We present methods of optical modulation employing electroabsorption in an interferometric structure. The interferometric operation enables electroabsorption modulators for phase-shift keying and allows generation of on-off-keying signals with much improved extinction ratios. We demonstrate 40-Gb/s phase-shift-keying modulation, and up to 8.7 dB improvement in extinction ratio in 40-Gb/s on-off-keying modulation with 0.8V driving voltage using a commercial electroabsorption modulator. Methods of applying electroabsorption modulators for high spectral-efficiency formats generation are also proposed.
© 2007 Optical Society of America
Semiconductor electroabsorption modulators (EAMs) offer the advantages of low drive voltages, small form factors, and the integrability with semiconductor lasers and amplifiers. EAM-integrated lasers are widely used in metro-distance communication networks and EAMs are key building blocks for highly integrated photonic circuits[1–3]. The low drive-voltage requirement is especially promising for emerging high bit-rate (>80 Gb/s) communication, where the wide-band electrical amplification required for driving lithium niobate modulators (LNMs) gets more challenging with increasing bit rates.
In view of the increasing importance of various phase-shift keying (PSK), including differential phase-shift keying (DPSK) and quadrature phase-shift keying (QPSK), in emerging high spectral-efficiency, high data-rate transmission, it is desirable to extend the advantages of EAMs for use with these modulation formats. In contrast to the LNMs implemented using a Mach-Zehnder interferometer (MZI) that are capable of both on-off keying (OOK) and PSK, EAMs traditionally have been used as a single OOK modulation element.
We demonstrate PSK modulation using an EAM. The PSK modulation is achieved at 40 Gb/s by combining electroabsorption modulation with an optical interferometer. More specifically, we use the interference between the TE and TM modes of a polarization-sensitive EAM.
We also demonstrate that the interferometric operation of EAMs can benefit OOK modulation as well by significantly improving the extinction ratio for the same drive voltage, or alternatively it allows EAMs to be operated at much lower drive voltages for a desired extinction ratio. One way to increase the electrooptic bandwidth of a lumped-element EAM is reducing the length of the EAM region and hence its capacitance, but this approach may compromise the extinction ratio of the device[3–5]. The interferometric operation could be used to enhance the electrooptic bandwidth of EAMs, providing an alternative method of overcoming the trade-off relationship between the extinction ratio and bandwidth.
2. Principle of operation
The principle of the interferometric operation of an EAM is schematically illustrated in the case of a polarization-sensitive interferometer in Fig. 1. EAMs based on multiple quantum wells show preferential TE-polarization absorption unless the anisotropy is relieved by strains. In Fig. 1(a), the electric field ECW of linearly polarized CW light launched into the EAM at an angle θ with respect to the TE axis is decomposed along the TE and TM axes. The projections of the non-return-to-zero (NRZ)-modulated fields ETE(t) and ETE(t) output from the EAM, onto an analyzing polarizer oriented at an angle ϕ, are shown using dotted and dashed lines, respectively. As implied in Fig. 1(a), the modulation depth of TE polarization is substantially larger than that of TM polarization for a given electric drive signal. For example, the TE modulation depth is ∼ 9 dB larger than the TM modulation depth for 2V drive signal amplitude and −1V bias voltage for the commercial device we used (OKIOM5753C-30B). The TE and TM mode transmissions plotted in Fig. 1(c) as a function of the bias voltage cause the different modulation depths of the two modes.
PSK generation or improvements of the extinction ratio (ER) can be achieved through adjustments of the relative magnitudes and modulation depths of the two projected fields at the output of the analyzer by controlling the two angles θ and ϕ. As shown in Fig. 1(b), PSK modulation (top) or OOK modulation with an extinction improvement (bottom) can be achieved by properly adjusting the TE and TM projections and interferometrically subtracting the latter from the former. PSK modulation is achieved when the mean values of the TE and TM projection fields are equalized. The extinction ratio of OOK can be improved by equalizing the levels of the 0’s of the TE and TM modes. The angle parameters can be determined for each modulation under the constraint that the excess optical loss due to the destructive interference be minimized.
3. PSK generation
The above-mentioned conditions for the interferometric operation can be calculated using the extinction ratios and linear transmittance of the EAM for a given driving condition. PSK generation is achieved when the angle variables satisfy
where ε TE (ε TM) is the extinction ratio, as defined by the power ratio between the ON and OFF states of the TE(TM) mode, respectively, and r is the ratio of the transmittance of TM mode to that of the TE mode for the ON state. A schematic of the experimental setup is shown in Fig. 2. In the experiment, the launch angle of the CW light (1553 nm) polarization is controlled using a polarization controller (HP8169A) and the polarization of the output of the EAM is controlled using a fiber-optic paddle polarization controller before the two polarization modes interfere at a fiber-optic polarizer. The polarization controllers are also used to compensate for the birefringence of the EAM and the fibers attached to it.
The EAM has a 3-dB electro-optic bandwidth of 30 GHz and we use 2.6-V peak-to-peak amplitude drive signal for PSK modulation. The magnitude of the drive signal is about half as that required to drive a commercially available lithium niobate Mach-Zehnder modulator to generate PSK signals, where the required signal amplitude is equal to Vπ for each arm of the modulator in the push-pull drive configuration. The PSK signal is demodulated by a 25-ps delay-line interferometer and the output from the destructive port is received and electronically demultiplexed to 10 Gb/s for measuring the bit error rates (BER). The photodetector is preceded by an EDFA with a noise figure of ∼5.6 dB and an optical filter with 1-nm bandwidth. We show in Fig. 3 the receiver sensitivities of the EAM-generated PSK signals and the PSK signals modulated using a lithium niobate modulator biased at a null point. The received powers are measured in front of the EDFA with a fixed gain. The optical powers received by the photodetector are maintained at a constant level (∼ 1 dBm) using a variable optical attenuator just in front of the detector. The receiver sensitivity of EAM-generated PSK at BER of 10−9 is −24.5 dBm. The observed non-smoothness of the BER curve for the EAM-generated PSK is attributed to the fluctuations of the polarization caused by the temperature change and vibration occurring during the BER measurements. These fluctuations degrade the PSK generation by making the interference between the TE and TM modes unstable. The implementations shown in section 5 using integrated optics should be immune to this type of fluctuations.
We show in Fig. 4 the eye diagrams of the 40-Gb/s PSK signal (a) and the eye diagrams of the signal after demodulation (b). The distortion of the temporal shape of the demodulated signal stems from the incomplete interference between the TE and TM electric fields, which have non-identical temporal shapes owing to the different responses of the two modes to the drive signal, as can be inferred from Fig. 1(c). For comparison, the eye diagrams of lithium-niobate modulator generated PSK after demodulation is shown in (c). The ∼3-dB relative power penalties of the EAM-generated PSK signals observed in Fig. 3 are likely to be caused by the distortion.
4. OOK data modulation
The extinction ratio of an EAM-modulated OOK signal in the conventional operation is determined by the drive-signal amplitude and lower-voltage drives are preferable for reducing the power consumption and complexity of the driving electronics . In Fig. 5(a), we show the electrical eye diagrams of OOK signal obtained directly from the EAM without the interference. In this case, the maximum extinction ratio (ER) is achieved when the polarization of the CW laser is aligned with the TE axis of the EAM, which is 5.7 dB for 1.1V drive voltage. In Fig. 5(b), we show a clear improvement of the ER using the interferometric operation with the optimization of the angles θ and ϕ. The angles can be calculated as similarly for the case of PSK and they need to satisfy θ =ϕ = Arctan [(ε TM/rε TE)1/4]. The improved ER is measured to be 12.5 dB for the same drive voltage. It is noted that the accuracy of the ERs as determined from the eye diagrams is limited for values higher than ∼14 dB due to the detector noise.
In Fig. 6(a), we plot the extinction ratio (ER) and receiver sensitivity (RS) (for BER=10−9) of the OOK signals obtained with and without the interferometric operation. For operations without the interference, ER and RS are optimized with the TE mode. With the interferometric operation, we observed the maximum ER improvement of 8.7 dB and RS improvement of 7 dB for the drive voltage of 0.8V. Note that this signal amplitude is among the lowest reported to date used for successfully generating 40-Gb/s signals.
We point out the fact that short optical pulse generation or optical demultiplexing using EAMs can benefit from the interferometric operation as well. It is well known that a short temporal optical gate can be generated using a deeply negatively biased EAM driven with a large amplitude sinusoidal drive [6, 7]. However, the extinction ratio of the optical gates generated in such a manner is inversely related to the pulse duration. On the contrary, short optical pulse or gate generation using interferometric operation of EAMs would not suffer from this type of extinction ratio degradation.
The destructive interference between the TE and TM modes causes reduction of the amplitude of the interferometrically generated signals. In Fig. 6(b), we show the excess loss of the peak signal powers of the interferometrically generated OOK signals for different drive voltages. The excess losses are normalized with respect to the peak signal power that would be obtained without interference by launching the CW polarization aligned with respect to the TE axis. Also shown in Fig. 6(b) are the calculated values using.
The measurements are in qualitative agreement with the calculation. The discrepancies observed for the large drive signals are due to the uncertainties in estimating the large extinction ratios from the eye diagrams: 1–2dB uncertainties in the measurements of ε TE could account for the discrepancies.
5. EAMs-integrated interferometric planar lightwave circuits
The interferometric operation of EAMs can be more efficiently realized in planar lightwave circuits (PLC) incorporating EAMs and interferometric structures. A simplest PLC structure is a Mach-Zehnder interferometer (MZI) having an EAM in at least one of the two arms (Fig. 7(a)). The interference between the optical modes of the two arms play the same role as the interference between the TE and TM modes described earlier. Although only one EAM in placed in one arm is sufficient to achieve PSK modulation or OOK modulation with enhanced extinction, where the optical mode propagating in the arm with the EAM plays the similar role as the TE mode, the push-pull operation as shown in Fig. 7(b) is preferable as it minimizes the excess optical loss due to the interference. In Fig. 7(b), the EAM in the upper arm is modulated by data signal D and the EAM in the lower arm is driven by the complementary data signal D̄. It can be seen by comparing Fig. 1 and Fig. 7 that the dual-drive arrangement is better at reducing the loss by minimizing the amount of light subtracted by the destructive interference. Furthermore, the EAM-MZI has an added benefit of the potential polarization insensitivity.
In optimizing the signal modulation using the EAM-MZI, the coupling ratio of the input (output) coupler of the MZI plays the same role as cos2(θ) (cos2(ϕ )) of the polarization-sensitive interferometer described earlier, respectively. The input and output coupling ratios should be the same to minimize the excess loss. For PSK generation, the coupling ratio is 50:50 regardless of the raw extinction ratio ε of the EAMs, and the excess loss due to the interference, normalized with respect to the input CW power, is (1−√1/ε)2/4. Note that we are not accounting for other sources of losses arising from the potential imperfections of the device including the coupling loss to the input and output optical fibers. For OOK generation, the coupling ratio between the upper and lower arms is 1+√1/ε:√1/ε and the excess loss in this case is (1−1/ε)2. The calculated excess losses are plotted in Fig 7(c). These relations along with the EAM transfer functions, such as shown in Fig. 1(c), can be used as a guide in designing EAM-MZIs for low-drive-voltage data modulation. For example, OOK signal generation with a nearly perfect extinction should be feasible with only ∼ 1V drive at 40 Gb/s with ∼ 7 dB extra loss due to the interference. Alternatively, the relations can be used to develop short active-length lumped-EAMs for high-speed operation as typically limited extinction ratio of the short EAMs can be overcome with the interference. The excess loss may be tolerated if very low drive voltages, especially at high bit rates, are demanded. The impact of the excess loss may be mitigated by integration of a source laser on chip or it can be compensated for by providing the needed optical gain using semiconductor optical amplifiers integrated on chip .
It is straightforward to apply the concept of interferometric operation of EAMs for generating Quadrature-PSK (QPSK) or Quadrature-Amplitude Modulation (QAM). For example, a nested MZI shown in Fig. 8(a) can be used to generate N-QAM signal, including QPSK. The constellation of 8-QAM signals corresponding to the digital binary drive signals to each of the EAMs is shown in Fig. 8(b). We expect the 8-QAM signals generated using the EAM-based device could have better noise properties than those generated employing a conventional dual Mach-Zehnder lithium-niobate modulator. Firstly, the EAM-based device should be less sensitive to the phase fluctuations that may be caused by the imbalances in the frequency responses of the two arms of the MZI or by the imperfection of the temporal shapes of the two drive signals not being exactly complementary to each other. Secondly, the phase shifters setting the phase rotation of the constellation, whether they are thermo-optic or electrooptic, are more stable than the phase shifters implemented in z-cut lithium niobate modulators, which tend to suffer from pyroelectric phase drift. Finally, highly nonlinear amplitude modulation characteristic of the EAMs with respect to the drive voltages can be utilized to reduce the amplitude fluctuation that may still arise owing to the aforementioned asymmetries of the device or drive signals.
We demonstrate the potentials of the interferometric operation of EAMs in generating PSK and OOK with enhanced extinction using much-reduced drive signals. The capabilities of a commercial polarization-sensitive EAM are greatly improved via the interferometric operation and we use the device to accomplish 40-Gb/s PSK modulation with a 2.6V drive voltage, and 40-Gb/s OOK modulation with 12.3 dB extinction ratio at 0.8 V drive voltage. We also propose planar lightwave circuits consisting of EAMs and interferometric structures as a promising device for enabling future high-speed, high spectral-efficiency transmission. Monolithic integration with other active devices such as lasers or semiconductor optical amplifiers are expected to further widen the applications of EAM-based devices.
References and links
1. R. Nagarajan, et al., “Large-scale photonic integrated circuits,” IEEE J. Sel. Top. Quantum Electron. 11,50–65 (2005). [CrossRef]
2. B. R. Koch, Z. Y. Hu, J. E. Bowers, and D. J. Blumenthal, “Integrated optical payload envelope detection and label recovery device for optical packet switching networks,” Opt. Express 14,5073–5078 (2006). [CrossRef] [PubMed]
3. H. Kawanishi, Y. Yamauchi, N. Mineo, Y. Shibuya, H. Murai, K. Yamada, and H. Wada, “EAM-integrated DFB laser modules with more than 40-GHz bandwidth,” IEEE Photon. Technol. Lett. 13,954–956 (2001). [CrossRef]
4. H. Fukano, T. Yamanaka, M. Tamura, H. Nakajima, Y. Akage, Y. Kondo, and T. Saitoh, “40Gbit/s electroabsorption modulators with 1.1V driving voltage,” Electron. Lett. 40,1144–1146 (2004). [CrossRef]
5. H. Fukano, T. Yamanaka, M. Tamura, and Y. Kondo, “Very-low-driving-voltage electroabsorption modulators operating at 40 Gb/s,” J. Lightwave Technol. 24,2219–2224 (2006). [CrossRef]
6. D. G. Moodie, A. D. Ellis, and C. W. Ford, “Generation of 6.3ps optical pulses at a 10 GHz repetition rate using a packaged electroabsorption modulator and dispersion compensating fiber,” Electron. Lett. 30,1700–1701 (1994). [CrossRef]
7. M. Suzuki, H. Tanaka, N. Edagawa, K. Utaka, and Y. Matsushima, “Transform-limited optical pulse generation up to 20-GHz repetition rate by a sinusoidally driven InGaAsP electroabsorption modulator,” J. Lightwave Technol. 11,468–473 (1993). [CrossRef]
8. B. Mason, et al., “40-Gb/s tandem electroabsorption modulator,” IEEE Photon. Technol. Lett. 14,27–29 (2002). [CrossRef]