TERAHERTZ IMAGING: Terahertz parametric oscillator enables sensitive imaging

A narrowband terahertz parametric oscillator creates high-peak-power picosecond terahertz pulses that are suitable for active detection and thus the capturing of high-contrast, real-time images without the use of background-subtraction methods.

PATRICK TEKAVEC

http://www.laserfocusworld.com/articles/print/volume-50/issue-01/features/terahertz-imaging-terahertz-parametric-oscillator-enables-sensitive-imaging.html

The terahertz region of the electromagnetic spectrum (0.1 to 10 THz) is very attractive for nondestructive evaluation applications in defense, security, biomedical imaging, and monitoring of industrial processes. However, performance of terahertz imaging systems is limited by the lack of high-power sources and sensitive detectors in this spectral range. Here we present a novel, high-power terahertz source and propose an imaging method based on upconversion of terahertz waves to infrared (IR) light that offers an orders-of-magnitude increase in contrast over existing systems. The proposed imaging method mixes the terahertz signal with IR light to create an upconverted image that can be detected with commercially available IR cameras.

Microtech Instruments has recently developed a high-power terahertz source, called a terahertz parametric oscillator (TPO), based on difference-frequency generation inside of an optical parametric oscillator (OPO) pumped by a fiber laser (see http://bit.ly/19XjzPh). With a narrow bandwidth of less than 0.1 THz, the output can easily fit through the atmospheric transmission window at 1.5 THz, minimizing attenuation when propagated over large distances. With an average output power of greater than 300 μW, this makes it a good source for near-single-frequency imaging. The pulse duration of 6 ps also gives high peak powers of greater than 450 mW, which makes it a good source to drive nonlinear interactions.

FIGURE. A schematic diagram depicts nonlinear terahertz 2D imaging. A terahertz beam in the far field illuminates the object. A lens is used to image this onto a nonlinear crystal, where the terahertz image is mixed with an optical probe beam and converted to an optical image. The strong probe beam is blocked by a suitable filter, and the image formed by the nonlinear interaction is detected with an array detector.

Active vs. direct detection

There are two main methods to detect terahertz radiation in free space: direct detection with a Golay cell or bolometer, for example, and active detection where terahertz is mixed with an optical field, and an optical detector is used. Direct detectors tend to be slow and lack sensitivity, making them cumbersome for imaging applications. Active detection relies on a nonlinear process, where the terahertz beam is converted to an optical signal. The conversion efficiency of terahertz to optical depends on the peak powers used; thus, short pulses are desirable.

Electro-optic sampling is a well-established active-detection method, where broadband terahertz pulses are mixed with femtosecond optical pulses (probe pulses) in a suitable nonlinear crystal. The terahertz field induces a birefringence, and the change in polarization of the optical field is detected with a polarized oriented to block the probe pulse. In this way the entire terahertz field can be sampled by scanning the delay between terahertz and probe pulses, giving complete amplitude and phase information. This method has been used with a single-element detector with a pixel-by-pixel raster scan to obtain terahertz images. By illuminating the object in the far field as shown in the figure, a complete 2D image can be obtained with an optical-array detector.¹

The quality of the image and speed at which it can be obtained depend of the power in the nonlinear signal and the amount that the probe beam can be attenuated. With broadband pulses, the incident terahertz power is low, and the use of a polarizer as filter won't remove the entire probe beam. This means that in general a weak signal is detected on top of strong background. While it is possible to use background-subtraction techniques to increase image quality, it is desirable to have the ratio of signal to background as high as possible.² This can be expressed as a figure of merit: FOM = (signal power)/(attenuated probe power). The method we propose is to increase this figure of merit by increasing the power in the signal beam and increasing the probe beam attenuation.

Upconversion process

The nonlinear interaction that samples the terahertz beam can be thought of as an upconversion process, where the terahertz and probe beam interact to produce frequency sidebands. If ω0 is the center frequency of the optical beam and ω_THz is the frequency of the terahertz beam, sidebands are generated at ω0 + ω_THz and ω0 - ω_THz. When broadband pulses are used, the sidebands are overlapped with the probe beam, and polarization must be used to remove the background. However, in practice, there is often residual birefringence and scattering that degrade the actual blocking efficiency. With the use of narrowband fields and an appropriate notch filter, spectrally well-separated sidebands can be spectrally.

Upconversion detection has been done with continuous-wave terahertz beams and nanosecond optical pulses.³ However, the resulting signal is weak due to the low peak power in the terahertz beam and low duty cycle of the optical pulses. By using the TPO as a terahertz source, high peak powers from picosecond pulses give a high terahertz-to-optical conversion efficiency, while the narrow spectrum enables efficient removal of background. We expect up to 1 μW of optical power to be generated in the upconversion signal, and a figure of merit improvement of two to three orders of magnitude over existing methods. With an upconverted signal now much larger than the background, high-contrast, real-time images are expected to be obtained without the use of background-subtraction methods.

ACKNOWLEDGEMENT

The work for this is currently underway with the recently awarded Phase 1 SBIR from the National Science Foundation (SBIR IIP-1315916).

REFERENCES

1. Q. Wu et al., Appl. Phys. Lett., 69, 8, 1026 (1996).

2. Z. Jiang et al., Appl. Opt., 39, 17, 2982 (2000).

3. M. J. Khan et al., Opt. Lett., 32, 22, 3248-3250 (2007).

Patrick Tekavec is a research scientist at Microtech Instruments, Inc., 858 West Park Street, Eugene, OR 97401; e-mail: tekavec@mtinstruments.com; www.mtinstruments.com.