Area detectors in chemical and protein crystallography

Professor J R Helliwell, Chemistry Department, University of Manchester, M13 9PL.

The scope of X-ray crystal structure analysis has improved greatly with the evolution of better detectors. Different technologies have sought to combine the 2-D nature of photographic film and the sensitivity and 'on-line' nature of the scintillation counter. Different detectors utilise different physical mechanisms to record the X-ray reflections. The multiwire proportional chamber (MWPC) uses an absorbing gas and delay line electronics for coordinate encoding of each X-ray photon. The solid state TV and charge coupled device (CCD) use a sensitive phosphor and charge scanning with analogue to digital (AD) readout. Image plates (IPs) record the signal via colour centres followed by a laser scan and also AD readout. The devices have various strengths, where no single detector offers an ideal performance over all parameters. Important parameters include detector quantum efficiency (DQE), dynamic range, number of resolution elements (derived from the aperture size, point spread factor and pixel size), readout time and wavelength range capability.

Novel opportunities have arisen for structure analysis with these devices. Several examples, especially involving synchrotron radiation, will be given. An 0.94A resolution study of the 25kDa protein concanavalin A has become possible. This involved use of intense synchrotron radiation and a fibreoptic type CCD, where the DQE proved critical to make good measurements beyond 1.05A resolution, compared with the use of an IP on the same beam. However, four crystals were needed, although cryocooled, because of the limited aperture of the CCD detector. In another study the CCD and an intense 0.5A wavelength synchrotron beam, as well as cryocooling, proved effective in extending the resolution of an octahexylphthalocyanine to 0.8A resolution, compared with 1.2A with a rotating anode CuKa single counter diffractometer. Anomalous dispersion applications at multiple wavelengths highlight the need not only for precise measurements but also fast and automatic read out, e.g. from online IP or CCD devices, as multiple data sets are measured. Examples of multiwavelength anomalous dispersion (MAD) studies of a brominated oligonucleotide, a selenomethionine labelled protein, manganese concanavalin A and mixed metal aluminophosphates will be given. In a final category of examples the advantage of fast readout greatly improves perturbation and dynamical crystallography studies. The determination of the space group transition temperature of the octahexylphthalocyanine liquid crystal referred to above will show the complementary characteristics of a fast readout CCD and a large aperture, but slower readout IP. A time-resolved Laue diffraction study of the K59Q slow mutant of hydroxymethylbilane synthase and its enzymatic reaction in the crystal will be described based on a large aperture and fast readout image intensified type CCD.

For the present many new opportunities exist then to harness the current generation of detector devices. Further device development will involve mosaic CCDs to provide increased aperture for the fibreoptic type CCD; parallel readout IP scanners to speed up these devices; novel pressurised MWPCs e.g. for use at longer wavelengths in protein microcrystal data collection; and finally the 'pixel detector', with readout electronics for each individual pixel, ideal for harnessing the highest count rates for fast data collection either in genome related structure studies or fast time-resolved work.

Area detectors in chemical crystallography: Experiences, opportunities and challenges. Comparison of single-crystal structure determinations with area and single point detector diffractometers.

Matthew S. Legge and A. Guy Orpen, School of Chemistry, University of Bristol, Bristol BS8 1TS, U.K.

The time required for data collection for single crystals may be cut drastically by use of an area detector diffractometer. Typically 6-15 hours are needed as opposed to several days with a conventional four-circle diffractometer. This study analyses the quality of results obtained using each method of intensity data acquisition. A comparison is presented of X-ray structures whose data collections have been made on both three-circle area detector (Siemens SMART) and four-circle (Siemens P4) diffractometers. Samples for analysis were chosen so as to represent a range of small molecule crystal types: organics, organometallics; different crystal systems and crystal qualities. Methods of absorption correction were compared for data collected on both diffractometer types. In common with other studies involving comparisons of data sets, the quality of each structure analysis was evaluated by examining the crystal and intensity data, refinement parameters and also the data collection conditions. Data have been analysed using a variety of statistical techniques (including normal and half-normal probability plots) on intensity data, coordinate and displacement parameters and uncertainty estimates.

Badly diffracting crystals and large molecules

Dr Harry Powell, Automation Office, University Library, West Road, Cambridge.

Area detectors in the laboratory rather than at a synchrotron offer the crystallographer the opportunity to collect data on samples which would previously have been rejected. The principal reasons for this lie in the ability to collect many data simultaneously; the implications of this will be discussed with reference to a number of data collections and structure determinations performed in the Chemistry Department in Cambridge, none of which would have been attempted with a conventional four-circle diffractometer. The examples range from a small organometallic through large organic species, main group and transition metal clusters, to synthetic "biological" molecules.

SYNCHROTRON CHEMICAL CRYSTALLOGRAPHY

Prof. William Clegg. Department of Chemistry, University of Newcastle upon Tyne and CCLRC Daresbury Laboratory

A new high-flux single-crystal diffraction facility, designated as Station 9.8, has been constructed at the Synchrotron Radiation Source, CCLRC Daresbury Laboratory, for use in chemical and materials crystallography. X-rays in the wavelength range are selected and focused by a bent triangular silicon monochromator and glancing-angle palladium-coated mirror, to achieve a highly collimated beam of size similar to, or smaller than, those generally used from conventional sources, but several orders of magnitude more intense. An Enraf-Nonius CAD4 diffractometer with scintillation detector is available for single-reflection measurements, and a Siemens SMART CCD detector with three-circle goniometer is installed for rapid data collection in most cases. Low-temperature facilities (Oxford Cryostream) are in routine use; high-temperature and high-pressure experiments are planned. Provision for sample handling includes a Schlenk line for air-sensitive materials and a high-power microscope with micromanipulator for very small crystals.

In six weeks of available beam time to date, a wide variety of samples has been studied, provided by several research groups. They include organic, inorganic, and organometallic compounds; molecular, supramolecular, polymeric, and microporous structures; medium to weak diffractors, very small and thin plate crystals, most of which are not amenable to examination with conventional laboratory X-ray equipment. Some samples previously rejected as unsuitable for area detector systems with rotating-anode X-rays have been successfully characterised, with fully refined structures of publishable quality.

A selection of results will be presented, together with a description of the station, and prospects for its future development, use and availability.

SOLVING MORE DIFFICULT STRUCTURES USING CCD TECHNOLOGY.

Mark R.J. Elsegood, Department of Chemistry, The University of Newcastle-upon-Tyne, Newcastle-upon-Tyne, NE1 7RU, UK.

No matter what the era and state of data collection technology there have always been limits as to what size of crystal or type of structure can be determined. The limits may be set by a number of factors including available X-ray flux, detector sensitivity, crystal sensitivity, software development and a host of others.

We have recently experienced the transition from one era to another as technology has advanced so radically as to significantly push back the fontiers of samples amenable to study. This transition involves the use of area detector technolgy, in this case Charge Coupled Device (CCD) technology, as as it has begun to replace the use of single point scintillation counters. The CCD detector offers considerably enhanced sensitivity, resulting in weaker data being measured more accurately. Having an area detector additionally results in data being acquired very rapidly with the inherant advantages of, often huge, data redundancy. A hemisphere of data is typically collected in 6-15 hours.

The talk will specifically aim to present results obtained over the past two years using the new technolgy where we firmly believe results would not have been achieved using a conventional 4-Circle instrument with a single point counter. These samples range from large polyoxometallate and calixarene species with nearly 200 non-H atoms in the asymmetric unit, to samples with awful reflection profiles, to tiny crystals with weak scattering characteristics, to highly sensitive crystals containing substantial solvent of crystallisation, to those cursed by substantial disorder, or indeed a combination of these.

Systematic Data Errors and their Correction

Dr. N.W. Alcock, University of Warwick

After many years of experience, achieving good data on four-circle diffractometers has become reasonably routine. This is not yet true for area detectors, and this talk examines some of the problems and some solutions.

The principal problem is absorption, and neglect of it can produce very obvious effects in the final structure. Both straightforward absorption by the crystal and absorption by the mounting materials need to be considered. It seems that the latter in particular may be reduced by careful crystal mounting and perhaps by the data collection strategy. Precise correction for crystal absorption (analytical or Gaussian) is, of course, possible in principle. However, it seems that no program to do this for area detector data has yet been widely distributed. In addition, new experimental techniques, such as oil-drop mounting, make it more difficult to determine precise crystal dimensions. When absorption is moderate, correction of area detector data for both types of absorption by a �psi-scan� procedure is very powerful, because reflections are sampled with substantial redundancy in many diffraction directions. Although this procedure affects the physical reality of the refined thermal parameters (because of uncertainty about the theta-dependence of the correction), it provides a good procedure in cases where absorption is not critical. The post-refinement pseudo-correction procedures (�DIFABS�, etc.) have the same severely adverse effects with area detectors as with four-circle data; their application is probably never justifiable, because of the effectiveness of the psi-scan method.

The ideal, for which the routine procedures still have to be developed, is the application of an analytical or Gaussian correction to a well-measured crystal, followed by a psi-scan correction to remove the small additional non-crystal absorption.

Coping with the flood of results

J.E. Davies, (University of Cambridge),

With the advent of area detectors, the time taken to collect a small-molecule dataset can be drastically reduced: two to three datasets per day are possible if suitable crystals are available. But this increase in data-collection efficiency can cause problems elsewhere ....

'Even now, the majority of single-crystal structures are not accessible in the databases because they have not yet been published. The barriers to publication include lack of time to prepare the publication; lack of desire to publish because the structure is not 'good enough' but the chemistry is adequately determined; or the compound is proprietary. To address this, worldwide electronic deposition of unpublished data, with validation safeguards, must be strongly encouraged.' R.A. Sparks et al. J.Res. Natl. Inst. Stand. Technol. 101, 295 (1996).

Although direct deposition is the most likely solution to this ever-growing problem, very few structures enter databases via this route. Attempts to directly-deposit some structures will be described and some serious problems with existing direct-deposit mechanisms will be highlighted.

Use of the whole diffraction pattern

Prof M.B. Hursthouse (University of Cardiff)

Over the last seven years we have demonstrated in some detail the enormous scope for small molecule work provided by use of a "thin- slice" area detector diffractometers (successful structure determinations now in excess of 2000). In addition to the possibility of rapid data collections of a few hours for routine samples, additional special applications have been data collections for small crystals, structures with large unit cells and crystals grown from liquids, and data collections for charge density studies (more than a dozen now completed).

More recently we have been concentrating on procedures which capitalise on the possibility of full recording of reciprocal space to tackle other problems such as twins (and triplets etc), diffuse scattering, diffraction patterns from deformed crystals, incommensurate phases etc. to extract useful structural information. The results of some of these studies will be described.

Exploratory electron density studies

Dr C. Frampton (Roche Products)

Experimental electron density studies on molecular crystals can yield meaningful information on the structure and inter/intramolecular bonding that exists in such materials

Necessary requirements for performeng a charge density are that the intensity data are collected at low temperature, with great care and to high resolution q/l 1.08Å^-1, (2q 100° for MoKa radiation).Additionally, symmetry equivalent data are also needed to improve statistics. These experimental conditions give rise to a large volume of data and, using a single-point detector, the data collection times can be as long as three months. It would therefore represent a great saving in time if an area detector could be utilised for these experiments.

To assess the feasibility of using an area detector, a model compound was chosen for which high resolution data collected on a single-point detector was available for comparison. A series of experiments were performed to optimise the data collection strategy. The results of these initial experiments will be presented.