Académique Documents
Professionnel Documents
Culture Documents
Emil J. Slowinski Citation: J. Chem. Phys. 23, 1933 (1955); doi: 10.1063/1.1740608 View online: http://dx.doi.org/10.1063/1.1740608 View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v23/i10 Published by the American Institute of Physics.
Downloaded 12 Apr 2012 to 133.41.91.47. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
THE
JOURNAL
OF
CHEMICAL
PHYSICS
VOLUME
23,
NUMBER
10
OCTOBER,
1955
New Form of Molecular Potential Function for Use in Molecular Vibrational Analysis
EMIL
J.
SLOWINSKI, JR.
C LASSICAL mechanics offers several methods for the analysis of the vibrations of linearly-coupled
systems. These methods involve the expression of the potential and kinetic energies of the system in terms of suitable coordinates. Given a system (see Fig. 1) consisting of several masses connected by conventional springs, one can immediately express the potential energy:
where the first sum is taken over springs between masses, and the second over springs in the angles. kil is the force constant of the spring between the ith and lth masses, Lil is the instantaneous distance between them, and Ailo is the distance in the event that there is no strain in the spring. Similarly, k j is the force constant of the spring in the jth angle, Aj is its instantaneous magnitude, and Cijo its magnitude when there is no strain in the spring. The form of the function is obvious, given the system. It allows for all possible strains within the system at equilibrium, and, indeed, one would not consider expressing the potential energy in the classical problem in any other way. The methods used in present-day problems in molecular vibrational analysis are essentially classical; atoms are considered to be mass points in a linearly-coupled force field and are assumed to move only infinitesimally from their equilibrium positions. This approach has been shown to be consistent with results as obtained from direct use of quantum-mechanical reasoning, and is almost universally employed. However, the potential functions which have been assumed for molecules are not of the form (1), which we would like to propose as being a reasonable function, but rather (in the simplest case) are of the following type:
(2)
rather than finite changes from the strain-free configuration. Basically, function (2) assumes what function (1) does not; namely, that at equilibrium, there is no potential energy in the molecule and hence no intramolecular strain. Therefore, even if the molecular force field were of the simple form shown in Fig. 1, function (2) would not be directly applicable to the vibrational analysis problem, except in the special situation where the molecule existed at equilibrium under no strain. To improve correlation between calculated and observed vibrational frequencies function (2) has been modified in various ways. One of the most successful has been a function of the type suggested by Urey and Bradley, l which takes essentially the following form:
where the sums are taken in the same way as in (1), FIG. 1. A simple linearly-coupled system. but 6.L il and 6.A j are infinitesimal changes in bond length and angle from the equilibrium configuration, 1 H. C. Urey and C. A. Bradley, Phys. Rev. 38, 1969 (1931). 1933
Downloaded 12 Apr 2012 to 133.41.91.47. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
1934
and angles. The (k')'s have no such simple meaning. Simanouti,2 in a treatment of a group of molecules with a Urey-Bradley field, never reports some of the k' values at all, but rather goes directly to the intramolecular forces. Also, the proposed function (1) attributes, in the first instance at least, intramolecular strain to the inability of the bond lengths and angles in the molecule to fit easily together, where the UreyBradley field introduces strain terms in a more formal way. Generalizing, one might say that the proposed function makes a more literal interpretation of the familiar model for the molecule than does the UreyBradley field. Application of the proposed function to problems of molecular vibrations yields some interesting results. The approach for XY 2 molecules of the H 20 type, shown in Fig. 1, will be outlined. The Lagrangian method is used with Cartesian displacement coordinates. The potential energy of the XY 2 molecule is taken to be, according to (1):
X(OLI3_ OLI3] )]+ksI:.AsOAs oqu oqu equit. oqu OAs )] . +k.[ (Aso-aso) ( - -OAs] oqu oqu equit.
Applying the Lagrangian equations to the system, and noting that at equilibrium a balance of forces exists within the molecule, we obtain for the generating equation for the secular determinants (general form) :
oLi! w2 ul:.qu= 'Lki!I:.Li!-m oqu OL' I OL'I] ) + 'Lkil(Lilo-Ailo) ( --' - - - ' . oqu oqu equit.
The I:.L's, I:.A s, and the various associated derivatives are fairly easily expressed in terms of the Cartesian displacement coordinates by means of well-known or readily-derived formulas and the secular determinants thereby formed. For the totally-symmetric class the resulting frequency equations are:
(WI2+W22)my=kxy[1+2my D2 mx y 2m + ( 1+--G2) ( 1 - mx L 120
AI20)] +2k yy
2 WI 2W 2. - - - mx+2my
(4)
k [1- A120]
Xy2
where mu is the mass of the uth atom, I:.qu is one of its Cartesian displacement coordinates, L ilo and A jo are magnitudes of bond lengths and angles at equilibrium, derivatives are taken in the instantaneous or, as noted, in the equilibrium configuration, and all other terms are as previously defined. From this equation it is apparent that the inclusion of possible strain has simply introduced extra terms into the force Eq. (4), and that in the absence of strain the equation reduces immediately, as expected, to the one obtained by application of the Lagrangian to function (2).
~
A120) ]
(A130)
Downloaded 12 Apr 2012 to 133.41.91.47. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
NEW FORM
OF
MOLECULAR
POTENTIAL FUNCTION
1935
TABLE
From the condition that a balance of forces must exist at equilibrium we obtain:
Molecule
hUO
P2
va
A,o o
kxy
--1
U80 0
L1202
Ll20
and
XI0 6 dynes/em
'--------..----'
a Vibrational frequencies are taken from G. Herzberg I nfrared and Raman Spectra (D. Van Nostrand Company, Inc., New York, 1945).
A.!oO
kxy
Lao 2
aBo o
H 2O 3825.32 1653.91 3935.59 10427' 8.317 0.931 121.0 D 20 2758.06 1210.25 2883.79 10427' 8.293 0.956 122.6 D 20 2763.3 1208.0 2883.6 (calc) em-1 X 105 dynes/em
'-----v----'
a Vibrational frequencies are those reported by Darling and Dennison, Phys. Rev. 57, 128 (1940).
to have a value of about 0.09X 105 dynes per cm' which is only about one percent of the 0- H bondstretching force constant. In the equations in which they appear the contributions of terms in k yy is also small, being about 1.5% in one case and about 5% in the other. The value of k yy is very sensitive to the vibrational frequencies which are used, and a change of one or two wave numbers in a single frequency is sufficient to change its size by a factor of two. This very small calculated size and effect of k yy has led to the postulate that in these molecules this force constant may reasonably be taken to be zero, though the force between the end atoms is not zero. This assumption, suggested by the results with the water molecule, has been extended to other XY2 molecules. Since that
assumption reduces the number of independent parameters by one, it becomes possible to calculate the vibrational parameters for a molecule in terms of its vibrational frequencies, with no need for the use of isotopic pairs. The results of such calculations are given in Tables I and II. In Table I, the parameters for H 20 and D 20 have been independently determined on the basis of the three-parameter function and the zero-order frequencies of the molecules. In the last line of the table the parameters of H 20 as obtained above have been used to calculate the zero-order frequencies of D 20. The mean deviation from the reported frequencies is 0.13%. In Table II, the parameters for several molecules are given on the basis of the fundamental frequencies and the three-parameter function. It is to be noted that in all cases studied, the bond angle in the molecule at equilibrium is found to be smaller than the angle would be in its strain-free state, indicating a moderate attraction exists between the nonbonded atoms in these molecules. In all calculations involving water and hydrogen selenide the strain-free bond angle Ci so appears to have the same value, about 120, although the angle magnitudes at equilibrium are known to differ by 15. Decrease in bond length due to the attraction between end atoms varies from about 4% for water to about 28% for F 20. The approach to molecular properties through the proposed potential energy function shows, we believe, some promise, and at present work is progressing on the application of the function to the vibrations of NH 3and CH4-type molecules.
Downloaded 12 Apr 2012 to 133.41.91.47. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions