MARTIN KOCH
While nearly all parts of the electromagnetic spectrum are in frequent use, the far-infrared, or terahertz region from 100 GHz to 10 THz, is still relatively unexploited. Like microwaves, terahertz radiation can easily penetrate plastics, paper, wood, and clothing. Metals and polar liquids like water are, however, opaque. Since terahertz waves have a much smaller wavelength than classical microwaves they can provide submillimeter spatial resolution that makes them interesting for many imaging applications including security checks of people, mail, or luggage, for process control of polymeric compounding, and for detection of contaminated food.
More challenging than inspection applications are terahertz short-range indoor communication systems with carrier frequencies of a few hundred gigahertz that could replace the next generation of wide local-area-network (WLAN) systems in 10 to 15 years.
Although a clear market potential for all these applications has been identified, terahertz systems, and in particular sources and detectors, must become more efficient and significantly drop in price for their potential to be fully realized.
Traditional terahertz sources
Discovered in the early years of the 20th century by Max Planck, terahertz waves are nothing special—they are emitted by all humans and by the matter around us because we are all “blackbodies.” However, the efficiency of a blackbody source is not very high. The spectral power density of the sun, for example, is 20,000 times higher at visible frequencies than at one terahertz. Using a globar—a bonded silicon carbide or carborundum rod heated to very high temperature that simulates a blackbody source—is not very practical because all unwanted frequencies emitted along with the terahertz waves must be carefully filtered out.
Terahertz emitters can be realized using microwave technology based on Gunn, Impatt, or resonant tunneling diodes. The fundamental frequencies of these devices, however, are typically not high enough for many terahertz applications and they need to be multiplied in special mixers. Although a microwave-based terahertz source can easily fit into a shoebox, its cost can exceed several tens of thousands of dollars.
A less-compact class of terahertz emitters based on an electron beam includes synchrotron sources, free-electron lasers, the Smith-Purcell emitter, and backward-wave oscillators. In a synchrotron and free-electron laser, electrons are sent through a region of alternating magnetic fields in which they jiggle back and forth. This oscillatory electron movement leads to the emission of terahertz radiation. Although both sources are very powerful and can be tuned over a wide frequency range, it is said that they “come with their own zip code” because of their very large dimensions.
In the somewhat more-compact Smith-Purcell emitter, the electron beam from an electron microscope propagates along the surfaces of a metallic grating, causing a magnetic-field effect similar to that of the free-electron sources. And at about the size of a football, backward-wave oscillators or carcinotrons are electro-vacuum devices in which electrons fly over a comb-like structure, grouping them in periodic bunches that emit terahertz radiation. Although these backward-wave oscillators are powerful sources that can produce 10 mW of monochromatic but tunable terahertz power, they are only produced in Russia and can cost $24,000 (€20,000) or more.
Laser sources
Terahertz waves can also be generated by traditional laser sources. One type is a molecular gas laser that relies on transitions between different rotational levels of a molecular species. These lasers can emit several milliwatts of power at discrete terahertz frequencies ranging from less than 300 GHz to more than 10 THz. Many frequencies can be chosen because the list of rotational transition-rich gases is not small and includes methanol, methyl fluoride, and formic acid, to name a few. Although Coherent (Santa Clara, CA) developed a class of surprisingly compact terahertz gas lasers, these systems are still prohibitively expensive for most applications.
Semiconductor lasers are more compact and less expensive. Unfortunately, the traditional ones need cryogenic cooling. Although somewhat out of fashion, lead-salt lasers can be good sources for higher terahertz frequencies. They rely on interband transitions in lead-containing semiconductors—lead selenide (PbSe), for instance. Introduced in the early 1980s, p-germanium lasers use hole transitions from the light-hole to the heavy-hole band and deliver strong terahertz pulses that can be tuned in frequency. But because the p-germanium laser only works at low temperatures and requires a magnetic field, it is not practical for applications outside the lab.
In the early 1990s, a big step in terahertz technology was made by the advent of reliable femtosecond lasers for the generation and detection of terahertz pulses. The oldest and probably most popular scheme involves photoconductive dipole antennas that are gated by the femtosecond pulses. These antennas consist of a semiconducting substrate onto which a metallic antenna structure is deposited by photolithography techniques. An external bias is applied to the antenna structure, which comprises a small gap to prevent a short. Current can only flow under the action of a laser pulse that optically excites the semiconducting antenna substrate in the gap. The generated electron-hole pairs are accelerated in the bias field and cause a short current pulse with a subpicosecond rise time. According to Maxwell’s law, this is the source for a short, and hence, broadband terahertz pulse. If an unamplified Ti:sapphire laser is used for excitation, the continuous-wave (CW) power level is in the microwatt range.
In a terahertz time-domain spectrometer, the terahertz pulse is detected with a second antenna of similar design. This generation/detection scheme is very powerful and ideally suited for imaging (see Fig. 1). Because the terahertz pulses are broadband, even spectroscopic information on the imaged object can be obtained. Furthermore, internal voids or interfaces can be visualized.
Improving terahertz sources
During the last ten years, many attempts have been made to make photoconductive terahertz antennas more efficient. Different semiconductor materials and antenna geometries have been tested. The most recent improvement was a large-area photoconducting emitter with an interdigitated electrode metal-semiconductor-metal (MSM) structure.1 A second metallization layer isolated from the MSM electrodes blocks optical excitation in every second period of the MSM finger structure. Because charge carriers are optically excited only in those periods of the MSM structure that exhibit a unidirectional electric field, the carriers are accelerated unidirectionally over the whole excited area and the terahertz radiation emitted through the substrate interferes constructively in the far field. This concept is scalable and emitters with sizes of a few square millimeters are conceivable.
Simultaneously, femtosecond lasers are becoming more compact and are dropping in price. By using different semiconductor materials such as indium gallium arsenide (InGaAs), antenna-based terahertz systems could work without Ti:sapphire lasers and use fiber lasers. Although this room-temperature technology will further advance, it will still cost more than $48,000 (€40,000), most of which has to be spent on the laser source.
A different scheme to excite photoconductive terahertz antennas uses two CW diode lasers that emit at slightly different frequencies. The emissions of these lasers are superimposed on the antenna and the resulting light beat is converted into an oscillating antenna current that is the source for a monochromatic terahertz wave. For the pulsed case, the power level of this CW radiation is in the microwatt range. In 2001, researchers made this so-called photomixing scheme more compact and less expensive by replacing the two diode lasers with a single laser diode.2 By placing this diode in a special external resonator it runs on two lines simultaneously.
Very recently, it was demonstrated that this “two-color laser” emits terahertz radiation via a nonlinear four-wave-mixing process.3 Terahertz emission was detected with a broadband indium antimonide (InSb) bolometer positioned directly behind the end facet of the laser. This new room-temperature terahertz source can be produced in a rather compact footprint and at a price below $12,000 (€10,000). Even though the detected terahertz power level is in the subnanowatt range, it can potentially be increased by using laser diodes with higher power and by modifying the laser geometry to provide waveguiding not only for the optical but also for the terahertz wave.
Another very promising technology to realize room-temperature, monolithic, and compact current-injected sources in the 1- to 5-THz range is represented by quantum-cascade lasers (QC lasers). First demonstrated in 1994, QC lasers rely on the emission from transitions between subbands in a quantum well.4 By applying a voltage to a complicated sequence of repeating quantum-well structures, electrically injected electrons stream down a potential staircase and emit a terahertz photon at each step. Early QC lasers needed cryogenic cooling, worked only in a pulsed mode, and emitted in the mid-infrared. Since then, the race has been toward CW operation, higher temperatures, and longer wavelengths.
Today, mid-IR CW QC lasers are ready for industrial applications.5 Until the late 1990s it was believed that working frequencies could never drop below 5 THz. Yet, in 2002, QC lasers working at 4.4 THz were demonstrated.6 Current QC laser sources can have working frequencies as low as 2.1 THz, and a CW 3.2-THz QC laser has been reported.7 Although compact terahertz sources can be built using portable, lightweight, closed-cycle Stirling coolers, room-temperature terahertz QC lasers still remain to be seen.
Finally, less-compact but clearly room-temperature terahertz-wave parametric generators were introduced in the mid-1990s.8 These sources use the parametric conversion of nanosecond laser pulses from Q-switched, diode-pumped Nd:YAG lasers. The conversion takes place in a nonlinear crystal, for example, magnesium oxide:lithium niobate (MgO:LiNbO3). The emitted terahertz radiation is narrowband, widely tunable, and so powerful that it can be detected with a pyroelectric detector operating at room temperature. With an estimated price of approximately $30,000 (€25,000), this system was extended for imaging in 2003 and identified different chemicals arranged in a pattern using spectroscopic information. Despite these advances, small, efficient, and inexpensive room-temperature terahertz sources, either pulsed or continuous wave, may remain a dream for at least the next ten years.
REFERENCES
1. A. Dreyhaupt et al., Appl. Phys. Lett. 86, 121114 (2005).
2. T. Kleine-Ostmann, et al., Elec. Lett. 37, 1461 (2001).
3. S. Hoffmann et al., Appl. Phys. Lett. 84, 3587 (2004).
4. J. Faist et al., Science 264, 553 (1994).
5. L. Mechold and J. Kunsh, Laser Focus World 40(5) 88 (May 2004).
6. R. Koehler et al., Nature 417, 156 (2002).
7. S. Kumar et al., Appl. Phys. Lett. 84, 2494 (2004).
8. Y. Watanabe et al., Appl. Phys. Lett. 83, 802 (2003).
Martin Koch is associate professor at the Institut für Hochfrequenztechnik at Technische Universität Braunschweig, Schleinitzstrasse 22, 38106 Braunschweig, Germany; e-mail: [email protected].