Report Industrial Group Autumn Meetings 7-8 November 2007

Link straight to the Day Two report

Pharmaceutical Special Interest Group - 7^th November 2007

This meeting was held at AstraZeneca's excellent conference facilities at their Macclesfield site, and attracted seventy delegates. The meeting was chaired by Roy Copley and myself. We had nine speakers and the day was split into two themes: the morning consisted of talks of general pharmaceutical interest, with considerable time devoted to polymorphism, and the afternoon's talks concentrated on structure solution and refinement. I would like to extend my thanks to the speakers for making this such a fine meeting, with very high quality and informative talks. The slides for these talks , when available, are linked within the report below.

Roger Davey (Manchester University) opened the meeting with a personal perspective on 'Polymorphism - what have we learnt?'. Examples from over 100 years of the scientific literature were given, illustrating the different solubilities of polymorphs, the solution-mediated transformation of the metastable to the stable form, the methods of discovering new forms, the insights into molecular packing and conformation, and the progress made in the prediction of crystal structures. In some cases, recent work has confirmed, or re-discovered what was known several decades ago. Roger questioned the use of high throughput methods with thousands of experiments and huge quantities of data. A more thoughtful approach might pay dividends: for example, using the output from structure prediction to identify the likely hydrogen bond motifs, and then devising experiments to produce those motifs. Roger went on to describe his group's recent work on understanding the links between molecular clusters present in solution, or the melt, the structure of the solid on nucleation, and the final crystal structure.

Gordon Barr (Glasgow University) spoke on the use of automated cluster analysis of XRPD and Raman data for use in solid form screens in a talk with a thought-provoking title: 'Is PXRD the Gold Standard in High Throughput Experiments?' Gordon described the benefits and disadvantages of XRPD and Raman spectroscopy: preferred orientation and degree of crystallinity are two disadvantages which can affect the cluster analysis of XRPD data, while Raman data can suffer from relatively small differences between polymorphs, variability in background level, and cosmic ray spikes. Gordon gave examples of cases in which cluster analysis of a combination of Raman and XRPD data outperformed the use of either method in isolation, in particular with respect to mixtures of forms. Gordon also showed how cluster analysis could be used on DSC data, and gave examples of quantitative analysis using Raman. He concluded that using a combination of datasets is the best way of understanding the properties of a set of samples.

Talbir Austin (AstraZeneca) gave a talk titled, 'Understanding relative polymorph stability through structure and thermodynamics'. She described the process of polymorph selection of a pharmaceutical and explained how it is essential to understand which polymorphs can form since their different physical properties may affect bioavailability, and hence efficacy, of the drug. Having identified which polymorphs exist, it is then necessary to understand the stability relationships between them, to ensure that the form chosen for development is robust and stable. Tal demonstrated how a combination of methods are used to understand the relationship between two forms, including DSC, construction of the E/T diagram, competitive slurries and sublimation experiments. In the absence of a single crystal, powder XRD data were used to obtain a crystal structure of one form (with refinement of the structure of the other form still in progress). Spectroscopy was used to show that hydrogen-bonding in the two forms was quite different. Tal concluded that a combination of structural, thermodynamic and kinetic studies is required in order to understand polymorphism.

Bill Jones (Cambridge University) described the use of grinding to produce new forms in his talk 'Screening for New Crystal forms Based on Mechanical Activation of Mixtures'. Grinding together of components, particularly in the presence of a drop of solvent, can produce novel co-crystals, salts, polymorphs and solvates, which solution experiments may (at least initially) fail to produce. Many examples of new solid forms produced by grinding were given, including: co-crystals of caffeine/acetic acid (grinding gives 1:1 as well as 1:2 stoichiometry; solution only gives 1:1), caffeine/TFA (grinding gives two polymorphs, and the seeds were then used in solution experiments). The presence of a few drops of solvent was found to be particularly beneficial and in the case of a salt screen on pyrimethamine the presence of methanol allowed a reaction in cases in which dry-grinding did not. Bill recommends including grinding experiments in solid form screens, particularly as a quick test for the ability to hydrate, and also recommends grinding as a possible route to interesting solid-state organic chemistry.

Jerry Heng (Imperial College) considered the importance of surface functional groups to the behaviour of pharmaceuticals in his talk, 'Crystal Engineering: Importance of Surface Properties'. Jerry described the use of macroscopic crystals of paracetamol, aspirin and ibuprofen (these crystals were several centimetres across!) to enable physical and chemical measurements to be made on individual faces. Jerry used a combination of methods: contact angle measurement, XPS, and approximations from knowledge of the crystal structure to understand how wetting properties are affected by the chemistry of different crystal facets. He showed that the most hydrophobic faces are those with weakest attachment energies (calculated from the crystal structure). Jerry went on to discuss the consequences of this finding for pharmaceutical processing, in which the bulk drug is milled - a process which probably reveals these hydrophobic surfaces.

After (a very tasty) lunch, the afternoon session concentrate on structure solution and refinement. Two themes ran through all the talks: a sense of confidence that powder data can be used for structure solution from powders, as long as the data are good quality, and a requirement that complementary techniques are used to confirm structures and avoid pitfalls.

Fred Vogt (GSK) spoke on 'X-ray diffraction, Computational Chemistry and NMR: A Multi-disciplinary Approach to Understanding the Pharmaceutical Solid State'. Fred demonstrated the use of the combined approach to maximise the understanding of the structure, dynamics and properties of pharmaceutical compounds, in order to enable improved control over the compound, and enable informed decision-making. Fred described the use of ssNMR in the determination of the crystal structure of a pharmaceutical compound for which single crystals were not available; ssNMR showed that an initial structure had a carboxylic acid group in the wrong orientation, and once this was corrected it was possible to solve the structure. In another case, ssNMR was used alongside XRPD, FTIR and GVS (gravimetric vapour sorption) to understand the mechanism of dehydration of a pharmaceutical, and the location of the water molecules in the crystal structure.

Jonathan Burley (Nottingham University) gave an introductory lecture - 'Cocrystals and Other Complex Pharmaceutical Materials: Structure Solution from Powder Diffraction'. He described the process of data collection and strongly recommended the use of the Debye-Scherrer (i.e. capillary) over Bragg-Brentano (flat plate) geometry. This was to avoid or reduce the errors associated with the latter: preferred orientation, X-ray absorption by the sample and sample height effects. Jonathan then described how the structure could be obtained in three stages: indexing, obtaining an approximate structure solution, and refinement to obtain an accurate structure; he identified suitable computer programs for each of these stages. The talk was illustrated by examples of structure solution including co-crystals, a hydrate and an organo-metallic. Jonathan concluded by assuring us that it is possible to solve a range of structures from laboratory data, as long as the data are high quality.
View Presentation... a 5.2MB PDF file.

Jacco van de Streek (Frankfurt University) spoke on 'Semi-automated Rietveld Refinement of Molecular Crystal Structures with DASH and TOPAS'. Jacco described the use of DASH to direct a Rietveld refinement in TOPAS, by generating input files, then taking the output files from TOPAS and adapting them to be used as new input files. This removes the need for tedious manual intervention, and going back and forth between DASH and TOPAS, and speeds up the process so that high-quality Rietveld refinements can be carried out routinely in a matter of minutes. Jacco confirmed Jonathan's confidence in the usefulness of powder data for structure solution......as long as the data are very high quality, and recommended keeping an instrument dedicated to structure solution. He cautioned that even a semi-automated structure solution process needs experienced, knowledgeable scientists to obtain accurate structures.

The first day ended with Maryjane Tremayne (Birmingham University) speaking on 'Powders and 'Peer-Pressure': Pitfalls and Progress'. Maryjane described her work on the development of the Cultural Differential Evolution (CDE) technique for structure solution from powder data. Space will not permit me to try to outline CDE and it's predecessor, Differential Evolution (DE) in this paragraph, but suffice to say that the 'biological' evolutionary approach to the generation of trial structures is enhanced with the addition of a 'cultural' driving force, or 'peer pressure' in which the distribution in values of structural parameters in each generation is used to guide and enhance the optimisation process. This enables a more rapid convergence to the global minimum. Maryjane illustrated the benefits of CDE over DE with several comparisons in which structure solution was achieved by CDE in far fewer generations than DE, and one example in which DE failed to obtain a structure (that of -methyl- -propyl succinimide). The pitfalls identified by Maryjane included the effects of preferred orientation and the discovery of a new polymorph when oxamic acid was deuterated.
View Presentation... a 2.8MB PDF file.

Anne Kavanagh

Autumn Meeting - 8^th November 2007, AstraZeneca, Macclesfield

Report BCA Industrial Group Autumn Meeting

Introduction

The British Crystallographic Associations (BCA) Industrial Group (IG) Autumn meeting was held at the AstraZeneca Silk Road Business Park site. The day long meeting gathered researchers and users of X-ray Diffraction (XRD) from industry, university departments and commercial organisations.

The meeting was an opportunity for BCA-IG members to present work to their peers and develop a better understanding of XRD techniques. The day consisted of a series of presentations split into two sessions. The morning session focussed on the Applications of Rietveld Refinement. The afternoon session focussed on Crystallography in Industry.

The meeting was opened with a welcome from Anne Kavanagh (Astra Zeneca) and Steve Norval (Intertek MSG), thanking all those attending the meeting and especially those giving presentations throughout the day.

Morning Session - Rietveld Applications

Chairs: Steve Norval and Jeremy Cockcroft

Rietveld: Pitfalls for the non-Expert
Jeremy K Cockcroft, University College, London

It is nearly 40 years since Rietveld developed the idea of whole pattern fitting of powder neutron-diffraction data in order to refine crystal and magnetic structures. Jeremy reviewed the method, its development and its uses. For the unwary or uncritical user, he also highlighted some of the many potential pitfalls.

H M Rietveld published a paper entitled 'A profile refinement of nuclear and magnetic structures' in the Journal of Applied Crystallography 1969 2 pp 65-71. Peak positions are determined by the unit cell and may be affected by instrument set-up. Peak Area Intensities are determined by the contents of the unit cell i.e. crystal structure. Peak width is determined by the measurement resolution and is a fundamental parameter factor derivation. Peak shape is a convolution of many symmetric functions - in reality it approximates very close to Gaussian. Jeremy pointed out that axial divergence gives asymmetry at low angle. Historically, the background was determined graphically (by eye) and substracted from the data. The modern approach is to add the background to the calibration model.

The refinement approach is to measure y_i(obs) and calculate y_i(calc) from the model and to use classical non-linear least squares to optimize the parameters.

Brief History

The uses of Rietveld have expanded. It can now be used for structural refinement, quantitative analysis, kinetic analysis and thermo-diffraction data-analysis. Modern computers, diffractometers and software are easier to use. However, beginners have limited crystallographic knowledge and are under more time pressure and so have less time to think!

To avoid most pitfalls

A big problem is trying to extract information not supported by the Powder X-ray Diffraction (PXRD) data. You could get the wrong model and the wrong unit cell. Powder diffraction does not measure symmetry since for high symmetry systems you have multiple possibilities from the Laue class. Jeremy's final message was for everyone to be self critical of refinements.

Time Resolved Synchrotron X-ray Powder Diffraction Study of Biogenic Nano-Magnetite
Tony Bell, Daresbury Laboratory

The presentation looked at the transformation of waste iron oxide into commercially useful materials, by using a type of bacteria known as Geobacter sulferreducens. Tony described how the bacterium can produce nanoparticulate magnetite (Fe₃O₄) by the reduction of amorphous Fe(III) oxyhydroxide. High resolution synchrotron X-ray powder diffraction data has been used to study samples taken at different stages of this reaction. Analyses were obtained at the Swiss Light Source (SLS). Topas Rietveld Refinement and Pair Distribution Function (PDF) software have been used to characterize the materials. PDF gives structural information on the transformations. The work shows that the initial amorphous phase first transforms to goethite (FeO(OH)), before undergoing a further transformation to magnetite. Tony said that magnetite is formed faster in the presence of the 'electron shuttling' compound disodium anthraquinone 2,6 disulphonate.

Versatile Rietveld - Applications to Real Materials
Steve Norval, Intertek MSG

Steve began by saying that a basic and common limitation for many laboratories, especially in industry, is the reality of non-ideal materials. Real world materials often contain several phases, some of which can be non-stoichiometric or have uncertain composition. The good news is that useful information can often be extracted from non-ideal samples using Rietveld techniques and the goodness-of-fit criteria considered necessary for establishing a crystal structure are often un-necessarily severe for other applications.

Steve went on to discuss the aluminium hydroxide bio-protein carrier. It has been found to be a less effective carrier after freezing/thawing. The aqueous gel was dried onto a silicon crystal surface. Rietveld refinement using Topas showed it to be very oriented. Freezing followed by thawing was found to produce colloidal clusters with random orientation. It is believed that during freezing the gluten on the films squashed the bio-protein plates. The orientation parameter in the Rietveld software was used as an indicator of carrier mobility.

A second example was the study of silver bromide crystals in photographic film. The crystals are AgBr(111) preferentially aligned with the film but it was found that after developing the film the AgBr(111) reflection was less intense and broader. Rietveld analysis is used to study the AgBr crystal size/distribution and to determine the presence of lattice distortions. It was found that the silver lattice contracted by 0.09%.

Rietveld Refinement of organic molecules: investigation of hydrogen-bond patterns
Michela Brunelli, ESRF, Grenoble

Michela has carried out work at station ID31 at the European Synchrotron Radiation Facility (ESRF). The synchrotron provides a high brightness, highly parallel, highly polarised beam with a wide energy spectrum; 1.5 - 300keV. The particular strengths of ID31 are accurate peak positions and very high 2-theta and energy resolutions. 'Hard' energies in the range 6-60keV are produced. The station can be used for structural studies, in-situ studies, anomalous scattering, high throughput (using an automatic sample changer), quantitative analysis, microstructural analysis, residual strain and Glancing Incidence and Reflectivity work. The mapping of residual stresses and strains generated by mechanical surface treatments has also been investigated e.g. those produced by laser peening, welding, riveting and machining. Once a good single-phase sample has been analysed it is indexed using Crysfire or N-Treor.

Michela has investigated hydrogen-bond patterns in organics and looked into temperature effects. A particular focus has been the analysis of new nitro-S proligands which have applications in Biometrics. The role played by water in the crystal structure of amino-acid has been found to be diverse.

Rietveld Fitting of High-throughput Data
Stuart Turner, Crystallography, Birkbeck College

Stuart described the experimental methods that have been used with the Rapid2 detector which is located at station 6.2 of the Synchrotron Radiation Source (SRS). Due to the high time resolution of this detector, it is possible to collect many 'frames' of data in a short time. This not only enables fast reactions to be monitored 'in-situ', but also presents the problem of dealing with large data sets which need to be refined in a reasonable time. Station 6.2 has the following

      - A computer controlled furnace is used to alter the sample environment by increasing or decreasing temperature during the measurement.

      - A liquid nitrogen cryojet (approximately 120K) which can be used to study negative thermal expansion.

      - A mass flow controller which switches gasses sequentially. An open-ended capillary with a 'gas-stub' lets gas through the sample.

      - A solvent can be used which is introduced from a solvent 'bubbler' inside a water bath. Air passes through the bubbler from an air line - this transports the solvent.

Reasearch into the dehydration of the industrial by-product, Gypsum, has been carried out. This has recycling issues. Kiln Feed raw materials, calcite, quartz, dolomite have also been examined. High temperature phase transitions have been observed and the 'setting' times of different types of cement have been determined. Stuart talked about the 'Le Bail' and 'Rietveld' refinement methods in the 'Fullprof' suite. He stressed the need to get good initial models before carrying out batch refinements.

Afternoon Session - Crystallography in Industry

Chair: Judith Shackleton

Inverse Pole Figures - their application to a detailed understanding of XRPD measurements
Ian F. Ferguson, UKAEA, retired

Ian discussed what inverse pole figures are and how they can be used to obtain a detailed understanding of anomalous intensity and 'line' positions in X-ray Powder Diffraction (XRPD). Inverse pole figures are produced by the projection of a pole figure onto an equatorial plane. A 'Nelson-Riley' function is used. They reveal information about texture and cell-edge variations. Specific emphasis was placed on the use of inverse pole figures in the metals industry. Inverse pole figures can present different characteristics of a metal such as the number of times it had been rolled and in what direction. Ian said that this type of information should be provided by any comprehensive Rietveld analysis program.

Utilising energy dispersive synchrotron X-ray diffraction for characterising stresses in two-phase titanium linear friction welds
Moataz Attallah, Materials Science Centre, University of Manchester

Linear Friction Welding (LFW) is a novel welding technology, which utlises a combination of frictional heating and plastic deformation to join difficult-to-weld materials. Moataz discussed the welding of the Ti-6246 two phase alloy used in the construction of parts of aircraft engines. The application environment is extreme with high pressures and temperatures. LFW is a solid state joining process where frictional heat is generated by rubbing two components together. The materials are not melted - only softened. Using LFW for Ti-6246 is of benefit since it has a more efficient material usage compared with other welding alternatives. In addition, less oxides are produced compared with 'fused' welding and the weld is 30% lighter than a traditional weld. Work was carried out at the ID15A station, ESRF, Grenoble. Synchrotron X-rays have excellent penetration of metals, enable short measurement times and offer high resolution. Multi peak fitting was carried out using GSAS. The phase fractions and the Young's Modulus and lattice parameter were determined for each phase. A Field Emission Gun (FEG) Scanning Electron Microscope (SEM) was also used to obtain the size of the grains. When joining high temperature materials high residual stresses are generated, which can be detrimental to the joint properties, requiring development of an appropriate post-weld heat treatment. Post heat treatment was found to significantly reduce the stresses in the Ti-6246 alloy caused by the LFW. Moataz concluded by saying that there are a number of limitations to the method. These arise from variations in residual stress across a weld, texture, deformation/re-crystallisation and the nature of the microstructure (grain size).

Industrial Applications of Residual Stress Measurement using XRD
Andrew Winn, Materials Testing and Analysis (MTA) Unit, Manchester Materials Science Centre

Andrew began by saying residual stresses are those stresses which reside within a material, component or assembly when all external forces have been removed. They add or substract from any applied stresses. To illustrate this he referred to steel rods in concrete. The rods are held under tension as the concrete sets. Residual stresses result from a shape mis-fit in the microstructure of materials (putting a peg in a round hole). Tensile residual stresses cause cracks to open-up and propagate. Compressive residual stresses cause cracks to close. More information can be found at http://www.metalimprovement.com/. Hole drilling is a destructive method whereas X-ray Powder Diffraction (XRPD), magnetics, Raman and Ultrasonics (wave speed measurement) are non-destructive. In XRPD it is strain that is measured and from this stress is calculated. The lattice planes are used as a strain gauge. Andrew uses the sin²-psi method to calculate stress.

Application of X-ray Diffraction In-Line to Monitor Solution Mediated Phase Transformations between Polymorphs during Batch Crystallisation Processes
Robert Hammond, University of Leeds

Robert began by saying in-line XRPD had been applied to a flow-through cell for the quantitative analysis of a polymorphic phase transformation, within a slurry, during crystallisation that converted the metastable alpha-form of L-Glutamic Acid (LGA) to the stable beta-form. Batch crystallisation, often the final stage in the synthesis of speciality materials and fine materials and is a quicker way to produce a 'large' amount of material. However, it can give rise to polymorphic forms. Crystallisation involves 'nucleation' (3D formation and assembly of molecular clusters) and 'growth' (2D growth of clusters to crystals). The crystals need to match the specification with respect to their size and shape. Early detection of polymorphs is key to the successful control of polymorphism The in-line instrument has a Bede Microsource with a curved Position Sensitive Detector (PSD). Profiles were recorded continuously with data acquisition times of 200 seconds. The area under representative peaks for each phase is proportional to the concentration of each phase at that time. Robert said that the rate limiting process was found to be the dissolution of the alpha phase of LGA. The study revealed the detection limits to be 0.4 wt% for the beta form (needle-like crystals) of LGA in LGA-methanol slurries containing mixtures of both polymorphs of LGA. It showed that In-Line XRD is a valuable technique for the measurement of dynamic phase transformations.

Analysis of Polymorphic Forms - Understanding and Quantifying Packing Similarity
James Chisholm, Cambridge Crystallographic Data Centre (CCDC)

The identification, control and characterisation of polymorphic forms is of clear importance in the pharmaceutical industry. XRD is the natural method to identify polymorphs during crystallisation experiments. As the physicochemical properties (e.g. solubility, melting point, hardness, chemical stability) can be highly dependant on the molecular packing pattern, an understanding of the similarities and differences between structures at the molecular level is crucial. James uses the new Materials module of the Mercury Crystal Structures Database (CSD) which has been used to study the packing patterns of polymorphs, hydrates and solvates. The Mercury CSD Visualizer program can compare crystal structures and identify similarities between polymorphs and specific regions that are similar. It can investigate packing to provide insights into the forces that direct crystal growth. Other new features in Mercury allow searching of the CSD for specific interaction motifs or more general packing features as well as comparison of the geometries of these features.

Mark Farnworth and John Kaniuka
Pilkington Group Limited

Last updated 06-Jan-2008

Report any errors or omissions on this page to the Industrial Group Webmaster, e-mail:djtaylor@lineone.net © Copyright 2008, BCA. All rights reserved.