Reactive intermediates in olefin oxidations with chromyl chloride. IR-spectroscopic proof for O=CrCl2-epoxide complexes

Article abstract of DOI:10.1021/ic981131s

After photolytic induction (411 nm) of the reaction between chromyl chloride and olefins in low-temperature matrixes, for the first time primary products of this system have been isolated and identified: the reaction of propylene yields O=CrCl₂?O=CHCH₂CH₃ and O=CrCl₂-OCH₂CHCH₃, while in the case of trans- and cis-2-butylene, OCrCl₂-O=C(CH₃)(CH₂CH₃) as well as OCrCl₂?(trans-2,3-butylene oxide) and OCrCl₂?(cis-2,3-butylene oxide) are obtained, respectively. Epoxide formation proceeds under complete retention of stereochemistry, and the regioselective reaction route leading to the carbonyl products can be suppressed if all hydrogen atoms at the olefinic bond are substituted: tetramethylethylene is selectively oxidized to tetramethyl-ethylene oxide, which forms a complex with OCrCl₂. All products were characterized by IR spectroscopy, which was further supported by DFT calculations in certain cases. The mechanism leading to the formation of these products, their electronic states, and the implications of the matrix results on the thermal reactions of chromyl chloride are discussed.

Full text of DOI:10.1021/ic981131s

2106
Inorg. Chem. 1999, 38, 2106-2116
Reactive Intermediates in Olefin Oxidations with Chromyl Chloride. IR-Spectroscopic Proof
for OdCrCl2-Epoxide Complexes
Christian Limberg*
Universita¨t Heidelberg, Anorganisch-Chemisches Institut, Im Neuenheimer Feld 270,
D-69120 Heidelberg, Germany
Ralf Ko1ppe
Universita¨t Karlsruhe, Institut fu¨r Anorganische Chemie, Engesserstrasse, Geb.-Nr. 30.45,
D-76131 Karlsruhe, Germany
ReceiVed September 18, 1998
After photolytic induction (411 nm) of the reaction between chromyl chloride and olefins in low-temperature
matrixes, for the first time primary products of this system have been isolated and identified: the reaction of
propylene yields OdCrCl2‚‚‚OdCHCH2CH3 and OdCrCl2‚‚‚OCH2CHCH3, while in the case of trans- and cis-
2-butylene, OCrCl2‚‚‚OdC(CH3)(CH2CH3) as well as OCrCl2‚‚‚(trans-2,3-butylene oxide) and OCrCl2‚‚‚(cis-
2,3-butylene oxide) are obtained, respectively. Epoxide formation proceeds under complete retention of
stereochemistry, and the regioselective reaction route leading to the carbonyl products can be suppressed if all
hydrogen atoms at the olefinic bond are substituted: tetramethylethylene is selectively oxidized to tetramethyl-
ethylene oxide, which forms a complex with OCrCl2. All products were characterized by IR spectroscopy, which
was further supported by DFT calculations in certain cases. The mechanism leading to the formation of these
products, their electronic states, and the implications of the matrix results on the thermal reactions of chromyl
chloride are discussed.
Introduction
workup7-11 to the reduction of these oxychlorides by zinc. To
study the influence of the solvent on the product distribution
Sharpless et al. further investigated the CrO2Cl2/olefin system
in CH2Cl2.13 Various types of olefins yielded chlorohydrins,
epoxides, and dichloridessformed by cis-additions in each
casesas the three main products. These were claimed to
generally occur in all CrO2Cl2 oxidations of olefins (being
modified in cases by individual experimental conditions), so
an attempt was made to fit them into a general reaction pat-
tern:13 a hypothetical olefin complex of CrO2Cl2 was suggested
to either undergo a Cr-Cl bond insertion reaction (generating
a Cr(VI) chloroalkyl compound) or rearrange to a chroma-
oxetane in the first step. The latter can decompose via the
mechanism shown in Scheme 1 where the substrate as well as
the reaction medium was proposed to determine the main
reaction pathway. However, none of these intermediates pos-
tulated has ever been proved, and, moreover, they cannot
account for all of the final products found1-11 by researchers.
Recently, the matrix isolation technique has been successfully
employed to investigate the ethylene oxidation by CrO2Cl2 at
an early stage: the species OdCrCl2‚‚‚OdCHCH3, 1 (see eq
1), was identified by IR spectroscopy.14 The inorganic part of
1, the OdCrCl2 molecule, had often been mentioned in the
literature7-11,13,15 as a potentially important species in such
oxidation reactions. Its formation had even been predicted to
For quite some time the olefin oxidation with chromyl chlor-
ide (CrO2Cl2) seemed to proceed completely unspecifically, and
reports with the most different product palettes accumulated.1-6
The first selective oxidation was reported in 1969 by F. Freeman
et al. claiming that reductive workup of the reaction mixtures
in the presence of zinc enables the isolation of carbonyl
compounds in good yields.7 However, mainly disubstituted,
electron-rich olefins had been employed in this investigation.7-11
In 1973 K. B. Sharpless et al. found that chloroketones are
formed selectively if the reactions are carried out in acetone at
-78 °C,12 and they related the findings made on reductive
* Corresponding author. Fax: +49 6221 54 57 07. E-mail: Limberg@sun0.
urz.uni-heidelberg.de.
(1) Etard, M. A.; Moissan, H. M. C. R. Hebd. Seances Acad. Sci. 1893,
116, 434-437.
(2) Cristol, S. J.; Eilar, K. R. J. Am. Chem. Soc. 1950, 72, 4353-4356.
(3) Stairs, R. A.; Diaper, D. G. M.; Gatzke, A. L. Can. J. Chem. 1963,
41, 1059-1064.
(4) Freeman, F.; Cameran, P. J.; Dubois, R. H. J. Org. Chem. 1968, 33,
3970-3972.
(5) Miyaura, N.; Kochi, J. K. J. Am. Chem. Soc. 1983, 105, 2368-2378.
(6) Freeman, F. in Organic Syntheses by Oxidation with Metal Com-
pounds; Mijs, W. J., de Jonge, C. R. H. I., Eds.; Plenum: New York,
1986; Chapter 2, pp 41-118.
(7) Freeman, F.; Duboius, R. H.; Yamachika, N. J. Tetrahedron 1969,
25, 3441-3446.
(8) Freeman, F.; McCart, P. D.; Yamachika, N. J. J. Am. Chem. Soc. 1970,
92, 4621-4626.
(9) Freeman, F.; Yamachika, N. J. J. Am. Chem. Soc. 1970, 92, 3730-
(12) Sharpless, K. B.; Teranishi, A. Y. J. Org. Chem. 1973, 38, 185-186.
(13) Sharpless, K. B.; Teranishi, A. Y.; Ba¨ckvall, J.-E. J. Am. Chem. Soc.
1977, 99, 3120-3128.
(14) Limberg, C.; Ko¨ppe, R.; Schno¨ckel, H. Angew. Chem. 1998, 110, 512-
515; Angew. Chem., Int. Ed. 1998, 37, 496-499.
3733.
(10) Freeman, F.; Yamachika, N. J. J. Am. Chem. Soc. 1970, 92, 1214-
1219.
(11) Freeman, F.; Arledge, K. W. J. Org. Chem. 1972, 37, 2656-2662.
10.1021/ic981131s CCC: $18.00 © 1999 American Chemical Society
Published on Web 04/15/1999

Reactive Intermediates in Olefin Oxidations
Inorganic Chemistry, Vol. 38, No. 9, 1999 2107
13
Scheme 1. Postulated Mechanism for the Thermal Olefin Oxidation with CrO2Cl2
represent the driving force of the latter15 although there had
never been proof of its existence. Bearing this in mind, it seems
surprising that, in spite of the frequent observation of carbonyl
compounds as products in CrO2Cl2 olefin oxidations (vide
supra), intermediates such as 1, i.e. complexes of carbonyl
compounds to OdCrCl2, had never been proposed and are
missing in Scheme 1. Instead, the carbonyl products were
preferentially thought to be a result of the decomposition of
chlorohydrins12 or rearrangements of the free epoxides during
workup.7-11
However, after proving the intermediacy of 1, at least under
photolytic matrix conditions, the following three questions arise:
(1) Do other olefins show the same reactivity toward chromyl
chloride under matrix conditions, so that the findings made can
be generalized?
Experimental Methods. A Leybold Displex closed-cycle helium
refrigerator gave temperatures of ca. 10 K inside a shroud maintained
at a pressure below 10-7 Torr. Mixtures were prepared in a vacuum
line with use of standard manometric techniques and then deposited
on the cold support. A copper block was used for all the IR experiments,
and the slow spray-on deposition was favored for CrO2Cl2 with the
sample held at -65 °C.
The photolysis source was a 200-W medium-pressure arc. The
radiation was filtered with a Balzers interference filter (bandpass ca.
15 nm). IR spectra were recorded in reflection within the ranges of
4000-300 and 1000-200 cm-1 and with a resolution of 1 cm-1 on a
Bruker IFS 66 v spectrophotometer, respectively.
CrO2Cl2 (99.99% pure) and tetramethylethylene (99% pure) were
supplied by Aldrich and further purified by fractional condensation in
high vacuum prior to use. Argon, propylene (99.999 and 99.95% pure,
Messer Griesheim), and cis- and trans-2-butylene (99% pure, Aldrich)
were used as received. The compounds 2-butanone, acetone, propanal,
and trans-2,3-butylene oxide (99% pure, Aldrich), which were matrix
isolated in order to obtain information concerning the intensity ratios
of the bands displayed and to compare literature values, were distilled
before dilution with argon.
(2) Can regioselectivity be observed when unsymmetrical
olefins are reacted?
(3) What happens if attempts are made to separate the two
steps needed for the formation of 1, i.e., oxygen transfer and
hydrogen shift?
18O-enrichment of CrO2Cl2 can be achieved by the following
procedure: 100 mg (0.6 mmol) of CrCl3 and a trace of Zn were placed
into a flamed-out ampule provided with a stopcock in a glovebox. A
0.5 mL amount of H218O was co-condensed, and the mixture was stirred
until all of the halide had dissolved (green color). Subsequently, the
solution was frozen to -150 °C, and in a stream of argon 116 mg (5.0
mmol) of sodium, freshly cut in a glovebox, was added. Afterward the
mixture was cooled with ice, initiating a vigorous reaction of sodium
with the water and evolution of hydrogen. Stirring overnight yielded a
bright green sludgy suspension of Na3[Cr(18OH)6] in water containing
1.4 mmol of Na18OH which was frozen to 77 K and treated with 1
mmol of chlorine gas. After it was annealed, the mixture was stirred
for a period of 2 h during which it turned bright yellow. All volatiles
were removed, and the resulting solid, consisting mainly of NaCl and
Na2Cr18O4, was reacted with an excess of dry HCl gas. After 15 min
of reaction, a subsequent flash distillation in high vacuum followed by
a fractional condensation yielded 50 mg (0.3 mmol, 54% yield) of pure
CrO2Cl2 containing 70-80% of 18O as judged by infrared. IR data:
Cr18O2Cl2 (Cr18O16OCl2): 960 (992) [νas(CrO2)], 943 (951) [νs(CrO2)],
In the present contribution we address these questions with
the aim of obtaining more detailed mechanistic insights.
Experimental Section
Computational Method. Density functional calculations were
carried out with the Gaussian/DFT16 or Turbomole/DFT17 series of
programs. The B3LYP formulation of density functional theory was
used employing the basis set LANL2DZ16 or SV(P).17 For the purpose
of comparison the structure of 8 was additionally optimized using BP/
SV(P).17 Harmonic vibrational frequencies and infrared intensities were
predicted at those levels by numerical second derivatives using
analytically calculated first derivatives. No scaling was applied.
(15) Rappe´, A. K.; Goddard, W. A., III. J. Am. Chem. Soc. 1982, 104,
3287-3294.
(16) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson,
B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G. A.;
Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewski,
V. G.; Ortiz, J. V.; Foresman, J. B.; Peng, C. Y.; Ayala, P. Y.; Chen,
Wong, W. M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin,
R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J.
P.; Head-Gordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94;
Gaussian Inc.: Pittsburgh, PA, 1995.
502 [νas(CrCl2)], 461 [νs(CrCl2)] cm-1
.
The reaction of propylene with CrO2Cl2 under photolytic conditions
was performed as follows: 100 mg (0.6 mmol) of CrO2Cl2 was
condensed into a high-pressure tube provided with a stirring bar and
sealed with a greaseless valve. An excess of liquid propylene (ca. 3
(17) Ahlrichs, R.; Baer, M.; Haeser, M.; Horn, H.; Ko¨lmel, C. Chem. Phys.
Lett. 1989, 162, 165-169.

2108 Inorganic Chemistry, Vol. 38, No. 9, 1999
Limberg and Ko¨ppe
Figure 1. Upper trace: Difference between the IR spectrum of CrO2Cl2 isolated in an argon matrix doped with 8 mol % of propylene at 10 K and
the spectrum recorded of the same matrix after photolysis at 411 nm for 20 min. Lower trace: The line diagrams for the IR spectra of propanal,
acetone, propylene oxide, and OdCrCl2.
mL) was co-condensed, and the CrO2Cl2 dissolved at 100 K. The red
solution was cooled to 77 K again and stirred at that temperature. A
photolysis lamp (Hg, medium arc pressure) was placed directly beside
the solution in the dewar with liquid nitrogen, and the mixture was
irradiated for 30 min. Afterward the propylene was pumped off with
the sample held at -100 °C, and unreacted CrO2Cl2 was removed at a
temperature of -60 °C. Subsequently the sample was annealed and all
other volatiles were removed, leaving behind a brown powder. A 0.5
mL amount of wet [D3]acetonitrile was co-condensed, in which the
solid was dissolved, and all volatile compounds delivered by this process
were distilled together with the solvent into an NMR tube which was
flame-sealed. The product distribution was determined NMR spectro-
scopically by comparison with the spectra of authentic samples. To
compare the result with the nonphotolytic reaction the experiment was
performed as above, the only difference being that the sample was
slowly annealed to room temperature without prior photolysis.
The reactions of TME and 2,4,4-trimethyl-1-pentene were performed
in CH2Cl2 solutions, which were slowly annealed to room temperature.
After workup as above an NMR sample containing a pinacolon/TMEO/
acetone mixture (45/30/25% yield) was obtained in the case of TME,
while 2,4,4-trimethyl-1-pentanal (65% yield) was essentially pure. The
yields were determined by means of a standard.
Reaction of Chromyl Chloride with Propylene. Co-con-
densation of a mixture of propylene in Ar (8%) with CrO2Cl2
(-65 °C) at 10 K yielded an IR spectrum of the isolated starting
materials, which was subtracted from a spectrum recorded after
subsequent photolysis at 411 nm. The result is shown in Figure
1 (the negative bands belong to consumed starting material).
As in the case of ethylene14 the conversion is relatively high
(30%), i.e., all collisional pairs must have reacted. The bands
at 1009, 447, and 380 cm-1 in Figure 1 can be assigned to
OdCrCl2 by analogy with the CrO2Cl2/ethylene case (1009
ν(CrdO), 452 νas(CrCl2), and 385 νs(CrCl2) cm-1). The
satellites/shoulders beside some of these bands and those in
spectra which will be discussed later are due to matrix splitting
caused by the residence of trapped species in different matrix
sites.19 Additionally, the formation of a new ν(CdO) band at
1662 cm-1 can be observed (1669 cm-1 in the ethylene reaction
for OdCHCH3 (eq 1)), which could belong to propanal, acetone,
Results
or both: free matrix-isolated acetone20,21 shows a ν(CdO) band
at 1721 cm-1, while propanal22,23 causes a matrix-split band at
1746/1738 cm-1, but an exact prediction of the shifts resulting
from their complexation to OdCrCl2 cannot be made.24
Therefore it is necessary to get an idea of the complete band
Product Identification. Concentrations were chosen so that
statistically18 ca. 30% of the CrO2Cl2 molecules had one olefin
molecule as nearest neighbor (i.e., Ar/olefin ) 96/4) or higher.
Since molecules do not undergo diffusion in the resulting solid
matrixes at 10 K, the reservoir of collisional reactant pairs
formed upon deposition is depleted as the photolytically induced
reaction proceeds. The resulting new bands observed in the IR
spectra reported below were occasionally somewhat broader than
usually found in matrix isolation experiments, but the line widths
proved to be independent of the CrO2Cl2 concentration, which
excludes aggregation processes as a possible reason. They rather
seemed to be correlated with the size of the molecules generated,
probably due to conformational isomers.
(19) This effect is also responsible for the splitting of the bands belonging
to consumed CrO2Cl2 (compare for instance the negative bands at 447/
460 cm-1 in Figure 1) and has been observed in some of our
experiments mainly for the bands belonging to chromyl chloride or
the OCrCl2 unit and independently of type and concentration of the
olefins.
(20) Han, S. W.; Kim, K. J. Phys. Chem. 1996, 100, 17124-17132.
(21) Dellepiane, G.; Overend, J. Spectrochim. Acta 1966, 22, 593-601.
(22) Sodeau, J. R.; Beeby, A.; Whyte, L. J. Spectrochim. Acta 1990, 46A,
887-901.
(23) Frankiss, S. G. Kynaston, W. Spectrochim. Acta 1972, 28A, 2149-
2161.
(18) Bandow, H.; Akimoto, H. J. Phys. Chem. 1985, 89, 845-849.

Reactive Intermediates in Olefin Oxidations
Inorganic Chemistry, Vol. 38, No. 9, 1999 2109
Table 1. Frequencies Obtained Experimentally by the Photolysis of CrO2Cl2 Isolated in a Propylene/Ar (8%) and trans-2-Butylene/Ar (4%)
Matrix, Respectively, as Well as the Frequencies Displayed by Propanal,22,23 Acetone,20,21 Propylene Oxide,25 2-Butanone, and
trans-2,3-Dimethyloxirane When Isolated in Argon Matrixes in cm-1 a Together with Those Calculated for 2 and OCrCl2‚‚‚OdC(CH3)2
(B3LYP/LANL2DZ)
trans-
2,3-butylene
oxide
assignment
(refs 20, 21,
and 25-27)
CrO2Cl2/
propylene
cis-
propylene
oxide25
calcd
OdCrCl2‚‚‚acetone
CrO2Cl2/trans-
2-butylene
propanal22,23 acetone20,21
calcd 2
2-butanone
1662 (0.18)
1746/1738 (1) 1721 (1)
1462 (0.41)
1684 (1)
1644 (1)
1662/1650
(0.08)
1731/1710
(0.5/1)
1455 sh
1439 (0.21) 1491 (0.70) δ(CH3)
1420 (0.30) δ(CH3)
1394 (0.15) 1415 (0.70) δ(CH3)
1367 (0.50) 1386 (0.85) δ(CH3)
1350 (0.25) 1336 (0.12) ω(CH2)
1258 (0.08) δ(CH)
ν(CdO)
1476 (0.02)
1456 (0.04)
1443 (0.06)
1409 (0.15)
1376 (0.04)
1349 (0.01)
1262 (0.03)
1456 (0.12) 1527 (0.07)
1444 (0.10) 1524 (0.08)
1429 (0.20) 1460 (0.18)
1500 (0.14)
1486 (0.25)
1480 (0.09)
1434 (0.36)
1430 (0.22)
1483 (0.06)
1461 (0.04)
1449 (0.14)
1432 (0.04)
1389 (0.42)
1348 (0.02)
1252 (0.01)
δ(CH3)
1420 (0.40)
1392 (0.50)
1378 (0.12)
1336 (0.25)
1408 (0.35) 1406 (0.4) 1457 (0.17)
1362 (0.66) 1369 (0.06) 1428 (0.02)
1354 (0.4)
1383 (0.04)
1264 (0.18)
1217 (0.5)
1316 (0.20)
νas(CCC)
1163 (0.05)
1146 (0.01)
1101 (0.02)
1094 (0.02)
1026 (0.15)
1143 (0.06)
1130 (0.03)
1103 (0.12)
1179 br (0.02) 1181 (0.20) 1157 (0.10) F(CH3) i.p.
1122 (0.06)
1111 (0.06)
1115 (0.70) C-H bend
1094 (0.15) F(CH3) i.p.
1120 (0.05)
1094 (0.49)
1092 (0.25)
1129 (0.06)
1143 (0.03)
1050 (0.70)
CH3 rock (oop)
1022 (0.27)
1027 (0.46)
1016 sh
1007 (1)
1027 (0.90) CH3 rock
1012 (0.70) CH3 rock
ν(CrdO)
1005 (0.01)
1009 (1)
1046 (0.34)
1040 (0.22)
997 (0.03) 960 (0.25) ν(CCC)
941 (0.18)
906 (0.04)
896 (0.03)
821/810 (0.89) 851 (0.50)
729 (0.03)
951 (0.25)
893 (0.03)
δ(CH3)/ν(CO)
916 (0.06)
876 (0.14)
798 (0.66)
727 (0.03)
668 (0.06)
941 (0.06)
897 (0.02) 886 (1)
853 (0.11) 810 (1)
ν(CC)
892 (0.05)
882 (0.02)
CH3 rock (i.p.)
829 (1)
748 (0.15)
867 (0.06)
Fs(CH3)/νs(COC)
719 (0.5) ν(CO)
781 (0.01)
529 (0.25)
587 (0.05) 470 (0.7) CO i.p. bend
683 (0.06)
651 (0.10)
447/460 (0.47)
395 (0.03)
380 (0.22)
683 (0.05)
668 (0.03)
435 (0.29)
δ(CCC)
563 (0.10)
422 (0.28)
653 (0.06)
446 (0.86)
409 (0.02)
δ(CCO)
νas(CrCl2)
367 (0.21)
361 (0.10)
380 (0.20)
νs(CrCl2)
a Intensities are given in parentheses and were mostly determined by us.
pattern expected for both molecules OCrCl2‚‚‚OdCHCH2CH3
and OCrCl2‚‚‚OdC(CH3)2. In Figure 1 line diagrams of the
spectra of free matrix-isolated cis-propanal22,23 and acetone20,21
are shown.
enhanced on complexation of the epoxide by OCrCl2 (the bands
of the latter fragment are expected to coincide with the cor-
responding ones of 2). This assignment is further supported by
experiments employing 18O2CrCl2: the 810 cm-1 band shifts
by 23 cm-1 which is characteristic for νs(COC) modes (compare
the experimental and calculated spectra of 7 and 8 discussed
below). Accordingly, a line diagram for the IR spectrum of free
matrix-isolated propylene oxide25 has been included in Figure
1, too. Apart from the νs(COC) band two modes of medium
intensity at 951 (CH3 bend/CO stretch) and 1022 cm-1 (CH3
rock) as well as a weak band at 1264 cm-1 (CH def) can be
noticed in the region between 800 and 1300 cm-1 for free
propylene oxide, and these are indeed observed in the spectrum
in Figure 1 at 941, 1026, and 1262 cm-1. It would be tentative
to compare the residual very weak features of the experimental
spectrum to weak bands displayed by free propanal or propylene
oxide and to thereby assign them as belonging to either 2 or 3.
However, as can be seen, even on this level the match is more
than agreeable, and concludingly the band pattern found in the
experimental spectrum fits nicely to the one a mixture of free
propylene oxide and propanal would display. Complexation
naturally weakens the bonds of the coordinating functional
groups leading to a red shift of ν(CdO) and νs(COC) (84/76
and 19 cm-1, respectively). The bands at 951 cm-1 (propylene
oxide) and 851 cm-1 (propanal) both show red shifts on
complexation (compare Figure 1 and Table 1), too, as the first
band has a partly ν(C-O) character,25 while the second one is
expected to involve a ν(CdO) motion.
As can be seen, acetone exhibits medium intense bands at
1217 and 529 cm-1 (νas(CCC) and δ(CCO), respectively) which
cannot be found in our experimental spectrum, so acetone is
excluded as a product at the outset. On the other hand, a strong
band (810 cm-1) and a band of medium intensity (821 cm-1
)
can be found in the region between 1000 and 600 cm-1, where
acetone displays only one band of negligible intensity. Propanal,
however, shows a mode at 851 cm-1, whose intensity would
fit to the one of the 821 cm-1 band. The intensity of the 810
cm-1 band is far too high for such an assignment and
consequently it finds no match in the spectrum expected for
OCrCl2‚‚‚OdCHCH2CH3, 2. It should therefore have its origin
in a second species generated by photolysis of the starting
materials. The only possible product, which could be anticipated
to show such a band, is a propylene oxide adduct OCrCl2‚‚‚
OCH2CHCH3, 3 (eq 2): free propylene oxide shows a strong
The identification of propanal as the organic oxidation product
of the matrix experiment could additionally be supported by
band at 829 cm-1 [νs(COC)]25 and its intensity will be further
(24) Vancik, H.; Gabelica, V.; Mihalic, Z.; Sunko, D. E. J. Chem. Soc.,
Perkin Trans 1994, 1611-1614.
(25) Lowe, M. A.; Alper, J. S., Kawiecki, R.; Stephens, P. J. J. Phys. Chem.
1986, 90, 41-50.

2110 Inorganic Chemistry, Vol. 38, No. 9, 1999
Limberg and Ko¨ppe
preparative experiments simulating the matrix conditions. Pro-
pylene “contaminated” by CrO2Cl2 is still fluid at 77 K and
can be stirred. After slow annealing of such a mixture to room
temperature and isolation of the precipitated Etard complex1-7
by removal of the volatiles, the hydrolytic decomplexation of
the latter yields propanal, dichloropropane, chloroacetone,
acrolein, 2-chloropropanal and allyl alcohol besides the main
product 1-chloro-2-propanol. However, if the propylene/
CrO2Cl2 mixture is photolyzed at 77 K while stirring and the
volatiles are removed at low temperatures, a similar work up
yields a product distribution in which the propanal proportion
is significantly enhanced.
Reaction of Chromyl Chloride with trans-2-Butylene. After
the reaction of trans-2-butylene, yet again the bands charac-
teristic for the OdCrCl2 core were observed (1007 ν(CrdO),
446 νas(CrCl2) and 380 νs(CrCl2) cm-1, compare Table 1). By
analogy with the propylene case a likely assignment for a matrix
split band at 1662/1650 cm-1 is a 2-butanone complex 4 as
shown in eq 3, while an intense band at 798 cm-1 suggests the
Reaction of Chromyl Chloride with Tetramethylethylene
(TME). After the photolytic reaction of this olefin with
CrO2Cl2 at 10 K no band in the carbonyl region of the difference
infrared spectrum can be found (see Figure 2), while the intense,
matrix-split bands of the OdCrCl2 fragment (1009, 451, and
383 cm-1) are present as in the spectra discussed above. An
additional intense band at 816 cm-1 gives a first hint to a
complexation of this OCrCl2 core by tetramethylethylene oxide
(TMEO) (free matrix-isolated TMEO shows an intense ν(COC)
absorption at 847 cm-1).28 Additional support for an assignment
to OdCrCl2‚‚‚TMEO, 7 (eq 4) comes from experiments with
18O-enriched CrO2Cl2 (compare Figure 2, upper trace): apart
from ν(CrdO), which is shifted by 44 cm-1 (43 cm-1 in the
case of 1) only the band at 816 cm-1 shows a significant shift
(28 cm-1) and can therefore confidently be assigned to the ring
stretch vibration (the satellites observed beside ν(CrdO) and
νs(COC) belong to the 16O isotopomer which simultaneously
formed due to incomplete enrichment). Furthermore, as can be
seen in Table 2, the 1500-1100 cm-1 region of the spectrum
of 7, where mainly bands of the TMEO ligand appear, fits nicely
to the one in the spectrum of free TMEO.29
DFT Calculations. In support of the assignments DFT
calculations were carried out, which had already helped the
identification of 1.14 Calculations concerning CrO2Cl2 itself had
previously established that at the B3LYP/LANL2DZ level of
theory14,30 deviations of 100 cm-1 are nothing unusual and have
their origins in very small differences between the theoretical
and real structures.14,31,32 It has further been shown,30 thats
although in comparison to ab initio studies DFT calculations
are more suitable for frequency calculations concerning species
of this typesthe sign of the errors made is not constant, making
the application of a scaling factor unreasonable. However, in
the case of 1 the general band pattern was found to be in
accordance with the experimental one.
(a) CrO2Cl2/CH3CHdCH2. According to qualitative argu-
ments, the pattern of those bands remaining in the experimental
spectrum of Figure 1, if the bands assigned to 3 are not con-
sidered, fits to the one which would be expected for a propanal
complex 2, while an acetone complex should give rise to a
different spectrum. Structures of 2 and OCrCl2‚‚‚OdC(CH3)2
were optimized33 (LANL2DZ) in form of their singlet and triplet
states with C1 symmetry. The triplet states proved to be much
more stable (vide infra), and the results of the subsequent
frequency calculations for those are shown in Table 1. It
becomes obvious that all of aboves statements and assumptions
concerning both frequencies and intensities are correct, so that
the outcome of the matrix experiment can definitely be
summarized by eq 2. However, it cannot be excluded that minor
amounts of OCrCl2‚‚‚OdC(CH3)2 have been formed simulta-
neously and are just not detected due to masking.
formation of a trans-2,3-butylene oxide complex 5 (the free
epoxide isolated in an argon matrix shows a strong νs(COC)
band at 810 cm-1 (see Table 1)). trans-2,3-Butylene oxide
further shows two strong bands at 1027 and 886 cm-1 (identified
as medium intense bands at 1027 and 876 cm-1 in case of 5),
two bands of medium intensity at 1012 and 719 cm-1 (observed
as a shoulder at 1016 and a weak absorption at 727 cm-1 in the
product spectrum) as well as a medium intense band around
470 cm-1 (the corresponding mode of 5 probably gets perturbed
by the subtraction of νs(CrCl2) of CrO2Cl2 at 467 cm-1).
Medium intense bands of free trans-2,3-butylene oxide and
2-butanone at 1115 and 1094 cm-1, respectively, appear for their
complexed versions at 1122 and 1111 cm-1 with decreased
intensities. The remaining bands of free 2-butanone, shown
within the region of 500-1300 cm-1, are only weak and are
easily masked by more intense bands in the experimental
spectrum.
Reaction of Chromyl Chloride with cis-2-Butylene. As
expected, bands which had previously been assigned to 4 also
appeared in this case after photolysis. However, none of the
intense bands of 5 could be detected. Instead the most intense
and therefore characteristic band of 5 at 798 cm-1 seemed to
have been shifted to 766 cm-1 (∆ ) 34 cm-1), which
corresponds well to the wavenumber difference between the
νs(COC) bands of free trans- and cis-2,3-butylene oxide (∆
) 32 cm-1). A pendant for the second predominant band of 5
at 1027 cm-1 was not observed when cis-2-butylene was
employed, probably due to masking: when the positions of
the corresponding bands displayed by the free molecules are
compared, again a shift (∆ ) 33 cm-1) to lower wavenumbers
can be observed on going from trans- to cis-2,3-butylene oxide;
if it is assumed that this is characteristic for the OCrCl2
complexes, too, the band would be placed in a region where
intense bands of the starting materials appear. As a conse-
quence, it is reasonable to suggest the formation of a complex
OCrCl2‚‚‚(cis-2,3-butylene oxide), 6, showing a spectrum
significantly different from the one that 5 displays.
(26) Durig, J. R., Feng, F. S.; Wang, A.; Phan, H. V. Can. J. Chem. 1991,
69, 1827-1844.
(27) Durig, J. R.; Nease, A. B.; Rizzolo, J. J. J. Mol. Struct. 1982, 95,
59-84.
(28) Nakata, M. Spectrochim. Acta A 1994, 50, 1455-1465.
(29) Note that the intensity ratios shift in advance to νs(COC) due to
complexation (compare for instance the ratios calculated for 7 (vide
infra) with those calculated for and displayed by free TMEO in Table
2).
(30) Torrent, M.; Gili, P.; Duran, M.; Sola`, M. J. Chem. Phys. 1996, 104,
9499-9510
(31) Be´rces, A.; Ziegler, T. J. Phys. Chem. 1995, 99, 11417.
(32) Be´rces, A. J. Phys. Chem. 1996, 100, 16538.
(33) The flexible C3 backbone of the propanal ligand in 2 opens up many
possibilities for structural arrangements, which all show slightly
different IR spectra. We have optimized four of them, and the
frequencies of the one with propanal in a cis conformation (the most
stable conformation of free propanal) are given in Table 1.

Reactive Intermediates in Olefin Oxidations
Inorganic Chemistry, Vol. 38, No. 9, 1999 2111
Figure 2. Middle trace: Difference between the spectrum of CrO2Cl2 isolated in an argon matrix doped with 4 mol % of tetramethylethylene at
10 K and the spectrum recorded of the same matrix after photolysis at 411 nm for 20 min. Upper trace: Difference spectrum obtained via the same
procedure as described above but using 18O-enriched CrO2Cl2. Lower trace: The line diagram calculated for the IR spectrum of OCrCl2‚‚‚TMEO,
7, shown in Figure 3 by means of DFT calculations using SV(P) basis sets (B3LYP).
Table 2. Frequencies Obtained Experimentally by the Photolysis of Cr16O2Cl2 and Cr18O2Cl2 Isolated in a Tetramethylethylene/Ar (4%)
Matrix, as Well as Calculated Frequencies (B3LYP, SV(P)) for OCrCl2‚‚‚TMEO, 7, and Free TMEO in cm-1 a
16O2CrCl2/TME
expt ∆(16O-18O)
calcd 16O-7
calcd ∆(16O-18O)
expt28 TMEO
calcd TMEO
assignment
1483 (0.12)
1444 (0.09)
1434 (0.12)
1396 (0.05)
1386 (0.50)
1214 (0.16)
1176 (0.37)
1117 (0.25)
1087 (0.06)
1016/1009 (1)
816 (0.60)
0
0
0
0
0
0
0
2
0
1502 (0.29)
1484 (0.05)
1469 (0.09)
1461 (0.06)
1410 (0.25)/1414 (0.15)
1232 (0.15)
1200 (0.33)
0
1
0
0
1/0
0
0
1484 (0.2)
1471 (0.1)
1438 (0.1)
1406 (0.1)
1381 (0.7)
1205 (0.3)
1173 (1.0)
1142 (0.3)
1501 (0.43)
1482 (0.13)
1437 (0.13)
1421 (0.10)
1409 (0.40)
1220 (0.58)
1199 (1)
CH3 bend
CH3 bend
CH3 bend
CH3 bend
CH3 bend
νas(CCC)
δ(CCC)
1160 (0.67)
2
1180 (0.50)
F(CH3)
44
28
1091 (1)
54
28
14
19
0
1
0
ν(CrdO)
νs(COC)
νas(COC)
δ(CrOC)
δ(CrOC), F(CCC)
νas(CrCl2)
ω(CCC)
F(CCC)
νs(CrCl2)
860 (0.93)
662 (0.19)
525 (0.14)
459 (0.29)
455 (0.14)
423 (0.13)
385 (0.19)
370 (0.14)
847 (0.2)
881 (0.40)
474 (0.60)
451 (0.43)
441 (0.50)
421 (0.43)
383 (0.50)
373 (0.40)
0
2
0
3
3
2
2
1
a Intensities are given in parentheses. The assignments are based on the spectrum of free TMEO28 whose experimental and calculated frequencies
are given for comparison, too, as well as the results of the calculation.
(b) CrO2Cl2/(CH3)2CdC(CH3)2. Calculations were also
performed in the case of 7, where the spectrum is complicated
by the relatively large organic moiety making a straightforward
interpretation difficult. During the structure optimization (SV-
(P))34 for a OCrCl2‚‚‚TMEO complex with Cs symmetry, the
red shift of νs(COC) (31 cm-1). Only few crystal structures of
metal epoxide complexes are known,35-37 and these display
M‚‚‚O distances of 2.2-2.4 Å. The angles between the M‚‚‚O
(34) SV(P) basis sets were used in this case, as the results of geometry
optimizations14 were very good when they were used in combination
with the BP functional of the Turbomole program,17 which, however,
is not suitable to determine the corresponding IR spectra.
(35) Groves, J. T.; Han, Y.; Van Eugen, D. J. Chem. Soc., Chem. Commun.
1990, 436-437.
(36) Darensbourg, D. J.; Holtcamp, M. W.; Khandelwal, B.; Klausmeyer,
K. K.; Reibenspies, J. H. J. Am. Chem. Soc. 1995, 117, 538-539.
(37) Harder, S.; Boersma, J.; Brandsma, L.; Katers, J. A.; Duisenberg, A.
J. M.; van Lenthe, J. H. Organometallics 1990, 9, 511.
3
small HOMO-LUMO gap indicated that the triplet state A′′
is significantly favored over the singlet state. Accordingly, the
3A′′ structure was optimized, and the result is shown in Figure
3. The structure shows qualitatively all features which could
have been expected considering the experimental spectrum: the
CrdO bond is shorter (1.552 Å) than in CrO2Cl2 and the bond
to TMEO is very strong (2.000 Å), explaining the significant

2112 Inorganic Chemistry, Vol. 38, No. 9, 1999
Limberg and Ko¨ppe
reactions observed in low-temperature matrixes, when they are
induced photochemically. In light of these findings a band38 of
medium intensity at 852 cm-1 in the spectrum,14 which had
been obtained after photolysis of the ethylene/CrO2Cl2 system
may also be interpreted in terms of an epoxide adduct OdCrCl2‚
‚‚OCH2CH2, 8. A calculation (B3LYP/LANL2DZ) proved39 that
indeed, besides the bands of the OCrCl2 fragment, which are
masked by those of 1, 8 should show just one intense band
(calculated at 835 cm-1) with the 18O shift found almost
exactly.38 Comparison to the corresponding band position of
the free epoxide25 yields a shift of 25 cm-1 as a result of
complexation, fitting nicely to the shifts found for 3 and 7 and
confirming finally the simultaneous formation of 8 with 1. A
comparison of the ν(CdO)/νs(COC) ratios of the pairs 1/8, 2/3,
and 4/5 reveals a steady decrease of these values within the
series of experiments involving ethylene/propylene/2-butylene.
This suggests that epoxide products are becoming more and
more important when the degree of methyl substitution at the
olefin is increased, although a quantitative statement is not
possible due to the unknown extinction coefficients of the
corresponding bands. This finding is in agreement with the
results of H. Frei et al. who studied the photooxidation of matrix-
isolated olefins by NO2.40 Our results obtained on employment
of CD2dCD2 further prove, that the H migration step during
carbonylation is not rate determining, as the product ratio of
8D/1D is not higher than the one of 8/1,5 as judged by the
ν(COC)/ν(CdO) intensity ratios. All original bonds at the olefin
should therefore still be mainly intact in the transition state
leading to the carbonyl products, so that the influence of methyl
substitution on this barrier should be confined to electronical
and steric factors becoming effective on approximation of the
two reactants. This suggested the further experiment employing
TME to test whether the aforementioned trend would lead to
exclusive epoxidation in this case. As described in the Results,
the selective formation of 7 could be confirmed. A methyl shift,
which would have yielded a pinakolon adduct, did indeed not
occur and the fact that no other product has been isolated instead
mean that the barrier leading to the carbonyl compounds can
no longer compete successfully with epoxide formation anymore
if the olefins are peralkylated. Our calculation (B3LYP/SV(P))
yielded a reaction enthalpy of ∆H ) -150 kJ/mol for eq 4 and
Figure 3. Optimized Cs (3A′′) structure calculated for OCrCl2‚‚‚TMEO,
7, by means of DFT calculations using SV(P) basis sets (B3LYP).
Selected bond lengths and angles are given in angstroms and degrees,
respectively.
vectors and the epoxide planes found there are in good
agreement with the angle found for 7 (125.8°). The calculated
IR spectrum based upon this geometry further confirmed the
assignment of the experimental IR spectrum to 7. As can be
seen in Figure 2 and Table 2, its band pattern and intensity
ratios could be reproduced nicely, especially if the general com-
ments made above and the fact that no scaling factor has been
applied are considered. The calculation predicted the position
of the ν(COC) band at 860 cm-1 with a theoretical 18O shift of
28 cm-1, the latter being in exact agreement with the experi-
mental result. To obtain information about the errors made in
the calculation of the theoretical IR spectrum of 7, additional
calculations concerning the vibrational frequencies of free
TMEO were performed. It is evident from Table 2 that the
deviations of the calculated frequencies from the experimental
ones are in excellent agreement with those obtained for 7.
Discussion
Mechanistic Inferences. To find answers to the questions 1
and 2 of the Introduction, propylene, which combines asym-
metry with simplicity, had been chosen as an appropriate
substrate first to be reacted with CrO2Cl2. The results show that,
in agreement with the observations13 made in thermal oxidations
of terminal olefins by CrO2Cl2, the carbonylation proceeds
regioselectively at the carbon with a lower degree of substitution,
yielding complex 2 as the thermodynamically less stable product
(the acetone complex was calculated to be energetically favored
by ca. 50 kJ/mol). This makes the existence of a transition state
likely which discriminates between the two possible products,
either by way of its formation or via a selective decomposi-
tion pathway. Furthermore, an epoxidation of propylene by
CrO2Cl2 producing 3 was observed, and in order to obtain
information about the stereochemistry of this reaction, trans-
and cis-2-butylene were employed as further reactants. The
bands for 5 could not be detected in the cis-butylene experiment,
and those of 6 were not present in the spectrum when trans-
butylene was reacted. This observation of complete stereochem-
istry retention in the epoxides (which again fits to the findings
made by Sharpless13 in the thermal oxidations with CrO2Cl2)
implies that if the reaction proceeds via an intermediate, internal
rotation around the central C-C bond is either not possible in
principle or is slow on the scale of the intermediates lifetime.
These results further clarify that H shift coupled oxygenation
and epoxidation of olefins by CrO2Cl2 are in fact typical
for purposes of comparison the one for the oxidation of ethylene
to yield 8 was calculated, too, employing two different func-
(38) This band, which shows an 18O shift of 26 cm-1 (compare 23 cm-1
for 3 and 28 cm-1 for 7) and which had not been in agreement14 with
the frequencies calculated for 1 with respect to its isotopic shift and
its intensity, had only tentatively been assigned to a δ(CCO) vibration
of 1, as seemingly there were not enough bands in the spectrum to
accommodate a second product and also since it proved impossible
(vide infra) to alter its intensity ratio with respect to ν(CdO) in various
photolytic cycles. In addition, all bands exhibited an identical growth
behavior which normally suggests an assignment to a single species.
(39) Calculated frequencies (intensities) for 8: 375 (0.34, νs(CrCl2)), 426
(0.39, νas(CrCl2)), 680 (0.09), 830 (0.13), 835 (1, νs(COC), ∆D 117
and ∆18O 29), 1051 (0.91, ν(CrdO), ∆18O 45), 1281 (0.24), 1501
(0.06).
(40) Frei, H. Chimia 1991, 45, 175-190.

Reactive Intermediates in Olefin Oxidations
Inorganic Chemistry, Vol. 38, No. 9, 1999 2113
Scheme 2. Mechanistic Possibilities for the Photolytic Olefin Oxidation with CrO2Cl2 under Matrix Conditions vs Relevant
Intermediates for the Thermal Process According to Calculations42,43
tionals and basis sets: B3LYP yielded -113.2 (SV(P)) and
-115.4 kJ/mol (LANL2DZ), while BP gave +6.5 kJ/mol (SV-
(P)) (compare +36.8 kJ/mol as found by T. Ziegler et al. in an
NL-P approach/STO, vide infra). The considerable differences
within these values stimulated further calculations concerning
the enthalpy of the reaction of CrO2Cl2 with ethylene to yield
free oxirane and 3[OdCrCl2], as a value of +104.4 kJ/mol can
be deduced from experimental data41 which allowed a validation
of the theoretical data in this case. B3LYP yielded +9.0 (SV-
(P)) and +120.2 kJ/mol (LANL2DZ), while BP gave +97.8
kJ/mol (SV(P)). All three calculated energies therefore agree
concerning the endothermic nature of this reaction, and two of
them are even close to the experimental value. Their comparison
with the data calculated for the formation of 8 provides an idea
about the energy, which is released on complexation of organic
O donors by OCrCl2 and which compensates the loss of energy
for the “pure oxidation process” (of course the oxidation product
is never really separated from the Cr center in the real
experiment). However, different functionals and basis sets seem
to give quite different values for these complexation energies
(-91.2 to -235.6 kJ/mol). Accordingly, it is difficult to judge
how exothermic these reactions really are, but, in any case, the
reactions proceed in the matrix due to selective excitation of
CrO2Cl2 (vide infra) and thermally due further exothermic
aggregation processes (formation of the Etard complex), through
which the species are constantly removed from any potential
equilibrium.
isomerization between them could be induced (it was impossible
to significantly change the ratios by photolysis at any wave-
length; photolysis with UV light decomposed both products).
Especially the latter finding is important, since it means that
the epoxides and carbonyl compounds are formed via indepen-
dent routes, i.e., the complexes of the carbonyl compounds were
not generated by rearrangement of the epoxide complexes. The
reaction paths shown in Scheme 2 for propylene could account
for all findings made so far. A species such as 3 had already
previously been suggested as one of the intermediates formed
in reactions of alkenes with CrO2Cl2.13 There it was proposed
to have its origin in a chromaoxetane (see Scheme 1), which
reductively eliminates the epoxide. However, recent DFT
calculations by T. Ziegler et al.42 have shown that a [2+3]
addition of ethylene to the CrdO bonds in CrO2Cl2 (which
would yield a glycolate G in the propylene case) is favored
over the [2+2] addition to one CrdO linkage both kinetically
and thermodynamically [on the singlet as well as the triplet
potential energy surface (PES)]. These DFT results suggest
exclusion of chromaoxetanes as the intermediates responsible
for product formation during our matrix experiment, too. Very
recently Ziegler et al.43 calculated the potential energy surface
for the thermal epoxidation of ethylene by CrO2Cl2 assuming
that 8, the species we have isolated and identified not only as
such but also in form of its methylated derivatives 3, 5, 6, and
7 as part of the present work, is an intermediate product. They
found that the kinetically and thermodynamically most favored
reaction route leads from the singlet starting materials via a
singlet transition state (see Scheme 3) to a triplet [2+3]
cycloadduct. The resulting five-membered ring G′ is sufficiently
flexible to allow for a facile rearrangement route to yield 8
via a modest activation barrier. In principle the products 1-8
isolated in the present study could therefore have their origins
The ν(CdO)/νs(COC) ratios within the pairs 1/6, 2/3, and
4/5 were not found to vary significantly with changes concerning
the external conditions (matrix apparatus and photolysis lamp).
This means that the two different productsscomplexes of
carbonyl compounds and epoxidessform within these systems
always roughly in the same relative yields (not overall yields!),
which additionally proved to be independent of the irradiation
wavelength used to generate the products. Furthermore, no
(42) Torrent, M.; Deng, L.; Ziegler, T. Inorg. Chem. 1998, 37, 1307-
1314.
(43) Torrent, M.; Deng, L.; Duran, M.; Sola´, M.; Ziegler, T. Can. J. Chem.,
(41) Ebbinghaus, B. B. Combust. Flame 1995, 101, 311-338.
submitted.

2114 Inorganic Chemistry, Vol. 38, No. 9, 1999
Limberg and Ko¨ppe
Scheme 3. Thermal Epoxidation Mechanism According to DFT Calculations43 (Enthalpies Given in kJ/mol)
in such glycolates, which would have to decompose photolyti-
cally via two independent routes. However, this does not seem
very likely as the reaction to G′ has been calculated to be
exothermic by -54.4 kJ/mol. Therefore such intermediates
should have been observable at some stage during our experi-
ments, as long as their lifetimes under photolytic matrix
conditions are not too short to prevent detection. Their UV/vis
absorption properties are not expected to be markedly different
from those displayed by the OCrCl2 complexes, meaning that
if a photolytic conversion of G′ to 8 is possible, the reverse
reaction should be possible, too. Consequently, such short
lifetimes would require very exothermic reactions of the
glycolates to those complexes (in all cases!) combined with very
low activation barriers for these processes. As can be seen in
Scheme 3, this is not at all the case according to the calcula-
tions43 for ethylene and the same is expected for propylene, so
it should have been possible to observe the glycolates at least
in these reactions if they had been the intermediates responsible
for epoxide formation. Their presence would have become most
obvious through ν(Cr-O) bands44 in the region between 750
and 500 cm-1; in our experiments, however, only in the case
of propylene a weak band has been observed.
How are the compounds 1-8 produced if not via glycolates
or chromaoxetanes? The starting situation in the matrix experi-
ments is different from the one under thermal conditions: at
the outset there are weak collisional CrO2Cl2/olefin complexes
formed upon deposition (neither in photolytical nor annealing
experiments were we able to observe classical olefin complexes
as suggested previously13 (compare Scheme 1) at any stage)
and only those react on photolysis as indicated by the yields
obtained. Formation of the epoxide complex OCrCl2‚‚‚
evidence for isomerization of 3 to 2 under matrix conditions,
so there should be an alternative route to 2. J. M. Mayer et al.
found a small loss of stereochemistry in an oxo transfer reaction
from an epoxide to a tungsten center, which they explained by
a ring-opened radical species containing the fragment W-OCH2-
• 45
CH2 . An analogous species Cl2(O)CrOCH2(CH3)CH• as
shown in Scheme 2 (path ii) should be considered as a transition
state or even intermediate, too: it has been established that the
CrdO double bond of CrO2Cl2 in its ground statesalthough
having no principle radical characterscan abstract a hydrogen
atom from organic substrates.48 If it attacked olefins in a similar
manner, this would produce the species mentioned above with
the necessary condition that the resulting C-O bond (whose
bonding energy is not known) is strong enough to represent
the driving force of the reaction. Addition of an organic radical
to CrO2Cl2 is likely to be very exothermic,48 so that the loss of
π-bonding energy could be compensated. The regioselectivity
found would then be explained by generation of the most stable
radical and a subsequent 1,2-H migration should yield 2.
According to DFT results43 a radical-like transition state as
shown in Scheme 2 should show a ring closure to a glycolate,
but the carbon carrying the unpaired electron would have to
come close to the CrdO bond, which might not be possible
within the rigid Ar lattice at 10 K, so that 1,2-hydrogen
migration becomes faster.
Consequently, a mechanism as in Scheme 2 could explain
all experimental findings but the electronic states of the products
still deserve some comments: in the case of 1 and 2 the energy
differences between triplet and singlet states ∆H(T-S) were
calculated (B3LYP/LANL2DZ) to amount to ca. -70 kJ/mol
(with slight deviations dependent on individual conformers),
while -118 kJ/mol was found for 8, in agreement with the
findings made for 7. This is what should be expected for d2
complexes with distorted tetrahedral geometries on grounds of
simple ligand field considerations: in the ground state the two
d orbitals of lowest energy are occupied with one electron each
and to reach a singlet state the spin pairing energy has to be
overcome, which is far higher than ∆Ed-d. Without calculating
the structures of products 3-6, it is therefore reasonable to
assume that in general the triplet states are energetically favored
OCH2CHCH3, 3, can then be imagined via a direct interaction
between one CrdO group and the olefin. This would account
for the retention of stereochemistry during epoxide formation
in our experiments, and there are already precedent cases in
the literature where intermediates/transition states with the
characteristics of the species in path i (Scheme 2) are
discussed.45-47 As pointed out above, we have found no
(44) Ault, B. S. J. Am. Chem. Soc. 1998, 120, 6105-6112.
(45) Atagi, L. M.; Over, D. E.; McAlister, D. R.; Mayer, J. M. J. Am.
Chem. Soc. 1991, 113, 870-874.
(46) Schiøtt, B.; Jørgensen, K. A.; J. Chem. Soc., Dalton Trans. 1989,
2099-2107.
(47) Samsel, E. G.; Srinivasan, K.; Kochi, J. K. J. Am. Chem. Soc. 1985,
107, 7606.
(48) Cook, G. K.; Mayer, J. M. J. Am. Chem. Soc. 1994, 116, 1855-1868;
1995, 117, 7139-7156.

Reactive Intermediates in Olefin Oxidations
Inorganic Chemistry, Vol. 38, No. 9, 1999 2115
over the singlet states. Further, the π* character of both SOMOs
in such complexes is not distinct, so that the rather strong
CrdO bond (compare ν(CrdO) of 1009 cm-1) is only slightly
weakened by the partial occupation of these orbitals. This
destabilizing effect is even less pronounced, if the orbital of
lower energy, which is less antibonding in comparison to the
one of higher energy, accommodates both electrons (i.e., in a
singlet state). This manifests in the IR spectra of 11 and 31, for
which, as an example, a ∆ν(CrdO) value of 70 cm-1 was
calculated. It was found that the deviations of the band positions
in the spectrum calculated for 31 from those in the experimental
spectrum are much smaller than in the case of 11, thereby
supporting the assignment of all spectra to the triplet molecules.
Reaction in the matrix occurs on photoexcitation at 411 nm
(mainly a ClfCr CT transition49), yielding an excited singlet
state initially, which may well be the state showing the reaction
with the olefins. However, as ethylene is a singlet molecule,
too, somewhere along the reaction coordinate certain transient
configurations have to change their spin state to yield the triplet
complexes. Although the selection rule forbids such intersystem
crossings, spin-orbit coupling occurring in transition metal
complexes can enhance the probability that a change of spin
state occurs at or near the crossover point of the two different
energy surfaces50 (note that the thermal oxidation of olefins by
CrO2Cl2 resulting in paramagnetic Cr(III/IV) species48 is forbid-
den, too). Absorption, emission, and excitation spectra have
further shown that CrO2Cl2 possesses a low-lying triplet
state51-53 and that it shows phosphorescence from this state on
excitation at 436 nm.54 Consequently, this state is accessible in
our experiments, too, and if this was responsible for product
formation a forbidden spin crossover would not even have to
be considered. In any case, after excitation the reactions appear
to be very fast, so that there must be two low-barrier reaction
paths (probably involving intermediate geometries such as the
ones depicted in Scheme 2) producing either complexes of
epoxides or carbonyl compounds. The probability of the excited
collisional complexes reacting via one route or the other should
depend on the electronic conditions of the olefins employed,
explaining the effects of methylation. The possibility that the
two different products are formed due to a “two-state reactivity”
cannot be excluded, but such an explanation seems to be
inconsistent with the relative product yields observed in the
series of olefins from ethylene to TME and the constancy of
these yields, implying that both paths have a common outset.
tions is probably different from the one the system moves on
when the reactions are induced thermally. Light of the wave-
length 411 nm is far below the CrdO dissociation threshold so
that CrO2Cl2, which shows a visible absorption band nearby,
takes up energy via this transition to reach an excited state
capable of reacting with olefins under matrix conditions.
Consequently the first transition states and intermediate geom-
etries which have to be passed will be different from the thermal
ones. However, as outlined above, in order to account for the
products of the thermal reactions a spin crossover has to occur
somewhere along the reaction coordinate, so that the thermal
and photochemical PES’s will meet at some stage. We consider
it as being very likely that the products isolated in the present
matrix study represent such touching points. This conclusion
can be deduced from our experimental study in combination
with the theoretical investigation by Ziegler et al., both of which,
for the case of ethylene, yielded the same intermediate species
8. This finding in turn suggests that the OdCrCl2‚‚‚OdCRR′
complexes we have identified are in fact intermediates during
thermal reactions, too. At least their incorporation into the Etard
complex would explain the frequent observation of carbonyl
compounds as products of such oxidations more nicely than
subsequent reactions of certain endproducts during workup.
Furthermore this statement is supported by the results of H.
Schwarz et al., who were able to show that, analogously to CrO2-
Cl2 in low-temperature matrixes, CrO2+ reacts with ethylene in
the gas phase55 (i.e., under thermal conditions) in a metal-
assisted oxidation linked to a 1,2-hydrogen migration to yield
acetaldehyde (OdCr+‚‚‚OdCHCH3 has been identified among
the products in the gas phase, while 1 was the matrix product).
After the formation of the aforementioned intermediates, of
course, other barriers, which are too high for matrix experiments,
can be crossed thermally, as for instance the one for the
isomerization of epoxides to carbonyl compounds catalyzed by
Lewis acidic centers.
In summary, the characterization of species isolated in such
matrix experiments is instructive for revealing the mechanism
of oxidation reactions with chromyl chloride, even if allowance
is made for different formation paths under photolytical and
thermal conditions. Proving the existence and intermediate
stability of compounds such as 1-8 provides strong evidence
that they are likely intermediates also in the “real” oxidation
processes. A very recent investigation44 by B. S. Ault concerning
the reaction of chromyl chloride with MeOH has proved, that
indeed the products generated via photolytic reaction in the
matrix are identical with those formed in the primary steps
thermally. Further support also comes from the fact that the
main features observed in the thermal reactions do correspond
quite well to the matrix results: as mentioned in the Introduction,
previous reports had claimed that olefin oxidation by CrO2Cl2
can be directed toward selective formation of carbonyl com-
pounds by reductive workup, for instance, in the cases of TME
and 2,4,4-trimethyl-1-pentene which are oxidized to pinakolon
and 2,4,4-trimethyl-1-pentanal in 50 and 78% yields, respec-
tively.7-11 However, in our hands these carbonyl compounds
were obtained in good yields without reductive workup. In the
case of TME the main side product was TMEO (eq 5) and it
follows, that oxidation of TME, as just one example, can in
principle yield both epoxide and ketone, the latter being favored
under the typical preparative conditions, but this might be
different for other olefins. This is exactly what we have found
studying the intermediates in our matrix experiments. It can
In our opinion, therefore, the results obtained in the present
matrix study are explained best by a reaction of triplet CrO2Cl2
with olefins in close proximity producing either epoxides via
route i or carbonyl compounds via path ii. Besides being
reasonable chemically as outlined above, this mechanism
accounts for retention of stereochemistry in epoxide formation,
the regioselectivity found for the generation of the carbonyl
compounds, and does not require a forbidden change of spin to
yield the most stable products.
Implications on the Thermal Oxidation of Olefins by
Chromyl Chloride. According to arguments above the PES
investigated in the matrix experiments under photolytic condi-
(49) Jasinski, J. P.; Holt, S. L.; Wood, J. H.; Asprey, L. B. J. Chem. Phys.
1975, 63, 760.
(50) Poli, R. Chem. ReV. 1996, 96, 2135-2204.
(51) Dunn, T. M.; Francis, A. H. J. Mol. Spectrosc. 1968, 25, 86.
(52) Spoliti, M.; Thirtle, J. H.; Dunn, T. M. J. Mol. Spectrosc. 1974, 52,
146.
(53) Bondybey, V. E. Chem. Phys. 1974, 18, 293-299.
(54) Lasko, C. L.; Miller, R. M.; Tinti, D. S. Chem. Phys. Lett. 1986, 130,
359-364.
(55) Fiedler, A., Kretzschmar, I.; Schro¨der, D.; Schwarz, H. J. Am. Chem.
Soc. 1996, 118, 9941-9952.

2116 Inorganic Chemistry, Vol. 38, No. 9, 1999
Limberg and Ko¨ppe
characterization of complexes OdCrCl2‚‚‚OdCRR′, whose
coexistence with other aggregated intermediates in the Etard
complex generated in olefin oxidations with CrO2Cl2 could
explain the frequent observation of carbonyl compounds among
the products after hydrolysis. It could be excluded that they
were generated in the matrix by isomerization of epoxide com-
plexes and on permethylation of the olefinic bond (i.e., employ-
ment of TME) their formation route was completely suppressed.
Moreover, first representatives for OCrCl2‚‚‚epoxide complexes,
often proposed as intermediates in the literature,1-11,13,15,43 have
been isolated and characterized. Both types of complexes should
be considered as respectable intermediates also under thermal
conditions. The stereo- and regiochemistry observed in the
common products of the thermal and photochemical processes
are in agreement and other characteristic principles are com-
parable, too, even if the initializations of the reactions are
different.
therefore be concluded in accordance with previous reports1-6,12
that carbonyl compounds can represent typical products of olefin
oxidations by CrO2Cl2 even without special workup procedures.
They form in concurrence to epoxides and can analogously be
found coordinated to Cr centers in the Etard complex after the
reaction. We have recently shown that complexation of epoxides
by CrO2Cl2 itself at -50 °C can induce a rearrangement to
chromium chloroalkoxides,56 yielding chlorohydrins on hy-
drolysis. From the spectroscopic investigation of this system at
low temperatures, therefore, a picture emerges which can explain
the variety of products resulting from such oxidations.
Acknowledgment. C.L. thanks the Deutsche Forschungs-
gemeinschaft as well as the Fonds der Chemischen Industrie
for scholarships as well as equipment. Further, we thank Prof.
H. Schno¨ckel for provision of matrix equipment and useful
discussions as well as Prof. G. Huttner for his generous support.
We are also grateful to Prof. T. Ziegler for providing us with a
preprint.
Conclusions
Cryogenic matrix studies concerning the olefin/CrO2Cl2
system have allowed for the first time the isolation and
(56) Limberg, C.; Cunskis, S.; Frick, A. Chem. Commun. 1998, 225-6.
IC981131S