MAcro-Electro-Mechanical Systems (MÆMS) based concept for microwave beam steering in reflectarray antennas

In recent years, there has been a growing interest in employing phased-array antennas in various applications ranging from satellite and airborne communications to radars and imaging systems. While a number of phased-array systems have been deployed, their extreme cost and complexity have limited their application only to the most expensive pieces of military hardware. Therefore, many systems that can potentially benefit from the capabilities offered by phased arrays are left behind.¹ Thus, new ideas for developing affordable phased arrays are needed to enable the widespread use of this technology. Beam steerable reflectarray antennas have been considered as a promising approach for realizing affordable phased arrays. Tuning approaches in the existing beam steering reflectarrays can be categorized into either feed switching techniques or element tuning techniques.² In the feed switching techniques, the spatial delay profile over the aperture can be tuned by displacement of the feed.³ Although this technique is simple to implement, it does not provide a continuous beam scanning and requires multiple feed antennas to illuminate the aperture. Using multiple feed antennas can reduce the aperture efficiency of the system, due to aperture blockage, as the number of feeds increases. In the element tuning technique, on the other hand, a tunable phase shifting mechanism needs to be incorporated into each spatial phase shifting element on the aperture of the reflectarray to achieve beam steering. Various designs of element tuning techniques have been reported in the literature over the years. Examples of these include using mechanically actuated patch antennas⁴ or integrating varactors,⁵ micro-electromechanical systems (MEMS) switches,⁶ PIN diodes,⁷ and functional materials^8–14 with the constituting elements of the aperture. The cost and complexity of the fabrication, reliability issues, and relatively low scanning rates are the primary challenges of the approach reported in Ref. 4. Varactors and MEMS switches, on the other hand, offer faster scanning speed and have lower power consumption compared to the previous technique. However, reflectarray antennas that use such electronically tunable elements to achieve tunability suffer from several major problems. First and foremost, these structures suffer from limited power handling capability due to the nonlinearity of these electronically tunable devices. In Ref. 15, it was demonstrated that the phase shift provided by a tunable spatial phase shifter used in a typical reflect- or transmit-array employing BST or GaAs varactors would drastically change as the incident RF power level of it was increased from low to moderate power levels. This is a major factor that limits the use of electronically reconfigurable reflectarrays to relatively low power applications. Moreover, a moderate-size electronically tunable reflectarray may have hundreds to thousands of pixels that need to be tuned individually. Varactors need to be integrated with each pixel and must be appropriately biased. Often times, it is required to have RF/dc isolation mechanisms (e.g., dc block capacitors) integrated with the unit cell as well. These factors increase the cost and complexity of the design and control of the reflectarray. Additionally, the ohmic losses of the electronically tunable elements as well as the bias line losses also deteriorate the radiation efficiency of the reflectarray and generate unwanted heat that needs to be dissipated, complicating the thermal managements of the aperture in high power applications. These issues are significantly exacerbated as we move from microwave frequencies to millimeter-wave (MMW) and sub-MMW frequencies. Functional materials including liquid crystals,^8–10 ferroelectric dielectrics,¹¹ photonically controlled materials,¹² and graphene^13,14 have also been used to design beam steerable reflectarrays. While these technologies have certain advantages that make them interesting for particular applications, they are not as widely studied as other techniques, and most of the demonstrations of these techniques have been done at the unit cell level as opposed to the full size apertures. Therefore, more research and development is required to determine if these techniques are suitable to design large-scale reflectarray antenna apertures with tunable responses.

Over the past few years, innovative mechanical reconfiguration techniques have been used to overcome the aforementioned challenges of using electronic tuning and to achieve tuning in reflective and transmissive type apertures.^16–24 In Ref. 16, a reconfigurable beam steering reflector based on mechanical motion of a movable board with respect to a fixed board was presented. In the proposed structure, a mushroom-like high impedance surface is topped with rotating periodic arrangement of capacitive patches. As the top layer moves, the phase gradient over the aperture changes which consequently steers the reflecting beam. In-plane displacement of layers was also used to achieve tunability in the response of periodic arrangement of coupled split-ring resonators in Refs. 17 and 18. This tunability of the response could be used to change the phase gradient over the aperture and achieve beam steering. In Ref. 19, a tunable high impedance surface was achieved by changing the displacement between two layers. Liquid tuning of the responses of transmissive spatial phase shifters is examined in Refs. 15 and 20. In Refs. 21, 22, and 23, stretching or folding of the layer was used to achieve reconfigurability in periodic structures. Finally, in Ref. 24, several techniques for designing large-scale tunable periodic structures are presented. One of these techniques is used to design a mechanically tunable reflecting type spatial phase shifter. It was shown that tunability of the response can be achieved through flexure of the surface. Such tunable spatial phase shifter can be used to design beam steering reflectarrays.

In this paper, we propose a new mechanical beam steering technique that exploits macro-scale mechanical movements over parts of a reflecting-type aperture to achieve beam steering.^25,26 Since we envision that these mechanical movements will ideally be performed using electro-mechanical actuators, we refer to this technique as MAcro-Electro-Mechanical Systems (MÆMS) based beam steering. Beam steering using this concept does not require the use of any solid state phase shifters, varactors, or switches. Consequently, the proposed technique is expected to address many of the shortcomings of electronically tunable reflectarray antennas discussed earlier in this section and is expected to enable the development of affordable, high-power capable phased-array antennas. Moreover, the spatial phase shifters constituting the reflectarrays do not need to be individually controlled. Rather, they are controlled collectively, over a macro scale, which significantly reduces the complexity of two dimensional (2D) beam steering in a reflectarray antenna. We present a design example where a MÆMS-based tuning technique is applied to an aperture composed of reflecting type spatial phase shifters. This spatial phase shifter is the unit cell of a non-resonant sub-wavelength periodic structure similar to the one reported in Ref. 3. The elements of such sub-wavelength periodic structures have been previously used in the design of reflectarrays,³ spatial filters,^27–30 transmitarrays,^31–33 and polarization converters.³⁴ Subsequently, the proposed tunable spatial phase shifter is used in the design of a beam steering reflective type aperture where only small tilting of the ground plane underneath the entire aperture is used to change the direction of the radiated beam in two dimensions. Finally, we discuss the fabrication and experimental characterization of a prototype of this MÆMS-based reflective aperture.

Fig. 1(a) shows a reflecting-type spatial phase shifter of the type used in Ref. 3. The structure is composed of an array of sub-wavelength capacitive patches separated from a ground plane with a thin dielectric substrate. To an incident electromagnetic wave, this surface acts as a resonator with the equivalent circuit model shown in Fig. 1(b). In this configuration, the patch array is modeled with a capacitor, and the small separation between the patch and the ground plane is modeled with two short transmission line sections representing the dielectric substrate and the separation between the dielectric substrate and the ground plane. Assuming that t, h ≪ λ, this short circuited transmission line has an inductive input impedance. Therefore, the structure shown in Fig. 1 can be modeled with a parallel LC resonator with the resonant frequency of $ω0=1LC$⁠. Far from the resonance frequency, this structure acts as a perfect electric conductor (PEC). Therefore, it reflects the wave with a ± π phase shift. At resonance, the phase of the reflection coefficient is zero and the structure acts as a perfect magnetic conductor (PMC). Assuming that the structure is lossless, the magnitude of the reflection coefficient is always equal to 1. The capacitance and inductance values of the equivalent circuit model shown in Fig. 1 can be calculated using the following formulas:

where D is the dimension of the unit cell, s is the gap spacing between two adjacent patches, t is the thickness of the substrate on which the patches are etched, ε_r is the dielectric constant of the substrate, and h is the spacing between the ground plane and the bottom of the substrate. Also, ε_r,eff is the effective dielectric constant for the capacitive patch array, ω is the operating frequency, and Z₀ = 377 Ω. For the structure under discussion in this section, $εr,eff≈1+εr2$⁠. To change the reflection phase at a certain fixed operating frequency, the resonant frequency of the effective LC resonator shown in Fig. 1 needs to be tuned. This can be done by tuning the inductance or the capacitance values of this structure. In the vast majority of periodic structures of this type, the literature, the response tunability, is achieved by tuning the capacitance values. In doing so, various techniques including loading of varactors and MEMS switches^5,35–38 and fluidic tuning³⁹ were employed. Unlike most previous studies, however, we propose to tune the response of this spatial phase shifter by changing its effective inductance as opposed to changing the capacitance values. We first reported this tuning technique in Refs. 25 and 26. To change the inductance value, the separation between the ground plane and the capacitive patch layer can be changed as shown in Fig. 1(c). Therefore, the tuning parameter that controls the value of the inductance in this case is h. The inductance value continuously changes as the separation between the capacitive patches and the ground plane changes. This, however, does not significantly impact the value of the capacitance of the structure, since most of the fringing effects are negligible and most of the fields in the capacitive layer are confined in the gap region between adjacent capacitive patches. Thus, a continuous tuning can be performed without using any solid-state devices. Moreover, since the separation between the ground plane and the patch is sub-wavelength, only small movements of the ground plane are needed to change of the phase shift over a wide range.

To demonstrate the capabilities of the proposed tuning technique, we examined a high-impedance surface (HIS) designed to operate at 9.5 GHz. The design parameters for this structure are the dielectric constant and thickness of the substrates, the unit cell size, the gap spacing between patches, and the maximum variation of the spacing between ground plane and the bottom of the patch. The choice of the unit cell size is arbitrary as long as it is small compared to the wavelength. The needed resonant frequency scanning to achieve a fixed phase shift is smaller for a resonator with higher quality factor. However, increasing the quality factor reduces the bandwidth of the structure. In our initial experiments, we focused on narrow-band operation around a single frequency (9.5 GHz) to demonstrate the fundamentals of beam steering using this concept. Therefore, a resonator with a moderate quality factor was chosen. In wideband applications, resonators with a higher-order response (e.g., using multiple capacitive patch layers) may be used in conjunction with similar tuning techniques. For a first-order parallel LC resonator, the quality factor is proportional to $C/L$⁠. Thus, to tune the response with minimized mechanical movements, a higher capacitance value and a smaller inductance value are needed. To achieve a higher capacitance, the gap spacing between the adjacent capacitive patches should be decreased for a fixed unit cell size, D, and relative permittivity of the substrate, ε_r. Smaller inductance values can be achieved by using a thinner substrate. Then, considering all these values including the substrate characteristics and the gap spacing between patches is known a priori; the required variation range for the distance between the ground plane and the capacitive patch array can be calculated. The unit cell size in this example is D = 6.5 mm or equivalently ≈0.2λ₀, where λ₀ is the free-space wavelength at the center frequency of operation. The dielectric constant and thickness of the substrate used in this design are ε_r = 3.4 (Rogers RO4003C) and 0.5 mm, respectively. The minimum gap spacing is limited by the fabrication technology. In this example, a relatively small gap spacing of s = 0.45 mm, which can be reliably fabricated, is considered. Knowing all these parameters, the range of the ground plane movements needed to achieve the desired phase shift range can be obtained. The capacitance value can be calculated using (1). Using this value and knowing the center frequency of operation, the inductance value can be determined. Then, h can be calculated using (2). For the center frequency of 9.5 GHz, h is calculated to be 0.75 mm. Therefore, the range of variations for the spacing between ground plane and the substrate is considered to be 0–1.5 mm. Fig. 2(a) shows the reflection phase of the structure shown in Fig. 1 as a function of frequency for different values of h. As can be observed, for a fixed operating frequency (9.5 GHz), the reflection phase changes as the value of h is changed. Also, the reflection phase at 9.5 GHz as a function of the tuning parameter, h, is shown in Fig. 2(b). Observe that small movements of the ground plane with a maximum variation of 1.5 mm allow for changing the phase shift over a wide range of ≈270°. As discussed, to achieve a wider phase shift range using the same mechanical movements, a larger capacitance value in the equivalent circuit model is needed. The higher capacitance can be achieved by either reducing the gap spacings between the patches or using a dielectric substrate with a higher dielectric constant.

The unit cell of the spatial phase shifter discussed in Section II can be used in the design of MÆMS-based beam-steerable reflectarrays. Fig. 3 shows the basics of beam steering in a MÆMS-based reflector. The structure is composed of a planar uniform reflector surface illuminated by a feed antenna. The reflector's aperture is populated by identical spatial phase shifters that locally manipulate the phase of the reflected wave. In this structure, the ground plane backing the reflectarray can be tilted freely by small distances compared to the wavelength. In doing so, the other parts of the structure including the capacitive patch layer and its supporting dielectric substrate remain fixed and do not move. The tilting of the ground plane locally changes the resonant frequencies of the spatial phase shifter (SPS) located at different points on the aperture. At a fixed operating frequency, this change in the resonant frequency of the SPSs changes the phase shift that each SPS provides. Therefore, the SPSs occupying different locations on the aperture can provide different phase shifts simply by tilting the ground plane underneath the structure while maintaining the remaining parts of the structure fixed. In the simplest form, if all SPSs are composed of capacitive patches with identical physical dimensions, the tilting of the ground-plane creates a phase shift gradient over the aperture of the reflectarray. This phase shift gradient determines the direction of maximum radiation (or the direction of main lobe of the radiation pattern) of the antenna in the far field. For an aperture with a linear phase gradient of $∂ϕ∂r$⁠, where $ϕ$ is the phase of the reflection coefficient from the aperture, a normally incident wave will be reflected towards the angle of $θ=±arcsin(λ2π∂ϕ∂r)$⁠. By dynamically changing this phase shift gradient $(∂ϕ∂r)$ over the aperture of the reflector, the direction of maximum radiation can be changed dynamically and the scattered beam can be steered. The beam scanning range in this structure is a function of several factors. In a narrow-band application as is the case studied here, the range increases as the reflector's aperture dimension decreases or the maximum phase shift provided by the SPSs increases. For a single, first-order resonant spatial phase shifter of the type examined in this work, the maximum variation of the phase is less than 2π. Therefore, the phase gradient in the extreme case can be approximated as $∂ϕ∂r≈2πD$⁠, where D is the aperture dimension. For this case, the scanning range is limited in the range of $±arcsin(λD)$⁠.

The spatial phase shifters discussed in Section II are used to design a flat, non-focusing beam-steerable reflector. The surface is designed to operate at 9.5 GHz and has aperture dimensions of approximately 5.6λ₀ × 5.6λ₀. The reflecting surface is illuminated with a horn antenna placed at a distance of ≈7λ₀ away from it. The structure is fabricated on a 0.5-mm thick RO4003C substrate (from Rogers Corp.) with a dielectric constant of 3.4. In this embodiment, all of the capacitive patches are identical. Therefore, unlike a reflectarray antenna, this reflecting surface does not perform any collimation of the beam. Nevertheless, the beam steering concepts can be demonstrated equally well without having beam collimation. The ground plane is located underneath the substrate separated by 0.75 mm from the bottom of the substrate. Assuming that the ground plane is fixed at its center, it can be tilted up or down along this pivot point by a maximum distance of 1.5 mm. This tilting allows for achieving a phase shift gradient over the surface of the reflector along any desired direction. This way, two dimensional continuous beam scanning can be performed. Fig. 4(a) shows the phase profile over the aperture for three different ground plane tilts. For the first case (State A), the ground plane is fixed and not tilted. For the second and third cases (State B and State C), the ground plane is tilted by the maximum distance along either edge of the aperture to steer the beam in the x – z plane along $+x̂$ or $−x̂$⁠. Fig. 4(b) shows the simulated co-polarized normalized radiation patterns of this structure for these three cases. Observe that by tilting of the ground plane up or down by a maximum distance of 1.5 mm, the direction of the reflected beam can be steered in the range of ±10°.

The beam scanning technique described in the previous paragraph relies on creating a phase shift gradient over the aperture by locally manipulating the resonant frequencies of its constituting spatial phase shifters. Because of the resonant nature of the structure, small shifts in the ground plane distance can cause significant changes in the resonant frequency of its spatial phase shifters. Consequently, large phase shift gradients can be achieved using this approach that cause a larger beam scanning field of view. In comparison, if one were to rotate the whole reflectarray structure (excluding its feed) and tilt it with the same maximum distance of 1.5 mm (or equivalently a rotation angle of α ≈ 1°), the beam steering would be only in the range of ±2°. On the other hand, if the reflector surface is maintained and the feed is rotated by an angle of ±1°, the maximum beam scanning range that can be achieved is limited to ±1°. Both of these scanning ranges are much smaller than the scanning range provided by tilting only the ground plane. This is due to the fact that neither of these two alternative techniques impact the resonant frequency of the SPSs occupying the aperture of the reflectarray. This ±10° scanning range is in agreement with the predicted result in which $θ=2 arcsin(λ2π∂ϕ∂r)$⁠, where $∂ϕ∂r≈2π5.6λ$⁠. The transition within this range is not abrupt, and the beam steers continuously as the ground plane tilts along the pivot point between state A and states B and C. The side lobe levels of the far field scattered pattern are generally above the expected values. This is mainly attributed to the aperture blockage caused by the feed antenna. The feed horn antenna used in the experiments has aperture dimensions of 3.21λ₀ × 2.76λ₀, which directly blocks the center part of the reflector's aperture. The issue of feed blockage can be resolved relatively easily by using offset feed to illuminate the reflector's aperture. Nonlinearity of the reflection phase as a function of displacement also contributes to enhance the level of side lobes.

An experimental prototype of the beam-steerable flat reflector surface analyzed in Section III was fabricated using standard printed circuit board (PCB) lithography. The prototype panel dimensions are 175.5 mm × 175.5 mm. Fig. 5 shows the photograph of the fabricated prototype. The feed antenna is a commercial X-band horn antenna located on the optical axis of the structure at a distance of ≈22 cm from its center. For initial characterization, dielectric spacers were used to fix the separation between the ground plane and the substrate panel, and the measurements were performed for a static prototype. To dynamically tilt the ground plane, these dielectric spacers can be replaced by spacers whose lengths can be electrically controlled (e.g., using piezoelectric actuators). Fig. 6 shows an illustration of a potential implementation of the structure with two-dimensional dynamic beam steering. In this structure, four piezoelectric actuators are placed at four corners of the reflecting surface and are employed to control the tilt angle of the ground plane. The displacement of each actuator from its idle position depends on the voltage applied to its two bias lines. When a positive voltage is applied, the actuator will displace in the positive z-axis, and when a negative voltage is applied, the actuators will displace the same distance in position direction from its idle position. By controlling the voltage applied to each actuator, the desired tilt vector of the ground plane can be achieved. In this arrangement, two dimensional beam steering can be performed by simply controlling the heights of the piezoelectric actuators, which determines the tilt vector of the ground plane. Based on the maximum displacement provided by actuators, bending actuators appear to be suitable for microwave frequencies (e.g., X-band), while stacked actuators are better options to be used at higher frequencies (e.g., MMW). Commercially available bending actuators (e.g., Ref. 40) provide maximum displacement of 2–3 mm with speeds up to few KHz. These characteristics make them a good candidate to be used in this design example.

The far-field scattering patterns of this structure were measured using a multi-probe spherical near-field system. For our static prototype, Fig. 4(b) shows the co-polarized normalized radiated fields from this structure at the center frequency of operation. In general, a good agreement is observed between the simulation and measurement results. Observe that the main beam is steered towards the expected direction by tilting the ground plane.

When all the parameters including the aperture dimensions, distance between feed and the aperture, and bandwidth are fixed, the scan range is only a function of the phase variation over the aperture. For a single resonant structure, the maximum variation of the phase is less than 2π. This limits the scanning range of the structure. However, the scan range of the array can still be improved if minor modifications are made to the architecture of the array shown in Fig. 3. An effective method for enhancing the phase shift range over the aperture is to use multiple independently controlled tiltable ground planes instead of the single tiltable ground plane. This way, a saw-tooth-shaped phase function over the aperture can be created. This broadens the achievable phase shift gradient over the aperture for given f and D parameters. However, since 2π discontinuities created by the adjacent ground planes are required to achieve a monotonic phase function, the beam scanning is abrupt and the resulting structure can be used as a beam switching reflector. To demonstrate the capabilities offered by this new feature, the same flat reflector surface discussed in Section III is examined again but this time with three independently controlled tiltable ground planes as shown in Fig. 7. Each ground plane can now be tilted freely up or down by a maximum distance of 1.5 mm. Fig. 8(a) shows the phase profile over the aperture for two states. For “State A,” all three ground planes are fixed and not tilted. For “State B,” all the ground planes are tilted by the maximum values to achieve beam-steering in x – z plane along $+x̂$⁠. Since the structure is symmetric, the state corresponding to “State C” of the structure with one ground plane scanning along in $−x̂$ is not shown for brevity. The simulated and measured co-polarized normalized radiation patterns of the reflector for both cases are shown in Fig. 8(b). As can be observed, using multiple independently controlled ground planes successfully steers the beam to larger angles by increasing the range of phase gradient over the aperture. For the case of three ground planes tilted, the maximum reflection occurs at ±30° when the aperture is illuminated under normal incidence. Other than nonlinearity of the reflection phase response as a function of displacement, the relatively high level of the side lobes here (compared to the case of single ground plane) is mostly due to the imperfect 2π phase discontinuities created by each two adjacent ground planes.

We investigated a new approach for designing passive phased arrays based on electro-mechanical beam steering. We discussed a specific design where MÆMS tuning techniques were applied to a planar, high-impedance surface (HIS) to achieve a beam-steerable flat reflector. We demonstrated that beam-scanning in this reflector can be achieved without the need for integrating individual electronic tuning elements with each unit cell of the structure. Rather, small, macro-scale mechanical movements of the ground plane of the high-impedance surface were exploited to achieve the same beam steering that would have been provided by integrating individual electronic tuning elements (e.g., varactors) with each unit cell of the HIS. If the proof-of-concept prototype structure demonstrated in this work was to be made tunable using varactors, 1458 varactors (the structure has 27 unit cells in each direction, and each unit cell needs two varactors to ensure dual-polarization operation; this results in a total of 1458 varactors) would have needed to be integrated with the unit cells of the structure and appropriately biased to achieve the same phase tuning and beam scanning range. In sharp contrast with this alternative electronic tuning technique, the proposed technique achieves two-dimensional beam steering by controlling significantly fewer variables, namely, the tilt vector(s) of the ground plane (segmented ground planes). Moreover, unlike other mechanical beam steering techniques where the entire reflecting surface of a reflector (or a reflectarray) antenna and its feed are rotated to scan the beam, in the proposed technique, most of the structure (including the feed horn and the capacitive patch array constituting the high-impedance surface) remains stationary and only the ground plane is moved by small distances.

This proposed concept is expected to make the task of designing large-scale tunable reflectarray and transmitarrays considerably simpler and more practical. Compared to their electronically tunable counterparts, MÆMS-based reflect- and transmit-arrays have several unique advantages. These include the capability to handle significantly higher power levels, reduced design and control complexity, reduction of losses associated with electronic tuning elements, and ease of thermal management. These attributes are expected to make this technology a promising candidate for development of affordable phased-array antennas at microwave, millimeter-wave, and THz frequency bands. One area where electronically tunable structures have an advantage over MÆMS-based structures is the tuning speed. However, because the mechanical movements involved in a MÆMS-based phased array are very small and the parts of the structure that need to be moved are light weight, mechanical movements are expected to be performed quite rapidly using commercially available electro-mechanical actuators (e.g., piezoelectric actuators). While further research and development is needed before, all practical issues involved in implementing phased-array antennas based on this technology are addressed; we expect MÆMS-based phased arrays to be capable of providing beam scanning speeds of at least several tens to several hundreds of Hertz. (This estimate is based on existing commercially available electro-mechanical actuators.) Finally, we like to emphasize that the proof-of-concept structure demonstrated in this work was meant to demonstrate the feasibility of using the proposed concepts to perform passive beam steering. Several practical engineering issues regarding the implementation, actuation, and control of such structures need to be addressed before MÆMS-based phased arrays can be commercialized. Many of these issues, however, are addressed in the field of adaptive optics where optical devices (e.g., mirrors) are reconfigured using electro-mechanical actuation techniques similar to the ones that would be needed to be used by a MÆMS-based phased array antenna.

This material is based upon work supported by the Office of Naval Research under ONR Award No. N00014-16-1-2308 and by the National Science Foundation under NSF Award No. ECCS-1101146.