Local formation of HArF in solid argon: Low-temperature limit and thermal activation

The H+Ar+F reaction leading to HArF formation in an argon matrix is studied at temperatures down to 8 K. The effects of the precursor concentration, deuteration, IR light, and deposition temperature as well as thermal activation of this reaction are studied. It is found that HArF molecules are formed slowly but efﬁciently at 8 K in a photolyzed HF / Ar matrix, supporting the previously reported results. The formation rate of HArF (cid:1) and DArF (cid:2) exhibits a low-temperature limit and enhances at elevated temperatures with activation energy of about 40 meV. All the data show that HArF is formed as a result of a local reaction of hydrogen atoms with the parent Ar–F centers and the tunneling mechanism is very probable here. The locality of the precursor photolysis required for this tunneling reaction is consistent with the partial HArF formation observed during photolysis of HF in an argon matrix. The decay mechanism of (cid:1) ArHAr (cid:2) + cations is also studied. The present results conﬁrm the previous conclusions that the decay of the cations is not essential to HArF formation. © 2010


I. INTRODUCTION
Noble-gas hydrides with the general formula HNgY ͑H = hydrogen atom, Ng= noble-gas atom, and Y = electronegative fragment͒ have been studied both experimentally and computationally. 1-5 These molecules have an ͑HNg͒ + Y − charge-transfer character leading to strong absorption of the H-Ng stretching mode. They can be prepared by photolysis of a hydrogen-containing precursor HY in a noble-gas matrix followed by thermal annealing promoting the H + Ng+ Y reaction of the neutral fragments. 6,7 Some of these molecules ͑HArF, HXeNCO, and HKrCl͒ are observed in relatively small amounts during UV photolysis as an intermediate species indicating locality of solid-state photodissociation. [8][9][10] HArF is a ground-state neutral molecule of argon. 9,11 It is synthesized experimentally using photolysis and annealing of HF in solid argon. HArF is an example of a noble-gas hydride forming during photolysis, hence indicating the locality of HF photolysis in solid argon. After thermal annealing, HArF can occupy two different matrix configurations referred as unstable HArF and stable HArF. 11,12 The unstable HArF configuration is formed in the largest amounts at about 20 K. Annealing above about 28 K decomposes unstable HArF leaving only the stable configuration visible.
The HArF molecule has been widely studied by theoretical methods. For example, the structure, potential energy surface, and vibrational properties have been examined. [13][14][15][16][17] A number of studies have simulated the trapping configurations of HArF in an argon matrix. [18][19][20] Recently Bochenkova et al. have reported theoretical and experimental results on libration motion of HArF in solid argon ͑Ref. 21͒ and on thermal reorganization of the unstable to stable HArF configuration. 22 It has been found experimentally that HArF molecules are formed slowly after photolysis of HF / Ar matrices at temperatures down to 8 K ͑without annealing͒. 23 It has been suggested that this «low-temperature» formation of HArF molecule occurs at a short-range scale and the reaction mechanism probably involves quantum tunneling of hydrogen. Local ͑short-range͒ processes have been often discussed in connection with noble-gas hydrides. Pettersson et al. have shown that the formation of HXeI molecule originates from a combination of local and global H atom mobilities. 24 Moreover, the recovery of HXeCC and HXeI after IR decomposition are local processes presumably involving quantum tunneling of hydrogen atoms. 6,7 The local mechanism of the HArF formation seems to be different from the cases of other HNgY molecules in Kr and Xe matrices formed mainly upon global mobility of hydrogen atoms activated by thermal annealing. 25 To investigate the question of the HArF formation scale is the main motivation of the present work. We study here the HArF formation at temperatures down to 8 K evaluating the formation kinetics, HF / Ar concentration dependence, effects of IR light and deposition temperature, H/D isotope effect, and activation energy of the formation. This information sheds light on the HArF formation mechanism.
Another subject related to the current work is the decay of the ͑NgHNg͒ + cations in noble-gas matrices. This phenomenon has been known for many years and various mechanisms have been suggested. 23,[25][26][27] It has been shown that the HArF formation and the decay of ͑ArHAr͒ + are kinetically different, suggesting that these two phenomena are fundamentally unconnected. 23 However, the decay of ͑ArHAr͒ + and ͑ArDAr͒ + occur at 8 K with similar rates to HArF and DarF formation, respectively; more studies in this direction are clearly needed. Here we report new experimental results on ion decay supporting the previous conclusions.

II. EXPERIMENTAL DETAILS
HF / Ar solid matrices were studied in a closed-cycle helium cryostat ͑APD, DE 202A͒ providing temperatures down to 8 K. The matrices were deposited onto a cold CsI sub-strate by passing argon gas ͑99.9999% AGA͒ through HFcontaining pyridine polymer ͑Fluka͒ using a method developed previously. 9 The deposition temperature varied between 8 and 16 K, most of the matrices being deposited at 12 K. A drop of sulphuric acid H 2 SO 4 ͑J.T. Baker 95-97%͒ was added to the deposition line to remove impurity water from the sample-gas flow. Deuteration of HF was made by using a drop of deuterated sulphuric acid D 2 SO 4 ͑Merck 96-98%, D-degree Ͼ99%͒ in the deposition line and deuteration was up to 90% was attained. 9 After deposition the matrices were photolyzed at 8 K with a krypton lamp ͑Opthos, microwave power 40 W͒ emitting 127-160-nm light. The ratio between HF and the matrix gas ͑Ar͒ was varied by changing the amount of pyridine polymer and the argon flow rate through the deposition line. The absolute concentrations of HF and DF in the matrices are difficult to control with this preparation method. However, the HF / Ar matrix ratio can be estimated using the integrated molar absorptivity of gaseous HF ͑99.8 km/ mol͒, 28,29 and the HF / Ar ratios were typically ϳ1 : 2000. In some experiments the HF and DF amounts were comparable and the HArF, ͑ArHAr͒ + , DArF, and ͑ArDAr͒ + species were studied simultaneously. The decomposition of HF after 30-60 min irradiation with a Kr lamp was typically ϳ20% and practically no progress was achieved for longer exposures. The HF decomposition is probably limited due to photolysis-induced absorbers and scattering in the optically thick matrix. 30 The IR absorption spectra in the 4000 to 400 cm −1 range with resolution of 1 cm −1 were measured with a Nicolet SX60 FTIR spectrometer.

III. EXPERIMENTAL RESULTS
The IR spectra of HF and DF in an Ar matrix are presented in Fig. 1. The monomeric HF bands are located at 3962.2 and 3953.6 cm −1 , the weak Q branch is at 3920.6 cm −1 , and the ͑HF͒ 2 absorption is at 3826.3 cm −1 , DF absorbs at 2895.1 cm −1 and ͑DF͒ 2 at 2803.5 cm −1 . These bands are in agreement with the published data. 31 Photolysis with a Kr lamp decreases these bands and builds up the ͑ArHAr͒ + absorptions. 32 Some HArF is already seen after photolysis as pointed out earlier. 9,12 Figure 2a shows the HArF formation upon annealing the photolyzed matrices. Unstable HArF has triplet absorption of the H-Ar stretching mode at 1965. 7, 1969.4, and 1972.3 cm −1 and the absorptions of stable HArF are at 2016.3 and 2020.8 cm −1 . 11,12 The unstable and stable HArF configurations are related to different local matrix morphologies. 20 Annealing at 20 K produces the maximum amount of unstable HArF. The thermally unstable configuration converts to the more stable configuration upon annealing at higher temperatures due to the relaxation of the matrix surrounding. 11,12,22 The broad feature marked with L has been recently assigned to librational motion of HArF in solid argon by Bochenkova et al. 21 A similar phenomenon was previously found for HXeBr and HKrCl in Xe and Kr matrices, respectively. 33 As reported previously, HArF forms slowly even at the lowest experimental temperature of 8 K. 23 In the present work, we study this process in more detail, evaluating its dependence on various experimental parameters. Figure 2b shows the formation of HArF and DArF at 8 K ͑top two spectra͒. No HF recovery is observed after a long time at low temperatures. Annealing of the matrix at 20 K after a long time at 8 K does not increase the HArF concentration much, i.e., the formation process can be mostly completed even at 8 K. In other words, the long-time low-temperature formation leads to nearly the same amount of unstable HArF as brief annealing at 20 K ͑see Fig. 2b͒. It should be noticed that the low-temperature ͑ϳ10 K͒ process produces HArF mainly in the unstable configuration. The thermally activated transition of unstable to stable HArF was studied elsewhere. 22 The formation of unstable DArF at 8 K is very slow compared to HArF ͑by a factor of ϳ50 at 8 K, see Fig. 3a͒, which agrees with the previous measurements. 23 As a result of this isotope effect it is difficult to finish the DArF formation process at 8 K in real experiments. The HArF and DArF formation rates increase at elevated temperatures ͑see Fig.  3a͒. The possible effect of broadband IR light was studied by measuring the HArF formation under Globar light and in the dark. The formation rates in these two cases were found to be very similar, as shown in Fig. 3b for formation of HArF at 10 K. The matrix deposition temperature affects the HArF formation rate at low temperatures. The HArF formation kinetics was studied at 10 K for matrices deposited at 8, 12, and 16 K. The HArF formation was fastest for deposition at 8 K and slowest for deposition at 16 K. The difference between the formation rates was substantial ͑several-fold͒. To eliminate this factor of uncertainty we analyzed mainly the data obtained for deposition at 12 K.
The ͑ArHAr͒ + ions are generated by photolysis and slowly decay after the formation ͑see Fig. 4 and this process shows a strong isotope effect. While the ͑ArHAr͒ + ions practically disappear after 8 h at 8 K, the ͑ArDAr͒ + decay is minor after 50 h at 8 K, remaining small even after annealing the sample for several minutes at 20 K. The decay of the ions is enhanced at elevated temperatures but less efficiently than the simultaneous HArF formation ͑see Fig. 5͒. Practically no effect of Globar irradiation on the ͑ArHAr͒ + decay is observed. The decay time of the ions is measured at the 1 / e level of concentration observed after photolysis. Figure 5 presents the Arrhenius plot for the HArF and DArF formation ͑unstable configuration͒ in matrices deposited at 12 K with the HF / Arϳ 1 ϳ1 / 2000 ratio. The formation time at various temperatures is calculated at the 0.63 level of the HArF concentration after additional annealing at 20 K similarly to the previous studies of other HNgY molecules. 34,35 It is a characteristic time for a single exponential process and can be used for more complicated kinet-ics functions as well. 36 The DArF formation time is presented at the 0.45 level of the largest value obtained by annealing at 20 K since the 0.63 level is not achieved in these experiments because formation is very slow. The Arrhenius plots for the HArF and DArF formation are quite different. The formation of DArF is much slower compared to HArF at 8 K whereas the formation rates are comparable at 15 K. The activation energy of the DArF formation was estimated by using four high-temperature data points of the Arrhenius plots yielding ͑322Ϯ 18͒ cm −1 and the three data points for HArF lead to the same value of ͑333Ϯ 25͒ cm −1 .
The concentration dependence of the HArF formation was studied by varying the amount of the precursor HF ͑see insert in Fig. 5͒. We found that the lower HF precursor concentrations ͑1:1400 to 1:6000͒ did not affect the formation kinetics. For highly concentrated samples ͑ϳ1:700͒ the formation is somewhat faster.

IV. DISCUSSION
In principle HArF formation can be a global or local process with respect to atomic mobility; thermal mobilization of both H and F atoms is possible. Local ͑short-range͒ formation efficiency is controlled by the local formation bar- Low-temperature formation of HArF and DArF. 23 The upper trace is measured after 30 min at 8 K and the middle trace after 50 h at 8 K after photolysis. Notice that the unstable configuration of HArF is mainly formed at 8 K. The lower trace shows the situation after additional annealing of the matrix at 20 K. The band marked as L is due to the librational motion of the HArF molecule in solid argon. 21 The spectra are measured at 8 K. The HF͑DF͒/Ar matrix was preliminarily photolyzed with a Kr lamp. The photolysis-induced HArF absorptions are subtracted from the spectra shown in panel ͑b͒. rier that can be intrinsic ͑characteristic of a molecule in vacuum͒ or matrix-induced. The intrinsic barrier can be caused by the avoided crossing between the ionic and neutral potential energy surfaces, which enables the HArF formation from the H + Ar+ F neutral fragments. The calculated intrinsic formation barrier is 0.18 eV by Runeberg et al. 14 and 0.44 eV by Li et al. 16 Bihary et al. found that the H + Ar + F reaction barrier is about 0.2 eV lower in solid argon than in vacuum and they estimated the barrier for the H + Ar+ F reaction in solid argon to be ϳ0.3 eV. 15 Reaction barriers in matrices are very complicated to model due to the many particles involved into the process. Intuitively, precursor photodissociation and HArF formation can be affected by the specific, local matrix morphology. Local processes in the solid state are energetically different from global ͑long-range͒ processes. 22,37 Global ͑long-range͒ formation occurs via atomic diffusion over relatively long distances ͑compared to the lattice parameter͒, and the formation reaction most probably occurs apart from the parent cage. Formation of HKrCl in a Kr matrix and formation of HXeCCH and HX-eBr in a Xe matrix have been shown to be mainly global processes, 25,35 although some contribution from local mobility can also be noticed. 24 The concentration dependence of the formation rate is a good probe for the local ͑vs. global͒ formation process. For a local process no dependence on the precursor concentration should take place. In contrast, the global formation should become slower for lower concentrations of the reacting species because more jumps are needed to reach the reactive center. The experimental dependence of the HArF formation time on the HF precursor concentration is shown in the insert in Fig. 5. It is observed that lower ͑from 1:6000 to 1:1400͒ HF precursor concentrations have practically no effect on the HArF formation time, which suggests a local formation mechanism. For example, the HKrCl formation time changed between these precursor concentrations by a factor of two, which was attributed to a global formation mechanism. 35 For higher HF / Ar concentrations ͑ϳ1:700͒, the formation time decreases by about 30%. To explain the concentration dependence one can speculate that matrices with higher HF concentrations have more defects, and this may lower the formation barrier compared to a defect-free matrix. This change is also much smaller than in the case of the HKrCl formation at similar precursor concentrations. It should be remembered that the concentration dependence of the formation time can also be weakened by high losses of mobile atoms, 25 which complicates the comparison of different experiments.
Local of HArF formation is reasonable because of the locality of HF photolysis in solid argon, which is evidenced by HArF absorptions observed after UV photolysis. In this model some dissociating H atoms stay very near the parent cage and they can be captured in the HArF energy minimum, thereby hence forming an intermediate. These HArF intermediate molecules can be destroyed by the next photons, providing large excess energy to H atoms and driving them further away from the parent cages. Another portion of the dissociating H atoms can in principle be stabilized in the matrix promptly after the first dissociation event. The ratio between these two channels of secondary ͑via HArF interme-  diate͒ and direct stabilization of H atoms in the matrix is difficult to measure and it presumably depends on the excess energy and matrix material ͑see Ref. 10 for more discussion͒. It is possible that the H atoms stabilized in the matrix via decomposition of HArF intermediates do not participate in the low-temperature formation of HArF due to a longer-scale separation from the parent cages. If the dissociation via intermediates is a major channel, a large fraction of the H atoms could be lost for HArF formation at low temperatures. On the other hand these H atoms may form HArF at elevated temperatures via longer range mobility; however, this formation channel is evidently not dominating. It should be understood that our claim on local formation of HArF does not reject the long-range mobility of hydrogen atoms in solid argon. It is possible that H atoms can globally move in a perfect argon crystal; however, no indication for this mobility has been presented, to our knowledge, in the literature and it is not evidenced by the present results either. In accord, Vaskonen et al. observed an efficient geminate recombination of HBr after UV photolysis in solid argon, which also featured the locality of solid-state photolysis and short-range recovery of HBr. 38 In the present case the recovery of HF does not occur because this channel is suppressed by the HArF formation.
The dependence of the formation time on the deposition temperature ͑slower formation for higher deposition tem-perature͒ is consistent with the concept of the local formation mechanism. In this model higher deposition temperatures lead to more regular matrix structures providing deeper energy wells for atoms stabilized after photolysis and consequently higher reaction barriers. In previous work the increase of deposition temperature led to a higher temperature of the unstable to stable HArF transformation, 11 which probably originated from local mobility of vacancies. 12,22 No formation of stable HArF is observed at the lowest experimental temperatures; hence, thermal mobilization of vacancies is required for the unstable to stable HArF transition with the experimental activation energy of about 70 meV. 22 The Arrhenius plot for the HArF and DArF formation is shown in Fig. 5. The dependence consists of two parts: a plateau for lower temperatures and a thermally activated part above 12 K. The temperature dependence with a lowtemperature limit is characteristic for a quantum tunneling mechanism. 39,40 The classical contribution to chemical reactions is negligible at low temperatures and quantum tunneling dominates. It was found out by Goldanskii et al. that chemical reactions involving light particles do not stop even at zero temperature, which is called a low-temperature limit of a chemical reaction rate. 39 The other support for the tunneling mechanism is the large H/D isotope effect, showing a much slower process for deuterium ͑see Fig. 3a͒. 40 The H/D isotope effect also shows that the HArF formation ͑unstable configuration͒ is not essentially contributed by mobility of F atoms and matrix vacancies.
Bihary et al. have calculated that the potential barrier for the formation of HArF in solid argon from the neutral H + Ar+ F fragments is ϳ0.3 eV with width 1.3 Å. 15 In our opinion this barrier is consistent with the observed HArF formation rates taking into account the hydrogen tunneling rates for the cis to trans process in carboxylic acids having similar reaction barriers. 40 Furthermore, it has been suggested previously that HXeI and HXeCC recover via tunneling of hydrogen after IR decomposition. 6,7 We suppose that the tunneling mechanism is very probable for the HArF formation, which contradicts with the theoretical conclusions of Bihary et al. 15 Importantly for the present study, tunneling of hydrogen atoms may occur only through a short distance, which further evidences the locality of both the HF photodissociation and HArF formation in solid argon. It looks plausible that HArF is formed at low temperatures from the same H atoms as upon annealing at 20 K. This conclusion is supported by the stability of the HArF concentration upon annealing at 20 K after a long period at 8 K. On the other hand some amount of unstable HArF is probably reorganized to stable HArF at this temperature, 22 but hopefully it is not the dominating channel.
The thermal activation of the DArF formation is observed at somewhat lower temperatures than that of HArF and at 14-15 K HArF and DArF are formed with similar rates. The Arrhenius plots give the activation energy of ϳ300cm −1 ͑40 meV͒ for the HArF and DArF formation. The mechanism of this thermal activation is not easy to interpret. This energy is much higher than that of a single-phonon process in solid argon ͑about 10 meV͒. At first glance it is inconsistent with the reaction over energy barrier calculated by Bihary et al. ͑0.3 eV͒. 15 However, this computational barrier may be inaccurate so that we cannot completely exclude an over-barrier reaction. Moreover, efficient tunneling can occur below the barrier ͑by about 0.1 eV͒, hence the effective barrier can be lower than the calculated one. 41 For the H + XeC 2 reaction, the experimental activation energy was remarkably similar to the present case. 7 Next, the tunneling reaction can be activated by population of some higher energy state, but the nature of such a state is unclear. Resonance with the accepting state of HArF may also play a role. Finally, the tunneling process might be enhanced by reorganization of the matrix medium. The relaxed surrounding geometries are different before and after tunneling reaction, and this mismatch can slow down the tunneling process, which is much faster than the reorganization of heavy atoms. Temperature may help the surrounding to find geometry better suitable for tunneling, which increases the tunneling rate. For instance, this reorganization may involve local mobility of vacancies. Bochenkova et al. have recently calculated barriers for vacancy mobility around HArF molecule in an argon cluster, and the smallest activation energy ͑45 meV͒ was close to the activation energy obtained here. 22 However, the formation of unstable HArF has not been modeled and more theoretical effort is required to understand this activation mechanism.
Another interesting question is the relationship between the formation of HArF and the decomposition of ͑ArHAr͒ + . The neutralization of ͑ArHAr͒ + upon electron transfer from F − might in principle ignite the formation of HArF via the H + Ar+ F reaction. 23 However, these two processes have different kinetics showing that they are not directly connected. In the present work we fitted the HArF formation and ͑ArHAr͒ + decay data at 8 K with a stretched-exponential function and found different parameters for these two processes, in agreement with the previous conclusions of Ref.
23 on their independency. Even stronger support of our model is provided by the Arrhenius plot shown in Fig. 5. It is clearly seen that the HArF formation and ͑ArHAr͒ + decay possess very different temperature dependences: the ͑ArHAr͒ + decay is much less sensitive to the matrix temperature. Furthermore, the decay of ͑ArHAr͒ + is much slower ͑by orders of magnitude͒ than the DArF formation at 20 K ͑see Fig. 4͒. Because of this very slow process, we could not obtain experimental data for the Arrhenius plot of the ͑ArHAr͒ + decay.

V. CONCLUDING REMARKS
We have studied experimentally the low-temperature formation of HArF in various experimental conditions, including the concentration dependence, deuteration, and effect of IR light. The thermal activation of the process was also measured. The decay of ͑ArHAr͒ + cations was studied and its hypothetical connection with the HArF formation was discussed and finally rejected.
It has been found that HArF molecules in the unstable configuration are formed in major amounts after a long period at 8 K in a photolyzed HF / Ar matrix, which is in basic agreement with a previous report. 23 The Arrhenius plot shows a strong H/D isotope effect ͑about two orders of mag-nitude͒ and a low-temperature limit of the HArF formation. The activation energy for the HArF and DArF formation is about 40 meV, and it is presumably connected with the mobility of hydrogen atoms ͑not F atoms or vacancies͒. Several mechanisms are possible for this enhancement including activation of vacancy mobility in the matrix surrounding. All the data show that HArF forms in a local reaction of H atoms with the parent Ar-F centers and tunneling mechanism is very probable as supported by the observed strong deuteration effect and the low-temperature limit of the reaction. The locality of the precursor photolysis required for this tunneling phenomenon is consistent with the HArF formation during photolysis of HF in solid argon. 12 The decay mechanism of ͑ArHAr͒ + cations in an Ar matrix has been recently discussed in terms of tunneling of an electron from an electronegative fragment to the ͑ArHAr͒ + cations. 23,25 This process was tentatively demonstrated for synchronous neutralization of ͑KrHKr͒ + and CCNN − in a Kr matrix. 42 In this model, the electron tunnels over a relatively long distance to the cation, and this process does not lead to recombination of the neutralized fragments to a noble-gas hydride. The present work supports the previous conclusions since the kinetic data for the formation of HArF and decay of ͑ArHAr͒ + and thermal activation of these processes are different, especially for the deuterated species.

ACKNOWLEDGMENTS
The Academy of Finland supported this work partially through the Finnish Center of Excellence in Computational Molecular Science. H. L. is a member of the graduate school Laskemo ͑Ministry of Education, Finland͒. We thank Mika Pettersson for helpful discussions and Toni Nurminen for technical assistance.