GaAs-based surface-normal optical modulator compared to Si and its wavelength response characterization using a supercontinuum laser

doi:10.1364/OE.19.004076

« Show journal navigation

GaAs-based surface-normal optical modulator compared to Si and its wavelength response characterization using a supercontinuum laser

Ojas P. Kulkarni, Mohammed N. Islam, and Fred L. Terry »View Author Affiliations

Optics Express, Vol. 19, Issue 5, pp. 4076-4084 (2011)
http://dx.doi.org/10.1364/OE.19.004076

View Full Text Article

Acrobat PDF (1082 KB)

Journal Search
Article Lookup

Article Tools

Citations

Alert me when this article is cited
Export Citation/Save

Full Text
Article Info
References (14)
Cited By
Figures (13)
Metrics

Abstract

A GaAs-based surface-normal optical modulator using the free-carrier effect is demonstrated for the first time to our knowledge. The device exhibits ~43% modulation depth compared to 24% for a previously demonstrated Si-based device with twice the interaction length. Simulations predict ~1.8 times the speeds for GaAs-based devices compared to Si. Operation in conjunction with a supercontinuum source is used to characterize the wavelength response of the modulator. Potential for colorless operation makes the modulator a candidate for wavelength-division multiplexed networks with broadband light sources.

1. Introduction

We demonstrate a surface-normal optical modulator based on the free-carrier effect (FCE) in GaAs. The device exhibits a peak modulation depth of ~43% with a 3-dB bandwidth of ~20 MHz. Compared to a previously demonstrated Si-based modulator [1

O. Solgaard, A. A. Godil, B. R. Hemenway, and D. M. Bloom, “All-silicon integrated optical modulator,” IEEE J. Sel. Areas Comm. 9(5), 704–710 (1991). [CrossRef]

], the fabricated device provides ~1.8 times the modulation depth in half the device interaction length. The relative response of the modulator as a function of incident wavelength is also characterized using a supercontinuum (SC) laser and modulation is observed over 1200-2400 nm wavelength range.

The device consists of two forward biased pin diodes, which independently modulate the phase of two halves of an incident beam. By coupling the differentially phase modulated beam in to a single mode fiber (SMF), phase to amplitude modulation conversion is achieved.

Simulation results are also presented, which suggest ~1.8 times the bandwidth in GaAs-based devices compared to Si. However, higher modulation speeds compared to Si could not be demonstrated due to parasitics in the fabricated device.

The demonstrated modulator has potential applications in low-cost access networks operating at few tens of MHz speeds. Also, in an array form, the device could be adapted to beam steering and shaping applications due to its phase modulation capabilities.

2. Motivation and background

Surface-normal optical modulators are attractive for applications such as mobile optical communication links [2

G. C. Gilbreath, W. S. Rabinovich, T. J. Meehan, M. J. Vilcheck, R. Mahon, R. Burris, M. Ferraro, I. Sokolsky, J. A. Vasquez, C. S. Bovais, K. Cochrell, K. C. Goins, R. Barbehenn, D. S. Katzer, K. Ikossi-Anastasiou, and M. J. Montes, “Large-aperture multiple quantum well modulating retroreflector for free-space optical data transfer on unmanned aerial vehicles,” Opt. Eng. 40(7), 1348–1356 (2001). [CrossRef]

], optical signal processing arrays [3

B. Noharet, Q. Wang, S. Junique, D. Ågren, and S. Almqvist, ““Multiple quantum well spatial light modulators for optical signal processing,” Integrated Optical Devices, Nanostructures, and Displays,” Proc. SPIE 5618, 146–155 (2004). [CrossRef]

], chip-to-chip optical interconnects [4

H. Liu, C. C. Lin, and J. S. Harris, “High-speed, dual-function vertical cavity multiple quantum well modulators and photodetectors for optical interconnects,” Opt. Eng. 40(7), 1186–1191 (2001). [CrossRef]

] and pulse-shaping [5

Y. Ding, R. M. Brubaker, D. D. Nolte, M. R. Melloch, and A. M. Weiner, “Femtosecond pulse shaping by dynamic holograms in photorefractive multiple quantum wells,” Opt. Lett. 22(10), 718–720 (1997). [CrossRef] [PubMed]

]. In communication systems, modulators operating over a broad wavelength range can enable the use of centralized broadband light sources in access networks and increase system throughput [6

B. C. Collings, M. L. Mitchell, L. Boivin, and W. H. Knox, “A 1021 channel WDM system,” IEEE Photon. Technol. Lett. 12(7), 906–908 (2000). [CrossRef]

]. Also, modulators capable of operation in the 2-2.4 μm wavelength region have applications in free-space systems due to the lower attenuation at these wavelengths in the atmosphere and their ability to pass through haze [7

A. Arnulf, J. Bricard, E. Curé, and C. Véret, “Transmission by haze and fog in the spectral region 0.35 to 10 microns,” J. Opt. Soc. Am. 47(6), 491–497 (1957), http://www.opticsinfobase.org/abstract.cfm?URI=josa-47-6-491. [CrossRef]

Amongst semiconductor-based surface-normal optical modulators, a MEMS device has been recently demonstrated with >15 dB contrast [8

T. H. Stievater, D. Park, M. W. Pruessner, W. S. Rabinovich, S. Kanakaraju, and C. J. K. Richardson, “A microelectromechanically tunable asymmetric Fabry-Perot quantum well modulator at 1.55 microm,” Opt. Express 16(21), 16766–16773 (2008), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-16-21-16766. [CrossRef] [PubMed]

]. However, the device has limited operating speeds of ~1 MHz. Systems based on electro-absorption (EA) modulators are also under investigation [3

] and results suggest high contrast ratio (~10 dB) with operating speeds >1 GHz. However, demonstrated tunability of EA modulators is only ~70 nm [9

H. Mohseni, W. K. Chan, H. An, A. Ulmer, and D. Capewell, “Tunable surface-normal modulators operating near 1550 nm with a high-extinction ratio at high temperatures,” IEEE Photon. Technol. Lett. 18(1), 214–216 (2006). [CrossRef]

]. In comparison, FCE based devices are of interest due to their ability to operate over larger wavelength range.

In this paper, we demonstrate a FCE-based surface-normal optical modulator in GaAs for the first time to our knowledge. A Si-based FCE device has been demonstrated with 24% typical modulation depth and 60 MHz electrical bandwidth in the past [1

O. Solgaard, A. A. Godil, B. R. Hemenway, and D. M. Bloom, “All-silicon integrated optical modulator,” IEEE J. Sel. Areas Comm. 9(5), 704–710 (1991). [CrossRef]

]. However, theoretical studies predict larger phase shifts per unit length from the FCE in materials with smaller carrier effective masses such as GaAs compared to Si [10

B. R. Hemenway, “Integrated silicon light modulator for fiber-optic interconnects at 1.3 micron wavelength,” Stanford University dissertation, Ginzton Lab. Report #4703, May 1990.

]. Techniques such as ion-implantation used in the fabrication of the Si-based device, are not suited for GaAs, due to the tendency of the GaAs crystal to decompose and emit arsenic at the high temperatures of 800-900°C required for these processes [11

R. E. Williams, Gallium arsenide processing techniques , (Artech House, Inc., 1984).

]. The demonstrated device is instead fabricated on n-type GaAs substrate with epitaxially grown p-type and intrinsic layers to avoid exposure of the device to high temperatures during processing.

3. Device operation

The device schematic is shown in Fig. 1 . The modulator consists of two pin diodes that produce phase modulation in the two halves of the incident beam based on the FCE, which occurs due to the polarizability of free carriers in semiconductors at wavelengths above the semiconductor band-gap. Each half of the beam is phase modulated by an amount proportional to the instantaneous carrier density across the pin diodes. Subsequently, phase modulation to amplitude modulation conversion is achieved by coupling to SMF as proposed in [1

O. Solgaard, A. A. Godil, B. R. Hemenway, and D. M. Bloom, “All-silicon integrated optical modulator,” IEEE J. Sel. Areas Comm. 9(5), 704–710 (1991). [CrossRef]

]. In this scheme, the phase profile of the modulated beam can either be maintained uniform, which results in efficient coupling in to the fundamental mode of the SMF, or can be distorted so as to achieve reduced coupling in to the SMF as illustrated in Fig. 2 . In addition, to maximize the amplitude change corresponding to a certain phase modulation, a π/2 radians phase bias offset is introduced between the two halves of the incident light beam.

Fig. 1 Two-diode split beam device geometry.

Fig. 2 Phase to amplitude conversion coupling.

Recombination of carriers in the diode invariably leads to generation of heat. Change in the junction temperature also leads to change in the device refractive index. In most semiconductors, the thermal effect is more dominant compared to the FCE, resulting in undesirable operation at low speeds. The split beam geometry used in the device allows for push-pull drive configuration of the two diodes, which reduces the impact of the thermal effect by maintaining similar average temperature in both diodes.

4. Device fabrication

The device is fabricated on n-type substrate with epitaxially grown intrinsic and p-type layers with doping concentrations described in Table 1 . Epitaxially grown p-type layer provides a sharp doping profile transition compared to ion-implantation techniques [11

R. E. Williams, Gallium arsenide processing techniques , (Artech House, Inc., 1984).

], and thus can provide larger free carrier density contrast in the intrinsic region. The device intrinsic layer thickness is chosen to be 3 μm to allow for reliable fabrication and keeping fabrication costs low. Thicker layers can be used at the expense of increased epi-wafer costs.

Table 1 Thickness of Epitaxially Grown Layers and Target Doping Concentrations

Layer	Thickness	Doping concentration (cm⁻³)
n-type substrate Intrinsic layer p-type layer	400 μm 3 μm0.5 μm	1-4 x 10¹⁸ (Silicon) < 1 x 10¹⁵ 1 x 10²⁰ (Carbon)

The fabrication steps are shown in Fig. 3 . Two p-type ohmic contacts of an alloy of Pd/Zn in the ratio 1:2 with dimensions of 10 μm x 4 μm each and separated by 2 μm are patterned on the wafer. Connecting pads to the p-contact with ~0.2 mm² area are laid down to allow for connections to external drive circuitry (Fig. 4(a) ). The surface area of the pads is an important factor that contributes to the parasitic capacitance of the device due to the only 3 μm thick intrinsic layer separating the highly conducting p- and n-type regions, and hence needs to be kept as small as possible. The semiconductor surface is then wet-etched until the buried n-type layer is exposed. Au/Ge eutectic contact to the n-type substrate is patterned by lift-off and the device is annealed at 400°C after passivation to reduce the contact resistance.

Fig. 3 Schematic of device fabrication steps.

Fig. 4 (a) Fabricated p-contacts and vias. The p-contacts are separated by a 2μm trench. (b) Front side device window with ~56 nm phase bias etch on one half of the window.

To reduce free-carrier absorption (FCA) of light in the highly doped substrate, provision in the fabrication needs to be made to allow for thinning down the substrate. To achieve this goal, the wafer is flip-chip bonded to a glass carrier before lapping operation (Fig. 3). Since the device has two diodes separated by only a 2 μm trench, conventional wafer bonding techniques cannot be used due to the possibility of shorting the two diodes at high process temperatures. Instead, we use solder bonding at ~200° C to mount the device onto a glass carrier while ensuring electrical isolation between the two diodes. The GaAs substrate is then thinned down to ~100 μm, with the glass carrier substrate providing mechanical strength for further device processing. Back-side aligning is used to define a device window on the exposed semiconductor surface as shown in Fig. 4(b). Approximately 56 nm deep dry etch is performed on one half of the device window to achieve π/2 radians phase bias between the two halves of the incident light beam and a single layer anti-reflection (AR) coating of SiO is evaporated on to the device designed for an operation wavelength close to 1.55 μm.

4. Experimental set-up

Two optical sources are used in our device characterization experiments. A 1530 nm continuous wave distributed feedback (DFB) laser diode is used to measure the modulation depth, frequency response and insertion loss of the modulator. An amplified nanosecond pulsed SC source, as described in [12

C. Xia, M. Kumar, M.-Y. Cheng, O. P. Kulkarni, M. N. Islam, A. Galvanauskas, F. L. Terry Jr, M. J. Freeman, D. A. Nolan, and W. A. Wood, “Supercontinuum generation in silica fibers by amplified nanosecond laser diode pulses,” IEEE J. Sel. Top. Quantum Electron. 13(3), 789–797 (2007). [CrossRef]

], is used to measure the wavelength response of the modulator. The SC source, as shown in Fig. 5 , consists of a nanosecond pulse driven 1553 nm DFB laser diode with variable repetition rate followed by three stages of fiber amplifiers. Pulses amplified to ~4 kW peak power are coupled in to ~2 m of SMF followed by 6 m of high nonlinearity fused silica fiber with a zero dispersion wavelength of 1544 nm. The SC extends from ~900 nm to 2700 nm with ~165 mW average power.

Fig. 5 Schematic of the SC source used for wavelength response measurement.

The experimental set-up is shown in Fig. 6 . An imaging spectrometer is placed in the beam path to select the wavelength of light. An AR coated CaF₂ 50:50 beam splitter is used to direct light in and out of the modulator. Fused silica aspherical lenses are used to focus light in to the modulator and the reflected modulated light in to 2 m long SMF. The detection system at the output end of the SMF consists of an InGaAs detector. For modulation depth measurement, a DC to 1 MHz variable-gain amplifier is used. The frequency response of the modulator is measured using a RF-amplifier with ~13 dB gain and a 2 GHz oscilloscope.

Fig. 6 Experimental set-up.

For measuring wavelength response, a pulsed SC source is used. To ensure overlap of the optical pulses and the modulating signal, the repetition rate of the SC source and the modulation signal are synchronized at 100 kHz. The modulated optical signal is detected by a strained InGaAs detector (0.8-2.6 μm operation) followed by a lock-in amplifier. The relative response of the modulator is measured by comparing the amplitude of the lock-in voltage normalized to the incident average power over the wavelength range of the source.

The device is operated in a push-pull configuration and the drive circuitry is shown in Fig. 7 . A sine wave generator is used to drive the input terminal of a hybrid coupler. The 180° out-of-phase outputs are fed to two bias tees which combine the AC signal with a DC bias current.

Fig. 7 Drive circuit schematic.

5. Experimental results

A maximum modulation depth of ~43% is observed with the 1530 nm DFB laser, as shown in Fig. 8 , at a drive current of 30 mA DC + 20 mA AC rms current at 100 kHz. The overall insertion loss of the modulator is ~7 dB, comprised of ~3 dB coupling loss to SMF in addition to ~4 dB of device loss. For a substrate with ~4x10¹⁸ cm⁻³ doping concentration, the FCA coefficient is ~30 cm⁻¹ [13

W. G. Spitzer and J. M. Whelan, “Infrared absorption and electron effective mass in n-type gallium arsenide,” Phys. Rev. 114(1), 59–63 (1959). [CrossRef]

], which corresponds to ~2.6 dB loss on double pass through a 100 µm substrate. The remaining 1.4 dB loss occurs due to light escaping through the isolation trench and diffraction of the beam.

Fig. 8 Modulation depth measured with a 1530 nm continuous wave laser.

Optical modulation is achieved in the modulator in two steps. First, the varying density of free carriers in the device active region produces a variable phase shift in the incident light beam. The resulting phase modulation is then converted to amplitude modulation by coupling to SMF. Figure 9 gives evidence of the phase to amplitude conversion process by showing the observed modulation depth in the optical beam before and after the ~2 m long SMF in the set-up. The data suggests that before coupling in to the SMF, there is only ~8% modulation on the light amplitude, which occurs due to interference between the modulated light beam and the weak reflection from the AR coating interface. At the output end of the SMF, ~43% modulation depth is observed due to the mode selectivity of the fiber between the device on and off states as outlined in Fig. 2. However, the SMF also adds ~3 dB of coupling loss to the overall insertion loss of the modulator.

Fig. 9 Modulation depth before and after ~2 m SMF.

The wavelength response of the modulator, measured using the SC source is shown in Fig. 10 . Modulation is observed over 1200-2400 nm with peak performance around 1500 nm and response falling off on either side. Our data suggests >1 dB modulation over 300 nm from 1450 to 1750 nm. Loss of modulation at the lower wavelengths is due to the cut-off wavelength of the SMF. Below 1200 nm, the fiber is multimode resulting in poor phase modulation to amplitude modulation conversion. At wavelengths above 2400 nm, the high absorption loss in the fiber prevents observation of modulation. The peak of the wavelength response around 1500 nm is due to the spectral characteristics of the single layer AR coating deposited on the device. Away from 1500 nm, the reflection from the front surface of the device increases and causes the observed modulation depth to fall. Also, the phase bias etch of ~56 nm corresponds to an optimal π/2 radians phase difference between the two halves of the incident beam only around 1530 nm wavelength. Away from this wavelength, the available amplitude change for a given phase change reduces, leading to reduced modulation depth.

Fig. 10 Wavelength response measured using SC source.

The frequency response of the modulator is shown in Fig. 11 . The relative modulation depth is measured by comparing the amplitude of the detected optical signal at higher frequencies compared to that at 100 kHz. We observe a 3-dB bandwidth of ~20 MHz and a maximum speed of operation of ~270 MHz. At higher frequencies, the modulation depth is too weak to be detected by the set-up. The degradation in the frequency response of the modulator beyond 20 MHz is attributed to poor RF isolation between the two diodes, which is confirmed by measuring the crosstalk between the two pin diodes versus frequency as shown in Fig. 12 . The figure shows the relative power parasitically coupled in to the neighboring diode versus input frequency. Below 20 MHz, the isolation between the two diodes is >-10 dB. However, at higher frequencies, more power from one diode leaks into the other, which results in the reduction of phase difference between the two halves of the incident light beam leading to the substantial drop in the observed frequency response.

Fig. 11 Frequency response of the modulator.

Fig. 12 Power coupled in diode 2 relative to power delivered to diode 1.

6. Simulation results

Simulations are used to compare the theoretical performance of similar devices in GaAs and Si. At near infrared wavelengths, the FCE can be modeled by Drude theory [10

B. R. Hemenway, “Integrated silicon light modulator for fiber-optic interconnects at 1.3 micron wavelength,” Stanford University dissertation, Ginzton Lab. Report #4703, May 1990.

], where the induced phase shift per unit length in an incident light beam is given by,

Δ φ = - \frac{q^{2} λ}{4 π c^{2} ε_{0} n} (\frac{Δ N_{e}}{m_{e}^{*}} + \frac{Δ N_{h}^{}}{m_{h}^{*}})

(1)

where, λ – wavelength of incident light
ΔN_e,h – change in electron (e) and hole (h) density
m_e*, m_h* - electron and hole effective mass

The simulation methodology is depicted in Fig. 13(a) . Using Sentaurus Device simulator tool, the charge continuity for the device structure is evaluated to solve the Poisson’s equation and develop a map of charged carrier density in the device. Simulations with voltage signals of up to 1 GHz are performed and the resulting carrier densities in the diode active region in the device on- and off-states are obtained. The phase shift experienced by the incident light beam along the wave front is calculated using Eq. (1) and material parameter values summarized in [14

S. L. Chuang, Physics of optoelectronic devices , (Wiley-Interscience Publication, 1995).

]. The coupling efficiency to SMF is determined by computing the Gaussian mode overlap integral in the on- and off- states of the diodes to give the expected modulation depth over a range of input frequencies.

Fig. 13 (a) Methodology of simulations (b) Predicted frequency response comparison for identical modulators in GaAs and Si.

Figure 13(b) compares the simulation results for identical modulators in GaAs and Si under similar bias conditions, suggesting an expected peak modulation depth of ~50% for the GaAs modulator and ~20% for a Si modulator in the 1550 nm region. The modulation depth from a GaAs modulator is expected to be larger than that of a Si modulator with identical structure and drive conditions due to the lower effective mass of the electrons (m_e*) in GaAs [10

B. R. Hemenway, “Integrated silicon light modulator for fiber-optic interconnects at 1.3 micron wavelength,” Stanford University dissertation, Ginzton Lab. Report #4703, May 1990.

]. From Eq. (1), it follows that for a fixed carrier density change between the device on- and off-state, direct band-gap material systems with lower m_e* values such as GaAs, should yield larger phase shift per unit device interaction length and hence larger modulation depth. The simulation results suggest the 3-dB bandwidth in GaAs to be ~600 MHz compared to ~350 MHz in the case of Si, implying ~1.8 times larger operation bandwidth in GaAs. The better speed performance in GaAs as suggested by the simulation can be attributed to the higher carrier mobility and twice as large carrier drift velocity in GaAs compared to Si.

7. Discussion

The measured results for the modulator frequency response are not in agreement with performance expected based on the simulations due to device parasitics. One of the main contributing factors to the device parasitics is the poor RF isolation between the two diodes in the fabricated device. Better RF isolation requires better control over the etch profile of the isolation trench, which has an aspect ratio >2 due to a depth of ~4.5 µm and width of only 2 µm. A taper in the profile of the isolation trench can lead to higher capacitive coupling between the two diodes. Strong evidence is not available to verify this hypothesis due to the small size of the isolation trench. However, the tapering of the isolation trench can be attributed to the complex dependence of etch profiles on process parameters in dry etching.

The observed peak modulation depth of ~43% is in agreement with the expected modulation depth calculated from simulations. Calculations suggest observation of ~0.56 radians peak phase shift between the two arms of the modulator and the experimental data corresponds to ~0.46 radians. The variation can be attributed to the residual reflection from the AR coating interface. Larger modulation depth can be achieved by increasing the intrinsic region thickness. For example, simulations suggest that by increasing the device intrinsic region thickness to 6 µm, ~75% modulation depth can be achieved in theory. However this would result in a significant increase in epi-wafer costs.

The observed wavelength response of the modulator is strongly dependent on the spectral characteristics of the single layer AR coating deposited on the device. For device insertion losses of ~4 dB, analytical calculations suggest that when the front face reflectivity of the device exceeds ~20%, the unmodulated component in the reflected light is stronger and subsequently the observed modulation depth falls below 3 dB from the maximum. The 3-dB wavelength cutoff can be made larger by using broadband multilayered TiO₂/MgF₂ AR coating. The wavelength of operation for the modulator observed to be from 1200 to 2400 nm is limited by the properties of the SMF. The designed modulator consists of the pin junction device that produces a differential phase shift in the incident light beam, and the SMF provides subsequent phase modulation to amplitude modulation conversion. Hence, even though the pin device may produce a phase shift on the incident light beam at wavelengths down to the bandgap of GaAs (~870 nm) and beyond 2400 nm, amplitude modulation cannot be observed at these wavelengths due to the multimode behavior at shorter wavelengths and the high absorption at longer wavelengths in standard SMF. In the absence of the AR coating response and wavelength-dependent SMF properties such as mode field diameter and absorption, Eq. (1) indicates that the phase shift due to free carriers should increase with wavelength. However, calculations suggest that due to the ~56 nm phase bias etch in the device design, the increased phase shift should not result in an appreciable increase in the observed modulation depth. For example, at 2.4 µm wavelength the phase-bias between the two halves of the beam is ~1 radian instead of the optimal π/2 radians necessary for maximum amplitude change for a given phase change. As a result, the expected modulation depth at 2.4 µm would increase only marginally to ~52% compared to ~50% at 1.55 µm in the presence of an ideal AR coating. However, due to the larger mode field diameter in SMF at 2.4 µm, the device insertion loss is also expected to be much higher. The analysis suggests that the wavelength response of the modulator should be largely governed by the spectral characteristics of the coating and the fiber properties.

Several feasible improvements would make the device attractive for practical applications in communication networks. The insertion loss of the modulator can be reduced by designing devices on semi-insulating substrates. Absorption loss in a 100 µm substrate with ~10¹⁶ cm⁻³ doping concentration should be less than 1 dB compared to ~2.6 dB in the ~1-4x10¹⁸ cm⁻³ n-type doping concentration of the substrate used in our fabrication. Better frequency response of the modulator can also be achieved by locating all contact pads onto a semi-insulating surface rather than highly doped GaAs substrate to reduce parasitic capacitance [11

R. E. Williams, Gallium arsenide processing techniques , (Artech House, Inc., 1984).

]. Vertical etch profiles in the isolation trench need to be achieved in order to minimize the crosstalk between the two pin diodes, which requires greater process control.

8. Summary

In summary, we have demonstrated an optical modulator with surface-normal geometry in GaAs based on the FCE. The modulator demonstrates a maximum modulation depth of ~43% obtained by varying the free carrier density across only 3 μm long device intrinsic region. The 3-dB modulation bandwidth is measured to be ~20 MHz and operation is demonstrated in conjunction with a broadband SC laser over a wavelength range of 1200-2400 nm.

Compared to a surface-normal modulator in Si, we observe ~1.8 times more modulation depth in half the interaction length in a GaAs modulator, largely due to the superior electrical properties of GaAs as pertinent to FCE. Simulations suggest better speed performance for GaAs-based modulators compared to Si. However, higher modulation speeds could not be demonstrated due to poor RF isolation between the two differentially modulating pin diodes.

The modulator can be a potential low-cost solution for access systems with >50 nm centralized light sources. With colorless modulators that operate over a broad wavelength range, more channels can be provided by a single central office and hence lead to higher capacity systems with potentially lower costs. The modulator can also lead to applications in free-space communication systems where 2-2.4 μm wavelengths are used due to higher penetration of these wavelengths through the atmosphere, smoke and fog.

Acknowledgements

The authors would like to thank Tellabs, Inc. for providing funding for this research and the staff at the Lurie Nanofabrication facility for providing assistance in the device fabrication.

References and links

1.	O. Solgaard, A. A. Godil, B. R. Hemenway, and D. M. Bloom, “All-silicon integrated optical modulator,” IEEE J. Sel. Areas Comm. 9(5), 704–710 (1991). [CrossRef]
2.	G. C. Gilbreath, W. S. Rabinovich, T. J. Meehan, M. J. Vilcheck, R. Mahon, R. Burris, M. Ferraro, I. Sokolsky, J. A. Vasquez, C. S. Bovais, K. Cochrell, K. C. Goins, R. Barbehenn, D. S. Katzer, K. Ikossi-Anastasiou, and M. J. Montes, “Large-aperture multiple quantum well modulating retroreflector for free-space optical data transfer on unmanned aerial vehicles,” Opt. Eng. 40(7), 1348–1356 (2001). [CrossRef]
3.	B. Noharet, Q. Wang, S. Junique, D. Ågren, and S. Almqvist, ““Multiple quantum well spatial light modulators for optical signal processing,” Integrated Optical Devices, Nanostructures, and Displays,” Proc. SPIE 5618, 146–155 (2004). [CrossRef]
4.	H. Liu, C. C. Lin, and J. S. Harris, “High-speed, dual-function vertical cavity multiple quantum well modulators and photodetectors for optical interconnects,” Opt. Eng. 40(7), 1186–1191 (2001). [CrossRef]
5.	Y. Ding, R. M. Brubaker, D. D. Nolte, M. R. Melloch, and A. M. Weiner, “Femtosecond pulse shaping by dynamic holograms in photorefractive multiple quantum wells,” Opt. Lett. 22(10), 718–720 (1997). [CrossRef] [PubMed]
6.	B. C. Collings, M. L. Mitchell, L. Boivin, and W. H. Knox, “A 1021 channel WDM system,” IEEE Photon. Technol. Lett. 12(7), 906–908 (2000). [CrossRef]
7.	A. Arnulf, J. Bricard, E. Curé, and C. Véret, “Transmission by haze and fog in the spectral region 0.35 to 10 microns,” J. Opt. Soc. Am. 47(6), 491–497 (1957), http://www.opticsinfobase.org/abstract.cfm?URI=josa-47-6-491. [CrossRef]
8.	T. H. Stievater, D. Park, M. W. Pruessner, W. S. Rabinovich, S. Kanakaraju, and C. J. K. Richardson, “A microelectromechanically tunable asymmetric Fabry-Perot quantum well modulator at 1.55 microm,” Opt. Express 16(21), 16766–16773 (2008), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-16-21-16766. [CrossRef] [PubMed]
9.	H. Mohseni, W. K. Chan, H. An, A. Ulmer, and D. Capewell, “Tunable surface-normal modulators operating near 1550 nm with a high-extinction ratio at high temperatures,” IEEE Photon. Technol. Lett. 18(1), 214–216 (2006). [CrossRef]
10.	B. R. Hemenway, “Integrated silicon light modulator for fiber-optic interconnects at 1.3 micron wavelength,” Stanford University dissertation, Ginzton Lab. Report #4703, May 1990.
11.	R. E. Williams, Gallium arsenide processing techniques , (Artech House, Inc., 1984).
12.	C. Xia, M. Kumar, M.-Y. Cheng, O. P. Kulkarni, M. N. Islam, A. Galvanauskas, F. L. Terry Jr, M. J. Freeman, D. A. Nolan, and W. A. Wood, “Supercontinuum generation in silica fibers by amplified nanosecond laser diode pulses,” IEEE J. Sel. Top. Quantum Electron. 13(3), 789–797 (2007). [CrossRef]
13.	W. G. Spitzer and J. M. Whelan, “Infrared absorption and electron effective mass in n-type gallium arsenide,” Phys. Rev. 114(1), 59–63 (1959). [CrossRef]
14.	S. L. Chuang, Physics of optoelectronic devices , (Wiley-Interscience Publication, 1995).

OCIS Codes
(060.2340) Fiber optics and optical communications : Fiber optics components
(250.4110) Optoelectronics : Modulators

ToC Category:
Optoelectronics

History
Original Manuscript: September 24, 2010
Revised Manuscript: February 5, 2011
Manuscript Accepted: February 8, 2011
Published: February 16, 2011

Citation
Ojas P. Kulkarni, Mohammed N. Islam, and Fred L. Terry, "GaAs-based surface-normal optical modulator compared to Si and its wavelength response characterization using a supercontinuum laser," Opt. Express 19, 4076-4084 (2011)
http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-19-5-4076

Sort: Author | Year | Journal | Reset

References

O. Solgaard, A. A. Godil, B. R. Hemenway, and D. M. Bloom, “All-silicon integrated optical modulator,” IEEE J. Sel. Areas Comm. 9(5), 704–710 (1991). [CrossRef]
G. C. Gilbreath, W. S. Rabinovich, T. J. Meehan, M. J. Vilcheck, R. Mahon, R. Burris, M. Ferraro, I. Sokolsky, J. A. Vasquez, C. S. Bovais, K. Cochrell, K. C. Goins, R. Barbehenn, D. S. Katzer, K. Ikossi-Anastasiou, and M. J. Montes, “Large-aperture multiple quantum well modulating retroreflector for free-space optical data transfer on unmanned aerial vehicles,” Opt. Eng. 40(7), 1348–1356 (2001). [CrossRef]
B. Noharet, Q. Wang, S. Junique, D. Ågren, and S. Almqvist, ““Multiple quantum well spatial light modulators for optical signal processing,” Integrated Optical Devices, Nanostructures, and Displays,” Proc. SPIE 5618, 146–155 (2004). [CrossRef]
H. Liu, C. C. Lin, and J. S. Harris, “High-speed, dual-function vertical cavity multiple quantum well modulators and photodetectors for optical interconnects,” Opt. Eng. 40(7), 1186–1191 (2001). [CrossRef]
Y. Ding, R. M. Brubaker, D. D. Nolte, M. R. Melloch, and A. M. Weiner, “Femtosecond pulse shaping by dynamic holograms in photorefractive multiple quantum wells,” Opt. Lett. 22(10), 718–720 (1997). [CrossRef] [PubMed]
B. C. Collings, M. L. Mitchell, L. Boivin, and W. H. Knox, “A 1021 channel WDM system,” IEEE Photon. Technol. Lett. 12(7), 906–908 (2000). [CrossRef]
A. Arnulf, J. Bricard, E. Curé, and C. Véret, “Transmission by haze and fog in the spectral region 0.35 to 10 microns,” J. Opt. Soc. Am. 47(6), 491–497 (1957), http://www.opticsinfobase.org/abstract.cfm?URI=josa-47-6-491 . [CrossRef]
T. H. Stievater, D. Park, M. W. Pruessner, W. S. Rabinovich, S. Kanakaraju, and C. J. K. Richardson, “A microelectromechanically tunable asymmetric Fabry-Perot quantum well modulator at 1.55 microm,” Opt. Express 16(21), 16766–16773 (2008), http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-16-21-16766 . [CrossRef] [PubMed]
H. Mohseni, W. K. Chan, H. An, A. Ulmer, and D. Capewell, “Tunable surface-normal modulators operating near 1550 nm with a high-extinction ratio at high temperatures,” IEEE Photon. Technol. Lett. 18(1), 214–216 (2006). [CrossRef]
B. R. Hemenway, “Integrated silicon light modulator for fiber-optic interconnects at 1.3 micron wavelength,” Stanford University dissertation, Ginzton Lab. Report #4703, May 1990.
R. E. Williams, Gallium arsenide processing techniques, (Artech House, Inc., 1984).
C. Xia, M. Kumar, M.-Y. Cheng, O. P. Kulkarni, M. N. Islam, A. Galvanauskas, F. L. Terry, M. J. Freeman, D. A. Nolan, and W. A. Wood, “Supercontinuum generation in silica fibers by amplified nanosecond laser diode pulses,” IEEE J. Sel. Top. Quantum Electron. 13(3), 789–797 (2007). [CrossRef]
W. G. Spitzer and J. M. Whelan, “Infrared absorption and electron effective mass in n-type gallium arsenide,” Phys. Rev. 114(1), 59–63 (1959). [CrossRef]
S. L. Chuang, Physics of optoelectronic devices, (Wiley-Interscience Publication, 1995).

Cited By

Alert me when this paper is cited

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.

Click here to see a list of articles that cite this paper