Since the GSAS manual has nothing about it here goes:
1) In the profile editing menu (the one where you set max 2-theta, etc.)
you should do the "A" command to make sure the "goniometer angles" are set
correctly. Assuming that you are doing conventional Bragg-Brentano or
Debye-Scherrer diffraction (the're the same for texture calcs.) then the
goniometer angles will be the default values (omega=0, chi=90, phi=0).
2) In the overall parameter menu under "O" (preferred orientation) select
"H" for spherical harmonics. Make sure you do not have any leftover
March-Dollase preferred orientation corrections.
3) Then you should use the default sample symmetry (cylindrical of "fiber"
texture) even if you didn't spin the sample. Spinning the sample makes no
difference for preferred orientation for Bragg-Brentano except to get rid
of any "grainyness" effects.
4) Then select a suitable & low order for the harmonics. The number of
terms will depend on the crystal symmetry. I.e. cubic will give only 1 term
at L=4 whle triclinic gives many more even at L=2. Only even orders of L
are allowed; the program will prevent selection of odd L's.
5) Set refinement flags only for the harmonic terms (not the angles - these
are needed only for full texture analysis with multiple data sets as
discussed in the paper).
6) If there is texture you should see immediate improvement in the fit.
Increase the order of the harmonic carefully in successive refinements
until there is no further meaningful improvement in the fit. For low
symmetry this may be at L=2 or 4. For cubics L might get as high as 6 or 8.
One shortcoming of the current software is that the "tetrahedral" cubics
are not available for the spherical harmonics texture. These are space
groups Pm3, Pa3, Im3, etc. The "octahedral" cubics work fine (Fm3m, Im3m,
I-43d, etc.).
7) You can then plot the axial distributions for each hkl using program
POLFPLOT. Use the "A" option after selecting the desired hkl.
The spherical harmonics are very robust in the refinement as they, by
definition, are orthogonal functions.
Have fun with them, feedback on their usefulness is welcome.
Bob Von Dreele