An interpretation is given of the interesting features of the unusual structures of the hexacarbonyls and the tetroxide. Photographs of the electron diffraction patterns of the various molecules were taken with the apparatus described by Brockway(l) and visually interpreted according to the methods currently used in these laboratories. Thick lines below the radial distribution curves indicate the distances and scattering weights of the best models by their positions and heights.
Curves for FCF angles of 107.5° with further variations of the temperature factors for C'F and C''F were also calculated. The scale factor becomes exactly unity for 108.5°N, again in general confirmation of the angle determination.
Numerical calculation of the difference in the rigid pattern distance C'''F and the average value of this distance leads to a shortening of nearly 0.0~. The truncation of the FF' distances is unimportant in the interpretation of the diffraction pattern and is neglected. The length of the triple bond is quite insensitive to the electronegativity of the substituents on the methyl group.
It seems likely that the triple bond is dominant in determining the adjacent C-C distance and the methyl group bond angles. Consideration of apparent shortening of CC''' and C'''F through bending of the carbon axis is taken into account in Model B.
The gross appearance of the photographs as well as the radial distribution curve made of visual I of Figure 3. The standard curve was considered a combination of visual I and visual II with significantly more weight given to the latter. Theoretical intensity curves were calculated for the entire plausible range of the six independent parameters described in Table XI for models containing three long and three short we bonds, with however only the w e, w o, and.
33 of the total of 179 models for which curves were calculated are reproduced in Table XII and their intensity curves are presented in Figures 3, 4, 5 and 6. The first letter in a curve designation indicates the set of relative temperature factor coefficients used (Table XIII), and the letter Z indicates that account has been taken of the variation in scattering power of C and 0 relative to W as a function of q . Badger's Rule< 22) and reasonable bending frequencies were used assuming the simple valence potential function to estimate the values of the relative temperature factor coefficients.
The lAZ, lBZ and 2CZ curves, each for sheep angles of 90°, are considered to be more satisfactory than any other model because they involve a minimum of variable parameters, where the overall quality of the fit is not improved by considering non- WCO linear or adjustable, as in 3C (short angle of w e, 105°; long angle of w e, 84°), of eight. Curve 4E, the best of 67 curves calculated for different values of the six parameters of Table XI for two short and four long us, is unsatisfactory. Furthermore, model 4 cannot be made acceptable by adding the neglected cc, co, and 00 scattering terms even for ram angles that vary widely from 90°, the value used in 4CZ.
Less than about two and a half WC short and more than about three and a half long, we cannot form the structure of W(C0)6•. This latter conclusion is subject to some extent, of course, to the limitations of Table XI for the 2:4 models. The final results are given in Table XV, in the derivation of which incomplete curves (eg, 50, lOE, etc.) were used except for the best models of cases A and B.
At Professor Pauling's suggestion, she set about re-determining the structure of molybdenum hexacarbonyl Mo(C0)6. 0.05~ • Although the symmetry of the crystal space group does not constrain the symmetry of the molecule, regular octahedral symmetry with MoC equal to 2.13~ and CO equal to about 1.15~ was found (2l) to be compatible with the X-ray data. A cursory study of the dissociation of Mo(C0)6 at temperatures of about 130° showed that the pressure of CO above the excess of solid Mo(C0)6 is less than about ten mm even after one hour at 112°, where the vapor pressure of the carbonyl is quickly reached is 81 mm.
But none of the many experiments produced a picture showing that Moe and MoO scattering was even one-fifth as important at about q equal to 70 as the clear CO scattering. This extraordinary decrease in MoC and MoO scattering relative to CO scattering can be attributed to a large one. This extraordinary decrease in MoC and MoO scattering relative to CO scattering can also be attributed to the existence of at least two different MoC distances, as in W(e0)6. In fact, the actual appearance of the photos, ash.
The radial distribution curve, however, is calculated from the part of the curve for q less than 65 and, accordingly, does not resolve the MoC and MoO groups. As for W(C0)6, a letter after the model number indicates the set of relative temperature factor coefficients used from Table XVI, and a letter Z indicates that the variation of the scattering power of Mo with respect to C and o. . The temperature factor coefficients were evaluated under the usual assumption of the simple valence potential function with the use of Badger's Rule (22) and reasonable bending frequencies.
The possibility of a 2:2:2 Moe split has not been investigated in detail, but it is likely that n2h . a model with MoC and 2.13~ distances would not prove entirely unsatisfactory, even to the very limit of the known diffraction pattern. Increasing the relative temperature factor coefficients for adjacent cc, co, and 00 would improve the fit obtainable for the Oh models for values of q less than 65, but would not explain the maximum of 11, minimum of 12, and maximum of 12. Determining the scale factor for the model 1 is presented in Table XVIII, and the final results in Table XIX.
A sample of Os04 (Amend Drug and Chemical Company, New York City), generously provided by the Shell Development Company, was purified by vacuum sublimation and photographed with temperatures of the 40 cc sample bulb ranging from 60° to 70°. The areas under the peaks, interpreted with the ZiZj approximation, indicate that the ratio of the number of long OsO to the number of short OsO is 1.5 and that the number of 00 is 7.5 times greater than in the tetrahedral Oso4- model, apparently partly because the De ZiZj approximation is far from valid and partly because the fine details of the pattern in the visual curve were strongly emphasized. The values used for the relative temperature factor coefficients were calculated based on the usual assumption of the simple valence potential function from the reported Raman spectrum <27) together with that of Badger.
In all models the angle change in the dispersive power of 0 with respect to Os is taken into account. Determined according to weights II of table XXI with the exception of model 4, for which all features were given equal weight, but missing 1 min, 1 max and 2 min. The plot of the intensity curve is fortunate for determining the overall arrangement of the oxygen atoms, but happens to be unfortunate for estimating the number of long and short OsO.
Because, minimum 7 of curve 2 is deepened either by increasing the 00 weight, as in curve 4, or by increasing the number of long Oso's, as in curve 3. On the other hand, if the measurements of maxima 5 and 6 reliable is -- they can suffer from the St. But in any case curve 4 is intolerable beyond q is equal to 70 except that it is physically unjustifiable, so that curve 1, with the ratio 1.5 of number of long OsO to number of short OsO and otherwise clearly the most satisfactory, should be judged best .
For a number of years there has been doubt about the value of the dissociation energy of carbon monoxide and about the heat of sublimation of graphite, a directly related quantity. The argument is based on the postulate that the linear extrapolation of the vibrational levels for the lower vibrational states of a molecule (a linear Birge-Sponer extrapolation) leads to an energy value corresponding to the dissociated atoms in a hypothetical electronic state that represents the called valence state. . Some substantiation of this value is provided by considering the normal state and the first excited state of the nitrogen molecule.
The valence state energy of the carbon atom can be similarly deduced from spectroscopic data for the molecules CH and C2. The energy of a carbon atom in its valence state and an oxygen atom in its valence state with respect to the normal state of the carbon monoxide molecule is found to be 11.23 e. Linear extrapolation of the vibrational levels of the lowest state of the CN molecule, X 2L:+, leads to 9.85 e.
The energies of the valence states of some atoms and the dissociation limits of the valence state of some. The significance of the difference in the dissociation energy found by linear extrapolation and the true dissociation energy has recently been pointed out by Pauling (3,4. The energy of the valence state of divalent oxygen as it exists in 02, OH, and H20 is evaluated by.
It is established that this energy of the valence state is almost independent of the individual molecule in which the divalent oxygen is located (4). And when several precise but very different values of the true dissociation energy are possible, the value predicted using the valence state energy concept should eliminate the ambiguity. Table I gives the values of the dissociation energies of several diatomic molecules determined by linear graphical extrapolation.
