Particle Formation
The aim of this project is to enhance our understanding of how solvents and supersaturation affect the nucleation of polymorphic forms of compounds. Polymorphism is very important to the pharmaceutical industry as a compound’s polymorph can exhibit different physical properties, some more desirable than others. Being able to control which polymorphic form is nucleated is thus very important. In the first year of this project, the dimorphic enantiotropic compound p-aminobenzoic acid was selected as the model compound due to the availability of its solubility data and the relatively low transition temperature between its and polymorphs. The second year of the project has been focused on the precise location of the transition temperature and measuring the relative nucleation rates of and at different temperatures.
Background
Research background
There have been few studies of the relative nucleation rates of two polymorphic forms. Figure 1 shows a schematic phase diagram for an enantiomorphic system of Polymorphs I and II, it is characterised by a transition temperature at which the relative solubility and hence stability of forms switches. At the transition temperature
The to-date literature search on encapsulation and micro-encapsulation produces over 37000 references, with the earliest results originating back in the early 1900-th and as early as XIX century. The interest in encapsulation, seemingly non-fading since its inception, has been recently prompted by new requirements from myriad of industrial applications. In the last decade, the interest in encapsulation and micro-encapsulation has sprung anew with novel materials and technologies available for it: for example, developments in the area of polymeric microcapsules. That, in turn, ignited research in the area of release of encapsulated materials. Encapsulation evokes a great variety of not only methods, materials, techniques but also a large number of various industries and research institutions became interested in using it. Perhaps most notably, the scope of industries ranges from pharmaceutical, food, perfume to agriculture and materials. Besides drug delivery, the emphasis in this milieu is focused on more effective storage and delivery of materials. Notwithstanding, the encapsulation methods are equally important in research centers where it can be used for studying intracellular processes, exploring bio-chemical reactions in confined volumes, etc.
The objective of this third year was to collect data from a pilot plant for control testing. A team of 4 visiting researchers joined the existing particle research group in Queensland (Jim Litster, Ian Cameron, Fu Yang Wang, David Page, Rachel Smith) for an experimental campaign: Jonathan Poon and Rohit Ramachandran, both PhD students at Imperial College London, supervised by Charles Immanuel; Thomas Glaser a diplomarbeit student from Stuttgart who stayed in Frank Doyle's group in Santa Barbara for a year; and Constantijn Sanders a post doc in the Doyle group. Thomas recently graduated and Jonathan is writing up his PhD thesis.
The team had 2 objectives while in Queensland:
- Operate the continuous drum granulator to improve the continuous plant model
- Study the chosen formulation in a batch granulator to extract rate constants and improve process understanding for the purpose of improved modeling.
After the experimental phase, the Imperial team returned to London to further analyze the experimental batch results and to improve the 3 dimensional process model (papers 2 and 4 in section 8: Publications Resulting from Support) in collaboration with Frantisek Stepanek. The UCSB team returned to Santa Barbara to analyze the continuous data and experiment with several model predictive control algorithms. The team reunited in November in Salt Lake City at the Annual AIChE meeting to present 2 papers and discuss overall progress.
The pilot plant is now ready to be used and fully controlled by MATLAB: it is a good test bed for controllers. However, Jim Litster moved to Purdue this year, and the future of the pilot plant is unclear. Frank Doyle and Constantijn Sanders started experimental work in collaboration with Hong Sing Tan (Newcastle) and Paul Mort (Cincinnati) both of Procter & Gamble. The aim for this collaboration is to demonstrate an MPC setup on an industrial process.
This report is divided in three sections:
- Batch granulation experiments and modeling
- Continuous granulation experiments and modeling
- Continuous granulation control
1. Introduction
A proposal of the Swiss Federal Institute of Technology Zurich (ETH), Laboratory of Food Process Engineering (years 4-6) on the topic "Quantitative Analysis of Structural Transformation in Extrusion Processing" has been accepted for support by IFPRI in June 2005 and started in April 2006. The present report covers the time frame from December 06 to May 2008.
The shape of a crystalline solid has a major impact on its downstream processing and on its end-use properties and product functionality, issues that are becoming increasingly important in the pharmaceutical and life science, as well as the specialty and fine chemical industries. Though it is widely known that improved crystal shapes can be achieved by varying the conditions of crystallization (e.g., solvent type, additive and impurity levels, etc), there is far less understanding of how to effect such a change. Until recently, most methods for predicting crystal shapes were based exclusively on the internal crystal structure, and hence could not account for solvent or impurity effects. New approaches, however, over the possibility of accurately predicting the effects of solvents. We review models for predicting crystal shape, and evaluate their utility for process and product design.
Crystalline solids are an essential part of our modern technological environment, since they are important components for many materials such as pharmaceuticals, foods, cosmetics, metals, ceramics and plastics. The customary way of forming crystals in the chemical industry is through suspension processes which rely on the use of solvents as media for homogenisation of the starting composition, and as an enabling environment for molecular assembly processes. Solvents have been found to influence crystallisation to the point of altering nucleation rates, crystal morphologies, crystal aggregation and the crystal structure of the end product. While significant progress has been made in understanding the origin of crystal morphology, our current knowledge of the molecular assembly processes leading to the nucleation of a particular crystal form in supersaturated solutions is poor. The work detailed in this final report deals with the role of solvent in nucleation processes and the mechanisms by which nucleation may be influenced by solvent choice.
The first part of the report deals with solution speciation in concentrated and supersaturated solutions and explores the link between the species present in solution and the crystal structure of the polymorph which nucleates. For this part of the study, carboxylic acids were selected as model materials and a combination of IR and Raman spectroscopy used identify the H-bonding motifs existing in their crystalline and solution phases. A significant portion of this work was devoted to the two monocarboxylic, tetrolic and benzoic acids and it was discovered that in some systems, for example tetrolic acid, there is a very clear link between the solution state of the molecule and its resulting crystal structures. In others, benzoic and mandelic acid it seem that irrespective of the state of the solute in solution it always nucleates the same structure.
The second portion of the work moves on from the nature of the solution to explore the impact of solvent on the nucleation process in the polymorphic system p-aminobenzoic acid (PABA). PABA has two crystalline polymorphic forms, and which are related enantiotropically having a transition temperature of 240C. It was thus envisaged that in this system the nucleation of the two forms could be studied at the transition temperature where thermodynamic effects would be identical for each and solvent influences could be systematically explored. Solvent selection was found to have a significant impact on the ability to nucleate the polymorph. Only in water was it possible to nucleate both forms; all other solvents favoured the form irrespective of the temperature. This outcome was rationalised in terms of the solution species and the impact of solvent on the growth of the phase.
This report is an integral part of an effort to develop a computational platform to virtually synthesize and test particle compacts based only on the bulk and surface properties of the particles prior to the consolidation process. This virtual manufacturing and testing facility (VMTF) includes die filling, compaction –particle rearrangement and particle deformation (elastic and inelastic)–, compact ejection and subsequent mechanical testing. The current simulation platform is based on a multiscale approach, which bridges systematically the micro and meso-scale. The VMTF will provide the ability to reproduce the behavior of current products but more importantly, it will enable the simulation of systems never yet manufactured, virtually screening the best manufacturing conditions and article/granule properties for a desired compact behavior or application. During this year we will continue the development of the subsequent modules of die-ejection and mechanical testing.
The specific content of this report includes a numerical study of the mechanical behavior of systems composed by particles with different sizes and materials subjected to consolidation. The simulation methodology is based on a mixed discrete/ continuum approach which allows to systematically bridge the microscale response (particle and inter-particle scale) with the mesoscale and macroscopic behavior (container/sample scale). The methodology is particularly suitable for describing the post-rearrangement regime where consolidation proceeds mostly by elastic and inelastic deformation. This formulation is able to provide quantitative estimates of the evolution of macroscopic variables, such as pressure and density, while following microlevel processes, such as local coordination number and loading paths. This methodology is applied to polydispersed systems composed by particles with different nonlinear properties. The predictions are in general agreement with the experimental data during both loading and unloading cycle.
This report addresses the properties of flow and jamming in a funnel or hopper. The first part of this project, to date, involves understanding flow properties using a quasi-2D hopper geometry, photoelastic techniques, and high speed video.
The second part of the project involves using theoretical approaches that, to some extent, borrow from recent work in the physics community on the so-called jamming transition.
The third aspect of this work, a direct application to an IFPRI industrial partner, is to better understand and interpret results using the Flowdex tester. This device is used to characterize flowability, and in particular to characterize that property for various powders of interest to P&G.
This work has benefited considerably from collaborative interactions with Dr. Paul Mort of P&G. IFPRI funds were used primarily for the support of Ph.D. student Junyao Tang, who is working fulltime on this project. In addition, an undergraduate Sepher Sagdiphour has carried out a significant number of measurements.
Over the period of the last year, we have investigated aspects of process model identification and model based control for granulation processes. Significant accomplishments from that period are reported in two journal publications which are highlighted here:
[Sanders, Hounslow, Doyle III, Powder Technology, in press, 2008]
The modeling work in this paper provides insight on improved control and design (including measurement selection) of a granulation processes. Two different control strategies (MPC and PID) are evaluated on an experimentally validated granulation model. This model is based on earlier work done at The University of Sheffield, UK and Organon, The Netherlands. The granulation kinetics were measured in a 10 liter batch granulator with an experimental design that included four process variables. The aggregation rates were extracted with a Discretized Population Balance (DPB) model. Knowledge of the kinetics was used to model a continuous (well mixed) granulator. The controller model for the Model Predictive Controller is a linearized state space model, derived from the nonlinear DPB model. It has the four process variables from the experimental design and a feed ratio as input variables. Since the DPB model describes the whole size distribution (GSD), different sets of output variables were chosen and compared. When measuring controller performance based on the full granule size distribution, a PID setup can actually produce results that fluctuate more than the open-loop response. An MPC controller improves stability on both process outputs and the full granule size distribution.
[Glaser, Sanders, Wang, Cameron, Litster, Poon, Ramachandran, Immanuel, Doyle III, Journal of Process Control, in press, 2008]
This paper details a methodology for the design of a Model Predictive Controller for a continuous granulation plant. The work is based on a nonlinear one-dimensional Population Balance Model (1D-PBM), which was parameterized using experimental step test data generated at a continuous granulation pilot plant installed at the University of Queensland, Australia. The main objective was to operate the granulator under optimal conditions while off-specification material was fed back into the granulator to increase the economy of the process. The final algorithm design combines elements of Model Predictive Control (MPC) with gain scheduling to cancel nonlinearities in the recycle flow. A model directly identified from the step test data was the basis for testing a model predictive controller. Simulations show that the efficiency and robustness of this granulation process can be improved by applying the proposed control strategy. Ongoing work focuses on the implementation of the proposed control strategy on a full scale industrial plant.
Our aims for the renewal period of this IFPRI project are reiterated here, and the main body of this document reports our progress against these aims.
We have employed Atomic Layer Deposition (ALD) to produce mineral like surfaces that are extremely smooth for use in surface force studies. Our investigations to date have focused on alumina and titania surfaces.
Alumina Surfaces
Alumina surfaces were found to be unstable in aqueous solution – they slowly dissolved. This prevented surface force investigations in simple aqueous solutions. However the surface could be passivated against dissolution through the adsorption of short chain carboxylic acids. These acids are of industrial interest as they have significant effects on the rheology of alumina dispersions. Examination of the surface forces between alumina surfaces in solutions of muconic acids revealed DLVO type forces under some conditions Non‐DLVO forces where a strong attraction was evident between the surfaces and is significantly stronger than van der Waals attraction. This attraction was attributed to the formation of a capillary consisting of an oil‐like muconic acid phase forming between the surfaces. This phase change is induced by the close proximity of the surfaces and is possible because the muconic acid is present at concentrations that approach the solubility limit in these solutions. The presence of a capillary between the surfaces results in a strong attraction. Attempts were made to form stable alumina surfaces that would enable surface force measurements to be conducted in water and electrolyte solutions. This included looking at much thicker layers and using different binding layers (such as titania). To date none have been successful. We are still pursuing this though it may be possible that all alumina surfaces – not just ALD surfaces – have this property. A slow rate of dissolution would not be revealed in many studies and therefore may have previously gone unnoticed. Evidence from Optical Reflectometer (OR) shows that the surface dissolves at a rate of ~8 nm per hour. So indeed the rate of dissolution is slow, but sufficient to prevent surface force or optical reflectometry measurements.
Titania Surfaces
In contrast, titania surfaces are stable and this has allowed us to perform a range of surface force studies at both low and high salt concentrations. At low salt concentrations a long range, pH dependent electrostatic force was observed. This data could be fit using the DLVO theory, which enabled the surface potential to be determined. This showed that the isoelectric point was between pH 5 and pH 6. At short range a repulsive interaction dominated the attractive van der Waals force. This is attributed to hydration forces. At high salt concentrations adhesion was seen that was dependent on both the pH and specific salt present. We find that this trend does not follow the Hofmeister series.