Open Access
Add to BibSonomyAbstract
High-resolution tomographic imaging is demonstrated using a reflection-type terahertz time-domain spectroscopy. To realize a practical system for general use, a robust all-fiber laser was used as the pump light source. Broadband terahertz waves were generated with the combination of optical pulses compressed to 17 fs using optical fibers and a DAST crystal. Using deconvolution signal processing, the wideband spectrum of the generated terahertz waves provided high-axial resolution leading to successful imaging of a multilayered structure containing a 2-μm-thin GaAs layer. To our knowledge, this is the first demonstration of terahertz tomographic imaging of such a thin layer.
© 2009 Optical Society of America
1. Introduction
Terahertz sensing and imaging technology has been investigated in various fields, including security, medicine, biochemical applications, and art [1–4]. One of the most powerful tools for terahertz imaging is terahertz time-domain spectroscopy (THz-TDS), with transmission-type THz-TDS widely used for measuring the dielectric response of various materials such as semiconductors, dielectrics, and biological tissues. Two-dimensional (2D) terahertz transmission images are easily obtained by transversely scanning such samples. Furthermore, THz-TDS’s spectroscopic imaging capabilities have recently been demonstrated. Opaque samples having either a high water content or metal backing can be measured with reflection-type THz-TDS, which can be used not only for spectroscopy but also for tomographic imaging [5,6]. Since the electromagnetic field of the sub-picosecond terahertz pulses is measured directly with THz-TDS, the multilayered structure of a sample can be imaged by detecting the echo pulses reflected from each layer. This technique is valuable in industry because the unique transmission characteristics of terahertz waves enables the inspection of multilayered paints in industrial products or tablet coatings [7,8], which are not measurable with optical coherence tomography (OCT) based on an optical light source [9].
As restricted by the temporal duration of the terahertz pulses used, the conventional axial resolution of terahertz tomography remains a few tens of microns, although an attempt to improve the axial resolution of time-of-flight measurements has been reported [10]. In the technique described here, a terahertz interferometer was added to the reflection-type THz-TDS configuration to attain a resolution of 12.5 μm. Although this method has proven effective for thickness measurement of single layer, it had not been investigated as a tomography system whereby multilayered structures are imaged. The common ∼10 μm thickness of paints in various industrial products and the stratum corneum of human skin require development of higher-resolution terahertz tomography for industrial, medical, and cosmetic applications important. One of the most effective and simplest ways to improve the axial resolution of tomographic imaging is by generating broadband and short terahertz pulses using shorter optical pulses.
Practicality is also a critical issue in terahertz tomography. Most terahertz tomography systems heretofore reported use a Ti: sapphire laser as a pump light source because they meet stringent specifications and are suitable for generating and detecting terahertz pulses. However, for real applications in medical practices or factories, a more robust system that counteracts vibration and temperature variations is preferable. An ultrashort-pulse fiber laser would be a strong candidate, except for the belief that generating broadband ultrashort terahertz pulses with it would be difficult because its pulse width is longer than that of the Ti: sapphire laser. However, with the appropriate control of the nonlinear and dispersion effects in optical fibers, sub-20-fs optical pulses can be generated [11–13]. We have thus developed an all-fiber laser system that produces 17-fs optical pulses at a wavelength of 1.56 μm and have generated and detected broadband terahertz pulses ranging from 0.1 to 27 THz [14].
In this paper, we describe our high-resolution terahertz tomography system based on reflection-type THz-TDS. In this system, an all-fiber ultrashort-pulse laser is used to realize a system sufficiently robust for practical applications. To achieve a high resolution, broadband terahertz pulses generated using a 17-fs fiber laser and a DAST crystal are used as a terahertz source. Since the temporal waveform of the generated terahertz pulses has several multiple peaks, the deconvolution signal processing with Gaussian window is applied, which results in single-peaked and ultrashort terahertz pulses. Finally, the high resolution of our system is demonstrated by successfully imaging a multilayered structure containing a 2-μm-thick GaAs layer.
2. Experiment
The time-of-flight terahertz tomography experimental setup is shown in Fig. 1. Output pulses from a 1.56-μm fiber laser oscillator were fed into an erbium-doped fiber amplifier. The pulses were then compressed using a large mode-area photonic crystal fiber and a highly-nonlinear fiber [14]. The inset of Fig. 1 shows the temporal waveform of the compressed optical pulses. The average power, pulse width, and repetition frequency of the generated optical pulses were 200 mW, 17 fs, and 48 MHz, respectively.
Fig. 1 Experimental setup for the fiber-laser, high-resolution time-of-flight terahertz tomography system. The inset shows the temporal waveform of the 17-fs fiber laser: BS: beam splitter; PPLN: periodically poled lithium niobate; DM: dichroic mirror; PCA: photoconductive antenna; Ge: germanium plate.
The optical pulses were divided into two beams for terahertz wave generation and detection using a 50:50 broadband beam splitter. To avoid the dispersion effect, the 1-mm thin plate beam splitter was used instead of a cube beam splitter. The terahertz waves were radiated from a 0.1-mm-thick DAST crystal. A germanium plate was inserted as a near-infrared filter to eliminate laser pulses transmitted through the crystal. The waves were collimated and focused onto the sample using a pair of 101.6-mm focal length parabolic mirrors whose location was optimized to achieve a minimum incident angle of 7°. The transverse resolution of the system was determined by the 0.9-mm spot size of the focused beam as measured by a knife-edge method. The terahertz pulses reflected from each layer of the sample were also focused on a 5-μm electrode gap LT-GaAs photoconductive antenna using parabolic mirrors. The temporal waveform of the terahertz waves was traced by scanning a delay stage whereby SHG pulses generated with a periodically-poled lithium niobate were used as probe pulses. The signal obtained was then fed into a preamplifier and detected with a lock-in amplifier. To reduce water vapor absorption effects during the tomography measurement, the terahertz wave path was purged with dry nitrogen gas. The 2D and 3D terahertz tomography images were obtained by scanning a sample in x- and y-directions. The maximum scan range was 30 mm in each direction.
Figure 2(a) shows the temporal waveform of the detected terahertz pulses with a flat aluminum mirror used as an example. The inset shows the corresponding spectrum obtained by Fourier transforming the temporal waveform. The maximum amplitude is observed at 1 ps with several small oscillations superimposed on this peak. The corresponding spectrum extended well beyond 20 THz; its trace is not flat due to either the absorption of the DAST crystal or phase mismatching. As shown in these figures, the temporal waveform of the detected terahertz pulses was distorted compared to those radiated from a photoconductive antenna. Signal processing is therefore required to reshape the temporal waveform and suppress the tomographic ghost images. In this work, the detected temporal waveform was analyzed using a deconvolution technique whereby the signal detected with another flat mirror was used as a reference. The detected signal reflected from the sample y(t) was expressed by the convolution of the reference signal reflected from a flat mirror x(t) and the transfer function of the multilayered sample h(t).
In the Fourier domain, this equation is expressed as follows. Therefore, the transfer function of the sample is obtained by the deconvolution process. Here, since the intensity of X(ω) is quite small at several frequency regions, we used a Weiner filter instead of the inverse filter to avoid divergence. The Weiner filter is defined as where a is the ratio of the signal power to the noise power, and we here assumed it as a constant. Using Weiner filter, the transfer function is calculated as follows. In addition, the window function was applied to this equation to achieve the single-peaked temporal waveform without sidelobes. To obtain clean half-cycle pulses, the Gaussian function was used as a window function. Since the Gaussian-shaped waveform with the pulse width at full width at half maximum TFWHM is expressed as the transfer function of the sample h(t) is denoted by the following equation. When TFWHM = 100 fs, the resultant temporal waveform is shown in Fig. 2(b). The corresponding spectrum is shown in the inset of Fig. 2(b) and has a Gaussian-like waveform. As shown in this figure, a single-peaked sidelobe-free pulse was obtained. A shorter pulse is obtained by broadening the Gaussian window bandwidth, but small side peaks arise in the temporal waveform due to an inverse relationship between temporal duration and sidelobe magnitude.Fig. 2 Temporal waveform (a) as detected and (b) after signal processing. The insets show the corresponding Fourier transforms. The gray curve represents a Gaussian-shaped waveform.
The Teflon film thickness was subsequently measured to evaluate the system’s axial resolution. Figure 3(a) and (b) show the measured temporal waveform and waveform after the deconvolution process for a 12-μm-thick Teflon film sample, respectively. Although the peaks corresponding to the front and back sides of the film were not distinguished in Fig. 3(a), two large peaks are clearly observed in Fig. 3(b) due to the effect of the deconvolution process. These peaks have opposite signs due to phase inversion by reflection at the back of the film. The film thicknesses were calculated using the delay time of these peaks and the refractive index of the film. The experimental results obtained for several thicknesses of film are summarized in Fig. 3(c). The horizontal and vertical axes show the actual and measured thicknesses, respectively, where a 1-μm-resolution micrometer was used to measure the actual film thickness. In this figure, film thicknesses of 5, 12, 20, 25, and 30 μm were measured correctly. That is, the axial resolution is sufficient to measure 5-μm-thick Teflon film. Since the refractive index of the Teflon film is 1.9, the resolution of our system is less than 10/n μm, where n is the refractive index of the sample. As respects the accuracy of this system, the standard deviation was smaller than 0.7 μm.
Fig. 3 Evaluation of axial resolution using Teflon films. (a) Typical measured temporal waveform, (b) waveform after deconvolution process, and (c) measured thickness of Teflon films as a function of actual thickness. The actual thickness was measured using a micrometer.
Lastly, terahertz tomography imaging was demonstrated. Figure 4 shows the three-dimensional (3D) tomography image of three sheets of 90-μm-thick paper. The scanning step of the sample was 0.1 mm, and the pixel size was 10 mm×10 mm. The front and reverse sides of each sheet can be seen clearly. With the high resolution of the system, not only are the thicknesses of the paper visible, but the gaps between the sheets are also seen.
Figure 5 shows terahertz tomography imaging of a semiconductor sample. An approximately 2-μm GaAs layer was adhered to a 300-μm silicon substrate using a polymer resin; the schematic diagram of this sample is shown in Fig. 5(a). The refractive indices of the GaAs layer, silicon substrate, and polymer resin at 1 THz are 3.6, 3.4, and 1.6, respectively. Figure 5(b) shows the temporal waveform after deconvolution process obtained to measure the thickness of the GaAs layer, which is reflected from the point indicated by the red arrow. Since the GaAs layer had a larger refractive index than the polymer resin, the phase of the second peak was inverted. The time delay between these two peaks, which correspond to the front and reverse sides of the GaAs layer, was 43 fs. Taking the refractive index of GaAs into account, this time delay corresponds to a 1.8-μm-thick GaAs layer. Because the refractive index of the GaAs layer is about twice that of the Teflon film, the thickness of a layer thinner than that shown in Fig. 3(b) was also measured. Figure 5(c) shows the 2D terahertz tomography image obtained by scanning the sample with 0.5 mm step along the direction of the blue arrow. The silicon substrate and the polymer resin are obviously distinguished. Then, to confirm that the thin GaAs layer was correctly imaged, the area around the GaAs layer was magnified and the color readjusted to more easily observe the thin layer as shown in Fig. 5(d). In this figure, the GaAs layer shown in Fig. 5(a) is observed at 0–3 mm. To the best of our knowledge, this is the first demonstration of tomographic imaging of such a thin-layered structure using terahertz waves.
Fig. 5 Terahertz tomography of a semiconductor sample: (a) schematic diagram of the sample, (b) temporal waveform after deconvolution process obtained to measure the thickness of GaAs, (c) measured tomography image, and (d) close-up around the GaAs layer.
3. Conclusion
In conclusion, the development of high-resolution time-of-flight terahertz tomography using a 17-fs fiber laser was presented. Broadband terahertz pulses obtained using a DAST crystal were injected into samples and the echo pulses reflected from each layer were detected. The deconvolution process then was carried out with a Gaussian window to obtain clean, high-resolution tomographic images. For a sample with a refractive index of n, the axial resolution was below 10/n μm, as determined by comparing with mechanical thickness measurements of PET films. The high-resolution capability of our system provided the first tomography images of multilayered structures built up of approximately 2-μm-thick layers. This high-resolution capability coupled with the practicality of an all-fiber-laser-based system will lead to further applications of terahertz tomography in the industrial, medical, and cosmetic fields. For these applications, the generation and detection of terahertz pulses with broader and more flat spectrum will be required.
Acknowledgment
This work was partly supported by a Grant-in-Aid for Scientific Research (18206009) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.
References and links
1. K. Kawase, Y. Ogawa, Y. Watanabe, and H. Inoue, “Non-destructive terahertz imaging of illicit drugs using spectral fingerprints,” Opt. Express11, 2549–2554 (2003), http://www.opticsinfobase.org/abstract.cfm?URI=oe-11-20-2549. [CrossRef] [PubMed]
2. Y. C. Shen, T. Lo, P. F. Taday, B. E. Cole, W. R. Tribe, and M. C. Kemp, “Detection and identification of explosives using terahertz pulsed spectroscopic imaging,” Appl. Phys. Lett. 86, 241116 (2005). [CrossRef]
3. Y. Ogawa, S. Hayashi, M. Oikawa, C. Otani, and K. Kawase, “Interference terahertz label-free imaging for protein detection on a membrane,” Opt. Express16, 22083–22089 (2008), http://www.opticsinfobase.org/oe/abstract.cfm?uri=oe-16-26-22083. [CrossRef] [PubMed]
4. K. Fukunaga, Y. Ogawa, S. Hayashi, and I. Hosako, “Terahertz spectroscopy for art conservation,” IEICE Electron. Express 4, 258–263 (2007). [CrossRef]
5. D. M. Mittleman, R. H. Jacobsen, and M. G. Nuss, “T-ray imaging,” IEEE J. Sel. Top. Quantum Elctron. 2, 679–692 (1996). [CrossRef]
6. D. M. Mittleman, S. Hunshe, L. Boivin, and M. G. Nuss, “T-ray tomography,” Opt. Lett. 22, 904–906 (1997). [CrossRef] [PubMed]
7. A. J. Fitzgerald, B. E. Cole, and P. F. Taday, “Nondestructive analysis of tablet coating thickness using terahertz pulsed imaging,” J. Pharm. Sci. 94, 177–183 (2005). [CrossRef] [PubMed]
8. T. Yasui, T. Yasuda, K. Sawanaka, and T Araki, “Terahertz paintmeter for noncontact monitoring of thickness and drying progress in a paint film,” Appl. Opt. 44, 6849–6856 (2005). [CrossRef] [PubMed]
9. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, C. A. Puliafito, and J. G. Fujimoto, “Optical coherence tomography,” Science 254, 1178 (1991). [CrossRef] [PubMed]
10. J. L. Johnson, T. D. Dorney, and D. M. Mittleman, “Enhanced depth resolution in terahertz imaging using phase-shift interferometry,” Appl. Phys. Lett. 78, 835–837 (2001). [CrossRef]
11. Y. Matsui, M. D. Pelusi, and A. Suzuki, “Generation of 20-fs pulses from a gain-switched laser diode by a four-stage soliton compression technique,” IEEE Photon. Technol. Lett. 11, 1217–1219 (1999). [CrossRef]
12. M. Tsuchiya, K. Igarashi, S. Saito, and M. Kishi, “Sub-100 fs higher order soliton compression in dispersion-flattened fibers,” IEICE Trans. Electron. E85-C, 141–149 (2002).
13. T. Hori, N. Nishizawa, T. Goto, “Generation of 14 fs ultrashort pulse in all fiber scheme by use of highly nonlinear hybrid fiber,” in Ultrafast Phenomena XIV, T. Kobayashi, et al., eds. (Springer-Verlag, Berlin, 2005), pp. 31. [CrossRef]
14. J. Takayanagi, S. Kanamori, K. Suizu, M. Yamashita, T. Ouchi, S. Kasai, H. Ohtake, H. Uchida, N. Nishizawa, and K. Kawase, “Generation and detection of broadband coherent terahertz radiation using 17-fs ultrashort pulse fiber laser,” Opt. Express16, 12859–12865 (2008), http://www.opticsinfobase.org/abstract.cfm?uri=oe-16-17-12859. [CrossRef] [PubMed]
