MEASUREMENT OF DENSITY DISTRIBUTION IN ALUMINA GREEN BODIES BY X-RAY TOMOGRAPHY Meulenyzer S.2, Gonon M.1, Descamps J-C.1, Cambier F.2 1 Faculté Polytechnique de Mons, Service de Science des Materiaux, 56 rue de l'Epargne, B-7000 Mons, Belgique. 2 Belgium Ceramic Research Centre, 5 avenue du gouverneur Cornez, B-7000 Mons (Belgique) ABSTRACT The objective of this study is to investigate the feasibility and the accuracy of density measurements on ceramics green bodies by using high-resolution X-ray CT (Computed Tomography). At a first stage, experimental parameters were investigated in order to define the best conditions to observe density gradients in single axe pressed alumina green samples. Then measurements were performed from 2D projection X-ray radiography. Axial and radial gradients in alumina cylinders shaped under various pressures an of various aspect ratios were characterised. Finally 3D X-ray CT were performed and compared to the results obtained in two dimensions. KEY WORDS X-ray Computed Tomography, Density gradients, Ceramics green bodies, Alumina. INTRODUCTION High-resolution X-ray CT is a non-destructive technique for visualising volume defects and heterogeneities in opaque solids. Information on the 3D geometries is translated into twodimensional slices of elemental thickness [1]. On that point high-resolution X-ray CT is equivalent to medical CAT-scanning but it differs in the resolution that reaches a few tens of microns, even when investigated parts are made of high density materials. For many advanced ceramic applications, the knowledge of the density and density gradients is of major importance. Indeed, density variations in green bodies may influence the final properties and also yields to uncontrolled shrinkage and deformation during sintering [2] [3]. Density gradients are linked to the shape and the green forming technique, they can be modelised but rare are the techniques that allow measurements as X-ray CT does [4]. EXPERIMENTAL PROCEDURES 1. X-ray CT System The X-Tech X-ray CT system used is composed of (figure 1): - a microfocus Tungsten X-ray source (5 µm diameter) 40–225 KeV, - a CsI scintillation X-ray detection screen coupled with a photomultiplier and a CCD camera (1024x1024 pixels), - a five axis manipulator (x, y, z, translations; y tilt; z rotation). Samples can take any position between X-ray source and detector (890 mm). However geometric shade due to source dimension must be considered for high enlargements (>1 pixel when enlargement is > x40). Geometric deformations due to polar projection are corrected with regard to a reference grid picture. Shading, variation of intensity between pixels detector due to different distances source-pixel, is corrected with regard to a white reference picture taken without sample. 3D CT reconstruction is operated by filtered backprojection [1] from series of 360 images taken at 1° steps. z Iy,z I0 x y x-ray source detector Figure 1: X-ray CT system 2. Density measurement 2.1 Measurement by conventional 2D projection radiography During a sample examination, the intensity I y , z of the X-ray beam hitting the screen detector at pixel y,z is attenuated with regard to the intensity I y0, z at the same pixel when no sample is introduced along the beam path (figure 1). This attenuation follows the Beer-Lambert’s law : I y,z I y0, z exp . y , z .t y , z y , z {1} ty,z : thickness of crossed sample, ()y,z: mean masse absorption coefficient of sample along ty,z. In the CT System, the intensity I y , z is converted into a grey levels N y , z ranging between 0 (black, no radiation) and 4096 (white, saturation of the detector). If there is a linear dependence between the intensity and the grey level and if the mass absorption coefficient is known, the mean density along the tx,y direction can be calculated. 2.2 Measurement by 3D CT In 3D CT the information in grey levels obtained is proportional to the linear absorption coefficients µ = (µ/). for all elemental volume associated to each voxel. Assuming an homogeneous chemical composition of the sample (µ/ = cste), the grey level is proportional to the local density. RESULTS AND DISCUSSION In a first stage, the system X-ray detection linearity was checked. As previously said, the grey levels given by the image must be proportional to the intensities striking the detector. Then, to carry out precise measurement of the density, the accurate value of the mass absorption coefficient () must be known. Therefore, attenuation measurements of the radiation according to the mass thickness were carried out on densified alumina samples. Thereafter, a first evaluation of density gradients present in green bodies was carried out by direct measurements in 2D radiography. Finally, 3D representation of density gradients of density were realised. 1. Linearity of the X-ray detector In this part of the study, no sample was introduced between the source and the detector. The grey level N introduced in the calculations corresponds to an average values measured around the centre of the detection screen on a surface y,z of ~ 10 x 10 mm2 (50 x 50 pixels). It corresponds to an average intensity I. For a given X-ray energy the X-ray tube current was set to obtain saturation of the detector (X-ray intensity I0, grey level N0 = 4096). The X-ray tube current was then decreased to reduce intensity I at the screen detector and the corresponding grey level N was measured. The experimental curves confirm the linear dependence between N/N0 and I/I0 (Figure 2), however the grey level Nb corresponding to no X-ray intensity is not 0 so that the following correction appears as necessary: I N Nb I0 N0 Nb {2} 2 Mass absorption coefficient This part of the study aims to determine accurate values of () for the alumina as function of the tube tension but also the effects of beam hardening. Dense alumina pellets of various mass thickness .t obtained by isostatic pressing and sintering were introduced in the CT system between source and detector. The grey level N.t introduced in the calculations corresponds to an average values measured around the centre of the detection screen on a surface y.z of ~ 10 x 10 mm2 (50 x 50 pixels). It corresponds to an average intensity I.t. Considering the distance source / detector (890 mm), the variations in t with the pixel position y,z are negligible and ty,z ~ t, thickness of the alumina pellet over all the y,z range of measurement. In that conditions, mass absorption coefficients are calculated according to equation: .t I0 N0 Nb ln .t ln .t b I N N .t .t {3} According to the experimental conditions, high variation of the mass absorption coefficient are observed due to beam hardening (figure 3). The use of Cu filters reduces this phenomenon and will avoid CT reconstruction artefacts [1] [9]. 3500 0.50 3000 0.45 2500 E= 60 kV E= 70 kV E= 80 kV E= 80 kV + Cu filter 0.40 2 (cm /g) Grey level N 4000 2000 1500 1000 NG NG correction 500 0.35 0.30 0.25 0 0 20 40 60 80 100 I/Imax (%) Figure 2: Linearity of X-ray detector 0.20 0 5 2 x (g/cm ) 10 Figure 3: Variation in µ/ 15 3 Measurement of density gradients by 2D radiography Alumina cylinders of two diameters (30 and 40 mm) uniaxially pressed under various pressures (10, 32 and 56 MPa) were introduced in the CT system. The X-ray tube setting was 80 kV and 290µA for all tests. Cu filter (0.4 mm) was used to reduce beam hardening (figure 3). Grey levels N were measured in the same conditions than in §2. Average densities along diameter at different heights of the cylinder and along the height at different points of the diameter were calculated (figure 4). Sample motion Sample motion I0 I0 I I X-rays X-rays Figure 4: Principle of density measurement by 2D radiography Densities were calculated with reference to the average grey level N to which is associated the average density of the sample : I I exp . .t ; 0 exp ..t 0 I I ; I N Nb exp .t. b I N N N Nb ln b N N .t {4} Axial gradients show that with single axe single effect die pressing the density is higher under punch that at the sample base. An axial density gradient of 0.20 is measured with a shaping pressure of 10 MPa and an aspect ratio H/D (height / diameter) of 1.16 (figure 5). The radial gradient show a minimum in density along the cylinder axis and a maximum near the surface (figure 5) The curve show a more or less high symmetry according to the measurement direction what can be linked to a non homogeneous distribution of the powder in the die before shaping. 1.92 2.00 1.90 1.95 1.90 1.86 Density Density 1.88 1.84 1.82 Average value geometricaly measured 1.85 1.80 1.80 1.75 Average value geometricaly measured 1.78 1.76 1.70 0 5 10 15 Position (mm) 20 25 30 0 5 10 15 20 25 30 35 Position (mm) Figure 5: Density gradients in alumina cylinders pressed at 10 MPa (a: axial; b: radial). Additional measurements show the decrease in axial and radial gradients with the increase in pressure and also with a decrease in aspect ratio H/D (figure 6). 2.20 density 2.10 2.00 1.90 10 MPa 32 MPa 32 MPa 56 MPa 1.80 1.70 0 5 H/D=1,16 H/D=1,15 H/D=0,67 H/D=1,1 10 15 20 25 30 35 Distance from top ( mm) Figure 6: Density gradients according to shaping pressure and aspect ratio (H/D) Density resolution depends on the experimental conditions throughout accuracy of the grey scale measurement (radiography quality) and the mass absorption coefficient (equation 4). In the present work the accuracy is about ± 0.01. 4. Measurement of density gradient by 3D X-ray CT At that point of the investigations, 3D characterisation of the density gradients have been done in a qualitative way. X-ray radiographies were collected using the same experimental parameters than previously for 2D characterisation. A 3D representation of the sample pressed at 10 MPa is given in figure 7. The grey levels distribution is decomposed into segments: from white to dark red according to increasing densities. These results confirm the density distribution characterised from 2D measurements but a low density zone is observed juste under the punch around the cylinder axis. Top (punch) Bottom Figure 7: Density gradients in an alumina cylinder pressed under 10 MPa (aspect ratio 1.16) (white = low density, black = high density) 5. Conclusions A methodology for density gradients characterisation in ceramic green bodies by X-ray 2D radiography has been investigated. The protocol proposed uses an autocalibration of the measurement with regard to the average density of the samples. Measurement realised on alumina green bodies uniaxilay pressed agree with theory. Qualitative 3D density distributions were characterised for the same samples. Results agree with the 2D measurements. Actual work focuses on the tridimentional quantitative measurement by computed tomography Acknowledgements The authors would like to express their sincere gratitude to Hakim Djerba 3rd year student in material science at UVMH – Maubeuge (France) for his help in the collection of the data used in this project. References [1] Baruchel J., Buffiére J., Maire E., Merle P., Peix G., X-rays tomography in material science, Hermes Science Publication, Paris, 2000 [2] Peizhen K. Lu and John J. Lannutti ,Density Gradients and Sintered Dimensional Tolerance in Compacts Formed from Spray-Dried Alumina, Journal of the American Ceramic Society Vol. 83, No. 6, June 2000 [3] Peizhen Lu and John J. Lannutti, Effect of Density Gradients on Dimensional Tolerance During Binder Removal Peizhen K. Journal of the American Ceramic Society Vol. 83, No. 10, October 2000 [4] Peizhen Lu, John J. Lannutti, Peter Klobes, and Klaus Meyer, X-ray Computed Tomography and Mercury Porosimetry for Evaluation of Density Evolution and Porosity Distribution. Journal of the American Ceramic Society Vol. 83, No. 3, march 2000, 518-522