THEO

CHEM
ELSEVIER Journal of Molecular Structure (Theochcm) 340 (1995) 63-71

XCXlocal spin density energy calculations

E. Vauthiera, M. Blaina, S. Odiot”, V. Baroneb,‘, M. Comeaub, S. Fli&rb,*
“D4partenzeni de Recherches Physiques,
Lahorutoire associi au CRNS, Universitb Pierre et Marie Curie, Tour 22, Sitmr,
4, Place Jussieu, Paris, France
hD4partement de Chimir, Universit4 de Mont&al, C.P. 6128, Sum. A, MontrPal, Qu&x H3C 3J7, Canada

Received 5 December 1994: accepted 20 February 1995

Abstract

Possible applications of density functional theory are examined with the intent of obtaining good-quality thermo-
chemical information, such as atomization energies and enthalpies of formation. The local spin density (LSD) approx-
imation is used. Exchange and correlation are treated in the Xol approach, where LYis a variable parameter. The tentative
hypothesis is that the Xol(LSD) method basically meets the required demands of accuracy, i.e. that the energy should be
correct, if Q is properly selected. A simple recipe for obtaining cy is suggested by the idea that non-local exchange-
correlation contributions should change in a more or less regular fashion when atoms join to form a molecule, i.e. in a
fashion largely dictated by the number and type of bonds that involve the individual atoms. Applications presented for
organic and inorganic systems using 6-3 lG** wavefunctions reveal a reasonable agreement, with a standard deviation of
4.4 kcal molY’, between calculated and experimental standard gas-phase enthalpies of formation.

1. Scope power of the number of electrons) is so heavy that

their use for chemically interesting molecules
A major objective of quantum chemistry is the appears to be impractical. Under such circum-
ab initio prediction of molecular structure and stances, group and bond equivalents have been
energy within chemical accuracy for systems of used [l] to correct binding energies obtained from
reasonable dimensions, i.e. not restricted to two Hartree-Fock computations with medium size
or three atoms. Despite impressive developments basis sets.
of algorithms and computer power, this target From another point of view, methods based on
is still unreached. In fact, sophisticated post- the density functional approach are more promis-
Hartree-Fock methods using huge basis sets are ing since they are able to include a large amount of
providing reliable results for small model systems, correlation energy in a formalism which requires in
but the scaling of the computational costs with the essence the same resources as the Hartree-Fock
dimensions of the systems (typically with the 7th (HF) method: methods rooted in density func-
tional theory (DFT) [2,3] are by far the most useful
non-empirical alternatives to configuration inter-
’ Visiting scientist, on temporary leave from the Dipartimento
di Chimica, liniversiti di Napoli ‘Federico II’, Via Mezzo-
action calculations. Surely, inclusion of generalized
cannone 4, I-80134, Naples, Italy. gradient corrections is pushing DFT methods to
* Corresponding author. achieve impressive levels of accuracy, but the

0166-1280/95/$09.50 c 1995 Elsevier Science B.V. All rights reserved

SSDI 0166-1280(95)04176-l
64 E. Vuuthier et al.,lJournal of Molecular Structure (Theochem) 340 i 19951 63-71

so-called “chemical accuracy” (average errors of N z 0.7 advocated by Baerends and Ros [9]
approximately 2.5 kcal mall’) is generally still out for molecular calculations. The familiar exchange
of reach. The local spin density (LSD) approxima- parameters given by Schwarz [lo], e.g. 0.77725 for
tion (or its Xa variant) corresponds in the DFT hydrogen and 0.75928 for carbon, obtained from
world to the same reference model as the HF one an adjustment of the statistical total energies to
in conventional computational chemistry. However, the HF total energies of the isolated atoms, also
LSD usually proves to be a much better starting fall in that range.
point than HF. We thought, therefore, it interesting With reference to the selection of the appropriate
to examine whether introduction of some sort of ~1:parameter, two approaches have been presented
“bond corrections” could lead to results compar- for use in polyatomic systems [ 111. The simplest
able to, or better than, those obtained starting one, rooted in the Q reproducing the conventional
off with HF computations. With this in mind, we HF SCF energies of the isolated atoms, leads to
shall examine the applicability of DFT to organic meaningful results in Xcy computations of molecu-
and selected inorganic systems, particularly to the lar properties which are not significantly affected
study of their atomization energies and enthalpies by an uncertainty of, say, 0.005 in the estimate of
of formation, paying attention to both the practical- o, namely, internuclear distances, vibrational fre-
ity and the physical content of approximate quencies and applications of Xa theory to energy-
solutions. The results reported in this study show, band calculations [6]. The second one, for use in
in our opinion, that these goals can be reached accurate calculations of dissociation energies,
and that we have at our disposal a method that requires a refinement which considers that the cys
can be routinely applied to provide reasonably to be used for the isolated atoms and the same
reliable thermochemical data for molecules of atoms incorporated into a molecule or a crystal,
large dimensions. though very similar, are not identical.
Here we use the LSD approximation [4] and The present work expands these refinements to
expand the molecular orbitals as a linear combina- complex polyatomic molecules in order to evaluate
tion of Gaussian-type orbitals (LCGTO) for solv- the potential of the simple Xcu approach in the
ing the LSD equations. The specific form of the study of dissociation energies. “Bond correction”
latter depends on the treatment of exchange and terms shall be postulated with the intent to predict
correlation in electron-gas calculations. A simple cy parameters that are sufficiently accurate to
approach is made possible by the use of Slater’s permit adequate energy calculations. Briefly, the
Xo exchange energy [5,6] idea is to explore the limits of simple LSD theory
and to extract whatever may facilitate its applica-
EXcv=-$o
0
81 ‘I3
__
47r s
(P;” + pii3)dy
tions to thermochemical problems.

where (Y is treated as a variable parameter. This 2. Outline of calculations

approximation heavily depends on the appropri-
ate selection of 0. Slater’s original treatment [5] In the present LSD calculations, the molecular
of the exchange potential amounts to taking orbitals are expanded as linear combinations
cy = 1 but Gaspar [7] and later Kohn and Sham of Gaussian-type orbitals. In this LCGTO-
[3] showed that Q = 5 surely is more appropriate. LSD(Xa) approach [12,13], full advantage is
The more recent treatment of the exchange- taken of the standard techniques of traditional
correlation potential by Gaspar and Nagy [8] ab initio calculations, namely as regards the
represents a generalization of the Slater and optimization of basis sets. Pople’s 6-3 lG** orbital
Gasp&KohnSham approaches and contains basis [14] was selected. The auxiliary basis for
those as special cases. The theoretical a’s deduced H was that of Dunning [ 151 plus Dunlap and
by Gaspar and Nagy are intermediate between coworkers’ standard 5p functions [13]. Finally, we
$ and 1 and offer a justification for the values used two s-type bond-centered auxiliary functions,
 E. Vauthier et al.JJournal of Molecular Structure (Theochem) 340 (1995) 63-71 6.5

Table 1 sufficiently accurate in energy calculations’. Thus

Xo(LSD) calculations reproducing the 6-31G** Hartree-Fock we consider replacing o+ by ok + Acuk where Aok
energies of selected atoms
“measures” how ok changes when atom k becomes
Atom aE,/L+a (a.u.) part of a molecule. So we get a0 + ACY’ instead of
EHF (a.=)
cy’, with
H -0.498233 0.117222 -0.405725
C -37.680860 0.754462 -6.716261
N -54.385442 0.746176 -8.854275
0 -74.183934 0.742676 -11.027134

following Dunlap et al. [13]. All the calculations are The problem is now to find the appropriate
completely spin unrestricted, meaning that for values of Aok, or, in other words, to obtain CY
spin-polarized systems different exchange poten- parameters defined for “atoms in the molecule”.
tials are used for spin-up and spin-down orbitals. To be of any practical value, these Aok corrections
Full details regarding the actual computations, should be transferable from molecule to molecule,
following in essence the methods developed by at least to a very good approximation’. Here we
Dunlap et al. [ 131, are given in Ref. [ 111. consider that the Aak applicable to an atom k in
a molecule depends on the immediate environment
2.1. The selection of a of that atom, i.e., on the type and number of bonds
it forms with its neighbors. So we define a new
The 6-31G** HF SCF energies obtained for quantity, Affkl, as the part contributed to Aok by
H, C, N, and 0 atoms are indicated in Table 1. the k-l bond and search for a weighted sum of
The LCGTO-LSD(Xa) total energies, however, these Aok, values over all atoms I bonded to k to
depend on the selection of cr which enters Slater’s get the Acrk value we are looking for. We opted for
exchange term, Eq. (1). The final value obtained the simplest possible approach: in a first step we
for E, the energy of an atom or a molecule, thus used the bond dissociation energies, Dk[, of the
depends on CY,with diatomics CH, NH, OH, CC CN, CO, NN, and
NO and calculated the corresponding derivatives
aE
-=- Exa
(2) aEk,/&%. Now, of course, except for the diatomics
da cy
themselves, the intrinsic energy, &k[,of a k-l bond
The “atomic” cy values an, cyc, . . ., ak, reproducing in a molecule at equilibrium differs from Dkl [17].
the HF SCF energies of the isolated atoms are So, rather than giving a weight of dEkl/da to the
given in Table 1, along with the corresponding individual Values Of Aok_, we used Tk[(d_&[/&)
derivatives dEH/&, dE,-/&, ., a&/&t. instead, with
For polyatomic systems, one conveniently
defines [11,16] a “molecular cy”, written here Q’, ykl 2/

(3) ’ In series of closely related compounds, however, such as

the alkanes, for example, an appropriate reparametrization
permits accurate energy calculations using o” parameters
given by Eq. (3) [16].
These a0 parameters are reasonable first approxi- ’ The Aok corrections selected once and for all for H, B, C, N,
mations. The idea behind this definition of a 0 and F in the molecule lead to an acceptable overall agreement
“molecular Q” is that it should not be too inaccu- with experimental dissociation energies of all the diatomics
rate to use, in essence, the same (Y for an isolated formed by these atoms [ll]: the approximate transferability of
these Aok parameters is noteworthy, considering the extreme
atom as that for the same atom in a molecule or
diversity of the bonding combinations covered by this selection
crystal [6]. There are reservations, however. The of molecules. The constraints considered in the present
cy3s given by Eq. (3), although adequate for use approach are somewhat less stringent, however, as explained
in geometry calculations [l I], are generally not in the text.
66 E. Vauthier et al.iJownal of Molecular Structure (Theochem) 340 (1995) 63- 71

Table 2
Bond-energy parameters, with Dk, and tk, in kcal mol-’ and ~k,(aEki/~~) in atomic units

k-l Bond &I fk/ ?k,(aEk//aCY) &l(k-l) Aa(l-k)

H-H 109.54 -0.865071 0.049885

Cm H 83.94 110 -9.449284 0.003935 0.023751
N-H 83.02 100 - 11.207587 0.000861 0.050548
O-H 106.54 115 -12.516125 -0.000733 0.053391
c--c 145.75 80 -7.490820 -0.004252
c=c 140 -13.108934 -0.000399
CEC 200 - 18.727049 0.001483
C: : :C 110 -10.299877 -0.002497
C-N 181.95 80 -6.898235 PO.006065 -0.001357
C=N 130 - 1 I .209636 0.002692 -0.002254
C=N 201 -17.331822 0.000504 0.000837
C-0 258.98 80 -5.582534 0.047818 0.006075
c=o 180 -12.560701 -0.003603 -0.001025
N-N 228.54 40 -3.106665 -0.002016
N=N 90 -6.989996 0.000475
NEN 156 -12.115994 0.001922
N-O 152.43 10 -9.185623 0.0075 15 -0.007166
N=O 115 - 15.090667 0.003344 -0.004010

The yk[ factor put in front of d&[/ao! tempers, in one obtains ACQ = 0.002224 for the ethane carbon
a way, the weight of this derivative by the impor- atom, or 0.002224 and -0.001828 for the primary
tance of the k-l bond in the molecular fabric. So and tertiary carbons of isobutane, respectively. For
we get the hydrogen atoms of these molecules, one finds
ACY~ = 0.023751. It is now easy to apply Eq. (4)
using the derivatives of Table 1. In these examples,
we obtain Acu” values of 0.005526 and 0.004168 for
(6) ethane and isobutane, respectively. The cv” values,
Eq. (3), calculated with the help of Table 1, com-
plete the information regarding the selection of
Table 2 reports experimental Dkl results [11,18] the appropriate N parameters.
and rough estimates of intrinsic ek/ values [ 1713.
(Note that both Dkl and tk[ are for molecules at
equilibrium, at 0 K, in their hypothetical vibration- 2.2. Xcl energy calculations
less state.) Using the calculated dEk,/dcr values we
thus obtain the ‘-,‘k[(a&,/&%) values indicated in All calculations are made for molecules at 0 K
Table 2. Finally, we deduce the Acvkl parameters when all the atoms are stationary in their equi-
from fits with ACY@values that reproduce experi- librium positions. The total ground-state energy
mental energies [I 11. At this point we have all is then
that is required to solve Eq. (4).
With the help of Table 2 and Eq. (6) we gain Emo’ = c ,Jgom _ AE;
access to the Aok values of interest. For example, k

-- where Eirom 1s the ground state energy of the

’ The intrinsic energy of any given bond is not a constant [17]:
isolated atom k - taken at its 6-31G** value indi-
it depends on the environment of that bond in its host molecule.
The values reported here are rough averages which. though
cated in Table 1, and AE; is the atomization
lacking precision, are adequate in our intended applications of energy of that molecule. AEZ is most convenient
Eq. (6). for comparisons with experimental results, using
 E. Vauthier et al./Journal of’kfolecular Structure (Theochem) 340 i 1WS) 63- 71 6-l

Table 3
Selected detailed examples of Xo calculations of atomization energies. a.u.

Molecule 0’ A@< (iX”“‘/do) AE,(NI) AE:(Xo) AZ

Methane 0.758891 0.00779 I -8.860425 0.648302 0.600229 0.669264

Ethane 0.757954 0.005526 --16.831415 1.128735 I .040684 1.133703
Butane 0.757448 0.004219 -32.771668 2.098082 1.930184 2.068445
Ethene 0.756915 0.004487 -15.853189 0.904277 0.832274 0.903412
Benzene 0.755759 0.000905 -44.9705 15 2.333262 2.133815 2.174531
Nitrobenzene 0.750541 -0.000780 -75.810303 2.742281 2.526237 2.467132
Methylamine 0.752917 0.004898 -18.330679 0.897928 0.834834 0.924683
1.2-Dimethylhydrazine 0.752343 0.003502 -35.715713 1.606417 I .492567 1.617632
Aniline 0.754294 0.001028 -54.468415 2.614624 2.397615 2.453634
Dimcthyl ether 0.751690 0.002353 -28.071998 1.305261 I.205096 1.271167
Diethyl ether 0.753565 0.002358 -44.094689 2.294534 2.108110 2.212108
Furan 0.752108 -0.000167 -41.219447 1.735652 1.590028 1.583162
Ammonia 0.749928 0.006866 -10.372193 0.427358 0.402318 0.473538
Water 0.745044 0.002977 -12.219109 0.363202 0.333381 0.369747
Carbon dioxide 0.745427 -0.001627 -29.405046 0.722433 0.672451 0.624612

the familiar relationship [17] say the least, as regards the amount of non-local
exchange corrections it involves [ll]. Rather
than AEi(ac), one should definitely consider the
following appropriately redefined dissociation
+ ZPE + (H, ~ H,) - ArHe (8) energy

where ArH” IS the standard enthalpy of forma- AE;(Xa) = AE;((v’)

tion of the molecule in the gas phase and
ArHfi(Ak) that of the isolated atom Ak, with
nk = number of atoms Ak found in the mol-
ecule. We have used [19] ArH*(H) = 52.09,
ArHe(C) = 170.89, ArH”(N) = 113.0, and
ArH*(O) = 59.54 kcal atg-I . ZPE is the zero-
which ensures that the exchange correction assumed
point energy of the molecule (usually evaluated in
for a molecule exactly equals, and thus cancels, that
the harmonic oscillator approximation) and
computed for the sum of its constituent atoms4
(H, - Ho) its heat content. The most important
[l 11. The term that measures by how much the
source of errors accompanying the evaluation of
non-local contributions to the exchange-correlation
AE; originates from the experimental uncertainty
are modified when the atoms join to form a molecule
associated with AfH fs. Possible errors associated
with ZPE + (HT - Ho) are expected not to exceed
’ In LSD calculations, with (1 = 2/3 everywhere. the non-local
about 0.2 kcal mol-‘.
exchange corrections assumed for a molecule are also identical
The Xn energy calculations were made in the to those taken for the sum of its atoms. namely 0. Hence, it must
6-31G** basis using the o” values given by Eq. (3) be anticipated that AE:(XQ) 2 AE:(LSD), which is, indeed,
and the parameters of Table 1. From Eq. (7) we the case except for minor differences which are fully accounted
for by the fact that the i)E/& functionals depend on (Y[ 1 I]. It is
get 1111
important to remember that AE:(LSD) is defined with respect

AE;(a”) = c E;“‘m(~k) ~ Emo’(a”) (9)

to “atoms” of much lesser quality than HF atoms, whereas our
AEl(Xo) is defined with respect to HartreeeFock atoms, We
!i could have selected esact atoms instead, with no change in
philosophy. and still obtained virtually the same AE;(XN)
The physical interpretation of AEz(a”) is fuzzy, to values [ll].
68 E. Vauthier et al./Journal of Molecular Structure (Theochem) 340 (1995) 63-71

Table 4
Xa(LSD) results for cYE”“‘/&Y (a.u.) and AE: (kcal mol-‘) and a comparison between calculated and experimental standard enthalpies
of formation, ArHO (kcal mol-‘)

Molecule ACU” i3Emo’l& AE; ZPE+H,-Ho AfHe

Calcd. Exptl

H2 0.049885 -0.86507 109.54 8.65 [20] 0.00 0.00

c2 -0.000399 -13.64734 130.89 5.02 [21] 212.95 200.22 [21]
N2 0.001922 -17.74993 228.54 5.74 [18] 0.00 0.00
02 -0.002824 -22.30254 120.15 4.63 [18] 0.00 0.00
CH 0.005063 -7.21066 89.04 6.45 [18] 137.30 142.40 [21]
NH 0.003038 -9.30454 82.74 7.06 [22] 90.28 90.00 [22]
OH 0.001188 -11.59537 106.41 7.71 [22] 9.45 9.32 [22]
CN 0.000693 -15.68918 181.93 5.33 [18] 104.01 103.99 [21]
NO -0.000734 -20.00235 153.04 5.09 [21] 20.97 21.58 [21]
CH2 0.006071 -7.77830 180.80 12.58 [21] 102.40 92.35 [21]
CH3 0.006975 -8.33677 303.33 20.66 [23] 38.57 34.00 [23]
HCrC 0.002535 -14.32784 283.10 13.54 [24] 119.87 114.00 [21]
CH2=CH 0.003613 - 15.28694 455.50 23.20 [25] 58.35 63.45 [21]
C2H5 0.004782 -16.31506 606.06 39.17 [23] 24.97 28.30 [23]
Methane 0.007791 -8.86043 419.97 29.50 [26] -18.62 -17.89 [27]
Ethane 0.005526 -16.83142 711.41 48.02 [26] -20.92 -20.04 [26]
Propane 0.004662 -24.80095 1004.53 65.94 [26] -25.50 -25.03 [26]
Butane 0.004219 -32.77167 1297.97 84.38 [26] -29.87 -30.05 [26]
Isobutane 0.004168 -32.77178 1299.70 83.89 [26] -32.09 -32.09 [26]
Cyclopropane 0.002841 -23.88607 846.07 51.9 [28] 17.71 12.73 [27]
Cyclobutane 0.002841 -31.85632 1139.66 70.4 [28] 13.24 6.78 [27]
Ethene 0.004487 -15.85319 566.90 33.36 [28] 7.71 12.45 [27]
Propene 0.003865 -23.83240 861.86 5 1.82 [28] 1.84 4.88 [27]
Cyclobutene 0.002119 -30.87806 998.48 55.16 [29] 37.97 37.45 [27]
trans-1,3-Butadiene 0.002965 -30.82636 1020.24 55.19 [30] 16.24 26.33 [31]
tram- 1,3-Pentadiene 0.002895 -38.79675 1313.23 73.32 [32] 12.00 18.77 [31]
Isoprene 0.002840 -38.79959 1312.55 72.85 [28] 12.22 18.06 [27]
Cyclopentadiene 0.001667 -37.91389 1183.77 59.29 [33] 26.22 31.94 [27]
Allene 0.002980 -22.85654 716.97 36.20 [28] 29.89 45.63 [27]
Benzene 0.000905 -44.97052 1364.54 66.22 [28] 21.79 19.81 [27]
Ethyne 0.003527 -14.88358 412.13 18.57 [28] 46.47 54.34 [27]
HCN 0.001788 -16.32817 311.40 12.28 [21] 32.41 32.30 [21]
CH3-CN 0.001915 -24.33478 612.01 30.45 [34] 20.60 20.9 [35]
Ammonia 0.006866 -10.37219 297.15 22.94 [36] -10.86 -11.02 [35]
Methylamine 0.004898 -18.33068 580.21 41.48 [28] -4.76 -5.50 [27]
Ethylamine 0.004269 -26.29609 875.11 59.41 [28] -11.10 -11.35 [27]
Dimethylamine 0.004047 -26.28333 865.51 59.44 [28] -1.47 -4.66 [27]
Isopropylamine 0.003880 -34.30104 1171.09 77.20 [28] -18.67 -20.02 [27]
Aniline 0.001028 -54.46842 1539.68 75.17 [37] 17.72 20.81 [27]
Pyrimidine 0.000371 -47.88715 1104.76 49.40 [38] 47.75 46.99 [27]

is written Acx~(~E~~‘/~Q),,,~, in XQ language. So 3. Results

we get the formally exact solution AElexact = AE:
from Ref. [l l] Table 3 reports detailed results for selected
examples. The (Y’ parameters are those given
AE; = AE;(XCX) - Aly”(dEmol/dc&=,~ (11) by Eq. (3) and Table 1 and the Aa’ are deduced
Eqs. (9)-(11) conclude the description of our Xa from Table 2 with the help of Eqs. (4) and (6).
energy calculations. The AE~(cx’) energies, Eq. (9), and the t3Emo'/da
 E. Vauthier et al./Joumal of Molecular Structure (Theochem) 340 (1995) 63-71 69

Table 4 continued

Molecule Aff” l3E ““‘/& AEI; ZPE+H=-H0 AfHe

Calcd. Exptl.

Pyrazine 0.000371 -47.88023 1102.45 49.07 [28] 49.73 46.86 [27]

Pyridazine 0.000571 -47.84975 1090.59 49.16 [28] 61.67 66.52 [27]
Pyridine 0.000844 -46.43 167 1240.83 57.29 [28] 28.07 34.55 [27]
Pyrrole 0.001161 -39.48090 1073.59 53.08 [39] 21.69 25.88 [27]
Imidazole 0.000785 -40.93759 947.87 46.30 [40] 32.13 40.38 [41]
Tetrazole 0.000215 -43.75752 666.97 31.93 [42] 81.66 79.94 [43]
co2 -0.001627 -29.40505 391.95 9.78 [44] -96.64 -94.05 [35]
H-CHO 0.000764 -19.11889 374.78 18.62 [36] -27.47 -25.95 [27]
CH,-CHO 0.001268 -27.08323 675.97 36.65 [26] -40.01 -39.73 [27]
CH3CH2-CHO 0.001621 -35.05504 968.10 54.95 [45] -43.21 -45.45 [27]
CH,(CH&CHO 0.001844 -43.10456 1261.50 73.25 [26] -47.68 -48.95 [27]
CH3-CO-CH? 0.001482 -35.06293 975.02 54.59 [26] -50.49 -51.90 [27]
CH3-CO-C2H5 0.001731 -43.03605 1265.91 72.89 [26] -52.46 -57.02 [27]
NO2 -0.001903 -31.33568 232.24 7.71 [44] 3.10 7.93 [35]
W’3 -0.001692 -5 1.27874 394.20 14.06 [21] 17.08 19.80 [21]
Nitric acid -0.001574 -42.93330 393.14 19.03 [46] -37.81 -32.10 [47]
CH3-N=O 0.000522 -28.46218 499.52 29.44 [48] 20.73 16.70 [49]
Nitromethane -0.000627 -39.84058 597.64 33.63 [50] -15.14 - 17.86 [27]
Nitroethane -0.000071 -47.66366 89 1.80 51.84 [51] -20.46 -24.38 [27]
2-Nitropropane 0.000299 -55.63669 1186.66 69.00 [52] -27.53 -33.21 [27]
Nitrobenzene -0.000780 -75.81030 1548.15 67.63 [53] 16.61 16.13 [54]
NH=CH-NO2 -0.000769 -48.15781 626.52 30.59 [55] 13.85 13.14 [55]
CH2=N-NO2 -0.000849 -48.11286 609.39 29.22 [55] 29.61 28.93 [55]
CH,=N(O)-OH -0.000492 -39.74198 585.82 33.46 [55] -3.49 -2.90 [55]
truns-HO-CH=N-OH -0.000056 -39.80939 620.94 33.78 [55] -38.29 -36.93 [55]
Hydrazine 0.004711 - 19.88393 438.53 33.94 [28] 20.89 22.80 [35]
Methylhydrazine 0.003937 -27.16126 730.3 1 52.00 [28] 17.79 22.60 [27]
I,1-Dimethylhydrazine 0.003398 -35.71134 1016.32 69.98 [28] 20.38 20.04 [27]
I ,2-Dimethylhydrazine 0.003502 -35.71571 1015.08 70.20 [28] 21.85 21.99 [27]
CH2=N-NH, 0.002635 -26.74737 587.88 37.82 [55] 44.82 40.32 [55]
cis-HN=NH 0.002896 -18.74179 292.08 18.81 [21] 50.99 50.90 [21]
tram-CH3 -N=N-CH3 0.002284 -34.66393 875.98 55.12 [56] 44.65 43.80 [57]
Urea 0.001183 -38.29511 748.11 42.78 [58] -52.39 -58.7 [27]
HN=CH-NH2 0.002849 -26.81879 622.92 38.24 [55] 10.20 11.02 [55]
H-0 0.002977 -12.21911 232.02 15.25 [44] -57.50 -57.80 [35]
Dimethyl ether 0.002353 -28.07200 797.67 52.55 1281 -44.59 -43.99 [27]
Methyl ethyl ether 0.002356 -36.02495 1092.16 70.19 [28] -50.81 -51.72 [27]
Diethyl ether 0.002358 -44.09469 1388.12 87.83 [28] -58.50 -60.26 [27]
Methyl n-propyl ether 0.002443 -43.99810 1388.38 87.83 [26] -58.77 -56.82 [27]
Furan -0.000167 -41.21945 993.45 45.50 [59] -9.82 -8.30 [27]
s,ck-Methylnitrite -0.000677 -39.67740 595.77 32.64 [60] -14.25 -15.31 [55]
s,trans-Methylnitrite -0.000677 -39.67341 592.33 32.17 [60] -11.28 -15.79 [27]

derivatives obtained for a” are indicated, as well the standard enthalpies of formation, Eq. (8).
as the corresponding AEz(Xcu) results deduced The relevant ZPE + HT - Ho energies are also
from Eq. (10). Finally, the theoretical atomization reported, as well as comparisons between
energies, AEZ, are from Eq. (11). calculated and experimental ArHo = AHr(gas,
The same protocol was followed to obtain 298 K) values. The root-mean-square error is
all the AE: results indicated in Table 4 and 4.4. kcal mol-‘.
70 E. Vauthier et al.:Journal of‘ Molecular Struc.iur.e ( Theorhem) 340 i 1995) 63-71

4. Conclusions of Eq. (6) and with the selection of basis sets.

As things are, here, Table 4 is surely capable of
Reasonable estimates of the (Y parameters for orienting us efficiently as to the quality of the
use in molecular XN calculations are most readily predictions regarding specific classes of molecules.
obtained from “atomic” ak values determined The underlying physical picture says that when
for the isolated ground state atoms k. These atoms join to form a molecule non-local exchange-
approximate “molecular” c): parameters are not correlation contributions seem to change in a
sufficiently accurate in energy calculations, how- more or less regular fashion, largely dictated by
ever. The ok values appropriate for isolated atoms the number and type of bonds that are created.
must give way to modified values, ok + Aok, These views could pave the way to further approxi-
appropriate for bonded atoms and the molecular mations capable of permitting simple but accurate
o must be recalculated using the modified future applications of Xcu theory. Yet we have to
o?k + A0k parameters. keep in mind that in the one-electron approxima-
The obvious question is: should the correction, tion, which is precisely what we are using here,
Aok, always be the same for a given atom k, only an approximate treatment can be attempted
regardless of its host molecule? (For example, along the philosophy suggested by this (or by any
should we put the C atoms of ethane, ethene, other) picture.
ethyne, etc., on the same footing, ignoring
hybridization changes, and, similarly, should the
same AN, apply to primary, secondary. tertiary, Acknowledgments
. alkane carbon atoms. disregarding any possi-
ble involvement of their differing bonding?) Our The financial support offered by le Fonds pour la
approach takes hybridization and bonding into formation de chercheurs et l’aide a la recherche
consideration and exchanges fixed, transferable (FCAR) Quebec, and by the National Sciences
“atomic” Acrk corrections for transferable Aak, and Engineering Research Council of Canada is
bond corrections (which seems sound) and sug- gratefully acknowledged.
gests a recipe to obtain Affk from a weighted sum
of Aa,, terms. Finally, we get AN? from these
individual Acvk corrections, i.e. the total correc-
References
tion applicable to the molecule that is needed for
the calculation of its energy. [l] M.J.S. Dcwar and B.M. O’Connor, Chem. Phys. Lett., 138
It turns out that our Aru” corrections represent (1987) 141. K.B. Wiberg, J. Comput. Chem., 5 (1994) 197.
less than 1% of the o3 value, x 0.75 i 0.007, given [2] P. Hohenberg and W. Kohn, Phys. Rev. B. 136 (1964) 864.
by the ok of the isolated atoms, thus suggesting [3] W. Kohn and L.J. Sham, Phys. Rev. A. 140 (1965) 1133.
[4] U. von Barth and L. Hedin, J. Phys. C., 5 (1972) 1629.
that Q@ surely represents a valid first approxima-
[5] J.C. Slater, Phys. Rev., 81 (195 I) 385: 82 (1951) 53X; Quan-
tion. Yet, although numerically small, these correc- tum Theory of Molecules and Solids, Vol. 3. McGraw-Hill.
tions are certainly most important in energy New York. 1967.
calculations: they contribute Aa”(dEm0’/8a) to [6] J.C. Slater and J.H. Wood, Int. J. Quantum Chem. Symp..
the final result, a quantity ranging from = -86.8 4 (1971) 3; J.C. Slater, Adv. Quantum Chem., 6 (1972) I;
The Self-consistent Field for Molecules and Solids. Vol. 4.
(in n-butane) to N 54.4 kcal mol-’ (in N,Os).
McGraw-Hill, New York, 1974.
Under the present circumstances, where tiny errors [7] R. Gaspar, Acta Phys. Acad. Sci. Hung.. 3 (1954) 263.
in the evaluation of Acr” rapidly translate into size- [8] R. Gaspir and A. Nagy, Theor. Chim. Acta, 72 (1987) 393.
able deviations in the final results, it is rewarding A. Nagy. Int. J. Quantum Chem., 31 (1987) 269.
to witness that our simple brute-force approach R. G.&par, Acta Phys.. 35 (1974) 213.
[9] E.J. Baerends and P. Ros, Chem. Phys.. 2 (1973) 52; Int.
still enables us to come up with a reasonable agree-
J. Quantum Chem. Symp., 12 (1978) 169.
ment between calculated and experimental energies. [IO] K. Schwarz. Phys. Rev. B, 5 (1972) 2466.
There is room for improvement, of course, per- [l l] S. Fliszir, N. Desmarais and M. Comeau, J. Mol. Struct.
haps with the weights to be given in applications (Theochem), 251 (1991) 83.
 E. Vauthier et al./Journul of Molecular Structure (Theochem) 340 (1995) 63-71 71

[12] H. Sambe and R.H. Felton, J. Chem. Phys.. 62 (1975) 1122. [35] CRC Handbook of Chemistry and Physics, 66th edition,
[13] B.I. Dunlap. J.W.D. Connolly and J.R. Sabin, J. Chem. CRC Press Inc., Boca Raton, FL (R.C. Weast et al., eds.).
Phys., 71 (1979) 3396; 71 (1979) 4393. (198551986).
[14] W.J. Hehre, L. Radom, P.v.R. Schleyer and J.A. Pople, [36] G. Herzberg, Molecular Spectra and Molecular Structure.
Ab Initio Molecular Orbital Theory, Wiley, New York, II. Infrared and Raman Spectra of Polyatomic Molecules,
1986. Van Nostrand, Princeton. NJ, 1945.
[15] T.H. Dunning, Jr.. J. Chem. Phys., 53 (1970) 2823; 55 [37] Zefu Nui, K.M. Dunn and J.E. Boggs, Mol. Phys., 55
(1971) 716. (1985) 421.
[16] S. Fliszir, S. Rioux, J. Andzelm, C. Minichino and (381 R.C. Lord, A.L. Marston and F.A. Millers, Spectrochim.
E. Vauthier, Can. J. Chem.. 66 (1988) 3166. Acta, 9 (1957) 113.
E. Vauthier, V. Barone, C. Minichino and S. Fliszar, Can. [39] C.A. Acevedo-Gonzales. M. Campos-Vallette and R.E.
J. Chem., 68 (1990) 1233; S. FliszBr. in S.N. Bulusu (ed.), Clavys-Campos, Spectrochim. Acta Part A, 42 (1985) 919.
Chemistry and Physics of Energetic Materials, NATO ASI [40] Kangnian Fan. Yaoming Xie and J.E. Boggs, J. Mol. Strut.
Ser. C, Vol. 309, p. 131, Kluwer, Dordrecht, 1990. (Theochem), 136 (1986) 339.
[17] S. Fliszar, Atoms, Chemical Bonds and Bond Dissociation [41] Value deduced from AH(combustion) = - 432.75
Energies, Lecture Notes in Chemistry, Vol. 63, Springer, kcal mall’ quoted in ref. [271 and from AH(subl.) = 20.4
Heidelberg. 1994. kcal mol-’ quoted in ref. [43].
[18] K.P. Huber and G. Herzberg, Molecular Spectra and [42] M.M. Sokolava, V.V. Mel’nikov, V.A. Ortrovski, G.I.
Molecular Structure. IV. Constants of Diatomic Mol- Koldobskii, A.A. Mcl’nikov and R.V. Gidaspov, Zh.
ecules, Van Nostrand Reinhold. New York, 1979. Org. Khim., I1 (1975) 1744.
[19] D.R. Stull and G.C.Sinke, Adv. Chem. Ser., 18 (1956) 1. [43] H. Zimmermann and H. Geisenfelder, Z. Elektrochem., 65
[20] G. Herzberg, Molecular Spectra and Molecular Structure. (1961) 368.
I. Spectra of Diatomic Molecules, Van Nostrand Reinhold, [44] G. Herzberg, Molecular Spectra and Molecular Structure.
New York, 1950. III. Electronic Spectra and Electronic Structure of Polya-
[21] M.W. Chase, C.A. Davies, J.R. Downey, D.J. Frurip, R.A. tomic Molecules, Van Nostrand Reinhold, New York,
McDonald and A.N. Syverud, JANAF Thermochemical 1966.
Tables, Vol. 1, Nat]. Stand. Ref. Data Ser., Nat]. Bur. [45] S. Fliszar and M.-T. Beraldin, Can. J. Chem., 60 (1982)
Stand., 14(Suppl. 1) (1985). 792.
[22] M.W. Chase, J.L. Curnutt, J.R. Downey, R.A. McDonald, [46] A. Palm, A. Castelli and C. Alexander, Spectrochim. Acta
A.N. Syverud and E.A. Valenzuela, JANAF Thermochemi- Part A, 24 (1968) 1658.
cal Tables, Nat]. Stand. Ref. Data Ser., Nat]. Bur. Stand. [47] D.R. Stull and H. Prophet, Natl. Stand. Ref. Data Ser.,
1982 (Supplement), J. Phys. Chem. Ref. Data. 11 (1982) 595. Natl. Bur. Stand., 37 (1971).
[23] S. Fliszir and C. Minichino, Can. J. Chem., 65 (1987) 2495. [48] M.E. Jacox, J. Phys. Chem. Ref. Data, 13 (1984) 945.
[24] M.E. Jacox, J. Phys. Chem. Ref. Data, 17 (1988) 269. [49] L. Batt and R.T. Milne, Int. J. Chem. Kinet., 5 (1973) 1067.
1251 M. Dupuis and J.J. Wendoloski, J. Chem. Phys., 80 (1984) [50] B.H. Rockney and E.R. Grant, J. Chem. Phys., 79 (1983) 708.
5696. [51] P. Groner, R. Meyer and H.H. Giinthard, Chem. Phys., 11
[26] S. Fliszir, Charge Distributions and Chemical Effects, (1975) 63.
Springer. New York, Heidelberg, 1983, and references [52] J.R. Durig, J.A. Smooter-Smith. Y.S. Li and F.M. Wasacz.
therein. J. Mol. Struct., 99 (1983) 45.
[27] J.D. Cox and G. Pilcher, Thermochemistry of Organic and [53] A. Kuwac and K. Machida, Spectrochim. Acta Part A, 35
Organometallic Compounds, Academic Press, London, (1979) 27.
1970. [54] J.P. Pedley, R.D. Naylor and S.P. Kirby, Thermochemical
[28] S. Fliszar. F. Poliquin, I. Badilescu and E. Vauthier, Can. J. Data of Organic Compounds, Chapman and Hall,
Chem., 66 (1988) 300. London, 1986.
[29] J.E. Baggott, H.J. Chase and I.M. Millis. Spectrochim. [55] C.F. Melius, in Chemistry and Physics of Energetic
Acta Part A, 42 (1986) 319. Materials, S.N. Bulusu (ed.), NATO-A%, Series C, 309,
[30] Yu. N. Panchenko, Spectrochim. Acta Part A, 31 (1975) 1201. p. 5 I, Kluwer Academic Publishers, Dordrecht, 1990, and
[31] S.W. Benson, F.R. Cruickshank, D.M. Golden, G.R. persona1 communication, 1990.
Haugen, H.E. O’Neal, A.S. Rodgers, R. Shaw and R. [56] N.C. Craig, M.N. Ackermann and R.A. MacPhail.
Walsh, Chem. Rev., 69 (1969) 279. J. Chem. Phys., 68 (1978) 236.
[32] D.A.C. Compton, W.O. George and W.F. Maddams, [57] S.W. Benson and H.E. O’Neal, NSRDS-NBS, 21 (1970).
J. Chem. Sot., Perkin Trans. 2, (1977) 1311. [58] Y. Saito, K. Machida and T. Uno, Spectrochim. Acta Part
[33] E. Castellucci, P. Manzelli, B. Fortunato, E. Gallinella and A, 27 (1971) 991.
P. Mirone, Spectrochim. Acta Part A, 31 (1975) 451. [59] J. Banki, F. Billes and A. Grofcsik, Acta Chim. Hung., 116
[34] T. Shimanouchi, Tables of Molecular Vibrational Frequen- (1984) 283.
cies, Part 2, Nat]. Stand. Ref. Data Ser., Natl. Bur. Stand. [60] P.N. Ghosh and H.H. Giinthard, Spectrochim. Acta Part
11 (1967). A, 37 (1981) 347.