Solvent Dependent Dynamics of Salicylidene Aniline in Binary

Jan 6, 2017 - Solvent Dependent Dynamics of Salicylidene Aniline in Binary Mixtures of Supercritical CO2 with 1-Propanol or Cyclohexane. Ryan D. Kieda...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/JPCB

Solvent Dependent Dynamics of Salicylidene Aniline in Binary Mixtures of Supercritical CO2 with 1‑Propanol or Cyclohexane Ryan D. Kieda,† Adam D. Dunkelberger,‡ Amanda S. Case,* and F. Fleming Crim* Department of Chemistry, University of WisconsinMadison, Madison, Wisconsin 53706, United States S Supporting Information *

ABSTRACT: The role of different solvent environments in determining the behavior of molecules in solution is a fundamental aspect of chemical reactivity. We present an approach for exploring the influence of solvent properties on condensed-phase dynamics using ultrafast transient absorption spectroscopy in supercritical CO2. Using supercritical CO2 permits adjustment of the density, by varying the temperature and pressure, whereas varying the concentration or identity of a second solvent, the cosolvent, in a binary mixture allows for adjustments of the degree of interaction between the solute and the solvent. Salicylidene aniline, a prototypical excited-state intramolecular proton-transfer system, is the subject of this study. In this system, the decay rate of the transient absorption signal decreases as the fraction of the cosolvent (for both 1-propanol and cyclohexane) increases. The decay rate also decreases with an increase in the viscosity of the mixture, but the effect is much larger for the 1-propanol cosolvent than for cyclohexane. These observations illustrate that the decay rate of the photoexcited salicylidene aniline depends on more than just the solvent viscosity, suggesting that properties such as polarity also play a role in the dynamics.

I. INTRODUCTION The environment in which a chemical reaction occurs can dramatically influence the course of the reaction, and reactions in solution can behave very differently than their gas-phase counterparts. On a fundamental level, a solvent alters the potential energy surface for the reaction, introduces new energy transfer pathways, and may even become a reaction partner itself.1 Condensed-phase photoisomerization reactions are good candidates for examining the influence of various solvent properties, such as viscosity and polarity.2−8 Additionally, supercritical fluids provide a unique solvent environment: a liquid-like density with a gas-like viscosity. Supercritical CO2 (scCO2) is an interesting solvent choice, as changes in pressure and temperature will significantly alter its density and viscosity. There is also the potential to fix the temperature and pressure and add a second solvent (the cosolvent) to tune the properties of interest. Using solvent mixtures that combine the unique properties of scCO2 with polar and nonpolar cosolvents, we can determine the influence of viscosity and polarity on the isomerization of salicylidene aniline (SA) in solution. SA is a well-studied, prominent example of excited-state intramolecular proton transfer (ESIPT)9−35 whose photochromic properties give it potential applications in optical switching and storage.36,37 Previous experimental and theoretical investigations provide a relatively comprehensive description of the dynamics of SA, and we adopt the general photodynamical scheme presented by Sliwa et al. to describe the dynamics we observe.25 Figure 1 shows a simplified schematic representation of the photoisomeriztion of SA, giving the various isomers of SA that we refer to in the discussion of © 2017 American Chemical Society

our data and the time scales for their interconversion. Timeresolved photoelectron spectroscopy confirms that, following excitation of trans-enol SA, ESIPT occurs to produce vibrationally excited, excited-state cis-keto SA in 30−50 fs.25,29,31 After vibrational relaxation, the excited-state cis-keto SA may relax to the cis-keto ground state by spontaneous emission or isomerize to the ground-state trans-keto species by internal conversion. The trans-keto isomer is stable in solution on the millisecond time scale, and eventually isomerizes back to the trans-enol global minimum.25 We explore the effect of solvent properties on the dynamics of photoexcited SA by exploiting the unique properties of scCO2, which provide a low viscosity but high-density environment for reacting molecules. We alter the viscosity (and polarity) of a solvent mixture by changing the relative proportion of scCO2 and a cosolvent (1-propanol or cyclohexane). We find that these adjustments of the solvent composition substantially change the rate of decay of the excited-state cis-keto SA, and the effect is particularly large for the 1-propanol cosolvent. We consider the decay rate of the excited-state cis-keto SA to be a measure of the cis-keto to transketo SA isomerization rate. The strong sensitivity of the isomerization rate to changes in the cosolvent fraction indicates the importance of the solvent viscosity on the photoisomerization dynamics of SA. In addition, altering the identity of the cosolvent changes the reaction rate, reflecting the Received: June 14, 2016 Revised: December 22, 2016 Published: January 6, 2017 835

DOI: 10.1021/acs.jpcb.6b05959 J. Phys. Chem. B 2017, 121, 835−842

Article

The Journal of Physical Chemistry B

Figure 1. Schematic layout of salicylidene aniline (SA) photoisomerization, adapted from Sliwa et al.25 and Ortiz-Sanchez et al.31 Solid lines represent the proton transfer coordinate connecting the trans-enol and cis-keto SA, while dashed lines represent rotation out of the plane of the molecule to form trans-keto SA. Arrows represent the excitation pulse (blue), transient absorption (red), and spontaneous or stimulated emission (purple).

containing scCO2. When using only a single solvent, the sample cell consists of a 1 mm spacer compressed between two calcium fluoride windows with a peristaltic pump circulating solutions of SA in 1-propanol or cyclohexane through the sample cell. We ensure that the rate of flow is sufficient to provide a fresh sample for each laser pulse, and we adjust the concentration of SA to maximize the signal while attenuating the probe light to about 0.3 OD at 350 nm. For the experiments that use binary mixtures, we use a dual syringe pump system (Teledyne Isco D260) as our sample handling system, as shown schematically in Figure 2. The high-

influence of solvent polarity on the photoisomerization dynamics of SA as well.

II. EXPERIMENTAL APPROACH II.A. Laser Apparatus. We use the 800 nm and roughly 100 fs pulses from a regeneratively amplified Ti:sapphire laser system to produce the excitation and probe pulses for this study. Our data acquisition techniques and optical configurations are similar to those described previously.38,39 A pair of nonlinear β-barium borate (BBO) crystals converts the fundamental light to an approximately 3 μJ pulse of 266 nm excitation light. Focusing a small portion of the fundamental light into a calcium fluoride plate generates the broadband continuum (330−700 nm) probe light. A parabolic mirror collimates the probe light, and a neutral density filter splits it into signal and reference beams. Another parabolic mirror (f = 100 mm) focuses the probe light into the sample, and a 250 mm lens focuses the excitation beam into the sample at a slight angle (∼10°) to the probe. The small aperture of the highpressure cell requires a small angle between the excitation and probe beams. An iris blocks residual excitation light as well as scattered light, and a 100 mm lens collimates the probe light into a home-built spectrometer. The spectrometer disperses the signal and reference pulses onto a pair of matched 1024 element photodiode arrays, which collect and integrate over 300 laser shots for each exposure pair using a set of fast shutters. An exposure pair consists of a set of signal and reference pulses with and without the excitation light present. In general, we collect five exposure pairs varying the arrival time of the excitation pulse (with respect to the probe pulse) over a range from −5 to 40 ps. We then average four to six of these scans to create a data set for each given series of sample conditions. II.B. Sample Preparation. We conduct experiments in solution both with a single solvent and with binary mixtures

Figure 2. Schematic of the sample handling system. Dual syringe pumps introduce a SA solution and scCO2 into a sample loop through a tee. The resulting two-solvent solution circulates through a highpressure and -temperature optical cell where pump and probe laser pulses cross at a small angle to interrogate the sample.

pressure and -temperature optical cell (Harrick HPL-TC-13-3) consists of a set of 2 mm sapphire windows compressed around a 1 mm spacer and uses a pair of o-rings (Kalrez 0090) with a stainless steel compression ring to form the seal. The use of a hard window substrate and an O-ring material that resists absorption of scCO2 is imperative for avoiding damage to the cell, particularly when altering the system pressure. A heater 836

DOI: 10.1021/acs.jpcb.6b05959 J. Phys. Chem. B 2017, 121, 835−842

Article

The Journal of Physical Chemistry B assembly controls the temperature via a variable voltage power supply, and a K-type thermocouple embedded into the cell body monitors the temperature. All measurements in this work use scCO2 at 1500 psi and 40 °C, conditions that give a calculated density of 0.651 g/mL.40 We obtain SA, 1-propanol, and cyclohexane from SigmaAldrich and use them without any further purification. Table 1 Table 1. Dipole Moment, Dielectric Constant, Viscosity, Density, and Molecular Weight of 1-Propanol, Cyclohexane, and scCO240 a

1-propanol cyclohexane scCO2

dipole moment (D)

dielectric constant (ε0)

viscosity (η, cP)

density (ρ, g/mL)

molecular weight (g/mol)

1.68 0.3 0

20.1 2.0 1.4

1.41 0.71 0.05

0.803 0.769 0.651

60.09 84.16 44.01

Figure 3. Static absorption spectra of SA in 1-propanol and scCO2 solutions having 0.25% (red) and 99% (blue) 1-propanol; we normalize the absorption at the excitation wavelength (266 nm). The excitation and a portion of the probe wavelengths are indicated by the orange and green boxes, respectively. The inset shows the boxed area around 266 nm, highlighting the solvatochromatic shift that occurs with the changing solvent environment.

a

All data displayed are for the conditions of the experiment at 1500 psi and 40 °C.

absorption wavelength for SA as a function of cosolvent ratio for both solvent mixtures in the Supporting Information. III.B. Transient Absorption. We acquire TA spectra to study the dynamic processes that SA undergoes following excitation in these various solutions. Figure 4 is a contour plot

gives the properties of the individual solvents. A cylinder of SFC/SFE grade CO2 (Airgas SFX15AS) equipped with a dip tube supplies the primary syringe pump, while the other syringe pump receives the SA in 1-propanol or cyclohexane solution from a Teflon tube equipped with a 10 μm inline filter. We combine the SA solution with scCO2 in a tee. After this initial mixing, the new two-solvent solution flows into the optical cell loop where it circulates by means of a high-pressure magnetically coupled gear pump (Micropump GAH). This procedure ensures that there is fresh, thoroughly mixed, and well equilibrated (in both temperature and pressure) sample for each laser pulse. For the various samples, we adjust the concentration of the SA solutions such that dilution with scCO2 results in the same final SA concentration. For example, we use an 80 mM solution of SA in 1-propanol for the 0.5:99.5% 1propanol:scCO2 by volume solution and a 40 mM solution of SA in 1-propanol for the 1:99% 1-propanol:scCO2 by volume solution. We choose these concentrations because they maximize the transient absorption (TA) signal of SA, while still allowing the majority of the probe light through the sample at wavelengths (20 ps, TA (not shown in Figure 4) between 420 and 540 nm. It arises from the trans-keto isomer, which is stable for ms in solution and, thus, is a photoproduct on our time scales. Additionally, we know that there is a contribution from the excited-state trans-enol molecules at early times. Sliwa et al. have used singular value decomposition (SVD) to identify the contributions to a similar feature appearing following the 266 nm excitation of SA in acetonitrile. They determine early time contributions from the excited-state trans-enol molecules and long-time contributions from the trans-keto photoproduct, as well as contributions from the cisketo isomer and an unstable, twisted trans-enol form.25 Our TA spectra are consistent with their conclusions, and we consider the decaying absorption feature between 375 and 525 nm as primarily arising from the excited-state cis-keto molecules created after ESIPT. Figure 5 shows the time evolution of the data for a 5 nm region at each of the four wavelengths marked as A−D in Figure 4. The temporal profile at 640 nm (position A) reports on the stimulated emission from the excited-state cis-keto isomer, 480 nm (position B) primarily on the trans-keto 838

DOI: 10.1021/acs.jpcb.6b05959 J. Phys. Chem. B 2017, 121, 835−842

Article

The Journal of Physical Chemistry B

converges with the longer time constant for smaller fractions of 1-propoanol. This fast decay may result from excited-state absorption of the initially populated trans-enol excited state. However, Sliwa et al. report an ESIPT lifetime for this state of less than 50 fs in acetonitrile. Thus, the lifetime would have to increase by an order of magnitude in 1-propanol for the transenol excited state to be responsible for this new feature. Furthermore, the time constant for this fast decay is close to that of vibrational relaxation of the excited-state cis-keto isomer produced by ESIPT. It seems more likely that the fast, 400 fs component of the decay in the 406 nm TA signal for the highest 1-propanol fractions arises from efficient energy transfer of the vibrationally excited, excited-state cis-keto molecules to the 1-propanol cosolvent. The presence of this efficient energy transfer in 1-propanol, and not in cyclohexane, is likely a consequence of the increased solute−solvent interactions (particularly hydrogen bonding) in the polar solvent.

function of the cosolvent ratio are substantially more pronounced with the 1-propanol:scCO2 solvent mixtures. We concurrently fit a set of time traces at a single wavelength for a given set of solvent conditions (rather than averaging all of the data sets taken at a given set of conditions) using a convolution of an instrument response function (∼170 fs) and a double exponential decay or rise to allow us to quantify the time evolution of the SA dynamics. We account for the longtime offset observed in these data by fixing the longer time constant associated with the trans-keto photoproduct at 1000 ps. Table 2 collects the decay constants (averaged over the Table 2. Table of Fitting Constants Obtained from the TA Signals of Various Solutions of SA in 1-Propanol or Cyclohexane and scCO2 for Given Cosolvent Percentagesa cosolvent %

viscosity (cP)

100 50 30 25 15 12.5 7.5 6.25 3.75 3.125 1.85 1.51 0.9 0.5

1.4 0.249 0.129 0.110 0.080 0.074 0.063 0.060 0.056 0.055 0.053 0.052 0.051 0.051

100 64 32 16 8 4 2 1 0.5

0.708 0.253 0.113 0.076 0.062 0.056 0.053 0.051 0.051

406 nm (ps) 1-Propanol 7.61 ± 0.39 3.66 ± 0.24 3.01 ± 0.13 2.09 ± 0.18 2.00 ± 0.05 1.85 ± 0.05 1.83 ± 0.04 1.69 ± 0.04 1.59 ± 0.04 1.52 ± 0.03 1.55 ± 0.09 1.44 ± 0.04 1.62 ± 0.04 1.57 ± 0.03 Cyclohexane 2.45 ± 0.04 2.00 ± 0.03 1.86 ± 0.06 1.71 ± 0.04 1.65 ± 0.04 1.63 ± 0.04 1.45 ± 0.04 1.26 ± 0.05 1.44 ± 0.05

480 nm (ps)

640 nm (ps)

4.58 2.78 1.86 1.77 1.31 1.46 1.36 1.34 1.31 1.42 1.19 1.35 1.30 1.30

± ± ± ± ± ± ± ± ± ± ± ± ± ±

1.21 0.63 0.12 0.12 0.07 0.09 0.05 0.07 0.05 0.07 0.05 0.13 0.04 0.05

5.24 ± 0.07

1.95 1.48 1.51 1.33 1.35 1.76 1.40 1.42 1.47

± ± ± ± ± ± ± ± ±

0.08 0.11 0.09 0.10 0.11 0.09 0.05 0.07 0.05

2.37 ± 0.51 2.39 ± 0.23 2.56 ± 0.55

IV. DISCUSSION The decay of the 406 and 480 nm TA signals (as well as the rise of the 640 nm signal) primarily comes from the nonradiative decay of the excited-state cis-keto population. Molecular motions orthogonal to the ESIPT reaction coordinate, likely including motions that twist one of the rings out of planarity, facilitate the relaxation responsible for the decay of these various signals. Theoretical results suggest that the vibrationally hot, excited-state cis-keto SA formed following ESIPT twists nearly 80° out of plane to form a more stable conformation.31,32 The rate at which the twisting motions of these bulky groups occurs should depend on the viscosity of the solvent, and we anticipate that in a less viscous solvent SA can more easily access the necessary geometries that facilitate the nonradiative decay from the cis-keto excited state. The longtime offset present in the 406 and 480 nm TA signals signifies isomerization of the excited-state cis-keto SA to the groundstate trans-keto SA. Because we cannot disentangle the rate of formation of the trans-keto photoproduct, we presume that the nonradiative decay from the cis-keto excited state, which we observe at 406 and 480 nm, reflects the isomerization dynamics. Figure 6 shows the decay rate obtained using the exponential decay coefficient from our fit of the 406 nm TA signal as a function of solvent viscosity in both 1-propanol:scCO2 (red) and cyclohexane:scCO2 (blue) solvent mixtures. (The solid lines are power law fits to the data, which represent the functional form of this dependence well but do not reflect a quantitative or modeled fit of the data.) Enhanced microscopic

2.29 ± 0.20 1.78 ± 0.18 1.61 ± 0.16 1.34 ± 0.18 1.52 ± 0.19 1.18 ± 0.07 1.29 ± 0.07

a

The reported errors are the standard deviations that resulted from the fitting procedure.

entire data set for a given set of conditions) for the entire range of cosolvent fractions for both solvent mixtures.41 The times for the absorption decay (at 406 and 480 nm) and the stimulated emission rise (at 640 nm) in the 1-propanol:scCO2 solvent mixtures change by a factor of 3−5 from the highest (100%) to the lowest (0.5%) fraction of 1-propanol, decreasing from 7.6 to 1.6 ps. The outcome is different for the cyclohexane:scCO2 solvent mixtures, where the time constant decreases by less than a factor of 2 over the entire range of cosolvent fractions studied. Additionally, we find that the overlapping contributions in the 400−500 nm region make establishing a rise time for the trans-keto photoproduct (λmax = 480 nm) difficult, but we expect the formation of the trans-keto isomer to be prompt compared to the fluorescence lifetime of the cis-keto excited state (within 6 ps after ESIPT). The 406 nm data have an additional fast component (not shown in Table 2) for the highest 1-propanol percentages (100, 50, and 30%), but this roughly 400 fs component disappears or

Figure 6. Viscosity dependence of the decay rate (1/time constant) of the TA signal at 406 nm of SA in 1-propanol (red) or cyclohexane (blue) and scCO2. Power-law fits (solid lines) to the data only represent the functional form of this dependence and are not a modeled fit. 839

DOI: 10.1021/acs.jpcb.6b05959 J. Phys. Chem. B 2017, 121, 835−842

Article

The Journal of Physical Chemistry B

isomers. Nevertheless, our data show that the decay rate out of the cis-keto excited state is slower in a polar solvent (1propanol, 7.6 ps) than in a nonpolar solvent (cyclohexane, 2.5 ps). Previous studies have investigated the influence of various solvent characteristics on the TA dynamics and the photochromic properties of SA. In particular, Mitra and Tamai used femtosecond TA spectroscopy to observe the dynamics of SA in a variety of solvents. They observed that the ESIPT of SA is itself slower in solvents that can participate in hydrogen bonding; for example, the proton-transfer rate in ethanol is almost one-half of that in cyclohexane.17 The authors considered the decrease in the ESIPT rate a consequence of the preformed hydrogen bond between SA and the solvent, which is in direct competition with the intermolecular hydrogen bond that forms during the proton-transfer reaction. In addition to the ESIPT rate, this study measures the IVR rate for the vibrationally hot, excited-state cis-keto molecules formed via ESIPT. The authors report that there is no direct correlation of either rate with the solvent viscosity or polarity.17 However, the study of Mitra and Tamai only takes into account the fast vibrational relaxation component of the transient dynamics. On a longer time scale, we find that decreasing the viscosity or the polarity increases the decay rate of the 406 nm TA signal that monitors the excited-state cis-keto SA and, thus, reports on the excited-state cis-keto SA to ground-state trans-keto SA isomerization dynamics. Other studies also show that there are competing pathways for the relaxation of the excited-state trans-enol SA.25,31 Rotation about the C−N bond in the initially excited transenol excited state can lead back to the trans-enol global minimum or to a twisted-enol configuration that can survive for seconds. The channel through the twisted-enol isomer is relatively minor for excitation to S1 but is more important for excitation to higher electronic states.25,29,31 Exciting to higher electronic states results in the dominance of the twisted-enol channel over the ESIPT channel because the additional internal energy in the torsional motion both promotes the formation of the twisted enol by rotating about the C−N bond and suppresses ESIPT, which requires a planar geometry.29 Excitation-wavelength-dependent nanosecond photolysis25 and photoelectron spectroscopy29 experiments, as well as high level theoretical calculations,29,31,32 provide evidence for the change in product branching ratios. In femtosecond studies, the primary direct evidence of a long-lived, twisted-enol species is the presence of a long-lived, ground-state bleach. While any photoproduct will result in a loss of ground-state population at long time delays, Sliwa et al. discuss that the formation of the trans-keto photoproduct cannot explain all of their long-lived bleach.25 While we may also see the absorption of this twistedenol species, we do not believe that this channel is involved in the solvent-dependent dynamics we describe. The twisted-enol isomer forms promptly (