PHOTON-INDUCED FORMATION OF MOLECULAR HYDROGEN FROM A NEUTRAL POLYCYCLIC AROMATIC HYDROCARBON: 9,10-DIHYDROANTHRACENE

Traditional cryogenic matrix isolation techniques were applied for trapping of DHA gas-phase molecules into solid Ar at 9–12K, deposited on an infrared transparent CsI cryostat window (Szczepanski et al. 1993; Espinoza et al. 2010). DHA purity of 97%+, anthracene purity of 99.9%+, both from Sigma-Aldrich, were used as supplied and Ar gas purity was 99.999%. After deposition of the DHA/Ar mixture, the matrix was subjected to 4–35 hr of irradiation time, using UV–visible photons from a 100 W medium pressure Hg lamp with energies of E ⩽ 5.5 eV and estimated spectral irradiance of 4.4 × 10¹⁴ photons cm⁻² nm⁻¹ s⁻¹ at 253.6 nm (at the matrix surface). Several long-wavelength pass filters (Veb Jena) were employed to establish the photon energy threshold for observing photofragment reactions in the DHA/Ar matrix. Infrared absorption spectra of the DHA/Ar matrix prior to and after photolysis were acquired using a Fourier transform infrared (FT-IR) absorption spectrometer (Magna 560 by NICOLET; and averaging of 1000 scans at 1 cm⁻¹ resolution).

A residual gas analyzer (RGA, by Extorr, Inc.) was connected to the cryostat vacuum chamber pumped by a 1100 l s⁻¹ (for N₂) turbo molecular pump (by Osaka) to monitor the increase in partial pressure of H₂ during matrix annealing (up to 35 K). In this experiment the photon-induced reaction product H₂, which was stored in the photolyzed matrix during several hours of UV–visible irradiation, was extracted from the matrix upon annealing. While the background partial pressure of H₂ was 4 × 10⁻¹⁰ Torr, the additional H₂ partial pressure due to release from the matrix was found to be 2–5 × 10⁻¹⁰ Torr in the 16–34 K temperature range of the matrix.

Molecular equilibrium geometries and their associated vibrational harmonic mode frequencies for DHA and its fragments were calculated using density functional theory with the Gaussian 09 program package (Frisch et al. 2009). Becke's three-parameter hybrid functional, the non-local correction functional of Lee, Yang, and Parr (B3LYP), with a 6–31 G basis set and d and p polarization functions on heavy and hydrogen atoms, respectively, were employed. The harmonic vibrational frequencies in the mid-IR range were scaled by a factor of 0.9614 and for CH stretching modes by 0.955 to correct for anharmonicity effects and functional/basis set deficiencies (Jolibois et al. 2005).

In the potential energy surface (PES) calculation for single hydrogen ejection, all degrees of freedom were optimized except for the R(CH) bond length, which was increased stepwise. Electronic energies were not zero-point energy (ZPE)-corrected in this case. For the H₂ ejection reactions, transition states (TSs) connecting the stable minima on the PESs were sought by applying the QST3 optimization procedure in Gaussian 09 (Frisch et al. 2009) at the B3LYP/6–31G(d,p) functional/basis level. The harmonic vibrational frequencies were calculated for each structure involved in the PES calculation in order to verify whether a first-order TS was present. Electronic energies for each structure (including the TS) of lowest-energy fragmentation reaction pathway were recalculated at the G3//B3LYP level (Frisch et al. 2009; Baboul et al. 1999) in order to improve the accuracy of the calculation (Curtiss et al. 1998). The G3 energy calculation protocol was modified in such a way that the geometries and ZPEs were applied from B3LYP/6–31G(d,p), rather than the geometries from second-order perturbation theory MP2(FULL)/6–31G(d) and zero-point energies from Hartree–Fock theory of HF/6–31G(d) (Baboul et al. 1999). Note that ZPE corrections for G3//B3LYP/6–31G(d,p) were scaled (as required by the G3 protocol) by the 0.9614 scaling factor adopted from the work of Jolibois et al. (2005),while energies at the B3LYP/6–31G(d,p) level were ZPE-corrected without scaling.

The mid-IR range of the vibrational absorption spectra of neutral DHA, isolated in an Ar matrix at 12 K, are displayed in Figure 1. The green spectrum was recorded before photolysis, while the blue and red spectra were scanned after 4 and 21 hr of irradiation with a UV–visible lamp, respectively. Clearly, in the blue and red spectra several new bands are apparent. The top spectrum represents the infrared absorption spectrum for anthracene trapped in an argon matrix, which was recorded in a separate experiment. A comparison to the top spectrum gives compelling evidence that many of the photo-induced IR bands are due to anthracene. The bands at 878.3, 883.5 (shoulder), 1001.1, 1317.9, and 1450.4 cm⁻¹, which are not overlapped with DHA bands, are marked by vertical dashed lines. The anthracene bands that overlap with existing DHA bands are indicated by open circles. The 725.8, 727.6, and 729.2 cm⁻¹ bands partially overlap with a low-energy shoulder of the 731 cm⁻¹ DHA band, but can be clearly distinguished (not shown for the sake of clarity). While the general agreement of the photofragment bands with the anthracene bands is credible in terms of position and intensity, there are a number of weaker photofragment bands that are left unassigned, such as 853.9 and 870.1 cm⁻¹ (marked by dots). So far, no candidate structures have been found to rationalize these bands, including photofragments associated with DHA ring opening reactions. We note that for the latter bands, the rate of appearance is markedly different to that of anthracene. In contrast to the anthracene bands, they grew fast in the initial 4 hr of photolysis, but did not increase further during the next 17 hr of irradiation (see Figure 1). One possibility is that the unknown carrier of those two bands is a photoproduct from an impurity in the sample (∼3%, as disclosed by the supplier).

Zoom In Zoom Out Reset image size

Control photolysis experiments employing a long-wavelength pass filter showed that no photofragments were observed for photon energies of E ⩽ 4.13 eV. In comparison, the lowest-energy electronic transition for DHA is located at 4.61 eV or 37219 ± 1 cm⁻¹, as reported from gas-phase jet experiments (Chakraborty & Chowdhury 1990). Time-dependent density functional theory calculations of vertical electronic transitions at the B3LYP/6–311++G(d,p) level suggest a dense manifold of such states, where the first state is predicted at E(S₁) = 4.595 eV and the four excited states S₁–S₄ are distributed over merely 0.48 eV energy, and having oscillator strengths of 0.0027, 0.0235, 0.0200, and 0.0087, respectively. This suggests that in these DHA photolysis experiments, with photon energies of E ⩽ 5.5 eV, several electronic excited states of DHA are likely accessible. Since the fluorescence in DHA is very weak (Chakraborty & Chowdhury 1990) the excitation energy can be deactivated in part by internal conversion to highly excited vibrational states of the electronic ground state S₀. Highly vibrationally excited molecules can fragment, provided that the dissociation rate is sufficiently high compared to the cooling rate in the matrix. Another possibility is that one (or more) electronic excited state(s) is a predissociation state leading to H₂ abstraction. The PESs of ground-state fragmentation pathways will be discussed in detail below Section 3.4.

The column density of anthracene photoproducts after 21 hr UV photolysis (Figure 1) can be estimated as follows. A matrix thickness of l = 80 μm after 1 hr deposition was estimated from an interference method (Szczepanski et al. 1993). The Ar column density in an l = 80 μm matrix was approximated at N(Ar) = 3.52 × 10⁻⁵ mol cm⁻², based on a density of 1.77 g cm⁻³ for solid Ar at 5 K (Dobbs & Jones 1957). The column density of DHA molecules for the unphotolyzed DHA/Ar matrix (Figure 1, green trace) was estimated from the formula

$$\begin{equation} N = A^{ - 1} \int \!\tau \;d\nu, \end{equation} \tag{ 1 }$$

where A is the calculated integral intensity IR band value, τ is the spectral absorbance of the experimental band, and ν is the frequency of the same band (Hudgins et al. 1994; Roser & Allamandola 2010). Using the well isolated 1208 cm⁻¹ band in this DHA spectrum, for which the experimental integral absorbance is 2.38 × 10⁻² cm⁻¹ and the calculated integral intensity is 15.9 km mol⁻¹ (at B3LYP/6–311++G(d,p) level), the column density of DHA molecules in this matrix is approximated at N(DHA) = 1.46 × 10⁻⁸ mol cm⁻². This equates to a ratio of N(Ar)/N(DHA) = 2.4 × 10³, insuring a reasonable isolation of DHA molecules in the Ar matrix.

Applying Equation (1) for the photoproduct anthracene doublet band at 878.3 and 883.5 cm⁻¹ for the photolyzed matrix at 21 hr (Figure 1, red trace), the column density of the anthracene photoproduct is calculated as N(anthracene) = 5.2 × 10⁻⁹ mol cm⁻². Note that this result is based on an experimental total integral intensity of 3.41 × 10⁻² cm⁻¹ and a theoretical integral intensity of 66 km mol⁻¹ (B3LYP/6–311++G(d,p)). In contrast, the depletion in column density of DHA during 21 hr of irradiation is estimated at 5.8 × 10⁻⁹ mol cm⁻², based on ∼40% decrease. As the DHA depletion is larger by ∼10% than the column density of the anthracene photoproduct (i.e., 5.2 × 10⁻⁹ mol cm⁻²), this suggests that 90% of depleted DHA molecules are photo-converted into anthracene. This is also consistent with additional minor photofragments that cannot be elucidated at this time (see unassigned bands in Figure 1).

An extrapolation of these laboratory photolysis results to astrochemical conditions in dark clouds suggests that the corresponding timescale for 40% photodepletion of DHA would be ∼770 million years, based on an estimated photon flux of 10³ photons cm⁻² s⁻¹ compared to the mercury lamp experiments (i.e., 21 hr at 4.4 × 10¹⁴ photons cm⁻² s⁻¹).

The infrared absorption measurements above confirmed that the major photoproduct of DHA is anthracene, which suggests a concomitant loss of molecular hydrogen, H₂. In order to confirm formation of molecular hydrogen in the photolysis experiment, a residual gas analyzer (RGA) was used to monitor the partial pressure of H₂ (m/z 2) upon annealing the matrix. The basic premise of this experiment rests on the assumption that photo-generated H₂ can be stored in the matrix for hours, prior to being released upon heating the matrix and being detected by the RGA.

Figure 2 shows the RGA H₂ partial pressures as a function of matrix temperature for photolyzed DHA/Ar and a control experiment of photolyzed Ar matrix. Since molecular hydrogen was present in the vacuum chamber at a background partial pressure of 4 × 10⁻¹⁰ Torr, this was not a background-free experiment, requiring care in establishing the same conditions for both experiments. To eliminate the production of H₂ molecules in pyrolysis of DHA on hot tungsten filaments, the vacuum gauges and RGA were off during DHA deposition into solid Ar (2 hr) and during photolysis (3 hr).

Zoom In Zoom Out Reset image size

The experimental results in Figure 2 show a large pressure increase in the range 9–16 K with a maximum centered at T = 13 K, and a second, much smaller peak in the 20–34 K range (bottom). The inset (blue plot) shows the corresponding data for another control experiment, where H₂ was introduced into the vacuum chamber as an H₂/Ar mixture (0.1%) to form a solid film of H₂ and Ar. In this control experiment, the bimodal nature of H₂ evaporation can be seen much more clearly. The onset temperatures for the two peaks correlate with the melting point of H₂ at 14 K and its boiling point at ∼20 K. The lower temperature peak at 13 K could thus be interpreted as the process of releasing H₂ to the vacuum chamber from the surface of the matrix and its cryostat window holder during melting of H₂, whereas the second broader peak from 20 to 34 K is likely ascribed to the slower process of leaking H₂ imbedded in the solid Ar matrix by diffusion (Szczepanski et al. 1992). Crucially, the second peak is only observed in the case of the DHA/Ar matrix (red), and not in the Ar-only control experiment (black). The increased H₂ partial pressure for DHA/Ar is thus consistent with the formation of molecular hydrogen during photolysis, even if the background level of H₂ complicates analysis.

Experimental X-ray crystallography found that the electronic ground state of DHA is non-planar with a folding angle of 145° along the C9–C10 axis (Ferrier & Iball 1954). The corresponding calculated equilibrium geometry at the B3LYP/6–31G(d,p) level indicates that DHA will adopt a C_2v symmetry with a folding angle of 142.9°. On the other hand, use of a larger basis set of 6–311++G(d,p) predicted a folding angle of 141.9° for the same point symmetry group. All of the PESs presented here were calculated first using the B3LYP/6–31G(d,p) level.

PES pathways for reactions of molecular hydrogen removal were explored for two significantly different cases. In the first case, both H's are supplied from the C9 (or equivalent C10) sp³ carbon (shown in Figure 3(a)), while in the second case, one H is supplied by C9 and another from the C10 carbon (displayed in Figure 3(b)). Figure 3(a) indicates that during scissoring vibrations at 1484 cm⁻¹ (unscaled, B3LYP/6–311++G(d,p); see Figure 4 for this normal mode) with large motion amplitude, the two H atoms on C9 sp³ carbon would interact to form a stable complex between H₂ and the anthracene-like structure B. To form this complex, DHA needs to overcome an energy barrier of 3.812 eV, as is indicated by the predicted transition state TS1, which connects precursor DHA (structure A) and the complex B. Structure B is a weakly bound complex, with a binding energy of merely 0.03 eV compared to the dissociated structure C. Due to the very flat nature of the PES for H₂ removal, attempts to locate a TS between B and C were unsuccessful, and this fact is indicated by a dashed line in Figure 3(a).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Structure C exhibits a "naked" C9 carbon, which is a highly reactive moiety. Structure C was not confirmed in the infrared absorption experiment, and may thus interconvert to another structure. If structure C has enough excess excitation energy available, the migration of one H from a CH aliphatic group around one of two external rings is possible. The PES of hydrogen migration from a C10 hydrogen, ultimately resulting in anthracene (structure H) and dissociated H₂ is shown in the remainder of Figure 3(a). For each step of the H "walk" around the ring, the energy barrier was calculated, as indicated by the TS2–TS6 transition states with relative energies versus the electronic energy of DHA. The highest barrier (TS2 = 5.692 eV) exceeds the photon energy of the mercury lamp. This suggests that the whole pathway would require absorption of multiple UV photons in a sequential way, thus representing an improbable route to generating abundant anthracene and H₂.

As a side note, the C–G molecular systems in Figure 3(a) exhibit one "naked carbon", which is analogous to recently reported results for the singly dehydrogenated naphthalene cation (Galue & Oomens 2011). Their IR vibrational spectra pointed out the triplet state multiplicity for the electronic ground state of the cation. This is also the case for structures C–G and their connecting transition states of TS3 and TS4 in Figure 3(a), as summarized in Table 1. Note that this is not reflected in the PES in Figure 3(a), which shows the PES for the singlet state throughout.

Figure 3(b) shows the PES of ejecting H9 and H10 from DHA in the form of molecular hydrogen. In contrast to the extended pathway in Figure 3(a), a single transition state, TS, connects precursor A and the anthracene–H₂ complex, structure B. Moreover, the low energetic barrier of this process of only 2.272 eV, or 2.304 eV when applying the G3//B3LYP/6–31G(d,p) level, makes this a highly credible mechanism for a one-photon photolysis reaction. The electronic energy of the dissociated products, structure C + H₂, is barely higher (i.e., 0.002 eV at G3), suggesting facile separation of the fragments. In Figure 4, the displacement vectors for the imaginary frequency of −1412 cm⁻¹ in the TS of Figure 3(b) are shown. It is interesting to note that a combination of two normal mode frequencies, 970 and 2981 cm⁻¹, yields similar displacement vectors to the reactive coordinate of this dissociation pathway. Moreover, the displacement vectors of the scissoring mode at 1484 cm⁻¹ rationalize an interaction of H9 and H9', in analogy with the H₂ loss mechanism in Figure 3(a).

Due to the symmetry of DHA, the PES pathways for a single H removal for four chemically distinct sites of DHA are considered in Figure 5. In two cases, H9 and H9' (or equivalently C10 and C10'), H's are bonded to an sp³ carbon. In the third and fourth cases, H1 and H2 are bound to sp² carbons. For each step of the PES, the respective C–H bond length was increased by 0.1 Å, optimizing all other degrees of freedom of the molecule. For H9 (blue) and H9' (red), their PESs converge after five steps. This happens because at step 5 the DHA skeleton flips along the C9–C10 axis during optimization, resulting in a ring inversion to convert H9' into H9. The relative energy keeps on increasing upon stretching the C–H bond further until ∼2.3 Å, when a dramatic drop in energy of 2.65 eV is observed. This correlates with H9 "grabbing" H10 to form a stable, but weakly bound complex between molecular H₂ and anthracene. The energy barrier for H₂ removal in this reaction is estimated at 3.424 eV, which is compatible with a one-photon absorption process.

Zoom In Zoom Out Reset image size

In hydrogen removal reactions from sp² carbons, the barriers are significantly higher (>6 eV), and thus beyond the energy available from a single photon in this experiment. For the removal of H1 (green), a similar drop in energy is predicted due to H1 "grabbing" the neighboring H9 to form H₂. Conversely, the removal of H₂ is predicted to proceed via a barrierless reaction, requiring more than 6.9 eV energy.