The Open-Chain Trioxide CF3OC(O)OOOC(O)OCF3

Article abstract of DOI:10.1002/chem.200304942

The open-chain trioxide CF₃OC(O)OOOC(O)OCF₃ is synthesised by a photochemical reaction of CF₃C(O)OC(O)CF ₃, CO and O₂ under a low-pressure mercury lamp at -40°C. The isolated trioxide is a colourless solid at -40°C and is characterised by IR, Raman, UV and NMR spectroscopy. The compound is thermally stable up to -30°C and decomposes with a half-life of 1 min at room temperature. Between -15 and +14°C the activation energy for the dissociation is 86.5 kJ mol^-1 (20.7 kcal mol^-1). Quantum chemical calculations have been performed to support the vibrational assignment and to discuss the existence of rotamers.

Full text of DOI:10.1002/chem.200304942

FULL PAPER
The Open-Chain Trioxide CF3OC(O)OOOC(O)OCF3
Stefan von Ahsen,[a] Pla¬cido GarcÌa,[a] Helge Willner,*[a, c] Maximiliano Burgos Paci,[b] and
Gustavo A. Arg¸ello[b]
Abstract: The open-chain trioxide CF3OC(O)OOOC(O)OCF3 is synthesised by a
photochemical reaction of CF3C(O)OC(O)CF3, CO and O2 under a low-pressure
mercury lamp at À408C. The isolated trioxide is a colourless solid at À408C and is
Keywords: fluorine ¥ matrix isola-
characterised by IR, Raman, UVand NMR spectroscopy. The compound is thermally
tion ¥ photochemistry ¥ trioxides ¥
stable up to À308C and decomposes with a half-life of 1 min at room temperature.
vibrational spectroscopy
Between À15 and ‡148C the activation energy for the dissociation is 86.5 kJmolÀ1
(20.7 kcalmolÀ1). Quantum chemical calculations have been performed to support
the vibrational assignment and to discuss the existence of rotamers.
mixtures with CO, whereby F atoms act as key species. In a
related reaction system, in which CF3O radicals catalyse
Introduction
There are many examples in chalcogen chemistry of mole-
cules such as polysulfanes, containing polyelement chains.
Single-bonded polyoxy compounds are less common, because
their stability decreases as the number of oxygen atoms
increases. Some trioxides with bulky organic substituents such
as tBuOOO-tBu,[1 3] as well as various primary ozonides, have
been studied for many years. Khalizov and co-workers
reviewed alkyl trioxides recently,[4] but perfluorinated polyoxy
the oxidation of CO to CO2,[12, 13] the new peroxide CF3-
OC(O)OOC(O)OCF3 was isolated,[13] studied kinetically[14]
and characterised structurally.[15] Traces of CF3OC(O)OOOC-
(O)OCF3 were detected as a by-product.[13]
Both the peroxide and the trioxide are good thermal
sources for CF3O,[16, 17] as is the trioxide for CF3OC(O)OO;[18]
both of these radicals are important intermediates in the
degradation of CFCs and their replacements.
compounds are still laboratory curiosities. Only a few
In this work, we describe improved reaction conditions that
lead to a high yield of bis(trifluoromethoxy)trioxodicarbonate
(CF3OC(O)OOOC(O)OCF3) as well as its isolation, spectro-
scopic characterisation and thermal decay in the presence of
excess N2 and CO. Molecular properties are predicted by
quantum chemical calculations, which also support the
interpretation of the observed spectra.
[5 9]
perfluoroalkyl trioxides such as CF3OOOCF3
and
[5, 6]
CF3OOOC2F5
have been isolated and well characterised,
whereas the first acyl compound FC(O)OOOC(O)F has been
mentioned only in recent years[10] and is still under inves-
tigation.[11]
Bis(fluoroformyl) trioxide is formed as a by-product in the
synthesis of FC(O)OOC(O)F by the reaction of F2/O2
Results and Discussion
[a] Prof. Dr. H. Willner, Dr. S. von Ahsen, P. GarcÌa
Fakult‰t 4, Anorganische Chemie
Synthetic aspects: In the reaction mechanism that leads to the
known peroxide CF3OC(O)OOC(O)OCF3,[13] CF3O radicals
generated by the reaction sequence represented by Equa-
tions (1) (3)[13] play a key role. The catalytic oxidation of CO
proceeds by the chain reaction given in Equations (4) (7)
and summarised in Equation (8).[12, 13]
Gerhard-Mercator-Universit‰t Duisburg
Lotharstr. 1, 47048 Duisburg (Germany)
Fax : (‡49)203-379-2231
E-mail: willner@uni-duisburg.de
[b] Dr. M. Burgos Paci, Prof. Dr. G. A. Arg¸ello
INFIQC, Depto. de FÌsico QuÌmica
Facultad de Ciencias QuÌmicas
¬
Universidad Nacional de Cordoba, Ciudad Universitaria
CF3C(O)OC(O)CF3 ‡ hn ! 2CF3 ‡ CO2 ‡ CO
2CF3 ‡ 2O2 (‡ M) ! 2CF3OO (‡ M)
2CF3OO ! 2CF3O ‡ O2
(1)
(2)
(3)
(4)
¬
5000 Cordoba (Argentina)
E-mail: gaac@isquim.fcq.unc.edu.ar
[c] Prof. Dr. H. Willner
Present address:
Bergische Universit‰t Wuppertal, FB 9, Anorganische Chemie
Gaussstr. 20, 42097 Wuppertal (Germany)
2CF3O ‡ 2CO (‡ M) ! 2CF3OCO (‡ M)
Chem. Eur. J. 2003, 9, 5135 5141
DOI: 10.1002/chem.200304942
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5135

H. Willner et al.
FULL PAPER
2CF3OCO ‡ 2O2 (‡ M) ! 2CF3OC(O)OO (‡ M)
2CF3OC(O)OO ! 2CF3OCO2 ‡ O2
2CF3OCO2 ! 2CF3O ‡ 2CO2
(5)
(6)
(7)
(8)
We conclude that radical radical recombination reactions,
which usually have no activation barrier and, therefore, no
temperature dependence, are not the only ones responsible
for the observations. Since unimolecular dissociation of
molecules does have an activation barrier, the observed
temperature dependence of the reaction system is now
rationalised. Moreover, as the dissociation of CF3OCO2
radicals requires a low activation energy, neither trioxide
nor peroxide is formed above room temperature.
Here, trapping of the trioxide at À408C, at which it is not
only thermally stable but also shows no significant vapour
pressure, gives evidence which strongly supports the reaction
system discussed above[13] and ensures that it proceeds in the
desired direction.
2CO ‡ O2 ! 2CO2
[8]
It was possible to isolate CF3OC(O)OOCF3 and CF3-
[13]
OC(O)OOC(O)OCF3
from the reaction mixture. These
compounds were formed by some of the termination reactions
[Eqs. (9) (11)].
2CF3OCO2 (‡ M) ! CF3OC(O)OOC(O)OCF3
CF3OCO2 ‡ CF3O (‡ M) ! CF3OC(O)OOCF3
CF3OCO ‡ CF3OO (‡ M) ! CF3OC(O)OOCF3
(9)
(10)
(11)
Spectroscopic properties: The IR spectrum of pure CF3-
OC(O)OOOC(O)OCF3 is very similar to the known IR
spectrum of CF3OC(O)OOC(O)OCF3,[13] since all the char-
acteristic bands appear in the same spectral region. The main
At a reaction temperature of 08C, a thermally labile
compound was detected in addition to these products, but it
could not be identified clearly, because its IR spectrum was
very similar to that of CF3OC(O)OOC(O)OCF3. The most
likely candidate for the unidentified compound was the
trioxide CF3OC(O)OOOC(O)OCF3,[13] which is formed by
recombination of the rather long-lived peroxy radical CF3O-
C(O)OO, formed through Equation (5), with the short-lived
CF3OCO2 radical, generated from Equation (6), in the
termination reaction [Eq. (12)].
ˆ
differences in the spectra can be found in the n(C O),
À
n(C O), n(OO) and n(OOO) modes. The IR and Raman
spectra of CF3OC(O)OOOC(O)OCF3 are presented in Fig-
ure 1.
CF3OCO2 ‡ CF3OC(O)OO (‡ M) ! CF3OC(O)OOOC(O)OCF3 (‡ M) (12)
At a lower reaction temperature (À408C) the trioxide can
be obtained as the main product, together with CO2 from the
catalytic cycle and the peroxydicarbonate as only a trace. As
the primary formation of CF3OC(O)OO depends on the
photolysis rate of trifluoroacetic anhydride (TFAA); its
concentration is, in a first approximation, independent of
temperature. Nevertheless, the formation of the short-lived
species CF3OCO2 [Eq. (6)] is strongly dependent on CF3-
OC(O)OO formation. Therefore, the trioxide CF3OC-
(O)OOOC(O)OCF3 is formed according to Equation (12)
as the primary product during the synthesis at any temper-
ature.
Figure 1. IR (gas-phase) and Raman (solid, À1968C) spectra of CF3O-
C(O)OOOC(O)OCF3.
The reaction given in Equation (9) is responsible for the
formation of the peroxide and is important only if the
concentration of the short-lived CF3OCO2 radical is effec-
tively increased. Here the postulated trioxide may act as a
reservoir for this radical since, once it is formed, the
regeneration of CF3OCO2 would then be dependent on the
reaction temperature as the decay of the trioxide is an
activated process [Eq. (13)].
Table 1 shows an assignment of the observed bands, which
are compared with the related peroxide. The assignment is
supported by quantum chemical calculations, which are
discussed below. Due to the similarity of the masses of all
the atoms and comparable bond strengths, all modes except
ˆ
the C O stretching are strongly mixed. Therefore, the
fundamental modes are not described in Table 1.
For CF3OC(O)OOOC(O)OCF3 only one main absorption
band occurs in the carbonyl region (1878 cmÀ1 in both the gas
phase and neon matrix), implying that the vibrational
CF3OC(O)OOOC(O)OCF3 ! CF3OC(O)OO ‡ CF3OCO2
(13)
ˆ
coupling of the two n(C O) bands is reduced in comparison
From Equation (13) together with Equation (6) it follows
that the formation of the peroxide CF3OC(O)OOC(O)OCF3
at higher reaction temperatures (À20to ‡108C) is only a
consequence of the thermal decay of the primarily generated
trioxide [Eq. (13)].
with CF3OC(O)OOC(O)OCF3 (where two separate bands at
1887 and 1866 cmÀ1 can be observed). This feature could be
interpreted in terms of the elongation and enhanced flexibility
of the enlarged oxygen chain. In general, any normal mode in
which a movement in the CF3OC(O) moieties is expected
5136
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 5135 5141

Polyoxy Compounds
5135 5141
Table 1. Fundamental modes of CF3OC(O)OOOC(O)OCF3, compared with CF3OC(O)OOC(O)OCF3.
CF3OC(O)OOOC(O)OCF3
CF3OC(O)OOC(O)OCF3
Gas phase
Ne matrix
n
Ar matrix
rel.
]
Raman
Calcd[a]
Assignment
Gas phase[b] Ar matrix[b] Calcd[a]
n
s
n
n
int.
n
IR int.
[kmmolÀ1
RA int.
[amuä4]
C2 sym.
n [cmÀ1
]
n [cmÀ1
]
n [cmÀ1
]
[cmÀ1
]
[10À20 cm2] [cmÀ1
]
[cmÀ1
int.[c]
[cmÀ1
]
[cmÀ1
]
]
1878
277
1878.3
1872
41
1866
s
1935.1
1930.9
448
1284.6
535
1240.9
114
1207.4
67
657
.6
129
0
10A
1.1
B
n1
1887
1866
1298
1884
1857
1296
1940.4
1911.9
1286.5
1288.1
1242.9
1242.5
1240.7
B
n24
n25
n2
1293
1257
1185
549
1295.01291
1255.3
47
1284.9
0
0.8
0.8
1.4
1.1
12
0.8
34
A
B
B
747
125060
1242.7
1224.3
.3
A
n3
1259
1251
254
n26
n4
1701185.1
3A.0
1181.8
1165.5
1178
34
99
n27
n5
1217.6
1185
1131
1048
988
1170174
1138
1169.6
348
2.9
A
1195
1135
1050
989
1177.2
1118.3
1069.2
987.5
958.9
905.5
876.8
815
230
35
1133.6
1010.3
975.6
1132
1009
974
100
1.7
25
1128.0 1900
B
A
B
A
B
A
n28
n6
1013
s
1025.0
973.0
956.3
901.6
883.2
34
603
1.1
122
0.55
5.9
.1
28
4.6
26
8.6
3.4
1.6
1.3
974
899
976
923
900
878
m
w
766
748
740
702
683
616
vw
vs
s
n29
n7
902.2
901
4.9
1.4
8.1
B
n30
n8
905
906
s
795
789
778
48
49
55
792.8
771.4
767.2
757.8
751.3
703.2
684.1
616.2
79013
7705.7
767
757
749
703
682
615
791
778
3.580 358
775.1
w
w
w
s
n31
n32
n9
32
769.5
757.8
0
B
756
742
774.4
774.7
756.0
746.3
667.2
723.3
0.2
0.5
0.3
2.5
1.1
1.0
1.7
A
A
B
A
B
A
B
1.1
n10
n33
n11
n34
n12
n35
n13
n36
n14
n37
n38
n15
n16
n39
n40
n17
n18
n41
n42
n19
n20
n43
n21
n44
n45
n22
n23
744
669
750.6
702.4
687.0
608.3
608.0
1.2
555.9
552.7
449.9
427.4
427.2
381.4
379.2
329.5
274.3
243.4
695
10
1.1
0.18
0.28
668
727
60
m
s
610614 7.6
606.7
546
547.05455.6130
0s
559.8
2.4
A
0.54
1.8
2.7
3.2
1.2
0.5
0.8
1.5
0.3
0.6
1.3
2.9
B
A
B
B
A
A
B
B
A
A
556.6
555.6
541
442
541.9
450.2
541
448
0.07
0.89
544
449
433
s
563[d]
431
19
.9460460460
m
s
0.21
0.16
0.03
1.1
0.90
0.29
0.50
426.1
428.5
383.7
379.8
327.1
340.4
231.8
156.0
386
383
339
286
252
180w
w
m
s
343[d]
242[d]
166.1
3.6
.1 0
0.08
B
141.3
116.6
100.5
88.8
70.9
57.7
29.8
27.8
21.4
0.1
0.6
0.7
0.5
0.4
0.4
0.1
0.2
0.3
B
A
A
B
A
B
B
A
A
121
72
s
s
0.04
0.34
0.47
0.12
0.07
0.23
0.20
0.00
161.5
111.4
100.7
74.7
56.7
33.0
108[d]
39.1
26.5
[a] B3LYP/6-311G(d,p), this work. [b] From ref. [13]. [c] Relative integrated band intensities. [d] Raman, solid at À1968C.
should give rise to two separate bands in the vibrational
spectrum as the movement can be in-phase or off-phase.
However, from our spectroscopic resolution (1 cmÀ1 in Ne
matrix, 2 cmÀ1 in the gas-phase IR spectra, 4 cmÀ1 in the
Raman spectrum), only overlapping bands of the correspond-
ing in-phase and off-phase modes are observed.
ns(OOO) is detected as a strong Raman band at 923 cmÀ1.
The antisymmetric mode na(OOO) is strongly mixed with the
carbon oxygen stretching modes, as is also calculated for
According to quantum chemical calculations, the most
stable isomer of the trioxide is the all-trans-sym-CF3-
OC(O)OOOC(O)OCF3 (Figure 2) that has C2 symmetry and
À
n(C O) and n(OO) in the case of the peroxide CF3-
OC(O)OOC(O)OCF3. The n(OO) fundamentals are useful
for distinguishing between the trioxide and peroxide, as the
trioxide absorbs at 974 cmÀ1 with moderate intensity while the
peroxide has its analogue at 989 cmÀ1 (both gas-phase values).
Figure 2. Calculated structure of CF3OC(O)OOOC(O)OCF3.
Chem. Eur. J. 2003, 9, 5135 5141
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5137

H. Willner et al.
FULL PAPER
a trans configuration (close to 1808) of all the independent
dihedral angles of the 11-atom F-C-O-C-O-O-O-C-O-C-F
chain except for the C-O-O-O and O-O-O-C angles, for which
b ˆ ‡90.138 (see Figure 2). The irreducible representation of
the 45 fundamental vibrations is given in Equation (14) and
they can be described approximately by 16 stretching, 19
deformation, two out-of-plane and eight torsion modes.
179.988, 179.438 and 177.438 for FCOC, COCO and OCOO,
respectively.
For the assignment of the vibrational spectrum, the
calculations predict only small differences in the IR spectra
for the four most stable rotamers. Irrespective of the state
(either gas-phase or solid), the broad bands in the trioxide
spectra do not allow a distinction to be made between the
different rotamers, so all observed bands can be accounted for
by the expected spectrum of the all-trans-sym-CF3-
OC(O)OOOC(O)OCF3 isomer as described in Table 1.
However, in principle, resolution of more fundamental
vibrations and discovery of different rotamers should be
possible in IR matrix spectra. A rough estimation of the
Boltzmann ratio gives percentages of the order of 20% for the
ttc-sym-ttt and 5% for the ttt-asy-ttt rotamer. Since the
detection limit of our experimental set-up is in the range of
a few per cent (that is, the band intensity of an impurity or
trace compound compared with the most intense band of the
major species) and since the spectra of the rotamers are very
similar, evidence of other isomers was observed, but no proof.
Bands at 1863 and 986 cmÀ1 that are also seen in the matrix IR
spectrum are in agreement with the calculated data for the ttc-
sym-ttt-CF3OC(O)OOOC(O)OCF3 isomer in position and
expected intensity; a very weak band at 1888 cmÀ1 in the matrix
IR spectrum may be due to the all-trans-asy-CF3OC(O)OOOC-
(O)OCF3 isomer. Nevertheless, all of these bands could also
be assigned to an impurity of CF3OC(O)OOC(O)OCF3, as its
IR spectrum differs from that of the trioxide especially at
these wavenumbers. Hence, it was not possible to prove
unambiguously the existence of isomers other than the most
stable one, although the appearance of an impurity in the
20% range is very unlikely according to the NMR analysis.
The UV spectrum of CF3OC(O)OOOC(O)OCF3 in the
375 200 nm region shows an unstructured absorption band
Gvib ˆ 23A(IR, Rap) ‡ 22B(IR, Radp)
(14)
The trioxide may exist in 72 possible rotamers: in each
CF3OC(O)O unit three independent dihedral angles can be
either 1808 or 08, leading to 23 ˆ 8 species for each unit. If both
parts of the molecule are identical, we get eight isomers. If
they are different, then 7‡6‡5‡4‡3‡2 ‡ 1 ˆ 28 additional
isomers are possible. The two dihedral angles in the COOOC
unit, each with a gauche configuration, may be symmetric
(both 908) or antisymmetric (908 and À908), which leads to a
total of 2(8‡28) ˆ 72 rotamers. Moreover, the non-symmetric
isomers can exist in two enantiomeric forms, which are not
counted here.
Full calculations were performed for rotamers that are
stable and differ from the most stable form in only a few
dihedral angles. Among the possible stable rotamers, the next
higher in energy is the ttc-sym-ttt-CF3OC(O)OOOC(O)OCF3
rotamer (cis configuration of one O-C-O-O dihedral, resulting
in C1 symmetry) and the energy difference from the most
stable isomer is 3.9 kJmolÀ1. The next is the ttt-asy-ttt-
CF3OC(O)OOOC(O)OCF3 (C1 symmetry), which is
7.1 kJ molÀ1 higher in energy than the most stable rotamer.
The most important calculated data are presented in Table 2.
The dihedral angles of the most stable all-trans-sym-
CF3OC(O)OOOC(O)OCF3 (Figure 2) were calculated to be
Table 2. Calculations on CF3OC(O)OOOC(O)OCF3 rotamers [B3LYP/6-311G(d,p)].
Rotamer[a]
ttt-sym-ttt
ttc-sym-ttt
ttt-asy-ttt
ttc-sym-ctt
tct-sym-ttt
tct-sym-tct
ctt-sym-ttt
Symmetry
C2
C1
C1
C2
C1
C2
E ‡ ZPC [hartree]
À 1278.1804
À 1278.1789
À 1278.1777
À 1278.1775
À 1278.1761
À 1278.1717
non-existent
#
DErel [kJmolÀ1
]
03.9
7.1
7.6
11.3
22.8
Ratio[b]
1
0.199
0.055
0.044
0.010
< 0.001
ttt-sym-ttt
[c]
ˆ
n(C O)
1935(67)
1931(657)
1285(448)
1285(129)
1243(535)
1241(254)
1224(114)
1207(99)
1170(348)
1128(1900)
1025(34)
973(603)
956(1)
1936(381)
1914(412)
1289(282)
1285(269)
1246(439)
1243(401)
1241(114)
1211(87)
1164(376)
1123(1847)
1008(124)
979(497)
959(8)
1940(531)
1912(134)
1288(290)
1281(274)
1244(397)
1241(381)
1230(183)
1209(42)
1169(721)
1126(1493)
1031(85)
992(408)
965(79)
1919(227)
1910(640)
1290(155)
1289(413)
1248(519)
1246(244)
1244(27)
1229(119)
1157(2)
1119(2471)
998(10)
983(323)
940(0)
1945(297)
1933(458)
1286(322)
1283(529)
1243(448)
1230(372)
1215(119)
1204(214)
1160(941)
1124(906)
1014(110)
963(228)
958(119)
893(55)
1948(48)
1944(735)
1289(315)
1279(961)
1235(504)
1230(256)
1204(574)
1204(16)
1146(215)
1126(1008)
996(37)
n (stretch)
967(2)
954(313)
860(3)
902(122)
83(1)
896(34)
879(1)
907(126)
878(22)
891(8)
872(0)
859(6)
857(0)
803(349)
782(160)
829(171)
765(73)
807(329)
808(351)
[a] Nomenclature: t ˆ trans (1808), c ˆ cis (08). Labelling order: 1st letter: FCOC dihedral; 2nd letter: COCO dihedral; 3rd letter: OCOO dihedral for first
half of molecule, second half the reverse; sym indicates that both central dihedral angles are identical (‡908), asy means one is ‡908, and one À908. Rotamer
ctt-sym-ttt has no energetic minimum; the CF3 unit rotates into the ttt-sym-ttt structure. [b] Ratio: derived from DE Boltzmann distribution at room
temperature (294 K). [c] Wavenumbers [cmÀ1] (intensities [kmmolÀ1]).
5138
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 5135 5141

Polyoxy Compounds
5135 5141
whose maximum should be below 200 nm. The absorption
cross sections could be estimated only approximately, because
of the experimental limitations of the matrix technique.
Nevertheless, from absorbance measurements at room tem-
perature in the gas phase, the cross sections are probably
higher than those obtained from the corresponding peroxide
CF3OC(O)OOC(O)OCF3,[13], showing that the photochemi-
cal decomposition of the trioxide can also be important in the
above-mentioned reaction scheme including the synthesis of
the trioxide or the peroxide.
The 19F and 13C NMR spectra of CF3OC(O)OOOC-
(O)OCF3 show the expected signals and coupling patterns
(Table 3).
A more complete characterisation could not be achieved
for several experimental reasons: reliable data for melting
point, boiling point, vapour pressure and UV cross sections
cannot be given as the trioxide CF3OC(O)OOOC(O)OCF3 is
stable only up to about À308C. At this temperature it is still
solid, with a vapour pressure in the 0.1 mbar region. Therefore
a structural characterisation of the trioxide in the gas phase
was not possible by gas electron diffraction. We were not able
to grow crystals by crystallisation of the amorphous solid
during several weeks at À308C.
Figure 3. Decay of CF3OC(O)OOOC(O)OCF3 in the presence of excess
N2.
Kinetic aspects: The IR cell, which was maintained at
different temperatures between À15 and ‡148C, was filled
with about 1 mbar of pure CF3OC(O)OOOC(O)OCF3 and
then mixed with a large excess of either N2 or CO. Time-
spaced spectra were then recorded.
Dissociation in N2 as bath gas gave mainly CO2, C(O)F2,
CF3OOOCF3 and small amounts of CF3OC(O)OOC(O)OCF3
as products, all of which were identified by their known IR
spectra. The rate of trioxide disappearance was measured at
À5, À3, ‡8 and ‡148C (Figure 3). Good straight lines were
obtained at any temperature and no pressure dependence was
observed.
Figure 4. Arrhenius plot of the decay reaction of CF3OC(O)OOOC-
(O)OCF3.
Except for the point at
Table 3. NMR data for CF3OC(O)OOOC(O)OCF3 and CF3OC(O)OOC(O)OCF3.
À58C, there is no superposition
ˆ
Species
dF CF3 [ppm]
dC CF3 [ppm]
1JCF [Hz]
dC C O [ppm]
of the measured temperatures
and good correspondence be-
[a]
[b]
CF3OC(O)OOOC(O)OCF3
CF3OC(O)OOOC(O)OCF3
CF3OC(O)OOC(O)OCF3
À 58.5
À 61.3
À 61.3
‡ 120.0
‡ 119.9
‡ 119.9
270.3
268.7
268.7
‡ 146.2
‡ 145.7
‡ 145.5
tween the two sets can be
observed. This points to a sim-
ple unimolecular dissociation
mechanism with an energy bar-
rier of 86.5 kJmolÀ1. Hence the
[b]
[a] This work: 20mg substance in 2 mL CD 2Cl2 with a few percent of CFCl3 measured at À788C, chemical shifts
related to dF CF35Cl3 (À788C) ˆ 0.0 and dC CD2Cl2 (À788C) ˆ 54.0, respectively. [b] Ref. [13], CF3OC(O)OO-
C(O)OCF3 with CF3OC(O)OOOC(O)OCF3 as impurity, neat at À308C, with CDCl3/CFCl3 mixture as external
lock and reference.
major pathway of the thermal
decay of CF3OC(O)OOOC-
Dissociation in the presence of CO gave CF3OC(O)-
C(O)OCF3 and C(O)F2 as principal products besides minor
quantities of CF3OC(O)OOC(O)OCF3 and the previously
unknown anhydride CF3OC(O)OC(O)OCF3 (which absorbs
strongly at 1056 cmÀ1).[19] In these experiments good straight
lines were also obtained at À15, À5, 0and ‡58C with similar
results to those shown in Figure 3.
As the Arrhenius plot gives the same activation energy
within experimental error for both bath gases, we have plotted
all the points in Figure 4.
(O)OCF3 follows Equation (13) and subsequent reactions.
The value derived for the activation energy is as expected. It is
around 40kJmol À1 lower than the energy needed to dissociate
a typical fluorinated peroxide,[14, 20] and it is also lower than
the energy required for the dissociation of CF3OOOCF3.[21]
The activation energies for hydrogenated dialkyl trioxides,
which are all in the range 80 90kJmol À1, show excellent
correspondence with those of other known trioxides.[4] Matrix
experiments and quantum chemical calculations also demon-
strate that the central oxygen oxygen bond is the weakest in
Chem. Eur. J. 2003, 9, 5135 5141
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5139

H. Willner et al.
FULL PAPER
the trioxides and that the thermal stability decreases in the
and FOO; however, the pentoxide CF3O5CF3 should not exist,
in accordance with the non-existent FOOOF.[23]
sequence
CF3OOOCF3 > CF3OC(O)OOOC(O)OCF3 ꢀ
FC(O)OOOC(O)F ꢁ FOOOF, which is demonstrated here
by the temperatures required for pyrolysis in matrix experi-
ments (500,[22] 160, 1608C,[22] for the first three) and by the fact
that FOOOF does not exist.[23] Predictions for the bonding
energy in fluorinated compounds were made through quan-
tum chemical calculations on the series of symmetrical
trioxides ROOOR with R ˆ CF3, FC(O), CF3OC(O), F,
CH3. Some properties of these trioxides were calculated
(Table 4).
Experimental Section
The polyoxides CF3OC(O)OOOC(O)OCF3 and CF3OC(O)OOC(O)OCF3
are potentially explosive, especially in the presence of oxidisable materials.
All reactions should be carried out in millimolar quantities only and it is
important to take safety precautions, especially when these compounds are
handled in the liquid or solid state.
Synthesis: Volatile materials were manipulated in glass vacuum lines, which
were equipped with two capacitance pressure gauges (221AHS1000 and
221AHS10; MKS Baratron, Burlington, MA), three U-traps used for trap-
to-trap condensation, and valves with PTFE stems. The vacuum line was
connected to an IR gas cell (20cm optical path length, Si windows) inside
the sample chamber of an FTIR spectrometer (Impact 400D; Nicolet,
Madison, WI). This enabled us to observe the course of reactions and the
purification process.
Table 4. Calculated[a] properties of selected symmetric trioxides ROOOR for
the most stable rotamer.
R ˆ CF3 R ˆ CF3OC(O) R ˆ FC(O) R ˆ F R ˆ CH3
[b]
DH(O-O) [kJmolÀ1
]
102.4
65.1
60.4
1.8
79.8
DG(O [kJmolÀ1
]
47.08.7
1.432
1.395
6.7
À 50.4 25.7
À
À O)
The synthesis of CF3OC(O)OOOC(O)OCF3 was a modification of the
published procedure for the generation of CF3OC(O)OOC(O)OCF3 at
d(O O) [ä]
1.423
1.383
106.7
À 0.01
À 0.16
‡ 0.54
1.427
1.375
106.6
0.00
1.3801.434
[13]
À
d(R O) [ä]
(1.445) 1.428
lower temperature. A Pyrex glass photoreactor (5 L) equipped with a
water-cooled mercury low-pressure lamp (15 W, TK15; Hereaus, Hanau,
Germany) in a quartz inlet was connected to a vacuum line. In a typical
experiment trifluoroacetic acid anhydride (5 mbar, 1.0mmol, 99%; Merck,
Darmstadt, Germany) was placed in the reactor together with carbon
monoxide (150mbar, 31 mmol, 99%; Messer Griesheim, Krefeld, Germa-
ny) and oxygen (300 mbar, 62 mmol, 99%; Messer Griesheim, Krefeld,
Germany). The reactor was cooled to À408C in an ethanol bath cooled
with a cryostat (Lauda-Wobser GmbH, Lauda, Germany). After about 2 h
of illumination with the mercury lamp, when the CO was nearly consumed,
more CO (150mbar) was added to the reactor. After 5 h most of the TFAA
had reacted. The reaction gas passed the three U-traps in the vacuum line,
cooled with liquid nitrogen, while the reactor was allowed to reach room
temperature slowly. Then the products were gathered in one trap and
fractionated by warming this trap to À408C, while the volatile material
passed the other traps held at À908C and À1968C. Pure solid trioxide
CF3OC(O)OOOC(O)OCF3 was left in the À408C trap. Small amounts of
the peroxide CF3OC(O)OOC(O)OCF3 could, if present, be separated from
the product by pumping it off at À378C.
Preparation of the matrices: A small amount of CF3OC(O)OOOC-
(O)OCF3 (approximately 0.05 mmol) was transferred into a small U-trap,
which was mounted on a quartz pyrolysis nozzle placed directly in front of
the matrix support held at 16 K for Ar or 7 K for Ne. The matrix gas Ne or
Ar was directed over the CF3OC(O)OOOC(O)OCF3 sample held at
À858C by means of an ethanol bath. For each matrix experiment 1
3 mmol Ne or Ar was used, passing the trap within 20 60min. To study
the thermal stability of the trioxide, the pyrolysis nozzle was heated to
1608C. Details of the matrix apparatus are given elsewhere.[26]
a(O-O-O) [8]
q(Ocenter) [e]
q (Oside) [e]
q (C) [e]
107.4
109.7
108.0
À 0.05
À 0.18
‡ 0.70
‡ 0.05 À 0.08
‡ 0.09 À 0.22
À 0.09
À 0.15
‡ 0.56
[a] B3LYP/6 311G(d,p). [b] Can be seen as an activation barrier of the
.
.
unimolecular decomposition of ROOOR into RO ‡ ROO .
Why does the fluorinated compound CF3OC(O)OOOC-
(O)OCF3 behave (in the case of its activation barrier) more
like a hydrocarbon trioxide (for example, CH3OOOCH3), and
why does CF3OOOCF3 differ so greatly from ROOOR when
R ˆ FC(O) and CF3OC(O)? A reasonable explanation lies in
the interaction between two effects that dominate the stability
of the trioxy bridge.
From the calculated values of charges on the OOO chain in
Table 4, it may be concluded that an optimum of accumulated
charge exists: the decreasing negative charge when going
from CH3OOOCH3 to CF3OOOCF3 leads to a strong
stabilisation of the molecule. The extremely electronegative
fluorine in the hypothetic molecule FOOOF even induces a
positive charge on the oxygen atoms that destabilises the
molecule.
A carbonyl moiety bonded to the trioxy group destabilises
the trioxide chain and favours dissociation, as a resonance-
stabilised carboxy radical is formed. Hence, an asymmetric
trioxide will always dissociate according to Equation (15) and
the trioxides ROOOR with R ˆ FC(O), CF3OC(O) are less
stable than CH3OOOCH3 and CF3OOOCF3.
Kinetic measurements: The kinetic measurements were performed in a
double-walled IR gas cell placed in the sample compartment of an IR
spectrometer. The cell was maintained at different temperatures between
À15 and ‡148C. Small samples (approximately 1 mbar) of pure CF3OC-
(O)OOOC(O)OCF3 were fed into the cell from the storage vessel and
subsequently mixed with N2 (99.99%; La OxÌgena, Argentina) or CO
(99.9%; Praxair, Argentina) up to 550mbar total pressure. Then time-
spaced spectra were recorded.
ROOOC(O)R' ! ROO ‡ R'CO2
(15)
The disappearance of CF3OC(O)OOOC(O)OCF3 was measured using the
bands at 1879 and 973 cmÀ1. At these wavenumbers the products do not
absorb. For the analysis of most of the products observed (CO2,
CF3OOOCF3, CF2O, CF3OC(O)OOC(O)OCF3), calibrated reference
spectra from pure samples were recorded.
Since a CF3O group behaves like a ™pseudo-fluorine
atom∫,[17] shown here by comparing CF3OC(O)OOOC-
(O)OCF3 with FC(O)OOOC(O)F[11] and also discussed in
studies of the pairs CF3OCO[17]/FCO[24] and CF3OC(O)OOC-
(O)OCF3[13]/FC(O)OOC(O)F,[25] some still-unknown proper-
ties of the species may be foreseen: the hypothetical tetroxide
CF3OOOOCF3 would probably be stable at liquid-nitrogen
temperature but would dissociate above this temperature into
CF3OOO and CF3O, in analogy to FOOF, which decays into F
Instrumentation: Gas and matrix IR spectra were recorded on an IFS28
FTIR or an IFS66v/S FTIR spectrometer (Bruker, Karlsruhe, Germany)
with resolutions of 2 cmÀ1 (gas phase) or 1 cmÀ1 (matrix) in the range
5000 400 cmÀ1 with DTGS detectors in combination with KBr beam
splitters; 64 scans were co-added for each spectrum.
Raman spectra were recorded with an RFS100/S FT Raman spectrometer
(Bruker, Karlsruhe, Germany) with a resolution of 4 cmÀ1. The sample was
5140
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
Chem. Eur. J. 2003, 9, 5135 5141

Polyoxy Compounds
5135 5141
condensed under high vacuum on a copper finger cooled with liquid
nitrogen and excited by a 500 mW Nd YAG laser (DPY301; ADLAS,
L¸beck, Germany).
[8] F. A. Hohorst, D. D. Desmarteau, L. R. Anderson, D. E. Gould, W. B.
Fox, J. Am. Chem. Soc. 1973, 95, 3866.
[9] K. I. Gobbato, M. F. Klapdor, D. Mootz, W. Poll, S. E. Ulic, H.
Willner, H. Oberhammer, Angew. Chem. 1995, 107, 2433; Angew.
Chem. Int. Ed. Engl. 1995, 34, 2244.
[10] G. Bednarek, G. A. Arg¸ello, R. Zellner, Ber. Bunsenges. Phys. Chem.
1996, 100, 445.
[11] H. Pernice, M. Berkei, H. Willner, G. Henkel, M. L. McKee, T. R.
Webb, G. A. Arg¸ello, unpublished results.
[12] F. E. Malanca, G. A. Arg¸ello, E. H. Staricco, R. P. Wayne, J. Phys.
Chem. A 1998, 117, 163.
[13] G. A. Arg¸ello, H. Willner, F. E. Malanca, Inorg. Chem. 2000, 39,
1195.
[14] M. A. Burgos, P. Garcia, G. A. Arg¸ello, H. Willner, Int. J. Chem.
Kinet. 2003, 35, 15.
UV spectra were recorded with a Perkin Elmer Lambda900 UV/Vis
spectrometer (Perkin Elmer, Norwalk, CT) with a resolution of 1 nm and
using quartz optics for both matrix-isolated and gas-phase samples. For the
latter a 10.5 cm optical pathlength, a 100 mL gas cell was used.
Calculations: All calculations were performed with the Gaussian98
software package[27] with density functional theory (DFT).[28] The molec-
ular geometries were first optimised to standard convergence criteria by
the DFT hybrid method with Becke×s nonlocal three-parameter ex-
change,[29, 30] the Lee, Yang and Parr correction[31] (B3LYP), and
a
6-311G(d,p) basis set. Then harmonic vibrational wavenumbers were
calculated using the optimised structure and an identical basis set and
method. The calculation of the most stable rotamers started with the (most
stable) all-trans-sym-CF3OC(O)OOOC(O)OCF3 and included only rotam-
ers with few changes in the molecule geometry, as other isomers show
strongly increasing energy. Relative energies of the rotamers (Table 2) and
¬
[15] D. Hnyk, J. Machacek, G. A. Arg¸ello, H. Willner, H. Oberhammer, J.
Phys. Chem. A 2003.
[16] G. A. Arg¸ello, H. Willner, J. Phys. Chem. A 2001, 105, 3466.
[17] S. von Ahsen, J. Hufen, H. Willner, J. S. Francisco, Chem. Eur. J. 2002,
8, 1189.
[18] S. von Ahsen, H. Willner, J. S. Francisco, Chem. Eur. J. 2002, 8, 4675.
[19] P. Garcia, H. Willner, M. A. Burgos, G. A. Arg¸ello, unpublished
results.
À
O O bond energies of different trioxides (Table 4) were evaluated for
standard conditions.
Nomenclature: E/Z nomenclature can be used, as well as cis/trans, to
determine the configuration of substituents at a bond. Usually this is found
to describe the connectivity around a double bond. The use of syn(-planar)/
anti(-periplanar) as part of the molecule name yields correct labelling of the
rotamer. In order to follow the nomenclature of related radicals such as
CF3OC(O)OO[18] (where the bond order in the peroxy unit is between 1
and 2, allowing the use of cis/trans) and to point out the importance of the
transoid configuration of the molecular chain, the cis/trans nomenclature is
preferred here. As the two dihedral angles in the COOOC bridge are in any
case around 908 (gauche configuration) the differentiation between
‡908, ‡ 908 or ‡908, À 908 is given by the prefix sym or asy (for symmetric
or antisymmetric), respectively.
[20] H. Sawada, Chem. Rev. 1996, 96, 1779.
[21] J. Czarnowski, H. J. Schumacher, Int. J. Chem. Kinet. 1981, 13, 639.
[22] S. Sander, H. Pernice, H. Willner, Chem. Eur. J. 2000, 6, 3645.
[23] M. W. Chase, J. Phys. Chem. Ref. Data 1996, 25, 551.
[24] D. E. Milligan, M. E. Jacox, A. M. Bass, J. J. Comeford, D. E. Mann, J.
Chem. Phys. 1965, 42, 3187.
[25] A. J. ArvÌa, P. J. Aymonino, H. J. Schumacher, Angew. Chem. 1960, 72,
327.
[26] G. A. Arg¸ello, H. Grothe, M. Kronberg, H. Willner, H. G. Mack, J.
Phys. Chem. 1995, 99, 17525.
[27] Gaussian 98 (Revision A.5), M. J. Frisch, G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, V. G. Zakrzewski, J. A.
Montgomery, R. E. Stratmann, J. C. Burant, S. Dapprich, J. M.
Millam, A. D. Daniels, K. N. Kudin, M. C. Strain, O. Farkas, J. Tomasi,
V. Barone, M. Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo,
S. Clifford, J. Ochterski, G. A. Petersson, P. Y. Ayala, Q. Cui, K.
Morokuma, D. K. Malick, A. D. Rabuck, K. Raghavachari, J. B.
Foresman, J. Cioslowski, J. V. Ortiz, B. B. Stefanov, G. Liu, A.
Liashenko, P. Piskorz, I. Komaromi, R. Gomperts, R. L. Martin,
D. J. Fox, T. Keith, M. A. Al-Laham, C. Y. Peng, A. Nanayakkara, C.
Gonzalez, M. Challacombe, P. M. W. Gill, B. G. Johnson, W. Chen,
M. W. Wong, J. L. Andres, M. Head-Gordon, E. S. Replogle, J. A.
Pople, Gaussian, Inc., Pittsburgh, PA, 1998.
Acknowledgement
Financial support by the Deutsche Forschungsgemeinschaft (DFG) as well
as a grant for the interchange of scientists supported jointly by DAAD
(Germany) and ANPCyT (Argentina) through the PROALAR program
are gratefully acknowledged.
[1] P. D. Bartlett, G. Guaraldi, J. Am. Chem. Soc. 1966, 89, 4799.
[2] J. E. Bennett, G. Brunton, R. Summers, J. Chem. Soc. Perkin Trans. 2
1980, 981.
[3] A. F. Khalizov, O. N. Makarova, S. L. Khursan, V. V. Shereshovets,
React. Kinet. Catal. Lett. 1995, 54, 427.
[28] W. Kohn, L. J. Sham, Phys. Rev. A 1965, 140, 1133.
[29] A. D. Becke, J. Chem. Phys. 1993, 98, 1372.
[30] A. D. Becke, J. Chem. Phys. 1993, 98, 5648.
[4] A. F. Khalizov, S. L. Khursan, V. V. Shereshovets, Kinet. Catal. 1999,
40, 194.
[31] C. Lee, W. Yang, R. G. Parr, Phys. Rev. B 1988, 41, 785.
[5] L. R. Anderson, W. B. Fox, J. Am. Chem. Soc. 1967, 89, 4313.
[6] J. Thompson, J. Am. Chem. Soc. 1967, 89, 4316.
[7] D. D. Desmarteau, Inorg. Chem. 1970, 9, 2179.
Received: March 13, 2003 [F4942]
Chem. Eur. J. 2003, 9, 5135 5141
www.chemeurj.org
¹ 2003 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5141