Photoelectron Spectroscopy Study of QuinonimidesEkram Hossain,† Shihu M. Deng,‡ Samer Gozem,§ Anna I. Krylov,§ Xue-Bin Wang,*,‡

and Paul G. Wenthold*,†

†The Department of Chemistry Purdue University West Lafayette, Indiana 47906, United States‡Physical Sciences Division, Pacific Northwest National Laboratory P.O. Box 999, MS k8-88 Richland, Washington 99352, UnitedStates§Department of Chemistry University of Southern California Los Angeles, 90089, United States

*S Supporting Information

ABSTRACT: Structures and energetics of o-, m-, and p-quinonimideanions (OC6H4N

−) and quinoniminyl radicals have been investigated byusing negative ion photoelectron spectroscopy. Modeling of thephotoelectron spectrum of the ortho isomer shows that the groundstate of the anion is a triplet, while the quinoniminyl radical has a doubletground state with a doublet-quartet splitting of 35.5 kcal/mol. The pararadical has doublet ground state, but a band for a quartet state is missingfrom the photoelectron spectrum indicating that the anion has a singletground state, in contrast to previously reported calculations. Thetheoretical modeling is revisited here, and it is shown that accuratepredictions for the electronic structure of the para-quinonimide anionrequire both an accurate account of electron correlation and a sufficientlydiffuse basis set. Electron affinities of o- and p-quinoniminyl radicals are measured to be 1.715 ± 0.010 and 1.675 ± 0.010 eV,respectively. The photoelectron spectrum of the m-quinonimide anion shows that the ion undergoes several differentrearrangements, including a rearrangement to the energetically favorable para isomer. Such rearrangements preclude ameaningful analysis of the experimental spectrum.

■ INTRODUCTION

Benzoquinones (1o, 1m, 1p) are important chemical structuresthat are found in a large variety of systems with differentapplications. As convenient electron acceptors, quinones playessential roles in critical biological processes including photo-synthesis and respiration.1−11 However, quinones are alsoinvolved in several harmful processes in biological systems.Ortho- and para-quinones can be easily formed by oxidation ofthe corresponding catechols, and can even be formed from theoxidation of aromatic molecules, such as benzene. Ortho-quinones formed in this manner have been proposed as thesources of benzene toxicity12 and carcinogenicity.13 Similarly,ortho-quinones formed in the oxidation of urushiol cause thecontact-dermatitis associated with plants like poison ivy.14

Benzoquinones are also utilized in various chemical applica-tions, serving as dyestuffs,15−17 battery electrodes,18−20 andhydrogen storage materials21 and as oxidizing agents22−24 anddienophiles25 in organic synthesis.

The electronic structures of benzoquinones are highlydependent on their topology. The stable quinones encountered

in the situations described above are the ortho- and para-isomers, and the stabilities are readily apparent in their Kekule ́valence structures, as shown in 1o and 1p. Consistent with thisview, both 1o and 1p have closed-shell singlet groundstates,26,27 with triplet-states lying 1.68 and 2.32 eV higher inenergy, respectively.28 In contrast, the meta-isomer, 1m, has anon-Kekule ́ structure and an open-shell, triplet ground-state28−31 9.0 kcal/mol lower in energy than the lowest-energysinglet state.30 The relative stabilities of the ortho- and para-isomers compared to 1m are reflected in the absoluteenthalpies of formation. The measured enthalpies of formationof 1o and 1p are −21.0 ± 3.1 and −29.3 ± 0.9 kcal/mol,respectively, whereas ΔHf,298(1m) = 8.1 ± 3.4 kcal/mol.29,32

The higher enthalpy of formation of 1o compared to 1p hasbeen attributed to electron-pair repulsion in the ortho-isomerthat is not present in the para-isomer.29

The benzoquinones 1o−1p are just one type within the classof molecules with quinoidal structures. As shown in Figure 1,the substituents on the aromatic ring can be varied to createquinone analogs. The most common of these would include thequinomethanes (X = O, Y = CR2, Figure 1)33−39 or thequinodimethanes (xylylenes),40−48 where both X and Y arecarbon-centered substituents. However, essentially all multi-

Received: May 19, 2017Published: July 21, 2017

Article

pubs.acs.org/JACS

© 2017 American Chemical Society 11138 DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

valent atoms can be incorporated and in almost anycombination. Whereas each variation will have unique proper-ties, the common theme is that the ortho- and para-isomers willhave stable,33,49 Kekule ́ structures with closed-shell singletelectronic states, whereas the meta-isomers are higher-energy,50

non-Kekule ́ structures with high-spin electron coupling in theground state, which is why meta-phenylene derivatives arecommonly used as the basis for high-spin materials.51−54

We recently reported a flowing afterglow study of quinoidalimides, including quinonimides (2, X = O, Y = N−) andquinomethanimides (3, X = CH2, N

−).55

Significant differences were observed in the reactions of 2pand 3p with nitric oxide. Whereas 3p reacts with NO merely byaddition, 2p reacts by addition and by a novel, N−O exchangereaction to form the semiquinone radical anion, as shown in eq1.The observed differences in reactivity were attributed to

differences in the electronic structures. The addition reactionthat occurs with 3p is consistent with what is expected for theclosed-shell, imide anion.56 In contrast, the N−O exchangereaction has been observed previously with nitrene radicalanions57 and is interpreted as indicating an open-shellelectronic structure.58 Consequently, our previous studyindicates that while 3p has a robust ground-state singlet, 2pis either a ground-state triplet or a singlet with a thermallyaccessible triplet state.55

A low-lying triplet state for 2p is somewhat surprisingconsidering that it is isoelectronic with p-benzoquinone, 1p,which is a singlet with a large (>2 eV) singlet−tripletsplitting.28 However, it can be rationalized by consideringthat 2p can alternatively be viewed as an oxo-substitutedphenylnitrene, as shown in Scheme 1. Considering that theenergies of closed-shell singlet states in aromatic nitrenes are

approximately 30 kcal/mol higher than the triplets,59,60 the factthat the singlet is close in energy to the triplet indicates that thesinglet quinonimide is highly stabilized, at least for a closed-shell singlet nitrene. Triplet nitrene 2p has also been proposedas an intermediate in the photolysis of p-azidophenoxide inaprotic conditions,61 affirming that the triplet state is chemicallyaccessible.The conclusions about the electronic structures of 2p and 3p

were consistent with reported electronic structure calculations55

that predict 3p to be a ground-state singlet with a singlet−triplet splitting of about 5 kcal/mol, with 2p a ground statetriplet, with a singlet that is 1−2 kcal/mol higher in energy.In this work, we describe a negative ion photoelectron

spectroscopy (NIPES) study of the quinonimides, 2o−p.Negative ion photoelectron spectroscopy provides a means toinvestigate the electronic properties of ions as well as those ofthe neutral molecules obtained upon electron detachment.62

Therefore, in addition to being a study of the quinonimides,this study also provides new spectroscopic information forquinoniminyl radicals, 4.

The quinoniminyl radicals are isoelectronic with thebenzoquinone radical cations.63,64 On the basis of isotropicproton coupling in ESR experiments, Chandra et al.63 proposedthat ionization of 1p occurs from a σ orbital. By anology, 4pwould be a σ-radical, with a closed-shell π system, leading to adoublet ground state, and the ortho-radical (4o) should besimilar. High-spin coupling of the meta-phenylene arrangementin 4m, however, should favor a quartet ground-state.65 Thepara-quinoniminyl radical, 4p, has previously been suggested asa possible intermediate in the Gibbs reaction between N-haloquinonimines and phenoxides.66

In this study, we use NIPES to examine the ortho-, meta- andpara-isomers of the quinonimides. In addition to determiningelectron affinities of the quinoniminyl radicals, we also usemodeling of the spectroscopic features to show that ortho-quinonimide, 2o, has a triplet ground state, consistent withelectronic structure calculations. For the para-isomer, 2p, thespectroscopy results are best interpreted as indicating a singletground state, in contrast to previous electronic structurepredictions, which have been revisited in this work. The spectraand calculations presented for 2p, combined with the reactivitystudies reported previously,55 suggest that whereas the isomerhas a singlet ground state, it also has a low-lying triplet statethat can be accessed during reaction.

■ EXPERIMENTAL METHODSAll NIPES experiments were carried out at PNNL using an apparatusconsisting of an electrospray ionization source (ESI), a cryogenic iontrap, and a magnetic-bottle time-of-flight (TOF) photoelectronspectrometer.67 Azidophenols were synthesized from o-, m-, and p-aminophenols by using procedures similar to those describedpreviously,68,69 and used without purification. A 5 × 10−4 Macetonitrile solution of each o-, m-, and p-azidophenol (N3−C6H4−OH) titrated dropwise with 0.01 M aqueous KOH was prepared in theglovebox under an N2 atmosphere. The electrospray ionizationconditions were optimized to afford maximum intensity for thedissociation (M−H−N2) product anions, that is, 2o, 2m, and 2p, m/z106. The anions generated by ESI were guided by quadrupole ion

Figure 1. Some variations of quinoidal molecules.

Scheme 1

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11139

guides into an ion trap, where they were accumulated and cooled for20−100 ms by collisions with cold buffer gas at 20 K, before beingtransferred into the extraction zone of a TOF mass spectrometer. Thecooling of the anions to 20 K eliminated the possibility of theappearance of hot bands in the NIPES spectra.Each of 2o, 2m, or 2p anions were then mass selected, and

maximally decelerated before being photodetached with 266 (4.661eV), 355 (3.496 eV), and 532 nm (2.331 eV) photons from a Nd:YAGlaser. The laser was operated at a 20 Hz repetition rate, with the ionbeam off at alternating laser shots to enable shot-to-shot backgroundsubtraction to be carried out.Photoelectrons were collected with ∼100% efficiency with the

magnetic bottle and analyzed in a 5.2 m long electron flight tube. Therecorded TOF photoelectron spectrum was converted into an electronkinetic energy spectrum by calibration with the known NIPE spectraof I− and OsCl2−. The electron binding energy spectrum was obtainedby subtracting the electron kinetic energy spectrum from the energy ofthe detaching photons. The energy resolution was about 2%, that is,∼20 meV for 1 eV kinetic energy electrons.

■ COMPUTATIONAL METHODSAll geometries and frequencies reported in this work are computedwith density functional theory (DFT) with the B5050LYP func-tional70,71 and 6-311++G(d,p) basis set. B5050LYP is a hybridfunctional that uses 50% Hartree−Fock, 8% Slater and 42% Beckeexchange, and 19% VWN+81% LYP correlation, and is used becauseextensive benchmarks indicate that it yields improved geometries,energies, and vibrational frequencies in combination with the spin-flip(SF)-DFT approach.71 We employ SF-DFT for low-spin (singlet ordoublet) states in ortho and para isomers (2o, 2p, 4o, 4p), but not formeta (2m, 4m) because in that case SF states were strongly spin-contaminated. For high-spin (triplet and quartet) states and for allmeta isomer calculations, standard DFT was used. The DFT or SF-DFT geometries were then used for subsequent energy calculationswith coupled-cluster-based methods.Singlet−triplet and doublet-quartet energy splittings were computed

with the spin-flip equation-of-motion coupled-cluster (EOM-SF-CC)method.72,73 The SF approach employs a high-spin reference state anduses a spin-flipping operator that allows description of both the high-spin and low-spin target states on an equal footing.72,74−76 As neededfor the systems presented in this study, the SF approach is capable ofdescribing multiconfigurational character encountered in open-shellsinglets and doublet radicals, which may not be described correctly in asingle-configuration formalism. Previous studies have shown that theSF variant of EOM-CC72,73,77 gives accurate singlet−triplet splittingsin aromatic nitrenes.EOM-SF-CC energies are calculated with singles and doubles

excitations, denoted EOM-SF-CCSD. In addition, the effect of tripleexcitations was included by perturbative triples correction, denotedEOM-SF-CCSD(dT).77,78 Coupled-cluster calculations were carriedout using B3LYP orbitals as the reference orbitals, as describedpreviously,60,72 to mitigate spin-contamination.79 Electron attachmentenergies were computed between the triplet and quartet states bytaking the difference in coupled-cluster energies with single and doubleexcitations and perturbative triples correction, CCSD(T). Electronattachment energies for other states are then derived on the basis oftriplet-quartet CCSD(T) energy difference and EOM-SF-CCSD(dT)splittings, as described above.80 All reported energies are adiabatic (i.e.,accounting for geometry relaxation of different states) and arecorrected for zero-point energies (ZPE) computed from DFT or SF-DFT unscaled vibrational frequencies at the corresponding geometry.For all coupled-cluster calculations, we employ Dunning’s correlationconsistent basis sets.81

Unless stated otherwise, the values for EOM-SF-CCSD(T)/(aug-)cc-pVTZ calculations were estimated by using the additive approachshown in eq 2.

= ‐ ‐

− ‐ ‐ + ‐ ‐

E E

E E

(CCSD/(aug )cc pVTz)

(CCSD/(aug )cc pVDZ) (CCSD(T)/(aug )cc pVDz)

projected

(2)

Previous studies,60 have found that this approach gives singlet−tripletenergy differences within ∼0.5 kcal/mol of those obtained by a directEOM-SF-CCSD(T)/cc-pVTZ calculation. In this work, the projectedenergies will be indicated as p-EOM-SF-CCSD(T)/(aug-)cc-pVTZ.All the calculations are performed using QCHEM ab initiopackage.82,83

Photoelectron spectra, including Franck−Condon vibrationalprogressions and detachment cross sections, have been calculatedusing procedures described previously. Franck−Condon factors arecalculated from DFT or SF-DFT normal modes and frequencies(depending on the isomer and state; see above). The Franck−Condonfactors for detachment are calculated using ezSpectrum.84

Detachment cross sections were calculated using ezDyson.85,86

Photoionization and photodetachment cross sections can formally becomputed from the photoelectron dipole matrix element, Dk

IF, eq 3,where ΨN and ΨN−1 are the wave functions of the initial (N-electron)and final (N−1-electron) states, respectively, ru is the dipole operatorfor the ionizing radiation, and Ψel is the wave function of the ejectedelectron.87,88

= ⟨Ψ | |Ψ ·Ψ ⟩−D rk IN

u FN

kIF 1 el

(3)

Owing to electron indistinguishability, DkIF can be expressed in terms of

a Dyson orbital, ϕd, such that it is now calculated from one-electronfunctions, as shown in eq 4.89,90

ϕ= ⟨ | |Ψ ⟩D rk u kIF d el

(4)

The Dyson orbitals are correlated extensions of Koopmans’ theorem.Here Dyson orbitals are calculated by using the ionization potentialEOM-CCSD (EOM-IP-CCSD) approach,85,90 which accounts fororbital relaxation and electron correlation. The wave function forreference triplet state of the anion was computed at the CCSD/aug-cc-pVDZ level of theory, and the neutral state wave functions werecalculated by using the EOM-IP-CCSD/aug-cc-pVDZ method.

To analyze diradical character, we also compute the Head-Gordon’sindex, using the EOM-SF-CCSD/aug-cc-pVDZ wave functions. TheHead-Gordon index is computed from state density matrices (or, moreprecisely, populations of natural orbitals) and therefore does notdepend on the choice of molecular orbitals. In this work, we use thenonlinear variant (eq 5),91,92

∑= −=

N n n(2 )i

M

i iu,nl1

2 2

(5)

where M is the total number of orbitals and ni is the occupationnumber for each natural orbital. The value of Nu,nl is a measure of theeffective number of unpaired electrons, such that for a closed shellsystem it is 0, a radical gives 1, and a perfect biradical gives 2, etc.

■ COMPUTATIONAL RESULTSBefore reporting the results of the photoelectron measure-ments, we provide a discussion of the electronic structures ofthe ions and corresponding neutrals. The electronic structuresof the anions, 2o, 2m, and 2p, can be understood from theframework of substituted phenyl nitrenes. The nonbondingmolecular orbitals in nitrenes usually include the benzylic-like πorbital and the σ (p) orbital localized on nitrogen (Scheme 2).An O− substituent adds an additional benzylic-like π orbital

that can interact with the π orbital of the nitrene. For the O− inthe ortho and para positions, the two π orbitals havesignificantly different energies, resulting from quinoidalbonding and antibonding combinations as shown at the topof Figure 2. Consequently, the bonding π orbitals are expectedto be significantly stabilized with respect to the σ orbitals, while

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11140

the antibonding orbitals are destabilized. For the meta isomer,the combinations of the two benzylic orbitals would benonbonding molecular orbitals, as shown at the bottom ofFigure 2, resembling the orbitals in other meta-phenylene-typesystems, such as 1,3-quinodimethane (m-xylylene). However,because they lack the bonding interactions present in theorbitals for 2o and 2p, there is no benefit to the combinationsfor 2m, such that the π orbitals in that isomer are generally thelocalized benzylic orbitals, as in Scheme 2.Anions 2o, 2m, and 2p are created by adding 4 electrons to

the three orbitals (σ and two π orbitals). For 2o and 2p, thetwo lowest energy electrons occupy the bonding π orbital, andthe other two electrons can occupy either the σ or π* orbitals.In our previous study, 2p was calculated to be a ground statetriplet, with the σ2 singlet state only 1−2 kcal/mol higher inenergy.55 Ion 2o is calculated to be a more robust ground-statetriplet, with a significantly larger singlet−triplet splitting of 8.7kcal/mol at the p-EOM-SF-CCSD(T)/aug-cc-pVTZ level oftheory. The increased singlet−triplet energy gap is likely theresult of destabilization of the σ2 state in the ortho isomerresulting presumably from repulsion between σ electron pairson the oxygen and nitrogen and, possibly, from 1,4-interactionin the π orbital. In fact, the destabilization is so significant thatit causes a nonplanar geometry for the singlet state when usinga spin-restricted-single determinant wave function, such asRHF, B3LYP, or BLYP. With unrestricted and spin-flipmethods, however, the state is predicted to be planar. Thisphenomenon is a manifestation of the pseudo Jahn−Tellereffect and is often encountered in open-shell species.93−95

The predicted electronic structure of the meta anion isomeris very different from those of the ortho and para isomers.

Because the orbitals are all very similar in energy, there aremultiple options for orbital occupancy. However, the electronicstate with the lowest energy is predicted to have two electronsin a phenoxide benzylic orbital, with the other two electronsoccupying the σ orbital and the π benzylic orbital on nitrogen.Therefore, in 2m the π orbitals localize into their benzylicforms (linear combinations of π1 and π2), presumably because itputs the negative charge predominantly on the oxygen.Consequently, 2m is best described as a phenoxide anionorthogonal to a phenyl nitrene. Finally, because it is a phenylnitrene distinct from the phenoxide, it is not surprising that ithas a triplet ground state. Similarly, the lowest energy singletstate of 2m is likely the singlet nitrene. The singlet−tripletenergy gap computed with EOM-SF-CCSD/aug-cc-pVTZ is17.3 kcal/mol.The difference in the electronic structure of 2m compared to

those of 2o and 2p is readily observed in Head-Gordon’sindices for the lowest-lying singlet state of each isomer. WhileNu,nl for both 2o and 2p is 0.04, indicating closed shell characterconsistent with the π2σ2 assignment discussed above, the sameindex is 2.04 for 2m, indicating that even the lowest energysinglet state has significant diradical character.The neutral quinoniminyl radicals, 4, formed upon detach-

ment of 2 are nominally triradicals, in that they have 3 electronsin three nearly degenerate orbitals. However, because of thestrong bonding interactions that stabilize the π orbitals in 4oand 4p, these isomers have doublet ground states, with high-energy quartet states (Table 1). In contrast, the ground state of4m is predicted to be a quartet, with multiple low-lying doubletstates. The relevant computed state energies are shown inTable 1, including computed electron affinities and doublet−quartet splittings.

■ EXPERIMENTAL RESULTSIn this section, we report the measured negative ionphotoelectron spectra for the [M−N2−H]− ions obtainedupon electrospray ionization of basic solutions of o-, m-, and p-azidophenol. Survey spectra, measured at 266 nm, are shown inFigure 3. Detailed descriptions of each spectrum are providedin the sections below.

ortho (2o). The 266 nm spectrum for the ortho isomer,Figure 3a, shows three identifiable features. The lowest energyfeature is a narrow band with an origin at 1.715 ± 0.010 eV.The highest energy feature is dominated by the origin peak at3.256 ± 0.005 eV, indicating a nearly vertical transition. Thecenter band begins at 2.54 eV. However, the relative intensityof this feature is variable and affected by source conditions.Consequently, the center band results from detachment of adifferent ion from that giving the other features in thespectrum.The 532 nm photoelectron spectrum of 2o (vide infra) only

contains the lowest energy feature. However, because thephotodetached electrons have lower energies, they are betterresolved than in the 266 nm spectrum. The identifiable peaksfor the ortho isomer are listed at the top of Table 1.

para (2p). The 266 nm spectrum of the para-substitutedion, shown in Figure 3c, has only a single intense feature, withan origin at 1.675 ± 0.010 eV. As with the ortho isomer, the532 nm spectrum (vide infra) shows a well-resolved band. Theobserved peaks are shown at the bottom of Table 1.

meta (2m). The 266 nm spectrum of 2m, shown in Figure3b, has three features. However, the lowest energy band isidentical to the lowest energy feature for the para isomer, in

Scheme 2

Figure 2. Symmetry-adapted π molecular orbitals in quinonimides.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11141

Figure 3c, indicating that the meta isomer undergoesrearrangement prior to photodetachment. The most intensefeature in the spectrum of 2m has an origin at 2.11 eV, withmany closely spaced peaks. The identifiable peaks are listed inTable 1. A third, weaker feature is also observed in the

spectrum. Although it is unresolved, it has a maximum at avertical detachment energy of about 3.15 eV.

■ DISCUSSION

In this section, we present an analysis of the measuredphotoelectron spectra of the ions derived from ortho-, meta-,and para-azidophenoxide, with a focus on determining theelectronic states of the anions and the radicals.

ortho. The spectrum of the ortho isomer contains featuresattributed to two different ions. The high and low energyfeatures result from a common ion, and are readily assigned toformation of the doublet and quartet states of 4o, respectively,formed upon detachment of 2o. The measured electron affinity,1.715 eV, agrees reasonably well with the calculated EA forformation of the triplet anion, 1.61 eV (Table 1). Moreimportantly, the energy difference between the bands, 35.5kcal/mol, is in very good agreement with the predicted doublet-quartet splitting of 36.0 kcal/mol, obtained at the p-EOM-SF-CCSD(T)/aug-cc-pVTZ level of theory (Table 1).The photoelectron spectrum of 2o shown in Figure 3a also

clearly demonstrates that the ion has a triplet electronic state.The most important evidence for the triplet state of 2o is thefact that detachment to both the doublet and quartet states areobserved. Given that the selection rule for photodetachment isΔS = ±1, the only way to form both the doublet and quartetstates of 4o is to detach from the triplet state of 2o.To confirm that the experimental peaks indeed are assigned

to doublet and quartet states formed from the triplet anion, werefer to quantum chemical calculations. In this way, we haveunequivocal evidence that the ground state is a triplet. In

Table 1. Observed Peaks in the Photoelectron Spectra of 2o, 2m, and 2p

ion and neutral electronic state peak eBE (eV) relative energy (cm−1) vibrational assignment calculated energya

2odoublet 1.715 ± 0.010 EA origin 1.61 eVb

1.780 525 ν8 5321.845 1050 2ν81.865 1210 ν20 11861.905 1530 ν23 15081.985 2160 ν8+ν23

impurity 2.5452.65

quartet 3.256 ± 0.005 +35.5 kcal/mol origin +36.0 kcal/molc

3.310 440 ν5 4293.430 1410 ν24 and ν25 1386, 1428

2mX 2.11 ± 0.02

2.222.352.48

A 3.15 (vertical)2p

doublet 1.675 ± 0.010 EA origin 1.47 eVb

1.730 445 ν5 4421.780 845 ν11, 2ν5 721, 8831.830 1250 3ν5, ν5+ν111.880 1655 2ν5+ν111.935 2100 ν5 + 2ν11

quartet +50.2 kcal/molc

aVibrational energies, in cm−1, calculated at the (SF)-B50H50/6-311++G(d,p) level of theory, scaled by 0.92; electron affinities and doublet-quartetsplittings are adiabatic energy differences and include ZPE correction bCalculated from p-CCSD(T)/aug-cc-pVTZ triplet-quartet energy differencesand p-EOM-SF-CCSD(dT)/aug-cc-pVTZ splittings. cp-EOM-SF-CCSD(dT)/aug-cc-pVTZ.

Figure 3. 266 nm photoelectron spectra of the m/z 106 ions derivedfrom (a) ortho-, (b) meta-, and (c) para-azidophenol. The asterisk (*)in panels a and b indicate bands whose relative intensity depends onsource conditions.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11142

particular, structural differences between the triplet and singletstates are manifested in the shapes of the features in thespectrum. The 532 nm photoelectron spectrum of the ion 2o,shown in Figure 4, has a well-resolved vibrational progression

for the doublet state. For comparison, we have calculated theexpected Franck−Condon profile for formation of the doubletfrom the triplet (Figure 4a) and singlet (Figure 4b) states of theion, using geometries and frequencies calculated at the DFT orSF-DFT B5050LYP levels of theory.84 The calculated spectrumobtained when using the triplet anion is in excellent agreementwith the measured spectrum, in both the positions of the peaksand in their intensities. In contrast, the calculated band fordetachment from the singlet state has poor resemblance to themeasured spectrum. Therefore, the Franck−Condon calcu-lation further confirms that the ion has a triplet ground state.Using this approach, we have also calculated the shape of the

quartet feature, shown in Figure 5. As with the doublet feature,there is outstanding agreement between the predicted relativepeak positions and intensities for detachment from the tripletanion, and those measured experimentally. The band for thedoublet state in Figure 5 is the same as that shown in Figure 4a,but with significantly broader peaks (450 cm−1 vs 150 cm−1 full-width at half-maximum) due to the lower resolution of theband in the 266 nm spectrum. Overall, these results create aconsistent picture for the photoelectron spectrum, indicatingdetachment of the triplet state of the anion forming the doubletand quartet states of the neutral.The middle band observed in the spectrum has an

identifiable origin peak at 2.54 eV, but the band has a relativeintensity that is dependent on source conditions, and thereforeresults from detachment of an isobaric ion. Given theazidophenoxide precursor, the most likely ionic formula iseither C6H4NO

− (same as 2o) or C5H4N3−. Either would result

from more complex dissociation of the precursor. We have not

been able to determine the actual structure of the ion leading tothe observed signal, but the electron binding energy suggests anoxygen-based anion, such as those shown in Figure 6.Cyanocyclopentadiene has been previously found to be formedby rearrangement of phenylnitrene in the gas-phase.96 SimpleB3LYP calculations predict cyclopentadienyl structures such asthose in Figure 6 to be up to 2 eV more lower in energy than2o.

meta. The photoelectron spectrum of the ion obtained fromthe meta isomer, shown in Figure 3b, contains a feature thatcan be identified as coming from detachment of the para anion,2p, which indicates that the ion is capable of rearrangement. Atthe p-CCSD(T)/aug-cc-pVTZ level of theory, 2p is calculatedto be 9.4 kcal/mol lower in energy than the meta-quinonimide,which is the driving force for the rearrangement. Unfortunately,the fact that the ion rearranges makes it difficult to analyze therest of the spectrum. The binding energy of 2.03 eV for themost intense band is significantly lower than the predictedvalue of 2.52 eV for the quartet state of 4m. Moreover, thetransition from 2m to the quartet state of 4m is predicted to beclose to vertical, but the observed band is much broader thanexpected for that transition. Therefore, it does not appear thatthe signal can be attributed to detachment of the tripletquinonimide, 2m. Interestingly, the computed detachment tothe doublet state is at 3.16 eV, which is in very good agreementwith the weak feature appearing at around 3.15 eV in theexperimental spectrum. This may be an indication that somefeatures from 2m appear in the spectrum but are overshadowedby peaks from rearranged products.As with the second band in the spectrum of 2o, the electron

binding energy of the main feature in the spectrum of the meta-isomer suggests an alkoxide, and there are rational possibilities.For example, unimolecular rearrangement of aromatic nitrenesis known to result in formation of cyclic ketenimines (eq 6a).97

A similar rearrangement (eq 6b) of 2m would lead to structuressuch as 5a and 5b. The computed electron binding energies of

Figure 4. Comparison of the low-energy feature in the 532 nmphotoelectron spectrum of 2o with the band predicted from (SF)-DFT B5050LYP/6-311++G(d,p) geometries and frequencies of the(a) triplet state and (b) singlet state of the anion and the doublet stateof the radical, 4o. Calculated harmonic frequencies were scaled by0.92.

Figure 5. Comparison of measured (266 nm) and calculated spectrafor 2o. The calculated spectra correspond to detachment of the tripletstate of 2o. Calculated harmonic frequencies were scaled by 0.92.

Figure 6. Possible structures of the second ion in the spectrum of 2o.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11143

5a and 5b are 2.06 and 1.71 eV, respectively, at the B3LYP/6-31+G* level of theory, and so are in the range of the measuredband. However, neither structure alone accounts for theobserved feature, as the bands for 5a or 5b are predicted tobe too narrow to account for the observed spectrum. Therefore,the low-energy feature in the spectrum of the meta isomer iseither a mixture of ions such as 5a and 5b or due to some otherundetermined rearrangement product. Although the keteni-mine structures 5a and 5b are calculated to be slightly higher inenergy than triplet 2m, they lower in energy than the singletstate.para. The spectrum of the ion obtained from p-

azidophenoxide is the simplest of the three, containing onlyone identifiable band, with an origin peak at 1.67 ± 0.01 eV.The measured electron affinity is in reasonable agreement withthe predicted value for the electron affinity of the doublet state(Table 1). Although there is broad signal at higher bindingenergy, there is no feature that can be clearly identified asresulting from detachment to form the quartet state, which ispredicted to be ∼2.2 eV higher in energy than the doublet(Table 1), and therefore within the energy range of theexperiment.The fact that there is only a single product formed upon

detachment of the para anion is significant. As noted in thediscussion of the ortho spectrum, detachment of the tripletstate of the anion would form both the doublet and quartetstate of the neutral product. Therefore, a possible explanationfor the absence of a peak for the quartet state is that the ion isin a singlet state. In order to confirm the singlet state, we havecarried out an analysis of the Franck−Condon progressionsimilar to what we did for 2o. Figure 7 shows a comparisonbetween the experimental 532 nm photoelectron spectrum of2p and those predicted for formation of the doublet upondetachment of the triplet (Figure 7a) and singlet (Figure 7b)states of the anion.Although the agreement between the calculated and

predicted spectra is clearly better for the singlet state, there isnot a sufficient difference in the quality of the agreementbetween the spectra for the singlet and triplet states for the paraisomer to make an unequivocal assignment of the electronicstate, and the differences observed for the triplet state in Figure7a are not sufficient to rule it out as the electronic state of theanion.Because the Franck−Condon calculation cannot sufficiently

distinguish between the triplet and singlet states of 2p, theassignment of the electronic state depends on the absence ofthe quartet band. However, this conclusion requires theassumption that the quartet band should be observable to asufficient extent. Although the spectrum for 2o, for example,

suggests that the quartet band should be strong, it is notconclusive. Therefore, we have calculated the expected relativecross sections for formation of the doublet and quartet statesusing the Dyson orbitals computed for correlated EOM-CCSDwave functions, as described above.85,86

The challenge in this calculation involves the choice for thewave function for the detached electron. For simple atomicsystems, the detached electron can be described as a Coulombwave in the case of photoionization,98,99 or as a plane wave inthe detachment of an atomic anion.88 To a first approximation,one can treat the detachment of 2p as similar to that of anatomic anion, and describe it as a plane wave. However, for theortho isomer, it was found that the relative intensity of thequartet feature was underestimated by using a plane waveapproximation (see Supporting Information). Alternatively, ithas been found previously that, for larger molecular systems,using a Coulomb wave with a partial charge between 0 and 1often gives the best agreement with experiment.99,100 Becausewe do not have a rigorous procedure to determine the besteffective charge, we have computed the cross section at Z valuesof 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0, and used the average values toobtain relative cross sections. This approach has been usedsuccessfully to predict the cross sections for ionization ofaldehydes and enols100 and is the approach used to determinethe relative intensities for the 2o shown in Figure 5. For thequinonimides, the relative intensities of the quartet statesobtained by using a modified Coulomb wave are consistentlylarger than those obtained by using a plane waveapproximation, regardless of the chosen Z value (seeSupporting Information).The average calculated total absolute cross sections for

detachment of triplet 2p to form the doublet and quartet statesof 4p are 4.7 and 6.2 megabarns, respectively, giving relativecross sections of about 1/1.3 for formation of the doublet andquartet, respectively. Therefore, as expected, detachment of the

Figure 7. Comparison of the low-energy feature in the 532 nmphotoelectron spectrum of 2p with the band predicted from (SF)-DFT B5050LYP/6-311++G(d,p) geometries and frequencies of the(a) triplet state and (b) singlet state of the anion and the doublet stateof the radical, 4p. Calculated harmonic frequencies were scaled by afactor of 0.92.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11144

triplet anion should lead to abundant formation of the quartetstate. Scaling the integrated Franck−Condon profiles by thecalculated total cross sections results in the predicted spectrashown in Figure 8, which confirm that the quartet is not being

formed as expected if 2p were a triplet anion. Although thepredicted relative cross-section for formation of the quartetstate is lower when using a plane wave approximation for thedetached electron, the band for the quartet state is nonethelesspredicted to be prominent in the spectrum (see SupportingInformation). Consequently, the photoelectron spectrum isbest interpreted as resulting from detachment of the singletstate of 2p.Vibrational Assignments. The calculation of the Franck−

Condon intensities from theoretical structures and frequenciescan be used to identify the vibrations active in thephotoelectron spectra. For example, as shown in Table 1, theactive modes in the doublet feature in the spectrum of 4oinclude ν8 (525 cm−1), ν20 (1210 cm−1), and ν23 (1530cm−1). The vibrational motions associated with these modesare shown in Scheme 3. The assigned modes are thosepredicted to be active in the Franck−Condon calculations. Themode ν8 is a ring-deformation mode, similar to the ν6a modein benzene, and is commonly active in negative ion photo-electron spectra of aromatic molecules.101−104 The mode ν20 ispredominately stretching of the CO and CN carbon atoms,whereas ν23 is the symmetric combination of the CO and CNstretching motions. In contrast, the active modes in the quartetstate of 4o, ν24 and ν25, are antisymmetric combinations of theCO and CN stretching motions. The active ring-deformationmode, ν5, involves the CN stretch.The active modes in the doublet state of 4p consist of the

ring-deformation mode, ν5, and a ring breathing mode, ν11.Comparison to Reactivity and Theoretical Results. As

noted in the introduction, the reactivity of 2p has beenpreviously interpreted as indicating that 2p is either a ground-state triplet or a singlet with a thermally accessible tripletstate.55 The 20K photoelectron spectrum indicates that the ionhas a ground state singlet, and so if reactivity is due to thetriplet state, it must be accessed during the reaction. Kenttam̈aaand co-workers have found that singlet diradicals are capable ofundergoing reactions involving the triplet state,105,106 andtherefore the presence of a singlet ground state for 2p is notinconsistent with the apparent radical reactivity.

The singlet ground state found for 2p is, however,inconsistent with predictions from electronic structure theoryreported previously.55 As noted above, all three quinonimideanions were predicted to have triplet ground states. Thecalculations leading to the prediction of the triplet ground statefor 2p reported in the previous work55 were carried out at theBLYP/6-311++G(3df,2p) and EOM-SF-CCSD(T)/cc-pVDZlevels of theory. The former is a DFT method employing alarge basis set with both polarization and diffuse functions,while the latter is method that accounts for static (via SFansatz) and dynamic (up to triple excitations) electroncorrelation. Nonetheless, both of these approaches failed topredict the correct ground state for 2p. Here we revisit thecalculations on the singlet−triplet splitting of 2p, summarizedin Table 2.We find that standard DFT methods give triplet ground

states, regardless of the basis set. As for EOM-SF-CCcalculations, we find that increasing the size of the basis setleads to preferential stabilization of the singlet state. Inparticular, augmentation of the basis set with diffuse functionssignificantly stabilizes the singlet state more than it does thetriplet, such that the predicted ground state when using singlyaugmented functions is the singlet state. At the best level oftheory used in this work, p-EOM-SF-CCSD(dT)/aug-cc-pVTZ, 2p is indeed predicted to have a singlet ground state,with a singlet−triplet energy splitting of 0.8−1.0 kcal/mol.Therefore, accurate prediction of the correct ground state in

2p requires both an accurate account of electron correlation butalso a very large basis set that includes diffuse functions.Interestingly, SF-DFT with B5050LYP/6-311++G(d,p) alsopredicts a singlet ground state, by 1.6 kcal/mol.The predicted energy difference between the singlet and

triplet states is sufficiently small that it is possible that therecould be some population of triplet state present at thermaltemperatures. The spectra shown above were all measured at atemperature of 20 K, and therefore, where the population of theexcited state should not be significant. We have also measuredthe 355 nm photoelectron spectrum of 2p at a temperature of299 K. Although there are some differences in the relative peakintensities of the doublet feature, they are not sufficient to be

Figure 8. Comparison of measured (266 nm) and predicted spectrafor 2p. The calculated spectra correspond to detachment of the tripletstate of 2p.

Scheme 3

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11145

attributed to triplet anion contribution. It is especially difficultto draw any conclusions due to the similarities of the doubletbands obtained from the singlet and triplet states of the anion(see the calculated spectra in Figure 7).

■ CONCLUSIONSNegative ion photoelectron spectroscopy studies have beenused to provide ground state assignments for quinonimideanions. Whereas previous experimental and computationalstudies55 suggested that para-quinonimide, 2p, either has tripletground state or a low-lying triplet, the photoelectron spectrumpresented in this work is best interpreted as showing a groundstate singlet. It is likely the case that the triplet state is relativelylow in energy, as very high levels of theory, involving anaccurate account of electron correlation and triple-ζ basis setswith diffuse functions, are required to predict correctly thesinglet ground state, while standard density functional methodsare insufficient. However, SF-DFT is able to produce singlet−triplet energy splittings in reasonably good agreement withEOM-SF-CCSD(dT) calculations, and correctly predict asinglet ground state for para.In contrast, the ortho isomer, 2o, is found spectroscopically

to have a triplet ground state, which is consistent withdestabilization of the singlet state due to electron-pairrepulsion, as proposed for ortho-quinone.29 The effect in 2ois likely more pronounced because of the diffuse nature of thelone pair of electrons on the anionic nitrogen.Removing an electron to form the radical, 4o, reduces the

electron repulsion. Any effects of repulsion of the in-planeunpaired electron in 4o appear to be minimal, as the relativeelectron affinities of 4o and 4p are similar to what has beenfound previously for o- and p-benzoquinone (1o and 1p), 1.90

and 1.85 eV, respectively.28 Similarly, the measured doublet-quartet splitting in 4o, 35.5 kcal/mol, is similar to that foundfor 1o, 36.0 kcal/mol. Consequently, the o- and p-quinoniminylradicals resemble the corresponding quinones.The meta-quinonimide anion, 4m, is found to rearrange to

the para- and other unidentified isomers. The instability of 4mis not surprising. Attempts to generate this anion under flowingafterglow conditions55 were unsuccessful, and reactivity studiesof the m-quinomethanimide anologue, 6, indicated that it alsolikely undergoes rearrangement.55

The m-quinonimide anion, 2m, and the radical, 4m, are bothmeta-phenylene structures and therefore would be expected tohave high-spin ground states, which is supported bycalculations.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/jacs.7b05197.

Calculated spectra for 2o and 2p obtained using a planewave approximation; comparison of 355 nm spectra of2p taken at 20K and 299 K; and calculated structures,energies, and frequencies used in this work (PDF)

■ AUTHOR INFORMATIONCorresponding Authors*Xuebin.Wang@pnnl.gov*pgw@purdue.eduORCIDXue-Bin Wang: 0000-0001-8326-1780Paul G. Wenthold: 0000-0002-8257-3907NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the National Science Foundation(P.G.W., CHE11-11777 and CHE15-65755). The photo-electron spectroscopy work done at PNNL was supported byU.S. Department of Energy (DOE), Office of Science, Office ofBasic Energy Sciences, the Division of Chemical Sciences,Geosciences, and Biosciences (X.-B.W. and S.H.D.), andperformed using EMSL, a national scientific user facilitysponsored by DOE’s Office of Biological and EnvironmentalResearch and located at Pacific Northwest National Laboratory,which is operated by Battelle Memorial Institute for the DOE.A.I.K. is supported by the Department of Energy through theDE-FG02-05ER15685 grant. A.I.K. is also a grateful recipient ofthe Bessel Research Award from the Alexander von HumboldtFoundation.

■ REFERENCES(1) Allen, J. P.; Feher, G.; Yeates, T. O.; Rees, D. C.; Deisenhofer, J.;Michel, H.; Huber, R. Proc. Natl. Acad. Sci. U. S. A. 1986, 83, 8589−93.(2) Lancaster, C. R. D.; Michel, H. J. Mol. Biol. 1999, 286, 883−898.

Table 2. Calculated Values for the Singlet-Triplet EnergySplitting in 2p

computational method ΔEST (kcal/mol)a,b

B3LYP/6-31+G* 3.2B3LYP/6-311++G(3df,2p) 3.3BLYP/6-31+G* 1.3BLYP/6-311+G(3df,2p) 1.5M06-2X/6-31+G* 2.4M06-2X/6-311++G(3df,2p) 2.5SF-B5050LYP/6-311++G(d,p) −1.6EOM-SF-CCSD/cc-pVDZ 7.8EOM-SF-CCSD(dT)/cc-pVDZ 2.9EOM-SF-CCSD/aug-cc-pVDZ 4.8EOM-SF-CCSD(dT)/aug-cc-pVDZ −0.2EOM-SF-CCSD/cc-pVTZ 5.7EOM-SF-CCSD(dT)/cc-pVTZ 0.5EOM-SF-CCSD/aug-cc-pVTZ 4.2p-EOM-SF-CCSD(dT)/aug-cc-pVTZ −0.8c

−1.0d

EOM-SF-CCSD(dT)/aug-cc-pVTZ −1.0eaSinglet-state energies, relative to triplet state. Positive values indicatea triplet ground state. bDFT energies are for geometries optimized atthe respective level of theory; spin-flip energies are for singletgeometries optimized with SF-B5050LYP/6-311++G(d,p) and tripletgeometries optimized with B5050LYP/6-311++G(d,p). cProjectedvalue calculated using the EOM-SF-CCSD/aug-cc-pVTZ energy withthe dT component calculated with the aug-cc-pVDZ basis set.dProjected value calculated using the EOM-SF-CCSD/aug-cc-pVTZenergy with the dT component calculated with the cc-pVTZ basis set.eDirect (nonprojected) calculation.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11146

(3) Deisenhofer, J.; Epp, O.; Miki, K.; Huber, R.; Michel, H. Nature1985, 318, 618−24.(4) Feher, G.; Allen, J. P.; Okamura, M. Y.; Rees, D. C. Nature 1989,339, 111−16.(5) Kamiya, N.; Shen, J.-R. Proc. Natl. Acad. Sci. U. S. A. 2003, 100,98−103.(6) Zouni, A.; Witt, H.-T.; Kern, J.; Fromme, P.; Krauss, N.; Saenger,W.; Orth, P. Nature 2001, 409, 739−743.(7) Ferreira, K. N.; Iverson, T. M.; Maghlaoui, K.; Barber, J.; Iwata, S.Science 2004, 303, 1831−8.(8) Jordan, P.; Fromme, P.; Witt, H. T.; Klukas, O.; Saenger, W.;Krauss, N. Nature 2001, 411, 909−17.(9) Krauss, N.; Schubert, W. D.; Klukas, O.; Fromme, P.; Witt, H. T.;Saenger, W. Nat. Struct. Biol. 1996, 3, 965−73.(10) Loll, B.; Kern, J.; Saenger, W.; Zouni, A.; Biesiadka, J. Nature2005, 438, 1040−1044.(11) Srinivasan, N.; Golbeck, J. H. Biochim. Biophys. Acta, Bioenerg.2009, 1787, 1057−1088.(12) Barreto, G.; Madureira, D.; Capani, F.; Aon-Bertolino, L.;Saraceno, E.; Alvarez-Giraldez, L. D. Environ. Mol. Mutagen. 2009, 50,771−780.(13) Cavalieri, E. L.; Rogan, E. G.; Chakravarti, D. Cell. Mol. Life Sci.2002, 59, 665−681.(14) Crosby, D. G. Plants Toxic to the Skin; Oxford University Press:North Carolina, USA, 2004.(15) Gordon, P. F.; Gregory, P. Organic Chemistry in Color; Springer-Verlag: Berlin, 1983.(16) Matsuoka, M. Quinone dyes, in Infrared Absorbing Dyes;Matsuoka, M., Ed.; Plenum: New York, NY, 1990; p 35−43.(17) Venkataraman, K.; Iyer, V. N. Chem. Syn. Dyes 1971, 5, 131−240.(18) Barres, A.-L.; Geng, J.; Bonnard, G.; Renault, S.; Gottis, S.;Mentre, O.; Frayret, C.; Dolhem, F.; Poizot, P. Chem. - Eur. J. 2012,18, 8800−12.(19) Chen, H.; Armand, M.; Courty, M.; Jiang, M.; Grey, C. P.;Dolhem, F.; Tarascon, J.-M.; Poizot, P. J. Am. Chem. Soc. 2009, 131,8984−8988.(20) Liang, Y.; Tao, Z.; Chen, J. Adv. Energy Mater. 2012, 2, 742−769.(21) Nagao, M.; Kobayashi, K.; Yamamoto, Y.; Hibino, T. J.Electrochem. Soc. 2015, 162, F410−F418.(22) Popp, B. V.; Stahl, S. S. Top. Organomet. Chem. 2007, 22, 149−189.(23) Shen, C.-Y.; Yang, T.-K. 1,4-Benzoquinone. In Handbook ofReagents for Organic Synthesis: Catalytic Oxidation Reagents; Fuchs, P.,Ed.; John Wiley & Sons, Ltd.: Chichester, UK, 2013; p 49−52.(24) Yang, T.-K.; Shen, C.-Y. 1,4-Benzoquinone. In e-EROSEncyclopedia of Reagents for Organic Synthesis; John Wiley & Sons,Ltd.: Chichester, UK, 2001.(25) Nawrat, C. C.; Moody, C. J. Angew. Chem., Int. Ed. 2014, 53,2056−2077.(26) Chiang, Y.; Kresge, A. J.; Zhu, Y. J. Am. Chem. Soc. 2001, 123,8089−8094.(27) Chiang, Y.; Kresge, A. J.; Zhu, Y. J. Am. Chem. Soc. 2002, 124,6349−6356.(28) Fu, Q.; Yang, J.; Wang, X.-B. J. Phys. Chem. A 2011, 115, 3201−3207.(29) Fattahi, A.; Kass, S. R.; Liebman, J. F.; Matos, M. A. R.; Miranda,M. S.; Morais, V. M. F. J. Am. Chem. Soc. 2005, 127, 6116−6122.(30) Chen, B.; Hrovat, D. A.; Deng, S. H. M.; Zhang, J.; Wang, X.-B.;Borden, W. T. J. Am. Chem. Soc. 2014, 136, 3589−3596.(31) Fort, R. C., Jr.; Getty, S. J.; Hrovat, D. A.; Lahti, P. M.; Borden,W. T. J. Am. Chem. Soc. 1992, 114, 7549−52.(32) Recalculated from the results from ref 29 but with the revisedvalue of EA(1m) taken from ref 30.(33) Chacko, S. A.; Wenthold, P. G. Int. J. Mass Spectrom. 2007, 267,277−283.(34) Yamamura, S. Polycyclic arenes, quinomethanes and large ringsystems, including heterocycles. In Rodd’s Chemistry of Carbon

Compounds, 2nd ed.; Topical Vols, Organic Electrochemistry Edition;Elsevier: Amsterdam, the Netherlands, 2002; Vol. 5, p 225−305.(35) Wan, P.; Brousmiche, D. W.; Chen, C. Z.; Cole, J.; Lukeman,M.; Xu, M. Pure Appl. Chem. 2001, 73, 529−34.(36) Thompson, D. C.; Thompson, J. A.; Sugumaran, M.; Moldeus,P. Chem.-Biol. Interact. 1993, 86, 129−62.(37) Toteva, M. M.; Richard, J. P. Adv. Phys. Org. Chem. 2011, 45,39−91.(38) Wang, P.; Song, Y.; Zhang, L.; He, H.; Zhou, X. Curr. Med.Chem. 2005, 12, 2893−2913.(39) Willis, N. J.; Bray, C. D. Chem. - Eur. J. 2012, 18, 9160−9173.(40) Segura, J. L.; Martin, N. Chem. Rev. 1999, 99, 3199−3246.(41) Yoshida, H.; Ohshita, J.; Kunai, A. Bull. Chem. Soc. Jpn. 2010, 83,199−219.(42) Atlas, S. M. Nuova Chim. 1971, 47, 41−6.(43) Charlton, J. L.; Alauddin, M. M. Tetrahedron 1987, 43, 2873−89.(44) Scaiano, J. C.; Wintgens, V.; Netto-Ferreira, J. C. Pure Appl.Chem. 1990, 62, 1557−64.(45) Yamamoto, K. Konan Joshi Daigaku Kenkyu Kiyo 1981, 18, 135−44.(46) Neuhaus, P.; Grote, D.; Sander, W. J. Am. Chem. Soc. 2008, 130,2993−3000.(47) Berson, J. A. meta-Quinonoid compounds. In Chemistry ofQuinoidal Compounds, 1 ed.; Patai, S., Rappoport, Z., Eds.; Wiley:Chichester, UK, 1988; Vol. 2, p 455−36.(48) Berson, J. A. Non-kekule molecules as reactive intermediates. InReactive Intermediate Chemistry; Moss, R. A., Platz, M. S., Jones, M., Jr.,Eds.; John Wiley & Sons, Inc.: Hoboken, NJ, 2004; p 165−203.(49) Pollack, S. K.; Raine, B. C.; Hehre, W. J. J. Am. Chem. Soc. 1981,103, 6308−6313.(50) Hammad, L. A.; Wenthold, P. G. J. Am. Chem. Soc. 2000, 122,11203.(51) Ichimura, A. S.; Lahti, P. M. Mol. Cryst. Liq. Cryst. Sci. Technol.,Sect. A 1993, 233, 33−40.(52) Nicolaides, A.; Tomioka, H. J. Photosci. 2003, 10, 165−73.(53) Olankitwanit, A.; Pink, M.; Rajca, S.; Rajca, A. J. Am. Chem. Soc.2014, 136, 14277−14288.(54) Rajca, A.; Olankitwanit, A.; Rajca, S. J. Am. Chem. Soc. 2011,133, 4750−4753.(55) Rau, N. J.; Welles, E. A.; Wenthold, P. G. J. Am. Chem. Soc.2013, 135, 683−690.(56) Rau, N. J.; Wenthold, P. G. Int. J. Mass Spectrom. 2015, 377,496−501.(57) Wijeratne, N. R.; Wenthold, P. G. J. Org. Chem. 2007, 72, 9518−9522.(58) Chacko, S. A.; Wenthold, P. G. Mass Spectrom. Rev. 2006, 25,112−126.(59) Winkler, M. J. Phys. Chem. A 2008, 112, 8649−8653.(60) Wenthold, P. G. J. Org. Chem. 2012, 77, 208−214.(61) Geiger, U.; Haas, Y.; Grinstein, D. J. Photochem. Photobiol., A2014, 277, 53−61.(62) Wenthold, P. G.; Lineberger, W. C. Acc. Chem. Res. 1999, 32,597−604.(63) Chandra, H.; Portwood, L.; Symons, M. C. R. J. Chem. Soc.,Faraday Trans. 1 1989, 85, 1801−7.(64) Liu, Z. L.; Wu, L. M.; Liu, Y. C. Magn. Reson. Chem. 1988, 26,577−80.(65) Krylov, A. I. J. Phys. Chem. A 2005, 109, 10638−10645.(66) Pallagi, I.; Toro, A.; Horvath, G. J. Org. Chem. 1999, 64, 6530−6540.(67) Wang, X. B.; Wang, L. S. Rev. Sci. Instrum. 2008, 79, 073108.(68) Ryu, B.-Y.; Emrick, T. Macromolecules 2011, 44, 5693−5700.(69) Tomioka, H.; Matsushita, T.; Murata, S.; Koseki, S. Liebigs Ann.1996, 1996, 1971−1980.(70) Grimme, S.; Waletzke, M. J. Chem. Phys. 1999, 111, 5645−5655.(71) Shao, Y.; Head-Gordon, M.; Krylov, A. I. J. Chem. Phys. 2003,118, 4807−4818.(72) Krylov, A. I. Acc. Chem. Res. 2006, 39, 83−91.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11147

(73) Krylov, A. I. Annu. Rev. Phys. Chem. 2008, 59, 433−462.(74) Slipchenko, L. V.; Krylov, A. I. J. Chem. Phys. 2002, 117, 4694−4708.(75) Krylov, A. I. Chem. Phys. Lett. 2001, 350, 522−530.(76) Levchenko, S. V.; Krylov, A. I. J. Chem. Phys. 2004, 120, 175−185.(77) Manohar, P. U.; Krylov, A. I. J. Chem. Phys. 2008, 129, 194105−194110.(78) Piecuch, P.; Włoch, M. J. Chem. Phys. 2005, 123, 224105.(79) Beran, G. J. O.; Gwaltney, S. R.; Head-Gordon, M. Phys. Chem.Chem. Phys. 2003, 5, 2488−2493.(80) Cristian, A.-M. C.; Shao, Y.; Krylov, A. I. J. Phys. Chem. A 2004,108, 6581−6588.(81) Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007−23.(82) Shao, Y.; Molnar, L. F.; Jung, Y.; Kussmann, J.; Ochsenfeld, C.;Brown, S. T.; Gilbert, A. T. B.; Slipchenko, L. V.; et al. Phys. Chem.Chem. Phys. 2006, 8, 3172−3191.(83) Shao, Y.; Gan, Z.; Epifanovsky, E.; Gilbert, A. T. B.; Wormit, M.;Kussmann, J.; Lange, A. W.; Behn, A.; et al. Mol. Phys. 2015, 113,184−215.(84) Mozhayskiy, V.A.; Krylov, A.I. ezSpectrum. http://iopenshell.usc.edu/downloads.(85) Oana, C. M.; Krylov, A. I. J. Chem. Phys. 2009, 131, 124114.(86) Gozem, S.; Krylov, A. I. ezDyson, 2017. http://iopenshell.usc.edu/downloads/ezdyson.(87) Bethe, H. A.; Salpeter, E. E. Quantum Mechanics of One and TwoElectron Atoms; Plenum: New York, 1977.(88) Reed, K. J.; Zimmerman, A. H.; Andersen, H. C.; Brauman, J. I.J. Chem. Phys. 1976, 64, 1368−75.(89) Ortiz, J. V.; Per-Olov, L. Adv. Quantum Chem. 1999, 35, 33−52.(90) Oana, C. M.; Krylov, A. I. J. Chem. Phys. 2007, 127, 234106.(91) Head-Gordon, M. Chem. Phys. Lett. 2003, 372, 508−511.(92) Plasser, F.; Wormit, M.; Dreuw, A. J. Chem. Phys. 2014, 141,024106.(93) Munsch, T. E.; Slipchenko, L. V.; Krylov, A. I.; Wenthold, P. G.J. Org. Chem. 2004, 69, 5735−5741.(94) Vanovschi, V.; Krylov, A. I.; Wenthold, P. G. Theor. Chem. Acc.2008, 120, 45−58.(95) Krylov, A. I. The Quantum Chemistry of Open-Shell Species. InReviews in Computational Chemistry; John Wiley & Sons, Inc.:Hoboken, NJ, 2017; pp 151−224.(96) Cullin, D. W.; Soundararajan, N.; Platz, M. S.; Miller, T. A. J.Phys. Chem. 1990, 94, 8890−6.(97) Karney, W. L.; Borden, W. T. J. Am. Chem. Soc. 1997, 119,1378−1387.(98) Landau, L. D.; Lifshitz, E. M. Quantum Mechanics: Non-relativistic Theory; Pergamon: Oxford, 1977.(99) Gozem, S.; Gunina, A. O.; Ichino, T.; Osborn, D. L.; Stanton, J.F.; Krylov, A. I. J. Phys. Chem. Lett. 2015, 6, 4532−4540.(100) Abplanalp, M. J.; Gozem, S.; Krylov, A. I.; Shingledecker, C.N.; Herbst, E.; Kaiser, R. I. Proc. Natl. Acad. Sci. U. S. A. 2016, 113,7727−7732.(101) Gunion, R. F.; Gilles, M. K.; Polak, M. L.; Lineberger, W. C.Int. J. Mass Spectrom. Ion Processes 1992, 117, 601−620.(102) Kim, J. B.; Wenthold, P. G.; Lineberger, W. C. J. Phys. Chem. A1999, 103, 10833−10841.(103) Leopold, D. G.; Miller, A. E. S.; Lineberger, W. C. J. Am. Chem.Soc. 1986, 108, 1379−1384.(104) Wenthold, P. G.; Squires, R. R.; Lineberger, W. C. J. Am. Chem.Soc. 1998, 120, 5279−5290.(105) Amegayibor, F. S.; Nash, J. J.; Kenttaemaa, H. I. J. Am. Chem.Soc. 2003, 125, 14256−14257.(106) Williams, P. E.; Jankiewicz, B. J.; Yang, L.; Kenttamaa, H. I.Chem. Rev. 2013, 113, 6949−6985.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b05197J. Am. Chem. Soc. 2017, 139, 11138−11148

11148