Time-resolved photoelectron and photoion fragmentation spectroscopy study of 9-methyladenine and its hydrates: a contribution to the understanding of the ultrafast radiationless decay of excited DNA bases

Abstract

The excited state dynamics of the purine base 9-methyladenine (9Me-Ade) has been investigated by time- and energy-resolved photoelectron imaging spectroscopy and mass-selected ion spectroscopy, in both vacuum and water-cluster environments. The specific probe processes used, namely a careful monitoring of time-resolved photoelectron energy distributions and of photoion fragmentation, together with the excellent temporal resolution achieved, enable us to derive additional information on the nature of the excited states (ππ*, nπ*, πσ*, triplet) involved in the electronic relaxation of adenine. The two-step pathway we propose to account for the double exponential decay observed agrees well with recent theoretical calculations. The near-UV photophysics of 9Me-Ade is dominated by the direct excitation of the ππ* (¹L_b) state (lifetime of 100 fs), followed by internal conversion to the nπ* state (lifetime in the ps range) via conical intersection. No evidence for the involvement of a πσ* or a triplet state was found. 9Me-Ade–(H₂O)_n clusters have been studied, focusing on the fragmentation of these species after the probe process. A careful analysis of the fragments allowed us to provide evidence for a double exponential decay profile for the hydrates. The very weak second component observed, however, led us to conclude that the photophysics were very different compared with the isolated base, assigned to a competition between (i) a direct one-step decay of the initially excited state (ππ* L_a and/or L_b, stabilised by hydration) to the ground state and (ii) a modified two-step decay scheme, qualitatively comparable to that occurring in the isolated molecule.

---

I. Introduction

The photodamage of DNA continues to be a subject of intense interest to biologists, chemists and physicists. DNA bases are the only nucleic acid components that can be electronically excited by the sun’s ultraviolet radiation. Therefore, the study of the conversion pathways of the substantial energy deposited by light absorption is of great importance.¹ The excited ππ* states of these complex aromatic systems mostly undergo non-radiative relaxation, namely either internal conversion or intersystem crossing to other dark or weakly allowed states, which could initiate photochemical processes. It is often stated that these latter processes are made highly improbable by a very fast (picosecond) internal conversion to the electronic ground state, where the electronic energy is eventually thermalised, providing nucleobases with a high level of photostability.^2–4 However, the detailed mechanism governing the excited state dynamics of the bases is far from being fully understood. In particular, the branching ratio between reactive and non-reactive relaxation pathways for each specific DNA base is not established and requires further investigation.

The present work is devoted to one of the most studied species: the purine base adenine (9H-Ade canonical isomer and its related methylated compound 9Me-Ade). Recent advances in time-resolved spectroscopic studies, in both the liquid and gas phase, and considerable theoretical efforts have recently given some insight on the nature and the properties of the lowest electronically excited states of this base.⁴ We shall begin with a short review of these recent results.

The electronic spectrum of the canonical form 9H-Ade is rather complex and consists of several closely spaced ππ*, nπ* and πσ* states. Theoretically, three low-lying singlet excited states have been found in the isolated base, giving rise to two electronic transitions of ππ* character (L_b in the near-UV and L_a at higher energy), and a very weak one of nπ* character.^5–9 The L_a state is the most strongly allowed state in the absorption spectrum. At higher energies, Rydberg states characterised by highly diffuse σ orbitals, localized mostly on the azine and amino groups, have also been predicted.^8,10–12 The molecular beam UV spectrum of 9H-Ade (and similarly for 9Me-Ade) is composed of a small number of resolved vibronic bands followed by an unstructured absorption at higher energy, which coincides with a fluorescence break-off.^13–17 Experimentally, both the frequency- and time-resolved observations in the gas phase indicate the existence of a rapid excited-state relaxation process with a low barrier (∼800 cm⁻¹).^13–23 The excited state dynamics exhibit a strong dependence on the excitation energy, with characteristic lifetimes ranging from a few tens of picoseconds near the origin to about one picosecond at excess energy of only ∼1300 cm⁻¹.^14,18–23 At this energy (267 nm), the excited molecule (9Me-Ade or 9H-Ade) mainly decays via a two-step mechanism with a short and long component of about 110 and 1300 fs,²³ assigned to the lifetimes of the optically bright ππ* L_b and the weakly allowed nπ* states, respectively. At higher excitation energies (250 nm), a similar temporal behaviour has been observed^20,24 as measured by both photoionisation and photoelectron spectroscopy. The localization of the excitation on the six-membered ring and the key role of the different vibrational modes (ring buckling, amino group inversion) in the relaxation process have been evidenced by the dynamical behaviour of different methyl-substituted adenines and 2-aminopurine (isomer of adenine).²³ Other studies on Ade,²¹ Ade–(H₂O)_n clusters²⁴ and Ade in water solution²⁵ have suggested that πσ* states might play an important role in the excited state relaxation. In room temperature aqueous solution, sub-picosecond lifetimes, found to be independent of the excitation energy, have been observed. After excitation at 267 nm, the decay is monoexponential for 9Me-Ade (0.22 ps), as observed by transient absorption,⁴ but monoexponential or biexponential as observed by fluorescence up-conversion for the ribonucleoside adenosine (∼0.31 ps)²⁵ and 2′-deoxyadenosine (dAdo) (∼100/420 fs) or 2′-deoxyadenosine monophosphate (dAMP) (∼100/520 fs), respectively.²⁶ Early time-resolved studies on adenine–water clusters reported monoexponential decays (∼60–220 fs).^24–27

On the theoretical side, several mechanisms were proposed to account for the ultrafast deactivation.^{10–12,28–31} Most of them involve a coupling between the optically bright, and therefore preferentially photo-excited, ππ* state and the nπ* state. Reactive and non reactive relaxation channels have been found theoretically. They proceed via conical intersections (CI) of the respective potential energy surfaces (PES), whose accessibility is governed by barriers of various heights. The N9–H loss or the opening of the five-membered ring involving Rydberg states,^8,10 for example, are predicted to occur at fairly high energy (∼230 nm and 200 nm, respectively) as compared to the almost-barrierless (∼0.1 eV) internal conversion to the ground state, which opens near the absorption threshold.¹⁰

This paper intends to bring together new experimental results as well as critically reviewed, previously published, material in order to provide a consistent set of data on the adenine system and its water clusters, to be compared with the theoretically proposed relaxation mechanisms. Among the low-energy excited states (L_b, L_a, πσ*, nπ*), the role of the state with the strongest oscillator L_a has not been considered so far. The contribution of this state to the dynamics involving the L_b and nπ*, and the branching ratio between these states in the dynamics of hydrogen bonded adenines (water solution or Ade–(H₂O)_n clusters) need further investigation. With these premises, considerable importance is attributed to precise experiments on the bases, in particular the biologically relevant species 9Me-Ade and its water-solvated species. Substitution of the N9 hydrogen atom by a methyl group blocks tautomerisation, thus eliminating the presence of 7H-Ade, and suppresses a non-relevant hydration site. It then provides a reasonable model for the base moiety in Ado, dAdo and dAMP, allowing a meaningful comparison with aqueous results on these species.⁴

Here, we report the first results obtained by time-resolved velocity map photoelectron imaging spectroscopy on 9Me-Ade. The mass selective photoionisation diagnosis performed in parallel guarantees the origin of the observed signals. In addition, the time-resolved measurements of fragmented species are shown to provide valuable information on the vibrational content of the system.^32,33 These complementary techniques provide a consistent set of data, allowing us to determine the nature of the electronic excited states in agreement with the theoretical approaches. Then, in an attempt to bridge the gap between isolated molecules and the biological environment, the photophysics of the model nucleoside 9Me-Ade in water clusters has been investigated by time-resolved photoionisation spectroscopy, with special attention given to the fragmentation process.

II. Experimental

The experimental setup combines a pulsed supersonic jet with femtosecond pulsed lasers and a detection system that can analyse either photoion signals through a time-of-flight mass spectrometer (TOF–MS) or photoelectron signals through a velocity map imaging device.

The beam is generated from a co-expansion of the carrier gas (argon or helium) with nucleobases produced in a temperature-controlled oven placed upstream of the heated pulsed valve (General Valve). Typical operating temperatures were about 140 °C. Commercial reagents from Aldrich were used without further purification. In order to form base hydrates, the carrier gas was seeded by passing through an upstream, temperature-controlled, water reservoir. The stagnation pressure P₀ (∼1–2 bars) and the temperature of the water reservoir (−10 °C) were kept low to avoid water aggregation and a large distribution of adenine–water clusters. Moreover, laser excitation occurred at a very short delay after opening the valve. The temperature was always kept as low as possible to ensure sufficient signal of isolated base, but avoid formation of base clusters.

The LUCA laser facility at Saclay delivers the second (400 nm) and third (267 nm) harmonics of the output of a Ti:Sapphire regenerative amplifier operating around 800 nm at a 20 Hz repetition rate. The two nearly collinear pump and probe laser beams can be focused independently onto the molecular beam. The power density of the probe laser was kept lower than 2 × 10¹¹ W cm⁻², which is enough to perform multiphoton ionisation of the electronically excited adenine. For all measurements, pump (267 nm) and probe (400 or 800 nm) laser energies and their focusing conditions were carefully adjusted to minimize one-colour ionisation. The IR fundamental light (800 nm) was eliminated when using the 400 nm probe beam by successive reflections on interferometric multilayer mirrors. The fundamental or second harmonic output of the probe laser goes through a computer-controlled delay line, allowing us to vary the time delay relative to the 267 nm pump laser pulse over about 600 ps. The temporal profiles were acquired either by randomly setting the pump–probe delays from a predetermined list of values (over a range of a few ps) or by scanning the range back and forth and averaging (for decays of several tens of picoseconds). This allowed us to minimize the effect of laser intensity fluctuations. Positive times correspond to the probe pulse (800 nm or 400 nm) coming after the pump pulse (267 nm).

The ions are detected by the microchannel plate detector of a TOF–MS perpendicular to the molecular beam and the lasers. When working with 800 nm probe laser, the carrier gas was switched from argon to helium because of an intense signal from ionisation of argon.

The photoelectrons are detected in the opposite direction by a velocity map imaging spectrometer based on the design of Eppink and Parker.³⁴ The electrostatic optics are adjusted to accelerate the electrons perpendicular to the molecular beam in a velocity mapping electric field. The polarisation of the lasers are both set parallel to the plane of the detector. The photoelectron signal is detected by a position-sensitive imaging detector composed of microchannel plates coupled with a phosphor detector monitored by a charge-coupled device camera (LaVision). This system allows us to project the spherical distribution of the photoelectrons’ speed (directly linked to their energy) on the planar surface of the camera detector. The treatment by Abel inversion of this projection with an adapted algorithm (Hansen & Law³⁵) after symmetrisation of each image enables us to recreate the original spherical distribution and to extract information on the radial and angular energy distribution of the photoelectrons. This finally gives access to the energy distribution of the photoelectrons for each time-step recorded. In the present conditions, the electron energy range observed was 0–3.4 eV, with an energy resolution of about 0.1 eV (at 1 eV). Prior to each photoelectron measurement, TOF mass spectra have been measured in order to check for the presence of trace compounds such as dimethylether (DME) and pyridine (used for internal calibration).

Data acquisition and analysis is achieved through home-made, Labview-based software. Each photoelectron image and mass spectrum was averaged over 1000 laser shots per pump–probe delay. In each case, a baseline was calculated and subtracted. For mass spectra, the baseline was calculated by averaging the signal before each peak of interest. This permits us to correct the influence of baseline drifts due to strong signals occurring at smaller masses (for example, the influence of the monomer signal on the clusters or dimers). In the case of the photoelectron images, an average image of the one-colour signal (negative times) was subtracted to all images within a series. The one-colour signal indeed presents features characteristic of the electronic states accessible by one-colour multiphoton absorption. The fitting procedure of the transients involves convolution of a multiple-exponential decay by the laser cross-correlation. The zero-delay time and the laser cross-correlation width are fixed parameters in the fitting procedure, and deduced from an independent in situ ionisation measurement of DME or water in the molecular beam. The cross-correlation widths of the lasers (assumed to be Gaussian-shaped) obtained are of the order of 80 fs (pump–probe: 267/400 nm) or 120 fs (267/800 nm). The relative precision of the values obtained for the lifetimes is estimated to be ±10%.

III. Results and discussion

A. Bare 9-methyladenine

Time-resolved photoionisation spectroscopy. The time-dependent ion transients of 9Me-Ade measured at the parent ion mass channel for two different pump–probe schemes (267/400 vs. 267/800 nm) are shown in Fig. 1b. The 267/400 nm pump–probe data, already published,²³ will be used as a reference for the new sets of data reported in the present study. The two 9Me-Ade signals show very similar temporal behaviours, which can be reproduced by a two-component exponential decay previously assigned to the initially excited ππ* (L_b) and nπ* state. A first ultrafast component of typically 100 fs is followed by a slower step in the picosecond time range. The lifetimes (95/1240 fs) are found to be the same within the measurement precision.

Fig. 1 Time-resolved photoion and photoelectron transients observed (dots) in 9Me-Ade with two pump–probe schemes (267/400 nm and 267/800 nm). (a) Decays measured by a photoelectron velocity imaging technique integrated over all electron energies, (b) Decays measured by photoionisation with detection in the parent mass channel, (c) Decays measured by photoionisation with detection in the m = 122 amu fragment channel. The dash-dotted and dotted curves show the convolution of the Gaussian cross-correlation with the first and second exponential decay components, respectively. The sum of these two components with a small offset (accounting for one-colour signal) gives the full line curve fitting the experimental data.

Another relevant parameter in time-resolved measurements is the relative weight of the two components, which depends on the nature of the excited states (and controls the photoionisation efficiency) and can therefore provide information on the deactivation pathways. In the present case, we observe similar relative weights of the two components (see the heights of the calculated components in Fig. 1) when comparing the two different pump–probe schemes, which suggests the absence of a specific effect caused by the probe process (400 vs. 800 nm photons). In the following, this parameter will be used extensively to characterise the excited state dynamics.

At this stage of the analysis, it should be mentioned that, in pump–probe experiments at high laser intensities, the signals measured may be perturbed by large one-colour signals and pump–probe signals due to strong laser intensities, as well as significant fragmentation of the parent 9Me-Ade ions following the photoionisation probe. These phenomena may have significantly affected previous measurements and corresponding fits.^21,22,24 In the present study, the laser intensities were kept as low as possible and the signal for negative delay times, corresponding roughly to one-colour background signals, was low (∼10–20%). This signal was found to be constant and equal to the signal at very long positive delays (>10 ps). This was checked prior to each measurement in order to ensure appropriate and steady experimental conditions (moderate laser intensity and focusing, good spatial overlap of the lasers, controlled expansion conditions). Owing to the good signal-to-noise ratio (and our excellent time resolution) together with the minimized one-colour signal, an accurate fit of the parent ion transient based on a biexponential model was obtained without addition of a third, very long-lived (ns), channel. Also, the absence of any multi-photon coherent absorption (laser cross-correlation), as already suggested,²³ was confirmed.

Concerning the second perturbing effect, fragmentation is commonly observed in time-resolved experiments using photoionisation as a probe. The large intensities of the probe lasers typically used mean it cannot be avoided, because of the large probability of ion photoexcitation during the probe laser pulse. The following analysis will nevertheless show that this phenomenon provides an interesting alternative way to monitor the excited state dynamics.

In the present experiment, the mass spectrum of 9Me-Ade shows the presence of many fragment ions (5–10% of the parent intensity) as well as the intense parent peak. The most intense fragment ion (122 amu) may correspond to the opening of the six-membered ring and loss of either HC2N3 or N1C2H (27 amu) (see Fig. 1a for the numbering). As a single photon excitation at 267 nm (4.64 eV) cannot cause prompt dissociation in 9Me-Ade (it would require the breaking of two C–N bonds), the fragmentation observed must occur in the ion after absorption of several probe (400 nm or 800 nm) photons (total energy >12 eV³⁶). It is noteworthy that the fragmentation in the ionic state may be relatively slow, on a timescale ranging from ps to tens of ns, depending on the excess energy and the size of the molecular system. As the fragment ions are created from the parent ions formed during the probe pulse, their pump–probe time dependence is expected to be the same. Interestingly, the pump–probe time profile measured at 122 amu clearly differs from that of the parent. The fragment decay signals (Fig. 1c) essentially correspond to the second component observed on the parent ion. It can be fitted by a rising exponential of 10–20 fs followed by a monoexponential decay almost identical to the second component value measured on the parent ion. This suggests that probing the dynamics from the fragment signal essentially focuses on the later part of the dynamics. It should be mentioned that this fragment signal is of course produced at the expense of the parent signal (namely, here, mainly the second component). The same type of observations have been previously reported on other, smaller systems^32,33,37 It has been shown that an increased vibrational content in the excited PES of the molecule can lead, after ionisation, to an enhancement of the fragment ion signals, thereby depleting the parent ion signal. Measuring differences between the temporal response of the parent and fragment ions can be a way to visualize the regions on the PES where the ionic fragmentation is favoured by vibrational excitations acquired in the excited neutral state. The comparison between the present parent and fragment ion transients nicely illustrates this effect on 9Me-Ade and provides evidence for the involvement of two excited electronic states with different vibrational contents.

These results demonstrate that the presence of fragmentation can significantly affect the temporal decay measured for the parent ion. The perturbation observed, i.e., the decrease of the second component’s weight, will be nicely confirmed by photoelectron spectroscopy (see next section). Of course, such a fragmentation effect is expected to be even more dramatic in the case of clusters characterised by smaller binding energies. The above discussion explains why it may be difficult to compare the present results with previously published decays acquired under different experimental conditions (strong fragmentation effects due to large laser intensities and/or presence of molecular clusters). With these limitations in mind, one can nevertheless note the consistency of the present results with previous work at different excitation energies (272, 263 and 250 nm)²² for Ade. In this former study, all the transients show biexponential decays with no significant change of the first lifetime but an increase of the second one, from ∼2.3 ps at 272 nm to ∼1 ps at 250 nm.

Time-resolved photoelectron velocity imaging of 9Me-Ade. Photoelectron spectroscopy is sensitive to both the molecular orbital configurations and the vibrational content, and is therefore the technique of choice to study the electronic relaxation processes in molecules. Moreover, the electron velocity imaging technique used in the present work is a very sensitive method because the electron counting rate may be very high and similar to that for the corresponding ions.

In a typical image (see Fig. 2d), radial (electron energy) and angular (spatial anisotropy, if present) dependencies can be observed. The images can be integrated over all angles (thus disregarding any angular dependence) in order to focus on their time dependence. Temporal profiles of photoelectron intensities integrated over the whole kinetic energy range are shown in Fig. 1a. The temporal profiles of the mass-resolved and photoelectron measurements show the same lifetimes (within the fit precision). The much stronger second component in the photoelectron measurement compared to the parent photoion decay is qualitatively consistent with a large depletion of the second component due to fragmentation, as discussed in the previous section. The qualitative agreement between the two lifetime measurements nevertheless supports the absence of perturbing trace compounds in the photoelectron experiments.

Fig. 2 Time-resolved photoelectron spectroscopy of 9Me-Ade using the velocity imaging technique. (a) Energy distributions of the 9Me-Ade photoelectrons measured with the 267/400 nm excitation scheme, corresponding to the first (dots and full line displaying a smoothing of the data) and second (dotted line) components of the decay. The two stars indicate the expected ionisation thresholds for the D₀ and D₁ ion states. (b) Typical photoelectron velocity imaging spectrum obtained after treatment of the raw data. (c) Transients observed by integration of selected electron energies, and calculated first and second components (dotted lines) of the fitting curve. (d) Example of a raw image, taken at zero time delay, and displaying the photoelectron intensity as a function of the position on the detector (increasing intensity from light to dark grey).

Time-resolved photoelectron kinetic energy distributions also provide spectral information on the electronic character of the excited states.^21,38 The electron energy distributions corresponding to the two components (95 fs and 1240 fs) in the 267/400 nm excitation scheme are shown in Fig. 2a. The energy distribution corresponding to the second component is easily extracted by integrating a time window where the first component has already decayed to zero (∼1 ps after the zero time delay). In contrast, the energy distribution of the first component cannot be obtained directly. In this case, we have considered only one time channel, close to the zero delay, where the relative weight of the first component compared to the second is the highest, and have determined for different energy windows the weight of the first component at this delay in the dynamics. It is then possible to extract the first component contribution from the total energy distribution obtained at this specific delay. The bottom scale of Fig. 2a shows the kinetic energy of the photoelectrons, whereas the upper scale gives the corresponding binding energies of the electrons assuming a 267/2 × 400 nm photoionisation scheme and a total energy of about 10.84 eV. The two curves have different envelopes, characterised by a large energy spread and very slow increase of the photoelectron intensity (reading from high to low kinetic energies). The first component is related to electron energies ranging from 0.2 to ∼2.5 eV, while the onset of the second component distribution is shifted to lower kinetic energies by ∼0.5 eV and shows a steep increase at 1.25 eV.

These results can be analysed based on the comparison of the experimental ionisation thresholds with those expected from Koopman’s ionisation correlations^39–41 combined with well characterised He(I) photoelectron spectra. The adiabatic ionisation energy of jet-cooled 9H-Ade has been determined by two-colour ionisation spectroscopy to be 8.606 eV.¹⁵ Comparing this value with He(I) photoelectron spectra of gas phase adenine and 9Me-Ade,⁴² we estimate the adiabatic ionisation energy of jet-cooled 9Me-Ade to be IE₀ = 8.5 eV (π hole) and IE₁ = 9.5 eV (n hole). When exciting at 267 nm (4.64 eV), the vibrational energy in the L_b ππ* state of 9Me-Ade (origin transition at 4.48 eV ¹⁸) is of the order of 0.164 eV and slightly more (∼0.24 eV) in the nπ* state, if we assume that 9Me-Ade and 9H-Ade follow a similar trend for the nπ* state (∼600 cm⁻¹ red shift).¹⁷ The Franck–Condon principle implies (for vertical transitions) that most of the vibrational energy in the excited neutral states is transferred to the ion. The estimated electron energy for threshold ionisation of the ππ* state to the D₀ state is then ∼2.18 eV (10.84 − 8.5 − 0.164). This value (represented by a star in Fig. 2a) corresponds fairly well to the half maximum of the ∼95 fs component curve. The estimated energy at threshold for the nπ* state (also represented by a star in Fig. 2a) is ∼1.1 eV if we suppose that it preferentially ionises to the D₁ state (n hole).²¹ The step-like intensity increase in the experimental curve (∼1.25 eV) corresponds fairly well to this value. The observed weak signal at threshold is located at higher energies (∼2eV), which may indicate that transitions from the nπ* state to the D₀ continuum are also weakly allowed. This is probably due to small deviations from the simple Koopman’s picture and the strong coupling between the two states.^43,44 The very slow rise of the Franck–Condon envelope observed for both states (i.e. not a step function) may indicate different geometries of the neutral S₁ and S₂ and ionic D₀, D₁ and D₂ states of 9Me-Ade.

The temporal profiles for energy-selected photoelectron intensities (Fig. 2c) show all biexponential decays fitted with the same lifetimes as for the all-photoelectron profiles. Different relative weights of the components are found, however, depending on the kinetic energy selected, for example, the first component’s weight is very small for electron energies in the 0.85–1.2 eV range, and increases for higher electron energies (1.9–2.5 eV). This observation is a direct consequence of the different energy distributions associated with the two components.

In a recent photoelectron study on Ade, an additional minor relaxation channel with a short lifetime (assigned to a πσ* state) was proposed^20,21 when exciting specifically at 267 nm (and not at 250 nm). In that case, the envelope of the electron kinetic energy distribution of the first component showed an additional low-energy electron signal (<0.7 eV) as compared to the present one. Indeed, a careful comparison of the time profiles obtained in this earlier study (integrated over all electron energies) with the present one (Fig. 1a) clearly indicates that the relative weight of the first component observed for the 267 nm excitation was anomalously intense in this former study, while the components’ ratio at 250 nm excitation is similar to the present photoelectron time profile, suggesting the absence of a marked dependence of the dynamics on the excitation wavelength. Such a feature would be in qualitative agreement with lifetimes measured at several wavelengths (272, 263 and 250 nm) in a previous series of photoion probe experiments (showing similar behaviour with excitation at 263 and 250 nm).²² A likely explanation for the additional low-energy photoelectron signal²¹ is the accidental presence of trace compounds in the jet during these 267 nm measurements, for example water and Ade–(H₂O)_n clusters, which are found to be very short-lived (see below). These species may have been much less populated or absent during the 250 nm experiment.

The present photoelectron results, namely the agreement of the kinetic energy distributions for the two temporal components with the expected vertical ionisation potentials and the excellent fit of the energy-selected temporal profiles with biexponential decays with the same lifetimes, are consistent with the complementary photoion data on both the parent and fragment ions, whatever the probe scheme used (400 vs. 800 nm). They support the involvement of only two electronic states in the deactivation of 9Me-Ade after 267 nm excitation. The directly excited short-lived ππ* electronic state (∼100 fs component) relaxes to a nπ* state (second component of ∼1200 fs), exhibiting a significant vibrational content. In addition, there is no evidence for the observation of a long-lived channel involving, for instance, a triplet state (in contrast to the earlier photoelectron study²¹).

At higher energy (250 nm), photoelectron studies²¹ on the relaxation of Ade excited states show similar temporal profiles, suggesting the involvement of electronic states of the same nature, even if different ππ* states are probably optically excited at this shorter wavelength (L_b and/or L_a).

B. 9Me-Ade–(H₂O)_n clusters

The time-dependence of the 9Me-Ade hydrates is reflected by the transient profiles recorded in the 9Me-Ade–(H₂O)_n (n = 0–10) mass channels (selected sizes are shown in Fig. 3). The carrier gas used for the expansion was argon for the measurement with the 400 nm probe and helium with the 800 nm probe. Great care was taken to prevent the formation of 9Me-Ade dimer (and higher clusters) and the corresponding hydrates. The molecular beam expansion conditions were optimised to produce narrow distributions of hydrates of the monomer only. The apparent cluster distribution vanishes for n > 10.

Fig. 3 Time-resolved photoionisation of 9Me-Ade–(H₂O)_n clusters displayed for two excitation schemes (267/400 and 267/800 nm). (Upper panel) The parent mass channel signal (dots) is compared to the 9Me-Ade–H₂O cluster (fine dotted line). Insert reveals a 20 fs shift between the maxima of the two curves in the case of 267/800 nm pump–probe scheme. (Following panels) Transients for the 9Me-Ade–(H₂O)_n, n = 1, 3 and 5. In all cases, the fitting curves are displayed in full line, and the mathematical first and second components in dotted lines.

All transients are fitted by a biexponential decay (∼100/∼1000 fs), with a much weaker second component as compared to the monomer measured under dry conditions (Fig. 1 and 3). The lifetime of the first component is found to decrease significantly with the cluster size, for example from 120 fs (110 fs) for n = 1 down to 100 fs (80 fs) for n = 6 when probed by the 267/400 nm (267/800) pump–probe schemes. The same trend is observed for the lifetimes of the second component, going from 980 fs (1000 fs) to 750 fs (500 fs), and its relative weights.

These results, in qualitative agreement with earlier studies,^24,45 show an apparent drastic change of the 9Me-Ade dynamics in a water cluster environment, in particular, a dramatic decrease of the second component weight compared to the isolated molecule. This new temporal features in clusters can have two possible origins. On one hand, it can be due to a specific solvent effect on the electronic dynamics of the 9Me-Ade molecule. On the other hand, it could be a consequence of an intense ion fragmentation, as described above for the isolated monomers, but much more important in these species because of their lower binding energy. Dissociation in the neutral excited state is energetically possible but its effect are probably small for two reasons: (i) the excess energy in the excited state is only sufficient to evaporate a few water molecules, since the corresponding cooling slows down the following evaporation events, and (ii) evaporation is probably negligible within the 1 ps timescale for small size hydrates (n > 3).⁴⁶

Evidence for an intense fragmentation in the ionic state can be found in the temporal behaviour of the ion signal measured in the 9Me-Ade⁺ mass channel (Fig. 3). As compared to the isolated monomer (Fig. 1b), the monomer transient with the 267/400 nm pump–probe scheme shows a longer first component and an apparently weaker second component. For comparison, the decay of the 9Me-Ade–H₂O complex is shown on the same figure. The apparent greater lifetime of the first component (140 fs) may be explained by contributions of cluster fragment signals, which reach a maximum value after a few tens of fs delay as compared to the corresponding parent. The smaller weight of the second component of this signal resulting from the sum of the isolated monomer and fragmented hydrates suggests that the relative weight of the second component in the intrinsic dynamics of the hydrates is small. This observation is further supported by the experiment with the 800 nm probe using helium as the carrier gas. In this case, a strong deviation electric field, perpendicular to the molecular jet, has to be used in the mass spectrometer flight tube in order to detect essentially higher mass species. Because of the larger kinetic energy in helium than in argon, this effect is very selective. The temporal signal recorded in the 9Me-Ade⁺ mass channel is very similar to that of the clusters in terms of the relative component weights, which suggests in this case a negligible contribution of isolated parent monomers to the signal. Moreover, this signal is shifted by ∼20 fs to longer times, as compared to the clusters transients’ rise time (see insert in Fig. 3). This observation ensures that, because of the electric deviation in the TOF mass spectrometer chosen to preferentially detect clusters, the signal in the monomer mass channel recorded in the helium jet with the 267/800 nm pump–probe scheme is essentially due to fragmented ionised hydrates and contains the dynamical contributions of a whole range of cluster sizes.

Under these very specific experimental conditions, the weakness of the second component in the monomer channel ensures both the existence and the weakness of this component in the clusters. It demonstrates the intrinsic character of the cluster dynamics observed and its qualitatively different nature as compared to the isolated monomer. From the comparison of the temporal shapes of the fragment monomer and 9Me-Ade⁺–(H₂O)_n, it is clear that a significant change in terms of the relative component weights is operative as soon as one water molecule is bound to 9Me-Ade. The existence of only two preferential hydration sites on the 9-methylated molecule, i.e., with the water molecule bridging the amino group and either the N1 or the N7 atom,^47–51 therefore suggests that the effect of the most strongly bonded solvent molecule is similar in these two sites. The progressive weak change in the dynamics lifetime with the cluster size can be simply understood by a progressive solvation of the nucleobase from these sites.

The present cluster study can be compared to two previous gas phase works on hydrated adenine.^24,45 In contrast to the present results, both studies reported monoexponential decays. In the first study, the similar lifetimes found for all the clusters (∼220 fs) may be easily explained by the low temporal resolution (400 fs) employed.⁴⁵ The results of the second study (60–80 fs and 110 fs for 263 and 250 nm excitations, respectively) suffered from strong laser intensities and therefore from a very intense fragmentation of the ionised Ade–(H₂O)_n clusters. This caused a loss of the second component in the cluster mass channels,²⁴ which should have been found in the monomer channel but could not be distinguished from the intense second component of the unsolvated 9Me-Ade monomer, which was detected in this experiment (as in the present measurement in the argon jet).

Finally, the present results on the 9Me-Ade–(H₂O)_n clusters can be compared with lifetimes of 9Me-Ade-related molecules (Ado, dAdo, AMP, dAMP) measured in aqueous solution. Double exponential decays with a weak second component have been observed by fluorescence up-conversion²⁶ on dAdo and dAMP. These results (in particular the lifetimes of 100/420 fs and 100/520 fs, respectively), are in excellent agreement with our cluster measurements. Other studies on these systems, using both fluorescence up-conversion and transient absorption techniques, also reported sub-picosecond signals, although fitted to monoexponential decays with lifetimes ranging from 0.22 to 0.53 ps.^4,25,52 The differences between the fluorescence results could be related to differences in the experimental conditions (temporal resolution, sensitivity). The different results obtained by transient absorption cannot be easily reconciled with fluorescence measurement and the absence of a second component in transient absorption measurements is not understood at present. In spite of this difference, the comparison of the results from both solution studies and the present water cluster approach clearly show great similarities, namely similar sub-picosecond lifetimes for the radiationless relaxation pathways.

C. Comparison between experimental results and theoretical models

Several recent theoretical papers^{8,10–12,29–31} provide a convenient theoretical framework to interpret the ultrafast deactivation pathways of adenine.

The energetics of the excited states, and the crossings between them or with the ground state via conical intersections (CI), where radiationless transitions can take place, have been calculated as well as the various barriers along the deactivation pathways. On the basis of these results, a qualitative picture can be drawn (Fig. 4).

Fig. 4 Two-dimensional schematic representation of the multidimensional potential energy surface, featuring the excited states involved in the electronic relaxation of isolated 9Me-Ade and 9Me-Ade hydrates. In the isolated base (upper scheme), L_a excitation may populate the L_b and nπ* states. The L_b state relaxes mostly to a nπ* state through a small (<0.1 eV) barrier. The relaxation of the nπ* state (directly to the ground state or through the L_a state) involves higher barriers. In the hydrates (lower scheme), both L_b and L_a states may be excited. The opening of a direct relaxation path from L_a to the ground state explain the dominant ultrafast first component observed, while the remaining L_b to nπ* state path (now minor) can account for the very weak second component detected.

Vertical excitation of the isolated 9Me-Ade populates the lowest excited state ¹L_b far from its equilibrium region, causing the initial displacement of the wave packet towards the ¹L_b ππ* local minimum, labelled (L_b)_Min. Calculations show the presence of one or two paths for the decay of this (L_b)_Min relaxed geometry with a small or virtually inexistent barrier to the (nπ*)_Min (via the (L_b/nπ*)_X CI^8,10,29–31) or to the ¹L_a state (via the (L_a/L_b)_X CI³¹). The competition (branching) between these decay channels may depend on the excess excitation energy.^10,31 Upon excitation of the ¹L_a state, there is a barrierless pathway from the FC geometry to the CI of ¹L_a with the ground state, labelled (L_a)_X.³¹ The state can therefore either decay straight to the ground state, or induce population of the ¹L_b state and then of the nπ* state when passing through the corresponding (L_a/L_b)_X and (L_a/nπ*)_X CI, which are found to be peak-shaped.⁵³ In contrast, the decay paths of the (nπ*)_Min through the conical intersections (L_a)_X and (nπ*)_X with the ground state have sizable barriers (0.1 eV or more).³¹ The conical intersections involved in these pathways are characterised by strongly bent geometries, and are accessed through out-of-plane deformation modes of the six-membered^8,10,29–31 ring. The presence of specific relaxation channels involving πσ* states (repulsive along the N9–H bond, for example) has been repeatedly suggested in theoretical works. These new channels have, however, recently been calculated to open up at higher energy (230–220 nm),^10–12 due to the existence of a high barrier (∼1 eV). Calculations also suggest that, from the (nπ*)_Min, a ³(ππ*) triplet state can be populated by intersystem crossing. This channel has not been observed experimentally either in the isolated or in the water cluster conditions, which is in agreement with the very low intersystem crossing quantum yield measured for dAdo in aqueous solution (<10⁻³).⁵⁴

The mass-resolved and photoelectron experimental data on the photophysics of the canonical tautomer of adenine (or 9Me-Ade) support the theoretical models proposed (qualitative scheme Fig. 4). The excited 9Me-Ade molecule shows an ultrafast decay of the initially excited ππ* L_b state (lifetime τ₁ ∼ 100 fs) to the nπ* state. The involvement of out-of-plane vibrational modes was experimentally evidenced²³ by longer timescale dynamics (∼200/3000 fs) measured in dimethyladenine, related to the lower frequency of the amino group inversion upon methylation. The nπ* state then decays to the ground state with a τ₂ lifetime ranging from ∼10–50 ps (near threshold) to 1.25 ps (267 nm excitation).

The short progression of vibronic lines extending only over 800 cm⁻¹ in the gas phase UV spectrum together with the lifetime dependence on the excitation energy are in qualitative agreement with the existence of a low barrier, which is easily crossed when exciting at UV wavelength below 267 nm. The large signal observed for the nπ* state indicates that this is the dominant relaxation channel and that other decay channels play only a minor role. At high excitation energy (250 nm), the relative amplitude of the two components changes weakly, which might be an indication of the involvement of the L_a state in the dynamics.

In the case of the hydrates, our very detailed experimental results provide enough data to propose a new pathway for the relaxation of their excited states, based on theoretical predictions combined to general considerations about hydration effect (Fig. 4, lower scheme). The excited 9Me-Ade in a water cluster environment does indeed show a dramatically different temporal behaviour, in particular a very weak second nπ* component is observed that may indicate that a new relaxation channel is operating for the initially excited ππ* state. This behaviour is probably related to the reordering of the states, as expected for these molecules in polar solvents. The ¹L_a state is the most polar state among the lowest singlet states (dipole moment 4.7 D¹²). This state (and to a lesser extend ¹L_b) is thus expected to be stabilised (red-shifted) in polar solvents. The opposite is expected for the less polar (relative to the ground state) nπ* state, which will be blue-shifted. The destabilisation of the nπ* state may induce a barrier for the decay of (L_b)_Min to (nπ*)_Min, blocking this channel and leaving the direct decay of (L_b)_Min through (¹L_a)_X as the dominant process. On the other hand, the most stabilised ¹L_a state may be accessible at 267 nm excitation and may decay in a barrierless manner to the ground state via the (L_a)_X CI and/or populate the (L_b)_Min and (nπ*)_Min through the peaked conical intersections³¹ (L_a/L_b)_X and (L_a/nπ*)_X, respectively. When exciting at 267 nm, the biexponential transient signal of the hydrates may thus contain the contributions of the L_a and L_b excited states, which cannot be experimentally distinguished. Moreover, the weak nπ* amplitude could reflect a competition between two different decay channels for the L_b (or L_a) ππ* state, namely (i) formation of the nπ* intermediate or (ii) direct decay to the ground state.

Finally, the solvation stabilised πσ* state might also play a role in the dynamics of hydrated species, as suggested in a recent study on adenine–water complex dynamics, for excitation at 267 and 250 nm.²⁴ However, such a relaxation pathway is expected to be negligible in the hydrates dynamics, since no signal corresponding to (9Me-Ade–H) species (hydrated or not) could be detected in our mass-resolved experiments in the presence of water.

IV. Conclusion

The present dynamics investigation allows a more refined description of the photophysics of the 9Me-Ade in isolated and microhydrated environments near the region of fluorescent-to-dark state switching. The combined use of two complementary detection techniques, namely time-resolved ion mass spectrometry on both the parent and fragments, and time-resolved photoelectron velocity imaging, provides a unique tool to probe the detailed characteristics (nature, lifetime, etc.) of the excited states involved in the relaxation. Based upon our results and previously reported studies, the following model for internal conversion pathways in this purine base is proposed.

In the isolated molecule, excitation at 267 nm populates the bright ππ* L_b state. The present results indicate that this state decays rapidly (∼100 fs) to the nπ* state, which has a lifetime of about 1 ps. This is evidenced by (i) the specific photoelectron energy distributions of these states, and (ii) the different fragmentation efficiencies following their photoionisation, illustrating their different vibrational contents. Excitation at 250 nm leads to a similar temporal behaviour, suggesting an initial excitation of the L_a state followed by internal conversion to the L_b and nπ* states. Finally, following 267 nm excitation, no other channel involving, for instance, a πσ* state or a long-lived (triplet) state could be detected in the present photoelectron study.

In the microhydrated molecule, our measurements proved the existence of biexponential dynamics that was only observed thanks to carefully controlled experimental conditions and a specific ion collection optimised for fragment detection. The temporal behaviour of the hydrates is dominated by a very strong ultrafast (∼100 fs) first component, followed by a minor longer second component. This intrinsic dynamics of the hydrates is discussed based on the expected solvent effects. The interaction with water will stabilise the L_a and L_b ππ* states according to their polarisability, suggesting that both states are excited simultaneously at 267 nm in the hydrates. Finally, the investigation of the 9Me-Ade–(H₂O)_n clusters in a supersonic expansion combined with time-resolved mass spectrometry allows the documentation of the gradual effects of solvation on the photophysics of a biomolecule, with a sensitivity that may be better than that achieved in condensed phase studies.

The models for the deactivation mechanism in the pyrimidine nucleobases, uracil and cytosine, as well as their relevant derivatives and hydrates, are currently being investigated using the comprehensive time-, energy- and mass-resolved spectroscopic approach.

Acknowledgements

The authors are happy to thank Drs B. Soep and L. Poisson for making available to us their photoelectron velocity imaging spectrometer and their help in data processing, as well as O. Gobert, the late P. Meynadier, M. Perdrix, F. Lepetit, J. F. Hergott, and D. Garzella, who are responsible for developing, maintaining, and running the femtosecond laser facility LUCA (Laser Ultra-Court Accordable) of the CEA, DSM/DRECAM at Saclay.

References

1. R. V. Bensasson, E. J. Land and T. Truscott, Excited States and Free Radicals in Biology and Medicine, Oxford University Press, Oxford, 1993.
2. M. Daniels and W. Hauswirt, Science, 1971, 171, 675.
3. P. R. Callis, Annu. Rev. Phys. Chem., 1983, 34, 329.
4. C. E. Crespo-Hernandez, B. Cohen, P. M. Hare and B. Kohler, Chem. Rev., 2004, 104, 1977.
5. A. Broo and A. Holmen, J. Phys. Chem. A, 1997, 101, 3589.
6. M. P. Fulscher, L. SerranoAndres and B. O. Roos, J. Am. Chem. Soc., 1997, 119, 6168.
7. B. Mennucci, A. Toniolo and J. Tomasi, J. Phys. Chem. A, 2001, 105, 4749.
8. A. L. Sobolewski and W. Domcke, Eur. Phys. J. D, 2002, 20, 369.
9. S. K. Mishra, M. K. Shukla and P. C. Mishra, Spectrochim. Acta, Part A, 2000, 56, 1355.
10. C. M. Marian, J. Chem. Phys., 2005, 122, 104314.
11. S. Perun, A. L. Sobolewski and W. Domcke, J. Am. Chem. Soc., 2005, 127, 6257.
12. S. Perun, A. L. Sobolewski and W. Domcke, Chem. Phys., 2005, 313, 107.
13. N. J. Kim, G. Jeong, Y. S. Kim, J. Sung, S. K. Kim and Y. D. Park, J. Chem. Phys., 2000, 113, 10051.
14. D. C. Luhrs, J. Viallon and I. Fischer, Phys. Chem. Chem. Phys., 2001, 3, 1827.
15. C. Plutzer and K. Kleinermanns, Phys. Chem. Chem. Phys., 2002, 4, 4877.
16. E. Nir, C. Plutzer, K. Kleinermanns and M. de Vries, Eur. Phys. J. D, 2002, 20, 317.
17. N. J. Kim, H. Kang, Y. D. Park and S. K. Kim, Phys. Chem. Chem. Phys., 2004, 6, 2802.
18. I. Hunig, C. Plutzer, K. A. Seefeld, D. Lowenich, M. Nispel and K. Kleinermanns, ChemPhysChem, 2004, 5, 1427.
19. H. Kang, B. Jung and S. K. Kim, J. Chem. Phys., 2003, 118, 6717.
20. S. Ullrich, T. Schultz, M. Z. Zgierski and A. Stolow, J. Am. Chem. Soc., 2004, 126, 2262.
21. S. Ullrich, T. Schultz, M. Z. Zgierski and A. Stolow, Phys. Chem. Chem. Phys., 2004, 6, 2796.
22. E. Samoylova, H. Lippert, S. Ullrich, I. V. Hertel, W. Radloff and T. Schultz, J. Am. Chem. Soc., 2005, 127, 1782.
23. C. Canuel, M. Mons, F. Piuzzi, B. Tardivel, I. Dimicoli and M. Elhanine, J. Chem. Phys., 2005, 122, 074316.
24. H. H. Ritze, H. Lippert, E. Samoylova, V. R. Smith, I. V. Hertel, W. Radloff and T. Schultz, J. Chem. Phys., 2005, 122, 224320.
25. T. Pancur, N. K. Schwalb, F. Renth and F. Temps, Chem. Phys., 2005, 313, 199.
26. T. Gustavsson, A. Sharonov, D. Onidas and D. Markovitsi, Chem. Phys. Lett., 2002, 356, 49.
27. H. Kang, K. T. Lee and S. K. Kim, Chem. Phys. Lett., 2002, 359, 213.
28. A. Broo, J. Phys. Chem. A, 1998, 102, 526.
29. H. Chen and S. H. Li, J. Phys. Chem. A, 2005, 109, 8443.
30. S. Matsika, J. Phys. Chem. A, 2005, 109, 7538.
31. L. Blancafort, J. Am. Chem. Soc., 2006, 128, 210.
32. W. Fuss, W. E. Schmid and S. A. Trushin, J. Chem. Phys., 2000, 112, 8347.
33. W. Fuss, W. E. Schmid and S. A. Trushin, J. Am. Chem. Soc., 2001, 123, 7101.
34. A. Eppink and D. H. Parker, Rev. Sci. Instrum., 1997, 68, 3477.
35. E. Hansen and P.-L. Law, J. Opt. Soc. Am. A, 1985, 2, 510.
36. H. W. Jochims, M. Schwell, H. Baumgartel and S. Leach, Chem. Phys., 2005, 314, 263.
37. J. M. Mestdagh, J. P. Visticot, M. Elhanine and B. Soep, J. Chem. Phys., 2000, 113, 237.
38. P. Asselin, F. Piuzzi, J. Lecalve, M. Mons and I. Dimicoli, Chem. Phys., 1994, 181, 271.
39. C. A. de Lange, High-Resolution Laser Photoionisation and Photoelectron Studies, ed. I. Powis, T. Baer and C. Y. Ng, Wiley, New York, 1995, p. 195.
40. J. Berkowitz, Photoabsorption, Photoionization, and Photoelecton Spectroscopy, Academic, New York, 1979.
41. S. T. Pratt, P. M. Dehmer and J. L. Dehmer, J. Chem. Phys., 1984, 81, 3444.
42. J. Lin, C. Yu, S. Peng, I. Akiyama, K. Li, L. K. Lee and P. R. LeBreton, J. Am. Chem. Soc., 1980, 102, 4627.
43. A. Stolow, A. E. Bragg and D. M. Neumark, Chem. Rev., 2004, 104, 1719.
44. C. P. Schick, S. D. Carpenter and P. M. Weber, J. Phys. Chem. A, 1999, 103, 10470.
45. N. J. Kim, Y. S. Kim, G. Jeong, T. K. Ahn and S. K. Kim, Int. J. Mass Spectrom., 2002, 219, 11.
46. G. Gregoire, M. Mons, I. Dimicoli, C. Dedonder-Lardeux, C. Jouvet, S. Martrenchard and D. Solgadi, J. Chem. Phys., 2000, 112, 8794.
47. J. D. Gu and J. Leszczynski, J. Phys. Chem. A, 1999, 103, 2744.
48. H. T. Kim, J. Mol. Struct. (THEOCHEM), 2004, 673, 121.
49. A. K. Chandra, M. T. Nguyen, T. Uchimaru and T. Zeegers-Huyskens, J. Phys. Chem. A, 1999, 103, 8853.
50. O. S. Sukhanov, O. V. Shishkin, L. Gorb, Y. Podolyan and J. Leszczynski, J. Phys. Chem. B, 2003, 107, 2846.
51. M. Hanus, M. Kabelac, J. Rejnek, F. Ryjacek and P. Hobza, J. Phys. Chem. B, 2004, 108, 2087.
52. J. Peon and A. H. Zewail, Chem. Phys. Lett., 2001, 348, 255.
53. G. J. Atchity, S. S. Xantheas and K. Ruedenberg, J. Chem. Phys., 1991, 95, 1862.
54. J. Cadet, W. Roth and I. Fischer, Bioorganic Photochemistry, ed. H. Morrison, John Wiley & Sons, New York, 1990, vol. 1, p. 1.

---