Development and future prospects of terahertz technology^*

Electromagnetic pulses are generated by irradiating the surface of semiconductors, devices, or nonlinear optical crystals (NLCs) with femtosecond lasers.^21,31) This phenomenon can be basically explained by nonlinear Maxwell equations,

$$\begin{equation} \nabla \times \boldsymbol{{H}} = \boldsymbol{{j}} + \varepsilon_{\mathbf{0}}\frac{\partial \chi^{(2)}\boldsymbol{{E}}\boldsymbol{{E}}}{\partial t} + \varepsilon \frac{\partial \boldsymbol{{E}}}{\partial t}. \end{equation} \tag{ 1 }$$

Here, j is the conduction current density, ε₀ is the vacuum permittivity, ε is the linear permittivity of a material, and χ⁽²⁾ is the second-order nonlinear electric susceptibility.³²⁾ First, I will explain a photoconductive switch [Fig. 3(a)], which is also called a photoconductive antenna because the electrodes in the figure function as an antenna. In a photoconductive switch, electrodes are formed on a semiconductor thin film and several tens of volts are applied between the electrodes. When the gap between the electrodes (usually, approximately 5 µm) is irradiated with a femtosecond laser pulse with an energy higher than the band gap of the semiconductor, transient current j flows to radiate electromagnetic pulses with a nearly single cycle of E ∼ dj/dt in a far field in accordance with Maxwell equations. When the time width of the change in current is on the order of picoseconds, an electromagnetic pulse with a width of picoseconds is radiated. This time width corresponds to frequencies in a very broad THz band (Fig. 3, lower left). Photoconductive switches are also called the Auston switches named after the inventor. It is known that the spectrum of THz radiation is strongly dependent on the shape of antennas, and the characteristics of photoconductive antennas were examined in detail by Tani and colleagues.^33,34) Currently, low-temperature-grown GaAs (LT-GaAs) exhibits good breakdown voltage characteristics and a high mobility of photoexcited carriers, and is used for photoconductive switches as standard THz radiation devices that are excited by light with a wavelength of approximately 800 nm, such as mode-locked Ti:sapphire lasers. In addition, photoconductive switches are also used to detect THz waves. Research on photoconductive switches has been steadily carried out with the aim of reducing the carrier lifetime to cover higher frequencies,³⁵⁾ enhancing various functions including the control of polarized radiation,³⁶⁾ and realizing the excitation and detection of radiation in the communication bands (1.5 and 1.3 µm).^37–39) In particular, the expansion to the communication bands is important because low-cost matured lasers and optical components can be used.

Zoom In Zoom Out Reset image size

In 1992, Zhang and Auston reported that THz pulses were generated by femtosecond laser excitation from the surface of a semiconductor that had not been processed for use as a device.⁴⁰⁾ They explained the radiation mechanism as follows: photogenerated carriers were accelerated in a strong built-in surface electric field generated by the bending of a band due to surface defects, as shown in Fig. 3(b). Later, however, another mechanism, i.e., the nonlinear optical effect [optical rectification, Fig. 3(c)], was also reported on the basis of the fact that the radiation intensity is periodically changed by rotating crystals during high-density excitation.⁴¹⁾ In addition, the above phenomenon was interpreted as radiation due to the photo-Dember effect, in which transient current is induced from a difference in the diffusion speed between electrons and holes generated on the surface, because such a strong surface electric field cannot be expected for THz radiation from semiconductors with a low band gap energy, i.e., narrow-gap semiconductors, such as InAs.⁴²⁾ It was also found that the radiation intensity was increased by applying a magnetic field.⁴³⁾ This is particularly effective for InAs with a small effective mass, which can be thereby used as a simple light source for spectrometers.⁴⁴⁾ In addition, various types of radiation from semiconductors related to elemental excitation have been observed including IR-active phonons,^42,45) cavity polaritons,⁴⁶⁾ coupled quantum wells,⁴⁷⁾ plasmons in a two-dimensional electron system at the semiconductor interface,⁴⁸⁾ Bloch oscillations in semiconductor superlattices,⁴⁹⁾ and longitudinal optical phonons in semiconductor superlattices.⁵⁰⁾ Moreover, THz waves are also radiated from high-temperature superconductors owing to the high-speed modulation of supercurrent by optical pulses.^51,52) However, most of these types of radiation are insufficient in terms of their intensity and frequency range for application as light sources for spectroscopy.

THz radiation based on the nonlinear optical effect in Fig. 3(c) is related to χ⁽²⁾ in the second term on the right-hand side of Eq. (1). The shorter the pulse width of the excitation light, the broader the band in which THz waves are radiated.⁵³⁾ Recently, Katayama et al. have observed mid-IR radiation with frequency up to 200 THz from an organic nonlinear optical crystal of 4-N,N-dimethylamino-4'-N'-methyl-stilbazolium tosylate (DAST) using a 5 fs laser.⁵⁴⁾

As mentioned above, tunable monochromatic THz waves can be generated using monochromatic laser light sources on the basis of the nonlinear optical effect. This method has two main approaches: one involves optical parametric effect [Fig. 3(d)]⁵⁵⁾ and the other involves difference frequency generation [Fig. 3(e)].⁵⁶⁾ In the former approach, a nanosecond monochromatic laser light (frequency, ω₁) is incident to a nonlinear optical crystal (e.g., LiNbO₃, DAST), and the phase-matching condition is changed by rotating the crystal to generate tunable monochromatic THz waves (ω₁ = ω₂ + ω_THz, where ω₂ is the frequency of the idler wave) (Fig. 3, lower right).²⁷⁾ Here, phase matching means that THz waves are efficiently generated when the conservation of wavenumber (or momentum), k₁ = k₂ + k_THz, holds by assuming that the incident and generated waves can be approximated as plane waves. The devices with a cavity for an idler wave are called THz optical parametric oscillators (THz-OPOs), and those without a cavity are called THz optical parametric generators (THz-OPGs). In the difference frequency generation, two monochromatic lights with frequencies having a difference of several terahertz (ω₁ and ω₂) are introduced to a nonlinear optical crystal to generate waves with a difference in frequency (ω_THz = ω₁ − ω₂). Also in this case, the phase-matching condition, k_THz = k₁ − k₂, must be satisfied to efficiently generate THz waves. Tunable THz waves can be generated when the light with ω₁ or ω₂ is tunable within the THz range (Fig. 3, lower right).²⁹⁾ Recently, Suizu and colleagues have developed a Cherenkov phase-matching method using a very large difference between the wavelength of excitation light and that of THz waves, and realized THz radiation at frequencies higher than 10 THz and with relatively flat frequency characteristics using LiNbO₃ and DAST crystals (Fig. 4).^57,58) In this phase-matching method, the refractive index of incident light in the nonlinear optical crystal, n_opt, is smaller than that of THz waves in the adjacent Si, n_Si, and the nonlinear polarization used as the light source propagates faster than the THz waves; therefore, the THz waves are radiated by a mechanism similar to that of Cherenkov radiation for high-speed charged particles. For the generation of THz waves using nanosecond lasers and their applications, please refer to the report by Minamide.⁵⁹⁾

Zoom In Zoom Out Reset image size

Explained above are broadband THz waves and tunable monochromatic waves generated by femto- and nanosecond laser excitation, respectively. There is another method of generating continuous wave (cw) THz waves called the photomixing method, in which photoconductive switches are irradiated with two continuously oscillated monochromatic laser lights (with frequencies having a difference of several terahertz).⁶⁰⁾ In this method, a light source emitting tunable monochromatic cw THz radiation can be obtained by sweeping the frequency of one of the two laser lights. As a device used for converting between light and THz waves other than photoconductive switches, uni-traveling-carrier photodiodes (UTC-PDs) developed by NTT Corporation are highly efficient and expected to be used as light sources for wireless communications below 500 GHz.^61,62) As shown in Fig. 5, a nonlinear optical crystal is placed in a cavity, and a vertical-external-cavity surface-emitting laser (VECSEL) diode based on difference frequency generation is excited by a 50 W fiber laser, obtaining a high output power of above 2 mW at 1.9 THz.⁶³⁾ Note that the insertion of a crystal used for THz radiation into a laser cavity has already been attempted by Sarukura et al.⁶⁴⁾

Zoom In Zoom Out Reset image size

First, I will explain THz-TDS, i.e., spectroscopy which employs broadband THz pulses radiated using femtosecond lasers as the light source.^{19,21,22,31,65,66)} For spectroscopy in the light range, lights with different frequencies are separated by gratings and the transmittance of the lights with different frequencies (wavelengths) through a sample is measured to obtain its transmission spectrum. In contrast, THz-TDS is based on the fact that a THz pulse with a very short time width contains a broadband frequency spectrum. Specifically, the strain of the waveform obtained after the pulse transmits through a sample is measured to deduce the transmission spectrum of the sample by performing a Fourier transform. Namely, the impulse response, which is well known in linear response theory, is measured to obtain the spectral response. The impulse response is first measured in a time domain, which is the reason for the name "time-domain spectroscopy". Figure 6 shows the principle of THz-TDS. When a radiation device is irradiated with approximately 100 fs laser pulses, THz pulses are radiated in accordance with the mechanism explained in the previous section. As the radiation device, photoconductive switches are used for relatively low laser intensities, whereas nonlinear optical crystals (mainly ZnTe) are used for relatively high laser intensities. Radiated THz pulses propagate in free space and are collected at a detector. As the detector, photoconductive switches or electro-optic (EO) crystals are used. In both detectors, the excitation laser pulses used to generate THz waves are separated by a semitransparent mirror and the detector is driven by time-delayed trigger pulses. In the case of photoconductive switches, photocarriers generated by the trigger pulses are accelerated in a weak THz electric field, amplified by a current amplifier, and detected. Because the current is proportional to the intensity of the THz electric field during the irradiation of the trigger light, the waveform of the electric field can be obtained by time-delay sweeping. EO crystal detection is based on a change in the polarization of the trigger light, which occurs when incident THz pulses and trigger light pulses are concentric and weak birefringence occurs in the THz electric field owing to the Pockels effect. Although step scanning using mechanical translation stages is usually performed for time-delay sweeping, rapid scanning using linear stages, which are used in IR Fourier spectrometers, and a special sweeping technique using high-speed rotational stages are also carried out. Yasui et al. have realized very rapid sweeping within milliseconds by optical sampling using two femtosecond lasers with slightly different repetition frequencies.⁶⁷⁾

Zoom In Zoom Out Reset image size

When the above system for generating and detecting THz waves is used for spectroscopy, for example, in the case of transmission spectroscopy, the time waveform is measured as a reference before a sample is inserted into a THz beam, and then the time waveform is measured after the sample is inserted into the beam. A complex Fourier transform is applied to these waveforms and the complex spectral amplitude after sample insertion is divided by the complex spectral amplitude before sample insertion, thus obtaining the amplitude transmittance and phase shift as functions of frequency. When the sample consists of parallel plates, the complex refractive index $\tilde{n} = n - i\kappa$ is obtained from the amplitude transmittance and phase shift. Here, n is the refractive index and κ is the extinction coefficient and generally corresponds to the absorbance of a sample. Moreover, the complex relative permittivity $\tilde{\varepsilon }$ and complex electric conductivity $\tilde{\sigma }$ have the following relationship with $\tilde{n}$,

$$\begin{equation} \tilde{n}^{2} = \tilde{\varepsilon} = \varepsilon_{\infty} - i\frac{\tilde{\sigma}}{\omega \varepsilon_{0}}. \end{equation} \tag{ 2 }$$

Here, $\varepsilon _{\infty }$ is the relative permittivity at high frequencies (e.g., mid-IR range) and ω is the angular frequency. Therefore, the real and imaginary parts of the permittivity and electric conductivity at high frequencies can be simultaneously obtained by transmission spectroscopy. In the evaluation of materials, it is extremely useful to obtain these electric properties without electrodes at a spatial resolution of approximately 1 mm by focusing a THz beam.

In addition to transmission spectroscopy, there are other types of method in THz-TDS, such as reflection,⁶⁸⁾ attenuated total reflection (ATR),⁶⁹⁾ and ellipsometry,⁷⁰⁾ similar to the case of spectroscopy in the usual frequency range. Although phase measurement in the light range is difficult, THz-TDS can easily measure phase shifts by inserting a sample into a beam and can simultaneously obtain the real and imaginary parts of optical constants, such as the refractive index and permittivity, without applying the Kramers–Kronig transformation. This advantage is very useful in ellipsometry. The amount of information obtained by THz-TDS is increased by applying magnetic fields to samples; for semiconductors, the carrier density, mobility, and effective mass can be independently obtained.^71,72) Recently, THz-TDS has been applied to electron spin resonance (ESR),⁷³⁾ and is expected to be further applied to the generation and disappearance of photogenerated radicals to realize ESR equipment with a temporal resolution much higher than that of conventional equipment.

In the above-mentioned THz-TDS, expensive femtosecond lasers are used to radiate THz waves. In 2000, Morikawa and colleagues demonstrated that subterahertz time-domain spectroscopy is even possible by replacing femtosecond lasers with inexpensive cw multimode laser diodes.^74,75) Their method is called cross-correlation spectroscopy because photoconductive switching detectors act as devices employing the cross-correlation between the amplitude of THz electric fields and the excitation laser intensity. The validity of this method was also confirmed by a German group in 2009, and this method is called quasi THz-TDS.⁷⁶⁾

Tunable monochromatic THz waves generated using nanosecond lasers can be used as the light source for conventional spectrometers. In this case, although bolometers cooled to liquid helium temperature are mainly used to detect THz waves, room-temperature pyroelectric detectors are also used because the intensity of THz radiation from difference frequency generation is relatively high. Spectrometers using nanosecond-laser-excited THz radiation only measure the output and cannot obtain the phase by an ordinary method but are reasonably applicable to general purposes. Moreover, such spectrometers are suitable for efficient spectroscopy that targets certain absorption lines of samples, in contrast to THz-TDS, which is suitable for broadband spectroscopy. In addition, these spectrometers enable measurements at a very high frequency resolution by injection seeding (a method of injecting both a high-intensity laser used for oscillation and a low-intensity but highly stable narrow laser to obtain high-intensity and highly stable narrow laser oscillation) in THz-OPGs⁷⁷⁾ and by narrowing the line width of laser lights with difference frequency generation using gratings.⁷⁸⁾ This advantage can also be obtained in spectroscopy using cw THz waves generated by photomixing using photoconductive switches as the light source.

THz spectroscopy and its applications are also described in the report by Kitagishi.⁷⁹⁾

Conventionally, large light sources, such as FELs, are required to induce the nonlinear effect in solids by excitation of THz waves. However, the peak intensity of THz pulses generated using femtosecond lasers is very high. Recently, methods of generating THz waves have been improved, enabling the generation of instantaneous electric fields stronger than 10 kV/cm excited by regeneratively amplified mJ-order femtosecond laser pulses.⁸⁰⁾ Media for generating THz waves include (1) photoconductive switches, (2) nonlinear optical crystals, and (3) laser-induced gas plasma.

Figure 7(a) shows a schematic of a photoconductive switch that has an interdigitated electrode on a semi-insulating GaAs (SI-GaAs) substrate. Every second tooth of the interdigitated electrode is masked so that THz electric fields in the same direction are intensified.^81,82) Although a high voltage is applied to the large gap between electrodes in conventional devices, the interdigitated electrode requires only a low voltage, thus suppressing thermal load and realizing a high repetition frequency. Using this device, THz electric fields with intensity higher than 10 kV/cm have been obtained at the focal point.

Zoom In Zoom Out Reset image size

Figure 7(b) shows a schematic of THz radiation from a LiNbO₃ crystal. Conventionally, ZnTe has mainly been used as nonlinear optical crystals to meet the requirement of phase matching, but it has a problem of saturation upon high-intensity excitation because of two-photon absorption. In contrast, LiNbO₃, which has a large band gap, has no such a problem but it is difficult to achieve phase matching using LiNbO₃. As shown in Fig. 7(b), Hebling et al. have realized Cherenkov phase matching by tilting the wave front of an excitation laser so that it deviates from the direction perpendicular to that of beam propagation, markedly increasing the radiation efficiency of THz waves.⁸³⁾ THz electric fields with an intensity of 250–1200 kV/cm have been realized at the focal point by this method.^84,85)

Figure 7(c) shows a schematic of THz radiation from gas plasma induced by a femtosecond laser. Fundamental waves with an angular frequency ω cause the ionization of the gas to induce plasma, and the second harmonic waves with angular frequency 2ω generated from a nonlinear optical crystal [this phenomenon is called second harmonic generation (SHG)] produce high-intensity THz waves owing to the third-order nonlinear optical effect of the plasma (two-color excitation).⁸⁶⁾ Although the physical mechanism underlying the generation of THz waves was initially interpreted to be perturbative, it is now understood that high-intensity optical electric fields distort the Coulomb potential of gas atoms and that electron–ion plasma is generated by tunnel ionization and accelerated in the optical electric fields. To date, an energy conversion efficiency higher than 10⁻⁴ and a THz pulse energy of 5 µJ have been obtained.⁸⁷⁾ Methods using gas plasma are advantageous because the gas used as the medium to generate THz waves is not damaged by high-intensity lasers and there are no frequency bands where the THz wave intensity becomes very low owing to the absorption by phonons, unlike in methods using solid plasma. Zhang's group proposed an ingenious system that that generates and detects THz waves using air.⁸⁸⁾ Recently, it has been found that THz waves are more efficiently radiated when using Ar clusters than when using gases in the case of excitation with monochromatic femtosecond lasers.^89,90) Such higher-efficiency radiation using Ar clusters is also expected to be realized in the case of two-color excitation (ω and 2ω).

Figure 8 shows the energies of THz waves generated by the above-mentioned methods, which are plotted as a function of excitation laser pulse energy.⁹¹⁾ The maximum energy efficiencies of the three methods are all approximately 0.1%. This is close to the limit derived from the Manley-Rowe relations for the nonlinear optical effect.

Zoom In Zoom Out Reset image size

For high-intensity THz pulses and their applications, please refer to the reports by Hirori and Tanaka.⁹²⁾

QCLs are based on the intersubband transition of semiconductor multiple quantum wells, and their oscillation frequency can be controlled by varying the width of the wells regardless of the band gap of the materials.^12,97–99) In QCLs, electrons injected into the active layers cascade from one sublevel to another within the well, thus inducing laser oscillation. Initially, QCLs were oscillated in the mid-IR range. In 2002, Kohler et al. succeeded in QCL oscillation at 4.4 THz.¹⁰⁰⁾ Later, the confinement of THz waves was achieved by metal–metal waveguides and the realization of population inversion was facilitated through the extraction of electrons from the post-transition level by resonant phonon scattering, thus realizing oscillation at 0.8–5 THz for GaAs-based materials. Efforts have been directed to the realization of oscillation at higher temperatures, and pulse and cw oscillations have been achieved at 186 and 117 K, respectively. Moreover, scientists have attempted to indirectly inject electrons into the active layers¹⁰¹⁾ and use new materials such as GaN¹⁰²⁾ with the aim of achieving oscillation at high temperatures and low frequencies. Although it was empirically considered difficult to realize laser oscillation at temperatures higher than or equal to that corresponding to the oscillation frequency ω, laser oscillation at a high temperature corresponding to an energy 1.9-fold higher than that corresponding to ω has been realized through the above attempts.¹⁰¹⁾ Although Japan started research on QCLs later than the U.S.A. and Europe, successful THz QCL oscillation has been achieved by the National Institute of Information and Communications Technology,¹⁰³⁾ Tohoku University,¹⁰⁴⁾ and RIKEN,¹⁰⁵⁾ and private companies have also started THz QCL research.

Unlike conventional lasers, THz QCLs oscillate in the active layers (multiple quantum wells) of metal–metal (or metal–semiconductor) waveguides narrower than the wavelength. Because THz waves are radiated from a small laser aperture, the divergence of radiation due to diffraction remains a major problem. New technologies have been introduced from other fields to resolve this issue. Thus, QCLs are of interest as a technology related to THz research, into which technologies from various fields have been introduced. I will give a detailed explanation of this matter below.

It is known that surface waves (surface plasmon polaritons, SPPs) formed by the hybridization of photons and plasma exist on the surface of a metal in the near-IR and visible ranges. Currently, research applying SPPs is being intensively carried out in the field of plasmonics. Meanwhile, metals behave as perfect conductors in the micro- and THz-wave ranges, in which no modes of surface waves exist. However, in 2004, Pendry et al. demonstrated that a localized mode similar to SPPs (called spoof or mimicking SPPs) exists in these frequency ranges when a periodic structure with a period shorter than the wavelength (e.g., a grating) is formed on the surface of a metal.¹⁰⁶⁾ Figure 9(a) shows high-directivity THz radiation into free space that is released from a laser aperture and introduced as spoof SPPs with a superperiod of λ₀, shown in pink in the figure.¹⁰⁷⁾ This is realized using a metallic substrate for a THz QCL and patterning the metal surface with subwavelength gratings. Figure 9(b) shows high-directivity THz radiation realized by forming holes on a metal–metal waveguide at regular intervals to allow the waveguide to act as a leaky wave antenna.¹⁰⁸⁾ In this setup, the direction of radiation has also been successfully controlled by varying the frequency. QCLs are also attracting attention as a gain medium to compensate for the loss in devices fabricated using metamaterials (see Sect. 4.2).¹⁰⁹⁾

Zoom In Zoom Out Reset image size

Generally, an external cavity is formed in the propagation direction of laser light and the cavity length is varied to change the oscillation frequency of lasers. However, this approach is difficult to realize for THz QCLs with a large beam divergence angle. On the basis of the fact that the lateral beam confinement is on the order of the wavelength, it has been demonstrated that the oscillation frequency is greatly changed by slightly moving a metallic or semiconductor plunger, as shown in Fig. 9(c).¹¹⁰⁾ Micro-electromechanical systems (MEMS) technology has been adopted to realize the movement of the plunger at low temperatures.¹¹¹⁾

As described above, new concepts and technologies, such as plasmonics, metamaterials, and MEMS, have been introduced into THz research, realizing the borderless fusion of technologies and accelerating the advancement of THz research.

Room-temperature-operating subminiature electronic devices, such as optical laser diodes, are essential for realizing a wide range of applications of THz technologies. The previously mentioned TUNNETT diodes are such an electronic device but have not yet achieved oscillation at a frequency of 1 THz. Resonant tunneling diodes (RTDs) are electronic devices based on the resonant tunneling of electrons due to the discrete levels in a double-barrier semiconductor quantum well and are expected to achieve oscillation at the highest reported frequencies. When a voltage is applied to an RTD, negative differential resistance is induced as a result of the generation of a current peak in the resonant tunneling state. In this case, oscillation is obtained by connecting an appropriate external resonant circuit to the RTD. Asada et al. obtained a high output power of 400 µW at 550 GHz by integrating a fine slot antenna with an RTD,¹¹²⁾ and in 2010 succeeded in producing oscillation with an output power of 7 µW at 1.04 THz, realizing the first electronic device with an oscillation frequency above 1 THz (Fig. 10).¹¹³⁾ Further advancements, such as array structures, are expected by using this device as a room-temperature oscillating electronic device.

Zoom In Zoom Out Reset image size

Oscillators, such as TUNNETT diodes and RTDs, which use the propagation of electrons, have a problem that the traveling time of electrons determines the upper limit of the oscillation frequency. To solve this problem, in 1993, Dyakonov and Shur proposed the use of plasma waves in two-dimensional electron systems at the interface of semiconductors for oscillation.¹¹⁴⁾ Electrons emitted from the source of a field-effect transistor (FET) interact with two-dimensional plasma waves until they reach the drain. They theoretically found that the two-dimensional plasma waves are amplified as a result of reflection at the source and drain in this process. Because of this instability, THz waves can be resonantly radiated by setting an appropriate gate length. Thus far, broadband THz radiation has been observed using mainly GaAlAs-based high-electron-mobility transistors (HEMTs). The underlying radiation mechanism is considered to be thermal excitation by traveling electrons rather than the mechanism proposed by Dyakonov and Shur.¹¹⁵⁾ Otsuji et al. succeeded in efficiently radiating THz waves into space by forming interdigitated gates and also applied this radiation technique to spectroscopy.¹¹⁶⁾ Two-dimensional plasma waves can also be used to detect THz waves, as described later.¹¹⁷⁾ Otsuji et al. have carried out ambitious research on the development of THz lasers using the unique band structure (Dirac electrons) of graphene, which is a two-dimensional electron system, and have recently observed stimulated THz emission from photoexcited graphene (negative electric conductivity).¹¹⁶⁾

In semiconductor superlattices, electrons move in the reciprocal lattice space at a constant velocity under a constant voltage when electron scattering is small. In real space, this movement appears as reciprocated movement with a time period determined by the applied voltage and the superlattice period. Therefore, it was considered that voltage-synchronous THz oscillators can be realized using this phenomenon, i.e., Bloch oscillation. However, recent studies have proved that such a simple explanation is incorrect; the correct explanation is that the electric conductivity becomes negative, i.e., amplification occurs, at a certain frequency range owing to the scattering of electrons, as clarified by Hirakawa and colleagues.^10,13,118) With this new interpretation of physics, research aiming to realize laser oscillation from superlattices is ongoing.

Gyrotrons based on the relativistic effect of electrons are known as high-output-power light sources and have been developed in two ways: by increasing the output power as the light source used for heating nuclear fusion plasmas and by increasing the frequency for use in new applications. In 2006, Idehara and colleagues at Fukui University realized the world's first gyrotron that generates radiation at a frequency exceeding 1 THz by applying a strong magnetic field of 19 T and using a higher-order mode [oscillation at 1.001 THz, Fig. 11(a)].¹¹⁹⁾ Gyrotrons are large because they require superconducting magnets and other large components.

Zoom In Zoom Out Reset image size

There are two methods of radiating ultrahigh-intensity THz waves using electrons accelerated by accelerators. One is to use free electron lasers (FELs), in which laser oscillation is induced by amplifying the interaction between electrons and electromagnetic waves using cavities. The other is to use coherent radiation based on the interaction of electrons having a bunch length shorter than the wavelength with magnetic fields or metallic thin films. FELs radiate tunable monochromatic THz waves, whereas coherent radiation generates broadband pulsed THz waves. Since the radiation from an FEL was first observed in 1976 at Stanford University, mainly IR FEL facilities have been constructed worldwide and used in joint research projects. Recently, the development of facilities involving X-rays and THz FELs has been a front-line target of research. Several THz FELs are currently in operation, among which the FEL at the University of California, Santa Barbara (UCSB) in the U.S.A.¹²⁰⁾ and the FEL for IR experiments (FELIX) of the Foundation for Fundamental Research on Matter (FOM, a plasma physics laboratory) in the Netherlands¹²¹⁾ are well known. A compact FEL oscillating at 100–300 µm has been constructed at the Korea Atomic Energy Research Institute (KAERI).¹²²⁾ Although Japan was a late starter in the development of THz FELs, the Institute of Science and Industrial Research, Osaka University, has succeeded in producing an FEL with oscillation at 70–100 µm, which has started to be used in joint research on a trial basis [Fig. 11(b)]. Moreover, a new amplification mechanism that uses spoof SPPs for Smith-Purcell FELs based on the interaction between electron beams with a relatively low energy of 100 keV or lower and metallic gratings, rather than large accelerators, has been proposed and is expected to be further developed in the future.^123,124)

Electromagnetic waves are radiated when a high-speed electron bunch is bent by a magnetic field. When the length of the electron bunch is smaller than the wavelength of the radiation, the electric field amplitudes of the electromagnetic waves radiated from each electron are coherently summed, and the intensity of the radiated electromagnetic waves becomes proportional to the square of the number of electrons (i.e., current) (coherent radiation). As a result, electromagnetic waves that are several orders of magnitude stronger than ordinary radiation are radiated.¹²⁵⁾ THz radiation with a very high intensity reaching 20 W has been obtained using a so-called energy recovery linac (ERL), which recycles the electrons it uses.¹²⁶⁾ In Japan, THz coherent radiation sources have been developed by the High Energy Accelerator Research Organization (KEK) as part of the development of ERLs.

In the radiation of THz waves by excitation with femtosecond lasers, photoconductive antennas and EO crystals are used as detectors, as mentioned in Sect. 2.2. Such detectors are generally activated by trigger light pulses that are obtained by separating an excitation laser light using semi-transparent mirrors, thus performing coherent detection. In this section, however, I will describe detectors that detect THz waves independently of their radiation sources rather than the above-mentioned detectors.¹²⁷⁾

Detectors are classified into thermal and nonthermal detectors. In the former, the device temperature increases as a result of the absorption of THz waves, and the increase is detected as a change in electric resistance. The response time is as long as several milliseconds because the detection is based on a change in temperature. In the latter, the absorption of THz waves directly changes the electron state, which is reflected in the change in current, allowing a rapid response of nanoseconds or shorter. Detectors are also classified into uncooled detectors, which can operate at room temperature, and cooled detectors, which operate at low temperatures (below the temperature of liquid helium). Moreover, there are two types of detection method, i.e., the incoherent method (direct detection), which detects only the intensity of THz waves, and the coherent method, which detects both the amplitude and phase (e.g., heterodyne detection using local oscillators). Detectors used for heterodyne detection are called mixers. Some detectors have been technically matured through their long development history, as exemplified by pyroelectric detectors, Golay cells, and Schottky barrier diodes. Other newer detectors include room-temperature detectors using two-dimensional plasma waves^117,128) and THz single-photon detectors using single-electron transistors.^129–131) Table I summarizes the various types of detector and their noise equivalent power (NEP), operating temperature, and response rate. NEP refers to the input power giving an output equivalent to the noise of detectors and is expressed in W Hz^−1/2. The lower the value of NEP, the higher the detection capability. The minimum NEP of room-temperature detectors is approximately 10⁻¹¹ W Hz^−1/2. For detectors cooled to the temperature of liquid helium or lower, the NEP is approximately 10⁻²⁰ W Hz^−1/2 when using a superconducting phenomenon¹³²⁾ and 10⁻²² W Hz^−1/2 when using semiconductor single-electron transistors, which is equivalent to the detection level for single photons.¹³⁰⁾

Detectors using two-dimensional plasma waves, which have recently been studied intensively, are highly promising for industrial applications because of their improved sensitivity and integration level. As mentioned above, these detectors are based on the instability of two-dimensional plasma waves and can operate at room temperature and are used at various resonance frequencies by varying the gate voltage and length. These detectors have been developed using GaAlAs-based HEMTs as well as Si complementary metal–oxide–semiconductors (CMOSs).¹³³⁾ The fabrication of multichannel detectors has also been attempted.¹³⁴⁾ CMOS-based detectors are highly compatible with the process of Si integration, enabling monolithic integration with reading devices and amplifiers. Recently, GaN-based two-dimensional plasmon detectors with sensitivity higher than that of conventional detectors have been reported.¹³⁵⁾

Bolometer microarrays are uncooled array detectors with relatively high sensitivity. NEC Corporation has developed 320 × 240 pixel bolometer microarrays and obtained good performance by combining them with QCLs.¹³⁶⁾ For details, please refer to the report by Oda.¹³⁷⁾

Conventionally, optical components used for THz radiation have been designed with particular specifications and are mostly expensive, and only a few types have been available. In recent years, however, research on various optical components such as polarizers, filters, wavelength plates, and waveguides has been actively carried out. In this section, I will present a brief summary of the current status of research on these components.

The polarizers most frequently used for THz radiation are freestanding wire grids, which are obtained by mechanically stretching approximately 10-µm-diameter metallic wires (tungsten is mainly used because of its high strength) with a pitch of approximately 30 µm. Freestanding wire grids are expensive and must be carefully handled. Another drawback of such grids is the decrease in extinction ratio at 3 THz or higher. Higher-performance polarizers are required in ellipsometry and the measurement of electron spin resonance, both of which require a high extinction ratio. To meet this requirement, metallic wire grids have been developed on insulating substrates by lithography.¹³⁸⁾ Figure 12(a) shows an aluminum wire grid polarizer fabricated on a thin plastic film by nanoimprinting. Since the wire pitch is only 140 nm, the extinction ratio of this wire grid is higher than that of freestanding wire grids in the THz range. The aluminum wire grid is transparent to visible light and also acts as a polarizer in the visible range.¹³⁹⁾ This wire grid is less affected by interference because of its thin substrate and is cost-effective because it can be mass-produced. Moreover, simple polarizers can also be fabricated on paper by printing using commercially available thermal printers and metallic color inks, although the extinction ratio is not high and the frequency range is limited to low frequencies.¹⁴⁰⁾

Zoom In Zoom Out Reset image size

Regarding filters usable in the THz range, thin metallic plates with periodically arranged holes have been used as frequency-selective surfaces (FSSs).¹⁴¹⁾ In particular, arrays with periodically arranged circular apertures (metallic aperture arrays) and structures called metallic meshes are used as band-pass filters.¹⁴²⁾ A dramatic increase in the transmittance of metallic aperture arrays was observed in the light range, which is explained as an effect of SPPs.¹⁴³⁾ A similar effect was also observed in the THz range through spoof SPPs, mentioned in Sect. 3.1. It was demonstrated that the high-sensitivity detection of trace substances is possible using this effect. FSSs with various unit structures or with a multilayer structure have been used on a trial basis and the understanding of the physical aspects of the filter effect has been deepened.

The development of wavelength plates has also started, although there was previously little need for them. An intriguing example is a wavelength plate based on the structural anisotropy of stacked papers, as shown in Fig. 12(b).¹⁴⁴⁾ Similar to the above-mentioned wire grids fabricated using printers, this wavelength plate will challenge the perception that THz optical components are complicated and expensive, providing a direction for future development.

THz waves propagating in free space are frequently used in applications. In the measurement using a far-IR spectrometer shown in Fig. 2(b), a polished brass pipe with an inner diameter of approximately 2 cm, called a light pipe, was used to introduce THz waves into a sample in a cryostat. Although this pipe performed satisfactorily for conventional applications, high-functionality THz waveguides with low loss will be necessary to meet various requirements, such as to freely extract THz waves, as could be done with optical fibers, and to introduce high-intensity THz waves into the human body for the treatment of cancer and other diseases. With this background, research on THz waveguides has become popular in recent years.¹⁴⁵⁾ Figure 12(c) shows a metallic wire waveguide (also known as a Goubau line in energy transmission), which can guide THz radiation at a low loss despite its simple structure.¹⁴⁶⁾ Without modification, however, this waveguide would allow the leakage of electromagnetic fields and be affected by obstacles around the waveguide. Therefore, plastic hollow waveguides (THz fibers) that can be bent to some extent have been proposed.¹⁴⁷⁾

The introduction of metamaterials into THz devices has been actively attempted.¹⁴⁸⁾ Metamaterials are artificial structures with novel properties, including permittivity and permeability, which are not exhibited by natural materials and are realized by arraying unit structures (mainly, metals) with a pitch shorter than the wavelength. The above-mentioned spoof SPPs on the surface of a metal with a microstructure can be considered to induce excitation on the surface of a metamaterial with an effective permittivity attributable to the microstructure. Moreover, conventional wire grids can be considered as planar metamaterials with an extraordinarily high anisotropic permittivity. Figure 13(a) shows a metamaterial in which gold split ring resonators (SRRs) are arrayed in a planar manner on an n-type GaAs thin film and Schottky junctions are formed between the SRRs and the thin film.¹⁴⁹⁾ When no voltage is applied, there are carriers in the gaps between the SRRs and the gaps are electrically closed. Applying a reverse-bias voltage expels the carriers from the gaps, which become open. This on–off operation can greatly modulate the transmittance of THz waves with a certain frequency [Fig. 13(b)]. When arrays of SRRs are arranged on a 4 × 4 pixel array [Fig. 13(c)], a THz wave modulator that can be switched on or off by each pixel can be obtained.¹⁵⁰⁾ Thus far, very thin reflection-free coatings, perfect absorber planes, and wavelength plates have been developed using metamaterials. Moreover, THz metamaterial devices can be easily fabricated using the superfine inkjet printer recently developed by Murata and colleagues at the National Institute of Advanced Industrial Science and Technology.^151,152)

Zoom In Zoom Out Reset image size

In the examination of human skin, compact probes are required to introduce beams to the human body and perform measurements with high stability and reproducibility. Figure 14 shows an example of such a probe. The THz radiation device in the probe is irradiated with a femtosecond laser to radiate THz waves, and the THz waves reflected at the skin are measured by a detector that is also inside the probe.¹⁵³⁾ Tomography images of the skin can be obtained from the time waveform of the reflected THz waves.

Zoom In Zoom Out Reset image size