The Tests folder has a number of sample Vibratz files for crystals, intended to reproduce published calculations.  The published calculated frequencies are entered in the Vibratz files as observed, so the differences can be seen clearly.

The Samples folder also contains a number of similar calculations on molecules (which generally have the actual observed frequency, instead of the calculated).  Many dozens of calculations on molecules have been done, in most cases with exact reproduction of calculated frequences.  See the SAMPLE.TXT file for some general discussion of force constants in silicates and some comparisons of different models 

There are two main reasons why Vibratz results may not agree with published calculations.  First, it is easy to make errors in manual calculations or even machine calculations which have to be set up by hand.  Second, details of force constant specifications, such as bond and angle limits, are not always given in published papers.  Frequently force constants such as "torsion" are not even defined.  Complete published results should probably include the number of each type of coordinate which is actually found in the structure.  These numbers are given in Vibratz in the output section "Internal Coordinates Located".

Since Vibratz agrees in most cases with published results, it would seem that it is doing the calculations correctly, and when there is not agreement it is likely that the other results have errors or that there is some difference in method or interpretation of force constants.  Nevertheless the possibility of errors in Vibratz in some types of calculations, or with some types of forces, cannot be ruled out, especially in some of the more complex forces such as 4-atom angles and interactions.

THIOSPINELS  - CdYb2S4, MgSc2S4, MgYb2S4

** Boldish & White, J. Solid St. Chem 25, 121 (1978)   Average deviation 0.2, 0.2, 0.4 wavenumbers.

These calculations use a simple valence potential and Vibratz reproduces them exactly (within round-off).

QUARTZ (LOW)

** Mirgorodskii, Lazarev & Makarenko, Sov. Phys. Krist (?) 1970(?), 289  Average deviation 0.2

This is a simple valence-force calculation with a moderate number of force constants and Vibratz reproduces it exactly.

** Etchpare, Merian and Smetankine, J. Chem. Phys., 60, 1873 (1974)  Average deviation 1.1

This is a valence-force calculation with 10 non-zero force constants including torsion.  They did not define the torsion coordinate.  There are 6 possible torsion (tau) angles at each bridging oxygen, and good agreement is obtained with Vibratz by using all of these with a force-constant value about half of theirs.  The torsion forces are small in any case and not very important in the calculated frequencies.

** Iishi & Yamaguchi, Am. Mineral 60, 907 (1975)  Average deviation 1.0

This uses a Urey-Bradley potential for SiO4 tetrahedra.  The "intramolecular tension", not otherwise defined, is presumably a torsion force, and Vibratz gets good agreement with a small torsion force constant.

QUARTZ (HIGH)

** Bates, J. Chem. Physics 56, 1910 (1972)  Average deviation 82

The agreement is very bad, presumably because of problems with the bond-angle interaction forces.  Bates specifies two types, with force constants of -.592 and +.330 md/A.  These different values are unreasonable for bonds and angles sharing two atoms within a tetrahedron, and if some other type of interactions are meant, the nature of the interactions is not clear.

CRISTOBALITE

** Etchepare, Merian and Kaplan, J. Chem. Phys., 68, 1531 (1978)  Average deviation 0.8

This uses the same force field as for quartz (above), and the results are similar.

FELDSPARS - Microcl, Lalbite  Average deviation 7.4, 8.1 (averaged forces),  0.8, 2.0 (with dependencies) 

** Von Stengel, Z. Krist. 146, 1 (1977)

These are simple valence calculations, but the M-O forces actually use a dependency on bond distances, and the bond-bond interactions through the bridging oxygen atoms use a dependency on the cosine of the T-O-T angle.  The .vbr files given in the TESTS folder have the interaction force constant adjusted for the average T-O-T angle, but do not otherwise take the dependencies into account.  The files MicroclB.out and LalbiteB.out give the results from a modified version of Vibratz which did include the dependencies.  This gives very good agreement, considering that Von Stengel did not give the unit-cell parameters or fractional atomic coordinates which he used - small differences are to be expected when the same crystallographic data are not used.

DIOPSIDE

** Tomisaka & Iishi, Mineral. J. Japan, 10, 84 (1980)  Average deviation 7.9

This is typical of later calculations by Iishi and colleagues, and the agreement is only fair.   Although the Urey-Bradley method used is supposed to be that of Shimanouchi (as used by Vibratz), no details are given and there seem to be differences with the method used for quartz (above).
