THz Radiation Measurement with HTSC Josephson Junction Detector Matched to Planar Antenna

Superconducting Josephson junctions have major advantages as detectors of millimeter wave radiation. Frequency of the radiation can be easily derived from the Shapiro steps of the current-voltage characteristics. However, system performance is highly sensitive to impedance mismatch between the antenna and the junction; therefore, optimization is essential. We analyzed and implemented an improved antenna structure, in which the junction is displaced from the antenna center and placed between the ends of two matching strips. Based on theoretical analysis and advanced electromagnetic simulations, we optimized strip dimensions, which affect both the detection magnitude and the frequency of the reflection coefficient dip. Accordingly, two Au bow-tie antennas with different matching strip widths were fabricated. Superconducting Yttrium Barium Copper Oxide (YBCO) thin films were deposited exactly at the bicrystal substrate misorientation points, forming Josephson junctions at the ends of two matching strips. We found a very high correlation between the simulations and the response to Radio Frequency (RF) radiation in the range of 145–165 GHz. Experimental results agree extremely well with the design, showing best performance of both antennas around the frequency for which impedance matching was derived.


Introduction
Due to the ever-increasing requirements of communication systems, there is an increasing demand for systems in the millimeter wave range [1]. This range is also applicable to medicine, biology, astronomy, defense and materials technology [2,3]. However, realization of high frequency communication systems in the THz range becomes extremely complex. Implementation of sources and detectors in this range is very complicated [4,5]. Moreover, conventional communication techniques are incompatible with these high frequencies. On the other hand, photon energies are too low for In order to examine the effects of impedance matching on the performance of the detection system, we implemented two bow-tie antennas with different matching strip width. The configuration of the two systems was based on the results of our simulations. The HTSC junctions were implemented by growing a thin film of YBa 2 Cu 3 O 7 on the substrate. Layers of YBCO thin films and Au were deposited over the (100)MgO bicrystal substrate, by Theva GmbH, Germany [32]. In the space between the two antenna poles, only the Au layer was etched, exposing a thin layer of YBCO. Thus, the YBCO Josephson junction was located exactly at the bicrystal misorientation point. The final structure of the planar antenna and the detector was obtained by etching of these layers, performed by STAR Cryoelectronics, USA [33]. In Section 3, we present experimental results, demonstrating the high performance of the fabricated antennas.

Josephson Junction as an RF Element in Cascade Reception Chain
In order to develop the model for the simulations of a superconducting antenna system, we investigated the circuit representation of the detector. The current flowing through the junction I S is the superposition between the DC current supplied to the junction, and the current generated by the RF radiation. Neglecting noise, I S can be expressed in terms of Bessel functions of the first kind J m [9]: The RF radiation generates steps in the I-V characteristics, known as the Shapiro steps. The height of the steps is proportional to the voltage amplitude of the radiation V RF [34,35]. It reflects the detection efficiency, as affected by the integration between the antenna and the Josephson junction detection element. The voltage separation V m between the steps gives an accurate measure of the radiation frequency ω RF , obtained from V m = m 2e ω RF .
The current flowing through the junction is a superposition of three components: the normal current, I R , the charging current, I ch , and the Josephson supercurrent, I JJ . Since the current supplied to the junction is larger than the critical current I c , the superconductor is not in its classic superconducting state, i.e., the junction has a parallel normal component that behaves as a resistor R n .
We analyze the interrelations between the input impedance, Z in , the antenna impedance, Z A , and the Josephson junction equivalent impedance, Z JJ . This relation is derived through a cascade chain connection of two "2 port network" models, using an equivalent ABCD matrix [27]. Figure 1 shows the layout of the networks. Network α represents the junction network, and network β represents the antenna network. system, we implemented two bow-tie antennas with different matching strip width. The configuration of the two systems was based on the results of our simulations. The HTSC junctions were implemented by growing a thin film of YBa2Cu3O7 on the substrate. Layers of YBCO thin films and Au were deposited over the (100)MgO bicrystal substrate, by Theva GmbH, Germany [32]. In the space between the two antenna poles, only the Au layer was etched, exposing a thin layer of YBCO. Thus, the YBCO Josephson junction was located exactly at the bicrystal misorientation point. The final structure of the planar antenna and the detector was obtained by etching of these layers, performed by STAR Cryoelectronics, USA [33]. In Section 3, we present experimental results, demonstrating the high performance of the fabricated antennas.

Josephson Junction as an RF Element in Cascade Reception Chain
In order to develop the model for the simulations of a superconducting antenna system, we investigated the circuit representation of the detector. The current flowing through the junction IS is the superposition between the DC current supplied to the junction, and the current generated by the RF radiation. Neglecting noise, IS can be expressed in terms of Bessel functions of the first kind Jm [9]: The RF radiation generates steps in the I-V characteristics, known as the Shapiro steps. The height of the steps is proportional to the voltage amplitude of the radiation VRF [34,35]. It reflects the detection efficiency, as affected by the integration between the antenna and the Josephson junction detection element. The voltage separation between the steps gives an accurate measure of the radiation frequency , obtained from V = m ℏ ω .
The current flowing through the junction is a superposition of three components: the normal current, IR, the charging current, Ich, and the Josephson supercurrent, IJJ. Since the current supplied to the junction is larger than the critical current Ic, the superconductor is not in its classic superconducting state, i.e., the junction has a parallel normal component that behaves as a resistor Rn.
We analyze the interrelations between the input impedance, Zin, the antenna impedance, ZA, and the Josephson junction equivalent impedance, ZJJ. This relation is derived through a cascade chain connection of two "2 port network" models, using an equivalent ABCD matrix [27]. Figure 1 shows the layout of the networks. Network α represents the junction network, and network β represents the antenna network.
One can express the integration between these networks through the voltage/current matrix relationship. Figure 1. Cascade, chain connection, of two "2 port network".
One can express the integration between these networks through the voltage/current matrix relationship.
Appl. Sci. 2020, 10, 6482 4 of 17 Equations (2) and (3) describe the network matrices, the junction ABCD matrix [α], and the antenna impedance matrix [β]. From these equations, we derive an expression for the input impedance, Z JJ , can be written in terms of electromagnetic wave propagation as [8]: where Z 0 = 377 Ω is the free space impedance. w and l are the width and length of the junction, ε rs is the relative dielectric constant of the junction, c is the speed of light in free space. The speed of light in a medium c, is given by: c = c l l ε rs , where the effective length of the junction, is given by l = l + λ L1 + λ L2 . The London penetration depth for both sides of the junction is given by λ L = m e µ 0 n s e 2 , where m e is the electron mass, µ 0 is the permeability of free space, n s is the density of superconducting charge carriers, and e is the electron charge. Using Equations (4) and (5) we can express the dependence of Z in on junction dimensions as: The optimal input impedance for impedance matching is obtained by selecting the proper dimensions of the antenna and matching strips, thus setting the value of Z A in Equation (6).

Antenna Optimization
The bow-tie antenna is composed of a two-part bow-tie shape, designed to reduce the reflection coefficient. It is characterized by broad-band impedance [15,20]. The performance of the detection systems is analyzed by electromagnetic simulations, carried out using High-Frequency Structure Simulator (HFSS) software by ANSYS. We begin with the analysis of the conventional design, in which the Josephson junction is placed in the gap between the two poles of the antenna [12,25]. Optimal results were obtained for a pole width of 115.5 µm and a length of 144.3 µm, with a separation between the two antenna poles, a gap width, of a 4.2 µm. The efficiency of such systems deteriorates rapidly at very high frequencies, due to impedance mismatch between the antenna and the junction. Figure 2 shows the simulation results of the reflection coefficient S 11 as a function of frequency. The simulation renders a bandwidth of about 4 GHz, at 3.57:1 voltage standing wave ratio (VSWR), showing a power ratio of up to 0.874 for the power supplied to the antenna, i.e., an efficiency of 87.4% not including ohmic losses. The reflectivity obtained is around −11 dB at the central frequency of 148 GHz.
We propose a new structure in which the junction is placed between the ends of two matching strips, as outlined in Figure 3. This structure is intended to enable, with proper optimization of the strip dimensions, high impedance matching at high frequencies between the antenna input impedance Z in and the junction impedance Z JJ . Figure 4 shows the structure of a two-pole antenna, as designed and built in the simulation software. The poles consist of a 250 nm thick Au layer on top of a 330 nm thick YBCO layer and 0.5 mm thick MgO substrate. Other parameters included in the simulations are the dielectric constant of the MgO substrate, ε r = 9.8, and the loss tangent, tanδ, approximately 10 −4 at 77 K [36,37]. The antenna dimensions were modified in order to obtain a high gain and a low reflection coefficient. Optimal results were obtained for a pole width of 75.5 µm and a length of 74 µm. The separation between the two antenna poles, the gap width, is 4.2 µm. Appl. Sci. 2020, 10, x FOR PEER REVIEW 5 of 18 We propose a new structure in which the junction is placed between the ends of two matching strips, as outlined in Figure 3. This structure is intended to enable, with proper optimization of the strip dimensions, high impedance matching at high frequencies between the antenna input impedance Zin and the junction impedance ZJJ.   We propose a new structure in which the junction is placed between the ends of two matching strips, as outlined in Figure 3. This structure is intended to enable, with proper optimization of the strip dimensions, high impedance matching at high frequencies between the antenna input impedance Zin and the junction impedance ZJJ.     Figure 5 shows the results of simulations for the reflection coefficient S11, as a function of frequency, with matching strips 4.6 μm wide each, for various values of length (L in Figure 3). In order to evaluate the junction equivalent impedance ZJJ, we used typical parameters for bicrystal Josephson junctions [31,38]. The light velocity ratio ̅ ⁄ is estimated to be around 0.03 [8,31] with a relative dielectric constant, , of 5 for the YBCO [38]. The junction width was set to 4.2 μm.
The lowest reflection coefficient was obtained for an antenna with L = 108 μm. The simulation renders a bandwidth of ~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not  Figure 5 shows the results of simulations for the reflection coefficient S 11 , as a function of frequency, with matching strips 4.6 µm wide each, for various values of length (L in Figure 3). In order to evaluate the junction equivalent impedance Z JJ , we used typical parameters for bicrystal Josephson junctions [31,38]. The light velocity ratio c/c is estimated to be around 0.03 [8,31] with a relative dielectric constant, ε rs , of 5 for the YBCO [38]. The junction width w was set to 4.2 µm. Inserting these values into Equation (3), one obtains Z JJ = 1 0.03 l 4.2·10 −6 ·5 ·377 Ω. Thus, for l 3.5 nm, Z JJ 2 Ω. The lowest reflection coefficient was obtained for an antenna with L = 108 µm. The simulation renders a bandwidth of~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not including ohmic losses. For this structure, the calculated reflection coefficient is −26 dB at the central frequency of 155 GHz.  Figure 5 shows the results of simulations for the reflection coefficient S11, as a function of frequency, with matching strips 4.6 μm wide each, for various values of length (L in Figure 3). In order to evaluate the junction equivalent impedance ZJJ, we used typical parameters for bicrystal Josephson junctions [31,38]. The light velocity ratio ̅ ⁄ is estimated to be around 0.03 [8,31] with a relative dielectric constant, , of 5 for the YBCO [38]. The junction width was set to 4.2 μm.
. • • • 377 Ω. Thus, for l ≅ 3.5 nm, ZJJ ≅ 2 Ω. The lowest reflection coefficient was obtained for an antenna with L = 108 μm. The simulation renders a bandwidth of ~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not including ohmic losses. For this structure, the calculated reflection coefficient is −26 dB at the central frequency of 155 GHz. Simulation results presented in Figure 5 were performed assuming a known value of the junction impedance: ZJJ. However, it should be pointed out that according to Equation (6), this impedance is most sensitive to the exact value of the length of the junction . Figure 6 shows the reflection coefficient as a function of frequency, for several values of ZJJ. Simulation results show that the system performance deteriorates rapidly for ZJJ values different from those required for optimal impedance matching conditions. The system efficiency drops from 99.75% Simulation results presented in Figure 5 were performed assuming a known value of the junction impedance: Z JJ . However, it should be pointed out that according to Equation (6), this impedance is most sensitive to the exact value of the length of the junction l. Figure 6 shows the reflection coefficient as a function of frequency, for several values of Z JJ . Simulation results show that the system performance deteriorates rapidly for Z JJ values different from those required for optimal impedance matching conditions. The system efficiency drops from 99.75% for Z JJ 2 Ω to 95% for Z JJ 1 Ω and down to 80% for Z JJ 0.6 Ω, excluding ohmic losses. The mismatch is also reflected in a shift of the central frequency.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 18 for ZJJ ≅ 2 Ω to 95% for ZJJ ≅ 1 Ω and down to 80% for ZJJ ≅ 0.6 Ω, excluding ohmic losses. The mismatch is also reflected in a shift of the central frequency.  Figure 7 shows the real (continuous blue line) and the imaginary (red dots) parts of the antenna input impedance Zin for a structure with matching strips 4.6 μm wide and 108 μm long, as a function of frequency. Clearly, Zin is highly correlated to the junction impedance ZJJ around 155 GHz, reaching high impedance matching at this frequency.  Figure 7 shows the real (continuous blue line) and the imaginary (red dots) parts of the antenna input impedance Z in for a structure with matching strips 4.6 µm wide and 108 µm long, as a function of frequency. Clearly, Z in is highly correlated to the junction impedance Z JJ around 155 GHz, reaching high impedance matching at this frequency. Figure 6. Reflection coefficient S11 vs. frequency for various values of Josephson junction characteristic impedance ZJJ for bow-tie antenna. Figure 7 shows the real (continuous blue line) and the imaginary (red dots) parts of the antenna input impedance Zin for a structure with matching strips 4.6 μm wide and 108 μm long, as a function of frequency. Clearly, Zin is highly correlated to the junction impedance ZJJ around 155 GHz, reaching high impedance matching at this frequency.     Figure 9 shows the results of simulations for the reflection coefficient S11, as a function of frequency, for antennas with a strips length of 108 μm, for various values of matching strips width W. Here, the lowest reflection coefficient was obtained for an antenna with a strip width of 4.6 μm. This simulation renders a bandwidth of ~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not including ohmic losses. The reflection coefficient obtained is −26 dB at a central frequency of 155 GHz. A very strong dependence of reflection coefficient on the width of the matching strips is observed. This correlation is more pronounced in determining the central frequency of the detection system, which emphasizes the need for a great precision in the production process of such a detection  Figure 9 shows the results of simulations for the reflection coefficient S 11 , as a function of frequency, for antennas with a strips length of 108 µm, for various values of matching strips width W. Here, the lowest reflection coefficient was obtained for an antenna with a strip width of 4.6 µm. This simulation renders a bandwidth of~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not including ohmic losses. The reflection coefficient obtained is −26 dB at a central frequency of 155 GHz. A very strong dependence of reflection coefficient on the width of the matching strips is observed. This correlation is more pronounced in determining the central frequency of the detection system, which emphasizes the need for a great precision in the production process of such a detection system.  Figure 9 shows the results of simulations for the reflection coefficient S11, as a function of frequency, for antennas with a strips length of 108 μm, for various values of matching strips width W. Here, the lowest reflection coefficient was obtained for an antenna with a strip width of 4.6 μm. This simulation renders a bandwidth of ~4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.75%, not including ohmic losses. The reflection coefficient obtained is −26 dB at a central frequency of 155 GHz. A very strong dependence of reflection coefficient on the width of the matching strips is observed. This correlation is more pronounced in determining the central frequency of the detection system, which emphasizes the need for a great precision in the production process of such a detection system. Figure 9. Reflection coefficient S11 vs. frequency for various matching strips widths without a lens. L=108 μm.
To improve the directivity of the antenna, a silicon (Si) hemispherical lens, 1 mm in diameter, was added in the simulations, as demonstrated in Figure 10 [39,40]. The lens was placed on the To improve the directivity of the antenna, a silicon (Si) hemispherical lens, 1 mm in diameter, was added in the simulations, as demonstrated in Figure 10 [39,40]. The lens was placed on the backside −Z direction of the MgO substrate, to prevent potential damage to the junction located in the center of the antenna.       Figure 11 shows the simulation results for the reflection coefficient, obtained for the antenna with a Si lens, with matching strips length L of 108 μm, for various values of strips width W. As in the previous case, the lowest reflection was obtained for an antenna with a width of 4.6 μm. The simulation renders a bandwidth of ~3.4 GHz, at 2.615:1 VSWR, with an efficiency of up to 99.2%, not including ohmic losses. The reflectivity obtained is −21 dB at the central frequency of 148.2 GHz. Figure 11. Reflection coefficient S11 vs. frequency for various matching strips widths of antennas with a Si lens. L = 108 μm. Figure 12 shows the radiation curve and gain power, 2-D far-field pattern, at the central frequency of 148.2 GHz. The radiation pattern indicates the lower gain power along the Z axis, i.e., the antenna performance is inferior in that direction, as compared with the gain of 6.2 dBi along the −Z axis. In order to obtain high performance, the detector should be positioned pointing in the −Z direction relative to the radiation source.

Experimental Results
Two bow-tie antennas, along with matching strips, were implemented. The widths of the strips, 4.6 μm and 5.4 μm, were set according to the best results obtained in our theoretical analysis. We tested the performance of the antennas with and without a Si lens. The frequency response of the antennas was measured using a Backward Wave Oscillator (BWO) THz source. The antennas were placed inside a quasi-optic cryogenic cooling system, thereby the Josephson junctions operated at 60 K, below their critical temperature. We measured the I-V characteristics of the antenna system by radiating it with RF signals in the range of 145-165 GHz.

Experimental Results
Two bow-tie antennas, along with matching strips, were implemented. The widths of the strips, 4.6 µm and 5.4 µm, were set according to the best results obtained in our theoretical analysis. We tested the performance of the antennas with and without a Si lens. The frequency response of the antennas was measured using a Backward Wave Oscillator (BWO) THz source. The antennas were placed inside a quasi-optic cryogenic cooling system, thereby the Josephson junctions operated at 60 K, below their critical temperature. We measured the I-V characteristics of the antenna system by radiating it with RF signals in the range of 145-165 GHz.

Experimental Setup
The measurement setup includes a BWO THz source, quasi-optical lenses to focus the radiated RF signal on the detector, a cryogenic cooling system with high-density polyethylene windows, and a Tektronix company Keithley Nano-voltmeter and current source ( Figure 13). The Cryo Industries cryogenic cooling system enables testing the performance of the device at low temperatures, down tõ 2 K. An important advantage of this cooling system is that the devices inside the chamber do not touch directly the liquid helium used for cooling, thus preventing contamination of the device. The BWO source is a miniature electrovacuum device placed in a metal packaging. When the BWO source is in a magnetic field and supplied with high voltage (up to 6.5 kV), it emits monochromatic electromagnetic radiation, with a power of up to~10 mW, depending on the supplied voltage and magnetic field frequency and strength [41]. A photo and a discerption of the measurement setup is presented in the Appendix A.

Measurement Results
We evaluated the sensitivity of the Josephson detector according to its NEP, based on the relation is the normal resistance, is the critical current in the presence of the RF radiation, and ωRF is the radiation frequency. The coupled power voltage responsivity ( ) was derived from ( ) = where is the amplitude of the RF power coupled into the junction.
can be estimated from the amplitude of the RF current, = ℏ [7][8][9]. is the unsuppressed critical current, without RF radiation. ℏ is Planck constant divided by 2π, and e is the electron charge. Δ = ( )Δ , where Δ is the change in the junction current due to the RF signal. ( ) is the dynamic resistance, Δ /Δ . Figure 14 shows the measured I-V and dI/dV characteristics for the antenna with a 4.6 μm wide matching strip. The data was taken at 60 K without a Si lens. The unsuppressed critical current is ≈1 mA and the normal resistance Rn

Measurement Results
We evaluated the sensitivity of the Josephson detector according to its NEP, based on the relation 4k B TR n . R n is the normal resistance, I c is the critical current in the presence of the RF radiation, and ω RF is the radiation frequency. The coupled power voltage responsivity η(V) was derived from η(V) = ∆V P RF where P RF is the amplitude of the RF power coupled into the junction. P RF can be estimated from the amplitude of the RF current, I RF = ω RF √ I c0 ∆I eR n I c [7][8][9]. I c0 is the unsuppressed critical current, without RF radiation. is Planck constant divided by 2π, and e is the electron charge. ∆V = R d (V)∆I, where ∆I is the change in the junction current due to the RF signal. R d (V) is the dynamic resistance, ∆V/∆I. Figure 14 shows the measured I-V and dI/dV characteristics for the antenna with a 4.6 µm wide matching strip. The data was taken at 60 K without a Si lens. The unsuppressed critical current is I c0 ≈ 1 mA and the normal resistance R n ≈ 3.5 Ω. The top curve shows the junction response to the RF signal at 155 GHz, showing a critical current I c ≈ 0.75 mA. The estimated electrical NEP for the detector is~4 × 10 −12 W·Hz −1/2 . The RF current, I RF , is estimated to be~0.1 mA, thus the voltage responsivity is η(V) ≈ 33 kV·W −1 . The Shapiro step at 0.32 mV is due to the presence of an electromagnetic radiation at~155 GHz. The dI/dV characteristics (bottom curve Figure 14 Figure 15 show the I-V and dI/dV characteristics of the same antenna, measured at 60 K, but with a Si lens. The top curve presents the junction response to an RF signal at 148.2 GHz, showing a critical current of ≈ 0.4 mA. The estimated NEP is ∼1×10 −13 W Hz −1/2 . The RF current, IRF, can be estimated as ~0.3 mA, thus the voltage responsivity is ( ) ≈ 55 kV · W −1 . The Shapiro step at 0.31 mV is due to the presence of electromagnetic radiation at ~148 GHz. These results are clearly superior to typical NEP and ( ) values reported for sensitive detection systems at 60 K, namely around 1 × 10 −12 W·Hz −1/2 and 10kV · W , respectively [13,25,42]. Once again, the high peak at 148 GHz in the dI/dV characteristics (bottom curve Figure 15) demonstrates the high agreement between simulation and experimental results. One can see that using Si lens the center frequency of the matched antenna has shifted, as predicted by the electromagnetic simulations.  Figure 15 show the I-V and dI/dV characteristics of the same antenna, measured at 60 K, but with a Si lens. The top curve presents the junction response to an RF signal at 148.2 GHz, showing a critical current of I c ≈ 0.4 mA. The estimated NEP is~1 × 10 −13 W·Hz −1/2 . The RF current, I RF , can be estimated as~0.3 mA, thus the voltage responsivity is η(V) ≈ 55 kV·W −1 . The Shapiro step at 0.31 mV is due to the presence of electromagnetic radiation at~148 GHz. These results are clearly superior to typical NEP and η(V) values reported for sensitive detection systems at 60 K, namely around 1 × 10 −12 W·Hz −1/2 and 10 kV·W −1 , respectively [13,25,42]. Once again, the high peak at 148 GHz in the dI/dV characteristics (bottom curve Figure 15) demonstrates the high agreement between simulation and experimental results. One can see that using Si lens the center frequency of the matched antenna has shifted, as predicted by the electromagnetic simulations. Figure 16 shows the measured I-V and dI/dV characteristic of the second antenna implemented with matching strips width of 5.4 µm. The measurements were taken at 60 K, with a Si lens. The unsuppressed measured critical current is I c0 ≈ 1.93 mA, a relatively high value for a bicrystal junction. The normal resistance is R n ≈ 3.8 Ω. This higher critical current reduces the sensitivity of the detector. The measured peak at 155 GHz agrees extremely well with the theoretical results shown in Figure 11, proving that the impedance matching between the antenna and the junction was not affected. The top curve shows the junction response to an RF signal at 155 GHz, demonstrating a critical current I c ≈ 1.52 mA. The estimated electrical NEP for the detector is~1 × 10 −12 W·Hz −1/2 . The RF current, I RF , can be estimated to be~0.1 mA, thus the voltage responsivity η(V) ≈ 40 kV·W −1 . The Shapiro step at 0.32 mV indicates the presence of electromagnetic radiation at~155 GHz.
The dI/dV characteristics (bottom curve Figure 16) show the junction RF response for signals at various frequency values. The high peak at 155 GHz demonstrates the high correlation between simulation and experimental results. It is clear that the performance of the antenna with 5.4 µm wide matching strips is inferior to the performances of the first antenna. This is most likely due to a combination of two reasons: the 4.6 µm width is the optimal design for minimal reflection coefficient, as seen in Figure 11, and the quality of the Josephson junction is not as high as the previous one. This results also in low quality data measured without a lens for this antenna.  Figure 16 shows the measured I-V and dI/dV characteristic of the second antenna implemented with matching strips width of 5.4 μm. The measurements were taken at 60 K, with a Si lens. The unsuppressed measured critical current is ≈ 1.93 mA, a relatively high value for a bicrystal junction. The normal resistance is Rn ≈ 3.8 Ω. This higher critical current reduces the sensitivity of the detector. The measured peak at 155 GHz agrees extremely well with the theoretical results shown in Figure 11, proving that the impedance matching between the antenna and the junction was not affected. The top curve shows the junction response to an RF signal at 155 GHz, demonstrating a critical current ≈ 1.52 mA. The estimated electrical NEP for the detector is ∼1 × 10 −12 W·Hz −1/2 . The RF current, IRF, can be estimated to be ~0.1 mA, thus the voltage responsivity ( ) ≈ 40 kV · W −1 .
The Shapiro step at 0.32 mV indicates the presence of electromagnetic radiation at ~155 GHz. The dI/dV characteristics (bottom curve Figure 16) show the junction RF response for signals at various frequency values. The high peak at 155 GHz demonstrates the high correlation between simulation and experimental results. It is clear that the performance of the antenna with 5.4 μm wide matching strips is inferior to the performances of the first antenna. This is most likely due to a combination of two reasons: the 4.6 μm width is the optimal design for minimal reflection coefficient,

Discussion
Several research groups report on sub-THz detection system based on YBCO Josephson junction detectors in a planar antenna configuration. Du et al. integrated a ring-slot [12,34,43] and log-periodic [44] planar antennas with relatively low input impedance (~30 Ω) for HTSC YBCO step-edge Josephson junctions, achieving a voltage responsivity ( ) ≈ 3.5 kV · W −1 and an estimated NEP of ∼1 × 10 −12 W·Hz −1/2 at 60 K. Additional structures were presented by Nakajima et al. [45] and Gao et al. [46,47]. They were able to obtain planar antennas with low input impedance (~13 Ω) combined with YBCO grain-boundary Josephson Junctions detectors, using coplanar waveguide feeding. In our previous report, we investigated a rounded bow tie antenna based on a bicrystal Josephson junction detector [25], with neither a Si lens nor an impedance matching unit. With that system, we obtained a voltage responsivity of η(V) ≈ 1 kV · W and an estimated NEP of ∼6 × 10 −12 W·Hz −1/2 at 60 K. In addition, we reported on a step-edge junction detector with a Si lens, without a matching unit, obtaining a voltage responsivity of ( ) ≈ 15 kV · W −1 [18]. In comparison with previous results, the performance obtained with impedance matching technique presented in this study are significantly superior, showing an estimated NEP of ∼1 × 10 −13 W·Hz −1/2 and voltage responsivity of ( ) ≈ 55 kV · W −1 .

Discussion
Several research groups report on sub-THz detection system based on YBCO Josephson junction detectors in a planar antenna configuration. Du et al. integrated a ring-slot [12,34,43] and log-periodic [44] planar antennas with relatively low input impedance (~30 Ω) for HTSC YBCO step-edge Josephson junctions, achieving a voltage responsivity η(V) ≈ 3.5 kV·W −1 and an estimated NEP of~1 × 10 −12 W·Hz −1/2 at 60 K. Additional structures were presented by Nakajima et al. [45] and Gao et al. [46,47]. They were able to obtain planar antennas with low input impedance (~13 Ω) combined with YBCO grain-boundary Josephson Junctions detectors, using coplanar waveguide feeding. In our previous report, we investigated a rounded bow tie antenna based on a bicrystal Josephson junction detector [25], with neither a Si lens nor an impedance matching unit. With that system, we obtained a voltage responsivity of η(V) ≈ 1 kV·W −1 and an estimated NEP of~6 × 10 −12 W·Hz −1/2 at 60 K. In addition, we reported on a step-edge junction detector with a Si lens, without a matching unit, obtaining a voltage responsivity of η(V) ≈ 15 kV·W −1 [18]. In comparison with previous results, the performance obtained with impedance matching technique presented in this study are significantly superior, showing an estimated NEP of~1 × 10 −13 W·Hz −1/2 and voltage responsivity of η(V) ≈ 55 kV·W −1 .

Conclusions
The performance of high frequency communication antennas for the sub-THz frequency range deteriorates rapidly by impedance mismatch. We designed a new structure in which a superior impedance matching can be obtained by placing the Josephson junction detector remote from the antenna center. Simulations in the range of 130 to 170 GHz for optimal configuration parameters demonstrate a high radiation gain and a low reflection coefficient, S 11 .
Based on these simulations, we fabricated two bow-tie antennas with different matching strips widths: 4.6 µm and 5.4 µm. The measured performances of these antennas at 60 K, with and without a silicon lens, were analyzed. The experimental results agree extremely well with the simulation results. The high peaks in the various dI/dV characteristics, as well as the measured values of the NEP and η(V), are a clear indication of the high sensitivity and frequency selectivity of the proposed detection systems. We have shown that this structure has the potential of improving significantly the detection system efficiency at the THz range.
The massive increase in use of communication networks, such as in the Fifth Generation (5G) technologies, leads to an ever-increasing demand for higher frequency communication ranges. New technologies must be implemented to enable handling the required communication capacity. Solving the problem of impedance mismatch between the HTSC detectors and the antenna presented in this work, is a significant step in the right direction.

Conflicts of Interest:
The authors declare no conflict of interest. Figure A1 is a photo image of the measurement setup described in Section 3. The setup includes the BWO THz source (1), the quasi-optical lenses (2), the cryogenic cooling system with the polyethylene windows (3), and the Keithley Nano-voltmeter and current source (4). For data acquisition and analysis, the measurement equipment was connected to a computer unit. Additionally seen on the optical bench are a chopper and power attenuators (5) (not used in the present experiments). The picture was taken in the applied superconductivity and optical spectroscopy lab in Ariel University, Israel.