Putting on the shift

Optical applications such as free-space optical communications, target identification, and tracking involve frequent repositioning of the beam. For many years radars have been able to jump directly from one angular position to another, while optical systems have used complex and expensive mechanical structures for this task. Finally, we are nearing the time when electro-optic systems will be able to jump from one angular position to another with no moving parts, or only micromotion.

We all know that a prism bends light by delaying light going through the thick side of the prism more than the light going through the thin side, since the index of refraction of light is higher within the prism than in free space or air. Light on the thin side of the prism gets ahead of light going through the thick side, which introduces a wavefront tilt. As a result, a collimated beam propagating through the prism will be "steered" in a direction opposite the tilt of the prism.

A slab of material of constant thickness that featured a greater refraction index on one side than the other, would constitute another form of a prism. If we could electronically change the index of refraction of a slab, we could write an index-of-refraction-based prism with no moving parts. Liquid crystal (LC) material allows us to do just that.

Figure 1. Voltage applied to LC film induces a phase shift (left). When an LC domain is oriented (right) to incident light, it induces a phase shift.

LC is made up of long, skinny molecules. By changing the voltage across an LC cell, it is possible to rotate the orientation of these long, skinny molecules and thus change the index of refraction for a given polarization (see figure 1 on page 16). The only difficulty with LC is the allowed thickness of the slab over which we can rapidly change the index of refraction. Traditionally, LC re-orientation speed is proportional to the thickness of the LC layer squared, so for rapid operation we would prefer to keep the LC layer thin.

Figure 2. Phase shifters (red) introduce a ramp in the phase profile (black). At the reset point, the shifters create a flyback region with opposite phase profile (blue).

One way to do this is to take advantage of the fact that electromagnetic radiation is in the form of a sine wave. We can create a phase ramp that builds to one wavelength of phase shift (2π), then we can subtract one wavelength (another 2π).¹ Since the waveform is a sine wave, subtracting one wavelength makes no difference in the unfolded, effective, phase front for this exact wavelength, (see figure 2). Using modulo 2π phase profiles does, however, make the beam steerer wavelength dependent (dispersive).

Figure 3. A set of crossed gratings (left) provides a serial 1-D method for steering in both azimuth and elevation.

If we electronically address this prism we want to have individual addressable elements separated by one half-wavelength or less. This allows steering a beam to large angles. For electro-optical phased arrays this is much more difficult than for microwave phased arrays. The difference between 10 GHz (3 cm) and 200 THz (1.5 µm) is a factor of 20,000. Phased arrays operating at 10 GHz have a half-wavelength spacing of 1.5 cm compared to phased arrays operating at 1.5 µm, which have a half-wavelength spacing of 0.75 µm. Because of this scale change, the best short-term solution is to steer the beam in one dimension at a time, due to addressing considerations. For example, a 10 cm x 10 cm aperture on 5-µm center-to-center spacing would contain 400 million phase-shifting elements compared to 40,000 phase-shifting elements in a crossed 1-D approach. A crossed set of gratings can steer in both azimuth and elevation (see figure 3).

LC material suffers from fringing fields and resistance to rapid change in orientation, which makes it difficult to electronically address an LC array at half-wavelength spacing. If we have a 1-µm wavelength and a birefringence of 0.25, we need to have light travel through 4 µm of LC to introduce an optical path delay of 1λ (a 2π phase shift). If we have two conducting plates separated by 4 µm and we have addressing electrodes spaced closer than 4 µm, the electric fields between the parallel plates will only change at about the 4-µm spacing. This difficulty does not manifest during the uniform phase-ramp portion of the addressing process. Instead, the difficulty comes at the reset point at which we would like to immediately drop the phase delay from 2pi to zero (to physically subtract one wavelength of optical path delay, or OPD, which keeps it small while not affecting the unfolded phase profile).

During the phase ramp, fringing field effects smooth the phase profile. At the reset point, the fringing field creates a spatial region with opposite phase profile. As indicated in figure 2, this results in a portion of the wave being steered to an opposite angular position (the blue portion). For small steering angles, this flyback region is a small percentage of the phase ramp and is of little consequence, but as angles get larger, it means a significant portion of the wavefront is steered to the wrong angle, resulting in a loss in the correct direction and high sidelobes. This is why we would like half-wavelength spacing. Spacing at the half-wavelength level would allow large-angle beam steering without creating this detrimental flyback effect. For optimal performance in the near term, the beam-steering capabilities of LC devices are limited to 5° or so in either direction in order to maintain high-efficiency beam steering.

Holographic glass and birefringent prisms offer two methods for widening the angle of a beam-steering system without obtaining half-wavelength spacing in LC beam steerers. In the holographic glass approach, multiple holograms are angularly multiplexed in glass to produce large-angle deflections. An LC beam steerer placed in front of the glass selects which hologram is addressed, allowing the user to select the desired large angle of deflection. Another LC beam steerer placed after the holographic glass steers to angles within the zone. We refer to this as filling the zone.

Raytheon (Lexington, MA) is pursuing this approach under the Steered Agile Beams Program (STAB), sponsored by the Defense Advanced Research Projects Agency. The team has demonstrated high-efficiency beam steering at 20° and near 40°. They have also demonstrated zone fill and continuous steering between zones near 20°. Because LC optical phased arrays are polarization- dependent, we can steer both polarizations by doubling the number of beam-steering elements.

Birefringent prisms provide another approach to wide-angle beam steering. This approach begins with an azimuth/ elevation LC beam steerer capable of continuous small- to moderate-angle beam steering. In this case larger angles are reached by a set of binary birefringent prisms. One polarization is deflected a given angular amount in one direction. The other polarization, orthogonal, is deflected the same amount in the opposite direction. LC waveplates are used to rotate polarization between prism layers. Rockwell Scientific (Thousand Oaks, CA) is pursuing this approach under the STAB program. Their design uses 3°, 6°, and 12° prism steering angles. Since these angles are each a factor of two larger than the one before, it is possible to come up with a combination allowing steering to any angle up to the maximum angle at which all prisms are steering the same direction. Their team has demonstrated steering angles of greater than ±24° using this approach.

As we've already noted, if we use modulo 2π beam steering, we have a dispersive system. This is fine for most laser-radar or laser-communications applications. If we wish to use the technique for wide-band beam-steering systems, however, we have to reduce the dispersion. One method for achieving this is to use larger phase resets.²

Phase resets work because the effective phase is the same after subtracting 2π phase as it was before the subtraction. When the wavelength differs from the design wavelength, each reset introduces a phase discontinuity. If, instead of a 2π reset, we increase the thickness of the LC layer and the OPD by a factor of a 100 to 200pi reset, we do not have to be far off the design wavelength before the new wavelength will divide into the OPD an exact number of wavelengths. Only a small wavelength shift will cause the new wavelength to divide into the 100X design wavelength OPD by either 99 or 101 times, for example, resulting in a smooth, unfolded phase profile. This is the reason that larger resets (e.g., 200π) yield better dispersion properties.

The challenge of using modulo 200π resets is that the system must be capable of introducing much larger OPDs. Typically, this is accomplished by increasing the thickness of the LC layer. Unfortunately, LC switching slows as a function of the square of the film thickness, making the approach unfeasible. Recent work at Kent State University (Kent, OH), however, has demonstrated a method that can accomplish this goal while maintaining required speeds. Researchers mixed polymers into the LC in a certain manner, which alleviated the speed problem. The remaining disadvantage appears to be that the thicker LC layers require increased voltage to maintain the same electric field.

In the long run, of course, people will learn how to use the inherent dispersion of optical phased-array devices to enhance multispectral sensors, since spectral-based sensors require some dispersion. At this time, we have not learned to make use of the dispersion of optical phased-array beam steerers in the design of spectral-based sensors.

Additional non-mechanical approaches for beam steering also are becoming available. These include microlens arrays, photonic crystal superdispersive devices, and micro-electro-mechanical systems (MEMS). Of these, MEMS technology is the most mature but has been developed mostly in very small devices for cross-bar switches in the telecom industry, not so much for laser beam steering in sensing, countermeasure, or free-space optical applications. Further work will be necessary to reap the benefits of both technologies. oe

References

1. P. McManamon, T. Dorschner, et al., Proc. IEEE 84(2), 268-298 (1996).

2. P. McManamon, E. Watson, Proc. SPIE Vol. 4369, 140-148 (2001).

Paul McManamon

Paul McManamon is the senior scientist for IR sensors at the Sensors Directorate of the Air Force Research Laboratory, Wright-Patterson Air Force Base, OH.