Probing the Biexponential Dynamics of Ring-Opening in 7

Aug 16, 2016 - James W. Snyder , B. Scott Fales , Edward G. Hohenstein ... Cecilia Cisneros , Travis Thompson , Noel Baluyot , Adam C. Smith , Enrico ...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/JPCA

Probing the Biexponential Dynamics of Ring-Opening in 7‑Dehydrocholesterol Broc D. Smith, Kenneth G. Spears, and Roseanne J. Sension* Department of Chemistry and Department of Physics, University of Michigan, 930 N. University Ave, Ann Arbor, Michigan 48109-1055, United States ABSTRACT: Our prior discovery of a novel biexponential photochemical ring-opening in 7-dehydrocholesterol (DHC) to previtamin D3 [Tang et al. J. Chem. Phys. 2011, 134, 104503] is further explored with ultrafast transient absorption spectroscopy, and the results are compared with recently reported high-level theoretical calculations. Three types of experiments are reported. First, variation of the excitation wavelength from 297 to 266 nm leaves the excited state dynamics unaffected. The biexponential decay of the excited state absorption is independent of excitation wavelength with time constants of 0.57 ± 0.06 and 1.88 ± 0.09 ps, in excellent agreement with the results reported earlier (0.56 ± 0.06 and 1.81 ± 0.15 ps) following excitation at 266 nm. Second, variation of the chirp of the excitation pulse influences the relative amplitude of the fast and slow decay components but has no influence on the photoproduct yield. Third, a 545 nm pulse delayed by 0.64 ps with respect to the initial 266 nm pulse was used to perturb the “slow” population and probe the influence of additional electronic or vibrational energy on the reaction process. The results show ultrafast internal conversion Sn → S1 on a ca. 150 fs time scale but no subsequent effect on the reaction dynamics. The experiments reported here are consistent with the recent state averaged complete active space self-consistent field ab initio multiple spawning (SA-CASSCF-AIMS) calculations of Snyder et al. [J. Phys. Chem. Lett. 2016, 7, 2444] that assign the biexponential decay to nonequilibrium dynamics related to the opening and closing motion of the cyclohexadiene ring moiety on the excited state surface. These new experiments support the model prediction that the biexponential dynamics does not involve multiple minima and demonstrate the direction for new experimental designs to manipulate the product yields and pathways.

■

INTRODUCTION The isomerization reactions of polyenes play a central role in a wide range of natural and synthetic photoactivated systems or devices. The function of these systems is controlled by photochemistryphoton energy produces action via a change in shape or cleavage of a bond. One such functional polyene is the 1,3-cyclohexadiene chromophore (CHD), a conjugated diene that undergoes a single photon electrocyclic ring-opening reaction to form a conjugated triene. Because of the intrinsic reversibility of this photochemical reaction, molecules containing the CHD chromophore are often considered as potential candidates for optically controlled switches.1−5 The actual and potential uses for this prototypical photochemical reaction have led to an extensive array of spectroscopic and theoretical investigations of the ring-opening reaction.5−7 These studies have ranged from fundamental studies of excited state dynamics to optical control of the photoproduct yields.8−13 Optical control mechanisms for electrocyclic ring-opening reactions have the potential to enhance the utility of this reaction in a variety of applications.14,15 The CHD chromophore is also the photoactive component of the precursor to vitamin D 3 , 7-dehydrocholesterol (Provitamin D3, DHC), the so-called “sunshine vitamin”. Following UV excitation (S1 ← S0), DHC undergoes a ring© 2016 American Chemical Society

opening reaction to form previtamin D3 (Figure 1). The photochemical ring-opening reaction occurs in a conrotatory

Figure 1. Photochemical ring-opening of 7-dehydrocholesterol to previtamin D3.

fashion according to the Woodward−Hoffmann rules.16 This previtamin D3 triene structure then undergoes a [1,7] sigmatropic hydride shift to form cholecalciferol (vitamin D3).17 In contrast to the ring-opening photoreaction, which occurs on a time scale of picoseconds,18−24 the thermal hydride shift occurs on the order of hours to days.25−27 The Received: July 12, 2016 Revised: August 4, 2016 Published: August 16, 2016 6575

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A

and co-workers. 36 The experiments reported here are consistent with state averaged complete active space selfconsistent field ab initio multiple spawning (SA-CASSCFAIMS) calculations of Snyder et al.36 The experiments support the model prediction that the biexponential decay does not involve multiple pathways with distinct quantum yields for ringopening. Rather, the biexponential decay is dominated by nonequilibrium dynamics along the reaction coordinate.

photochemistry of vitamin D has received much experimental20,22−24,28−33 and theoretical7,34−36 attention due to the fundamental biological importance of the reaction and even its potential use in photoactivated devices.4 Early ultrafast transient absorption studies on DHC by Anderson et al. identified a strong excited state absorption (ESA, Sn ← S1) in the visible region of the spectrum.21,22 More extensive studies by Tang et al. showed that the ESA is characterized by a strong peak at about 485 nm as well as a long tail extending to 770 nm.19,20 The decay of the ESA is clearly not single exponential but can be modeled using a biexponential decay with time constants dependent on temperature and solvent. At room temperature the time constants are ∼0.4−0.65 and ∼1.0−1.8 ps for alkane and alcohol solvents. The temperature dependence of the decay constants implies a small intrinsic barrier of ∼2 kJ/mol (167 cm−1) as well as a larger extrinsic, solvent dependent barrier in the excited state along the reaction coordinate.19 The overall decay of the excited state is also dependent on the linear chirp of the excitation pulse with positive chirp resulting in a somewhat faster decay than negative chirp.20 The biexponential decay of the excited state was assigned to parallel reaction pathways from the Franck−Condon region to two nonsymmetry equivalent intermediates as the ring begins to distort.19 In this model the dominant ESA observed in the transient absorption experiment comes from these distinct intermediates rather than from the initially excited Franck− Condon region. This model is consistent with theoretical simulations by Tapavicza and co-workers7,34 that coupled excited state dynamics to product formation. In contrast, recent theoretical simulations by Snyder et al. propose a competing mechanism assigning the biexponential population decay to nonequilibrium excited state dynamics.36 In the work reported here three sets of experiments were performed to change the excited state dynamics of DHC by modifying the excited state population. (1) Transient absorption measurements were performed exciting DHC at a range of wavelengths between 297 and 266 nm indicated in Figure 2. (2) UV transient absorption measurements probed

■

EXPERIMENTAL METHODS Pulses for the tunable pump measurements were generated using a home-built, passively mode-locked Ti:sapphire oscillator centered at 800 nm. These pulses were then amplified using a pair of home-built amplifiers (regenerative amp followed by 3-pass amp) yielding pulses with energy of ∼750−800 μJ. Pulses were then split into pump and probe arms. A non-collinear optical parametric amplifier (NOPA) was tuned between ∼530 and 625 nm and then doubled using a Type I BBO crystal to generate the UV pump pulses ranging between ∼270 and 300 nm. The 266 nm pump pulses were generated by doubling the fundamental laser pulses and overlapping these with the residual fundamental pulses. The NOPA was also used to produce the second excitation pulse at 545 nm for pump−repump−probe measurements. For experiments as a function of linear chirp, the 266 nm pulse was passed through an acoustooptic programmable dispersive filter (AOPDF, Dazzler, Fastlite) to output a pulse with programmed linear chirp as measured at the sample position. For all experiments typical pulse energies at the sample were between ∼220 and 300 nJ. Focusing the fundamental pulses into a 5 mm CaF2 plate generated broadband white light probe pulses. The plate was continually kept in motion in order to prevent damage. The spectrum of these pulses spanned between about 340 and 650 nm. Residual 800 nm pulses were filtered out with either a BG23 or KG3 Schott glass filter depending on the experiment. The UV continuum spanning the range from 270 to 600 nm was generated by focusing the 400 nm second harmonic of the laser into the CaF2 plate. Pump pulses were switched on and off at the sample by an optical chopper (Terahertz Technologies Digirad C-980) at a frequency of 100 Hz (5 on, 5 off). The transient absorption difference spectra were calculated using the pump on and pump off spectra. The pump−probe delay was adjusted using a mechanical delay stage. The angle between pump and probe beams at the sample was ∼9°. Pump pulses were focused into the sample with a fused silica lens (FL = 175 mm), and probe pulses were focused into the sample with a UV-enhanced aluminum spherical mirror. After probing the sample, the white light pulses were focused into a 200 μm UV−vis optical fiber from Avantes and coupled into a single-channel spectrometer (Avantes Avaspec, 2048-USB2-UA). For all measurements, the pump and probe polarizations were kept at magic angle (54.7°) in order to minimize contributions to the signal due to rotational diffusion. After the completion of each data set, the chirp of the white light continuum was corrected by fitting the coherent spike at t = 0 ps with a third-order polynomial. 7-Dehydrocholesterol (98%) was acquired from SigmaAldrich and used without further purification. A gravity driven wire-guided flow setup based on a design described in the literature37 was used in order to reduce the amount of crossphase modulation between pump and probe. The thickness of the solvent sheet was adjusted by changing the flow rate to

Figure 2. Steady state absorption spectrum of 7-dehydrocholesterol in 2-butanol with arrows indicating the excitation wavelengths used in these experiments.

the photoproduct yield as a function of the chirp of a 266 nm excitation pulse. Previous measurements only probed the excited state dynamics.20 (3) Finally, a three-pulse experiment was performed, pumping the sample at 266 nm, repumping the excited state with a 545 nm pulse, and probing with a broadband visible continuum. The results of these experiments are discussed in the context of the recent theoretical simulations by Tapavicza and co-workers7,34 and by Snyder 6576

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A

decay constants. Like the previously reported results, all data collected here required at least two exponential decay components to accurately model the excited state decay. Some data sets also required a long time component to account for multiphoton excitation of the solvent by the UV pump pulse. The data and fits are plotted in Figure 4a,b. The time constants for decay of the excited state are summarized in Table 1. The values for these time constants are plotted in Figure 4c along with previously reported values for 266 excitation in 2butanol (0.56 ± 0.06 and 1.81 ± 0.15 ps).19 Within the error, the time constants for each excitation wavelength are in good agreement with the previously reported results. The relative amplitudes of the two components for each excitation wavelength are shown in Figure 4c as the fraction of the total amplitude belonging to the slow component. The amplitudes of the fast and slow decay components have previously been shown to depend on the chirp of the excitation pulse.20 Since chirp was not directly monitored in the current experiments (and almost certainly changed somewhat with excitation frequency due to changes in the tunable NOPA), it is unknown whether any differences in the fraction of the total amplitude of the slow component are a result of the chirp of the excitation pulse or the frequency of the pulse. However, these numbers show that in general the relative contributions of the fast and slow components are not significantly changed upon varying the excitation frequency. The wavelength independence of the biexponential excited state decay of DHC demonstrates that the excess vibrational energy introduced into the molecule is either dissipated rapidly among the bath of available vibrational states of DHC or sequestered in the Franck−Condon active modes decoupled from the reaction coordinate. The excitation wavelengths (266−297 nm) differ in energy by 3900 cm−1 or 46.9 kJ/mol. DHC (C27H44O) contains 72 atoms, resulting in a total of 210 (3N − 6) normal modes within the molecule. If there is rapid dissipation of the excess energy among all 210 vibrational modes of DHC, the internal vibrational temperature of the molecule will be raised by many tens of degrees. Assuming a bath of harmonic modes with frequencies similar to those calculated for the ground state DHC molecule, an additional 3900 cm−1 will raise the temperature about 88 K. This temperature increase would be sufficient to modify the lifetime of an excited state having a small barrier to internal conversion. The internal barrier in DHC is estimated to be ∼2 kJ/mol or 240 K,19 and an 88 K temperature rise would be anticipated to decrease the excited state lifetime by 10%−20%. Energy dissipated over a smaller subset of modes localized in the vicinity of the initially excited cyclohexadiene chromophore would be expected to raise the local temperature higher and have a larger effect. A 10% difference in the lifetime over the range from 297 to 266 nm is within the error of the data, but there is no evidence for a larger change in the excited state lifetime. If energy dissipation is slow and the excess energy remains sequestered in the Franck−Condon active modes of the molecule, the influence on excited state dynamics will depend on the nature of these Franck−Condon active modes. Energy placed directly into vibrational modes along the reaction coordinate may be expected to have a significant influence on the excited state dynamics, either in the lifetime of the excited state or in the branching between the two excited state pathways. The dominant vibrational progression in the DHC absorption spectrum is assigned to a 1350 cm−1 CC

achieve the highest amount of white light coupling into the fiber and spectrometer after passing through the sheet. All ultrafast measurements discussed here were performed in 2butanol. The concentration of the samples was ∼0.8 mM. The steady state absorption spectrum of DHC was collected in 2butanol. The data were analyzed using a simplex fitting algorithm developed in our laboratory, the freely available global analysis program Glotaran,38 or a more flexible program, VarPro,39 developed in our laboratory and capable of modeling pump− repump experiments.

■

RESULTS AND DISCUSSION Dependence on Excitation Wavelength. Transient absorption (TA) spectra of DHC in 2-butanol were collected using five different excitation wavelengths: 266, 272, 277, 284, and 297 nm (Figure 2) for probe wavelengths from 350 to 650 nm and time delays out to 30 ps. This range allows characterization of both the decay dynamics and the spectral structure of the excited state absorption. Data obtained following excitation at 284 nm are plotted in Figure 3a. The

Figure 3. (a) Chirp-corrected transient excited state absorption observed following excitation of DHC at 284 nm. (b) Characteristic excited state absorption spectra for a range of excitation wavelengths. Spectra were averaged between 1 and 2.5 ps.

excited state spectra for four probe wavelengths, integrated from 1 to 2.5 ps, are plotted in Figure 3b. Each spectrum in the lower panel is scaled to a maximum intensity of 1 for comparison. The peak intensity for each pump wavelength is 54 mOD (266 nm), 14 mOD (272 nm), 12 mOD (277 nm), 52 mOD (284 nm), and 19 mOD (297 nm) reflecting the variation in the intensity of the pump pulse. The kinetic analysis of each data set was done by integrating over the excited state absorption peak from 450 to 500 nm (with the exception of the 272 data, which was integrated from 480 to 500 nm). This noise reduction procedure was used by Tang et al. where it was shown to yield results identical to those obtained as a function of individual wavelengths.19,20 There is no evidence for probe wavelength dependent variation in the 6577

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A

Figure 4. (a) Kinetic traces and biexponential fits with residuals for four excitation wavelengths used in this study. (b) Kinetic trace and a comparison between the monoexponential and biexponential fit to the data at 297 nm. (c) Time constants for the biexponential fits to the integrated spectra plotted alongside the previously reported values for decay following 266 nm excitation and their error bars (solid and dashed horizontal lines, respectively).19 The fraction of the slow decay component relative to the total amplitude of the two fitting components is plotted along the right axis. The solid, horizontal line is the average value of the five data points.

then modify the product yield by changing the branching between these two pathways.20 The photoproduct yield following excitation of DHC at 266 nm was measured as a function of the linear chirp of the incident pulse spanning the range from −8000 to 10 000 fs2. Measurements of the excited state absorption over this range provided evidence for an influence of chirp on the decay of the excited state absorption.20 UV transient absorption measurements of product formation were performed on two separate occasions, and the chirp sequence was varied randomly to avoid an influence of systematic drift. The averaged results for positive and negative chirp are summarized in Figure 5. There is no evidence for any influence of the chirp on the product yield, although there remains evidence for a small influence of chirp on the ESA.41 These measurements challenge the hypothesis that the fast and slow decay dynamics represent parallel pathways on the excited state surface with different

Table 1. Parameters from Fits to the Decay of the Excited State Absorption of DHC in 2-Butanol λexcitation (nm) 266 272 277 284 297 weighted average

τ1 (ps) 0.54 0.55 0.60 0.64 0.56 0.57

± ± ± ± ± ±

0.12 0.14 0.13 0.16 0.09 0.06

τ2 (ps)

fraction slow

± ± ± ± ± ±

0.41 0.54 0.31 0.56 0.52

1.87 1.93 1.64 1.88 1.97 1.88

0.16 0.22 0.26 0.30 0.17 0.09

stretching mode. This mode is not directly involved in the ringopening reaction, and excess energy placed into this mode may be decoupled from the reaction coordinate. The lack of dependence on excitation wavelength is similar to the trend observed by Sajadi et al. following excitation of cis-stilbene at 283 and 267 nm.40 Energy is not easily funneled from the CC stretching modes into reaction coordinates. Influence of Linear Chirp on the Ring-Opening Reaction. Tapavicza et al. reported nonadiabatic molecular dynamics simulations of the DHC ring-opening reaction using a linear response time dependent density functional theory surface hopping method (LR-TDDFT-SH).7,34 In contrast to the expectation that curve crossing from the bright S1 state to a dark state occurs in the process of the excited state ringopening reaction, they find no evidence for the participation of more than one excited state. The nature of this state changes along the reaction pathway.7 Of most significance for the experimental measurements reported here, Tapavicza and coworkers find that the reactive trajectories on average decay to the ground state on a time scale faster than the unreactive trajectories by a factor of ∼1.8. This suggests the possibility that the biexponential decay of the excited state population, if it represents population of two separate excited state intermediates (i.e., parallel pathways), could be coupled to photoproduct yield. The chirp of the excitation pulse could

Figure 5. UV difference spectra measured following excitation of DHC in 2-butanol with positively chirped (1000, 4000, and 10 000 fs2) and negatively chirped (−1000, −4000, and −8000 fs2) excitation pulses at 266 nm. The product spectrum, >50 ps, represents 3 times as many individual measurements as the 1.2 and 2.0 ps difference spectra. There is no evidence for an impact of linear chirp on the product yield. 6578

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A

excited state capable of increasing the yield for the ring-opening reaction by as much as a factor of 3. The excited state lifetime of DHC is significantly shorter than diarylethene, and the intrinsic quantum yield for the ring-opening reaction is significantly larger (ca. 34% rather than 2%). We have found no evidence for a second excitation pulse in the visible modifying the reaction dynamics. The lack of any observed change in the rate of internal conversion to the ground state is also additional confirmation that excess vibrational energy placed into the molecule is largely decoupled from the excited state reaction coordinate. Excess energy placed into the molecule in the excitation process will then have no impact on the excited state decay. The secondary pump pulse at 545 nm places an additional 18 350 cm−1 of excess energy into the electronically excited molecule. Initially this places the molecule into a higher electronic excited state, but the S1 state is repopulated within a few hundred femtoseconds. In the limit of fast IVR, this excess energy will raise the vibrational temperature of the DHC molecule by several hundred degrees and should speed up a barrier crossing process substantially. Assuming a 2 kJ/mol barrier and an Arrhenius barrier crossing process, the additional energy would be expected to speed up the process by as much as 50%. No such change in the decay of the excited state is observed. The only lasting impact of the 545 nm excitation is a small blue-shift in the spectrum. Thus, this result may also indicate that the only significant barrier for internal conversion through the S1/ S0 conical intersection is the extrinsic solvent dependent barrier. An Alternative Model for the Biexponential Excited State Decay. More recent calculations by Snyder et al. used state averaged complete active space self-consistent field (SACASSCF) and ab initio multiple spawning (AIMS) to explore the ring-opening reaction of DHC.36 This method avoids many of the limitations of TD-DFT based methods. As in the simulations by Tapavicza et al.,7,34 and as predicted by the constant polarization anisotropy in transient absorption measurements,19 only one excited electronic state is involved. The SA-CASSCF-AIMS calculations reproduce the observed biexponential decay of the excited state population, but find no difference between the time scales for reactive and nonreactive trajectories. Rather than the parallel or sequential models proposed in our initial paper,19 these calculations suggest that the biexponential decay reflects nonequilibrium excited state dynamics. The bond breaks rapidly in the excited state, within 25 fs. The kinetic energy imparted in the excitation process causes the molecule to overshoot the S0/S1 conical intersection. As the molecule rebounds into the S0/S1 coupling region on a ca. 350 fs time scale there is substantial population transfer to the ground state. The remaining population transfer to the ground state occurs on a ca. 1 ps time scale from a substantially relaxed population. These calculations shed important new light on the experimental observations. Linear chirp can serve to focus or defocus an excited state vibrational wave packet depending on the details of the excited state surface.42 The bandwidth available in our experiments, ca. 3 nm around 266 nm, or 425 cm−1, is sufficient to prepare a wave packet in the low frequency modes in the excited state coupled to the ring-opening reaction. The modest dependence of the excited state dynamics on positive linear chirp20 may reflect wave packet focusing that enhances the rapid coupling through the S0/S1 conical intersection and decreases the relative contribution of later

product yields. The absence of an influence of chirp on product yield also rules out an intrapulse pump−dump mechanism, in agreement with our earlier analysis of the chirp dependence of the ESA.20 Manipulating the Excited State Population with a Second Visible Pump Pulse. A final probe of the excited state dynamics was performed using a second visible pulse (a “repump” pulse) to perturb the excited state population. A time delay >0.5 ps was chosen for the visible pulse on the hypothesis that the long-lived population corresponded to a less reactive intermediate. Thus, Sn ← S1 excitation of the less reactive intermediate might be expected to impact excited state dynamics and the product yield following internal conversion. For these measurements a visible pulse at 545 nm was timed to arrive 0.64 ps after the initial 266 nm excitation pulse to excite the molecule to a higher electronic state, Sn ← S1. The kinetic traces obtained after excitation at 266 nm with and without the second visible pump pulse averaged between 380 and 630 nm are displayed in Figure 6. Following the arrival of the 545 nm

Figure 6. Kinetic traces of both pump only and pump−repump scans averaged between 380 and 630 nm omitting the region of the solvent stimulated Raman band from 460 to 490 nm. The difference between the two traces is plotted on the right vertical axis and fit to an exponential decay. The inset shows double difference spectra as the repump pulse at 0.64 ps (∼0.05 ps width) interacts with the sample.

pulse, there is a strong stimulated Raman signal also observed in solvent-only scans, along with a short-lived increase in the total intensity of the visible absorption on the blue side of the spectrum only observed with DHC present. There may also be a very small net decrease in the intensity of the DHC ESA on longer time scales, but this is within the error of the measurement. The internal conversion of DHC Sn → S1 fits to a 150 fs time constant and is essentially complete. This is in contrast to measurements on stilbene where the second pulse produces a permanent depletion of the excited state absorption with a portion of the population returning to the ground state through another conical intersection.41 Once the population has returned to S1, there is a small blue-shift of the excited state spectrum, but no significant change in shape or decay. Elles and co-workers have used a pump−repump sequence to modify the photochemical reaction of the related 1,3-cyclohexadiene ring-opening reaction of a diarylethene.14,15 A second excitation pulse at 500 nm at short time delays is found to have no significant effect, resulting only in ultrafast internal conversion to S1, while excitation at time delays >3 ps or using 800 nm at time delays >0.3 ps accesses a reactive 6579

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A

(3) Geppert, D.; Seyfarth, L.; de Vivie-Riedle, R. Laser control schemes for molecular switches. Appl. Phys. B: Lasers Opt. 2004, 79, 987−992. (4) Rangel, N. L.; Williams, K. S.; Seminario, J. M. Light-activated molecular conductivity in the photoreactions of vitamin D3. J. Phys. Chem. A 2009, 113, 6740−6744. (5) Arruda, B. C.; Sension, R. J. Ultrafast polyene dynamics: the ring opening of 1,3-cyclohexadiene derivatives. Phys. Chem. Chem. Phys. 2014, 16, 4439−4455. (6) Deb, S.; Weber, P. M. The ultrafast pathway of photon-induced electrocyclic ring-opening reactions: The case of 1,3-cyclohexadiene. Annu. Rev. Phys. Chem. 2011, 62, 19−39. (7) Tapavicza, E.; Bellchambers, G. D.; Vincent, J. C.; Furche, F. Ab initio non-adiabatic molecular dynamics. Phys. Chem. Chem. Phys. 2013, 15, 18336−18348. (8) Kotur, M.; Weinacht, T.; Pearson, B. J.; Matsika, S. Closed-loop learning control of isomerization using shaped ultrafast laser pulses in the deep ultraviolet. J. Chem. Phys. 2009, 130, 134311. (9) Petrovic, V. S.; Schorb, S.; Kim, J.; White, J.; Cryan, J. P.; Glownia, J. M.; Zipp, L.; Broege, D.; Miyabe, S.; Tao, H.; et al. Enhancement of strong-field multiple ionization in the vicinity of the conical intersection in 1,3-cyclohexadiene ring opening. J. Chem. Phys. 2013, 139, 184309. (10) White, J. L.; Kim, J.; Petrovic, V. S.; Bucksbaum, P. H. Ultrafast ring opening in 1,3-cyclohexadiene investigated by simplex-based spectral unmixing. J. Chem. Phys. 2012, 136, 054303. (11) Kim, J.; Tao, H. L.; White, J. L.; Petrovic, V. S.; Martinez, T. J.; Bucksbaum, P. H. Control of 1,3-cyclohexadiene photoisomerization using light-induced conical intersections. J. Phys. Chem. A 2012, 116, 2758−2763. (12) Carroll, E. C.; White, J. L.; Florean, A. C.; Bucksbaum, P. H.; Sension, R. J. Multiphoton control of the 1,3-cyclohexadiene ringopening reaction in the presence of competing solvent reactions. J. Phys. Chem. A 2008, 112, 6811−6822. (13) Carroll, E. C.; Pearson, B. J.; Florean, A. C.; Bucksbaum, P. H.; Sension, R. J. Spectral phase effects on nonlinear resonant photochemistry of 1,3-cyclohexadiene in solution. J. Chem. Phys. 2006, 124, 114506. (14) Ward, C. L.; Elles, C. G. Cycloreversion dynamics of a photochromic molecular switch via one-photon and sequential twophoton excitation. J. Phys. Chem. A 2014, 118, 10011−10019. (15) Ward, C. L.; Elles, C. G. Controlling the excited-state reaction dynamics of a photochromic molecular switch with sequential twophoton excitation. J. Phys. Chem. Lett. 2012, 3, 2995−3000. (16) Woodward, R. B.; Hoffmann, R. Conservation of orbital symmetry. Angew. Chem., Int. Ed. Engl. 1969, 8, 781−853. (17) Okamura, W. H.; Elnagar, H. Y.; Ruther, M.; Dobreff, S. Thermal 1,7 -sigmatropic shift of previtamin-D3 to vitamin-D3 synthesis and study of pentadeuterio derivatives. J. Org. Chem. 1993, 58, 600−610. (18) Arruda, B. C.; Smith, B.; Spears, K. G.; Sension, R. J. Ultrafast ring-opening reactions: a comparison of α-terpinene, α-phellandrene, and 7-dehydrocholesterol with 1,3-cyclohexadiene. Faraday Discuss. 2013, 163, 159−171. (19) Tang, K.-C.; Rury, A.; Orozco, M. B.; Egendorf, J.; Spears, K. G.; Sension, R. J. Ultrafast electrocyclic ring-opening of 7-dehydrocholesterol in solution: The influence of solvent on excited state dynamics. J. Chem. Phys. 2011, 134, 104503. (20) Tang, K. C.; Sension, R. J. The influence of the optical pulse shape on excited state dynamics in provitamin D-3. Faraday Discuss. 2011, 153, 117−129. (21) Anderson, N. A.; Sension, R. J. In Liquid Dynamics: Experiment, Simulation, and Theory; Fourkas, J. T., Ed.; American Chemical Society: Washington, DC, 2002; Vol. 820, pp 148−158. (22) Anderson, N. A.; Shiang, J. J.; Sension, R. J. Subpicosecond ring opening of 7-dehydrocholesterol studied by ultrafast spectroscopy. J. Phys. Chem. A 1999, 103, 10730−10736. (23) Fuss, W.; Hofer, T.; Hering, P.; Kompa, K. L.; Lochbrunner, S.; Schikarski, T.; Schmid, W. E. Ring opening in the dehydrocholesterol-

population transfer from the relaxed population. This has no measurable effect on the ultimate quantum yield for the ringopened product. In the two-pulse experiment, the 545 nm pulse coming 0.64 ps after the initial 266 nm excitation interacts with population remaining on the excited state surface following the initial nonequilibrium passage through the region of the S0/S1 conical intersection. Rapid internal conversion back to the S1 state replenishes the relaxed population, with no significant effect on the subsequent transfer of population to the S0 state. Finally, excess energy placed primarily into vibrational modes assigned to CC stretching in the excited state plays no role in the initial ring-opening dynamics. This energy is largely sequestered during the excited state lifetime. Both the quantum yield for the reaction33 and the dynamics on the excited state are insensitive to the nonequilibrium distribution of extra energy. This observation is consistent with both LR-TDDFTSH and SA-CASSCF-AIMS simulations of the excited state ring-opening reaction.

■

CONCLUSIONS The photochemical ring-opening of previtamin D3 was shown to have novel biexponential kinetics in previous work from this group.19 To further probe the possibility of multiple surfaces, we report results for variable excitation energy and also repumping of transient excited states to affect yields or dynamics. There is no excitation energy dependence for wavelengths from 266 to 297 nm, which suggests little primary coupling of Franck−Condon active modes into a reaction coordinate. A visible pulse repump of the excited state at 545 nm and 0.64 ps has little influence on the excited state dynamics and no effect on product yield, which suggests fast internal conversion into modes that do not change the pathway. Finally, prior phase control of optical excitation at 266 nm that showed dynamics effects from pulse chirp20 is extended with the new experiments probing product yield and discussed in the context of very recent high level theory. The new SA-CASSCFAIMS simulations of the excited state ring-opening reaction predict that the observed biexponential decay has a dynamical basis. This prediction is consistent with all of our prior results and these new results and does not require parallel channels to account for the observations. The new calculations suggest that a pump−repump or pump−dump pulse sequence timed to intercept the highly distorted nonequilibrium ring-opened excited state molecule may provide the opportunity to manipulate the yield of previtamin D3.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected], phone 734-763-6074 (R.J.S.). Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by the National Science Foundation through grants NSF-CHE 1150660 and 1464584. REFERENCES

(1) Yokoyama, Y.; Gushiken, T.; Ubukata, T. Fulgides and Related Compounds; Wiley-VCH Verlag GmbH: Weinheim, 2011. (2) Irie, M. Diarylethenes for memories and switches. Chem. Rev. 2000, 100, 1685−1716. 6580

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581

Article

The Journal of Physical Chemistry A previtamin D system studied by ultrafast spectroscopy. J. Phys. Chem. 1996, 100, 921−927. (24) Meyer-Ilse, J.; Akimov, D.; Dietzek, B. Ultrafast circular dichroism study of the ring opening of 7-dehydrocholesterol. J. Phys. Chem. Lett. 2012, 3, 182−185. (25) Webb, A. R.; Holick, M. F. The role of sunlight in the cutaneous production of vitamin-D3. Annu. Rev. Nutr. 1988, 8, 375−399. (26) Tian, X. Q.; Holick, M. F. A liposomal model that mimics the cutaneous production of vitamin D3 - Studies of the mechanism of the membrane-enhanced thermal isomerization of previtamin D-3 to vitamin D-3. J. Biol. Chem. 1999, 274, 4174−4179. (27) Holick, M. F.; Tian, X. Q.; Allen, M. Evolutionary importance for the membrane enhancement of the production of vitamin-D3 in the skin of poikilothermic animals. Proc. Natl. Acad. Sci. U. S. A. 1995, 92, 3124−3126. (28) Arruda, B. C.; Peng, J.; Smith, B.; Spears, K. G.; Sension, R. J. Photochemical ring-opening and ground state relaxation in αterpinene with comparison to provitamin D3. J. Phys. Chem. B 2013, 117, 4696−4704. (29) Lu, J. K.; Sun, Y. B.; Miao, J. Y.; Deng, L.; Wang, F.; Tan, T. W. Photochemical production of vitamin D2, scale-up and optimization. Chem. Eng. Technol. 2007, 30, 994. (30) Terenetskaya, L. P.; Repeev, Y. A. Specific features of laser photosynthesis of previtamin D upon high-intensity picosecond laser irradiation. High Energy Chem. 1996, 30, 418−422. (31) Jacobs, H. J. C. Photochemistry of conjugated trienes: Vitamin D revisited. Pure Appl. Chem. 1995, 67, 63−70. (32) Koolstra, R. B.; Cornelisse, J.; Jacobs, H. J. C. Studies on vitamin-D and related-compounds 0.31. Photochemistry of estra-5,7diene-3-beta,17-beta-diol 17-acetate and its 9,10-seco isomer, 10desmethyl analogs of provitamin-D and previtamin-D. 1. Reversible isomerization-reactions. Recl. Trav. Chim. Pays. Bas. 1987, 106, 526− 533. (33) Jacobs, H. J. C.; Gielen, J. W. J.; Havinga, E. Studies on vitaminD and related-compounds 0.29. Effects of wavelength and conformation on the photochemistry of vitamin-D and related conjugated trienes. Tetrahedron Lett. 1981, 22, 4013−4016. (34) Tapavicza, E.; Meyer, A. M.; Furche, F. Unravelling the details of vitamin D photosynthesis by non-adiabatic molecular dynamics simulations. Phys. Chem. Chem. Phys. 2011, 13, 20986−20998. (35) Bearpark, M. J.; Larkin, S. M.; Vreven, T. Searching for conical intersections of potential energy surfaces with the ONIOM method: Application to previtamin D. J. Phys. Chem. A 2008, 112, 7286−7295. (36) Snyder, J. W. J.; Curchod, B. F. E.; Martínez, T. J. GPUaccelerated state-averaged CASSCF interfaced with ab initio multiple spawning unravels the photodynamics of Provitamin D3. J. Phys. Chem. Lett. 2016, 7, 2444−2449. (37) Tauber, M. J.; Mathies, R. A.; Chen, X. Y.; Bradforth, S. E. Flowing liquid sample jet for resonance Raman and ultrafast optical spectroscopy. Rev. Sci. Instrum. 2003, 74, 4958−4960. (38) Snellenburg, J. J.; Laptenok, S.; Seger, R.; Mullen, K. M.; van Stokkum, I. H. M. Glotaran: A Java-based graphical user interface for the R package TIMP. J. Stat. Soft. 2012, 49, http://dx.doi.org/10. 18637/jss.v18049.i18603. (39) Spears, K. G. VarPro; unpublished fast global fitting code that uses a variable projection algorithm for pump and repump experiments in Python language. (40) Sajadi, M.; Dobryakov, A. L.; Garbin, E.; Ernsting, N. P.; Kovalenko, S. A. Time-resolved fluorescence spectra of cis-stilbene in hexane and acetonitrile. Chem. Phys. Lett. 2010, 489, 44−47. (41) Smith, B. D. Ultrafast Optical Control of Ring-Opening and Isomerization in 7-Dehydrocholesterol and Stilbene. Ph.D. Dissertation, University of Michigan, 2015. (42) Cao, J. S.; Wilson, K. R. A simple physical picture for quantum control of wave packet localization. J. Chem. Phys. 1997, 107, 1441− 1450.

6581

DOI: 10.1021/acs.jpca.6b06967 J. Phys. Chem. A 2016, 120, 6575−6581