New Potential Function for Alkali Halide Molecules

Yatendra Pal Varshni and Ramesh Chandra Shukla

Citation: J. Chem. Phys. 35, 582 (1961); doi: 10.1063/1.1731973

View online: http://dx.doi.org/10.1063/1.1731973
View Table of Contents: http://jcp.aip.org/resource/1/JCPSA6/v35/i2
Published by the American Institute of Physics.

Additional information on J. Chem. Phys.

Journal Homepage: http://jcp.aip.org/
Journal Information: http://jcp.aip.org/about/about_the_journal
Top downloads: http://jcp.aip.org/features/most_downloaded
Information for Authors: http://jcp.aip.org/authors

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
THE JOURNAL OF CHEMICAL PHYSICS VOLUME 35, NUMBER 2 AUGUST,1961

New Potential Function for Alkali Halide Molecules

Y ATENDRA PAL V ARSHNI*
Division oj Pure Physics, National Research Council, Ottawa 2, Canada
AND

RAMESH CHANDRA SHUKLA

Department oj Physics, Allahabad University, Allahabad, India

(Received December 12, 1960)

A new potential-energy function has been suggested for the alkali halide molecules,
U = (-&/r) +P exp( -kr2 ) ,
where P and k are constants. Values of a e, "'eX., and the ionic binding energy Di have been derived for the
following three functions: (1) simple Born-Mayer potential, (2) Rittner potential, and (3) the above
suggested potential. The results obtained with the new function are much better than those obtained
with the Born-Mayer potential, but slightly inferior to those obtained with the Rittner potential.

INTRODUCTION used two types of repulsion terms, viz., exponential

NUMBER of attempts have been made to ap-
A proximate the potential energy curves of the
alkali halide molecules assuming the molecule to be
constituted of ions. For earlier developments, refer-
ence may be made to Rittnerl and Varshni. 2 Some years
ago Rittnerl presented a theory of the alkali halide
molecules in the spirit of the Born-Mayer lattice theory.
This theory has been found to be reasonably successful
in describing the ionic bond in the alkali halide mole-
cules.
The binding energy is usually represented as con-
sisting of the following terms: (a) charge-charge
interaction, which gives a term [-e2/rJ, (b) terms
arising due to charge-dipole interaction, dipole-dipole
interaction and quasi-elastic energy stored in the in-
duced dipoles, [-e2(at+a2)/2r4-2e2ala2/r7J, where at
and a2 are the polarizabilities of the alkali and halide
ions: (c) a van def Waals attraction term [ -c/,.oJ,
(d) kinetic energy terms representing the difference
in translational, rotational, and vibrational energy
between the molecule and the free iom; from which it is
composed (these are usually quite small and may be
neglected), (e) a short-range repulsion term. Rittner
used a Born-Mayer exponential repulsion term, A
exp( -r/p), and found satisfactory results for a number
of properties of alkali halide molecules. Honig, Mandel,
Stitch, and Townes3 have compared the coefficients of
higher powers of (r-r.) in the expansion of the energy
expression in terms of (r-re) as obtained from experi-
ment, with those from Rittner's theory. They found
satisfactory agreement considering the large un-
certainties in some of the experimental data. Varshni2
* National Research Council Postdoctorate Fellow. On leave
from the Department of Physics, Allahabad University Allaha-
bad, India. ' 10 M. Eisenstadt, G. M. Rothberg, and P. Kusch, J. Chern.
1 E. S. Rittner, J. Chern. Phys. 19, 1030 (1951).
2 Y. P. Varshni, Trans. Faraday Soc. 53,132 (1957).
SA. Honig, M. Mandel, M. L. Stitch, and C. H. Townes,
Phys. Rev. 96, 629 (1954).
582
A exp( -rip) and inverse power b/rn and calculated
the rotational constant a e and vibrational constant
w.x. for the two types of repulsion terms and found
that the exponential one was better. Similar results
were obtained by Klemperer and Rice. 4 5
Pauling6 has given a simple treatment of alkali halide
molecules taking into account only terms of the type
(a) and (e). However, Varshni and Shukla7 have shown
that this does not give satisfactory values of a. and
We X
Recently, Frost and Woodson8 have shown a very
interesting connection between the potential-energy
functions of two rare gas atoms and corresponding
alkali halides whose ions have similar closed shells.
As a first approximation they have put
U (alkali halide) = (-e 2/r) +U (between rare gas
atoms having some closed shells as alkali halide ions).
Considering the crude approximation, the results ob-
tained by Frost and Woodson are quite satisfactory.
One important limitation of this approach is that the
same curve is predicted for two alkali halide molecules,
such as KBr and RbCl, regardless of which inert gas
structure is contained in which ion.
In recent years the existence of polymeric species in
the vapors of alkali halides has been demonstrated in
molecular beam experiments,9-12 mass spectrometric
4 W. Klernperer and S. A. Rice, J. Chern. Phys. 26, 618 (1957).
5 S. A. Rice and W. Klemperer, J. Chern. Phys. 27,573 (1957).
6 L. Pauling, Proe. Natl. Acad. Sci. India A25, 1 (1956).
7 Y. P. Varshni and R. C. Shukla, Proc. Phys. Soc. (London)
76,794 (1960).
8 A. A. Frost and J. H. Woodson, J. Am. Chern. Soc. 80, 2615
(1958).
9 R. C. Miller and P. Kusch, J. Chern. Phys. 25, 860 (1956) j
27, 981 (1957).
Phys.29,797 (1958).
uP. Kusch, J. Chern. Phys. 30, 52 (1959).
12 M. Eisenstadt, V. S. Rao, and G. M_ Rothberg, J. Chern.
Phys. 30, 604 (1959).

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 POT E N T I A L FUN C T ION FOR AL K AL I HAL IDE MOL E C U L E S 583

studies,13-19 and other methods. 20 Correspondingly a TABLE 1.-

number of theoretical investigations have also been
made on the subject. O'Konski and Higuchi21 .22 were the w, k, T. B.
Molecules em-I (105/crn) (10- 8 em) em-I
first to investigate (NaClh using the Rittner model.
Milne and Cubicciotti23 have made calculations on LiF 906.2 g 2.4586 1. 545 g 1.378"
alkali halide dimer molecules using the Pauling; model. LiBr 576b 1.2478 2. 1704d 0.5555 d
LiI 501 b 0.9728 2.392d 0.4429 d
Berkowitz24 has extended the theory of O'Konski and
Higuchi and has calculated the molecular structure and NaCl 366b 1.1004 2.361d 0.2179 d
NaBr 302b 0.9591 2.502 d 0.1511 d
infrared-active vibrational frequencies of the dimer NaI 258b 0.7631 2.711 d O.l177d
molecules.
KF 400' 1.205 2.171 0.2799
Alkali halide ions of the type M2X+ (Milne and KCl 281 b 0.8648 2.667 d 0.1285 d
Cubicciotti25 ) and trimers (Milne and Cubicciotti26 ) KBr 213b 0.7015 2.821d 0.0811 d
have also been treated on the basis of the Pauling model KI (173)b 0.5269 3.048d 0.0608 d
for the monomer. RbF 390' 1.361 2.265 1 0.2107 1
In these studies the potential-energy function of the RbC] 228b 0.7673 2.787 d 0.0876d
RbBr (166)b 0.6703 2.945 d 0.0475 d
alkali halide monomer has been used as the starting RbI (128)b 0.4929 3.177d 0.0328 d
point. Investigations have been made without polariza-
CsF 385' 1.451 2.345 d 0.184d
tion terms (e.g., Pauling model treatments23 ,25,26) and CsC] 209b 0.7201 2.906 d O.072 d
also with polarization terms (O'Konski and Higuchi,22 CsBr (139)b 0.5679 3.072 d 0.036 d
Berkowitz24 ) . However, it is known that the Pauling CsI (101)b 0.3900 3.515 d 0.0236 d
model is not wholly satisfactory for the alkali halide
- ( ) indicate estimated values.
monomers (Varshni and Shukla7). On the other hand b See works cited in references 4 and 5.
there has been some disagreement as to the proper C See work cited in reference 32.

d See work cited in reference 3.

values of the polarizabilities of alkali and halide ions
that should be used in the extended potential-energy
function. Pauling27 has calculated the polarizability of
the free ions from the theory of the quadratic Stark
effect, and his values have been checked by Mayer
and Mayer28 using spectral data. Tessman, Kahn,
and Shockley29 have calculated the polarizabilities of
ions in ionic crystals from a study of the indices of
refraction of alkali halide crystals (see also Hanlon and
Lawson30 ). There are considerable differences between
these two sets of values. Klemperer and Rice4,6 have
argued that the polarizabilities of the ions used in
calculations of the energy curve should be chosen to
give agreement with the observed molecular dipole.
e G. W. Green and H. Lew, Can. J. Phys. 38, 482 (1960).
1 H. Lew, D. Morris, F. E. Geiger, and J. T. Eisinger, Can. J. Phys. 36, 171
(1958).
G. L. Vidale, J. Phys. Chern. 64, 314 (1960).
We have found that it is possible to account for the
potential-energy curve of the diatomic alkali halide
molecules by using only two tenus: an attractive term
[ -e2/r ] and a new type of repulsive term P exp( -kr2),
this type of tenu is often referred to as the "Gaussian
function." Besides its utility for the monomer, a simple
and compact function has obvious advantages in deal-
ing with dimers and other polymers. We have carried
out calculations for the following three potentials:

13 L. I<'riedrnan, J. Chern. Phys. 23, 477 (1955). U = (-e 2/r) +Be-r1u (Born-Mayer) (1)
14 N. A. lonov, Doklady Akad. Nauk. S.S.S.R. 59, 467 (1948).
16 T. A. Milne, H. M. Klein, and D. Cubicciotti, J. Chern. Phys.
U = - (e2/r) -[e2(al+a2) /2r4J- (2e2ala2/r7) - (c/r6)
28, 718 (1958).
16 J. Berkowitz and W. A. Chupka, J. Chern. Phys. 29, 653
(1958) .
+A exp( -rip) (Rittner) (2)
17 R. F. Porter and R. C. Schoonmaker, J. Chern. Phys. 29, 2/r)+P
1070 (1958). U= (-e exp( -kr2 ). (3)
18 R. C. Schoonmaker and R. F. Porter, J. Chern. Phys. 30,
283 (1959). In all the three cases we have determined the con-
19W. A. Chupka, J. Chern. Phys. 30, 458 (1959). stants, viz, B, rr, A, p, P, and k by using the conditions
20 A. C. Pugh and R. F. Barrow, Trans. Faraday Soc. 54, 671
(1958) .
21 C. T. O'Konski, J. Chern. Phys. 23,1174 (1955).
(dU/dr)r=r.=O (4)
22 C. T. O'Konski and W. 1. Higuchi, J. Chern. Phys. 23,1175
(1955) . (d2 U/dr2)r.=r.=k., the force constant. (5)
23 T. A. Milne and D. Cubicciotti, J. Chern. Phys. 29, 846
(1958) . We have referred above to the conflicting opinions
24 J. Berkowitz, J. Chern. Phys. 29, 1386 (1958).
26 T. A. Milne and D. Cubicciotti, J. Chern. Phys. 30, 1418 regarding the values of polarizabilities that should be
(1959) . used with such calculation,>. Berkowitz24 has found
26 T. A. Milne and D. Cubicciotti, J. Chern. Phys. 30, 1625
(1959) .
that the differences in polarizability do not change the
27 L. Pauling, Proc. Roy. Soc. (London) A114, 191 (1927). results very much, but somewhat better agreement of
28 J. E. Mayer and M. G. Mayer, Phys. Rev. 43, 605 (1933). vibrational frequencies was obtained with Pauling's
29 J. R. Tessman, A. H. Kahn, and W. Shockley, Phys. Rev.
92, 890 (1953). values of a. Hence we have also used the Pauling set of
30 J. E. Hanlon and A. W. Lawson, Phys. Rev. 113, 472 (1959). values as given in Rittner. The van der Waals attractive

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
584 Y. P. VARSHNI AND R. C. SHUKLA

TABLE II. Values of the repulsion constants. Calculations were made for the rotational constant
(X" vibrational constant W'X" and the binding energy
<T p x k
Di for the above three potentials. The method of calcu-
Molecules (10- cm)
8 (10- cm)
8 (10 16 cm-2) lating (x. and WeXe has been explained in Varshni. 31 The
relevant equations are
LiF 0.2605 0.2739 6.9305 1.4517
LiBr 0.2882 0.3435 7.5302 0.7993 (Xe= -[(Xr./3) +1](6BNw.) (7)
LiI 0.3078 0.3758 8.7714 0.7665
NaCl 0.2852 0.3328 9.2779 0.8322 W.Xe= [~-X2- Y]re2 (W /JLAr.2), (8)
NaBr 0.2939 0.3049 9.5118 0.7597 and
NaI 0.3156 0.3364 9.5909 0.6525
KF 0.2956 0.3184 8.3449 0.8853 Di= -U(r.), (9)
KC] 0.2927 0.3430 10.1115 0.7108
KBr 0.3196 0.3731 9.8266 0.6174 where W =2.1078X 10-16 and
KI 0.3600 0.4199 9.4676 0.5095
RbF 0.2558 0.2938 9.8555 0.9605 X = UIII (re) /Ull (r.)
RbC] 0.3029 0.3558 10.2002 0.6566
RbBr 0.3126 0.3709 10.4216 0.6008 Y = UIV (r.) /Ull (r.).
RbI 0.3589 0.4218 9.8514 0.4880
CsF 0.2319 0.2852 11.1109 1.0103 The experimental data used are recorded in Table I
CsC] 0.3008 0.3614 10.6604 0.6312 and are discussed below.
CsBr 0.3362 0.3964 10.137 0.5371 The expressions for (X" WeX., and D i by the three
CsI 0.3762 0.4414 10.3422 0.4185
potentials are summarized below.
POTENTIAL (1)
constant c can be obtained from
C=!(X1(X2[I2E/ (12+E)], (6) The potential (1) gives

where 12 is the second ionization potential of the alkali (X.=~ ~(r'/CT-3)6B/ (10)
atom, and E is the electron affinity of the halogen atom. 3 CT r./CT-2 We
Varshni2 has given the calculated values of "c" and
these have also been utilized here.
A number of earlier workers have made calculations =~[ (r./CT )4+3(r./CT ) 3-30 (r./CT )2+36 (r./CT ) +18]~
with the Rittner potential (2). Some of these were 3 (r./CT-2) 2 JLAr.2
based on the older data and in some it was assumed (11)
that p is constant for all molecules. As shown below, and
such an assumption is not quite correct. For these Di= (e2/r.) [1- (CT/r.)]. (12)
reasons we have performed completely new calcula-
tions with the best available data for the Rittner The values of CT are given in Table II and other results
potential also. are shown in Tables III, IV, and V.
RITTNER POTENTIAL (2)
Rittner's potential yields the following:

X= (13)

(14)

whence (x. and .x. can be calculated. Also2

Di=~{ 1-~)+ 2e (xl;(x2) (!_~)+ 14e2~1(x2(!_~)+ 6C6(!_~) (15)
;::\ r. re 4 r. r. 7 re r. 6 rc
values of p are recorded in Table II and other results are given in Tables III to V.
31 Y. P. Varshni, Revs. Modern Phys. 29, 664 (1957); 31, 839 (1959).

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 POTENTIAL FUNCTION FOR ALKALI HALIDE M OLE:C U LE S 585
TABLE III.

lQ4Xa. a.XIQ4 a.XI04 a.XIQ4

obs. calc. % calc. % calc. %
Molecules em-I pot. (1) error pot. (2) error pot. (3) error

LiF 197.1a 185.3 -6.0 145.3 -26.3 100.7 -48.9

LiBr 56.41 b 66.09 +17.2 50.41 -10.6 47.63 -15.6
LiI 4O.90b 50.31 +23.0 37.6 -8.1 37.05 -9.4
NaCl 16.1b 18.06 +12.2 14.9 -7.5 13.81 -14.2
NaBr 9.4b 10.89 +15.9 11.0 +17.0 8.45 -10.1
NaI 6.5 b 7.83 +20.5 7.67 +18.0 6.10 -6.2
KF 23.35 c 23.39 +0.2 19.91 -14.7 16.54 -29.2
KCl 7.90b 9.20 +16.5 7.70 -2.5 7.37 -6.7
KBr 4.05 b 4.65 +14.8 3.87 -4.4 3.67 -9.4
KI 2.70b 3.06 +13.3 2.51 -7.0 2.37 -12.2
RbF 15.18d 17.22 +13.4 14.49 -4.5 13.61 -10.3
RbCl 4.50b 5.33 +18.4 4.42 -1.8 4.29 -4.7
RbBr 1.86b 2.16 +16.1 1.83 -1.6 1.80 -3.2
RbI 1.10b 1.27 +15.5 1.04 -5.5 1.00 -9.1
CsF 11.00b 15.59 +41.7 12.49 +13.5 12.96 +17.8
CsCI 3.30b 4.17 +26.4 3.38 +2.4 3.41 +3.3
CsBr 1.20b 1.47 +22.5 1.19 -0.8 1.17 -2.5
CsI 0.68b 0.89 +30.9 0.75 +10.3 0.72 +5.9
Average % error 18.0 8.7 12.1
Average (excluding fluorides) 18.8 7.0 8.0

G. L. Vidale, J. Phys. Chern. 64,314 (1960).

b See work cited in reference 3.
e G. W. Green and H. Lew, Can. J. Phys. 38,482 (1960).
d Lew, Morris, Geiger, and Eisinger, Can. J. Phys. 36, 171 (1958).

POTENTIAL (3) Then we have

The results for the new potential (3) can be con-
a.=[(x2-6x+3) /3(x-3) ] (6BHw.) (17)
veniently expressed in terms of another parameter x
defined by w.x.=[2X
L
78x2-3X3+279X-36J~
(16) (18)
3(x-3)2 J.i.Ar.2

TABLE IV.

L.

/;.
0
we X" WeXe weXe weXe 0.44 o
Molecules (est.) calc. calc. calc. N.
em-I pot (1) pot(2) pot(3) + t<
.42
" Rb
LiF 7.9- 7.787 6.896 6.620 o Cs
LiBr 4.28 b 4.71 3.66 4.12 .40
LiI 3.35 b 3.97 2.997 3.49
NaCI 2.05b 2.14 1. 75 1.89 .38
NaBr 1. SOb 1.56 1.64 1.39
NaI 1.08b 1.23 1.25 1.10
P .36
KF 1.45e 2.25 1.99 1.96
. KCl
KBr
KI
RbF
RbCl
RbBr
1.30b
0.80b

1.3"
0.92b
1.47
0.885
0.621
2.33
1.02
0.576
1.21
0.727
0.507
1.95
0.830
0.465
1.32
0.792
0.553
2.09
0.916
0.520
.34

.32
:
~
,',:/
//
,,
RbI
CsF 1.23 e
0.360
2.54
0.292
1.97
0.323
2.31
.30
,, "
CsCI 0.75b 0.910 0.722 0.824 .28 o ,,"
CsBr 0.416 0.335 0.374 cf
CsI 0.253 0.208 0.229
0.26
F CL Br
a G. L. Vidale, J. Phys. Chern. 64, 314 (1960).
b See works cited in references 4 and 5. FIG. 1. Behavior of the constant p occurring in the Rittner po-
C See work cited in reference 32. tential (2). Broken line is interpolation.

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
586 Y. P. VARSHNI AND R. C. SHUKLA

TABLE V.

Molecules D.obs." D; calc. % Di calc. % Di calc. %

kcal/mole pot. (1) error pot. (2) error pot. (3) error

LiF 177.8 178.8 +0.6 185.8 +4.5 184.0 +3.5

LiBr 142.7 132.7 -7.0 142.1 -0.4 132.7 -7.0
LiI 130.7 121.0 -7.4 130.6 -0.1 123.0 -5.9
NaCI 127.9 123.7 -3.3 130.0 +1.6 125.5 -1.9
NaBr 124.4 117.2 -5.8 127.7 +2.7 118.8 -4.5
NaI 114.5 108.3 -5.4 118.9 +3.8 109.8 -4.1
KF 131.6 132.2 +0.5 136.5 +3.7 134.7 +2.4
KCI 113.3 110.9 -2.1 116.2 +2.6 112.2 -1.0
KBr 109.1 104.4 -4.3 109.5 +0.4 105.8 -3.0
KI 101.9 96.1 -5.7 101.1 -0.8 97.5 -4.3
RbF 135.8 130.1 -4.2 135.7 -0.1 131.8 -2.9
RbCI 111.1 106.2 -4.4 111.4 +0.3 107.5 -3.2
RbBr 106.7 100.8 -5.5 106.0 -0.7 102.0 -4.4
RbI 98.7 92.7 -6.1 97.6 -1.1 94.0 -4.8
CsF 133.3 127.6 -4.3 135.4 +1.6 128.9 -3.3
CsCI 107.6 102.5 -4.7 108.1 +0.5 103.6 -3.7
CsBr 103.2 96.3 -6.7 101.3 -1.8 97.5 -5.5
CsI 93.2 84.4 -9.4 88.3 -5.3 85.4 -8.4
Average % error 4.9 1.8 4.1

8 See work cited in reference 33.

and of lXe, WeXe, and binding energy D i are given in Tables

Di= (e 2/re) [1- (l/x)]' (19) III, IV, and V, respectively. The We values for KI,
RbBf, RbI, CsBr, and CsI given here are those esti-
Values of k and x are tabulated in Table II. Calculated mated by Rice and Klemperer. 5 Though experimental
results for lXe, WeXe, and Di along with percentage errors values for these molecules are available from the work
are shown in Tables III to V. of earlier workers, these estimated values are believed
DATA
to be more reliable and consistent. 5 It is very difficult
to evaluate WeXe from the experimental data, so investi-
The experimental data used in the calculations, gators have used semi empirical methods to e<;timate
together with the sources, are given in Table I. Values WeXe and to adjust We accordingly. Barrow and Caunt32
have estimated WeXe from the relation
1.5
q
xeMA!=0.013. (20)
\
o Li
1.4 \
\
No However, the constant in this relation IS rather un-
\
\
+ K certain and it may be as high a" 0.018.
1.3 \ A Rb
\
\ o Cs
On the other hand, Klemperer and Rice4 ,5 have used
\
1.2 \
\
the relation obtained from the Morse curve,
\
\
1.1
\
\ weXe=[(welXe/6) + B n2/Be 3, (21)
\
\

1.0
\
\ We have followed the policy of taking We and WeXe from
\
\ the same source.
k .9 \""" The binding energy (or the ionic dissociation energy)
D; was calculated from the relation

~"
.8
D;=De+I-E,
.7
~ No
where De is the dissociation energy (obtained from
Gaydon33 ), I is the ionization potential of the alkali

~:.
6
atom (taken from Herzberg 34 ), and E iE the electron
., affinity of the halogen atom (taken from Pritchard35 )
cCs 32 R. F. Barrow and A. D. Caunt, Proc. Roy. Soc. (London)
.4 A219, 120 (1953).
33 A. G. Gaydon, Dissociation Energies and Spectra of Diatomic
0.3 Molecules (Chapman and Hall, Ltd., London, 1953).
F CL 8r I
34 G. Herzberg, Atomic Spectra and Atomic Structure (Dover
FIG. 2. Behavior of the constant k occurring in the potential Publications, New York, 1944).
(3). Broken line is interpolation. 36 H. O. Pritchard, Chern. Revs. 52, 529 (1953).

Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
 POT E N T I A L FUN C T ION
Most of the De values are uncertain by about 4
kcal and cOI'sequently there are corresponding un-
certainties in the Di values.
DISCUSSION
It will be noted from Table II that neither ()" nor p
are constant for all of the molecules; ()" varies between
0.23 to 0.38 and p varies between 0.27 to 0.45. It is clear
that the earlier practice of assuming these to be con-
stant has to be abandoned. Baughan36 has al'lo recently
concluded that ()" varies by about 15% for different
molecules. The behavior of p has been shown in Fig. 1.
In a group of molecules having the same alkali atom, p
is seen to increase in the order of F, CI, Br, I, but the
Na series appears to be an exception.
It is interesting to note that parallel results have also
been obtained in the case of alkali halide crystals.
(Cubicciotti37 )
The behavior of the constant k has been shown in
Fig. 2. For a given alkali atom, k is found to decrease
as we pass from fluoride to iodide.
For a e : The fluorides appear to give high percentage
errors. Hence we have shown average percentage errors
for both cases: (i) including fluorides and (ii) excluding
fluorides.
From the average percentage errors it will be ob-
served that the resultL obtained with Eq. (3) are much
better than those obtained with Eq. (1) and only
slightly inferior to those obtained with potential (2).
For WeXe: As noted above, except for LiF, the re-
ported values are estimated ones and it is not possible
to draw any conclusions regarding the errors of the
calculated results from the three potentials, hence we
have not given the percentage errors. Suffice it to say
that the calculated values by the three potentials
usually lie in the following order: pot. (1) > pot. (3) >
pot.(2).
Binding Energy D i : We have noted above that most
of the experimental values are uncertain by above 4
kcal, which corresponds to ,,-,3%. As the percentage
errors using the various functions are also of the same
order, no strict conclusion can be drawn about the
relative performances. Nevertheless it may be seen
that the average percentage error using potential (3) is
less than that using potential (1) but greater than that
using potential (2).
To summarize, we may say that the results obtained
with potential (3) are much better than those obtained
with potential (1) but slightly inferior to those ob-
tained with potential (2).
Certain points regarding the form of the repulsive
term may be noted:
As compared to the inverse-power term, the ex-
potential term is certainly superior as has been found
by Varshni2 and by Rice and Klemperer. 5
Another argument that has sometimes been ad-
36 E. C. Baughan, Trans. Faraday Soc. 55, 737 (1959).
37 D. Cubicciotti, J. Chern. Phys. 31, 1646 (1959).
Downloaded 29 Mar 2012 to 200.16.16.13. Redistribution subject to AIP license or copyright; see http://jcp.aip.org/about/rights_and_permissions
FOR ALK ALI HAL IDE MOL E C U L E S 587
vanced in support of the exponential term is that the
quantum mechanical calculations predict such a term.
However, it should be noted that the situation on this
point is not so simple. About 30 years ago, Unsold 38
and Bruck39 showed that the repulsive energies between
a point cation and a closed-shell anion should be of the
form IT(r) exp( -')'r) where IT(r) is a polynomial in r.
Approximation of such an expression merely by an
exponential term may not be quite satisfactory.
Further evidence on this point is available from the
quantum-mechanical calculations of the mutual po-
tential between two atoms having a rare-gas electron
configuration. The He-He repulsive potential was first
calculated by Slater4 and later by several others. A
recent treatment is due to Sakamoto and Isiguro. 41
Bleick and Mayec42 have used a generalized Heitler-
London method to compute the repulsive potential
between two atoms (or ions) having closed-shell con-
figurations of eight electrons each, and have reported
numerical results for two neon atoms. Their method
ha'l been used by Kunimune43 to calculate the re-
pUlsive potential in the case of two argon atoms. The
theoretical results of Sakamoto and Isiguro, Bleick
and Mayel, and of Kunimune were plotted as logY
versus r. The points in each case could reasonably ap-
proximated by a straight line, indicating that the
exponential term was satisfactory. However, it may
also be pointed out that some of the theoretical results
are in serious disagreement with the experimental data
(e.g., see the comparison of the theoretical and experi-
mental results in the paper of Sakamoto and Isiguro).
The fact that pot.(3) gives a better agreement with
experiment than the Born-Mayer potential, does not
nece~f,arily mean that a Gaussian term more accurately
describes the repulsive part of the potential than
does a simple exponential term. Rather, the Gaussian
repulsive term appears to simulate the combined effect
of the terms (b), (c), and (e).
The available evidence suggests that the simple
exponential term is satisfactory for the repulsive
potential alone, while the Gaussian more adequately
(or fortuitously) takes into account polarization and
van der Waals effects.
ACKNOWLEDGMENTS
The authors are thankful to the Council of Scientific
and Industrial Research (India) for financial as-
sistance.
One of the authors (Y.P.V.) is highly grateful to
Dl. Ta-You Wu and Dr. G. Herzberg, F.R.S. for their
kind interest in the work and further thanks the
National Research Council for the award of a Post-
doctorate Fellowship.
38A. Unscild, Z. Physik 43, 563 (1927).
39H. Briick, Z. Physik 51, 707 (1928).
40 J. C. Slater, Phys. Rev. 32, 349 (1928).
41 M, Sakamoto and E. Isiguro, Progr. Theoret. Phys. (Kyoto)
15,37 (1956).
42 W. E. Bleick and J. E. Mayer, J. Chern. Phys. 2,252 (1934).
43 M. Kunimune, Progr. Theoret. Phys. (Kyoto) 5,412 (1950).