Quantum Chemical Characterization of Low-Lying Excited States of an

Mar 4, 2010 - ... however, came under renewed scrutiny subsequent to UV pump/IR probe ..... Previous models invoked the presence of an intermediate ...
0 downloads 0 Views 2MB Size
Download PDF
J. Phys. Chem. A 2010, 114, 4289–4295

4289

Quantum Chemical Characterization of Low-Lying Excited States of an Aryl Peroxycarbonate: Mechanistic Implications for Photodissociation† Seth Olsen,*,‡ Dirk Schwarzer,§ Ju¨rgen Troe,§,| and Sean C. Smith*,⊥ Centre for Organic Photonics and Electronics, School of Mathematics and Physics, The UniVersity of Queensland, Qld 4072, Australia, Abteilung Spektroskopie und Photochemische Kinetik, Max-Planck-Institut fu¨r Biophysikalische Chemie, 37070 Go¨ttingen, Germany, Institut fu¨r Physikalische Chemie, UniVersita¨t Go¨ttingen, Tammannstrasse 6, 37077 Go¨ttingen, Germany, and Centre for Computational Molecular Science, Australian Institute for Bioengineering and Nanotechnology, The UniVersity of Queensland, Qld 4072, Australia ReceiVed: March 31, 2009; ReVised Manuscript ReceiVed: January 19, 2010

Recent experiments have revealed the existence of an excited state dissociative mechanism for certain peroxycarbonates, with the demonstration that the lifetime of the excited state matches the picosecond time scale for appearance of nascent carbon dioxide product. The data infer that the photoreaction proceeds via an effectively concerted three-body dissociation within the lifetime of the singlet excited state. Many other arylperoxides decay sequentially via [(aryloxy)carbonyl]oxy radical intermediates on nanosecond-microsecond time scales. Uncertainty as to the lifetime of the excited state relates to the character and the relative energetic ordering of states of the parent molecule, since the spectra and photochemistry imply that low-lying states may exist on each of the aryl, carbonate, and peroxide chemical functionalities. We employ many-body electronic structure calculations to determine the energies and characters of the low-lying valence states of a minimal aryl peroxycarbonate model germane to the above-mentioned experiments, methyl phenyl peroxycarbonate (MPC). Our results indicate that the lowest-lying state is an intrinsically nondissociative aryl ππ* excited state. We identify additional low-lying states that are expected to be dissociative in nature and propose that the time scales observed for the dissociation reaction may correspond to the time scale for transfer of excited state population to these states. Introduction Organic peroxycarbonates with aromatic or conjugated chromophores are useful as thermal initiators of free radical polymerization processes. The photochemical decomposition of these and similar organic peroxides has also been studied in considerable detail, both for fundamental understanding and for providing an alternative means of radical initiation. Two possible mechanisms have been postulated,1 contrasting in terms of the timing of the cleavage of the two labile bonds. The first is a sequential mechanism involving initial cleavage of the peroxide bond to yield (i) an alkyloxy radical R2O• and (ii) an [(aryloxy)carbonyl]oxy radical R1OC(O)O•. This is followed subsequently by decarboxylation of the latter to yield CO2 and the aryloxy radical R1O•. The second postulated mechanism is photochemically concerted, involving simultaneous cleavage of the two labile bonds to yield CO2 and the two complementary radical products. The question as to whether a photochemically concerted or a sequential mechanism operates, and if sequential then the stability and lifetime of the intermediate carbonyloxy radical, is an important issue for unraveling the kinetics of radical initiator efficiencies in polymerization.2 Long-lived (nanoseconds-microseconds) radical intermediates have been † Originally submitted for the ″Walter Thiel Festschrift″, published as the October 29, 2009, issue of J. Phys. Chem. A (Vol. 113, No. 43). * Corresponding authors. Phone: +61 7 3365 2816. Fax: +61 7 3365 1242. E-mail: S.O., [email protected]; S.C.S., [email protected]. ‡ Centre for Organic Photonics and Electronics, School of Mathematics and Physics, The University of Queensland. § Max-Planck-Institut fu¨r Biophysikalische Chemie. | Universita¨t Go¨ttingen. ⊥ Centre for Computational Molecular Science, Australian Institute for Bioengineering and Nanotechnology, The University of Queensland.

observed previously in the sequential photodecomposition of benzoylperoxides,3 coumarin peroxyesters,4 and benzophenone peresters.5 The studies we describe here have been motivated by a series of recent experimental observations of phenyl and naphthyl peroxycarbonates,6 methylfluorenepercarboxylates,7,8 and peroxyphenylacetates9 suggesting (i) that the lifetime of the intermediate is decreased by the introduction of an oxygen or methylene carbon between the aryl chromophore and the carbonyloxy moiety9 and (ii) that this effect can be sufficiently pronounced as to remove the intermediate entirely, rendering complete photodecomposition within the lifetime of the singlet excited state.6,8 Pioneering UV pump/VIS probe experiments on the 9-methylfluorene-9-percarboxylate system (TBFC) by Falvey and Schuster10 recorded a transient absorption feature attributed to methylfluorenyl radical with a rise time of 55 ps. From this it was inferred that fast cleavage of the peroxide bond was followed by decarboxylation of the resulting [(methylfluorenyl)carbonyl]oxy radical on the slower 55 ps time scale, with the implication that a sequential mechanism was operational. On the basis of the stability of methylfluorenyl radical, this was reasoned to represent a lower limit for the photoreaction time scale. This mechanism, however, came under renewed scrutiny subsequent to UV pump/IR probe experiments by Aschenbru¨cker et al.,11 which revealed effectively instantaneous (within the 2 ps resolution of the experiments) production of hot CO2. This implied that the photodissociation should be concerted or, if not, that any assumed initial cleavage of the peroxide bond must happen on a still-shorter femtosecond time scale. Subsequently, femtosecond resolved UV pump/VIS+NIR probe experiments by Abel and colleagues7,9,12 identified transient

10.1021/jp9029166  2010 American Chemical Society Published on Web 03/04/2010

4290

J. Phys. Chem. A, Vol. 114, No. 12, 2010

Olsen et al.

SCHEME 1: Aryl Peroxycarbonates Used in Ultrafast Spectroscopy Studies (TBPC and TBNC) and in This Study (MPC)

absorptions attributed to the carbonyloxy radical, which decayed on a picosecond time scale matching the previously measured appearance of nascent CO2. Coupled with extensive electronic structure calculations of the stabilities and barriers for decarboxylation of the ground state carbonyloxy radicals, this work led to an apparently comprehensive sequential model for the photodissociation mechanism. In this model it was assumed that ultrafast initial fragmentation of the peroxy bond from the S1 excited state would yield a carbonyloxy radical predominantly vibrationally excited in its ground electronic state, which would subsequently undergo decarboxylation (in competition with collisional relaxation) to produce hot CO2 on the measured picosecond time scale. Most recently, however, the interpretation of the photodissociation dynamics has taken another turn as a result of the femtosecond UV pump/IR probe experiments of Reichardt et al.,6 which revealed that the lifetime of the S1 excited state in fact matches the picosecond time scale for appearance of nascent carbon dioxide product with quantum yield close to unity. It may consequently be inferred that, in the systems studied by Reichardt et al., there cannot be any ultrafast fragmentation of the peroxide bond on a femtosecond time scale. The operational mechanism appears to be a concerted three-body dissociation from the excited state, yielding hot CO2 and the two complementary radicals. Reichardt et al. suggest that the previously measured transient absorptions in the visible and near-infrared regions, which were ascribed as electronic absorptions of the ground state intermediate [(aryloxy)carbonyl]oxy radicals, must instead be reassigned as S1-Sn absorptions of the excited state of the parent molecule.6 Reichardt et al. have also reinvestigated the decomposition of tert-butyl-9-methylfluorene-9-percarboxylate and confirmed, by unambiguous assignment of the modes of the resulting methylfluorenyl radical, that full decomposition was complete within the excited state lifetime.8 They have also found similar fast decay for the aryl peroxycarbonates tert-butyl 2-naphthyl peroxycarbonate (TBNC) and tert-butyl phenyl peroxycarbonate (TBPC), which decay with excited state lifetimes of 25 and 6 ps, respectively. The observation of concerted fast decomposition in peroxycarbonates and peroxyesters suggests a mechanism very different from that suggested by data on coumarin peroxyesters,4 benzophenone peresters,5 and dibenzoyl peroxides.3 These compounds decay sequentially on longer (nanosecond-microsecond) time scales, corresponding to decarboxylation of a long-lived aryloxy radical. For coumarin peroxyesters and benzophenone

peresters, there appears to be a more significant contribution from intersystem crossing processes.4,5 In contrast, Reichardt et al. found no evidence to support substantial intersystem crossing for the peroxycarbonates TBPC and TBNC,6 nor for TBFC.8 The current picture regarding mechanism of photodissociation of the peroxycarbonate systems is clouded by a lack of knowledge regarding the nature of the excited states of the parent peroxycarbonates. It is clear that there should be low-lying excited states on the aryl species, the carbonyl (or carbonate), and the peroxide, all of which may lie near the excitation wavelengths employed in the experiments. Furthermore, they are all implicated to some degree in the spectra or the subsequent photochemistry. The absorption spectra suggest dominance by aryl ππ* states in the initial excitation event.9 On the other hand, fast peroxide dissociation implicates nσ* excitations localized to the peroxide,13 while cleavage of a bond in the R position relative to carbonyl immediately suggests involvement of an nπ* state localized on that moiety. Hydrogen peroxide photodissociation occurs following excitation at 266 nm and has been recorded at wavelengths down to 405 nm.14 The nπ* excitation of acetone has a somewhat higher λmax of 280 nm.15 For peroxycarbonates TBPC, TBNC, and MPC, the planes of the aryl and percarbonate moieties are not parallel, and so mixing of aryl ππ* states and carbonoxy nπ* states will not be symmetry-forbidden. Both aryl and carbonate states could interact with the peroxide states, but overlap/proximity arguments suggest that the aryl-peroxide mixing should be smaller than carbonate-peroxide mixing. Given that kinetic models have been proposed on the basis of the assumption that the S1 state is dissociative16 whereas the more recent data implies that this assumption is not correct, clarifying the nature of the low-lying states is essential at this point to move forward with understanding the dissociation dynamics. Following this motivation, the present paper reports the results of many-body electronic structure calculations performed to characterize the low-lying valence states of methyl phenyl peroxycarbonate (MPC, Scheme 1). We report the nature and relative energies of low-lying excited states of MPC and characterize simple dipole-related properties. We then perform a brief characterization of nearby features of interest on the potential surfaces of the molecule to cast further light on the early time photodynamics. It will be seen that the lowest singlet state, populated initially in the photodissociation experiments, is an aryl ππ* state, which does not appear to be dissociative.

Characterization of Aryl Peroxycarbonate

J. Phys. Chem. A, Vol. 114, No. 12, 2010 4291

We find evidence supporting dissociation via a peroxidelocalized nσ* state and further characterize the nπ* state on the intervening carbonate moiety which can lead to R-cleavage. This study constitutes the first reported application of ab initio wave function-based methods to a representative peroxycarbonate system shedding light on the nature of the excited states most likely to be involved in the photodissociation of such molecules. Method Identification and Characterization of Low-Lying Valence States. To clarify the nature and energies of the low-lying valence states of MPC (Scheme 1), we have made use of two different types of many-body electronic structure computation. The first is equation of motion coupled cluster singles and doubles (EOM-CCSD).17 The second is internally contracted second order multireference Raleigh-Schro¨dinger perturbation theory (MR-RSPT2)18 built upon a state-averaged19 complete active space self-consistent field20 (SA-CASSCF) reference space. We used a 6-31G* contracted Gaussian basis set.21 The calculations were carried out at a geometry that was optimized using Møller-Plesset second order perturbation theory (MP2)22 built upon an RHF reference with the same basis set. The MP2 optimization was started from the lowest-energy geometry indicated by a conformer search performed with a molecularmechanics force field. The EOM-CCSD calculation was performed on the RHF reference wave function. Excitations from the core shell of all heavy atoms were not considered. Properties were evaluated as matrix elements between left and right-hand eigenvectors of the similarity transformed Hamiltonian. The reported values are averages between conjugate matrix elements. The excitation energies did not vary between the left- and right-hand spectra of the similarity transformed Hamiltonian. The MS-RSPT2 calculations were built upon model spaces generated using state-averaged23 complete active space selfconsistent field (SA-CASSCF)20 methodology. The active space contained fourteen electrons distributed over twelve orbitals and was averaged over the lowest-lying four electronic states. We will refer to this active space by shorthand as SA4-CAS(14,12). Excitations out of the lowest-lying 17 orbitals were neglected in the MR-RSPT2. We include graphical images of SACASSCF state-averaged natural orbitals and occupation numbers (which specify the wave function) in the Supporting Information. Features of Interest on the SA-CASSCF Surfaces. With the same SA4-CAS(14,12) model space used to determine the states as above, we conducted optimizations on the respective SA-CASSCF state surfaces. Optimizations on the S0 and S1 states converged to nondissociated structures, which we describe below. Optimization on the S2 state led promptly to peroxide dissociation. This optimization was stopped once the behavior was clear and the resulting structure is reported here as a model for an early stage dissociated structure. We also conducted searches for minimal energy intersections (MECIs) between the S1 and S2 states and between the S2 and S3 states, starting from the S0 minimum energy structure. The results of these optimizations are also outlined below. MR-RSPT2 calculations were conducted as above at the geometries generated by the optimizations. It is important to note that the variant of multireference perturbation theory used here is not a truly multistate perturbation theory (i.e., it is of the “diagonalize-perturb” variety). As such, the perturbation contribution to the coupling is not described and therefore the method cannot describe conical intersections. We do not report MR-RSPT2 energies for states involved in an intersection.

Figure 1. Excitation energies and transition dipole strengths for lowlying valence excitations of our methyl phenyl peroxycarbonate model (MPC, Scheme 1). EOM-CCSD results are on the left, MR-RSPT2 results on the right. In the center are the SnS0 difference densities of the MR-RSPT2 states calculated using the reference wave function. An analogous figure, displaying results obtained with a correlationconsistent, polarized double-ζ basis, is included in the Supporting Information. The results there displayed are semiquantitatively identical.

Figure 2. Geometry of the model used to calculate the excitation energies of the low-lying valence states. Bond lengths (Å) are listed next to the bonds they represent on the structure.

All of the data reported in this paper was generated with the Molpro quantum chemistry software.24 Results Identification and Characterization of Valence States. Our results for the low-lying states of methyl phenyl peroxycarbonate (MPC) are summarized in Figure 1. The figure describes excitation energies calculated at the MR-RSPT2 and EOMCCSD levels, and dipole strengths for the S0-Sn and S1-Sn transitions evaluated at the SA-CASSCF and EOM-CCSD levels of theory. In addition, Sn-S0 difference densities, calculated at the SA-CASSCF level, are shown. The geometry used in the calculation of the low-lying states is displayed in Figure 2. Both of the methods used to examine the states agree on the nature and order of the two lowest valence states. The lowest lying state is a ππ* state localized on the aromatic chromophore. The excitation energy obtained via MR-RSPT2 is within a reasonable range of the absorption maximum of the related compound TBPC, which has been the subject of ultrafast spectroscopic experiments.2,6,16,25 The energy for this state predicted by EOM-CCSD is not as close to the absorption band, which is probably because the state in question has significant

4292

J. Phys. Chem. A, Vol. 114, No. 12, 2010

Olsen et al.

Figure 3. Geometries obtained by optimizations on the SA-CASSCF potential energy surfaces. Bond lengths (Å) are listed next to the bonds they represent. The angle (θσπ-90°), a measure of pyramidalization, was calculated as described by Haddon in ref 26.

double excitation character. The MR-RSPT2 and EOM-CCSD calculations also agree on the character of the S2 state, which is an nσ* excitation on the peroxide, as well as on the energy of a higher nπ* state on the carbonyl. As can be seen from the difference density (and also from the orbital expansion), the localization of these excitations is not strict. The EOM-CCSD calculation indicates the presence of a ππ* (S3) excitation between the nσ* (S2) and S4 nπ* (S4) states. Direct participation of the higher-lying ππ* state in the photochemistry is unlikely. The S0-Sn and S1-Sn dipole strengths calculated by the two methods used are also displayed in Figure 1. The S0-Sn dipole strengths are in good agreement between the calculations. The S1-Sn strengths differ. We are inclined to place more trust in the SA-CASSCF dipole strengths in these cases because the calculation of matrix elements of the (one-body) dipole operator between doubly excited configurations will require triple excitations, which cannot be described in the EOM-CCSD calculation. In the context of variational electronic structure methods, expansion of the basis set should provide a better approximation to the exact Born-Oppenheimer electronic energies for the molecule in isolation. A figure analogous to Figure 1, but calculated using a correlation-consistent polarized valence double-ζ basis set, is included in the Supporting Information. The results shown there indicate an identical ordering of the excited states in question, with nearly identical excitation energies, and lead to identical conclusions to those described here. Features of Interest on the SA-CASSCF Surfaces. Relevant features of the SA-CASSCF optimized structures, as well as SA-CASSCF and MR-RSPT2 (where appropriate) energies of the states at these structures are depicted in Figures 3 and 4, respectively. The geometry to which the S1 optimization converged differs from the S0 minimum geometry only by expansion of the bond lengths of the aromatic ring and a slight contraction of the aryl-O bond. This provides support for (albeit not conclusive proof of) the contention that the S1 state at the Franck-Condon structure is not dissociative. The character of the states at this geometry is preserved relative to the S0 minimum. The energy of the S1(ππ*) state is lower as expected. The energies of the

other states are not changed by much, which suggests that the “active” modes of the other states are disjoint from those of the ππ* state. The S1/S2 MECI is distinguished from the S1 optimized geometry by elongation of the peroxide bond. This is consistent with the notion that the nσ* state, which participates with the ππ* state in the intersection, favors dissociation of this bond. The SA-CASSCF energy of the intersection is ∼2.5 kcal/mol above the S1 energy of the S1-optimized structure. The energy of the S3 state has not changed significantly relative to its energy at the S1-optimized structure and is somewhat higher than at the S0-minimum geometry. The S2/S3 MECI differs from the other structures by a pronounced pyramidalization of the carbonate carbon and an elongation of the carbonyl bond. The pyramidalization can be quantified by the π-orbital axis vector (POAV1) defined by Haddon.26 The angle θσπ-90° (which is 0° for one of the carbons in benzene and 19.5° for methane) is evaluated in this case to be 9.9°. Very interestingly, the energies of all three excited states in the ππ*, nσ*, and nπ* triad are nearly degenerate at this point. Furthermore, the characters of the states is inverted, and the nπ* state has assumed the S1 position. The latter behavior raises the suggestion that a three state intersection may not be far away, but adiabatic mixing of the nπ* and nσ* is also plausible. The S2 optimization finished without convergence at a structure with a dissociated peroxide bond. The energies of the S0, S1, and S2 states are quite close at this geometry, with the energy of the S3 state lying slightly higher. Discussion Given that the results of previous femtosecond spectroscopy studies for a number of other arylperoxyesters have been interpreted to support a sequential photodissociation mechanism, a significant implication of the recent experimental results6,8 as well as the present computational study is that the photodissociation mechanism for a certain class of such molecules may require revision. The chemistry (peroxide dissociation, R-cleavage) as well as the femtosecond time-resolved spectroscopic

Characterization of Aryl Peroxycarbonate

J. Phys. Chem. A, Vol. 114, No. 12, 2010 4293

Figure 4. Relative energies of the four states calculated via SA-CASSCF and MR-RSPT2 at the geometries obtained by optimizations on the SA-CASSCF potential energy surfaces. All energies are referenced to the ground state energy of the S0-optimized structure. SA-CASSCF energies are indicated by dashed lines; MR-RSPT2 energies are indicated by solid lines. MR-RSPT2 energies of states that intersect at a particular geometry are not reported. The characters of the states are indicated next to the lines representing them. SA-CASSCF Sn-S0 difference density isosurfaces for nonintersecting states are displayed in line with the energy bars (difference densities are not distinguishable for the S0 and S1 optimized structures, so one set is used for both of these geometries).

data imply the participation of multiple low-lying valence states attributable respectively to the aryl, carbonate, and peroxide moieties in the overall photodissociation. What has been lacking is a demonstration of how the electronically excited states characteristic of localized excitations of each of these chemical functionalities interact to give rise to the electronic states, and by inference the photochemical behavior, of the parent peroxycarbonate or peroxyester molecules. Our results indicate that the lowest electronic state of the system (S1) is a ππ* state localized on the aromatic chromophore. The excitation energy calculated using MR-RSPT2 supports the contention that this is the state populated by the 266 nm excitation (the excitation used in studies of the closely related tert-butyl model).6 We have also identified two other low-lying valence states that are by their nature implicated in the subsequent photodissociation. The first of these is an nσ* excitation (S2), which corresponds to the dissociative state in simpler peroxides.13 The second, higher state is an nπ* state on the carbonyl moiety, which separates the aryl and peroxide moieties. Carbonyl nπ* states often lead to R-cleavage,15 such as would lead to fast production of CO2 in the photoreaction. Our calculations indicate a large energy gap between the initially excited (aryl) ππ* state and the (carbonyl) nπ* and (peroxide) nσ* states. This indicates that the initial excitation populates only the lowest-lying ππ* state. Our brief survey of the SA-CASSCF potential surface described above suggests that the S1(ππ*) state is not dissociative. Geometric changes (Figure 3) suggest that the ππ* state represents a trap for the photoexcited biradical structure, localizing it on the aryl ring. However, SA-CASSCF optimizations on the S2(nσ*) state do lead to O-O bond scission. This is consistent with previous characterizations of the dissociative states of small peroxide models.13 Low-lying aromatic ππ* states can fluoresce, so one

might ask if transient fluorescence has been observed in aryl peroxycarbonates. The answer is yes: transient fluorescence has been observed in TBNC and decayed on a time scale of 51 ps, comparable to that characterizing the dissociation process (55 ps).6 If the initially populated state is not intrinsically dissociative, then the natural question is: how does population of the dissociative states occur? In a sequential mechanism, the biradical structure on the aryl ring makes its way to the peroxide, causing ultrafast dissociation to an alkyl radical and [(aryloxy)carbonyl]oxy radical (possibly in an excited state), followed by decarboxylation on a longer time scale. In a concerted mechanism, spin repairing on the CO2 could lead directly to a biradical interaction between aryloxy and alkyl fragments (separated by molecular CO2). The resulting state correlates with a carbonate nπ* excitation. Since the nσ* excitation is implicated in the ultrafast peroxide dissociation in sequentially decomposing arylperoxides, it seems that all three localized excited states should be included in a model space that can span both sequential and concerted chemistries. In percarbonates such as MPC the plane of the aryl ring and the plane of the carbonate are not parallel, consistent with their union at an sp3-hybridized center. This means that interactions of the π states of the aryl and the n states of the carbonate are not forbidden by symmetry. This point should also apply to percarboxylates such as TBFC. It is conceivable that the accentuation of π-n interactions in these compounds facilitate fast, complete decomposition within the singlet excited state lifetime. Our survey of the SA-CASSCF surfaces indicates that there are modes which lead to a substantial mixing of the spatially disjoint states and which induce intersections of the upper states with the ππ* state. The intersection located at the SA-CASSCF level is ∼2.5 kcal/mol above the optimized structure of the ππ*

4294

J. Phys. Chem. A, Vol. 114, No. 12, 2010

state, hence access of the intersection itself may be possible. One might regard the energy of the MECI as a crude upper limit to a transition state energy for adiabatic passage, although this assumes that the average energy of the intersecting states is not greatly higher than the intersection at the transition state and indeed that the transition state approximation at an adiabatic saddle-point is dynamically appropriate. It appears plausible to suggest that such activated exit from the ππ* state may be the origin of the single time scale observed in the experiments by Reichardt et al.6 This would explain why the time scales for transient fluorescence of TBNC are similar to the time scale for production of CO2. It also can explain the longer decomposition time constant of TBNC relative to TBPC, since substitution of naphthyl for phenyl will lower the ππ* state without substantially affecting the energy of the nσ* state and (invoking an ergodic model for the dynamics) will slow the rate for activated passage from the ππ* due to the greater density of states. Comparison to other peroxyesters is more complicated, because then we expect that altering the nature of the intervening group will alter the coupling between the ππ* and nσ* states and may alter the energy of the nπ* state. Our results suggest that coupling between the aryl ππ* and peroxide nσ* states may arise through mutual projection of the configuration interaction vector onto a local ππ* excitation of the carbonate moiety. This partial carbonate ππ* excitation is consistent with the observed bleach of the ground state CO stretch (1806 cm-1) and appearance of a red-shifted vibration (1799 cm-1) in the excited state at short time scales.6 The isodensity plots in Figures 1 and 4 show that the peroxide nσ* excited state has significant projection onto carbonate excitations. Vibration of the CO bond, and other motions that break the local planarity of the carbonate (for example, rehybridization of the carbonate center) can be expected to modulate couplings and splittings between the participating states. This partially explains the ordering that we find here: the nπ* state is raised by coupling to both aryl and peroxide states which couple less strongly among themselves. The projection of the nσ* state onto carbonyl-localized excitations is consistent with the notion that peroxide dissociation via the nσ* channel would leave any [(aryloxy)carbonyl]oxy product in an electronically excited state. Subsequent development of nπ* character may induce R-cleavage on an ultrafast time scale. Alternatively, strong mixing of the nσ* and nπ* states may lead to concerted peroxide bond dissociation and R-cleavage. In either scenario the outcome is, in terms of the photochemical kinetics, an effectively concerted three body dissociation. Reichard et al. have suggested6 that near-IR transient absorption recorded9 by Abel et al. for the photodissociation of TBNC and assigned to an [(aryloxy)carbonyl]oxy intermediate should be reassigned to S1-Sn transitions of the parent peroxycarbonate. Our results suggest additionally that excited states of the parent peroxycarbonate may occur in this range and that they are indeed dipole-allowed, lending some inferential support to the tentative reassignment of the transient absorptions. Clearly, there is more work to be done to elaborate the essential features of the mechanism proposed here. In particular, although circumstantial evidence has been presented that the ππ* state is metastable, this has not been conclusively proven at this stage. Vibrational analysis would be required at a level that can reliably treat the energies of the relevant states as well as their coupling. The analysis of transition states arising from the avoided crossings of the states would clearly be nice, as would an analysis of the branching spaces of the associated intersections.

Olsen et al. Conclusions We have reported the first EOM-CCSD and MR-RSPT2 analyses of the excited states of an aryl peroxycarbonate. The excited state structure of such molecules is of significant current interest owing to suggestions that the proposed mechanism for their photoinduced decomposition may need to be revised. Previous models invoked the presence of an intermediate [(aryloxy)carbonyl]oxy radical created by peroxide scission, which was suggested to undergo decarboxylation on a picosecond time scale. Recent data suggest that complete decomposition to form molecular CO2 and the complementary aryloxy and alkyloxy radicals occurs with a single time constant within the lifetime of the excited singlet state. This mechanism contrasts with the established mechanism for other organic peroxides2-5,9,12,27 and may indicate a new mechanistic regime for these systems. We have computed the relative energies of the initially excited aryl ππ* state, as well as nσ* and nπ* states on the peroxide and carbonate moieties. While the involvement of each of these low-lying valence states in the overall photodissociation is implied respectively by the absorption spectrum, the peroxide bond dissociation and carbonyl R-cleavage, our results clearly demonstrate that the lowest lying (S1) singlet state populated by UV absorption is an intrinsically nondissociative aryl ππ* state. This result carries the ramification that the observed single time scale for fragmentation most likely correlates with the time scale for exit from the aryl ππ* state via adiabatic or nonadiabatic means. Acknowledgment. Calculations were performed on the NCI Supercomputer Facility, Canberra. We are grateful to the NCI staff for their consistent and expert support. Time on the NCI machines was generously provided by a grant from the NCI Merit Allocation Scheme (MAS, Project M03). Figures were generated with the aid of the VMD28 and ChemBioDraw Ultra29 software packages. S.C.S. gratefully acknowledges support in the form of a Bessel Research Award of the Alexander von Humboldt Foundation, which facilitated the development of this work. S.O. is supported by funds from the Australian Research Council Discovery Project DP0877875. We are grateful to R. H. McKenzie for helpful critique of an early version of the manuscript. Supporting Information Available: Cartesian coordinates (Å), SA-CASSCF, MR-RSPT2, EOM-CCSD, MP2, and RHF energies and dipole moments (au), SA-CASSCF state-averaged natural orbitals, corresponding occupation numbers, a complete ref 24, and excitation data calculated with a larger basis set. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Bartlett, P. D.; Hiatt, R. R. J. Am. Chem. Soc. 1958, 80, 1398– 1405. (2) Buback, M.; Kling, M.; Schmatz, S.; Schroeder, J. Phys. Chem. Chem. Phys. 2004, 6, 5441–5455. (3) Chateauneuf, J.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1988, 110, 2877–2885. Chateauneuf, J.; Lusztyk, J.; Ingold, K. U. J. Am. Chem. Soc. 1988, 110, 2886–2893. (4) Polyansky, D. E.; Neckers, D. C. J. Phys. Chem. 2005, A109, 2793– 2800. (5) Morlino, E. A.; Bohorquez, M. D.; Neckers, D. C.; Rodgers, M. A. J. J. Am. Chem. Soc. 1991, 113, 3599–3601. (6) Reichardt, C.; Schroeder, J.; Schwarzer, D. J. Phys. Chem. 2007, A111, 10111–10118. (7) Abel, B.; Buback, M.; Kling, M.; Schmatz, S.; Schroeder, J. J. Am. Chem. Soc. 2003, 125, 13274–13278.

Characterization of Aryl Peroxycarbonate (8) Reichardt, C.; Schroeder, J.; Schwarzer, D. Phys. Chem. Chem. Phys. 2008, 10, 5218–5224. (9) Abel, B.; Assmann, J.; Buback, M.; Grimm, C.; Kling, M.; Schmatz, S.; Schroeder, J.; Witte, T. J. Phys. Chem. 2003, A107, 9499–9510. (10) Falvey, D. E.; Schuster, G. B. J. Am. Chem. Soc. 1986, 108, 7419– 7420. (11) Aschenbrucker, J.; Buback, M.; Ernsting, N. P.; Schroeder, J.; Steegmuller, U. J. Phys. Chem. 1998, B102, 5552–5555. (12) Abel, B.; Assmann, J.; Botschwina, P.; Buback, M.; Kling, M.; Oswald, R.; Schmatz, S.; Schroeder, J.; Witte, T. J. Phys. Chem. 2003, A107, 5157–5167. (13) Liu, Y.-J.; Persson, P.; Lunell, S. Mol. Phys. 2004, 102, 2575– 2584. (14) Brouard, M.; Martinez, M. T.; Milne, C. J.; Simons, J. P.; Wang, J. X. Chem. Phys. Lett. 1990, 165, 423–428. (15) Turro, N. J. Modern Molecular Photochemistry; University Science Books: Sausalito, CA, 1991; p 224. (16) Assmann, J.; Kling, M.; Abel, B. Angew. Chem. Int. Ed. 2003, 42, 2226–2246. (17) Stanton, J. F.; Bartlett, R. J. J. Chem. Phys. 1993, 98, 7029–7039.

J. Phys. Chem. A, Vol. 114, No. 12, 2010 4295 (18) Celani, P.; Werner, H.-J. J. Chem. Phys. 2000, 112, 5546–5557. (19) Docken, K. K.; Hinze, J. J. Chem. Phys. 1972, 57, 4928–4936. (20) Roos, B. O. In Ab Initio Methods in Quantum Chemistry II; Lawley, K. P., Ed.; John Wiley and Sons: New York, 1987; p 399. (21) Frisch, M. J.; Pople, J. A.; Binkley, J. S. J. Chem. Phys. 1984, 80, 3265–3269. Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257–2261. (22) Møller, C.; Plesset, M. S. Phys. ReV. 1934, 46, 618. (23) Stalring, J.; Bernhardsson, A.; Lindh, R. Mol. Phys. 2001, 99, 103– 114. Werner, H.-J.; Meyer, W. J. Chem. Phys. 1981, 74, 5794–5801. (24) Werner, H.-J., MOLPRO, 2008.1 (25) Buback, M.; Kling, M.; Seidel, M. T.; Schott, F. D.; Schroeder, J.; Steegmuller, U. Z. Phys. Chem. 2001, 215, 717–735. (26) Haddon, R. C. J. Phys. Chem. 2001, A105, 4164–4165. (27) Abel, B.; Assmann, J.; Buback, M.; Kling, M.; Schmatz, S.; Schroeder, J. Ang. Chem. Int. Ed. 2003, 42, 299–303. (28) Humphrey, W.; Dalke, A.; Schulten, K. J. Mol. Graph. 1996, 14. (29) CambridgeSoft, I., ChemBioDraw Ultra, 11.0.

JP9029166