1288
J. Phys. Chem. 1986, 90, 1288-1293
viations have been analyzed in terms of a maximum to minimum length ratio in the initial pattern chosen to generate the fractal. Moreover, we have shown that these deviations from the earlier result (eq 4) should exhibit the same kind of periodicity in the time scale, the donor intensity scale, and the acceptor concentration scale, but this is no longer true when the space is not fractal. This periodicity has been analyzed in terms of the initial number No of sites used to generate the pattern and the dilation factor b in the length scale. In particular, this means that the knowledge of the time dependence of the donor decay at a given concentration should enable the experimentalist to predict the time dependence at other concentrations and/or time scales. This is not at all the case when dealing with microstructures which do not have dilation symmetry.
Conclusions From the simulations on the fractal lattice with discrete dilation symmetry, eq 4 is found to describe most of the time-dependent behavior when the ratio of D / d is close to 1 (or /3 is small, see eq 8). In this case, the plot of In of negative logarithm intensity vs. In of time shows a straight line, the slope of which gives the value of the fractal dimension D. However, as the D / d ratio decreases (or the value of /3 increases), the deviation from the straight line behavior becomes large, and an oscillatory behavior is observed. Fortunately, from the oscillatory behavior, information regarding the geometry of the fractal structure being studied can be obtained. For example, from Figure 3, the oscillation period is equal to the logarithm of the characteristic factor bD. The change in the logarithm of negative logarithm of the intensity corresponding to this period is equal to the logarithm of the number of points in the generating pattern, In No. Furthermore, D can still be determined from the slope of the best straight line fit through the oscillating function. Lastly, the amplitude of the oscillation, as measured from the best straight line fit, is related to the value of (3 as defined in eq 8. In principle, the results obtained for the case of small D / d
should enable experimentalists to test whether the anomalous decay is due to a real fractal space or to a regular structure with excluded volume. For fractal structure, it is possible, at least in principle, to characterize the fractal dilation symmetry by determining the value of No and of b. Unfortunately, the range of the time and the concentration which can be studied experimentally in energy-transfer processes is, in general, much smaller than those used in our simulation calculation. Due to the relatively short time of the donor excited state, only a small time range can be studied experimentally. This, together with the oscillatory behavior of the In (-In I ( ? ) ) vs. In tDI6, could yield error in the value of D experimentally determined. Furthermore, the value of D determined could depend on the time range plotted (Le., on different portion of the oscillation used). Thus, due to the experimental limitation imposed by the lifetime of the donor excited state, as well as the distance dependence of the dipolar coupling mechanism, the determination of the real fractal dimension from the one-step dipolar energy-transfer might be met with uncertainty. For example, the smallest length scale for an intermolecular one-step energy transfer is the intermolecular distances (-4 8,).A transfer time at this distance is expected to be 10-'2-10-13 s. At a transfer s (the radiative lifetime), a distance of R, = ( r 2 / time T ~ ) ~= /(10-9/10-13)'/6 ~R~ X 4.0 = 18.6 8, is expected. This is only a few length scale. Even if one takes the well accepted characteristic dipolar transfer distance of 50 A, this is at best 12 times the smallest possible length scale in this kind of experiment. In a real fractal, one expects that the length scale of the structure to be larger than the intermolecular distance in a packed crystal (otherwise the structure is not disordered). One might then ask about the value of the minimum distance spanned by the energy-transfer process over the structure in order for the observed results to be able to distinguish between a real fractal and an organized structure with excluded volumes.
-
-
Acknowledgment. The authors thank Mr. T. Corcoran for careful reading of the manuscript and the Office of Naval Research for financial support. P.E. thanks NATO for a travel grant.
UV Multiphoton Dissociation of Volatile Iron Complexes, As Revealed by MPI Ion-Current and Photoelectron Spectroscopy Yatsuhisa Nagano, Yohji Achiba, and Katsumi Kimura* Institute for Molecular Science, Okazaki 444, Japan (Received: August 26, 1985; In Final Form: October 18, 1985)
Gas-phase multiphoton ionization (MPI) ion-current spectra were measured in the laser wavelength region 366.5-370.0 nm for several volatile iron complexes such as ferrocene (Fe(Cp)J, iron pentacarbonyl (Fe(CO)5), iron tris(acety1acetonate) (Fe(Acac),), and iron trichloride (FeCI,). Remarkable differenceswere found in the ion-current spectra among these complexes. However, from photoelectron spectroscopic measurements, it was found that all these ion-current spectra are due to twoor three-photon ionizations of Fe atoms. It is concluded that Fe(Cp), and Fe(CO)s provide two extreme cases of producing Fe atoms in multiphoton dissociation: (a) The Fe atoms produced from Fe(Cp), are populated mostly in the ground state, whereas (b) those produced from Fe(CO)5 are distributed among various electronic excited states up to 3 eV. In this paper we also discuss the mechanism of formation of excited Fe atoms. It is also concluded that the extent of distribution of excited Fe atoms largely depends on the energy transfer taking place from the molecular electronic excited state to the ligands during fragmentation.
Introduction R ~ ~multiphoton ~ ~ ~ionization I ~ (MPI) , studies for many volatile metal complexes in the gas phase have been carried out by ion-current and mass spectroscopic measurements with visible and UV l a ~ e r s . l - ~Most ~ of the ion-current spectra show sharp ( I ) G. J. Fisanick, A. Gedanken, T. S. Eichelberger IV, N. A. Kuebler, and M. B. Robin, J. Chem. Phys., 75, 5215 (1981).
0022-3654/86/2090-1288$01.50/0
lines attributable to resonant-enhanced multiphoton ionizations of metal atoms. From these studies, it has been suggested that (2) D. P. Gerrity, L. J. Rothberg, and V. Vaida, Chem. Phys. Lett., 74, ( 3 ) D. A. Lichtin, R. B. Bernstain, and V. Vaida, J . Am. Chem. SOC.,104, 1830 (1982). (4) P. C. Engelking, Chem. Phys. Left., 74, 207 (1980). ( 5 ) R. L. Whetten, K . 4 . Fu, and E. R. Grant, J . Chem. Phys., 79, 4899
(1983).
0 1986 American Chemical Society
The Journal of Physical Chemistry, Vol. 90, No. 7 , 1986 1289
UV MPI of Volatile Iron Complexes neutral metal atoms are efficiently generated in the multiphoton dissociation of gaseous metal complexes. It has also been indicated that the several metal complexes give rise to different ion-current spectra in spite of the formation of identical metal atoms. Fisanick et a1.l have indicated for Cr complexes that resonant MPI lines due to Cr atoms appear with intensities characteristic of the parent complexes. Welch et al.ls have considered for Fe and Ru complexes that concurrence of MPI ion-current lines provides evidence for formation of identical photofragments. Furthermore, Karny et a1.I6 have indicated that fluorescence spectra due to Fe atoms produced by UV multiphoton dissociation of iron pentacarbonyl, 1,1’-dimethylferrocene, and butadieneiron tricarbonyl are different from one another. In the course of our MPI experiments on several iron complexes, we found that their MPI ion-current spectra observed in the visible region 400-480 nm are mainly composed of identical lines with different intensities. The appearance of the identical ion-current lines is obvious evidence for the resonant MPI signals due to Fe atoms. However, in the shorter wavelength region 366.5-370.0 nm, we have found extreme differences in the MPI ion-current spectra between iron pentacarbonyl (Fe(CO)5) and ferrocene (Fe(Cp),). There are no common features between the two ion-current spectra, in spite of the mass spectroscopic evidence that Fe+ ions are dominantly formed in this wavelength region. In the present work, in order to clarify the mechanism of multiphoton dissociation of Fe complexes, we have carried out MPI ion-current and photoelectron measurements with the following six compounds: (a) ferrocene (Fe(CsH5),), denoted Fe(Cp),; (b) iron tris(acety1acetonate) (Fe(C5H70J3), denoted F e ( A ~ a c )(c) ~ ; iron pentacarbonyl (Fe(CO)S);(d) iron trichloride (FeCI,); (e) 1,l’-benzoylferrocene, and (f) iron tris(heptafluoro)dimethyloctadionate. MPI studies of these Fe complexes have not been reported yet, except for (a) and (c). In this paper, we want to demonstrate the following remarkable features for the Fe complexes. (1) The whole ion-current spectra studied here can be interpreted in terms of multiphoton ionizations of Fe atoms from photoelectron analysis. (2) The observed extreme differences in the ion-current spectra of Fe(CO)S and Fe(Cp), are attributed to remarkable differences in the distribution of the electronic states of Fe atoms. We also want to discuss here the influence of the metal-ligand bond dissociation energies as well as the ligand vibrational-state densities on the mechanism of the multiphoton dissociation of the Fe complexes.
Experimental Section A nitrogen laser pumped dye-laser system (Molectron UV-24, DL-14) was used in the present experiments. A laser beam with a power less than 1 mJ at each pulse (6-8 ns in width) was focused by a quartz lens cf = 25-30) on the ionization region 1 cm away from the sample nozzle. MPI ion-current spectra were measured as a function of laser wavelength in the region 366.5-370.0 nm. Photoelectron spectra were measured by tuning the laser wavelength to various significant ion-current peaks. The apparatus used here is described elsewhere.” A timeof-flight (TOF) analyzer was used for measurements of photo-
-
(6) Y . Nagano, Y . Achiba, and K. Kimura, J . Chem. Phys., 84, 1063 (1986). (7) V. Vaida, N. J. Cooper, R. J. Hemley, and D. G. Leopold, J . Am. Chem. Soc., 103, 7022 (1981). (8) S. Leutwyler, U. Even, and J. Jortner, Chem. Phys. Lett., 74, 11 (1980). (9) S. Leutwyler, U. Even, and J. Jortner, J . Phys. Chem., 85,3026 (1981). (10) S. Leutwyler, U. Even, and J. Jortner, Chem. Phys., 58, 409 (1981). (11) P. A. Hackett and P. John, J . Chem. Phys., 79, 3593 (1983). (12) S. A. Mitchell and P. A. Hackett, J . Chem. Phys., 79,4815 (1983). (13) A. Gedanken, M. 9. Robin, and N. A. Kuebler, Inorg. Chem., 20, 3340 (1981). (14) A. Gedanken, M. 9. Robin, and N. A. Kuebler, J . Phys. Chem., 86, 4096 (1982). (15) J. A. Welch, V. Vaida, and G.L. Geottroy, J . Phys. Chem., 87, 3635 (1983). (16) Z. Karny, R. Naaman, and R. N. Zare, Chem. Phys. Lett., 59, 33 ( 1 978). (17) Y. Achiba, K. Sato, K. Shobatake, and K. Kimura, J . Chem. Phys., 78, 5474 (1983).
I
b)
I
Laser Pulse
Heater C u r r e n t
L
generator
control unit
Nozzle
1
Heater c u r r e n t \
\
Vacuum chamber
Figure 1. (a) Block diagram of the temperature-control system of the nozzle as well as the laser system. Pulses of 10 and 30 Hz are generated by a pulse generator to trigger the laser and the heater current. (b) The pulses of the laser and the heater current are schematically shown, switched on and off at opposite times to keep the field-free condition during the photoelectron measurements. TABLE I: Electronic Terms and Configurations of Fe and Fe+ electronic electronic electronic notation term configuration energy. eV Fe 3d64s2 1” a5D 0.00-0.12 3d7(4F)4s 0.86-1.01 2“ aSF 3d7(4F)4s 1.48-1.61 a3F 3” 3d7(4P)4s asp 2.18-2.22 4” 3d64s2 a3~2 2.28-2.48 5” 3d6(SD)4s4p(3P)o 2.40-2.48 6” Z’DO 3d64s2 2.40-2.45 7“ a3H 3d64s2 2.56-2.6 1 8” b3~2 3d7(2G)4s a3G 9/! 2.69-2.76 10” Z’FO 3d6(sD)4s4p(3P)o 2.81-2.89 3d7(4P)4s b’P 2.83-2.86 11” 3d6(SD)4s4p(3P)0 2.94-3.04 12” z7p0 3d64s2 2.95-3.02 13” b3G 1+ 2+ 3+ 4+ 5+
a6D a4F a4D a4P a2G
Fe+ 3d6(5D)4s 3d’ 3d6(sD)4s 3d’ 3d7
0.00-0.12 0.2 3-0.3 9 0.99-1.10 1.67-1.72 1.96-2.03
electron spectra and mass spectra. A set of fi-metal plates was used for shielding the vacuum chamber from the external field so as to keep the field-free condition for photoelectron measurements. An effusive nozzle was used to introduce a gaseous sample into the ionization region. The nozzle was surrounded with a resistance heater made of a carbon-coated glass tube for heating the sample, its temperature being controlled in a range between room temperature and 300 O C . The nozzle-heating system as well as the pulse laser system are schematically shown in Figure 1. The present measurements for Fe(CO)s, Fe(Cp),, Fe(Acac),, and FeC13 were carried out at 25, 90, 180, and 225 O C , respectively. The heater current was switched on and off at different times after the laser irradiation, in order to maintain the field-free condition during the photoelectron measurements.
Nagano et al.
1290 The Journal of Physical Chemistry, Vol. 90, No. 7 , 1986
TABLE 11: Laser Wavelengths Used for Photoelectron Measurements, Comparisons of Experimental and Calculated Photoelectron Energies ( K ) , and Possible Assignments for the Ionization Processes N 14
7” L
0
Y -
oLc
I
‘
,‘ laser wavelength, initial state final state nm i’ (J”) j + ( P ) expt calcd 366.80
3” ( 2 )
2’
(7/2)
367.05 367.1 1
1”
4b.c 5b
1+ 3+
2.1 1.2
2.3-2.1 1.3-1.1
367.21 367.31
I”
367.36 367.44
2” 1”
2.1 1.2 3.1 2.1 1.2 1.8 3.1 3.4 0.38 0.14 0.08
2.3-2.1
6b 76
1+ 3+ 1+
8b 9b 1Ob 1I Q
368.55 369.04 369.1 1 369.48
I+
1”
2” 2” 3” (3)
3+ 2+ 1+ 1+
1’ 2’
(9/2)
2+
(7/2)
1.3-1.1
3.2-3.0 2.3-2.1 1.3-1.1 2.1-1.8 3.2-3.0 3.2-3.0 0.37-0.25 0.14 0.07
Two-photon resonant ionizations. Three-photon resonant ionizaI
I
tions. ‘See Figure 6 concerning the assignments. N=l
J
IN.3
I
01
8
I
369
2+ (s/2) 2’ ( 3 / 2 )
0.37 0.47-0.35 0.24 0.24 0.17 0.17 0.12 0.12 0.08 0.08
(9/2)
2b.c 3b
c C
370
I+ 2’
5
368
367
Wavelength (nm) Figure 2. MPI ion-current spectra observed for (a) F ~ ( C P )(b) ~ , Fe(Acac)3, (c) Fe(CO)S,and (d) FeCI,. The numbers indicate the MPT ion-current lines or peaks at which photoelectron spectroscopic measurements were carried out.
-
The time resolution of the electron analyzer is 10 ns. The energy resolution depends on the photoelectron kinetic energy ( K ) ; it is better than 20 meV in the range K C 0.5 eV, while it is more than 200 meV at K = 3 eV. Commercial samples of Fe(CO)5, Fe(Cp),, Fe(Acac),, and FeCI3 were used without further purification. A sample of 1,l’-benzoylferrocene was kindly provided by Prof. Matsumura (Nara University of Education, Nara, Japan). A sample of iron tris(heptafluoro)dimethyloctadionate was prepared according to the literature.18
conditions in the laser wavelength region 366.5 - 370.0 nm are shown in spectra a 4 in Figure 2. In this wavelength region, so far, no MPI ion-current spectra have been reported for the Fe complexes, except for Fe(C0)5.’9 Spectra a, b, and c in Figure 2 are very different in spectral pattern from one another, although spectrum d somewhat resembles spectrum c. Spectrum a observed for Fe(Cp), is the simplest in Figure 2, mainly consisting of four lines indicated by N = 3, 5, 7, and 8. Such sharp lines are often observed in MPI ion-current spectra of many metal complexes, considered as atomic resonant transitions. In fact, in some cases, ion-current lines have been successfully interpreted in terms of atomic resonances. For spectrum a in Figure 2, however, we cannot find any suitable transitions from the atomic energy table of Fe.,O In such cases, photoelectron spectroscopic analysis should be helpful for assignments of MPI ion-current lines, as described later. Spectrum b obtained for Fe(Acac), in Figure 2 shows several other ion-current lines ( N = 1 , 6, 9, 10, and 11) and weak broad bands, in addition to the lines observed in spectrum a . Among these ion-current peaks, only the two lines N = 1 and 11 were tentatively assigned to the following two-photon resonant ionizations of Fe atoms by comparing the laser photon energies with the available atomic transition energies.20 N = 1 (366.8 nm): Fe(a3F2) Fe(x5Da2) Fe+ N = 1 I (369.5 nm):
-
Fe(a3F3)
-
Fe(x5Do,)
-
Fe’
Results and Discussion For simplicity, the electronic terms of Fe and Fe+ are indicated by i”(l”, 2”, 3”, ...) andj+ (I+, 2+, 3+, ...), respectively, throughout this paper. Table I summarizes several low-lying electronic terms of Fe and Fe+ together with their electronic configurations and energies. MPI Ion-Current Spectra. MPI ion-current spectra observed for Fe(Cp),, Fe(Acac),, Fe(CO)>,and FeCI, under similar laser
These assignments will further be confirmed by the photoelectron energy measurements described later. Spectra c and d in Figure 2 observed for Fe(C0) and FeCI,, consisting of many broad bands, are in striking contrast to spectra a and b. For such broad bands, it is not immediately obvious which species contribute to the ion-current spectra, atomic or molecular species. However, the similarity in the broad bands between the two spectra c and d may provide a clue to this question. Since there are no common molecular fragments expected from Fe(CO)j and FeCI,, it is suggested that the broad bands are also attributable to Fe atoms. The origins of the broad bands are discussed later in more detail on the basis of photoelectron data. It should also be mentioned that a significant background signal appears in spectrum c in Figure 2. From our previous photoelectron spectroscopic work, the background signal has been in-
(18) R. E. Severs, .I. W. Connolly. and W . D. Ross, J . Gas Chromatogr., 5, 241 (1967).
(19) Y . Nagano, Y. Achiba, and K. Kimura, J . Phys. Chem., in press. ( 2 0 ) C. Corliss and J. Sugar, J . Phys. Chem. R e j Dura, 11, 1 3 5 (1982).
UV MPI of Volatile Iron Complexes
The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 1291
2’
32 1 05
0 2 01 0 0 5
K (eV) Figure 4. Photoelectron spectrum observed at the MPI ion-current line N = 1 1 of Fe(CO)5,showing bands due to the two-photon ionizations of producing 1’ and 2’ (J+ = 9 / 2 and 7/2) from 3”. The resonant intermediate state is Fe(x5DoJ;J’ = 4).
1’
where hu is the photon energy (3.35-3.38 eV), I the first ionization energy of atomic Fe (7.90 eV),23E(i”) the energy of the initial state of the atomic Fe, and EG3) the energy of the ionic state above the ground-state ion.
Considering the fine structure levels of i f fand j3, the photoelectron energies ( K ) are evaluated to be 2.06-2.30 eV for the 1” and 1.08-1.31 eV for the three-photon ionization 1+ three-photon ionization 3+ lf’, consistent with the experimental values 2.1 and 1.2 eV, respectively. The three peaks N = 3, 5, and 7 of spectrum a in Figure 2 can thus be interpreted in terms of the three-photon ionizations of producing 1 + and 3+ from the neutral ground-state 1”. The reason the ionic state 2+ is not formed in spectra a-c in Figure 3 may be explained by considering its electronic configuration as follows. The ionic states 1+ and 3+ are the same in the electronic configuration (3d64s), but different from 2+ (3d7). Roughly speaking, the electronic configurations of the resonant intermediate states allowing the formations of 1’ and 3+ should inhibit the production of 2’. It should also be mentioned that a pair of photoelectron bands due to 1+ and 3+ is observed in He I (58.4 nm) photoelectron spectra of atomic Fe by Dyke et aLZ3 The N = 8 ion-current peak in Figure 2a has almost the same intensity and sharpness as the N = 3 peak, a fact suggesting three-photon ionization of 1”. The photoelectron spectrum obtained at this ion-current peak is shown by spectrum d in Figure 3, indicating a single peak at 1.8 eV. This may well correspond to the calculated photoelectron energy (1.81-2.08 eV) expected for the three-photon ionization of producing 2+ from 1”. Interestingly, the 2’ formation observed here is not followed by formation of 1’ and 3’. Such remarkable differences in the production of 1+, 2+, or 3’ may be probably due to differences in the electronic configurations of the resonant intermediate states concerned. Let us next discuss the photoelectron spectra observed at the ion-current lines N = 1 and 1 1. Essentially the same photoelectron patterns were observed at these ion-current peaks. A photoelectron spectrum obtained at N = 1 1 is shown in Figure 4, in which three bands appear at 0.38, 0.14, and 0.08 eV. The photoelectron energies evaluated for the two-photon ionizations of producing 1’ and 2’ from 3’’ are compared with the experimental energies in Table 11. As seen from Table 11, reasonable agreements have been obtained between the experimental and calculated photoelectron energies. Therefore, from photoelectron energy analysis, it is concluded that the ion-current peaks N = 1 and 1 1 are due to the two-photon ionizations of producing 1 + and 2+ from 3“. It should be mentioned here that the photoelectron bands at 0.14 and 0.08 eV may correspond to the S = 9 / 2 and 7 / 2 levels of 2+. The simultaneous formation of 1’ and 2’ suggests that the production of 2’ is due to nondirect ionization. This will be discussed in more detail elsewhere.24 Furthermore let us discuss the photoelectron spectra observed at the ion-current lines N = 6, 9, and 10 in Figure 2, which are relatively intense in spectrum b but very weak in spectrum a. The resulting photoelectron spectra are compared in Figure 5 , showing a band at 3.1 eV for N = 6 or 9 and a band at 3.4 eV for N =
(21) R. Hippler, H.-J. Humpert, H. Schwier, S. Jetzke, and H. 0. Lutz, J . Phys. E , 16, L713 (1983). (22) K. H. Welge, J . Phys. E , 15, 1663 (1982).
(23) J. M. Dyke, B. W. J. Gravenor, R. A. Lewis, and A. Morris, J . Phys. E , 15, 4523 (1982). (24) Y. Nagano and K. Kimura, to be published.
--
1+
3 2 1 0.5 0.2 K (eV) Figure 3. Photoelectron spectra observed at the MPI ion-current lines (N = 3 , 5 , 7 , and 8) of Fe(Cp),. The photoelectron bands corresponding to the ionic states are indicated by lt, 2+, and 3+.
terpreted in terms of Fe atoms populated in various higher excited states.I9 MPI Photoelectron Spectra. Photoelectron spectra were measured at the main ion-current peaks (N = 1-1 1) at the laser wavelengths summarized in Table 11. Photoelectron spectra obtained for Fe(Cp), at the four ion-current peaks N = 3, 5, 7, and 8 in Figure 2 are shown in Figure 3. Each of the spectra a-c in Figure 3 shows two photoelectron bands at 2.1 and 1.2 eV, while spectrum d shows a single peak at 1.8 eV. The most prominent result observed in spectra a-c in Figure 3 is that the two photoelectron bands appear with a separation of 0.9 eV. This energy separation suggests that these two bands may correspond to the 1+ and 3+ ions which are,pssibly formed. This suggestion may be rationalized by the follo*ig consideration. Since no photoelectron bands appear in the2egion above the one-photon energy (hu = 3.4 eV), we do not need to consider any continuum-continuum transition2’ or autoionization ladder22in the present MPI processes. Therefore, the two photoelectron bands appearing at 2.1 and 1.2 eV in Figure 3 strongly suggest that the corresponding ionic states should be lower than 3’ and 6+, respectively. From Table I, we may conclude that only the combination of 1+ and 3+ among the energetically allowed ionic states gives rise to the observed energy separation of 0.9 eV. The photoelectron energy to be observed is calculated by the relationship K = nhu - I E(i”) - EO’+) (1)
+
-
Nagano et al.
1292 The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 I
~
1‘
‘ a)NIL L
32105 0 2 K (eV)
Figure 5. Photoelectron spectra observed at the MPI ion-current peaks ( N = 6 , 9, and 10) of Fe(Cp),, each showing a band due to 1’.
10. As seen from Table 11, the three-photon ionization of producing 1’ from 2” gives rise to a photoelectron energy of 3.0-3.2 eV, which is consistent with the experimental photoelectron energies. In Figure 2b the line widths of the N = 1 and 11 ion-current peaks are about twice as large as those of the other sharp peaks. Power broadening occurs more easily in the one-photon than in the two-photon resonant ionization transitions. This suggests that the latter sharp peaks are due to two-photon resonant three-photon ionizations, whereas the former broader peaks are due to onephoton resonant two-photon ionizations (a3F xSDo Fe’). Therefore, such information on the line width supports that the N = 6, 9, and 10 ion-current peaks are due to three-photon ionizations, since their line widths are half of those for the N = 1 and 11 peaks. From the photoelectron assignments, the main ion-current peaks of Fe(Acac)3 have been assigned to the threephoton ionizations of 1” and 2” as well as the two-photon ionizations of 3”. Let us further consider the origins of the ion-current peaks N = 2 and 4 observed in both spectra c and d in Figure 2. The photoelectron spectra obtained at these ion-current peaks for Fe(CO), are shown by spectra a and b in Figure 6, largely differing in bandwidth from those in Figures 3-5. The possible photoelectron energies calculated here for several two-photon ionizations j’( lf-3+) i”(4”-13”) are shown by horizontal lines in the upper part of Figure 6, for comparison with the observed spectra on the same energy scale. As seen from the comparison in Figure 6, band I with a maximum at 1.23 eV may be regarded as a i”(4”-13”). congestion of several two-photon ionizations 1’ Bands I1 and 111 may correspond to the two-photon ionizations 3+ 12” (1 3”) and 3’ 9”, respectively. From a similar photoelectron analysis, it has been found that other ion-current peaks observed for Fe(CO)5 are also attributed to two-photon ionizations occurring from the excited Fe atoms distributed between 4” and 13”. A series of the weak photoelectron bands indicated by 3’ 6” in Figure 6b may be interpreted in terms of three fine-structure of the 3+ state produced by twolevels (J’ = 7 / 2 , 5 / 2 , and photon ionizations from 6” ”,( = 2), since the energy resolution in such a low energy region is high enough to distinguish the J+ levels. It is also interesting to notice that septet states (6”, IO”, and 12”) are produced in the multiphoton dissociation of Fe(CO)5 in addition to quintet states (I”, 2”, etc.) and triplet states (3”, 5”, etc.). The formation of septet states has been found for the first time in the present work, in contrast to the UV multiphoton dissociation study of Karny et a1.I6 and the vacuum-UV photolysis study of Horak et al.25
- -
-
-
-
-
-
I
1
1
32
I
1
1
0.5
1
1
1
I
0.2
K (eV) Figure 6. Photoelectron spectra observed at the MPI ion-current peaks N = 2 and 4 for Fe(CO)5. The calculated photoelectron energies are shown schematically by horizontal lines in the upper part, in which the ionizations to 1’ and 3’ are shown by solid lines and those to 2’ are shown by broken lines on the same energy scale as the photoelectron
spectra. Distribution of Fe Atoms in Excited States. From a comparison of the MPI ion-current spectra, it has been found that Fe(Cp)2 and Fe(CO)5 are two extreme examples which give rise to the remarkably different patterns in the ion-current spectra. The present photoelectron spectroscopic analysis clearly indicates that the remarkable differences in the ion-current spectra are due to the large differences in the distribution of Fe atoms among the excited states. In one extreme case of F ~ ( C P ) Fe ~ , atoms are produced predominantly in the ground state (1”), whereas in the other extreme case of Fe(CO)5, Fe atoms are broadly distributed among the excited states up to 13”. The case of Fe(Acac), is an intermediate between the two extreme cases in the sense of the distribution of Fe atoms. Many other Fe complexes are probably in between the two extreme cases. The reasons no sharp peaks due to the three-photon ionizations of 1” appear in the ion-current spectrum of Fe(CO)5 may be explained by the fact that the amount of Fe atoms populated in lower electronic states is relatively small. This has been confirmed in the following way. Increasing the laser power ten times higher than usual, we have obtained the MPI spectra shown in Figure 7 for both Fe(Cp), and Fe(CO)5. The ion-current peaks corresponding to the N = 3, 5, and 7 peaks shown in Figure 2 appear in the spectrum of Fe(C0)5. Role of Ligands in Multiphoton Dissociation. From MPI ion-current spectra, Fisanick et al.’ have reported that identical atomic resonant peaks are observed for Cr(C0)6,Cr(C0),(C6H6), and Cr(C6H& with characteristic intensities of the precursor molecules. This result has been successfully explained in terms of differences in the metal-ligand bond dissociation energies. In the present work, at first we attempted to explain our results by a similar idea based on the dissociation energy differences. The adiabatic dissociation energies of the metal-ligands bonds are estimated to be nearly 6 eV from the photodissociation and thermodynamic data available for Fe(CO),,2S,26Fe(Cp)2,27,28 and ( 2 5 ) D. V. Horak and J. S . Winn, J . Phys. Chem., 87, 265 (1983). ( 2 6 ) F. A. Cotton, A. K. Fischer, and G.Wilkinson, J . Am. Chem. SOC., 81, 800 (1959).
The Journal of Physical Chemistry, Vol. 90, No. 7, 1986 1293
UV MPI of Volatile Iron Complexes
I
367.5
I
367.0 Wavelength (nm)
Figure 7. MPI ion-current spectra observed at a high laser power in the region 366.7-367.5 nm for (a) Fe(Cp), and (b) Fe(CO)5. The numbering is the same as in Figure 2.
F e ( A ~ a c ) , . , ~The dissociation energy of the FeCI, molecule is evaluated to be 10.7 eV from its thermodynamic data.30 Therefore, it is difficult to explain the observed difference in the excited-state distribution of Fe atoms in terms of differences in the adiabatic dissociation energies. Instead of the above consideration based on the adiabatic dissociation energies, we want to propose a mechanism in which the excited state energy of a parent complex is rapidly transfered to the vibrational modes of the ligands. The dissociation of the metal-ligand bond seems to occur from a ligand-to-metal charge-transfer state or ligand-field states above the dissociation limit. It is considered that energy transfer competes with the bond dissociation, and the rate of the energy transfer largely depends on the density of the vibrational states. In the series FeCl,, Fe(C0)5, Fe(Acac),, and Fe(Cp),, the number of low-frequency bending modes increases with the total number of vibrational modes. Therefore, roughly speaking, the density of the vibrational states increases with the number of vibrational modes. In the present work, we have evaluated the ratios of the number of ligand vibrational modes to the number of metal-ligand stretching modes in the Fe complexes. The ratios thus calculated for FeCl,, Fe(CO)5, Fe(Acac),, and Fe(Cp), are 2.0 (2.0), 5.4 (5.4), 22.0 (18.0), and 28.5 (23.5), respectively, where the values in parentheses are those obtained when the C-H stretching modes were excluded. The rate of the energy transfer to the central Fe atom is expected to be inversely proportional to these ratios. The observed tendency of the excited-state distribution of Fe atoms seems to be consistent with this expectation. In earlier flash photolysis studies of Fe(C0)531and Fe(Cp)*,,, it has been reported that Fe atoms in l”-l2” are detected in optical (27) J. A. Connor, Top. Current Chem., 71, 72 (1977). (28) G. Wilkinson, P. L. Pauson, and F. A. Cotton, J . Am. Chem. SOC., 76, 1971 (1954). (29) I . K. Igumenov, Mater. Vses. Semin. ‘Str., Suoistua Primen. P-Diketonatou Met.”. 2nd 1976. 45-9 (1978) (in Russian). (30) JANAF Thermochemical ‘Tables,‘2nd ed, Nhtl. Stand. Ref Data Ser., Natl. Bur. Stand., 5 , 37 (1971). (31) A. B. Callear and R. J. Oldman, J . Chem. SOC.,Faraday Trans. 1 , 63, 2888 (1967). (32) B. A. Thrush, Nature (London), 178, 155 (1956).
absorption spectra for flashed Fe(CO),, while only those in 1” and 2” are observed for flashed Fe(Cp),. These results are consistent with the present results of multiphoton dissociation. In general, two processes of UV multiphoton dissociation processes are considered.I0 One is stepwise process, in which a portion of the excitation energy is spent to eliminate a certain metal-ligand bond during an interval of sequential photoabsorption. The other is an explosive process, in which simultaneous stripping of all the ligands occurs after multiphoton absorption. Both processes depend on the rates of up-pumping and energy transfer to nuclear motion, competing with each other. The up-pumping rate is proportional to the nth power of the photon density for simultaneous n-photon absorption. Therefore, which process (explosive or stepwise) is dominant depends on the photon density. Yardley et al.33have indicated that UV photodissociation of Fe(CO)5 occurs efficiently at 193 and 248 nm to produce all energetically possible Fe(CO), ( n = 1-4) fragments. This has been interpreted with a nonstatistical model of fast ligand elimination. In the MPI study of Fe(CO)5 in the 320-280-nm region, Whetten et aL5 have estimated the upper limit of the dissociative lifetime of the photoexcited Fe(CO), to be 0.6 ps by comparing the ion currents of Fe+ and Fe(C0)5+. From these results, it is suggested that multiphoton dissociation of Fe(CO)5 would be a stepwise process rather than explosive. In the present MPI study, no parent ions have been observed for all the Fe complexes studied in the 460-360-nm region. This fact strongly suggests that all multiphoton dissociation processes concerned here are stepwise under the present laser power conditions (