2644
J. Phys. Chem. 1991, 95. 2644-2652
dropped to 56% of its maximum value. As noted earlier, the Liu and Dykstra red-shift predictions also appear to go to zero for R of this magnitude.24 Thus these small clusters indicate that the Ar matrix red shift may be accounted for by the nearest neighbors alone, with little or no contribution from atoms past the first coordination shell.
V. Conclusions The first rotationally resolved, high-resolution infrared spectra of Ar,HF, Ar3HF, and Ar4HF have been recorded in the H F stretching region near 3950 cm-I. Combined with the earlier observation of ArHF, these data allow us to examine the red shifting of the H F stretching motion as a function of complex size and geometry. By n = 3, the red shift of the Ar,HF is half way to the matrix value of 3919 cm-’ from the uncomplexed value of 3961 cm-l. We find that the magnitude of the red shift depends sensitively on the position of the Ar atoms with respect to the HF dipole moment. Hence the largest effect occurs for the addition of the first Ar atom in a linear ArHF geometry. With the addition of a fourth Ar in a position equivalent to that of an atom in a second solvation shell, the red shift increases only 2.2% beyond the value for Ar3HF. This radial dependence to the red shifts suggests that only the nearest neighbors of the HF contribute significantly to the vibrational red shift of the H F stretch and thus some inert gas condensed-phase properties may be well approximated in relatively small clusters which close the first solvation shell. No line broadening beyond the instrumental limit is observed for any of the spectra, indicating a long lifetime (516 ns) for the H F vibration in these clusters. The relatively high signal to noise ratios observed for these large clusters offer the opportunity to extend the considerable high-
resolution work on two-particle van der Waals complexes into a regime where the transition to condensed-phase properties may be investigated. These single-mode IR laser experiments have the advantage of rotational resolution which allows the determination of cluster geometries and provides a probe of the temperature of the expansion. This last point may prove particularly important since previous slit jet experiments have shown essentially complete equilibration of internal and rotational degrees of freedom. This equilibration indicates that the rotational temperature is likely a good probe of the internal “temperature”, an important parameter in many molecular dynamics simulations of clusters. To extend this work, we are now investigating the applicability of the H F v = 1 ArHF pair potential of Nesbitt et al. to these larger clusters. Combined with the observation of combination bands of the H F stretch plus intermolecular vibrations, these calculations should provide insight into the importance of three-body forces and additivity of two-body potentials. A clear focus of these computations will be to attempt to reproduce the free-rotor behavior observed in the octahedral matrix site. To complement this last calculation, we are also pursuing the spectroscopy of still larger clusters in order to explore the transition from the large-amplitude bender of ArHF to the free-rotor behavior proposed in the matrix work.
Acknowledgment. Support from this research by National Science Foundation Grant No. CHE86-05970 through the University of Colorado as well as grants from the Henry and Camille Dreyfus Foundation and the Sloan Foundation is gratefully acknowledged. Registry No. Ar, 7440-37-1; HF, 7664-39-3.
Ultraviolet-Visible and Raman Spectroscopy of Diatomic Manganese Isolated in Rare-Gas Matrices A. D.Kirkwood: K. D. Bier, J. K. Thompson,*T. L. Haslett, A. S. Huber, and M. Moskovits* Department of Chemistry and the Ontario Laser and Lightwave Research Centre, University of Toronto, Toronto, Ontario MSS 1 A l , Canada (Received: June 4, 1990)
Numerous absorption bands in the UV-vis spectrum of matrix-isolated Mnz are assigned to transitions between singlet, triplet, quintet, and septet electronic spin states of the diatomic. These assignments are based on the temperature dependence of the intensities of these absorptions resulting from population redistribution among the low-lying spin states of the Mn2molecule. The exchange energy (J) obtained from the observed temperature dependence ranges from -8 to -1 1 c d . Fluorescence spectra of Mnz in krypton matrices are also reported. The resonance Raman spectrum of diatomic manganese isolated in solid xenon matrices was observed with IZg+ ground-state vibrational constants: w,” = 68.1 cm-’ and w/x,” = 1.05 cm-l.
Introduction
The characterization of small metal clusters is an important link in understanding the chemical and physical properties of the respective bulk materials. Transition-metal clusters are particularly interesting because of their enhanced reactivity due to the presence of d-orbital electrons. Diatomic manganese is unique in that it is a weakly bound van der Waals molecule similar to the group 1IA metal dimers rather than its neighboring first-row transition-metal diatomics. In this paper more of the spectroscopy of dimanganese will be examined. Nesbetl was the first to propose that Mn2 was an antiferromagnetic diatomic with a lZg+ground state based on an a b initio ‘Resent address: Schlumberger Well Services, 5000 Gulf Freeway,P.O. Box 2175, Houston, TX 77252-2175. *Resent addms: Dow Chemical, Patents Department, 1776 Bldg., Midland, MI 48674.
Hartree-Fock calculation with exchange interaction. The Heisenberg exchange Hamiltonian describing such systems can be written as
H = -.IS,&, where Sa and S, are the electron spin operators of the two manganese atoms and J is the exchange energy. The eigenenergies of this Hamiltonian are given to first order by an equation that is similar to the Land&interval expression E(S) = -(J/Z)[S(S + 1 ) - 2s(s + l ) ] Here S represents the total spin of the molecule and s is the atomic spin. Assuming that the bonding in manganese dimer results solely (1) Nesbet, R. K. Phys. Rev. 1964, A135,460.
0022-3654191 12095-2644%02.50/0 0 1991 American Chemical Society
Diatomic Manganese Isolated in Rare-Gas Matrices
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2645
from the antiferromagnetic exchange interaction, one obtains
E(S)
-(J/2)[S(S
+ 1) - 35/21
where the total spin, which ranges from S = 0 to 5, and the quantity 35/2 result from the coupling of the 6S atoms. However, Nesbet's calculations assumed a single su molecular orbital created from d6s' atomic states, resulting in eigenenergies
E ( S ) = - ( J / 2 ) [ S ( S + 1) - 121 with S = 0-4. From this configuration Nesbet predicted that the molecule should be weakly bound (0.79 eV) with a Heisenberg exchange energy (J)of -4.13 cm-l. Some of the most informative work on matrix-isolated dimanganese comes from the ESR experiments of Van Zee et al., and Baumann et aL3 They confirmed that the bonding in Mn, involves exchange coupling. Also, by assuming that the molecule was antiferromagnetic and that the Land6 interval for successive spin state energies was obeyed, they assigned observed bands to transitions within 32and 2 states. From the temperature dependence of the intensity of transitions within the ?2 state, they estimated the Heisenberg exchange parameter J to be -9 i 3 cm-l, which is in rough agreement with that postulated by Nesbet.' Subsequent ESR4 studies performed on Mn, isolated in cyclopropane matrices have shown the existence of 9Z and "L: spin states, indicating that the bonding in the diatomic is totally due to antiferromagnetic exchange interaction. Rivoal et al.5,6assigned Following the earlier ESR two ultraviolet absorption bands (331.5 and 347 nm) to Mn, in an argon matrix through the use of temperature variation and magnetic circular dichroism. By studying the temperature dependencies of the intensity of these bands, they postulated that the 347-nm absorbance (and possibly the 331.5-nmband) originated from the ?2 spin state of Mn,. From this assumption a value of the ground-state exchange energy was found to be -10.3 f 0.6 cm-I. Other absorptions at 375, 380, and 386 nm with different temperature dependencies were also observed and tentatively assigned to transitions between 'Z states. Despite the apparent consistency between the J values of Rivoal et al. and those of Van Zee et al., we call the identification of the carrier of the bands at 331.5 and 347 nm into question. This will be discussed below. DeVore et aL7 were the first to report the visible spectrum of dimanganese in an argon matrix. Here, a banded absorption system centered at 15 400 cm-I (649 nm) was assigned to the A X transition of Mn2 The fine structure of the system, which had an average spacing of 11 1 cm-I, was suggested as being the excited-state vibrational progression. Klotzbiicher and Ozins saw this same system in matrices of both argon and krypton (band center shifted to 14900 cm-' (671 nm)). They also observed numerous nonatomic bands in the ultraviolet and blue regions of the spectrum, some of which showed temperature-dependent intensity variations. More recently, the resonance Raman spectra of Mn2 in both argon and krypton matrices were r e p ~ r t e d . ~In Kr matrices a vibrational l2, progression for the IZ,ground state was seen with q," = 76.4 cm-'. In Ar matrices the spectra were complicated
-
(2) Van Zee, R. J.; Baumann, C. A,; Weltner Jr., W. J . Chem. Phys. 1981, 74,6977. ( 3 ) Baumann, C. A.; Van Zee, R. J.; Bhat, S.V.; Weltner Jr., W. J . Chem. Phys.. 1983, 78. 190. (4) Cheeseman, M.; Van Zee, R. J.; Flanagan, H. L.; Weltner Jr., W. J . Chem. Phys. 1990, 92, 1553. (5) Rivoal, J. C.; Shakhs Emampour, J.; Zeringue. - K. J.; Vala, M. Chem. Phys: Lerr. 1982,92, 313. (6) Rivoal, J. C.; Grisolia, C.; Vala, M. In The Physics und Chemisrry of Small Clusters; NATO AS1 Series B: Phvsics. Vol. 158; Jena, P., Rao. B. K.,Khanna, S. N., Eds.; Plenum: New Ymk, 1987, p 617. (7) DeVore, T. C.; Ewing, A.; Franzen, H.F.; Calder, V. Chem. Phys. Lerr. 1915, 35. 78. (8) KlotzbGcher, W. E.; Ozin, G. A. Inorg. Chem. 1980, 19, 3776. (9) Bier, K. D.; Haslett, T. L.;Kirkwood, A. D.; Moskovits, M. J. Chem. Phys. 1988, 89. 6.
25QX
2 3 W
21,033 19,m 17,030 W A V E NU M BE R S ( c m-' 1
15,030
13,000
Figure 1. Visible absorption spectra of manganese in (A) solid krypton, (B) solid argon, and (C) solid argon but with a higher metal concentration.
due to the presence of several progressions. Two of the progressions based on 59 and 68 cm-l were assigned to the 'Z ground state of the dimer isolated in two different matrix sites. A third progression based on 71 cm-' was assigned to the 32ground state. The dissociation energy of dimanganese has also been estimated by a variety of methods, giving values of De between 0.02 and 0.1 5 eV.IO
Experimental Section The experiments were performed by co-condensing vaporized manganese atoms with inert gas onto a cooled, highly polished aluminum paddle. This was done under ultrahigh-vacuum conditions (=8 X Torr) in order to prevent the codeposition of impurities onto the matrix. Cooling was achieved by means of a DISPLEX closed cycle refrigerator. The Raman spectra were obtained by using a Coherent Model CR-599 dye laser with DCM dye, pumped by a Spectra Physics Ar+ laser for excitation, and collecting the resulting scattered light through a SPEX 1400 double-pass monochromator equipped with photon counting. The instrumentation was interfaced appropriately to a Tektronix 4052 computer. The absorption spectra were obtained through reflectance experiments by replacing the laser with a focused tungsten lamp source. In the case of the absorption temperature difference experiments, spectra were collected by means of an optical multichannel analyzer (OMA) with a SPEX 1877 triple spectrograph (visible region) and a HEATH EU-700 single monochromator (UV region). In the low-temperature absorption experiments the DISPLEX was replaced by a liquid helium transfer system (Heli-Tran). Results and Discussion Absorption Temperature Difference Spectra. Visible absorption spectra of manganese in argon and krypton matrices at low temperature are shown in Figure lA,B. In both of these cases the metal concentration in the matrix is low (only atomic and diatomic Mn present). Since atomic absorptions" only occur above 25 OOO cm-I, these spectra show that the dimer is rich in electronic absorptions. In order to differentiate between absorptions due to Mn, clusters and absorptions due to other sized clusters, one can take advantage of the temperature dependence of the population distribution among the Mn, ground spin states. Provided that the magnitude of the exchange energy is small enough, the pop ulation variation should be substantial, even in the temperature range used in matrix isolation. Since the absorption intensities of manganese atomic transitions and of Mn, do not show any (10) Haslett, T. L.; Moskovits, M.; Weitzman, A. L. J . Mol. Spectrosc. 1989, 135, 259.
(11) Schnepp, 0. J . Phys. Chem. Solids 1961, 17, 188.
Kirkwood et al.
2646 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991
w
u
2
a
m 0:
0 v)
m
a
500
I
B
I
I
I
600
700
WAVELENGTH (nm)
000
1
5I - - -
W
u
z a m a: 0 v)
m
a
1 200
I I I 300 400 WAVELENGTH ( n m ) 1
I
Figure 2. (A) Visible region of the difference between the absorption spectra of Mn2 isolated in solid argon recorded at the temperatures shown and that recorded at 13.5 K. (B) as in (A) but in the ultraviolet region.
temperature dependence, it is possible to determine the exchange energy and assign transitions to particular spin states from the temperature effect on the Mnz absorption intensities by keeping the matrix metal concentration low so as to restrict the matrix content to at most the trimer. To make the fitting of the diatomic manganese bands more reliable, difference spectra were recorded (i-e., differences between spectra recorded a t two temperatures) in order to eliminate the effect of a sloping baseline and to remove temperatureindependent spectral features. The spectra were obtained by first taking the reflectance spectrum of the matrix at the higher temperature, then immediately cooling back down to the base temperature of the displex (13.5 K), and obtaining the other. The difference in absorption was then defined as the log of the ratio of the hightemperature spectrum to that of the base temperature spectrum. An optical multichannel analyzer (OMA) was used as the detector in order to allow rapid data acquisition. This “cycling” procedure was used in order to avoid artifacts due to changes in lamp power and other sources of drift. All studies were carried out below the diffusion temperatures of the matrix hosts so that differences in absorption due to structural changes in the samples would not occur. A series of temperature difference spectra in the ultraviolet and visible regions are shown for manganese in argon, krypton, and xenon in Figures 2, 3, and 4, respectively. Part A in each figure represents the visible wavelength portion of the spectrum while part B shows the UV wavelength portion. The temperature behavior of the spectra of Mn2 in the argon matrix is slightly different from that in the other two matrix gases in the visible region of
300 400 WAVELENGTH ( n m ) Figure 3. As in Figure 2 but for a krypton matrix. 200
the spectra. In the argon matrix changing the temperature also changed the baseline, presumably due to the decrease in intensity of an extremely broad band originating in the far-red portions of the spectrum and extending to approximately 375 nm (see Figure 2A). This particular matrix was more concentrated in metal than the others, and the broad negative red feature most likely arises from a manganese molecule with more than three atoms, possibly the high-spin Mn5 complex.I2 In order to determine the temperature dependencies of the various often overlapping spectral features, the more prominent absorption bands were fitted to Gaussians. Although the absorption band profiles are not Gaussian, it was found that in most cases consistent results could be obtained in this way. The temperature dependence of the bands was assumed to follow the conventional Boltzmann distribution taking into account, of course, the appropriate degeneracy of each level. Hence
+ 1) exp(-1.43879S(S + l)J/(2?7)]/ [ x ( 2 S ’ + 1 ) exp(-1.43879S’(S’+ 1)J/(2T))J [(2S + 1) exp(-1.43879S(S + 1)J/(2T0))]/[C(2S’+ 1 ) X exp(-l.43879S’(S’+ 1)J/(2T0))]J ( 1 )
AA = 4[(2S
where A is the absorbance, J is the exchange energy, F is a scaling parameter that is proportional to the absorption cross section, To (12) Van Zee, R. J.; Baumann, C. A.; Bhat, S.V.;Wcltner Jr., W.J . Chem. Phys. 1982, 76, 5636.
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2647
Diatomic Manganese Isolated in Rare-Gas Matrices
W
u
z a m E
0
w m
a
1
1
500
B
I
1
W 700 I ,
,
600 WAVELENGTH (nm) I
I
1
I
800
I
w V
z a
m 1
U 0 v,
m
a
LL
c 2
a a
a C (r
a Q
I 2 00
I
I
1
I
300 400 WAVELENGTH ( n m) Figure 4. As in Figure 2 but for a xenon matrix.
is the base temperature of the displex ( ~ 1 3 . 5K), and 1.43879 is h / k in units of cm-'/K. The summations in the denominators were for S' = 0-5. The fit was carried out using a nonlinear least-squares program based on the Fletcher-Powell algorithm and provided by Harwell (VA05). An analysis of the expected variation in the intensity of the difference bands with temperature, assuming a value of J = -10 cm-l, predicts that, above 13.5 K, only the intensities of IL: and 'I: absorptions difference bands are expected to decrease with increasing temperature. Difference bands originating from higher spin states increase in intensity with increasing temperature. Although good fits are obtained (as shown in Figure 5 ) , a unique assignment of the bands to their spin states is obviously not possible on the basis of the temperature variation alone. This is because for absorptions belonging to to "I: spin states equally good fits are obtained with more than one assignment, yielding, of course, different values of J in the various cases. Unique assignments are possible if J can be determined even approximately from the temperature variation of, say, a IZ band. This is done in the section below. Assignments are also aided by associating spectral features with similar temperature behavior in different matrices. Table 1 shows the results of fitting the observed temperature dependence of various features in the absorption difference spectra to e 1. The average value of the exchange energy from the sZ and Z spin states is -10.3 cm-l, which is in good agreement with the aforementioned ESR experiment^.^*^ No features could definitely be assigned to transitions involving S > 3. It can be
9
Figure 5. Experimental (points) and recalculated (lines) values of the difference between the absorbance of various bands at the temperatures indicated as the abscissa and 13.5 K: (A) 483-nm band of Mn, in solid Ar assuming a 5Z assignment, (B) 420-nm band of Mn2 in solid Ar assuming a 'Z assignment, (C) 206-nm band of Mn, in solid Kr assuming a 5Z assignment, (D) 399-nm band of Mn2 in solid Kr assuming a '2 assignment, (E) 497-nm band of Mn, in solid Xe assuming a 5Zassignment, (F) 267-nm band of Mn2 in solid Xe assuming a 'Z assignment, (G) 665-nm band of Mn, in solid Xe assuming a '2 assignment, (H) 665-nm band of Mn, in solid Xe assuming a 'Z assignment, (I) 401-nm band of Mn, in solid Kr assuming a 'Zassignment, (J) 401-nm band of Mn, in solid Kr assuming a 'Z assignment.
argued, however, that some of the assignments that give a slightly larger magnitude of J for a 72spin state assignment (e.g., the 261-nm band in xenon matrices) could in fact be reassigned 9Z states. For this assignment a value of J is obtained which is lower in magnitude than for the lower states. This is entirely possible, of course, since the energy of the spin states need not follow the Land6 interval rule strictly. The nonobservance of transitions originating from 92and IIZ bands might also be due to the low occupation of these states a t our temperatures. Even at 40 K (and J = 10 cm-I) the expected relative populations of the 92 and I 1 L : states are 4% and 1% respectively, whereas those of the and 71:states are 29% and 14%. The negative features in the temperature difference spectra, some of which show vibrational structure (Figures 2A, 3A, and spin 4A), can only arise from transitions originating in IZ,or 3Zu states. Fits of the negative bands to eq 1, assuming they are due to transitions between singlets, give values of J whose magnitudes are too low (-4.1 to -8.1 cm-I), whereas assuming transitions between triplets produces values with magnitudes that are too high (-20.2 to -49.9 cm-l). This is shown in the next section to be due to the fact that these bands correspond, at least in part, to overlapping transitions involving both singlet and triplet states. This would lead to a higher apparent value of J or a lower value of J , by assuming that the intensity in that region is due entirely
2648 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991
Kirkwood et al.
TABLE I: Values of (4) Obtaioed by Fitting tbe Temperature Dependence of the Absorptions Indicated, Assuming the Assignments Given
matrix (band, nm) Kr (206) Kr (247) Xe (261)
'Z,
3Z
52
8.7 12.5 10.3 9.4
Kr (258)
Xe (267) Xe (269) Kr (263) Xe (272) Ar Ar Ar Ar Kr Kr Ar Kr
(296) (31 I ) (325) (350) (368) (370) (369) (374) Xe (384) Ar (389) Kr (399) K r (401) Ar (420) Kr (423) Xe (431) Ar (454) K r (452) Ar (483) Kr (497) Xe (497) Ar (517) K r (517) Xe (517) Ar (552) K r (559) Xe (607) K r (632) Xe (639) Ar (649) K r (680) Xe (665) Kr (739) Xe (741) Ar (avd K r (a%)
'2
92
10.8 7.4
7.2, 40.9 9.5 9.5
W 0
z a
11.4 12.8 8.6
8.7 6.5
m
a
0
m
m
7.9
a
8.1, 49.9 7.1 9.4 8.6 9.0 9.6
5.1, 27.9 9.7 12.0 9.3
7.3
13.5 13.3 11.1 9.9 10.5 10.4 10.5 13.4 7.7 13.9
I
IkfX
I
15ooo
I
I
14503 1403.7 WAVENUMBER (cm-')
I
l 3 m y m
Figure 6. The red portion of the visible absorption spectrum of Mn, in solid Xe recorded at the temperatures shown. Peaks 1-6 represent vibrational bands whose intensity variations with temperature were fit. 10.7
12.9 7.4 4.1, 22.1 6.0, 33.4 4.2, 20.2
Xe ( a v d
10.8 11.9 10.4
6.8 7.4 9.3 9.3 9.5
avg
11.1
9.4
to a 'Z or a triplet transition, respectively. This problem can be reduced by performing experiments at lower temperatures where the ' Z dominates. Low-Temperature Absorption Spectra. The negative going difference bands at the red end of the temperature difference spectra (about 670 nm) must correspond to singlet or triplet absorptions. By use of low enough temperatures only the singlet contribution will survive. Accordingly, experiments were camed out with a liquid helium transfer system capable of temperatures down to 6.5 K. Figure 6 shows a portion of the visible absorption spectrum of dimanganese in xenon at various matrix temperatures. It is evident that the central portion of the spectrum (near 6 7 0 nm) increases more rapidly in intensity as the temperature is lowered than the portions below 650 nm and above 690 nm. Therefore, the cental portion of this spectrum is assigned to a 'Z transition. The portion to longer wavelengths is dominated by a triplet transition while the part below 650 nm is likely a 51; transition. The temperature variations of the intensities of the bands were fit to the equation A = F[(2S
I
16030 15500
+ 1) exp(-1.43879S(S + l)J/(223)]/ [x(ZS'+1) exp(-1.43879S'(S'+ 1)J/(27'))] (2)
over the temperature range 6.5-18 K assuming S = 0 and the sum over S'runs from 0 to 5. Corrected intensities were obtained by subtracting a straight-line baseline determined by extrapolating
TABLE 11: Exchange Energy Values Obtained from Absorption Spectra Measured at L o w Temperatures peak no. or spin state spectral region assumed J . cm-l 1 (Mn/Xe) 'Z -1 7.3 2 (Mn/Xe) 'Z -11.2 3 (Mn/Xe) '2 -11.1 4 (Mn/Xe) 'Z -13.0 5 (Mn/Xe) 'Z -8.7 6 (Mn/Xe) '2 -7.0 integrated 735-763 nm (Mn/Xe) 'Z -9.7 highest part =680 nm (Mn/Kr) 'Z -7.6 integrated 735-763 nm (Mn/Kr) 'Z -8.6 between the outer edges of the absorption bands shown in Figure 6. For the band centered at 660 nm (IZ),six of the central peaks due to the vibrational fine structure of the band were fit to Gaussians. The intensities of the resultant bands were then individually fit to the temperature dependence expected of a singlet by using eq 2. The intensity of the band centered at approximately 710 nm was found to exhibit the temperature dependence characteristic of a triplet absorption. It was fit by integrating the intensity of the higher wavelength portion of the band in order to avoid interference with the overlapping singlet band at 660 nm. In krypton matrices a similar procedure was applied to the band centered at 720 nm, which also shows the temperature dependence expected of a triplet transition. The singlet band in krypton matrices was centered a t 680 nm. The exchange energy in this case was determined by monitoring the intensity of the middle portion of the band as a function of temperature. Table I1 shows the results of the above fits. From the table it can be seen that fitting the six central vibrational peaks of the singlet absorption gave Jvalues ranging from -17.3 to -7.0 cm-I. The value of -17.3 cm-I seems to deviate significantly from the others, signaling the fact that at that wavelength significant overlap with an absorption due to a spin state of higher multiplicity occurs. By omitting this value and averaging the others, one finds the exchange energy to be -10.2 cm-' for the singlet ground spin state
Diatomic Manganese Isolated in Rare-Gas Matrices I
'
I
'
I
'
I
'
I
'
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2649
I
I
I
I
I
18000-
I
I
19.5
15000-
-
0
a 0 v
t
-
Iv,
z
0I8500
w
c
18003
17500
17000
16500
16EO I
15xx)
I5300
WAVENUMBER (cm-')
z -
Figure 8. Temperature dependence of the emission spectrum of Mn2 in solid Kr obtained with 514-nm excitation. Temperatures (K) are indicated in the figure.
-cn a
u t
L cn
z W
I-
z Y
'
I
laooo
I
I , 11030
I I I 16033 "I
I
I
1
14ooo
I
*I
Ism
WAVEN UMBER (cm-')
Figure 7. Emission spectra of Mn, in solid Kr resulting from argon ion laser excitation with (A) 5.1 mW of 457.9 nm, (B) 5.1 mW of 467 nm, (C) 5.1 mW of 473 nm, (D) 5 mW of 476 nm, (E) 5.2 mW of 488 nm, (F) 5 mW of 496.5 nm, (G)5 mW of 501.7 nm, and (H) 5.1 mW of 514 nm.
of dimanganese in a xenon matrix. Assuming that the state centered at 710 nm may be assigned to a triplet transition, one obtains a value of J = -9.7 cm-I when the high-wavelength portion of the absorption band is integrated and fit to the appropriate expression (eq 2). Similarly, the temperature dependence of the intensity of the triplet band (at 720 nm) in krypton matrices suggests an exchange energy of -8.6 cm-l, while the temperature-dependent intensities of the singlet band for Mn2 in krypton matrices (680 nm) produced a J value of -7.6 cm-I. All of these values agree well with those obtained from the previously discussed
absorption spectra involving other bands. Fluorescence Spectra. Fluorescence studies were performed for manganese dimer in krypton matrices alone. Figure 7 shows a series of emission spectra between 12 500 and 18 500 cm-' (800 and 541 nm) excited with various visible lines from the argon ion laser. Excitation with the shorter wavelengths produces a fluorescence centered at about 15 000 cm-' (667 nm). As the excitation wavelength is increased, another weak feature located at about 14 500 cm-l (690 nm) appears. These two fluorescences correspond in position to the singlet and triplet absorptions noted in the temperature difference studies. They could possibly arise from relaxation to these excited states followed by emissions to the ground-state manifold. It should be noted that the excitations which give rise to these bands (457.9-496.5 nm) fall upon a negative band in the Kr matrix temperature difference spectra. (This negative feature is more apparent at the lower temperatures in Figure 3A.) With 488-nm excitation yet another fluorescence centered a t approximately 16 250 cm-' (615 nm) is observed. With 496.5-nm excitation an adjacent emission a t 15 950 cm-I (627 nm) also becomes apparent as does a band at 17 700 cm-' (565 nm). Both grow strongly in intensity as one goes to 514-nm excitation. Finally at 514 nm, the band at 16250 cm-' (615 nm) seems to have been replaced by a much broader band at about 16600 cm-I (602 nm). The source of these emissions does not seem apparent from the temperature difference spectra. They are likely transitions between excited states. A temperature study was performed on the emissions at 15 950 and 17 700 cm-l using the 5 14-nm argon ion laser line as the source of excitation (Figure 8). Attempts to fit the temperature dependence of the intensities of these emissions to eq 2, assuming the temperature dependence to be entirely due to population changes in the ground state, produced either poor fits or unreliable values of J or both. For example, for the lower frequency fluorescence, values of J = -19.9 and -6.9 cm-I were obtained if one assumed the lower state to be a singlet or triplet state, respectively. Very poor fits were obtained for the 17 700cm-l emission, yielding values of J = -4.9 and -2.9 cm-'. Clearly, the assumption that the temperature dependence of the intensities of these bands results from the temperature-induced changes in the population of the lower state is incorrect. A possible explanation takes into account both the temperature dependence of the lower state population and other depopulating mechanisms of the fluorescing state. The excited-state manifold, which is populated by the argon laser radiation, is rich in states. Because of this it is possible that there are many state crossings and interstate transitions. The rate of crossing is acutely dependent on accidental resonances that occur in the crossing states; for example, if the two crossing states have vibrational levels that are coincidentally close in energy the rate of interstate crossing will be great. The crossing rate can be altered if the state matching is changed. It is well-known that the energy of electronic states is affected by the matrix environment. Moreover, different states show different
2650 The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 I
I
I
I
Kirkwood et al. 150
I
I
I
I
1
A
i t: i
60000
t
- 30000 0
20030
Lxm 1
22000 2xDo WAVENUMBER (cm-')
21000
24000
2-
Figure 11. Anti-Stokes emission spectrum of Mn, in solid Kr obtained with 5.2 mW of 514-nm excitation. 1.5
I
1
I
I
1
Figure 9. Emission spectra of Mn, in solid Kr obtained with R6G dye laser excitation: (A) 10 mW of 17 300 cm-l (578 nm), (B) 10 mW of 17 350 cm-I (576 nm), (C) 10 mW of 17400cm-I(575 nm), (D) 9.3 mW of 17450 cm-I (573 nm). RAMAN SHIFT (cm-')
Figure 12. Resonance Raman spbctrum of Mn, in solid Xe at 13.5 K obtained with 8 mW of 15 1 IO-cm-I (662-nm) excitation.
Figure 10. Comparison of absorption and emission spectra of Mn, in solid Kr: (A) absorption spectrum showing 675-nm band; (B)emission spectrum obtained with 10 mW of 17450-cm-l (573-nm) excitation.
sensitivity to matrix environment. The expansion of the matrix about the guest molecule as the matrix is heated can change the energies of the excited state and hence the crossing point in the crossing states. This might explain the difference in temperature sensitivity of the intensities of the 17 700- and 15950-cm-' states. The former might have been close to resonance at crossing and the latter less so. Hence, the thermal expansion of the matrix might affect the former more than the latter. Emission studies were also performed using rhodamine 6G dye laser excitation between 17 300 and 17450 cm-' (Figure 9). The two fluorescences at 15 950 and 16600 cm-' are again present with the higher frequency one being relatively much more intense with this excitation. In addition, an extremely sharp and asymmetric band appears at 17 050 cm-I. Because of its uncharacteristic shape in comparison to the other features, it might be suspected that this fluorescence results from a different species. Indeed it corresponds very well with the atomic emission of the 6Dstate of manganese" which occurs at about 17 300 cm-I. Since there are no accessible atomic absorptions in this region (the nearest is at approximately 24600 cm-I), the fluorescence is likely
due to the dissociation of Mn2 to two Mn atoms, of which at least one is in the 6Dexcited state. One of the more interesting emissions is the one that occurs at 14500 cm-I (Figure 9 and lo), which is also a result of R6G laser irradiation. This feature exhibits the exact same fine structure as the excited-state absorprion at 14 800 cm-l (singlet state) as seen in Figure 10. But since it is lower in energy it cannot be a result of a fluorescence from this state. This, plus the fact that in emission the fine structure is that of the lower state, suggests that this process is due to a two-photon absorption followed by an emission to the state that is the upper state of the 14 800-cm-' absorption. This would also imply that the lifetime of the state at 14800 cm-'(676 nm) is relatively long. By adding the band position to that of the corresponding absorption, one can determine that the uppermost excited state involved in this fluorescence occurs a t about 29 300 cm-' (341 nm). This corresponds to be a negative feature in the uitraviolet temperature difference spectra of Mn2 in krypton. Excitation with 514-nm laser light also gave rise to intense "anti-Stokes" emissions (Figure 1I). The more intense of these bands appear at 22 750 and 23 050 cm-' and a group of at least four between 23 300 and 24 500 cm-I. There are also three much weaker bands which appear at about 20 200,20 400, and 20 600 cm-' (495,490, and 485 nm). It should be noted that all of these emissions can only arise from two-photon processes. There is the possibility that the group of these bands between 23 300 and 24500 cm-I corresponds to atomic fluorescence arising as before from the dissociation of Mn2 to excited Mn atoms. (Atomic Mn ('D) emissions" occur at 23 297,23 549,23 720, and 23 819 an-'.)The assignment of the other lines is unclear. Resonance Raman Spctra. Resonance Raman spectra of Mn in argon and krypton matrices have been previously reported. The resonance Raman spectrum of Mn2 in a xenon matrix at 13.5 K is shown in Figure 12. As with krypton matrices, only a single resonance Raman progression was observed, in this case consisting of five lines based on approximately 66 cm-I. A vibrational analysis was performed using the spectrum giving values of a/
a
The Journal of Physical Chemistry, Vol. 95, No. 7, 1991 2651
Diatomic Manganese Isolated in Rare-Gas Matrices 1.2,
*
I
I
I
I
I
1 I
\\
1 0 2 0 3 0 4 0 5 0 6 0 TEMPERATURE (K) Figure 13. Experimental (points) and recalculated (solid curves) tem-
perature dependence of the second Stokes line of the resonance Raman spectrum of Mn, in solid Xe. Curves obtained by assuming (*) population redistribution in the ground-state manifold only; (through points) thermalization in both the ground and excited-statespin state manifolds.
= 68.1 cm-l and w,"x," = 1.05 cm-I. These values indicate that Mn, is a weakly bound molecule, agreeing with the suggestion that it is a van der Waals dimer. The intensity of the third member of the progression is much lower than the others. This is probably due to the fact that the frequency of this feature corresponds precisely to an intense absorption feature in the singlet band of Mn, in xenon matrices at 670 nm (Figure 6, peak 4). Thus, the weakness of this feature is most likely due to a reabsorption of the resonance Raman emission. Temperature studies were performed over a range of 13.5-57.5 K on the resonance Raman spectrum in order to derive an exchange energy value for comparison with what was obtained from the absorption spectra. The temperature dependence of the intensities of the second Stokes line of the resonance Raman spectra were fit by again using eq 2. The observed temperature dependence of the intensity suggests that the ground state involved in the resonance Raman spectrum is either a singlet or perhaps a triplet state. Neither assumption produced a good fit. The better fit was obtained by assuming a IZ, ground state. This produced a value of J of -6.67 cm-I. The fit however (Figure 13) indicated that there was a systematic error between the experimental points and the best fit assuming eq 2. The poorness of the fit may be due to the fact that the resonance Raman intensity does not follow the population of the lower state alone but rather the population of the excited state. If there is rapid thermalization of the population between the resonant excited state and neighboring spin states, then the resonance Raman intensity will be proportional to the product of the fractional population in the lower state and the excited state, i.e.
IRR = FQgrQex
(3)
where F is a proportionality factor and Qer and Qexrepresent the ground-state and excited-state Boltzmann factors, respectively. Using eq 3 (and assuming both ground- and excited-state spin-state energies follow a Landt form), a good fit is obtained to the Raman intensity data as a function of temperature (Figure 13) but with a somewhat lower ground-state exchange energy magnitude than before. The values found were Jgr= -5.50 cm-' and J,, = -61.05 cm-' where the subscripts refer to ground and excited states. This implies that the Mn, molecule is more strongly bound in the excited state, a fact previously suspected in seeing that the ground-state vibrational frequencies are smaller than those of the excited statese9 An equally good fit was obtained by assuming that, in the excited state, thermalization occurs between only two states; the state excited and one c cm-I above it. In that case the best fit is obtained with Jgr = -6.2 cm-I and c = 60 cm-I. The failure to observe resonance Raman signals involving other spin states of the ground-state manifold (though attempts were made to excite into the other parts of the banded absorption system around 15 000 cm-I (667 nm)) is puzzling. This occurred in
krypton matrim as well, while in argon only a weak '2 progression (based on 71 cm-I) was ~ b s e r v e d . This ~ might arise from the aforementioned rapid thermalization. In that case, however, one expects to see relaxed fluorescence from the excited singlet transitions. This was not observed. Alternatively, the lack of resonance Raman in the higher spin s t a t 9 could be due to a state crossing in the excited-state manifold; perhaps the same state that is responsible for the reduction in the population of the excited singlet state. If so, its relaxation is either nonradiative or radiative to a state that is less than 11 000 cm-' below it (the limit of our detection). Comparison of Exchange Energies. In comparing the exchange energies obtained with the ESR experimental values," one should note that in the ESR work the temperature dependence of 5Z signals were studied in order to derive a value for the exchange energy. The ESR intensity was fit as a function of tem rature to an expression similar to eq 2 but multiplied by (1 - e-GxT)/( 1 e-hY/kT) in order to correct for the thermal population of the split magnetic spin states which give rise to the observed transition. This normally works well when the magnitude of J is large compared to the Zeeman splitting. But in this case the Zeeman splitting of about 3 cm-I (9 GHz) is of the same order as the exchange energy. A proper analysis would have taken the actual magnetic field splittings into account for all of the electronic spin states in deriving the partition function. As a result, the observed fits of the data for the quintet state were not very good, thereby placing a large uncertainty on the value of the exchange energy, 9 f 3 cm-'. The MCD value of the exchange energy5 (-10.3 f 0.6 cm-I) which at first glance appears to be in agreement with our values was obtained by monitoring the temperature dependence of an absorption of manganese in an argon matrix at 345 nm that was assigned to a 5Z absorption. The absorption spectra showed this band and another at 33 1.5 nm to be clearly isolated from any other neighboring absorptions. This does not agree well with the spectra reported in this work. The nearest bands we observe in this region of the spectrum (Figure 2B) are quintet states that occur at 350 and 325 nm. However, we see overlapping singlet- or triplet-state absorptions in the same range. The authors of ref 5 suggest that the triplet state was not observed due to the 13 K lower limit of their cryostat. (The singlet state cannot be observed by either MCD or ESR techniques.) However, the singlet state has the greatest population throughout this temperature range. In another publication on the same work6 a more complete spectrum is reported that shows the presence of species that are larger than the diatomic. It is highly possible that the results reported in refs 5 and 6 are actually due to a manganese cluster that is larger than Mn, or perhaps a paramagnetic complex of manganese with an impurity.
+
Conclusion The average values of the exchange energies for Mn2 obtained by fitting the temperature dependence of absorption difference spectra are as follows: for the quintet states -10.8, -1 1.9, and -1 0.4 cm-I for argon, krypton, and xenon matrices, respectively; for the septet states -9.3, -9.3, and -9.5 cm-I. Values obtained from the low-temperature absorption experiments were -7.6 and -10.2 cm-I for the singlet state of Mn, in krypton and xenon matrices, respectively, and -8.6 and -9.7 cm-' for the triplet. All of these values agree with each other quite well. Differences are almost certainly due to experimental uncertainty. Nevertheless, discrepancies in the value of J could arise from the possibility that the Landt interval rule is not perfectly obeyed. The average exchange energy from the absorption experiments (singlets to septets) is -10.0 f 2.0 cm-'. This result is in good agreement with the value of J = -9 f 3 cm-I obtained from ESR experiThe temperature dependence of the intensities of visible and UV absorptions was also used to assign numerous bands of Mn2 in argon, krypton, and xenon matrices to particular spin states of manganese dimer (Table I). The resonance Raman spectrum in the ground-state singlet spin state of Mn, in xenon was also reported from which the
2652
J . Phys. Chem. 1991, 95, 2652-2661
ground-state vibrational frequency was determined to be 68.1 cm-l. This compares well with the vibrational frequency of the singlet state of dimanganese in krypton matrices (76.4 cm-l) as well as that in argon matrices (approximately 59 and 68 cm-’ for the two different matrix sites).I0 All of the above results clearly indicate that Mnz is a weakly bound antiferromagnetic van der Waals dimer.
Acknowledgment. The authors thank NSERC and the Connaught Fund for financial support. M.M. thanks the Canada Council Killam Program for a research’fellowship. T.L.H. and A S H . thank the Ontario Government for graduate research fellowships. Registry No. Mn2, 12596-53-I .
Photochemistry of Cycioaikene+O, Collisional Pairs in a Cryogenic Matrix: Chemical Trapplng of Cycioalkene Oxirane Biradicai Conformers, and Comparison of Product Control for Excitation above and below the NO2 Dissociation Threshokl Donald J. Fitzmaurice and Heinz Frei* Chemical Biodynamics Division, Lawrence Berkeley Laboratory, University of California, Berkeley, California 94720 (Received: June 29, 1990)
Oxygen atom transfer from NOz to cyclohexene and from NOz to cyclopentene was induced by excitation of reactant pairs isolated in solid Ar by light in the wavelength range between 610 and 355 nm. Continuous wave dye, Ar ion, and Nd:YAG lasers were used as photolysis sources, and the chemistry was monitored by Fourier transform infrared spectroscopy. The 0 atom transfer path accessible at wavelengths longer than the 398-nm NOz dissociation limit led to cycloalkene oxide as the only final oxidation product. The reaction threshold was at 610 nm. In the case of cyclohexene + NOz, two cyclohexyl nitrite radical diastereomers were produced concurrently with the epoxide. Infrared analysis based on I8O and 15Nisotopic substitution, visible light induced wavelength-selective photodissociation, and matrix annealing experiments indicate that the two stereoisomers are cyclohexyl nitrite radical chair conformers with an equatorial and an axial C-O bond, respectively. Since these are transient cyclohexene oxirane biradicals chemically trapped by reaction with NO cage coproduct according to the previously established reaction mechanism, it is concluded that cyclohexeneoxidation proceeds along two diastereomeric paths. In the case of the cyclopentene + NOz reaction only a single cyclopentyl nitrite radical stereoisomer is observed, presumably because of the very low barrier to pseudorotation of the pentyl ring. Loss of product specificity upon photolysis of cyclohexene/NOZ/Arand cyclopentene/NOZ/Ar matrices above the NOz dissociation threshold with 355-nm light is evident from the appearance of two additional products aside from cycloalkene oxide, namely, cyclohexanone (cyclopentanone) and 2-cyclohexen-1-01 (Zcyclopenten-1-01). Absorbance growth behavior of these products indicated that the ketone and epoxide originate from the same reservoir of sustained cycloalkene.NOz collisional pairs that reacts at visible wavelengths by large-amplitude 0 atom transfer, while cycloalkenol stems from dissociation of NOz, followed by reaction of free O(’P) with cycloalkene. It is proposed that large-amplitude oxygen atom transfer remains ;he dominant pathway for collisional cycloalkene.NOzpairs excited above the NO2 dissociation threshold. Reaction by NO2 predissociation is operative for reactant pairs with orientation or distances (including separation by Ar host atoms) that make the large-amplitudereaction inaccessible.
I. Introduction We have recently observed product-specific oxidation of small, unfunctionalized alkenes by vibronic excitation of N02.alkene pairs in a cryogenic matrix with red light. Photons at these wavelengths excite NOz to vibronic levels that lie 25 kcal below the 398-nm dissociation threshold of this reactant. cis- and trans-2-butene NOz give exclusively 2-butene oxides as final oxidation products with a high degree of stereochemical Alkenes with a terminal methylene group, such as ethylene’ and isobutylene: gave aldehyde as an additional product. No products that would result from breakup of the carbon skeleton were observed. Concurrent trapping of alkyl nitrite radicals allowed us to map the detailed reaction path of these mild oxidations. Strong evidence was obtained from product stereochemical analysis and photolysis wavelength dependence of the reaction kinetics for transfer of an 0 atom to the C C double bond by vibronically excited NOz to form a transient oxirane biradical. A fraction of these biradicals are chemically trapped by combination with concurrently produced N O cage neighbor to yield the observed alkyl nitrite radicals. Chemical trapping of transient oxirane biradicals in their nascent
+
( I ) Nakata, (2) Nakata, (3) Nakata, (4) Nakata,
M.; Frei, H.J. Am. Chem. Soc. 1989, I l l , 5240. M.; Frei, H. J . Phys. Chem. 1989, 93, 7670. M.; Shibuya, K.; Frei, H. J . Phys. Chem. 1990, 94, 8168. M.; Frei, H. J . Chem. Soc. Jpn. 1989, 1412.
conformation by reaction with NO coproduct is a unique feature of vibronically induced chemistry of NOz*alkenesustained collisional pairs in a matrix, revealing in unprecedented detail the stereochemistry of alkene oxidation paths.5 We have employed this method of controlled 0 atom transfer to cyclohexene and cyclopentene in order to explore in detail the stereochemical path of cycloalkene epoxidation. CyclohexeneN02 and cyclopenteneNOz pairs were isolated in solid Ar a t 12 K, and the photolysis wavelength dependence of the long wavelength visible light induced chemistry was investigated. The reaction was initiated by tuned continuous wave (CW) dye laser radiation, and the chemistry monitored by Fourier transform infrared spectroscopy. Photolysis at wavelengths shorter than 398 nm was also performed in order to compare the chemistry of sustained collisional pairs6 above and below the NO2 dissociation threshold. This comparison is of special interest with regard to product control. In principle, reactant excitation above the lowest dissociation limit permits access to dissociative surfaces of the isolated reactant, in addition to the large-amplitude motion path accessible to the NOZ.alkene sustained collisional pairs from vibronic levels (5) For a rccent review of intermediates of hydrocarbon oxidation, see: Hucknall, D. J. Chemistry of Hydrocarbon Combustion; Chapman and Hall: London, 1985. (6) Frei, H.; F’imentel, G. C. In Chemistry and Physics of Matrix Isolated Species; Andrews, L., Moskovits, M., Us.; Elsevier: Amsterdam, 1989; p 139.
0022-3654/9 1/2095-2652%02.50/0 0 199 1 American Chemical Society
