I have refined data from SRM 660 which were collected on:
1) a Siemens D500 with an IBM and PSD, the incident slit was .7 degrees and
the PSD window was set to 4.6 degrees, there were no incident sollar slits
(the beam path is 550mm long an 8mm wide (.83 degrees divergence), the
recieving sollar were of 1.5 degrees divergence, and 2) a Siemens 5000
with automated slits, incident sollars of unknown divergence (probably
about 1 degree), a .05 degree recieving slit, and a radius of 217mm. I used
GSAS to obtain the following:
FROM the D500
1LaB6 #27 Disc PSD .12--1--.3 GENLES Version 6.2 17-JUN-94 13:04:00 Page10
Powder data statistics Fitted All Average
Bank Ndata Sum(w*d**2) wRp Rp wRp Rp DWd Integral
Hstgm 1 PXC 1 13399 1.11486E+05 0.0526 0.0258 0.0526 0.0258 0.207 0.970
Powder totals 13399 1.11486E+05 0.0526 0.0258 0.0526 0.0258 0.207
No serial correlation in fit at 90% confidence for 1.949 < DWd < 2.051
Cycle 54 There were 13399 observations. Total before-cycle CHI**2 = 1.1149E+05
Reduced CHI**2 = 8.331 for 17 variables
Profile coefficients for histogram no. 1 and for phase no. 1:
Coeff. : GU GV GW LX LY
Value : 4.724E+00 -8.760E+00 8.380E+00 1.424E+00 1.426E+00
Sigmas : 1.295E-01 2.376E-01 1.069E-01 4.542E-02 9.474E-02
Shift/esd: 0.05 -0.02 0.01 0.01 -0.02
From the D5000 with the incident slit fixed at .85 degrees:
1LaB6 .2,.01--6--.85 GENLES Version 6.2 19-JUL-94 09:03:58 Page 16
Powder data statistics Fitted All Average
Bank Ndata Sum(w*d**2) wRp Rp wRp Rp DWd Integral
Hstgm 1 PXC 1 13599 59550. 0.1706 0.1248 0.1706 0.1248 0.549 0.983
Powder totals 13599 59550. 0.1706 0.1248 0.1706 0.1248 0.549
No serial correlation in fit at 90% confidence for 1.950 < DWd < 2.050
Cycle 32 There were 13599 observations. Total before-cycle CHI**2 = 5.9550E+04
Reduced CHI**2 = 4.385 for 19 variables
Atom parameters for phase no. 1
frac x y z 100*Uiso
LA ( 1) Values : 1.000 0.000000 0.000000 0.000000 0.307
Sigmas : 0.011
Shft/esd: 0.00
LA(1) moved 0.00A sum(shift/e.s.d)**2 : 0.00
B ( 2) "B(2) " not refined.
Atomic parameter sum(shift/error)**2 for phase 1 : 0.00
Diffractometer coefficients for powder data:
Hist. : 1 PXC
Dif A/Pola : 0.55025
Sigmas : 0.00362
Shift/esd : 0.00
Profile coefficients for histogram no. 1 and for phase no. 1:
Coeff. : GU GV GW LX LY
Value : 9.052E-01 -8.691E+00 3.324E+00 3.727E+00 6.073E-01
Sigmas : 1.238E+03 4.216E-01 1.238E+03 8.015E-02 1.298E-01
Shift/esd: 0.00 -0.04 0.00 -0.02 0.02
From the D5000 with the inc. slit set at "small" (6mm width), the data were
then corrected with a 1/sin(theta) function:
1LaB6 small/smspec m6lab3 GENLES Version 6.2 6-JUL-94 15:50:44 Page 16
Powder data statistics Fitted All Average
Bank Ndata Sum(w*d**2) wRp Rp wRp Rp DWd Integral
Hstgm 1 PXC 1 13389 23244. 0.1645 0.1159 0.1645 0.1159 0.467 0.983
Powder totals 13389 23244. 0.1645 0.1159 0.1645 0.1159 0.467
No serial correlation in fit at 90% confidence for 1.949 < DWd < 2.051
Cycle 33 There were 13389 observations. Total before-cycle CHI**2 = 2.3244E+04
Reduced CHI**2 = 1.738 for 18 variables
The value of the determinant is 0.1114*10.0**( -12)
Atom parameters for phase no. 1
frac x y z 100*Uiso
LA ( 1) Values : 1.000 0.000000 0.000000 0.000000 0.142
Sigmas : 0.010
Shft/esd: 0.00
LA(1) moved 0.00A sum(shift/e.s.d)**2 : 0.00
B ( 2) "B(2) " not refined.
Diffractometer coefficients for powder data:
Hist. : 1 PXC
Dif A/Pola : 0.57706
Sigmas : 0.00362
Shift/esd : 0.00
Profile coefficients for histogram no. 1 and for phase no. 1:
Coeff. : GU GV GW LX LY
Value : 2.806E+00 -7.232E+00 6.361E+00 2.623E+00 2.137E+00
Sigmas : 1.547E-01 3.155E-01 1.472E-01 4.712E-02 8.466E-02
Shift/esd: 0.00 0.00 0.00 0.00 0.00
I don't use the GSAS asymetry correction, it doesn't work well with profiles
from this equipment and generally leads to instability. I am pestering Bob
incessantly about incorporation of the Larry Finger (et.al) model for
asymetery into GSAS. I do not feel the Cagliotti function is well suited
to XRD data, Langford presented a more suitable one in ACCURACY IN POWDER
DIFFRACTION II, Bob Cheary and I are also working on a better function for
FWHM vs. two theta, I will speak of it at Denver '94.
The second two data sets compare the results from data collected with and
without the use of the variable slits on our D5000. Both these data are
acceptable. However, one notes the difference in the temperature and LP
factors. The application of the correction to data collected at a constant
area of illumination to yield that which would result from a constant volume
of illumination is predicated by an assumption of uniform flux density with
respect to the angle the beam makes with the tube anode. This is not a
reliable assumption; as the tube ages the electron beam will etch a trench
in the tube anode. Thus, as the tube ages the flux density with respect
angle will most certainly change. Problems with this approach will appear
in the temperature and LP factors. Owing to their large impact on the scale
factor, I do not use the method unless the shape or size of the specimen
demand it. I have additional data collected on SRM 1976, using NBS*QUANT,
during its certification which also indicates problems with the use of
"corrected" data from theta compensating slits.
I have never heard of any observation of peak splitting with the use of
SRM 660. If anyone else has, please waste no time in reporting it to me.
Jim Cline
Ceramics Division
NIST