Investigation of Terra Cotta artefacts with terahertz

Transcription

1 Appl Phys A (2011) 105:5 9 DOI /s x RAPID COMMUNICATION Investigation of Terra Cotta artefacts with terahertz Julien Labaune J. Bianca Jackson Kaori Fukunaga Jeffrey White Laura d Alessandro Alison Whyte Michel Menu Gerard Mourou Received: 23 May 2011 / Accepted: 21 July 2011 / Published online: 31 August 2011 Springer-Verlag 2011 Abstract Terahertz Time Domain Imaging has been used in the last few years for the investigation of cultural heritage. In this article, the authors demonstrate the possibility to apply it for the investigation of clay artifacts. Tomographic images were obtained of a model in reflection, and an Egyptian vessel in transmission. Introduction In this paper, we propose to use terahertz time domain spectroscopic imaging (THz-TDSI) as a nonionizing, nondestructive, and noncontact technique for the investigation of clay artifacts. This technique has already been used for the investigation of multilayer papyrus [1], canvas [2], wood panel [3], mummies [4] or pigment identification [5], and more generally in art work. We report a study of clay artefacts by using terahertz in transmission and reflection configurations. From the earliest J. Labaune J.B. Jackson G. Mourou Institut de Lumière Extrême, Chemin de la Hunière, Palaiseau 91120, France J. Labaune J.B. Jackson M. Menu ( ) Centre de Recherche et de Restauration des Musées de France, quai François Mitterrand, Paris 75001, France michel.menu@culture.gouv.fr K. Fukunaga National Institute of Information and Communications Technology, Nukui-Kitamachi Koganei, Tokyo , Japan J. White Picometrix, 2925 Boardwalk, Ann Arbor, MI 48104, USA L. d Alessandro A. Whyte Oriental Institute, 1155 East 58th Street, Chicago, IL 60637, USA civilizations, man has used clay from the earth to create objects of necessity and decoration from tools to writing surfaces, from dishware to storage containers, from religious idols to burial urns. Archaeologists have recovered many of these objects from their excavations, some of which are still sealed. X-ray radiography and computed tomography are the most common techniques with which conservation scientists can investigate the contents of closed containers while preserving its structure. However, due to the potential risk of the radiation affecting [6, 7] the results of thermoluminescence dating and the protection necessary for radiography, there is an interest in developing alternative techniques. This paper will present preliminary work on identifying objects in clay enclosures, which include phantoms hidden behind clay tablets and a sealed Egyptian clay vessel found in the Youya and Touya tomb from the 14th century BC. To the best of the authors knowledge, this is the first observation of a large dimension, ceramic craft object using terahertz imaging. Additionally, the authors have applied a corrective algorithm to compensate for the distortions in the reconstructed image due to the thickness and shape of samples. 1 System and sample description Terahertz image were recorded by spatially scanning the sample on the focal point of a standard THz Time Domain Spectrometer system (TDS). For these experiments, a Picometrix T-Ray 4000 with 100-Hz waveform acquisition rate was used. The broadband THz pulses were generated and detected by photo-conductive antennas fiber-coupled to a femtosecond laser. The measured signal includes all of the amplitude and phase information of the electric field and gives spectroscopic access to the dielectric properties of the sample. It can be used in the transmission configuration or in

2 6 J. Labaune et al. the reflection configuration at normal incidence using a pellicle beamsplitter. The frequency range is between 0.1 and 2.5 THz in transmission and 1.5 THz in reflection. Silicon and HDPE lenses of different focal length were used to focus the terahertz beam. For the phantoms, modern white clay (Beck-ceramique) was used to produce planar tablets of different thickness, which were baked in a kiln by a professional in pottery workshop. At the end of this process four square tablets of cm 2, of thicknesses physically measured to be between 4 and 14 mm, were used for the experiments. The Egyptian clay vessel was 53 cm in height with a maximum diameter of 28 cm; and since it was sealed, it was not possible to physically measure the thickness of the walls, nor determine its contents. 2 Spectroscopy In preparation for the imaging experiment, we measured the complex index of refraction of the baked clay. Each tablet was placed into the transmission setup at the focal plane between two 3-inch focal length lenses. Each was measured twice over six months at ten different positions with an average of 1000 waveforms. A reference measurement of air was taken under the same experimental conditions. The plot in Fig. 1a of the transmission of clay tablets (5, 7, 10, and 14 mm thick) as a function of frequency shows that the bandwidth decreases significantly as a function of the increasing clay thickness. While it is possible to use 1 THz for a 5-mm-thick layer, for 14 mm, only 0.5 THz is outside the noise floor and usable for imaging. This is not an insignificant factor when considering the spatial resolution restrictions for even diffraction limited spot sizes. The extraction of the complex refractive index [8, 9], plot on Fig. 1b, ñ(ν) = n(ν) + iκ(ν) of the clay used gives the real index, n, which is constant near 1.95 between 0.1 and 1.2 THz and comparable to the value found by Piesiewicz et al. [10]. On the other hand, the imaginary part of the index, κ, increases with the frequency; however, above 1.2 THz, the signal loss due to scattering was too high, and it was not possible to extract reliable dielectric index information from these samples. 3 Reflection imaging In the normal incidence reflection configuration, the terahertz emitter and receiver are perpendicular to each other with a pellicle beamsplitter at 45 degrees between the two antennas. The collimated terahertz beam was focused by a single lens, either 1 or 3 in focal length. The lens selection can be optimized as a function of the dimensions of the object for a better ratio of spot size to depth resolution or penetration depth. In this case, the wider angle lens provided better spatial resolution; however, the Rayleigh range was also shorter, which restricted the depth of analysis. Mittleman et al. [11] demonstrated that it is feasible to reconstruct a 3D time model, or tomogram, of a multilayered sample by using the delay between the reflections and amplitude to know the refractive indices of the elements in the system. This technique can only be used with low absorbent materials. In the case of clay, the index is too high to use this algorithm. The tomogram in Fig. 2a shows the general arrangement between the clay top layer and the phantom (in this case, about 4.7 mm of clay and a 5 coin). The separation distance and relative orientation between the clay and the phantom were varied as a function of this experiment. The focal plane remained constant at the top surface of the phantom, even as the optical path length changed, and was determined by the optimization of the reflected terahertz signal. Fig. 1 Index and absorption of clay tablet for different thickness

3 Investigation of Terra Cotta artefacts with terahertz 7 Fig. 2 THz image of a coin through a layer of clay a tomogram, b tomogram of the coin, c tomogram of the coin after correction Fig. 4 Fusion of terahertz and optical image Fig. 3 a THz image of the unobscured coin and b THz image of the coin through 5 mm of clay in a nonplanar configuration A 3D reconstruction was rendered (Fig. 2b) by plotting the amplitude of the interface reflection at its respective temporal positions over the scan area. The time domain signal was denoised using a THz bandpass filter before applying a peak detection algorithm to find to temporal locations of each surface-air interface (a, b, c). However, there are distortions in the surface reconstruction of the coin due to the variations in thickness and surface flatness of the clay top layer. It is possible to extract the precise thickness of the clay at each pixel, d i, by using the time-of-flight between the front and back surface reflections and the refractive index of the clay: d i = c(t ib t ia ) n bulk where c is the speed of light, a and b are the first and second clay-air interfaces, respectively, and n bulk is the real index of refraction between 0.1 and 1.2 THz. Therefore, with this information, it is possible to correct the wave front of the terahertz pulse after it has propagated through the clay. As with the flattened coin surface in Fig. 2c, this technique can be used to correct the shape of an unknown internal object. An unobscured terahertz image of the surface of the coin can be seen in Fig. 3a. Figure 3b shows the surface image through a clay layer positioned at an angle with respect to the coin. It still is possible to distinguish some of the texture from the coin s relief despite the millimeter spatial resolution of the system. If the top layer is more complicated than a flat layer, the previous algorithm may be applied necessary to produce a corrected picture. Attenuation of the terahertz signal can be significant due to its high absorbance by clay. For example, a 350 microns difference in thickness for the top layer results in a 3% difference in amplitude at 0.32 THz. By using the extracted clay thickness at each pixel, it is possible to correct the spectral amplitude by dividing it by the absorbance (thickness times the absorption). With a mask, a correction of the image is possible to increase the contrast. This correction is not limited to clay, but will be useful for other artwork absorbent material. For example, in the case of real walls with frescos covered by a nonuniform layer of plaster, the difference of reflectivity between pigments is small, and the correction will be necessary. 4 Transmission imaging Last experiments were performed on the Egyptian vessel in transmission with a system in the conservation laboratory at the Oriental Institute. The emitter and the receiver were moved around the object for the purpose of protecting the artifact. Regrettably, the 12-inch diameter of the ob-

4 8 J. Labaune et al. Fig. 5 Transmission pictures for different orientation: a 0, b 20, c 40 of rotation Fig. 6 Signal for different angles ject required the 8-inch focal length lenses to have a separation greater than 18 inches. The vessel was scanned in five cm sections with a step size of 1.25 mm using a motorized translation stage in under two hours. Figure 4 shows a fusion of a photograph and the terahertz images of the container. At the top of the neck, the terahertz beam was absorbed by the mud plug which seals the vessel. We observed a low-density, lightly scattering material near the bottom of the neck, which, based on similar artifacts, is believed to be some kind of textile. The contents of the jar s body are very dense and highly scattering of the terahertz radiation. Thickness of the container walls was estimated by assuming that the refractive index of this baked clay is comparable to our clay models. The delay between a reference pulse without the vessel and a pulse propagating through a section of the vessel with no apparent content is about 55 ps. This corresponds to 18 mm of material, or approximately 0.9-cm-thick walls. A dark line appears in the image of the right side of the container; therefore that area was scanned at two additional angles (Fig. 5) in order to provide further insight into whether the phantom is a two-dimensional crack in the clay or a three-dimensional object within the container [12, 13]. Generally, the contents appear to be mostly granular, and clusters have asymmetrically shifted within the vessel as a result of horizontal transport and a slow vertical resettling. The time of flight of the terahertz pulse through the container was extracted for one horizontal line scan at the same vertical position for each rotation (Fig. 6). The increasing time delay as the scan approaches the edge of the container (near the 68th pixel) of the plots can be attributed to the increase in material at the edge due to the curvature of the walls. Deformations in the three curves, at pixels corresponding to the location of the phantom, indicate an extra temporal delay as a consequence of extra material. This is a promising sign that the phantom is not a crack. The location of the phantom was triangulated, using its position at each of the three angles, and compared to the radius of the vessel to further corroborate the likelihood that it is a physical object within the container. Conclusion In this article, we demonstrated that terahertz Time Domain Imaging can be a powerful tool to image large clay artifacts. We performed spectroscopic measurements on clay models and imaged modern phantoms through thick clay material. We used reflection measurements and signal processing to produce tomograms and enhance surface images of objects enclosed in clay. We were able to image the contents of an Egyptian funeral vessel, in transmission, and determine the internal location of a phantom. This work was supported by CHARSIMA pro- Acknowledgements gram. References 1. J. Labaune, J.B. Jackson, S. Pags-Camagna, I.N. Duling, M. Menu, G.A. Mourou, Appl. Phys. A 100(3), 607 (2010) 2. K. Fukunaga, I. Hosako, C. R. Phys. 11(7 8), 519 (2010) 3. J.B. Jackson, M. Mourou, J. Labaune, J.F. Whitaker, I.N. Duling, S.L. Williamson, C. Lavier, M. Menu, G.A. Mourou, Meas. Sci. Technol. 20(7), (2009)

5 Investigation of Terra Cotta artefacts with terahertz 9 4. L. Öhrström, A. Bitzer, M. Walther, F.J. Rühli, Am. J. Phys. Anthropol. 142(3), 497 (2010) 5. K. Fukunaga, M. Picollo, Appl. Phys. A 100, 591 (2010) 6. M.S. Tite, J. Waine, Archaeometry 5(1), 53 (1962) 7. M.S. Tite, Archaeometry 9(1), 155 (1965) 8. I. Pupeza, R. Wilk, M. Koch, Opt. Express 15(7), 4335 (2007) 9. L. Duvillard, F. Garet, J.L. Coutaz, IEE 2(3), 739 (1996) 10. R. Piesiewicz, T. Kleine-Ostmann, N. Krumbholz, D. Mittleman, M. Koch, T. Kurner, Int. J. Infrared Millim. Waves 8(5), 363 (2007) 11. W.L. Chan, J. Diebel, D. Mittleman, Rep. Prog. Phys. 70, 1325 (2007) 12. S. Wang, X.-C. Zhang, J. Phys. D 37(4), 1 (2004) 13. B. Recur, A. Younus, S. Salort, P. Mounaix, B. Chassagne, P. Desbarats, J.-P. Caumes, E. Abraham, Opt. Express 19, 5105 (2011)