1334
J. Phys. Chem. A 2010, 114, 1334–1337
Topological Control of Spin States in Disjoint Diradicals† Matthew J. Lenington and Paul G. Wenthold* Department of Chemistry, Purdue UniVersity, 560 OVal DriVe, West Lafayette, Indiana 47907-2084 ReceiVed: June 18, 2009; ReVised Manuscript ReceiVed: August 10, 2009
The meta- and para-bis-allylbenzene radical anions have been generated and investigated using mass spectrometry. The ions are formed by reaction of the corresponding bis-2-propenylbenzenes with atomic oxygen anion. Reactivity of the ions indicates that the ions most likely have a bis-allylbenzene structure. Reaction of the ions with carbon disulfide creates CS2 adducts, which, upon collision-induced dissociation, decompose to regenerate the bis-allylbenzene anion or carbon disulfide radical anion. The branching ratios for the two products indicate differences in the electronic structures of the neutral bis-allylbenzene diradicals. The difference in branching ratios and corresponding estimated electron affinities is interpreted in terms of different electronic states being formed, with the para diradical a singlet and the meta diradical either a groundstate triplet or a singlet with a very small singlet-triplet splitting. The difference in electron affinities is used to estimate a singlet-triplet splitting of 0.06 eV for the para diradical. The studies show that topology can be used to control the electronic properties of disjoint, tetramethyleneethane (TME)-like diradicals. Introduction Tetramethyleneethane (TME) is a famous disjoint1 organic diradical that has been characterized by matrix isolation2,3 and gas-phase spectroscopy4 and computationally.5-15 TME consists of two allyl radicals joined at the nodal carbon of each allylic fragment. The unpaired diradical electrons in TME occupy the symmetry-adapted pair of nonbonding molecular orbitals shown in Figure 1a. However, this pair of orbitals can be localized onto the two separate groups of atoms, as shown in Figure 1b, and therefore are technically disjoint orbitals having no atoms in common.1 As such, TME is expected have singlet and triplet spin states that are very close in energy. Simple qualitative models, however, predict that TME should have a ground-state singlet. For example, Ovchinikov16,17 has shown that ground states can be predicted in alternate hydrocarbons by using S ) |n* - n|/2, where S is the overall spin and n* and n are the number of starred and unstarred carbon atoms, respectively. With an equal number of starred and unstarred atoms, TME is predicted to have S ) 0, indicating a singlet state. Similarly, a “spin polarization” approach, wherein a local high spin preference around an atom causes polarization of the electron spin in the σ bond,17-20 also predicts that the unpaired electrons in TME are singlet coupled (Figure 2). The qualitative predictions are consistent with high-level ab initio calculations,5-15 which predict the singlet to be the ground state of TME, and these predictions are confirmed by photoelectron spectroscopy studies carried out by Lineberger and coworkers,4 which estimated the energy difference between the ground-state singlet and the triplet (∆EST) to be 2 kcal/mol. Other disjoint diradicals have been postulated or shown to have singlet ground states, including square cyclobutadiene,1 D8h cyclooctatetraene,21 and R,3-dehydrotoluene.22 In this work, we investigate the properties of TME analogues, para- and meta-bis-allyl benzenes, 1 and 2 (Figure 3). Adding an annulenic spacer between the allylic fragments does not †
Part of the “W. Carl Lineberger Festschrift”. * Corresponding author. Fax: (+1)765-494-0239. E-mail: pgw@ purdue.edu.
Figure 1. (a) Symmetry adapted and (b) localized nonbonding molecular orbitals in TME.
Figure 2. Spin-polarization assessment of the spins of the unpaired electrons in TME.
change the disjointedness of orbitals of the TME-like system.23 However, placing the allyl groups in different positions on the aromatic ring allows for topological control of the spin-state preferences for the two isomers. As shown in Figure 3, spin polarization methods predict that isomer 1 should resemble TME and have a ground state singlet. However, whereas the orbitals in 2 are still similar to those in TME, it is predicted by the spin polarization approach to have a ground-state triplet. Therefore, meta-bis-allylbenzene is predicted to be a rare example of a disjoint diradical with a high spin ground state
10.1021/jp905757p 2010 American Chemical Society Published on Web 09/10/2009
Spin States in Disjoint Diradicals
J. Phys. Chem. A, Vol. 114, No. 3, 2010 1335 TABLE 1: Branching Ratios for CID of CS2 Adducts of Ions 1- and 2- and Anionic References ion/CS2branching ratioa meta-bis-allylphenyl diradical para-bis-allylphenyl diradical 2-phenylallyl radical allyl radical 2-(2-propenyl)allyl radical benzyl radical para-fluorobenzyl radical
Figure 3. Nonbonding molecular orbitals and spin polarization models for para - and meta-bisallyl benzene.
similar to that of pentamethylenepropane, reported previously by Lahti et al.20 The properties of 1 and 2 are studied by using negative ion chemistry and mass spectrometry. Kass and coworkers previously described how an electron can serve as a “protecting group” for reactive species.24-26 In this work, we have compared the properties of the radical anions, 1- and 2-, which are the protected versions of the diradicals. In particular, the electron affinity of 2 is found to be significantly higher than that for 1, and the results are interpreted as reflecting the difference in spin states between the diradicals, which is consistent with the predictions from qualitative methods and high-level electronic structure calculations.
19 8.3 9.1 0.08 0.51 15 44
EAb 0.90 0.84 0.84 0.481 0.654 0.912 0.937
( ( ( (
0.008c 0.010c 0.006c 0.008c
a At a collision energy of 25 V (lab frame-of-reference) and argon target, with P(Ar) ) 0.225 mTorr. b Electron affinities estimated from CID branching ratios in electronvolts, unless otherwise noted. Day-to-day deviation in branching ratios indicates relative uncertainties of (0.03 eV. The absolute uncertainties in kinetic method measurements are estimated to be ca. ( 0.15 eV. c Ref 50.
anion, O-, as shown in eq 1.29 Reactivity of 1- and 2- is consistent with that expected for bis-allylic ions, as opposed to other isomers.
Experimental Section All experiments were conducted in a flowing afterglow triple quadrupole mass spectrometer that has been described in detail elsewhere.27,28 Atomic oxygen anion and fluoride ion were generated in the source by electron ionization (70 eV) of nitrous oxide (grade 2.0, BOC gases, Murray Hill, NJ) and molecular fluorine (5% in He, Spectra Gases, Branchburg, NJ), respectively. Hydroxide ion was generated by ionization of a mixture of nitrous oxide and methane (99.3%, Matheson). The ionic reagents were carried by helium buffer gas (0.400 Torr, flow (He) ) 190 STP cm3/s) through a 1 m flow tube, where it was allowed to react with neutral precursor vapors introduced through valve inlets. The bisallyl benzene radical anions were synthesized by reaction of atomic oxygen anion with the respective 1,3- and 1,4-diisopropenylbenzene precursors. Other ions, including 2-phenylallyl, 2-(2-propenyl)allyl, benzyl, and para-fluorobenzyl anions, were synthesized by deprotonation of the commercially available acid precursors by hydroxide, whereas allyl anion was synthesized by the reaction of fluoride with allyl trimethylsilane. Carbon disulfide adducts were formed by termolecular addition with CS2 vapors introduced into the flow tube through a micrometering valve. Ions passed through a 1 mm orifice into a differentially pumped chamber containing an Extrel triple quadrupole mass filter. CID was carried out on selected adducts using a typical collision energy of 25 V (laboratory frame-of-reference) and argon target (P(Ar) ) 0.225 mtorr). Materials. 1,3-Diisopropenylbenzene (97%) was purchased from Sigma Aldrich. 1,4-Diisopropenylbenzene was prepared by acid-catalyzed dehydration of R,R,R′,R′-tetramethyl-1,4benzenedimethanol (99%, Sigma Aldrich). Results and Discussion Ionized diradicals 1- and 2- were generated by reaction of the corresponding bis-2-propenylbenzenes with atomic oxygen
For example, whereas o-benzyne radical anion (C6H4-), formed by reaction of O- with benzene, reacts with SO2 and NO by electron transfer and loss of signal, respectively,30 ions 1- and 2- react with these reagents to form addition products. Moreover, the chemical properties (including CID branching ratios, vide infra) do not change when water is added to the ion source, ruling out the presence of other high-energy isomers, such as R,n-radical anions resulting from an abstraction from the ring, vinylidene anions resulting from H2+ abstraction from the methylene position,31-33 or vinyl carbene anions resulting from the abstraction from a single methyl group. These ions are all calculated to be 18-25 kcal/mol higher in energy than 1- and 2- and therefore should isomerize to the lower-energy structures with acid catalysis.34 The properties of the neutral diradicals have been investigated by using collision-induced dissociation (CID) of the CS2 adducts of 1- and 2- (eq 1). CID produces a mixture of the diradical anions and CS2-. The measured branching ratios (r ) I(1- or 2-)/I(CS2-)) at a collision energy of 25 V (laboratory frameof-reference) are shown in Table 1. The most important observation is that the branching ratio for the meta isomer is more than twice than that for the para isomer. We interpret this result to indicate that the meta diradical, 2, has a higher electron affinity than diradical, 1, such that it is more likely to retain the electron.
1336
J. Phys. Chem. A, Vol. 114, No. 3, 2010
Lenington and Wenthold
EA(R) ) 0.0742 ln(r) + 0.68 eV
Figure 4. Energy diagram showing the effect of spin state on the electron affinities of bis-allylbenzene diradicals. The state energies for each structure are shown relative to that for the radical anion, with the approximation that the triplet states have similar EAs. Electronic states refer to the ion, singlet (S), and triplet (T). Electron affinities are taken from Table 1.
Figure 5. Kinetic method plot of EA versus (r) for CS2 adducts of the carbanions. The black triangles refer to the data for the references shown in Table 1, used to create the calibrations shown in eq 2. The red and blue lines indicate the results for 1 and 2. The error bars reflect the uncertainties described in Table 1.
A higher electron affinity for 2 is consistent with what is expected if 1 is a singlet and 2 is a triplet. To a first approximation, the electron affinities of the triplet states of 1 and 2 would be expected to be similar. However, if 1 is a singlet, then the electron affinity will be lowered compared with that for 2, as shown in Figure 4. The difference in the electron affinities does not require that 2 has a triplet state but is consistent with that prediction. However, if 2 was to have a singlet state, its higher electron affinity requires that it have a much smaller singlet-triplet splitting than does 1. The lower electron affinity for 1 is even more striking considering that calculations at the B3LYP/6-311++G(3df,2p)//B3LYP/6-31+G* level of theory predict that the adiabatic electron affinity for the triplet state of 1 is slightly higher than that for the triplet state of 2 by 0.03 eV. The measured CS2 branching ratios (r) can be used with the kinetic method35,36 to obtain a quantitative estimate of the electron affinities. Although photodetachment methods for measuring electron affinities are more accurate,37-40 the simple kinetic method has been found to give reasonably accurate EA estimates,34,41-49 particularly for the relative EAs of structurally similar substrates.34,41,42,44 Branching ratios for a series of references50 are shown at the bottom of Table 1. From a plot of ln(r) versus EA (Figure 5), the relationship between electron affinity and CS2 branching ratio is found to be that shown in eq 2.
(2)
By using this expression with the measured branching ratios for 1- and 2-, we estimated the electron affinities of 1 and 2 to be 0.84 and 0.90 eV, respectively. According to the energy diagram in Figure 4, if 2 is a ground-state triplet and the triplet diradicals have similar electron affinities, then the difference in the electron affinities of 1 and 2 (∆EA ) 0.06 ( 0.03 eV) is an estimate of the singlet-triplet energy splitting in the para diradical. Moreover, if, as the calculations predict, the electron affinity of triplet 1 is higher than that of 2, then ∆EA is a lower limit to the singlet-triplet splitting. If, however, 2 is a groundstate singlet, then ∆EA would be the difference in the ∆EST values for the diradicals, with singlet 1 being relatively more stable than 2 by 0.06 ( 0.03 eV. The use of electron affinities to estimate singlet-triplet energy splittings in diradicals is similar, in principle, to Chen’s approach of estimating the singlet-triplet splittings by using the deviation in ionization energies (IEs).51 However, EAs are more difficult to use in this regard because, unlike what is found for IEs, the monoradicals are not good models for the triplet state. For example, as shown in Table 1, the electron affinity of 2-phenylallyl radical, obtained in this work by using the kinetic method, is found to be comparable to that of the singlet diradical, 1. Similarly, the electron affinity of 2-(2-propenyl)allyl radical (0.654 eV, Table 1) is much lower than that of the corresponding TME diradical, which has EA ) 0.855 eV.4 In the current case, a reasonable comparison can be made because we are using structurally related, disjoint diradicals of similar size. The triplet EA calculations are consistent with this assumption, indicating that the EAs agree within 0.03 eV. Moreover, the triplet state of 1 is predicted to be higher than that for 2, so to the extent that the assumption is incorrect, the calculations suggest that it would imply a larger singlet-triplet splitting for 1. Electronic structure calculations also provide evidence of the ground states of 1 and 2. At the B3LYP/6-31+G* level of theory, the optimized geometries for 1 and 2 (and for the anions) are nonplanar, with the allyl groups rotated out of the plane of the benzene ring. There are two nearly degenerate, stable structures for each isomer, depending on the relative orientation of the allyl groups. For the para isomer, the structures have D2 or C2h symmetry, and they are C2 and Cs symmetry for the meta isomer (Scheme 1). The allyl groups are rotated out of the plane by ca. 45° in all of the structures. The optimized geometries of the triplet states and radical anions are provided as Supporting Information. The energies of the structures for each isomer differ by only a small amount (
