A 220–260-GHz Silicon-Micromachined Waveguide MEMS Crossover Switch

This article presents a novel subterahertz (sub-THz) crossover waveguide switch concept operating in the 220–260-GHz frequency band. The crossover switching circuit is implemented by two hybrid couplers and two single-pole-single-throw (SPST) switching mechanisms, utilizing microelectromechanically reconfigurable switching surfaces. The silicon-micromachined crossover switch prototype is very compact, with a total footprint of $5.6\times 5\times1.2$ mm, including four standard WR-3.4 waveguide ports and the waveguide routing to these ports. The measured insertion loss (IL) is 0.9–1.4 dB in the crossover state and 0.8–1.3 dB in the straight state from 220 to 260 GHz, and the isolation (ISO) is better than 29.3 and 29 dB, respectively, for these states. The measured return loss (RL) is better than 14 dB in the crossover state and better than 13.6 dB in the straight state. Besides, the measured input-to-input ISO is better than 13.7 and 34 dB in the crossover and straight states, respectively. The measurement results are in excellent agreement with the simulation data. Moreover, the signal paths are fully symmetric for all input-to-output signal paths, making the crossover switching circuit suitable for redundancy applications.


I. INTRODUCTION
T HE subterahertz (sub-THz) frequency bands have recently attracted considerable scientific and industrial attention.The short wavelength above 100 GHz leads to compact and low-weight devices, even for complex waveguide and antenna systems.Besides, the large available bandwidth at the sub-THz range makes this frequency spectrum well-suited for different applications, such as short-range car radars [1], [2], [3], [4], high-resolution imaging systems for biomedical imaging [5], [6], [7] and security screening [8], [9], high The authors are with the Micro and Nano Systems Division, School of Electrical Engineering and Computer Science, KTH Royal Institute of Technology, 100 44 Stockholm, Sweden (e-mail: arminka@kth.se;umers@kth.se;suxian@kth.se;joachimo@kth.se).
Color versions of one or more figures in this article are available at https://doi.org/10.1109/TMTT.2024.3373870.
As of today, computer-numerical-control (CNC) milling is still the dominant fabrication method for sub-THz systems, despite its disadvantages of fabricating bulky and heavy devices and its nonsuitability for volume manufacturing.In contrast to that, silicon micromachining, using deep-reactive-ion-etching (DRIE) and subsequent sidewall metallization, offers several improvements for fabricating devices and systems operating at the sub-THz frequency bands [15], [16], [17].This technique offers precise fabrication of micrometer-size features and complex 3-D structures.This precision, combined with volume-manufacturing capability and high product uniformity, enables the exploitation of the sub-THz frequency spectrum on a larger scale for various applications.
Rectangular waveguides are preferred for devices operating at the sub-THz frequencies, compared to planar transmission lines, due to their low insertion loss (IL).The surface roughness is the most critical parameter affecting the IL of a rectangular waveguide at these frequencies [18].Silicon micromachining achieves a surface roughness of a few nanometers, superior to any other fabrication techniques, leading to the lowest IL waveguides reported in the literature.For instance, WR-3.4 and WR-1.5 silicon micromachined rectangular waveguides were characterized with an IL of 0.02-0.07and 0.05-0.07dB/mm, respectively [15], [18].Many siliconmicromachined components, such as single-pole-single-throw (SPST) switches [19], [20], phase shifters [21], antenna arrays [22], [23], transitions [24], and filters [25] operating at the sub-THz frequencies, are successfully demonstrated in the literature with outstanding performance.
Many applications require reconfigurable circuits, for which switches are critical for adding this reconfigurability to the systems, for instance, for frequency band selection in filter banks or radiation pattern reconfiguration in antenna arrays and radar sensors [23], [26], [27].Conventional waveguide switches, though achieving excellent RF performance, are based on macroscale mechanics and comprise a waveguide section to be moved or rotated by a stepper motor [28], [29] and thus are bulky, heavy, slow, and have high power consumption.Besides, circuits based on such switching elements cannot achieve high integration density.For instance, although the mechanical single-pole-double-throw (SPDT) switch recently reported in [28] has excellent isolation (ISO) and IL performance, it is unsuitable for applications with volume manufacturing requirements, and its switching speed, size, and power consumption do not fulfill the requirements in many applications.
In contrast to macromechanical waveguide switches, microelectromechanical-based (MEMS-based) switches integrated in silicon-micromachined waveguides have the potential to outperform the existing switching technologies since they exhibit low IL, high ISO, high linearity, large operation bandwidth, low power consumption, switching speed in the microsecond range, and high degree of miniaturization.The SPST MEMS switch operating in the 500-750 GHz [19] was the first miniaturized rectangular waveguide-based switch operating at the sub-THz frequencies.Karimi et al. [20] have recently reported on an improved SPST switch concept operating in the 220-290-GHz frequency band with ISO, IL, and return loss (RL) of better than 30, 1, and 17 dB, respectively.
This article presents a novel crossover switch concept, which is a generalized symmetric switching circuit with four ports, i.e., two input and two output ports.The crossover switching mechanism is implemented by two hybrid couplers and two SPST switches controlled by MEMS comb-drive actuators.Due to its symmetric nature, leading to entirely balanced input-to-output signal paths for the different input ports, the proposed crossover switch concept can be used for redundancy switching circuits, and due to its high ISO and low IL, it can be used for receiver calibration.The concept is validated by fabricating and characterizing a crossover switch prototype operating in the 220-260-GHz frequency band, in which the measured results are in excellent agreement with the simulated data.

II. CONCEPT AND DESIGN
Fig. 1 shows the operation principle of the proposed crossover switch, which is a generalized switching circuit with two input and two output ports in different states.In the straight state (referring to Fig. 1), input ports #1 and #2 are connected directly to output ports #1 and #2, and in the crossover state, the signal paths are crossed, connecting the input ports #1 and #2 to the output ports #2 and #1.As shown in the schematic block diagram of Fig. 2(a), the proposed crossover switch concept is implemented by two hybrid couplers and two SPST switches.The signal fed to input #1 propagates through the first hybrid coupler, whose outputs are connected to two SPST switches that are operating in the same state, i.e., either both are in propagating or reflecting state.In the crossover state, the SPST switches are in the propagating state, and the −3-dB signals propagate through both SPST switches and combine in the second hybrid coupler at output #2 since the 90 • phase difference between the output ports of the first coupler is compensated by the same phase difference in the second coupler.In the straight state, the SPST switches are in the reflecting state and block the signal paths between the two couplers.Thus, the reflected signals from the SPST switches are then recombined at output #1.Besides, as can be seen from the block diagram, the proposed crossover switching concept is fully symmetric regarding the input-tooutput signal paths, i.e., the input #1 to output #1 or to output #2 paths are symmetric to the input #2 to output #2 or to output #1 paths, respectively.Overview of the MEMS-RSs in the propagating and reflecting states with detailed dimensions (d: distance between the switching cantilevers in the propagating state, g: gap between the switching cantilevers in the reflecting state, t: thickness of the switching cantilevers, and l ov : vertical overlap between the switching cantilevers).
1) In the propagating state (referring top Fig. 3), the RL of the SPST switches is minimized by two sequential MEMS-RSs in the waveguide path, which are designed for destructive interference for the reflection signals in the frequency band of interest [20].Any reflected power from the MEMS-RSs in the propagating state is problematic for the crossover switching circuit as the reflected power recombines in the unwanted output port (the isolated port), thus degrading the ISO.2) In the reflecting state (referring top Fig. 3), the two sequential MEMS-RSs increase the ISO of the SPST switch without impacting the IL [20].Thus, utilizing two sequential MEMS-RSs improves the ISO of the switch as any leaked power combines in the unwanted output port (the isolated port).
Fig. 3 shows the MEMS-RSs in the propagating and reflecting states.Each MEMS-RS consists of two sets of switching cantilevers.The first set is fixed, and the second set is suspended and controlled by a comb-drive MEMS actuator.In the propagating state, there is a 20-µm distance between the fixed and suspended cantilevers, allowing the wave to propagate through the MEMS-RSs with minimal interference, as the cantilevers are significantly shorter compared to the wavelength.In the reflecting state, the cantilevers are in contact, shortcircuiting the E-field of the rectangular waveguide's dominant mode (TE 10 ) and thus effectively reflecting the signal.According to a comprehensive study performed in [19], a large number of E-plane cantilevers increase the reflecting-state ISO but negatively influence the propagating-state RL and IL.Furthermore, a large number of H -plane cantilevers lead to better propagating-state RL and IL and deteriorate the reflecting-state ISO.Considering this tradeoff, the MEMS-RSs are designed with five E-plane and ten H -plane switching cantilever rows, resulting in 60 contact points in the reflecting state [20].
The crossover switch prototype is designed to be implemented in four vertically stacked silicon-on-insulator (SOI) chips composed of a handle, device, and buried-oxide layer thickness of 275, 30, and 3 µm, respectively.The in-plane waveguides, referring to the chip surface, are etched in the handle layers; thus, they are slightly shallower compared to the standard WR-3.4 waveguide height of 432 µm.The MEMS-RSs and comb-drive actuators for their operation are etched in the device layer.Furthermore, the designed crossover switch needs 12 H -plane bends for signal routing and four additional E-plane transitions at the input-output ports that facilitate direct connection of the reduced-height in-plane waveguides (849 × 275 µm) to standard-sized (864 × 432 µm) out-of-plane WR-3.4 interfaces.Fig. 4 shows an exploded view of the individual chips composing the crossover switch prototype and the concept of the assembled device mounted between two custom-made dual-channel WR-3.4 waveguide interposers.The distance between the adjacent channels of the interposer, which is made of brass and fabricated by CNC milling, is 550 µm on one side and diverted away on the opposite side to be able to connect it to two waveguide flanges in the measurement setup.As can be seen, chips #1 and #4 contain the waveguiding structures, including the hybrid couplers, waveguides of the SPST switches, and routing to the standard WR-3.4 interfaces, and chips #2 and #3 contain the MEMS-RSs, comb-drive actuators, and vertical coupling slots.The subcomponents of the designed switch prototype are optimized separately before fine-tuning the overall device, and all the full-wave simulations are performed by CST Microwave Studio Suits.Besides, according to the frequency band requirements outlined in the CAR2TERA project, 238-248-GHz band, the crossover switching circuit is designed to operate properly within a broader bandwidth of 220-260 GHz to ensure optimal performance and robustness of the switching circuit.Fig. 5(a) and (b) shows the simulated performance of the crossover switch, without the E-plane transitions and waveguide routings, for different switching cantilever distances (d: referring to Fig. 3) in the crossover state.The simulated crossover-state ISO, IL, and RL are better than 32, 0.3-0.6, and 26 dB, respectively, between 220 and 260 GHz for a cantilever distance of 20 µm.Moreover, Fig. 6(a) and (b) shows the simulated performance of the crossover switch, without the E-plane transitions and waveguide routings, for different contact gaps (g: referring to Fig. 3) in the straight state.The switching mechanism is designed to be robust against variations in the contact gap and achieves its performance even when the switching cantilevers are not in full contact but in capacitive mode with a contact gap of up to 250 nm.The ISO, IL, and RL are better than 32, 0.2-0.5, and 12.5 dB, respectively, from 220 to 260 GHz for a contact gap of 250 nm. A. Hybrid Coupler Fig. 7(a) shows an overview of the designed multistep hybrid coupler utilized to implement the crossover switch, with detailed dimensions.The coupler is optimized for having high ISO, low RL, amplitude balanced, and 90 • out-of-phase outputs in the design bandwidth of 220-260 GHz.According to Fig. 7(c), which shows the simulated performance of the hybrid coupler, the RL and ISO are better than 30 dB, and the output phase error is better than 1 • from 220 to 260 GHz.As can be seen, the amplitude of the output ports is not entirely balanced for the whole frequency range, i.e., there is a slight difference between the coupled and thru ports, which is typical for such a multistep coupler geometry.

B. SPST Switch
The SPST switches are the core components in the proposed crossover switch concept.The complete design procedure of the utilized SPST switches, which are based on two MEMS-RSs that are positioned sequentially in the signal path in order to improve the ISO, as well as the IL and RL, has previously been published by Karimi et al. [20].Fig. 7(b) shows a perspective view of the SPST switch with detailed dimensions, and Fig. 7(d) shows the simulated performance for a 20-µm distance between the switching cantilevers in the propagating state and a 250-nm contact gap in the reflecting state.

C. E-Plane Transition
An overview of the E-plane transition, utilized for the input-output waveguide port interfaces, is shown in Fig. 7(e), along with its detailed dimensions [24].According to Fig. 7(e), the simulated insertion and return losses are better than 0.1 and 20 dB between 200 and 280 GHz, respectively.

D. Electrostatic Actuator
The actuation mechanism consists of four comb-drive push-pull actuator sections on each side of the MEMS-RSs, in which every section consists of 50 interdigitated comb fingers.Besides, the restoring mechanism consists of eight parallel 1.5-folded beam mechanical springs, also placed on both sides of the MEMS-RSs.The push-pull actuator is designed to displace the MEMS-RSs by ±10 µm at a nominal actuation voltage of 40 V. Fig. 8 shows a scanning electron microscope (SEM) image of the actuating system with zoomed-in views of the stopping mechanism and interdigitated comb fingers.Moreover, the natural resonance frequency of the mass-spring system, which defines an upper bound for the switching speed, is calculated numerically and is around 2.1 KHz.Thus, the switching speed is expected to be in the order of a few hundred microseconds.

III. FABRICATION AND ASSEMBLY
The designed crossover switch prototype consists of four vertically stacked SOI chips.Three masks for the handle layer and one mask for the device layer are used to etch different features on the SOI wafer.Starting from a thermally oxidized SOI wafer, the device layer and the first handle layer mask Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.patterns are transferred on the wafer by photolithography and dry etching of the thermal oxide layers.The first handle layer mask includes clearance areas above the dc voltage lines and the moving parts to avoid short-circuiting and to enable a free movement of the mechanical parts.Next, 150 nm of amorphous silicon is deposited on the wafer as an etch-stop layer.Then, a further 1.5-µm-thick silicon dioxide layer is deposited on the handle layer as the second handle layer mask, with plasma-enhanced-chemical-vapor-deposition (PECVD) and patterned by photolithography and oxide dry etching.The second handle layer mask is used for etching the steps inside the waveguides.The same process is repeated for the third handle layer mask, which is for etching the fullheight waveguides.After transferring all the mask patterns onto the silicon dioxide layers on both sides of the wafer, first, the handle layer and then the device layer are etched by DRIE process.Afterward, the moving parts are released by vapor hydrofluoric acid (HF).Next, the chips are metalized with 200 and 1000 nm of gold on the device and handle layers, respectively.Finally, the metalized chips are bonded by thermocompression bonding at 200 • C.
Fig. 9(a) and (b) shows the photographs of the fabricated chips before and after bonding.As can be seen, for accessing the dc voltage pads, the top and bottom chips (chips #1 and #4) are smaller compared to the middle ones (chips #2 and #3).Moreover, Fig. 10 shows the four SEM images of the fabricated chips.Fig. 10(a) shows a top view of the chips containing the waveguide structures (chips #1 and #4); Fig. 10(b) shows a zoomed-in view of the switching junction, in which the E-plane step and the clearance above the dc lines and moving parts can be seen clearly, and the debris visible in the cross section view AA ′ is due to the dicing process.Fig. 10(c) shows the H -plane bends and the E-plane transitions to standard WR-3.4 waveguide ports; Fig. 10(d) shows a top view of the MEMS actuators and MEMS-RSs with a close-up view of the switching cantilevers on chips #2 and #3.

IV. CHARACTERIZATION AND ANALYSIS
The RF performance of the fabricated crossover switch prototype is measured by a Rohde & Schwarz ZVA-24 VNA with ZC330 frequency extenders, calibrated by a standard calibration kit with thru-offset-short-match (TOSM) method.Fig. 11(a) shows the assembled device mounted between the two interposers and aligned to them by inner alignment pins.One of the inner alignment holes on the fabricated device is tightly fit with a circular hole, and the other is a loose elliptical hole, ensuring a repeatable alignment precision of ±5 µm [25], [30].The overall structure, which is fixed on a custom-made holder [shown in Fig. 11(a)], has four standard WR-3.4 interfaces, in which the unwanted ports are loaded by sub-THz match loads in every measurement to reduce the effect of the reflected power from other ports.Moreover, Fig. 11(b) and (c) shows the configuration of the measurement setup, where the interposers are connected to the frequency extenders in the measurement setup, and the switch is actuated with five micrometer-size needles, which are fixed on precise dc probe positioners.Fig. 11(b) depicts the measurement setup for measuring input #1 to output #2 (referring to Fig. 2), and Fig. 11(c) shows the measurement setup measuring input #1 input #2 Fig. 2).
After all the flange-to-flange measurements, which include the and the crossover switch sandwiched them, the S-parameters of the interposers are measured and de-embedded from the measured flange-to-flange results by postprocessing the data in MATLAB.Thus, the reference (calibration) planes are shifted to the surface of the crossover switch device [referring to Fig. 2(b)].between 220 and 260 GHz at the actuation voltage of 50 V, which is in excellent agreement with the simulation data shown in Fig. 6(b).Furthermore, Fig. 14 shows the measured and simulated RL of the crossover switch in the crossover and straight states.According to Fig. 14, the RL is better than 14 and 13.6 dB from 220 to 260 GHz for the crossover and straight states, respectively, and is better than 15 dB for more than 95% of the bandwidth.After analyzing the measured results, it can be concluded that the fabricated crossover switch prototype achieves an excellent performance from 220 to 260 GHz, 16.6% fractional bandwidth, with excellent agreement between the measured results and the simulated data.
According to the simulations shown in Figs.distance between the switching cantilevers, d in the crossover state, and g in the straight state (referring to Fig. 3).Thus, the input-to-output ISOs are measured at different actuation voltages, which are corresponding to different distances between the switching cantilevers and shown in Fig. 15(a) and (b).Furthermore, the performance of the fabricated crossover switch prototype summarized and compared to other state-of-theart mm-wave and sub-THz switches in Table I.According to Table I, presented crossover switch outperforms all the other switches except for the macromechanical SPDT switch presented in [28] by JPL.Compared to the latter switch, the presented crossover switch in this article, though having lower ISO and slightly higher IL, is three orders of magnitude compacter, has near-zero power consumption, and is faster by at least three orders of magnitude due to its electrostatic MEMS actuator nature.Besides, the presented crossover switch with two input and two output ports is a generalized form of SPDT switches and has more advanced circuit functions compared to SPDT switches, i.e., loading either input port #1 or #2 [referring to Fig. 2(b)] transforms the crossover switch into an SPDT switch with a single input and double outputs.

V. CONCLUSION
A novel crossover switch, consisting of two hybrid couplers and two SPST switches, is successfully demonstrated in this article.The proposed crossover switch concept is validated by a switch prototype operating in the 220-260-GHz frequency band, with excellent agreement between the measured results and the simulated data.In essence, the function of the proposed switch can be explained as a switchable resonance circuit composed of a resonating section between two MEMS-RSs.Based on this principle, it is possible to compose more complex elements in the circuit, for instance, for achieving larger bandwidth by adding more of such sections and MEMS-RSs.The prototype is fabricated by silicon micromachining and implemented with four vertically stacked SOI chips.The switching is performed by two microelectromechanically reconfigurable switching surfaces, which are controlled by two microelectromechanical comb-drive actuators.Due to the symmetry of the design, the presented switch is well-suited for applications with redundancy requirements.Besides, it can be used for receiver calibration due to excellent ISO, IL, and RL performance with the measured ISO/IL/RL of better than 29.3/0.9-1.4/14dB and 29/0.8-1.3/13.6 dB for the fabricated prototype in the crossover and straight states, respectively.The novelty and excellent performance of the proposed crossover switch make it a significant progress toward developing more sophisticated switching circuits.

Manuscript received 14
December 2023; revised 11 February 2024; accepted 27 February 2024.Date of publication 21 March 2024; date of current version 5 September 2024.This work was supported in part by CAR2TERA Project from European Union's Horizon 2020 Research and Innovation Program through Kungliga Tekniska högskolan (KTH) under Grant 824962, in part by the Swedish Foundation for Strategic Research under Grant CHI19-0027, and in part by the Marie Sklodowska-Curie under Grant 811232.(Corresponding author: Armin Karimi.)

Fig. 1 .
Fig. 1.Operation principle of the proposed crossover switch concept in different states.

Fig. 2 .
Fig. 2. (a) Schematic block diagram of the proposed crossover switch, illustrating the thru (green) and isolated (red) ports' signal path in the crossover and straight states.(b) Perspective view of the crossover switch implementation, including the hybrid couplers and SPST switches connected to four E-plane transitions to standard WR-3.4 waveguide.

Fig. 2 (
b) shows a perspective view of the micromachined waveguide implementation of the crossover switch.The couplers are implemented by multistep hybrid couplers, and the SPST switches are implemented by two sequential microelectromechanically reconfigurable switching surfaces (MEMS-RSs)[20].There are two reasons for having two sequential MEMS-RSs in each SPST switch.

Fig. 3 .
Fig. 3.Overview of the MEMS-RSs in the propagating and reflecting states with detailed dimensions (d: distance between the switching cantilevers in the propagating state, g: gap between the switching cantilevers in the reflecting state, t: thickness of the switching cantilevers, and l ov : vertical overlap between the switching cantilevers).

Fig. 4 .
Fig. 4. Exploded view of the vertically stacked SOI chips of the crossover switch prototype and a concept view of the measurement setup showing the assembled device mounted and aligned to two custom-made interposers with inner alignment pins.

Fig. 5 .
Fig. 5. Simulated performance of the designed crossover switch, without the E-plane transitions and waveguide routings, with different distances between the switching cantilevers (d: referring to Fig. 3) in the crossover state.(a) ISO.(b) IL and RL.

Fig. 6 .
Fig. 6.Simulated performance of the designed crossover switch, without the E-plane transitions and waveguide routings, with different gaps between the switching cantilevers (g: referring to Fig. 3) in the straight state.(a) ISO.(b) IL and RL.

Fig. 7 .
Fig. 7. Perspective view and detailed dimensions of the designed.(a) Multistep hybrid coupler and (b) SPST switch.Simulated performance of (c) hybrid coupler and (d) SPST switch.(e) Simulated performance of the E-plane transition, utilized for the input-output ports, along with its perspective view and detailed dimensions.

Fig. 8 .
Fig. 8. SEM image of the microelectromechanical comb-drive actuator with zoomed-in view of the stopping mechanism and interdigitated comb fingers.

Fig. 9 .
Fig. 9. Photograph of the fabricated chips (a) before and (b) after the thermocompression bonding.As can be seen in (b), the top and the bottom chips (chips #1 and #4) are smaller compared to the middle ones (chips #2 and #3), facilitating access to the dc voltage pads.

Fig. 12 (
a) and (b) shows the measured and simulated input-to-output (|S 21 | in the crossover state and |S 31 | in the straight state) and input-to-input (|S 41 |) ISOs of the fabricated crossover switch prototype in the crossover and straight states, respectively.The applied actuation voltage is 44 V in the crossover state and 50 V in the straight state.The measured input-to-output ISOs in the crossover and straight states are better than 29.3 and 29 dB in the 220-260-GHz frequency band, respectively, and are in excellent agreement with the simulation data.Besides, as expected from the simulation data, the input-to-output ISO in the straight state remains uniform across the operating band, and by comparing the measured performance with the simulation data, it can be inferred that

Fig. 10 .
Fig. 10.SEM images of the fabricated chips after metallization.(a) Top view of the waveguide structures.(b) Overview of the switching junction and a cross section view AA ′ of a diced chip.(c) Overview of the H -plan bends and E-plane transitions to standard WR-3.4 waveguide interfaces.(d) Top view of the MEMS actuators and MEMS-RSs, and a zoomed-in view of the switching cantilevers.

Fig. 11 .
Fig. 11.(a) Assembled crossover switch prototype mounted between two interposers and fixed on a custom-made holder with screws, along with an illustration of the device aligned to an interposer with inner alignment pins.Configuration of the setup for measuring (b) input #1 to output #2 (referring to Fig. 2) and (c) input #1 to input #2 (referring to Fig. 2).

13. 7
and 34 dB from 220 to 260 GHz in the crossover and straight states, respectively.Fig. 13(a) and (b) shows the measured and simulated IL in the crossover and straight states, respectively (|S 31 | in the crossover state and |S 21 | in the straight state).The crossover-state IL is between 0.9 and 1.4 dB in the 220-260-GHz frequency band at the actuation voltage of 44 V.According to the inset in Fig.13(a), the crossover-state IL does not change considerably by displacing the reconfigurable switching surfaces from the nonactuated state to the fullactuated state, which was expected according to the simulation data shown in Fig.5(b).On the contrary, as can be seen Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

Fig. 12 .
Fig. 12. Measured and simulated input-to-output and input-to-input ISOs of the crossover switch in (a) crossover and (b) straight states.

Fig. 13 .
Fig. 13.Measured and simulated IL of the crossover switch in (a) crossover and (b) straight states.

Fig. 14 .
Fig. 14.Measured and simulated RL of the crossover switch in the crossover and straight states.

Fig. 15 .
Fig. 15.Measured and simulated input-to-output ISO of the crossover switch at different actuation voltages in (a) crossover and (b) straight states.
5(a) and 6(a), the input-to-output ISOs (|S 21 | in the crossover state and |S 31 | in the straight state) are changing significantly by changing the

TABLE I COMPARISON
OF THE PRESENTED CROSSOVER SWITCH WITH OTHER STATE-OF-THE-ART MM-WAVE AND SUB-THZ WAVEGUIDE SWITCHES.FOR COMPARISON, THREE STATE-OF-THE-ART SOLID-STATE SEMICONDUCTOR-BASED SWITCHES ARE ALSO ADDED AT THE BOTTOM OF THE TABLE