• Detector calibration
• Use of available calibration files (most common option)
• You make your own calibration
• Wavelength / 2Theta offset calibration
• Angle Calibration

D20 detector consists in 1600 independent cells whose signal is transmitted to the data acquisition system through the same number of amplifiers which are adjusted individually. As the detection efficiency may vary slightly from cell to cell, one must normally correct patterns by multiplying the raw data by a correction factor _i called efficiency coefficient:

• I_i(corr.) = alphai*Ii(raw) [i = 0 -> 1599]

Efficiency coefficients are calculated from measured diffraction patterns of a vanadium (constant wavelength, better choice), water or plexiglass rod (some inelastic scattering contribution due to hydrogen, so no constant wavelength). This calibration is performed by instrument responsibles typically at the beginning of every reactor cycle, after changing amplifieres or discriminator thresholds on the detector or after change of wavelength. However, in some cases, you may prefer to make your own calibration. You should discus this with your local contact.

The calibration files for angle calibration and efficiency correction are stored on d20lnx.ill.fr (and everywhere else where you access the ILL fileservers) in /home/cs/lambda/CALIBRATION. They had till February 1998 a name in the format cycnumorn.d20 where cyc correspond to the cycle and numorn the first numor to be read with this calibrations. Now they have a name in the format instrument_year_month_numor.wavelength*100. Reading data numors by LAMP automatically read in the right file and performs the corrections if you want so. You may choose another calibration file as well. In the case of an experiment using the radial oscillating collimator (ROC) you even have to do so. See also the corresponding LAMP-Macros.

Efficiencies are (only) slightly depending on wavelength: Relative efficiencies (so the efficiency correction coefficients) of some detection cells may change if these are not exactly geometrically equivalent. As the total detection efficiency changes with wavelength, the position of neutron capture in the detection gap changes as well. This would not change relative efficiencies if all detection cells are geometrically totally equivalent.

First set up a vanadium (or plexiglass) rod or vanadium cylinder filled with water within the vacuum vessel and measure its diffraction pattern. For a proper determination of the efficiency coefficients, all cells of the detector should scan the same part of the diffraction pattern. That is not possible as the 160 deg-PSD scans only up to 30 deg but there are enough overlap to do an efficiency calculation and powder pattern reconstitution by a TwoTheta scan (as large as possible) with a stepwidth of 0.1 deg, the width of detector cells. Proceed as shown in the following example for MAD:

par mut User LocalContact (User and local contact names)

par sub Vacuum Vessel noROC (Commentary)

par mco Calibration V rod VanaChieux 10mm (Commentary)

scan 2theta -32 -2.5 0.1 60 1 1

scan 2theta-32 -6 0.1 5 1 1 (if the ROC is used)

Calculate the correction file by running the function mkcal6.pro in LAMP. This program will yield a workspace containing the efficiencies for each cell. You have to store the inverted efficiencies as a calibration file.

Of course, you are not obliged to do a long scan, the only problem will be the region of the detector that is below the beamstop all the time - efficiencies won't be correct. However, sometimes this does not matter. Once you have a reconstituted, correct powder pattern of a calibration sample, you may re-do calibrations just by comparing it to just one re-measured diagram under exactly the same conditions.

The zero-shift depends on the adjustment of 2theta encoders of the detector. It should normally be refined from a standard pattern. To do so, run a standard sample (e.g. NIST-Silicon), read and export the data with LAMP (e.g. flag,/eff & wn=rdrun(numor) & export_dat,wn) and analyse the data with a profile refinement program such as FullProf or GSAS.

Common standards :

All lattice constants at 25°C
YIG is used essentially to measure the resolution curve of the instrument

Control files to run FullProf are available in /home/cs/lambda/RIETVELD/takeoff/[monochromator/][alpha1/][monochromator-aperture/]. They are identified by the subdirectory (takeoff-angle, incident divergence, horizontal monochromator aperture) and a name which reminds both the chemical composition of the standard and the numor, e.g. /home/cs/lambda/RIETVELD/42/hopg/nac_1645179.pcr for a control file corresponding to a Na2Ca3Al2F14 standard measured with a wavelength of 2.41Å (HOPG monochromator) with no incident Soller collimators and the monochromator fully visible.

There is no automatic wavelength and zero-shift correction during reading in the data by LAMP. It's because both may be affected by changes to the geometry of the beam, even by your sample setting. So there is no file structure foreseen in the moment to read in wavelength and zero-shift for a given configuration.

As the PSD is curved and the 50 microstrip plates of each 32 detector cells are arranged polygonally, the with of each cell - defined by the electrical field - is not constant in 2Theta over the whole detection gap of 5 cm. Close to the Aluminium entrance window detector cells of the mid of microstrip plates covers a slightly larger angular range because they are closer to that window, whilst detector cells near the plate junctions are farer away and covering less angular range. Closer to the microstrip plates this effect becomes less important, it might even be inverted at last due to a paralaxis effect of projection. So not only the width of each cell is slightly different and so their position deviated slightly form the ideal one, supposing each cell to cover 0.1 deg in 2 Theta, but also this deviation is wavelength-dependent : If detection appears closer to the entrance window (high efficiency - high wavelength) the deviation is more important than for detection in the mid of the detection gap (low efficiency - low wavelength).

The structure of calibration files foresees an angle calibration and we will make use of it eventually, but if it won't change a lot on the data, as the effect is even for high resolution barely noticeable. Until further notice we recommend thus to switch this correction off in LAMP (command flag,/noang)