Optical Phased Array Technology
`
`PAUL F. MCMANAMON, SENIOR MEMBER, IEEE, TERRY A. DORSCHNER, MEMBER, IEEE,
`DAVID L. CORKUM, LARRY J. FRIEDMAN, MEMBER, IEEE, DOUGLAS S. HOBBS,
`MICHAEL HOLZ, SERGEY LIBERMAN, HUY Q. NGUYEN, DANIEL P. RESLER,
`RICHARD C. SHARP, MEMBER, IEEE, AND EDWARD A. WATSON
`
`Optical phased arrays represent an enabling new technology
`that makes possible simple, affordable, lightweight, optical sensors
`offering very precise stabilization, random-access pointing, pro(cid:173)
`grammable multiple simultaneous beams, a dynamic focus/defocus
`capability, and moderate to excellent optical power handling ca(cid:173)
`pability. These new arrays steer or otherwise operate- on an
`already formed beam, as compared to modem microwave phased
`arrays which both generate a beam and direct it in a specific
`direction. A phase profile is imposed on an optical beam as it
`is either transmitted through or reflected from the phase shifter
`array. The imposed phase profile steers, focuses, fans out, or
`corrects phase aberrations on the beam. The array of optical
`phase shifters is realized through lithographic patterning of an
`electrical addressing network on the superstrate of a liquid crystal
`waveplate. Refractive index changes sufficiently large to realize
`full-wave differential phase shifts can be effected using low ( < 10
`V) voltages applied to the liquid crystal phase plate electrodes.
`High efficiency large-angle steering with phased arrays requires
`phase shifter spacing on the order of a wavelength or less; con(cid:173)
`sequently addressing issues make 1-D optical arrays much more
`practical than 2-D arrays. Orthogonal oriented 1-D phased arrays
`are used to deflect a beam in both dimensions. Optical phased
`arrays with apertures on the order of 4 em by 4 em have been
`fabricated for steering green, red, 1.06 f1m, and 10.6 {1m radiation.
`Steering efficiencies of about 60% at 4° and 85% at about 2°
`have been achieved to date with switching times as short as
`a few milliseconds in the visible. Fluences of several hundred
`W/cm 2 have been demonstrated at 10.6 11m with nonoptimally
`engineered devices. Higher fiuences can be handled at shorter
`wavelengths. Larger apertures are feasible, as is operation at
`other wavelengths and significantly faster switching times. System
`concepts that include a passive acquisition sensor as well as a
`laser radar are presented.
`
`Manuscript received June 30, 1995; revised November 14, 1995. Tills
`work was supported in part by Raytheon internal funds, and in part by
`the Air Force Wright Laboratory at Wright Patterson AFB, Dayton, OH.P.
`F. McManamon and E. A. Watson are with Wright Laboratory, Wright
`Patterson AFB, OH 45433 USA.
`D. L. Corkum is with Texas Instruments, Att1eborough, MA 02703
`USA.
`T. A. Dorschner, L. J. Friedman, D. S. Hobbs, M. Holz, D. P. Resler, and
`R. C. Sharp are with Raytheon Company, Electronic Systems, Lexington,
`MA 02173 USA.
`H. Q. Nguyen is with Kopin Corp., Taunton, MA 02173 USA.
`S. Liberman is with SemiTest, Billerica, MA 01821 USA.
`Publisher Item Identifier S 0018-9419(96)01390-4.
`
`I.
`
`INTRODUCTION
`Currently optical sensor systems, including laser radar,
`are often limited in performance and cost by mechanical
`beam directing and stabilization mechanisms. The requisite
`pointing and stabilization usually requires precise, rapid,
`mechanical motion, and is often associated ·with substan(cid:173)
`tial masses. Submicroradian steering precisions are often
`desired, but are usually impractical for available, afford(cid:173)
`able, mechanical beam directing systems. Most mechanical
`systems do not facilitate rapid random pointing. Further(cid:173)
`more, the rapid steering of a large aperture optical sensor
`often requires a prohibitive amount of power. Despite the
`considerable accumulated manufacturing experience in this
`field, mechanical beam steering for optical sensors remains
`complex, precise, and expensive.
`Optical phased arrays appear to have the potential to
`overcome many of the limitations of mechanical beam
`steering. Liquid-crystal-based phased arrays require very
`little prime power, even for large apertures, thereby opening .
`up application areas such as missile interceptors, satellite
`communications, and portable sensors of all types. Phased
`arrays are inherently random-access devices, a distinct
`advantage when regions of interest are distributed widely
`across a sensor field of regard (FOR). Unlike mechanical
`systems, liquid crystal devices are generally insensitive
`to accelerations, and their costs can drop rapidly with
`volume production, as is the general case for the electronic
`devices they resemble. Flat panel displays, fabricated using
`technologies that are similar to those required for liquid(cid:173)
`crystal optical phased arrays, are now inexpensive enough
`to be in every notebook computer.
`In the related microwave radar arena, phased arrays are
`rapidly displacing conventional horn antennas. The clear
`benefits of random-access, rapid beam pointing with no
`moving parts have made phased arrays the technology of
`choice, despite their high cost. Fortunately, cost trades for
`optical radars using optical phased arrays promise to be
`more favorable since the optical arrays are monolithically
`fabricated with no discrete elements, consist of an array
`of phase shifters rather than individual transmiUreceive
`
`PROCEEDINGS OF THE IEEE, VOL. 84, NO. 2, FEBRUARY 1996
`
`0018-9219/96$05.00 © 1996 IEEE
`
`268
`
`FINISAR 1009
`
`

`

`modules, and are designed to use low-cost addressing
`electronics. The optical phased arrays discussed here are
`passive arrays, consisting soley of phase shifters, and are
`operated as space-fed arrays, meaning that an already
`formed beam is fed to the array of phase shifters, which
`then effects steering of that beam. This contrasts to an active
`phased array in which individual transmit modules form a
`beam as it exits a large array of transmitters.
`There are many application areas that can benefit from
`the performance/cost benefits made possible by optical
`phased arrays. Inexpensive, reliable laser radar for target
`detection, wind profiling, and gas cloud identification are
`examples of high interest. Laser communication, whether
`effected with directed beams in free space or by switching
`of guided beams within fiber links, is another application
`area. Defense against infrared guided missiles benefits from
`directed laser energy, and is another potential optical phased
`array application area. Later in the paper issues associ(cid:173)
`ated with steering broadband optical energy are addressed.
`Passive infrared sensors for imaging or point detection
`applications can also benefit from phased array optical beam
`steering, but to a more limited degree at this time; however,
`future applications are expected to expand as techniques for
`reducing the influence of dispersion are developed.
`Optical beam steering by means of phased elements is
`a rich area heavily researched by prior workers. As early
`as 1971, Meyer [1] had developed a 1-D optical phased
`array using bulk, lithium tantalate phase shifters. The
`array comprised 46 phase shifters on one-half millimeter
`spacings. The number of addressable beam positions, beam
`widths, scan angles, and beam spacings all were shown to
`agree with theory as developed for microwave phased array
`antennas. Shortly thereafter Ninomiya [2] demonstrated a
`1-D array of lithium niobate electrooptic prism deflectors.
`The resolving power of the array was shown to be N
`times that of a single prism, where N is number of arrayed
`prisms. The array successfully demonstrated 50 resolvable
`spots with 600 V applied. Both discrete and continuous
`steering were demonstrated. The power required was noted
`to be similar to that for acousto-optic deflectors. Although
`these early phased arrays clearly demonstrated the concept,
`they were neither developed for high performance nor
`were intended for practical application. Large phase shifter
`spacings of hundreds of wavelengths were unavoidable,
`given the state of the technology, and precluded achieving
`efficient large angle beam steering. The small aperture fill
`factors also guaranteed large insertion losses. However,
`many of the key advantages of the phased array approach to
`beam steering were well appreciated by these early workers.
`Ninomiya pointed out that a phased array offers random
`access, that the resolving power of a phased array is high,
`that the steering angle is very accurate, and that there is no
`shift of optical frequency as with acousto-optic deflectors.
`Beam steering of visible light has recently been reported
`using a liquid crystal television panel as an elementary
`phased array [3]. Although liquid crystal displays are usu(cid:173)
`ally configured to effect intensity modulation, when the po(cid:173)
`larizers are removed the accompanying phase shift becomes
`
`observable. The display pixels are programmed to effect
`a discrete blazed-grating phase ramp across the aperture.
`However, the relatively large pixel spacing (several hundred
`waves), the nonunity array fill factor, and the limited
`available phase modulation depth ( 1.3?r) have severely
`limited the achievable steering efficiency and angle ( <0.1 °).
`Other workers in the field have attempted to develop
`higher performance optical phased arrays by greatly re(cid:173)
`ducing the phase shifter spacings. Vasey et al. [ 4] have
`developed an integrated optics approach comprising a 1-
`D phased array based on a linear array of closely spaced
`AlGaAs waveguides, the relative phases of which can be
`electrically adjusted using the electrooptic effect in the
`waveguiding material itself. Beams are coupled into the
`guided structure and launched into free space using grating
`couplers. Continuous steering is achieved by electrically
`imposing a linear phase ramp of adjustable slope across
`the aperture. Addressing is accomplished via a fine/coarse
`architecture, somewhat similar to the approach discussed
`in Section IV. Continuous steering of a 900 nm beam over
`a ±7.5 mrad field has been reported. Element spacings
`are orders of magnitude less than those in earlier bulk
`demonstrations, but remain multiple ( 13-14) wavelengths.
`Consequently, maximum steering angles are limited and
`efficiencies are low due to the large number of radiated
`diffraction orders (so-called grating lobes). Although the
`electrooptic effect used in this approach is inherently fast
`(ns), achieving the steering angles and efficiency levels
`required for laser radar is expected to be difficult, as is
`scaling to required aperture sizes and obtaining steering in
`two-dimensions.
`Another approach reported recently [5] uses a thin, 2-D
`array of liquid crystal phase shifting elements configured to
`operate as coherent microprisms with a relatively high fill
`factor. The individual elements are multiple wavelengths
`in extent and spacing, but are constructed to produce a
`linear phase ramp of adjustable slope across each element
`face, thereby simulating a discrete blazed grating with
`programmable blaze angle. The current device steers in
`one dimension only, although in principal two devices
`could be cascaded to steer in both dimensions. Unlike
`most preceding optical arrays, this device was specifically
`designed for laser radar application and is, in principle,
`capable of high efficiency at large angles (20°) and of being
`fabricated with large apertures. However, to date, only
`small apertures (2 mm square), moderate steering angles
`(5°), and low efficiencies (1 %-9%) have been achieved
`owing to fabrication difficulties inherent to the approach.
`One of the difficulties is that this approach requires using
`only the linear portion of the liquid crystal versus voltage
`curve, resulting in a limited use of available birefringence.
`A more classic approach to optical phased arrays has been
`under development by the authors. This work has resulted in
`development of a true optical phased array with 1-D phase
`shifter spacings smaller than a single free-space wave(cid:173)
`length, 100% aperture fill factors over significant apertures,
`and performance approaching theoretical predictions for
`small to moderate angles. The unity fill factor, small phase
`
`MCMANAMON et al.: OPTICAL PHASED ARRAY TECHNOLOGY
`
`269
`
`

`

`shifter spacings, and careful fabrication techniques used
`result in very low sidelobe levels. These are the first optical
`phased arrays which are capable of redirecting a single input
`beam into essentially a single, diffraction limited, output
`beam with negligible sidelobes. Using this approach, high
`performance optical phased array based steering of carbon
`dioxide laser beams (10.6 p,m) was first demonstrated in
`1989, with demonstrations of Nd:YAG steering (1.06 p,m)
`following soon thereafter [6], [7]. That work will soon
`appear in the open, reviewed, literature [8], [9].
`These new optical phased arrays are direct functional
`analogs of the well known microwave phased array anten(cid:173)
`nas [10] that make possible the agile, inertialess steering of
`microwave beams. The underlying fundamental concepts
`are identical to those for a microwave array. However,
`due to the orders-of-magnitude difference in wavelengths
`between the microwave and optical worlds, these new opti(cid:173)
`cal phased arrays have been implemented quite differently.
`The current optical devices are 1-D, space-fed, passive,
`phase-only, apertures. This differs from modern microwave
`phased arrays with which a beam is usually both formed
`and steered in two dimensions by a 2-D array of active
`elements. The field intensity across the aperture of an active
`microwave array is generally tapered at the edges in order
`to achieve low sidelobe levels. This is not an option with
`a passive, phase-only array. However, being space-fed, if
`the input optical beam is Gaussian in spatial profile, as is
`the usual case, additional tapering is not needed. The 1-D
`phase-only array steers an optical beam in one dimension
`only. Unlike modern microwave arrays, and most other
`optical phased arrays to date, these new arrays are designed
`to be easily cascaded. This allows simple mounting of
`orthogonal1-D arrays to steer the beam in two dimensions.
`Microwave arrays are built using discrete phase shifters,
`as have been most early optical phased arrays. However,
`since a vast number of phase shifters is needed to realize a
`high performance optical array, distributed liquid crystal
`phase shifters have been implemented, as described by
`Huignard et al. [11]; however, it has proven essential to
`implement additional innovative addressing means to avoid
`the otherwise impractical numbers of interconnects.
`The organization of this paper is as follows. In Section II
`liquid crystal optical phased array technology is summa(cid:173)
`rized. In Section III we briefly discuss alternative can(cid:173)
`didates to optical phased arrays for eliminating complex
`and expensive mechanical motion from laser radar optical
`systems. A more detailed description is presented in section
`IV. Section V summarizes performance levels achieved
`and predicted performance potential. Section VI considers
`the pointing of an acquisition sensor, often a passive in(cid:173)
`frared (IR) sensor. Section VII discusses laser radar system
`concepts that incorporate target acquisition and tracking
`capabilities. Section VIII contains conclusions.
`
`II. OVERVIEW OF LIQUID CRYSTAL OPTICAL
`PHASED ARRAY CONCEPTS
`A prism inserted into the aperture of an optical system
`introduces a linear gradient of optical path delay (OPD)
`
`across the aperture which tilts the phase front and thereby
`steers the optical beam. For a given wavelength a phase
`shift of 21r (corresponding to an OPD of one wavelength)
`can be subtracted periodically from the phase front without
`influencing the far-field pattern produced by the phase front
`[12]. The "folded" phase profile represents a blazed grating.
`The phase ramp, or its equivalent modulo-21r sawtooth
`phase profile, further can be approximated by a series of
`discrete phase steps, as long as the steps are small.
`Fig. 1 illustrates the use of nematic liquid crystal cells
`as phase shifters. With no applied fields, the liquid crystal
`molecules align with an average orientation parallel to
`the substrates, according to the liquid crystal alignment
`layer applied at the substrate interface. Application of a
`relatively low voltage, on the order of 1-10 V, reorients
`the liquid crystal molecules and changes the effective index
`of refraction as seen by light polarized along the direction
`of quiescent molecular orientation. The maximum phase
`shift available is proportional to the thickness of the liquid
`crystal layer. The case of a 21r phase retarder is illustrated.
`The switching speed of a nematic liquid crystal phase
`shifter is generally inversely proportional to the square of
`the thickness of the nematic liquid crystal layer [13]. For
`steering angle/aperture size combinations that require phase
`resets, the minimum thickness of the liquid crystal layer
`to produce efficient steering requires a liquid crystal layer
`sufficiently thick to produce a full wavelength of OPD and
`allow modulo 21r operation. Only a combination of very
`small angles, or very small aperture size, allows practical
`beam steering without the use of resets. The liquid crystal
`layer thickness, t, for a 21r phase shift is given by
`
`(1)
`
`where An= (ne- no) is the birefringence of the material
`and A is the free space wavelength. As an example, the ne(cid:173)
`matic liquid crystal E7 has a birefringenceof approximately
`0.2 in the visible and near infrared spectrum. It requires a 5
`p,m layer thickness to achieve a relative phase delay of 21r
`radians at a 1 p,m wavelength. If a reflective-mode design is
`used, allowing two passes through the liquid crystal layer,
`a full wave OPD is created using only half that thickness,
`or 2.5 p,m.
`The diffraction efficiency, TJ, of a grating' with a stair(cid:173)
`step blaze designed to maximize energy in the. first order
`is given by [23]:
`
`17
`
`(2)
`
`= (sin(n) q)) 2
`7r I q
`where q is the number of steps in the blaze profile. From
`(2) it can be seen that an eight-step approximation gives a
`theoretical efficiency of approximately 95%. Fig. 2 shows
`a step approximation to the wavefront deflected by a prism,
`including the 21r phase resets. Note that a 21r phase reset has
`that value only for the design wavelength. Fig. 3 shows the
`deviation from a straight· line in the unfolded phase profile
`when a wavelength other than the design wavelength is
`used. This variation in phase reset values causes dispersion
`
`270
`
`PROCEEDINGS OF THE IEEE, VOL. 84, NO. 2, FEBRUARY 1996
`
`

`

`Polarization )
`
`Optical Beam
`
`OFF
`
`~~~
`~
`~~~~~~~~~
`~~~
`~~~
`~~~
`~~~
`~48~
`~48~
`~~~
`~~~
`~48~
`~~~
`~48~
`~~~
`~48~
`~41D~
`
`~~ ~ 48 41D~
`
`= :c
`
`0
`
`.c
`=
`a.
`
`RMS Voltage
`
`5
`
`---------
`,,,,,,,,,
`---------
`,,,,,,,,,
`ON "' V 111111111
`.,.,.,.,,.,,.,.,
`--------(cid:173)
`
`~~~48~~~~
`
`Fig. 1. Nematic liquid crystal phase shifters. The liquid crystal molecules are birefringent. Light
`polarized along the long axis of the molecule will experience a different index of refraction than light
`polarized along the short axis of the crystal. The molecules will rotate when a voltage is applied,
`producing an effective index change for light polarized perpendicular to the long axis of the crystal.
`
`Unfolded Phase
`Profile
`
`Design
`Wavelength
`
`Steered
`Wavelength
`
`Undisturbed
`Phase Front
`
`Individual
`Phase Shifters
`
`Fig. 2. Optical phased array agile beam steering. The optical
`phase delay introduced by a prism in an aperture can be approx(cid:173)
`imated by a series of stair-step ramp phase delays. When a ramp
`has an optical path difference equal to or larger than the design
`wavelength one design wavelength of optical path difference is
`subtracted from the ramp. At the design wavelength, the phased
`array effectively reproduces the steering caused by a prism.
`
`[14], which will be discussed further in Section VI. As
`shown in Fig. 3, the unfolded OPD is in error by (>.->.d)
`after each reset, and the l,lnfolded phase is in error by
`21r ( >. - >.d)/>. after each reset, where >. is the actual
`wavelength and >.d is the design wavelength.
`Practical factors can cause the measured efficiency of
`an actual phased array beam steerer to deviate from the
`theoretical value given by (2). One such factor, evident in
`liquid crystal phased arrays currently being developed, is a
`spatial "ftyback" in the molecular orientation of the liquid
`crystals which results from the minimum spatial extent
`required to change from the orientation for a phase shift
`of 21r to that for a phase shift of zero. The actual ftyback
`transition is a complex function of device design and liquid
`crystal visco-mechanical properties. Fig. 4 depicts phase
`versus position for a simple ftyback model. As a result of
`
`Unfolded
`
`Fig. 3. Unfolded phase profile. This figure shows the influence
`on the unfolded phase profile of operation at a wavelength other
`than the design wavelength.
`
`ftyback, only a portion of the grating imposes the correct
`phase distribution to steer a beam in the design direction.
`That portion of the grating over which ftyback occurs can
`be thought of as steering the beam in a different direction.
`The resulting diffraction efficiency 'T/ into the desired grating
`order can be approximated by [15]
`
`(3)
`
`where Ap is the width of the ftyback region and A is the
`period of the programmed grating. The energy that is not
`directed into the desired grating order is distributed among
`numerous other grating orders, causing a loss in efficiency
`for the primary order. The overall steering efficiency is
`given by the product of (2) and (3). Depending on the
`grating period (which affects both the number of steps in
`the blaze profile as well as the relative size of the ftyback),
`
`MCMANAMON et al.: OPTICAL PHASED ARRAY TECHNOLOGY
`
`271
`
`

`

`operated modulo 21r, such adaptive optic elements would
`be dispersion limited to narrow band applications.
`
`III. BEAM STEERING APPROACHES· USING
`LIMITED MECHANICAL MOTION
`An optical phased array is not the only approach to
`realizing rapid beam steering without the use of con(cid:173)
`ventional mechanical systems. Some of the more viable
`alternate options are briefly reviewed here. All of the
`options discussed here potentially allow the redirection
`of the field-of-view of an optical sensor without the use
`of complex, costly, mechanical mechanisms. Unlike the
`optical phased array, most of these alternate options do
`not eliminate mechanical motion, but instead minimize the
`degree of mechanical motion required. To date, none. of
`these alternate approaches have demonstrated the scope of
`performance characteristics desired for laser radar and most
`other optical sensors.
`One such option is the use of cascaded microlens arrays
`[19], [20] an example of which is shown in Fig. 5. Each
`microlens array consists of a (generatly) close packed,
`periodic array of miniature lenses which can be fabricated
`in either diffractive or refractive forms. Beam steering is
`effected by translating one microlens array with respect
`to the other. The concept can be understood by first
`considering a single microlens pair from a set of aligned
`afocal arrays. A collimated input beam is focused to the
`back focal point of the first microlens, which is also the
`front focal point of the second microlens, resulting in
`an unsteered, collimated output beam. However, if the
`second microlens is offset, then the back focal point of the
`first microlens appears as an off-axis point to the second
`microlens. The point remains in the front focal plane of the
`second microlens, so the second microlens still recollimates
`the light, but the beam is redirected to a nonzero field
`angle. A paraxial ray trace shows that the tangent of this
`field angle is equal to the amount of •offset divided by
`the focal length of the second microlens. Maximum useful
`steering occurs with an offset equal to the radius of a
`microlens. It may be noted that it does not matter if the
`second microlens has a positive or negative focal length
`so long as the condition of overlapping focal planes is
`met. If the individual microlenses of the arrays are aligned,
`periodically spaced, and designed to fill the aperture, the
`output beam replicates the input beam. If the offset is small,
`the steered beam approximates a simple· redirection of the
`input beam.
`However, if the offset is large, significant fractions of the
`input beam are coupled into other grating modes. This can
`be appreciated by noting that phased arrays and microlens
`arrays both approximate blazed gratings [21]. If the periodic
`quadratic phase profiles of two offset • micro lens arrays
`are superimposed, the result is a (generally asymmetric)
`triangular waveform, which approximates a blazed grating.
`If the composite phase profile were a sawtooth, the approx(cid:173)
`imation would be exact. Motion of the lenses alters the
`slope(s) of the phase profile, thereby changing the blaze
`
`Duty Cycle, A
`
`'-y-'
`Fly
`Back
`Region
`AF
`
`Fig. 4. Flyback. When one design wavelength of optical path
`difference is subtracted it requires finite spatial extent. This region
`is referred to as the flyback region. The steering efficiency into a
`given order is influenced by the relative size of the flyback region
`with respect to the grating period.
`
`either (2) or (3) may dominate the overall steering efficiency
`of an optical phased array beam steerer.
`For a normally incident input beam the steered angle is
`given by [16]
`
`(4)
`
`.
`Ao
`()
`sm =A
`where A0 is the design wavelength for the beam steerer,
`A = qd is the period of the staircase ramp, q is the number
`of phase shifters between resets, and d is the center-to(cid:173)
`center spacing between phase shifters, which is assumed
`to equal the width of the phase shifter as well. Large
`steering angles correspond to high spatial frequencies (small
`periods) and vice versa. From (3) and (4) it can be seen
`that the steering efficiency decreases monotonically with
`steering angle, for fixed flyback.
`Two-dimensional beam steering can be achieved using
`two orthogonally oriented 1-D liquid crystal phase gratings.
`In addition, any optical distortion that is separable in
`Cartesian coordinates can be fully compensated, modulo
`27r. Spherical aberrations can be fully compensated with a
`crossed grating system. For a full adaptive optics capability,
`a third layer, with a 2-D array of phase shifters, would be
`required. This would add the ability to clean up an arbitrar(cid:173)
`ily aberrated beam and adapt for atmospheric turbulence.
`Such a liquid crystal adaptive optics layer has recently
`been discussed [17]. The spacing of elements on such an
`adaptive optics layer would be orders of magnitude courser
`than the spacing required for large angle beam steering.
`Current adaptive optics mirror systems have on the order
`of 50-400 elements correcting for turbulence while using
`apertures up to a few meters [18]. However, the adaptive
`optic element is usually used prior to final beam expansion
`and is much smaller in aperture. Pixelated phase shifters
`of about 1 mm square would probably suffice for most
`applications and could be readily fabricated with the current
`technology. Thus liquid crystal phase shifter technology
`could replace the current piezoelectrically driven adaptive
`optic components, resulting in a single three-layer compo(cid:173)
`nent that both deflects and phase compensates a beam. If
`
`272
`
`PROCEEDINGS OF THE IEEE, VOL. 84, NO. 2, FEBRUARY 1996
`
`

`

`profile of the equivalent grating, and shifting the light to
`different grating orders. To the extent that the composite
`phase profile approximates a true sawtooth, light is steered
`to a single direction. However, the offset of two lens arrays
`inherently causes each input lenslet to illuminate adjacent
`output lenslets, resulting in the multiple-slope profile of
`the triangular wave, and steering to multiple directions. To
`mitigate this effect, designs using a third microlens array
`as a field lens have been put forth, but demonstrations have
`not yet been reported.
`The agile steering of a beam using the microlens array
`concept requires the agile motion of one microlens array
`with respect to the other. Microlens arrays inherently have
`small focal lengths (typically on the order of a few lenslet
`diameters, usually a millimeter or less); consequently, the
`amount of mechanical offset required to achieve a desired
`steering angle can also be quite small. Compared to steering
`via a displaced bulk lens having the same aperture as
`the microlens array, the reduction of motion required to
`steer to a given angle is proportional to the ratio of the
`individual microlens diameter to the array diameter. Due
`to its essentially planar structure, a microlens array can be
`made much lighter than a bulk lens of equivalent aperture.
`The combination of low mass and small motion allows agile
`positioning (and agile beam steering) to be accomplished
`with more simplified mechanical drivers than would be
`required for macroscopic lenses. The microlens arrays can
`be designed to effect substantial steering with mechanical
`motions that can be achieved with piezoelectric transducers.
`However, small errors in mechanical positioning are ampli(cid:173)
`fied by the same optical leverage that makes possible the
`reduction in mechanical motion. This means the amount of
`energy at the desired steering angle will be influenced by
`a small amount of mechanical motion. Thus fine angular
`beam steering with this approach generally requires very
`precise motion.
`Microlens arrays can be programmed onto the liquid(cid:173)
`crystal based optical phased arrays reported here, thereby
`making possible an electronic translation of one lens array
`with respect to the other, and complete elimination of
`all mechanical motion. Since only one microlens array
`must move to achieve beam steering, only a single array
`would have to be programmable. Owing to the precise
`displacement control available with an optical phased array,
`this option may be preferable to piezoelectrically driven
`motion for applications requiring precision pointing.
`Flexure beam micromirror technology is another ap(cid:173)
`proach with large numbers of small apertures arranged in
`regular arrays [22]. The individual apertures are lithographi(cid:173)
`cally fabricated mirror "pixels" on hinges with micromotion
`effected by an electrostatic field. The field attracts the ele(cid:173)
`ment and moves it rapidly, on the order of a microsecond.
`This can create a piston phase shift for the individual
`aperture. These devices have demonstrated 211' phase shifts
`at 633 nm wavelength with a 60-75% fill factor. Much of
`the same phased array theory discussed later in this paper
`applies to these array structures, although the physical im(cid:173)
`plementation is significantly different. The implementation
`
`Fig. 5. Decentered micro/ens array beam steering. The figure
`shows two decentered microlens arrays, and their influence in
`steering an incoming beam. If one array is moved with respect
`to the other array it causes the beam to steer.
`
`of this approach makes individual apertures with sizes on
`the order of visible or near infrared wavelengths imprac(cid:173)
`tical, thus limiting maximum steering angles. A second
`limitation is the nonunity fill factor. Hidden hinge concepts
`are being considered to address the fill factor issue [23].
`Different approaches are represented by electro-optical
`(EO) and acoustic-optical (AO) beam deflection [24]. There
`are no moving parts with either approach. EO deflec(cid:173)
`tion can occur in nanoseconds, while AO beam deflection
`is generally effected on the order of microseconds, the
`time for an acoustic wave to propagate across the crystal
`aperture. AO beam steering requires high drive power
`at longer wavelengths, has an aperture size limited by
`available crystal dimensions, and shifts the frequency of the
`transmitted, or receive, beam. EO beam deflectors require
`high voltages for large apertures, typically of the order
`of 10 kV/cm. Beam deflection angles for both EO and
`AO deflectors are typically limited by practical issues to a
`few milliradians; consequently, these devices are generally
`restricted in applications to fast, fine beam steering of small
`b