Cork quality estimation by using Compton tomography
Introduction
The products obtained from cork bark represent an important economical resource for several countries. Among such products, stoppers for wine and champagne represent the most lucrative product [1], [2]. The principal requirements for cork stoppers are the homogeneity of the cork and the lack of cavities and/or cracks. The quality control are performed in several steps. However, most of these controls are visual checks performed by experienced people and/or by electronic cameras [3], [4]. All these kinds of inspection allow to analyze only the external surface. Thus cracks or holes inside the stopper will not be detected.
The computed tomography is a non-destructive technique that allows one to reconstruct cross-sections of the sample analyzed starting from a set of X-ray radiographs taken at different angles all around the sample. The most diffused tomographic technique is the transmission tomography. It is performed by placing an X-ray tube and a detector aligned at opposite sides of the sample. The detector records that part of radiation from the X-ray tube that does not interact with the sample. This technique have been well developed for industrial and medical applications along the last 30 years [5], [6]. The reconstructed section represents a map of absorption coefficients of the material. Thus, the contrast inside the image will be given from different values of absorption. The attenuation of the radiation can be expressed bywhere I0 is the radiation emitted by the X-ray source, I is the radiation recorded by the detector, μ(E,x) is the linear absorption coefficient and it is a function of the position, i.e. composition and of the energy E.
The absorption coefficient depends on the chemical nature of the sample. Although μcork>μair, it is not, unfortunately, very much larger at X-ray energy above 20 keV, i.e. these usually utilized in X-ray industrial tomographs. Thus, the contrast will be poor.
Another possible source of information is that part of the radiation that interacts with the sample. This radiation can be divided in scattered radiation and fluorescent radiation. The radiation can be scattered elastically (or Rayleigh) and inelastically (or Compton). Both these kinds of radiation can be used. The fluorescent part can also be used but it needs materials able to emit fluorescence lines and this is not the case of the cork, where the energy of the fluorescence lines is too low.
In order to maximize the probability to have coherent scattering the detector must be placed at low angles with respect to the X-ray beam, avoiding to detect the direct (not interacting) radiation. The radiation recorded on the detector depends on the molecular structure of the material. The incoherent or Compton tomography is performed by placing the detector at 90° with respect to the X-ray source. The signal is proportional to the electronic density of the sample. Both techniques will provide a good contrast on the reconstructed images, but the Compton tomography gives a higher count rate than coherent tomography.
The experimental setup used in this paper is shown in Fig. 1. The sample is rotated from 0° to 180° at uniform angular steps. At each angle the sample is also translated, the spectrum emitted at each position is acquired and stored. The reconstruction problem is more easily described in terms of a stationary sample and a moving source/detector system. In Fig. 1, the two (x,y) axes are fixed in the reference frame of the sample. The s and u axes are parallel to the translation direction and to the beam respectively, and they are related to (x,y) by a rotation of an angle θ. The input data-set is represented by the count rates measured for a complete set of translational and rotational steps (si,θj). The recorded data are used as input to a reconstruction algorithm. Several reconstruction algorithms have been reported in the literature [7], [8], [9], [10], [11], [12]. The present work uses the Hogan et al. algorithm [7], [13], which is appropriate for samples with low absorption coefficient. This algorithm is described in some details in the next paragraph.
Section snippets
Methods
The aim of our technique is to obtain a finer quality control of visually selected cork slices. Some samples of different quality cork bark are reported in Fig. 2.
The natural cork bark passes through several selections and manipulations. After the slice collection, the first step is to put the slices in a steam flux. This allow one to relax the cell of cork and so to straighten up the originally bent barks. Then the slices are visually inspected and selected by skilled workers. The stoppers are
Results
Our system is composed by an X-ray tube produced by EIS s.r.l. (a maximum of 40 kV with 1mA of current) and a Oxford X-ray tube (30 kV, 0.2 mA), PCA-P MCA card, a Physik Instrumente rotation–translation system and two NaI(Tl) detectors. The sample position and data acquisition are controlled by a custom software. A set of 60 projection each formed by 40 up to 60 translations was used. The Compton measurements were performed by using the EIS tube at 38 kV and 1 mA, while the transmission
Conclusions
In this paper a new application of Compton tomography is reported. This technique has been used for density analysis of cork stopper sample. It demonstrated to be particularly useful for cork stoppers used in high quality wine. These stoppers are considerably more expensive than the normal high quality stoppers. Therefore X-ray tomographic equipment can be conveniently used. By means of the Compton tomography reconstruction the visual classification can be considerably refined. Moreover, our
References (13)
- et al.
Forest Ecol. Manage.
(2001) - et al.
Forest Ecol. Manage.
(1997) - et al.
Nucl. Instr. and Meth. A
(1989) - et al.
Nucl. Instr. and Meth. B
(1993) - et al.
Comput. Phys. Commun.
(2001) Industrie alimentari
(1988)