The changes in the strength of the interaction between the polycyclic aromatic hydrocarbon, coronene, and graphite as a function of the degree of super-hydrogenation of the coronene molecule are investigated using temperature programmed desorption. A decrease in binding energy is observed for increasing degrees of super-hydrogenation, from 1.78 eV with no additional hydrogenation to 1.43 eV for the fully super-hydrogenated molecule. Density functional theory calculations using the optB88-vdW functional suggest that the decrease in binding energy is mostly due to an increased buckling of the molecule rather than the associated decrease in the number of π-electrons.
I. INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) are generally thought to be the dominant carriers of the unidentified/aromatic IR bands1–3 in the interstellar medium (ISM) and are expected to account for 5%-10% of the interstellar carbon abundance.4 PAHs play an important role in determining a range of physical parameters of the ISM, such as the ionization fraction and heating rates.5 Furthermore, PAHs may play a key role as catalysts for interstellar chemical complexity,6–13 as well as for the formation of larger carbon clusters14,15 and carbonaceous grains.16 Observations of photo-dissociation regions (PDRs) have shown a correlation between the abundance of PAHs and the formation rate of molecular hydrogen.6,17 Based on this it has been suggested that PAH molecules may act as efficient catalysts for H2 formation.6 This hypothesis has been supported by theoretical calculations7–9 and experimental measurements10–12 probing both neutral and cationic PAH species, reflecting that the PAH charge state of particular interstellar regions is observed to be dependent on the UV flux.18
Density functional theory (DFT) calculations have shown that neutral PAHs can catalyze the formation of H2 through the addition of excess H-atoms to the PAH molecule leading to the creation of super-hydrogenated PAH species.7 Subsequent impinging H-atoms can then abstract these excess H-atoms to form H2 via the Eley-Rideal abstraction mechanism. In low UV-flux regions, the presence of such super-hydrogenated PAH species has been shown to be in accord with IR observations.18,19 Neutral PAHs are expected to exist both in the gas phase20 and condensed on interstellar grain surfaces.21 Experiments investigating the hydrogenation of the PAH molecule coronene (C24H12) adsorbed on a graphite surface have confirmed that super-hydrogenated coronene acts as a catalyst for H2-formation11 and that completely super-hydrogenated coronene—perhydrocoronene (C24H36)—can be formed in this process.10 Recently, similar results have been demonstrated for the coronene cation in the gas phase.22
In this article, the effect of super-hydrogenation on the interaction between coronene and graphite is investigated experimentally through temperature programmed desorption (TPD) measurements of monolayer coronene films super-hydrogenated with an atomic deuterium beam. The coronene molecules in this experiment are thus in effect super-deuterated. The experimental results are compared with DFT calculations, using the optB88-vdW functional. This functional has been shown to describe well the interaction between the pristine coronene and the graphite surface yielding a monolayer structure in good agreement with LEED and STM studies and a binding energy in good agreement with TPD measurements.23 We present TPD measurements that reveal a significant decrease in binding energy with increasing super-deuteration, from 1.78 eV for coronene with no excess deuterium atoms (C24H12) to 1.43 eV for fully super-deuterated coronene (C12D36), where all available sites have been deuterated and the H-atoms present on the pristine coronene molecule have been replaced by D-atoms. The DFT calculations suggest that the decrease in binding energy is predominantly due to an increased buckling of the molecule rather than the decrease in the number of π-electrons resulting from the super-hydrogenation/deuteration.
II. METHODS
A. Experimental methods
The thermal desorption measurements were performed in an ultrahigh vacuum (UHV) chamber with a base pressure of 10−10 mbar. A highly oriented pyrolytic graphite (HOPG; SPI grade 1) substrate, mounted on a Ta sample holder, was used. The sample was heated through electron bombardment of the backside of the Ta sample holder. The temperature of the sample was measured with a C-type thermocouple pressed between the sample holder and the front face of the graphite sample. To obtain a clean and well-defined surface, the graphite sample was cleaved prior to mounting and annealed to 1100 K in UHV. Coronene (C24H12; Sigma-Aldrich, 99%) films were grown on the graphite sample at 290 K using a home-built Knudsen evaporation cell held at 438 K. A monolayer of coronene was prepared by exposing the sample to coronene for 60 s, growing a 2-3 layer film, and subsequently annealing the sample to 400 K to desorb the multilayers.23 The coronene monolayer was subsequently exposed to a D-atom beam produced by dissociating D2 (Air Liquide, >99.9%) with a hot capillary source24–26 operated at ∼2300 K. The D-atom flux was estimated to be 2.9 × 1014 atoms cm−2 s−1, based on operational parameters (capillary temperature, feed pressure, and distance to sample). D-atoms were used rather than H-atoms to allow us to confirm the exchange of the outer edge H-atoms with D-atoms. After exposing the coronene monolayer to atomic D, temperature programmed desorption (TPD) measurements were performed by heating the sample with a linear temperature ramp (β) of 1 K s−1, with desorbing molecules being detected with a quadrupole mass spectrometer (QMS; Extrel CMS LLC) scanning in the range from 290 to 365 amu e−1. It is assumed that only singly ionized ions are detected; therefore, amu is used hereafter, rather than amu e−1.
B. Theoretical methods
The density functional theory (DFT) calculations were carried out using the projector-augmented wave (PAW) method, as implemented in the real space grid-based method, GPAW.27 For the description of exchange-correlation effects, we used the optB88-vdW functional,28 since we have shown in a previous publication that this functional compares well with experimental data for the binding of coronene on graphite.23 From previous studies of the packing of coronene molecules on graphite, it has been determined that the unit cell for the monolayer structure is (√21 × √21) R ± 10.9°,23,29 which is also predicted by the optB88-vdW functional to be the most stable structure.23 We therefore also employed this structure in the present study. Two layers of graphene in AB stacking were used to model the graphite surface. 2D periodic boundary conditions were employed parallel to the surface, and a vacuum region of 6 Å separated the slab from the cell boundaries perpendicular to the surface. We used a grid spacing of 0.16 Å and a (2 × 2) k-point sampling. The structures were relaxed until the maximum force on every atom was below 0.01 eV/Å. The optB88-vdW optimized graphene lattice constant of 2.465 Å was used throughout the study. For the calculation of binding energies, zero-point energies were not taken into account since a previous study on a similar system, namely, the benzene crystal, found that this correction amounts to only 10–35 meV.30 The calculations were carried out by adding H-atoms, rather than D-atoms, to the coronene molecule.
III. RESULTS AND DISCUSSION
Figure 1 shows an intensity map, representing TPD traces measured for masses from 295 to 365 amu after a monolayer of coronene was exposed to varying 2300 K D-atom fluences. Figure 1(a) shows the desorption traces for a monolayer of coronene without any exposure to D. The dominant peak (58% of the signal) is observed at a peak desorption temperature (Tp) of 465 K and consists of the parent ion with a mass (m) of 300 amu. A significant contribution from other masses ranging from m = 295–302 amu is also observed. Here the lower mass contribution (295 amu: 1%, 296 amu: 3%, 297 amu: 2%, 298 amu: 9%, and 299 amu: 7%) is considered to be due to the fragmentation (H loss) of the molecule in the ionization source, while the higher mass contribution (301 amu: 14% and 302 amu: 1%) is due to the natural abundance of 13C. Two examples of desorption traces from a coronene monolayer exposed to 2300 K D-atoms are shown in Figures 1(b) and 1(c). In both cases higher mass species are observed due to the addition of D-atoms to the molecule as well as the substitution of H with D on the outer edge of the molecule, i.e., due to the super-hydrogenation/deuteration of the coronene molecule, as observed in earlier experiments.10 In Figure 1(b), desorption traces following exposure to a 2300 K D-atom fluence of 1.0 × 1017 atoms cm−2 show that the dominant peak is still located at 300 amu and Tp remains at 465 K, representing coronene molecules that have not been deuterated. However, higher mass species form a significant fraction of the desorbing molecules and masses as high as 350 amu are observed. The higher mass desorption peaks are observed to shift to a lower desorption temperature compared to the 300 amu peak, reaching a value of Tp = 350 K for 348 amu. In Figure 1(c) the desorption traces following exposure to a 2300 K D-atom fluence of 3.1 × 1018 atoms cm−2 are shown. Here the majority of the desorbing molecules have masses above 350 amu and masses up to 360 amu are observed corresponding to fully super-deuterated coronene (C24D36). The strongest peak is located at 358 amu for which Tp = 376 K.
Temperature programmed desorption traces for masses from 295 to 365 amu for a monolayer of coronene on HOPG exposed to 2300 K atomic D fluences of (a) no exposure to atomic D, (b) 1.0 × 1017 atoms cm−2, and (c) 3.1 × 1018 atoms cm−2.
Temperature programmed desorption traces for masses from 295 to 365 amu for a monolayer of coronene on HOPG exposed to 2300 K atomic D fluences of (a) no exposure to atomic D, (b) 1.0 × 1017 atoms cm−2, and (c) 3.1 × 1018 atoms cm−2.
In Figure 2 a selection of desorption traces for coronene monolayers exposed to two different 2300 K D fluences are shown, together with the 300 amu trace from the pristine coronene monolayer plotted in grey. In Figure 2(a) the three largest even mass desorption traces for a short D dose with a fluence of 1.7 × 1016 atoms cm−2 are displayed. The blue curve is the desorption trace for 300 amu and the peak desorption temperature is observed to be unchanged at 465 K. The green and red curves are for 302 amu and 304 amu, respectively, corresponding to the singly and doubly super-deuterated coronene molecules. A significant decrease in the peak desorption temperature is observed with Tp = 448 K for 302 amu and Tp = 432 K for 304 amu. In Figure 2(b) the three largest even mass desorption traces for a longer D dose with a fluence of 3.1 × 1018 atoms cm−2 are illustrated. The blue, green, and red curves are for 356 amu, 358 amu, and 360 amu, respectively, and, as mentioned earlier, all show a significantly lower peak desorption temperature compared to the 300 amu peak. Furthermore, even the largest peaks for the long D dose are more than 20 times less intense than the 300 amu peak from the pristine coronene film. This is mainly due to a broader distribution of molecules, but can also partly be attributed to a loss of molecules during deuteration. We have previously reported a loss of up to 75% of molecules during super-deuteration31 as a result of coronene desorption induced by the D addition. Similar results have been observed for benzene on graphite, where only about 10% of the molecules initially adsorbed on the surface were detected with TPD after deuteration.32
The three TPD traces for the most significant equal mass desorption traces are plotted together with the TPD trace for the parent coronene ion for a monolayer of pristine coronene (grey). (a) TPD traces for a monolayer of coronene exposed to a D fluence of 1.7 × 1016 atoms cm−2 for 300 amu (blue), 302 amu (green), and 304 amu (red). (b) TPD traces for a monolayer of coronene exposed to a D fluence of 3.1 × 1018 atoms cm−2 for 356 amu (blue), 358 amu (green), and 360 amu (red).
The three TPD traces for the most significant equal mass desorption traces are plotted together with the TPD trace for the parent coronene ion for a monolayer of pristine coronene (grey). (a) TPD traces for a monolayer of coronene exposed to a D fluence of 1.7 × 1016 atoms cm−2 for 300 amu (blue), 302 amu (green), and 304 amu (red). (b) TPD traces for a monolayer of coronene exposed to a D fluence of 3.1 × 1018 atoms cm−2 for 356 amu (blue), 358 amu (green), and 360 amu (red).
The peak desorption temperature for the observed desorption peaks has been determined for monolayers of coronene exposed to 16 different fluences of 2300 K D-atoms ranging from 0 to 3.1 × 1018 atoms cm−2. Figure 3 shows the peak desorption temperature as a function of mass. These data allow the extraction of the strength of the binding between the deuterated coronene and the graphite for the whole mass range: it has previously been reported that a coronene monolayer follows a first order desorption process as described by the Polanyi-Wigner equation.23 Assuming that this extends to the deuterated coronene species, the barrier for desorption (Edes) can be determined through the Redhead equation,33
where ν is a pre-exponential factor. The pre-exponential factor is often thought of as the attempt frequency for desorption and typically values of the order of 1013 s−1 are used. This model is, however, only appropriate for small molecules, while for larger molecules, entropic effects play an important role. The pre-exponential factor for coronene on graphite has previously been estimated for the coronene monolayer through transition state theory to lie between 8.9 × 1017 s−1 and 4.8 × 1018 s−1.23 Under the assumption that this does not change significantly for the deuterated coronene species, Edes is here determined for the desorbing molecules over the entire mass range using an average value of ν = 2.9 × 1018 s−1.23 In Figure 4 Edes is determined from the average Tp for each mass and plotted relative to mass. The error bars are based on the uncertainty in the pre-exponential factor, as described above. Initially Edes decreases rapidly with increasing mass from 1.78 eV for non-deuterated coronene to 1.55 eV at 318 amu. After this Edes decreases more slowly to a minimum of 1.34 eV at 347 amu and then increases to 1.43 eV at 360 amu. The general decrease in Edes is expected due to the reduction of the number of π-electrons due to the conversion of the carbon hybridization from sp2 to sp3. A further reduction can be expected due to buckling of the molecule in the case of single sided super-deuteration as predicted by DFT calculations.10 The possible origin of the observed increased binding energy for very high masses is discussed below.
Desorption peak temperature (Tp) determined for the most significant desorption peaks for monolayers of coronene exposed to 16 different fluences of 2300 K atomic D ranging from 0 atoms cm−2 to 3.1 × 1018 atoms cm−2. Tp is plotted relative to the molecular mass of the desorbing species.
Desorption peak temperature (Tp) determined for the most significant desorption peaks for monolayers of coronene exposed to 16 different fluences of 2300 K atomic D ranging from 0 atoms cm−2 to 3.1 × 1018 atoms cm−2. Tp is plotted relative to the molecular mass of the desorbing species.
The energy of desorption (Edes) determined from the most significant desorption peaks for super-deuterated coronene plotted relative to the molecular mass of the desorbing species. Edes was determined from desorption traces for monolayers of coronene exposed to 16 different fluences of 2300 K atomic D.
The energy of desorption (Edes) determined from the most significant desorption peaks for super-deuterated coronene plotted relative to the molecular mass of the desorbing species. Edes was determined from desorption traces for monolayers of coronene exposed to 16 different fluences of 2300 K atomic D.
Figure 5 displays the corresponding theoretical binding energies from DFT calculations as a function of the hydrogenation state, i.e., the total number of additional H-atoms attached to the coronene molecule. Each hydrogenation state can be realized in different configurations, which all contain the same number of H-atoms, but differ from each other with respect to the positions of the H-atoms, i.e., to which C-atoms and from which side of the molecule the H-atoms are attached. The coronene binding energy has been determined for 25 different configurations covering 10 different hydrogenation states; thus, each hydrogenation state can contain several different binding energies depending on the realized configuration. The grey area marks the boundaries of the calculations. The details for all of the individual configurations are shown in the supplementary material. The highest binding energies come from those molecules that experience the least amount of buckling at high coverage and limited hydrogenation of the substrate-facing side of the molecule at low coverage, thus allowing a more optimal graphite–molecular carbon skeleton distance. The superhydrogenated coronene configurations for these calculations are shown in the top of Figure 5. For low degrees of hydrogenation, these configurations generally involve only edge hydrogenation. For higher hydrogenation states, hydrogenation of both sides of the molecule results in a high binding energy. For 24 additional H-atoms, this corresponds to fully trans-hydrogenated coronene—perhydrocoronene (C24H36). The lowest binding energies come from those molecules that experience the largest amount of buckling at high coverage and hydrogenation of the bottom side of the molecule at low coverage, in all cases resulting in increased graphite–molecular carbon skeleton distances. The configurations for these calculations are shown in the bottom of Figure 5. For up to 4 additional H-atoms, the lowest binding energy is achieved through hydrogenation of center sites on the bottom side of the molecules. For the higher hydrogenation states, hydrogenation of only one side of the molecule leads to the lowest binding energy, which for 24 additional H-atoms corresponds to fully cis-hydrogenated coronene. The calculations show that the reduction in the number of π-electrons in the molecule caused by conversion of the carbon hybridization from sp2 to sp3 only results in a small decrease in binding energy, since the fully trans-hydrogenated coronene has a binding energy of 1.80 eV, compared to 1.92 eV for pristine coronene. Molecular buckling or hydrogenation of the substrate-facing side of the molecule, on the other hand, results in a much larger decrease, i.e., the fully cis-hydrogenated coronene has a binding energy of only 1.09 eV.
Binding energies calculated with DFT are plotted relative to the number of additional H-atoms attached to the coronene molecule (black dots). The binding energy has been determined for 22 different configurations covering 10 different super-hydrogenation states. The grey area drawn in the plot covers the boundaries of the calculation. The configuration leading to the higher and lower boundary of the binding energies is shown above and below the plot. The details for all of the individual configurations are shown in the supplementary material.
Binding energies calculated with DFT are plotted relative to the number of additional H-atoms attached to the coronene molecule (black dots). The binding energy has been determined for 22 different configurations covering 10 different super-hydrogenation states. The grey area drawn in the plot covers the boundaries of the calculation. The configuration leading to the higher and lower boundary of the binding energies is shown above and below the plot. The details for all of the individual configurations are shown in the supplementary material.
When comparing the experimentally obtained binding energies to the DFT calculations as in Figure 6, several issues arise. Although the optB88-vdW functional in previous studies was shown to yield the best agreement in binding energy with TPD experiments on pristine coronene,23 the DFT calculations yielded a binding energy which was 0.14 eV higher than the experimental value. Although this is generally an acceptable error for a DFT vdW functional, it nevertheless complicates the detailed comparison with experiments that we are aiming for here. In lack of any knowledge about how this error varies with the degree of super-hydrogenation, we chose to subtract 0.14 eV from all DFT calculated binding energies. Furthermore, due to the exchange of H to D, the predicted energies for a given super-hydrogenation state can cover up to 13 amu since the amount of H exchanged to D is unknown. In other words, the binding energy of a given hydrogenation state n can apply to all the molecules described by the formula C24H12−iDn+i, where i = 0–12 and describes the number of H exchanged with D. Each DFT calculated point relating a binding energy to a hydrogenation state n is used to describe the masses of the molecular species covered by the range i = 0–12 in the above formula. The application of these two corrections gives rise to the light grey area plotted in Figure 6 as the possible energies predicted by the DFT calculations. The experimentally determined binding energies have been plotted in Figure 6 as black dots. A loss of H and D in the ionization source of the QMS is also expected, just as with pristine coronene. The degree to which this occurs is not known as we do not have the fragmentation patterns for the deuterated species, but this will add an additional small error to the measured mass for each binding energy point. We note that the decrease in energy observed from 301 to 310 amu lies below what can be achieved through hydrogenation of only the top-side of the molecule (the side facing away from the substrate), even for the configurations yielding the maximum amount of buckling, i.e., center side addition. For these masses the measured binding energies lie between those found for top-side hydrogenation of the molecule and those for center site hydrogenation of the substrate-facing side of the molecule. This is, however, in contradiction with the low barrier hydrogen addition route predicted by previous DFT calculations.7 However, the 2300 K D-atoms have sufficient energy to deuterate the graphite, and experiments have shown that coronene molecules are able to abstract D-atoms adsorbed on the graphite surface, which would be expected to lead to super-deuteration on the substrate-facing side of the molecule.34 Hence the additions might rely on surface abstraction instead of gas phase impingement. For higher masses the observed binding energies lie between the values found for pure cis- or trans-hydrogenation, indicating that a combination of single sided and double sided super-deuteration is taking place. The increase in binding energy from 347 amu to 360 amu could suggest a decrease in the buckling of the molecule occurring through the final deuterium addition reactions.
Comparison between experimental and theoretical values. The light grey area shows the area within which the DFT calculations predict that the binding energies should fall, after subtracting 0.14 eV and taking all molecular species for each super-hydrogenation/deuteration state into account. The experimentally found values are plotted as black dots.
Comparison between experimental and theoretical values. The light grey area shows the area within which the DFT calculations predict that the binding energies should fall, after subtracting 0.14 eV and taking all molecular species for each super-hydrogenation/deuteration state into account. The experimentally found values are plotted as black dots.
IV. SUMMARY
In summary, the change in binding energy between coronene and graphite was investigated experimentally as a function of the degree of super-deuteration using temperature programmed desorption of super-deuterated coronene monolayers. A decrease in Edes was observed from 1.78 eV at 300 amu, corresponding to non-deuterated coronene, to 1.34 eV at 348 amu. A subsequent increase was then observed to 1.43 eV at 360 amu (fully super-deuterated coronene). DFT calculations of the binding energy were performed using the optB88-vdW functional on 25 different configurations covering 10 different super-hydrogenation states. Since several configurations can exist for each super-hydrogenation state, calculations reveal several binding energies for each super-hydrogenation state. The highest binding energies tend to be the ones that promote the smallest amount of buckling of the molecule, while the lowest binding energies are the ones that promote the largest amount of buckling. Comparison of the binding energies found experimentally with those found with DFT indicates that the coronene molecules are not purely cis- or trans-superdeuterated, but rather show mixed configurations with binding energies in between these two extremes. The increase in binding energy from 347 amu to 360 amu, however, suggests that a decrease in the buckling of the molecule occurs through the final addition reactions, possibly as a result of surface deuterium abstraction, which would result in a conversion towards a more trans-superdeuterated structure.
SUPPLEMENTARY MATERIAL
See supplementary material for details about all of the individual configurations used for the DFT calculations.
Acknowledgments
We acknowledge financial support from the European Research Council under ERC starting Grant “HPAH,” No. 208344, and the European Commission’s seventh Framework Programme through the “LASSIE” ITN under Grant Agreement No. 238258. We also acknowledge support from the Danish Research Councils and Danish Center for Scientific Computing.