Open Access
Add to BibSonomy
Abstract
In this work, we present an optically subharmonic pumped WR3-mixer for enabling photonic coherent frequency-domain terahertz (THz) imaging and spectroscopy systems in the future. The studied mixer operates within the upper range of the WR3-band from 270 GHz to 320 GHz. High-power uni-travelling carrier photodiodes (UTC-PDs) are developed for providing the subharmonic local oscillator (LO) signal within the corresponding WR6-band in the range between 135 GHz and 160 GHz. The proposed THz mixer module consists of a gallium arsenide (GaAs)-based low barrier Schottky diodes (LBSDs) chip and an indium phosphide (InP)-based UTC-PD chip. For integrating the UTC-PD with the WR6 at the mixer’s LO input, an E-plane transition and a stepped-impedance microstrip line low pass filter (MSL-LPF) are developed and monolithically integrated with the UTC-PD chip on a 100 µm thick InP substrate. The E-plane transition converts the quasi-TEM mode of the grounded coplanar waveguide (GCPW) to the dominant TE10 mode of the WR6 and matches the GCPW’s impedance with the WR6’s impedance. According to full-wave EM simulations, the transition exhibits a 1 dB bandwidth (BW) of more than 30 GHz (138.8-172.1 GHz) with a corresponding return loss (RL) better than 10 dB, whereas the minimum insertion loss (IL) is 0.65 dB at a frequency of 150 GHz. Experimentally, the 1 dB BW of the fabricated transition is found to be between 140 GHz and 170 GHz, which confirms the numerical results. The minimum measured IL is 2.94 dB, i.e., about 2 dB larger than the simulated value. In order to achieve the required LO power for successfully pumping the mixer in a direct approach (i.e., without an additional LO amplifier), the design of the epitaxial system of the UTC-PD is optimized to provide a high output power within the WR6-band (110-170 GHz). Experimentally, at 150 GHz, the output power of the fabricated UTC-PD chip is measured to be +3.38 dBm at a photocurrent of 21 mA. To our knowledge, this is the highest output power ever achieved from a UTC-PD at 150 GHz. Finally, the developed high-power UTC-PDs are used as LO source to pump the subharmonic WR3-mixer. Experimentally, the conversion loss (CL) is determined in dependency of the LO power levels within the RF frequency range between 271 GHz and 321 GHz for a fixed IF at 1 GHz. The achieved results have revealed an inverse relation between the CL and LO power level, where the average minimum CL of 16.8 dB is achieved at the highest applied LO power level, corresponding to a photocurrent of 10 mA. This CL figure is promising and is expected to reach the CL of electronically pumped and commercially available THz mixers (∼12 dB) after packaging the LO source with the mixer. Furthermore, an average CL of 17.2 dB is measured at a fixed LO frequency of 150 GHz and a tuned RF frequency between 301 GHz and 310 GHz, i.e., IF between 1 GHz and 10 GHz.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Modern terahertz (THz) applications, such as high resolution spectroscopy, imaging, high data rate communications, and sensing systems, require high sensitive THz receivers [1]. Since the THz frequency range offers several advantages, such as wide signal bandwidth, and high spatial resolution, THz systems are gaining a lot of interest. Consequently, a number of THz detection technologies and approaches have been studied. In general, THz signals can be detected using either a direct or a heterodyne approach [2]. In a heterodyne THz detector, the incoming THz signal is mixed with a reference local oscillator (LO) signal coming from a frequency-stable or pulsed THz source [1,3,4]. In a direct detector, the incoming THz signal is directly received without any LO signal [3].
Direct THz detectors comprise thermal detectors such as Golay cells [5,6] and bolometers [7], barrier detectors such as Schottky barrier diodes (SBD) [8] and superconductor-insulator-superconductor (SIS) [3,9], and quantum well detectors [10]. Usually, direct THz receivers detect THz signals over a wide spectral range, which restricts their use in THz systems that require ultra-high spectral resolution [2]. In addition, direct detectors are not optimum for detecting weak THz signals due to their comparably high noise equivalent power (NEP) which is limited by the background noise in addition to the detector’s own thermal fluctuations [2].
In a heterodyne receiver, the incoming THz signal is first down-converted to an intermediate frequency (IF) or DC before it is being detected [1,3]. In contrast to direct THz detectors, heterodyne mixers allow to detect also weak THz signals because the LO signal enhances the detected IF signal in the mixing process and narrow band IF filters can be used to further reduce the noise bandwidth [1]. Both effects effectively enhance the NEP of a heterodyne THz receiver [1]. In a heterodyne THz mixer the nonlinear behavior of diodes such as SBD, SIS junctions, hot electron bolometers (HEB) and field effect transistors (FET) is exploited for mixing the RF and LO signals [1,11]. For achieving ultra-high sensitivity, heterodyne mixers are usually employed in a coherent approach, which requires the phase of the THz signal to be locked to the phase of the LO signal [12]. This is not the case in incoherent heterodyne THz mixers utilizing non-locked phase LOs. To avoid confusion, it is noted that especially in optical communications often the term “coherent detection” is used although the LO-laser in the receiver is not locked to the optical input signal.
For generating the LO signal required to pump a heterodyne THz mixer, several electronic-based solid-state sources such as Gunn diodes, impact ionization avalanche transit-time (IMPATT) diodes or resonant tunneling diodes (RTDs) are used [1,12]. Another common approach is using a low frequency reference oscillator whose frequency is subsequently multiplied to the desired LO frequency [13]. However, the latter approach suffers from a high phase noise (+6 dB per frequency doubling) [14], while other approaches such as RTD-based THz oscillators are limited in terms of frequency tenability [12]. Alternatively, the LO signal can be also generated by optical means, e.g. by direct LO signal generation using for example gas lasers [11] or quantum cascade lasers (QCL) [15,16] or via optical heterodyne THz LO signal generation in photoconductors or photodiodes [17–19]. Optical heterodyne LO generation is considered as being beneficial with respect to heterodyne THz mixers as the approach offers extreme wide frequency tenability, room temperature operation and, perhaps most importantly, opens up the potential for future heterogeneous or even monolithic integration. In addition, optical heterodyne signal generation using optical combs has recently been reported to achieve record low phase noise level below -170 dBc/Hz for a 12 GHz microwave signal [20]. Although the lasers used in [20] require complex stabilization of the laser’s free-spectral range and carrier offset frequency which cannot be chip-integrated as of now, it shows the general potential of the approach. Recently, QCLs have been used to optically generate the LO signal for pumping HEB mixers [21–23] and Schottky diode mixers [24]. Also photomixers have been employed as LO source to pump HEB mixers [18] and SIS mixers [19,25]. Even uni-travelling carrier photodiodes (UTC-PDs) were already successfully employed for pumping a 183 GHz subharmonic Schottky diode mixer. However, so far, room temperature Schottky barrier subharmonic THz mixers could not be pumped optically since the saturation output power of state-of-the-art UTC-PDs was not high enough given the mixer’s LO power requirements for minimum conversion loss.
In this paper, we report on an optically pumped subharmonic THz mixer operating in the upper frequency domain of the WR3-band, where the LO signal in the WR6-band is generated directly from a high-power UTC-PD. To our knowledge, this is the first demonstration of an optically pumped heterodyne THz WR3-mixer using a high-power PD as LO source. To eliminate the need for lossy electrical interconnectors and for enabling future packaging of the high-power PDs together with the LBSDs mixers in a common split-block module, the PDs are monolithically integrated with E-plane transitions to couple the output signal from the PD to a rectangular waveguide (WR6). The developments pave the way for high sensitive coherent photonic THz systems, which utilize PDs for both, the generation and detection of the THz signals.
The manuscript is organized as follows, section 2 discusses the proposed integration concept for a future optically pumped THz mixer module. A split-block package is proposed that features a WR3 for an RF input of the THz signals (220-320 GHz), a fiber-optic input for providing the optical LO signal and a coaxial V-type connector for the IF output. Sections 3 and 4 describe the design, fabrication, and experimental characterization of the high-power UTC-PD chips with monolithically integrated GCPW-to-WR6 E-plane transitions for LO frequencies between 140 GHz and 180 GHz. Section 5 describes the measurement setup and experimental characterization of the optically pumped subharmonic THz mixer.
2. Integration concept of a photodiode-pumped subharmonic THz-mixer
Figure 1 shows the proposed integration concept for the novel optically pumped subharmonic THz mixer, where the local oscillator (LO) signal is provided using a high-power photodiode (PD) chip. The PD chip and the low barrier Schottky diodes (LBSDs)-based mixer chip are both integrated in a common split-block package that features a WR3-input. A WR6 is used to interconnect the output of the PD chip to the LO input of the LBSD mixer chip. The module also features an optical fiber input for providing the optical LO signal and a coaxial V-connector for the IF port.
Fig. 1. Integration concept of an optically pumped subharmonic THz mixer. The local oscillator signal is generated by means of a high-power photodiode.
The PD chip is grown on a thin indium phosphide (InP) substrate to suppress the substrate modes. Besides an optimized layer structure for achieving a high saturation output power level within the WR6-band, a passive optical waveguide (POW) is used for high-efficient optical fiber-to-chip coupling. Typically, PDs feature planar grounded coplanar waveguide (GCPW) outputs with an impedance in the range of 30 Ω to 50 Ω. For a wide operational bandwidth (BW) and low electrical coupling losses, the impedance of the PD but also those of the LBSD mixer chip have to be matched to the WR impedance, i.e., transformed from 30-50 Ω up to 377 Ω. To achieve this, the PDs are monolithically integrated with a GCPW-to-WR6 E-plane transition and a stepped-impedance low-pass filter (LPF) for biasing the PD. On the other side, the mixer chip is consisting of an anti-parallel pair of LBSDs that are integrated with a microstrip line (MSL)-based stepped-impedance LPF and two WR-to-MSL E-plane transitions to interconnect the mixer to the WR3-RF and WR6-LO ports. Using LBSDs-based mixer reduces the required LO power for pumping the mixer as compared to the required one for pumping SBDs-based mixers. Generally, the function of the E-plane transition is to convert the dominant mode of the GCPW or the MSL structures to the dominant mode of WR, i.e., from quasi-TEM mode to TE10 mode.
3. High-power photodiode-based local oscillator
3.1 Design and fabrication
To achieve minimum conversion loss and noise-equivalent power (NEP), the utilized PD has to provide an output power above the minimum required LO power for the LBSD THz mixer. For the LBSD THz mixer used in this work, the minimum LO power for WR3-band operation is in the range between 150 µW and 500 µW. To achieve such a high output power over the second-subharmonic frequency range between 135 GHz and 160 GHz (WR6-band), we utilized waveguide-type modified uni-traveling carrier photodiodes (UTC-PDs) [26]. The epitaxial layers structure of the UTC-PDs is summarized in Table 1. A hybrid InGaAs absorber with a depleted and undepleted section was placed as a sandwich between two regions of InP. The highly doped p-type InP layers on top are used as diffusion blocker, while the lower undoped InP layer functions as the depleted collector and cladding for the waveguide. Energy band discontinuities are smoothed by InGaAsP spacer layers at both hetero junctions. For the p- and n-contact layers, p + -InGaAs and n + -InP are used, respectively. Furthermore, the epitaxial layer structure of the UTC-PDs was improved with respect to the internal electrical field management for reducing carrier screening effects and thus achieve a high saturation photocurrent. This allows to achieve a sufficient LO power to operate the LBSDs mixer. The layer structure was further improved to achieve an enhanced evanescent optical coupling from the passive optical waveguide (POW) into the active absorber. This led to an improved spectral responsivity of 0.25 A/W without anti-reflection coating.
Table 1. Epitaxial layers of waveguide modified UTC-PDs.
For fabrication, the epitaxial layer were grown on an iron-compensated InP substrate (Fe-InP). The layer structure consists of an InGaAs p-contact layer followed by an InP layer serving as a diffusion blocker as well as an undepleted and depleted InGaAs absorber layer. For improving the electric field confinement in the carrier transport region, a highly doped InGaAs p-cliff layer, two intrinsic InGaAsP spacer and an InP n-cliff layer were used. The UTC-PDs chips were fabricated using selective wet chemical etching for the p- and n-mesa. For low-loss ohmic metallic p- and n-contacts, Ti/Pt/Au and Ni/Ge/Au layers were deposited, respectively.
3.2 Experimental results
The fabricated UTC-PDs were measured by using a system setup that consisted of two tunable telecom lasers of 1.55 µm wavelength, which allowed to set the heterodyne-generated optical beat frequency over the frequency of LO frequency band. A lensed optical fiber mounted on a fully automated fiber mount was used for efficient fiber-to-chip coupling. An on-chip WR5-probe (FORMFACTOR, model I220-S-GSG-100-BT) together with a power meter (VDI Erickson, model PM5) were utilized to measure the output power of the fabricated UTC-PDs.
Figure 2 shows the measured output power versus the DC photocurrent for various UTC-PDs chips with different p-mesa areas of 7 × 10 µm2, 8 × 10 µm2, 8 × 15 µm2, and 8 × 20 µm2. The measurement was carried out at 150 GHz to study the potential of pumping the LBSDs mixer at 300 GHz. The maximum achieved output power provided by a 7 × 20 µm2 UTC-PD was found to be +3.38 dBm at a photocurrent 21 mA. To our best knowledge, this is the highest output power ever achieved from a PD chip at 150 GHz. The longer UTC-PDs showed slightly lower output power, where the maximum output power of the 8 × 20 µm2 PD area was + 2 dBm at a photocurrent of 23 mA. For the 8 × 15 µm2 and 8 × 10 µm2 PDs, the saturation output power level were +1.12 dBm and +2 dBm, respectively. Despite the fact that the 7 × 20 µm2 PD has a larger RC limitation than the smaller 8 × 10 µm2 or 8 × 15 µm2 PDs, its saturation output power is higher. This is traced back to a more homogeneous optical power distribution in the longer PD, leading to an improved o/e-power conversion. Photodiodes that are shorter than the minimum length required for the light to be coupled from the POW into the absorber, suffer from an inhomogeneous power distribution along the absorber, which leads to a reduced saturation photocurrent. On the other hand, PDs that are longer suffer from increased RC limitations.
The frequency response of the fabricated UTC-PDs was measured at a photocurrent of 6 mA for the entire LO-band (140-220 GHz). As can be seen from Fig. 3, the highest output power was -3.5 dBm at 150 GHz, whereas the minimum one was -11.77 dBm at 210 GHz. Generally, the output power decreases as the frequency increases, due to RC limitation and increasing electrical RF transmission losses. Internal transit frequency limitations were simulated using Silvaco TCAD ATLAS and were numerically calculated to be around 250 GHz and thus are expected to have only a minor impact on the frequency response.
4. Monolithically integrated GCPW-to-WR6 E-plane transition with stepped-impedance low-pass filter on InP substrate
4.1 Design of GCPW-to-WR6 E-plane transition with stepped-impedance LPF chip
The final design of the InP-based GCPW-to-WR6 E-plane transition with a monolithically integrated stepped-impedance microstrip line low-pass filter (MSL-LPF) is shown in Fig. 4. It consists of a GCPW section of a length l1, a MSL section of a length l2, and the E-plane antenna of a length l3. The GCPW length (l1) was intentionally chosen to be relatively long for better mechanical stability and to provide space for DC pins in future packaging. To gradually match the signal width of the GCPW at the output of the UTC-PD to the MSL signal width and match the imaginary part of the UTC-PD’s impedance, the GCPW section features a taper structure at its input. The length of this taper is 190 µm, which approximately corresponds to λg/4 at 150 GHz. The impedance of the GCPW was designed to be 50 Ω. The gap width g, signal width s and all other design parameters are given in Table 2. The InP-substrate thickness was selected to be h = 100 µm to avoid substrate modes within the entire WR6-band. To match the 50 Ω impedance of the MSL to the 377 Ω of the WR6, the signal width s of the MSL is increased in a stepped-manner using a short MSL section of a width w1 in front of the E-plane antenna of a width w2. This stepped-impedance E-plane antenna approach provides an efficient mode conversion from the quasi-TEM mode of the MSL to the TE10 mode of the WR6.
Table 2. Numerical values of the design parameters of the monolithically integrated GCPW-to-WR6 transition with stepped-impedance MSL-LPF on an InP substrate.
The layout of the monolithically integrated stepped-impedance MSL-LPF is also shown in Fig. 4. The filter is employed as an RF-choke to enable biasing the UTC-PD chip on the one hand and to avoid RF leakage into the DC bias supply on the other hand. The key advantages of the stepped-impedance MSL-LPF are its compact size and the high isolation it provides, when compared to similar LPFs using planar stub structures [27].
Fig. 4. Design parameters of the monolithically integrated GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF.
Figure 5 shows the equivalent circuit of the monolithically integrated UTC-PD with GCPW-to-WR6 E-plane transition and stepped-impedance MSL-LPF on an InP substrate. It consists of the UTC-PD (ZUTC-PD = 50 - j . 20 Ω) with a 50 Ω GCPW output (ZGCPW), the impedance of the stepped E-plane antenna (ZE-plane) and a 6th order (N = 6) stepped-impedance MSL-LPF. The filter is connected to a DC supply from one side and to the E-plane transition from the other side. The impedance of the E-plane antenna ZE-plane acts as an impedance transformer between the impedance of the UTC-PD (ZGCPW) and the 377 Ω impedance of the WR6. To achieve an isolation between the RF port and DC port of higher than 20 dB, the length of the high-impedance line has been optimized.
Fig. 5. Equivalent circuit of the monolithically integrated UTC-PD with a GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF.
Maximum isolation was achieved at a line length of 3λg/4. To calculate the physical length and width of the low- and high-impedance MSLs, the characteristic impedance of the filter was chosen to be 50 Ω, while the lowest (capacitor) and highest (inductor) impedance of the low- and high-impedance lines were selected to be 25 Ω and 120 Ω, respectively.
4.2 Packaging concept
To integrate the developed high-power UTC-PD used as LO in a split-block package together with the LBSDs mixer chip, a transition from GCPW-to-WR6 is required. Furthermore, to fulfill the packaging requirements of the proposed subharmonic mixer that is illustrated in Fig. 1, the GCPW-to-WR6 transition has to be inserted into the WR6 and aligned along the direction of the E-field propagation in the WR6, i.e., an E-plane antenna is required.
Figure 6 demonstrates the packaging concept of the proposed GCPW-to-WR6 transition. The substrate material was chosen to be the same as for the UTC-PDs, i.e., InP to enable the monolithic integration of the transition with the UTC-PDs. This is to avoid the use of lossy electrical bonding wires or other high-frequency interconnect technologies. The UTC-PD chip with the monolithically integrated GCPW-to-WR6 transition is placed inside the WR6 through an opened window in its broad-walls at a height t from the bottom wall and a backshort depth d from the closed wall. For low-loss and wideband operation, both, mode conversion from the quasi-TEM mode of the GCPW to the TE10 mode of the WR6 and impedance matching from 50 Ω of the GCPW to 377 Ω of the WR6 have to be performed. The correct positioning of the InP chip in the WR6 is crucial, to avoid the reflection from the opposite WR6’s wall toward the PD chip and to achieve impedance matching within the entire WR6-band. The window’s height t and depth d depend on the WR6’s height a and the guided wavelength λg, respectively. Here, they are t ≈ a/2 and d ≈ λg/4. Finally, to hermitically seal the package and ensure a long life-time for the PD chip, a metallic box is mechanically assembled with the broad-walls of the WR6 at the same position of the opened window.
Fig. 6. Packaging concept of the monolithically integrated GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF.
4.3 Numerical analysis
The S-parameters of the GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF were numerically analyzed within the WR6-band by means of ANSYS Electronics Desktop software. The design parameters of the E-plane transition and the stepped-impedance MSL-LPF, the substrate thickness and the dimensions of the side window were optimized to achieve minimum insertion and reflection losses, a high isolation loss, and a sufficient bandwidth. The numerical values are listed in Table 2.
In the simulation, the input port (GCPW) was excited by means of an active “wave port” with an input impedance of $\textrm{50}\; - \; \textrm{j}\; \cdot {\; 20\; }\mathrm{\Omega }$, while the output port (WR6) and DC port were set to passive “wave port” to collect the excited signal. Furthermore, the properties of the InP substrate, such as the dielectric constant (εr = 12.4) and dielectric loss tangent (tan δ = 0.0012) were considered.
Figure 7 shows the numerically determined S-parameters of the monolithically integrated GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF from 110 GHz to 180 GHz. The simulation results for the insertion loss (IL) (i.e., S21 from the LO to WR6) in Fig. 7. reveal a 1 dB operational bandwidth in the frequency range between 138.8 GHz and 172.1 GHz. The corresponding reflection loss (RL) at the LO input of the transition (S11) and the isolation loss between the input port and the DC port (S31) were less than 10 dB and 30 dB, respectively. The IL was found to be 0.65 dB at 150 GHz and the corresponding S11 and S31 were 15.9 dB and 48.8 dB, respectively.
Fig. 7. Numerical analysis of the monolithically integrated GCPW-to-WR6 E-plane transition with stepped-impedance MSL-LPF.
4.4 Experimental results
For fabricating the monolithically integrated THz UTC-PDs with the designed GCPW-to-WR6 E-plane transition and stepped-impedance MSL-LPF on InP substrate, first the UTC-PDs were fabricated (as discussed in 3.1). The E-plane transition and MSL-LPF were then fabricated using additional metallization steps. Next, the thickness of the InP substrate was reduced down to 100 µm by a mechanical lapping process and finally, the fabricated chips were diced by means of a micro diamond wafer scriber.
To measure the IL of the GCPW-to-WR6 E-plane transition, first the output power of a UTC-PD with a monolithically integrated E-plane transition and stepped-impedance LPF was measured. These results were then compared to the measured ones for a UTC-PD of the same size but without the integrated E-plane transition. Figure 8 shows a photo of the optical heterodyne measurement system that was used to characterize the transition chips. As can be seen from the inset in Fig. 8, the fabricated UTC-PD with integrated E-plane antenna and MSL-LPF was placed on a metallized Rogers laminate serving as the bottom ground and aligned in front of the WR5 facet. Two needles were used for biasing the PD and an adjustable metal plate was used as a backshort. To connect the ground of the PD’s GCPW with the bottom ground (Rogers), the adjustable backshort was carefully connected to the ground DC needle, i.e., the adjustable backshort was considered as a common ground. Note that when the chip is integrated in a real package, the package serves as a common ground. A lensed optical fiber was used to illuminate the UTC-PD and a WR5-SBD (VDI, model WR5.1-ZBD) was utilized to measure the generated RF output power. The WR5-SBD was mounted on a micro-positioner (MPI) to be aligned as close as possible to the GCPW-to-WR6 transition. For determining the IL of the E-plane transition, the RF output power of the UTC-PD without and with integrated GCPW-to-WR6 E-plane transition were measured and plotted in Fig. 9. Both measurements were carried out using an UTC-PD size of 8 × 10 µm2 at a photocurrent of 6 mA and a DC-bias of -2 V. As can be seen from Fig. 9, the maximum achieved output power was found to be - 3.5 dBm and - 6.45 dBm at a frequency of 150 GHz for the UTC-PD without and with monolithically integrated E-plane transition, respectively.
Fig. 8. Photograph of the measurement setup with an enlarged view of the lensed fiber, the adjustable backshort, the DC needles, and the fabricated UTC-PD with a monolithically integrated GCPW-to-WR6 transition and stepped-impedance MSL-LPF.
Fig. 9. Measured RF output power of the fabricated UTC-PD without and with the monolithically integrated E-plane transition and MSL-LPF.
The difference between the two measurements, i.e., the IL of the E-plane transition with LPF is shown in Fig. 10 for the WR6-band. The minimum and maximum de-embedded IL were found to be 2.94 dB and 5.45 dB at a frequency of 150 GHz and 180 GHz, respectively, while the average de-embedded IL was found to be 3.9 dB within the WR6-band. The additional IL can be traced back to the non-perfectly aligned chip with the WR6 aperture and backshort. To study the impact of the misalignment of the chip with the WR5-SBD and backshort, the IL was simulated considering the approximated relative positions of the back-short and WR5-SBD from the chip. In the simulation, the gap between the chip and the backshort as well as between the chip and the WR5 flange was considered to be 0.5 mm and 0.2 mm, respectively. The average resulted IL was found to be 3.04 dB within the WR6-band, as can be seen in Fig. 10. Overall, there is a reasonably good agreement between the simulated IL and the measured IL.
Fig. 10. Simulated and de-embedded insertion loss of the monolithically integrated E-plane transition with the UTC-PD.
5. Optically pumped subharmonic WR3-mixer
As previously mentioned, the aim of this work is to develop high-power PDs serving as LO source for optically pumping a subharmonic THz mixer. Such optically pumped mixer would allow to benefit from the unprecedented low-phase noise and frequency-stability of optically generated LO signals [14] and furthermore, enables the development of optically coherent transponder consisting of a UTC-PD THz transmitter and an optically pumped THz mixer as receiver. Figure 11 and Fig. 12. Show a schematic diagram and a photograph of the measurement setup that was utilized for measuring the conversion loss (CL) of the optically pumped subharmonic WR3-mixer, respectively. The THz mixer used for the experiments is developed by ACST (model 300 GHz-SHM) and it consists of a pair of antiparallel low barrier Schottky diodes which allow to pump the mixer at relatively low LO power between - 8 dBm and - 3 dBm. The packaged mixer features a WR3.4 interface for the RF signals in the frequency range between 270 GHz and 320 GHz and a WR6.5 input for the LO frequencies between 135 GHz and 160 GHz as well as a coaxial SMA connector (DC-18 GHz) for the IF.
Fig. 11. Schematic diagram of the measurement setup of an optically pumped subharmonic WR3-mixer. Fiber connections are shown in red, and electronic connections in blue.
Fig. 12. Photograph of the measurement setup of an optically pumped subharmonic WR3-mixer with an enlarged view of the WR3-mixer, the RF-, LO- and IF-port.
The system setup for measuring the conversion loss of the mixer consists of three free-running tunable lasers (TL) (Pure Photonics, model PPCL700). The polarization of each laser was set by a polarization controller (PC) to align the optical polarization of all three lasers. In the experiments, one laser was used as common optical reference, whereas the other two lasers were set to generate beat frequencies in the WR3-band for the RF and in the WR6-band for the LO. For generating the WR3 RF signal, the optical beating signal was first amplified by an erbium-doped fiber amplifier (EDFA) (KEOPSYS) and then converted into an RF signal using a commercial WR3 PD-module (J-band photomixer module, model IOD-PMJ-13001). The WR6-band LO signal was generated using our own fabricated high-power UTC-PD described in this paper after amplifying the optical beat signal first by using an EDFA (THORLABS, model EDFA100P). The RF signal was directly coupled to the WR3-input of the mixer, while the LO signal from the UTC-PD was coupled to the mixer via an on-chip WR6-probe (FORMFACTOR, model 170-S-GSG-100-BT). To measure the IF output power, an electrical spectrum analyzer (ESA) (HEWLETT PACKARD, model 8565E) was used, where the resolution bandwidth (RBW) of the ESA was fixed at 100 kHz during the measurement.
Figure 13 shows the measured CL for RF frequencies between 301 GHz and 310 GHz. Here, the LO frequency was fixed at 150 GHz, i.e., IF varies from 1 GHz up to 10 GHz. The photocurrents of the LO and RF source were kept constant at 10 mA and 2 mA, respectively. The LO power and RF power were measured using a power meter (VDI Erickson, model PM5). As can be seen from Fig. 13, the measured CL is in the range between 15.7 dB at IF = 1 GHz and 18.7 dB at IF = 10 GHz. The average CL was calculated to be 17.2 dB. Note that RF losses of the WR6 probe (∼2.5 dB) are not de-embedded but are included in the shown CL values. This means, one can expect a CL improvement after packaging the mixer and UTC-PD chips using the presented transition. The variation of the CL of about 3 dB is traced back to the performance of the mixing diodes and the responsivity of the RF source over the frequency, since the LO frequency and LO power level were kept constant during the whole measurement.
Fig. 13. Measured conversion loss of the subharmonic WR3-mixer versus IF at LO frequency of 150 GHz and RF between 301 GHz and 310 GHz.
Next, the CL of the optically pumped subharmonic WR3-mixer was measured versus the RF frequency range between 271 GHz and 321 GHz. For this measurement, the IF frequency was fixed at 1 GHz, i.e., the RF frequency was set to $\textrm{2}\cdot {\textrm{f}_{\textrm{LO}}}\textrm{ + 1}\; \textrm{GHz}$ for each measurement. To investigate the minimum required LO power level for pumping the subharmonic WR3-mixer and study the impact of the LO power level on the CL, the photocurrent (IPH) of the LO source was gradually increased from 2 mA up to 10 mA and the corresponded CL was measured, as shown in Fig. 14. Note that 2 mA was the minimum required photocurrent to measure a CL with the noise level range. It can be concluded from Fig. 14 that the CL is inversely proportional to the LO power level. Here, the CL is maximum at the minimum applied LO power level (at IPH = 2 mA) and decreases when the LO power level increases, until it reaches its minimum values at the maximum applied LO power level (at IPH = 10 mA). The average maximum measured CL (at IPH = 2 mA) and average minimum measured CL (at IPH = 10 mA) were found to be 42 dB and 16.8 dB, respectively. At LO photocurrent of 4 mA, 6 mA, and 8 mA the average measured CL was found to be 28.47 dB, 22.03 dB, and 18.47 mA, respectively. Quantitatively, all the plots of the measured CL versus the RF frequency range have the same behavior, where the CL is minimum at the lowest edge of the RF band (at RF = 271 GHz) and maximum at its highest edge (at RF = 321 GHz).
Fig. 14. Measured conversion loss of the subharmonic WR3-mixer as a function of the UTC-PD photocurrent (LO power level) versus RF.
Figure 15 shows the measured conversion loss of the subharmonic WR3-mixer as a function of the LO power level at LO = 150 GHz and RF = 301 GHz. As can be seen from Fig. 15, when the LO power is increased from -20.45 dBm to -15.37 dBm, the CL is decreased from 44.76 dB to 33.76 dB, i.e., an increased LO power of about 5 dB corresponds to an improved CL of about 11 dB. Similarly, when the LO power is increased from -15.37 dBm to -9.35 dBm, the corresponding CL is decreased from 33.76 dB to 20.76 dB. At LO power of -11.9 dBm, a reasonable CL of 22.76 dB was achieved. The minimum CL of 16.76 dBm was measured at the maximum LO power of -7.37 dBm. Even there was still a room to increase the LO power level, higher LO power was not applied to avoid damaging the mixer.
Fig. 15. Measured conversion loss of the subharmonic WR3-mixer as a function of the LO power level at LO frequency of 150 GHz and RF of 301 GHz.
6. Conclusion
This manuscript presented a packaging concept and technological advancements for developing an optically pumped THz mixer using a high-power uni-travelling carrier photodiode (UTC-PD) to provide the required LO power. To our knowledge, this is the first demonstration of an optically pumped THz mixer using a UTC-PD. To meet the mixer’s LO power requirements, high-power UTC-PDs for WR6-band (110-170 GHz) operation were developed and experimentally characterized. At 150 GHz, the fabricated 7 × 20 µm2 UTC-PD delivered a maximum RF output power of + 3.38 dBm at a photocurrent of 21 mA. According to our knowledge, this is the highest ever reported output power at 150 GHz for a UTC-PD. Furthermore, a novel packaging approach was presented. It is based on a monolithic integration of the indium phosphide (InP)-based high-power UTC-PDs with a GCPW-to-WR6 E-plane transition and a stepped-impedance microstrip line low pass filter (MSL-LPF) for biasing the UTC-PD. The aim of the monolithic transition is to enable the integration of both chips, the PD and the mixing diode chips in a common split-block package. The fabricated InP-based transitions with monolithically integrated UTC-PDs exhibited an average insertion loss of 3.9 dB for an operating frequency range from 140 GHz to 180 GHz. Finally, a subharmonic WR3-mixer operating within the frequency range between 271 GHz and 321 GHz was optically pumped using the fabricated high-power UTC-PDs. The conversion loss (CL) as a function of the LO power was investigated by pumping the subharmonic WR3-mixer at different LO power levels. The experimental results exhibited an improved CL with increased LO power levels. The average minimum measured CL for IF of 1 GHz was found to be 16.8 dB at the maximum employed LO power level corresponding to a photocurrent of 10 mA. After packaging the mixer and the UTC-PD using the fabricated GCPW-to-WR6 transition, the LO power is expected to be improved by about 2.5 dB and correspondingly the CL.
The achievements presented in this paper are considered as a fundamental technological step towards coherent THz transponder for frequency domain THz systems in which photodiodes are not only used to generated the THz signal but also to provide the LO signal for pumping a coherent THz receiver. In contrast to all electronic THz transponder, the optical receiver presented here is expected to benefit from ultralow-low phase noise performance of optically generated LO signals using optical combs.
Funding
Bundesministerium für Wirtschaft und Energie(ZF4047824PR9); NRW/EFRE Terahertz-Integrationszentrum (THzIZ) (EFRE-0400215); Deutsche Forschungsgemeinschaft (287022738– CRC/TRR 196); Bundesministerium für Bildung und Forschung (16KISK017, 16KISK039).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. Y.-J. Lin and M. Jarrahi, “Heterodyne terahertz detection through electronic and optoelectronic mixers,” Rep. Prog. Phys. 83(6), 066101 (2020). [CrossRef]
2. A. Rogalski and F. Sizov, “Terahertz detectors and focal plane arrays,” Opto-Electron. Rev. 19(3), 346–404 (2011). [CrossRef]
3. R. Lewis, “A review of terahertz detectors,” J. Phys. D: Appl. Phys. 52(43), 433001 (2019). [CrossRef]
4. M. Tani, Y. Hirota, C. T. Que, S. Tanaka, R. Hattori, M. Yamaguchi, S. Nishizawa, and M. Hangyo, “Novel terahertz photoconductive antennas,” Int. J. Infrared Millimeter Waves 27(4), 531–546 (2006). [CrossRef]
5. S. Hargreaves and R. Lewis, “Terahertz imaging: materials and methods,” J. Mater. Sci.: Mater. Electron. 18(S1), 299–303 (2007). [CrossRef]
6. N. Karpowicz, H. Zhong, J. Xu, K.-I. Lin, J.-S. Hwang, and X.-C. Zhang, “Non-destructive sub-THz CW imaging,” Proc. SPIE 5727, 132–142 (2005). [CrossRef]
7. A. W. Lee, Q. Qin, S. Kumar, B. S. Williams, Q. Hu, and J. L. Reno, “Real-time terahertz imaging over a standoff distance (> 25 meters),” Appl. Phys. Lett. 89(14), 141125 (2006). [CrossRef]
8. S. Nadar, M. Zaknoune, X. Wallart, C. Coinon, E. Peytavit, G. Ducournau, F. Gamand, M. Thirault, M. Werquin, and S. Jonniau, “High performance heterostructure low barrier diodes for sub-THz detection,” IEEE Trans. Terahertz Sci. Technol. 7(6), 780–788 (2017). [CrossRef]
9. J. Li and J. Li, “Terahertz (THz) generator and detection,” Elect. Sci. Eng. 2 (2020).
10. A. Mobini and M. Solaimani, “A quantum rings based on multiple quantum wells for 1.2–2.8 THz detection,” Physica E: Low-dimensional Systems and Nanostructures 101, 162–166 (2018). [CrossRef]
11. H.-W. Hubers, “Terahertz heterodyne receivers,” IEEE J. Sel. Top. Quantum Electron. 14(2), 378–391 (2008). [CrossRef]
12. Y. Nishida, N. Nishigami, S. Diebold, J. Kim, M. Fujita, and T. Nagatsuma, “Terahertz coherent receiver using a single resonant tunnelling diode,” Sci. Rep. 9(1), 1–9 (2019). [CrossRef]
13. A. Maestrini, B. Thomas, H. Wang, C. Jung, J. Treuttel, Y. Jin, G. Chattopadhyay, I. Mehdi, and G. Beaudin, “Schottky diode-based terahertz frequency multipliers and mixers,” C. R. Phys. 11(7-8), 480–495 (2010). [CrossRef]
14. K. Saleh, R. Henriet, S. Diallo, G. Lin, R. Martinenghi, I. V. Balakireva, P. Salzenstein, A. Coillet, and Y. K. Chembo, “Phase noise performance comparison between optoelectronic oscillators based on optical delay lines and whispering gallery mode resonators,” Opt. Express 22(26), 32158–32173 (2014). [CrossRef]
15. B. S. Williams, “Terahertz quantum-cascade lasers,” Nat. Photonics 1(9), 517–525 (2007). [CrossRef]
16. R. Köhler, A. Tredicucci, F. Beltram, H. E. Beere, E. H. Linfield, A. G. Davies, D. A. Ritchie, R. C. Iotti, and F. Rossi, “Terahertz semiconductor-heterostructure laser,” Nature 417(6885), 156–159 (2002). [CrossRef]
17. S. Verghese, E. K. Duerr, K. McIntosh, S. M. Duffy, S. D. Calawa, C.-Y. Tong, R. Kimberk, and R. Blundell, “A photomixer local oscillator for a 630-GHz heterodyne receiver,” IEEE Microw. Guid. Wave Lett. 9(6), 245–247 (1999). [CrossRef]
18. I. Cámara Mayorga, P. Muñoz Pradas, E. Michael, M. Mikulics, A. Schmitz, P. van der Wal, C. Kaseman, R. Güsten, K. Jacobs, and M. Marso, “Terahertz photonic mixers as local oscillators for hot electron bolometer and superconductor-insulator-superconductor astronomical receivers,” J. Appl. Phys. 100(4), 043116 (2006). [CrossRef]
19. I. C. Mayorga, A. Schmitz, T. Klein, C. Leinz, and R. Gusten, “First in-field application of a full photonic local oscillator to terahertz astronomy,” IEEE Trans. Terahertz Sci. Technol. 2(4), 393–399 (2012). [CrossRef]
20. X. Xie, R. Bouchand, D. Nicolodi, M. Giunta, W. Hänsel, M. Lezius, A. Joshi, S. Datta, C. Alexandre, and M. Lours, “Photonic microwave signals with zeptosecond-level absolute timing noise,” Nat. Photonics 11(1), 44–47 (2017). [CrossRef]
21. J. L. Kloosterman, D. J. Hayton, Y. Ren, T.-Y. Kao, J. Hovenier, J.-R. Gao, T. M. Klapwijk, Q. Hu, C. K. Walker, and J. L. Reno, “Hot electron bolometer heterodyne receiver with a 4.7-THz quantum cascade laser as a local oscillator,” Appl. Phys. Lett. 102(1), 011123 (2013). [CrossRef]
22. S. Shiba, Y. Irimajiri, T. Yamakura, H. Maezawa, N. Sekine, I. Hosako, and S. Yamamoto, “3.1-THz heterodyne receiver using an NbTiN hot-electron bolometer mixer and a quantum cascade laser,” IEEE Trans. Terahertz Sci. Technol. 2(1), 22–28 (2012). [CrossRef]
23. J. Gao, J. Hovenier, Z. Yang, J. Baselmans, A. Baryshev, M. Hajenius, T. Klapwijk, A. Adam, T. Klaassen, and B. Williams, “Terahertz heterodyne receiver based on a quantum cascade laser and a superconducting bolometer,” Appl. Phys. Lett. 86(24), 244104 (2005). [CrossRef]
24. M. C. Wanke, E. W. Young, C. D. Nordquist, M. J. Cich, A. D. Grine, C. T. Fuller, J. L. Reno, and M. Lee, “Monolithically integrated solid-state terahertz transceivers,” Nat. Photonics 4(8), 565–569 (2010). [CrossRef]
25. A. Stöhr, R. Heinzelmann, K. Hagedorn, R. Güsten, F. Schäfer, H. Stüer, F. Siebe, P. Van der Wal, V. Krozer, and M. Feiginov, “Integrated 460 GHz photonic transmitter module,” Electron. Lett. 37(22), 1347–1348 (2001). [CrossRef]
26. V. Rymanov, B. Khani, S. Dülme, P. Lu, and A. Stöhr, “InP-based waveguide triple transit region photodiodes for hybrid integration with passive optical silica waveguides,” in Photonics, (MDPI, 2015), 1152–1163.
27. D. M. Pozar, Microwave engineering (John wiley & sons, 2011).
