizumoto, “Efficient light-to-heat conversion by optical absorption of a metal on an Si microring resonator,” J. Lightwave Technol. 37(10), 2223–2231 (2019). [CrossRef]
22. D. Dai, H. Wu, and W. Zhang, “Utilization of field enhancement in plasmonic waveguides for subwavelength light-guiding, polarization handling, heating, and optical sensing,” Materials 8(10), 6772–6791 (2015). [CrossRef]
23. N. Ota, T. Miyauchi, and H. Shimizu, “221 K Local photothermal heating in a Si plasmonic waveguide loaded with a Co thin film,” Sensors 21(19), 6634 (2021). [CrossRef]
24. Y. Maegami, M. Okano, G. Cong, M. Ohno, and K. Yamada, “Completely CMOS compatible SiN-waveguide-based fiber coupling structure for Si wire waveguides,” Opt. Express 24(15), 16856–16865 (2016). [CrossRef]
25. K. Tanizawa, K. Suzuki, M. Toyama, M. Ohtsuka, N. Yokoyama, K. Matsumaro, M. Seki, K. Koshino, T. Sugaya, S. Suda, and G. Cong, “Ultra-compact 32× 32 strictly-non-blocking Si-wire optical switch with fan-out LGA interposer,” Opt. Express 23(13), 17599–17606 (2015). [CrossRef]
