(421d) Understanding the Role of Cake Structure in the Filtration of Needle-like Crystals in the Pharmaceutical Industry
AIChE Annual Meeting
2019
2019 AIChE Annual Meeting
Pharmaceutical Discovery, Development and Manufacturing Forum
Predictive Scale-up/Scale-down for Production of Pharmaceuticals and Biopharmaceuticals I
Tuesday, November 12, 2019 - 4:30pm to 4:50pm
1. Motivation
The
filtration time in the pharmaceutical industry can be critical for the
feasibility of a process [1]. In this field, filtration is used to recover
crystals of active pharmaceutical ingredients or intermediates resulting from
purification processes [2]. The time required by this step is heavily
influenced by the size and the shape of the crystals being filtered, which is a
result of the nature of the compound and the way the particles are generated
[3,4]. Crystals often have non-equant morphologies and, even though there is considerable
knowledge on how to predict and control the final shape and size they will have,
there is no direct way to forecast how they will filter. The size and shape of
the particles are important for filterability as they influence the way particles
pack, therefore the internal structure of the filter cake. This structure, in
the end, determines the cake resistance, and hence the filtration time. Being able
to predict how populations of particles with different sizes and shapes will
behave during filtration would enable us to find the optimal crystallization
conditions that require the minimum combined processing time. It would improve
and speed up process development allowing new drugs to get to the market sooner. 2. Methodology Determining
the way that particle size and shape affect the filter cake structure is
fundamental in order to predict filterability. This prediction becomes
increasingly difficult to be carried out for non-equant particles: here we
present a work on needle-like crystals, which is a type of morphology that is
often encountered in the pharmaceutical industry and strongly differs from
ideal/model shapes such as spheres or cubes. We here propose two methods to
tackle this understanding: the first is an experimental approach that takes advantage
of X-ray tomography to investigate the internal structure of real filter cakes and
the second one is a computational approach that simulates the packing of
different size and shape distributions of particles. The two approaches
together can be used to validate one another and to explore many scenarios
without the need of further experiments.![](https://www.aiche.org/sites/default/files/aiche-proceedings/conferences/269951/papers/572541/Paper_572541_abstract_153678_0.png)
2.1. X-ray tomography
Micro X-ray
tomography is a technology that enables us to recognize the composition of an
object based on the density of its different components: elements with
different densities will indeed cause different attenuation of the X-rays. In
our work, this enables us to determine the position of particles and pores
inside filter cakes, as we can detect objects down to the size of 1 micron. The
output of these measurements is a series of slices that represent subsequent
cross-sections of the cake. These slices together with a segmentation algorithm
let us gather information on the single particles of the cake, as well as on
the shape and connectivity of pores. 2.2. Random
packing simulations We aim to
study systematically the effect of particle morphology (size, shape and
polydispersity) on the cake structure. For this purpose, we use Monte Carlo
simulations to model the final packing of particles in filter cakes. The
resulting structures can be generated starting from a random configuration of
dispersed particles to which a gravitational field (pressure gradient) is
applied. For the simulation of needle-like systems, the spherocylinder (a
cylinder topped by two spherical caps) was selected as model shape, as it is known
to be an appropriate representation of elongated particles and to be computationally
efficient [5]. 3. Results X-ray tomography
has enabled us to notice the preferential orientation of needle-like particles
inside filter cakes and how this impacts on the porosity of the system. Figure
1 shows a typical example of what can be extracted from such measurements. On
the left two orthogonal slices of the cakes are shown, highlighting the
contrast between the particles (the lighter elements) and the pores (the darker
parts). The image on the right shows the reconstructed 3-D structure of a small
section of a cake, where different colors are used to distinguish the single
segmented particles. The results
of Monte Carlo simulations also enable us to understand the way that particles
tend to orient in the cake and how this changes by modifying the distribution of
sizes and shapes, in particular its polydispersity. Figure 2 shows an example
of how a system of particles resembling the experimental ones can be packed.
After having checked the agreement between the experimental results and the
computational ones we can use the knowledge gathered from the simulations to
predict the cake structure of different populations of needle-like crystals. References:
[1] MacLeod C.
S. and Muller F. L. (2012) Org. Process Res. Dev., 16, 3, 425434. [2]
Murugesan S. (2010) Chemical Engineering in the Pharmaceutical Industry,
315-345. [3] Perini G.
et al. (2019) Sep. Purif.
Technol.,
211, 768781. [4] Wakeman
R. (2007) Sep. Purif. Technol.,58, 2, 234241. [5]
Ferreiro-Córdova C. et al. (2014) Chem. Eng. Data, 59, 10,
30553060. Acknowledgments: The authors
would like to thank EPSRC and AstraZeneca for funding the PhD project and the
Henry Moseley X-ray Imaging Facility.
filtration time in the pharmaceutical industry can be critical for the
feasibility of a process [1]. In this field, filtration is used to recover
crystals of active pharmaceutical ingredients or intermediates resulting from
purification processes [2]. The time required by this step is heavily
influenced by the size and the shape of the crystals being filtered, which is a
result of the nature of the compound and the way the particles are generated
[3,4]. Crystals often have non-equant morphologies and, even though there is considerable
knowledge on how to predict and control the final shape and size they will have,
there is no direct way to forecast how they will filter. The size and shape of
the particles are important for filterability as they influence the way particles
pack, therefore the internal structure of the filter cake. This structure, in
the end, determines the cake resistance, and hence the filtration time. Being able
to predict how populations of particles with different sizes and shapes will
behave during filtration would enable us to find the optimal crystallization
conditions that require the minimum combined processing time. It would improve
and speed up process development allowing new drugs to get to the market sooner. 2. Methodology Determining
the way that particle size and shape affect the filter cake structure is
fundamental in order to predict filterability. This prediction becomes
increasingly difficult to be carried out for non-equant particles: here we
present a work on needle-like crystals, which is a type of morphology that is
often encountered in the pharmaceutical industry and strongly differs from
ideal/model shapes such as spheres or cubes. We here propose two methods to
tackle this understanding: the first is an experimental approach that takes advantage
of X-ray tomography to investigate the internal structure of real filter cakes and
the second one is a computational approach that simulates the packing of
different size and shape distributions of particles. The two approaches
together can be used to validate one another and to explore many scenarios
without the need of further experiments.
![](https://www.aiche.org/sites/default/files/aiche-proceedings/conferences/269951/papers/572541/Paper_572541_abstract_153678_0.png)
![](https://www.aiche.org/sites/default/files/aiche-proceedings/conferences/269951/papers/572541/Paper_572541_abstract_153679_0.png)
tomography is a technology that enables us to recognize the composition of an
object based on the density of its different components: elements with
different densities will indeed cause different attenuation of the X-rays. In
our work, this enables us to determine the position of particles and pores
inside filter cakes, as we can detect objects down to the size of 1 micron. The
output of these measurements is a series of slices that represent subsequent
cross-sections of the cake. These slices together with a segmentation algorithm
let us gather information on the single particles of the cake, as well as on
the shape and connectivity of pores. 2.2. Random
packing simulations We aim to
study systematically the effect of particle morphology (size, shape and
polydispersity) on the cake structure. For this purpose, we use Monte Carlo
simulations to model the final packing of particles in filter cakes. The
resulting structures can be generated starting from a random configuration of
dispersed particles to which a gravitational field (pressure gradient) is
applied. For the simulation of needle-like systems, the spherocylinder (a
cylinder topped by two spherical caps) was selected as model shape, as it is known
to be an appropriate representation of elongated particles and to be computationally
efficient [5]. 3. Results X-ray tomography
has enabled us to notice the preferential orientation of needle-like particles
inside filter cakes and how this impacts on the porosity of the system. Figure
1 shows a typical example of what can be extracted from such measurements. On
the left two orthogonal slices of the cakes are shown, highlighting the
contrast between the particles (the lighter elements) and the pores (the darker
parts). The image on the right shows the reconstructed 3-D structure of a small
section of a cake, where different colors are used to distinguish the single
segmented particles. The results
of Monte Carlo simulations also enable us to understand the way that particles
tend to orient in the cake and how this changes by modifying the distribution of
sizes and shapes, in particular its polydispersity. Figure 2 shows an example
of how a system of particles resembling the experimental ones can be packed.
After having checked the agreement between the experimental results and the
computational ones we can use the knowledge gathered from the simulations to
predict the cake structure of different populations of needle-like crystals. References:
[1] MacLeod C.
S. and Muller F. L. (2012) Org. Process Res. Dev., 16, 3, 425434. [2]
Murugesan S. (2010) Chemical Engineering in the Pharmaceutical Industry,
315-345. [3] Perini G.
et al. (2019) Sep. Purif.
Technol.,
211, 768781. [4] Wakeman
R. (2007) Sep. Purif. Technol.,58, 2, 234241. [5]
Ferreiro-Córdova C. et al. (2014) Chem. Eng. Data, 59, 10,
30553060. Acknowledgments: The authors
would like to thank EPSRC and AstraZeneca for funding the PhD project and the
Henry Moseley X-ray Imaging Facility.