088 by nuhman10



     Morphological characterisation of particles as a tool for process
                                                         F. Stepanek and M. A. Ansari

       Department of Chemical Engineering, Imperial College London, South Kensington campus
               London SW7 2AZ, United Kingdom; E-mail: f.stepanek@imperial.ac.uk


The external shape of a particle is a direct consequence of the processing history that has lead to the
formation of the particle. For example, granules produced by low-shear agglomeration (such as fluid-
bed granulation) tend to have a different sphericity than those formed in high-shear mixer granulation,
which in turn are morphologically distinct from spray-dried granules of the same composition. Particle
shape can also be a good indicator of attrition, breakage, or re-crystallisation that may have occurred
during a processing step such as pneumatic conveying or contact drying. Shape characterisation
therefore lends itself as a tool for process monitoring and identification of ‘hot-spots’ within a process,
i.e. unit operations where an undesirable transformation has occurred, or for reverse engineering, i.e.
identification of the sequence of processing steps or the manufacturing route used for the production
of the particles. This ‘particle fingerprinting’ is potentially useful in applications such as identification of
counterfeit drugs or branded consumer products.

In this work we will present a methodology for particle fingerprinting based image shape analysis
(Garboczi, 2002). A large number (from the point of view of statistical significance) of particle images
are first obtained using a digital camera attached to a microscope. The images (Figure 1a) are then
segmented and thresholded (i.e., converted from greyscale to binary from) in image analysis software
to extract individual particle contours, as shown in Figure 1b. The radius of gyration as function of the
rotation angle (Figure 1c) is then obtained from each individual particle by means of in-house software
and the following quantities are calculated: the particle aspect ratio, the equivalent area and equivalent
perimeter radii, the mean gyration radius, the standard deviation of the gyration radius, the surface
roughness amplitude, and the surface roughness correlation length. The correlation length is the zero-
order moment of the auto-correlation function (Figure 1d), defined as

                      (r(   )  r )  (r(   )  r )d
                                    g                       g
         R( )      0

                                (r( )  r )
       rg 
                      r( )d                                                                            (2)

is the mean gyration radius. The surface roughness correlation function has recently been used for the
computer reconstruction of particle populations for the purpose of virtual granulation (Stepanek, 2004)
and the formation of random particle assemblies (Kohout et al, 2004). A population average is then
calculated and a 7-component vector composed of the characteristics mentioned above is then used
as a ‘fingerprint’ of the particle population. Provided that the distance between the fingerprint vectors
of two particle populations in the Euclidean space is sufficiently large, these populations can be
positively identified as having a distinct origin, e.g. having been manufactured by a different sequence
of processing steps or having been subjected to a different shear history. As a demonstration of the
methodology we have analysed and calculated fingerprints of several lactose samples (lactose is
commonly used as a pharmaceutical excipient) from different suppliers and produced by different
routes (spray drying, crystallisation and milling). We show that the fingerprinting method is sensitive
enough to distinguish not only between different processing routes, but also between different
suppliers who use nominally the same processing route.


                                               QuickTime™ and a
                                      TIFF (Uncompressed) decompressor
                                         are need ed to see this picture.                         QuickTime™ and a
                                                                                        TIFF (Uncompressed) decompressor
                                                                                           are needed to see this picture.

                       (a)                                                        (b)

     (c)                                                                    (d)

Figure 1: (a) Original image of particles; (b) Single particle contour after image segmentation and thresholding; (c) Gyration
radius function of that particle; (d) Autocorrelation function calculated from the gyration radius function.

GARBOCZI, E.J., (2002), Three-dimensional mathematical analysis of particle shape using x-ray
tomography and spherical harmonics: application to aggregates used in concrete, Cement and
Concrete Research, 32, pp. 1621-1638.

KOHOUT, M., COLLIER, A.P., AND STEPANEK, F., (2004), Effective thermal conductivity of wet
particle assemblies, International Journal of Heat and Mass Transfer, 47, pp. 5565-5574.

STEPANEK, F., (2004), Computer-aided product design: granule dissolution, Chemical Engineering
Research and Design, 82, A11, pp. 1458-1466.


To top