Open Access
Add to BibSonomyAbstract
An optical phase modulator based on novel n-i-n InGaAlAs/InAlAs multiple-quantum-barrier (MQBs) is proposed to reduce the waveguide loss and enhance the modulation efficiency. Efficient bias loading with low leakage current is realized by taking advantages of the large conduction band discontinuity in InGaAlAs/InAlAs heterostructure and high electron reflection by MQBs. Low optical transmission loss of 2.3 dB/cm and improved modulation performance with a voltage length product of 1.06 V⋅cm are demonstrated for TE mode at 1550 nm. The linear and quadratic electro-optic coefficients of the fabricated device are estimated to be 1.0 pm/V and 1.4 × 10−19 m2/V2, respectively.
© 2013 Optical Society of America
1. Introduction
Mach-Zehnder modulators (MZMs) are of great importance in high-speed fiber communication systems, especially in high spectral efficiency networks utilizing advanced modulation formats [1]. In such applications, low drive voltage is crucial to improving the system performance and reducing power consumption. Compared with LiNbO3 or silicon based modulators, III-V compound semiconductor based electro-optic modulators are advantageous in their reduced drive voltage, compact size, and the potential for integration with a laser source. Conventional semiconductor electro-optic modulators adopt a reversely biased p-i-n structure [2]. However, p-i-n waveguides exhibit significant waveguide loss due to free carrier absorption in the p-type cladding layer [3, 4]. As the electrical and optical propagation loss of the n-type cladding layer is 20 times smaller, it would be advantageous to replace the p-type cladding layer with n-type one. Both n-SI-i-n and n-p-i-n MZMs were proposed to eliminate the waveguide loss caused by the p-cladding layer, where a semi-insulating (SI) layer or a thin p-doped layer was adopted to block the electron current [5–7]. However, the SI layer in the n-SI-i-n structure consumes a considerable portion of the modulation voltage, thus hampering further reduction of the modulator drive voltage. On the other hand, a thick undoped cladding layer is often inserted between the p-doped layer and the core layer in the n-p-i-n structure to reduce the optical loss, which also impairs the modulation efficiency. Moreover, such a structure requires strict doping control to ensure efficient bias loading.
In this work, we propose a simple InGaAlAs/InAlAs multiple-quantum-barrier (MQB) based n-i-n heterostructure for the fabrication of low drive voltage InP MZMs. The novel n-i-n structure does not include any SI or p-doped layer, and can efficiently block the electron current by taking advantage of the large conduction band discontinuity of InGaAlAs/InAlAs heterostructure and electron reflection by MQBs. At the same time, the modulation voltage is directly applied to the intrinsic waveguide core, thereby minimizing the half-wave voltage of the electro-optic modulator. An n-i-n InGaAlAs/InAlAs MQB electro-optic phase modulator is fabricated, and its electro-optic phase shift and propagation loss is characterized by an improved Fabry-Perot (FP) contrast method.
2. Theoretical analysis and simulation
Unlike a p-i-n structure, there is no reversely biased p-n junction in a normal n-i-n structure to block the flow of leakage current, making it impossible for the modulation signal to be effectively applied to the electro-optically active intrinsic region. Furthermore, leakage current also gives rise to significant electrical loss at high frequencies, thus limiting the bandwidth of the device. Three different InP-based n-i-n heterostructures are analyzed in the following to verify the effectiveness of MQB structure in blocking the leakage current.
Figure 1(a) depicts the band diagram of an InGaAsP/InP n-i-n heterostructure, in which a 390-nm-thick intrinsic InGaAsP core layer (lattice-matched to InP, Eg = 0.95 eV) is sandwiched between two n-type InP cladding layers. The doping level of the n-InP cladding layers is taken to be 5 × 1017 cm−3. It is clear that there is a barely 0.09 eV energy barrier for electrons formed by the conduction band discontinuity between InGaAsP and InP. Such a small barrier is insufficient to block the electron current when bias voltage is applied to the n-i-n structure.
Fig. 1 Band diagrams of (a) InGaAsP/InP (b) InAlAs/InP and (c) InGaAlAs/InAlAs MQB based n-i-n heterostructures.
To suppress the leakage current in the n-i-n structure, we take advantages of the band discontinuity to form heterojunction barrier to minimize the influences of leakage current on the microwave signal. In Fig. 1(b), the InGaAsP core layer is replaced by a 390-nm-thick intrinsic InAlAs layer (lattice matched to InP, Eg = 1.45 eV). The type II hetero-junction formed between InAlAs and InP provides a much stronger current blocking capability than the InGaAsP/InP based n-i-n hetero-structure. Unfortunately, the low refractive index of InAlAs layer does not provide any optical field confinement. To circumvent the problem, two InGaAlAs confinement layers (lattice matched to InP, Eg = 0.89 eV) are inserted between the InAlAs/InP interface. As shown in Fig. 1(b), an electron barrier of 0.43 eV is formed in the conduction band which would lead to a better current blocking behavior.
The superiority in the carrier-confinement effect of the MQB over the bulk barrier was reported in GaAs-based n-i-n structures [8, 9]. Moreover, multiple-quantum-wells (MQWs) are often adopted in semiconductor electro-optical modulators to enhance the modulation efficiency by the high quadratic electro-optic coefficients related to quantum confined Stark effect (QCSE) [10]. In Fig. 1(c), the bulk InAlAs layer is replaced with 20-periods of InGaAlAs/InAlAs MQBs with λPL = 1370 nm. Both the 10-nm-thick InGaAlAs well and the 10-nm-thick InAlAs barrier are lattice matched to InP, and the total thickness of the MQB layer is 390 nm. An electron barrier of 0.51 eV is formed in the conduction band of the MQB/InP n-i-n structure.
To investigate the current blocking capability of the heterostructures shown in Fig. 1, the leakage current densities of the three n-i-n structures are calculated using Synopsys TCAD Sentaurus. As clearly shown in Fig. 2, significant leakage current is generated in the InGaAsP/InP n-i-n heterostructure. On the other hand, the leakage current is more than four orders of magnitude reduced in the InAlAs/InGaAlAs n-i-n heterostructure, thanks to the much larger electron barrier. The n-i-n structure with the MQB core shows much lower leakage current density, due to the high electron barrier and the electron reflection by the MQB [11].
Next we investigate the microwave performance of the proposed low leakage n-i-n heterostructure waveguides. In our simulations, a coplanar-waveguide-type traveling-wave electrode structure is assumed for the modulator [5]. In the presence of leakage current, the traveling-wave electrodes behave like a lossy transmission line, and the microwave signal loss due to leakage current must be taken into account. The frequency response of the coplanar electrodes is calculated using finite element method (FEM), with the conductance of the waveguide core at a given bias voltage determined from Fig. 2. In this way, the influence of microwave loss due to the leakage current as well as the electrode metals and epitaxial layers can be investigated.
Figure 3 plots the calculated microwave response of the traveling-wave electrode structure on a 1-mm-long and 2-μm-wide high-mesa MQB/InP waveguide. Both the signal and the ground electrodes are assumed to be 2 μm thick gold. The signal electrode on top of the waveguide is 2 μm wide, while the lateral signal-to-ground spacing is 10 μm. According to the calculated transmission coefficient S21, the structure exhibits a 6-dB electrical bandwidth of 80 GHz. This is basically the same as the case when leakage current is fully excluded from the simulation model, indicating that the influence of the leakage current is negligible. The relatively high reflection coefficient S11 shown in Fig. 3 is mainly due to impedance mismatch, which can be improved by using the series push-pull drive configuration [10].
3. Device fabrication
Figure 4 shows the epitaxial structure of the novel n-i-n modulator, in which an SI-InP substrate is adopted to minimize the microwave propagation loss. The n+-InGaAs lower ohmic contact layer, n-InP lower cladding layer, undoped MQB core layer, n-InP upper cladding layer and n+-InGaAs upper ohmic contact layer are sequentially grown on the SI-InP substrate by metal organic chemical vapor deposition (MOCVD). The undoped core layer is the same as given in Fig. 1(c).
To process the wafer into electro-optic phase modulators, high-mesa ridge waveguides are formed along the [110] direction by inductively coupled plasma (ICP) dry etching and subsequent selective wet etching with HCl:H3PO4 (3:1). The width and height of the mesa are 1.5 μm and 2.8 μm, respectively, to ensure operation in fundamental mode for both TM and TE polarized light. Ohmic contacts are formed by sputtering Ti(20nm)/Pt(50nm)/Au(400nm) on both the top n+-InGaAs layer and the bottom n+-InGaAs etch stop layer, so that both forward and reverse bias can result in vertical electric fields across the MQB core. The forward bias here means positive bias voltage on the top n+-contact layer, whereas the reverse bias is the other way round. The wafer is cleaved into phase modulator chips with length ranging between 0.6 mm and 1.9 mm, and the facets of the devices are left uncoated.
4. Experiments results
Figure 5 shows the experiment setup for characterizing the n-i-n MQB phase modulators. The output of a tunable laser is coupled into one of the cleaved facets of the waveguide under test through a tapered fiber. A polarization controller is used to ensure TE or TM polarization. The output optical power from the waveguides is coupled into another tapered fiber and measured with a power meter. The current flowing through the device is monitored by a current meter.
Fig. 5 Experimental setup for optical loss and electro-optic phase shift measurements with n-i-n heterostructure phase modulators.
Figure 6 shows the typical I-V curve of the n-i-n modulator without light input. For a 1-mm-long waveguide, the leakage current is less than 15 μA for both forward and reverse bias up to 4V, confirming effective electron blocking by the n-i-n heterostructure.
FP contrast method is adopted to characterize the optical loss and the electro-optic phase shift of the device [12, 13]. FP fringes formed by reflections at the end-facets are analyzed to provide a measurement of the fringe contrast K and the phase shift Δϕ as a function of reverse bias, polarization, and wavelength for a given waveguide. The contrast of the FP fringes is used to estimate the waveguide propagation losses as follows [14]:
where R is the facet reflectivity, α is the waveguide propagation loss, L is the waveguide length, while Tmax and Tmin are the maximum and minimum transmitted optical power through the waveguide.Figure 7 shows a plot of the quantity vs. the waveguide length measured at the wavelength of 1550 nm. Plotted in this way, the waveguide loss α is given by the slope of the curve. A least-squares fit to the data gives a propagation loss of 2.3 and 1.0 dB/cm for the TE and TM light, respectively. These values are one order of magnitude lower than that of conventional p-i-n structures [3, 15], and the 2.3 dB/cm propagation loss for TE mode is also lower than that of the n-p-i-n structure reported in [7].
Fig. 7 FP method for measuring propagation loss in the MQB n-i-n heterostructure waveguides. The lines are linear fits to the data.
The electro-optic effect is measured by the change of the fringe phase as a function of the bias voltage [14]:
where T(V) is the transmitted optical power through the FP waveguide resonator at different biases.The electro-optic effect induces a change in the optical phase in a waveguide as
where Δn is the change in the refractive index and λ is the operation wavelength. For TE mode, Δn comprises contributions from both the linear and the quadratic electro-optic effects, whereas it only depends on the quadratic effect for the TM mode. It should be noted that in semiconductor waveguide modulators, effects other than the linear and quadratic electro-optic effects, such as plasma effect [16] and thermo-optic effect [17], can play an important role. However, in our n-i-n heterostructure phase modulator, the refractive index variation due to plasma effect and thermo-optic effect are found to be one order of magnitude smaller than that of the linear electro-optic effect, as a result of the suppressed carrier leakage induced heat generation at low bias voltage less than 8V. So both the plasma effect and the thermal-optic effect can be neglected for low bias voltage. Therefore, Δn of both TE and TM modes at different biases can be written as [18]: where neff is the effective refractive index, r41 is the linear electro-optic coefficient, R12 and R11 is the quadratic electro-optic coefficient of the TE and TM mode, respectively. E is the electric field within the intrinsic core layer, and E0 is the built-in electric field strength. ψTE(x) and ψTM(x) are the normalized optical modal field of the TE and TM mode, respectively.The phase shift at different voltage in a 1930-μm-long device at 1550 nm is plotted in Fig. 8 for both TE and TM polarizations. It is found that no linear electro-optic effect is associated with the TM polarization, whereas the phase shift is most significant for TE polarization when forward biased, as can be expected from Eqs. (4)-(7). It is interesting to notice that the TM phase shifts for the forward and reverse biases deviate somewhat from each other. This is attributed to the asymmetrical overlap between the electric field and the optical mode due to the non-vertical waveguide sidewall formed by combined dry and wet etching.
Fig. 8 Measured phase shift vs. bias voltage for (a) TE and (b) TM polarized light at 1550nm in a 1930-μm-long phase modulator.
According to Fig. 8, the minimum voltage length product VπL is 1.06 and 1.67 V⋅cm for TE and TM polarization in forward bias, respectively. This value represents an order of magnitude improvement over typical LiNbO3 devices [19]. The novel n-i-n heterostructure phase modulator has the potential to realize low half-wave voltage, as the modulation bias almost completely falls on the electro-optically active core layer. The bias loading efficiency, defined as the portion of bias voltage over the core layer, is close to 99% in our n-i-n phase modulator, whereas the value is estimated to be only 71% and 52% according to the active layer structures in previously reported n-SI-i-n [5] and n-p-i-n [7] semiconductor electro-optic modulators.
Figure 9 depicts the dependence of phase shift on the operation wavelength measured with a 630-μm-long phase modulator. Under a given forward bias, the phase shift becomes more prominent as the wavelength deviation from the band edge of the MQW core decreases. For reverse bias, the voltage to compensate the linear electro-optic effect induced phase shift decreases as the operation wavelength approaches the band edge.
Fig. 9 Measured FP phase shift vs. bias voltage for the TE mode for 1525nm, 1550nm and 1600nm in a 630μm long waveguide.
The values of r41 and R12 are estimated from the measured TE phase shifts using Eqs. (5) and (6), while R11 is estimated from the measured TM data according to Eq. (7). We adopt the abrupt-junction approximation to estimate the electric field strength and the finite element method to calculate the optical electric field distribution. The electro-optic coefficients thus determined are shown in Fig. 10 for different waveguide lengths. The estimated r41 is around 1.0 pm/V, which is similar to the previously reported values [20, 21]. And the values of R11 and R12 are all around 1.4 × 10−19 m2/V2 at 1550nm. R12 and R11 exhibit strong wavelength dispersion, and is estimated to be 1.8 × 10−19 and 4.5 × 10−20 m2/V2 at 1525 and 1600 nm, respectively.
Figure 11 depicts the linear electro-optic (LEO) and the quadratic electro-optic (QEO) coefficients for TE mode at different wavelengths. At low external voltages, the contribution from the linear electro-optic effect, which is largely wavelength-independent, is dominant. The quadratic electro-optic effect becomes significant when we operate close to the MQW bandgap, where the QCSE is stronger. For operation wavelength larger than 1525 nm, the QCSE induced propagation loss is relatively small at a bias voltage below 6V (< 1 dB/mm at 1550 nm). However, the propagation loss becomes significant at higher voltages due to the red-shift of the band edge and the free-carrier absorption.
Fig. 11 Calculated linear and quadratic electro-optic effects at different wavelengths for TE mode from the estimated electro-optic tensor coefficients.
5. Conclusions
A novel n-i-n InGaAlAs/InAlAs MQB phase modulator is proposed to reduce the waveguide loss and enhance the modulation efficiency. By taking advantages of the large conduction band discontinuity of the InGaAlAs/InAlAs heterostructure, the leakage current can be efficiently suppressed and its influence on high-speed performance can be made negligible. The simple n-i-n structure allows high bias loading efficiency, and an improved modulation performance with a voltage length product of 1.06 V⋅cm is demonstrated. The novel n-i-n electro-optic modulator is believed to have promising potential in future high-speed optical communication systems.
Acknowledgments
This work was supported in part by the National Basic Research Program of China (Grant Nos. 2012CB315605, and 2014CB340002), the National Natural Science Foundation of China (Grant Nos. 61176015, 61176059, 61321004, and 51002085).
References and links
1. P. J. Winzer and R.-J. Essiambre, “Advanced modulation formats for high-capacity optical transport networks,” J. Lightwave Technol. 24(12), 4711–4728 (2006). [CrossRef]
2. M. Jarrahi, T. H. Lee, and D. A. B. Miller, “Wideband, low driving voltage traveling-wave Mach–Zehnder modulator for RF photonics,” IEEE Photon. Technol. Lett. 20(7), 517–519 (2008). [CrossRef]
3. L. Zhang, J. Sinsky, D. Van Thourhout, N. Sauer, L. Stulz, A. Adamiecki, and S. Chandrasekhar, “Low voltage high speed travelling wave InGaAsP/InP phase modulator,” IEEE Photon. Technol. Lett. 16(8), 1831–1833 (2004). [CrossRef]
4. P. W. Juodawlkis, F. J. O’Donnell, R. J. Bailey, J. J. Plant, K. G. Ray, D. C. Oakley, A. Napoleone, and G. E. Betts, “Sub-volt-Vπ InGaAsP electrorefractive modulators using symmetric, uncoupled quantum wells,” in Proceedings of 16th Annual Meeting of IEEE (Lasers and Electro-Optics Society, Tucson, 2003), pp. 788–789.
5. K. Tsuzuki, T. Ishibashi, T. Ito, S. Oku, Y. Shibata, R. Iga, Y. Kondo, and Y. Tohmori, “40 Gbit/s n-i-n InP Mach-Zehnder modulator with a π voltage of 2.2 V,” Electron. Lett. 39(20), 1464–1465 (2003). [CrossRef]
6. T. Yasui, Y. Shibata, K. Tsuzuki, N. Kikuchi, M. Ishikawa, Y. Kawaguchi, M. Arai, and H. Yasaka, “10-Gb/s 100-km SMF transmission using InP Mach-Zehnder modulator monolithically integrated with semiconductor optical amplifier,” IEEE Photon. Technol. Lett. 20(13), 1178–1180 (2008). [CrossRef]
7. N. Kikuchi, Y. Shibata, K. Tsuzuki, H. Sanjoh, T. Sato, E. Yamada, T. Ishibashi, and H. Yasaka, “80-Gb/s low-driving-voltage InP DQPSK modulator with an n-p-i-n structure,” IEEE Photon. Technol. Lett. 21(12), 787–789 (2009). [CrossRef]
8. I. Hase, H. Kawai, K. Kaneko, and N. Watanabe, “Current-voltage characteristics through GaAs/AlGaAs/GaAs heterobarriers grown by metalorganic chemical vapor deposition,” J. Appl. Phys. 59(11), 3792–3797 (1986). [CrossRef]
9. T. Takagi, F. Koyama, and K. Iga, “Electron-wave reflection by multi-quantum barrier in n-GaAs/i-AlGaAs/n-GaAs tunneling diode,” Appl. Phys. Lett. 59(22), 2877–2879 (1991). [CrossRef]
10. N. Dagli, High-speed photonics device (Taylor & Francis, 2007), Chap. 4.
11. K. Iga, H. Uenohara, and F. Koyama, “Electron reflectance of multiqutum barrier (MQB),” Electron. Lett. 22(19), 1008–1009 (1986). [CrossRef]
12. E. Kapon and R. Bhat, “Low-loss single-mode GaAs/AIGaAs optical waveguides grown by organometallic vapor phase epitaxy,” Appl. Phys. Lett. 50(23), 1628–1630 (1987). [CrossRef]
13. P. W. A. McIlroy, P. M. Rodgers, J. S. Singh, P. C. Spurdens, and I. D. Henning, “Low-loss single-mode InP/InGaAsP waveguides grown by MOVPE,” Electron. Lett. 23(13), 701–703 (1987). [CrossRef]
14. R. J. Deri and E. Kapon, “Low-loss III-V semiconductor optical waveguides,” IEEE J. Quantum Electron. 27(3), 626–640 (1991). [CrossRef]
15. T. H. Stievater, D. Park, W. S. Rabinovich, M. W. Pruessner, S. Kanakaraju, C. J. Richardson, and J. B. Khurgin, “Enhanced electro-optic phase shifts in suspended waveguides,” Opt. Express 18(2), 885–892 (2010). [CrossRef] [PubMed]
16. J. Mendoza-Alvarez, L. Coldren, A. Alping, R. Yan, T. Hausken, K. Lee, and K. Pedrotti, “Analysis of depletion edge translation lightwave modulators,” J. Lightwave Technol. 6(6), 793–808 (1988). [CrossRef]
17. C. Cocorullo, M. Iodice, I. Rendina, and P. M. Sarro, “Silicon thermo-optical micro-modulator with 700 kHz 3dB bandwidth,” IEEE Photon. Technol. Lett. 7(4), 363–365 (1995). [CrossRef]
18. S. S. Lee, R. V. Ramaswamy, and V. S. Sundaram, “Analysis and design of high-speed high-efficiency GaAs-AlGaAs double-heterostructure waveguide phase modulator,” IEEE J. Quantum Electron. 27(3), 726–736 (1991). [CrossRef]
19. E. L. Wooten, K. M. Kissa, A. Y. Yan, E. J. Murphy, D. A. Lafaw, P. F. Hallemeier, D. M. Maack, D. V. Attanasio, D. J. Fritz, G. J. McBrien, and D. E. Bossi, “A review of lithium niobate modulators for fiber-optic communications systems,” IEEE J. Sel. Top. Quantum Electron. 6(1), 69–82 (2000). [CrossRef]
20. S. Nishimura, H. Inoue, H. Sano, and K. Ishida, “Electrooptic effects in an InGaAs/InAlAs multiquantum well structure,” IEEE Photon. Technol. Lett. 4(10), 1123–1126 (1992). [CrossRef]
21. S. Adachi, Physical Properties of III–V Semiconductor Compounds: InP, InAs, GaAs, GaP, InGaAs, and InGaAsP (Wiley-VCH, 1992).
