The concept of a modulating retro-reflector is an old one. However, until recently, modulators have not been available that could support a link at reasonable communications data rates. To be viable, the shutter must have a high switching speed, low power consumption, large area, wide field-of-view, and high optical quality. It must also function at wavelengths where good laser sources are available, be radiation-tolerant (for space applications), and be rugged. Mechanical shutters, for example, are too slow and heavy, and ferroelectric liquid crystal (FLC) devices are too slow (kilobits per second) and are not robust enough. To extend modulating retro-reflector links to data rates of megabits per second (Mbps) and higher, and to payloads that must operate over large temperature swings characteristic of installation out-of-doors and in space, NRL has pursued the use of a different type of electro-optic shutter. Specifically, we have been developing a semiconductor-based optical switch based on GaAs Multiple Quantum Wells (MQW). Figure 1 illustrates the concept of a modulating retro-reflector.

Multiple Quantum Well Modulators

Semiconductor MQW modulators are one of the few technologies that meet all the requirements described above. When used as a shutter, MQW technology offers many advantages: it is robust and all solid state, operates at low voltages (less than 20 V) and low power (tens of milliWatts), and is capable of very high switching speeds. MQW modulators have been run at Gbps data rates in fiber optic applications.

The MQW modulators used in this program were grown at NRL by molecular-beam epitaxy (MBE). The modulators consist of about 75-100 very thin (~10 nm) layers of several semiconductor materials, such as GaAs, AlGaAs, and InGaAs that are epitaxially deposited on a large (7.62 cm diameter) semiconductor wafer. Electrically, they take the form of a P-I-N diode. Optically, the thin layers induce a sharp absorption feature at a wavelength that is determined by the constituent materials and the exact structure that is grown.

When a moderate (~15V) voltage is placed across the shutter in reverse bias, the absorption feature changes, shifting to longer wavelengths and dropping in magnitude. Thus, the transmission of the device near this absorption feature changes dramatically. Figure 2 shows absorbance data for an InGaAs MQW modulator designed and grown at NRL for use in a modulating retro-reflector system. The figure illustrates how the application of a moderate voltage shifts the transmittance. Hence, a signal can be encoded in an On-Off-Keying format onto the carrier interrogation beam.

This modulator consists of 75 periods of InGaAs wells surrounded by AlGaAs barriers. The device is grown on an n-type GaAs wafer and is capped by a p-type contact layer, thus forming a P-I-N diode. This device is a transmissive modulator designed to work at a wavelength of 980 nm, compatible with many good laser diode sources. GaAs/AlGaAs modulators that work at 850 nm have also been grown. These materials have very good performance and operate in reflection architectures. Choice of modulator type and configuration architecture is application-dependent.

Once grown, the wafer is fabricated into discrete devices using a multi-step photolithography process consisting of etching and metallization steps. The NRL experimental devices have a 5-mm aperture, though larger devices are possible and masks for one centimeter devices are currently being designed as well. It is important to point out that while MQW modulators have been used in many applications to date, modulators of such a large size are uncommon and require special fabrication techniques. Figure 3 shows a block diagram and photo of a wide aperture MQW shutter designed, grown, and fabricated at NRL.

MQW modulators are inherently quiet devices, faithfully reproducing the applied voltage as a modulated waveform. An important parameter is contrast ratio, defined as Imax/Imin. This parameter affects the overall signal-to-noise ratio. Its magnitude depends on the drive voltage applied to the device and the wavelength of the interrogating laser relative to the exciton peak. The contrast ratio increases as the voltage goes up until a saturation value is reached. Typically, the modulators fabricated at NRL have had contrast ratios between 1.75:1 to 4:1 for applied voltages between 10 V and 25 V, depending on the structure.

There are three important considerations in the manufacture and fabrication of a given device: inherent maximum modulation rate vs. aperture size; electrical power consumption vs. aperture size; and yield.

Figure 1: A modulating retroreflector using a transmissive device is illustrated, where (1) is the interrogation beam; (2) is the retroreflected beam; (3) is the drive signal from the data source; (4) is the modulator; and (5) is a retroreflector.
Figure 1: A modulating retroreflector using a transmissive device is illustrated, where (1) is the interrogation beam; (2) is the retroreflected beam; (3) is the drive signal from the data source; (4) is the modulator; and (5) is a retroreflector.

     





Figure 2: Absorbance vs. Frequency is shown. In its quiescent state the MQW shutter blocks the transmission of incident light. When a moderate voltage is applied, the absorbance shifts and light is transmitted through to the retroreflector.
Figure 2: Absorbance vs. Frequency is shown. In its quiescent state the MQW shutter blocks the transmission of incident light. When a moderate voltage is applied, the absorbance shifts and light is transmitted through to the retroreflector.



     

Figure 3a: A schematic of a MQW modulator is shown. The GaAs material is grown in alternating layers with active regions about one micron thick.
Figure 3a: A schematic of a MQW modulator is shown. The GaAs material is grown in alternating layers with active regions about one micron thick.



     

Figure 3b: A fabricated 0.5cm transmissive device is shown. The NRL large aperture MQW devices operate between 850 nm and 1.06 microns to date.
Figure 3b: A fabricated 0.5cm transmissive device is shown. The NRL large aperture MQW devices operate between 850 nm and 1.06 microns to date.

Inherent Maximum Modulation Rate vs. Aperture Size

The fundamental limit in the switching speed of the modulator is the resistance-capacitance limit. A key trade then is area of the modulator vs. area of the clear aperture. If the modulator area is small, the capacitance is small, hence the modulation rate can be faster. However, for longer ranges on the order of several hundred meters, larger clear apertures are needed to close the link. For a given modulator, the speed of the shutter scales inversely as the square of the modulator diameter.

Electrical Power Consumption vs. Aperture Size

When the drive voltage waveform is optimized, the electrical power consumption of a MQW modulating retro-reflector scales as:

Dmod4•V2B2RS

Where Dmod is the diameter of the modulator, V is the voltage applied to the modulator (fixed by the required optical contrast ratio), B is the maximum data rate of the device, and RS is the sheet resistance of the device. Thus a large power penalty may be paid for increasing the diameter of the MQW shutter.

Yield

MQW devices must be operated at high reverse bias fields to achieve good contrast ratios. In perfect quantum well material this is not a problem, but the presence of a defect in the semiconductor crystal can cause the device to break down at voltages below those necessary for operation. Specifically, a defect will cause an electrical short that prevents development of the necessary electrical field across the intrinsic region of the PIN diode. The larger the device the higher the probability of such a defect. Thus, If a defect occurs in the manufacture of a large monolithic device, the whole shutter is lost.

To address these issues, NRL has designed and fabricated segmented devices as well as monolithic modulators. That is, a given modulator might be "pixellated" into several segments, each driven with the same signal. This technique means that speed can be achieved as well as larger apertures. The "pixellization" inherently reduces the sheet resistance of the device, decreasing the resistance-capacitance time and reducing electrical power consumption. For example, a one centimeter monolithic device might require 400 mW to support a one Mbps link. A similar nine segmented device would require 45 mW to support the same link with the same overall effective aperture. A transmissive device with nine "pixels" with an overall diameter of 0.5 cm was shown to support over 10 Mbps. A representative trace is shown in Figure 4. A Photograph of a modulator segmented into 9 pixels is shown in Figure 5.

This fabrication technique allows for higher speeds, larger apertures, and increased yield. If a single "pixel" is lost due to defects but is one of nine or sixteen, the contrast ratio necessary to provide the requisite signal-to-noise to close a link is still high. There are considerations that make fabrication of a segmented device more complicated, including bond wire management on the device, driving multiple segments, and temperature stabilization.

An additional important characteristic of the modulator is its optical wavefront quality. If the modulator abberates the beam, the returned optical signal will be attenuated and insufficient light may be present to close the link. In Figure 6, an infrared interferometric measurement of a one-cm piece of the InGaAs modulator is shown. As can be seen, the optical quality of the device is very good and should not deleteriously impact system performance.

Figure 4: Amplitude vs time for a nine-"pixel" modulator is shown. The trace shows that modulation rates on the order of 10 Mbps can be supported easily with the device.
Figure 4: Amplitude vs time for a nine-"pixel" modulator is shown. The trace shows that modulation rates on the order of 10 Mbps can be supported easily with the device.
Figure 5: Segmentation of NQW modulator into 9 pixels. The individual wire bonds are evident.
Figure 5: Segmentation of NQW modulator into 9 pixels. The individual wire bonds are evident.
An interferogram of an InGaAs transmissive MQW modulator is shown. The ripples around the square are edge effects around the sample. It can be seen that the surface is relatively good and should not deleteriously impact system performance.
An interferogram of an InGaAs transmissive MQW modulator is shown. The ripples around the square are edge effects around the sample. It can be seen that the surface is relatively good and should not deleteriously impact system performance.