J . Phys. Chem. 1992, 96, 2561-2567 way to generate high concentrations of vibrationally excited IF, IC1, and I2 molecules in a flow reactor. The secondary interaction of NF(a) with the vibrationally excited I2 and IC1 molecules probably gives I atoms. The formal NF(a) quenching kinetics of these iodine-containing molecules are intriguing because pseudo-first-order kinetics are observed even though [NF(a)] > [Q],because the 12, IF, and IC1 molecules are recycled. Qual:itative measurements gave the quenching rate constant for NF(a)
2561
by I atoms as -1.8 X lo-" cm3 s-I; this excitation-transfer constant seems to be 2-3 times smaller than for 02(a) 1.31 At least, CI2 and I F do not react rapidly with NF(X).
+
Acknowledgment. This work was supported by the Air Force Office Of Scientific Research (Grant 88-0279)* (31) Young, A. T.; Houston, P. L. J . Chem. Phys. 1983, 78, 2319.
Diode Laser Probes of the Product Distribution of Coordinatively Unsaturated Iron Carbonyls Produced following Excimer Laser Photolysis of Fe(CO), in the Gas Phase Robert J. Ryther and Eric Weitz* Department of Chemistry, Northwestern University, Evanston, Illinois 60208-31 13 (Received: September 9, 1991; In Final Form: December 2, 1991)
The distributions of photoproductscreated by single photon photolysis of Fe(CO)5at 193,248, and 35 1 nm have been determined by using a tunable infrared diode laser to monitor the concentration of Fe(CO)5 consumed in reactions with each photoproduct. The relative concentrations of the iron carbonyl photofragments that form are as follows: on 193-nm photolysis, 90 i 5% Fe(C0)2and 10 f 5% Fe(C0); on 248-nm photolysis, 64 f 7% Fe(CO)3 and 36 f 7% Fe(C0)2; and on 351-nm photolysis, 39 5% Fe(C0)4 and 61 i 5% Fe(C0)3. A pressure dependence of the 248-nm product yield is observed, and the reported branching ratio is for 3 Torr of total pressure. Peak absorption coefficients of these species are reported. In addition to these products, which are produced in their ground electronic states, species best assigned as excited electronic states of Fe(C0)4 and Fe(CO), are produced following 35 1- and 248-nm photolysis, respectively. These product distributions are compared to those reported in other studies, and evidence is presented that for at least some of the wavelengths employed in this study the dissociation of Fe(CO)5 involves multiple potential energy surfaces.
*
I. Introduction A number of recent studies have focused on the photodissociation dynamics of metal carbonyls in the gas phase. Of particular interest is the observation that species with multiple site of coordinative unsaturation can be produced by absorption of a single photon.l-I8 To date, iron pentacarbonyl is the best studied of such systems. Yardley et al. photolyzed Fe(CO)5 in the ultraviolet and used PF3to chemically trap the unsaturated photoproducts with product yields determined by G C analysis.] Vernon and co-workers used a crossed laser-molecular beam system to study (1) Yardley, J. T.; Gitlin B.; Nathanson, G.; Rosan, A. M. J . Chem. Phys. 1981, 74, 370. (2) Ray, U.; Brandow, S. L.;Bandukwalla, G.;Venkataraman, B. K.; Zhang, Z.; Vernon, M. J . Chem. Phys. 1988,89, 4092. (3) Venkataraman, B. K.; Bandukwalla, G.;Zhang, Z.; Vernon, M. J . Chem. Phys. 1989, 90, 5510. (4) Waller, I. M.; Hepburn, J. W. J . Chem. Phys. 1988, 88, 6658. ( 5 ) Waller, 1. M.; Davis, H.F.; Hepburn, J. W. J. Phys. Chem. 1987,91, 506. (6) Seder, T. S.; Ouderkirk, A. J.; Weitz, E. J . Chem. Phys. 1986, 85, 1977. (7) Ryther, R. J.; Weitz, E. J . Phys. Chem. 1991, 95, 9841. (8) Ryther, R. J. Ph.D. Thesis, Northwestern University, 1991. (9) Bogdan, P. L.; Weitz, E. J. Am. Chem. SOC.1989, 111, 3163. (IO) Bogdan, P. L.; Weitz, E. J . Am. Chem. SOC.1990, 112, 639. (1 1) Seder, T. A.; Church, S. P.;Weitz, E. J . Am. Chem. SOC.1986,108, 4721
(12) Seder, T.; Ouderkirk, A.; Church, S.; Weitz, E. ACS Symp. Ser. 1987, No. 333, 8 1. (13) Seder, T. A.; Church, S. P.: Weitz, E. J . Am. Chem. SOC.1986, 108, 1084. (14) Ishikawa, Y.; Hackett, P. H.; Rayner, D. M. J. Phys. Chem. 1988, 92, 3863. ( 1 5) Fletcher, T. R.; Rosenfeld, R. N. J . Am. Chem. Soc. 1988,110, 2097. (16) Rosenfeld, R. N.; Ganske, J. A. J . Phys. Chem. 1989, 93, 1959. (17) Rayner, D. M.; Ishikawa, Y.; Brown, C. E.; Hackett, P. A. J . Chem. Phys. 1991, 94, 5471. (18) Ishikawa, Y.; Brown, C. E.; Hackett, P. A,; Rayner, D. M. J. Phys. Chem. 1990, 94, 2404.
0022-36S4/92/2096-2561$03.00/0
the photodissociation of Fe(C0)5,2,3detecting the iron carbonyl photofragments using time-of-flight mass spectrometry. Hepbum and co-workers also studied the photolysis of Fe(CO)S in a molecular beam, employing vacuum ultraviolet laser-induced fluorescence to detect vibrational, rotational, and translational distributions of the photoejected CO. They then used the energy distribution of the ejected CO, along with literature values for the iron420 bond energies for the different iron carbonyl species to calculate a product distribution for each UV photolysis ~avelength.~,~ Although there are some significant differences in product distributions reported in these studies, they all concur that the dissociation mechanism involves stepwise loss of CO ligands which results in an increase in the degree of unsaturation of the photoproducts with increasing photon energy. This picture is supported by transient infrared spectroscopic studies of product distributions as a function of photolysis wavelengthe6 Further, both molecular beam studies conclude that sequential loss of CO can be rationalized by a statistical model. The current study makes use of fast time-resolved IR spectroscopy (TRIS) to quantitatively determine the UV wavelength dependent product distribution of the photofragments produced by the gas-phase photolysis of Fe(CO)S. A diode laser has been employed in this study instead of the more commonly used C O laser, which has typically been the probe source of choice for gas-phase studies of coordinatively unsaturated metal carbonyls. The CO laser by necessity lases at frequencies corresponding to transitions of excited-state CO. Thus transient IR spectra of metal carbonyls obtained with a CO laser may contain contributions, at a given wavelength, from both metal carbonyls and internally excited CO. As pointed out previously, this makes quantitative determinations of product distributions and absorption coefficients of coordinatively unsaturated species difficuk6 However, with a diode laser probe it is possible to easily distinguish metal carbonyl absorptions from absorptions of internally excited CO. This has 0 1992 American Chemical Society
2562 The Journal of Physical Chemistry, Vol. 96, No. 6,1992
Ryther and Weitz Faraday Cage
In%
Detector
I
Excimer
Laser
Parabola Minor
;
uv Reaction Flow Cell
ccst4d
IR Windows
CO/ He Discharge
Figure 1. Schematic of the time-resolved infrared spectroscopy apparatus used in the iron pentacarbonyl product distribution studies.
allowed for an accurate determination of the products as a function of photolysis wavelength. A simple method based on reaction kinetics has been employed to quantitatively determine the product distributions and the absorption coefficients of the unsaturated metal carbonyls produced in this study. This information coupled with observations on the pressure dependence of the product distributions in this system provides further information about the photodissociation dynamics of Fe(CO)S. 11. Experimental Section
The experimental apparatus, shown schematically in Figure 1, is described in more detail in ref 7. Fe(CO), is continuously passed through mass flow controllers (MKS, Tylan) and into a 35 cm X 1.5 cm Pyrex cell fitted with CaF, windows with the cell pressure monitored using a capacitance manometer (MKS). Unfocused UV radiation from an excimer laser (Questek 2200 series) operating at 193, 248, or 351 nm is incident on the cell at a 1-Hz repetition rate which is approximately equal to the flow cell evacuation rate. UV fluences were measured, using a Scientec power meter, to be approximately 3 mJ/cm2 for ArF and 10 mJ/cm2 for KrF and XeF excimer pulses, well below the fluence limits resulting in two photon processes as determined by Hepburn and c o - ~ o r k e r s . ~ The coordinatively unsaturated iron carbonyl species produced on the UV photolysis of Fe(CO)S were detected by monitoring the attenuation of the continuously tunable, C W IR diode laser (Laser Photonics) probe beam which is detected by a liquid nitrogen cooled InSb photovoltaic detector (SBRC) with a measured l / e rise time of -60 ns. Wavelength was determined by passing the diode laser output through a 0.5-m monochromator (f/ 11 Czemey-Turner). Transient signals were amplified (Perry X loo), digitized and averaged using a Lecroy 9400 digital oscilloscope, and stored on an IBM-AT computer for analysis. Data points taken from the transients obtained at different wavelengths at a common delay time were connected to produce a spectrum a t a given delay time. Fe(CO)S pressures were monitored spectroscopically. The absorption coefficient for Fe(CO)5 a t desired frequencies was obtained in a static cell using the diode laser probe by acquiring between five and ten Fe(CO), absorbance versus pressure measurements (with the pressure being measured simultaneously using a capacitance manometer). This same diode laser frequency was then used to monitor the Fe(CO), pressures in the flow cell allowing for an accurate determination of the Fe(CO)s concentration for each experimental flow setting. Possible density gradients due to mass flow in the cell have been characterized and shown not
to effect kinetic studies, if obtained under pseudo-first-order conditions.8 The Fe(CO)S pressures used in generating the time-resolved spectra were 12.5 mTorr for 193-nm photolysis, 10 mTorr for 248-nm photolysis, and 80 mTorr for 351-nm photolysis. An argon buffer gas was used in most experiments in order to increase the heat capacity of the system to damp shock waves which can result from inhomogeneous deposition of photolysis energy or release of energy by relaxation or reaction of photoproducts. The argon buffer gas also provided a convenient third body for the stabilization of association reactions. 111. Results
Initial assignments of the coordinatively unsaturated iron carbonyls produced on single photon UV photolysis was made in ref 6. Reference 7 expands on this work and provides a complete discussion of the TRIS based assignments of the initial photoproducts formed on the 193-, 248-, and 351-nm photolysis of Fe(CO), as well as a determination of the rate constants for formation of polynuclear iron carbonyls formed on reactions of these photolysis products with Fe(CO),. These initial photoproduct assignments are briefly recounted here as they pertain to the determination of quantitative photoproduct yields. Figure 2 displays single time spectra of the 193-, 248-, and 351-nm UV laser photolysis of Fe(CO)Staken with the diode laser. Figure 2A displays the 193-nm Fe(CO)s photolysis spectrum obtained using the diode laser as a probe. In ref 7 the major photoproduct of 193-nm photolysis of Fe(CO)S was assigned as Fe(CO), (labeled as feature 2). A tentative assignment was made of feature 1 as an absorption of Fe(C0) with the less likely possibility that the species could be due to an excited electronic state of Fe(C0)2 or Fe(CO)3. Figure 2C is the spectrum obtained following 351-nm photolysis of Fe(CO)s. The initial photoproducts are assigned to Fe(CO)3 (labeled as feature 3) and Fe(C0)4 (labeled as feature 4). A third initial photoproduct, assigned to an excited electronic state of Fe(C0)4 (labeled Fe(CO).,*)’ is not observable in Figure 2C as Fe(C0)4* is rapidly quenched to ground-state Fe(CO)4 in the presence of a high buffer gas pressure. However, its effect on the product distribution on 351-nm photolysis will be discussed in more detail in section 1II.B. Figure 2B is a single time spectrum obtained using the diode laser following 248-nm photolysis of Fe(CO)5. The initial photoproducts are assigned as Fe(CO)* (feature 2) and Fe(C0)3 (feature 3). The absorption bands of these initial photoproducts have the same position and shape, and these species have the same rate constant for reaction with CO and Fe(CO)s as the respective
The Journal of Physical Chemistry, Vol. 96, No. 6,1992 2563
Iron Carbonyl Photolysis Products of Fe(CO)5
1.1
Photolytic Depletion
4 Iri. L
.
2100
,
.
,
,
2060
,
#
.
2020
.
#
.
1980
,
,
#
.
.
1940
I
,
1900 \
Reactive Decay
3
0.0
2
I . ' , ' , . . , . . . ' . . . ' 2100
2060
2020
1980
1940
t
2100
2060
2020
1980
.
1940
Fe(C0)5 Decay af 2016 em-'
10.0
20.0 Time in Microseconds
30.0
40.0
Figure 3. Fe(CO), decay signal obtained with the diode laser probe at 2016 cm-' after 193-nm photolysis of 0.0125 Torr of Fe(CO)5in 10 Torr of argon. The top bracket represents the initial Fe(CO)S decay on photolysis. The bottom bracket represents the Fe(C0)5depleted due to the reaction of Fe(CO), + Fe(CO)Sand reaction of Fe(C0) + Fe(CO)S. The best fit (shown) to the reactive decay is a double exponential curve (see text for discussion).
1900
,
1900
Wavenumbers (cm-')
Figure 2. (A) Time-resolved spectra generated upon 193-nm photolysis of Fe(CO)5after 1.0 ps. The photolysis flow cell contained 12.5 mTorr of Fe(CO)5with 10 Torr of argon buffer gas. (B) Time-resolved spectra generated upon 248-nm photolysis of Fe(CO)Safter 0.4 ps. The photolysis flow cell contained 10 mTorr of Fe(CO)5with 3 TOKof argon buffer gas. (C) Time-resolved spectra generated upon 351-nm photolysis of Fe(CO)5after 0.2 ps. The photolysis flow cell contained 80 mTorr of Fe(CO)5 with 30 Torr of argon buffer gas. See text for assignments.
photoproducts produced on 193-nm photolysis of Fe(CO), and the 351-nm photolysis of Fe(CO),. A third initial photoproduct, which has decayed due to rapid collisional quenching on the time scale of Figure 2B, is observed at low buffer gas pressure, is assigned to a singlet excited electronic state of Fe(CO)3, and is discussed in more detail in section III.A.3.b. A. Quantitative Determination of Product Distributions. 1. 193-nmPhotolysis. When Fe(CO)5 is in large excess, the amount of photoproduct consumed in a reaction producing a dinuclear species will be equal to the amount of Fe(CO)5 consumed in the reaction. With determination of the quantity of Fe(CO), that is consumed in each reactive process involving a coordinatively unsaturated metal carbonyl, the initial concentration of each of these photoproducts can be obtained. This procedure is most easily implemented when the rates of reaction of the kinetic processes that compete for the Fe(CO), reactant are easily separable in time. For 193-nm photolysis, the only reactions of photoproducts with parent are those involving Fe(CO)2 and Fe(C0). Since Fe(CO)2 is by far the major photoproduct, it reacts 4 times more rapidly with parent than Fe(CO), and the rate constants for reaction of Fe(C0)2 and Fe(C0) are independently known (with reaction and 9.2 X lo-" cm3 molecule-I s-l, rate constants of 3.7 X respectively7),a good quality double exponential fit of the depletion
of Fe(CO), due to these reactions is obtainable. These rate constants can be fixed, and only the preexponentials remain to be determined by the fitting procedure. A signal obtained following 193-nm photolysis is shown in Figure 3, where the loss of Fe(CO), is monitored at 2016 cm-'. The initial very fast decay in this signal is due to photolysis of Fe(CO),. From the part of the signal corresponding to reactive loss of Fe(CO),, a concentration corresponding to the amount of Fe(CO)5 consumed by reactive processes can be calculated either relative to the photolytic loss of Fe(CO)5or in absolute terms using the measured absorption coefficient of 0.39 cm-' Torr-' for Fe(CO), at 2016 cm-I. Clearly, the amount of Fe(C0)5 photolyzed and the amount of Fe(CO), that is depleted due to reaction should be equal if all photoproducts react with parent and if there is no further reaction of the polynuclear species with parent. In Figure 3,2.8 mTorr of the initial concentration of 12.5 mTorr of Fe(CO), is photolyzed by the 193-nm laser pulse. (A single 193-nm laser pulse photolyzed 24% of the Fe(CO)5 at the entrance of the reaction cell and 20% at the exit with an average of 22.4%.) Further depletion of Fe(CO), due to reactions of Fe(CO)S with the initial photoproducts resulted in the loss of an additional 2.6 mTorr of Fe(CO)S. The 2016-cm-' probe frequency was chosen because it is a wavelength at which there is minimal Fe2(C0), absorption, which is the expected product of the Fe(CO)2 Fe(CO), reaction. Nevertheless, the apparent 8% difference between the Fe(C0)5 photolysis loss and the Fe(CO)5 loss due to reactions is probably due to some overlap, at 2016 cm-I, of the Fe(CO)5 decay and the rise of the Fe2(C0)7dinuclear product (with absorptions at 2045,2025,2000, and 1950 an-').' Reactions of the photoproducts with nascent CO can be discounted since, on the basis of the measured rate constants for reaction of the initial photoproducts with parent and CO, the percentage of Fe(CO), photolyzed, and the observed product d i ~ t r i b u t i o nthese ~~ reactions result in the removal of less than 0.25% of the total photoproduct on the time scale of the reaction with parent. The amount of C O in the flow system before photolysis is negligible. A diode laser scan of a CO absorption in the Fe(CO), flow system (in the absence of photolysis) failed to detect any CO, indicating that the amount of CO in the system before photolysis is less than 1% of the CO formed during photolysis. Figure 4 compares the experimental decay signal (at 2016 an-]) subtracted from calculated fits for five different possible Fe(CO)* and Fe(C0) concentrations. The best fit is signal C with 10% Fe(C0) and 90% Fe(C0)2. Including experimental error as well as the error in the fit yields a relative concentration of 10 f 5% Fe(C0) and 90 f 5% Fe(C0)2 following 193-nm photolysis of
+
2564
The Journal of Physical Chemistry, Vol. 96, No. 6, 1992
I
I 10.0
20.0
30.0
40.0
50.0
Time in Microseconds
Figure 4. Comparison of the residuals of the time-resolved signal representing the reactive decay subtracted from calculated fits for five different combinations of relative concentrations of Fe(CO)*and Fe(CO), respectively, as follows: A, 95:5; B, 9 2 5 7 . 5 ; C, 90:lO; D, 87.5:12.5;E, 85:15. The diode laser probe was at 2016 cm-'.
Fe(CO)5. Taking into account the effect of the Fe2(C0), absorption at 2016 cm-' will only slightly increase the calculated Fe(CO)2concentration, leaving it well within the range of the error in the fit. From these product distributions, absorption coefficients can be calculated at the peak of the absorption bands. Fe(C0)2 has an absorption coefficient at 1920 cm-I of 0.33 f 0.03 cm-' Torr-' and Fe(C0) has an absorption coefficient of 0.2 f 0.1 cm-I Torr-] at 1946 cm-]. The relatively large error for the Fe(C0) absorption is due to overlap by the much larger Fe(C0)2 absorption. 2. 351 -nmPhotolysis. a. Ground-State Products. The determination of the product distribution at 351 nm is simplified by the more than 2 orders of magnitude difference in reaction rate constants of the two major photoproducts, Fe(C0)3 and Fe(C0)4, with Fe(C0)5 (2.9 X and 5.2 X cm3 molecule-' s-], re~pectively).~Due to this large difference in rate constants, less than 1% of the Fe(CO), will have reacted with parent by the time Fe(C0)3 has completely reacted with parent. The quantity of Fe(CO)5 consumed by the reaction of Fe(C0)3 + Fe(CO)5 is measured to be 61 f 5% of the quantity of Fe(CO), lost due to photolysis. Once again, this 61% corresponds directly to the concentration of Fe(CO)3 relative to the total photoproduct concentration if the initial photoproduct reacts exclusively with parent and the reaction product does not further react with Fe(CO),. The 2032-cm-I frequency was chosen for the determination of Fe(CO), decay amplitudes since, within the region of parent absorption (2040-2000 cm-I), the Fe2(CO)8absorption is near a minimum at this freq~ency.~ However, on the much longer time scale of the Fe(CO), + Fe(CO)5 reaction, an independent measurement of the Fe(CO), concentration is hindered by overlapping absorptions of Fe2(C0)9(the product of the Fe(CO), + Fe(CO)5 reaction with absorptions at 2066, 2060, 2045, and 2036, cm-l).' The measurement is further complicated by the overlap between Fe(CO)4 and parent between 2016 and 2000 cm-I, a region where the Fe,(CO)9 absorptions are not as intense. Therefore, the relative amount of Fe(CO), following 351-nm photolysis is reported as the difference between the Fe(C0)5 loss on photolysis and the Fe(CO)5 loss in the reaction of Fe(C0)3 + Fe(C0)5, or 39 5%. Using these product distributions the absorption coefficient at 1952 cm-', which corresponds to the peak amplitude for Fe(C0)3, is 0.20 f 0.03 cm-I Torr-]. The absorption coefficient for Fe(CO), at 1995 cm-l is 0.28 f 0.06 cm-l Torr-]. (The error limits for the area of the Fe(C0)4 absorption band are estimated to be f10% due to some overlap of the Fe(CO), and Fe(CO), absorptions.) 6. Electronically Excited Fe(CO),. As discussed in ref 7, another species best assigned as an excited electronic state of Fe(CO), (Fe(CO),*) is produced on 351-nm photolysis of Fe(CO),. Fe(CO),* reacts rapidly with Fe(C0)5 with a rate con-
*
Ryther and Weitz stant of (1.8 f 0.3) X lo-'' cm3 molecule-' SKI (slightly slower than the reaction of Fe(CO), + Fe(CO)5). The reaction of Fe(CO),* + Fe(CO), occurs in competition with collisional relaxation of Fe(CO),* to ground-state Fe(CO),. The rate constant for relaxation of Fe(CO),* by argon was measured to be (6 f 3) X cm3 molecule-' SKI, and the rate constant for the sum of relaxation and reaction of Fe(CO),* with CO was measured to be (1.5 f 0.3) X lo-' cm3 molecule-I s-]. Since ground-state Fe(CO), is formed highly vibrationally excited, the IR spectrum of this photofragment is broadened (narrowing and moving toward higher frequencies as Fe(CO), relaxes). Addition of CO as a vibrational relaxer of Fe(CO), was used to show that Fe(CO), is an initial photoproduct and not just formed by relaxation of Fe(CO),*. (Without CO, the Fe(CO),* absorption is masked by growth in the same frequency region of the Fe2(CO)8dinuclear species from the Fe(CO), + Fe(CO), reaction.) Since ground-state Fe(C0)4 reacts slowly with CO, while Fe(CO),* and Fe(C0)3 are rapidly depleted by CO, using a high enough CO pressure, the rate of formation of vibrationally cold Fe(CO), is observed to grow in at a faster rate than the depletion of either Fe(C0)3 (at 1950 cm-I) or Fe(CO),* (at 1980 cm-I). In other words Fe(CO), forms before Fe(C0)3 can react with CO and before Fe(CO),* is quenched to the ground-state Fe(CO), species. The amount of the Fe(CO),* photoproduct can be measured at low buffer gas pressure by determining the amount of Fe(CO)5 consumed in the Fe(CO),* + Fe(CO), reaction in the same manner as that used to obtain the amount of Fe(CO), formed in section III.B.l above. At low buffer gas pressures, Fe(CO)5 will be removed by the two processes (Fe(CO), + Fe(CO), and Fe(CO),* + Fe(CO),) on approximately the same time scale. The best fit to the 2032-cm-' decay signal incorporated the measured reaction rate constants for the Fe(C0)3 + Fe(CO)5 and Fe(CO),* + Fe(CO), and varied the preexponentials for these processes obtained a value of 8 f 4% for the amount of Fe(CO), depleted by reaction with Fe(CO),* relative to the amount of Fe(CO), depleted by reaction with Fe(CO),. Extrapolating to zero rare gas pressure increases the percentage of Fe(CO),* such that an upper limit for the Fe(CO),* formed upon 351-nm photolysis of Fe(CO), is 10 f 5% of the initial Fe(C0)3 concentration or 6 f 3% of the total initial photoproduct distribution. 3. 248-nm Photolysis. a. Ground-State Products. The determination of the product distribution following 248-nm photolysis of Fe(CO)5, using a kinetic analysis similar to that used for 193and 351-nm photolysis, is complicated by the fact that both photoproducts, Fe(C0)3 and Fe(C0)2, react with parent at very similar rates. However, the relative amounts of each species can be calculated as 64 f 7% Fe(CO), and 36 f 7% Fe(C0)2 using the already determined absorption coefficients for Fe(C0)3 and Fe(C0)2 obtained from the 351- and 193-nm photolysis data, respectively. Interestingly, unlike the behavior observed with the products of 193- and 351-nm photolysis, where argon pressure was varied from 5 to 100 Torr with no observable change in the product distribution (except for quenching of Fe(CO),*), changing the pressure of argon over the range of 3-10 Torr has an effect on the 248-nm product distribution, increasing the amplitude of the Fe(C0)3 absorption and decreasing the amplitude of the Fe(CO)2 absorption. The relative amounts of Fe(CO)3 and Fe(C0)2 measured at 3 Torr of argon are 64 f 2% for Fe(C0)3 and 36 f 2% for Fe(C0)2 and at 10 Torr of argon are 71 f 2% for Fe(C0)3 and 29 f 2% for Fe(C0)2. This dependence of the product distribution on the inert gas pressure is discussed further in section IV below. b. Electronically Excited Fe(CO)3. Reference 7 assigns an absorption formed on 248-nm photolysis to an electronically excited Fe(CO)3 species (most likely singlet Fe(CO),*). Because Fe(CO)3* is collisionally quenched (after 10 collisions with argon buffer gas) much more rapidly than Fe(CO),*, it was not possible to accurately monitor reactions of the species. The relative amount of Fe(C0)3* is also difficult to obtain, even though Fe(C0)3* is collisionally quenched to the ground-state Fe(CO), species. Since
The Journal of Physical Chemistry, Vol. 96,No. 6,1992 2565
Iron Carbonyl Photolysis Products of Fe(CO)S TABLE I: Bond Dissociation Energies (BDE) for Fe(CO), (x = 1-5)
BDE(Fe(CO),/eV lit. ref Engel king et aLZo
2.4 f 0.5
Waller and Hepburn4
1.8
Poliakoff and we it^'^^ current studyd Lewis et aL2' Vernon et a1.2f
1.8 1.8
x=5
* 0.5
x = 2
x = l
1.4 f 0.3 1.4 f 0.3
1.0 f 0.3 1.0 f 0.3
1.0 f 0.3 1.0 f 0.3
1.4 f 0.42 [0.8 f 0.421
1.0 f 0.34
1.1
1.3
x=3
x = 4 0.2 f 4 0.8 4 0.2 f I
*
>0.2 0.4
1.6
Obtained from a thermodynamic cycle containing measurements of mass spectrometric appearance potentialsz2and electron detachment value of Lewis et aL2' for first BDE ( x = 5 ) and values for x = 3, 2, and 1 from ref 17; the second BDE ( x thresholds for [Fe(CO),I]- ions. = 4) is the difference between the sum of these energies and the experimental energy for Fe(C0)5 Fe + 5CO dissociation of 6.08 f 0.07 eV.23 cAssessmentof error bars in ref 20 bond dissociation energy data. dAn extension of error bar assessment from c for BDE values from Engelking et al. data. eExperimentallydetermined by pulsed laser pyrolysis. 'Used data in ref 21 for first BDE and chose deviations from data in ref 20 for the remaining vaiues on the basis of experimental results. +
Fe(CO),* quenching is more rapid than the vibrational relaxation of the initially formed ground-state Fe(CO), photoproduct, it should be possible to separate the formation of Fe(CO), resulting from Fe(CO),* decay from Fe(CO), formed by vibrational relaxation of vibrationally excited Fe(CO),. Unfortunately, the vibrational relaxation of ground-state Fe(C0)3 monitored at 1956 cm-l (where vibrationally relaxed Fe(CO), absorbs) is not characterized by a single exponential, as demonstrated in Figure 5. The formation of Fe(CO)3 (before 1.0 ps in Figure 5) is not even fit well with a double exponential. The fastest portion of the signal is due to the formation of Fe(CO), by relaxation of the excited electronic state species, Fe(CO),*, and appears to be a single exponential rise. The slower portion of the signal is assigned to Fe(CO), formation due to relaxation of vibrationally excited Fe(CO), but is not fit well by simply adding a second exponential. A good fit to this second portion of the Fe(CO), rise requires the use of two exponentials. The necessity for two additional exponentials to fit the Fe(CO), rise can be explained in terms of an induction time for the vibrational relaxation of Fe(CO),. This is not surprising since vibrational relaxation of Fe(CO), would not be expected to occur in a single collision event and Fe(CO), species with different degrees of vibrational excitation would be expected to proceed at differing rates. The best fit to the Fe(C0)3 formation signals monitored at 1956 cm-I (Figure 5) imply that Fe(CO),* represents 15 f 10%of the total Fe(CO), amplitude or approximately 10% of the total photoproducts produced after 248-nm photolysis of Fe(CO)S in the presence of 3 Torr. The error represents the uncertainty in the amplitude represented by the Fe(CO), formed by the quenching of the vibrationally excited Fe(CO), during the induction time. Due to the overlap between internally excited Fe(CO), and Fe(CO),, which is more severe and lasts longer at lower buffer gas pressure, it was difficult to make detailed measurements of the Fe(CO),* branching ratios at Ar pressures lower than 3 Torr. However, it is expected that 15 f 10% is a lower limit for the amount of Fe(CO),* produced at zero buffer gas pressure. B. Absorption Coefficients. The peak absorption coefficients for the Fe(CO), species are as follows: Fe(CO), 0.2 f 0.1; Fe(C0)2, 0.33 f 0.03; Fe(CO),, 0.20 f 0.03; and Fe(CO),, 0.28 f 0.06. As previously discussed, knowledge of these peak absorption coefficients has allowed for the determination of the branching ratio among Fe(CO), products following 248-nm photolysis and should allow for facile determination of product branching ratios at any other photolysis wavelength. For Fe(CO),, which has C,, symmetry, this absorption coefficient represents a superposition of the absorptions of the A,, B,, and B2 modes. Though some questions regarding the geometry of Fe(CO)3 have been raised in ref 7, the accepted symmetry of Fe(CO), is C,,, and for this symmetry the corresponding measured absorption coefficient is for the E mode. The geometry of Fe(C0)2 has not been characterized in matrix isolation studies. However, it is reasonable to expect it to have C2, symmetry and to expect that the two CO stretchings do not exactly overlap and thus that the absorption being observed is due to a single CO stretch. A similar conclusion would be reached for Dmhsymmetry. Though
z
fra
se 0
4
0.0
1.0
2.0
3.0
4.0
s.0
6.0
7.0
8.0
ao
10.0
Microseconds Figure 5. Time-resolved transient signal obtained after 248-nm photolysis of Fe(CO), at 1956 em-' and fit to a double exponential rise and a single
exponential fall. little is known about Fe(CO), the measured absorption coefficient is representative of the one CO stretching mode. Little attention has been devoted toward calculation or even rationalization of the magnitudes of absorption coefficientsin metal carbonyls. Thus, there is little in terms of detailed prior expectations regarding the magnitude of the absorption coefficients of the various stretching modes of metal carbonyls. However, since the Fe(CO), and Fe(CO), absorption bands have similar halfwidths and shapes, it is clear by comparison of the absorption coefficients of these two species that the absorption coefficient does not simply scale with the number of CO's involved in the vibrational mode. Reproducing the magnitudes of these absorption coefficients represents another challenge to the successful theoretical treatment of these systems.
IV. Discussion A. Product Distributions. Figure 6 schematically illustrates the energies of the coordinatively unsaturated iron carbonyl species relative to ground-state Fe(CO)5 and also schematically depicts the energies of the photolysis wavelengths used in this and other studies of the iron carbonyl product distributions. The iron carbonyl states in Figure 6 represent the bond dissociation energies for the given species (see Table I), while the brackets represent the amount of energy attributable to the photoejected CO mole c u l e ~ .The ~ ground-state energies of the unsaturated species in Figure 6 are shown as boxes to represent the error in these energy values. No error bars are shown for excited-state iron carbonyl energies or for the photoejected C O energies. Table I1 is a compilation of the product distributions reported in these experiments and in the literature for the single photon UV photolysis of Fe(CO)5. From a comparison of these Fe(CO)S product distributions to the energy diagram in Figure 6, it is apparent that the sum of literature values of the bond energy and
2566 The Journal of Physical Chemistry, Vol. 96, No. 6, 1992
Ryther and Weitz
TABLE 11: Comparison of Product Distributions Formed on UV Photolysis of Fe(CO)5 product distribution at given wavelength/% current study” Waller and Hepburnb Vernon et al. 193 nm 248 nmc 351 nm 193 nm 248 nm 266 nm 351 nm 193 nm 248 nm 0 38 39 0 0 0 0 Fe(C0)4 0 0 0 61 13 (0) 87 (47) 98 62 6 Fe(CO)3 0 64 0 86 (98) 13 (53) 100 2 0 94 Fe(CO), 90 36 0 0 0 Fe(C0) 0 0 0 1 (2) 10 0
193 nm 9 9 81 1
Yardley et al. 248 nm 332 nm 10 23 35 45 55 31 0 0
For comparison purposes Fe(CO),* species are lumped with the respective Fe(CO), species. bReference 4 distributions using bond energies from ref 20 (no parentheses) in electron volts: Fe(CO)5, 1.8; Fe(CO)4, 0.8; Fe(CO),, 1.4; Fe(C0)2, 1.0; Fe(CO), 1.0. Terms in parentheses represent the effect of a change in bond energies to indicated values in electron volts as calculated from ref 4: Fe(CO),, 1.8; Fe(C0)4, 0.8; Fe(CO),, 0.9; Fe(CO)*, 1.5; Fe(CO), 1 .O. ‘Reported for 3-Torr total pressure. I
I
I
6.0
I
248 5.0 266 nm 4.0
351 nm 3.0
c
aI ‘ O’O
I Fe(C0I5
F i p e 6. Diagram of the energies of the mrdinatively unsaturated iron carbonyl species relative to ground-state Fe(CO),. The ground-state energies are represented by boxes in order to depict the error in these energy values (see Table I Q 4 The excited-state energies are values obtained from refs 22 and 23 and are not shown with error bars. The energies of the 193-, 248-, 266-, and 351-nmphotons are shown as broken lines. The brackets represent the amount of energy carried away by the photoejected CO molecules (which is reported to be between 0.4 and 0.6 eV, depending on the UV photolysis ~avelength).~ The energy of photoejected CO is also represented by brackets above the Fe(C0)4* and Fe(CO)3* energy levels.
the energies for photoejected CO for an iron carbonyl species are at least qualitatively compatible with the experimental product distribution. The energy of formation for any photoproduct generated at a given UV wavelength is at most only slightly above the photolysis energy (and well within the error limits). The agreement between our product distribution data and that of Hepburn and co-workers is good, especially a t 35 1 nm where not only is there agreement on the photoproducts produced but the relative amounts of these photoproducts agree within experimental error.4 However, it should be noted that Hepburn does not report the formation of excited states of Fe(CO)3and Fe(CO),; thus agreement is based on the overall yield of these species as reported in both studies. Additionally, since the branching ratio at 248 nm is pressure dependent and the collision frequencies in the two environments that are beiig compared differ significantly, it is reasonable to expect only the qualitative agreement that is achieved. However, a larger yield of more unsaturated species would be anticipated for the lower pressure environment. This expectation is realized with the alternative choice of bond energies discussed in ref 4. This alternative choice also leads to better agreement for 193-nm photolysis. Nevertheless even for the primary choice of bond energies, the agreement between the product distributions reported in this study and those calculated by Hepburn for 193-nm photolysis is also good. In both 193-nm studies, Fe(CO)* is the major photoproduct. Formation of Fe(C0) and Fe(CO), is predicted by Hepburn while the one minor photoproduct in this study is best assigned as Fe(CO).’ The iron carbonyl photofragment distributions calculated by Hepburn for two possible sets of dissociation bond energies are shown in Table 11. What is currently known about the iron
carbonyl bond dissociation energies is summarized in Table I. Only the Fe(C0)4-C0 bond energy and the energy required to remove all CO ligands have been directly measured experimentally. The other four bond dissociation energies have been calculated from secondary experimental and/or theoretical data, and questions have been raised about these values and their associated error limits.I9 Our results and those of Hepburn are also qualitatively similar to, but differ quantitatively from, the product distributions reported by Vernon and co-workers for 193- and 248-nm photolysis wavelengths of Fe(CO)S (Table 11). Vernon reports only one photoproduct following 193-nm photolysis, which is assigned as Fe(CO)2.2 Both Fe(CO)2 and Fe(CO)3 were reported by Vernon as initial photoproducts after 248-nm photolysis of Fe(C0)5.3 However, the differences between our product distributions and those of Vernon et al. are significant with considerably more Fe(C0)2 being reported in ref 2 and 3. Though the pressure dependence of the 248-nm product yield that we observe would be expected to favor production of more Fe(CO)2in a low-pressure environment, the difference between 36% (at 3 Torr) and 94% Fe(CO)2 production (from ref 3) is still quite significant, especially in light of the relatively modest pressure dependence of the product branching ratios between 3- and 10-Torr total pressure. It should also be recognized that Vernon et al. report that it was not possible to use the experimental timeof-fight distributions to independently determine the branching ratios of Fe(CO)3 and Fe(CO), following 248-nm photolysis of Fe(C0)5. In this case the product distributions were determined on the basis of experimentally optimized iron carbonyl bond dissociation energies (Table I). The distributions calculated from the chemical trapping studies of Yardley and cuworkers’ are in some cases significantly different than those determined from our study and the molecular beam studies of Hepburn and mworkers4 and Vernon and ~ + w o r k e r s . ~ ~ The most obvious difference is that Fe(C0)4 is reported as a product even following 193-nm photolysis, while our TRIS study and the beam studies do not observe evidence for Fe(CO)4 even following 248-nm photolysis. Yardley et al. determined their photoproduct distributions by a GC analysis of chemically trapped photofragments using PF3 as the reactive trapping reagent. It is possible that an exchange of CO with PF3 may have occurred on the column and perturbed the nascent Fe(CO), distributions. Since the lowest PF3 pressure used was 9 Torr and product distributions were obtained from data at high PF3 pressure where the composition of the trapped species no longer changed, it is also possible that collisional quenching as recently observed by Rayner et al. affected their reported distribution (see bel0w).”3’~ However, this latter possibility seems an unlikely explanation for all of the differences since even though we observed collisional effects on the product distribution at 248 nm, we did not observe significant changes in the distribution of photoproduct at pressures (19) Poliakoff, M.; Weitz, E. Acc. Chem. Res. 1987, 20, 408. (20) Engelking, P. C.; Lineberger, W. C . J . Am. Chem. SOC.1979, 101, 5569. (21) Lewis, K. E.; Golden, D. M.; Smith, G . P. J . Am. Chem. SOC.1984, 106. 3905. (22) Compton, R. N.; Stockdale, J. A. D. I n f . J . Mass Specfrom. Ion Phys. 1976. 22, 41; (23) Housecraft, K . W.; Smith, B. C . J . Organomef. Chem. 1979.170, C1.
Iron Carbonyl Photolysis Products of Fe(CO)5 up to L 100 Torr of argon following 35 1- and 193-nm photolysis. B. Pressure Dependence of Product Distributions. Rayner et al., using TRIS, have recently reported a significant effect of inert gas pressure on the branching ratios of the group 6 metal carbonyls.”~’* For some coordinatively unsaturated species, they observed shifting of the product distribution at some wavelengths from more unsaturated to the more saturated photoproduct as argon was added to the system over a range of 2-40 Torr. Since the photodissociation process can produce a distribution of internal energies in the photoproducts which can be broadened by successive elimination of CO ligands in an RRKM or statistical picture, there will be a range of rate constants, k ( E ) , for loss of the next CO ligand from an internally excited coordinatively unsaturated photofragment. Collisions can inhibit loss of further C O S by collisional relaxation of the internal energy distribution. For a given photodissociation pathway, when two ‘photofragments” are observed, it implies that the distribution of k(E)’s is competitive with collisions at the experimental pressure: some of the excited molecules have enough energy to react further and some do not. Thus in the model in ref 19, increasing the pressure of quenching gas is in general predicted to lead to further quenching and a shifting of the distribution of k(E)’s toward the more saturated photoproduct. As discussed in section III.A.3.a, the photoproducts of Fe( C 0 ) 5 following 248-nm photolysis show the same trend over a range of 3-10 Torr of argon. However, there is no change in the product distribution over a range of argon pressure from 5 to 100 Torr following 193- and 351-nm photolysis of Fe(CO),. This lack of a pressure dependence for the product distribution could result from a combination of factors that distinguish the Fe(CO)5 system from the group VI hexacarbonyls. The photochemistry of the Fe(C0)5 system is different from the group VI carbonyls in that there is ample evidence of the involvement of multiple photodissociation pathways and, therefore, the participation of multiple potential energy surfaces. The involvement of multiple potential energy surfaces provides a mechanism for production of excited states of coordinatively unsaturated species.’ Preparation of molecules on each surface can result in production of these speices with varying amount of internal energy. In addition, there will be less total energy to dispose among internal degrees of freedom for a speices prepared on an excited-state surface than for the same species prepared on a ground-state surface. Collisional relaxation can lead to a competition with dissociation on each surface with each surface providing a potentially pressure dependent mix of products. However, photodissociation on each surface could also produce primarily or even exclusively a single product over the experimentally accessed pressure range. For smaller molecules the energy dependence of k ( E ) is expected to be steeper. The steeper k ( E ) is beyond the threshold for dissociation, the less effectively collisions can compete with dissociation. This effect could also help in explaining the lack of an observed pressure dependence in the Fe(C0)5 system for 193- and 351-nm dissociations though it is a priori a more likely explanation for 193-nm photodissociation. However, since multiple surfaces are almost certainly involved in the photodissociation of Fe(CO)5at these wavelengths, we feel these surfaces are also likely to play a role in the determination of the final product distribution. The process that leads to production of Fe(CO)2 following 248-nm
The Journal of Physical Chemistry, Vol. 96, No. 6, 1992 2567 photolysis of Fe(CO)5 apparently takes place near the energy boundary between the Fe(C0)2 and Fe(CO), species, such that inert gas pressure quenching of the Fe(CO),