DE DETERMINADOS BIENES DEL PATRIMONIO MUNDIAL

When comparing observed and calculated frequencies, many factors need to be considered. Problems arise with inaccuracies in the observed values, incorrect

assignments, and the inadequacies o f the theoretical models used for the calculations. Most o f the assignments o f the vibrational fundamentals found in the literature may be considered reliable, since they were made using both Raman and infrared spectra, and the study and identification o f the combination bands. However, the observed values may vary by a few wavenumbers, due to inaccuracies in the recording o f the spectra and the difficulties in the selection o f the correct centre o f a band when two or more bands overlap. The assignment o f inactive modes is difficult, and was usually based on the interpretation o f combination bands. Further misinterpretation of the spectra may have been caused by the presence o f weak bands due to contamination o f the samples used.

It can be seen from tables 6.1 to 6.19 that reasonable agreement is obtained between the ab initio results and the observed frequencies. It should be noted however, that the ab initio calculations are based on several approximations as described in section 1.6.2. The neglect o f electron correlation by the HF method will result in errors in the computation, however, comparison o f the results obtained from the HF and MP2 levels o f theory indicate that at least for the lighter HFCs the differences were small. It is possible that the results for the heavier molecules may be improved by the use o f a higher level o f theory, however, considerable computer time would have been required

to complete these calculations, and generally the HF results seem acceptable for this work.

The scaling factors used to adjust the frequencies have been determined from a large number o f ab initio calculations (Pople et al. 1993). Another source o f error may arise from the calculation o f the equilibrium geometry o f the molecule. Fogarasi and Pulay (1985) state that SCF wave functions ( as used for the HF calculations) generally yield bond lengths which are too short This is confirmed by the results for CF4 and C2F6 calculated by Cooper et al. (1989). The effects o f anharmonicity for the C-H stretching modes may increase the error for the calculated values, but even with errors in the range o f 20-30cm*1, Fogarasi and Pulay (1985) suggest that the results are suitable for checking vibrational assignments.

The Urey-Bradley calculations have not agreed as w ell as the ab initio

calculations with the fundamental assignments. The programs used in this work allow only the simple Urey-Bradley potential field to be used. Shimanouchi (1963) divided various molecules into three groups, those for which the simple UB field was

successful, such as CX4 (X=halogen); those for which the UB field was successful, but the resultant force constants were not transferable to similar molecules, such as the halomethanes with at least one hydrogen atom; and those for which the force field had to be modified, such as the dichloroethanes. The modification o f the UB force field includes the incorporation o f a 'trans' interaction constant, a 'gauche' interaction constant and an angle interaction constant. The 'trans' and 'gauche' interaction

constants given by Shimanouchi (1963) involve interactions between atoms attached to adjacent carbon atoms, and the angle interaction constant involves interactions between atoms attached to the same carbon, but not directly affected by the rocking or twisting motions involved. The use o f the same force constants for bonded atoms in different environments may also have introduced errors, for example, the C-F stretching force constants for fluorine atoms sharing the same carbon atom may not all be the same if the atoms on the other carbon atom are not identical to each other. For example, the C-F stretching force constants for CFC115, shown in table 6.23, are not all the same, even for the three fluorines attached to the same carbon atom.

Errors in the experimental geometry will have introduced errors in the frequencies, since a small change in bond length or angle results in a change o f frequency by a few wavenumbers.

The most time-consuming problem arose from the difficulty o f starting with two sets o f 'approximate' values, the initial force constants and the assigned frequencies. While most o f the assigned frequencies may be considered to be accurate within a few wavenumbers, the force constants, in particular for the non-bonded interactions were not reliable. The assumption was made that the assignments given in tables 6.1 to 6.19

were correct, and the force constants adjusted accordingly. Difficulties arose for those frequencies where the assignments were doubtful or missing, as, a change in a

particular force constant may change the calculated frequency to correspond to any specified value within a range of approximately 100 wavenumbers. This problem was also encountered by Guirgis and Crowder (1984) when they were calculating the frequencies for HFC152a.

Further refinements of the force constants could have been carried out, however, the possible reduction of the overall errors by tenths of a percent was not considered to be beneficial, since it is the trends in force constants that are o f more interest here, as the simple UB force field used will not give exact values.