Article pubs.acs.org/accounts
Singlet Oxygen Photophysics in Liquid Solvents: Converging on a Unified Picture Mikkel Bregnhøj,† Michael Westberg,† Boris F. Minaev,‡ and Peter R. Ogilby*,† †
Department of Chemistry, Aarhus University, DK-8000 Aarhus, Denmark Department of Natural Sciences, Bogdan Khmelnitsky National University, Cherkassy 18031, Ukraine
‡
S Supporting Information *
CONSPECTUS: Singlet oxygen, O2(a1Δg), the lowest excited electronic state of molecular oxygen, is an omnipresent part of life on earth. It is readily formed through a variety of chemical and photochemical processes, and its unique reactions are important not just as a tool in chemical syntheses but also in processes that range from polymer degradation to signaling in biological cells. For these reasons, O2(a1Δg) has been the subject of intense activity in a broad distribution of scientific fields for the past ∼50 years. The characteristic reactions of O2(a1Δg) kinetically compete with processes that deactivate this excited state to the ground state of oxygen, O2(X3Σ−g ). Moreover, O2(a1Δg) is ideally monitored using one of these deactivation channels: O2(a1Δg) → O2(X3Σ−g ) phosphorescence at 1270 nm. Thus, there is ample justification to study and control these competing processes, including those mediated by solvents, and the chemistry community has likewise actively tackled this issue. In themselves, the solvent-mediated radiative and nonradiative transitions between the three lowest-lying electronic states of oxygen [O2(X3Σ−g ), O2(a1Δg), and O2(b1Σ+g )] are relevant to issues at the core of modern chemistry. In the isolated oxygen molecule, these transitions are forbidden by quantum-mechanical selection rules. However, solvent molecules perturb oxygen in such a way as to make these transitions more probable. Most interestingly, the effect of a series of solvents on the O2(X3Σ−g )− O2(b1Σ+g ) transition, for example, can be totally different from the effect of the same series of solvents on the O2(X3Σ−g )− O2(a1Δg) transition. Moreover, a given solvent that appreciably increases the probability of a radiative transition generally does not provide a correspondingly viable pathway for nonradiative energy loss, and vice versa. The ∼50 years of experimental work leading to these conclusions were not easy; spectroscopically monitoring such weak and low-energy transitions in time-resolved experiments is challenging. Consequently, results obtained from different laboratories often were not consistent. In turn, attempts to interpret molecular events were often simplistic and/or misguided. However, over the recent past, increasingly accurate experiments have converged on a base of credible data, finally forming a consistent picture of this system that is resonant with theoretical models. The concepts involved encompass a large fraction of chemistry’s fundamental lexicon, e.g., spin−orbit coupling, state mixing, quantum tunneling, electronic-to-vibrational energy transfer, activation barriers, collision complexes, and charge-transfer interactions. In this Account, we provide an explanatory overview of the ways in which a given solvent will perturb the radiative and nonradiative transitions between the O2(X3Σ−g ), O2(a1Δg), and O2(b1Σ+g ) states.
■
INTRODUCTION The study of molecular oxygen and systems involving oxygen is a cornerstone of chemistry, physics, and biology. In part, this reflects the important role of oxygen on earth.1 Over the years, this seemingly simple homonuclear diatomic has provided challenging problems, the solutions of which define turning points in the progress of science.1 One example is the seminal work of Mulliken, in which molecular orbital arguments equate the paramagnetism of oxygen to a ground electronic state with triplet spin, O2(X3Σ−g ).2,3 It was soon acknowledged that the two lowest-energy excited electronic states of oxygen were singlet spin states, O2(a1Δg) and O2(b1Σ+g ), and that the energy differences between these three states provided a quantitative framework for Hund’s rules on the occupancy of degenerate orbitals (Figure 1).3 © 2017 American Chemical Society
Another scientific turning point was the recognition that O2(a1Δg), commonly called “singlet oxygen”, exhibits a unique set of chemical reactions through which organic molecules are oxygenated.4 These reactions play important roles in a wide range of systems, particularly those exposed to light.4−7 In contrast, analogous chemical reactions involving the other singlet state of oxygen, O2(b1Σ+g ), have yet to be found; interactions with a second molecule instead result in the efficient deactivation of O2(b1Σ+g ) to O2(a1Δg).8,9 When studying and exploiting the chemical reactions of O2(a1Δg), one must likewise account for kinetically competing processes that deactivate O2(a1Δg) to O2(X3Σ−g ), and solvent Received: April 5, 2017 Published: July 21, 2017 1920
DOI: 10.1021/acs.accounts.7b00169 Acc. Chem. Res. 2017, 50, 1920−1927
Article
Accounts of Chemical Research
solutes that quench or react with O2(a1Δg) (denoted as Q and R, respectively). For example, O2(a1Δg) is readily formed in a photosensitized experiment and, depending on the experimental conditions, the sensitizer itself can deactivate O2(a1Δg).8 To avoid this problem, O2(a1Δg) can be formed in sensitizer-free systems upon excitation via the X → b transition at 765 nm followed by b → a internal conversion.17,18 In most common solvents, the rate constant for a → X radiative deactivation, kaX r , is much smaller than that for nonradiative deactivation, kaX nr . Thus, the nonradiative channel generally defines τΔ, yielding solvent-dependent values that range from ∼3 μs to ∼200 ms.14 aX It should be noted that in writing kaX r and knr , we refer to −1 −1 bimolecular rate constants with units of M s . Over the years, many have instead used this same notation to denote pseudo-first-order rate constants with units of s−1. As discussed below, the difference between first- and second-order rate constants may have mechanistic implications. One can obtain kaX r by quantifying the number of photons emitted by O2(a1Δg), as reflected by the integrated intensity of the time-resolved O2(a1Δg) phosphorescence signal, IaX (eq 2):
Figure 1. Diagram of the three lowest-energy electronic states of oxygen with approximate transition energies. Radiative transitions are shown as solid arrows, whereas nonradiative transitions (i.e., loss of energy as heat) are shown as wavy arrows.
molecules play a pronounced role in this regard. In themselves, these solvent-mediated deactivation processes provide a wonderful model by which a number of fundamental issues can be investigated. In short, by examining the effect of the solvent on the radiative and nonradiative transitions between O2(X3Σ−g ), O2(a1Δg), and O2(b1Σ+g ), we can use oxygen as a tool to address problems that challenge our understanding of many chemical principles. Radiative transitions between the O2(X3Σ−g ), O2(a1Δg), and O2(b1Σ+g ) states are forbidden as electric dipole processes in the isolated oxygen molecule.3,10,11 However, interactions with the solvent increase the transition probabilities of not just the radiative processes but also the nonradiative processes. Quantifying the extent to which a given solvent influences a given transition has proven to be a challenge for the community, principally because the experiments involve monitoring weak transitions in picosecond to millisecond time-resolved experiments. Moreover, as can be seen in Figure 1, the pertinent transitions of O2(a1Δg) occur in the near-IR, a spectral region for which optical detectors often have limitations. Nevertheless, recent experiments have yielded increasingly credible data that can now be used to explain the different roles simultaneously played by the solvent in influencing oxygen’s electronic transitions.
I aX = γ ΦΔτΔk raX[M]
(2)
Different techniques can be used to produce O2(a1Δg), and the parameter ΦΔ quantifies the efficiency of the process used. The 1 product τΔkaX r [M] yields the fraction of O2(a Δg) molecules produced that are deactivated via the radiative channel. For −6 many solvents, τΔkaX r [M] ∼ 10 , indicating that the nonradiative channel indeed dominates the a → X transition. The parameter γ is difficult to quantify in an absolute sense; it reflects parameters that include, for example, the efficiency of collecting the emitted light. As such, most experiments are performed under conditions where relative values of kaX r are obtained, and this has been sufficient to quantify the effect of the solvent. Nevertheless, attempts have been made to determine absolute values of kaX r , and to date the community −1 in has generally relied upon a value of kaX r [M] = 1.5 ± 0.5 s 19 benzene. Expressions corresponding to eqs 1 and 2 can likewise be written for the b → a and b → X channels of O2(b1Σ+g ) deactivation. However, the lifetime of O2(b1Σ+g ), τΣ, is very short (
