*3.1. Measurement Setup*

The setup used to measure the performance of the proposed detector is illustrated in Figure 8. The 200-GHz signal generated by the gyrotron source demonstrates Gaussian beam characteristics, whose focal plane is aligned by off-axis parabolic (OAP) mirrors [20] with a focal length of 15.24 cm. A physical chopper located in the focal plane reduces the flicker noise by transmitting the detected output DC voltage along with an AC signal. The measurements were conducted with a modulation frequency of 200 Hz for avoiding the effect of switching noise from the power supply. To monitor the constant voltage output of the LDO regulator, digital 4-bit control signals were applied using the National Instruments (NI) data acquisition board (DAQ). Before reaching the monitoring equipment, the detector IC output was amplified with a gain of 5 through a bandpass filter of 100–300 Hz using an SR560 low-noise voltage amplifier manufactured by Stanford Research Systems. *RV* was measured using an oscilloscope, and *NEP* was measured using the Keysight N9010B signal analyzer. The unamplified detector performance was measured by dividing the measurement with the amplifier gain.

**Figure 8.** Measurement setup to analyze the performance of the proposed detector IC. An oscilloscope to measure *RV* and a signal analyzer to obtain noise-equivalent power have been used.

In the THz imaging system test, as shown in Figure 9a, the distance between the mirror and the sample was 420 mm, and that between the sample and the proposed CMOS detector IC was 40 mm; the sample was placed on the XY stage. During image acquisition, the DAQ generated a digital 4-bit control code and analyzed the measurement data. The output signals were acquired using the NI DAQ hardware and NI LabVIEW software tools. The sample was moved at intervals of 1 mm in the measurement environment, as shown in Figure 9b, and the final output image was obtained using 2-D raster scanning.

**Figure 9.** Measurement setup: (**a**) block diagram for THz imaging using raster scanning; (**b**) experimental setup.

#### *3.2. Proposed Detector IC Performance*

The performance of sub-THz CMOS detectors is determined by *RV* and *NEP*. *RV* is defined as the change in the output voltage based on the presence or absence of the incident signal with the specific power applied to the detector; it is calculated as:

$$R\_V = \frac{V\_{OUT} - V\_{DCOFF}}{P\_{IN}} = \frac{V\_{OUT} - V\_{DCOFF}}{P\_D \cdot A\_{EFF}} \text{ [ $V/W$ ]},\tag{1}$$

where *VOUT* is the output voltage when the input power *PIN* is applied to the detector, *VDCOFF* is the output voltage without the incident signal, *PD* is the power density incident to the detector IC, and *AEFF* is the effective antenna area, which includes the integrated antenna and wavelength characteristics [9,13]. *NEP* is defined as the input power level that becomes equal to the noise generated from the detector itself; it is expressed using *RV* as:

$$NEP = \frac{\sqrt{N\_V}}{R\_V} \left[ W / \sqrt{Hz} \right]\_{\prime} \tag{2}$$

where *NV* denotes the noise spectral density [9].

When using a receiver antenna to transmit a THz signal to a detector, it is vital to determine the characteristics of the detector antenna and consider the input power equation of the detector to accurately analyze and measure the detector performance. The

unit power density of the gyrotron measured at the detector position is 0.5 W/m2, and the effective area considering the receiving antenna gain at 200 GHz is 9.62 × 10−<sup>8</sup> m<sup>2</sup> [13]. The input power was calculated based on the measured performance, considering the difference in antenna gain according to the radiation area and similar antenna simulation values [21]. Figure 10 shows the measured *RV* and *NEP* values of the proposed CMOS detector with different gate bias voltages. The results exhibit high *RV* and low noise when the gate bias is lower than the threshold voltage. *RV* and *NEP* were 14.13 MV/W and 34.42 pW/ √Hz, respectively, under the gate bias condition of 150 mV. Table 1 lists the performance comparison of the proposed CMOS detector with previously developed detector core configurations. The proposed CMOS detector exhibits higher *RV* compared with the other detectors. As a result of comparing detectors with the same minimum gate length, the proposed detector showed the lowest *NEP* and highest *RV*. In previous studies, while calculating the effective area of an antenna, the difference in the radiation area between the antenna simulations and measurements was not considered. The performance of the proposed detector was calculated using the simulation data of the integrated antenna, which includes the ground guard ring in the simulation model for providing the same radiation area as the fabricated IC.

**Figure 10.** Measurement results of the voltage responsivity and the noise-equivalent power using the proposed CMOS detector IC at 200 GHz.



1 Measured in a Faraday cage.

#### *3.3. Images Obtained Using the Proposed Detector IC*

Copper foil tapes of different thicknesses were placed on the Styrofoam substrate to measure the resolution of the proposed detector, as shown in Figure 11a. The sample size was 50 mm × 50 mm. Considering that the wavelength of 200 GHz in the air is 1.5 mm, the sample was manufactured considering a thickness of ≥2 mm. The real sample was digitized, as shown in Figure 11b, for a digital area comparison using MATLAB. The sub-THz imaging at 200 GHz yielded results that are 63.6% identical to those of the digitalized sample. As shown in Figure 12, imaging results obtained using the proposed detector demonstrated that a 2-mm thick conductive target could be distinguished from the background. The measurement image is more distributed than the physical sample, as the passing waves are dispersed over the distance of 40 mm between the detector and sample. Considering the difficulty in identifying the wavelength width of the metal using a CMOS detector owing to the distance between the sample and detector, the measurement result exhibits a high-resolution image. All the images were compared by normalizing them to either maximum or minimum ratios.

**Figure 11.** Sample target for sub-THz imaging: (**a**) photograph of a sample with different copper widths; (**b**) digitalized image of the sample to image correlation.

**Figure 12.** Measurement image using the proposed CMOS detector IC.

As illustrated in Figure 13a, the sample is used to compare the effect of the difference in the detector core circuit, and the individual detectors are considered under the optimal detection performance conditions [23]. Figure 13b shows the image obtained using the model capacitor, which is supported by the process design kit (PDK). Figure 13c shows an image obtained from a detector designed to achieve optimal detection performance using a customized capacitor in the same core circuit structure. The proposed detector, as shown in Figure 14, exhibited a 59.37% match with the normalized image, whereas previous studies demonstrated correlations of 41.4% and 53.7%. The concurrent-mode detector containing the cross-coupled capacitors better resolves the inner plus-shaped copper foil, which was impossible to identify in previous studies. The image results show that the image SNR

of a single frame is 49 dB, which is 9 dB higher than that of the image obtained using the previous detector in one frame.

**Figure 13.** Sub-THz imaging in the previous study [23]. (**a**) Photograph of a real sample; (**b**) digitized image; (**c**) using a common-source detector circuit with the standard capacitors in the process design kit; (**d**) using a common-source detector circuit with the customized capacitors.

**Figure 14.** Imaging measurement using the proposed concurrent-mode detector at 200 GHz.
