Greg T.
Dunning
a,
Thomas J.
Preston
a,
Andrew J.
Orr-Ewing
*a,
Stuart J.
Greaves
b,
Gregory M.
Greetham
c,
Ian P.
Clark
c and
Michael
Towrie
c
aSchool of Chemistry, University of Bristol, Cantock's Close, Bristol BS8 1TS, UK. E-mail: a.orr-ewing@bristol.ac.uk; Tel: +44 (0)117 9287672
bSchool of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, UK
cCentral Laser Facility, Research Complex at Harwell, Science and Technology Facility Council, Rutherford Appleton Laboratory, Harwell Science and Innovation Campus, Didcot, Oxfordshire, OX11 0QX, UK
First published on 23rd June 2014
Transient electronic absorption measurements with 1 ps time resolution follow XeF2 photoproducts in acetonitrile and chlorinated solvents. Ultraviolet light near 266 nm promptly breaks one Xe–F bond, and probe light covering 320–700 nm monitors the products. Some of the cleaved F atoms remain in close proximity to an XeF fragment and perturb the electronic states of XeF. The time evolution of a perturbed spectral feature is used to monitor the FXe–F complex population, which decays in less than 5 ps. Decay can occur through geminate recombination, diffusive separation or reaction of the complex with the solvent.
Condensed-phase photolysis can parallel its gas-phase analogue,7,8 but the presence of the solvent can modify photodissociation dynamics in several ways. The surrounding molecules restrict the system from reaching complete separation, which can lead the nascent species to sample metastable complexes that can be non-intuitive to chemists. The solvent molecules can also affect the photochemistry by altering the electronic states involved in photolysis or by exchanging energy with the reactive complex during its separation. With small molecules like XeF2 that have known spectroscopic features9,10 and relatively few reaction pathways,4,5,11 we can build on our chemical intuition of condensed-phase chemistry.
Our current effort uses time-resolved transient absorption spectroscopy to determine the condensed-phase dynamics and products of XeF2 from 1 ps to 1.5 ns after photodissociation. This work provides ready analysis of XeF2 as a source of F atoms for use in our on-going studies of liquid-phase bimolecular reaction dynamics.12 Following absorption of a 266 nm photon by XeF2, homolytic fission of one Xe–F bond is predicted to dominate the gas-phase dynamics.6 Facile migration of photolytically released F atoms occurs in rare-gas matrices,13 but reaction with the solvent may limit the F-atom mobility in organic solvents.
Application of XeF in excimer lasers provided some original impetus for studying the ground and electronically excited XeF states.10,14–17Fig. 1 shows the electronic states of XeF that are important in 266 nm photolysis of XeF2. One-photon absorption by XeF2 leads to ground-state XeF through Xe–F bond cleavage, and its strong absorption to the B state near 345 nm provides a probe of the XeF photoproduct.2 The ground state is bound by about 1200 cm−1 in the gas phase,10 and a large fraction of the XeF products remain bound well beyond the timescale of our measurement.2 The electronically excited states of XeF are ionic in character, and an F atom in the vicinity of XeF perturbs the optically bright B state.18 We use this known shift of the XeF spectrum as a marker for incomplete separation of the condensed-phase products, allowing us to follow with detail the condensed-phase dynamics that occur within the first few ps after photolysis of XeF2.
Fig. 1 Schematic diagram of the bound, gas phase, electronic states of XeF relevant in this work, adapted from Zerza et al.18 The ion-pair character of the B state makes it susceptible to electronic perturbations. The red curve shows the shift of the B state caused by a neighbouring F atom, first identified by Apkarian and co-workers.18 |
Imperfect alignment of the beam along the optical delay line that controls the relative timing of photolysis and probe pulses can affect signal intensities at Δt ≥ 100 ps. To compensate for this we use the known excited-state lifetime of phenol in cyclohexane21 to apply a linear correction factor to the transient XeF2 photolysis data we present. This correction yields long-time kinetics in the XeF2 systems that are consistent with previous measurements.2 The spectra are not corrected for the temporal chirp in the WLC: instead, data for time delays <1 ps are excluded from analysis.
The solute XeF2 is fragile, and keeping samples colder than 290 K before and during measurements helps ensure its stability. Dried glassware, solvents, and sample holders prevent unwanted side reactions between XeF2 and water. Rapid thermal reaction of the possible contaminant XeF4 with solvent5 ameliorates its effect on our measurements. Acquiring a pre-determined set of time delays between photolysis and probe pulses in random order further reduces the effects of longer timescale changes in the sample on our measurement. Dilute solutions of 0.25 M XeF2 (>99.9% trace metal basis, Sigma Aldrich) in dried CH3CN, CCl4, CDCl3, or CD2Cl2 pass through a sealed flow system. A rastering sample cell, consisting of CaF2 windows separated by a 0.2 mm PTFE spacer, ensures that each laser pulse interrogates a fresh sample volume.
Fig. 2 Sketch of the XeF2 absorption spectrum, electronic states, and energy of photoproducts, adapted from Kono et al.27 One- and two-photon absorption (indicated by the horizontal dashed lines) by XeF2 give many energetically accessible photoproducts, but a linear power dependence of the XeF(B–X) feature has confirmed that only one photon processes occur within our experimental setup. |
Previous work has identified the condensed-phase XeF absorption spectrum in both steady-state cryogenic matrices16,28 and in liquid samples of photolysed XeF2.2 We use these prior assignments to guide the interpretation of our measurements, which are the first to follow this process on the ps to ns timescale. Ground-state XeF isolated in cryogenic rare gas matrices shows a broad absorption feature near 320 nm. This feature is the B ← X transition, which lies approximately 4000 cm−1 lower in energy than in the gas phase.16,28 Photolysis of XeF2 at 266 nm in CD3CN produces XeF with a sharp feature near 345 nm.2 XeF has a 25 μs lifetime in acetonitrile, and our transient absorption spectrum at Δt = 500 ps matches the previously measured spectrum at Δt = 100 ns.2 Because the spectrum is largely unchanged between our longest time delay and ns to μs intervals, we take our long-time spectrum to represent the equilibrated XeF photoproduct.
It is necessary to verify that one-photon absorption dominates our condensed-phase study. Photolysis with ns laser pulses at 266 nm and intensities greater than 9 × 1012 W cm−2 saturates the transition to the Σu state and ensuing absorption of a second photon produces electronically excited XeF(B) and an F atom.29 Our photolysis intensity of about 1 × 1012 W cm−2 is ten times smaller than the onset of two-photon absorption.29 The integrated area of the transient absorption feature at Δt = 500 ps, which measures XeF from photolysis of XeF2, as a function of photolysis pulse energy,2 produces a linear dependence within the range 0.25–2.5 × 1012 W cm−2, spanning more than an order of magnitude of attenuation below the onset of two-photon absorption.29 We therefore conclude that transient signals measured and shown here result from one-photon absorption leading to the XeF(X) + F photodissociation pathway.
Decomposition to Xe + F2 is a lower energy channel than XeF + F production and so might compete following UV excitation of XeF2. We discount contributions to the transient spectra from this pathway because F2 absorbs only weakly in the observable, 320–700 nm region30 and the strong Xe − F2 → Xe+ − F2− charge transfer band lies to much shorter wavelength. In liquid xenon, for example, the band is observed below 185 nm.31,32
(1) |
Fig. 3 Stabilisation energy of the XeF B state as a function of the solvent dielectric constant, εr. The filled circles are the energy difference from the gas phase as determined by the maximum of the XeF(B–X) absorption at Δt = 500 ps. The solid line shows that the model of eqn (1) reproduces the trend in the data. |
One factor perturbing the XeF spectrum is the electronegative F-atom photoproduct of XeF2. Because some photolysis events lead to incomplete separation of the XeF and F fragments, we observe the spectral signatures of the metastable FXe–F complex. This complex is stable in rare-gas matrices at 20 K and indeed influences the B state.18 Apkarian and co-workers first described how the F atom decreases the energy gap between the XeF(X) and XeF(B) states and labelled this altered state XeF(B*),18 shown by the red line in Fig. 1.
Fig. 4 shows the results of our calculations of the isolated FXe–F complex, as detailed in the Experimental Approach, and they provide an extension of the original description of the electronic states in the FXe–F complex.18 We label the states of the complex as they correlate to the X, A, and B states of XeF at large FXe–F separation. The figure displays the results of calculations with linear FA–Xe–FB geometries and a fixed FA–Xe bond length at its calculated minimum of 2.3 Å, which is in agreement with previous work.14,15 These calculations show no barrier to association back to bound XeF2, but the FXe–F complex is known to be stable in Ar and Ne matrices.18 Stabilisation of the B-state by approach of an F atom collinear to the diatomic axis, and a reduction in the gap between the B and X states are consistent with the red shifted B*–X absorption band.
Fig. 4 Calculated electronic states of the FXe–F complex in linear F–Xe–F geometries. The labels for the states of the complex match the states of isolated XeF to which they correlate. The B state in the complex is the same as the XeF(B*) of Zerza et al.18 Association to bound XeF2 is barrierless at this level of theory. Bent F–Xe–F configurations lead to repulsive states. |
Avoided crossings with repulsive states derived from the XeF(X) and XeF(A) potentials are observed in the A and B states of the complex. Calculations with an FXe–F bond angle of 90° produce purely repulsive potentials over all states. These repulsive states show that the complex has a preferred linear geometry and that the stabilization is likely to occur through the overlap of the XeF 2Σ+ state with the F pz orbital.
The spectrum of the feature at 425 nm and its evolution into the 345 nm XeF band are consistent with prompt formation and decay of the FXe–F complex, but not with alternative assignments such as to Xe–F2 for reasons outlined earlier. We make further arguments for the assignment here and comment on its implications in condensed-phase reactions below. The 5500 cm−1 shift to lower energy of XeF(B*–X) relative to the XeF(B–X) transition is larger than the 2740 cm−1 shift reported for FXe–F in Ne and 4245 cm−1 in Ar.18 The difference between these two matrices indicates that the more polarisable medium of Ar more effectively stabilises the B* state, lowering the energy of the transition. It is therefore reasonable to propose that the greater polarisability of acetonitrile decreases the XeF(B*–X) energy gap further.
The shape of the prompt feature, which extends from the visible to the ultraviolet, is consistent with electronic states of FXe–F shown in Fig. 4. At an FXe–F distance of about 2.9 Å, the calculated B*–X energy gap goes through a minimum that corresponds to a 361 nm long-wavelength absorption cutoff. Longer and shorter distances have larger energy gaps, which could account for the broad extension of the feature to shorter wavelength. The Franck–Condon-active mode of the absorption is likely to be stretching between the FAXe pair, which is orthogonal to the plotted coordinate. Thus, the intensity of the transition could remain relatively constant over all Xe–FB distances, reflected by the flat-topped prompt absorption. The overestimate of the energy gap, however, is attributed to the gas-phase nature of the calculation. Exploring the effects of other FXe–F degrees of freedom or the solvent is beyond the scope of the current work, but we expect the general trends presented in these isolated molecule calculations to apply to the current condensed phase study.
Absorption to the A state of XeF may be discounted as a source of the longer wavelength absorption, because it has a gas-phase band origin near 1400 nm.16 Other candidates for the prompt absorption that we rule out include vibrationally excited XeF, charge-transfer transitions between XeF or F and the solvent or between Xe atoms and F2 (see earlier), and excited-state absorptions of XeF or XeF2. A spectral simulation of the XeF X and B states using previously derived spectroscopic constants10,18 shows that populating up to the highest bound vibrational level of XeF shifts the gas phase spectrum by only 14 nm, which is incongruous with our observation. Both XeF-to-CD3CN and F-to-CD3CN charge-transfer transitions have calculated high intensities outside our probe range as discussed in the ESI.†
The B state of XeF has a 14 ns gas phase lifetime,35 raising the possibility of an excited state absorption. As Fig. 2 describes, the absorption of 1 photon at 266 nm does not allow for population of an electronically excited state of XeF, which discounts the absorption out of this excited state. Equally excited state absorptions of XeF2 may be removed as a possible source of the long wavelength absorption, as no energetically accessible bound states of XeF2 are known to exist.27
Global fitting of the transient absorption spectra in Fig. 5 through deconvolution of the prompt XeF(B*–X) and long-lived XeF(B–X) spectral signatures provides ready analysis of the kinetics of the system. At all times after photolysis t, we fit the spectrum, S(λ,t) to a sum of two spectral features,
S(λ,t) = A(t)P(λ) + [B(t)erf(λ) + S0(t)]. | (2) |
Because the spectrum at Δt ≥ 500 ps results from only XeF(B–X) absorption, and our spectral simulation described in the ESI† indicates that the XeF(B–X) spectrum changes inappreciably with excess internal energy, we treat the long-time spectrum as the basis for fitting the time-invariant XeF(B–X) spectrum, P(λ). The function P is a Voigt profile fit to the average 1000 ps ≤ Δt ≤ 1500 ps spectra. We use an error function to model the evolution of the prompt spectrum not to describe a particular model, but because it requires few floating parameters and captures the general appearance of this feature that extends beyond our probe range. The amplitude of the long-time absorption, A(t), directly reports on the free XeF population, and an integral of the error function with combined offset and amplitude scaling factor B(t) provides an estimate of the FXe–F complex population.
The deconvolved time evolution of the XeF and FXe–F features, shown in Fig. 6 as A(t) in black and in red, faithfully represents the kinetics of the raw spectra of Fig. 5. The rapid rise of the XeF(B*–X) feature shows that the FXe–F complex forms within our time resolution, which is about 1 ps. It is also possible for the liberated F atom to escape the vicinity of the XeF co-fragment directly, similar to the facile migration reported for F atoms in rare gas matrices.13 The kinetic scheme used to model the data is described below, with fits to the model shown as solid lines in Fig. 6.
Fig. 6 Time-evolution of FXe–F complex (red) and XeF (black) formed from the photolysis of XeF2 at 266 nm in CD3CN and CD2Cl2 solvents. Intensities are extracted by a global fit to eqn (2) and subsequently normalised. Solid lines represent the fit to the kinetic model described within the text. The complex rises promptly after photolysis of XeF2 and decays during the first 5 ps. Bare XeF grows with a similar time constant. |
The kinetic model described allows for initial photodissociation of XeF2 to produce either the FXe–F complex (with rate coefficient k1) or XeF(X) and an escaped F atom (k2). The complex can subsequently dissociate to form XeF(X) + F or react with the solvent (k3), and we observe both a rise in XeF(B–X) absorption and build-up of DF (the latter by transient IR absorption spectroscopy36). The model is fitted to the time-dependent relative yields of XeF(X) and FXe–F by numerical integration.
The above scheme neglects some plausible competing loss processes for the FXe–F complex. For example, work in Ar and Ne liquids observed XeF2 growth from FXe–F decay, although this pathway was cut off upon matrix isolation. This recombination of the FXe–F complex is not included explicitly in our analysis because we have no spectral signature for XeF2 recovery. Similarly, our experiments are insensitive to reactive decomposition of FXe–F to Xe + F2 because the charge-transfer band is outside our spectral window, while the near-UV band of F2 is weak and we see no evidence for build-up of this stable molecule in our spectra. However, we recognise that these unobserved pathways could contribute to the magnitude of k3 as additional loss pathways for FXe–F. Fits of the experimental absorption band intensities to the kinetic model produce the rate coefficients reported in Table 1.
Solvent | k 1/ps−1 | k 2/ps−1 | k 3/ps−1 |
---|---|---|---|
CD3CN | 4.2 ± 0.3 | 0 | 0.25 ± 0.01 |
CD2Cl2 | 0.73 ± 0.66 | 0.63 ± 0.33 | 0.22 ± 0.04 |
First, if very few F atoms escape the solvent cage ballistically, and nearly all remain trapped within the cage, most of the prompt photoproduct will appear as the FXe–F complex. Small atoms and fragments such as H and CN are known to find their way directly through the solvent cage,7,37,38 and F atoms show high mobility in inert matrices,13 so we expect some fraction of direct F-atom separation. However, different solvents have a significant influence upon the quantum yield of in-cage recombination. Highly polar solvents such as H2O show large in-cage recombination quantum yields of ∼70% for ICN photolysis, in comparison to ∼30% in the less polar CHCl3.39,40 It is thus reasonable to expect a large proportion of the XeF and F fragments to become trapped within the CD3CN solvent cage immediately after dissociation.
Second, the impulse of photolysis may impart so much rotational and vibrational energy in the XeF fragment that broadening to the XeF(B–X) absorption, estimated to be as large as 40 nm FWHM, may occur (ESI†). We are unable to separate any such broadened XeF signal from the FXe–F complex feature and therefore cannot make a direct assessment of the importance of this process.
Loss of FXe–F complexes may follow a few competing pathways, of which one is geminate recombination to XeF2. In the previous work that identified the FXe–F complex,18 the rigid Ar or Ne matrix at 20 K prevented relaxation to linear XeF2, suggesting a small barrier to recombination. Our time-resolved measurements detect FXe–F that is prevented from reforming XeF2, but cannot discriminate between dynamical or energetic causes. However, the commensurate decay of the FXe–F band and growth of the XeF feature, and the near-isosbestic point evident in Fig. 5 hint that a significant proportion of the FXe–F survives recombination and instead decays by other routes.
The delayed cage escape and reactive pathways would be expected to have individual rates that depend upon the complex stability and any barrier to reaction. Time-resolved infra-red absorption spectra, reported elsewhere,36 show build-up of DF product on a timescale that is consistent with the decay of the FXe–F band and rise of the XeF(X) band intensities. We conclude that the reaction of the F atom with the solvent is a substantial loss pathway for the complex, and the rate coefficient k3 represents the sum of this reactive loss, the dissociation of the complex and its recombination to XeF2.
We fit the transient spectra in CD2Cl2, but not the other two solvents, to eqn (2) because there is definite separation between the two transient features. The integrated intensities are then simultaneously fitted to the kinetic model described above, as shown in Fig. 6. Unlike in acetonitrile, the XeF(X) feature near 350 nm appears promptly after photolysis in these solvents and the FXe–F complex is not as evident in CDCl3 or CCl4. Less clearcut signatures of the complex in these two solvents may be explained by shifts in the FXe–F band to shorter wavelength, a decrease in the branching ratio of complex formation, or an increase in the rate of the loss of the complex.
The rate coefficients extracted from the fit of the integrated band intensities following photolysis of XeF2 in CD2Cl2 are displayed in Table 1. The direct production of XeF(X) shows a significant rate coefficient of 0.63 ± 0.33 ps−1, while the complex decay shows a time constant similar to that for XeF2 photolysis in acetonitrile. The near identical timescales for complex loss suggest that the stability of the complex to decay (whether by reaction or other pathways discussed earlier) is unchanged upon changing solvent. Reaction with the CD2Cl2 is most likely to produce DF because we see no evidence for the B–X absorption band of ClF near 480 nm43 in the transient spectra. The known selectivity of gas-phase reactions of F atoms with chloroform and dichloromethane supports this deduction.44
Any difference in the observed FXe–F complex yields between the chlorinated solvents and acetonitrile may result from two distinct mechanisms:
(1) The photolytically produced F atom immediately escapes from the solvent cage. Ballistic escape can be affected by the ability of the solvent to quench the excess energy from the system, the mobility of F atoms in the solvent, and the strength of the solvent cage.
(2) The probability of prompt recombination to XeF2 increases within the chlorinated solvents. However, the origins of any barrier to recombination are uncertain, so we draw no conclusion about this trend.
The XeF molecule is observed to have a nanosecond or longer stability in CD3CN and CCl4 solvents but there is a substantial loss channel in CDCl3 and CD2Cl2. We suggest that the XeF reacts with the solvent to produce DF, and our transient IR absorption studies support this interpretation.36 Time constants for the loss of the XeF molecule are found to be 800 ± 200 ps in CD2Cl2 and 350 ± 50 ps within CDCl3.
The XeF B–X absorption in CCl4 solvent is observed to shift to shorter wavelength with a 4.1 ± 0.3 ps time constant, which may be indicative of vibrational cooling of the XeF. The XeF(X) ground state potential supports ∼8 vibrational levels. However, in CD3CN and CD2Cl2 solvents the FXe–F complex feature obscures any spectral signatures of vibrational relaxation dynamics.
The XeF molecule is stable in CD3CN and CCl4 solvents for nano- to microseconds. Substantial loss channels of XeF in CDCl3 and CD2Cl2 solvents are suggestive of reaction with the solvent.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4cp01854k |
This journal is © the Owner Societies 2014 |