This study presents an optical frequency switching scheme for a high-speed broadband terahertz (THz) measurement system based on the photomixing technique. The proposed system can achieve high-speed broadband THz measurements using narrow optical frequency scanning of a tunable laser source combined with a wavelength-switchable laser source. In addition, this scheme can provide a larger output power of an individual THz signal compared with that of a multi-mode THz signal generated by multiple CW laser sources. A swept-source THz tomography system implemented with a two-channel wavelength-switchable laser source achieves a reduced time for acquisition of a point spread function and a higher depth resolution in the same amount of measurement time compared with a system with a single optical source.
© 2017 Optical Society of America
Recently, terahertz (THz) radiation has received much attention due to its unique features. THz radiation can penetrate dielectric substances that are opaque to light waves . Compared with radio frequency (RF) and microwave signals, THz radiation has a much shorter wavelength which enables a higher spatial resolution for the characterization of substances . Unlike x-rays which are harmful to biological tissues due to their high ionization energy, THz radiation is less harmful due to its lower energy . For these reasons, application of THz technologies to THz spectroscopy, THz vibrometers, and THz tomography systems has been actively investigated [4–6]. In particular, THz tomography and 3D THz imaging have attracted much interest in various fields.
In THz tomography and 3D THz imaging systems, the THz sources can be categorized into pulsed type and continuous wave (CW) type THz sources. The pulsed type THz source, in which a photoconductive antenna is excited by a femtosecond laser, provides a very wide spectral coverage up to a few THz, but its industrial utilization is hindered by its bulky size and high system costs . On the other hand, the CW type THz source is more compact and cost-effective than that of the pulsed THz source . Among the many CW THz generation techniques, the photomixing technique, in which a CW THz signal is generated through the beating of two CW laser sources, provides a high dynamic range, high signal-to-noise ratio, narrow spectral linewidth, and wide frequency tunability required for broadband spectroscopy . It is also a potential candidate for the implementation of a compact THz measurement system using silicon photonics technology .
For broadband measurements using CW THz radiation, a multi-mode THz source and a frequency swept-THz source are used in a photomixing-based THz system. The multi-mode THz source generated by multiple CW laser sources, which is called the quasi TDS (Time-domain spectroscopy) approach, has an extremely low cost compared with the pulsed THz source using a femtosecond laser [11,12]. However, its frequency resolution is limited by the modes of the laser sources, typically in the range of several Gigahertz. In addition, it requires the scanning of a mechanical delay line with an inevitable increase in the complexity and size of the system , and its output power of the individual THz frequency component decreases sharply as a function of the number of modes due to the limited maximum optical input power of the photomixer. In contrast, the use of a frequency swept-THz source based on photomixing provides a higher THz output power and a good frequency resolution as small as a few megahertz .
The frequency swept-THz source suffers from the disadvantage of a long frequency sweep time due to the limited frequency sweep speed of the laser sources . Especially, the frequency sweep time is a critical issue in a THz tomography system requiring high-resolution measurements because its theoretical depth resolution is inversely proportional to the measurement bandwidth [14,15]. In a THz tomography system with a long frequency sweep time, the data acquisition time is prohibitively increased by repetitive measurements to construct a 3D image.
In this paper, an optical frequency switching scheme is proposed to reduce the data acquisition time for a photomixing-based THz application requiring broadband measurements. The proposed scheme uses a tunable laser source and a wavelength-switchable laser source to generate a CW THz signal based on the photomixing technique. The proposed scheme can generate a broadband CW THz signal despite the narrow optical frequency sweep of the tunable laser source and consequently reduce the data acquisition time by reducing the time required for the frequency sweep. As a proof-of-concept, an optical frequency switching scheme in a phase-noise-compensated self-heterodyne detection system  for a swept-source THz tomography application was implemented and its performance including the signal-to-noise ratio and depth resolution were investigated.
2. Principle of the optical frequency switching scheme for high-speed broadband measurements
Figure 1 shows a simplified block diagram of a self-heterodyne system using the optical frequency switching scheme to reduce the data acquisition time of the broadband CW THz tomography application. The wavelength switchable laser source (WSLS) consists of multiple (N) laser sources with an optical frequency space (Δfspc) and an N × 1 optical switch and generates a periodic sequence of an optical signal with N different wavelengths controlled by the rectangular pulse (fSW) from the trigger source. The output of the WSLS is combined with the output of the tunable laser source (TLS) and directed to the O-E converter and photomixer. In the O-E converter, a periodic sequence of the swept-CW THz signal shown in the inset of Fig. 1 is generated through the beating of the optical signals from the WSLS and TLS and then radiated. The radiated periodic sequence of the swept-CW THz signal is detected and down-converted by the photomixer. The optical local oscillator (LO) signal, which is used as an LO signal for down-conversion by the photomixer, consists of the output of the wavelength switchable laser source (WSLS) shifted by the optical frequency of fs using the frequency shifter (FS) and the output of the TLS.
For proper down-conversion by the photomixer, the propagation delays of the periodic sequence of the swept-CW THz signal and the optical LO signal should be similar at the photomixer which can be expressed as follows:
The frequency of the rectangular trigger pulse (fSW) has to be much larger than the maximum frequency (fmax) of the photomixer output. In a conventional swept-source THz tomography system, the maximum frequency of the photomixer output is proportional to the maximum path length difference between lLO and lprob . Because the path length of lprob is dependent on the thickness of the sample under test, the maximum frequency of the photomixer output can be estimated as follows:Fig. 1, the maximum path length difference (lprob - lLO), the velocity of light in the free space, the sweep bandwidth of the TLS, and the time required for sweeping the optical frequency of the TLS, respectively. Using the 1 × N electronic switch, which is synchronized with the N × 1 optical switch in the WSLS, discrete-time signals associated with the multiple (N) laser sources are extracted from the output of the photomixer and then filtered by anti-imaging filters (AIFs). The output signals (N) of the AIFs represent the responses of the THz self-heterodyne system for different THz frequency bands associated with the multiple (N) laser sources and can be expressed as follows:Fig. 1. Consequently, because the proposed optical frequency switching scheme simultaneously takes measurements for N different THz frequency bands with a single optical frequency sweep of the TLS, a high-speed broadband measurement (N × optical-sweep-bandwidth) can be achieved using this scheme. While the proposed frequency switching scheme can take measurements for N different THz frequency bands with a single optical frequency sweep of the TLS, the photomixer that generates the THz signal takes only two optical signals at a time and thus can generate high THz signal power, which improves the signal-to-noise ratio (SNR). On the other hand, in THz systems that generate multiple THz frequency tones using a single mixer and multiple optical signals, the photomixer can generate reduced THz signal power due to its limited saturation power resulting in a degraded SNR [11, 12].
The principle of the optical frequency switching scheme is shown graphically in Fig. 2, assuming that there are two laser sources (N = 2) in the WSLS. The optical frequencies of each channel (Ch1 and Ch2) in the WSLS are set to f1 and f2 with a frequency difference of Δfspc, respectively, and the optical frequency of the TLS is changed from f0 to f0-Δf0 shown in Fig. 2(a). Note that the frequency difference of Δfspc is identical to the sweep bandwidth of Δf0. In this case, the frequency of the CW THz radiation at the output of the O-E converter related to Ch1 is changed from (f1-f0) with the sweep bandwidth (Δf0) and that related to Ch2 is changed from (f2-f0) with the sweep bandwidth (Δf0). The frequency of the periodic sequence of the CW THz signal as a function of time is shown in Fig. 2(b). The frequency-switched CW THz signal is detected and down-converted by the photomixer, and the resulting output signal with interleaved components associated with the two laser sources (Ch1 and Ch2) in the WSLS is shown in Fig. 2(c). As shown in Fig. 2(c), the two down-converted signals corresponding to the CW THz radiation starting from (f1-f0) and (f2-f0) are multiplexed in the output of the photomixer.
The 1 × 2 electronic switch splits the output of the photomixer into two discrete-time signals shown in Fig. 3.
Assuming that all the conditions expressed in the previous description are satisfied, the two discrete-time signals can be reconstructed without any distortion induced by the optical frequency switching scheme. Figure 4 shows the output of the post signal processor from Ch1 and Ch2 as a function of the THz signal frequency, which can be used to estimate the path length difference (i.e., thickness) associated with the sample under test. Note that in the optical frequency switching scheme with N = 2, the spectrum of the THz signal used for the measurements is two times wider than the optical frequency sweep bandwidth (Δf0) of the TLS, which offers an improved depth resolution in the estimation of the thickness of the sample under test. Note also that the optical frequency switching scheme with N = 2 requires a frequency sweep time half of that for the case of N = 1 (i.e., without optical frequency switching), which indicates that it is appropriate for high-speed broadband measurements.
For a proof-of-concept experiment of the proposed optical frequency switching scheme, a THz tomography system using two optical channels (N = 2) was investigated. Figure 5 shows the configuration of a phase-noise-compensated self-heterodyne THz tomography system using the two-channel optical frequency switching scheme. For phase-noise-compensation, a photodiode and an 1 × 2 electronic switch were used to generate phase noise identical to that which exists at the output of the photomixer and then used it to cancel the phase noise from the output of the photomixer. Details of the phase-noise-compensation are reported in .
The WSLS consisted of two laser sources (81662A and 81949A, Agilent) and a 2 × 1 optical switch (FOS1200-4400, Boston Applied Technologies Incorporated). The TLS (81640A, Agilent) was used to sweep the frequency of the CW THz radiation. The optical switch had a switching frequency (fSW) of 100 kHz. Two optical frequency shifters (FS1 and FS2) implemented with sawtooth-modulated electro-optic phase modulators (LN65S, COVEGA) were used to produce a fs1 and fs2 of 18 kHz (blue shift) and 20 kHz (red shift), respectively. The blue-shifted optical signal was combined with the optical signal from the WSLS, amplified with a polarization maintaining EDFA (CEFA, KEOPSYS), and then directed to the O-E converter (EK-000724, Toptica) to generate a CW THz radiation with a frequency that corresponded to the beating frequency of the two optical signals. The red-shifted optical signal was combined with the optical signal from the TLS, amplified with a polarization maintaining EDFA, and then directed to the photomixer (EK-000725, Toptica) to down-convert the CW THz radiation by the self-heterodyne detection. The 1 × 2 electronic switches (ADG5209F, Analog Devices) were synchronized with the 2 × 1 optical switch. Low-pass filters, which were connected to the electronic switches, were used as an anti-imaging filter. A lock-in amplifier (LIA) with multi-demodulators (MFLI, Zurich Instruments) was used to cancel out the common phase noise component from the photomixer output signal. Finally, the phase-noise compensated signal from the two channels was rearranged to construct the amplitude of the signal as a function of the THz signal frequency which was used to estimate the path length difference (i.e., thickness) associated with the sample under test.
The SNR of the phase-noise compensated signal with and without the optical frequency switching scheme was compared to investigate the effect of the optical frequency switching scheme of the phase-noise-compensated self-heterodyne system. In the optical frequency switching scheme, the optical frequency of the laser sources (f1, f2) in the WSLS was set to 193.63 THz and 193.93 THz corresponding to wavelengths of 1549.32 nm and 1546.95 nm, respectively. The optical frequency (f0) of the TLS was changed from 193.33 THz (1551.75 nm) to 192.43 THz (1559.01 nm) with a step of about 0.623 GHz (0.005 nm). In this condition, the frequency of the CW THz radiation related to the optical channel 1 was changed from 300 GHz to 1200 GHz, and the frequency of the CW THz radiation related to the optical channel 2 was changed from 600 GHz to 1500 GHz. Figure 6 shows the measured SNR of the phase-noise compensated signal of optical channel 1 with and without the optical switching scheme as a function of the THz signal frequency. As shown in Fig. 6, the optical switching scheme does not have any significant loss in the SNR. The slight decrease in the SNR of the phase-noise compensated signal is attributed to the increased noise floor due to the additional electronic circuits for the optical frequency switching.
To demonstrate the capability of the proposed optical frequency switching scheme in the phase-noise-compensated self-heterodyne THz system for high-speed broadband measurements, a swept-source THz tomography system with a two-channel optical frequency switching scheme shown in Fig. 5 was investigated. In this experiment, the optical frequency (f0) of the TLS was swept from 193.33 THz to 193.03 THz with a sweep rate (Δf0/τsweep) of 634 GHz/sec. In this case, the THz signal frequency associated with optical channel 1 (f1 = 193.63 THz) was swept from 300 GHz to 600 GHz, and the THz signal frequency associated with optical channel 2 (f2 = 193.93 THz) was swept from 600 GHz to 900 GHz. With this sweep rate, the time required for the measurement within the THz bandwidth of 600 GHz was about 0.48 sec for N = 2. A THz reflector was used as the sample under test shown in Fig. 5, so that a single reflection of THz radiation from the surface of the THz reflector could be measured. Figure 7 shows the phase-noise-compensated signals corresponding to Ch1 and Ch2.
The phase-noise-compensated signals in the time domain shown in Fig. 7 can be converted and rearranged to a phase-noise-compensated signal as a function of the CW THz frequency ranging from 300 GHz to 900 GHz using the sweep rate of 634 GHz/sec. While combining the phase-noise-compensated signals in the time domain, there may exist a waveform discontinuity at the center of the combined signal since the rectangular trigger pulse for switching of the wavelength-switched laser and the sweeping signal of the wavelength-tuned laser are not synchronized. The temporal discontinuity may degrade the depth resolution of the tomography system. However, the time discontinuity is less than a period (< 10 μsec) of the rectangular trigger pulse (fSW = 100 kHz). Since this value is much smaller than the period (4.74 ms) of the phase-noise-compensated signal (fmax - fs ≈211 Hz), its effect on degradation of the depth resolution is negligibly small. Note that the fmax was estimated using the maximum path length difference (Δlmax) of 100 mm. Figure 8(a) shows the point spread function (PSF) calculated by a Fourier transform of the frequency domain phase-noise-compensated signal from optical channel 1 (for the THz frequency range from 300 GHz to 600 GHz). Figure 8(b) shows the PSF calculated from the combined phase-noise-compensated signal (optical channel 1 and channel 2 and thus for the THz frequency range from 300 GHz to 900 GHz). After Fourier transformation, the depth (i.e., thickness) information of the sample under test can be obtained from the PSF. Note that there exists only one peak because only a single reflection of the THz radiation from the THz reflector is measured. The location of the THz reflector extracted from the PSF is 11.18 mm. The PSF from the phase-noise-compensated signal associated with optical channel 1 only has a full-width half maximum (FWHM) of about 0.93 mm, and the PSF from the combined phase-noise-compensated signals (channels 1 and 2) has a FWHM of about 0.52 mm. Therefore, the swept-source THz tomography system with the two-channel optical frequency switching scheme shows an improved depth resolution (approximately half the value) for the same amount of measurement time. The depth resolution of 0.52 mm is larger than the theoretical value (0.25 mm) estimated for the frequency sweep bandwidth of 600 GHz, assuming a constant THz signal power within the sweep bandwidth , which is attributed to the THz signal power spectrum shown in Fig. 6.
To show the effect of the improved depth resolution by the optical frequency switching scheme, the thicknesses of opaque rigid plates were measured with the proposed swept-source THz tomography system. The plates consisted of a photopolymer (RGD835) manufactured with a 3D printer (Object30 pro, Stratasys). Prior to tomographic imaging, the transmittance of the plates with different thicknesses was investigated over a spectral range from 0.3 THz to 1.0 THz. Figure 9(a) shows the experimental setup for the transmittance measurement, and Fig. 9 (b) shows the THz transmission characteristics of the plates with thicknesses of 0.2 mm and 1.2 mm. Note that the transmittance of the plate decreased as the frequency of the THz radiation was increased as shown in Fig. 9(b).
A two-axis motorized stage, shown in Fig. 10(a), was used for three-dimensional (3D) tomographic imaging of a plate that is composed of a photopolymer. The plate was 2 mm thick and had rectangular concave regions with a thickness ranging from 0.2 mm to 1.2 mm formed on the rear of the plate shown in Fig. 10(c). If the maximum path length difference (Δl) is assumed to be 100 mm, the maximum frequency of the photomixer output is 2.21 kHz. Because the switching frequency (fSW) of 100 kHz is much larger than the maximum frequency (2.21 kHz) of the photomixer output signal, the depth information can be reconstructed without any distortion. Figure 11(a) and (b) show the PSFs of the phase-noise-compensated signal measured at the point marked with an arrow in Fig. 10(b) without and with the optical frequency switching scheme, respectively. The thickness of the plate marked with the arrow in Fig. 10(b) was 0.4 mm, and double reflections of THz radiation from the front and rear surface of the plate are supposed to be observed. However, the PSF obtained without the optical frequency switching scheme shown in Fig. 11(a) exhibits only one peak, which is attributed to a large depth resolution of 0.93 mm. On the other hand, the PSF obtained with the optical frequency switching scheme shown in Fig. 11(b) clearly exhibits two peaks associated with the front and rear surface of the plate due to its improved depth resolution of 0.52 mm. The measurement results indicate that the proposed swept-source THz tomography system with the optical frequency switching scheme can offer a substantially improved depth resolution.
Figure 12 shows slices of depth images along the x-axis for different positions on the y-axis including the lines (top, middle, and bottom) indicated in Fig. 10(b) without and with the optical frequency switching. Comparison of the depth images measured at the top and bottom positions can reveal the depth resolution of the swept-source THz tomography system without and with the optical frequency switching. In the case without the optical frequency switching scheme, the minimum layer thickness that can be resolved is 1 mm. On the other hand, in the case with the optical frequency switching scheme, the minimum layer thickness that can be resolved is 0.4 mm. The results indicate that the optical frequency switching scheme offers an improved depth resolution for the swept-source THz tomography system in the same amount of measurement time, which is suitable for high-speed broadband measurement applications.
This study presented an optical frequency switching scheme for a high-speed broadband THz measurement system based on the photomixing technique. The proposed system can achieve high-speed broadband THz measurements because measurements of samples in multiple THz bands can be done simultaneously using narrow optical frequency scanning of a tunable laser source combined with a wavelength-switchable laser source. In addition, this optical frequency switching scheme can provide a larger output power for an individual THz signal compared with that of a multi-mode THz signal generated by multiple CW laser sources . A swept-source THz tomography system implemented with a two-channel wavelength-switchable laser source did not show any degradation in the signal-to-noise ratio and offers an improved depth resolution with only half the measurement time required for that of a THz tomography system without the optical frequency switching scheme. An increase in the number of the optical channels using an optical comb signal and an arrayed waveguide grating can offer a higher depth resolution with a reduced measurement time, which makes the proposed scheme attractive for the implementation of real-time THz measurement systems.
Brain Research Program (NRF-2015R1A2A1A15055838, NRF of Korea).
References and links
1. D.-S. Yee, J. S. Yahng, C.-S. Park, H. Don Lee, and C.-S. Kim, “High-speed broadband frequency sweep of continuous-wave terahertz radiation,” Opt. Express 23(11), 14806–14814 (2015). [CrossRef] [PubMed]
2. Q. Song, Y. Zhao, A. Redo-Sanchez, C. Zhang, and X. Liu, “Fast continuous terahertz wave imaging system for security,” Opt. Commun. 282(10), 2019–2022 (2009). [CrossRef]
3. S. Jones, J.-Y. Kim, Y. Doi, T. Yamada, N. Koshobu, and H. Togo, “Ultra-Wideband Tunable Dual-Mode Laser for Continuous Wave Terahertz Generation,” J. Lightwave Technol. 32(20), 3461–3467 (2014). [CrossRef]
4. A. Roggenbuck, H. Schmitz, A. Deninger, I. C. Mayorga, J. Hemberger, R. Güsten, and M. Grüninger, “Coherent broadband continuous-wave terahertz spectroscopy on solid-state samples,” New J. Phys. 12(4), 043017 (2010). [CrossRef]
5. H. Song and J.-I. Song, “Terahertz-Wave Vibrometer Using a Phase-Noise-Compensated Self-Heterodyne System,” IEEE Photonics Technol. Lett. 28(3), 363–366 (2016). [CrossRef]
6. T. Nagatsuma, H. Nishii, and T. Ikeo, “Terahertz imaging based on optical coherence tomography [Invited],” Photon. Res. 2(4), B64–B69 (2014). [CrossRef]
8. H.-J. Song, N. Shimizu, T. Furuta, K. Suizu, H. Ito, and T. Nagatsuma, “Broadband-Frequency-Tunable Sub-Terahertz Wave Generation Using and Optical Comb, AWGs, Optical Switches, and a Uni-Traveling Carrier Photodiode for Spectroscopic Applications,” J. Lightwave Technol. 26(15), 2521–2530 (2008). [CrossRef]
10. J.-Y. Kim, H. Nishi, H.-J. Song, H. Fukuda, M. Yaita, A. Hirata, and K. Ajito, “Compact and stable THz vector spectroscopy using silicon photonics technology,” Opt. Express 22(6), 7178–7185 (2014). [CrossRef] [PubMed]
11. M. Scheller, K. Baaske, and M. Koch, “Multifrequency continuous wave terahertz spectroscopy for absolute thickness determination,” Appl. Phys. Lett. 96(15), 151112 (2010). [CrossRef]
13. A. J. Deninger, T. Göbel, D. Schönherr, T. Kinder, A. Roggenbuck, M. Köberle, F. Lison, T. Müller-Wirts, and P. Meissner, “Precisely tunable continuous-wave terahertz source with interferometric frequency control,” Rev. Sci. Instrum. 79(4), 044702 (2008). [CrossRef] [PubMed]
15. E. D. Moore and R. R. McLeod, “Phase-sensitive swept-source interferometry for absolute ranging with application to measurements of group refractive index and thickness,” Opt. Express 19(9), 8117–8126 (2011). [CrossRef] [PubMed]
16. H. Song, S. Hwang, and J.-I. Song, “Terahertz tomography system using self-heterodyne detection with phase noise compensation,” in Proceedings of CLEO:2016, OSA Technical Digest (Optical Society of America, 2016), paper SM1L.7. [CrossRef]