Artículos
10. ANTES DEL BIG BANG: ALGO O NADA (15.01.2013)
For X-ray CT scanning, intact soil cores collected in the field using polythene pipes were used. The X-ray CT scanner was Nanotom, Phoenix X-ray system made by GE Sensing and Inspection Technologies GmbH, Germany (Fig. 2.1). This Micro CT system is characterised by the presence of high resolution detectors with the detectability of 1 µm. The X-ray tube is characterised by nanofocus <800 nm spot size with maximum voltage of 180 kV and a maximum output of 15W.
Fig. 2.1. Phoenix nanotom X-ray Computed Tomography scanner (http://www.ge-mcs.com)
2.2.2.1 Sample preparation and acquisition of CT data
The samples were scanned over a range of angular orientations using the X-ray beam generated by passing high energy current (expressed in µA) over a tungsten target. The energy levels of X-ray beam generated is described as the peak X-ray energy in kV. The soil core was placed on the sample stage and the position was adjusted to ensure that the sample was within the field of view and was fitted firmly to avoid sample movement during the scan. After placing the soil core in the sample stage, the energy levels and current were adjusted to obtain good quality images in a reasonable time period (detector time). This was done by looking into the histogram to get a grey scale value of ≥20% of
the dynamic range of detector, in the densest part of the soil core (centre). Use of a copper filter to control the incident X-ray beam aided getting good quality images. The possibilities of changing detector position from the X-ray gun were also tested to get best possible resolution. Once the scanning parameters such as energy current, detector timings, binning, spin and resolutions were decided, the sample was removed from the sample stage to calibrate the X-ray signals. Two calibrations were undertaken namely offset and gain. During offset calibration the X-ray was switched off while X-ray was switched on during the scanning for “gain” calibration. These calibrations are done to standardise the detector with respect to X-ray signals being generated and it serves as the baseline from which all sample scan data are subtracted. Then the core sample is introduced back to the sample stage and CT scanning was performed. Scanning was done at energy levels of 130 kV and a current of 110 µA. The soil cores were scanned in a vertical upright position. A total of 2000 images of resolution 27.5 µm were recorded over 60 minutes for each core.
2.2.2.2 Image reconstruction
Image reconstruction is a mathematical process to generate images from projection data obtained by CT scanning. The reconstruction of images was performed by datos|x software (GE Sensing and Inspection Technologies GmbH, Germany) and then using VG Studiomax (volume graphics); an image processing software. In the datos|x software the raw projection intensity data are converted to CT numbers in a range of grey scales (12 bit) which in turn correspond to the X-ray attenuation coefficient which is a function of density, atomic number and X-ray energy (Ketcham, 2005). A total of 2000 images were acquired for each scan..
2.2.2.3 Artefacts
Using the scan optimiser option in datos|x the difference between first and last image was computed and the value was accepted. This step eliminates artefacts caused by movement, if any, during scanning.
2.2.2.4 Beam hardening
Since the size of sample used in all the experiments was large, hardening of X- ray beam was expected. Beam hardening makes the edges of sample brighter than the centre parts. It is caused by an increase in mean energy of X-ray as it passes through the sample to be scanned; since lower energy X-rays in a polychromatic beam get attenuated more readily. To reduce this artefact a copper filter of thickness 0.1 mm was used in front of X-ray tube to pre-harden the X-ray beam, before beginning of scanning.
2.2.2.5 Image resolution
By scanning the soil cores through a 360orotation, image data is recorded in the form of stack of slices. To account for the thickness element of each slice, which provides the three-dimensional capabilities for CT images, the pixels in CT images are referred to as voxels. The resolution of CT images is given as voxel size in µm which indicate the size of a 3D pixel that can be identified as an independent entity. The image resolution varied with each experiment and is given separately in each chapter.
2.2.2.6 Image visualisation and saving for analysis
The reconstructed data of each scan was opened in VG Studio Max software and saved as image stack for further analysis (Fig. 2.2)
Fig. 2.2. Selected original and gray scale images of (a) clay loam tilled (b) clay
loam grass strip (c) sandy loam tilled (d) sandy loam grass strip from Sutton Bonington and (e) Random traffic ( f) Intermediate traffic and (g) No-traffic soils of Bedford (Pore space is shown in black).
a) b)
c) d)
e) f)
g)
2.2.2.7 Image analysis
Images analysis was carried out using ImageJ software (Rasband, 2002) to study the soil pore characteristics. ImageJ is an open source software written in Java. A rectangular region of interest (33 x 33 mm2) was selected in the reconstructed CT images to exclude pores adjacent to the sample edges. A total of 1800 images were used in the analysis excluding 100 images from the start and the end which are more prone to cone beam artefacts. A suitable image routine was developed after trying different filters and image enhancement techniques. The contrast of all images was enhanced, normalised and equalised. The function ‘sharpen’ increases the contrast and accentuates details in the image. A median filter was used reduce noise. The differentiation of pores from solids was made by thresholding with a suitable automated algorithm and the image was converted to an 8-bit gray scale image. Thresholding is used to convert a gray scale image into binary by defining a segmentation point on a histogram. This step facilitated classifying the image into features of interest (pores) and background (solids). The thresholding algorithms used were ‘minimum’ ‘MinError’ and ‘MaxEntropy’. The noise in the subsequent binary image was then removed by the ‘remove outlier’ option which replaces a pixel by the median of the pixels in the surrounding if it deviates from the median by more than the value assigned for threshold (ImageJ, 2012).The statistics on pore characteristics of each individual pore were generated using the ‘analyse particles’ option in ImageJ. The information on number of pores, average pore size, porosity, pore size distribution, surface area and circularity of pores were obtained. A coefficient of uniformity was calculated to statistically compare the pore size distribution. This was
determined as the ratio distribution (Atkinson experiment varied slightly
chart showing the image analysis is depict
Fig. 2.3. Flow chart
analysis.
as the ratio of size of pores at 10% and 60% of total Atkinson et al., 2009). The image routine followed varied slightly and is given in respective chapters. A general showing the image analysis is depicted in Fig. 2.3.
Flow chart showing the procedures used in image acquisition of total macro pore
followed in each chapters. A general flow