____________________________________________ | | | Professor G. Stuart PAWLEY, FRSE, FRS | | Professor of Computational Physics, | | Department of Physics, | | Kings Buildings, | | Edinburgh University, EH9 3JZ, | | UK. | | tel (44) 131 650 5300 | | fax (44) 131 650 7165 | | e-mail g.s.pawley@ed.ac.uk | |__________________________________________| This describes the structure of the input data for ALLHKL Each numbered box is a line of input, and when there is no decimal point in the sub-box the number to be placed there is an integer; as the input is formatted (to catch errors) you must make sure the integers are correctly placed in their fields. The position of the decimal point is not important. <1> MINPT is the minimum point in the scan, measured from zero in units of STEP; MAXPT is the maximum. NPEAKS is the number of entries in box 8; the reading of box 8 is avoided if NPEAKS=0. WAVETC is the wavelength in Angstrom, and NEXT determines what is read for box 10; NEXT=0 avoids read of box 10. The title is anything you want. <2> If MSTEPU is (say) 4, then the scan intensities will be averaged in fours, STEP will be increased by 4, and therefore the numbering of the scan is changed; the part of the scan to be used is from MINPTU to MAXPTU in the old units, and can be a smaller scan than the total scan defined by box 1. <3> NC cycles are done, output of sorts on cycle NOUT, more output concerning the generated reflections is given if KOUT=1, IW=1 selects unit weights (otherwise you do it yourself), IRHOMB=1 only if the structure is rhombohedral, LOT=0 if only the input peaks (NPEAKS of them) are to be refined, LOT=1 refines all the peaks that can exist in the scan, NSIDE is part of the decorrelation procedure and this number should be increased from the value in the example if there are correlation problems. <4> Use the values of SIG and REFSIG in the example, as these again relate to the decorrelation procedure, though SIG=0.0 does produce a sensible value. This procedure is a place where some research could usefully be done; my work finished when the procedure was proven. <5> Unit cell parameters, angles in degrees. <6> A scale is needed, a flat background, two blank sub-boxes, the scan zero in (I think) the new two-theta step units, the Rietveld peak shape parameters (start with those in the example; they are fiercely correlated), and finally Howard's peak shape function which has its effect only on the low-angle reflections. <7> LATTIS is 1: triclinic, 2/3: monoclinic P/C, 4/5/6/7: orthorhombic P/C/F/I, 8/9: tetragonal 4/m P/I, 10/11: tetragonal 4/mmm P/I, 12/13/14: cubic m3 P/F/I, 15/16/17: cubic m3m P/F/I, 18: rhombohedral Rbar3, R3, Rbar3c, 19: higher symmetry rhonbohedral, 20: bar3, 21: bar3m, 22: 6/m, 23: 6/mmm, ISCREW(I)=0 for no screw along I-axis, =I for screw along I-axis (I=1,3), IGLIDE(I)=0 for no glide in I-plane, =J for glide in J-direction in I-plane, =4 for n-glide in I-plane (I=1,3). <8> (See box 1 for NPEAKS=0) There are NPEAKS entries here, each with a single reflection h,k,l and m (multiplicity factor), and intensity. There seems to be an inconsistency here involving m-values, and the usage of the program has been with NPEAKS=0, see box 10. <9> MAXPT-MINPT+1 entries here, the scan intensities. A negative entry will be masked out. <10> (See box 1 for NEXT=0). This (and box 8) are restart facilities which are in disrepair; with NEXT=0 there is no need to try to restart refine- ment. The idea here was to edit the last output file and use it as an input file, but this is not important with the fast computers of today. <11> NC entries here for the NC cycles. Here we have the keys which, when set 0/1 switch refinement off/on for the NPQ parameters, namely the 15 parameters of box 6 followed by box 5, plus the NPEAKS peak intensities if LOT=0; all the intensities will be automatically refined if LOT=1.