We demonstrate the first sub-100 μm silicon Mach-Zehnder modulators (MZMs) that operate at >10 Gb/s, by exploiting low-dispersion slow-light in lattice-shifted photonic crystal waveguides (LSPCWs). We use two LSPCW-MZM structures, one with LSPCWs in both arms of the MZM, and the other with an LSPCW in only one of the arms. Using the first structure we demonstrate 10 Gb/s operation with a operating bandwidth of 12.5 nm, in a device with a phase-shifter length of only 50 μm. Using the second structure, owing to a larger group index as well as lower spectral noise, we demonstrate 40 Gb/s operation with a phase-shifter length of only 90 μm, which is more than an order-of-magnitude shorter than most 40 Gb/s MZMs.
©2012 Optical Society of America
Silicon Mach-Zehnder optical modulators (MZMs) are generally considered to be versatile compared to resonator and electro-absorption type silicon modulators, due to their amplitude- and phase-modulation capabilities while also having a large working spectrum . One distinctive challenge of the MZMs, however, has been their long device lengths of millimeter-order. Although there are carrier-injection type MZMs as short as 200 μm, they come at the expense of modulation speed and require complex signal pre-emphasis to operate at 10 Gb/s [2, 3]. Speeds of 40 Gb/s and above have been reported in other MZMs, but because they operate by carrier-depletion or accumulation, their device lengths are mostly 1 mm or longer [4–8]. While the length of MZMs can be reduced by employing inter-digitated  and other p-n structures that require complex doping profiles , the length can also be reduced significantly by using slow-light structures. By enhancing the light-matter interaction, optical phase-shift can be achieved in much shorter propagation lengths.
Photonic crystal waveguides (PCW) are one such structure that can generate slow-light. As PCWs can easily exhibit a large group index ng ≈30 – almost 10 times larger than standard rib-waveguides – PCW modulators can be potentially an order-of-magnitude shorter. In fact, we recently demonstrated a 10 Gb/s PCW-MZM with a length of only 200 μm [10, 11], which operated by carrier-depletion and, at the same time, with a peak-to-peak drive voltage (Vpp) smaller than that of the shortest carrier-injection MZM . Such PCW-MZMs can be fabricated by standard CMOS-compatible process in the case that the PCW is cladded by SiO2. Compared with Si rib MZMs, the excess loss related with the PCW can be almost negligible − the propagation loss and junction loss between the PCW and Si wire can be both less than 1 dB . Thus we can take advantage of slow light with minimal insertion-loss penalty. The PCW-MZM above operated at ng = 20 – 30, and exhibited a narrow operating bandwidth of <1 nm, partly because of the asymmetric MZM that was employed. The operating bandwidth of PCW-MZMs can be widened by employing lattice-shifted PCWs (LSPCWs), where the spatial shift of certain holes can modify the band structure to produce low-dispersion slow-light [13–16]. As the bandwidth of the constant-ng region increases, the bandwidth of the LSPCW-MZM would increase also.
Meanwhile to shorten the device further, one obvious solution is to operate at an even larger ng, in order to compensate for the reduced phase-shifter length. This is not so trivial, however, for the following reason. Due to the narrower bandwidth at large ng (which are inversely proportional) , imperfections in the PCWs can cause random phase-difference between the two arms of the MZM, making it difficult for the eye to open. One could expect the low-dispersion slow-light in LSPCWs to mitigate this issue, however low-dispersion with large ng ≳ 30 is not easily attainable in SiO2-clad LSPCWs . One solution, which we propose below, is to use MZMs with a LSPCW in only one of the MZM arms. Hereafter, we refer to such structure as Single-LSPCW-MZMs, and use the term Dual-LSPCW-MZMs for devices with LSPCWs in both arms.
In this paper, we demonstrate the first sub-100 μm MZMs to operate at >10 Gb/s using Dual- and Single-LSPCW-MZMs, with wide-band and high-speed operations, respectively. We demonstrate 10 Gb/s operation of a Dual-LSPCW-MZM with only 50 μm length, and with 12.5 nm bandwidth made possible by the LSPCW. We also demonstrate 40 Gb/s operation of a 90 μm Single-LSPCW-MZM, which is less than 1/5th of the next-smallest MZM capable of 40 Gb/s . This paper is structured as follows. Section 2 details the device structures, including the optical properties of the LSPCSWs that are employed. Section 3 describes the experimental setup. Section 4 describes the DC optical and 10 Gb/s modulation characteristics of the 50 μm Dual-LSPCW-MZM. Section 5 describes the basic optical properties of the 90 μm Single-LSPCW-MZM, and reports on its RF modulation at 10 Gb/s and 40 Gb/s. Finally we discuss and make concluding remarks in Section 6.
2. Device structure
Figure 1(a) shows the schematic of the Dual-LSPCW-MZM. It comprises of a symmetric Mach-Zehnder interferometer, with each arm containing a p-n doped LSPCW phase-shifter of length LPCW. The LSPCW itself is a SiO2-clad PCW, where the PCW is completely buried in SiO2 after the holes are etched, all by standard CMOS process. It has a triangular lattice of holes, but with the third rows of holes from the center shifted along the direction of optical propagation, as shown in Fig. 2(a) . The band structure of the LSPCW can be modified and low-dispersion slow-light of different ng can be generated, by adjusting the shift-parameter s . The parameters of the LSPCW used here are LPCW = 50 μm, lattice constant a = 400 nm, target hole diameter 2r = 215 nm, s = 95 nm and slab-thickness of 210 nm. The p-n diode is formed by moderately-doped p (1 × 1013 cm−2 dose) and n (6 × 1012 cm−2), and highly-doped p+ and n+ (both 4 × 1015 cm−2) regions, the latter separated by 4 μm. There are also thermo-optic (TO) phase-tuners integrated into the MZM to enable fine-tuning of phases at the device output. These are 240 μm long to ensure phase-shifts of >2π can be applied easily, although they can be shortened significantly. The TO phase-tuners are so-called to distinguish them from the LSPCW p-n phase-shifters. The lumped RF electrodes are laid-out in a GSGSG configuration: the widths of the G and S electrodes are 100 μm and 10 μm respectively, they are separated by 19 μm and are 0.75 μm thick. We note that these electrodes are not optimized for our short devices, and we believe they can be significantly improved and more compact.
Figure 1(b) shows the schematic of the Single-LSPCW-MZM. It comprises of an asymmetric Mach-Zehnder interferometer, with a p-n doped LSPCW phase-shifter in only one of the arms. The LSPCW design parameters are similar to that described above, but with s = 80 nm and LPCW = 90 μm, where the smaller s is to obtain larger ng but with moderately-low dispersion. The other MZM arm contains a delay line, which is designed to match the group delay between the two arms at a specific, target group index (ntarget) of the LSPCW. In other words, Ldelay = ngtarget∙LPCW/ngdelay, where Ldelay and ngdelay are the length and group index of the delay line waveguide, respectively. Here, we design Ldelay = 643 μm with ngdelay = 4.2 so that the group delays are matched when ngtarget = 30 and LPCW = 90 μm. In this device we also integrate TO phase-tuner and variable attenuator, in order to fine-tune the phase and amplitude differences between the arms. The heater length of the phase-tuner and attenuator are 240 μm and 150 μm respectively, although again they can be shortened significantly.
2.3 LSPCW characteristics
Figure 2(b) shows the transmission and ng spectra of stand-alone LSPCWs with a length of 200 μm. When s = 0 nm, the transmission band is ~25 nm wide, bound by the light-line and the band-edge on the short- and long-wavelength sides, respectively. Within this band, ng increases monotonically and rises more sharply towards the band-edge, which is consistent with the band structure in Ref . As s increases, the ng spectrum becomes flatter within the transmission-band, particularly when s = 95 nm which exhibits a low-dispersion bandwidth of 16 nm with ng = 23 ± 10%. In terms of the MZMs, low-dispersion is desirable for wide-band operation but there is a clear trade-off between the value of ng and the low-dispersion bandwidth. In Sections 4 and 5, we demonstrate an operating bandwidth of 12.5 nm in the Dual-LSPCW-MZM with s = 95 nm, while the bandwidth is reduced for a larger ng in the Single-LSPCW-MZM with s = 80 nm.
3. Experimental setup
TE-polarized light from a CW laser is coupled to the device through a lensed-fiber and an on-chip spot-size converter. In the modulation experiments, the device is driven electrically with 231 − 1 non-return-to-zero pseudo-random bit-sequence signals, in push-pull and single-ended configurations for the Dual- and Single-LSPCW-MZMs, respectively. The drive signal from a pulse-pattern generator (PPG) is electrically amplified and combined with a bias offset through a bias-tee, before being delivered to the device through RF probes. No RF termination is applied at be back-end of the device, as we have observed that this lowers the required drive voltage . A separate probe is used to deliver DC power to the TO phase-tuners and attenuators. The modulated optical signal is amplified optically and O/E converted, then observed on a Agilent 86100C sampling oscilloscope or a bit-error rate (BER) tester.
In the 10 Gb/s modulation experiments, we use Anritsu MP1800A for the PPG and BERT, and Agilent 11982A for the O/E converter. In the 40 Gb/s experiment, Alnair Labs SeBERT-1040C and Alnair Labs ORC-400 are used for the PPG and O/E converter, respectively, while the BER is calculated from the eye diagram.
4. Device characterization: Dual-LSPCW-MZM
4.1 Fabricated device
Figure 3 shows the optical and scanning electron micrographs of the device fabricated by CMOS-compatible process. Currently the overall device footprint is much larger than the actual LSPCW phase-shifters, since our focus has been more on the short phase-shifter length than a compact footprint. The footprint can be reduced significantly, for example by removing the outer G electrodes and narrowing the central G electrode. The TO phase-tuners can also be shortened by more than a factor of 5, and overall we believe that the footprint of the optical components can be reduced to <3000 μm2.
4.2DC optical characteristics
Figure 4 shows the transmission spectrum of the 50 μm Dual-LSPCW-MZM with s = 95 nm. The passive fiber-to-fiber and on-chip insertion losses are 14.8 dB and 9.1 dB, respectively. Of the on-chip losses, ~2.4 dB are from the two MMI couplers, and 5.7 dB are from the coupling and propagation losses of the p-n doped LSPCWs. These losses, including those related to the LSPCWs, can be reduced further with improved design and fabrication [12, 19, 20]. Figure 4 also shows the ng spectrum of the LSPCW within the MZM, inferred from the group-delay spectrum of the whole structure. ng = 16.5 ± 10% over a roughly 14 nm bandwidth within the transmission band, while ng increases to >20 at the edges.
4.3RF Modulation at 10 Gb/s
Figure 5(a) shows the 10 Gb/s eye pattern of the Dual-LSPCW-MZM, showing an open eye even with a device length of only 50 μm. Here the device is operated at λ = 1562.4 nm where ng = 28, and is driven by a drive signal with DC bias (VDC) and peak-to-peak (Vpp) voltages of −4.0 V and 4.5 V, respectively. The dynamic extinction ratio (ER) is 1.2 dB, which was obtained by comparing the one-, zero- and noise-levels on the sampling oscilloscope, taking into account the spontaneous-emission noise floor from the optical amplifier. The excess optical loss is 11 dB at the device output, but this is reduced to 4.2 dB after amplification due to the gain saturation of the amplifier. Both the ER and loss will be improved with increased modulation efficiency. Here the BER is measured to be 2 × 10−5at a received power of 1.0 dBm.
Figure 5(b) shows the eye patterns at various wavelengths across a 12.5 nm bandwidth, between λ = 1544.9 − 1557.4 nm as indicated in Fig. 4. Here, only the laser wavelength has been changed, while all other experimental conditions including Vpp, TO phase-tuner and sampling oscilloscope display settings were kept constant. Even then, eye patterns with the same mid-voltage levels and with similar eye-openings can be observed across this bandwidth. There are wavelengths inside this bandwidth at which the eye patterns are offset vertically within the eye diagram, compared to those in Fig. 5(b), due to small oscillations in the transmission spectrum. However, such DC offsets can be adjusted easily by the TO phase-tuner, or even better, they can be removed automatically using a DC-block at the output of the O/E converter. Identical operation would then be possible over the 14 nm or larger range mentioned in Section 4.2. It is clear that the combination of LSPCW, integrated TO phase-tuners and symmetric MZM enables wide-band operation in a compact device.
5. Device characterization: Single-LSPCW-MZM
5.1 DC optical characteristics
Figure 6(a) shows the transmission spectrum of the 90 μm Single-LSPCW-MZM with s = 80 nm, and a passive on-chip insertion loss of 6.2 dB. Since this is an asymmetric MZM structure, there are spectral oscillations in the transmission spectrum. In this case the free-spectral range (FSR) of the peaks depends on the difference in group delay between the two arms: FSR(λ) = λ2/(LPCW |ng(λ) −□ngtarget|). Therefore the transmission spectrum is flat in the spectral region where the group delays match, while the oscillation peaks will merge closer as the group delays become mismatched. In principle the flat spectral region can be as wide as the bandwidth of the low-dispersion slow-light, by choosing appropriately the ngtarget and Ldelay of the device. Here, since the device is designed with ngtarget = 30, we can infer that ng(λ) ≈30 in the flat region of the transmission spectrum around λ = 1558 – 1563 nm. At other wavelengths, the spectral oscillations indicate that |ng(λ) – ngtarget| > 0, and we can infer from the ng spectrum in Fig. 2(b) that ng(λ) < 30 for λ < 1558 nm and ng(λ) > 30 for λ > 1563 nm.
We can determine directly whether ng(λ) is greater than or less than ngtarget, by measuring the transmission spectrum as a function of the heating power of the TO phase-tuner, as shown in Fig. 6(b). As the heating power is increased, the peaks blue-shift and red-shift for λ < 1558 nm and λ > 1563 nm, respectively. For the structure in Fig. 1(b), with the TO phase-tuner and the LSPCW on the same arm, it can be shown analytically that the peaks blue-shift and red-shift, when ng(λ) < ngtarget and ng(λ) > ngtarget, respectively. So if we want to operate the MZM at ng > 30, we know to choose λ > 1563 nm.
5.2 RF modulation at 10 Gb/s
Figure 7(a) shows the 10 Gb/s eye pattern of the 90 μm Single-LSPCW-MZM, at λ = 1565.4 nm where ng > 30 as described above. We note that although the device was designed with ngtarget = 30, it is still possible to operate at other ng’s. The device is driven at VDC = −4.0 V and Vpp = 4.5 V, and we achieve error-free operation with BER < 10−9, as well as a dynamic ER of 7.3 dB. These are improved modulation characteristics compared to the Dual-LSPCW-MZM, noting that here LPCW is almost doubled but it is driven in single-ended mode. This suggests that the single-LSPCW structure and the large ng played a large factor in the device performance.
Although we propose this device structure primarily for high-ng operation, it can also be used for lower-ng wide-band operation. Figure 7(b) shows the eye pattern at λ = 1560.65 nm and 1562.72 nm, indicating that similar eye pattern and hence modulation characteristics can be obtained over a 2 nm bandwidth within the “flat” region of the transmission spectrum in Fig. 6(a). This is a smaller bandwidth than the Dual-LSPCW-MZM above, due to the small bandwidth of the ng ≈30 region of the LSPCW itself. In principle the bandwidth can be as wide as that of the low-dispersion slow-light, by designing ntarget to equal ng of the slow-light and setting the delay-line length appropriately.
5.3 RF modulation at 40 Gb/s
Figure 8 shows the eye pattern of the 90 μm Single-LSPCW-MZM operating at 40 Gb/s, with VDC = −5.0 V, Vpp = 5.3 V and a wavelength similar to the 10 Gb/s case. The eye is still open, with a BER calculated from the eye pattern of 10−5 order, so this is the first demonstration of 40 Gb/s modulation in a sub-100 μm MZM. Although there is increased noise, we note that this is achieved with very rough control over the drive voltage, and with un-optimized electrodes. Even then, by appearance this is of a level similar to the 500 μm device in Ref , which is the next smallest 40 Gb/s MZM that we are aware of. Therefore the device performance will be improved with the drive electronics and electrode design.
In summary we have experimentally demonstrated the first sub-100 μm Mach-Zehnder modulators that operate at >10 Gb/s, by incorporating slow-light photonic crystal waveguides. We demonstrated 10 Gb/s operation with 12.5 nm bandwidth in a Dual-LSPCW-MZM structure of only 50 μm phase-shifter length, while we also demonstrated 40 Gb/s operation in a 90 μm Single-LSPCW-MZM. Figure 9 compares the phase-shifter length and Vpp of our devices [11, 21] with other MZMs: carrier- injection devices that operate at 10 − 30 Gb/s [2, 3], and carrier-depletion or accumulation devices that operate at 10 − 30 Gb/s [9, 22–28] and at >40 Gb/s [4–8, 18]. The results indicate the reduced lengths in our device without increasing Vpp, particularly in the case of 40 Gb/s operation. While there are other performance metrics such as operating bandwidth, ER and losses that need to be considered and improved, it is exciting nevertheless that MZMs can be sub-100 μm in length.
This brings the size of MZM phase-shifters down to a scale that is comparable to micro-ring and electro-absorption modulators, while retaining operating bandwidth that the latter devices usually lack . While slow-light plays a significant role in making this possible, electrical factors may have also played a role, given that the our phase-shifter lengths are ≲1% and a few percent of the 10 Gb/s and 40 Gb/s RF signal wavelengths, respectively. Certainly, our device operates even at 40 Gb/s despite the use of simple, un-optimized RF electrodes, whereas other MZMs have required more complex travelling-wave electrode designs [4, 6, 8, 18]. A better understanding of such electrical aspects may improve the device performance, and truly bring the device footprint down close to that of micro-ring and electro-absorption modulators.
This work was partly supported by the FIRST Program of JSPS.
References and links
1. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thomson, “Silicon optical modulators,” Nat. Photonics 4(8), 518–526 (2010). [CrossRef]
3. S. Akiyama, T. Baba, M. Imai, T. Akagawa, M. Takahashi, N. Hirayama, H. Takahashi, Y. Noguchi, H. Okayama, T. Horikawa, and T. Usuki, “12.5-Gb/s operation with 0.29-V·cm V(π)L using silicon Mach-Zehnder modulator based-on forward-biased pin diode,” Opt. Express 20(3), 2911–2923 (2012). [CrossRef] [PubMed]
4. L. Liao, A. Liu, D. Rubin, J. Basak, Y. Chetrit, H. Nguyen, R. Cohen, N. Izhaky, and M. Paniccia, “40 Gbit/s silicon optical modulator for highspeed applications,” Electron. Lett. 43(22), 1196–1197 (2007). [CrossRef]
5. D. J. Thomson, F. Y. Gardes, J. M. Fedeli, S. Zlatanovic, H. Youfang, B. P. P. Kuo, E. Myslivets, N. Alic, S. Radic, G. Z. Mashanovich, and G. T. Reed, “50-Gb/s silicon optical modulator,” IEEE Photon. Technol. Lett. 24(4), 234–236 (2012). [CrossRef]
6. M. Ziebell, D. Marris-Morini, G. Rasigade, J. M. Fédéli, P. Crozat, E. Cassan, D. Bouville, and L. Vivien, “40 Gbit/s low-loss silicon optical modulator based on a pipin diode,” Opt. Express 20(10), 10591–10596 (2012). [CrossRef] [PubMed]
8. H. Yu, M. Pantouvaki, J. Van Campenhout, D. Korn, K. Komorowska, P. Dumon, Y. L. Li, P. Verheyen, P. Absil, L. Alloatti, D. Hillerkuss, J. Leuthold, R. Baets, and W. Bogaerts, “Performance tradeoff between lateral and interdigitated doping patterns for high speed carrier-depletion based silicon modulators,” Opt. Express 20(12), 12926–12938 (2012). [CrossRef] [PubMed]
9. M. R. Watts, W. A. Zortman, D. C. Trotter, R. W. Young, and A. L. Lentine, “Low-voltage, compact, depletion-mode, silicon Mach-Zehnder modulator,” IEEE J. Sel. Top. Quantum Electron. 16(1), 159–164 (2010). [CrossRef]
11. H. C. Nguyen, Y. Sakai, M. Shinkawa, N. Ishikura, and T. Baba, “Photonic crystal silicon optical modulators: carrier-injection and depletion at 10 Gb/s,” IEEE J. Quantum Electron. 48(2), 210–220 (2012) (Invited Paper). [CrossRef]
12. M. Shinkawa, N. Ishikura, Y. Hama, K. Suzuki, and T. Baba, “Nonlinear enhancement in photonic crystal slow light waveguides fabricated using CMOS-compatible process,” Opt. Express 19(22), 22208–22218 (2011). [CrossRef] [PubMed]
13. Y. Hamachi, S. Kubo, and T. Baba, “Slow light with low dispersion and nonlinear enhancement in a lattice-shifted photonic crystal waveguide,” Opt. Lett. 34(7), 1072–1074 (2009). [CrossRef] [PubMed]
14. A. Hosseini, X. C. Xu, H. Subbaraman, C. Y. Lin, S. Rahimi, and R. T. Chen, “Large optical spectral range dispersion engineered silicon-based photonic crystal waveguide modulator,” Opt. Express 20(11), 12318–12325 (2012). [CrossRef] [PubMed]
15. A. Y. Petrov and M. Eich, “Zero dispersion at small group velocities in photonic crystal waveguides,” Appl. Phys. Lett. 85(21), 4866–4868 (2004). [CrossRef]
16. J. Li, T. P. White, L. O’Faolain, A. Gomez-Iglesias, and T. F. Krauss, “Systematic design of flat band slow light in photonic crystal waveguides,” Opt. Express 16(9), 6227–6232 (2008). [CrossRef] [PubMed]
17. T. Baba, “Slow light in photonic crystals,” Nat. Photonics 2(8), 465–473 (2008). [CrossRef]
18. A. Brimont, D. J. Thomson, P. Sanchis, J. Herrera, F. Y. Gardes, J. M. Fedeli, G. T. Reed, and J. Martí, “High speed silicon electro-optical modulators enhanced via slow light propagation,” Opt. Express 19(21), 20876–20885 (2011). [CrossRef] [PubMed]
19. S. Rahimi, A. Hosseini, X. C. Xu, H. Subbaraman, and R. T. Chen, “Group-index independent coupling to band engineered SOI photonic crystal waveguide with large slow-down factor,” Opt. Express 19(22), 21832–21841 (2011). [CrossRef] [PubMed]
20. M. Spasenović, T. P. White, S. Ha, A. A. Sukhorukov, T. Kampfrath, Y. S. Kivshar, C. M. de Sterke, T. F. Krauss, and L. K. Kuipers, “Experimental observation of evanescent modes at the interface to slow-light photonic crystal waveguides,” Opt. Lett. 36(7), 1170–1172 (2011). [CrossRef] [PubMed]
21. H. C. Nguyen, M. Shinkawa, N. Ishikura, and T. Baba, “1.6 Vpp, 10 Gb/s Si photonic crystal optical modulators without RF termination,” in JSAP Spring Meeting, (Tokyo, Japan, 2012), pp. 18a-F14–18.
22. T. Y. Liow, K. W. Ang, Q. Fang, J. F. Song, Y. Z. Xiong, M. B. Yu, G. Q. Lo, and D. L. Kwong, “Silicon modulators and germanium photodetectors on SOI: monolithic integration, compatibility, and performance optimization,” IEEE J. Sel. Top. Quantum Electron. 16(1), 307–315 (2010). [CrossRef]
23. D. M. Gill, S. S. Patel, M. Rasras, K. Y. Tu, A. E. White, Y. K. Chen, A. Pomerene, D. Carothers, R. L. Kamocsai, C. M. Hill, and J. Beattie, “CMOS-compatible Si-ring-assisted Mach-Zehnder interferometer with internal bandwidth equalization,” IEEE J. Sel. Top. Quantum Electron. 16(1), 45–52 (2010). [CrossRef]
24. G. Rasigade, M. Ziebell, D. Marris-Morini, J. M. Fédéli, F. Milesi, P. Grosse, D. Bouville, E. Cassan, and L. Vivien, “High extinction ratio 10 Gbit/s silicon optical modulator,” Opt. Express 19(7), 5827–5832 (2011). [CrossRef] [PubMed]
25. L. Liao, D. Samara-Rubio, M. Morse, A. S. Liu, D. Hodge, D. Rubin, U. D. Keil, and T. Franck, “High speed silicon Mach-Zehnder modulator,” Opt. Express 13(8), 3129–3135 (2005). [CrossRef] [PubMed]
26. A. M. Gutierrez, A. Brimont, G. Rasigade, M. Ziebell, D. Marris-Morini, J. M. Fedeli, L. Vivien, J. Marti, and P. Sanchis, “Ring-assisted Mach-Zehnder interferometer silicon modulator for enhanced performance,” J. Lightwave Technol. 30(1), 9–14 (2012). [CrossRef]
27. T. Baehr-Jones, R. Ding, Y. Liu, A. Ayazi, T. Pinguet, N. C. Harris, M. Streshinsky, P. Lee, Y. Zhang, A. E. J. Lim, T. Y. Liow, S. H. G. Teo, G. Q. Lo, and M. Hochberg, “Ultralow drive voltage silicon traveling-wave modulator,” Opt. Express 20(11), 12014–12020 (2012). [CrossRef] [PubMed]