Enthalpy of formation of selected carbonyl radicals from theory and

comparison with experiment

Be la Viskolcza and Tibor Be rces*b

a Department of Chemistry, J. Gy. T eacherÏs T raining College, H-6701 Szeged, P.O. Box 396,
Hungary
b Chemical Research Center, Institute for Chemistry, Hungarian Academy of Sciences,
H-1525 Budapest, P.O. Box 17, Hungary

Received 7th June 2000, Accepted 10th October 2000

First published as an Advance Article on the web 7th No¿ember 2000

Enthalpies of formation and entropies are calculated for 13 carbonyl radicals. Radicals of atmospheric
importance are selected for study. The CBS-4 (Complete Basis Set 4) and G2(MP2,SVP) ab initio molecular
orbital methods are used to obtain the thermodynamic properties of the radicals. In addition to the enthalpy
of formation determinations based on atomization energy computations, enthalpies of formation are derived
for CH CO, CF CO and CCl CO by studying the isodesmic reactions CH ] CH CHO \ CH ] CH CO,
3 3 3 3 3 4 3
CH ] CF CHO \ CH ] CF CO and CH ] CCl CHO \ CH ] CCl CO. Good agreement is found
3 3 4 3 3 3 4 3
between the results obtained by these two approaches. For comparison, carbonyl radical enthalpies of
formation are estimated by using recent experimental room temperature rate coefficients for reactions of a
bromine atom with a series of aliphatic aldehydes. Systematic comparison is made of the theoretical results
with these experimentally based data and recommendations for the enthalpies of formation of carbonyl
radicals and the formyl CÈH bond dissociation energies in the appropriate aldehydes are presented. Finally,
theoretical results are used to obtain reaction heats for carbonyl radical decomposition by CO elimination,
and the atmospheric implications of the results are discussed.

1. Introduction CH OCO for which no data or reliable thermodynamic

Accurate values for bond dissociation energies of molecules
and free radicals are essential in the modelling of atmospheric
and combustion processes. In such studies, the enthalpy of
formation of the appropriate free radical is used (i) in the cal-
culation of the rate coefficient using kinetic data for the
reverse reaction and (ii) in the estimation of the branching
ratio for competitive processes.
Carbonyl free radicals (RÈC2O) are important short-lived
intermediates in the atmospheric degradation of volatile
organic compounds. Carbonyl radicals are formed mainly by
hydrogen abstraction from aldehydes and, to a smaller extent,
by halogen extrusion from acyl halides. The tropospheric
removal of these radicals can occur in a number of ways
including photolysis, hydrolysis or thermal reactions. Little
quantitative knowledge is available on the atmospheric fate of
the carbonyl free radicals. The major chemical reaction of
these radicals is either decomposition by loss of a CO mol-
ecule or combination with O giving a peroxy radical. These
2
two reaction paths have di†erent impact on atmospheric
ozone formation and on the degradation product distribu-
tion.1,2 Recently the thermal decomposition of acyl radicals
was investigated by direct Ñash photolysis1 and the com-
petition between thermal decomposition and O addition has
2
been studied2 in a photochemical reaction chamber.
In this work, the thermochemistry of the decomposition of
carbonyl radicals by CO elimination is investigated using ab
initio molecular orbital methods. In order to test the consis-
tency and reliability of the computed geometries and energies,
two di†erent methods were used, which are described in the
next section. Computations were carried out for carbonyl rad-
icals of atmospheric importance. Thus, Ðrst we dealt with rad-
icals CF CO, CCl CO, CH 2CHCO, HOCH CO and
3 3 2 2
5430 Phys. Chem. Chem. Phys., 2000, 2, 5430È5436
This journal is ( The Owner Societies 2000
3
properties are known. CF CO is a degradation product of the
3
alternative halocarbons of the type CF CX H, such as HFC-
3 2
143a (CF CH ), HFC-134a (CF CFH ), HCFC-141b
3 3 3 2
(CF CClH ) and HCFC-123 (CF CCl H), while CCl CO is a
3 2 3 2 3
degradation product of chloroethane (CCl CH ). In order to
3 3
test the reliability of the computational results obtained for
these radicals, our study was extended to simple RCO free
radicals (where R is an alkyl group) for which experimental
heats of formation are available.
2. Computational methods
Two type of calculations were used : (i) CBS-4,3 CBS-Q3 and
in a few cases CBS-APNO ;4 (ii) G2(MP2, SVP)5,6 and
G3MP27 for the simpler free radicals. All closed and open shell
species were calculated using restricted and unrestricted wave-
functions, respectively.
Petersson and co-workers have developed a series of theo-
retical models,3,4 based on complete basis set extrapolation,
with the aim of determining the projected second-order (MP2)
energy in the limit of a complete basis set. Among these
methods CBS-APNO4 is the most accurate, while CBS-4 is
the most widely applicable treatment. The geometries were
obtained at the UHF level of theory using the 3-21G* basis
set. The harmonic vibrational frequencies were calculated at
the same level of theory. Thermodynamic functions were
obtained by standard thermodynamic methods8 using fre-
quencies scaled by 0.9167.3 Internal rotations were treated as
vibrational motions which may introduce errors in the calcu-
lated entropies, heat capacities, thermal corrections and zero-
point energies.
The G2(MP2,SVP) method5 is a variant of the G2(MP2)
theory in which the size of the basis set used in the QCISD(T)
DOI : 10.1039/b004548i
energy calculations is reduced from 6-311G(d,p) to 6-31G(d). approximation, the classical approximation for the trans-
The geometries and harmonic vibrational frequencies were lational (3/2 RT ) and rotational (3/2 RT and RT for the non-
calculated at the MP2(full)/6-31G(d) level of theory. The fre- linear and linear species, respectively) contributions and the
quency scaling factor of 0.9427, recommended by Scott and PV term. The heat capacity corrections for the atoms were
Radom,9 was used. Thermodynamic functions were obtained taken from the JANAF tables.13
in the same way as for the CBS computations. In the energy Version A and version B calculations were carried out in
calculation, a higher-level empirical correction (HLC) of 4.94 order to see how the results agree. Which of them gives more
mE per electron pair (as suggested by Curtiss et al.10) was accurate results depends on the accuracy of the calculated and
h
used in order to accommodate remaining deÐciencies. In addi- experimental atomization energies of the appropriate diatomic
tion, a spinÈorbit correction to the total energy was applied10 molecules used in Version A and Version B, respectively.
for species containing halogen atoms (i.e. [0.61 mE per Ñuo- Version A was recently used to obtain enthalpies of formation
h
rine atom and [1.34 mE per chlorine atom). The G3(MP2) of acyl radicals and in this study we test it for the calculation
h
energies were obtained as in case of G2(MP2,SVP) except that of the enthalpies of formation for hetero-atom-containing
the MP2(fc)/6-311&G(3df,2p) basis set was replaced by the radicals.
G3MP2 basis.7
The assessment, on a total 148 molecules in the G2 neutral 3. Results and discussion
test set, gave average absolute deviations between computed
and experimental enthalpies of formation for the CBS-4,11 3.1 Geometries and vibrational frequencies
G2(MP2,SVP)10 and G3(MP2)7 methods of 3.06, 1.93 and
At the HF/3-21G* level of theory, two stable conformers were
1.18 kcal mol~1, respectively. In this paper we use CBS-4 for
found for all RCO radicals where R is a saturated group : a
all studied species, G2(MP2,SVP) for all RCO radicals, as well
cis- and a trans-conformer with CÈCÈCÈO (or XÈCÈCÈO)
as G3(MP2) and CBS-Q for some selected carbonyl radicals.
dihedral angles of 0¡ and 180¡, respectively. However, with the
All ab initio calculations were performed with the GAUSS-
MP2(full)/6-31G* method, only the cis-conformers proved to
IAN 94 molecular orbital packages.12
be stable species. For the optimized trans-conformers one of
The enthalpies of formation of the free radicals were
the frequencies was negative, thus the trans-conformers may
obtained in two di†erent ways which are designated as
be identiÐed most probably as rotational transition states.
““ Version A ÏÏ and ““ Version B ÏÏ.
Similar observations were made for CF CO in the
In Version A, the enthalpy of formation for a carbonyl 3
BAC-MP4 calculations14 where the cis-conformer (with
radical, with the formula C X CO, is obtained from the
n 2n`1 FÈCÈCÈO dihedral angle equal to 0¡) proved to be the stable
theoretical reaction enthalpy of the hypothetical reaction
species, while the trans-conformer (with FÈCÈCÈO dihedral
2n ] 1 1 angle equal to 180 degree) had an imaginary vibrational fre-
(n ] 1)C ] X ] O ]C X CO (1) quency.
2 2 2 2 n 2n`1
The geometries published by Francisco and Abersold15 for
according to the equation CH CO and CF CO correspond to the trans-conformers (XÈ
3 3
* H¡ (C X CO) \ H¡ (C H CO, g) [ (n ] 1) CÈCÈO dihedral angle equals 180¡). However, the CF CO
f 298 n 2n`1 298 n 2n`1 3
geometry optimized at the UMP2/6-31G* level of theory by
] [H¡ (C , s) [ * H¡ (C, g)]
298 graphite f 298 Dibble and Francisco16 is the geometry of the stable cis-
2n ] 1 1 conformer (with 0¡ for the FÈCÈCÈO angle).
[ H¡ (X , g) [ H¡ (O , g) (2) Geometries of the carbonyl radicals were computed in this
2 298 2 2 298 2
work at the HF/3-21G* and the MP2(full)/6-31G* level of
where the H¡ designations represent the computed total theory. The results of the two calculations are similar, with a
298
energies including zero-point energies and thermal correc- few percent shorter MP2 bond lengths.
tions, and * H¡ (C, g) \ 171.29 ^ 0.1 kcal mol~1 is the ele- Selected bond lengths and bond angles of the cis-
f 298
mental correction for the C atom taken from the JANAF conformers, optimized at the MP2(full)/6-31G* level, are given
tables.13 in Table 1. Since no experimental bond lengths and angles are
Version B is based on the atomization reaction
Table 1 Selected bond lengths in AŽ and bond angles in degrees of
CX CO ] (n ] 1)C ] (2n ] 1)X ] O. (3)
n 2n`1 the RCO radicals optimized at the MP2(full)/6-31G* level
The enthalpy of atomization for radical C X CO at 0 K is
n 2n`1
obtained from the computed total energies including zero-
point energies :
* H¡ \ (n ] 1)H¡(C) ] (2n ] 1)H¡(X)
r 0 0 0
] H¡(O) [ H¡(C X CO) (4)
0 0 n 2n`1
With this theoretical enthalpy of atomization and the experi- Radicals r r a a a b(X CCO)
mental enthalpies of formation for the appropriate atoms 1 2 1 2 3 1
(taken from the JANAF tables13), the 0 K enthalpy of forma- CH CO 1.196 1.513 127.5 111.1 108.6 0.0
tion of the radical is calculated : C H3 CO 1.198 1.517 126.9 113.5 106.6 0.0
2 5H CO
n-C 1.198 1.517 126.9 113.7 106.8 0.0
* H¡(C X
f 0 n 2n`1
CO) \ (n ] 1)* H¡(C) ] (2n ] 1)
f 0 i-C 3H 7CO 1.197 1.525 127.5 108.7 107.4 0.0
3 7
tert-C H CO 1.198 1.531 127.5 110.9 107.0 0.0
] * H¡(X) ] * H¡(O) [ * H¡ . (5) 4 9
f 0 f 0 r 0 CH 2CHCO 1.197 1.485 129.4 118.9 117.8 23.1
Finally, the enthalpy of formation at 298.15 K is obtained CH22C(CH )CO 1.205 1.485 127.9 116.6 116.8 24.1
from the 0 K value using the heat capacity correction (given in CF 2CO 3 1.186 1.551 123.2 112.7 108.6 0.0
CCl3 CO 1.183 1.554 127.3 112.1 106.0 0.0
square brackets) : 3
HOCH CO 1.191 1.538 126.4 113.3 106.2 169.9a
2
* H¡ (C X CO) \ * H¡(C X CO) CH OCO
3
1.201 1.324c 130.2 114.6 È 0.0b
f 298 n 2n`1 f 0 n 2n`1 HC(O)CO 1.195 1.557 122.6 113.6 121.8 81.3a
] [H¡ [ H¡] [ ; [H¡ [ H¡] (6) CH C(O)CO 1.196 1.552 124.6 114.2 119.2 83.7a
298 0 CnX2n`1CO 298 0 atoms 3
The heat capacity correction for C X CO is calculated by a OCCO angle. b COCO angle. c CÈO bond length.
n 2n`1
using the scaled vibrational frequencies in the harmonic

Phys. Chem. Chem. Phys., 2000, 2, 5430È5436 5431

Table 2 Harmonic vibrational frequencies in cm~1 and rotational constants (in square brackets) obtained at the MP2(full)/6-31G* level, as well
as ZPE in kcal mol~1 and heat capacities (C ) in cal mol~1 K~1 for carbonyl radicals
v
Radicals Frequencies and rotational constants ZPE C
v
CH CO 81, 469, 899, 991, 1093, 1421, 1525, 1525, 1979, 3110, 3217, 3220 26.3 10.7
3
[2.744, 0.331. 0.313]
C H CO 103, 252, 253, 633, 761, 852, 1023, 1082, 1132, 1312, 1374, 1471, 44.0 15.1
2 5
1518, 1560, 1562, 1954, 3121, 3122, 3168, 3212, 3220
[0.609, 0.197, 0.158]
n-C H CO 72, 160, 194, 254, 348, 656, 735, 870, 874, 961, 1094, 1108, 1160, 61.4 19.8
3 7 1290, 1328, 1362, 1426, 1475, 1514, 1555, 1565, 1569, 1949, 3105,
3109, 3123, 3153, 3173, 3196, 3197 [0.541, 0.086, 0078]
i-C H CO 44, 231, 248, 278, 361, 446, 618, 797, 937, 965, 1005, 1110, 1184, 61.3 18.5
3 7
1233, 1343, 1382, 1457, 1476, 1549, 1553, 1565, 1569, 1949, 3109,
3110, 3112, 3199, 3203, 3210, 3211 [0.261, 0.138, 0.102]
tert-C H CO 21, 231, 253, 281, 288, 351, 355, 400, 458, 576, 750, 878, 983, 990, 78.2 26.1
4 9
999, 1067, 1087, 1282, 1286, 1307, 1452, 1454, 1482, 1540, 1546,
1550, 1561, 1564, 1578, 1940, 3100, 3101, 3104, 3190, 3191, 3195,
3200, 3202, 3204 [0.153, 0.093, 0.093]
CH 2CHCO 270, 323, 592, 691, 970, 1013, 1031, 1107, 1343, 1483, 1705, 1965, 31.6 12.9
2
3220, 3274, 3320 [0.892, 0.205, 0.168]
CH 2C(CH )CO 157, 199, 276, 382, 469, 590, 717, 912, 950, 1040, 1053, 1104, 47.1 18.4
2 3 1325, 1463, 1493, 1540, 1556, 1730, 1975, 3108, 3190, 3209, 3212,
3309 [0.330, 0.135, 0.098]
CF CO 43, 241, 405, 417, 541, 545, 673, 813, 1233, 1267, 1285, 2230 13.1 16.9
3 [0.186, 0.100, 0.100]
CCl CO 38, 195, 273, 277, 331, 403, 456, 601, 837, 871, 890, 2116 9.8 20.3
3
[0.062, 0.056, 0.052]
HOCH CO 132, 253, 363, 521, 857, 908, 1128, 1213, 1366, 1430, 1525, 1939, 29.2 14.2
2 3066, 3180, 3774 [1.505, 0.143, 0.135]
CH OCO 139, 287, 389, 763, 949, 1192, 1195, 1231, 1509, 1550, 1554, 1867, 30.0 13.0
3
3137, 3239, 3275 [0.746, 0.228, 0.181]
HC(O)CO 91, 346, 559, 818, 1010, 1399, 1903, 2637, 3079 [1.295, 0.179, 16.0 11.7
0.167]
CH C(O)CO 74, 118, 118, 373, 502, 502, 744, 983, 983, 1212, 1439, 1439, 1527, 32.4 17.2
3
1899, 1899, 3111, 3197, 3197 [0.313, 0.141, 0.104]
CH CHO 147, 515, 799, 926, 1168, 1170, 1439, 1467, 1527, 1535, 1801, 33.7 11.0
3
2993, 3107, 3187, 3237 [1.886, 0.337, 0.302]
CF CHO 68, 259, 314, 434, 528, 530, 708, 868, 992, 1244, 1259, 1374, 1437, 20.1 17.4
3
1806, 3074 [0.181, 0.099, 0.098]
CCl CHO 77, 212, 260, 284, 332, 339, 461, 652, 792, 902, 1035, 1082, 1424, 17.1 20.6
3
1805, 3065 [0.062, 0.055, 0.051]

available for the carbonyl radicals, the calculated data are
compared with the results of previous theoretical studies.
Dibble and Francisco16 published UMP2/6-31G* optimized
geometries for CF CO, while Bauschlicher17 reported geo-
3
metrical parameters for CH CO which were computed at the
3
MP2/6-311&G(3df,2p) level of theory. As expected, the
UMP2(fc)/6-31G(d) results reported by Dibble and Francisco
agree reasonably with the similar (MP2(full)/6-31G(d) results
obtained in this work. The MP2 geometry of CH CO
3
obtained with the larger, 6-311&G(3df,2p) basis set di†ers from
that calculated with the smaller basis set by about 0.01 AŽ in
bond length and 1¡ in bond angle.
The geometry of the RCO type carbonyl radicals depends
on the nature of the R moiety. If R is an alkyl group, the
r (C2O) distance is roughly constant at around 1.197 AŽ for all
1
studied carbonyl radicals, while the r (CÈCO) distance
2
increases from 1.513 to 1.531 AŽ as the C-atom number (and
simultaneously the size) of the R group increases from C to
1
C . This increase of the CÈCO bond length parallels the
4
decrease of the CÈCO bond energy (see later). The a (OCC)
1
angle is around 127¡, while the a (CCH) and a (CCH) angles
2 3
are 112¡ and 107¡, respectively. The bond angles do not
change signiÐcantly with the nature of R.
Introduction of halogen or oxygen atoms in the alkyl R
group, shortens the r (C2O) and lengthens the r (CÈCO) dis-
1 2
tances. An a (OCC) angle, signiÐcantly lower than the
1
average, is obtained for CF CO.
3 2
If the R group is an unsaturated hydrocarbon group (as in
CH 2CHCO and CH 2C(CH )CO), a considerably shorter
2 2 3
r (CÈCO) interatomic distance is obtained compared to that
2
in the radicals with saturated R groups. This is obviously the
result of conjugation in the former case. For these radicals, the
geometries obtained in the HF/3-21G* and in the MP2(full)/6-
31G* geometry optimizations are signiÐcantly di†erent : all
heavy atoms lie in a plane in the geometry derived at the HF/
3-21G* level, while at the MP2(full)/6-31G* level, the carbon-
yl oxygen atom is outside the plane determined by the rest of
the heavy atoms.
A large di†erence was found between the MP2/6-31G* and
the HF/3-21G* geometries in the case of the HC(O)CO
radical. Geometry optimizations for the HC(O)CO radical
carried out at the HF/6-311G(d,p) and QCISD/6-311G(d,p)
level gave similar results to those of the MP2(full)/6-31G(d)
calculations, i.e. non-planar geometry. (The OCCO dihedral
angle was about 90¡.) In addition, the total energy of the non-
planar HC(O)CO radical obtained in the CBS-4//HF/6-
311G(d,p) computations was found to be lower by about 5
kcal mol~1 than that of the planar species obtained with the
CBS-4//HF/3-21G* method. Moreover, the energy of the non-
planar HC(O)CO radical derived with the CBS-APNO
method was in excellent agreement with that obtained in the
G2(MP2,SVP) computations.
On the basis of these experiences we prefer to use, in the
case of the carbonyl radicals with unsaturated R groups (i.e.
CH 2CHCO, CH 2C(CH )CO, HC(O)CO and CH C(O)CO),
2 2 3 3
the non-planar geometry derived at the MP2(full)/6-31G(d) or
HF/6-311(d,p) level of theory.
Although CH 2CHCO and HC(O)CO, as well as
CH 2C(CH) CO and CH C(O)CO are isoelectronic free rad-
2 3 3
icals, with unsaturated R groups, there are considerable di†er-
ences in the structures of these RCO radicals depending upon
whether R contains an oxygen atom or not : Compared to the

5432 Phys. Chem. Chem. Phys., 2000, 2, 5430È5436

r (CÈCO) distance in RCO radicals with saturated alkyl R
2
groups, the r (CÈCO) interatomic distance is considerably
2
shorter if R is an unsaturated hydrocarbon group, while this
bond is longer if R is an oxygen-containing unsaturated
group.
The spin contamination in the equilibrium structures, calcu-
lated at the UHF/6-311&G(3df,2p) level of theory, was found
to be low and the expectation value SS**T was 0.75È0.80, i.e.
close to the correct value of 0.75. In addition, the occupation
numbers of the UHF natural orbitals did not di†er signiÐ-
cantly from 2 or 0 below and above the Fermi level, respec-
tively, indicating that the single-conÐguration UHF
wavefunctions provide an acceptable description.
Unscaled harmonic vibrational frequencies and rotational
constants, calculated at the MP2(full)/6-31G* level, are given
in Table 2. The frequencies, scaled with a factor of 0.9669,
were used to obtain the zero-point energies (ZPE) and heat
capacities (C ), as well as heat capacity corrections (HCC). The
v 2
MP2(full)/6-31G* zero-point energies are higher, by 0.4 kcal
mol~1 on average than those obtained at the HF/3-21G* level
of theory.
Our MP2(full)/6-31G* frequencies calculated for CH CO
3
and CF CO are in good agreement with those obtained by
3
Bauschlicher17 with the MP2/6-311]G(3df,2p) method and
the frequencies computed by Melius14 using the BAC-MP4
technique, respectively.
3.2 Enthalpies of formation for carbonyl radicals and formyl
C–H bond dissociation energies for aldehydes
The computed total energies, together with the ZPE and HCC
values, were used to obtain the enthalpy of formation for the
carbonyl free radicals. The calculated standard enthalpies of
formation, derived according to Version A and Version B, are
summarized in Table 3. This Table also lists the enthalpies of
formation for three selected aldehydes. Among these, no ther-
modynamic properties can be found in the literature for
CF CHO and CCl CHO, while CH CHO may serve as a test
3 3 3
compound for which the thermodynamic properties are well
established.
In addition to the enthalpies of formation, standard
entropies are also given in Table 3 for the carbonyl radicals.
In the calculation of these values, internal rotations were
treated as harmonic oscillators. Since these low frequency
motions make signiÐcant contributions to the entropies, the
given entropy values may be considered only as rough
approximations. If accurate entropies are required, more
sophisticated treatments have to be used for the low frequency
motions (for such treatments see for instance ref. 18 and 19).
Comparing the enthalpies of formation obtained by Version
A and Version B, one Ðnds good agreement for most species.
However, signiÐcant deviations occur for the CBS-4 results of
the Ñuorine-containing species CF CO and CF CHO, as well
3 3
as for the G2(MP2,SVP) results of the RCO radicals where R
is an oxygen-containing group (i.e. HOCH CO, CH OCO,
2 3
HC(O)CO and CH C(O)CO). In all these cases lower enth-
3
alpies of formation are obtained with Version A than with
Version B. Since Version A is based on the computation of the
enthalpy of formation of RCO from diatomic molecules (and
C), we have checked both with the CBS and the G2 methods
the accuracy of the enthalpies of formation calculation of the
diatomic molecules. The calculated enthalpies of formation at
298 K are given, in kcal mol~1, below in brackets. CBS-4
method : H ([0.37), O (1.18), F (4.57), Cl (0.87). G2(MP2,
2 2 2 2
SVP) method : H ([0.61), O (3.03), F (0.26), Cl ([1.03). As
2 2 2
can be seen, the enthalpies of formation of F and O are
2 2
signiÐcantly overestimated by the CBS-4 and G2(MP2,SVP)
methods which explains most of the di†erence between the
Version A and Version B results. Therefore, in our further dis-
cussions only Version B results will be considered.
For comparison of the CBS-4 enthalpies of formation with
the G2(MP2,SVP) results (all Version B calculations), the
RCO radicals and RCHO molecules may be divided into four
groups :
(i) For species where R is an alkyl group, the G2(MP2,SVP)
results are systematically higher than the CBS-4 enthalpies of
formation. As a check, higher level computations were carried
out. Thus, CBS-Q (Version B) calculations gave [2.5 kcal
mol~1 for the enthalpy of formation of CH CO while G3MP2
3
(Version B) computations resulted in [22.1 kcal mol~1 for
the enthalpy of formation of the tert-C H CO radical.
4 9
Although with these results the deviation decreased some-
what, however, the basic problem remained.
Examining the series of RCO radicals with alkyl R group,
one Ðnds that the di†erence between the G2(MP2,SVP) and
CBS-4 results increases with increasing size of the R group. It
has been found recently,20 in a theoretical study of alkyl rad-
icals, that the CBS-4 method underestimates the enthalpy of
formation of these radicals. In order to compensate for this,
an empirical correction has been suggested. Such an under-
estimation of the enthalpies of formation by the CBS-4
method may explain (at least partly) the di†erence between
the CBS-4 and G2 results.
The consistency of the results obtained by the two methods

Table 3 Standard entropies (in cal mol~1 K~1) and heats of formation (in kcal mol~1) for carbonyl radicals and selected aldehydes, calculated
by di†erent theoretical models

CBS-4//HF/3-21G* G2(MP2,SVP)//MP2(full)/6-31G(d)
* H¡ * H¡
f 298 f 298
Radicals S¡ Version A Version B S¡ Version A Version B

CH CO 64.0 [3.5 [3.6 64.8 [3.2 [3.0

C H3 CO 72.2 [9.6 [10.1 71.3 [7.3 [8.0
2 5H CO
n-C 79.8 [15.5 [16.4 78.8 [11.6 [13.0
i-C 3H 7CO 79.2 [16.6 [17.6 79.0 [12.8 [14.3
3 7H CO
tert-C 83.0 [25.8 [27.3 85.9 [20.6 [22.9
4 9
CH 2CHCO 67.1 25.2 23.7 66.8 23.7 23.8
2
CH 2C(CH )CO 75.3 8.7 8.1 74.8 14.0 13.3
CF 2CO
3
77.6 [152.9 [145.6 77.8 [153.5 [151.9
CCl3 CO 83.7 [16.9 [15.1 85.9 [13.9 [14.1
3 CO
HOCH 69.4 [36.3 [35.8 69.8 [39.2 [37.5
2
CH OCO 70.1 [38.4 [37.9 68.8 [42.5 [40.8
3
HC(O)CO 66.5 [11.7 [10.9 68.3 [17.5 [15.1
CH C(O)CO 74.9 [29.7 [29.3 76.8 [31.5 [30.4
CH3CHO 62.4 [40.9 [41.2 62.7 [41.0 [41.2
CF 3CHO 75.3 [192.9 [185.8 75.9 [195.0 [193.7
CCl3 CHO 81.8 [58.1 [56.6 82.8 [55.8 [56.4
3

Phys. Chem. Chem. Phys., 2000, 2, 5430È5436 5433

has been tested by means of BensonÏs group additivity rule.21
With a reasonable assumption for the group value of the
group adjacent to the radical center and with published group
contributions,20,22 the unknown group additivity value (GAV)
of [C0 O-C] could be derived. Using the enthalpies of forma-
tion of carbonyl radicals obtained by the G2(MP2,SVP)
method (Version B), a GAV of 7.2 ^ 0.1 kcal mol~1 was
determined. However, when testing the CBS-4 results, strong
dependence of the GAVs on size of the R group was observed ;
i.e. in the series from CH CO to tert-C H CO, the GAVs
3 4 9
varied as 6.5, 4.9, 4.0 and 2.8 kcal mol~1, respectively.
Among the species belonging to this group, the experimen-
tal enthalpy of formation for CH CHO is known23 to be
3 3
[39.73 kcal mol~1, and the recommended value for the
CH CO radical, derived in a recent critical evaluation,24 is
3
[2.9 ^ 0.7 kcal mol~1. Both of these are somewhat closer to
the G2(MP2,SVP) results. However, the reported enthalpy of
formation for C H CO25 (i.e. [10.2 ^ 1 kcal mol~1) agrees
2 5
better with the CBS-4 calculations.
(ii) In the case of the RCO radicals with unsaturated hydro-
carbon R group, again the G2(MP2,SVP) results are higher
than the CBS-4 enthalpies of formation. Considering the
CH 2CHCO enthalpy of formation of 23.7 kcal mol~1
2
obtained in our CBS-Q calculation, a value around 24 kcal
mol~1 appears to be well established by theory. This is higher
by almost 7 kcal mol~1 than the experimental value.25 A
larger di†erence is found between the G2(MP2,SVP) and
CBS-4 results for CH 2C(CH )CO. The comparison of the
2 3
data obtained for the two unsaturated radicals is in favor of
the former one.
(iii) In the third set of radicals with oxygen-containing R
group (i.e. HOCH CO, CH OCO, HC(O)CO and
2 3
CH C(O)CO), relatively small di†erences are found between
3 3
the CBS-4 and G2(MP2,SVP) results. However, in contrast to
the results obtained for radicals belonging to the Ðrst and
second set, the enthalpies of formation computed by the
CBS-4 method for radicals with oxygen-containing R group
are systematically higher than those obtained by the G2(MP2,
SVP) procedure. Calculations were carried out with higher
level CBS methods. Thus, CBS-Q computations for
HOCH CO and CH OCO gave enthalpies of formation of
2 3
[36.1 and [39.8 kcal mol~1, respectively. CBS-APNO com-
putations resulted in the value of [15.4 kcal mol~1 for
HC(O)CO. These values are higher than the CBS-4 results
and bring the CBS enthalpies of formation close to the
G2(MP2,SVP) results. No experimental enthalpies of forma-
tion are available in the literature for this group of radicals.
(iv) The Ðnal set consists of free radicals and aldehydes with
halogen atoms in the R group. The greatest di†erence between
the CBS-4 and G2(MP2,SVP) results are found in this set,
especially for the Ñuorinated species CF CO and CF CHO. It
3 3
was found6,10 in G2 calculations that spinÈorbit correction
([0.61 mE per Ñuorine atom and [1.34 mE per chlorine
h h
atom) is required for molecules containing two or more
halogen atoms. This correction, which decreases the di†erence
in the calculated results, is still far from explaining the signiÐ-
cant discrepancies. Although no reliable experimental enth-
alpies of formation are available, however, BAC-MP4
computations were carried out for these species.14 The calcu-
lated enthalpies of formation at 298 K are [145.8 and
[185.1 kcal mol~1 for CF CO and CF CHO, respectively,
3
which are in excellent agreement with the CBS-4 data but
di†er considerably from the G2(MP2,SVP) results. Computa-
tions with the G3MP2 method (Version B calculations) yield
[150.3 and [12.5 kcal mol~1 for CF CO and CCl CO,
3 3
respectively, i.e. the discrepancy still exists with the CBS-4
results.
In order to obtain further information on the enthalpies of
formation of the halogenated carbonyl radicals CF CO and
3
CCl CO, calculations were carried out for three isodesmic
3
reactions, i.e. for reactions
CH ] CH CHO \ CH ] CH CO (7)
3 3 4 2
CH ] CF CHO \ CH ] CF CO (8)
3 3 4 3
CH ] CCl CHO \ CH ] CCl CO (9)
3 3 4 3
The reaction enthalpies at 298 K, computed by the CBS-4
method, are : * H¡ \ [15.18 kcal mol~1, * H¡ \ [11.57
r 7 r 8
kcal mol~1 and * H¡ \ [11.14 kcal mol~1. Using the enth-
r 9
alpy of formation for CH ,24 CH ,23 CH CHO,23 as well as
3 4 3
those for CF CHO and CCl CHO (this work, CBS-4 results),
3
the following enthalpies of formation are obtained :
* H¡ (CH CO) \ [1.9 kcal mol~1, * H¡ (CF CO) \
f 298 3 f 298 3
[143.7 kcal mol~1 and * H¡ (CCl CO) \ [14.8 kcal
f 298 3
mol~1. These values also support the computed CBS-4 enth-
alpies of formation for CF CO and CCl CO.
3 3
Summarizing the above considerations, we conclude that
the theoretical enthalpies of formation of RCO radicals and
RCHO molecules are best represented by the G2(MP2,SVP)
results if R is a saturated or unsaturated hydrocarbon group.
The enthalpies of formation of radicals with oxygen-
containing R groups are equally well represented by the CBS
and G2 results. Finally, the enthalpies of formation of species
with halogen-containing R groups are best characterized by
the CBS-4 values. The data given in Table 4, under the
heading ““ Theory ÏÏ, are selected in accordance with this con-

Table 4 Enthalpies of formation for carbonyl radicals, * H¡ (RCO), and formyl CÈH bond dissociation energies of aldehydes,a D
H¡ (RC(O)ÈH), in kcal mol~1 f 298
298
Experiment Theory Recommendation
Species * H¡ D H¡ * H¡ D H¡ * H¡ D H¡
f 298 298 f 298 298 f 298 298
CH CO [3.5 88.3 [3.0 88.8 [3.2 ^ 0.2 88.6
C H3 CO [9.7 87.9 [8.0 89.6 [8.7 ^ 0.7 88.9
2 5
n-C H CO [13.0 88.0 [13 ^ 1 88
i-C 3H 7CO [16.7 87.7 [14.3 90.1 [15.3 ^ 00.9 89.1
3 7H CO
tert-C [27.1 84.0b [22.9 88.2b [24.6 ^ 1.5 86.5b
CH 2CHCO4 9 23.8 90.5b 24 ^ 3 91b
CH22C(CH )CO 13.3 88.4b 14 ^ 4 89b
CF 2CO
3
[144.5 93.4 [144.7c 93.2 [144.6 ^ 0.4 93.3
CCl3 CO [7.0 101.7 [15.0c 93.7
HOCH 3 CO [37.5 90.0b
2
CH OCO [40.8 96.5b
3
HC(O)CO [15.1 87.7
CH C(O)CO [30.4 86.5
3
a The enthalpies of formation of the aldehydes were taken from ref. 23, or estimated b by the additivity method. c Average of CBS-4 values given
in Table 3 and derived from calculations for isodesmic reactions.

5434 Phys. Chem. Chem. Phys., 2000, 2, 5430È5436

clusion : i.e. CBS-4 results are preferred for the halogen- ation of the theoretical and experimental based enthalpies of
containing species and G2MP2 computations for the rest of formation.
the radicals. (Naturally, due attention was paid in the evalu-
ation to the results derived from the higher level computations 3.4 Carbonyl radical decomposition by CO elimination and
and from the study of isodesmic reactions. This explains the atmospheric implications
small di†erence in the case of a few theoretical values given in
Table 4 and Table 3.) Carbonyl radical decomposition products, formed by CO
elimination, have also been studied by ab initio computations.
Calculated CBS-4 results obtained in this work, together with
3.3 Comparison of theoretical and experimental results results of similar calculations taken from the literature, are
In the absence of reliable experimental enthalpies of formation listed in the second column of Table 5. The comparison of
for the great majority of the studied carbonyl radicals, estima- these computational results with experimental data shows
tions were made using recently determined room temperature excellent agreement. The only exception is the trichloromethyl
rate coefficients for reactions of a bromine atom with a series radical where the theoretical value is considerably lower than
of aliphatic aldehydes : 26 the experimental one. Similar disagreement was observed also
for the trichloroacetyl radical (see Table 4). It appears that the
Br ] RCHO ] HBr ] RCO (10) theoretical methods used in this study underestimate the enth-
Activation energies for the bromine atom reactions with alde- alpies of formation of the chlorinated species. It is to be men-
hydes, E , were estimated from the room temperature rate tioned, however, that the experimental enthalpy of formation
10 of CCl indicated in Table 5 ( just like the experimental value
coefficients k by assuming A \ 7.8 ] 1012 cm3 mol~1 s~1 3
10 10 of CCl CO given in Table 4) is rather uncertain. Therefore,
(the recommended A-factor27 for reaction Br ] CH CHO) to 3
3 new theoretical and especially experimental investigations of
be valid for the whole reaction series studied. Finally, second-
law values for the RCO enthalpies of formation were obtained the thermochemistry of these chlorinated radicals are highly
with the estimated E values and assumed E activation desirable.
10 ~10 The recommended enthalpies of formation of the carbonyl
energies for the RCO ] HBr reactions. The assumptions
made were : E \ 0.0 ^ 2.0 kcal mol~1 and E \ 2.0 radicals (from Table 4) and the experimental values of the
~10 ~10 decomposition products (from Table 5) were used to estimate
^ 2.0 kcal mol~1 for reactions of RCO radicals without and
with halogen (F or Cl) substituents, respectively. (A higher the reaction enthalpies for the CO elimination reactions of the
activation energy appears reasonable for the attack of the carbonyl radicals. The results are given in Table 6. The
electrophilic CF CO and CCl CO radicals on the hydrogen decomposition enthalpies are seen to depend strongly on the
3 3 structure of the R group of the RCO radical. Among the RCO
atom of HBr carrying a partial positive charge.) The results of
estimations are given in Table 4 under the heading radicals with alkyl R group, the calculated stability strongly
““ Experiment ÏÏ. decreases with increasing branching at the C atom adjacent to
In addition to the enthalpies of formation of carbonyl free the carbonyl group. This observation is in full agreement with
radicals, the formyl CÈH bond dissociation energies for the the result of a very recent experimental study2 and is related
appropriate aldehydes are also given in Table 4. In the calcu-
lation of the bond dissociation energy, the enthalpy of forma- Table 5 Enthalpies of formation for carbonyl radical decomposition
tion of the aldehyde is taken from ref. 23 or is estimated by products in kcal mol~1
the group additivity method.22 (For CF CHO and CCl CHO,
3 3 Species Theory (CBS-4) Experiment
the values obtained in this work were used.)
The theoretical enthalpies of formation of carbonyl radicals, CH 34.720 35.124
where R is an alkyl group, are systematically higher than the C H3 28.820 28.424
2 5H
n-C 24.120 23.924
experiment-based estimations. Considering the formyl CÈH
bond dissociation energies, the experimental values show the i-C 3H 7 21.020 21.024
3 7H
tert-C 12.320 11.524
decreasing trend expected by intuition, while the theoretical 4 9
CH 2CH 70.7 71.524
values are practically constant within the group. This results CH22C(CH ) 58.4
in an increased di†erence between theory and experiment as CF 2 3 [112.4 [112.413
the complexity of the R group increases. CCl3 9.3 17.027
3
Among the halogenated carbonyl radicals, the theoretical HOCH [3.4 [4.029
and experimental enthalpy of formation for CF CO are in CH O 2 6.6 4.930
3 3
good agreement and the corresponding formyl CÈH bond dis- HCO 10.2 10.413
CO [26.2 [26.413
sociation energies in CF CHO are close to the 91.1 ^ 1.9 kcal
3
mol~1 value derived by Amphlett and Whittle28 from the
study of thermal bromination of triÑuoroacetaldehyde. Thus,
the enthalpy of formation of CF CO appears to be well estab- Table 6 Reaction enthalpy (in kcal mol~1) at 298 K of carbonyl
3 radical decomposition by CO elimination
lished. However, signiÐcant di†erence occurs between the
theoretical and experimental heat of formation of the CCl CO Decomposition * H¡ (RCO2R ] CO)
3 r 298
radical. Since only a rough value could be determined26 by
experiment for the rate coefficient of the reaction of Br CH CO ] CH ] CO 11.9
] CCl CHO, we prefer the theoretical value, although an C H3 CO ] C 3H ] CO 10.7
3 2 5H CO ]2n-C 5 H ] CO
n-C 10.5
underestimation in the computations can not be excluded. 3 7 3 7] CO
i-C H CO ] i-C H 9.9
In the last two columns, recommended values for the car- 3 7H CO ] tert-C
tert-C 3 7 H ] CO 9.7
bonyl radical enthalpies of formation and the formyl CÈH 4 9 ] CH 2CH 4 9] CO
CH 2CHCO 21.1
bond dissociation energies of the appropriate aldehydes are CH22C(CH )CO ]2CH 2C(CH ) ] CO 18.0
given. These are selected by taking into account both the CF 2CO ] CF3 ] CO 2 3 5.8
experimental and theoretical values, with somewhat more CCl3 CO ] CCl 3 ] CO 5.6
3 CO ] HOCH
3
HOCH ] CO 7.1
weight given to the latter ones (i.e. the weighting factors of 1 2 ] CH O ] 2CO
CH OCO 19.3
and 1.5 were used for the experimental and theoretical values, 3 3
HC(O)CO ] CHO ] CO [1.0
respectively). In the case of the CCl CO radical, no recom- CH C(O)CO ] CH CO ] CO 0.8
3 3 3
mended value is given in Table 4 due to the signiÐcant devi-

Phys. Chem. Chem. Phys., 2000, 2, 5430È5436 5435


Fig. 1 Correlation between the reaction enthalpies of CO elimi-
nation, * H¡ , and the length of the ruptured bond, r (CÈCO).
r 298 2
to the decrease of the elimination barrier in the reaction
series.1
Comparing the reaction enthalpy of CO elimination with
the length of the splitting bond in RCO, an inverse relation-
ship is found between * H¡ , and r (CÈCO) as shown in Fig.
r 298 2 12
1. This empirical correlation indicates the inverse trend in the
reaction series between the change of the decomposition
barrier and the length of the bond to be ruptured.
On the basis of the reaction enthalpies of CO elimination,
the studied carbonyl radicals may be divided into three
groups :
(i) Free radicals CH 2CHCO, CH 2C(CH )CO and
2 2 3
CH OCO are stable with respect to CO elimination. The
3
atmospheric fate of the Ðrst two species is association reaction
with oxygen forming RC(O)OO type acyl-peroxy radicals. For
the CH OCO radical, another decomposition path (CO
3 2
elimination) exists, which is exothermic by 18.2 kcal mol~1.
Thus, the CH OCO radical will preferably decompose by
3
CO elimination in the atmosphere.
2
(ii) Free radicals HC(O)CO, CH C(O)CO, HOCH CO,
3 2
CF CO and CCl CO are relatively unstable and are expected
3 3
mainly to decompose in the atmosphere. Following CO elimi-
nation, the R product radical will enter into an association
reaction with O and participate as an ROO radical in the
2
atmospheric processes.
(iii) Finally, the RCO radicals where R is an alkyl group,
are of intermediate stability. The stability decreases with
increasing branching of the alkyl group. For these radicals,
the competition between decomposition and association with
oxygen has to be considered :
RCO ] M ] R ] CO ] M (11)
RCO ] O ] M ] RC(O)O ] M (12)
2 2
On the basis of the reaction thermochemistry, the k /k [O ]
11 12 2
ratio is expected to increase with increasing branching at the
C atom next to the CO group. However, using kinetic data for
reactions (11) and (12) it can be shown1,2 that decomposition
is negligible in the atmosphere (at 298 K in 1 atm air) com-
pared to reaction with O .
2
Acknowledgements
This work was supported by the European Commission
project entitled AFCAR (contract no. IC20-CT97-0037) and
5436 Phys. Chem. Chem. Phys., 2000, 2, 5430È5436
by the Hungarian Science Foundation (OTKA, contract No. :
T029722 and F030436). B. V. is grateful for the award of a
Magyary Zoltan Postdoctoral Fellowship (AMFK 535/2).
References
1 A. Tomas, E. Villenave and R. Lesclaux, Phys. Chem. Chem.
Phys., 2000, 2, 1165.
2 S. Jagiella, H. G. Libuda and F. Zabel, Phys. Chem. Chem. Phys.,
2000, 2, 1175.
3 J. W. Ochterski, G. A. Petersson and J. A. Montgomery, J. Chem.
Phys., 1996, 104, 2598.
4 J. A. Montgomery, Jr., J. W. Ochterski and G. A. Petersson, J.
Chem. Phys., 1994, 101, 5900.
5 B. J. Smith and L. Radom, J. Phys. Chem., 1995, 99, 6468.
6 L. A. Curtiss, P. C. Redfern, B. J. Smith and L. Radom, J. Chem.
Phys., 1996, 104, 5148.
7 L. A. Curtiss, P. C. Redfern, K. Raghavachari, V. Rassolov and J.
A. Pople, J. Chem. Phys., 1999, 110, 4703.
8 D. A. McQuarrie, Statistical T hermodynamics, Harper and Row,
New York, 1973.
9 A. P. Scott and L. Radom, J. Phys. Chem., 1996, 100, 16502.
10 L. A. Curtiss, K. Raghavachari, P. C. Redfern and J. A. Pople, J.
Chem. Phys., 1997, 106, 1063.
11 L. A. Curtiss, K. Raghavachari, P. C. Redfern and B. B. Stefanov,
J. Chem. Phys., 1998, 108, 692.
M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G.
Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. A. Peterss-
on, J. A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G.
Zakrzewski, J. V. Ortiz, J. B. Foresman, J. Cioslowski, B. B. Ste-
fanov, A. Nanayakkara, M. Challacombe, C. Y. Peng, P. Y.
Ayala, W. Chen, M. W. Wong, J. L. Andres, E. S. Replogle, R.
Gomperts, R. L. Martin, D. J. Fox, J. S. Binkley, D. J. Defrees, J.
Baker, J. P. Stewart, M. Head-Gordon, C. Gonzalez and J. A.
Pople, GAUSSIAN 98, Gaussian, Inc., Pittsburgh, PA, 1995.
13 M. W. Chase, Jr., C. A. Davies, J. R. Downey, Jr., D. J. Frurip, R.
A. McDonald and A. N. Syverud, J. Phys. Chem. Ref. Data, 1985,
14, Suppl. No. 1.
14 M. R. Nyden, G. T. Linteris, D. R. F. Burgess, Jr., P. R. West-
moreland, Wing Tsang and M. R. Zachariah in, Evaluation of
Alternative In-Flight Fire Suppressants for Full-Scale T esting in
Simulated Aircraft Engine Nacelles and Dry Bays, ed. W. L. Gros-
shandler, R. G. Gann and W. M. Pitts, NIST SP 861, April 1994,
section 5.
15 J. S. Francisco and N. J. Abersold, Chem. Phys. L ett., 1991, 187,
354.
16 T. S. Dibble and J. S. Francisco, Chem. Phys. L ett., 1993, 215,
409.
17 C. W. Bauschlicher, Jr., J. Phys. Chem., 1994, 98, 2564.
18 A. L. L. East and L. Radom, J. Chem. Phys., 1997, 106, 6655.
19 P. Y. Ayala and H. B. Schlegel, J. Chem. Phys., 1998, 108, 2314.
20 I. Marsi, B. Viskolcz and L. Seres, J. Phys. Chem. A, 2000, 104,
4497.
21 S. W. Benson, T hermochemical Kinetics, Wiley, New York, 2nd
edn., 1976.
22 N. Cohen and S. W. Benson, Chem. Rev., 1993, 93, 2419.
23 J. D. Cox and G. Pilcher, T hermochemistry of Organic and
Organometallic Compounds, Academic Press, London, 1970.
24 Wing Tsang, in Energetics of Organic Free Radicals, ed. J. A. M.
Simoes, A. Greenberg and J. F. Liebman, SEARCH Series,
Blackie, London, 1996, vol. 4.
25 D. F. McMillen and D. M. Golden, Annu. Rev. Phys. Chem.,
1982, 33, 493.
26 L. A. Curtiss and P. C. Redfern, J. Chem. Phys., 1996, 104, 5148.
27 R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A.
Kerr and J. Troe, J. Phys. Chem. Ref. Data, 1999, 28, 191.
28 J. C. Amphlett and E. Whittle, T rans. Faraday. Soc., 1970, 66,
2016.
29 S. Dobe, T. Berces, T Turanyi, F. Marta, J. Grussdorf, F. Temps
and H. Gg. Wagner, J. Phys. Chem., 1996, 100, 19864.
30 D. L. Osborn, D. J. Leahy, E. M. Ross and D. M. Neumark,
Chem. Phys. L ett., 1995, 235, 484.