One important consideration on the issue of thermal parameters vs
occupancies is the Q-range (Q=4pisin(theta)/lambda) where useful information
exists. This is a function of the wavelength used, the resolution, the
complexity of the structure and the radiation (x-rays vs neutrons). The
best results are obtained for simple structures using good-resolution TOF
neutron diffraction. I include a comparison between single-crystal
synchrotron x-ray and TOF neutron powder data on Y2O3 (3g sample, Q up to 25
A-1). Clearly, one obtains very good *anisotropic* D-W factors within one
or a few error bars of the SX values. Also, notice the error bars on oxygen
coordinates, which are 1/10 of the x-ray ones. Short data acquisition times
also yield sensible results. The oxygen occupancy was also refined at the
same time, yielding 1.013(1) for the long run, so there is a little problem
at the 1% level.
Another issue is that of the absolute value of the D-W factors, which is
critical to distinguish between static and dynamic disorder. In order to
get accurate values, one has to perform a careful attenuation correction.
Paolo
SXXD* GEM-8 hours GEM-1 min GEM-10
sec
Y1 U11*100 0.32(1) 0.357(6) 0.36(2)
0.34(2)
U12*100 0.056(7) 0.063(9) 0.04(2)
-0.08(5)
Y2 X 0.96764(3) 0.96745(2) 0.96750(4)
0.9676(8)
U11*100 0.27(1) 0.285(7) 0.33(2)
0.34(4)
U22*100 0.27(1) 0.282(9) 0.29(2)
0.28(4)
U33*100 0.27(1) 0 .28(1) 0.40(3)
0.22(5)
U23*100 -0.026(6) -0.035(8) -0.02(2)
-0.01(4)
O X 0.3907(2) 0.39065(2) 0.39065(6)
0.3908(1)
Y 0.1518(2) 0.15187(2) 0.15188(6)
0.1520(1)
Z 0.3801(2) 0.38009(2) 0.38017(6)
0.3802(1)
U11*100 0.51(5) 0.49(1) 0.44(2)
0.32(5)
U22*100 0.53(5) 0.457(9) 0.41(2)
0.45(5)
U33*100 0.41(5) 0.348(9) 0.25(2)
0.25(4)
U12*100 -0.03(4) -0.034(7) -0.01(2)
-0.04(3)
U13*100 -0.06(4) -0.060(6) -0.07(2)
-0.09(3)
U23*100 -0.05(4) -0.050(8) -0.08(2)
-0.03(4)
*E.N. Maslen et al. Acta Cryst. B52,(1996) P. 414-422 Photon Factory data