Experimental and Computational Analysis of Para-Hydroxy Methylcinnamate following Photoexcitation

Abstract

Para-hydroxy methylcinnamate is part of the cinnamate family of molecules. Experimental and computational studies have suggested conflicting non-radiative decay routes after photoexcitation to its S₁(ππ*) state. One non-radiative decay route involves intersystem crossing mediated by an optically dark singlet state, whilst the other involves direct intersystem crossing to a triplet state. Furthermore, irrespective of the decay mechanism, the lifetime of the initially populated S₁(ππ*) state is yet to be accurately measured. In this study, we use time-resolved ion-yield and photoelectron spectroscopies to precisely determine the S₁(ππ*) lifetime for the s-cis conformer of para-hydroxy methylcinnamate, combined with time-dependent density functional theory to determine the major non-radiative decay route. We find the S₁(ππ*) state lifetime of s-cis para-hydroxy methylcinnamate to be ∼2.5 picoseconds, and the major non-radiative decay route to follow the [¹ππ*→¹nπ*→³ππ*→S₀] pathway. These results also concur with previous photodynamical studies on structurally similar molecules, such as para-coumaric acid and methylcinnamate.

1. Introduction

Cinnamates and coumaric acids have been widely studied to elucidate their remarkable light absorbing properties, driven by efficient non-radiative decay (NRD) pathways [1,2,3,4,5]. In nature, these chromophores are embedded in proteins to enable vitally important processes to occur. A prime example is phototaxis in the photoactive yellow protein (PYP). PYP was first discovered in the Ectothiorhodospira halophila bacterium [6,7,8] and utilizes the photoisomerization of para-coumaric acid (p-CA) to facilitate a negative phototactic response towards blue light, thus protecting the bacterium from potentially harmful photodamage [9,10,11,12]. Similar families of molecules are used in sunscreen formulations due to their photoprotective properties against potentially harmful ultraviolet A (UVA, 315–400 nm) and B (UVB, 280–315 nm) radiation from the Sun [13,14,15].

Para-hydroxy methylcinnamate (p-HMC; Figure 1a) is the simplest ester derivative of p-CA and exhibits a strong absorption peak in the UVB region of the electromagnetic spectrum. Upon UV irradiation, the optically bright S₁(ππ*) state is predominantly populated and undergoes relaxation via multiple NRD processes, returning to the electronic ground S₀ state. Although the decay dynamics of p-HMC following photoexcitation to the ¹ππ* state are well-studied, there have been some conflicting conclusions regarding the dominant NRD route in the gas-phase. An early theory suggested that decay via the ¹πσ* dissociative state may dominate, analogous to what is observed in phenol [16]. The lowest lying ¹πσ* state in phenol is dissociative along the O–H bond and ultimately leads to a conical intersection (CI) with the S₀ state, providing rapid relaxation of the molecule [17,18]. Upon deuteration of the OH moiety, a dramatic change in excited-state lifetime is observed, indicating that the excited-state decay in phenol is mediated via tunnelling [19,20]. In contrast, deuteration of the OH moiety in p-HMC results in very little change in excited-state lifetime, implying that O–H fission is unavailable to p-HMC following photoexcitation to the S₁ state [21].

Following this observation, Smolarek et al. proposed an alternative NRD route for p-HMC, mirroring that of p-CA. This was based on the similarities in linewidths (and hence excited state lifetimes) of the photoexcitation spectra of p-HMC and p-CA, implying a comparable NRD route in both systems [21,22].

p-CA and methylcinnamate (MC), both structurally very similar to p-HMC, are photo-chemically and photo-physically well understood. It therefore stands to reason that they can be used as starting models for unravelling the NRD dynamics of p-HMC. Upon photoexcitation of p-CA to the ¹ππ* state in the gas-phase, the NRD process shown in Figure 2a dominates, with the observed lifetime of the ¹ππ* and ³ππ* states measured as ≤10 picoseconds (ps) and 20 ± 4 nanoseconds (ns), respectively [5]. For MC, the same NRD process dominates with the ¹ππ*, ¹nπ* and ³ππ* states having been experimentally observed. The observed lifetimes are also remarkably similar to p-CA, being 4.5 ± 1.1 ps and 20 ns for the ¹ππ* and ³ππ* states, respectively [23,24].

In contrast, a study by Yamazaki et al. for para-methoxy methylcinnamate (p-MMC), a close structural derivative of p-HMC, suggests that multistep intersystem crossing (ISC) from the ¹ππ* state to the ³ππ* state could involve an intermediate ³nπ* state [25]. Similarly, a recent computational study by Kinoshita et al. suggests a major NRD route for p-HMC and p-MMC, based on MC, that supports the involvement of the aforementioned intermediate ³nπ* state [1]. Their rationale is based on the fact that OH and OMe are π donors; the highest occupied molecular orbital (π) is destabilised by these π-donors, reducing the π-π* energy gap and consequently decreasing the energy of the ππ* states while having minimal effect on the nπ* states. This relative decrease in energy of the ππ* states reverses the adiabatic ordering of the ¹ππ* and ¹nπ* states, opening new major NRD routes. For p-HMC and p-MMC, Kinoshita et al. suggest the major NRD shown in Figure 2b.

There are four conformers of p-HMC, shown in Scheme 1. The S₁$\leftarrow$S₀ origin bands of these four conformers are located at 32,711.0, 32,715.9, 32,879.5, and 32,887.7 cm⁻¹ for, respectively, (a), (b), (c) and (d) [26]. From the same study, Tan et al. used the linewidths from UV-UV depletion spectroscopy to propose lower lifetime limits of 1.5 and 2.5 ps for the s-cis and s-trans conformers, respectively. Since this work, Kinoshita et al. and Shimada et al. have carried out time-resolved pump-probe studies on p-HMC. In both of these studies, the ¹ππ* lifetime was inferred to be ≤9 ps [27,28]. The ³ππ* state has also been experimentally observed at 19,020 cm⁻¹ and shown to have an average lifetime of 26 ± 4 ns [28]. However, the optically dark intermediate ¹nπ* or ³nπ* state is lacking a direct observation, and therefore opens speculation for the major NRD route. This lack of observation may be attributed to weak nπ*←S₀ absorption that is swamped by the strong S₁(ππ*)←S₀ absorption in combination with a large ionisation energy to remove an electron from an nπ* state [28].

In the present study, our aim is to: (a) precisely determine the gas-phase lifetime of the S₁(ππ*) state in p-HMC; and (b) contribute towards determining the major NRD route from the S₁(ππ*) state. To address (a), we use femtosecond (fs) to ps time-resolved ion-yield (TR-IY) and time-resolved photoelectron spectroscopy (TR-PES). The approximate 200 cm⁻¹ energy difference in the origin bands of the s-cis and s-trans conformers allows us to selectively excite and precisely measure the S₁(ππ*) state lifetime of the s-cis p-HMC conformer. To address (b), time-dependent density functional theory (TDDFT) is used to calculate the possible relaxation pathways originating from the S₁(ππ*) state in the s-cis conformer of p-HMC.

2. Results and Discussion

2.1. Experimental

Figure 3b,c show TR-IY transients for the s-cis conformer of p-HMC including the associated time constants and errors for the forward dynamics (pump-probe time delay (Δt) > 0), see Section 3.1. For these transients, the pump beam was centred at 308.5 nm (32,415 cm⁻¹) to ensure s-cis (syn and anti) conformer-specific excitation of p-HMC to the S₁(ππ*) state [26,27,28]. The spectral overlap of the pump beam with the S₁(ππ*) state of relevant conformers can be seen in Figure 3a. For the TR-IY transient in Figure 3b, the probe beam was centred at 240 nm (p-HMC’s minimum absorption, Figure 1a) to minimise both the probe-only ion-signal and reverse dynamics (Δt < 0). The ionisation potential (IP) and the energy of the ³ππ* state are located around 65,020 and 19,020 cm⁻¹, respectively [27,28]. As a result, a probe beam at 240 nm (41,667 cm⁻¹) ensures that ionisation is only optically accessible from the S₁(ππ*) state (according to the proposed NRD paths shown in Figure 2), since the first two ionisation levels are of π⁻¹ character and there is insufficient energy to reach the D₂(n⁻¹) ionisation state from any possible nπ* states [28]. This is evidenced by the total kinetic energy release (TKER) spectra in Figure 4 (discussed below), derived from reconstructed photoelectron images (Figure 4 inset), which shows no real structural change upon varying Δt, apart from a fast (<5 ps) decay. Knowing this, we can reasonably assume that the excited state lifetimes extracted from the TR-IY transient using a 240 nm probe in Figure 3b are directly related to the S₁(ππ*) state lifetime. For completeness, the probe beam in Figure 3c is centred at 200 nm (50,000 cm⁻¹) to enable ionisation from the S₁(ππ*) state, any possible intermediate states, and the final ³ππ* excited state located 46,000 cm⁻¹ below the IP. Since the ³ππ* state is known to have an average lifetime of 26 ± 4 ns, Figure 3c is expected to include an additional long-lived lifetime component (assuming it is formed within the time-window of our experiment, 900 ps) that would not be seen with the 240 nm probe due to the insufficient energy required to ionise from that state [28]. However, both transients exhibit the same dynamics, with three extracted time-constants τ_1–3 using the procedure outlined in Section 3.1.

To start our discussion, we will address the first time constant, τ₁ (∼150 fs). This lifetime is much shorter than the lower lifetime limit of 1.5 ps suggested by Tan et al. from the depletion spectra linewidth of the S₁(ππ*) state in s-cis p-HMC, and is therefore likely not the lifetime of the S₁(ππ*) state [26]. As mentioned supra, when using a 240 nm probe, the S₁(ππ*) is the only state available for ionisation. Knowing this, we assign τ₁ to initial evolution out of the Franck–Condon (FC) region, combined with any initial intramolecular vibrational redistribution, in keeping with previous studies [3,29,30,31].

The second time constant, τ₂, from the TR-IY transients in Figure 3b,c is 2.15 ± 0.18 ps and 2.46 ± 0.35 ps, for a 240 nm and 200 nm probe, respectively. These lifetimes broadly agree with the lower limit of 1.5 ps suggested by Tan et al. and the ≤9 ps suggested by Kinoshita et al. for the lifetime of the S₁(ππ*) state in s-cis p-HMC [26,28]. Additionally, TR-PES (with a 240 nm probe) derived TKER spectra, shown in Figure 4, almost returns to baseline by 2 ps with no real structural change or shift in the maximum TKER (TKER_max). This implies that the ps decay is solely related to the decay of the initially populated excited S₁(ππ*) state. The TKER_max is also in agreement with the expected TKER_max for p-HMC, originating from excitation to the origin of s-cis p-HMC. Using Equation (1), the expected TKER_max can be calculated:

$$TKE{R}_{max}=E_{OB}+E_{probe}-IP$$

(1)

where E_OB is the energy of the S₁(ππ*) origin band for the s-cis conformers, E_probe is the energy of the probe (41,667 cm⁻¹), and IP is the ionisation potential (65,020 cm⁻¹). As a result, the expected TKER_max is 9358 cm⁻¹ and 9362.9 cm⁻¹ for the two s-cis conformers, indicated by the black arrow in Figure 4 [26,27]. Encouragingly, this appears in excellent agreement with the observed TKER_max from the TKER spectra shown in Figure 4. Considering these results and their agreement with previous results, τ₂ is assigned to the lifetime of the S₁(ππ*) state of p-HMC in its s-cis configuration.

Finally, the third time constant, τ₃, is long-lived in both 240 nm probe and 200 nm probe transients and extends beyond the maximum time-window of our measurements, 900 ps (ESI, Figures S1 and S2). When probing at 240 nm, this result is surprising, since the only optically ionisable state is the S₁(ππ*). As such, this merited further investigation. The relative amplitude of this feature is dependent on the helium backing pressure, which can be seen in Figure 5a for the TR-IY transients at 1.5 and 5 bar helium. An increase in the backing pressure increases collisional cooling, which leads to formation of water clusters, shown in Figure 5a inset. At 1.5 bar helium, most of the ion-signal is attributed to p-HMC, but at 5 bar helium, the data includes a large ion-signal that is assigned to p-HMC + H₂O clusters. Furthermore, TKER spectra taken at 5 bar helium (Figure 5b), derived from reconstructed photoelectron images (Figure 5b inset), are clearly different to those at 1.5 bar helium (Figure 4). The first notable difference is the increase in TKER_max from approximately 9360 cm⁻¹ to >12,000 cm⁻¹. The TKER_max in Figure 5b accords with the expected TKER_max, 12,497 cm⁻¹, when using an IP of 61,585 cm⁻¹, determined by Smolarek et al., for a p-HMC + H₂O cluster (due to non-selective excitation, the expected TKER_max is obtained using the pump energy and not the energy of the S₁(ππ*) origin band for p-HMC + H₂O) [21]. From this, we can confirm the presence of water clusters at higher backing pressures. Additionally, it is clearly evident that the TKER spectra obtained at 5 bar helium do not tend to zero (cf. at 1.5 bar helium); there is still population of the first excited state at 50 ps. This correlates strongly with the long-lived S₁(ππ*) state lifetime for the p-HMC + H₂O cluster, previously measured by ps pump-probe spectroscopy [27]. Considering the above, the marked baseline offset in the TR-IY transients (notably with a 240 nm probe) described by the long-lived component τ₃, is attributed to the S₁(ππ*) state in p-HMC + H₂O clusters.

We end this section with a caveat: When probing at 200 nm, the mild baseline offset is expected since we can, potentially, ionise from both the ¹nπ* and ³ππ* states. Consequently, we are unable to determine whether τ₃ from the 200 nm probe TR-IY transient (see Figure 3c) is a result of cluster dynamics, ionisation from an nπ* intermediate state, a ³ππ* state, or any of these combinations.

2.2. Computational

As the experiments were carried out on the s-cis conformers of p-HMC, we have carried out calculations using both OH-syn and OH-anti conformations (Scheme 1a,b), respectively); for brevity we will refer to these as syn and anti in what follows.

Initially the ground state geometries of the two conformers were optimised using density functional theory (DFT) and the excited state energies evaluated at both points. It was found that the anti conformation had the lower S₀ energy at the minimum by 38 cm⁻¹. As such, all energies reported here are given relative to this minimum, whichever conformation they are. We are interested in the NRD mechanism of this molecule, specifically whether the ¹nπ* or ³nπ* state plays an intermediate role in that decay, and, as such, we note that, in both conformations, the ¹nπ* is the ninth lowest energy excited state (see data in ESI, Tables S1 and S2). There are five triplets below the bright S₁/¹ππ* state, followed by a sixth triplet before two more singlets, S₂ being a less bright (by a factor of over 4) ¹ππ* state and S₃ being the dark ¹nπ* of interest (the non-bonding orbital contains the lone pair on the carbonyl oxygen, see Scheme 1). Above these two singlets is another triplet state, which is the limit of our interest. We note that T₅ in both conformations is of nπ* character, whilst all other triplets have ππ* character. It is thus clear that excitation will be predominantly to the S₁ state, that being by far the brightest in both cases. The vertical excitation energies are 39,993 cm⁻¹ (anti) and 39,844 cm⁻¹ (syn), meaning that, whilst the absolute values of the excitations are about 7000 cm⁻¹ above the experimental values (which are adiabatic, so not directly comparable), they are close in energy with the correct ordering. In both cases, we also note that there is less than 3000 cm⁻¹ between all three singlet excited states.

Given the two optimised, FC geometries, they were then used as starting points for optimisation on the excited states: minimum energy geometries on all ten excited states were sought (using restricted Kohn–Sham (RKS)), as were the CIs between all energy adjacent states of the same multiplicity (using RKS), and finally all crossings between energy adjacent states, regardless of multiplicity (using unrestricted Kohn–Sham (UKS), which does not distinguish states based on multiplicity in ORCA). We note that not all optimisations were successful, with some minima not being located in spite of restarted calculations at the most recently visited geometry. All data is included in Tables S1–S5 in the ESI (where the data is presented in energy order and labelled S_x and T_y for clarity, because the character of the states change at different geometries), which gives energies relative to the anti S₀ minimum, but the main data to be discussed in what follows is shown in Figure 6 and Figure 7. Here, Figure 6 contains information about geometries optimised from the syn conformer and Figure 7 contains information about points found starting with the anti conformer. We will focus our discussion on these selected points that best relate to the NRD of s-cis p-HMC. With these selected points, we have also tried to locate the transition states (TSs) between relevant pairs of conformers in order to assess any energy barriers between them. To note, there appears to be no energy barrier from the anti FC to the syn FC point. Therefore, initial excitation to the anti conformer could be followed by rotation of the OH group to the syn conformer, prior to any excited state crossings.

Looking at the deactivation of p-HMC by crossing to lower energy states, there are two proposed mechanisms, both of which involve ISC to the triplet manifold through which deactivation to S₀ occurs; the difference between the two is whether the first step is internal conversion to a ¹nπ* state before the ISC step or whether the excited molecule goes from ¹ππ* to ³ππ* via ³nπ*. To assess this, we first look at the ISC process to see what feasible paths there are by examining the spin-orbit couplings (SOCs) between the singlet and triplet states at the minimum energy crossing points (MECPs) found using UKS TDDFT. A total of ten singlet–triplet MECPs were found, four optimisations starting in the anti conformer and six in the syn. In what follows, we will denote the MECPs found by the initial conformer (anti or syn) and the crossed states, so anti-1¹ππ*/3³ππ* is the crossing of 1¹ππ* and 3³ππ* starting from the anti-FC point. The optimised anti-1¹ππ*/3³ππ* geometry finished as a syn conformer, whilst all other nine systems stayed in their original anti or syn conformation. The anti-1¹ππ*/3³ππ* MECP geometry is non-planar with carbonyl C (Scheme 1) having a dihedral angle of 165° with the ring. The state energies at this MECP are similar to those found at syn-1¹ππ*/3³ππ* point, being only a few hundred cm⁻¹ different in all cases. The syn-1¹ππ*/3³ππ* geometry has a dihedral angle of −178.5° between the carbonyl C and the ring, so the geometries of the MECPs are different, suggesting they lie on different seams of intersection.

Looking at the SOCs between the singlet and triplet states, shown in

Table 1. Spin-orbit couplings between singlet and triplet states of p-HMC calculated at MECPs found by optimising geometries from both the anti and syn conformer FC points. Matrix elements for the x, y and z components (all imaginary) given along with the magnitudes.

Initial Geometry	MECP	Spin Orbit Coupling Values (cm−1)
		S→T (x)	S→T (y)	S→T (z)	Magnitude
Anti	11ππ*/33ππ* (S1/T3)	0.66i	0.52i	−0.95i	1.27
	11ππ*/63ππ* (S1/T6)	0.01i	0.00	−0.02i	0.02
	21ππ*2/63ππ* (S2/T6)	0.00	0.00	0.01i	0.01
	31ππ*/73ππ* (S3/T7)	0.00	0.00	−0.01i	0.01
Syn	11ππ*/33ππ* (S1/T3)	0.02i	0.71i	−0.06i	0.71
	11ππ*/43ππ* (S1/T4)	0.19i	−0.53i	−0.02i	0.56
	11nπ*/53ππ* (S1/T5)	4.29i	10.28i	1.58i	11.25
	21ππ*/53ππ* (S2/T5)	−0.59i	4.57i	0.99i	4.71
	21ππ*/63ππ* (S2/T6)	0.01i	0.00	0.00	0.01
	31ππ*/73ππ* (S3/T7)	−0.01i	0.00	0.00	0.01

, the only couplings with a magnitude of greater than 1 cm⁻¹ are at the anti-1¹ππ*/3³ππ* crossing (1.27 cm⁻¹), the syn-2¹ππ*/5³ππ* MECP (4.71 cm⁻¹) and the syn-1¹nπ*/5³ππ* geometry (11.25 cm⁻¹). As already noted, the anti-1¹ππ*/3³ππ* geometry is actually syn-like. We would expect the greatest rates of ISC to occur at these three crossing points, so let us consider the characters of the states there in more detail. At the anti-1¹ππ*/3³ππ* MECP, both intersecting states have ππ* character. At the syn-1¹nπ*/5³ππ* MECP, the singlet is over 84% nπ* with a little ππ* mixed in, whilst at syn-2¹ππ*/5³ππ*, the singlet is predominantly ππ*, but with over 20% nπ* character included. From this analysis, we conclude that the largest SOCs at the MECPs are between singlets with significant nπ* components in their character (the more the better) and ππ* triplets. This is in accordance with El-Sayed’s rule and therefore, unsurprising [32]. Pictures of the orbitals involved in the excitations, with largest components, which characterise the states that cross at the two geometries with largest SOCs are given in Figure S6 of the ESI. Even the anti-1¹ππ*/3³ππ* MECP noted here has a SOC an order of magnitude lower than that at the syn-1¹nπ*/5³ππ* geometry. It thus appears that the rate of ISC will be greater going from a ¹nπ* state to a ³ππ* state in the triplet manifold, rather than from the ¹ππ* state, suggesting that the NRD of p-HMC is mediated by the ¹nπ* state. In order for this to occur, there must be a route from the initially excited ¹ππ*state to the ¹nπ*, i.e., we need a CI and also a path to the crossing points from the FC point with a low barrier or none at all. We note that none of the other MECPs between singlet and triplet states, as listed in Table 1, involved the ³nπ* state, meaning that we have not located a path from the optically bright ¹ππ* state to the ³nπ* state, which was the alternative NRD pathway proposed (Figure 2b). If such a crossing could be located, NRD by this route would have to compete with internal conversion to the ¹nπ* state, but we would expect this spin-allowed transition to occur more quickly than the spin-forbidden ISC to a ³nπ* state.

A total of three CIs between the two lowest energy excited singlet states were found, two directly by searching for one starting at the FC geometries (anti-1¹ππ*/2¹ππ* and syn-1¹ππ*/2¹nπ*) and the third in the search for crossing using UKS TDDFT (anti-1¹nπ*/2¹ππ*). Considering first the anti-1¹ππ*/2¹ππ* point, this has an energy directly between the energies of the 1¹ππ* and 2¹ππ* states at the anti-FC geometry, corresponding to 2800 cm⁻¹ above the 1¹ππ* excitation point. The crossing would thus likely be energetically accessible in the experiment, however, we see that the character of the two states is ¹ππ*, i.e., it does not allow for crossing to the ¹nπ* state. We also note that the final geometry has a syn conformation; the OH bond now points in the opposite direction to its original position. Next, we consider the CIs labelled syn-1¹ππ*/2¹nπ* and anti-1¹nπ*/2¹ππ*, both of which have very similar energies (separated by about 50 cm⁻¹) suggesting that the main difference between them is the orientation of the OH bond. The crossings are below the energies of the 1¹ππ* state at either FC point, so are definitely energetically accessible after excitation. With regard to the state characters, both crossings are between the ¹nπ* state (greater than 85% components in both cases) and the bright ¹ππ* state, so they represent an accessible route from the initially excited state and the ¹nπ* state which can, later on, give access to the triplet manifold.

We have thus found a possible path from the bright 1¹ππ* state to the ³ππ* manifold and we also note that there are decreasing energy CIs between the triplet states which provide a path from T₅ to T₂, so there is a path towards the ground state. We have been unable to locate a crossing between the lowest two energy triplet states (except a very high energy crossing at a distorted geometry with the alkene-ester chain having migrated around the ring); however, that is not to say that one does not exist. With regard to the characters of the states at these crossing points: only the anti-3¹nπ*/4³nπ* crossing involves the ³nπ* state. For the syn conformations, all states are ³ππ* at the triplet CIs, apart from the fourth triplet at the syn-4³nπ*/5³ππ* CI. As the crossing from the singlet manifold is ¹nπ* to ³ππ*, it is not necessary for the decaying molecule to cross to the ³nπ* at these CIs, the nuclear wavepacket may stay on the ³ππ* state as it makes its way down towards the lower states. Transfer of density to the ³nπ* state may compete with the decay through the ³ππ* states, slowing the NRD.

As a final set of calculations, we searched for barriers between relevant geometries. Of first interest was any barrier between the CIs found between the two lowest energy singlets (syn-1¹ππ*/2¹nπ* and anti-1¹nπ*/2¹ππ*) and their corresponding FC geometries on the first singlet excited state potential energy surface. In both cases, the energies found were very close to that of the 1¹ππ* minimum (∼2500 cm⁻¹ below the 1¹ππ* energy at the FC point), suggesting that no barrier is present. We note that there is also no barrier to the 1¹ππ* minimum energy geometry, thus there is likely to be competition between relaxation for the 1¹ππ* minimum and to the singlet CIs, potentially slowing the NRD. A barrier was also sought between the syn-1¹ππ*/2¹nπ* geometry and the syn-1¹nπ*/5³ππ* point (i.e., the location of the largest SOC) and, again, none was found. This indicates that the path from the FC excitation geometry to the point of most likely ISC (1¹nπ* to 5³ππ*) is barrierless, the rate of NRD being determined by coupling between the states and the rate of competitive processes including relaxation to the 1¹ππ* minimum.

Bringing together the experimental and computational work, Figure 8 schematically illustrates the calculated major NRD route for s-cis OH-syn p-HMC along with the associated lifetimes, τ₁ and τ₂, determined by TR-IY and TR-PES. We focus specifically on the syn conformer since it shows a clearer path, with strong SOC, down through the inter-state crossings. Upon photoexcitation using a 308.5 nm pump, we propose that s-cis p-HMC evolves out of the FC region with no energy barrier in ∼150 fs. Subsequent to this, internal conversion to the ¹nπ* state occurs within ∼2.5 ps, and finally ISC to the ³ππ* state with no energy barrier.

3. Materials and Methods

3.1. Experimental Setup

The experimental setup has been described in detail elsewhere [23,33]. To briefly summarise, a Ti:Sapphire oscillator (Spectra-Physics Tsunami) and regenerative amplifier (Spectra-Physics Spitfire XP) are used to produce ∼40 fs laser pulses at a rate of 1 kHz and centred around 800 nm. The output laser beam is ∼3 W and is subsequentially split into three equal ∼1 W parts. One 1 W part pumps an optical parametric amplifier (Light Conversion TOPAS-C) to produce the pump pulse, centred at 308.5 nm to ensure s-cis conformer selective excitation of p-HMC to the S₁(ππ*) state. Another 1 W part either pumps a second TOPAS-C to generate the 240 nm probe or is successively frequency converted using a series of type I, type II and type I β-barium borate crystals to produce the 200 nm probe. For simplicity, the polarisations of the pump and probe are parallel with respect to each other, and the dynamics appear unaffected upon changing the probe to magic angle (54.7°) with respect to the pump—indicating the absence of rotational artifacts (ESI, Figure S3).

Both beams intersect a molecular beam produced by seeding p-HMC (99.9% purity by HPLC, Apollo Scientific) heated to 170 °C into 1.5, 3 and 5 bar helium. This gas mixture is expanded into vacuum via an Even-Lavie pulsed solenoid valve and is subsequentially passed through a 2 mm diameter skimmer [34,35].

For TR-IY, the photoions are accelerated, via an electric field, towards a detector which consists of a 25 mm active area MCP with a metal anode detector (Del Mar Photonics MCP-MA25/2). The signal output is measured by a digital oscilloscope (LeCroy LT372 Waverunner) and the ion-signal relating to the parent ion (p-HMC⁺) is monitored as a function of Δt; this creates the TR-IY transients.

A velocity map ion imaging setup, based off a design by Eppink and Parker, is used for TR-PES [36]. The photoelectrons are accelerated and focused onto a position-sensitive temporally-gated detector, consisting of two MCPs coupled to a phosphor screen (Photek VID-240) and imaged by a CCD camera (Basler A-312f). In this configuration, electrons with the same initial velocity are mapped onto the same point on the detector. The two-dimensional images generated are used to reconstruct the original three-dimensional Newton sphere using a polar onion peeling algorithm, from which the desired one-dimensional photoelectron spectrum is derived [37]. Xenon is used to calibrate the detector by which the obtained photoelectron spectrum is scaled to coincide with the energy of its well-known ionisation states [38].

Dynamical information from the TR-IY transients (S(Δt) vs. Δt) was extracted via a non-linear curve fitting algorithm consisting of a sum of exponential decay functions convoluted with the instrument response function (IRF):

$$S\left(\mathsf{\Delta }t\right)=S_0+G\left(\mathsf{\Delta }t\right)\ast \left[{\displaystyle \sum }_i{A}_i{e}^{\left(-\frac{\mathsf{\Delta }t-t_0 }{{\tau }_i}\right)}H(\mathsf{\Delta }t-t_0)\right]$$

(2)

Here, S₀ is the signal baseline, G(Δt) is the Gaussian function to account for the IRF, A_i is the amplitude of the i-decay, t₀ is time zero, τ_i is the lifetime of the i-decay and H(Δt − t₀) is a step function that is used to define time zero as the starting point of all observed dynamics; such that:

$$H\left(\mathsf{\Delta }t-t_0\right)=\left\{\begin{array}{c}0 if \left(\mathsf{\Delta }t-t_0\right)<0,\\ 1 if \left(\mathsf{\Delta }t-t_0\right)\ge 0.\end{array}\right.$$

(3)

The instrument response (IR) is estimated from the cross-correlation of the 308.5 nm pump and 240/200 nm probe with ammonia (ESI, Figures S4 and S5). From this, the IR is determined to be 117 ± 3 fs and 154 ± 6 fs with the 240 nm and 200 nm probe, respectively.

3.2. Computational Details

To further investigate the NRD of p-HMC, we have performed a series of electronic structure calculations to explore the ground- and excited-state potential energy surfaces of the molecule. To do so, we have used DFT for the ground state and TDDFT for the excited states, as implemented in the ORCA quantum chemistry package (version 5.0.1) [39,40,41], specifically using the ωB97X functional, both in RKS and UKS forms allied with the 6-311G(d,p) basis set [42,43]. The molecular geometry has been optimised using RKS DFT and TDDFT in order to locate energy minima on the electronic ground and excited states (both singlet and triplet); at such points, frequency calculations were carried out to ensure that the geometries were indeed minima of the potential energy surfaces, rather than saddle points. Additionally, MECPs between excited states were located, these being CIs in the case of crossing between states of the same multiplicity (found using RKS TDDFT). Once these minima and MECPs had been found, further calculations were performed to locate TSs between selected geometries, on the states of interest, where such barriers exist. The initial guess for the TS structure was found by taking the central geometry in a linear interpolation between the two selected geometries (using their respective Z-matrices). Upon convergence of the TS structures using the energies of the state of interest, frequency calculations were performed on that state to ensure that all frequencies were real except for one; the single imaginary frequency indicates that the structure is indeed a TS. All such optimised geometries were indeed found to be TSs using this condition. At points of crossing between singlet and triplet states, SOCs were calculated to assess the probability of ISC at such geometries. These points were located using UKS TDDFT, but were then checked using RKS TDDFT, where state energies were found to be the same as when using UKS; the SOCs of these RKS states found with quasi-degenerate perturbation theory [44]. Tight convergence criteria were used both for the geometry optimisations (convergence criteria were: 10⁻⁶ Eh change in energy; 10⁻⁴ Eh/Bohr maximum absolute gradient component; 3 × 10⁻⁵ Eh/Bohr root mean square (RMS) of gradient components; 10⁻³ Bohr maximum coordinate displacement and 6 × 10⁻⁴ Bohr RMS displacement) and the underlying electronic structure calculations (convergence criteria were: 10⁻⁸ Eh energy change; 10⁻⁵ Eh change in the one-electron component of the energy; 10⁻⁵ orbital gradient, 10⁻⁵ orbital rotation angle; and 5 × 10⁻⁷ error in the direct inversion of the iterative subspace).

4. Conclusions

We have investigated the gas-phase excited state dynamics of s-cis para-hydroxy methylcinnamate with both experiment and computation. Using pump-probe femtosecond time-resolved ion-yield and time-resolved photoelectron spectroscopies, the gas-phase S₁(ππ*) excited state lifetime was determined. Time-resolved ion-yield transients produced S₁(ππ*) lifetimes of 2.15 ± 0.18 ps and 2.46 ± 0.35 ps when probing at 240 nm and 200 nm, respectively. These lifetimes sensibly compare with time-resolved photoelectron spectroscopy measurements. Furthermore, the possible relaxation pathways were calculated by time-dependent density functional theory. We found the major non-radiative route, following photoexcitation to the S₁(ππ*) state, is mostly likely intersystem crossing to the ³ππ* state, mediated by an optically dark ¹nπ* state. Overall, the excited state lifetime and major non-radiative decay route from the first excited state in s-cis para-hydroxy methylcinnamate correlates strongly with structurally similar molecules, such as para-coumaric acid and methylcinnamate. This conclusion brings us closer to developing a greater understanding for the photodynamical behaviour of cinnamates and coumaric acids, to ultimately tailor similar molecules for applications including sunscreens and molecular heaters.