DataCollectionAndTreatment

Data Collection and Data Treatment

Hints for Collecting Data

Synchrotron versus Laboratory Data

• DASH is capable of solving structures from both synchrotron and laboratory X-ray data.

• Synchrotron X-ray powder diffraction stations offer better instrumental resolution and positional accuracy, coupled with a vastly superior incident flux. These benefits manifest themselves best with nicely crystalline samples, where peaks that are overlapped in the laboratory X-ray pattern become well resolved in the synchrotron pattern.

• The collection of synchrotron X-ray powder diffraction data is indicated when laboratory data has failed to provide a solution, or when a precise high-resolution structure solution is required.

Choice of Detector Type

• Of the two common types of detector, scintillation detectors give better resolution but linear PSDs (position sensitive detectors) offer vastly better counting statistics. Use of either is likely to yield good results.

• If the sample line widths are well matched to the resolution of the PSD, there is little to be gained by switching to a scintillation detector.

• Most high-resolution powder diffractometers at synchrotrons currently use one or more scintillation detectors. Increasingly, though, image plates are being used to shorten data collection times and provide better counting statistics. The choice of which to use depends very much upon the prevailing instrumental set up; the station scientist is in the best position to advise you on such matters.

Required Resolution

• Global optimisation methods of structure solution do not require data collected to such a high angle as do direct methods of structure solution. Typically, if data can be collected to approximately 1.5 Å resolution, then structure solution will be feasible (Note that we are speaking here of spatial resolution within the data set). The 2θ value corresponding to this resolution can be easily calculated from:

  2θ_1.5Å = 2*sin^-1(λ / 3.0 Å)

• Data should be collected to as low a 2θ value as possible since the low-resolution reflections help with the indexing process. Generally, this low angle limit is imposed by the diffractometer, because there is a risk of damage to the detector from the straight-through beam at 2θ values close to zero.

Obtaining Monochromatic Radiation in the Laboratory

• The radiation source should ideally be monochromatic.

• In decreasing order of preference, monochromatisation can be achieved by:

• Use of a primary monochromator, i.e. one that lies between the X-ray tube and the sample. For a copper X-ray tube, this eliminates CuK_α2 and CuK_β, leaving only CuK_α1.

• Use of a secondary monochromator, i.e. one that lies between the sample and the detector. These usually eliminate only CuK_β, leaving both CuK_α1 and CuK_α2. The CuK_α2 contribution then needs to be stripped out algorithmically, which is a complication best avoided.

• The use of filters to achieve monochromatisation is considered to be inappropriate for structure solution work.

Choice of Wavelength

• It is doubtful if any particular wavelength offers an advantage when dealing with organic powder samples. Whilst longer wavelengths spread out the pattern and would seem to decrease the chances of peak overlap, the peaks themselves widen and thus no advantage is gained.

• In situations where it is possible to select a wavelength (e.g. at a synchrotron), it should be chosen to maximise the incident flux, unless compelling reasons (such as absorption) dictate otherwise.

Data Collection Geometry

• Transmission geometry is recommended, with the sample in a rotating capillary.

• It is also possible to collect diffraction data in transmission mode when the sample is held as a thin film in a suitable attachment. Reflection geometry may be used, but there is a high risk of preferred orientation having a significant impact on the diffraction pattern. Whilst a small degree of preferred orientation can be tolerated in a structure solution, it is a complication that is best avoided.

Background Reduction

You should endeavour not to introduce any additional background scattering. For example, it may be appropriate to use borosilicate capillaries rather than glass or quartz, in order to avoid seeing the amorphous scattering from the capillary manifest itself as a background hump.

Zero-Point Calibration

Although it is possible to refine instrumental zero-point errors in the whole pattern fitting stage, it is always preferable to calibrate the instrument prior to a structure solution using a well-defined standard sample, e.g. NBS silicon.

One Long Scan versus Several Short Scans

It is generally better to perform several short scans and sum them together using the data reduction software, rather than collecting a single long run. For example, four one-hour runs are preferable to one four-hour run. Prepare each sample fresh in order to reduce preferred orientation.

Choice of Step Size

• Ideally, you should have plenty of points across every peak in the diffraction pattern in order to accurately describe the underlying peak shape.

• If you use a step size that is too large, you risk missing subtle features such as peak shoulders that may be critical at the indexing stage.

• If there is any doubt, it is better to collect data on a finer grid, since a coarser grid can always be constructed later by re-binning the data; the converse is obviously not true.

How Long to Count For

• Obviously, the longer the time spent on collecting data, the more closely it will resemble the true diffraction pattern.

• For each doubling of the collection time, the estimated standard deviations are improved by a factor of 2^1/2. Eventually, a stage is reached where substantial increases in collection times are required in order to achieve modest improvements in the signal-to-noise ratio.

• As a general guide, the data should be collected sufficiently long that reflections can be clearly distinguished from the background at around 1.5 spatial resolution. Not all samples will diffract strongly to this resolution, but it remains a useful rule of thumb.

Optimising Use of Data Collection Time

• For a given data collection time, the question arises of how to optimise the use of that time. A common formula is:

  Time per step = (expt duration in seconds) / ((2θ*_max* – 2θ*_min*) / step size)

• This scheme gives equal weighting to all data points and takes no account of the fact that the diffracted intensities at low angle will be much stronger than those at high angle. In many cases, this may be sufficient, but a more sophisticated data collection strategy is described in J. Mater. Chem. (1997) 7, 569-572. Implementing such a data collection scheme is a simple matter and is strongly recommended when using scintillator detectors on a laboratory or a synchrotron source.

Neutron Data

DASH does not currently handle neutron diffraction data.

Hints for Treating Data

Lorentz Correction

DASH always corrects the data for Lorentz effects, so no correction should be applied to the data in advance.

Polarisation Correction

• In the case of synchrotron data, DASH assumes that the incident radiation is vertically polarised and, provided that the Synchrotron radiation option is turned on when the radiation wavelength is entered, no pre-processing of the data is necessary.

• In the case of laboratory X-ray data, DASH applies an appropriate polarisation correction provided that the Laboratory radiation option is turned on when the radiation wavelength is entered, and no pre-processing of the data is necessary.

• The exact form of the polarisation correction applied is suitable for instruments equipped with a primary monochromator. Whilst not exactly correct for different instrumental geometries, it is still a good enough approximation to be useful.

• It is always possible to fully correct for the polarisation effects of particular geometries, if your data processing software allows it, before inputting the corrected data into DASH. Within DASH, the data should then be treated as having been obtained using monochromatic synchrotron radiation.

• A Lorentz correction should never be applied in advance.

K_α2 Stripping

• DASH is able to handle data collected using monochromatic radiation only. The use of more than one incident wavelength is a serious complication that should be avoided when tackling problems of structure solution.

• However, it is possible that your diffractometer software may provide suitable K_α2 stripping routines that allow you to export a data file from which the K_α2 contribution has been removed algorithmically. In such cases, the exported file may be treated within DASH as a monochromatic laboratory X-ray data set.

• NB: Stripping algorithms inevitably introduce some degree of uncertainty into the data.

ESDs

• Each (2θ, count) data point must be accompanied by an estimated standard deviation (ESD). Ideally the diffraction data set input into DASH should consist of three columns of data:

  <2θ> <count> <estimated standard deviation>

• Many diffractometers will output such a listing. However, if only

  <2θ> <count>

  are available in the input file, DASH will automatically calculate the ESD values from counting statistics.

Background Subtraction

• Some diffractometer software may offer the possibility of background subtraction. However, it is better to leave modelling of the background to DASH, unless there is a good reason to do otherwise e.g. if you have an appropriate physical model for the background and therefore can remove the background with confidence.

• DASH provides a robust Monte Carlo background fitting option that is recommended for use with most data sets.

Checklist for Diffraction Data

• Use synchrotron or monochromatic laboratory X-ray radiation.

• If possible, collect data to at least 1.5 Å resolution.

• Use transmission capillary geometry.

• Do not apply Lorentz or polarisation corrections, or subtract the background, before entering DASH. DASH will assume raw data and perform these steps itself.

• ESDs are preferable in the input file, i.e.

  <2θ> <count> <estimated standard deviation>

  If only <2θ> <count> is available, DASH will automatically calculate ESDs.