Open Access
Add to BibSonomyAbstract
On-chip optical isolators not requiring the use of magneto-optical materials has become a long-standing challenge in integrated optics. Here, we demonstrate that a traditional travelling-wave modulator can effectively function as an optical isolator, when driven under a prescribed modulation condition. By using an off-shelve lithium niobate modulator, we achieve more than 12.5 dB isolation over an 11.3-THz bandwidth at telecommunication wavelengths with a fiber-to-fiber insertion loss of 5.5 dB, by employing only a single radio-frequency drive signal. We also verify that the proposed active isolator can be functional in a laser system to effectively prevent instability due to strong back reflections. Since travelling-wave modulators are common devices in III-V and silicon photonics, our simple but efficient architecture may provide a practical solution to non-reciprocal light routing in photonic integrated circuits.
© 2015 Optical Society of America
1. Introduction
The direction-dependent light propagation in an isolator requires the breaking of Lorentz reciprocity, which holds for linear and time-independent devices with symmetric permittivity and permeability [1,2]. In order to achieve optical isolation, current bulk and fiber optical systems utilize magneto-optical materials with asymmetric magnetic permeability. Although some waveguide-based isolators have been demonstrated by incorporating such materials [3–11], it is difficult to integrate them with other silicon and III-V photonic integrated circuits (PICs). Isolators not requiring the use of magneto-optical materials and with easy on-chip integration capability would be highly demanded. This has become a long-standing challenge in integrated optics. In recent years, there has been increasing interest to achieve isolation based on nonlinear optical effects [12–15] and time-varying modulation [16–21]. The use of nonlinear optics leads to unwanted power-dependent effects and the previously demonstrated actively modulated isolators still suffer from a variety of issues including low isolation, high insertion loss, or narrow operational bandwidth.
Time-dependent modulation of the refractive index in a linear system can break Lorentz reciprocity since the active modulation introduces the direction-dependent perturbation in the permittivity. Photonic transitions [17,22] and Aharonov-Bohm-like effects [23,24] can be also used to explain such a mechanism. Smith demonstrated the use of non-reciprocal frequency shift of an acousto-optic modulator and the filtering effect of a laser cavity to effectively function as an optical isolator [25]. Bhandare et al. experimentally demonstrated broadband optical isolation when two travelling-wave Mach-Zehnder modulators (MZMs) were driven to produce a single side-band modulation [16]. However, this configuration suffers from a 6-dB intrinsic loss and at least two driving signals are needed. Yu et al. proposed a single waveguide or a micro-ring system to produce photonic transitions between a pair of forward-propagating modes but not in backward-propagating modes [17]. Subsequent experimental demonstration has been presented with limited isolation [18]. Optical isolators can also be realized using two or more tandem phase modulators [19–21] with certain delays between the modulators but at least two modulators are needed to achieve certain isolation. In this paper, we propose and experimentally demonstrate that a single travelling-wave MZM can effectively work as an optical isolator without residual modulation on the forward-transmitted light.
2. Travelling-wave MZM functioning as isolator
In order to achieve the isolator, the typical optical output port located at the side of RF output of a MZM is now used as the optical input port. In this scenario, the forward transmission corresponds to the case of counter-propagation between the RF and light and backward to co-propagation. The MZM is biased at maximum transmission, while modulated by a square-wave RF signal with a peak-to-peak voltage of 2Vπ (Vπ is the voltage applied to achieve π-phase shift between two arms of an MZM). The working principle can be explained as follows. If the electro-optic phase change is linearly proportional to the applied voltage V(t), where V(t) is the voltage at time constant of t applied on the RF input, the total phase shifts accumulated along the whole length L for co- and counter-propagation are expressed as [16]
where +/− represent co- and counter-propagation, respectively; g = π/(LVπ) is a constant relating voltages and phase shifts; co and ce are optical and RF propagation velocities, respectively. To better understand the above equation, we consider the case of that co and ce are identical or very close. The total phase shifts for co-propagation isFor counter-propagation, it isAs shown in the above equations, the device experiences different phase changes for co- and counter-propagation. For co-propagation, the total phase shift along the whole length L only depends on the voltage at the particular time as this voltage signal is co-propagating with the light with similar velocities. For the counter-propagation, the total phase shift will be the integral for the voltage signal over time of L/co + L/ce. If one can find a drive voltage waveform so that the phase change is φ = 0 (or 2mπ in general case where m is an integer) in one direction and φ = (2m + 1)π, where m is an integer in the other direction, the light will pass in the first direction but be blocked in the other direction. This can be achieved by a square-wave RF signal with a peak-to-peak voltage of 2Vπ. Under this condition, for co-propagation (or backward transmission), the phase changes along the phase shifter can be either -π or π except at the transition moments, leading to very low transmission. In the counter-propagation (or forward transmission), if the integral of the voltage is carried out over 2nT, where T is the bit period and n is a nonzero positive integer, the total phase change becomes zero because of the sum of exact positive and negative voltages. Therefore, the condition to achieve zero-phase change becomes:
The above equation states that the sum of the times that the light and the RF propagating through the phase shifter equals to 2n time of the bit period. Under this condition, the forward-transmitted light experiences zero-phase change even with the modulation. Therefore, the active modulation does not degrade the transmitted light nor introduce excess loss under this condition.
We present simulation results in Fig. 1, where we assume co = ce and the condition of Eq. (4) is satisfied. It is to be noted that the exact velocity match is not required. For the forward light (counter-propagating with the RF), the transmission stays constant with a value of 1 as illustrated in Fig. 1(d), while for the backward light, the transmission is zero but with some transition spikes, whose peak values depend on the rise/fall times of the square waves and also the modulation bandwidth of the MZM. However, if the spikes’ pulse duration is much shorter than that a laser can response, the laser will see the average back reflection from the isolator, which makes the isolator work even with the presence of these spikes. Moreover, cascading more MZMs with asynchronous drives can reduce the peak value of the spikes if needed.
Fig. 1 Schematic and working principle of using a single travelling-wave modulator as an isolator. (a) Working diagram of an MZM-based isolator. The green and red arrows represent forward and backward transmissions. The RF input is at the side of the optical output. A square-wave RF signal actively modulates the MZM. (b) MZM transmission as a function of applied voltage, assuming it is biased at maximum transmission at 0 V. (c) A square-wave voltage with a peak-to-peak of 2Vπ drives the MZM to achieve isolation between forward and backward transmissions. (d) Simulated forward and backward transmissions while the MZM is modulated with the voltage shown in (c). In this simulation, the modulator bandwidth is assumed as 30 GHz. High forward transmission and low backward transmission demonstrate that the device acts as an optical isolator.
3. Experimental results
For a proof-of-concept demonstration, we use an off-shelf lithium niobate (LN) modulator with a 3-dB bandwidth more than 25 GHz and a phase length of ~5 cm. Assuming the optical and RF group indices are around 2.2, the bit period should be ~366 ps for n = 1 in Eq. (4), leading to the RF frequency of ~2.7 GHz. Figure 2(a) presents the normalized transmission as a function of applied DC voltage for three different wavelengths. The Vπ is ~2.3 V under the DC condition, but will increase with higher frequency and has a slight dependence on wavelength. With this information, we drive the MZM using an electrical amplifier that amplifies a square wave from a pattern generator to a peak-to-peak voltage output of ~5 V. We measured the transmission spectra with a low-speed photodetector, which averages the optical power over a period of 20 milliseconds. Figure 2(b) plots the fiber-to-fiber transmission spectra of the device when modulated at 2.75 GHz and also when the modulation is off. Comparing the forward transmission between the modulation on and off states, there is no observable difference, demonstrating that the active modulation does not introduce excess loss. The forward transmission also indicates that the total fiber-to-fiber insertion loss of MZM is about 5.5 dB over 90 nm (the bandwidth is limited by our test equipment). For the backward transmission with the RF on, the transmission drops to less than −18 dB, indicating that the isolation is more than 12.5 dB over the entire 90 nm wavelength range (11.3 THz).
Fig. 2 (a) Normalized transmission with different bias voltages for a commercial LN modulator at three different wavelengths. (b) Average transmission for the actively modulated isolator for a 90-nm wavelength span in the telecommunication wavelengths. More than 12.5 dB isolation has been achieved over the whole 90-nm wavelength range.
Next, we observed that the modulation on the forward-transmitted light could be minimized under the matching condition of Eq. (4). To do so, a continuous-wave (CW) laser with a power of 10 dBm at a wavelength of 1552.52 nm is launched in the MZM. The optical output power is about 4.5 dBm. By sweeping the RF frequency from 0.5 GHz to 11 GHz, the optical output response is monitored by a high-speed photodetector and an electrical spectrum analyzer (ESA). The results in Figs. 3(a) and (b) display the superimposed electrical spectra and also the peak power. Here the RF power measured from the ESA is proportional to the modulation depths. The results clearly demonstrate that the RF power drops significantly at frequencies of 2.75 GHz, 5.37 GHz, 8.05 GHz, and 10.59 GHz, corresponding to n = 1-4 in Eq. (4). At 2.75 GHz, the power drops to −75 dBm, only slightly larger than the noise level of −77 dBm.
Fig. 3 Forward transmission response to active modulation. (a) Superimposed electrical spectra with different RF driving frequencies. While the modulator is driven by a square-wave signal at a particular RF frequency, the electrical spectrum is collected with a center frequency at this frequency with a span of 100 MHz. (b) Peak electrical power as a function of RF driving frequency. The power dips indicate that the matching conditions in Eq. (4) are satisfied so that the active RF driving has minimum modulation on the MZM for the forward transmission.
We experimentally observed the time-domain isolation by measuring the optical output response with an oscilloscope. Figure 4(a) presents the modulated optical waveforms under different RF frequencies for the forward transmission (counter-propagation with RF). For a frequency of 2.75 GHz, the time-domain transmission is almost identical to the one with RF power off, while for 2.00 GHz and 4.00 GHz, the unwanted modulation become significantly larger. Figure 4(b) presents both the forward and backward optical transmission responses while the MZM is modulated at 2.75 GHz. For the backward optical transmission (co-propagation with the RF drive), the transmission is very low with only some transition spikes. This verifies the device is blocking the light from the backward transmission. These results confirm the simulation results presented in Fig. 1(d).
Fig. 4 Time-domain transmission for actively modulated isolator. (a) Forward transmission under different RF driving frequencies. The inset shows the voltage signal applied on the device at 2.75 GHz. (b) Forward and backward transmission under the driving frequency of 2.75 GHz. The discrepancy of the transmissions between forward and backward propagation implies successful isolation.
An isolator based on active modulation schemes could degrade the intensity and phase performance of a CW laser, which is transmitting through it. Here we confirm that our MZM-based isolator would not introduce measureable penalty for communication applications if driven under the matching condition shown in Eq. (4). In order to do so, the forward-transmitted light from the MZM is modulated by an electro-absorption modulator at 10 Gb/s and the bit error ratios (BER) are measured against the received optical power into error detection equipment. Figure 5 shows the BERs and eye diagrams at different driving frequencies. Comparing BERs at 2.75 GHz with that of RF power off (no active modulation), there is no observable optical power penalty. For non-matching driving frequencies such as 2.00 GHz and 4.00 GHz, 4-5 dB power penalties can occur at a BER of 10−9, which is attributed to intensity modulation from the isolator. Eye diagrams correlate well with the BER measurement. While this experiment confirms that electro-optic active isolators can be used in communication applications, we must be aware that other applications may require extremely low phase and intensity noises, which makes the active isolators less acceptable.
Fig. 5 BER and eye diagram measurement to verify that, for the transmitted light, the active modulation does not introduce the penalty for communication applications. When the MZM is driven at 2.75 GHz, no observable penalty has been found, compared the red curve (no modulation) with blue curve (2.75 GHz square waves). Under other RF frequencies, significant penalties have been found due to the intensity modulation imposed on the transmitted light.
In the final experiment, we demonstrate that the isolator can effectively block the back reflections and restore stable lasing operation of a distributed Bragg reflector (DBR) semiconductor laser. The DBR laser is coupled into an optical fiber and then launched into the MZM-based isolator. A back reflection is intentionally introduced through an optical loop using a 2x2 3-dB power coupler, as shown in Fig. 6(a). When a strong back reflection of −12 dB and −9 dB are present, the laser output, measured by a fast detector and oscilloscope, shows a chaotic intensity distribution [26], observed in the eye diagrams in Figs. 6(b) and 6(d). After we turn on the RF drive signal at 2.75 GHz, the laser output immediately becomes constant and stable in Figs. 6(c) and 6(e). This demonstration verifies that the isolator is functional, even though some reflection spikes are present. To the best of our knowledge, this is the first experiment to demonstrate that active isolators can be used to prevent back reflection in a laser system.
Fig. 6 The MZM-based isolator can block back reflections and restore stable lasing in a DBR laser system. (a) Optical setup. (b) and (d) Time-domain intensity of a laser system with −12 dB and −9 dB reflections (the reflection is referenced to the laser output power). Here, no any active modulation is applied on the MZM. Chaotic intensities indicate that the laser is operating in a very unstable status due to back reflections. (c) and (e) After the RF power is turned on, isolator is functional, so that the light intensities become constant and stable. The isolator hence restores the stable lasing operation.
4. Conclusion
In conclusion, we have demonstrated that a traditional single traveling-wave MZM can be used as an electro-optic isolator. The achieved isolation of >12.5 dB over an 11.3-THz bandwidth with an insertion loss of 5.5 dB signifies that this type of isolators can find practical applications. Furthermore, we verify that it is possible to find driving frequencies under which the active modulation has very little residual modulation on the transmitted light. We also experimentally demonstrated that such active isolators could be used in a laser system to prevent strong reflection.
The achieved isolation of 12.5 dB is mainly limited by transition spikes from the switching between the low and high voltages of the applied square waves. If the switching time is shorter than the MZM’s response time, these spikes can be greatly reduced. Therefore, a proper optimization of the drive signal can further increase the isolation. The eventually achievable isolation would be determined by the extinction ratios of the MZMs. In addition, although the active modulation in the proposed isolator does not introduce excess loss in theory, a practical MZM has coupling loss to fibers and waveguide propagation loss, as demonstrated in the LN MZM used in this paper. It is critical to have low insertion loss for on-chip isolators in applications.
Since MZMs are common in silicon and III-V PICs, on-chip isolators would be viable using the same techniques. Nevertheless, one must consider that for silicon and III-V MZMs, the length of phase shifter is typically a few millimeters [27], one order of magnitude shorter than that of LN modulators, which requires much higher modulation frequency. However, since the group index of silicon/III-V waveguides can be close to 4, two times that of LN waveguides, the driving frequency can be around 10 GHz. Moreover, passive optical and/or electrical delays can be intentionally introduced into the MZM to increase the propagation time, and hence a reduced driving frequency of less than 10 GHz is possible. In recent years, there is high demand to develop low-loss, high-bandwidth silicon and III-V MZMs with very low Vπ. This makes the active isolator proposed here more feasible and practical as the performance of MZMs is improved in the future.
Appendix: experimental methods
The single-drive MZM used here is a Fujitsu 7938EZ, with more information available at http://www.fujitsu.com. For the measurement in Fig. 2, the MZM is first set to the maximum transmission point, then a DC power supplier is used to apply a voltage from −4 V to 4 V. A CW laser at 10 dBm is launched into MZM, and the output is measured using a low-speed photodetector.
For high-speed measurement in Figs. 3 and 4, the MZM driver is an electrical amplifier (SHF 810) with a voltage swing of ~5V. The electrical square wave signals are generated by a pattern generator driven by a clock source. The output from the pattern generator has a voltage swing of 0.5 V. The optical output, with a power around 4.5 dBm, is measured using a high-speed photodetector with a 3-dB bandwidth of 30 GHz and a responsivity of 0.5 A/W. The output of the photodetector is connected to an electrical spectrum analyzer, which measures the RF spectrum up to 40 GHz. The clock frequency is swept from 0.5 GHz to 11 GHz. At each clock frequency, the spectrum over a 100-MHz span and centered at the aforementioned clock frequency is collected, and the peak power is measured. Figure 3(a) shows superimposed electrical spectra from 0.5 GHz to 11 GHz with a frequency step of 0.15 GHz. The real-time output power of the system in forward and backward configurations in Fig. 4 is measured by a fast oscilloscope (Agilent DCA 86100A) with an electro-optic module at 40 GHz.
For the BER measurement in Fig. 5, the output from the MZM is amplified by an erbium-doped fiber amplifier (EDFA) to ~15 dBm, which is launched into an electro-absorption modulator (EAM). The EAM is driven by another electrical amplifier with a voltage swing of ~3 V. The drive signal is a 10 Gb/s pseudorandom bit sequence (PRBS) of a length 231 – 1 from a pattern generator. The output of EAM is detected by an optical receiver, whose output connects to an error detection system.
Acknowledgments
We greatly thank Larry Buhl and S. Chandrasekhar for providing the DBR laser and modulator used in this paper, Young-Kai Chen, David Neilson, Jeffrey Sinsky, Guillermo Acevedo, Chia-Ming Chang, and Martin Zirngibl for helpful discussion, paper revision, and great support.
References and links
1. B. Saleh and M. Teich, Fundamentals of Photonics, Wiley Series in Pure and Applied Optics (John Wiley, 1991).
2. D. Jalas, A. Petrov, M. Eich, W. Freude, S. Fan, Z. Yu, R. Baets, M. Popovic, A. Melloni, J. D. Joannopoulos, M. Vanwolleghem, C. R. Doerr, and H. Renner, “What is - and what is not - an optical isolator,” Nat. Photonics 7(8), 579–582 (2013). [CrossRef]
3. H. Shimizu and Y. Nakano, “Fabrication and characterization of an InGaAsp/InP active waveguide optical isolator with 14.7 dB/mm TE mode nonreciprocal attenuation,” J. Lightwave Technol. 24(1), 38–43 (2006). [CrossRef]
4. W. Van Parys, B. Moeyersoon, D. V. Thourhout, R. Baets, M. Vanwolleghem, B. Dagens, J. Decobert, O. L. Gouezigou, D. Make, R. Vanheertum, and L. Lagae, “Transverse magnetic mode nonreciprocal propagation in an amplifying AlGaInAs/InP optical waveguide isolator,” Appl. Phys. Lett. 88(7), 071115 (2006). [CrossRef]
5. Y. Shoji, T. Mizumoto, H. Yokoi, I. Hsieh, and R. M. Osgood Jr., “Magneto-optical isolator with silicon waveguides fabricated by direct bonding,” Appl. Phys. Lett. 92(7), 071117 (2008). [CrossRef]
6. T. R. Zaman, X. Guo, and R. J. Ram, “Semiconductor waveguide isolators,” J. Lightwave Technol. 26(2), 291–301 (2008). [CrossRef]
7. L. Bi, J. Hu, P. Jiang, D. H. Kim, G. F. Dionne, L. C. Kimerling, and C. A. Ross, “On-chip optical isolation in monolithically integrated non-reciprocal optical resonators,” Nat. Photonics 5(12), 758–762 (2011). [CrossRef]
8. M. C. Tien, T. Mizumoto, P. Pintus, H. Kromer, and J. E. Bowers, “Silicon ring isolators with bonded nonreciprocal magneto-optic garnets,” Opt. Express 19(12), 11740–11745 (2011). [CrossRef] [PubMed]
9. Y. Shoji and T. Mizumoto, “Magneto-optical non-reciprocal devices in silicon photonics,” Sci. Technol. Adv. Mater. 15(1), 014602 (2014). [CrossRef]
10. L. Bi, J. Hu, P. Jiang, H. S. Kim, D. H. Kim, M. C. Onbasli, G. F. Dionne, and C. A. Ross, “Magneto-optical thin films for on-chip monolithic integration of non-reciprocal photonic devices,” Materials 6(11), 5094–5117 (2013). [CrossRef]
11. S. Ghosh, S. Keyvavinia, W. Van Roy, T. Mizumoto, G. Roelkens, and R. Baets, “Ce:YIG/Silicon-on-Insulator waveguide optical isolator realized by adhesive bonding,” Opt. Express 20(2), 1839–1848 (2012). [CrossRef] [PubMed]
12. J. Hwang, M. H. Song, B. Park, S. Nishimura, T. Toyooka, J. W. Wu, Y. Takanishi, K. Ishikawa, and H. Takezoe, “Electro-tunable optical diode based on photonic bandgap liquid-crystal heterojunctions,” Nat. Mater. 4(5), 383–387 (2005). [CrossRef] [PubMed]
13. M. Krause, J. Muller, and E. Brinkmeyer, “Measurement of nonreciprocal stimulated Raman scattering in silicon photonic wires,” in Proceedings of 2012 IEEE 9th International Conference on.Group IV Photonics (GFP) (San Diego, 2012), pp. 6–8. [CrossRef]
14. L. Fan, J. Wang, L. T. Varghese, H. Shen, B. Niu, Y. Xuan, A. M. Weiner, and M. Qi, “An all-silicon passive optical diode,” Science 335(6067), 447–450 (2012). [CrossRef] [PubMed]
15. L. Fan, L. T. Varghese, J. Wang, Y. Xuan, A. M. Weiner, and M. Qi, “Silicon optical diode with 40 dB nonreciprocal transmission,” Opt. Lett. 38(8), 1259–1261 (2013). [CrossRef] [PubMed]
16. S. Bhandare, S. K. Ibrahim, D. Sandel, H. Zhang, F. Wust, and R. Noe, “Novel nonmagnetic 30-dB traveling-wave single-sideband optical isolator integrated in III/V material,” IEEE J. Sel. Top. Quantum Electron. 11(2), 417–421 (2005). [CrossRef]
17. Z. Yu and S. Fan, “Complete optical isolation created by indirect interband photonic transitions,” Nat. Photonics 3(2), 91–94 (2009). [CrossRef]
18. H. Lira, Z. Yu, S. Fan, and M. Lipson, “Electrically driven nonreciprocity induced by interband photonic transition on a silicon chip,” Phys. Rev. Lett. 109(3), 033901 (2012). [CrossRef] [PubMed]
19. C. R. Doerr, N. Dupuis, and L. Zhang, “Optical isolator using two tandem phase modulators,” Opt. Lett. 36(21), 4293–4295 (2011). [CrossRef] [PubMed]
20. C. R. Doerr, L. Chen, and D. Vermeulen, “Silicon photonics broadband modulation-based isolator,” Opt. Express 22(4), 4493–4498 (2014). [CrossRef] [PubMed]
21. Y. Yang, C. Galland, Y. Liu, K. Tan, R. Ding, Q. Li, K. Bergman, T. Baehr-Jones, and M. Hochberg, “Experimental demonstration of broadband Lorentz non-reciprocity in an integrable photonic architecture based on Mach-Zehnder modulators,” Opt. Express 22(14), 17409–17422 (2014). [CrossRef] [PubMed]
22. P. Dong, S. F. Preble, J. T. Robinson, S. Manipatruni, and M. Lipson, “Inducing photonic transitions between discrete modes in a silicon optical microcavity,” Phys. Rev. Lett. 100(3), 033904 (2008). [CrossRef] [PubMed]
23. K. Fang, Z. Yu, and S. Fan, “Realizing effective magnetic field for photons by controlling the phase of dynamic modulation,” Nat. Photonics 6(11), 782–787 (2012). [CrossRef]
24. E. Li, B. J. Eggleton, K. Fang, and S. Fan, “Photonic Aharonov-Bohm effect in photon-phonon interactions,” Nat. Commun. 5, 3225 (2014). [PubMed]
25. R. G. Smith, “Use of the acoustooptic light deflector as an optical isolator,” IEEE J. Quantum Electron. 9(5), 545–546 (1973). [CrossRef]
26. J. Mork, B. Tromborg, and J. Mark, “Chaos in semiconductor lasers with optical feedback: theory and experiment,” IEEE J. Quantum Electron. 28(1), 93–108 (1992). [CrossRef]
27. P. Dong, L. Chen, and Y.-K. Chen, “High-speed low-voltage single-drive push-pull silicon Mach-Zehnder modulators,” Opt. Express 20(6), 6163–6169 (2012). [CrossRef] [PubMed]
