J. Phys. Chem. 1988, 92, 6914-6919
6914
Triplet-Tftpbt Fluorescence and Spin Polarizatlon. rn-Xylylene Type BIradicals A. Despres, V. Lejeune, E. Migirdicyan,* Laboratoire de Photophysique MolPculaire du CNRS,t Biitiment 21 3, Universiti Paris-Sud, 91 405 Orsay Cedex, France
and W. Siebrand Division of Chemistry, National Research Council of Canada,$ Ottawa, Canada KIA OR6 (Received: July 18, 1988)
The fluorescence decays of the whole series of methyl-substituted benzyl monoradicals and m-xylylene biradicals have been measured between 4.2 and 77 K. While the fluorescence decay is a single exponential for monoradicals, it is a sum of two exponential functions for biradicals. The lifetimes are in the range of 30-200 ns for the fast component and 180-1400 ns for the slow component. In the presence of a magnetic field, the monoradical decay remains unchanged, whereas the biradical decay is drastically modified. The biexponential decay is attributed to the fluorescence from the triplet sublevels of m-xylylene biradicals. This requires that the fluorescence from the triplet sublevels be faster than the spin-lattice relaxation between the different sublevels. This interpretation is supported by the magnetic field effect on the decay curves of biradicals. The triplet sublevels decay with different rates because intersystem crossing (ISC) is competing with the radiative process. Purely electronic spin-orbit coupling (SOC) is responsible for the slow decay from one sublevel. Spin-orbit-vibronic coupling is involved in the fast decay from the two other sublevels. The fact that the ISC rates from two sublevels are found to be nearly equal can be traced back to the localized character of SOC in these planar biradicals. It is predicted that nonexponential decay is a characteristic property associated with triplet-triplet fluorescence.
Introduction
In recent years, much information has become available from theoretical, photochemical, and spectroscopic studies on m-xylylene1.2 (m-quinodimethane), the prototype of aromatic nonKekulC biradicals. This species as well as its methyl-substituted derivatives can be photolytically generated in low-temperature matrices and observed directly by electron paramagnetic resonance (EPR)’” and optical s p e c t r o ~ c o p y . ~ ~ ~ We have previously reported**9a spectroscopic and theoretical study on the whole series of methyl-substituted m-xylylenes. The quasi-line fluorescence and excitation spectra of these biradicals in Shpolskii matrices at 5-10 K have been obtained by site-selective laser experiments. The electronic symmetries, transition energies, and oscillator strengths have been determined by the self-consistent field molecular orbital configuration interaction (SCF-MO-CI) method?*’0 The calculations indicate that m-xylylene and its methylated derivatives have a triplet ground state. This assignment has been confirmed by EPR experiments for m - ~ y l y l e n e ~and - ~ for 3methyl-m-xylylene.’ In the Mulliken axis convention, the planar C, biradicals have a triplet ground state of B2 symmetry and the two lowest excited triplet states have A, and B2 symmetries. These triplets are close in energy so that their ordering is theoretically uncertain. Moreover, this ordering is sensitive to methyl substitution on the ring. Calculations show that the first excited triplet is A, for all the biradicals except m-xylylene and 2-methyl-mxylylene, which have B2 symmetry. The first electronic transition is therefore 2’B2-13B2 for the two latter biradicals and l3AI-l3Bz for the others. The calculated oscillator strengthsg are about 5 X IO-’ and 30 X lo-’ respectively for B2-B2 and AI-B2 transitions. The biradicals have emitting levels of either B2 or A, symmetries and are therefore expected to have significantly different fluorescence lifetimes. This prediction could be checked by measurements of fluorescence decay. In a recent paper,” we have described decay measurements performed on the triplet-triplet fluorescence of the perprotonated 3-methyl-m-xylylene biradical (M-h,,) and its perdeuteriated homologue (M-d,,) produced from mesitylene-hI2 and -d12in Shpolskii matrices. For both species, the decay curves can be analyzed as a sum of two exponential functions in the whole AssociE B I’UniversitE Paris-Sud. Issued as NRCC No. 29474.
temperature range where the biradical fluorescence is detectable. A significant deiterium effect is observed on the lifetimes of the two components. For a given fragment, these lifetimes are independent of variations in excitation and observation wavelengths, temperature, and local environmental conditions. They are, however, drastically altered in the presence of a magnetic field. One possible interpretation is to attribute the biexponential decay to the fluorescence from two individual sublevels of the lowest excited triplet of biradicals. This requires that the fluorescence decay from the triplet sublevels is faster than the spin-lattice relaxation between the different sublevels. In the present paper, we will extend the fluorescence decay measurements to the whole series of methyl-substituted m-xylylenes. If the above interpretation is correct, the fluorescence decays of all the biradicals are expected to be nonexponential since all the measured emissions correspond to triplet-triplet transitions. Experimental Section
n-Pentane and n-hexane (Merck UVASOL) were used without further purification. Impurity luminescence could not be observed when these solvents frozen at 10-20 K were photolyzed and excited under conditions identical with those used for mixed crystals. The purest samples commercially available of the methyl-substituted benzenes were the same as those used previo~sly.~.~ The m-xylylene biradicals and benzyl-type monoradicals were generated in situ by UV photolysis (unfiltered 200-W Osram Hg lamp) of methyl-substituted benzenes (concentrations ranging from low3to M) dispersed in Shpolskii matrices such as n-pentane or n-hexane frozen at 10-20 K. The samples were mounted on ( 1 ) Platz, M. S. In Diradicnls; Borden, W. T., Ed.; Wiley-Interscience: New York, 1982; pp 195-258. (2) Berson, J. A. In Chemistry of Functional Groups;Patal, S . , Rappoport, Z., Eds., in press. (3) Wright, B.; Platz, M. S. J . Am. Chem. SOC.1983, 105, 628. (4) Goodman, J. L.; Berson, J. A. J . Am. Chem. SOC.1985, 207, 5409. (5) Haider, K.; Platz, M. S.; Despres, A,; Lejeune, V.; Migirdicyan, E.; Bally, T.; Haselbach, E. J . Am. Chem. SOC.1988, 110, 2318. ( 6 ) Migirdicyan, E. C. R. Hebd. Seances Acad. Sci. 1968, 226, 156. (7) Migirdicyan, E.; Baudet, J. J . Am. Chem. SOC.1975, 97, 7400. (8) Lejeune, V.; Despres, A.; Migirdicyan, E. J . Phys. Chem. 1984, 88, 2719. (9) Lejeune, V.; Despres, A,; Migirdicyan, E.; Baudet, J.; Berthler, G.J . Am. Chem. SOC.1986, 108, 1853. (10) Baudet, J. J . Chim. Phys. Phys.-Chim. Biol. 1971, 68, 191. (1 1) Lejeune, V.; Despres, A.; Fourmann, B.; Benoist d’Azy, 0.;Mi?irdicyan, E. J . Phys. Chem. 1987, 91, 6620.
:022-3654/88/2092-6914$01 SO10 0 1988 American Chemical Society
Triplet-Triplet Fluorescence and Spin Polarization
The Journal of Physical Chemistry, Vol. 92, No. 24, 1988 6915
T=MK
1
TABLE I: Observed Lifetimes and Ratio AJA, of the Initial Intensities of the Two Components Determined from the Fluorescence Decays of m-Xylylene Biradicals in Shpolskii Matrices'
\ biradical
theor lifetimes, ns
0
I
2
I
I
I
3
t(CLS)
Figure 1. Fluorescence decays of duryl-d13monoradicals (upper curve) and D-d12biradicals in two different sites of n-hexane at 14 K. The biradical decays are excited with laser lines at 4438.5 (A) and at 4449 A (B) coincident with the fluorescent origins. They are measured on the corresponding vibronic bands at 4535.5 (A) and at 4546.3A (B). The monoradical decay is excited at 4444.5A and measured on the fluorescence origin at 4865.2A.
a holder inside an Air Liquide cryostat with helium gas used as coolant. The excitation source was a home-built tunable dye laser pumped with a (Lambda Physik M 2000) nitrogen laser. Coumarin 440 and 480 dyes from Exciton Co. were used in the dye laser. The pulse width at half-maximum was 6 ns. Fluorescence was analyzed with a T H R 1500 Jobin-Yvon spectrometer. The signal was detected by a XP 2020 photomultiplier (RTC) whose output was fed to a LeCroy 9400 digital oscilloscope equipped with signal-averaging WPOl software. The decay curves, averaged over 5000 pulses, were analyzed by the Simplex rnethod.l2 Emission from the fragments trapped in specific sites of the Shpolskii matrix was excited with laser band widths of 4-6 cm-'. The decays were measured at selected spectrometer wavelengths with narrow spectral widths. The cryostat was placed between the poles of a magnet which, with the pole spacing of 110 mm, produced a maximum magnetic field of lo00 G. The field strengths were determined within 10%.
Results The UV photolysis of methyl-substituted benzenes in Shpolskii matrices at low temperatures gives rise to benzyl-type monoradicals together with 0-and/or rn-xylylene biradicalar species. These fragments result from the photodissociation of CH bonds on one or two different methyl groups of the parent molecule. They can be generally distinguished by the energies and vibrational structures of their excitation and fluorescence spectra. In this paper, we are concerned with the fluorescence decays of m-xylylene biradicals excited with a dye laser in the 440460-nm wavelength region. This radiation cannot excite the o-xylylenes, which absorb at shorter wavelengths (410-420 nm), but they produce fluorescence from monoradicals whose electronic energy levels are lower than those of biradicals. Fluorescence from the two species is always excited by a laser wavelength that is coincident with a strong biradical excitation band. Fluorescence from one species is isolated from that of the other by a spectrometer tuned to strong emission bands. The spectrometer band width is 2-20 cm-' depending on the fluorescence intensity. The fluorescence decays are measured in the absence and in the presence of a magnetic field. Fluorescence Decays in the Absence of a Magnetic Field. The decay characteristics of mono- and biradicals produced from all (12) Demas, J. N. In Excited State Llfetime Measurements; Academic: New York, 1983; Chapter 5 , pp 80.
T~
A2/A1
(30)
390
(0.25)
3000
100
285
0.65
'7'
35
200
0.55
50
260
0.65
230
(90)
460
(0.65)
90
105 130
730 645
0.65 0.65
200 235
1370 1070
0.55 0.55
120
45
230
0.55
1000
110
300
0.65
90
160
340
0.80
90
80
185
0.55
80
95
185
0.65
180
I
T,
600
3 I
obsd lifetimes, ns
*V'
'The theoretical lifetimes are calculated for To
-
TI transitions.
the methyl-substituted benzenes dispersed either in n-pentane or in n-hexane at 10-20 K are qualitatively the same as those described in our previous paper" for mesityl monoradicals and M-hlo and M-dto biradicals. As typical examples, the decays of the duryl-d', monoradical and 2,4-dimethyl-rn-xyly1e11e-d~~ (D-dlz) biradical produced from perdeuteriated durene in n-hexane at 14 K are displayed in Figure 1. This semilogarithmic scale shows clearly that the decay of the monoradical is exponential while that of the biradical is nonexponential. The decay curves of biradicals have been analyzed as a sum of two exponential functions. We eliminate spurious contributions having apparent lifetimes of 10 ns, as determined by a dummy scatterer. The plots of the weighted residuals versus time for the Poisson-weighted least-squares fit are randomly distributed, which indicates that the biexponential model describes our experimental results correctly. The observed lifetimes and the ratio Az/ALof the initial intensities of the two components determined for all the m-xylylene biradicals are listed in Table I. The lifetimes of the slow ( T 2 ) and fast (7') components are determined within uncertainties of respectively 7 and 20%. As for the M-hlo and M-dlo biradicals previously studied, a significant deuterium effect is observed on the decay components of D-d12 biradicals when (D-hI2). compared to those of 2,4-dimethyl-m-~ylylene-h~~
Despris et al.
6916 The Journal of Physical Chemistry, Vol. 92, No. 24, 1988
5
Y
c
-I
i 1 2 t(iLs) Figure 2. Magnetic field effect on the fluorescence decays of the monoradical (F3)and biradicals (F,and Fz)produced from pentamethylbenzene in n-hexane at 15 K. The field strengths are 0 (a), 220 (b), 400 (c), and 1000 G (d).
0
We have found that the decay characteristics of m-xylylene biradicals are essentially independent of temperature in the whole temperature range where the fluorescence can be detected. Let us now consider the dependence of the fluorescence decays on the specific lattice site where a biradical is trapped. The insert in Figure 1 shows the structure of the fluorescence origin band of D-d12biradicals in n-hexane at 17 K under broad-band excitation. This sharp and multisite structure results from the interaction of biradicals with their environment, leading to a distribution of transition enmgia that is characteristic of the n-hexane Shpolskii matrix. Figure 1 presents the fluorescence decays of D-d12 biradicals excited with laser lines coincident with the sharp origin bands at 4438.5 (curve A) and a t 4449 A (curve B) with a spectral width of 14 cm-I. The two decay curves measured on the corresponding vibronic bands at 4535.5 (A) and at 4546.3 A (B) are slightly different, leading to component lifetimes of 200 and 1370 ns for curve A and of 235 and 1070 ns for curve B. Similarly the decay curves measured for D-hlz biradicals trapped in two distinct sites are different. The component lifetimes are 105 and 730 ns for one site and 130 and 645 ns for the other site. This shows that the relaxation law of D-hlz and D-d12biradicals is controlled by their environment in the Shpolskii matrix. For all the other m-xylylene biradicals, we have found that the decay curves are independent of the site, within the experimental errors. Fluorescence Decays in the Presence of a Magnetic Field. The fluorescence decays of all the methyl-substituted m-xylylene biradicals and the corresponding benzyl-type monoradicals have been studied as a function of magnetic fidd from 0 to lo00 G. In all the cases, we have found that the biexponential decays of biradicals are drastically modified, whereas the exponential decays of monoradicals remain unchanged upon application of the field. These observations suggest that the fluorescence decay curves as well as the magnetic field effects can distinguish between mono- and biradicals. These two criteria are verified when the parent molecule gives rise to only one monoradical and one biradical which are spectroscopically well-identified. They will be particularly useful in ambiguous cases when several fragments produced from a given molecule cannot be clearly assigned as monoand biradicals by the energies and vibronic structures of their fluorescence spectra. For example, the photodissociation of one or two CH bonds from different methyl groups in pentamethylbenzene leads to the formation of three tetramethylbenzyl monoradicals and three trimethyl-m-xylylene biradicals. Only three fluorescence spectra F,,F2, and F3 with origin bands at respectively 4510,4653.7,and 5089 A (see Figure 2 of ref 9) have been detected out of the six expected emissions, thus creating some ambiguity in the assignments. The correlations between calculated and observed transition energies9 suggest attribution of spectra Fl and F, to bi-
I
I
1 T
1 0
T
500
1 IO00
Gauss
Figure 3. Magnetic field effect on (a) the observed lifetimes of the slow (72) and fast (71) components and (b) the fraction of the initial intensity Al/(A, + A,) of the fast component of M-d,,, biradicals in n-pentane: (0)sample 1 studied at 4.2 K; (A) sample 2 studied at 15-20 K; (0) sample 1 annealed by warming to 80 K and freezing to 16 K.
radicals and F3 to a monoradical. These attributions are defmitely confirmed by the fluorescence decays displayed in Figure 2. The decay of F, is exponential with a lifetime of 450 f 10 ns, while the decays of F1 and F2 are biexponential with substantially different component lifetimes: 160 and 340 ns for F, and 80 and 180 ns for F2. In addition, the magnetic field has no effect on the decay of F,, while it perturbs drastically the decays of F, and F2. This is illustrated in Figure 2,where the decay curves of F1 and F2 corresponding to 750 G are not presented since they are superimposed on those corresponding to 1000 G. Magnetic field studies of emissions from individual triplet sublevels to the ground state have to be carried out on species oriented in single crystals. Unfortunately, we have not yet been able to produce m-xylylene biradicals in single-crystal hosts. All our samples contain randomly oriented biradicals in polycrystalline Shpolskii matrices. In such systems, the relaxation law is expected to become multiexponential (see Discussion). For convenience sake, however, the decay curves obtained in the presence of a magnetic field are still fitted with a sum of two exponential functions. Each apparent lifetime might well be the average of a distribution. The magnetic field dependence of the fluorescence decays has been studied in detail for M-dIoin n-pentane at 4.2-20 K. We have collected in Figure 3 the results obtained in three sets of experiments performed with two different photolyzed samples and one of these samples annealed by warming up to 80 K and refreezing down to 16 K. The variations of the slow T~ and the fast T~ component lifetimes as a function of magnetic field are presented in part a of the figure. The uncertainties involved in each value are given by the length of the bar used. As the field increases to 300 G, T~ decreases drastically while T~ remains practically unchanged. Between 300 and 1000 G, the variations of T , and
The Journal of Physical Chemistry, Vol. 92, No. 24, 1988 6917
Triplet-Triplet Fluorescence and Spin Polarization
with calculated energies for both singlet and triplet states is depicted in Figure 4. As far as state symmetries are concerned, the diagram of 2-methyl-m-xylyleneis analogous. For all the other C, biradicals, the order of the energy levels is similar to that presented in Figure 7 of ref 11 for mesitylylene whose TI state has Al symmetry. The triplet states of m-xylylene biradicals are split into T*, T?, and T , components. For m-xylylene and mesitylylene, the zerofield splitting (ZFS) parameters determined by EPR*$ are IDl/hc = 0.01 1 cm-' and J E ( / h c2 0.001 cm-' in To and not known in T1. For all the other biradicals, the ZFS parameters have not been determined in either To or T1. In the C, point group, the three triplet spin functions T ~ T,,,, and T , belong to the B2, B1,and A2 irreducible representations, respectively, whereas the singlet spin function is of AI symmetry. The total symmetries of the Toand T1sublevels are given III Figure 4 for m-xylylene and 2-methyl-m-xylylene and in Figure 7 of ref 11 for all the other biradicals. Theory predicts that only 7,(TO) T,(T?), ry(T0) T,,(T,), and 7,(TO) * T ~ ( T Jtransitions are electronically allowed and the corresponding transition moments are equal. In the absence of spin-orbit coupling (SOC) and nonradiative decay, the fluorescence emitted from the T1 state is expected to be exponential. The corresponding lifetime can be evaluated from the approximate relationship T ( S ) = 1.5/f3 where v is the transition energy in cm-' and f the oscillator strength calculated for the first electronic T1 To transition? These theoretical lifetimes are listed in Table I for all m-xylylene biradicals. They are not correlated with the lifetimes of the fast or slow components which in turn are not correlated with each other. By analogy with the quinoxaline and pyrazine model, the biexponential decay of biradicals would be consistent with fluorescence from independent T1 sublevels at a rate faster than the rate of equilibration by spin-lattice relaxation. Furthermore, the decay must not occur with the same rate for each sublevel populated in the To TI excitation. A number of temperaturedependent processes such as the direct and the Raman processes16are responsible for the increase of the sublevel equilibration rate by spin-lattice relaxation with the temperature T. As a result of these processes, the quinoxaline and pyrazine nonexponential decays turn into exponential decays of an equilibrated sublevel population above 4.2 and 10 K, respectively. The fact that nonexponential decay persists at higher T for pyrazine is simply due to its shorter phosphorescence lifetime (18 ms for pyrazine, 237 ms for quinoxaline). If the sublevel equilibration rates in our biradicals are like those of quinoxaline and pyrazine, we would expect our nonexponential decays to persist to much higher temperature. The persistence is purely due to the fast (30-1350 ns) decay of the spin-allowed biradical T1 Totransition. The biradical fluorescence remains nonexponential even a t 77 K. The consistency of the quinoxaline-pyrazine model with the biexponential decay of m-xylylene biradicals is supported by the drastic modification of the decay curves in the presence of a magnetic field. Upon application of the field, the wavefunctions describing the TI sublevels get mixed." The depopulation rate of each mixed sublevel becomes a linear combination of the three zero-field rates, with weights proportional to the squared mixing coefficients and thus depends upon the orientation of the molecular axes with respect to the field direction. The relevant algebra has been presented for the radiative rates of pyrazine phosphorescence.18 It applies equally well to nonradiative rates from triplet states in a magnetic field, provided that the molecules are randomly oriented and that spin relaxation is negligible. The striking feature is an acceleration of the slow decay. Since most of the admixture between the slow and fast rates is apparently obtained for field intensities around 300 G (see Figure 3), the ZFS pa-
-
I
Figure 4. Energy level diagram of m-xylylene biradical with calculated energies of both triplet and singlet states. The order of T ~ 7,,,,and T, sublevels is arbitrary. 7 2 are difficult to determine because of the large uncertainties as indicated by the bars. The field dependence of the fraction of initial density A1/(A1 A2) for the fast component is presented in part b of the figure. Within the uncertainties of our data, this fraction remains unchanged as the field increases.
+
Discussion The triplet character of the ground state of m-xylylene and its methyl-substituted derivatives has been established by SCFMO-CI calculations9 and confirmed by EPR experiments for both m-~ylylene*~ and 3-methyl-m-xylylene.3 The correlations between spectroscopy and theoryg indicate that the fluorescence of all the biradicals correspond to triplet-triplet transitions. The component lifetimes listed in Table I are consistent with transitions that are spin-allowed. The present results show that all biradicals have biexponential fluorescence decay that is drastically modified by the presence of a magnetic field. These new findings have common features with o b ~ e r v a t i o n s ~made ~ J ~ many years ago on the TI So phosphorescence of aromatic compounds. In both cases, the nonexponential decays can be related to the triplet character of the emitting state. Specific examples of nonexponential decays occur in the phosphorescence of quinoxaline in durene at temperatures lower than 4.2 K13 and of pyrazine in cyclohexane below 10 K.I4 Both decays become exponential either by increasing the temperature or by applying a magnetic field. These results have k e n explained as due to competition between spin-lattice relaxation and phosphorescence among the three spin sublevels of the lowest triplet state. In our previous paper," we have used this model to explain the decay dynamics of M-hlo and M-dlo biradicals. Let us now examine whether such a picture can be generalized to the whole series of methyl-substituted m-xylylenes. The planar C, m-xylylene biradicals have a triplet ground state To of B2symmetry, in Mulliken's convention. The first excited triplet T1 has AI symmetry for all the biradicals except m-xylylene and 2-methyl-m-xylylene in which T1 has B2symmetry. Three and possibly four singlet states are predicted to lie below the emitting triplet T1.l5 The energy level diagram of m-xylylene
-
-
-
-
~~~
(13) De Groot, M. S.; Hesselmann, I. A. M.; Van der Waals, J. H. Mol. Phys. 1967, 12, 259. (14) Hall, L.;Armstrong, A.; Moomaw, W.; El-Sayed, M. A. J. Chem. Phys. 1968, 48, 1395.
-
~~~
(1 5) Lejeune, V., unpublished results. (16) Waller, I. Z . Phys. 1932, 79, 370. (17) Van der Waals, J. H.; De Groot, M. S. The Triplet State; Zahlan, A. B., Ed.; Cambridge University Press: 1967; p 101. (18) Hall, L.; Owens, D.; El-Sayed, M. A. Mol. Phys. 1971, 20, 1025.
6918
DesprEs et al.
The Journal of Physical Chemistry, Vol. 92, No. 24, 1988
TABLE II: Grwp Table indicating the Representations of the Siglet States S, Coupled to TI,of the SOC Operator Component Rp ( p = x , Y, or z)and of the Out-of-PlaueMode vt Inducing the Coupling TI P Ro S" vk
B2 B2 B2 B2 B2 AI
Y
BI
z
'42
X
AI
z
AI AI AI
X
B2 BI A2 B1 A2 B2 B1 A2
Y z
Y Y z
SOC integral as L(0,) f L(0,). The sign depends on the symmetry of the electronic wave functions relative to the mirror plane uy between the two methylene groups: ( 1AlIHso13BZ)=
- LA(O2)
AI AI
a2
bl
(1B21Hso13B2)= LB(01) + LB(02)
B2 B2 B2 B2
bl
A1 AI
The angles 0, and O2 are governed by out-of-plane vibrations of the methylene groups, which typically appear in pairs of modes of comparable frequencies, one being symmetric (namely, b,) and one antisymmetric (a2) relative to uy;hence 0, = f 4 for bl and 0, = -e2 for a2, so that
bl
a2 a2
bl
[('AllHso13B2(Ty))la2= LA(e) - LA(-0)
a2
rameter ID1 in TI is of the order of 0.03 cm-I, which corresponds to the pure Zeeman term. The D value in TI is therefore of the same order of magnitude as IDllhc N 0.01 1 cm-' in T0.3-5 Spin-Orbit-Vibronic Mechanism. This interpretation applies only if the TI sublevels decay with different rates. However, as pointed out above, in the absence of spin-orbit coupling (SOC), their rates should be equal. On general physical grounds we expect SOC in these planar aromatic radicals to be much too small to affect the T1 Totransitions directly. In ref 11 we have therefore suggested that the biexponential decay might be due to a competing spin-forbidden process induced by SOC, namely, tripletto-singlet intersystem crossing. Symmetry dictates that SOC mixes only singlet and triplet levels belonging to the same representation of the molecular point group. For the biradical represented by the diagram of Figure 4, this means that only the T~ triplet level can mix with singlet states, specifically with S1and S3that have energies lower than T1. This might suggest that 7, is subject to fast intersystem crossing while T~ and T~ decay more slowly by radiative transition to To. However, this interpretation would lead to an A2/A1 ratio of about 2, whereas the observed ratios are close to 0.5. To resolve this apparent contradiction, we note that purely electronic SOC is expected to be very weak in planar aromatic biradicals such as m-xylylene. In their pioneering study, Salem and Rowlandlg have indicated which factors influence this SOC. They have shown, in particular, that it is proportional to the degree to which the two singly occupied atomic orbitals are nonparallel. In m-xylylene, these two orbitals have pure ?r character and are perfectly parallel, leading to SOC as small as in aromatic hydrocarbons. The accepted dominant mechanism of SOC in these hydrocarbons involves vibronically induced coupling between ?r and o orbitals, the inducing modes being out-of-plane vibrations. We therefore expect that a similar vibronic mechanism will be dominant in biradicals. In particular out-of-plane bending or twisting of the methylene groups will lead to nonparallel ?r orbitals as well as to overlapping between R and o orbitals. Both of these mechanisms may induce SOC. In the biradicals considered, out-of-plane vibrations belong to the representations a2 and bl of the C2, point group. As shown in Table 11, they can induce SOC of TY(A2)and T~(B,)with &(AI) and S3(A1)as well as with S2(B2) and S4(B2), but they cannot induce SOC of T ~ ( Awith ~ ) any of these singlets. Thus if vibronic SOC is much stronger than purely electronic SOC, we expect that T~ and 7, will be subject to rapid intersystem crossing, whereas T~ will decay radiatively to To,a complete reversal of the situation resulting from purely electronic SOC. In principle, we would now expect T ~ T, ~and , 7, to decay each with a different rate constant, leading to triexponential decay. The observation of biexponential decay is compatible with the observation that A2/A1 is close to 0.5 and indicates similar decay rates for T~ and 7,. To explain this similarity, we assume that SOC in these biradicals is largely localized in the methylene groups. Denoting the local SOC integral of such a group, associated with an outof-plane bend or twist angle 0 by L(0), we can express the total +-
= LA(0) - LA(e) [(1AIIHso13B2(7~))Ib, [ ( 'B21ffsoI3B2(7Z)) l a , = L B ( ~+ ) LB(-~)
[('B21H~01%(7~))1b,= L B ( ~+ ) LB(@
If the SOC is between a ?r orbital, which changes sign upon reflection through a,, and a u orbital, which does not, we have L(0) = -L(-0); if, on the other hand, it is between two 7r orbitals, we have L(0) = L(4). These two possibilities lead to the following results: T U - SOC: [('AIIHsoI3B2(7y))Ia2 = 2LA(@ [ ( ' B ~ I H W I ~ B ~= ( ~~LB(O) ~))I~~ ?r*
- SOC: [ ( 'B2Iffsol3B2(7,))la2
2LB(0)
[('B21H~13B2(7y))lb,= 2LB(0)
The other terms vanish identically. It follows that only the latter alternative offers an explanation for the observation A2/A1 c! 0.5. If vibronic ?rc SOC were dominant, we would expect only the T~ level to decay rapidly, leading to A2/AI = 2. The observed values are mostly in the range 0.55-0.65and would be compatible with only a small contribution of TU SOC relatively to the dominant 7r?r SOC. A similar argument applies to the biradicals with a Tl(A,) excited state. In that case the dominant TU SOC will be with A, singlets. In all biradicals considered, there is a singlet state of the same symmetry as TInearly degenerate with it. Of course, this does not prove that this singlet state dominates the SOC with TI. In principle, one might expect the magnitude of the deuterium effect on the observed rate constants to give an indication of the energy gap between the coupled states. In the two biradicals for which deuteriated samples were available, both T~ and 72 increase significantly upon deuteriation. Since the vibronic coupling integral required may itself be subject to a deuterium effect?O this does not prove that the coupled S, and TI states are separated by a substantial gap. There is no obvious reason why 7 2 should have roughly the same deuterium effect as T ~ If. the T~ level decays purely radiatively, 7 2 should be insensitive to deuteriation. The observed sensitivity of 7 2 to deuteriation suggests that the To TIemission is in competition with To TI internal conversion. The large Tl-To energy gap would then indicate that the radiative decay channel remains dominant; otherwise a much larger deuterium effect would be expected. In summary, we have shown that triplet-to-triplet fluorescence, observed in aromatic biradicals, shows a temporal behavior different from the conventional singlet-to-singlet and doublet-todoublet fluorescence, since the triplet sublevels do not communicate on the timescale of the fluorescence. As a result nonexponential
--
+-
~
(19) Salem, L.;Rowland, C . Angew. Chem., Int. Ed. Engl. 1972, 2, 92.
(20) Lawetz, V.;Orlandi, G.;Siebrand, W. J . Chem. Phys. 1972.56, 4058.
J. Phys. Chem. 1988, 92, 6919-6923
decay will be observed if there is a competitive nonradiative process with different rates for different sublevels. This is a general conclusion that should not be limited to the aromatic biradicals considered here. Registry No. D2, 7782-39-0; m-xylene biradical, 32714-83-3; 2methyl-m-xylene biradical, 100859-21-0;3-methyl-m-xylene biradical,
6919
57384-02-8; 4-methyl-m-xylene biradical, 90046-34-7; 4,6-dimethyl-mxylene biradical, 90046-30-3; 2,5-dimethyl-m-xylene biradical, 900463 1-4; 4,5-dimethyl-m-xylene biradical, 90046-32-5; 2,4-dimethyl-mxylene biradical, 100859-22-1; 2,4,6-trimethyl-m-xylenebiradical, 1 16996-88-4; 2,4,5-trimethyl-m-xylene biradical, 116996-89-5; 2,4,5,6tetramethyl-m-xylene biradical, 100859-23-2; 2,4,5-trimethylbenzyI monoradical, 15220-27-6.
Oxidation of Small Boron Agglomerates: Formation of and Chemiluminescent Emission from BBO T.C. Devore,+J. R. Woodward,* and J. L. Gole* High Temperature Laboratory. Center for Atomic and Molecular Science, and School of Physics, Georgia Institute of Technology. Atlanta, Georgia 30332 (Received: October 20, 1987; In Final Form: April 26, 1988)
A source configuration that lies intermediate to a low-pressure effusing molecular beam and a high-pressure flow device is used to generate boron cluster molecules in a highly oxidizing environment. Using this source operating in an NO2 or N20oxidative environment, we generate a chemiluminescentemission spectrum, which we attribute to the asymmetric BBO molecule. The observed spectrum is characterized by a strong Av = 0, Av = 40 cm-I sequence grouping and a weaker Au = +1 sequence (Au = 40 cm-I), 440 cm-’to higher energy. A second sequence with Av 142 cm-’is also observed. Combining the 440-cm-’ upper-state frequency with the 142-cm-’ sequence structure implies a lower-state frequency of -582 cm-I for the B-B stretch, consistent with ab initio calculation.
-
Introduction Because of its high heat of combustion per kilogram, boron represents an attractive high performance fuel’ whose use may substantially improve the performance of volume-limited vehicles. However, in order to realize this potential, both the heterogeneous gas-surface and homogeneous gas-phase boron chemistry associated with particle ignition, flame stability, and combustion efficiency must be addressed. Investigations of the combustion of boron slurries and large particles2 and of boron atomss5 have now been reported. Unfortunately these impressive studies neglect the bridge between the combustion chemistry of boron particles and the oxidation of compounds containing a single boron atom. There is growing evidence that the clustering of reactant and product species and the reactions of these clusters may make a significant contribution to the combustion chemistry2and hence influence the ability to extract the potential energy content of boron fuels. In an assessment and investigation of this important intermediate cluster regime, it will be necessary to determine the identity and signatures of those species contributing to the combustion reactions, thus learning more about their chemical and physical properties. Recently, we reported preliminary results from the investigation of the oxidation of boron atoms and boron cluster molecules with NO2 and N20? The optical signature of a new species formed from the oxidation of boron clusters was tentatively associated with the B 2 0 molecule. This paper presents the results of a more complete analysis of the observed band system whose location is ideally suited for further kinetic analysis of boron cluster oxide combustion. We present evidence that the obse~edemission should be correlated with the molecule BBO as opposed to the thermodynamically more stable BOB. The vaporization of mixtures of the heavier group 13 metals and M203is known to produce M 2 0 molecules.6 The infrared spectra of A120, Ga20, In2O, and T1207have been extensively investigated in rare gas matrices. From this work, it is known that each of the “equilibrium” generated molecules has a symmetric structure. A120 is linear, while the heavier compounds in this series are bent. In contrast, little is known about B20. Permanent address: James Madison University, Department of Chemist ?Present address: General Electric Lamp Division, Cleveland, OH.
.
Theoretical calculations indicate that, like A120, the linear symmetric structure, BOB, is the most stable configuration.8-I0 The BBO structure is calculated to be less stable by 100 kJ mol-’. +lo However, since these calculations also show an activation barrier of 125 kJ mol-’ for interconversion of the isomers, it is reasonable to expect that either isomer could be formed in a kinetically controlled reaction and subsequently maintained due to the high barrier to interconversion. For example, the oxidation of BZ, via sideways attack, forming a ringlike intermediate
-
-
B I
B
;o
would be expected to produce the more stable B-0-B isomer. Also, since B-O-B is linear, the visible emission spectrum should show a progression in the bending mode. In contrast, end-on oxidation to give a B-B-O intermediate favors formation of the dynamically stable BBO isomer. Any progressions, if present, should be largely localized to stretching modes. Hence a vibrational analysis of the “B20” band can be used to indicate which (1) Burdette, C. W.; Lander, H. R.; McCoy, J. R.J. Energv 1978,2, 289. (2) King, M. K.; Combust. Sci. Tech. 1972,5,155. King, M. K.Combust. Sci. Tech. 1974,8, 255; Mohen, G.; Williams, F. A. AIAA J. 1972,10,776. King, M. K. J. Spacecr. Rockets 1982, 19, 294. (3) Parker, W. A.; Clark, A. H.; Vamper, M. Prog. Asrronaur. Aeron. 1964, 15, 327. Miller, W. J. Aerochem TP, 1976, 348. (4) Woodward, R.; Le, P. N.; Temmen, M.; Gole, J. L. J. Phys. Chem. 1987, 91, 2637. ( 5 ) Green, G. J.; Gole, J. L. Chem. Phys. Lert. 1980,69,45. Hanner, A. W.; Gole, J. L. J. Chem. Phys. 1980, 73, 5025. (6) For example see: Cubucciotti, D. High Temp. Sci. 1969, 1, 11. Drowart, J.; Dcmaria, G.; Burns, R.P.; Inghram, M.G. J . Chem. Phys. 1960, 32, 1366. Porter, R. F.; Schissel, P.; Inghram, M. G. J. Chem. Phys. 1955, 23, 339 and references therein. (7) For example see: Douglas, M. A.; Hauge, R. H.; Margrave, J. L. High Temp. Sci. 1983, 16, 35. Lynch, D. A., Jr.; &he, M. J.; Carlson, K.D. J . Phys. Chem. 1974,78,236. Makowiecki, D . M.; Lynch, D. A., Jr.; Carlson, K. D. J. Phys. Chem. 1971, 75, 1963 and references therein. (E) Dekock, R. L.; Barbachyn, M. R. J. Inorg. Nucl. Chem. 1981, 43, 2645. (9) Zyubine, T. S.;Charkin, 0. P.; Zyubin, A. S.; Zakzhevokii A. S . Zh. Neorg. Khim. 1982,27, 558. (10) Gina, H.; Jones, L.; Shillady, D., private communication.
0022-365418812092-6919$01.50/0 0 1988 American Chemical Society
