Open Access
Add to BibSonomyAbstract
We have proposed and developed a new type of electroabsorption modulator (EAM) that employs both optical absorption and interferometric extinction. The EAM operates at a record low voltage of 0.2 V at 25.8-Gbit/s modulation, which can reduce optical transmitter power consumption and allows the adoption of cost-effective CMOS drivers.
© 2014 Optical Society of America
1. Introduction
Electroabsorption modulators (EAMs) are being used and developed for middle-distance optical communication for both line-side and client-side systems such as local area network-wavelength division multiplexing (LAN-WDM) systems [1–5]. And thanks to its compact size, EAMs are also suitable for monolithic integration with laser diodes (LDs). Furthermore there are some reports of EAM based monolithically integrated light source arrays, which make optical transmitters very compact [6, 7].
An issue with EAMs, namely the reduction of driving-voltage swing, Vpp, has been attracting interest with a view to reducing the power consumption of optical transmitters. Moreover, if we can reduce the Vpp sufficiently, we can adopt cost-effective CMOS drivers with a low power consumption. However, it has been difficult to further reduce the driving voltage other than by making the electrode longer, which degrades the modulation speed. Thus, the only approach to enhancing EAM performance has been improving their multi-quantum well (MQW) characteristics.
To overcome this problem, we focused on the interferometric extinction of the Mach-Zehnder (MZ) structure using the refractive index change of an EAM. By combining MZ interference and conventional optical absorption, we can obtain a large extinction ratio (ER) without modifying the MQW structure. There have already been some reports concerning the integration of EAMs with interferometric waveguides to realize on-off keying (OOK) modulators with a high ER or compact phase-shift keying (PSK) modulators [8–10]. However, we newly designed an MZ-type EAM (EA-MZ) that consists of two asymmetric couplers whose coupling ratios are designed taking account of the refractive index change of an EAM.
In this paper, we propose an EA-MZ and describe its experimental characteristics. The fabricated EA-MZ exhibited a steeper extinction curve than that of a conventional EAM. This means that for a given ER we can reduce the driving voltage or shorten the electrode length of the EAM. In addition, we found that the EA-MZ was tolerant to changes of input light wavelength if we adjusted the bias phase shift of the phase-shifter arm of the MZ configuration. Furthermore, we realized operation at a very-low Vpp of 0.2 V at 25.8-Gbit/s modulation with a dynamic ER (DER) of greater than 4 dB thanks to the sharp extinction characteristics due to combination of MZ interference and conventional electroabsorption. To the best of our knowledge 0.2 V is the lowest voltage at this bit rate and DER and less than half that of our conventional EAM [7], despite the MQW structures of the EA-MZ and the conventional EAM being the same. The very-low-voltage characteristics even allow the adoption of cost-effective and low-power consumption CMOS drivers [11, 12]. We believe that the EA-MZ can help to reduce the power consumption and/or enhance the modulation speed of EAM based optical transmitters.
2. Design and fabrication of EA-MZ
The waveguide structure of an EA-MZ (Fig. 1) consists of a splitting and combining coupler and arm waveguides on which an EAM and a phase shifter are placed. In the design, we assume that these 1st and 2nd couplers are directional couplers (DCs) and that they are asymmetric, namely, their coupling ratios are not 1:1. In this section, we describe the way we determine suitable coupling ratios for obtaining an EA-MZ with a high ER.
Fig. 1 Schematic waveguide structure of an EA-MZ. “K” represents a power-coupling coefficient of the couplers.
The ratios of the two DCs are chosen taking two output states into account, namely the on and off states (“1” and “0” level optical output, respectively) of the modulator. If we define TA1B2 as the transmittance from port A1 to port B2, TA1B2 is expressed as follows:
In Eq. (1), cosθi and sinθi (i = 1, 2) are electrical field transmittances to the bar and cross port of the 1st and 2nd DCs. And, α(V) and ϕEAM(V) represent the power attenuation factor and the amount of light-phase shift in the EAM arm when a bias, V, is applied to the EAM. And ϕph expresses the amount of phase shift in the phase-shifter arm.First, we focus on maximizing the output power in the on state. When no electrical signal is input into either the EAM or the phase shifter arm (α = 1 and ϕEAM = ϕph = 0), TA1B2 is expressed as follows:
Equation (2) means that the EA-MZ corresponds to a simple DC in this situation, and this is a natural result because we can ignore the influence of the two arm waveguides. To obtain the maximum output power in the on state, we should design the 1st and 2nd DCs so that θ1 and θ2 satisfy θ1 + θ2 = π/2. So, as shown in Fig. 1, if we define the power-coupling coefficient from port A1 to the EAM arm as K (0≤K≤1), we can rewrite Eq. (1) with the K as follows:
Next, we determine the optimal K so that the output power in the off state is minimized, that is, TA1B2 = 0. So, K and the other parameters should satisfy
The two equations comprising Eq. (4) mean that the absorbed light intensity from the EAM arm should equal that from the phase-shifter arm and the phase difference of the two beams should be π. In this situation, optical absorption and interferometric extinction occur simultaneously in the EA-MZ. As a result a larger ER is obtained than with a conventional EAM.
Figure 2 shows a vector diagram of the beams from the EAM and the phase-shifter arm in the EA-MZ. Since ϕEAM is generally less than π in actual EAMs, we need a finite ϕph to satisfy the condition of Eq. (4) as shown in Fig. 2. However, such a finite ϕph causes fundamental loss in the on state (α = 1 and ϕEAM = 0) of the EA-MZ due to destructive interference in the 2nd coupler. Therefore we need to determine a suitable V = Voff that maximizes ϕEAM. Once we obtain the Voff value, K is determined from the first equation of Eq. (4).
To obtain Voff, we investigated the relationship between α(V) and ϕEAM(V) of our conventional EAM as shown in Fig. 3. Here α(V) is normalized so that α(0) = 1. First, we experimentally measured the wavelength dependence of α(V) when we applied a voltage. Next, using the α(V) spectrum, we calculated ϕEAM(V) by employing the Kramers-Kronig relation. We found that we should adopt a Voff value of around 2.5 V to obtain a sharp transmission change with an EA-MZ, taking account of the steepness of both the α and ϕEAM transitions. Figure 3 also shows the calculated K corresponding to each α(V) and ϕEAM(V). When Voff = ~2.5 V, K and ϕph should be ~0.8 and ~-0.8π, respectively.
Fig. 3 Attenuation factor (α), phase-change (ϕEAM), and required power-coupling coefficient (K) for design of an EA-MZ.
We fabricated the EA-MZ with an InAlGaAs MQW wafer grown on an InP substrate by metalorganic vapor-phase epitaxy. The waveguides were formed by dry etching and embedded in benzocyclobutene to reduce the capacitance of the electrode pads.
The fabricated EA-MZ (Fig. 4) is ~700 μm long and smaller than conventional MZ modulators whose lengths are typically a few millimeters. The entire device structure is a ridge mesa and is made with the same MQW for ease of fabrication. Therefore, we can further reduce the device length by fabricating a more compact bent waveguide with a deeply etched mesa, and improve the insertion loss by replacing the lossy MQW semiconductor of the device waveguide except for the active arms with a semiconductor with a wide bandgap.
3. Experimental characteristics of EA-MZ
First, we measured and compared the static extinction characteristics (Fig. 5) of two sorts of EA-MZs and conventional EAMs as follows: (i) EA-MZ with a 150-μm-electrode length (LE = 150 μm), (ii) EA-MZ with LE = 50 μm, (iii) EAM with LE = 150 μm, and (iv) EAM with LE = 50 μm. They were fabricated on the same wafer, and so have the same MQW structure. And we formed them into chips of the same length.
Fig. 5 Static extinction characteristics of fabricated EA-MZs and conventional EAMs with electrode lengths of (a) 150 μm and (b) 50 μm.
In the experiment, since we designed the EA-MZ for a LAN-WDM modulator, we input transverse-electric polarized light at a wavelength of 1300 nm. We found that the ERs of the EA-MZs were much larger than that of the conventional EAMs. As a result, we could reduce the driving voltage for a certain ER. For example, 20-dB ERs were obtained when 3.0 V and 4.5 V are applied to conventional EAMs with LE = 150 μm and LE = 50 μm, respectively. On the other hand, we obtained ERs with 2.0 V and 2.5 V for EA-MZs with LE = 150 μm and LE = 50 μm, respectively. Furthermore, the result means that we can improve the trade-off between the driving voltage and the modulation speed, which is limited by the parasitic capacitance of the electrode. Since the required driving voltage for the ER of an EA-MZ is smaller even if we shorten the electrode length by one-third compared with a conventional EAM, we can design an EA-MZ operating at higher speed with a lower driving voltage than a conventional EAM.
The large insertion losses of about 20 dB of the four devices, including the coupling losses between a lensed fiber and a chip facet (~3 dB/facet), were caused by the absorption in their lossy waveguides which consist of MQWs for electroabsorption. We estimated propagation loss of 1300-nm-wavelength light in the MQW waveguide at ~8 dB/mm. So, as described in section 2, if we adopt a semiconductor with a wide bandgap for waveguides other than active ones with electrodes, we can improve the loss caused by the absorption of the MQW semiconductor, as reported in [6, 7]. Therefore, the intrinsic excess losses of the EA-MZs compared with that of the conventional EAM equal the transmittance difference when no driving voltage is applied, and the difference is ~3 dB, which is caused by the phase shift in the phase shifter. This can be understood from Fig. 2, where a certain amount of input-light power is extinguished at the cross port of the EA-MZ even if no driving voltage is applied. Although we injected 0.3 mA and 0.6 mA into the phase-shift arms of the two EA-MZs as phase adjust current, we estimated that there was little absorption loss caused by the carrier effect with such very low currents.
While the extinction characteristics of a conventional EAM are unambiguously determined for an input light wavelength, the EA-MZ is also tolerant to changes of the wavelength since we can adjust the phase shifter current, Iph, so that Eq. (4) is satisfied. We measured the static extinction characteristics of the EA-MZ and the conventional EAM for input light wavelengths of 1295, 1305 and 1310 nm (Fig. 6), which we chose in accordance with the LAN-WDM wavelength grid. The measured EA-MZ and EAM are the same as those shown in Fig. 5(a) (LE = 150 μm). By adjusting Iph, EA-MZ showed a steep extinction curve compared with the conventional EAM. The tolerance is beneficial, for example, for the EAM of monolithically integrated light source array. Although the modulation performance of an EAM-based monolithically integrated WDM device tends to degrade due to the detuning dependence of the EAMs, we can compensate for this degradation by adjusting Iph for each EAM.
Fig. 6 Static extinction characteristics of fabricated EA-MZs and conventional EAMs for input lightwavelengths of (a) 1295 nm, (b) 1305 nm, and (c) 1310 nm.
Next, we measured the dynamic characteristics of the same EA-MZ as shown Fig. 5(a) using 25.8 Gbit/s, non-return to zero, 231-1 pseudo-random bit stream signals. Figure 7(a) and 7(b) show eye diagrams and DERs for back-to-back transmission with various Vpp values. We adjusted the bias point for each measurement so that we could obtain the largest DERs. The relatively large noises in the eye diagrams at low Vpp are caused by the amplification used to compensate for the large insertion loss due to the deep bias applied to the device. We found that the EA-MZ requires only half the Vpp of a conventional EAM for an ER. Even if Vpp = 0.2 V, we obtained eye openings and a DER exceeding 4 dB, which is an important criterion for a LAN-WDM application. These results mean that we can obtain a sufficiently large practical ER even if we drive the EA-MZ with recent low-power-consumption CMOS drivers operating at 25 Gbit/s [11, 12].
Fig. 7 Dynamic extinction characteristics at 25.8-Gbit/s modulation. (a) Eye diagrams (b) DERs as a function of driving voltage.
4. Conclusion
To reduce the driving voltages of conventional EAMs without changing the MQW structure, we proposed an EA-MZ that consists of two asymmetric couplers. These couplers are designed so that the device has steep extinction characteristics. For a given ER, the fabricated EA-MZ required a lower driving voltage or a shorter electrode length than those of a conventional EAM. And, the EA-MZ showed higher tolerance to changes of input light wavelength due to the adjustability of the current to the phase-shifter arm. Although it has a large insertion loss of ~20 dB, this can be reduced by fabricating passive waveguides with a wide-bandgap semiconductor and/or integration with LDs. Thanks to its very low voltage characteristics, we achieved 0.2-V operation with a DER exceeding 4 dB at 25.8-Gbit/s modulation, which to the best of our knowledge is the lowest voltage operation at this bit rate and DER. The result indicates that we can reduce power consumption and transmitter cost by adopting cost-effective CMOS drivers. We achieved these EA-MZ performance improvements despite employing the same MQW as that of conventional EAMs, which means that we have realized a new approach for improving EAM performance without modifying the MQW structure. We believe that the EA-MZ is a feasible candidate for a future modulator with low power consumption for use in middle-distance optical communication systems.
References and links
1. S. Makino, K. Shinoda, T. Kitatani, H. Hayashi, T. Shiota, S. Tanaka, M. Aoki, N. Sasada, and K. Naoe, “High-speed EA-DFB laser for 40-G and 100-Gbps,” IEICE Trans. Electron. E92-C(7), 937–941 (2009). [CrossRef]
2. H. Oomori, H. Ooe, M. Seki, Y. Fujimura, K. Matsumoto, and Y. Murakami, “An extremely compact electro-absorption modulator integrated DFB laser module for 100gbps ethernet over 75km SMF reach,” in Proc. of ECOC 2008, P.2.0.7 (2008). [CrossRef]
3. T. Saito, T. Yamatoya, Y. Morita, E. Ishimura, C. Watatani, T. Aoyagi, and T. Ishikawa, “Clear eye opening 1.3µm-25/43 Gbps EML with novel tensile-strained asymmetric QW absorption layer,” in Proc. of ECOC 2009, P.8.1.3 (2009).
4. H. Takahashi, T. Shimamura, T. Sugiyama, M. Kubota, and K. Nakamura, “High-power 25-Gb/s electroabsorption modulator integrated with a laser diode,” IEEE Photon. Technol. Lett. 21(10), 633–635 (2009). [CrossRef]
5. T. Fujisawa, M. Arai, N. Fujiwara, W. Kobayashi, T. Tadokoro, K. Tsuzuki, Y. Akage, R. Iga, T. Yamanaka, and F. Kano, “25 Gbit/s 1.3 μm InGaAlAs-based electroabsorption modulator integrated with DFB laser for metro-area (40 km) 100 Gbit/s ethernet system,” IET Electron. Lett. 45(17), 900–902 (2009). [CrossRef]
6. S. Kanazawa, T. Fujisawa, A. Ohki, H. Ishii, N. Nunoya, Y. Kawaguchi, N. Fujiwara, K. Takahata, R. Iga, F. Kano, and H. Oohashi, “A compact EADFB laser array module for a future 100-Gb/s ethernet transceiver,” IEEE J. Sel. Top. Quantum Electron. 17(5), 1191–1197 (2011). [CrossRef]
7. T. Fujisawa, S. Kanazawa, K. Takahata, W. Kobayashi, T. Tadokoro, H. Ishii, and F. Kano, “1.3-μm, 4 × 25-Gbit/s, EADFB laser array module with large-output-power and low-driving-voltage for energy-efficient 100GbE transmitter,” Opt. Express 20(1), 614–620 (2012). [CrossRef] [PubMed]
8. I. Kang, “Phase-shift-keying and on-off-keying with improved performances using electroabsorption modulators with interferometric effects,” Opt. Express 15(4), 1467–1473 (2007). [CrossRef] [PubMed]
9. C. R. Doerr, L. Zhang, P. J. Winzer, and A. H. Gnauck, “28-Gbaud InP square or hexagonal 16-QAM modulator,” in Proc. of OFC/NFOEC 2011, OMU2 (2011). [CrossRef]
10. C. Kazmierski, D. Carrara, K. Ławniczuk, G. Aubin, J.-G. Provost, and R. Guillamet, “12.5GB operation of a novel monolithic 1.55μm BPSK source based on prefixed optical phase switching,” in Proc. of OFC/NFOEC 2013, OW4J.8 (2013).
11. Y. Lee, D. Kawamura, T. Takai, K. Kogo, K. Adachi, T. Sugawara, N. Chujo, Y. Matsuoka, S. Hamamura, K. Yamazaki, Y. Ishigami, T. Takemoto, F. Yuki, H. Yamashita, and S. Tsuji, “25-Gb/s 100-m MMF transmission using a prototype 1.3-μm-range CMOS-based transceiver for optical interconnections,” IEEE Photon. Technol. Lett. 24(6), 467–469 (2012). [CrossRef]
12. N. Chujo, T. Kawamata, K. Ohhata, and T. Ohno, “A 25Gb/s laser diode driver with mutually coupled peaking inductors for optical interconnects,” in Proc. of CICC 2010, 1–4 (2010). [CrossRef]
