The precision to which C can be estimated by this method is very dependent on the precision and the extent of the intensity data sets. The data sets must be collected with CuKα radiation to as high a 2θ as possible. Athough the nature of the crystal used in the data collection is not important for this process, it must be very stable and the secondary extinction effects should be relatively small. The same crystal must be used to collect a data set using a Ni filter, and an identical data set using the monochromator for which the perfection factor is to be estimated.

The processing of these two data sets prior to the execution of PERFAC is as follows. Note the example data inputs given below.

(1) Process the intensity data set collected with a Ni filter using the routines STARTX, DIFDAT, ADDREF, SORTRF and GENEV. Note in particular that the reduce parameters in ADDREF are 'itof rlp2'. The bdf output from GENEV after this sequence will be referred to below as bdf fil.

(2) Process the intensity data set collected with the graphite monochromator using an identical command sequence to (1), except that the reduce parameters in ADDREF must be 'itof rlp3 xray 26.2 0.0'. The bdf output from GENEV after this sequence will be referred to below as bdf mn0.

(3) Process the same data set as used in (2) except that the reduce parameters in ADDREF must be 'itof rlp3 26.2 1.0'. The bdf output from GENEV after this sequence will be referred to below as bdf mn1.

(4) PERFAC requires (and complains bitterly otherwise) that the order, and the number, of reflections stored in the three bdf's be identical! If this is not the case (it can be surprisingly difficult with present-day auto-aligning procedures to collect two identical and independent data sets) then it will be necessary to use MODHKL to delete any superfluous reflections. See the example input given below.

(5) Execute PERFAC after ensuring that the three bdf's are correctly named. Be very careful about this as PERFAC cannot distinguish between these files and the results from this calculations are critically dependent on the correct F's being on the correct file!

(6) Change the default F and ratio thresholds (entered on the PERFAC line) only if absolutely necessary. These thresholds are intended to prevent very small, and the occasional badly measured, F values from strongly influencing the average ratios. Read Hakon Hope's paper carefully and plot out for several values of C the F(mono)/F(filter) ratio versus mean 2θ. The value of C that gives the flattest curve (especially in the 2θ range 20° to 70°) will be the most appropriate C to use for this monochromator.

Reads the F(filter) values from bdf fil

Reads the F(mono C=0.) values from bdf mn0

Reads the F(mono C=1.) values from bdf mn1

This is a typical sequence of runs needed to do the PERFAC calculation.

(1) This run produces the bdf fil.

title C15H19Co P21/a PERFAC run : Cu rad. with filter LP

STARTX

CELL 11.5885 8.6455 12.0101 90 102.027 90

cellsd .0007 .0003 .0005 0 .005 0

sgname -p 2yab

celcon co 4

celcon c 60

celcon h 76

exper 1 1.54179

DIFDAT cad pri 50 excl inst 1 obst 2 sig

attenu 29.8

genscl 3

ADDREF lpin list 20

reduce itof rlp2

bdfin all

SORTRF print 20 aver 1

GENEV

copybdf a fil

(2) This run produces the bdf mn0.

title C15H19Co P21/a PERFAC run :Cu rad, monochr & kinematic Lp

...... as above

reduce itof rlp3 26.2 0.0

......as above

copybdf a mn0

(3) This run produces the bdf mn1.

title C15H19Co P21/a PERFAC run :Cu rad, monochr & dynamic LP

......as above

reduce itof rlp3 26.2 1.0

......as above

copybdf a mn1

(4) This run corrects differences between the data sets.

title remove extra data from mono data

MODHKL

delete 0 1 6

delete 1 7 10

delete 1 10 5

delete 6 4 -1

delete 7 2 11

delete 12 2 -1

delete 13 4 1

(5) The PERFAC calculation reads the FIL, MN0 and MN1 bdf's.

title use 'PERFAC' to calculate mono perf. factor copper rad.

PERFAC

For the above PERFAC run the perfection factor was estimated to be 0.80 to 0.85.