This is my second shot at the SMRR-2002. In these days I belong to 
the "developers" category of users, by education I am a 
mineralogist-geologist.

Software used: PHILIPS Analytical X'Pert HighScore v.1.0b, with ICDD 
               PDF-2 database release 2001.
               PHILIPS Analytical X'Pert Plus v.1.0 for crystallographic 
               analysis (sample2)


I am disappointed with the examples 2 and 3! This is not because I have 
difficulties to find an answer, but because they are not very well suited 
to judge the quality of a search-match procedure.


Any search-match program can only match unknown data against existing 
reference patterns. When the match is poor, this does result inevitably 
in a low score. If there is a good quality reference pattern present, 
(even with extra large sample displacement or zero shift), it is usually 
found. But if there is no good reference data available, this does not 
tell lot about the search-match algorithm.


More examples to test search-match:
Real-life examples were prepared and measured at the Department of 
Chemistry, North Dakota State University. These measurements were used 
at the 1993 Denver Conference for search-match applications in a workshop 
by Prof. G.J. McCarthy, NDSU, Fargo, North
Dakota (mostly minerals).
Other examples can be found in the "tutorial" section of X'Pert HighScore, 
covering a wide range of materials including some "difficult" examples too.


====================================================
Sample 1: solved, phosphate minerals


The main phase is: Gormanite, Fe3 +2 Al4(PO4)4(O H)6 *2H2O, additional 
phases are Quartz, SiO2 and Corundum, Al2O3. (see figure SMRR1_2a). 
Still there are a lot of peaks remaining unexplained and I am very sure 
this is due to the major phase Apatite,
Ca5(PO4)3F, even if there has no Calcium been detected in the chemical 
analysis. (see figure SMRR1_2b).


These four phases explain all peaks and all major intensities above the 
background.


====================================================
Sample2: solved - bad example


This is a monophase material. Unit cell search (TREOR) and subsequent 
refinement revealed a tetragonal cell with a=9.082 and c=13.458 [Angstrom] 
and a zero offset of -0.01 [deg. 2Theta]. The closest match in the ICDD 
PDF-2 is no. 26-1993, b-Uranium bis(tetraethylammonium) fluoride, 
C8H20F6NU. Peak positions match rather good, but the intensities are 
very different. The "true" phase is not in the ICDD PDF-2 database.

The given chemistry and the only entry in the ICDD, 48-0475 , 
20SiO2 * 2 (C7H12NF), do not fit very well. Even when allowing for 
extra zero shift (0.136 deg. 2Theta), only half of all present peaks 
and net intensities are matched, resulting in a rather
low score. (see figure SMRR2_2).


A simple 2Theta-zero-shift or a sample displacement fit do not help 
here. This example requires an independant change/adaption of the 
a- axis and of the c-axis in order to fit better, which is not realized 
in the search-match software. As you can see, the
results from a unit cell refinement (as given above) are still 
different from the ICDD-card 48-0457: a=9.194 and c=13.396 [Angstrom], 
explaining the bad fit.


===================================================
Sample 3: not solved


There is no solution present in the database. The only entry with the 
given chemistry (C13H10N2O4, ICDD 02-0201) has strong unmatched peaks a
nd explains only a little part of the experiment. (see figure SMRR3_2a). 
This is an organic substance, but no drug.


The best fitting drug substance is Thalidomide, kind of a bad joke... 
(see figure SMRR3_2b).


====================================================
Sample4: completely solved (phases from a car battery)


The phases are: Massicot, PbO; Litharge, PbO; lead oxide sulfate, 
Pb5O4SO4 and lead oxide sulfate hydrate, Pb4O3SO4*H2O.


There could be very little lead oxide carbonat, Pb2OCO3, present, 
which explains one more small peak with respect to my first answer. 
Also lead carbonate, Cerussite, PbCO3, could be present in minor amounts, 
but this does not explain more peaks or more
net intensity. (figure SMRR4_2).


====================================================