UV-Photochemistry of the Disulfide Bond: Evolution of Early

Article Cite This: J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

pubs.acs.org/JACS

UV-Photochemistry of the Disulfide Bond: Evolution of Early Photoproducts from Picosecond X‑ray Absorption Spectroscopy at the Sulfur K‑Edge Miguel Ochmann,† Abid Hussain,† Inga von Ahnen,† Amy A. Cordones,‡ Kiryong Hong,§ Jae Hyuk Lee,‡ Rory Ma,§ Katrin Adamczyk,† Tae Kyu Kim,*,§ Robert W. Schoenlein,*,‡ Oriol Vendrell,*,∥ and Nils Huse*,† †

Department of Physics, University of Hamburg and Max Planck Institute for the Structure and Dynamics of Matter, Center for Free Electron Laser Science, 22761 Hamburg, Germany ‡ Ultrafast X-ray Science Lab, Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720, United States § Department of Chemistry and Chemistry Institute of Functional Materials, Pusan National University, Busan 46241, South Korea ∥ Center for Free-Electron Laser Science, DESY and The Hamburg Centre for Ultrafast Imaging, 22607 Hamburg, Germany S Supporting Information *

ABSTRACT: We have investigated dimethyl disulfide as the basic moiety for understanding the photochemistry of disulfide bonds, which are central to a broad range of biochemical processes. Picosecond timeresolved X-ray absorption spectroscopy at the sulfur K-edge provides unique element-specific insight into the photochemistry of the disulfide bond initiated by 267 nm femtosecond pulses. We observe a broad but distinct transient induced absorption spectrum which recovers on at least two time scales in the nanosecond range. We employed RASSCF electronic structure calculations to simulate the sulfur-1s transitions of multiple possible chemical species, and identified the methylthiyl and methylperthiyl radicals as the primary reaction products. In addition, we identify disulfur and the CH2S thione as the secondary reaction products of the perthiyl radical that are most likely to explain the observed spectral and kinetic signatures of our experiment. Our study underscores the importance of elemental specificity and the potential of time-resolved X-ray spectroscopy to identify short-lived reaction products in complex reaction schemes that underlie the rich photochemistry of disulfide systems.

■

INTRODUCTION Sulfur is an important element in biochemistry and is found in many proteins and enzymes either as metal−sulfur active sites1 or as one of two amino acids, L-methionine or L-cysteine, incorporated into the protein backbone.2 The latter is very important in structure determination of a protein’s tertiary structure as the side chains of two L-cysteine amino acids can be covalently linked via the oxidative formation of a disulfide bond. Since disulfide bonds strongly influence the tertiary structure of many proteins, they play an important role in the protein folding process and for structure retention. The stability of a protein’s structure is pivotal in maintaining its biological function. However, their structure and thus their function is dependent on many environmental factors such as temperature,3 pH,4 oxidizing or reducing environment,5 solvent,6 and radiation exposure to cosmic radiation,7 hard Xrays8,9 near-ultraviolet and visible radiation.10 It is well-known that the disulfide bond is most prone to cleavage when exposed to UV radiation. Investigation of the early time scale kinetics of © XXXX American Chemical Society

the UV induced disulfide bond breakage is therefore crucial in order to understand the photostability and photodamage repair mechanisms in proteins. Dimethyl disulfide (DMDS), the simplest organic disulfide, can serve as a model system for the photochemistry of disulfide bonds, and is a good starting point for developing a bottom-up understanding of the early time scale photochemistry of the disulfide bond motif in more complex molecules. Moreover, it is important to understand the contribution of DMDS and its UV-generated photoproducts to the atmospheric sulfur cycle, as DMDS is one of the major volatile organic sulfur compounds (VOSCs) and is considered very toxic for all organisms.11 Naturally, DMDS occurs as a decomposition product in the microbial degradation pathway of sulfur-containing amino acids12 and is also found as a luring agent in dead-horse arum (Helicodiceros muscivorus).13 DMDS also plays an Received: December 19, 2017

A

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

measurements and electronic structure calculations (RASSCF, DFT) yields a powerful method to follow sulfur chemistry with elemental specificity. In previous studies by Kennepohl and coworkers, static X-ray absorption spectroscopy was used to investigate radiation damage and photochemistry in biomolecules and smaller biologically relevant molecules.28,29 Another steady-state spectroscopic study of frozen solution and of room-temperature powder samples by Sneeden et al. has identified signatures of sulfur-centered radicals generated by hard X-ray exposure on a time scale of minutes to hours.30 These very long-lived species provide valuable information for crystallography and X-ray scattering of molecules in amorphous and crystalline solids subjected to hard X-ray photons. However, such sample conditions are very different from those found in living organisms at ambient conditions subjected to UV radiation where photochemical processes occur on ultrafast time scales via reaction pathways accessible by resonant excitations with few eV photons. In contrast, photons of several keV can generate large amounts of solvated electrons and may give rise to very long-lived matrix-trapped radicals. Herein, we quantify the lifetimes of different sulfur radicals under ambient conditions from the first time-resolved sulfur Kedge spectroscopy of a disulfide model system in a solvent environment. We establish the basic photochemical behavior of disulfide-containing linear molecules upon UV irradiation at 267 nm and address major discrepancies in the literature regarding product formation on tens of picosecond to submicrosecond time scales.

important role in the atmospheric sulfur cycle as a sulfur reserve and is primarily found emanating from swamps and salt marshes.14 DMDS has already been studied with both static and timeresolved optical methods, both in the gas phase15,16 and in solution.17 Early on, organic disulfides were found to form free radicals upon irradiation with suitable light, undergoing various subsequent chemical transformations rather than reforming the S−S bond.18 It was also established that the two major initial chemical reaction steps are breakage of the S−S and the C−S bonds, which yield thiyl and perthiyl radicals as reactive species, respectively.17 Furthermore, early time-resolved studies of the S−S bond breakage in 4-aminophenyl disulfide showed that the initial bond cleavage is ultrafast and takes places on a subpicosecond time scale.19 The first time-resolved optical studies on microsecond time scales of pure DMDS in the gas phase reported two dominant reaction pathways upon excitation with either mid-infrared multiphoton absorption or with UV light. Accordingly, it was assumed that DMDS would either decay via a molecular fourcentered transition state producing S2 and C2H6, or dissociate homolytically to yield CH3S· thiyl radicals.20 Further studies on the UV-photodissociation of aliphatic disulfides in the gas phase using time-of-flight mass spectrometry revealed two major photoproducts: (i) S−S bond cleavage, producing thiyl radicals or (ii) C−S bond cleavage, yielding perthiyl radicals.21 At excitation wavelengths shorter than 250 nm, S−S bond cleavage was reported to dominate, and C−S bond cleavage becomes competitive only at longer excitation wavelengths. From these findings Bookwalter et al. concluded that dissociation has to occur from an excited state and not from the ground state because the bond enthalpy of the C−S bond (235 kJ/mol, 2.4 eV) is lower than that of the S−S bond (280 kJ/mol, 2.9 eV). Resonance Raman spectroscopy with 267 nm excitation and calculations of potential energy surfaces for the lowest excited singlet states led Rinker et al.14 to conclude that excitation into the lowest singlet excited state favors a C−S bond cleavage due to a barrier for S−S bond dissociation, while the situation is reversed for second excited singlet state, featuring a barrier for C−S bond cleavage and favoring S−S bond dissociation. Rinker et al. concluded that 267 nm excitation accesses the first excited state and leads primarily to C−S bond cleavage, while excitation at 250 nm wavelengths promotes the system into the second excited state. More recently, theoretical studies with high-level SA-CASSCF and MS-CASPT2 approaches were applied by Luo et al. to investigate both the vertical excitation energies and the photodissociation mechanism of DMDS.22 These calculations indicate that both the S1 and S2 states strongly favor S−S bond fission due to a barrier for C−S bond cleavage, in disagreement with Rinker et al.14 According to Luo et al. C−S bond fission after 267 nm excitation should occur on the ground state potential energy surface. Clearly, the reaction pathways are not fully understood, nor are all products and their lifetimes unambiguously identified for a given excitation wavelength. Recently, time-resolved X-ray absorption spectroscopy (TRXAS) at the sulfur K-edge has been employed as a new tool for understanding photochemical reactions of sulfurcontaining compounds.23,24 TRXAS is very sensitive to the local electronic and chemical environment of sulfur. This sensitivity is reflected in sulfur-1s transitions shifting much more substantially than in other atoms upon changes in the binding state of the respective sulfur atom.25−27 Combining this spectral fingerprint of sulfur chemistry with time-resolved

■

METHODS

Chemicals and Materials. Dimethyl disulfide and cyclohexane were purchased from Sigma-Aldrich and were used as received without further purification. DMDS solutions (100 mM) were prepared by diluting 1.8 mL DMDS with cyclohexane to 200 mL. The sample was flowed and continuously measured for up to 6 h before being replaced by a fresh sample. Experimental Setup. All measurements were conducted at beamline 6.0.1 of the Advanced Light Source in Berkeley, CA. Details and capabilities of the experimental setup have been reported previously.23,31,32 Photoexcitation of the sample was achieved by 267 nm laser pulses of 100 fs duration obtained by third harmonic generation (THG) of the 800 nm fundamental wavelength of a Ti:Sa regenerative amplifier. A fluence of 50 mJ/cm2 was used for all measurements. Long delay scans at specific energies were corrected for electronic background variation by subtracting these electronic artifacts with no X-rays on the sample (see also Supporting Information). Static spectra were recorded with an integrating photodiode in total fluorescence yield (TFY) mode. Time-resolved spectra were recorded with an avalanche photodiode which was shielded from scattered laser light with a 100 nm thick aluminum foil. The sample was delivered through a 100 μm thick sapphire-nozzle liquid jet driven by a gear pump with flow rates of 3 m/s. Kinetic Modeling. We have modeled transient delay scans at three characteristic spectral positions with a sum of exponential decays multiplied by an error function with appropriate approximations to the argument of the error function (see the Supporting Information of ref 23):

ΔAa =

ΔAb = B

⎛ t − t0 ⎞⎫⎡ 1⎧ ⎨1 + erf⎜ ⎟⎬⎢a + ⎝ 2 σ ⎠⎭⎢⎣ 0 2⎩

∑ ai et− t0/ τai⎥

⎤

⎛ t − t0 ⎞⎫⎡ 1⎧ ⎨1 + erf⎜ ⎟⎬⎢b + ⎝ 2 σ ⎠⎭⎢⎣ 0 2⎩

∑ bi et− t0/ τbi⎥

i=1

⎥⎦

⎤

i=1

⎥⎦

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society ΔAc =

⎛ t − t0 ⎞⎫⎡ 1⎧ ⎨1 + erf⎜ ⎟⎬⎢c + ⎝ 2 σ ⎠⎭⎢⎣ 0 2⎩

⎤

∑ ci et− t0/ τci⎥ i=1

⎥⎦

The fit parameter t0 is global and was fit to all three transients at once. The width σ of the response function equals the X-ray pulse duration and was sufficiently well-known: σ = (30 ± 2) ps. The amplitudes {ai, bi, ci} were varied independently for each transient with the amplitudes {a0, b0, c0} being the signal amplitude at each of the three spectral positions at long delays, >150 ns. We considered two ways of fitting the data: (1) The transients were fit with two individual time constants each: {τai, τbi, τci}, i = 1,2. (2) The transients were fit with partial global time constants{τ12,τ23,τ13} where the time constant τij has been fit simultaneously to transients i and j. This was done because of pairwise similarities of the obtained time constants {τai, τbi, τci}, i = 1 and 2, when fitting the transients with individual biexponential decay models. Results of all model fits to the experimental data are shown in the Supporting Information. Rate constants kij are obtained from the fitted lifetimes τij via the relation kij = τij−1. Theoretical Calculations. The equilibrium geometries of all studied molecular species were obtained at the second order of Møller−Plesset perturbation theory33 (MP2) with the Dunning correlation consistent basis set aug-cc-pvtz.34 The vibrational frequency calculations confirm that the computed structures are at true energy minima. Geometry optimizations were obtained using very tight convergence criteria in the Gaussian09 suite of programs.35 The X-ray absorption transitions were simulated using the firstprinciples multireference restricted active space self-consistent field (RASSCF) method with no symmetry imposed. The active space for the simulated sulfur-1s transitions for dimethyl disulfide and the methylperthiyl radical comprised 14 electrons distributed over 12 orbitals. The active space for the methylthiyl radical was reduced to 11 electrons in 11 orbitals. The aug-cc-pvdz basis was used for all atomic centers. The active space in the RASSCF calculations is divided into two subspaces, namely, RAS2 containing all orbitals without occupational constraints and RAS3, comprising the two sulfur 1-s core electrons with at most one electron occupancy (thus suppressing configurations with a doubly filled sulfur-1s core orbital). The wave function during the RASSCF calculations and the presence of irrelevant low-energy configurations during the calculation of core excitations are avoided by grouping the excitations in this way. The energies of the core excitation obtained by RASSCF calculations were further improved by applying multistate secondorder perturbation theory (MS-RASPT2).36 For the ground-state energy, single-state CASPT237,38 was used with 0.25 hartree for the ionization potential electron affinity (IPEA) shift and an imaginary shift of 0.3 hartree to avoid intruder states. The RASSI module is used for the calculations of the transition dipole moments on the wave functions obtained from the RASSCF calculations. RASSCF/RASSI calculations were performed with the Molcas 7.8 program suite.39 The transition energies of all calculated species were shifted by −11.30 eV such that the lowest calculated DMDS transitions match the experimental DMDS spectrum.

■

Figure 1. (A) Sulfur K-edge spectrum of DMDS along with the computed lowest sulfur-1s transitions. (B) Differential absorption spectra at time delays of 0.1, 5, 50, and 150 ns after excitation with 267 nm pulses. (C) Time evolution of the absorbance changes at three distinct photon energies, indicated by color-coded arrows in panel B (spheres and dotted lines) along with exponential decays (solid lines) fit to the experimental data.

RESULTS AND DISCUSSION additional region of induced absorption centered at 2474.5 eV. The observed differential pump−probe signals develop fully within the time resolution of our experiment (∼70 ps) and recover in a multimodal fashion on time scales of subnanoseconds to tens of nanoseconds. Notably, even after 150 ns, some differential signals still persist. The reasons for this behavior are discussed below. We will first present a kinetic analysis of the transients in Figure 1C before we turn to the nature of the induced absorption at lower energy than groundstate bleaching. The induced absorption centered at 2474.5 eV is the net difference signal of ground-state bleaching and the sulfur-1s continuum absorption edges of multiple photo-

Figure 1A shows the static X-ray absorption spectrum of DMDS (1) in cyclohexane solution at the sulfur K-edge. Two major transitions are observed at 2471.6 and 2473.1 eV along with less pronounced transitions above 2475.0 eV, which merge into the continuum edge. Below, panel B shows the differential absorption spectrum at delays of 0.1, 5, 50, and 150 ns after 267 nm (4.64 eV) excitation. The negative absorbance change at 2471.5 and 2473.0 eV coincides with the main absorption peaks of the DMDS spectrum, signaling loss of ground-state absorption. Two distinct induced absorption peaks are observed at 2466.5 and 2467.4 eV which are accompanied by a broader shoulder between 2468.0 and 2470.0 eV and an C

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society products. It is therefore very congested and cannot provide chemically useful information. Figure 1C shows the recorded time traces at the three probe energies of 2466.5, 2467.4, and 2471.5 eV along with fits of a rate equation model to the experimental data. We have used a sum of exponential decays along with a long-lived signal to quantify the decay of the most prominent induced absorbance signals at 2466.5 eV (orange arrow) and 2467.4 eV (violet arrow). The underlying model (explained in the Methods section) takes into account similar pairwise decay behavior for the three transients when fit independently with biexponential decays: (i) The positive transient at 2466.5 eV and the negative one at 2471.5 eV indicate a subnanosecond decay. (ii) The positive transient at 2467.4 eV and the negative one at 2471.5 eV both exhibit a decay constant of a few nanoseconds. (iii) The positive transients at 2466.5 and 2467.4 eV decay on tens of nanoseconds. Assuming that this decay behavior is related, the model in Figure 1C contains three time constants: τ13 = (0.3 ± 0.2) ns, τ23 = (4.5 ± 1.0) ns, and τ12 = (21 ± 4) ns. Transient 1 at 2466.5 eV is modeled by a biexponential decay with time constants τ13 and τ12. Transient 2 at 2467.4 eV is modeled by a biexponential decay with time constants τ23 and τ12. Transient 3 at 2471.5 eV is modeled by a biexponential decay with time constants τ13 and τ23. The induced absorption signals persisting beyond the longest measured delay time of 150 ns define an upper limit for a slower rate constant k∞ = 1/τ∞ ≤ 7 μs−1. Alternative fitting schemes (as presented in the Supporting Information) result in similar time constants and reflect an overall behavior: The bleach recovery occurs on nanosecond time scales (τ23) and after about 20 ns reaches a value that is nearly constant for hundreds of nanoseconds, while the induced absorption signals also decay on longer time scales (τ12). The subnanosecond time constant τ13 is marginally significant and it will be interesting to see if such a nanosecond constant manifests in other disulfides. We will discuss the observed kinetics in the context of different sulfur species. The sulfur K-edge spectrum of DMDS in Figure 1A is welldescribed by three distinct resonant sulfur 1s → 3p transitions and a weaker line that produces the shoulder at 2473.8 eV. The calculated transitions are convolved with a Voigt line shape function with a best match to experiment for a Gaussian X-ray monochromator broadening of σ = 0.38 eV and Lorentzian lifetime broadening of Γ = 0.35 eV. Our RASSCF electronic structure calculations provide an accurate approach to describing the bound−bound transitions including all electronic correlations required for core-excited state calculations.40 To describe the Rydberg series and the atomic absorption edge, other theoretical approaches (time-dependent density functional and scattering theory) are more suitable. Two major initial reaction pathways in the UV induced photochemistry of DMDS (1) have previously been proposed: S−S and C−S bond cleavage, yielding methylthiyl and methylperthiyl radicals, henceforth abbreviated as thiyl (I) and perthiyl (II) radical, respectively. Figure 2 shows the species that have been discussed in the literature along with the calculated sulfur-1s transitions overlaid with the differential absorption spectrum at 0.1 ns delay. The thiyl radical (I) exhibits a theoretical sulfur-1s spectrum that fits our experimentally observed induced absorption feature at 2466.5 eV similar to our previous findings for an aromatic thiol.23 The theoretical sulfur-1s spectrum of the perthiyl radical (II) exhibits two dominant transitions of which the energetically

Figure 2. (A) Possible sulfur species are displayed in a color-coded fashion with roman numerals assigned to transient species, i.e., radicals. (B) RASSCF-computed energetically lowest sulfur-1s transitions for all species are overlaid with the differential absorption spectrum at a time delay of 0.1 ns after 267 nm excitation. The dominant species are thiyl and perthiyl radicals.

lower one matches the induced absorption feature at 2467.4 eV. The second predicted transition at 2471.5 eV overlaps with the lowest DMDS transition and would therefore be masked by the ground-state bleaching signal. Our theoretical results show very good agreement with the two prominent radicals, I (thiyl) and II (perthiyl), and we can already conclude that these two radical species exist at 100 ps delay, thereby settling the question for 267 nm excitation of which product is created because I and II do not interconvert. S−S bond cleavage produces two thiyl radicals, while breaking the C−S bond creates only one perthiyl radical. The calculated transitions strengths in Figure 2B would predict a quantum yield for C−S bond cleavage about twice as high as S−S bond cleavage (cf. Supporting Information). However, we note that the accuracy of the RASSCF transition strengths is only as good as the wave function modeling the quantum system. Product yields are therefore estimates based on theoretically predicted transition strengths. From a spectroscopic point of view, the character of the lowest sulfur-1s transitions in DMDS as well as the thiyl and perthiyl radicals is illustrated in Figure 3. The lowest sulfur-1s transitions in DMDS is well-described by a 1s-to-π* transition of the S−S bond orbital. The second transition is dominated by a 1s-to-σ* transition of the C−S bond with the excited state formally being a superposition state of LUMO+1 and LUMO +2. The third pronounced sulfur-1s transition in DMDS is situated closest to the continuum edge and is mainly a LUMO +4 excitation. For the radicals, the dominant transitions are excitations into the singly occupied molecular orbital (SOMO). This interpretation is consistent with an isolated lowest transition of thiyl radicals reported by Sneeden and coworkers.30 An additional significant transition is predicted for the perthiyl radical as a SOMO+4 excitation. Further D

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

spontaneous decay of hot II as well.45 Several other studies concluded that disulfur (2) is a decay product of UV excited perthiyl radical II.46−48 It is important to note that conventional flash photolysis studies employ lasers which provide photons for many nanoseconds during which primary product formation occurs. Therefore, the use of nanosecond lasers cannot generally rule out sequential photolysis. We calculated S2 as a possible product of the photoreaction. Four theoretical transitions of disulfur (2) emerge, of which two are found at 2468.8 and 2469.3 eV, while the higher-lying transitions would again be masked by the bleach signal of the missing DMDS ground state absorption. The energetically lowest transitions of S2 could explain the observed shoulder between 2468.5 and 2470.0 eV. Because S2 is not a stable species under ambient conditions, experimental spectra for comparison are scarce. However, a publication by Rühl et al. reports on induced absorption interpreted as the signature of S2 which shows very good agreement in spectral position to our RASSCF predictions.49 The data also suggests a weaker lowest transition strength compared to signatures of thiyl radicals as published recently by Sneeden et al.,30 when considering the continuum absorption edge as reference in both publications. Such disulfur formation would also point to the conclusions from earlier studies20,43,45 of spontaneous disulfur (2) elimination from hot II since our initial UV excitation pulses are 100 fs in duration, too short for perthiyl generation and subsequent photolysis by our pump pulses. Since the bond enthalpy of a C−S bond has been reported to be about 2.4 eV while the exciting photons carry 4.6 eV, it is certainly valid to assume that the nascent radical species are vibrationally hot enough for a perthiyl radical to decay into disulfur and a methyl radical. The latter combining with another methyl radical to form ethane would also explain the findings of Kumar et al. via a sequential process. Disulfur may combine with other disulfur molecules to form elementary sulfur (S8). Both sulfur species feature considerably weaker resonant transitions50 than DMDS which could contribute to the persistent ground state bleaching signal on time scales beyond 150 ns. Thiyl and perthiyl radicals can also combine to form trisulfides and tetrasulfides, both of which will also have X-ray absorption near-edge spectra different from DMDS. These larger sulfur-dimerized molecules can form through diffusion-limited bimolecular reactions. The time scale set by the diffusion limit is indeed sufficiently short (ca. 6 ns) that detectable concentrations of such species are plausible (see the Supporting Information for details). As very similar decay kinetics of the thiyl (I) and perthiyl (II) radicals on tens of nanoseconds are absent in the bleach recovery kinetics and both, I and II, have comparable size and hence diffusivity, we interpret the rate constant k12 as an effective rate constant for diffusion-limited bimolecular reactions involving thiyl and perthiyl radicals. We end our discussion with the decay of the perthiyl radical II with a suggestion by Cole-Filipiak et al.47,51 which considers another reaction channel that would lead to the formation of a hydrogen sulfide radical (III) and thioformaldehyde (3). Our calculated sulfur-1s absorption spectrum for III shows one major transition at 2465.1 eV and another one with roughly half the intensity at 2472.6 eV. The latter one would again be masked by the ground state bleach signal of DMDS, but the first transition at 2465.1 eV overlaps with a small signal at the rising edge of the induced absorption of the thiyl radical (I). This signal is barely significant, but the simultaneously

Figure 3. Molecular orbitals and corresponding isosurfaces relevant to the lowest sulfur-1s transitions in the three indicated species: DMDS (1), thiyl radical (I), and perthiyl radical (II)).

information can be found in the Supporting Information. We note that these transitions have been described in a oneelectron picture by time-dependent density functional theory (TDDFT) with the same character of the lowest transitions.28 The differential spectra in Figure 1B and the transients in Figure 1C provide interesting additional information: The thiyl radical appears to exhibit a small subnanosecond relaxation (τ13 = 0.3 ns) that we tentatively attribute to geminate recombination. The dominant relaxation occurs over the course of tens of nanoseconds (τ12 = 21 ns). The perthiyl radical (II) appears to decay more rapidly (τ23 = 4 ns) than the majority of the thiyl radicals (although without an apparent subnanosecond component) which−given the similarity to the nanosecond bleach recovery−possibly contributes to the latter one. It too exhibits a slower decay on tens of nanoseconds that seems absent in the bleach recovery. Overall, thiyl radicals persist longer than perthiyl radicals. This behavior seems to contradict the notion that perthiyl radicals are more stable than thiyl radicals. For instance, acid−base equilibria of hydropersulfides have been explained by the relative stability of the perthiyl radical41 due to a partial double-bond character of the S−S bond.42 However, Callear and Dickson argue that statistical partitioning of excess energy will lead to a less stable perthiyl radical compared to the thiyl radical when being photogenerated from DMDS.43 Their study reports S2 formation upon UV excitation of DMDS which we will discuss in the following. Our differential absorption spectrum in Figure 1B exhibits a pronounced shoulder (∼2469 eV) that cannot be explained by sulfur-1s transitions in either thiyl or perthiyl radicals. One possible explanation is the formation of DMDS cations by twophoton absorption with a predicted dominant sulfur-1s transition at 2469.1 eV (Figure S3). However, two-photon absorption (TPA) cross sections from a more recent study44 point to insignificant TPA in DMDS (see the Supporting Information for details). Microsecond time-resolved studies by Kumar et al. suggested that a possible reaction pathway could lead to the production of S2 and ethane.20 Another earlier study promotes the idea of disulfur (2) being generated by E

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society Scheme 1. Reaction Pathways of Dimethyl Disulfide upon 267 nm Excitation

dithiane takes place within picoseconds via an internal conversion process.54,55 For comparison, this time scale is slower than excited state quenching in some DNA sequences56−58 but comparable if not faster than that of intramolecular vibrational energy redistribution (IVR). The latter would allow the carbon backbone of the protein to move the two radicals away from each other resulting in an alteration of the tertiary structure. Comparison of these findings for the cyclic disulfide 1,2-dithiane with the linear diethyl disulfide (DEDS) further points toward the relative photostability of the disulfide bonds in proteins arising from a structural property that spatially confines the sulfur radical and permits efficient energy dissipation.53,54 The disulfide bond may therefore act as radiation shield in proteins, protecting the integrity of the proteins tertiary structure by absorption of harmful radiation and benign dissipation of energy by ultrafast radical recombination and subsequent vibrational energy redistribution.54

generated thione (3) would have a predicted transition at 2467.8 eV. Such a formation process would explain the broad induced absorption between the predicted perthiyl and disulfur transitions. Scheme 1 summarizes our data analysis and discussion on the 100 ps to submicrosecond time scales of the photochemistry of DMDS observed with sulfur K-edge TRXAS. Upon illumination with 267 nm light, two dominant reaction pathways are accessible: S−S and C−S bond cleavage to yield methylthiyl (I) and methylperthiyl (II). Both I and II have distinct lowestenergy sulfur-1s transitions as shown in Figure 2B. Disulfur (2) is one of the likely decay products of II as is the pairwise formation of a hydrogen sulfide radical (III) and thioformaldehyde (3). We interpret the largest relaxation rate k1,3 that we associated with the thiyl radical (I) as geminate recombination of two thiyl radicals. We assume that this component is missing for the perthiyl radical (II) due to the relative small size of the ejected methyl radical which quickly diffuses into the solvent shell. We associate the intermediate rate constant k2,3 with decay reactions of the perthiyl radical (II) including ground state recovery through recombination with a methyl radical. Finally, we interpret the slowest rate constant k1,2 as an effective rate for diffusion-limited irreversible secondary reactions such as polysulfide formation, contributing to the incomplete ground state recovery. The relatively complex reaction scheme highlights the need for advanced spectroscopies. It would certainly be interesting to extend this study to other disulfide systems that relate to important research fields such as aerosol chemistry or radical migration in proteins. Furthermore, femtosecond X-ray sources in the 2−3 keV photon energy range are able to access the very early dynamics of the reactions, allowing better study of the mechanisms by which product formation occurs. It will also be interesting to build upon our findings and devise small protein systems with disulfide bonds in welldefined configurations to study the sulfur photochemistry for ambient protein conditions and connect to work such as recent time-resolved mass spectrometry (TRMS) and nonadiabatic multiconfigurational molecular dynamics studies of model disulfide compounds.52−54 These studies indicate that structural restrictions render disulfide bonds in proteins more stable because the photogenerated radicals remain in the vicinity of each other to geminately recombine. For instance, the restoration of the disulfide bond in the model system 1,2-

■

CONCLUSIONS We have identified the thiyl and perthiyl radicals as the primary reaction products of a model disulfide system (dimethyl disulfide) in solution under ambient conditions excited with 267 nm femtosecond pulses. These primary photoproducts are generated through S−S and C−S bond cleavage, respectively. Additional spectral absorption features provide evidence for disulfur and thio-formaldehyde formation on picosecond time scales as secondary photoproducts. Kinetic modeling of the experimental data provides effective rate constants which we associate with reversible and irreversible reactions subsequent to generation of the primary photoproducts. This study underscores the potential and usefulness of time-resolved Xray spectroscopy of heteroatoms under ambient conditions23,24,59−63 by virtue of its elemental specificity and sensitivity to electronic structure changes and altered chemical bonding. With an increasing number of X-ray free-electron lasers providing new instrumentation, it will be possible to carry our study into the femtosecond regime to observe the early time evolution of the photoinduced reaction with less ambiguity and formation times. The latter will help unravel the reaction mechanisms that lead to the rich photochemistry of sulfur-containing systems such as (bio)macromolecules where the reaction pathways are altered by the surroundings. These include proteins, which seem to lack perthiyl formation, F

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society

thank Bruce Rude for his continuous support of the experimental hardware.

or other polymer blends in materials applications, e.g., photovoltaics or molecular electronic devices.

■

■

ASSOCIATED CONTENT

S Supporting Information *

(1) Bingaman, J. L.; Kohnhorst, C. L.; Van Meter, G. A.; McElroy, B. A.; Rakowski, E. A.; Caplins, B. W.; Gutowski, T. A.; Stromberg, C. J.; Webster, C. E.; Heilweil, E. J. J. Phys. Chem. A 2012, 116, 7261−7271. (2) Creighton, T. E. BioEssays 1988, 8, 57−63. (3) Matsuura, Y.; Takehira, M.; Joti, Y.; Ogasahara, K.; Tanaka, T.; Ono, N.; Kunishima, N.; Yutani, K. Sci. Rep. 2015, 5, 15545. (4) Miesenbock, G.; De Angelis, D. A.; Rothman, J. E. Nature 1998, 394, 192−195. (5) Davies, M. Biochem. J. 2016, 473, 805−825. (6) Yasuda, S.; Oshima, H.; Kinoshita, M. J. Chem. Phys. 2012, 137, 135103. (7) Parihar, V. K.; Allen, B. D.; Caressi, C.; Kwok, S.; Chu, E.; Tran, K. K.; Chmielewski, N. N.; Giedzinski, E.; Acharya, M. M.; Britten, R. A.; Baulch, J. E.; Limoli, C. L. Sci. Rep. 2016, 6, 34774. (8) Utschig, L. M.; Chemerisov, S. D.; Tiede, D. M.; Poluektov, O. G. Biochemistry 2008, 47, 9251−9257. (9) Weik, M.; Ravelli, R. B. G.; Kryger, G.; McSweeney, S.; Raves, M. L.; Harel, M.; Gros, P.; Silman, I.; Kroon, J.; Sussman, J. L. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 623−628. (10) Baxter, R. H. G.; Seagle, B.-L.; Ponomarenko, N.; Norris, J. R. J. Am. Chem. Soc. 2004, 126, 16728−16729. (11) Ito, T.; Miyaji, T.; Nakagawa, T.; Tomizuka, N. Biosci., Biotechnol., Biochem. 2007, 71, 366−370. (12) Buzzini, P.; Gasparetti, C.; Turchetti, B.; Cramarossa, M. R.; Vaughan-Martini, A.; Martini, A.; Pagnoni, U. M.; Forti, L. Arch. Microbiol. 2005, 184, 187−193. (13) Stensmyr, M. C.; Urru, I.; Collu, I.; Celander, M.; Hansson, B. S.; Angioy, A.-M. Nature 2002, 420, 625−626. (14) Rinker, A.; Halleman, C. D.; Wedlock, M. R. Chem. Phys. Lett. 2005, 414, 505−508. (15) Yokozeki, A.; Bauer, S. H. J. Phys. Chem. 1976, 80, 618−625. (16) Grabowski, J. J.; Zhang, L. J. Am. Chem. Soc. 1989, 111, 1193− 1203. (17) Morine, G. H.; Kuntz, R. R. Photochem. Photobiol. 1981, 33, 1− 5. (18) Lyons, W. E. Nature 1948, 162, 1004. (19) Ernsting, N. P. Chem. Phys. Lett. 1990, 166, 221−226. (20) Kumar, A.; Chowdhury, P.; Rao, K. R.; Mittal, J. Chem. Phys. Lett. 1992, 198, 406−412. (21) Bookwalter, C. W.; Zoller, D. L.; Ross, P. L.; Johnston, M. V. J. Am. Soc. Mass Spectrom. 1995, 6, 872−876. (22) Luo, C.; Du, W.-N.; Duan, X.-M.; Liu, J.-Y.; Li, Z.-S. Chem. Phys. Lett. 2009, 469, 242−246. (23) Ochmann, M.; von Ahnen, I.; Cordones, A. A.; Hussain, A.; Lee, J. H.; Hong, K.; Adamczyk, K.; Vendrell, O.; Kim, T. K.; Schoenlein, R. W.; Huse, N. J. Am. Chem. Soc. 2017, 139, 4797−4804. (24) Van-Kuiken, B. E.; Ross, M. R.; Strader, M. L.; Cordones, A. A.; Cho, H.; Lee, J. H.; Schoenlein, R. W.; Khalil, M. Struct. Dyn. 2017, 4, 044021. (25) George, G. N.; Gorbaty, M. L. J. Am. Chem. Soc. 1989, 111, 3182−3186. (26) Pickering, I. J.; Prince, R. C.; Divers, T.; George, G. N. FEBS Lett. 1998, 441, 11−14. (27) Rompel, A.; Cinco, R. M.; Latimer, M. J.; McDermott, A. E.; Guiles, R. D.; Quintanilha, A.; Krauss, R. M.; Sauer, K.; Yachandra, V. K.; Klein, M. P. Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 6122−7. (28) Martin-Diaconescu, V.; Kennepohl, P. J. Am. Chem. Soc. 2007, 129, 3034−3035. (29) Karunakaran-Datt, A.; Kennepohl, P. J. Am. Chem. Soc. 2009, 131, 3577−3582. (30) Sneeden, E. Y.; Hackett, M. J.; Cotelesage, J. J. H.; Prince, R. C.; Barney, M.; Goto, K.; Block, E.; Pickering, I. J.; George, G. N. J. Am. Chem. Soc. 2017, 139, 11519−11526. (31) Cho, H.; Hong, K.; Strader, M. L.; Lee, J. H.; Schoenlein, R. W.; Huse, N.; Kim, T. K. Inorg. Chem. 2016, 55, 5895−5903.

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/jacs.7b13455. Temporal modeling, possible cation formation, relative yield analysis, diffusion-limited product formation, energy calibration, and RASSCF calculations (PDF)

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: *E-mail: *E-mail: *E-mail:

REFERENCES

[email protected]. [email protected]. [email protected]. [email protected].

ORCID

Tae Kyu Kim: 0000-0002-9578-5722 Oriol Vendrell: 0000-0003-4629-414X Nils Huse: 0000-0002-3281-7600 Present Addresses

A.H.: Department of Physics and Applied Mathematics, Pakistan Institute of Engineering and Applied Sciences, Nilore, Islamabad, Pakistan. I.v.A.: Coherent LaserSystems GmbH & Co. KG, Seelandstrasse 9, 23569 Lübeck, Germany. A.A.C.: PULSE Institute, SLAC National Accelerator Laboratory, Menlo Park, CA 94025, United States. K.H.: Center for Gas Analysis, Division of Chemical and Medical Metrology, Korea Research Institute of Standards and Science, Daejeon 34113, Republic of Korea. J.H.L.: Pohang Accelerator Laboratory, San-31 Hyoja-dong Pohang, Kyungbuk 790−784, South Korea. O.V.: Department of Physics and Astronomy, Aarhus University, 8000 Aarhus, Denmark. R.W.S.: Linac Coherent Light Source, SLAC National Accelerator Laboratory, Menlo Park, CA 94025, United States. Author Contributions

M.O. and A.H. contributed equally to this work. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, the Chemical Sciences, Geosciences, and Biosciences Division under the Department of Energy, Contract No. DE-AC02-05CH11231 (A.C., K.H., J.L., and R.W.S.). This research has been supported by grants of the National Research Foundation of Korea (NRF) funded through the Ministry of Science and ICT (No. 2016R1E1A1A01941978, 2014R1A4A1001690, and 2016K1A4A4A01922028) to K.H., R.M., and T.K.K. M.O., I.v.A., K.A., and N.H. acknowledge funding from the Max Planck Society and the City of Hamburg. M.O. and N.H. gratefully acknowledge financial support by the German Science Foundation (DFG) through the SFB 925 “Light induced dynamics and control of correlated quantum systems”. This research used resources of the Advanced Light Source (LBNL), which is a DOE Office of Science User Facility. We G

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Article

Journal of the American Chemical Society (32) Van Kuiken, B. E.; Huse, N.; Cho, H.; Strader, M. L.; Lynch, M. S.; Schoenlein, R. W.; Khalil, M. J. Phys. Chem. Lett. 2012, 3, 1695. (33) Møller, C.; Plesset, M. S. Phys. Rev. 1934, 46, 618−622. (34) Dunning, T. H. J. Chem. Phys. 1989, 90, 1007−1023. (35) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin, K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09, revision E.01; Gaussian, Inc.: Wallingford, CT, 2009. (36) Finley, J.; Malmqvist, P.-Å.; Roos, B. O.; Serrano-Andrés, L. Chem. Phys. Lett. 1998, 288, 299−306. (37) Andersson, K.; Malmqvist, P. Å.; Roos, B. O.; Sadlej, A. J.; Wolinski, K. J. Phys. Chem. 1990, 94, 5483−5488. (38) Andersson, K.; Malmqvist, P.-Å.; Roos, B. O. R. J. Chem. Phys. 1992, 96, 1218−1226. (39) Aquilante, F.; De Vico, L.; Ferré, N.; Ghigo, G.; Malmqvist, P. Å.; Neogrády, P.; Pedersen, T. B.; Pitonák, M.; Reiher, M.; Roos, B. O.; Serrano-Andrés, L.; Urban, M.; Veryazov, V.; Lindh, R. J. Comput. Chem. 2010, 31, 224−247. (40) Hussain, A.; Huse, N.; Vendrell, O. Struct. Dyn. 2017, 4, 054102. (41) Ono, K.; Akaike, T.; Sawa, T.; Kumagai, Y.; Wink, D. A.; Tantillo, D. J.; Hobbs, A. J.; Nagy, P.; Xian, M.; Lin, J.; Fukuto, J. M. Free Radical Biol. Med. 2014, 77, 82−94. (42) Benson, S. W. Chem. Rev. 1978, 78, 23−35. (43) Callear, A. B.; Dickson, D. R. Trans. Faraday Soc. 1970, 66, 1987−1995. (44) Dragonmir, A.; McInerney, J. G.; Nikogosyan, D. N. Appl. Opt. 2002, 41, 4365−4376. (45) Martinez-Haya, B.; Bass, M. J.; Brouard, M.; Vallance, C.; Torres, I.; Barr, J. J. Chem. Phys. 2004, 120, 11042−11052. (46) Nourbakhsh, S.; Liao, C.-L.; Ng, C. Y. J. Chem. Phys. 1990, 92, 6587−6593. (47) Cole-Filipiak, N. C.; Negru, B.; Just, G. M. P.; Park, D.; Neumark, D. M. J. Chem. Phys. 2013, 138, 054301. (48) Harrison, A. W.; Ryazanov, M.; Sullivan, E. N.; Neumark, D. M. J. Chem. Phys. 2016, 145, 024305. (49) Rühl, E.; Flesch, R.; Tappe, W.; Novikov, D.; Kosugi, N. J. Chem. Phys. 2002, 116, 3316−3322. (50) Hayter, C. E.; Evans, J.; Corker, J. M.; Oldman, R. J.; Peter Williams, B. J. Mater. Chem. 2002, 12, 3172−3177. (51) Cole-Filipiak, N. C.; Shapero, M.; Haibach-Morris, C.; Neumark, D. M. J. Phys. Chem. A 2016, 120, 4818−4826. (52) Stephansen, A. B.; Brogaard, R. Y.; Kuhlman, T. S.; Klein, L. B.; Christensen, J. B.; Sølling, T. I. J. Am. Chem. Soc. 2012, 134, 20279− 20281. (53) Stephansen, A. B.; Larsen, M. A.; Klein, L. B.; Sølling, T. I. Chem. Phys. 2014, 442, 77−80. (54) Rankine, C. D.; Nunes, J. P. F.; Robinson, M. S.; Lane, P. D.; Wann, D. A. Phys. Chem. Chem. Phys. 2016, 18, 27170−27174. (55) Wolf, T. J. A.; Myhre, R. H.; Cryan, J. P.; Coriani, S.; Squibb, R. J.; Battistoni, A.; Berrah, N.; Bostedt, C.; Bucksbaum, P.; Coslovich, G.; et al. Nat. Commun. 2017, 8, 29. (56) Crespo-Hernandez, C. E.; Cohen, B.; Kohler, B. Nature 2005, 436, 1141−1144. (57) Crespo-Hernandez, C. E.; Cohen, B.; Kohler, B. Nature 2006, 441, E8. (58) Schwalb, N. K.; Temps, F. Science 2008, 322, 243−245.

(59) Wernet, P.; Gavrila, G.; Godehusen, K.; Weniger, C.; Nibbering, E. T. J.; Elsaesser, T.; Eberhardt, W. Appl. Phys. A: Mater. Sci. Process. 2008, 92, 511−516. (60) Wen, H.; Huse, N.; Schoenlein, R. W.; Lindenberg, A. M. J. Chem. Phys. 2009, 131, 234505. (61) Van Kuiken, B. E.; Cho, H.; Hong, K.; Khalil, M.; Schoenlein, R. W.; Kim, T. K.; Huse, N. J. Phys. Chem. Lett. 2016, 7, 465−470. (62) Eckert, S.; et al. Angew. Chem., Int. Ed. 2017, 56, 6088−6092. (63) Fondell, M.; et al. Struct. Dyn. 2017, 4, 54902.

H

DOI: 10.1021/jacs.7b13455 J. Am. Chem. Soc. XXXX, XXX, XXX−XXX