A novel facet of carbonyliron-diene photochemistry: The η4-s-trans isomer of the classical Fe(CO)3(η4-s-cis-1,3-butadiene) discovered by time-resolved IR spectroscopy and theoretically examined by density functional methods

Article abstract of DOI:10.1021/om020785c

The photolysis of Fe(CO)₃(η⁴-s-cis-1,3-butadiene) (1) and Fe(CO)₄(η²-1,3-butadiene) (2), formerly studied in low-temperature matrixes, is reexamined in cyclohexane solution at ambient temperature using time-resolved IR spectroscopy in the v(CO) region. Flash photolysis of 2 (λ_exc = 308 nm) generates Fe(CO)₃(η⁴-s-trans-1,3-butadiene) (5) as a transient product, which then rearranges to form the classical η⁴-s-cis-1,3-butadiene complex 1. Species 5, previously addressed as the coordinately unsaturated Fe(CO)₃(η²-1,3-butadiene) (3), is also photogenerated from 1, in this case along with the very short-lived CO loss fragment Fe(CO)₂(η⁴-1,3-butadiene) (τ < 4 μs under CO atmosphere). It decays by temperature-dependent first-order kinetics (τ = 13 ms at 25°C; ΔH^? = 17.3 kcal·mol^-1) with nearly complete recovery of 1. According to density functional calculations at the BP86 level of theory, 5 resides in a distinct energy minimum, 20.3 kcal·mol^-1 above 1 and separated from it by a barrier of 15.0 kcal·mol^-1. Its computed structure involves a diene dihedral angle of 129°. Species 3 (with a diene dihedral angle of -150.1°), by contrast, is predicted to exist in a rather flat minimum, which makes it too short-lived for detection with our instrumentation. Flash photolysis of Fe(CO)₅ generates the very short-lived (<1 μs) doubly unsaturated Fe(CO)₃(solv) species in addition to the familiar Fe(CO)₄(solv) fragment (τ = 10-15 μs), Fe₂(CO)₉ being the ultimate product in the absence of potential trapping agents other than CO. Deliberate contamination of the system with water gives rise to the formation of Fe(CO)₄(H₂O) as a longer lived transient (ca. 1 ms). In the presence of 1,3-butadiene, both 2 and 5 appear almost instantaneously. The latter decays, again in the millisecond time range, with formation of 1, thus providing clear evidence of a single-photon route from Fe(CO)₅ to 1 in addition to the established two-photon sequence via the monosubstituted complex 2.

Full text of DOI:10.1021/om020785c

1696
Organometallics 2003, 22, 1696-1711
A Novel F a cet of Ca r bon ylir on -Dien e P h otoch em istr y:
Th e η4-s-tr a n s Isom er of th e Cla ssica l
F e(CO)3(η4-s-cis-1,3-bu ta d ien e) Discover ed by
Tim e-Resolved IR Sp ectr oscop y a n d Th eor etica lly
Exa m in ed by Den sity F u n ction a l Meth od s
Vinzenz Bachler,* Friedrich-Wilhelm Grevels,* Klaus Kerpen,
Gottfried Olbrich, and Kurt Schaffner
Max-Planck-Institut fu¨r Strahlenchemie, Stiftstrasse 34-36, Postfach 10 13 65,
D-45413 Mu¨lheim an der Ruhr, Germany
Received September 18, 2002
The photolysis of Fe(CO)3(η4-s-cis-1,3-butadiene) (1) and Fe(CO)4(η2-1,3-butadiene) (2),
formerly studied in low-temperature matrixes, is reexamined in cyclohexane solution at
ambient temperature using time-resolved IR spectroscopy in the ν(CO) region. Flash
photolysis of 2 (λexc ) 308 nm) generates Fe(CO)3(η4-s-trans-1,3-butadiene) (5) as a transient
product, which then rearranges to form the classical η4-s-cis-1,3-butadiene complex 1. Species
5, previously addressed as the coordinately unsaturated Fe(CO)3(η2-1,3-butadiene) (3), is
also photogenerated from 1, in this case along with the very short-lived CO loss fragment
Fe(CO)2(η4-1,3-butadiene) (τ < 4 µs under CO atmosphere). It decays by temperature-
dependent first-order kinetics (τ ) 13 ms at 25 °C; ∆Hq ) 17.3 kcal‚mol-1) with nearly
complete recovery of 1. According to density functional calculations at the BP86 level of
theory, 5 resides in a distinct energy minimum, 20.3 kcal‚mol-1 above 1 and separated from
it by a barrier of 15.0 kcal‚mol-1. Its computed structure involves a diene dihedral angle of
129°. Species 3 (with a diene dihedral angle of -150.1°), by contrast, is predicted to exist in
a rather flat minimum, which makes it too short-lived for detection with our instrumentation.
Flash photolysis of Fe(CO)5 generates the very short-lived (<1 µs) doubly unsaturated
Fe(CO)3(solv) species in addition to the familiar Fe(CO)4(solv) fragment (τ ) 10-15 µs),
Fe2(CO)9 being the ultimate product in the absence of potential trapping agents other than
CO. Deliberate contamination of the system with water gives rise to the formation of
Fe(CO)4(H2O) as a longer lived transient (ca. 1 ms). In the presence of 1,3-butadiene, both
2 and 5 appear almost instantaneously. The latter decays, again in the millisecond time
range, with formation of 1, thus providing clear evidence of a single-photon route from
Fe(CO)5 to 1 in addition to the established two-photon sequence via the monosubstituted
complex 2.
In tr od u ction
photochemical steps, Scheme 1, involving the interme-
diate formation of the monosubstituted Fe(CO)4(η2-1,3-
bd) (2), which is independently accessible from Fe2(CO)9
and 1,3-butadiene.6
Tricarbonyl(η4-diene)iron complexes are accessible by
various routes1 and have found a wide range of applica-
tions in metal-assisted organic syntheses.2 The parent
member of this class of compounds, Fe(CO)3(η4-1,3-bd)
(1, bd ) butadiene), was first prepared by heating
Fe(CO)5 and 1,3-butadiene in a pressure vessel.3,4 An
alternative photochemical procedure, explored and elabo-
rated in this laboratory, takes advantage of the high
efficiency of photolytic CO substitution under thermally
mild conditions and affords high yields of 1 in hydro-
carbon solution at ambient temperature.5 The overall
process was formulated as a sequence of two consecutive
The thermal stability of 1 and the existence of the
analogous 1,3-cyclohexadiene complex led to the pro-
posal that the 1,3-diene ligand is η4-coordinated in its
s-cis conformation.7 This has been verified by a single-
crystal X-ray structure analysis of 1,8 which, moreover,
revealed a square-pyramidal coordination geometry
with the CdC units of the diene in two basal positions.
These characteristics were found as recurrent features
in the structures of many other Fe(CO)3(η4-1,3-diene)
compounds and in the related Fe(CO)(η4-1,3-diene)2
complexes as well.9,10
* To whom correspondence should be addressed. E-mail: grevels@
mpi-muelheim.mpg.de.
(1) Kno¨lker, H.-J . Chem. Rev. 2000, 100, 2941-2961.
(2) Gre´e, R.; Lellouche, J . P. Adv. Metal-Org. Chem. 1995, 4, 129-
273.
(3) Reihlen, H.; Gruhl, A.; v. Hessling, G.; Pfrengle, O. Liebigs Ann.
Chem. 1930, 482, 161-182.
(4) King, R. B. Organometallic Syntheses, Vol. 1, Transition-Metal
Compounds; Academic: New York, 1965; pp 128-129.
(5) Koerner von Gustorf, E.; Pfajfer, Z.; Grevels, F.-W. Z. Naturfor-
sch. B 1971, 26, 66-67.
(6) Murdoch, H. D.; Weiss, E. Helv. Chim. Acta 1962, 45, 1156-
1161.
(7) Hallam, B. F.; Pauson, P. L. J . Chem. Soc. 1958, 642-645.
(8) Mills, O. S.; Robinson, G. Acta Crystallogr. 1963, 16, 758-761.
10.1021/om020785c CCC: $25.00 © 2003 American Chemical Society
Publication on Web 03/15/2003

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1697
Sch em e 1
Resu lts a n d Discu ssion
F la sh P h ot olysis of F e(CO)3(η4-1,3-b u t a d ien e)
(1). The electronic absorption spectrum of 1 exhibits a
maximum at 282 nm (ꢀ ) 2550 L‚mol-1‚cm-1) as the
only distinct feature on a steep declivity from the UV
to longer wavelengths. It ends in a long tail absorption
which extends into the visible region and is responsible
for the yellow color of the complex. Concrete information
on the underlying transitions is lacking, apart from the
observation that the efficiency of both CO and diene
photosubstitution decreases significantly on going from
shorter to longer wavelengths. At 313 nm, for example,
ΦCO ) 0.16 and Φdiene ) 0.015 were measured with
trimethyl phosphite as the entering ligand.15 Thus, flash
photolysis of 1 at λ ) 308 nm, using the XeCl excimer
laser routinely available in our experimental setup,
seems appropriate for the present time-resolved IR
spectroscopic study. Most clear results were obtained
under CO atmosphere, where the system proved largely
reversible.19
High yields of the latter compounds are accessible
from photogenerated Fe(CO)3(η4-1,3-diene) by further
CO photosubstitution occurring upon extended irradia-
tion in the presence of excess diene.5,11 Other photoin-
duced reactions of Fe(CO)3(η4-1,3-diene) complexes in-
clude CO/13CO exchange,12 metal-assisted diene-olefin
coupling,13,14 and CO substitution by phosphine ligands
yielding Fe(CO)2(η4-1,3-diene)(L) derivatives,12,15 whereas
replacement of the diene plays only a minor role.15
Mechanistic aspects have been addressed in low-
temperature matrix studies using conventional IR spec-
troscopy in the CO stretching vibrational region.16,17
Photolysis of Fe(CO)4(η2-1,3-bd) (2) initially yields a
tricarbonyliron intermediate, at that time assigned as
the coordinately unsaturated species Fe(CO)3(η2-1,3-bd)
(3), which upon continued irradiation is converted into
Fe(CO)3(η4-1,3-bd) (1).16 The same tricarbonyliron spe-
cies is also generated by photolysis of 1,16,17 in this
case along with the dicarbonyl fragment Fe(CO)2(η4-
1,3-bd) (4), thus indicating that two primary photopro-
cesses of Fe(CO)3(η4-1,3-diene) complexes may be opera-
tive, detachment of CO and dechelation of the η4-diene
ligand.
Transient IR difference spectra in the ν(CO) region
have been recorded in two time ranges, 100 µs (Figure
1) and 40 ms (Figure 2). The spectrum displayed in
Figure 1A represents the situation at a time (1.5 µs) at
which our instrumentation shows full response. In
addition to the depletion pattern of the starting material
1 (2056, 1988, and 1980 cm-1; cf. Table 1 for reference
data), five product absorptions have emerged, which are
readily grouped into two sets on the basis of their
distinctly different decay kinetics, 2024/1962/1944 and
2002/1938 cm-1
.
The latter pair of bands is in close agreement with
the data previously reported for the CO loss product of
1 in low-temperature matrixes (2009 and 1943 cm-1 in
Ar; 1999 and 1936 cm-1 in Xe),16 such that the assign-
ment as the dicarbonyl fragment Fe(CO)2(η4-1,3-bd)-
(solv) (4) is straightforward. It decays very rapidly
within a few microseconds with concomitantly occurring
partial recovery of the depletion bands associated with
1, as exemplarily illustrated in Figure 1B by the signals
recorded at 2002 and 2056 cm-1. Obviously, fragment
4 immediately takes up carbon monoxide from the CO-
saturated solution to re-form a corresponding amount
of the parent 1, Scheme 2.18,20
The difference spectrum left behind after 4 µs (Figure
1C) comprises the aforementioned three-band pattern
(marked as shaded areas) along with a weaker, new
absorption which has emerged at 1920 cm-1 during the
fast decay of 4. It vanishes within a period of 25-30 µs,
during which further recovery of depleted 1 takes place,
as documented by the ∆A signals recorded at 1920 and
2056 cm-1, respectively (Figure 1B). Apparently, a
certain amount of 4 is temporarily trapped by a weakly
binding agent present as an impurity in the solution,
water being a likely candidate by analogy with similar
observations made in our early investigations into the
The laser flash photolysis experiments, reported in
the present paper, aim at elucidating the kinetics and
reactivity of the above photoproducts of 1 and 2 in
hydrocarbon solution at ambient temperature by means
of time-resolved IR spectroscopy in the micro- and
millisecond time domain. As a most interesting outcome,
our experimental findings demand a reassignment of
the tricarbonyliron species “Fe(CO)3(η2-1,3-bd)” (3) as
a novel isomer of 1, Fe(CO)3(η4-s-trans-1,3-bd) (5), which
finds support from the results of a density functional
theory (DFT) study. In this context we also examine the
photofragmentation of Fe(CO)5 in hydrocarbon solution
in the presence and in the absence of 1,3-butadiene. As
a noteworthy result of this part of the study, the doubly
unsaturated species Fe(CO)3(solv) is recognized as a
single-photon product of Fe(CO)5 in addition to the
expected Fe(CO)4(solv).18
(9) Kru¨ger, C.; Tsay, Y.-H. Angew. Chem. 1971, 83, 250-251; Angew.
Chem., Int. Ed. Engl. 1971, 10, 261.
(10) Whiting, D. A. Cryst. Struct. Commun. 1972, 1, 379-381.
(11) Koerner von Gustorf, E.; Buchkremer, J .; Pfajfer, Z.; Grevels,
F.-W. Angew. Chem. 1971, 83, 249-250; Angew. Chem., Int. Ed. Engl.
1971, 10, 260-261.
(12) Warren, J . D.; Clark, R. J . Inorg. Chem. 1970, 9, 373-379.
(13) Kerber, R. C.; Koerner von Gustorf, E. A. J . Organomet. Chem.
1976, 110, 345-357.
(14) Grevels, F.-W.; Feldhoff, U.; Leitich, J .; Kru¨ger, C. J . Orga-
nomet. Chem. 1976, 118, 79-92.
(15) J aenicke, O.; Kerber, R. C.; Kirsch, P.; Koerner von Gustorf,
E. A.; Rumin, R. J . Organomet. Chem. 1980, 187, 361-373.
(16) Ellerhorst, G.; Gerhartz, W.; Grevels, F.-W. Inorg. Chem. 1980,
19, 67-71.
(17) Astley, S. T.; Churton, M. P. V.; Hitam, R. B.; Rest, A. J . J .
Chem. Soc., Dalton Trans. 1990, 3243-3253.
(18) In the schemes, the solvent molecule(s) occupying the vacant
coordination site(s) of the fragments Fe(CO)4, Fe(CO)3, Fe(CO)3(η2-1,3-
bd) (3), and Fe(CO)2(η4-1,3-bd) (4) are omitted for the sake of concise-
ness. As regards the doubly unsaturated Fe(CO)3, this species is
formulated as Fe(CO)3(solv) in the running text, since concrete
information on the number of coordinated solvent molecules is lacking.
(19) Under argon or argon-diluted atmosphere (Ar/CO 9:1), the
reversibility is diminished due to the formation of ill-defined, long-
lived (>1 s) side products.
(20) The dotted arrows refer to processes of minor importance.

1698 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
F igu r e 2. Flash photolysis (λexc ) 308 nm) of Fe(CO)3(η4-
1,3-bd) (1, 1.5 mM in cyclohexane solution) under CO
atmosphere at ambient temperature with time-resolved IR
detection in the 40 ms time range. (A) Transient ν(CO)
difference spectrum recorded after 0.2 ms showing the
depletion bands of 1 together with the absorptions of
Fe(CO)3(η4-s-trans-1,3-bd) (5, shaded areas). (B) Decay of
5 (monitored at 2024 cm-1) and recovery of 1 (monitored
at 2056 cm-1). (C) ν(CO) difference spectrum recorded after
35 ms showing residual depletion of 1 together with a weak
absorption centered at 2004 cm-1 (hatched area), assigned
(see text) as Fe(CO)4(η2-1,3-bd) (2).
F igu r e 1. Flash photolysis (λexc ) 308 nm) of Fe(CO)3(η4-
1,3-bd) (1, 1.5 mM in cyclohexane solution) under CO
atmosphere at ambient temperature with time-resolved IR
detection in the 100 µs time range. (A) Transient ν(CO)
difference spectrum recorded after 1.5 µs showing the
depletion bands of 1 (2056, 1988, 1980 cm-1) together with
the absorptions of the photogenerated fragment Fe(CO)2-
(η4-1,3-bd)(solv) [4 (B); 2002, 1938 cm-1] and the species 5
(shaded areas; 2024, 1962, 1944 cm-1), assigned (see text)
as Fe(CO)3(η4-s-trans-1,3-bd). (B) Selected kinetic traces.
(C) Difference spectrum recorded after 4 µs showing the
depletion bands of 1 together with the absorptions of 5
(shaded areas) and Fe(CO)2(η4-1,3-bd)(H2O) [ 6 (2); 1920
cm-1].
canceled due to overlap with the depleted absorption of
1 in this region.
The following discussion focuses on the assignment
of the three prominent bands at 2024, 1962, and 1944
cm-1, marked as shaded areas in Figure 1A and 1C. The
same pattern was previously observed upon photolysis
of 1 in low-temperature matrixes (2029, 1970, and 1953
cm-1 in Ar; 2023, 1963, and 1944 cm-1 in Xe)16 and, at
that time, was attributed to the coordinately unsatur-
ated tricarbonyliron species Fe(CO)3(η2-1,3-bd) (3). How-
ever, this assignment does not seem consistent with its
behavior in CO-saturated solution. By analogy with the
very fast reaction of the dicarbonyl fragment 4 with CO,
one would expect that the vacant coordination site of 3
is likewise rapidly occupied by carbon monoxide, which,
in this case, would lead to the tetracarbonyl complex
Fe(CO)4(η2-1,3-bd) (2). In striking contrast to this
expectation, the three-band species survives virtually
untouched from ca. 2 µs to 90 µs (as illustrated in Figure
1B by the ∆A trace recorded at 2024 cm-1) and even
beyond (0.2 ms, Figure 2A). It then decays gradually
with concomitant and nearly complete recovery of 1
(Figure 2B), while the take-up of CO with formation of
2 is, at best, an insignificant side reaction (which for
the sake of completeness is nevertheless included in
photolysis of group 6 metal hexacarbonyls.21,22 In sup-
port of this notion, the intensity of the 1920 cm-1 band
is significantly increased upon deliberate contamination
of the solution with water, which, moreover, slows down
the decay to such an extent that it takes 60-70 µs until
the absorbance returns to the baseline. Hence it seems
that, despite all efforts made in the purification of the
cyclohexane solvent, traces of moisture are still avail-
able for trapping the photogenerated fragment 4 in the
form of an equilibrium with the species Fe(CO)2(η4-1,3-
bd)(H2O) (6), as suggested in Scheme 2. A second CO
stretching mode of 6, expected around 1980 cm-1, is
(21) Hermann, H.; Grevels, F.-W.; Henne, A.; Schaffner, K. J . Phys.
Chem. 1982, 86, 5151-5154.
(22) Church, S. P.; Grevels, F.-W.; Hermann, H.; Schaffner, K. Inorg.
Chem. 1985, 24, 418-422.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1699
Ta ble 1. F r equ en cies a n d In ten sities of th e CO Str etch in g Vibr a tion s of F e(CO)3(η4-s-cis-1,3-bd ) (1),
F e(CO)4(η2-1,3-bd ) (2), F e(CO)3(η2-1,3-bd ) (3), F e(CO)3(η4-s-tr a n s-1,3-bd ) (5), F e(CO)5, F e(CO)4(solv), a n d
F e(CO)3(solv)i
frequency [cm-1
]
integrated intensity [km‚mol-1]a)
exptl
calcd
exptl
calcd
ꢀ [L‚mol-1‚cm-1
]
1b
(Cs)
2054.8 (A′)
1988.5 (A′)
1978.6 (A′′)
2044.0
1992.8
1985.4
274.1 (0.248)
462.4 (0.419)
367.3 (0.333)
475.2 (0.248)
816.4 (0.426)
625.3 (0.326)
5460
8380
6510
2b
(C2v
2082.5 (A1)
2011.5 (A1)
2003.6 (B1)
1983.2 (B2)
102.7 (0.075)
189.1 (0.138)
595.6 (0.433)
487.1 (0.354)
2210
2550
11660
6840
c
)
3d
2036.6
1987.0
1980.7
561.4 (0.278)
862.8 (0.427)
596.3 (0.295)
5e
(C1)
2024
1962
1944
2023.6
1975.7
1959.5
(0.293)f/(0.217)g
(0.210)f/(0.262)g
(0.496)f/(0.522)g
555.8 (0.280)
423.2 (0.213)
1009.0 (0.508)
b
Fe(CO)5
(D3h
2022.0 (A2′′)
1999.0 (E′)
9920
13220
)
Fe(CO)4‚Sh
(C2v
2084 (A1)
1986 (B1)
1970 (A1)
)
1950 (B2)
1926 (E)
Fe(CO)3‚Sh
(C3v
)
a
Relative integrated intensities are given in parentheses. b Experimental data from FTIR spectra. c Fe(CO)4 skeleton assumed to possess
local C2v symmetry. Data calculated for singlet 3; data for triplet 3 are listed in ref 52. e Experimental data from transient IR spectra
d
recorded after flash photolysis of 1 (Figure 2A) and 2 (Figure 11A), respectively. f Data from Figure 2A. Data from Figure 11A.
g
h
Experimental data from transient IR spectra recorded after flash photolysis of Fe(CO)5 (Figure 9A); S ) solvent cyclohexane.
Experimental and calculated data are taken from measurements in cyclohexane solution and from BP86 DFT calculations, respectively.
i
transition metals in higher oxidation states,28-34 but is
Sch em e 2
a novelty in diene-substituted zerovalent metal carbo-
nyls. The ν(CO) pattern of 5 (Figures 1C and 2A) bears
some resemblance to that of trigonal-bipyramidal fac-
Fe(CO)3(η2-ethene)(PR3) complexes.35 This leads us to
suggest a similar coordination geometry for 5 with one
CdC unit of the diene residing in an axial and the other
one in an equatorial coordination site.
The ∆A traces recorded in the 40 ms time range show
satisfactory fits to first-order kinetics, regardless of
(24) Erker, G.; Wicher, J .; Engel, K.; Kru¨ger, C. Chem. Ber. 1982,
115, 3300-3310.
(25) Kai, Y.; Kanehisa, N.; Miki, K.; Kasai, N.; Mashima, K.;
Nagasuna, K.; Yasuda, H.; Nakamura, A. J . Chem. Soc., Chem.
Commun. 1982, 191-192.
Scheme 2). This is evident from the difference spectrum
recorded after 35 ms (Figure 2C), which exhibits no
more than minor remnants of the depletion pattern of
1 along with a rather weak product absorption around
2004 cm-1, attributable to the strongest of the four
ν(CO) bands of 2 (cf. Table 1 for reference data).
Hence we conclude that we are actually dealing with
a coordinately saturated isomer of 1, a structure con-
taining η4-coordinated 1,3-butadiene in its s-trans con-
formation (5) being an appealing proposal. This coor-
dination mode of a 1,3-diene has some well-documented
precedents among early transition metal (η4-diene)-
metallocenes23-27 and other 1,3-diene complexes of late
(26) Erker, G.; Kru¨ger, C.; Mu¨ller, G. Adv. Organomet. Chem. 1985,
24, 1-39.
(27) Yasuda, H.; Tatsumi, K.; Nakamura, A. Acc. Chem. Res. 1985,
18, 120-126.
(28) Hunter, A. D.; Legzdins, P.; Nurse, C. R.; Einstein, F. W. B.;
Willis, A. C. J . Am. Chem. Soc. 1985, 107, 1791-1792.
(29) Hunter, A. D.; Legzdins, P.; Einstein, F. W. B.; Willis, A. C.;
Bursten, B. E.; Gatter, M. G. J . Am. Chem. Soc. 1986, 108, 3843-
3844.
(30) Christensen, N. J .; Hunter, A. D.; Legzdins, P. Organometallics
1989, 8, 930-940.
(31) Benyunes, S. A.; Green, M.; Grimshire, M. J . Organometallics
1989, 8, 2268-2270.
(32) Benyunes, S. A.; Day, J . P.; Green, M.; Al-Saadoon, A. W.;
Waring, T. L. Angew. Chem. 1990, 102, 1505-1507; Angew. Chem.,
Int. Ed. Engl. 1990, 29, 1416-1417.
(33) Ernst, R. D.; Melendez, E.; Stahl, L.; Ziegler, M. L. Organome-
tallics 1991, 10, 3635-3642.
(23) Erker, G.; Wicher, J .; Engel, K.; Rosenfeldt, F.; Dietrich, W.;
Kru¨ger, C. J . Am. Chem. Soc. 1980, 102, 6344-6346.
(34) Benyunes, S. A.; Binelli, A.; Green, M.; Grimshire, M. J . J .
Chem. Soc., Dalton Trans. 1991, 895-904.

1700 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
whether the decay of 5 or the recovery of depleted 1 is
monitored, as exemplarily illustrated in Figure 2B.
Variable-temperature measurements, carried out in a
thermostated cuvette, yielded kobs ) 76.4 ((4.6) s-1 at
24.6 °C (with 0.5 mM 1), 76.9 ((3.4) s-1 at 25.5 °C (with
1.5 mM 1), 129 s-1 ((4) at 30.3 °C (with 1.5 mM 1),
and 556 ((24) s-1 at 46.0 °C (with 1.5 mM 1). These
data are average values from curve fittings at 18 or
more different monitoring wavenumbers in each set of
measurements. At any temperature, nearly complete
recovery of 1 is observed; that is, the side reaction of 5
with CO with formation of 2 can be considered as
negligible. The Arrhenius plot of the above rate con-
stants is a straight line yielding Ea ) 17.9 kcal‚mol-1
(A ) 9.6 × 1014 s-1; ∆Hq ) Ea - RT ) 17.3 kcal‚mol-1).36
This activation barrier is of the same order of magnitude
as the values reported for the s-trans f s-cis isomer-
ization of Cp2Zr(η4-1,3-diene) complexes (∆Gq ) 18.2-
22.7 kcal‚mol-1 at t ) -25 to 10.5 °C),23,24,26 thus
providing further support for the assignment of 5 as
Fe(CO)3(η4-s-trans-1,3-bd).
Den sity F u n ction a l Stu d y of F e(CO)3(1,3-bu ta -
d ien e) Sp ecies w it h Va r yin g Dien e Dih ed r a l
An gles. In search of complementary information on the
transient Fe(CO)3(η4-s-trans-1,3-butadiene) (5) and re-
lated species we examined this system by means of
density functional theory (DFT)37-39 methods, which
have become useful tools for the study of molecular
structures, transition states, and force fields of a wide
variety of complexes including, for example, metal
carbonyls.40,41
We first revisit Fe(CO)3(η4-s-cis-1,3-bd) (1), which was
recently studied by Bu¨hl and Thiel41 with particular
emphasis on the CO stretching frequencies in the
context of rotational CO site exchange. Our results
compare well with that study. The computed geometry
of 1 (Figure 3), obtained without imposing symmetry
restrictions, shows no significant deviations from the
structure found in the crystal.8 Satisfactory agreement
between calculated and experimental data is also found
for the CO stretching vibrational pattern of 1 (Table 1)
with respect to both the frequencies and the relative
intensities, while the absolute intensities show a sys-
tematic deviation by a factor of 1.7.
information on isomeric species with different confor-
mations of the 1,3-butadiene ligand. Before going into
detail, we wish to premise that our study not only
corroborates the proposed structure of the observed
transient Fe(CO)3(η4-s-trans-1,3-bd) (5) but also hints
at the existence of the coordinately unsaturated species
Fe(CO)3(η2-1,3-bd) (3) in another minimum, albeit rather
flat, on the potential energy surface of the Fe(CO)3(1,3-
bd) system.
In a series of partial geometry optimizations, the
C10-C9-C8-C11 dihedral angle φd of the diene is
varied, in steps of 20°, from 100° to (180° and further
to -100°,42 with the C11dC8 double bond remaining
attached to the metal. In each of these calculations,43
the angle φd is kept fixed while all other coordinates
are optimization parameters. The plot of the calculated
energies versus φd (Figure 4) shows a well-pronounced
minimum in the range 120-140°, while a nearly flat
region is seen around -160°.
A full geometry optimization, in which the restriction
of φd ) 140° is released, leads to a structure that in
essence shows the features proposed for the transient
species Fe(CO)3(η4-s-trans-1,3-bd) (5). It resides in a
minimum located 20.3 kcal‚mol-1 above the parent
Fe(CO)3(η4-s-cis-1,3-bd) (1) (Figure 5). This energy dif-
ference is smaller than previously predicted (1.3 eV )
30 kcal‚mol-1 27,44
on the basis of extended Hu¨ckel
)
calculations, which, however, used merely a rough
estimate of the geometry of 5.
The computed structure of 5 (Figure 3) contains a fac-
Fe(CO)3 skeleton as part of a slightly distorted trigonal
bipyramid with an angle of 123.0° between the two
equatorial CO groups and with angles of 90.8° and 89.7°
between the CO group in the axial position and the
equatorial carbonyls. The remaining two coordination
sites of 5, one axial and one equatorial, are occupied by
the CdC double bonds of the diene. These two vinyl
units are significantly twisted away from coplanarity,
as indicated by the computed C10-C9-C8-C11 dihe-
dral angle of 129°. Note that similar characteristics of
1,3-diene ligands were experimentally found for the
compounds Cp2Zr(η4-s-trans-1,4-diphenyl-1,3-butadi-
ene),25,27 CpMo(NO)(η4-s-trans-2,5-dimethyl-2,4-hexa-
diene),28 and Ru(η4-s-trans-2,4-dimethyl-1,3-pentadiene)-
(acac),33 which show torsion angles about the central
C-C bond of ca. 126°, 124.8°, and 122.6°, respectively.45
Regarding the carbon-carbon bond lengths, it is inter-
esting to note that the alternation, present in free
s-trans-1,3-butadiene (1.337 Å, CdC; 1.483 Å, C-C),46
is still preserved in Fe(CO)3(η4-s-trans-1,3-bd) (5), al-
In light of this encouraging test, we are confident that
the applied level of theory is sufficient to yield reliable
(35) Lindner, E.; Schauss, E.; Hiller, W.; Fawzi, R. Chem. Ber. 1985,
118, 3915-3931.
(36) With a barrier of this height, it should be possible to slow the
thermal 5 f 1 rearrangement at low temperatures to such an extent
that NMR spectroscopy can be employed for complementary charac-
terization (we thank one reviewer for drawing our attention to this
issue), by analogy with the study of a thermally labile carbonyliron
complex, HFe(CO)3(η3-C3H5), photogenerated by continuous irradiation
of Fe(CO)4(η2-C3H6) (Barnhart, T. M.; De Felippis, J .; McMahon, R. J .
Angew. Chem. 1993, 105, 1134-1136; Angew. Chem., Int. Ed. Engl.
1993, 32, 1073-1074). Complex 5, generated by continuous irradiation
of 1 in 2-methylpentane solution at -120 °C, proved indeed sufficiently
long-lived for detection by FTIR spectroscopy. However, the conversion
of 1 does not exceed 1-2%, such that the concentration of 5 remains
at a stationary level which is much too low for recording its NMR
spectrum. It seems that the system is photoreversible, by analogy with
previous observations made in low-temperature matrixes.16
(37) Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density
Functional Theory; Wiley-VCH: Weinheim, 1999.
(42) Recall that the two vinyl units of η4-s-cis-coordinated 1,3-
butadiene are bound to a metal with opposite enantiotopic faces, while
the same enantiotopic faces are involved in the case of η4-s-trans-
coordinated 1,3-butadiene. The numbering of the diene carbon atoms
chosen in Figure 3 for Fe(CO)3(η4-s-cis-1,3-bd) (1) involves the R
configuration for C9 and the S configuration for C8. The vinyl unit
containing C8 is kept attached to the metal when the C10-C9-C8-
C11 dihedral angle is varied in the series of partial geometry optimiza-
tions performed in search of the structure of the η4-s-trans isomer 5.
In consequence of this choice, the configuration of C9 is inverted from
R to S on going from 1 to 5 and, moreover, the diene dihedral angle φd
has a positive value in 5.
(43) A spin multiplicity of 1 was chosen throughout, unless otherwise
noted.
(38) Labanowski, J . K., Andzelm, J . W., Eds. Density Functional
Methods in Chemistry; Springer: New York, 1991.
(39) Ziegler, T. Chem. Rev. 1991, 91, 651-667.
(40) J onas, V.; Thiel, W. J . Chem. Phys. 1995, 102, 8474-8484.
(41) Bu¨hl, M.; Thiel, W. Inorg. Chem. 1997, 36, 2922-2924.
(44) Tatsumi, K.; Yasuda, H.; Nakamura, A. Isr. J . Chem. 1983, 23,
145-150.
(45) The geometry of the diene ligand in the zirconocene derivative
Cp2Zr(η4-s-trans-1,3-butadiene)23,24 remained ambiguous due to unusu-
ally high thermal parameters of the central carbon atoms.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1701
F igu r e 3. BP86 optimized geometries42,43 of Fe(CO)3(η4-s-cis-1,3-bd) (1), the fragment Fe(CO)3(η2-s-trans-1,3-bd) (3), Fe-
(CO)3(η4-s-trans-1,3-bd) (5), and the transition states TS[5-1], TS[5-3], and TS[3-1] along with the respective energies
and selected bond distances and angles. For comparison, data from the X-ray structure analysis8 of 1 are given in
parentheses.
though much less pronounced (1.407/1.408 Å, CdC;
1.461 Å, C-C). Recall that for the η4-s-cis complex 1
nearly equal CdC and C-C distances are experimen-
tally found (1.46 ( 0.05 Å, CdC; 1.45 ( 0.06 Å, C-C)8
and computed (1.435 Å, CdC; 1.426 Å, C-C; see Figure
3). This is consistent with the notion27 of more effective
back-donation from the metal into the π3* acceptor
(46) Sutton, E. L., Ed. Tables of Interatomic Distances and Config-
uration in Molecules and Ions. Supplement 1956-1959; The Chemical
Society: London, 1965; Sect. B, M 109s.

1702 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
As regards the CO stretching vibrations, the observed
transient product spectrum (Figure 2A, Table 1) is much
better reproduced by the calculated pattern of Fe(CO)3-
(η4-s-trans-1,3-bd) (5) (Table 1) than by the calculated
pattern of either singlet or triplet Fe(CO)3(η2-1,3-bd)
(3).52 This is particularly true for the relative intensities,
as convincingly demonstrated in Figure 6, where all four
spectra are displayed in schematic form. This further
corroborates the assignment of the transient species as
5 rather than 3.
Fragment 3 may nevertheless exist in an equilibrium
with 5, albeit in a concentration far too low for spec-
troscopic detection. We suppose that 3 is trapped by
carbon monoxide, Scheme 3,20 to yield the minor amount
of Fe(CO)4(η2-1,3-bd) (2) observed as the side product
of the thermal reconversion of photogenerated 5 into the
stable η4-s-cis-1,3-bd parent complex 1 in CO-saturated
solution (vide supra, cf. Scheme 2).
One easily realizes that the 5 f 1 rearrangement
involves rotation about the central C-C bond of the
diene, either clockwise or counterclockwise, and requires
the temporary detachment of one of the two vinyl units
of the diene from the metal followed by reattachment
with its opposite enantiotopic face.42 The clockwise
rotation proceeds via the species 3, while the counter-
clockwise rotation directly converts 5 into 1. For the
respective transition states, designated as TS[5-3],
TS[3-1], and TS[5-1] in Figure 5, one vibrational
mode with an imaginary frequency is found (data listed
in Figure 3) as requested for first-order saddle points
on the potential energy surface.
The computed structures of TS[5-1] and TS[3-1]
(Figure 3) show a nearly orthogonal orientation of the
two vinyl units of the diene ligand relative to each other.
The C10-C9-C8-C11 dihedral angle has a positive
sign for TS[5-1] (φd ) 95.7°), whereas the sign is
negative for TS[3-1] (-84.7°). It is the C11dC8 unit
of 5 that remains coordinated in either case and experi-
ences merely a moderate increase of its CdC distance.
The carbon-carbon distance of the detached C10dC9
unit is substantially shortened and assumes a value
that is almost identical with the CdC distance in free
1,3-butadiene46 and with that of the uncoordinated vinyl
group in 3 as well.
F igu r e 4. Energies of partially optimized geometries of
Fe(CO)3(1,3-bd) computed at the BP86 level of DFT with
various C10-C9-C8-C11 dihedral angles φd of the diene
ligand.42
orbital of the η4-s-cis-coordinated 1,3-butadiene in 1
compared with the η4-s-trans-coordinated 1,3-butadiene
in 5, which seems to be largely responsible for the
energy difference between the two compounds.
A further minimum is found in the nearly flat region
of the diagram shown in Figure 4. In this case, a full
geometry optimization is started with the partially
optimized structure obtained with φd ) -160°. The
computed structure (Figure 3) shows a dihedral angle
of φd ) -150.1°, and the Fe-C(diene) and the CdC
distances indicate that only one vinyl group of the 1,3-
butadiene ligand is bound to the Fe(CO)3 unit; that is,
we are dealing with the coordinately unsaturated spe-
cies Fe(CO)3(η2-1,3-bd) (3). The minimum is located 7.4
kcal‚mol-1 above Fe(CO)3(η4-s-trans-1,3-bd) (5) and 27.7
kcal‚mol-1 above Fe(CO)3(η4-s-cis-1,3-bd) (1) (Figure 5).
Fragments of type Fe(CO)3(η2-olefin)47,48 and Fe(CO)3-
(η2-diene)49 have been suggested to possess a triplet
ground state, by analogy with (naked) Fe(CO)4.50,51
Hence one might argue that the electronic ground state
of 3 is also a triplet rather than a singlet, as tacitly
assumed in the above treatment.43 To address this
point, a comparative calculation is performed for the
triplet case.52 The results place the triplet above the
singlet, although the energy gap is as small as 0.3
kcal‚mol-1 (0.5 kcal‚mol-1 with ZPE) or 2.2 kcal‚mol-1
(1.5 kcal‚mol-1 with ZPE), depending on whether the
calculation is performed at the unrestricted or restricted
BP86 level of DFT. In either case, the η2-coordinated
1,3-butadiene shows an almost coplanar arrangement
of the two vinyl units (φd ) 178.3° and 179.5°, respec-
tively).
These transition states are rather close in energy,
TS[5-1] and TS[3-1] being located 15.0 and 18.6 (7.4
+11.2) kcal‚mol-1, respectively, above 5 (Figure 5). Of
the two alternative pathways, the direct conversion via
TS[5-1] may be slightly favored over the two-step route
F igu r e 5. Energies [kcal‚mol-1] (with ZPE correction) and C10-C9-C8-C11 diene dihedral angles φd of Fe(CO)3(η4-s-
trans-1,3-bd) (5), the fragment Fe(CO)3(η2-s-trans-1,3-bd) (3), and the transition states TS[5-1], TS[5-3], and TS[3-1]
in relation to Fe(CO)3(η4-s-cis-1,3-bd) (1), as computed at the BP86 level of DFT.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1703
F igu r e 7. FTIR ν(CO) spectrum recorded after single-flash
irradiation (λexc ) 308 nm, 150 mJ incident flash energy)
of Fe(CO)5 (1.5 mM in cyclohexane containing 130 mM 1,3-
butadiene) in an IR cell (d ) 0.5 mm), illustrating the
formation of Fe(CO)3(η4-1,3-bd) (1) and Fe(CO)4(η2-1,3-bd)
(2) (for frequency data see Table 1) in 1:4 ratio. Note that
the absorptions associated with unphotolyzed Fe(CO)5
(78%) are removed by computer-assisted subtraction; small
artifacts arising from this treatment are marked by the
asterisk.
of its photochemical synthesis from Fe(CO)5 and 1,3-
butadiene, thought to involve the monosubstituted
Fe(CO)4(η2-1,3-bd) (2) as an intermediate product
(Scheme 1).5
Using FTIR spectroscopy for monitoring the progress
of this conversion, we note that complex 2 does not
accumulate under continuous irradiation (λexc ) 313
nm). It rather remains at a relatively low level of
concentration, thus indicating that the secondary, in-
tramolecular CO photosubstitution leading to 1 occurs
with high efficiency.53
F igu r e 6. (A) Schematic representation of the observed
transient ν(CO) product pattern attributed to Fe(CO)3(η4-
s-trans-1,3-bd) (5) (Table 1; cf. Figure 2A). (B) Calculated
ν(CO) spectrum of 5 (Table 1). (C) and (D): Calculated
ν(CO) spectra of Fe(CO)3(η2-1,3-bd) (3, singlet and triplet,
respectively; data from footnote 52).
Sch em e 3
The result of single-flash irradiation under similar
conditions (λexc ) 308 nm; 1.5 mM Fe(CO)5; 130 mM 1,3-
butadiene in cyclohexane solution under argon atmo-
sphere) is shown in Figure 7. The product spectrum is
dominated by the well-resolved four-band ν(CO) pattern
of the tetracarbonyl complex 2, which, however, is
accompanied by additional absorptions indicative of the
tricarbonyl complex 1 (frequencies listed in Table 1).
This means that, surprisingly enough, a single-photon
route from Fe(CO)5 to 1 is operative in addition to the
previously suggested (Scheme 1) two-step sequence
Fe(CO)59 9
hν8 2 hν8 1
proceeding via TS[5-3], 3, and TS[3-1]. In either case,
the computed barrier compares well with the experi-
mentally determined activation energy for the 5 f 1
rearrangement (Ea ) 17.9 kcal‚mol-1; ∆Hq ) 17.3
kcal‚mol-1; vide supra), thus further substantiating the
assignment of 5.
F la sh P h otolysis of F e(CO)5 in th e P r esen ce of
1,3-Bu ta d ien e. Having accomplished the characteriza-
tion of the transient photoproducts of Fe(CO)3(η4-s-cis-
1,3-bd) (1), we now deal with the mechanistic aspects
This additional route accounts for not less than 20% of
the product mixture.
(52) We thank one reviewer for drawing our attention to this issue.
The geometry optimization for the triplet was started with the
structure of singlet 3 shown in Figure 3. If performed at the un-
restricted BP86 level of DFT (the application of which seems justified
by an S2 expectation value of 2.03), it yielded an energy of -620.064492
au (-619.957052 au with ZPE correction). All frequencies are real. CO
stretching vibrational frequencies and intensities: 2015.5 cm-1 (523.1
km‚mol-1; 0.248 rel), 1968.9 (741.1; 0.352), and 1964.4 (842.9; 0.400).
Selected angles [deg] and bond distances [Å] (for the numbering scheme
see Figure 3): 178.3 (C10-C9-C8-C11), 100.3 (C6-Fe1-C2), 98.2
(C6-Fe1-C4), 97.0 (C2-Fe1-C4), 123.7 (C11-C8-C9), 124.1
(C10-C9-C8); 2.169 (Fe1-C11), 2.168 (Fe1-C8), 2.680 (Fe1-C9),
3.611 (Fe1-C10), 1.399 (C11-C8), 1.457 (C8-C9), 1.366 (C9-C10).
The alternative calculation at the restricted BP86 level of DFT yielded
an energy of -620.061484 au (-619.953931 au with ZPE correction).
All frequencies are real. CO stretching frequencies and intensities:
2014.9 cm-1 (524.8 km‚mol-1; 0.249 rel), 1969.0 (729.5; 0.346), and
1964.8 (854.9; 0.405). Selected angles [deg] and bond distances [Å]:
179.5 (C10-C9-C8-C11), 100.2 (C6-Fe1-C2), 98.2 (C6-Fe1-C4),
97.0 (C2-Fe1-C4), 123.8 (C11-C8-C9), 124.1 (C10-C9-C8); 2.170
(Fe1-C11), 2.174 (Fe1-C8), 2.719 (Fe1-C9), 3.655 (Fe1-C10), 1.398
(C11-C8), 1.459 (C8-C9), 1.363 (C9-C10).
(47) Barnhart, T. M.; Fenske, R. F.; McMahon, R. J . Inorg. Chem.
1992, 31, 2679-2681.
(48) Barnhart, T. M.; McMahon, R. J . J . Am. Chem. Soc. 1992, 114,
5434-5435.
(49) Gravelle, S. J .; van de Burgt, L. J .; Weitz, E. J . Phys. Chem.
1993, 97, 5272-5283.
(50) Poliakoff, M.; Weitz, E. Acc. Chem. Res. 1987, 20, 408-414, and
references therein.
(51) Poliakoff, M.; Turner, J . J . Angew. Chem. 2001, 113, 2889-
2892; Angew. Chem., Int. Ed. 2001, 40, 2809-2812, and references
therein.

1704 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
Sch em e 4
More detailed information on this route is gathered
from studies employing flash photolysis in combination
with time-resolved IR spectroscopy, which reveal that
the novel isomer of 1, Fe(CO)3(η4-s-trans-1,3-bd) (5), is
involved as a transient species, Scheme 4.20 The differ-
ence spectrum displayed in Figure 8A illustrates the
situation after 1.5 µs. It shows the depletion bands of
Fe(CO)5 along with a product ν(CO) pattern which is
dominated by the absorptions associated with the main
product, Fe(CO)4(η2-1,3-bd) (2). Note that the strongest
ν(CO) band of 2 (B1), expected at 2003.6 cm-1 (Table
1), appears slightly shifted at 2006 cm-1 and with lower
intensity than in Figure 7. We explain this apparent
misrepresentation by taking into account the lower
spectral resolution of the time-resolved IR instrumenta-
tion (7-8 vs 2 cm-1 of the FTIR spectrometer). This
gives rise to some foot-and-tail overlap with the depleted
F igu r e 8. Flash photolysis (λexc ) 308 nm) of Fe(CO)5 (1.5
mM, in cyclohexane solution containing 130 mM 1,3-
butadiene) under argon atmosphere at 25 °C. (A) Transient
ν(CO) difference spectrum recorded after 1.5 µs showing
the depletion bands of Fe(CO)5 (2020, 1998 cm-1) and the
absorptions of photogenerated Fe(CO)4(η2-1,3-bd) (2) (2084,
2006, 1984 cm-1) and Fe(CO)3(η4-s-trans-1,3-bd) (5) (1962,
1944 cm-1; shaded area) along with a weak feature at 2056
cm-1 attributed to Fe(CO)3(η4-s-cis-1,3-bd) (1). (B) Decay
of 5 monitored at the band positions listed in Table 1 and
marked in spectra A and D by a solid triangle (2). (C)
Growth of 1 monitored at the band positions listed in Table
1 and marked in spectra A and D by a solid circle (B). (D)
Spectrum recorded after 18 ms.
E′ band of Fe(CO)5 in close proximity (1999.0 cm-1
,
Table 1) and, hence, causes noticeable mutual cancel-
lation.
The two product absorptions at lower frequencies,
rendered prominent as shaded areas in Figure 8A, are
readily attributed to Fe(CO)3(η4-s-trans-1,3-bd) (5) by
comparison with the complete ν(CO) transient spectrum
of 5 observed upon flash photolysis of 1 (Figures 1C and
2A, Table 1). The third band of 5, expected at 2024 cm-1
,
is obscured by the depleted A2′′ absorption of Fe(CO)5
near this position (2022.0 cm-1, Table 1). At this early
time (1.5 µs), 1 is present in rather small concentration.
Only the high-frequency A′ ν(CO) band of 1 is just barely
discernible at 2056 cm-1, while the other two ν(CO)
modes of this complex (A′ and A′′, expected at 1988.5
and 1978.6 cm-1, see Table 1) are masked by the much
stronger B2 absorption of 2 in the center of this region
(1983.2 cm-1, Table 1).
The subsequently occurring s-trans f s-cis rearrange-
ment of the η4-coordinated butadiene ligand at the
Fe(CO)3 unit, 5 f 1 (Scheme 4), is evident from the
spectral changes monitored within a period of about 20
ms, which ultimately lead to the difference spectrum
shown in Figure 8D. The gradual disappearance of 5
and the concomitant formation of 1 are exemplarily
illustrated by the ∆A traces displayed in Figure 8B and
8C, respectively. Note the additional depletion at 2024
cm-1, which confirms that the missing third band of 5
is indeed located at this position. The concentration of
the main product 2 does not change during this period,
as indicated by the constancy of ∆A at 2084 and 2006
cm-1. The apparent increase of absorbance around the
position of the B2 ν(CO) band of 2 at 1984 cm-1 just
reflects the growth of the closely spaced absorptions of
1 at 1988 and 1978 cm-1 (Figure 8C).
The absorbance changes recorded at a wide range of
monitoring wavenumbers, including those displayed in
Figure 8B and 8C, invariably show satisfactory fits to
first-order kinetics, yielding an average value of kobs
)
250 ((24) s-1. Comparative measurements under CO
atmosphere instead of argon yield almost the same rate
constant within the margins of error, kobs ) 274 ((18)
s-1. In other words, the presence of added CO (9.2 mM
in CO-saturated cyclohexane)54 has no noticeable influ-
ence on the decay of 5. It does nevertheless change the
(53) An investigation into the gradual conversion of Fe(CO)5 into
Fe(CO)3(η4-s-cis-1,3-bd) (1) under continuous irradiation, aiming at the
determination of quantum yields for the individual steps at various
wavelengths, is in progress and will be published in a forthcoming
paper.
(54) Wilhelm, E.; Battino, R. J . Chem. Thermodyn. 1973, 5, 117-
120.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1705
Sch em e 5
system. It shows the depleted ν(CO) absorptions of
Fe(CO)5 along with a product spectrum that is domi-
nated by the four-band pattern of the Fe(CO)4(solv)
fragment58 (C2v symmetry; 2 A1, B1, and B2 ν(CO)
modes). The frequencies of the latter species, as well as
the relative intensities, are in good agreement with the
data reported for the closely related (singlet) Fe(CO)4-
(CH4), photogenerated from Fe(CO)5 in a low-tempera-
ture methane matrix.56
In addition, a product band of moderate intensity,
marked as a shaded area, is observed at 1926 cm-1. This
frequency is not far from the position of the E ν(CO)
mode of the Fe(CO)3 fragment in a methane matrix
(1930.4 cm-1),56,62 such that the assignment to the
species Fe(CO)3(solv) is justified. The much weaker A1
ν(CO) mode of this species, expected around 2036 cm-1
(2040.1 cm-1 in solid methane),56,62 is just barely
discernible at this wavenumber. This doubly unsatur-
ated fragment decays very rapidly. As illustrated by the
inset in Figure 9A, the absorption band at 1926 cm-1
has largely vanished within <1 µs, i.e., before the
detection system has fully responded. It is no longer
observable in the difference spectrum recorded after 1.2
µs (Figure 9B), which shows the well-developed ν(CO)
pattern of Fe(CO)4(solv).
initial product distribution in a sense that it reduces
the amount of the tricarbonyliron complex 5 in propor-
tion to the main product 2 by a factor of about 0.6-0.7.
Somewhat puzzling is the much higher velocity of the
5 f 1 rearrangement in comparison with the data (kobs
≈ 77 s-1) evaluated from those experiments where 5 was
generated by photoisomerization of 1 in CO-saturated
solution at nearly the same temperature (about 25 °C).
This large acceleration (by a factor of 3.5) is certainly
due to the presence of added 1,3-butadiene in high
concentration, though we are not yet in a position to
present a well-founded interpretation of this effect.55
The simplest way to rationalize the single-photon
route from Fe(CO)5 to the Fe(CO)3(η4-1,3-bd) complexes
5 and 1 would be to invoke the Fe(CO)3(solv) as a
directly photogenerated species, Scheme 5.18,20 This
would also explain the aforementioned influence of
added CO on the initial product ratio, since competitive
capturing of Fe(CO)3(solv) by carbon monoxide on one
hand and 1,3-butadiene on the other should ultimately
favor the formation of the main product 2 at the expense
of 5 (and 1).
P h otofr a gm en ta tion of F e(CO)5 in th e Absen ce
of 1,3-Bu ta d ien e. We are aware that the above pro-
posal conflicts with the conventional belief that multiple
CO dissociation as a single-photon reaction of Fe(CO)5
does not occur in condensed media,50,56-58 although it
has been observed in the gas phase.59-61 Therefore, we
set out to reexamine the flash photolysis of Fe(CO)5 in
hydrocarbon solution at ambient temperature, where
Fe(CO)4(solv) has been detected by means of time-
resolved IR spectroscopy in the milli- and microsecond
time range.57,58 In the following we concentrate on the
flash photolysis of Fe(CO)5 in CO-saturated cyclohexane
solution. Observations made in comparative experi-
ments under argon atmosphere are nevertheless briefly
mentioned, if relevant to the discussion.
At this time, the absorption at 2036 cm-1 has grown
significantly, but is of course no longer attributable to
the Fe(CO)3(solv) fragment. Rather, it can be assigned
to the unbridged form of the dinuclear species Fe2(CO)8,
supposed to originate from the reaction of Fe(CO)3(solv)
with unphotolyzed Fe(CO)5, Scheme 6,18 by analogy with
related findings in gas phase experiments.61 The as-
signment is based on comparison with literature data
available from low-temperature matrix studies, accord-
ing to which the unbridged Fe2(CO)863,64 species exhibits
two prominent bands at 2038 and 2006 cm-1 (in a CO-
doped argon matrix). Although the latter position in our
spectrum is obscured by the depleted E′ ν(CO) mode of
Fe(CO)5, this does not completely prevent the observa-
tion of an overlaid transient absorption centered at 2004
cm-1, the growth and decay of which parallel the ab-
sorbance changes at 2036 cm-1 (Figure 9E). We assume
that a certain amount of Fe(CO)3(solv) is concomitantly
trapped by CO with formation of additional Fe(CO)4-
(solv). Unfortunately, we are not in a position to present
direct spectroscopic evidence of the occurrence of this
latter process, since it takes place within the response
period of our instrumentation, during which the ν(CO)
absorption signals of Fe(CO)4(solv) continue to grow
anyhow. However, the above assumption seems never-
theless justified, since in the absence of added CO it
takes much longer (10-15 µs) until the absorption band
of Fe(CO)3(solv) at 1926 cm-1 has completely disap-
peared.
The transient IR difference spectrum displayed in
Figure 9A is recorded at a very early time (0.4 µs),
shorter than the full response time of our detection
(55) The apparent influence of various added trapping agents L,
including both strongly binding ligands and weakly bound stand-in
ligands, on the decay kinetics of photogenerated 5 is currently under
investigation, a functional relationship of the type kobs ) kiso(1 + [L])
+ kL[L] being used as a working hypothesis.
The fragment Fe(CO)4(solv) vanishes with kobs ≈ 2.7
× 105 s-1, as evaluated from the ∆A trace at 1986 cm-1
(inset in Figure 9B). A substantial amount, if not the
predominant portion (vide infra), reacts with Fe(CO)5
to form the dinuclear product Fe2(CO)9,65 Scheme 6. This
is evident from the ν(CO) bands emerging at 2056 cm-1
(barely discernible already after 1.2 µs, Figure 9B), in
(56) Poliakoff, M.; Turner, J . J . J . Chem. Soc., Dalton Trans. 1973,
2276-2285.
(57) Church, S. P.; Grevels, F.-W.; Hermann, H.; Kelly, J . M.;
Klotzbu¨cher, W. E.; Schaffner, K. J . Chem. Soc., Chem. Commun. 1985,
594-596.
(58) Grevels, F.-W. In Photoprocesses in Transition Metal Complexes,
Biosystems and other Molecules. Experiments and Theory; Kochanski,
E., Ed..; Kluwer: Dordrecht, 1992; pp 141-171.
(59) Yardley, J . T.; Gitlin, B.; Nathanson, G.; Rosan, A. M. J . Chem.
Phys. 1981, 74, 370-378.
(62) Poliakoff, M. J . Chem. Soc., Dalton Trans. 1974, 210-212.
(63) Poliakoff, M.; Turner, J . J . J . Chem. Soc. (A) 1971, 2403-2410.
(64) Fletcher, S. C.; Poliakoff, M.; Turner, J . J . Inorg. Chem. 1986,
25, 3597-3604.
(60) Seder, T. A.; Ouderkirk, A. J .; Weitz, E. J . Chem. Phys. 1986,
85, 1977-1986.
(61) Ryther, R. J .; Weitz, E. J . Phys. Chem. 1991, 95, 9841-9852.

1706 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
F igu r e 9. Flash photolysis (λexc ) 308 nm) of Fe(CO)5 (2 mM, in CO-saturated cyclohexane solution) at 25 °C. (A) Transient
ν(CO) difference spectrum recorded after 0.4 µs showing the depletion bands of Fe(CO)5 [2020 (A2′′), 1998 cm-1 (E′)] and
the absorptions of the photogenerated fragments Fe(CO)4(solv) [2084 (A1), 1986 (B1), 1970 (A1), 1950 (B2) cm-1; B)] and
Fe(CO)3(solv) [1926 (E) cm-1; shaded area]. (B) Spectrum recorded after 1.2 µs showing the fully developed ν(CO) pattern
of Fe(CO)4(solv) (B) along with weak absorptions at 2056 and 2036 cm-1, attributed (see text) to Fe2(CO)9 and Fe2(CO)8.
(C) Spectrum recorded after 10 µs showing the ν(CO) pattern of Fe2(CO)9 (hatched area, for details see text) along with
two residual absorptions at 1950 cm-1 [(*), assigned to Fe(CO)4(H2O)] and 1970 cm-1 (associated with an unidentified
species X). (D) Growth of Fe2(CO)9 monitored at the positions marked in spectra B, C, F by a solid square (9). (E) Growth
and decay of Fe2(CO)8 monitored at the positions marked in spectra B, C, F by a solid triangle (2). (F) Spectrum recorded
after 45 µs showing the ν(CO) pattern of Fe2(CO)9 (hatched area) along with the residual absorption at 1950 cm-1 [(*),
assigned to Fe(CO)4(H2O)].
matrix),64 allowing for some shifts in frequency arising
Sch em e 6
from the difference in temperature and media.66
Trapping of Fe(CO)4(solv) by the added carbon mon-
oxide with re-formation of Fe(CO)5 is included in
Scheme 6 as a competing process.67 A reasonable esti-
(65) Recall that the photochemical conversion of Fe(CO)5 into Fe2-
(CO)9, discovered a century ago as the first light-induced reaction in
carbonylmetal chemistry, is a very efficient process providing high-
yield access to Fe2(CO)9 on a preparative scale: (a) Mond, L.; Langer,
C. J . Chem. Soc., London 1891, 59, 1090-1093. (b) Dewar, J .; J ones,
H. O. Proc. R. Soc., London (A) 1905, 76, 558-577. (c) Speyer, E.; Wolf,
H. Chem. Ber. 1927, 60, 1424-1425. (d) Brauer, G., Ed. Handbuch
der Pra¨parativen Anorganischen Chemie, 3rd ed.; Enke: Stuttgart,
Germany, 1981; Vol. 3, pp 1827-1828.
(66) It is unfortunate that, due to the insolubility of Fe2(CO)9, data
from spectra in solution are not available. The ν(CO) pattern of Fe2-
(CO)9 embedded in a potassium iodide pellet differs somewhat from
the 2030-2012 cm-1 region, and at 1826 cm-1, which
are rendered prominent as hatched areas in the spec-
trum recorded after 10 µs (Figure 9C). This pattern
the low-temperature matrix data with respect to the frequencies (2082,
closely resembles the absorptions observed for Fe2(CO)9
2019, and 1829 cm-1) and to the relative intensities as well: Cotton,
under low-temperature matrix isolation conditions63
F. A.; Liehr, A. D.; Wilkinson, G. J . Inorg. Nucl. Chem. 1955, 1, 175-
186.
(2064, 2037, and 1847.4 cm-1; in a CO-doped argon

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1707
mate of the branching ratio of these two reaction
channels of Fe(CO)4(solv) is obtained by using comple-
mentary data from comparative measurements in the
absence of added CO. Taking into account that under
argon atmosphere the decay of Fe(CO)4(solv) occurs with
kobs ≈ 1.8 × 105 s-1, we conclude that in CO-saturated
solution (kobs ≈ 2.7 × 105 s-1) only one-third of Fe(CO)4-
(solv) is trapped by CO to re-form the parent Fe(CO)5,
while two-thirds react with Fe(CO)5 to yield Fe2(CO)9,
although CO is present in higher concentration (9.2
mM)54 than Fe(CO)5 (2 mM). This means that the
reaction of Fe(CO)4(solv) with Fe(CO)5 (kFe(CO)5 ≈ 9 ×
107 L‚mol-1‚s-1) is faster than the reaction with CO (kCO
≈ 107 L‚mol-1‚s-1) by about 1 order of magnitude. It is
interesting to note that this ratio has also been found
in the gas phase,61 where the reaction of photogenerated
Fe(CO)4 with the parent Fe(CO)5 occurs with kFe(CO)5
)
5.2 × 10-13 cm3‚molecule-1‚s-1, while trapping by
added CO takes place with kCO ) 5.2 × 10-14 cm3‚mole-
cule-1‚s-1
.
The growth of Fe2(CO)9, exemplarily illustrated by the
∆A traces displayed in Figure 9D, is not terminated by
the time (10 µs) when Fe(CO)4(solv) has completely
vanished. It rather proceeds further, though by irregular
kinetics, thus hinting at other route(s) leading to
Fe2(CO)9 as the ultimate product. The most obvious one
involves the species Fe2(CO)8, which is converted into
Fe2(CO)9 by taking up CO from the solution,68 Scheme
6. This reaction occurs with kobs ≈ 1.6 × 105 s-1, as
evaluated from the decay of the ∆A traces recorded at
2036 and 2004 cm-1 (Figure 9E).
As shown in Figure 9C, the decay of Fe(CO)4(solv) in
CO-saturated solution leaves behind two residual ab-
sorptions at 1970 (×) and 1950 cm-1 (*), which before
were masked by the strong absorptions of Fe(CO)4(solv)
at these positions. The first one has not yet been
identified, whereas the second one (which is still observ-
able after 45 µs, Figure 9F) is assigned to the species
Fe(CO)4(H2O), formed from Fe(CO)4(solv) by reaction
with apparently unavoidable traces of moisture.
This assignment rests on a comparative experiment
carried out in water-saturated cyclohexane (0.01% ( 5.5
mM;69 3 mM70). The transient IR ν(CO) difference
spectra recorded shortly after the flash photolysis of
Fe(CO)5 (Figure 10A) in essence still show the four-band
product pattern of Fe(CO)4(solv). However, at this stage
already we note that the relative intensities of the B1
and B2 modes differ slightly from the picture familiar
from measurements in rigorously dried solvent (Figure
9A and 9B). On going from 1.2 µs (spectrum displayed
with closed circles, B) to 2.0 µs (spectrum displayed with
open circles, O), the intensity at 1950 cm-1 increases
F igu r e 10. Flash photolysis (λexc ) 308 nm) of Fe(CO)5 (2
mM, in water-saturated cyclohexane solution) under CO
atmosphere at 25 °C. (A) Transient ν(CO) difference spectra
recorded after 1.2 µs (B) and 2.0 µs (O) showing the deple-
tion bands of Fe(CO)5 [2020 (A2′′), 1998 (E′) cm-1] and the
absorptions of photogenerated Fe(CO)4(solv) [2084 (A1),
1986 (B1), 1970 (A1), 1950 (B2) cm-1]. (B) Spectrum recorded
after 10 µs showing product absorptions attributed to
Fe(CO)4(H2O) (2 A1, E) and Fe2(CO)9 along with the
depleted bands of Fe(CO)5. (C) Decay of Fe(CO)4(solv) and
Fe(CO)4(H2O) (monitored at 1986 and 1950 cm-1) and
growth of Fe2(CO)9 (monitored at 2056 cm-1). (D) Spectrum
recorded after 1100 µs showing the ν(CO) pattern of
Fe2(CO)9 (hatched area) along with the depleted bands of
Fe(CO)5.
(67) Recall that the photoinduced exchange of CO for isotopically
labeled carbon monoxide is a well-known reaction of Fe(CO)5: (a)
Noack, K.; Ruch, M. J . Organomet. Chem. 1969, 17, 309-322. (b) Bor,
G. Inorg. Chim. Acta 1969, 3, 191-195.
(68) Under neat argon atmosphere, where merely the photodisso-
ciated carbon monoxide is available, this additional formation of Fe2-
(CO)9 seems negligible. What occurs instead is the growth of a strong
absorption in the 2050-2040 cm-1 region, which fills the gap between
the apparent maxima of Fe2(CO)9 at 2056 and 2026 cm-1. We attribute
this new absorption to Fe3(CO)12, the strongest ν(CO) band of which
in hydrocarbon solution reportedly appears at 2046 cm-1: Noack, K.
Helv. Chim. Acta 1962, 45, 1847-1859.
noticeably at the expense of the absorption at 1986 cm-1
The latter vanishes completely within 10 µs, whereas
the former persists for a much longer time.
The spectrum recorded after 10 µs (Figure 10B) is
dominated by the strong band at 1950 cm-1, which
exhibits a distinct shoulder at the high-frequency side
(1964 cm-1). In addition, three other absorptions have
grown in at 2056, 2032, and 1826 cm-1, which are
.
(69) Duve, G.; Fuchs, O.; Overbeck, H. Organische Chemikalien
Hoechst, Lo¨semittel Hoechst, Ein Handbuch fu¨r Laboratorium und
Betrieb; Hoechst AG: Frankfurt a. M., 1976.
(70) Goldman, S. Can. J . Chem. 1974, 52, 1668-1680.

1708 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
Sch em e 7
readily attributed to Fe2(CO)9. They grow further with
concomitant decrease of the absorption at 1950 cm-1
within about 1100 µs (Figure 10C). In the last spectrum
(Figure 10D), this band has completely vanished and
Fe2(CO)9 is now the only product. Its formation in two
stages is clearly recognizable in Figure 10C, the right-
hand part of which shows the absorbance changes on
an expanded time scale. The rapid growth of the first
batch of Fe2(CO)9 goes along with the complete decay
of the Fe(CO)4(solv) absorption at 1986 cm-1 and with
the concomitant disappearance of a part of the absorp-
tion at 1950 cm-1. The much slower growth of the
second batch of Fe2(CO)9 mirrors the further decay at
1950 cm-1. Hence it follows that this latter absorption
is associated with a reservoir form of the Fe(CO)4
fragment, Fe(CO)4(H2O) being the obvious candidate.
It apparently exists with Fe(CO)4(solv) in an equilibri-
um, from which it is slowly released, Scheme 6. The
competitive reactions with Fe(CO)5 and CO yield Fe2-
(CO)9 on one hand and lead to the re-formation of some
Fe(CO)5 on the other.
The species Fe(CO)4(H2O), like other Fe(CO)4L com-
plexes with n-donor ligands,71,72 should possess a trigo-
nal-bipyramidal structure with the ligand L in an axial
position (C3v symmetry; three IR active ν(CO) modes:
2 A1, E). Taking Fe(CO)4(PEt3) as a typical example
[three bands at 2048.6 (A1), 1974.5 (A1), and 1935.7 (E)
cm-1 with 1:1.3:8.8 intensity ratio],73 we assign the
strong band at 1950 cm-1 and the shoulder at 1964 cm-1
to the E and the low-frequency A1 ν(CO) modes of
Fe(CO)4(H2O), respectively. The high-frequency A1 mode
of this species, expected around 2050-2060 cm-1, is
hidden underneath the strong absorption of Fe2(CO)9
centered at 2056 cm-1 (Figure 10B). Finally, we note a
barely discernible shoulder at the low-frequency side of
the E mode of Fe(CO)4(H2O), which we tentatively
attribute to a small amount of trans-Fe(CO)3(H2O)2 [D3h
symmetry; one IR active ν(CO) mode: E′], taking into
account that the E′ mode of trans-Fe(CO)3L2 complexes
commonly appears at a lower frequency than the E
mode of the corresponding Fe(CO)4L.73
changes in the intensities of the ν(CO) IR spectrum, if
kept in the dark. Therefore, such a stock solution is used
for the intended investigation into the photolytic be-
havior of 2.
Time-resolved IR spectroscopy with measurements in
the micro- and millisecond time ranges reveals that the
flash photolysis (λexc ) 308 nm) of 2 initially generates
Fe(CO)3(η4-s-trans-1,3-bd) (5) as the major product,
which then undergoes η4-s-trans f η4-s-cis rearrange-
ment to form 1, Scheme 7.20
The ν(CO) pattern of 5 appears almost instanta-
neously along with the depleted absorptions of the
starting material, as observed in an IR difference
spectrum recorded 1.2 µs after the flash photolysis of
2. Virtually no spectral changes occur during the
following period of 100 µs, such that a spectrum
recorded after 0.1 ms (Figure 11A) is also representative
of the initial situation. The assignment of the dominant
product absorptions in this spectrum (shaded areas) to
Fe(CO)3(η4-s-trans-1,3-bd) (5) is straightforward by
comparison with Figures 1C and 2A. The weak absorp-
tion at 2054 cm-1 is attributed to the high-frequency
A′ ν(CO) mode of 1, which up to this stage is present
only in small concentration. The low-frequency A′ and
A′′ ν(CO) absorptions of 1, expected at 1988 and 1978
cm-1 (cf. Table 1), are canceled due to foot-and-tail
overlap with the strong B2 depletion band of 2 at
1984 cm-1
.
F la sh P h ot olysis of F e(CO)4(η2-1,3-b u t a d ien e)
(2). What remains to be addressed is the intramolecular
photosubstitution of one CO by the free vinyl unit of
the η2-coordinated butadiene ligand in the tetracarbonyl
complex 2, which ultimately yields Fe(CO)3(η4-s-cis-1,3-
bd) (1). Compound 2 is most easily accessible in pure
state if the preparation from Fe2(CO)9 and 1,3-butadiene
and its isolation from the reaction mixture is performed
under milder conditions than previously reported.6 The
improved preparation, along with the complementary
A considerable increase in the concentration of 1
occurs subsequently at the expense of 5, which vanishes
completely within a period of about 20 ms, as illustrated
in Figure 11B. All three ν(CO) bands of 1 appear well
resolved in the resulting difference spectrum (Figure
11C). There is no recovery of the starting material, as
manifested by the persisting depletion of 2 at 2082 (A1)
and 2004 cm-1 (A1/B1).
However, the depletion of the B2 band of 2 at 1984
cm-1 is largely canceled, thus indicating that a new
absorption has grown near this position. Bearing in
mind that the strongest ν(CO) band of, for example,
1
characterization of 2 by H and 13C NMR spectroscopy,
is given in the Experimental Section.
Fe(CO)3(η2-ethene)2 appears at 1981 cm-1 74
we assume
,
The moderate thermal stability of 2 in hydrocarbon
solution at ambient temperature is greatly improved in
the presence of added 1,3-butadiene. Even after several
days, a solution of 2 in cyclohexane containing, for
example, 85 mM free 1,3-butadiene shows virtually no
that a product of type Fe(CO)3(η2-1,3-bd)2 (7) is formed
by addition of 1,3-butadiene to 5 in a side reaction of
the 5 f 1 rearrangement, Scheme 7. This side reaction
may involve the species Fe(CO)3(η2-1,3-bd) (3) as a
reactive intermediate, by analogy with the proposal
made in Scheme 3 for the formation of a minor amount
of Fe(CO)4(η2-1,3-bd) (2) as a side product of the 5 f 1
rearrangement under CO atmosphere.
(71) Shriver, D. F.; Whitmire, K. H. In Comprehensive Organome-
tallic Chemistry; Wilkinson, G., Stone, G. A., Abel, E. W., Eds.;
Pergamon: Oxford, England, 1982; Vol. 4, pp 289-291.
(72) Chen, Y.; Hartmann, M.; Frenking, G. Z. Anorg. Allg. Chem.
2001, 627, 985-998.
(73) Reckziegel, A.; Bigorgne, M. J . Organomet. Chem. 1965, 3, 341-
354.
(74) Wuu, Y.-M.; Bentsen, J . G.; Brinkley, C. G.; Wrighton, M. S.
Inorg. Chem. 1987, 26, 530-540.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1709
F igu r e 12. FTIR difference spectrum recorded several
minutes after single-flash irradiation (XeCl excimer laser,
λexc ) 308 nm; 95 mJ incident flash energy) of Fe(CO)4(η2-
1,3-bd) (2, 0.5 mM, in cyclohexane solution containing 85
mM 1,3-butadiene) under argon atmosphere at ambient
temperature in an IR cell (d ) 1 mm), showing Fe(CO)3-
(η4-1,3-bd) (1) as the only product (for frequency data see
Table 1).
The changes of ∆A in the 20 ms time window,
recorded at 2054 and 2024 cm-1 (Figure 11B) and a wide
range of other monitoring wavenumbers, show satisfac-
tory fits to first-order kinetics, yielding an average value
of kobs ) 185 ((10) s-1. These data underline the
influence of the concentration of added 1,3-butadiene
(85 mM) on the decay of 5. It is smaller compared with
the value found in the presence of 130 mM 1,3-butadiene
(250 s-1), where 5 was generated from Fe(CO)5, but
much larger than the one evaluated for the 5 f 1
rearrangement in CO-saturated solution in the absence
of added 1,3-butadiene (77 s-1, at 25 °C). The accelera-
tion by 1,3-butadiene may in part be due to the ad-
ditional reaction channel leading to the type 7 product,
but this is probably not sufficient to rationalize this
effect entirely. One might invoke, for example, a de-
crease of the activation barrier of the 5 f 1 rearrange-
ment arising from a stabilization of the transition
state(s), TS[5-1] and/or TS[5-3] (Figure 6), by inter-
action with the diene.55
F igu r e 11. Flash photolysis (λexc ) 308 nm) of Fe(CO)4-
(η2-1,3-bd) (2, 0.5 mM, in cyclohexane solution containing
85 mM 1,3-butadiene) under argon atmosphere at 25 °C.
(A) Transient ν(CO) difference spectrum recorded after 0.1
ms showing the depletion bands of 2 [2082 (A1), 2004 (A1/
B1), 1984 cm-1 (B2)] and the absorptions of photogenerated
Fe(CO)3(η4-s-trans-1,3-bd) (5) (2024, 1962, 1944 cm-1
;
shaded area) along with a weak absorption attributed to
Fe(CO)3(η4-s-cis-1,3-bd) (1) [2054 cm-1 (A′); (B)]. (B) Growth
of 1 (monitored at 2054 cm-1) with concomitant decay of 5
(monitored at 2024 cm-1). (C) Difference spectrum recorded
after 18 ms showing the depletion bands of 2 along with
the product pattern of 1 [2054 (A′), 1990 (A′), 1976 cm-1
(A′′); (B)]; for further details see text.
Con clu d in g Rem a r k s
Surveying the observations made in this work, we can
distinguish two highlights: the discovery of Fe(CO)3-
(η4-s-trans-1,3-bd) (5) as a transient isomer of the
classical Fe(CO)3(η4-s-cis-1,3-bd) (1) and the detection
of the doubly unsaturated fragment Fe(CO)3(solv) as a
single-photon product of Fe(CO)5 in addition to Fe(CO)4-
(solv). Armed with this information, we are now in a
position to present Scheme 818,20 as a substantially
expanded and revised version of the earlier proposals
made for the photochemical synthesis of 1 from Fe(CO)5
and 1,3-butadiene (Scheme 1)5 and the photolytic be-
havior of 1 as well.15,16,77
Fe(CO)3(η2-olefin)2 complexes are known for their
relatively low stability.74-76 Hence one would expect that
a compound of type Fe(CO)3(η2-1,3-bd)2 (7) undergoes
intramolecular displacement of one of the two η2-1,3-
butadiene ligands with ultimate formation of an ad-
ditional amount of 1, Scheme 7. This is indeed recog-
nizable from a comparison of the transient difference
spectrum shown in Figure 11C with an FTIR difference
spectrum recorded several minutes later (Figure 12).
The intensity of the ν(CO) pattern of 1 in relation to
the depleted absorptions of the starting material 2 has
increased significantly, while the depletion of the low-
frequency B2 band of 2 is no longer canceled by the
absorption associated with species 7.
The involvement of the η4-s-trans-1,3-bd complex 5
as a kinetically favored intermediate in either of the two
(77) Trapping of the CO loss fragment 4 by potential ligands other
than CO is included in the scheme, though such reactions are not dealt
with in the present paper. They will be communicated in a forthcoming
publication along with the results of current investigations into the
photolytic behavior of other Fe(CO)3(η4-1,3-diene) complexes. Prelimi-
nary observations indicate that compounds containing acyclic 1,3-
dienes (isoprene, 2,3-dimethyl-1,3-butadiene) expectedly undergo both
photolytic CO detachment and η4-s-cis f η4-s-trans photoisomerization.
Here again, the latter process is largely reversible and, hence, has little
net effect apart from diminishing the efficiency of CO photosubstitution
in accord with previous quantum yield measurements.15
(75) Fleckner, H.; Grevels, F.-W.; Hess, D. J . Am. Chem. Soc. 1984,
106, 2027-2032.
(76) Weiller, B. H.; Miller, M. E.; Grant, E. R. J . Am. Chem. Soc.
1987, 109, 352-356.

1710 Organometallics, Vol. 22, No. 8, 2003
Bachler et al.
Fe(CO)3(η2-(Z)-cyclooctene)2.75 This precaution is par-
ticularly necessary in cases where a large excess of the
entering ligand is prohibitive for economic reasons.
It is unfortunate that the response time of our
instrumentation prevents us from gathering information
on the early events in the photofragmentation of
Fe(CO)5, prior to the appearance of (singlet) Fe(CO)4-
(solv) and Fe(CO)3(solv) (Figure 9A). Picosecond studies
conducted elsewhere82 have nevertheless shown that the
naked (triplet) Fe(CO)4 is almost instantaneously pho-
togenerated from Fe(CO)5 in heptane solution (where
it persists for not less than 600 ps), but no evidence was
found for the doubly unsaturated fragment Fe(CO)3. All
the more exciting are most recent findings according to
which both triplet Fe(CO)4 and Fe(CO)3 are clearly
observable in heptane at a time as early as 10 ps.83
The occurrence of 2-fold CO photodetachment from
Fe(CO)5 is also relevant to reactions other than those
specifically dealt with in the present work. Obvious
examples are photoinduced carbonyliron-catalyzed pro-
cesses, such as olefin isomerization or hydrogenation,85
where the Fe(CO)3 moiety plays a key role as the
repeating unit in the catalytic cycle. Another interesting
case is the single-photon conversion of Fe(CO)5 into
mixtures of mono- and disubstituted derivatives of type
Fe(CO)4L and Fe(CO)3L2 (L ) mono-olefin or phospho-
rus ligands),58,84-87 where the addition of two ligands L
to the doubly unsaturated Fe(CO)3(solv) fragment can
now be considered as the most simple route to the
Fe(CO)3L2 products. Competitive trapping of Fe(CO)3-
(solv) by Fe(CO)5 may occur in case of high Fe(CO)5:L
ratios. This would yield the dinuclear species Fe2(CO)8
(cf. Scheme 6), which upon reaction with two ligands L
can form two Fe(CO)4L product molecules as the ulti-
mate result of the absorption of one single photon, thus
providing a new, simple rationale of quantum yields
exceeding unity.87
Sch em e 8
routes from Fe(CO)5 to 1 is readily rationalized as
follows. The s-trans form is by far the predominant
conformer of 1,3-butadiene,78,79 such that the Fe(CO)3-
(solv) fragment, if generated in the presence of the
diene, preferentially adds this s-trans conformer to yield
Fe(CO)3(η2-1,3-bd) (3) (with φd ) -150.1°; Figure 3) as
the primary adduct.80 The same species 3 will also be
formed by photolytic CO detachment from Fe(CO)4(η2-
1,3-bd) (2), which contains the diene in a nearly s-trans
conformation.81 Separated from 5 by a barrier as low
as 0.9 kcal‚mol-1 (Figure 5), species 3 will immediately
move downhill in energy with formation of 5, which
subsequently undergoes η4-s-trans f η4-s-cis rearrange-
ment to yield 1 as the thermodynamically more stable
product.
Exp er im en ta l Section
Trapping of the fragments Fe(CO)4(solv) and Fe(CO)3-
(solv) by unphotolyzed Fe(CO)5 or by trace impurities
such as water (Scheme 6) is omitted in Scheme 8, since
these processes are negligible for kinetic reasons as long
as 1,3-butadiene is present in sufficiently large excess.
However, one has to consider undesired Fe2(CO)9 for-
mation when exploiting the photosubstitution of
Fe(CO)5 for preparative purposes in runs with high
complex concentrations. It is for this reason that in the
preparation of 1 (see Experimental Section) the parent
Fe(CO)5 is added in several successive portions to the
solution of 1,3-butadiene, by analogy with the procedure
proven successful in the synthesis of, for example,
Gen er a l Rem a r k s. All reactions and manipulations of the
air-sensitive carbonyliron complexes are carried out under
argon and in argon-saturated solvents, unless otherwise noted,
using standard Schlenck techniques. Spectra are recorded on
the following instruments: IR, Perkin-Elmer 1600 FTIR
(operating with 2 cm-1 resolution), Bruker IFS 66 (operating
with 0.5 cm-1 resolution); UV-vis, Shimadzu UV-2102PC;
NMR, Bruker ARX 250, Bruker DRX 400.
Ma ter ia ls. Synthetic grade n-pentane (95%, Merck), 1,3-
butadiene (99%, Messer Griesheim), and chloroform-d1 (99.8%,
Deutero GmbH) are used as received. High-purity gases used
for the flash photolysis experiments (Ar, g99.9999%; CO,
g99.997) are purchased from Messer Griesheim. Cyclohex-
ane (>99.6%, Fisher Scientific) is refluxed over LiAlH4 for 24
h followed by distillation under argon atmosphere through a
5 m glass-packed column until no impurities are detectable
by GC. Fe(CO)5 (>97%, Fluka) is purified by trap-to-trap
vacuum distillation prior to use (IR ν(CO) data: see Table 1).
(78) Huber-Wa¨lchli, P. Ber. Bunsen-Ges. Phys. Chem. 1978, 82,
10-12.
(79) Mui, P. W.; Grunwald, E. J . Am. Chem. Soc. 1982, 104, 6562-
6566.
(80) It is worth mentioning in this context that the involvement of
the s-trans conformer of a free 1,3-diene was not considered in related
gas phase studies.49 Rather, photogenerated Fe(CO)3 was suggested
to add the 1,3-diene in the form of its s-cis conformer, yielding an (s-
cis-1,3-diene)-Fe(CO)3 type complex as the only initially formed adduct.
(81) A geometry optimization of Fe(CO)4(η2-1,3-bd) (2) at the BP86
level of DFT was started with a guess derived from the structure of
[Fe(CO)4]2(η2:η2-1,3-bd) (Klanderman, K. A. Ph.D. Dissertation, Uni-
versity of Wisconsin, 1965; Dissertation Abstr. 1965, 25, 6253) by
removal of one Fe(CO)4 unit. The computed structure exhibits a diene
dihedral angle of φd ) -161.2°.
(82) Snee, P. T.; Payne, C. K.; Kotz, K. T.; Yang, H.; Harris, C. B. J .
Am. Chem. Soc. 2001, 123, 2255-2264.
(83) George, M. W. Personal communication.
(84) Angermund, H.; Bandyopadhyay, A. K.; Grevels, F.-W.; Mark.
F. J . Am. Chem. Soc. 1989, 111, 4656-4661.
(85) Schroeder, M. A.; Wrighton, M. S. J . Am. Chem. Soc. 1976, 98,
551-558.
(86) Nayak, S. K.; Burkey, T. J . Inorg. Chem. 1992, 31, 1125-1127.
(87) Nayak, S. K.; Farrell, G. J .; Burkey, T. J . Inorg. Chem. 1994,
33, 2236-2242.

Carbonyliron-Diene Photochemistry
Organometallics, Vol. 22, No. 8, 2003 1711
Fe2(CO)9,65d Fe(CO)3(η4-1,3-bd) (1),5 and Fe(CO)4(η2-1,3-bd) (2)6
are prepared according to the published procedures, but with
some modifications as described in the following.
Com p u ta tion a l Meth od s. All DFT calculations are carried
out by means of the Gaussian 98 package of programs,91 using
Becke’s gradient-corrected exchange-energy functional92 in
combination with Perdew’s density-functional approximation
for the correlation energy,93 referenced as BP86. For the
central iron atom a pseudopotential developed by Stoll and
co-workers is used,94 which simulates the 10 neon shell core
electrons of the iron atom, while the remaining 16 valence
electrons are described by means of a (8s7p6d1f/6s5p3d1f)
basis set. Carbon, oxygen, and hydrogen atoms are treated by
means of the 6-31G* basis set.95 Geometry optimizations are
performed in redundant internal coordinates96 using the GDIIS
algorithm (geometry optimization by direct inversion in the
iterative subspace).97 Transition states are searched for by
means of the STQN method98 and finally refined using the
regular transition state optimization. Energies are reported
with and without zero point energy (ZPE) correction. Frequen-
cies are calculated from analytical second derivatives.
F e(CO)3(η4-1,3-bu ta d ien e) (1). A solution of Fe(CO)5 in
1,3-butadiene-saturated n-pentane (400 mL), which replaces
the toxic benzene solvent used in the published procedure,5 is
irradiated at ambient temperature in a water-cooled immer-
sion-well apparatus88 (Solidex glass, λ > 280 nm) equipped
with a Philips HPK 125 W high-pressure mercury lamp. The
total amount of Fe(CO)5 (15 mL, 21.8 g, 0.11 mol) is added in
three portions, according to the progress of the conversion (as
monitored by IR), to suppress the formation of Fe2(CO)9. The
irradiation is continued until the starting material is largely
consumed (14 h). Vacuum evaporation of the volatiles (e10-2
hPa) at ambient temperature, followed by trap-to-trap vacuum
distillation of the residue at 40 °C, yields a yellow oil (14.8 g),
which upon redistillation using a concentric tube column
affords spectroscopically pure 1 (8.1 g, 38%), yellow oil, bp 51
°C at 9 hPa. IR ν(CO): see Table 1. 13C NMR (CDCl3, 300 K):
1
1
δ 211.82 (CO), 85.59 (d, J CH 170 Hz; dCH-), 40.82 (t, J CH
161 Hz; CH2d).
F e(CO)4(η2-1,3-bu ta d ien e) (2). A suspension of Fe2(CO)9
(1.21 g, 3.32 mmol) in a solution of 1,3-butadiene (ca. 6 g, ca.
0.11 mol) in n-pentane (150 mL) is stirred at ambient tem-
perature in the dark until the insoluble starting complex has
vanished (3 h). Evaporation of the volatiles at e-30 °C under
reduced pressure (e10-2 hPa) is continued (ca. 10 h) until
Fe(CO)5 is completely removed from the residue (monitored
by IR). Vacuum sublimation yields spectroscopically pure 2
as a yellow oil (0.67 g, 90%). IR ν(CO): see Table 1. 13C
Ack n ow led gm en t. This paper is dedicated to Pro-
fessor Karl Wieghardt on the occasion of his 60th
birthday. Skillful technical assistance by P. Bayer, D.
Merkl, R. Schrader, and the NMR spectroscopic staff of
the MPI fu¨r Strahlenchemie is gratefully acknowledged.
The authors wish to thank L. J . Currell, Dr. H. Go¨rner,
G. Klihm, Dr. W. E. Klotzbu¨cher, and Dr. B. Weimann
for their help and advice concerning flash photolysis in
combination with time-resolved IR spectroscopy and
computer-assisted signal analysis. Particular thanks are
due to Dr. M. W. George (University of Nottingham, UK)
for disclosing his recent findings prior to publication.
1
NMR (CDCl3, 300 K): δ 211.34 (CO), 142.23 (d, J CH 153 Hz;
CH2dCH-), 112.47 (t, 1J CH 156 Hz; CH2dCH-), 60.46 (d, 1J CH
1
155 Hz; η2-CH2dCH-), 36.13 (t, J CH 159 Hz; η2-CH2dCH-).
1H NMR (CDCl3, 263 K): δ 5.70 (m, 1 H), 5.27 (m, 1 H), 4.88
(m, 1 H), 4.02 (m, 1 H), 2.65 (m, 1 H), 2.59 (m, 1 H).
F la sh P h otolysis w ith Tim e-Resolved IR Sp ectr os-
cop y. The basic design of our instrumentation89 and details
of the actual configuration and performance90 (HgCdTe pho-
todiode, system response time 1-1.2 µs, 7-8 cm-1 spectral
resolution) have been described previously. In the present
study, the flash energy of the Lambda Physik EMG 200
excimer laser, operating with XeCl for λ ) 308 nm emission
(pulse duration 20 ns), is routinely attenuated to 25-30
mJ /pulse. With the chosen concentrations of Fe(CO)3(η4-1,3-
bd) (1) (1.5 mM), Fe(CO)4(η2-1,3-bd) (2) (0.5 mM), and
Fe(CO)5 (2 mM) in the 1 mm sample cell, the absorbance at
308 nm ranges from 0.25 to 0.35. Data acquisition for point-
by-point construction of transient spectra (every 2 cm-1) and
kinetic analysis (using homemade PC software for fitting
to a monoexponential function) routinely involves signal
averaging of three single-shot experiments. After each laser
shot, the thermostated IR cell is emptied and refilled with a
fresh portion of the respective stock solution from the reser-
voir, which is kept under the desired gas atmosphere at
1060 hPa.
OM020785C
(91) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.,
J r.; Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Millam, J . M.;
Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .;
Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo,
C.; Clifford, S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.;
Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J . B.; Cioslowski, J .; Ortiz, J . V.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.;
Fox, D. J .; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; J ohnson, B.; Chen, W.;
Wong, M. W.; Andres, J . L.; Gonzalez, C.; Head-Gordon, M.; Replogle,
E. S.; Pople, J . A. Gaussian 98, Revision A.5; Gaussian Inc.: Pitts-
burgh, PA, 1998.
(92) Becke, A. D. Phys. Rev. A 1988, 38, 3098-3100.
(93) Perdew, J . P. Phys. Rev. B 1986, 33, 8822-8824.
(94) See, for example: (a) Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim.
Acta 1993, 85, 441-450, and references therein. (b) Wedig, U.; Dolg,
M.; Stoll, H. In Quantum Chemistry: The Challenge of Transition
Metals and Coordination Chemistry; Veillard, A., Ed.; Reidel: Dor-
drecht, 1986; pp 79-89. (c) Leininger, T.; Nicklass, A.; Stoll, H.; Dolg,
M.; Schwerdtfeger, P. J . Chem. Phys. 1996, 105, 1052-1059, and
references therein.
(88) Grevels, F.-W.; Reuvers, J . G. A.; Takats, J . Inorg. Synth. 1986,
24, 176-180.
(89) Schaffner, K.; Grevels, F.-W. J . Mol. Struct. 1988, 173, 51-65.
(90) Grevels, F.-W.; Kerpen, K.; Klotzbu¨cher, W. E.; Schaffner, K.;
Goddard, R.; Weimann, B.; Kayran, C.; O¨ zkar, S. Organometallics
2001, 20, 4775-4792.
(95) Hehre, W. J .; Ditchfield, R.; Pople, J . A. J . Chem. Phys. 1972,
56, 2257-2261.
(96) Peng, C.; Ayala, P. Y.; Schlegel, H. B.; Frisch, M. J . J . Comput.
Chem. 1996, 17, 49-56.
(97) Csa´sza´r, P.; Pulay, P. J . Mol. Struct. 1984, 114, 31-34.
(98) Peng, C.; Schlegel, H. B. Isr. J . Chem. 1994, 33, 449-454.