Terahertz (THz) spectroscopy has emerged as a powerful tool for nondestructive inspection applications, such as biomedical diagnosis, device inspection, and characterization of materials [1–12]. In particular, THz imaging allows for the characterization of the internal structures of objects, which is possible because nonconducting materials are transparent to THz waves [13–23]. Imaging using continuous-wave (CW) sources has increased in popularity, owing to recent technological advances in the development of sources and detectors [24–32]. Although coherent THz tomography based on frequency-modulated CW (FMCW) methods has been developed using tunable or ensemble sources, it is limited in terms of the general depth and longitudinal resolution [33–37]. In contrast, the use of time-domain spectroscopy (TDS) enables the collection of information such as the depth and range of objects with superior resolution, without the need to sweep the frequency [6, 38, 39]. In other words, THz pulsed imaging provides direct 3D mapping of an object by using the time of flight (TOF) of the reflected THz pulses. Importantly, the longitudinal resolution of a THz-TOF imaging system is determined by the temporal width of the pulsed THz source, which yields resolution as good as 150 μm.

Recently we introduced a high-speed nondestructive THz-TOF imaging technique for the inspection of packaged integrated circuits and metal corrosion beneath paint layers [38, 39]. An imaging rate of 100 Hz/pixel has been achieved based on the THz–TDS technique using the optical sampling by cavity tuning (OSCAT) system [40, 41], which delivers a rapid time delay with a full range of 50 ps. The OSCAT system uses a single pulsed fiber source split into two pulses, used for the THz antenna and receiver respectively, and the rapid time delay between the two pulses is generated by varying the repetition rate of the fiber laser. Although the OSCAT system delivers stable THz–TDS signals with relatively low jitter between the two pulses, the range of the time delay is limited, especially for fast scanning (>100 Hz), because it is based on a single laser source. In addition, because the time delay varies sinusoidally, the linear-response range reaches only approximately 20 ps; this corresponds to an optical path length of 3 mm. Therefore, a THz-TOF system based on OSCAT methods is restricted in nondestructive imaging applications that require large depth profiling.

In contrast, THz-TDS systems with asynchronous optical sampling (ASOPS) use two fiber lasers, with repetition rates that can be adjusted independently [42, 43]. In this case, the range of the time delay is limited by the time window between consecutive pulses; thus the time delay reaches 10 ns for a laser repetition rate of 100 MHz. Conversely, the A-scan rate (100–500 Hz) is determined by the repetition-rate difference between the two fiber lasers. This has numerous advantages, because 3D TOF images with virtually unlimited depth range can be obtained. However, it is still restricted by the optical geometry of the imaging system, such as the depth of focus of the region of interest. THz-TOF imaging with the electronically controlled optical sampling (ECOPS) technique has also been introduced, using two independent femtosecond lasers; however, it allows a finite range of time delay, as in the case of the OSCAT system [44, 45].

In this study, we report rapid THz-TOF imaging results based on the ASOPS technique, which delivers a large vertical scan range of more than 30 mm when a focused lens of 50 mm is used. The lateral resolution is addressed as a function of the longitudinal location of the object. Further, we obtain nondestructive TOF images of metallic objects embedded in insulating enclosures, such as polyethylene foam and chocolate.

A schematic of the THz-TOF imaging equipment is illustrated in Fig. 1. For the rapid THz–TDS system we use a commercial ASOPS instrument (TERA-ASOPS; Menlo Systems GmbH, Planegg, Germany) consisting of two independent femtosecond fiber lasers with a wavelength of 1,560 nm, operating with a repetition rate of 100 MHz. The ASOPS system allows the full time-delay range of 10 ns with a rate of 100–500 Hz; we remained fixed at 100 Hz throughout the experiments. In contrast, in the OSCAT system the time-delay scan range is limited to 50 ps, and in particular the region of linear response is restricted to 20 ps, as mentioned previously. Therefore, THz-TOF with ASOPS has a great advantage in terms of large depth profiling.

Figure 1. Experimental setup: A schematic illustration of the THz- time of flight (TOF) imaging setup based on the asynchronous optical sampling (ASOPS) method.

We incorporate a galvano scanner (Scanlab GmbH, Puchheim, Germany), which enables us to obtain images with a speed that is limited by the ASOPS THz-TDS system, whereas the combined galvano-scanning (x-axis) and slow-stage-scanning (y-axis) method can be used for larger samples. The THz signal is focused using a convex lens with a focal length of 5 cm and a diameter of 3.8 cm. The reflected THz signal is collected using a 50:50 beam splitter (Tydex Inc., St. Peterburg, Russia) that is fed into the THz receiving antenna (denoted as Rx). The current signal from the antenna is amplified using a fast current preamplifier (FEMTO Messtechnik GmbH, Berlin, Germany) and digitized with a data acquisition card (Alazar Tech. Inc., Quebec, Canada) with a rate of 10 MHz, allowing us to obtain a time-delay step of 0.1 ps. The collected data contains the phase-sensitive THz amplitudes as a function of the 3D parameters, such as the time delay and x-axis and y-axis positions. We apply point averaging to improve the signal-to-noise ratio, leading to a measurement rate of 5–10 Hz/pixel, while the system delivers 100 Hz/pixel. The images are recorded as a single binary file, which can be reconstructed using home-built analysis software. The phase-sensitive THz amplitudes can be converted into the envelope signal by Hilbert transformation; in other words, from the complex THz signal ${\tilde{E}}_{THz}=\mathrm{Re}\left({\tilde{E}}_{THz}\right)+iH\left[\mathrm{Re}\left({\tilde{E}}_{THz}\right)\right]$, where H is the Hilbert transformation [46].

We first characterize the focused beam, to address the vertical extent available for THz-TOF imaging. As shown in Fig. 2(a), we use a metal plate with a knife edge to establish the lateral resolution of the system. We obtain THz images for different heights z of the metallic plate, from z = +25 to z = −25 mm, where z = 0 indicates the location of the focus plane, 5 cm below the lens. Figure 2(b) shows a series of A-scan data for the three different z locations of the metal plate. We use a delay-scan range of ∆T = 500 ps, which corresponds to a vertical range of 75 mm in the case of free-space propagation. Here the time delay can be converted into depth information, in which 1 ps corresponds to 0.15/n mm, where n is the refractive index of the material [38, 39]. The position of the main peak varies according to different z locations on the reflecting surface. We note that a series of echoed signals appears with an interval of 77.6 ps; these may be inherent to the ASOPS system, because they exist even when the reflective optics are used for testing, instead of lenses. They can interfere with the TOF measurement if the optical path length in terms of the vertical range corresponds to a time delay larger than 77 ps. However, most of our experiments are not greatly affected because the echoes are weaker than the primary reflection.

Figure 2. Spatial resolution as a function of longitudinal position. (a) A schematic for addressing the resolution in the THz imaging setup. (b) Plots of the phase-sensitive THz reflection amplitude as a function of time-delay, for different vertical locations z of the metal plates. (c) THz signal (black squares) as a function of position x when the metal plate is located at z = 0 mm. The blue solid line is the curve fitted using the error function. (d) The spatial resolution w as a function of the vertical position z of metal plate, extracted from the fitting result in (c). Plotted as black squares is the strength of THz amplitude as a function of z for the reflection at the metal surface.

Figure 2(c) illustrates the line profile of the reflected amplitude signal as a function of lateral position x while the vertical location is fixed at z = 0 mm, as schematically illustrated in Fig. 2(a). The reflection signal is high when it is reflected by the metal plane (x < 3.87 mm), whereas it is suppressed when the focused beam is positioned outside of the metal (on the right side of the knife edge). The line profile is fitted with the error function to obtain the resolution w of the focused THz beams [47]. The resolution of the reflected THz pulse is 0.79 mm, in terms of the amplitude. The extracted resolution as a function of the vertical position z of the metal plane is plotted in Fig. 2(d). The minimum resolution is achieved at approximately z = 0 mm, whereas it increases when the metal plane is away from the focal plane. For instance, it increases up to w = 1.75 mm for z = 25 mm. In other words, we can readily achieve 3D TOF imaging with a depth range of 50 mm with a spatial resolution in the range of 0.8–1.8 mm, in terms of the THz amplitude. Conversely, the strength of the reflected THz amplitude [shown as black squares in Fig. 2(d)] decreases as the sample plane moves away from the focal plane, because the reflected beam becomes out of focus at the position of the Rx antenna.

Figure 3 shows THz-TOF results to prove the usefulness of the ASOPS system. The test sample is a metal block with three rectangular holes of lateral size 10 mm × 10 mm and depth 11.5 mm, as shown in Fig. 3(a). In Fig. 3(b), the reflected THz signals are shown as a function of the lateral position (C-scan), with a scan range of 15 × 45 mm² and a pixel number of 50 × 150. Here we show the time-integrated THz amplitudes; in other words, the THz envelope signal (obtained via Hilbert transformation) has been time-integrated for each pixel over the entire time-delay range of 120 ps. This corresponds to an optical path length of 18 mm. We clearly identify in Fig. 3(b) the bright regions for both the top (green) and bottom (red) metal surfaces, whereas the dark region (blue) appears in the middle because of the presence of a circular hole. The bottom surface is brighter than the top surface because the focal plane is closer to the bottom layer. We also note that reflection is suppressed when the reflecting surface is inclined, as shown at the periphery of the circular hole [the region enclosed by the dashed orange line in Fig. 3(a)].

Figure 3. Time of flight (TOF) imaging of a layered metallic structure with a large depth range. (a) Photograph of a metal sample containing square holes with a depth of 11.5 mm. (b) Time-integrated C-scan image (in terms of x- and y-axes). (c) Cross-sectional B-scan image (THz reflection amplitude in terms of y and T). (d) A reconstructed 3D image of the sample. The color scale represents the time-delay T, different from that in (b).

THz–TDS allows us to address the depth profiling directly from the B-scan images, as shown in Fig. 3(c). The THz signal is plotted as a function of position y [along the dashed line in Fig. 3(b)] and time delay T. The THz reflection is dominated by the signal from the top (12 ps) and bottom (88.5 ps) metal surfaces, whose difference in terms of vertical location is consistent with that expected from the rectangular-hole depth (11.5 mm). The image is reconstructed in 3D by plotting the peak amplitude as a function of x, y, and T, and the results are shown in Fig. 3(d). Here the color scale indicates the position z at which the reflection occurs. Both the top (blue) and bottom (red) surfaces are clearly identified, even when displaced more than 10 mm from each other.

Figure 4 shows the THz-TOF imaging result for one of the test samples, in which metallic objects are embedded in a polyethylene-foam enclosure. As shown in the photograph in Fig. 4(a), the test sample consists of a series of five metal blades (size 10 × 4.5 mm²) stacked vertically and displaced by 5 mm with respect to one another. They are placed between sheets of low-density-polyethylene foam (thickness 5 mm). The size of the enclosure surrounding the metal blades is 25 mm × 45 mm × 30 mm. We record THz imaging with the reflected signals as a function of the position (x and y) and time-delay (T), with a scan range of 15 × 50 mm² and a pixel number of 60 × 200. The time-delay is 200 ps, which corresponds to an optical path length of 30 mm. Figure 4(b) shows the time-integrated C-scan data, in which the THz envelope signal is time-integrated in each pixel. We are able to clearly identify the images of the five metal blades; however, the signal is relatively weak for the blades that are inclined with respect to the incident THz waves. The signals are reconstructed as a 3D image, as shown in Fig. 4(c). Our system identifies the locations of the five blades successfully, both in the horizontal and vertical directions. We also note that the refractive index of the polyethylene foam is close to unity, because of its porous structure; hence 10 ps corresponds to 1.5 mm in terms of z position.

Figure 4. Imaging of metallic blades embedded in polyethylene foam. (a) Photograph of the test sample, in which five metal blades at different z locations are embedded in the polyethylene-foam enclosure. (b) The time-integrated C-scan image (in terms of envelope function) for (a). (c) A reconstructed 3D image of the sample.

Recently there has been increasing demand for finding conducting and nonconducting foreign substances in food via THz imaging with CW THz sources [29–32]. This has been mostly performed in transmission geometry; therefore, locating these substances in terms of the distance from the surface has not been accomplished. We first show the transmission imaging results as a C-scan in Fig. 5(a) (top) for thick chocolate (4.5 mm) in which a metal blade has been intentionally embedded in the middle, by melting the chocolate. The THz transmission image is measured using a focusing lens of 5 cm focal length, by moving the motorized stage containing the chocolate sample in both x and y directions. The dark area in the middle of the image indicates the presence of a metallic object blocking the THz transmission. On the other hand, no noticeable indication of a foreign object is observed in the B-scan image [bottom of Fig. 5(a)] taken along the dashed line in the C-scan image. Therefore, locating objects in terms of the vertical position is restricted in transmission geometry.

Figure 5. Time of flight (TOF) imaging of a metallic object embedded in a chocolate bar. (a) C-scan image (top) for the THz transmission amplitude through a chocolate sample with a metal blade embedded in the middle. B-scan image (bottom) along the dashed line in the C-scan image. (b) C-scan image (top) for the THz reflection amplitude through the sample. B-scan image (bottom) along the dashed line in the C-scan. (c) Photograph of the chocolate sample. (d) Reconstructed 3D image of the chocolate sample.

Conversely, in reflection geometry, which allows the 3D mapping of the object, we can identify the metal structure in the B-scan image [bottom of Fig. 5(b)], whereas it is not identified in the time-integrated C-scan image [top of Fig. 5(b)]. This is partly because the reflected signal from the metal blade interferes with the engraved letters on the top surface. The metal blade is found approximately 30 ps from the top surface of the chocolate, which indicates that it is located 2.5 mm beneath the surface (considering that the refractive index of the chocolate is 1.8). This result is consistent with the actual location of the blade in the chocolate, confirmed by the photograph in Fig. 5(c), taken after the chocolate has been cracked. The 3D plot shown in Fig. 5(d) clearly demonstrates the location of the metal blade in both horizontal and vertical directions; this means that the TOF image allows one to locate a foreign object in terms of both horizontal and vertical positions. We note that the reflection image can be obtained for a host material with relatively low attenuation, such as chocolate, which has low water content [48, 49]. In addition, the detection of a metallic object is restricted when the object is inclined with respect to the surface, as in the case of reflection geometry.

We have developed a 3D THz time-of-flight imaging technique based on the ASOPS method, which allows us to address large vertical ranges. For a focal length of 5 cm, we measured the spatial resolution for different vertical locations; the resolution reached 0.8 mm at the focal plane, and increased up to 1.8 mm when the reflecting surface was 25 mm away from the focal plane. We obtained THz-TOF images for samples with multiple reflection layers; the horizontal and vertical locations of the objects could be successfully determined from 2D cross-sectional images or from reconstructed 3D images. We successfully identified metallic objects embedded in insulating enclosures with a vertical depth range greater than 30 mm. In addition, we used our technique to locate a metallic object in a thick chocolate bar, which is not accessible via the conventional transmission geometry. Furthermore, the THz-TOF system has proven to be an effective candidate for finding foreign substances in foods such as thick chocolate bars; in other words, we can successfully locate a metallic object in terms of both horizontal and vertical positions. Therefore, THz-TOF imaging based on the ASOPS system will work as a powerful means of nondestructive testing with a large depth range, for applications in the fields of security, foods, packaged microchips, and infrastructure.

The authors declare no conflict of interest.

Data underlying the results presented in this paper are not publicly available at the time of publication, but may be obtained from the authors upon reasonable request.

Midcareer Researcher Program (2020R1A2C1005735); and Basic Science Research Program (2021R1A6A1A1 0044950) through a National Research Foundation grant of Korea Government; GRRC Program (GRRCAJOU2022B01, Photonics-Medical Convergence Technology Research Center) of Gyeonggi province.