Unraveling the photochemistry of Fe(CO)5 in solution: Observation of Fe(CO)3 and the conversion between 3Fe(CO)4 and 1Fe(CO)4(solvent)

Article abstract of DOI:10.1021/ja048411t

The photochemistry of Fe(CO)₅ (5) has been studied in heptane, supercritical (sc) Ar, scXe, and scCH₄ using time-resolved infrared spectroscopy (TRIR). ³Fe(CO)₄ (³4) and Fe(CO)₃(solvent) (3) are formed as primary photoproducts within the first few picoseconds. Complex 3 is formed via a single-photon process. In heptane, scCH₄, and scXe, ³4 decays to form ¹4·L (L = heptane, CH₄, or Xe) as well as reacting with 5 to form Fe₂(CO)₉. In heptane, 3 reacts with CO to form ¹4·L. The conversion of ³4 to ¹4·L has been monitored directly for the first time (L = heptane, k_obs = 7.8(±0.3) x 10⁷ s^-1; scCH₄, 5(±1) x 10⁶ s^-1; scXe, 2.1(±O.1) x 10⁷ s^-1). In scAr, ³4 and 3 react with CO to form 5 and ³4, respectively. We have determined the rate constant (k_CO = 1.2 x 10⁷ dm³ mol ^-1 s^-1) for the reaction of ³4 with CO in scAr, and this is very similar to the value obtained previously in the gas phase. Doping the scAr with either Xe or CH₄ resulted in ³4 reacting with Xe or CH₄ to form ¹4·Xe or ¹4·CH₄. The relative yield, [³4]:[3] decreases in the order heptane > scXe > scCH₄ ? scAr, and pressure-dependent measurements in scAr and scCH₄ indicate an influence of the solvent density on this ratio.

Full text of DOI:10.1021/ja048411t

Published on Web 08/10/2004
5
Unraveling the Photochemistry of Fe(CO) in Solution:
3
3 4
Observation of Fe(CO) and the Conversion between Fe(CO)
1
and Fe(CO)
4
(Solvent)
†
†
†
†,§
‡
Peter Portius, Jixin Yang, Xue-Zhong Sun, David C. Grills, Pavel Matousek,
‡
‡
,†
Anthony W. Parker, Michael Towrie, and Michael W. George*
Contribution from the School of Chemistry, UniVersity of Nottingham, UniVersity Park,
Nottingham, NG7 2RD, UK, and Central Laser Facility, CCLRC Rutherford Appleton
Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX, UK
Received March 19, 2004; E-mail: Mike.George@nottingham.ac.uk
Abstract: The photochemistry of Fe(CO)
scCH using time-resolved infrared spectroscopy (TRIR). Fe(CO)
as primary photoproducts within the first few picoseconds. Complex 3 is formed via a single-photon process.
5
(5) has been studied in heptane, supercritical (sc) Ar, scXe, and
3
3
4
4 3
( 4) and Fe(CO) (solvent) (3) are formed
3
1
In heptane, scCH
4
, and scXe, 4 decays to form 4‚L (L ) heptane, CH
9
4
, or Xe) as well as reacting with
1
3
1
5
to form Fe (CO)
2
. In heptane, 3 reacts with CO to form 4‚L. The conversion of 4 to 4‚L has been
7
-1
6
-1
monitored directly for the first time (L ) heptane, kobs ) 7.8((0.3) × 10 s ; scCH
4
, 5((1) × 10 s
;
7
-1
3
3
scXe, 2.1((0.1) × 10 s ). In scAr, 4 and 3 react with CO to form 5 and 4, respectively. We have
7
3
-1 -1
3
determined the rate constant (kCO ) 1.2 × 10 dm mol
s ) for the reaction of 4 with CO in scAr, and
this is very similar to the value obtained previously in the gas phase. Doping the scAr with either Xe or CH
4
3
1
1
3
resulted in 4 reacting with Xe or CH
order heptane > scXe > scCH . scAr, and pressure-dependent measurements in scAr and scCH
an influence of the solvent density on this ratio.
4
to form 4‚Xe or 4‚CH
4
. The relative yield, [ 4]:[3] decreases in the
4
4
indicate
Introduction
Scheme 1. Photochemistry of Fe(CO)5 in CH4 and Xe Matrices,
Adapted from Ref 5
The photochemistry of Fe(CO)5 has been studied intensely,
1
with applications ranging from photocatalysis and synthesis
2
to coherent control of gas-phase reactions. There is considerable
interest in the implication of crossing between different spin
3
surfaces in transition metal chemistry. Fe(CO)5 has been widely
investigated as a model system in organometallic photochem-
4
istry, using techniques such as matrix isolation, flash photolysis,
time-resolved infrared spectroscopy (TRIR), and electron dif-
fraction. However, there still remain intriguing questions re-
garding the mechanism of Fe(CO)5 photolysis.
In low temperature (12 K) matrices, Poliakoff and Turner
9
1
5
also generated C3V symmetric Fe(CO) (3). In Ne and Ar
showed that UV photolysis of Fe(CO)5 ( 5, Scheme 1)
3
3
3
6-8
matrices, near-IR irradiation or annealing the matrix could cause
the reaction of 4 with CO to regenerate Fe(CO)5 ( 5). In Xe
and CH4 matrices, 4 could be converted to Fe(CO)4‚L ( 4‚L,
L ) Xe or CH4) by near- or mid-infrared radiation. The gas-
phase photochemistry of Fe(CO)5 has been probed at room
generated C2V symmetric Fe(CO)4 ( 4). Prolonged irradiation
3
1
3
1
1
†
University of Nottingham.
CCLRC Rutherford Appleton Laboratory.
Current address: Chemistry Department, Brookhaven National Labora-
‡
§
tory, Upton, NY 11973-5000.
(
1) (a) Asinger, F.; Berg, O. Chem. Ber. 1955, 88, 445-451. Frankel, E. N.;
Emken, E. A.; Peters, H. M.; Davison, V. L.; Butterfield, R. O. J. Org.
Chem. 1964, 29, 3292-3297. (b) Velichko, F. K.; Vinogradova, L. V. IzV.
Akad. Nauk SSSR, Ser. Khim. 1970, 1628-1631. (c) Sigman, M. S.; Eaton,
B. E. J. Am. Chem. Soc. 1996, 118, 11783-11788.
temperature using TRIR, a combination of UV flash photolysis
10-13
and fast IR detection.
Fuss and co-workers investigated the
(7) Poliakoff, M.; Turner, J. J. J. Chem. Soc., Dalton Trans. 1974, 2276-
2285;
(
2) Bergt, M.; Brixner, T.; Kiefer, B.; Strehle, M.; Gerber, G. J. Phys. Chem.
A 1999, 103, 10381-10387.
(8) Barton, T. J.; Grinter, R.; Thomson, A. J.; Davies, B.; Poliakoff, M. J.
Chem. Soc., Chem. Commun. 1977, 841-842.
(
3) Harvey, J. N.; Aschi, M. Faraday Discuss. 2003, 124, 129-143; Harvey,
J. N.; Poli, R. J. Chem. Soc., Dalton Trans. 2003, 4100-4106.
4) Poliakoff, M.; Turner, J. J. Angew. Chem., Int. Ed. 2001, 40, 2809-2812.
(9) Poliakoff, M. J. Chem. Soc., Dalton Trans. 1974, 210-212.
(10) Ouderkirk, A. J.; Wermer, P.; Schultz, N. L.; Weitz, E. J. Am. Chem. Soc.
1983, 105, 3354-3355.
(
(
5) The numbering scheme starting with 3 (3, Fe(CO)
3
; 4, Fe(CO)
) was adapted from ref 5.
6) Poliakoff, M.; Turner, J. J. J. Chem. Soc., Dalton Trans. 1973, 1351-
357;
4
; 5,
Fe(CO)
5
; 8, Fe
2
(CO)
8
; 9, Fe
2
(CO)
9
(11) Seder, T. A.; Ouderkirk, A. J.; Weitz, E. J. Chem. Phys. 1986, 85, 1977-
(
1986.
1
(12) Ryther, R. J.; Weitz, E. J. Phys. Chem. 1991, 95, 9841-9852.
10.1021/ja048411t CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 10713-10720
9
10713

A R T I C L E S
Portius et al.
reactivity of Fe(CO)4 by pumping with femtosecond UV pulses
absorption bands are monitored using a cw IR source (M u¨ tek diode
laser) and a fast HgCdTe (MCT) detector.25 The change in IR ab-
14
and detecting with time-of-flight mass spectrometry. Ultrafast
time-resolved electron diffraction was used by Zewail and co-
sorbance at one particular IR frequency is measured following exci-
tation, and IR spectra are built up on a “point-by-point” basis by
repeating this measurement at different infrared frequencies. The
picosecond TRIR studies were performed on the PIRATE facility in
the Rutherford Appleton Laboratory, details of which are described
1
1
15
workers to determine the gas-phase structure of Fe(CO)4 ( 4),
which was found to be almost identical to the structure deter-
6
1
mined from early matrix work. In solution, 4‚L (L ) benzene,
cyclohexane) was characterized by Grevels and co-workers
26
elsewhere. Briefly, a part of the output from a 1 kHz, 800 nm, 150
16-18
18
using µs-TRIR.
They recently proposed, for the first time,
fs, 2 mJ Ti:Sapphire oscillator and regenerative amplifier was used to
pump a white light continuum seeded BBO OPA (â-barium borate
optical parametric amplifier). The signal and idler produced by this
that Fe(CO)3(solvent) (3‚L, L ) cyclohexane) is formed in
solution. Neither the spin state (singlet or triplet) of 3‚L nor
the number of solvent molecules or matrix atoms (one or two)
attached to it (L) is known. The formation of 3 is unusual, since
in the condensed phase (matrix and solution) it is normally
expected that metal carbonyls will only lose one CO group
following excitation; i.e., only Fe(CO)4 should be formed in
OPA were difference frequency mixed in a type I AgGaS
2
crystal to
generate tuneable mid-infrared pulses (ca. 150 cm fwhm, 1 µJ). The
67 nm pump pulses were the third harmonic of the 800 nm regenerative
-
1
2
amplifier output. Changes in infrared absorbance were recorded by
normalizing the outputs from a pair of 64-element MCT linear-infrared
array detectors on a shot-by-shot basis. Normally, several thousand shots
were averaged for one delay time. 300 lines/mm gratings were used in
19
solution. However, the absorption of one photon may provide
sufficient energy to eject more than one CO in a stepwise
-
1
spectrographs to achieve a spectral resolution of approximately 4 cm
process, and this is often observed in the gas phase (i.e.,
-1
5
in the 2200 cm region. The Fe(CO) concentration was adjusted to
generation of Fe(CO)n fragments, n ) 4-1).2
0-22
From the
-3
-10 mol dm-3. Different path lengths (ca. 0.5-1 mm for ultrafast
-4
10
above-mentioned experimental work and theoretical investiga-
tions, it was suggested that for the photoreaction of Fe(CO)5 a
experiments; 1-15 mm for the nanosecond point-by-point experiments)
were used to ensure that the UV absorbance at the particular laser
excitation wavelength and the maximum IR absorbance at the position
of the ν(CO) absorption bands were below unity. n-Heptane (Aldrich
HPLC grade, referred to as heptane throughout the text) was distilled
1
1
1
singlet pathway has to be given preference ( 5 f [ 5]* f [ 4]*
3
1
1
3
f 4) as opposed to a triplet pathway ( 5 f [ 5]* f [ 5]* f
3
4,14
1
1
4
).
In solution this means that hot 4 (i.e., [ 4]*) does not
from CaH
2
and degassed prior to use, and Ar (99.994%, BOC), CH
and CO (Aldrich), and Xe (99.995%,
4
form an intermediate complex with the solvent but rather forms
(99.995%, BOC), Fe(CO)
5
3
4
which undergoes a slow triplet-singlet conversion, generating
Spectragases) were used as supplied.
1
the solvent complex ( 4‚L, L ) solvent molecule). An ultrafast
TRIR experiment by Harris and co-workers revealed that, in
Results and Discussion
3
heptane solution, 4 is formed within 33 ps and is long-lived
up to 660 ps.23 Neither the formation of 14‚L nor 3‚L was
(a) Picosecond and Nanosecond TRIR Studies of Fe(CO)5
in Heptane. Selected spectra from picosecond TRIR experi-
ments with 267 nm excitation in heptane under CO (2 atm) are
shown in Figure 1b. The two parent Fe(CO)5 bands are bleached
instantaneously, and there are new broad transient peaks
apparent at lower wavenumbers after the first 5 ps. As is often
observed in picosecond TRIR experiments performed upon
metal carbonyls, these bands narrow and blue-shift, producing
three clearly resolved transient ν(CO) bands at 1988, 1967, and
1
observed. In the microsecond TRIR experiments, 4‚L and 3‚L
were observed within the rise time of the M u¨ lheim TRIR
3
18
apparatus (ca. 0.4 µs) but no 4 was found. Consequently, a
3
1
comparably slow reaction of 4 with solvent to form 4‚L has
to be assumed, but this has never been directly monitored.
Some intriguing questions remain regarding the photochem-
istry of Fe(CO)5 in solution. What is the time scale for the
3
1
conversion of Fe(CO)4 to Fe(CO)4(solvent), and how quickly
is Fe(CO)3(solvent) formed? In this paper we report the results
of our investigation into the time scale of the 4 to 4‚L
conversion and the rate of formation of 3‚L in conventional
and supercritical fluids using fast and ultrafast TRIR.
-
1
1
926 cm . The shift of the ν(CO) bands over the first 50 ps is
3
1
consistent with a vibrational cooling of the newly formed
species. The cooling time is similar for all three bands and is
2
7
estimated to be ca. 10 ps. The ν(CO) bands at 1988 and 1967
-
1
3
cm can readily be assigned to 4 by comparison with the
Experimental Section
9
23
previous matrix isolation and picosecond TRIR results,
-
1
9,18
The Nottingham ns-TRIR apparatus has been described elsewhere.24
In these experiments, a pulsed Nd:YAG laser (Quanta-Ray GCR-12S;
whereas the band at 1926 cm is due to 3‚heptane.
The
ultrafast formation of 3‚heptane is unexpected, since in solution
the excess energy which remains in a molecule following the
ejection of one CO group is normally lost to the solvent rather
than leading to loss of a second ligand. We have examined this
question further by performing power-dependent TRIR mea-
surements. We found that there was a linear dependence between
2
66 nm) initiates the photochemical reactions and the transient IR
(
13) Ryther, R. J.; Weitz, E. J. Phys. Chem. 1992, 96, 2561-2567.
(
14) Trushin, S. A.; Fuss, W.; Kompa, K. L.; Schmid, W. E. J. Phys. Chem. A
2
000, 104, 1997-2006.
(15) Ihee, H.; Cao, J.; Zewail, A. H. Angew. Chem., Int. Ed. 2001, 40, 1532-
1
536.
3
(
16) Church, S. P.; Grevels, F. W.; Hermann, H.; Kelly, J. M.; Klotzb u¨ cher,
W. E.; Schaffner, K. J. Chem. Soc., Chem. Commun. 1985, 594-596.
17) Grevels, F. W. NATO ASI Ser., Ser. C 1992, 376, 141-171.
18) Bachler, V.; Grevels, F.-W.; Kerpen, K.; Olbrich, G.; Schaffner, K.
Organometallics 2003, 22, 1696-1711.
the yield of 3‚heptane and 4 and the laser power. This strongly
(
(
(25) Both the rise time of the detecting system (ca. 7-8 ns) and the bandwidth
of the Nd:YAG UV laser (ca. 5-6 ns at 266 nm) contribute to the TRIR
traces. Deconvolution was used to obtain the sub-50 ns kinetics. Decon-
volution required both the response function of the detector, which was
obtained using a femtosecond IR pulse, and the instrument response,
obtained by measuring the light emitted from a Ge wafer.
(26) Towrie, M.; Grills, D. C.; Dyer, J.; Weinstein, J. A.; Matousek, P.; Barton,
R.; Bailey, P. D.; Subramanium, N.; Kwok, W.; Ma, C.; Phillips, D.; Parker
A. W.; George, M. W. Appl. Spectrosc. 2003, 57, 367-380.
(27) Band areas and line widths at half-height were estimated by multi-Lorentzian
curve fitting of the spectral points. The change in line width was then fitted
to a first-order exponential decay function.
(
19) For a detailed review see Leadbeater, N. Coord. Chem. ReV. 1999, 188,
3
5-70 and ref 5.
(
20) Weitz, E. J. Phys. Chem. 1987, 91, 3945-3953.
(
21) Rayner, D. M.; Ishikawa, Y.; Brown, C. E.; Hackett, P. A. J. Chem. Phys.
1
991, 94, 5471-5480.
(
(
(
22) Ishikawa, Y.; Brown, C. E.; Hackett, P. A.; Rayner, D. M. J. Phys. Chem.
1
990, 94, 2404-2413.
23) Snee, P. T.; Payne, C. K.; Kotz, K. T.; Yang, H.; Harris, C. B. J. Am.
Chem. Soc. 2001, 123, 2255-2264.
24) George, M. W.; Poliakoff, M.; Turner, J. J. Analyst 1994, 119, 551-560.
10714 J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004

Photochemistry of Fe(CO)
5
in Solution
A R T I C L E S
The lifetime of 3‚heptane was found to be dependent on both
the CO and starting Fe(CO)5 concentrations. Thus, 3‚heptane
1
reacts with CO to generate 4‚heptane and with unphotolized
Fe(CO)5, forming 8. These results are complementary to the
previously published picosecond and microsecond TRIR data,
since neither of these experiments were performed on a time
3
1
scale that allowed the conversion of 4 to 4‚heptane to be
1
monitored. The formation of 4‚heptane from the reaction of 3
with CO occurs on a slower time scale than the conversion of
3
1
4
to 4‚heptane and contributes to ca. 10% of the total
1
formation of 4‚heptane in the presence of CO. In the absence
of CO on the nanosecond time scale, 4‚heptane was observed
to form solely from 4. However, a close inspection of Figure
1
1
3
-1
reveals that there is an additional, small peak at ca. 1950 cm ,
1
and this possibly indicates that a small fraction of 4‚heptane
is formed as an initial photoproduct. In other words, a small
1
amount of the originally formed 4 could be directly converting
1
3
into 4‚heptane rather than crossing over to form 4. However,
the possibility that this absorbance is due to some other minor
photoproduct cannot be ruled out.
3
1
A conversion of 4 into 4‚heptane is consistent with the
early matrix isolation work of Poliakoff and Turner, in which
6
they demonstrated that coordination of Xe or CH4 to the
Fe(CO)4 moiety could be observed. However, no such coordina-
tion occurred following photolysis of Fe(CO)5 in Ar matrices.
We have recently used a combination of TRIR and supercritical
fluids to characterize organometallic noble gas complexes in
Figure 1. (a) FTIR spectrum and (b) TRIR spectra obtained between 5
and 500 ps following irradiation (267 nm) of a solution of Fe(CO)5 in
heptane saturated under 2 atm of CO.
2
9
solution at room temperature. This approach gives a unique
supports a single-photon process where Fe(CO)5 loses two CO
ligands to generate 3‚heptane within the first 5 ps. An even
weaker A1 ν(CO) mode for 3‚heptane is expected at around
1
opportunity to explore whether 4‚L can be formed in more
weakly coordinating solvents such as Xe or Ar. In particular,
the use of supercritical fluids may allow us to probe the role of
vibrational relaxation on the formation of Fe(CO)3 (3) from
initially formed Fe(CO)4 (4), since supercritical solvents have
a significant effect on vibrational relaxation rates for metal
carbonyls. For example, the vibrational relaxation lifetime of
the ν ) 1f0 ν(CO) transition in W(CO)6 varies significantly
-
1 9
2
036 cm , but it is hardly discernible in our experiments (band
3
positions are summarized in Table 1). Both 4 and 3‚heptane
are stable up to 1 ns. From Figure 2, it can be seen that the
TRIR spectrum obtained after 25 ns is similar to the 500 ps
spectrum in that the vibrational bands due to 4 (1989 and 1969
cm ) and 3‚heptane (1926 cm ) are clearly present. However,
an additional band at ca. 1950 cm becomes apparent. As the
bands at 1989, 1969, and 1926 cm
absorption bands appear (see the 150 ns spectum in Figure 2).
In light of Grevels’ microsecond TRIR work,
CO) peaks at 2083 cm (A1), 1989 cm (B1), 1973 cm
3
-1
-1
30
as the solvent is changed from n-hexane (140((20) ps) to scAr
-
1
3
1
(1200 ps).
-
1
decay, many new
(b) Picosecond and Nanosecond TRIR Studies of Fe(CO)5
in Supercritical Xe (scXe). Selected TRIR spectra obtained
following 267 nm photolysis of Fe(CO)5 in scXe in the presence
of CO are shown in Figure 4. The two parent ν(CO) bands are
slightly blue-shifted relative to their positions in heptane. There
are three bands clearly visible after the initially formed pho-
toproducts have undergone vibrational cooling. The 1993 and
16,18
the new ν-
-
1
-1
-1
(
(
-1
A1), and 1953 cm (B2) are unambiguously assigned to the
singlet complex, 4‚heptane.
1
The formation of Fe2(CO)8 (8) and Fe2(CO)9 (9) is responsible
for the other new absorption bands and is consistent with the
-
1
3
1
972 cm bands can be readily assigned to 4 in scXe, and
3
previously published microsecond TRIR studies. 4 decays at
-
1
the vibrational band at 1931 cm is tentatively ascribed to
‚Xe. The vibrational relaxation rates for 4 and 3‚L in scXe
1
the same rate as 4‚heptane is formed (Figure 3, Table 2). The
3
3
3
decay of 4 and 3‚heptane is complex, since both of these
are slightly longer (ca. 25 ps) than those obtained from the
2
8
species can react with CO in addition to forming 9 and 8.
3
aforementioned heptane experiments. 4 and 3‚Xe in scXe are
3
However, the decay rate of 4 is an order of magnitude faster
than that of 3‚heptane. We have repeated the nanosecond
TRIR experiment in the presence of argon (2 atm) instead of
carbon monoxide, and it was found that the rate for the
conversion of 4 (kobs ) 8.1((1.2) × 10 s ) into 4‚heptane
kobs ) 7.5((1.0) × 10 s ) was unchanged. Changing the
Fe(CO)5 concentration (from 6.6 × 10 to 5.5 × 10 mol
dm ) did not affect the rate of the 4 to 4‚heptane conversion.
stable up to 1 ns. It is interesting to note that there is relatively
3
2
3
more 3‚L produced in scXe. The ratio of [ 4]:[3‚L] is ca. 5:1
in scXe compared to 17:1 in heptane. The TRIR spectrum
obtained 25 ns after irradiation (Figure 5) is analogous to the
3
7 -1
1
-
1
spectrum at 500 ps. The bands at 1992 and 1973 cm show
7
-1
(
-
4
-3
(
29) Grills, D. C.; George, M. W. AdV. Inorg. Chem. 2001, 52, 113-150.
-3
3
1
(30) Arrivo, S. M.; Dougherty, T. P.; Grubbs, W. T.; Heilweil, E. J. Chem.
Phys. Lett. 1995, 235, 247-254.
(31) Myers, D. J.; Shigeiwa, M.; Fayer, M. D.; Cherayil, B. J. J. Phys. Chem.
(
28) The bands from dimers in the experiment are of very low intensity compared
B. 2000, 104, 2402-2414.
1
to the band from 4‚heptane. The intensities of dimer bands can be
(32) These branching ratios are calculated by the sum of the areas of the two
3
enhanced by increased Fe(CO)
5
concentration.
most intense bands of 4 divided by the area of the 3‚L band.
J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004 10715

A R T I C L E S
Portius et al.
Table 1. ν(C-O) IR Band Positions (cm-1) of Iron Carbonyl Species
ν(CO) band
ν(CO) band
complex
conditionsg
positions (cm-1)
complex
conditionsg
scXe
heptane
cyclohexane
vapor
Ar matrix
scAr (Xe doped)
scXe
positions (cm-1)
1
2032, 2018a
1931c
Fe(CO)5 ( 5)
vapor
Fe(CO)3(solv) (3‚L)
b
c
Ar matrix
Xe matrix
CH4 matrix
scAr
scXe
heptane
2025, 2010, 2005, 1998
1926
1926
2023, 2000, 1996, 1994b
2025, 2003sh, 2001, 1996
f
b
a
Fe2(CO)8 (8)
2046, 2016, 1982
2042, 2009, 1987vw, 1979w
2045c
c
2030, 2009
c
2025, 2004
c
c
2023, 200
2000, 1984
2043
3
3
a
2035c
Fe(CO)4 ( 4)
vapor
heptane
d
f
Ar matrix
Xe matrix
CH4 matrix
scAr
scXe
heptane
Xe matrix
CH4 matrix
cyclohexane
scAr (Xe doped)
scXe
1996, 1992, 1973
cyclohexane
Ar matrix
scAr (Xe doped)
scXe
heptane
cyclohexane
vapor
Ar matrix
Xe matrix
CH4 matrix
scAr
2036, 2004
1989, 1983, 1967b
Fe2(CO)9 (9)
Fe(CO)3 (3)
2066, 2038, 2018w, 1855w
b
2067c
1992, 1985, 1967
c
c
1999, 1980
1993, 1972
1988, 1967
2061
c
2057c
c
f
2056, ∼2020, 1826
1
1
2085w, 1990, 1984, 1952b
1950a
Fe(CO)4(solv) ( 4‚L)
2090w, 1998, 1984, 1954b
2084w, 1986, 1970, 1950
-, 1998, 1980, 1966c
2042w, 1936b
2036w, 1927b
b
2040w, 1930
2085w, 1990, 1974, 1961c
2083w, 1989, 1973, 1953c
1940c
heptane
a
Reference 12. b Reference 9. c This work. d Reference 38. e Reference 16. f Reference 18. g Matrix 12 K, fluids 298 K.
Figure 3. TRIR traces obtained following irradiation (266 nm) of a solution
of Fe(CO)5 in heptane saturated under 2 atm CO showing (a) the formation
1
-1
3
of 4‚heptane at 1953 cm and (b) the decay of 4 overlapped with the
1
-1
formation of 4‚heptane at 1989 cm
.
3
3
Table 2. Rates of Decay of Fe(CO)4 ( 4) and Fe(CO)3(solvent)
1
1
(
3‚L) and Formation of Fe(CO)4(solvent) ( 4‚L) in the Solvents,
scAr, scXe, scCH4, or Heptane (All Data at 298 K)
a
process
solvent
k
obs (s-1)
decay of 34
scAr (CO doped)
scXe (CO doped)
scCH4 (CO doped)
heptane (CO saturated)
heptane (Ar saturated)
scAr (CO doped)
5.3 × 10
2.2 × 10
6
7
Figure 2. TRIR spectra obtained 25 ns and 150 ns following excitation
6
7
7
6.2((0.6) × 10
8.1((0.7) × 10
8.1((1.2) × 10
(266 nm) of a solution of Fe(CO)5 in heptane saturated under 2 atm of CO.
the presence of 4, and the band at 1931 cm-1 is assigned to
3
8
decay of 3
decay of 3‚L
7.2 × 10
7
3
‚Xe. At 200 ns, these bands have decayed and several new
scXe (CO doped)
5.3 × 10
6
heptane (CO saturated)
heptane (Ar saturated)
scXe (CO doped)
8.5((1.0) × 10
ν(CO) bands are formed. Three of these bands (1990, 1974,
6
1.99((0.02) × 10
-
1
1
and 1961 cm ) can be assigned to 4‚Xe by comparison with
1
7
formation of 4‚L
2.0 × 10
matrix isolation studies7 and the analogous experiments carried
,9
4.4((0.4) × 106
scCH (CO doped)
4
3
heptane (CO saturated)
heptane (Ar saturated)
7.6((0.1) × 107
out in heptane, which were described above. 4 decays at the
7
1
7
-1
7.5((1.0) × 10
same rate as 4‚Xe forms (kobs ) 2.0((0.2) × 10 s ). By direct
comparison of the nanosecond traces in the region 2030-2090
a
The heptane solutions have been saturated at 30 psi.
-1
1
cm , a fourth, very weak ν(CO) absorption of 4‚Xe was found
by a single trace experiment at 2085 cm , which is, however,
-
1
shows that, analogous to the experiments in heptane, there is a
small peak (at ca. 1959 cm ) apparent at early time, possibly
suggesting a rapid formation of a small amount of 4‚Xe in the
-
1
to small to be visible in the transient spectrum. Bands due to 8
3
1
and 9 in scXe were also observed to form as 4 and 3‚Xe decay
(Table 1). Close examination of the picosecond TRIR spectra
first few picoseconds following excitation.
10716 J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004

Photochemistry of Fe(CO)
5
in Solution
A R T I C L E S
Figure 4. (a) FTIR spectrum and (b) TRIR spectra obtained between 10
and 500 ps following excitation (267 nm) of Fe(CO)5 in scXe (1500 psi)
doped with CO (30 psi).
Figure 6. (a) FTIR spectrum and (b) selected TRIR spectra obtained
between 10 and 500 ps after irradiation (267 nm) of Fe(CO)5 in scAr (4500
psi) doped with CO (30 psi).
vibrational cooling rate is slower in scXe. Also, the lifetime of
3
4
is longer in scXe. It is tempting to conclude that the more
strongly coordinating heptane ligand is driving a faster conver-
3
1
sion from 4 to 4‚L. We have investigated this further by
performing TRIR experiments in scAr.
(c) Picosecond and Nanosecond TRIR Studies of Fe(CO)5
in Supercritical Ar (scAr). Selected spectra from the picosec-
ond TRIR investigation in scAr solution in the presence of CO
are shown in Figure 6. In scAr, the rate of vibrational cooling
3
is slower and bands which are readily assigned to 4 and
Fe(CO)3 (3) sharpen at a much longer time delay relative to
the experiments in scXe and heptane. Again the observed
3
vibrational cooling times of 4 and 3 are very similar (ca. 60
34
ps). The yield of 3 is higher in scAr, but a precise ratio of
3
[
4]:[3] is difficult to estimate, since 3 is not stable and starts
to decay before the vibrational relaxation is complete, see Figure
. A value of 3:1 is obtained from the early time spectra. The
lifetime of 3 (τ ) 0.3 ns) is much shorter than that in scXe (τ
7
3
4
)
18 ns) or heptane (τ ) 117 ns). A possible explanation is
that the strength of the Fe-solvent bond in the solvent
coordinated species (3‚L) increases in the order Ar , Xe <
33) Under CO, 34 decays rapidly to form 14‚heptane. The formation of 4‚
1
(
heptane from 3 was observed in a CO-saturated heptane solution of 5
1
1
resulting in a biphasic rise of 4‚heptane. Rapid formation of 4‚heptane
was followed by a further slow growth which matched the decay of 3.
Figure 5. TRIR spectra obtained at (a) 25 ns and (b) 200 ns after excitation
1
Under Ar, the formation of 4‚heptane only occurred from the decay of
(
266 nm) of an Fe(CO)5 solution in scXe (1500 psi) doped with CO (30
3
4
. A reaction of 4 with CO cannot be assumed due to the fact that 5 does
3
1
1
psi). TRIR traces showing (c) the formation of Fe(CO)4Xe ( 4‚Xe) at 1961
not recover after the vibrational cooling has ceased and while the triplet-
singlet conversion takes place. The formation of 8 starts within the first 40
ns, whereas the formation of 9 takes place within the first 1 µs. Compared
-
1
3
3
cm and (d) the decay of Fe(CO)4 ( 4) overlapped with the growth of
1
-1
4
‚Xe at 1990 cm .
1
with 4‚heptane, only small fractions of 8 and 9 are formed while the
triplet singlet conversion takes place.
(
34) In this work, the spectra 100 ps after the laser flash are used to estimate
It is clear from these results that the basic photochemistry is
3
.
3
the intensity ratio, [ 4]: [3 L] or [ 4]: [3], respectively. At this time, we
have assumed that there is a minimum contribution from vibrationally
excited states. However, it must be noted that 3 has started to decay in
scAr. The ratios obtained ranged from 3:1 in scAr to 5:1 in scXe to 17:1
in heptane.
3
similar in scXe and heptane in that both 4 and 3‚L are formed
within 5 ps and 4 decays to form 4‚L, whereas 3‚L decays to
form 4‚L and 8. The yield of 3‚L is increased and the
3
1
1
33
J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004 10717

A R T I C L E S
Portius et al.
Figure 7. Change in intensity of the IR band centered at 1940 cm-1 between
0
.1 and 2.0 ns after 267 nm excitation of Fe(CO)5 in scAr (4500 psi) doped
with CO (30 psi). Each data point represents the integral under the
Lorentzian curve fitted transient peak. The curve represents a fit according
to a first-order rate law for the decay of Fe(CO)3 (3).
heptane. The long lifetime of 3 in heptane and scXe casts little
doubt on the assignment of the solvated species, 3‚heptane and
3
‚Xe. However, the coordination of Ar to 3 is more tentative.
Figure 8. TRIR spectra obtained at 50, 100, 150, and 200 ns after irradiation
(266 nm) of Fe(CO)5 in scAr (4500 psi) doped with CO (30 psi).
We have repeated the experiment at a much higher CO
concentration (200 psi). This causes 3 to decay much more
rapidly (τ ) 50 ps). Using five points at different CO partial
pressures (30-200 psi), we have obtained an estimate for the
10
3
-1
bimolecular rate constant (kCO ) 3.2((0.9) × 10 dm mol
-
1
s ) for the reaction of 3 with CO in scAr. This value is an
order of magnitude lower than the expected diffusion controlled
1
1
3
-1 -1
rate (kdif ) 2.0 × 10 dm mol s ), which suggests that 3 is
solvent coordinated even in scAr solution. In scAr, 3 reacts with
CO to form 4. Nanosecond TRIR spectra of Fe(CO)5 in scAr
3
following photolysis in the presence of CO are shown in Figure
3
8
. 4 is much longer lived in scAr (τ ≈ 1 µs). There was no
1
evidence for the formation of the singlet solvated species 4‚
3
35 3
6
-1
Ar from 4. 4 decays (kobs ) 5.2((0.5) × 10 s ) at the
1
6
-1
same rate as 5 is reformed (kobs ) 4.5((0.5) × 10 s ) (see
Figure 9), and the formation of 8 and 9 was not detected in
3
3
3
Figure 9. TRIR traces corresponding to the Fe(CO)4 ( 4) decay at 1980
scAr. The decay rate of 4 in scAr depends linearly on the CO
-
1
1
-1
cm (a) and the recovery of Fe(CO)5 ( 5) at 1998 cm (b) obtained from
concentration (see inset in Figure 9), and this allows an
the experimental results shown in Figure 8. (c) The CO concentration-
dependent decay rate of 4 is shown.
3
3
estimation for the reaction rate of 4 with CO in scAr (kCO )
7
3
-1 -1
1
.2 × 10 dm mol s ). This value is similar to that obtained
7 3 -1 -1
by Weitz and co-workers (3.1 × 10 dm mol s ) for the
50 ns and 500 ns after excitation of Fe(CO)5 in scAr doped
with Xe, and a small amount of CO are shown in Figure 10.
The parent bleaches are at very similar positions to those
observed in pure scAr. The TRIR spectrum obtained after 50
3
reaction between 4 and CO in a gas phase study using Ar buffer
gas.11
Although these studies show that Fe(CO)5 has different
photochemistries in scAr and scXe, it does not prove conclu-
sively that the conversion is solely due to coordination of Xe
to the metal center. Therefore we have repeated our TRIR
experiments in scAr doped with Xe to probe this reaction further.
-
1
ns shows that new bands are formed at 1998 and 1980 cm ,
3
which are assigned to 4 in the Xe-Ar mixture. These bands
1
-1
decay to form 4‚Xe (1998, 1980, and 1966 cm ), by
1
comparison with the experiments carried out in pure scXe. 4‚
Xe grows at the same rate as 4 decays. Bands due to 8 and 9
3
(d) Picosecond and Nanosecond TRIR Studies of Fe(CO)5
in Supercritical Ar Doped with Xe. TRIR spectra obtained
are also detected in the Xe-Ar mixture (see Table 1), but the
band of 9 is too small to become visible in the spectrum in
(
35) Close inspection of the picosecond TRIR spectra reveals an additional peak
3
Figure 10. In the presence of Xe and CO, 4 can react to form
at 1969 cm- which decays rapidly (τ ) 730((60) ps). It is tempting to
1
1
1
1
assign this to rapidly decaying 4‚Ar. This would be consistent with the
either 5 or 4‚Xe. We have repeated this experiment, keeping
1
1
observation that in other solvents a small amount of 4‚heptane or 4‚Xe
appears to be produced in the first few picoseconds. However, such an
assignment must remain as very tentative.
the Xe concentration constant (450 psi) at two CO pressures
3
(10 and 30 psi). We find that the rate of conversion between 4
10718 J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004

Photochemistry of Fe(CO)
5
in Solution
A R T I C L E S
Figure 12. TRIR spectra obtained 150 ps after 267 nm excitation of
Figure 10. TRIR spectra obtained at 50 ns and 500 ns after excitation
Fe(CO) in scAr, doped with 30 psi of CO at different overall pressures.
5
(266 nm) of Fe(CO)5 in scAr (4500 psi) doped with Xe (400 psi) and CO
30 psi).
The fit curves result from a multicomponent analysis of the six spectra at
1550 (a), 2140 (b), 2840 (c), 3450 (d), 4330 (e), and 5550 psi (f). The
spectra are normalized with the sum of the areas of the parent bleaches.
(
1
Fe(CO)4(CH4) in the mixed CH4-Ar solvent, thus suggesting
that a “heavy atom” effect is not being observed.36
(e) Exploring the Solvent-Dependent Yield of Fe(CO)3 (3).
The yield of Fe(CO)3 (3 in scAr and 3‚L in scXe, scCH4, and
3
heptane) relative to 4 was found to be solvent dependent. The
3
ratio [ 4]:[3] increases in the order scAr (3.5:1 at 1500 psi) <
scCH4 (4:1 at 3500 psi) < scXe (5:1 at 1500 psi) < heptane
(17:1). One possible explanation is that 3 is formed from a
vibrational excited state of Fe(CO)4, i.e., Fe(CO)3 is only
1
3
generated when Fe(CO)4 is vibrationally excited ([ 4]* or [ 4]*),
and the yield of 3 depends on the vibrational relaxation rate.
3
Another contribution to the relative yield of 3 and 4 might be
3
3
the solvent cage affecting the ratio of [ 4]:[3]. 4 would not be
1
expected to react with CO in the cage to form 5. It is possible
that 3 could react with CO to form 4 and that the nature of the
3
1
Figure 11. TRIR traces showing the formation and decay of Fe(CO)4Xe
(
solvent cage would then have a direct effect on the out-of-cage
ratio of [ 4]:[3].
1
4‚Xe), monitored at 1966 cm-1 following irradiation (266 nm) of
Fe(CO)5 in scAr, doped with Xe (450 psi) and (a) 10 psi of CO or (b) 30
psi of CO. The overall pressure was 5000 psi.
3
In supercritical fluids, the solvent concentration can be
changed. A series of experiments at different pressures have
been carried out to investigate the effects of Ar and CO
concentration on the reaction mechanism of Fe(CO)5 photolysis.
We have performed picosecond TRIR experiments starting with
a partial pressure of ca. 30 psi CO. As the total pressure was
1
7
-1
and 4‚Xe is identical (kobs ) 2.0((0.2) 10 s ) at both CO
concentrations. However, the lifetime of 4‚Xe increases as the
CO concentration is lowered, Figure 11. Although the Xe doped
experiments strongly suggest that it is the coordinating ability
of solvents, which is responsible for the conversion of 4 to
‚Xe, it does not exclude the possibility that the additional Xe
1
3
3
dropped from 4500 to 1500 psi, the [ 4]:[3] ratio decreased from
1
4
7:1 to 3.5:1. At the same time, 3 became longer lived (τ changes
could be acting as a “heavy atom”, facilitating the triplet-singlet
conversion. The use of supercritical fluids again gives us an
opportunity to test this theory by doping scAr with CH4 rather
than Xe. The photochemistry of Fe(CO)5 in pure scCH4 was
very similar to that observed in heptane and scXe. TRIR
experiments in scAr (4500 psi) doped with CH4 (400 psi) show
from 0.5 to 1.2 ns), which is consistent with the lower CO
concentration. In a separate experiment at a fixed CO pressure
of 30 psi, the total pressure was varied from 1600 to 5600 psi
4
(36) In the scAr doped with CH experiment, we have observed a band growth
1
at 1964 and 2086 cm- , which can be assigned to 4‚CH
1
by comparison
4
with the previous matrix IR results. These bands are not observed in the
experiment in pure scAr.
3
6
-1
that Fe(CO)4 decays (kobs ) 4.6((0.5) × 10 s ) to form
J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004 10719

A R T I C L E S
Portius et al.
Scheme 2. UV Photolysis of Fe(CO)5 in Solution (Heptane,
scCH4, scXe, and scAr)
Conclusions
a
We have provided direct evidence for the conversion of
3
photogenerated triplet Fe(CO)4 ( 4) to singlet Fe(CO)4(solvent)
1
(
4‚L) (L ) CH4, Xe and heptane) in conventional and
supercritical solution (Scheme 2). The existence of the solvent
1
complexes of Fe(CO)3 (3‚L) and Fe(CO)4 ( 4‚L) was demon-
strated in solution for all solvents except scAr. In scAr, we have
1
no evidence for the formation of 4‚Ar.
3
1
The conversion of 4 to 4‚Ar was not observed in scAr, but
1
there is tentative evidence for the formation of 4‚Ar in the
first few picoseconds. In coordinating solvents, 3‚L reacts with
1
CO to form 4‚L, whereas, in scAr, 3 reacts with CO to form
3
3
4
. The relative yield of [ 4]:[3] is solvent and pressure
a
Reactions which have major contribution are indicated by thick arrows.
dependent. The next challenge is to elucidate fully the formation
of Fe(CO)4 and Fe(CO)3(solvent) in solution.
The dashed line is a tentative suggestion based upon a small band observed
at early time.
3
Acknowledgment. We thank Professors F.-W. Grevels, M.
Poliakoff, and J. J. Turner for helpful discussions. We thank
the EPSRC (GR/M40486) for support. M.W.G. is grateful for
the support of the Sir Edward Frankland Fellowship of the Royal
Society of Chemistry, and P.P., for the Feodor Lynen Fellowship
of the Alexander-von-Humboldt Foundation.
by adding Ar to the mixture. Figure 12 shows TRIR spectra
obtained at a range of pressures. The ratio changes significantly,
and this is consistent with the solvent cage affecting the out-
of-cage yield of 3. However, it is also possible that the change
in pressure is significantly affecting the vibrational relaxation
rates. Unfortunately, analysis of the lower pressure data proved
problematic, since both band position and width changed with
pressure and accurate vibrational relaxation rates were not
obtained.37 We have repeated such experiments in scCH4 and
JA048411T
(
37) The problem arises from the restricted number of spectral points compared
to the number of bands in the spectral region of interest. This is combined
with the especially broad and overlapped bands that are obtained at low
pressure, thus preventing a quantitative analysis.
3
also found that the [ 4]:[3] ratio had changed from 3:1 (1300
psi) to 6:1 (4300 psi) even though the observed vibrational
relaxation is faster (ca. 10 ps).
(38) Poliakoff, M. Chem. Phys. Lett. 1981, 78, 1-3.
10720 J. AM. CHEM. SOC.
9
VOL. 126, NO. 34, 2004