Ultrafast Excited-State Dynamics of 2,4-Dimethylpyrrole† Michael Staniforth,‡,∥ Jamie D. Young,‡,∥ Daniel R. Cole,‡ Tolga N. V. Karsili,§ Michael N. R. Ashfold,*,§ and Vasilios G. Stavros*,‡ ‡

Department of Chemistry, University of Warwick, Library Road, Coventry, CV4 7AL, U.K. School of Chemistry, University of Bristol, Cantock’s Close, Bristol, BS8 1TS, U.K.


S Supporting Information *

ABSTRACT: The dynamics of photoexcited 2,4-dimethylpyrrole (DMP) were studied using time-resolved velocity map imaging spectroscopy over a range of photoexcitation wavelengths (276−238 nm). Two dominant H atom elimination channels were inferred from the time-resolved total kinetic energy release spectra, one which occurs with a time constant of ∼120 fs producing H atoms with high kinetic energies centered around 5000− 7000 cm−1 and a second channel with a time constant of ∼3.5 ps producing H atoms with low kinetic energies centered around 2500−3000 cm−1. The first of these channels is attributed to direct excitation from the ground electronic state (S0) to the dissociative 11πσ* state (S1) and subsequent N−H bond fission, moderated by a reaction barrier in the N−H stretch coordinate. In contrast to analogous measurements in pyrrole (Roberts et al. Faraday Discuss. 2013, 163, 95−116), the N−H dissociation times are invariant with photoexcitation wavelength, implying a relatively flatter potential in the vertical Franck− Condon region of the 11πσ* state of DMP. The origins of the second channel are less clear-cut, but given the picosecond time constant for this process, we posit that this channel is indirect and is likely a consequence of populating higher-lying electronic states [e.g., 21πσ* (S2)] which, following vibronic coupling into lower-lying intermediary states (namely, S1 or S0), leads to prompt N−H bond fission.

I. INTRODUCTION In order to gain a better understanding of the photochemistry and photophysics of complex biochemical systems, spectroscopists have, for decades, utilized bottom-up methodologies, wherein the spectroscopy of isolated subunits of larger biomolecules are studied.1,2 This allows for a more intimate understanding of the intrinsic properties of these subunits and, thus, a better foundation on which to develop a wider understanding of biochemical processes as we build in molecular complexity.3−5 Exemplar works using such methodologies, of particular interest to the present study, involve the molecule pyrrole, 6−17 a subunit found in heme and tryptophan.18,19 Such studies of the ultraviolet (UV) photochemistry and photophysics of pyrrole have gone some way to providing benchmark spectroscopic data to be exploited for more complex pyrrole-like systems. In this work, we utilize this benchmark data to continue the bottom-up theme and study a derivative of pyrrole, 2,4-dimethylpyrrole (DMP). In particular, we explore the role of N−H bond fission mediated by dissociative 1πσ* states of DMP and compare this, where appropriate, to its parent molecule, pyrrole.16 The role of dissociative 1πσ* states in the relaxation dynamics of photoexcited molecules was highlighted by Sobolewski et al.20 and since then has received growing attention both experimentally and theoretically.1,11,16,21−26 The authors proposed that 1πσ* states localized along heteroatom hydride (X−H, where X is commonly O or N) bond coordinates within a subunit may be accessed through © 2014 American Chemical Society

photoexcitation with UV radiation from the ground electronic state (S0) of a molecule. Traditionally “optically dark”, these states can be populated either (i) directly8,23,27−29 or (ii) indirectly through “optically bright”, bound 1ππ* states;1,20,21 after excitation to 1ππ* states, population may flow into the 1 πσ* state through internal conversion (IC) mediated by a 1 ππ*/1πσ* conical intersection (CI). Once populated, 1πσ* states may facilitate nonradiative relaxation back to S0 via a 1 πσ*/S0 CI at extended X−H bond distances or undergo X−H bond fission.11,20,21 While pyrrole in S0 has comparatively high (C2v) symmetry, substituting the hydrogen atoms at ring-positions 2 and 4 by methyl groups results in a reduced Cs symmetry for DMP (parts a and b of Figure 1, respectively). This has direct consequences for the UV absorption spectrum, shown in Figure 1c, with the highlighted region indicating the photoexcitation window of the present measurements. Relative to pyrrole, photoexcitation to the first two 1πσ* states of DMP [i.e., the 11πσ*(1A″, S1) ← 1ππ(1A′, S0) and 21πσ*(1A″, S2) ← 1ππ (1A′, S0) transitions] becomes electric dipole allowed (given the selection rule A″ ⊗ Γμ ⊗ A′ ⊇ A′, where Γμ is the symmetry of the transition dipole moment operator, which transforms as either a′ or a″ in Cs symmetry). Thus, the UV spectrum of DMP shows a discernible absorption feature spanning the range Received: September 3, 2014 Revised: October 21, 2014 Published: October 24, 2014 10909 | J. Phys. Chem. A 2014, 118, 10909−10918

The Journal of Physical Chemistry A


that take place within DMP following excitation between 276 and 238 nm using a combination of time-resolved velocity map imaging (TR-VMI) and detailed theoretical calculations using the complete active space self-consistent field (CASSCF) theoretical method, together with its second-order perturbation theory extension (CASPT2).

II. METHODS a. Experimental Methods. The experimental setup of the TR-VMI spectrometer has been described in detail elsewhere,32,33 and as such, only a brief overview is given here. A commercially available Ti:sapphire oscillator and regenerative amplifier system (Spectra-Physics Tsunami and Spitfire XP, respectively) produce 3 mJ laser pulses of ∼40 fs duration centered around 800 nm at a repetition rate of 1 kHz. Two optical parametric amplifiers (Light Conversion, TOPAS-C), each pumped with a fraction of the 800 nm fundamental (1 mJ/ pulse), produce tunable UV pump and probe pulses (hνpu and hνpr respectively). hνpu is tuned between 276 and 238 nm (5−7 μJ/pulse) and is temporally varied with respect to the probe pulse by means of a hollow gold retroreflector mounted on a motorized stage, allowing for a maximum temporal delay of Δt = 1.2 ns. hνpr is set to 243 nm (∼7 μJ/pulse) and is resonant with the two-photon allowed 2s ← 1s transition in hydrogen. Both beams are then focused near collinearly into the VMI spectrometer, where they intersect perpendicularly a molecular beam seeded with the sample molecule. The molecular beam is produced by passing 2 bar of He gas over the sample molecule (DMP, Sigma-Aldrich, 99%) heated to 50−55 °C and then expanding the seeded He into vacuum (∼10−7 mbar) via an Even Lavie pulsed solenoid valve,34 operating at 125 Hz. Typical opening times for the valve were 12−14 μs. The subsequent supersonic jet expansion passes through a 2 mm conical skimmer, into the VMI spectrometer. Temporal overlap of the pump and probe pulses (Δt = 0) and characterization of the instrument response function (IRF) are achieved via multiphoton ionization of methanol, yielding an IRF of ∼120 fs at full width at half-maximum (FWHM), which provides a minimum temporal resolution of ∼30 fs (∼25% of IRF). The spectrometer itself, which is a modified version of our previous setup,32,33 is now in line with the molecular beam and follows the standard Eppink and Parker design.35 After photolysis of the parent molecule with hνpu, any resulting H atoms are ionized with hνpr using 2 + 1 resonance-enhanced multiphoton ionization (REMPI). The VMI ion-optics project the 3-D velocity distribution of H+ ions toward a position sensitive detector consisting of a pair of microchannel plates and a P-43 phosphor screen (Photek, VID-240). The emitted light is captured using a CCD camera (Basler, A-312f). The initial 3-D velocity distribution is reconstructed from the resulting 2-D image using a polar onion-peeling (POP) algorithm,36 resulting in both energy and angular information. Radial pixels on the image can be converted into total kinetic energy release (TKER) using the appropriate Jacobian, calibration factor, and cofragment mass, enabling us to obtain the desired 1-D TKER spectra. The calibration factor is obtained through photolysis of HBr at 200 nm, which produces H atoms with well-characterized kinetic energies (KEs).37 Comparative TR-VMI studies were carried out on monodeuterated DMP [2,4-dimethylpyrrole-d1 (DMP-d1)]. DMP-d1 was synthesized according to ref 38 by repeated stirring of undeuterated DMP in excess D2O in a light-sealed vessel for

Figure 1. Molecular structures of (a) pyrrole and (b) DMP. (c) Vapor-phase UV absorption spectrum of DMP between 200 and 290 nm. The blue arrow indicates the excitation wavelength range investigated in this study.

285−245 nm that has been attributed to photoexcitation to the S1 state.14 The corresponding 1πσ*-state absorption in pyrrole is much weaker16 and both theory30 and photofragment translational spectroscopy studies8 suggest that this (weak) transition strength derives from vibronic interaction with higher-lying 1ππ* states. The steep rise in the DMP absorption spectrum below 245 nm may be attributable either to the 21πσ*(1A″, S2) ← 1ππ (1A′, S0) transition intensity-borrowing from the higher-lying electronic 11ππ*(1A′, S3) and 21ππ*(1A′, S4) states or to direct population of these latter states.14 Our previous studies on pyrrole have demonstrated the role of tunneling out of the quasibound 3s Rydberg well of the 11πσ* state [in the vertical Franck−Condon (vFC) region, the 11πσ* state has notable 3s Rydberg character associated with the N atom].16 Measurements on pyrrole and pyrrole-d1 (selective deuteration of the N−H bond) returned a kinetic isotope effect (KIE, kH/kD) of ∼11. This observation, coupled to the significant reduction in the appearance time constant for H atom elimination at excitation energies above the threshold required to surmount the barrier associated with the Rydberg well, led us to conclude that tunneling was very likely operative. Electronic structure calculations for DMP return a more pronounced barrier associated with 3s(Rydberg)/σ*(valence) mixing in the 11πσ* state of DMP (∼0.35 eV, defined relative to the local S1 minimum at an equilibrium N−H bond length; cf. ∼0.18 eV in pyrrole).31 Relative to pyrrole, therefore, we might anticipate a longer appearance time constant for H atom elimination following photoexcitation of DMP in this quasibound region. In the present study, we test this hypothesis by building on our previous work on pyrrole.16 Specifically, we investigate the role methyl substitution has on the relaxation dynamics in DMP and compare the dynamics to that of pyrrole. To this end, we present the first comprehensive study of the dynamics 10910 | J. Phys. Chem. A 2014, 118, 10909−10918

The Journal of Physical Chemistry A


∼24−48 h. The deuterated product was then separated from the D2O and dried over Na2CO3. Time of flight mass spectroscopy was used to check for deuteration and returned a 3:1 ratio of deuterated to nondeuterated photoproduct. In addition to the above, a vapor-phase UV absorption spectrum of DMP was measured by placing a drop of DMP in a fused silica sample cell and then recording the spectrum using a commercially available UV−visible absorption spectrometer (PerkinElmer, Lambda 25, 0.2 nm resolution). b. Theoretical Methods. Using the Gaussian 09 computational package,39 the minimum energy geometry of groundstate DMP was optimized using Møller−Plesset second-order perturbation theory (MP2)40 coupled with the aug-cc-pVTZ (AVTZ) basis set. Given the optimized MP2 ground-state geometry, the Molpro 2010.1 computational package41 was used to calculate unrelaxed (rigid body) potential energy curves (PECs) along the N−H stretch coordinate (RN−H). These were calculated using CASPT2,42,43 based on a fully state-averaged CASSCF reference wave function (SA5-CASSCF) comprising the lowest five equally weighted singlet states (three A′ and two A″ states) and coupled with the AVTZ basis set with additional even-tempered sets of s and p diffuse functions on the N atom (ratio = 2). The (10,10) active space comprised 10 electrons arranged in the following 10 orbitals: three occupied π and two virtual π* molecular orbitals, two σ and two σ* orbitals (one of each of which is centered around the N−H bond), and the 3s Rydberg orbital centered on the N atom. Using this same active space and basis set, the geometries of the ground state and first four excited electronic states of the parent molecule were then reoptimized using CASSCF and corrected with CASPT2 in order to determine the vertical and adiabatic excitation energies. CASSCF/CASPT2 calculations using the AVDZ basis set were also undertaken on the 2,4-dimethylpyrrolyl radical (C6H8N in Figure 2a, labeled DMPr henceforth), in

performed within Molpro, at the MP2-optimized ground state geometry with the AVTZ basis set, to calculate the transition dipole moment (TDM) vectors and oscillator strengths (f) for transitions from S0 to the first four singlet excited states. All calculations were nonsymmetry-adapted and assumed Cs symmetry.

III. RESULTS a. Ab Initio Calculations. 1. Vertical and Adiabatic Excitation Energies. Table 1 lists the calculated vertical and Table 1. Vertical and Adiabatic (relaxed) Excitation Energies from S0 to the S1, S2, S3 and S4 States of DMP Calculated at the CASPT2(10,10)/aug-cc-pVTZ (AVTZ) Levela transition S1−S0 S2−S0 S3−S0 S4−S0

(σ*←π) (σ*←π) (π*←π) (π*←π)

vertical energy/eV

adiabatic energy/eV

4.91 5.66 5.87 6.04

4.58 5.45 5.49 5.51

(253) (219) (211) (205)

(271) (228) (226) (225)

oscillator strength f 0.0019 0.0012 0.0218 0.1447


The values in parentheses are wavelength (nm) to aid direct comparison with the UV absorption spectrum of DMP (Figure 1c). Associated oscillator strengths were calculated using EOM-CCSD.

adiabatic excitation energies and oscillator strengths for transitions from S 0 to the S 1 (1 1 πσ*), S 2 (2 1 πσ*), S3 (11ππ*), and S4 (21ππ*) electronic states of DMP. The vertical excitation energies are in good accord with previous calculations (at the EOM-CCSD level).14 Reference to Figure 1 confirms that the calculated S1−S0 adiabatic energy matches well with the experimentally observed origin [∼279 nm (4.44 eV)]. The S1−S0 vertical excitation energy returned by the CASPT2 calculations exceeds the experimental S1−S0 energy gap, by ∼0.47 eV. This is an inevitable consequence of “freezing” all bar the N−H stretch coordinate at the optimized ground-state geometry and accords with (i) the expectation that promoting a π-bonding electron should result in some expansion of the ring and (ii) the observed breadth of the S1← S0 (σ*←π) absorption. 2. Potential Energy Cuts (PECs). Figure 2a shows the calculated 1-D “unrelaxed” PECs along RN−H (i.e., PECs calculated with all other coordinates fixed at the corresponding ground-state value), demonstrating the dissociative nature of the S1 and S2 states. Both PECs show a small barrier to N−H bond dissociation, formed through an avoided crossing between the diabatic 1π3s Rydberg and 1πσ* valence states. The calculated height of this barrier in the 11πσ* PEC is ∼0.35 eV (not zero-point-corrected, and defined relative to the S1 minimum in the vFC region). Upon increasing RN−H, both the 11πσ* and 21πσ* PECs show a CI with the S0 PEC that will control the branching between ground (X) or excited state DMPr products (plus an H atom). We note, as is often the case, that the present “unrelaxed” CASPT2 calculations significantly overestimate the lowest dissociation energy [De(N−H) ∼ 36 000 cm−1 (cf., the experimentally determined value of 31200 ± 50 cm−1 in ref 14)]. Indeed, the nature of these rigid body scans also leads to the apparent radical splitting of ∼0.2 eV (Figure 2), which is significantly smaller than that predicted by relaxed CASPT2/aug-cc-pVDZ calculations. These latter calculations return an adiabatic excitation energy of ∼0.85 eV between the ground and first excited electronic states of DMPr, a result to which we return in section IV.

Figure 2. (a) Calculated unrelaxed PECs for the S0, S1, S2, S3 and S4 states of DMP. The gray circle highlights the S1/S0 and S2/S0 CIs. (b) Calculated TDMs for transitions from S0 to the S1, S2, S3 and S4 excited states, the last two of which lie in the plane of the ring.

order to calculate the adiabatic (and vertical) excitation energies between the ground and first excited state of this species. The (10,8) active space in this case comprised the following eight orbitals: three occupied π and two virtual π* molecular orbitals, one ring-centered σ and one σ* orbital, and the N-centered pz orbital. Equation of motion coupled cluster single and double (EOM-CCSD) calculations were also 10911 | J. Phys. Chem. A 2014, 118, 10909−10918

The Journal of Physical Chemistry A


b. Total Kinetic Energy Release (TKER) Spectra. Figure 3a shows an example TKER spectrum with the image from

color pump−probe signal. Each spectrum possesses a common high-KE feature peaking around 5000−7000 cm−1, the intensity of which remains constant beyond Δt = 1 ps. A second, lower KE feature is present at pump wavelengths ≤254 nm and appears subsequent (in time) to the high-KE feature. The high-KE feature can be understood by considering the PEC of the 11πσ* state, which correlates nonadiabatically to H + DMPr(X) products (Figure 2). TKER spectra obtained at short time delays when exciting at wavelengths

Ultrafast excited-state dynamics of 2,4-dimethylpyrrole.

The dynamics of photoexcited 2,4-dimethylpyrrole (DMP) were studied using time-resolved velocity map imaging spectroscopy over a range of photoexcitat...
1MB Sizes 0 Downloads 6 Views