Summary on the Subject 'Low FWHM and Rp'

Powder diffractometers performances as estimated by the Rietveld method

Posted 20-July-1994 to sci.techniques.xtallography and the Rietveld mailing list
Updated 15-October-1996 (Updated parts are in dark-blue)
Updated 3-December-1998 and 14-January-1999 (in red-brown)
Updated  15-June-1999 for NAC with variable slits
Updated May-2002 with Philips X'Celerator data
A. Le Bail



What for ?

The main purpose of this quest was to get an idea about in-laboratory current performances (1994) of powder diffractometers on the point of view of both the resolution and the possibility of fine whole pattern fitting by the Rietveld method: in what conditions to have both low FWHM and Rp.

It was intended to buy the best powder diffractometer of the moment, according to the above requirements. A very difficult task. Another requirement was that any sample could be studied without any modification of the experimental conditions (avoid fluorescence problem). Several combinations were tested mainly by Siemens, Philips and Rigaku (not all combinations by all manufacturers):

Some tests were made by using a rotating anode. Both Siemens and Philips proposed variable slit systems for having the best current resolution on Bragg-Brentano-type instruments.

SRM standards were particularly designed for this quest, and particularly LaB6 SRM 660 which is the recommended standard for calibration by the JCPDS-ICDD (Powder Diffraction 3, 209-218, 1988). However, tests were also on Al2O3 SRM 1976, quartz and Na5Cr3F14 choosen for medium fluorescence problems with a copper target in the absence of back monochromator.

Those who have pushed their diffractometer to their limits (for instance using 0.018 scattering slit or so on) know that very low FWHM may be obtained but that correspondingly, at the Rietveld refinement stage, Rp (background subtracted) may be > 20%. In such a case, the figure showing Iobs, Icalc and intensity difference in a manuscript submitted for publication may not be esthetically acceptable. So one may prefer to use, say 0.05 scattering slit, and have Rp<13% or less. A good equilibrium between FWHM and Rp is essential not only to refine a structure but also to determine a structure ab initio from powder diffraction data. What interest to have minimal FWHM < 0.04 (2-theta degrees) if your data cannot be reasonably fitted. Of course, to decide for a given compromise is subjective. My own quest was for the best resolution authorizing Rp<10% (Of course, Rp may be a function of the Rietveld program used, so this may be considered as a curious combination of criteria for selection of equipment that concern the hardware and the software).

Among various tests, the best results according to these requirements corresponded to the data presented in the original posting. No need to say that several days were generally necessary to obtain one pattern with good statistics (step 0.02 to 0.01 2-theta or less; time>10sec/step).


Origin of data sent for comparison
As responses to the questions contained in the original posting, data were obtained from 3 sources: Thanks to all of them.

Finally, the results on FWHM and Rp proposed in the posting were from data measured on a Siemens D5000 with variable incident and scattering slits set to V6/V6 (6mm), Peltier cooled Si(Li) detector. Note that these measurements were made for me by Siemens with the objective to have both "low FWHM and Rp". I have not the material, my intention was to buy the best one...


FWHM estimations

Two programs were used to fit the powder diffraction data : GSAS and FULLPROF, either with or without the structure constraint.

definitions : THX=TAN(THETA) ; COX=COS(THETA)
 

- FULLPROF : All fits with pseudo-Voigt.
3 parameters refined for the width : U V W.
FWHM = ( U*THX**2 + V*THX + W )**(1/2).

- GSAS : All fits with the Thompson, Cox, Hastings formula.
Eventually 6 parameters refined: GU, GV, GW, LX, LY, GP.
FG = [( GU*THX**2 + GV*THX + GW + GP/COX**2) * 8Ln2]**(1/2)
FL = LY*THX + LX/COX
FWHM = (FG**5+2.69269*FG**4*FL+2.42843*FG**3*FL**2+4.47163*FG**2*FL**3+0.07842*FG*FL**4+FL**5)**0.2

With FG, FL and FWHM in 0.01 degrees 2-Theta being the Gaussian, Lorentzian and full widths at half maximum. FWHM/100 are given below accordingly.

Note: These FWHMs may not well reflect the reality if Rps are high.

J. Cline furnished the FWHMs obtained for LaB6 measured with the D500-PSD, fitting the lines one by one, they are listed below.

J. Kaduk gave approximate formulae:
normal PAD V : FWHM = 0.006 + 0.065/COX
narrow PAD V : FWHM = -.065 + 0.1042/COX
X3B1 : FWHM = 0.000 + 0.0626*THX

results according to these formulae are also given.


Results


A) CeO2 and m-ZrO2 data treated with FULLPROF in Fobs extracting mode:

Lachlan's comments about his data:

> While the PW1710 had the traditional fixed slits. At the time the m-ZrO2 data was collected, the X'Pert had a ADS (automatic divergence slit) installed but was emulating a fixed 1 degree slit. We dumped it later as the Sollers slits inside the ADS assembly did not have as good beam collimation as the tradition Philips Sollers.

All data with CuKa1-a2 and diffracted beam monochromator. The Siemens D501 data are for the same ZrO2 sample in conditions not optimized for the best resolution (0.15 slit), because of line broadening intrinsic to the sample.

              CeO2 (SRM 674a)            m-ZrO2
              PW1710  X'PERT     PW1710  X'PERT  Siemens D501 
                                                 0.15 Slit
U             0.00995 0.01517    0.01403 0.02193 0.03303 
V             -.01241 -.02747    0.01493 -.00537 -.02084 
W             0.01154 0.02397    0.01921 0.03311 0.05315
Rp  %         13.5    12.9       7.35    7.39    5.99   Background 
Rwp %         22.7    20.5       9.39    9.65    8.23   subtracted 
Re  %         6.88    3.44       4.08    4.51    2.51

Rp  %         12.5    11.3       5.64    5.65    4.90    Background 
Rwp %         21.8    18.9       7.54    7.74    6.94       not 
Re  %         6.60    3.18       3.28    3.61    2.11    subtracted

2-Theta       25-140  15-150     15-143  15-154  10-150 
range                                                    2-Theta 
                                 0.1481  0.1811  0.2251    18 
FWHM values                      0.1515  0.1816  0.2241    24 
in 2-Theta    0.0945  0.1330     0.1556  0.1823  0.2235    30 
degrees       0.0890  0.1202     0.1709  0.1881  0.2250    50 
              0.0879  0.1103     0.1912  0.2003  0.2340    70 
              0.0953  0.1080     0.2195  0.2229  0.2556    90 
              0.1188  0.1252     0.2630  0.2649  0.3013   110 
              0.1752  0.1866     0.3402  0.3499  0.4005   130 
                      0.3105             0.5002  0.5817   146

Note that the m-ZrO2 of the RRRR is not a very good choice for looking for low FWHM because of its mean grain size estimated to be 620 Angtroems.
 
 

B) For memory, the results posted originally to sci_tech_xtal :

Also with FULLPROF in Fobs extracting mode. D5000, variable slits V6/V6, Peltier cooled Si(Li) detector.

                  LaB6             Al2O3
               NIST SRM 660      NIST SRM 1976
U                0.00310          0.00444 
V                0.00112          0.00149 
W                0.00112          0.00156
Rp               8.9%             8.1%  (background subtracted, 
Rwp              12.6             10.7       peak only) 
Re               7.0              6.8
Rp               7.9%             5.9%  (background not 
Rwp              11.5             8.5     subtracted, peak only) 
Re               6.4              5.4
2-Theta range    18-148           24-147.5 
                                            2-Theta Degrees 
FWHM values :    0.0381           0.0452       24 
in 2-Theta       0.0405           0.0480       30 
degrees          0.0481           0.0569       50 
                 0.0585           0.0693       70 
                 0.0731           0.0867       90 
                 0.0951           0.1130      110 
                 0.1333           0.1589      130 
                 0.1948           0.2327      146


No formula combining FWHMs and Rp was given, so that a winner having both "low FWHM and Rp" could be hard to designate.

If the definition is now "lowest FWHMs first and then Rp at least <10%", the winner is LaB6 cited just above, up to now.
 

C) from posts of Cline and Kaduk using GSAS on LaB6 SRM 660 :

All refinements with the structure constraint, Cu Kalpha with the exception of X3B1.

Comments of J. Cline about his data:

> 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 and 8mm wide (.83 degrees divergence), the receiving 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 receiving slit, and a radius of 217 mm. I used GSAS to obtain the following:
..........
> From the D5000 with the inc. slit set at "small" (6mm widrh), the data were then corrected with a 1/sin(theta) function:
..........
> 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 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 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.
.......
 

               LaB6 RESULTS from Cline and Kaduk with GSAS: 
               -------------------------------------------- 

                     Cline                      Kaduk

              D500   D5000   D5000     Scintag  PAD V   X3B1 
              PSD   .85 slit   V6     .05 slit  normal  Brookhaven
GU            4.780   1.110   2.815     0.000   0.000   7.479 
GV           -8.727  -8.704  -7.244     0.000   0.000   -.676 
GW            8.342   3.477   6.360     2.877   7.432   0.000 
LX            1.461   3.744   2.625     0.000   0.000   0.375 
LY            1.351   .5854   2.136     4.454   5.098   0.000 
trns         -3.195  -2.081  -.6622 
asym          0.000   0.000   0.000     0.000   0.749   1.171 
shft          7.088   8.151   6.689 
GP            0.000   2.846   0.000 
stec          0.000   0.000   0.000 
ptec          0.000   0.000   0.000 
sfec          0.000   0.000   0.000

Rp  %         2.48    12.38   11.52             not given 
Rwp %         5.16    16.93   16.38     37.50   13.02   17.62 
           (R determined without background subtraction in all cases)
Special       very    fixed  variable   very            peak 
Comment       high    slit    slit      weak            splitting 
           background                   peaks
                                                                2-Theta 
FWHM values   0.0708  0.0749  0.0709    0.0451  0.0700  0.0125  24 
in 2-Theta    0.0697  0.0735  0.0702    0.0465  0.0716  0.0162  30 
degrees       0.0669  0.0698  0.0688    0.0520  0.0775  0.0292  50 
              0.0663  0.0683  0.0693    0.0590  0.0850  0.0445  70 
              0.0699  0.0716  0.0737    0.0687  0.0954  0.0643  90 
              0.0833  0.0868  0.0879    0.0836  0.1114  0.0925 110 
              0.1191  0.1278  0.1254    0.1107  0.1402         130 
              0.1878  0.2045  0.2284    0.1564  0.1896         146


FULLPROF results on D500-PSD, fit in Fobs extracting mode : 
  U=0.00532 V=-0.00621 W=0.00611 
  Rp=11.0 Rwp=10.7 Re=3.36 background subtracted, peak only 
  Rp=3.36 Rwp=5.18 Re=1.63 background not subtracted, peak only
                           2-Theta Degrees 
FWHM values :    0.0709    24 
in 2-Theta       0.0695    30 
degrees          0.0661    50 
                 0.0661    70 
                 0.0722    90 
                 0.0900   110 
                 0.1314   130 
                 0.2067   146

FWHM values obtained from individual profile refinements on the D500-PSD 
LaB6 data by J. Cline (with split Pearson 7, via Diffrac AT): 
------------------------------------------------------------- 
         2-Theta   FWHM
         21.3046   0.077 
         30.341    0.065 
         37.4027   0.063 
         43.4692   0.06 
         48.9189   0.06 
         53.9535   0.06 
         63.1821   0.061 
         67.5124   0.062 
         71.7101   0.062 
         75.8138   0.064 
         79.8405   0.067 
         83.8193   0.067 
         87.7679   0.07 
         95.6475   0.072 
         99.6208   0.075 
        103.6339   0.078 
        107.7257   0.083 
        111.9135   0.087 
        116.2258   0.089 
        120.705    0.098 
        130.393    0.112 
        135.782    0.123 
        141.7594   0.138 
        148.6659   0.162

From the approximate equations given by J. Kaduk : 
--------------------------------------------------
              Scintag PAD          X3B1 
            .05 slit  normal     Brookhaven 
                                           2-Theta 
FWHM values  0.0415   0.0724      0.0133    24 
in 2-Theta   0.0428   0.0733      0.0168    30 
degrees      0.0500   0.0777      0.0292    50 
             0.0622   0.0854      0.0438    70 
             0.0824   0.0979      0.0626    90 
             0.1167   0.1193      0.0894   110 
             0.1816   0.1598               130 
             0.2913   0.2283               146

To compare results with and without the structure constraint may be difficult. The Rp and Rwp values may decrease 2-3% in the Fobs extracting mode. Also, the conventional Rietveld Rp and Rwp (background subtracted) are generally much higher than the R calculated without subtracting the background (the more the bacground is high, the more the non-conventional Rp or Rwp are lower than the conventional ones).

The case of Al2O3 SRM 1976 poses some problems for Rietveld refinement with structure, because of a strong preferred orientation (although there is no problem in "Fobs" extracting mode). This may be the reason why nobody has shown any data on this sample up to now.


Conclusions

The questions were :

> 1- One can hear sometimes that variable slits lead to irreproducible measurements, is that true ?

About the suspicion of irreproducibility, there are some comments in the well known book "Modern Powder Diffraction", Chapter 3 from Ron Jenkins (Reviews in Mineralogy, Vol. 20, Bish & Post, editors, 1989) :

"Significant intensity differences are apparent among users of variable divergence slit diffractometers. This problem can likely be corrected by proper alignement of the divergence slit."
 

One response was obtained from Bob Snyder: X-rayX-press<SNYDER@xray.alfred.edu>

> Just a few words on variable slits and Rietveld refinement: Using a 1 degree incident slit will cause the incident beam to splash over the ends of most sample holders at two theta angles somewhat less than twenty degrees (for example it is 17 degrees in the old vertical Philips diffractometers). The conventional intensity equation for a diffractometer assumes a constant volume of sample is illuminated and this is the equation implemented in all of the Rietveld codes with which I am familiar. Therefore peaks below the splash over angle will have less incident beam on them and will fail the constant volume of irradiation assumption resulting in their observed intensities being systematically too low. To use these peaks in a Rietveld refinement the pattern must be run with a narrower incident slit and/or a longer sample holder. If, on the other hand, you wish to use a variable incident slit to ensure that the beam never splashes over the sides of your sample holder, then you must modify the intensity equation in the Rietveld program to allow for the now variable volume of diffracting material in the beam as a function of 2-theta. Since it is only relative intensities we care about you can use a very simple routine like the one that Cam Hubbard and I used in the AUTO/NBS*QUANT84 system years ago (included below). Or, you can put in the more formally derived equation of the effect of variable receiving slit on intensity.
............
Followed the SUBROUTINE TCSLIT
...........

> 2- Has anybody better results with an in-laboratory apparatus (eventually Kalpha1) on both aspects (FWHM and Rp) for these well known standards ?

From the FWHM and Rp listed above, there was no better results on both aspects than those I gave.

The winner according to FWHMs is undoubtedly x3b1 (synchrotron) but is not an in-laboratory apparatus and has Rwp quite high. Moreover, near 110 2-theta degrees, the FWHMs are larger than with a lab instrument.

LaB6 best results

For in-lab results, none of the FWHM curves as a funtion of two-theta is the best in the whole 2-theta domain. Concerning the samples, the best results are for LaB6 (other recommended compound with similar quality, provided it has been correctly annealed is BaF2, according to D. Louer).

Selecting the three possible winners :

                Posting     Cline     Kaduk 
                Siemens    Siemens    Scintag 
                 V6/V6     D500 PSD   .05 slit 
                                              2-Theta Degrees 
FWHM values :   0.0381     0.0708     0.0451     24 
in 2-Theta      0.0405     0.0697     0.0465     30 
degrees         0.0481     0.0669     0.0520     50 
                0.0585     0.0663     0.0590     70 
                0.0731     0.0699     0.0687     90 
                0.0951     0.0833     0.0836    110 
                0.1333     0.1191     0.1107    130 
                0.1948     0.1878     0.1564    146

Rp              8.9%       11.0*            (background subtracted, 
Rwp             12.6       10.7*                 peak only) 
Re              7.0        3.36*
Rp              7.9%       2.48              (background not 
Rwp             11.5       5.16       37.5       subtracted, peak only) 
Re              6.4        1.63*
*: with FULLPROF

Improvements due to variable slits are mainly at low angle where this can be interesting for indexation purpose. For structure determination and refinement, FWHMs at large angles that would be lower than those shown here would be of the highest interest. The Scintag results on the point of view of FWHM are fine, they are the best at large angle, however, Rwp is really too high (may be the reason is not only difficulties in profile fitting, very poor statistics may have also such a consequence).

Al2O3 SRM 1976 best results

Results obtained with a Philips X-Pert MPD equipped also with variable slits (presented below and compared with Siemens) were almost exactly identical with those given in the original posting for Al2O3, SRM 1976 (on both FWHM and Rp):

    Results from fits without structure constraint (Le Bail method) by FULLPROF,
Full data, base of peak=9*FWHM, 11 parameters refined (background estimated manually)
             Measurements on two different Al2O3 samples 
                            NIST SRM 1976
                     Philips           Siemens 
                    X-Pert MPD          D5000 
                          Variable slits
  a                  4.75879(1)       4.75883(2) 
  c                 12.99245(2)      12.99301(4) 
 zero                0.0215(2)        0.0051(4) 
TRANSP               -.0239(4)        -.0076(7) 
Asym1                0.185(5)         -.069(9) 
Asym2                0.028(1)         0.007(1) 
X                    0.0007(1)        0.0009(2) 
Eta(p-V)             0.621(9)         0.68(1) 
U                    0.0046(1)        0.0044(1) 
V                    0.0020(1)        0.0015(2) 
W                    0.00096(5)       0.00157(8)
Rp                   8.0%             8.1%     (background subtracted, 
Rwp                  10.8             10.7           peak only) 
Re                   6.8              6.8
Rp                   6.4%             5.9%     (background not 
Rwp                  9.3              8.5        subtracted, peak only) 
Re                   5.8              5.4

2-Theta range        24-147.5         24-147.5 
step                 0.01             0.02
                (measured:0.005) 
               (mean of 2 points)
                                                2-Theta Degrees 
FWHM values :        0.0400           0.0452       24 
in 2-Theta           0.0427           0.0480       30 
degrees              0.0537           0.0569       50 
                     0.0679           0.0693       70 
                     0.0869           0.0867       90 
                     0.1148           0.1130      110 
                     0.1625           0.1589      130 
                     0.2382           0.2327      146 

Rigaku on Quartz

Finally, Rigaku (which has no variable slits to propose to date) presented data on quartz with low FWHMs (incident slit 0.05mm, scattering slit 0.05mm, receiving slit 0.15mm with CuKalpha1+2-60kV-300mA, goniometer radius 285mm). Without the complete pattern, a Rietveld fit was not possible, so the results are from separated fits on independent reflexions (as given by Rigaku):

             hkl    2-theta     FWHM
             100     20.84     0.0450 
             101     26.83     0.0571 
             110     36.53     0.0574 
             112     50.12     0.0688 
             211     59.93     0.0805 
             302     75.64     0.0826 
             205    106.57     0.1076 
             412    122.57     0.1433 
             206    136.39     0.1813 
             413    137.85     0.1986

So, the comments made by Lachlan Cranswick in an oldest posting : "all diffractometers are approximately the same so you can concentrate on going for the best price" may be true at least for Siemens/Philips latest XRD systems. However, not a lot of data coming from these variable slits new systems have been published up to now. Since the money to buy a new system is reported now to 1996-97 in Le Mans, I have some time for more reflexion... This seems to be also the Lachlan's position since he writes to me:

> I am not very trusting of computer controlled slits at the moment. How do you know that when you choose 1 degree it is one degree? And if/when the slits go out of whack - how much dodgy data will be collected before you find out? Until at least a few years in the future - I think I will be sticking to fixed slits as these are inherently reliable.

Manufacturers seem confident in the quality of their apparatus with variable slits (both Siemens and Philips). Yes the patterns may be well fitted with low FWHM and Rp at least with a Rietveld-type program in Fobs extracting mode. It is interesting now to compare refinements with structure constraint performed on LaB6 with and without variable slits :

          fixed slits                     variable slits 
        ------------------                 ----------------- 
        D500         D5000                  D5000       D5000 
        PSD          .85 slit               V6          V6/V6 
        -----------------------GSAS-----------------    FULLPROF 
B/U(La) 0.23(1)      0.30(1)                0.14(1)     0.202(9) 
z(B)    0.1896(7)    0.1906(8)              0.1939(8)   0.1911(8) 
B/U(B)  1.00(8)      1.10(8)                0.68(7)     0.34(6) 
a       4.15695(1)   4.15647(1)             4.15646(1)  4.15666(1) 
        (100*Uiso)   (100*Uiso)             (100*Uiso)  (B) 
RB(%)          not given                    not given   4.89 
RF(%)          not given                    not given   2.93

Large differences on B/U thermal parameters are detected. This is also the case when comparing different measurements made with fixed slits. The fact is that a satisfying fit on the point of view of the Bragg R-factor is obtained in the case of variable slits systems.

The fit with structure constraint is not possible with SRM 1976 (Al2O3). However, the "|Fobs|" extracted from the Siemens and Philips independent measurements were very similar (also the FWHM and Rp). So this allows to conclude that SRM 1976, in spite of strong preferred orientation, is a good standard, and that good alignment of the slits has been achieved by both manufacturers. May be variable slits will be the system for the future of powder diffractionists, after all.


Two years later

The project to buy a new powder diffractometer is still a project...

New Cu-K-alpha-1 data from Philips (X'Pert) on Al2O3 NIST SRM 1976
(more information could be obtained on demand).

                           Pattern 1         Pattern 2
DivergenceSlit,            Fixed,1,          Automatic, 7,
ReceivingSlit,                        0.05,
MonochromatorUsed,                    YES ,
GeneratorVoltage,                      37,
TubeCurrent,                           55,
FileDateTime,           27-apr-1996 7:44,   1-may-1996 6:45,
DataAngleRange,                 20.0000, 157.0000,
ScanStepSize,                         0.010,
ScanType,                             STEP,
ScanStepTime,                         10.00,
Results from fits without structure constraint (Le Bail method) by FULLPROF,
Full data, base of peak=5*FWHM, 17 parameters refined (6 for background):
              Pattern 1        Pattern 2
  a            4.75868(2)       4.75883(1)
  c           12.99240(4)      12.99283(4)
 zero         -0.0200(8)       -0.0046(8)
TRANSP        -0.0084(8)       -0.0228(8)
Asym1          0.234(8)         0.264(9)
Asym2          0.057(2)         0.056(3)
X              0.016(2)         0.0015(1)
Eta(p-V)       0.087(11)        0.097(12)
U              0.0058(2)        0.0048(1)
V             -0.0015(3)        0.0013(2)
W              0.028(1)         0.00144(9)

Rp             10.8             11.0  (background subtracted,
Rwp            15.0             14.4       peak only)
Re             11.3             10.8

Rp             8.9              8.5   (background not
Rwp            13.3             12.2    subtracted, peak only)
Re             10.0             9.2
2-Theta range         22-157
step                  0.01

                                       2-Theta Degrees
FWHM values :  0.0524           0.0440     24
in 2-Theta     0.0531           0.0462     30
degrees        0.0580           0.0556     50
               0.0678           0.0686     70
               0.0843           0.0868     90
               0.1112           0.1144    110
               0.1620           0.1622    130
               0.1956           0.1934    138
               0.2448           0.2388    146
               0.3242           0.3117    154
These results remain similar with those obtained two years before on the point of view of the FWHMs. However, this time we cope with Alpha-1 instead of Alpha-1+2. The reliability factors appear a bit higher by nearly 2%. In fact, the counting time was slightly too small (10sec/point) for ensuring sufficient statistics. In order to demonstrate this, intensities were summed, two points giving one and the fit was repeated this time with a 0.02 step. Reliabilities were considerably improved without large modification of the fit parameters :
              Pattern 1        Pattern 2
  a            4.75876(2)       4.75882(2)
  c           12.99292(5)      12.99307(4)
 zero         -0.0481(11)      -0.0356(9)
TRANSP         0.0162(9)       -0.0001(8)
Asym1         -0.045(17)       -0.123(13)
Asym2          0.033(4)         0.032(4)
X              0.020(2)         0.0022(2)
Eta(p-V)       0.022(11)        0.0009(112)
U              0.0054(2)        0.0047(1)
V             -0.0010(3)        0.0010(2)
W              0.029(1)         0.00204(9)

Rp             8.3              8.0   (background subtracted,
Rwp            11.4             10.6       peak only)
Re             8.1              7.8

Rp             6.8              6.1   (background not
Rwp            10.0             9.0     subtracted, peak only)
Re             7.1              6.5
2-Theta range         22-157
step                  0.02 (two points 0.01 apart summed)

                                       2-Theta Degrees
FWHM values :  0.0542           0.0495     24
in 2-Theta     0.0550           0.0514     30
degrees        0.0601           0.0594     50
               0.0696           0.0710     70
               0.0854           0.0880     90
               0.1117           0.1143    110
               0.1600           0.1606    130
               0.1922           0.1912    138
               0.2396           0.2358    146
               0.3160           0.3075    154
It is not excluded that you could obtain a better fit with another Rietveld program or even with FULLPROF by selecting other profile shapes and so on. Have a look at the fits as GIF files for Pattern 1 (parts 20-80 and 80-157) and Pattern 2 (parts 20-80 and 80-157).

Try by yourself : download Pattern 1 and Pattern 2.

Note that if you use DMPLOT (shareware), you should contact the author (Henryk Marciniak) for obtaining the program version allowing to plot more than 9600 points...

Conclusions : if your synchrotron can't do really better, stay at home !

However, staying at home you may have fluorescence problems with this configuration (monochromator in the incident beam). Joke apart, the definite synchrotron advantage remains the possibility to tune the wavelenght and to obtain really low FWHMs (Full Width at Half Maximum in 2-Theta degrees <0.01 recently at ESRF). The time needed, if calculated on the basis of 20 sec/point would have been 76 hours for the 13700 points of pattern 1. This remains manageable and may be compared to not so low counting times needed at some synchrotron sources. For low symmetry samples, I had been unable to obtain good patterns at Daresbury on station 2.3 by counting less than 12 to 24 hours (3 shifts) for 7000 points.

New data coming soon, maybe next month.


Data you may download



Four years later

The diffractometer, whatever it is (D8 - Bruker), is there.

Fortunately, the contract was mentioning that a test on NIST 1976 should be presented before and after delivering, and should be as satisfying as above (background subtracted Rp lower than 10% - and why not near of 8% -, minimal FWHM lower than 0.045 2-theta degrees)...

I would recommend you to do the same ! Do not buy any diffractometer without those tests.

NIST-1976
If you are insterested in the test pattern, you may download it here (it is a zipped .uxd file). Recording conditions are : Bragg Brentano geometry, Cu-Kalpha, graphite monochromator in the diffracted beam, Soller 2°, divergence slit 0.3°, antiscatter slit 0.3°, Soller 2°, receiving slit 0.1mm. And the summary of the Le Bail fit (without structure) by using the FULLPROF program is below :

 

        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 12/01/1999  Time: 10:07:42.560                              

 => Title:Al2O3                                                                 

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*    8.00
 ==> Angular range, step and number of points:
     2Thmin:  20.0000  2Thmax: 140.0000  Step:   0.0100  No. of points:  12001
 => Pattern Matching (fixed scale) for phase: 1
 => Scor: 1.5788

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   14


------------------------------------------------------------------------------
 => Phase No.  1 Al2O3                                   R -3 C              
------------------------------------------------------------------------------

 =>  No. of reflections:    115


 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                              4.75881   0.00002
                              4.75881   0.00002
                             12.99295   0.00005
                             90.00000   0.00000
                             90.00000   0.00000
                            120.00000   0.00000
 
 => overall scale factor :    0.000000100   0.000000000
 => Eta(p-v) or m(p-vii) :    0.39955   0.00988
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00458   0.00017
                              0.00037   0.00022
                              0.00171   0.00007
 => Preferred orientation:    0.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.02931   0.01305
                              0.03607   0.00158
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00190   0.00014
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00407   0.00025
                              0.00255   0.00034
                              0.00100   0.00010



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0054  0.0009
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :  -0.0077  0.0006

 ==> RELIABILITY FACTORS WITH ALL NON-EXCLUDED POINTS:

 => Cycle: 30 => MaxCycle: 30
 => N-P+C: 11987
 => Rp: 7.33     Rwp: 11.1     Rexp:    9.88 Chi2:  1.27      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 10.5     Rwp: 13.8     Rexp:   12.24 Chi2:  1.27    
 => Deviance: 0.154E+05     Dev*  :  1.278    
 => DW-Stat.:    1.6624     DW-exp:     1.9458
 => N-sigma of the GoF:   21.204

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  6698
 => Rp: 6.08     Rwp: 9.49     Rexp:    8.09 Chi2:  1.38      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 7.34     Rwp: 10.8     Rexp:    9.19 Chi2:  1.38    
 => Deviance: 0.914E+04     Dev*  :  1.358    
 => DW-Stat.:    2.7548     DW-exp:     1.9285
 => N-sigma of the GoF:       21.747
 => Phase:  1
 => Bragg R-factor:  2.65       Vol:  254.821( 0.002)  Fract(%):    0.00( 0.00)
 => Rf-factor= 1.58             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 12/01/1999  Time: 10:08:01.730                              

NAC
 One may also be interested in a true Rietveld fit, with structure. Here is one : the well known NAC used as standard for calibrating INEL curve detectors (Na2Ca3Al2F14 : G. Courbion & G. Ferey, J. Solid State Chem. 76 (1988) 426-431).
The pattern shows minimum FWHM = 0.05 (2-theta degrees). Note that a 1.5mm thick sample gave 0.08 (2-theta degrees) as minimal FWHM, so that the thickness is 0.5mm in the present case, and the sample was packed and rotating. The following reliability factors are produced either with fixed slits or variable slits :
                               %         RP        RWP        RB
Fixed slits
Le Bail method                          10.1      11.9      
Rietveld with structure constrained :   10.6      12.4       5.4
Rietveld with structure refined     :   10.0      12.0       4.5
Variable slits V6/V6                    
Le Bail method                          9.28      11.1            uncorrected by 1/sin(theta)
Rietveld with structure constrained :   10.1      12.8       5.5  corrected by 1/sin(theta)
Rietveld with structure refined     :   8.75      11.1       4.2  corrected by 1/sin(theta)
From these results, it seems that no preferred orientation occurs with NAC. Seems that the thermal parameters, as estimated by the Rietveld method from these Bruker D8 data, are quite similar to those produced by the previous single crystal study. The slightly high RP and RWP values are probably due to the presence of a few percent of impurities, the step too large (0.02) regarding the minimal FWHM (0.05) and the statistics slightly too low, needing to increase the counting time (was 3 days already !).

It seems also that the correction in 1/sin(theta) for variable slits works in a satisfying way, at least for this pattern (however, the F3 B factor is a bit small when the structure is refined - anyway the F3 weight is small due to the special position occupied). Of course it would be better to use the raw data and to fit by calculating a pattern multiplied by sin(theta) rather than to use a modified observed pattern divided by sin(theta). At least one Rietveld software can do that : WinMProf (the example is exactly the pattern shown here). The interest is to obtain more counts at large angles, improving statistics (but less at small angle...).

If you are interested in the test patterns, you may download here (zipped .uxd file) the fixed slit pattern. Recording conditions are : Bragg Brentano geometry, Cu-Kalpha, graphite monochromator in the diffracted beam, Soller 2°, divergence slit 0.5°, antiscatter slit 0.5°, Soller 2°, receiving slit 0.1mm.

The variable slits patterns are there (nac-test-5.uxd = uncorrected data ; nac5stru3.dat = corrected data).

And the summaries of the 6 above fits by using the FULLPROF program are below :
 

Le Bail method :
 FIXED SLITS

        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 12/01/1999  Time: 12:01:44.410                              

 => PCR file code: NACtest2
 => DAT file code: NACtest2
 => Title:   Na2Ca3Al2F14      NAC-test-2 sur D8 fentes 0.5 0.5 0.1             

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*    8.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 140.0000  Step:   0.0200  No. of points:   6501
 => Pattern Matching (fixed scale) for phase: 1
 => Scor: 1.8653

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   19


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    400


 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25197   0.00005
                             10.25197   0.00005
                             10.25197   0.00005
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.010000000   0.000000000
 => Eta(p-v) or m(p-vii) :    0.54614   0.01238
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00516   0.00027
                             -0.00264   0.00030
                              0.00298   0.00008
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.07488   0.00526
                              0.04089   0.00097
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00277   0.00024
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00497   0.00044
                             -0.00141   0.00049
                              0.00276   0.00013



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0234  0.0007
 => Background Polynomial Parameters ==>
                              96.7082    0.523397    
                             -43.0679     2.25469    
                              226.432     10.3165    
                              306.700     14.4346    
                             -692.570     38.0027    
                             -872.613     39.6150    
 
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :   0.0137  0.0009

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4373
 => Rp: 6.59     Rwp: 8.86     Rexp:    5.70 Chi2:  2.41      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 10.1     Rwp: 11.9     Rexp:    7.63 Chi2:  2.41    
 => Deviance: 0.109E+05     Dev*  :  2.489    
 => DW-Stat.:    1.3072     DW-exp:     1.9150
 => N-sigma of the GoF:       65.961
 => Phase:  1
 => Bragg R-factor:  3.59       Vol: 1077.510( 0.009)  Fract(%):    0.00( 0.00)
 => Rf-factor= 2.39             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 12/01/1999  Time: 12:02:07.370


 VARIABLE SLITS

        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 17/05/1999  Time: 14:34:21.150                              

 => PCR file code: NACtest5
 => DAT file code: NACtest5
 => Title:   Na2Ca3Al2F14      NAC-test-5 sur D8 fentes V6-V6                   

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*    8.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 140.0000  Step:   0.0200  No. of points:   6501
 => Pattern Matching (fixed scale) for phase: 1
 => Scor: 1.8857

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   19


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    408


 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25308   0.00004
                             10.25308   0.00004
                             10.25308   0.00004
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.010000000   0.000000000
 => Eta(p-v) or m(p-vii) :    0.51346   0.01321
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00181   0.00017
                              0.00195   0.00022
                              0.00146   0.00007
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.05549   0.00904
                              0.03697   0.00118
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00405   0.00021
                              0.00000   0.00000
 => Strain parameters    :    0.00000   0.00000
                              0.00000   0.00000
                              0.00000   0.00000
 => Size   parameters    :    0.00000   0.00000
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00219   0.00030
                              0.00286   0.00039
                              0.00144   0.00011



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0675  0.0009
 => Background Polynomial Parameters ==>
                              116.591    0.575055    
                              7.36134     2.31746    
                              189.861     10.0476    
                              356.360     13.0238    
                             -493.454     32.8539    
                             -613.677     32.6398    
 
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :   0.0011  0.0008

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4384
 => Rp: 6.04     Rwp: 8.18     Rexp:    5.31 Chi2:  2.38      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 9.28     Rwp: 11.1     Rexp:    7.17 Chi2:  2.38    
 => Deviance: 0.107E+05     Dev*  :  2.438    
 => DW-Stat.:    1.3091     DW-exp:     1.9151
 => N-sigma of the GoF:       64.462
 => Phase:  1
 => Bragg R-factor:  3.38       Vol: 1077.862( 0.007)  Fract(%):    0.00( 0.00)
 => Rf-factor= 2.04             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 17/05/1999  Time: 14:35:03.600


Rietveld with structure constrained :

 FIXED SLITS

 

        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 12/01/1999  Time: 12:02:50.480                              

 => PCR file code: NACstru2
 => DAT file code: NACstru2
 => Title:   Na2Ca3Al2F14      NAC-test-2 sur D8 fentes 0.5 0.5 0.1             

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*    8.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 140.0000  Step:   0.0200  No. of points:   6501
 => Crystal Structure Refinement for phase: 1
 => Scor: 2.0467

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   20


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    408


 ==> ATOM PARAMETERS:

  Name    x     sx       y     sy       z      sz      B   sB  occ.  socc.

 CA    0.46670(   0)  0.00000(   0)  0.25000(   0)  0.660(  0)  0.120(  0)
 AL    0.24820(   0)  0.24820(   0)  0.24820(   0)  0.600(  0)  0.080(  0)
 NA    0.08470(   0)  0.08470(   0)  0.08470(   0)  2.160(  0)  0.080(  0)
 F1    0.13870(   0)  0.30620(   0)  0.12060(   0)  0.970(  0)  0.240(  0)
 F2    0.36400(   0)  0.36270(   0)  0.18730(   0)  1.140(  0)  0.240(  0)
 F3    0.46140(   0)  0.46140(   0)  0.46140(   0)  0.820(  0)  0.080(  0)

 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25199   0.00005
                             10.25199   0.00005
                             10.25199   0.00005
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.003888535   0.000008927
 => Eta(p-v) or m(p-vii) :    0.52535   0.01303
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00451   0.00028
                             -0.00197   0.00030
                              0.00281   0.00008
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.07579   0.00587
                              0.04019   0.00101
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00377   0.00026
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00429   0.00045
                             -0.00059   0.00049
                              0.00253   0.00013



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0238  0.0008
 => Background Polynomial Parameters ==>
                              95.6832    0.552610    
                             -41.8855     2.35543    
                              175.038     10.7676    
                              275.914     15.0733    
                             -466.592     39.6650    
                             -653.621     41.3530    
 
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :   0.0135  0.0009

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4365
 => Rp: 6.94     Rwp: 9.27     Rexp:    5.70 Chi2:  2.64      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 10.6     Rwp: 12.4     Rexp:    7.61 Chi2:  2.64    
 => Deviance: 0.119E+05     Dev*  :  2.734    
 => DW-Stat.:    1.1910     DW-exp:     1.9154
 => N-sigma of the GoF:       76.783
 => Phase:  1
 => Bragg R-factor:  5.35       Vol: 1077.517( 0.010)  Fract(%):    0.00( 0.00)
 => Rf-factor= 4.02             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 12/01/1999  Time: 12:03:15.690
 VARIABLE SLITS
        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 17/05/1999  Time: 15:53:09.410                              

 => PCR file code: NAC5stru
 => DAT file code: NAC5stru
 => Title:   Na2Ca3Al2F14  NAC-test-5 sur D8 V6-V6 corrected 1/sin(theta)       

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*   10.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 140.0000  Step:   0.0200  No. of points:   6501
 => Crystal Structure Refinement for phase: 1
 => Scor: 2.3955

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   20


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    408


 ==> ATOM PARAMETERS:

  Name    x     sx       y     sy       z      sz      B   sB  occ.  socc.

 CA    0.46670(   0)  0.00000(   0)  0.25000(   0)  0.660(  0)  0.120(  0)
 AL    0.24820(   0)  0.24820(   0)  0.24820(   0)  0.600(  0)  0.080(  0)
 NA    0.08470(   0)  0.08470(   0)  0.08470(   0)  2.160(  0)  0.080(  0)
 F1    0.13870(   0)  0.30620(   0)  0.12060(   0)  0.970(  0)  0.240(  0)
 F2    0.36400(   0)  0.36270(   0)  0.18730(   0)  1.140(  0)  0.240(  0)
 F3    0.46140(   0)  0.46140(   0)  0.46140(   0)  0.820(  0)  0.080(  0)

 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25336   0.00005
                             10.25336   0.00005
                             10.25336   0.00005
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.010537170   0.000027404
 => Eta(p-v) or m(p-vii) :    0.60831   0.01411
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00369   0.00025
                             -0.00038   0.00025
                              0.00195   0.00007
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.15396   0.00337
                              0.03526   0.00098
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00331   0.00027
                              0.00000   0.00000
 => Strain parameters    :    0.00000   0.00000
                              0.00000   0.00000
                              0.00000   0.00000
 => Size   parameters    :    0.00000   0.00000
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00321   0.00042
                              0.00196   0.00043
                              0.00150   0.00010



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0785  0.0004
 => Background Polynomial Parameters ==>
                              159.758     1.24920    
                             -161.811     5.48362    
                              421.106     22.3883    
                              654.290     37.4726    
                             -868.236     72.8059    
                             -1396.62     87.0274    
 
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :  -0.0056  0.0007


 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4690
 => Rp: 7.29     Rwp: 10.2     Rexp:    3.76 Chi2:  7.37      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 10.1     Rwp: 12.8     Rexp:    4.72 Chi2:  7.37    
 => Deviance: 0.334E+05     Dev*  :  7.123    
 => DW-Stat.:    1.0588     DW-exp:     1.9181
 => N-sigma of the GoF:      308.408
 => Phase:  1
 => Bragg R-factor:  5.48       Vol: 1077.951( 0.009)  Fract(%):    0.00( 0.00)
 => Rf-factor= 3.58             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 17/05/1999  Time: 15:54:08.840


Rietveld with structure refined :

 FIXED SLITS

        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 12/01/1999  Time: 14:41:44.450                              

 => PCR file code: NACstru3
 => DAT file code: NACstru3
 => Title:   Na2Ca3Al2F14      NAC-test-2 sur D8 fentes 0.5 0.5 0.1             

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*    8.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 136.8000  Step:   0.0200  No. of points:   6341
 => Crystal Structure Refinement for phase: 1
 => Scor: 1.9465

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   36


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    400


 ==> ATOM PARAMETERS:

  Name    x     sx       y     sy       z      sz      B   sB  occ.  socc.

 CA    0.46574(  13)  0.00000(   0)  0.25000(   0)  0.747( 25)  0.120(  0)
 AL    0.24841(  25)  0.24841(  25)  0.24841(  25)  0.799( 39)  0.080(  0)
 NA    0.08448(  21)  0.08448(  21)  0.08448(  21)  2.069(118)  0.080(  0)
 F1    0.13911(  27)  0.30639(  25)  0.12159(  37)  0.884( 73)  0.240(  0)
 F2    0.36297(  37)  0.36250(  25)  0.18818(  27)  1.344( 81)  0.240(  0)
 F3    0.46175(  19)  0.46175(  19)  0.46175(  19)  0.608(113)  0.080(  0)

 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25201   0.00005
                             10.25201   0.00005
                             10.25201   0.00005
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.003919237   0.000014435
 => Eta(p-v) or m(p-vii) :    0.55900   0.01349
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00437   0.00027
                             -0.00173   0.00030
                              0.00273   0.00008
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.07963   0.00568
                              0.04008   0.00097
 => X and y parameters   :    0.00307   0.00028
                              0.00000   0.00000

 => U,V,W for Lambda(2)  : 
                              0.00411   0.00044
                             -0.00027   0.00048
                              0.00243   0.00012



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0244  0.0008
 => Background Polynomial Parameters ==>
                              96.4012    0.578133    
                             -40.2696     2.34532    
                              182.282     10.6013    
                              284.332     14.7851    
                             -493.580     39.1065    
                             -685.849     40.6327    
 
 => Sin(2theta)-shift parameter :   0.0131  0.0009


 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4349
 => Rp: 6.51     Rwp: 8.97     Rexp:    5.69 Chi2:  2.48      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 9.97     Rwp: 12.0     Rexp:    7.60 Chi2:  2.48    
 => Deviance: 0.112E+05     Dev*  :  2.572    
 => DW-Stat.:    1.2606     DW-exp:     1.9228
 => N-sigma of the GoF:       69.171
 => Phase:  1
 => Bragg R-factor:  4.45       Vol: 1077.523( 0.009)  Fract(%):    0.00( 0.00)
 => Rf-factor= 3.98             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 12/01/1999  Time: 14:42:24.490
 VARIABLE SLITS
        ********************************************************
        ** PROGRAM FULLPROF.98 (Version 0.2  - Mar98-LLB JRC) **
        ********************************************************
           Rietveld, Profile Matching & Integrated Intensity
                Refinement of X-ray and/or Neutron Data


    Date: 17/05/1999  Time: 15:57:24.980                              

 => PCR file code: NAC5stru3
 => DAT file code: NAC5stru3
 => Title:   Na2Ca3Al2F14  NAC-test-5 sur D8 V6-V6 corrected 1/sin(theta)       

 ==> CONDITIONS OF THIS RUN:

 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected
 => Data supplied in free format 
 => Wavelengths:  1.54056 1.54439
 => Cos(Monochromator angle)=   0.7998
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*   10.00
 ==> Angular range, step and number of points:
     2Thmin:  10.0000  2Thmax: 140.0000  Step:   0.0200  No. of points:   6501
 => Crystal Structure Refinement for phase: 1
 => Scor: 2.1577

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   36


------------------------------------------------------------------------------
 => Phase No.  1 Na2Ca3Al2F14                            I 21 3              
------------------------------------------------------------------------------

 =>  No. of reflections:    408


 ==> ATOM PARAMETERS:

  Name    x     sx       y     sy       z      sz      B   sB  occ.  socc.

 CA    0.46570(  11)  0.00000(   0)  0.25000(   0)  0.635( 21)  0.120(  0)
 AL    0.24735(  21)  0.24735(  21)  0.24735(  21)  0.623( 34)  0.080(  0)
 NA    0.08451(  20)  0.08451(  20)  0.08451(  20)  2.125(105)  0.080(  0)
 F1    0.13627(  25)  0.30508(  23)  0.12097(  33)  0.940( 65)  0.240(  0)
 F2    0.36309(  32)  0.36031(  22)  0.18728(  24)  1.060( 68)  0.240(  0)
 F3    0.46193(  17)  0.46193(  17)  0.46193(  17)  0.139( 90)  0.080(  0)

 ==> PROFILE PARAMETERS:

 => Cell parameters      :
                             10.25335   0.00004
                             10.25335   0.00004
                             10.25335   0.00004
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.010484128   0.000037432
 => Eta(p-v) or m(p-vii) :    0.52325   0.01215
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00266   0.00022
                              0.00085   0.00022
                              0.00165   0.00006
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.09947   0.00569
                              0.03949   0.00092
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00460   0.00024
                              0.00000   0.00000
 => Strain parameters    :    0.00000   0.00000
                              0.00000   0.00000
                              0.00000   0.00000
 => Size   parameters    :    0.00000   0.00000
                              0.00000   0.00000
 => U,V,W for Lambda(2)  : 
                              0.00274   0.00037
                              0.00228   0.00039
                              0.00150   0.00009



 ==> GLOBAL PARAMETERS:

 => Zero-point:   0.0688  0.0006
 => Background Polynomial Parameters ==>
                              157.592     1.17995    
                             -161.154     4.98089    
                              445.652     20.0857    
                              657.984     33.5124    
                             -940.810     65.4198    
                             -1464.48     78.0686    
 
 => Cos( theta)-shift parameter :   0.0000  0.0000
 => Sin(2theta)-shift parameter :   0.0033  0.0007
 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  4673
 => Rp: 6.35     Rwp: 8.84     Rexp:    3.76 Chi2:  5.54      L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 8.75     Rwp: 11.1     Rexp:    4.70 Chi2:  5.54    
 => Deviance: 0.260E+05     Dev*  :  5.565    
 => DW-Stat.:    1.2730     DW-exp:     1.9249
 => N-sigma of the GoF:      219.395
 => Phase:  1
 => Bragg R-factor:  4.18       Vol: 1077.946( 0.008)  Fract(%):    0.00( 0.00)
 => Rf-factor= 3.28             ATZ:    0.000          Brindley:  1.0000

 => Run finished at:     Date: 17/05/1999  Time: 15:58:53.840


Philips X'Celerator data
Refinements performed by using FULLPROF, in the |F| extraction mode (no structure). 
Using pseudo-Voigt profiles and asymmetry by using the classical asy1 and asy2
parameters (in FULLPROF). The main problem seems to be due to asymmetry. Improved 
reliabilities would probably be obtained with an approach that would better take account 
of this asymmetry at low angle.
Some reliabilities :
Al2O3 - step 0.004 - plot

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:
 =>  R-factors (not corrected for background) for Pattern:  1'
 => Rp: 8.20     Rwp: 12.1     Rexp:   11.16 Chi2:  1.17     
 => Conventional Rietveld R-factors ==>
 => Rp: 11.0     Rwp: 14.3     Rexp:   13.22 Chi2:  1.17    

Al2O3 : step 0.008 - plot (summation of 2 points in one).
This improves statistics, and gives better R values :

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS FOR PATTERN:  1
 =>  R-factors (not corrected for background) for Pattern:  1
 => Rp: 7.46     Rwp: 10.7     Rexp:    7.89 Chi2:  1.83  
 => Conventional Rietveld R-factors for Pattern:  1
 => Rp: 10.0     Rwp: 12.7     Rexp:    9.38 Chi2:  1.83    

And some FWHM (=Hw below)
 No.  Code     H   K   L  Mult      Hw      ETA/M    2theta/TOF    
    1   1      0   1   2     6  0.045745  0.104477      25.577 
    2   1      1   0   4     6  0.047820  0.117819      35.150 
    3   1      1   1   0     6  0.048575  0.121483      37.778 
    4   1      0   0   6     2  0.049844  0.126915      41.675 
    5   1      1   1   3    12  0.050448  0.129256      43.355  
    6   1      2   0   2     6  0.051540  0.133193      46.180 
    7   1      0   2   4     6  0.054368  0.142075      52.552 
    8   1      1   1   6    12  0.056927  0.148970      57.499 
    9   1      2   1   1    12  0.058197  0.152098      59.744 
   10   1      1   2   2    12  0.059019  0.154034      61.133 
   11   1      0   1   8     6  0.059120  0.154268      61.300  
   12   1      2   1   4    12  0.062472  0.161544      66.520 
   13   1      3   0   0     6  0.063648  0.163903      68.213 
   14   1      1   2   5    12  0.065248  0.166976      70.418  
   15   1      2   0   8     6  0.068263  0.172386      74.300 
   16   1      1   0  10     6  0.070407  0.175970      76.871 
   17   1      1   1   9    12  0.070720  0.176476      77.234 
   18   1      2   1   7    12  0.073573  0.180915      80.419 
   19   1      2   2   0     6  0.073838  0.181312      80.704 
   20   1      0   3   6     6  0.076251  0.184817      83.219  
   22   1      2   2   3    12  0.077392  0.186408      84.360 
   23   1      1   3   1    12  0.078193  0.187502      85.145 
   24   1      3   1   2    12  0.079458  0.189191      86.356 
   25   1      1   2   8    12  0.079614  0.189395      86.503 
   26   1      0   2  10     6  0.082336  0.192869      88.996 
   27   1      0   0  12     2  0.084294  0.195251      90.704 
   28   1      1   3   4    12  0.084867  0.195931      91.192 
   29   1      3   1   5    12  0.089343  0.200991      94.823 
   30   1      2   2   6    12  0.089898  0.201589      95.252 
   31   1      0   4   2     6  0.094147  0.205975      98.399 
   32   1      2   1  10    12  0.098026  0.209704     101.074 
   33   1      1   1  12    12  0.100702  0.212139     102.821 
   34   1      4   0   4     6  0.101492  0.212837     103.322 
   35   1      1   3   7    12  0.103624  0.214678     104.643 
   36   1      3   2   1    12  0.112232  0.221514     109.548 
   37   1      1   2  11    12  0.112819  0.221948     109.859 
   38   1      2   3   2    12  0.114690  0.223309     110.835 
   39   1      3   1   8    12  0.114996  0.223528     110.992 
   40   1      2   2   9    12  0.121284  0.227819     114.071 
   41   1      3   2   4    12  0.125763  0.230656     116.106 
   42   1      0   1  14     6  0.126887  0.231340     116.597 
   43   1      4   1   0    12  0.129853  0.233099     117.859 
   44   1      2   3   5    12  0.135726  0.236388     120.219 



SRM 660a, step 0.004 - plot :

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:
 =>  R-factors (not corrected for background) for Pattern:  1
 => Rp: 7.54     Rwp: 10.1     Rexp:    5.21 Chi2:  3.78  
 => Conventional Rietveld R-factors ==>
 => Rp: 14.4     Rwp: 14.6     Rexp:    7.53 Chi2:  3.78    


SRM 660a step = 0.008 - plot :

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS FOR PATTERN:  1
 =>  R-factors (not corrected for background) for Pattern:  1
 => Rp: 7.11     Rwp: 9.70     Rexp:    3.69 Chi2:  6.91   
 => Conventional Rietveld R-factors for Pattern:  1
 => Rp: 13.6     Rwp: 14.1     Rexp:    5.34 Chi2:  6.91    

And some FWHM (=Hw below)

 No.  Code     H   K   L  Mult      Hw      ETA/M    2theta/TOF    

    1   1      1   0   0     6  0.040438  0.178903      21.357   
    2   1      1   1   0    12  0.040657  0.207317      30.384   
    3   1      1   1   1     8  0.041144  0.229530      37.440    
    4   1      2   0   0     6  0.041805  0.248619      43.505    
    5   1      2   1   0    24  0.042607  0.265777      48.956   
    6   1      2   1   1    24  0.043538  0.281613      53.987  
    7   1      2   2   0    12  0.045772  0.310664      63.216 
    8   1      3   0   0     6  0.047083  0.324291      67.545 
   10   1      3   1   0    24  0.048534  0.337504      71.742 
   11   1      3   1   1    24  0.050138  0.350405      75.841 
   12   1      2   2   2     8  0.051910  0.363076      79.866 
   13   1      3   2   0    24  0.053872  0.375590      83.842 
   14   1      3   2   1    48  0.056049  0.388010      87.787 
   15   1      4   0   0     6  0.061190  0.412811      95.666 
   16   1      4   1   0    24  0.064251  0.425310      99.637 
   18   1      3   3   0    12  0.067730  0.437958     103.655 
   20   1      3   3   1    24  0.071723  0.450826     107.743 
   21   1      4   2   0    24  0.076362  0.463995     111.927  
   22   1      4   2   1    48  0.081837  0.477564     116.237 
   23   1      3   3   2    24  0.088426  0.491657     120.715 
   24   1      4   2   2    24  0.106956  0.522141     130.399 
   25   1      5   0   0     6  0.120900  0.539107     135.789  
   26   1      4   3   0    24  0.120901  0.539107     135.789 
   27   1      5   1   0    24  0.141032  0.557909     141.762 
   29   1      5   1   1    24  0.173952  0.579630     148.662  


These results are not better than previous ones above, nor really worst. 
However, I repeat that a better account of asymmetry would probably lead 
to better reliabilities.
Refinement with structure for LaB6 - plot
Doing fits with structure is not possible with SRM 1976 due to 
high preferred orientation. But this should be possible with 
LaB6 SRM 660a. Using CTHM=0.8247 in FULLPROF (2-theta for the Ge
111 reflection is 24.749), here is the .sum FULLPROF file :
        **********************************************************
        ** PROGRAM FullProf.2k (Version 1.9c - May2001-LLB JRC) **
        **********************************************************
                       M U L T I -- P A T T E R N                
            Rietveld, Profile Matching & Integrated Intensity
                 Refinement of X-ray and/or Neutron Data


    Date: 17/04/2002  Time: 09:28:42.510                              

 => PCR file code: 660a-2s
 => DAT file code: 660a-2s              -> Relative contribution: 1.0000                                                            
 => Title:    LaB6 NIST 660a Philips - X'celerator                                    

 ==> CONDITIONS OF THIS RUN FOR PATTERN No.:  1

 => Global Refinement of X-ray powder diffraction data                          
 => Global Refinement of X-ray powder diffraction data                          
    Bragg-Brentano or Debye-Scherrer geometry
 => The    5th default profile function was selected

 => Data supplied in free format for pattern:  1
 => Wavelengths:  1.54056 1.54056
 => Cos(Monochromator angle)=   0.8247
 => Absorption correction (muR-eff):   0.0000
 => Base of peaks: 2.0*HW*   20.00
 ==> Angular range, step and number of points:
     2Thmin:  18.0030  2Thmax: 151.9950  Step:   0.0080  No. of points:  16750
 =>-------> Pattern#  1
 => Crystal Structure Refinement for phase: 1
 => Scor: 2.7954

 ==> RESULTS OF REFINEMENT:


 => No. of fitted parameters:   19


------------------------------------------------------------------------------
 => Phase No.  1    La B6  NIST                          P M 3 M             
------------------------------------------------------------------------------

 =>  No. of reflections for pattern#:   1:    30  


 ==> ATOM PARAMETERS:

  Name    x     sx       y     sy       z      sz      B   sB  occ.  socc.  Mult

 LA    0.00000(   0)  0.00000(   0)  0.00000(   0)  0.112(  7)  0.100(  0)    1
 B     0.50000(   0)  0.50000(   0)  0.19850(  67)  0.131( 53)  0.600(  0)    6

 ==> PROFILE PARAMETERS FOR PATTERN#  1

 => Cell parameters      :
                              4.15697   0.00001
                              4.15697   0.00001
                              4.15697   0.00001
                             90.00000   0.00000
                             90.00000   0.00000
                             90.00000   0.00000
 
 => overall scale factor :    0.000572586   0.000001496
 => Eta(p-v) or m(p-vii) :    0.11607   0.00572
 => Overall tem. factor  :    0.00000   0.00000
 => Halfwidth parameters :    0.00257   0.00005
                             -0.00099   0.00007
                              0.00174   0.00002
 => Preferred orientation:    1.00000   0.00000
                              0.00000   0.00000
 => Asymmetry parameters :    0.02044   0.00464
                              0.04150   0.00109
                              0.00000   0.00000
                              0.00000   0.00000
 => X and y parameters   :    0.00299   0.00010
                              0.00000   0.00000
 => Strain parameters    :    0.00000   0.00000
                              0.00000   0.00000
                              0.00000   0.00000
 => Size parameters (G,L):    0.00000   0.00000
                              0.00000   0.00000

 ==> GLOBAL PARAMETERS FOR PATTERN#  1


 => Zero-point:   0.0144  0.0002
 => Background Polynomial Parameters ==>
                              330.888    0.801991    
                              29.0268     4.14131    
                              29.3728     10.0818    
                              122.108     29.7537    
                              171.935     23.0625    
                             -384.967     49.7109    
 
 ==> RELIABILITY FACTORS WITH ALL NON-EXCLUDED POINTS FOR PATTERN:  1

 => Cycle: 20 => MaxCycle: 20
 => N-P+C: 16731
 =>  R-factors (not corrected for background) for Pattern:  1
 => Rp: 6.42     Rwp: 8.69     Rexp:    4.30 Chi2:  4.09      L.S. refinement
 => Conventional Rietveld R-factors for Pattern:  1
 => Rp: 18.1     Rwp: 15.3     Rexp:    7.59 Chi2:  4.09    
 => Deviance: 0.711E+05     Dev*  :  4.250    
 => DW-Stat.:    0.5594     DW-exp:     1.9544
 => N-sigma of the GoF:  282.495

 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS FOR PATTERN:  1

 => N-P+C:  8320
 =>  R-factors (not corrected for background) for Pattern:  1
 => Rp: 7.30     Rwp: 9.82     Rexp:    3.69 Chi2:  7.06      L.S. refinement
 => Conventional Rietveld R-factors for Pattern:  1
 => Rp: 14.0     Rwp: 14.2     Rexp:    5.36 Chi2:  7.06    
 => Deviance: 0.613E+05     Dev*  :  7.362    
 => DW-Stat.:    0.6514     DW-exp:     1.9367
 => N-sigma of the GoF:      390.956

 => Global user-weigthed Chi2 (Bragg contrib.):8.22    
 => Phase:  1
 => Bragg R-factor:  2.32       Vol:   71.834( 0.000)  Fract(%):  100.00( 0.02)
 => Rf-factor= 1.18             ATZ:  203.770          Brindley:  1.0000

 
              CPU Time:    15.928 seconds
                            0.265 minutes

 => Run finished at:     Date: 17/04/2002  Time: 09:28:58.500                              


RBragg = 2.32%, that seems OK. The thermal parameters factors are a bit 
small. The CTHM value would have to be better estimated, Celeste Reiss suggests
to use 0.88.
The main advantage of the X'Celerator is the gain in time with no loss of resolution.
For the two data sets (the original with 0.004 step) the total times were:
           SRM1976      16:00:34 h:m:s
           SRM660a      17:48:50 h:m:s
For those interested in the data, they are gathered together with the
.pcr and .sum FULLPROF files in a file compressed by WinZip : x-celerator.zip



Thanks to all those who made a response. Opinions developped here belong respectively to their authors.
Armel Le Bail - Laboratoire des Fluorures, CNRS-URA-449, Université du Maine, 72017 Le Mans Cedex, FRANCE -