© ESO, 2011

1. Introduction

The presence of polycyclic aromatic hydrocarbons (PAHs) in many phases of the interstellar medium is evidenced by their strong and ubiquitous mid-infrared (mid-IR) emission features (Smith et al. 2007; Draine et al. 2007; Draine & Li 2007; Tielens 2008). Mid-IR features are efficiently emitted by a PAH after excitation by an energetic photon. Toward dense clouds, however, these mid-IR features are strongly quenched. Here, most volatile molecules are frozen out on grains forming layers of ice (e.g., Pontoppidan et al. 2004; Boogert et al. 2008; Öberg et al. 2008; Bottinelli et al. 2010). Under such conditions, less volatile molecules such as PAHs condense on interstellar grains as well, incorporating them in interstellar ices.

Indeed, weak mid-IR absorption features in the spectra of dense clouds point to the presence of PAHs in these ices as well, with PAH:H₂O ice concentrations on the order of a few percent (Smith et al. 1989; Sellgren et al. 1995; Brooke et al. 1999; Chiar et al. 2000; Bregman et al. 2000). Based on these observations, Bouwman et al. (2011) deduced PAH:H₂O ice ratios toward MonR2/IRS3 to be about 2%, a ratio comparable to common interstellar ice species such as H₂CO and OCN⁻. However, since astronomical PAHs are expected to contain on the order of 50 to 100 C atoms and some 20 to 30 reactive C–H peripheral sites, their influence on interstellar ice chemistry and ice physics is likely important.

Experimental studies on the effect of vacuum ultraviolet (VUV) irradiation of interstellar ice analogues have shown that more complex molecules form in the simplest mixed ices (e.g. Gerakines et al. 1995; Öberg et al. 2009a; Muñoz Caro & Dartois 2009). Laboratory studies on VUV irradiated PAH containing ices indicated that PAHs are easily ionized (Gudipati & Allamandola 2003; Gudipati 2004; Bouwman et al. 2009, 2010). These experiments also showed the formation of new species. Kinetic information on these chemical reactions, however, is largely lacking.

Here, we present the time evolution of the use-up of four PAHs, anthracene (Ant, C₁₄H₁₀), pyrene (Py, C₁₆H₁₀), benzo[ghi]perylene (B_ghiP, C₂₂H₁₂), and coronene (Cor, C₂₄H₁₂) in H₂O ice together with the formation and destruction of the ionized PAH⁺ species. This paper is part II of a series which aims to quantify and understand the time-dependent chemistry of PAH:H₂O ice mixtures upon VUV irradiation and the resulting photoproducts. Part I deals with the mid-IR spectroscopy of the photoproducts and its applications to astronomical observations of low and high mass objects and has been published separately (Bouwman et al. 2011). Because of severe band overlap, the infrared experiments are not capable of providing the kinetic information reported here. The two papers together provide a complementary (vibrational and electronic) study of PAH containing H₂O ice, as well as their photoproducts and dynamic behavior upon VUV excitation.

Here, the chemical evolution is tracked by means of near-UV/VIS absorption spectroscopy at two temperatures, 25 K and 125 K. This work is an extension of a previous optical study of pyrene in H₂O ice (Bouwman et al. 2010) and draws more general conclusions on PAH photochemistry in ices based on a larger set of different PAHs. Furthermore, the present work extends the PAH:H₂O photochemistry to larger and astrophysically more relevant members of the PAH family.

The outline of this paper is as follows. In Sect. 2 the experimental setup is briefly discussed, together with details of supporting calculations. Section 3 describes spectra of the PAH and PAH⁺ cations, the assignments of the observed transitions and presents the (relative) oscillator strengths. The fitted time-dependent data are discussed in detail in Sect. 4, after which the astrophysical implications are discussed in Sect. 5. The conclusions are summarized in Sect. 6.

2. Experimental technique

The experimental details are available from Bouwman et al. (2009). The heart of the setup is of a high-vacuum (~10^-7 mbar) chamber. In the center of the vacuum chamber a MgF₂ sample window is suspended, which is cooled by a closed cycle He cryostat to a temperature of 25 K. Temperatures as low as 11 K can be realized. PAH containing H₂O ices are grown onto the sample window by vapor deposition. Milli-Q H₂O is further purified by three freeze-pump-thaw cycles and the PAHs are used as commercially available (Ant, Aldrich  ≥99%, Py, Aldrich, 99%, B_ghiP, Aldrich, 98%, Cor, Aldrich, 99%). The thickness of the ice samples is determined by laser interference and the amount of deposited PAH is monitored in absorption.

The inlet system has been modified for depositing the larger PAHs in our sample, B_ghiP and Cor. A sample container is mounted in the vacuum chamber and located adjacent to the H₂O deposition tube. The sample container is heated with polyimide insulated nichrome heater wire. The H₂O exposure is set to result in a growth rate of approximately 4 μm h^-1. The current through the heater wire determines the PAH sample temperature and thus the resulting sample concentration. It should be noted that, in view of the higher sensitivity of the optical detection scheme compared to the infrared, these concentrations are much lower than in Paper I.

After deposition on the 25 K window, the sample is heated to the desired temperature. Subsequently, the sample is subject to VUV radiation, which is produced by a H₂ flow microwave (MW) discharge lamp. The lamp operates at a static H₂ pressure of 0.4 mbar and a MW power of 100 W, resulting in an effective VUV flux of  ~10¹⁴ photons cm^-2 s^-1 at the sample surface (Muñoz Caro et al. 2002). In the present experiment this value may be higher as a reflecting tube has been positioned behind the lamp also directing diverging photons towards the ice sample. As an upper limit we use a flux of 10¹⁵ photons cm^-2 s^-1 in this paper.

Near UV/VIS absorption spectra are taken during VUV processing of the samples. To this end, a Xe-lamp is used as a broadband light source and a spectrometer equipped with a 1024  ×  256 pixel CCD camera is used as detector. The CCD camera is read out in vertical binning mode by a computer on which the raw data are converted into optical depth (OD = ln(I/I₀)). Spectra ranging from  ~280 to 800 nm are taken at a rate of 0.1 s^-1, which is sufficient to monitor chemical changes in our ice samples. Each spectrum is the result of co-adding 229 individual spectra, resulting in a high signal-to-noise ratio.

The integrated absorbance of the deposited neutral PAH signal, ${}^{\mathrm{\int }}\mathit{\tau }_{\mathit{\nu }}\mathrm{d}\mathit{\nu }$, is converted into a PAH column density, N_PAH, via the oscillator strength of the neutral PAH, f, and the conversion factor 8.88 × 10^-13 taken from Kjaergaard et al. (2000): ${\mathit{N}}_{\mathrm{PAH}}\mathrm{=}\frac{\mathrm{\int }{\mathit{\tau }}_{\mathit{\nu }}\mathrm{d}\mathit{\nu }}{\mathrm{8.88}\mathrm{\times }{\mathrm{10}}^{-13}\mathit{f}}\mathrm{·}$(1)Together with an ice thickness measurement based on the interference pattern in the reflection of a HeNe laser as described in Bouwman et al. (2009), this allows for a determination of the PAH:H₂O concentration. Sample deposition is stopped at thicknesses of  ~2 μm, resulting in comparable ice samples.

In a typical 4 h experiment, as many as 1400 spectra are obtained. The spectra are all baseline corrected by fitting a second-order polynomial through points where no absorptions occur and subtracting the polynomial. Additionally, absorption bands are integrated and, if necessary, corrected for contributions by atomic H lines of the H₂ MW discharge lamp. All the data handling is performed in a LabView program.

In order to support the assignment of the measured absorption photoproduct bands, density functional theory (DFT) calculations are performed using the gaussian09 software (Frisch et al. 2009). The B3LYP functional is used in conjunction with the 6-311++G(2d,p) basis set to determine the ground state geometry and electronic structure of PAH neutrals and cations as a starting point to predict the electronic transitions. Excited states are investigated within the framework of the time-dependent density functional theory (TDDFT) applying the same level of theory.

Fig. 1 The correlation between the amount of cation produced and the amount of neutral consumed, both relative to the total amount of deposited neutral precursor; Ant, Py, BghiP, or Cor.

3. PAH:H₂O spectroscopy

Long duration photolysis experiments are performed on a set of four PAHs (Ant, Py, B_ghiP, and Cor) in H₂O ice at low (25 K) and high (125 K) sample temperatures. An overview of the used mixture concentrations, the temperature at which the samples are photolyzed, and the oscillator strength values (f) of the neutral PAHs adopted from the literature is given in Table 1.

For each of the PAHs under investigation, the oscillator strength of the cation absorption bands relative to that of the neutral precursor is determined. This allows for a full quantitative study of the formation and destruction, which will be presented in Sect. 4. Relative oscillator strengths are determined by plotting the time evolution of the integrated absorbance of the cation transition against the integrated absorbance of the strongest electronic transition of the neutral species (see Fig. 1). Both integrated absorbances are normalized to the amount of deposited neutral PAH, which is determined during the preparation of the ice sample. A quantitative conversion of the PAH molecule into its cationic species is assumed to occur during the first photolysis stage: $\mathrm{PAH}\stackrel{\mathrm{VUV}}{\mathrm{-}0\mathrm{\to }}\mathrm{PA}{\mathrm{H}}^{\mathrm{+}}\mathrm{+}{\mathrm{e}}^{\mathrm{-}}\mathit{.}$(2)Linear fits are made to the first data points, and the slope directly reflects the oscillator strength of the cation relative to that of the neutral molecule. None of the PAH species in our sample substantially deviates from a one-to-one conversion during the first 100 s of photolysis, making the assumption valid and the resulting relative oscillator strength values reliable starting points for further analysis.

Figure 2 shows typical absorption spectra for the 25 K PAH:H₂O ice mixtures taken at the maximum cation absorption. The position of the band origin, the range used for integration and oscillator strength value relative to the strongest neutral absorption are listed in Table 2. The assignments of the neutral and photoproduct bands are discussed for each individual PAH below.

Fig. 2 The 280 to 800 nm spectra of the PAHs anthracene a), pyrene b), benzo[ghi]perylene c), and coronene d) in H2O ice, photolyzed at 25 K. Negative features indicate that the corresponding band carrier is destroyed, positive bands indicate that a species is formed. Hydrogen emission lines (sharp negative signals) originating from the VUV lamp are labeled with an H. The mixing ratios are 1:700, 1:5.000, 1:2.500, and 1:4.000 (PAH:H2O) for anthracene, pyrene, benzo[ghi]perylene, and coronene, respectively. The molecular structures are indicated in the corresponding spectra. Note that the y-axes for the different spectra are adjusted to facilitate the display and consequently band intensities of different species cannot be directly compared.

3.1. Anthracene (C₁₄H₁₀)

The negative signal between  ~310 and 380 nm (Fig. 2a) is assigned to the ${}^{\mathrm{1}}\mathit{B}_{\mathrm{2}\mathit{u}}$  ←  ${}^{\mathrm{1}}\mathit{A}_{\mathit{g}}$ Ant transition (Bak et al. 2000) and is caused by the photodestruction of neutral Ant molecules. The positive absorption features are caused by species that are produced from the parent PAH. A vibronic progression arises between 500 and 760 nm with its maximum at 719.6 nm. This progression has previously been assigned to the ${}^{\mathrm{2}}\mathit{A}_{\mathit{u}}$  ←  ${}^{\mathrm{2}}\mathit{B}_{\mathrm{2}\mathit{g}}$ transition of the singly ionized Ant species (Ant⁺) in an argon matrix (Szczepanski et al. 1993). For this transition we derive an oscillator strength value of 0.59 relative to that of the neutral. A sharp and strong absorption feature previously assigned to the ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}\mathit{u}}$  ←  ${}^{\mathrm{2}}\mathit{B}_{\mathrm{2}\mathit{g}}$ transition of Ant⁺ appears at 351.1 nm. This absorption has a relative oscillator strength of 0.15. Likewise, an absorption feature which is assigned to the ${}^{\mathrm{2}}\mathit{A}_{\mathit{u}}$  ←  ${}^{\mathrm{2}}\mathit{B}_{\mathrm{2}\mathit{g}}$ transition of Ant⁺ is found at 313.7 nm.

Two additional absorptions are apparent in the spectra of our photolyzed sample. One sharp absorption appears around 445.8 nm and is probably due to photolysis of small Py contaminations in our ice sample, resulting in a Py⁺ absorption. Additionally, a broad feature spanning the range from 380 to 470 nm is found. This band does not correlate with the cation features and is hence thought to be caused by a mixture of Ant+H and/or Ant+OH addition reaction products. These reaction products have previously been identified mass spectrometrically in VUV photolyzed Ant:H₂O (1: ≥ 100) mixtures (Ashbourn et al. 2007). Also, mid-IR absorption bands were reported in Paper I (Bouwman et al. 2011) that point to the existence of such species in photolyzed ices.

3.2. Pyrene (C₁₆H₁₀)

The VUV photolysis of Py:H₂O mixtures has been described in detail by Bouwman et al. (2010). Here the band assignments are briefly described for completeness.

The negative bands that appear between 290 and 345 nm are assigned to the ${}^{\mathrm{1}}\mathit{B}_{\mathrm{2}\mathit{u}}$ ← ${}^{\mathrm{1}}\mathit{A}_{\mathit{g}}$ electronic transition of neutral pyrene (S₂ ←    S₀) (Vala et al. 1994; Halasinski et al. 2005). Most of the positive bands that form upon VUV photolysis of the Py containing H₂O ice are ascribed to the Py⁺ species. The system ranging from  ~411–470 nm is the strongest Py⁺ transition and assigned as ${}^{\mathrm{2}}\mathit{A}_{\mathit{u}}$ ← ${}^{\mathrm{2}}\mathit{B}_{\mathrm{3}\mathit{g}}$. The weaker absorption bands between 350 and 370 nm are assigned to the ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}\mathit{u}}$ ← ${}^{\mathrm{2}}\mathit{B}_{\mathrm{3}\mathit{g}}$ Py⁺ vibronic transition. Finally, the band on the red-wing of the strongest Py⁺ transition, situated around 490.1 nm, is due to the ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}\mathit{u}}$ ← ${}^{\mathrm{2}}\mathit{B}_{\mathrm{3}\mathit{g}}$ transition. Besides the rather strong Py cation absorptions, two weak bands which do not correlate with the cation features are detected around 400 and 405 nm. The band at 400 nm has been previously found to originate from an electronic transition in PyH^· and the band at 405 nm has been tentatively assigned to an electronic transition of ³Py (Bouwman et al. 2010).

3.3. Benzo[ghi]perylene (C₂₂H₁₂)

The negative bands between  ~280 and 390 nm in the spectrum of the VUV photolyzed B_ghiP:H₂O reflect the loss of B_ghiP. The absorption bands have previously been assigned to the S₂(¹B₂)  ←    S₀(${}^{\mathrm{1}}\mathit{A}_{\mathrm{1}}$) transition (Rouillé et al. 2007). In turn, new bands appear upon VUV photolysis of the B_ghiP containing H₂O ice. A very strong absorption, which has previously been assigned to a B_ghiP⁺ absorption by Salama et al. (1995) arises at 509.7 nm. Another, much weaker, absorption appears in the mid-IR at 762.2 nm. This band has been assigned in previous argon matrix work to a low energy electronic transition of the B_ghiP⁺ species (Hudgins & Allamandola 1995). Here we report a new absorption band at 404.3 nm, which shows a clear correlation with the other B_ghiP⁺ absorptions. This band is likely due to a higher electronic state of B_ghiP⁺.

An attempt has been made to assign the new observed cation transitions by means of DFT calculations. The optimized geometry of the B_ghiP cation has C_2v symmetry. The calculations are based on the molecule in the x–z plane with the z-axis coinciding with the C₂ symmetry axis. The electronic ground state is ${}^{\mathrm{2}}\mathit{A}_{\mathrm{2}}$ making dipole-allowed transitions to A₂, B₂, and B₁ states possible. In the observed wavelength range, several transitions are predicted by TDDFT calculations. A transition to a ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}}$ state is calculated to be at 673 nm (f_calc. = 0.048), which may correspond to the observed band at 762.2 nm. Another transition calculated at 472 nm (f_calc. = 0.21) is relatively close in energy to the strongest observed band at 509.7 nm and we tentatively attribute this to the corresponding transition also involving a ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}}$ state. The absorption at 404.3 nm corresponds with a transition either to the ${}^{\mathrm{2}}\mathit{A}_{\mathrm{2}}$ or to the ${}^{\mathrm{2}}\mathit{B}_{\mathrm{1}}$ state. The calculations also predict a transition to the ${}^{\mathrm{2}}\mathit{A}_{\mathrm{2}}$ state around 300 nm, which overlaps with an absorption from the neutral molecule and consequently cannot be discriminated.

3.4. Coronene (C₂₄H₁₂)

The neutral Cor molecule is of D_6h symmetry. From the A_1g ground state, dipole-allowed transitions are only possible to electronic states of A_2u or E_1u symmetry. The observed absorption spectrum in cryogenic inert gas matrices strongly resembles the spectrum of hexa-peri-hexabenzocoronene (HBC, Rouillé et al. 2009). The weak S₁(¹B_2u) ←    S₀(¹A_1g) transition expected near 390 nm is not seen in our spectrum, but the S₂(¹B_1u) ←    S₀(¹A_1g) transition appears at 337.6 nm. As with HBC, it gains intensity due to vibronic interaction with the first allowed transition S₃(¹E_1u) ← S₀(¹A_1g) found at 301.4 nm. The TDDFT calculations predict this transition at 303 nm (f_calc. = 1.3).

Fig. 3 Neutral PAH decay and the rise and fall of the corresponding cation signals for the four PAHs, anthracene a), pyrene b), benzo[ghi]perylene c), and coronene d) trapped in H2O ice for two different temperatures (left – 25 K and right – 125 K) as a function of VUV photolysis time. The molecular structure and fitted curves (derived from the model in Fig. 4) are indicated in the plots as well.

As already noted by Oomens et al. (2001), the degenerate ground state of the coronene cation causes Jahn-Teller interaction and leads to an effective reduction of the point group from D_6h, as found for the neutral species, to D_2h. It also complicates the assignment of measured absorption bands on the basis of DFT calculations. However, the DFT geometry optimization predicts a slightly elongated structure for the coronene cation with an elongation of only 0.7% of the total diameter. Therefore, we can only provide tentative band assignments assuming D_2h to be the correct point group. In that case, the electronic ground state is ${}^{\mathrm{2}}\mathit{B}_{\mathrm{3}\mathit{u}}$. The feature seen in H₂O ice at 463.7 nm corresponds with a transition to an excited state of B_1,   2g symmetry. This band was found at 462 nm in an Ar matrix (Szczepanski & Vala 1993) and at 459 nm in a Ne matrix with an oscillator strength of 0.012 (Ehrenfreund et al. 1992). The calculation predicts three states between 390 nm and 460 nm with somewhat higher oscillator strengths between 0.02 and 0.05. Likewise, the broad cation absorption at 687.1 nm could belong to transitions to states of B_1,   2g symmetry. Further dipole-allowed transitions to B_1,   2g electronic states are predicted around 310 nm, overlapping with the strongest absorption of neutral Cor. These bands are probably very broad, leading to a rise in the baseline around 310 nm as visible in Fig. 2.

4. PAH ionization rates

The previously derived oscillator strength of the strongest cation band relative to that of the strongest neutral parent PAH transition (Fig. 1) is now used for quantification of the reaction channels that are driven by the VUV photolysis. The evolution of the column density in the ice is tracked as a function of time. As stated earlier, this is far more accessible in the optical work presented here than in the infrared work (Paper I). Cation relative oscillator strengths are used to convert the integrated absorbance into a column density relative to the amount of the neutral species. The analysis is based on the strongest cation bands listed in Table 2. The resulting time evolution of the number densities, relative to the deposited amount of neutral PAH, is shown in Fig. 3 for the four systems studied here.

In the analysis we consider a channel for ionizing the PAH with rate k₁₁, a back-channel for recombination of PAH⁺ species with electrons with rate k₁₂, a channel for the formation of products P₁ directly from the parent neutral PAH with rate k₁, and the formation of products P₂ from the PAH⁺ species with rate k₂. The reaction scheme is schematically displayed in Fig. 4. This reaction scheme effectively describes the processes observed in the laboratory experiment for a binary PAH:H₂O ice mixture. Caution is needed when using the reaction rate constants derived here in an astrochemical model, since the measurements are performed for specific binary mixtures only. For the Py:H₂O sample, a reaction scheme with one more channel was used in Bouwman et al. (2010) since this molecule clearly follows an additional reaction path involving PyH^·. The contribution of PyH^· to the total amount of reaction products is low and for the sake of consistency all data, including pyrene, are fitted with the same rate equations following from the reaction scheme indicated in Fig. 4.

The time-dependent chemistry of PAHs in H₂O ice is studied for two temperatures, 25 K and 125 K. The fits are co-plotted with the experimental data (left and right column of Fig. 3, respectively). Fits to the time evolution curve of the PAH with the strongest cation absorption, B_ghiP, are made twofold; keeping the ionization channel, k₁₁, temperature dependent and temperature independent. This yields insight in the effect of the temperature on the process. In our previous publication (Bouwman et al. 2010) we noted a temperature dependence in the Py ionization channel. In the case of Py⁺, the signal is very weak and thus an accurate fit is hard to obtain. From the two fits to the B_ghiP data in Fig. 3 (both T-dependent and T-independent) it is clear that temperature actually does not have a large influence on the quality of the fit, and inherently the ionization rate k₁₁ turns out to be independent of temperature. The ionization rate k₁₁ of the other species, Ant and Cor, also does not exhibit a large temperature dependence. Table 3 gives an overview of the fit parameters which are obtained while keeping all parameters free, i.e. dependent of temperature. The photoionization rate (k₁₁) and recombination rate (k₁₂) are in agreement with previously published rates, while product formation rates show significant differences (Bouwman et al. 2010; Gudipati & Allamandola 2003).

Fig. 4 The reaction scheme used to fit the experimental time evolution of PAHs, their cations and other photoproducts upon VUV irradiation is indicated in the dotted box. The total reaction scheme including the accretion of PAH species from the gas phase into the ice is used for modeling the astrophysical case as described in Sect. 5.

From the constants in Table 3 it is found that the recombination channel, k₁₂, increases with temperature, except in the case of Cor. The rate of product formation directly from the parent PAH, k₁, exhibits some temperature dependence. In the case of Ant:H₂O and Cor:H₂O photolysis it increases by a factor of  ~1.5 and for Py and B_ghiP it increases by a factor of 2. This behavior indicates that reactions of a PAH molecule with H₂O ice photoproducts – H and OH – become more efficient at higher temperatures, presumably because of a larger mobility of these species. The formation rate of products from cation species, k₂, on the other hand, drops to zero at 125 K for all PAHs except for B_ghiP. What is most striking about the data presented in Table 3, is that the ionization rate (k₁₁) for all PAHs is of the same order.

By the end of an experiment, typically after about 14 000 s of VUV photolysis, both the neutral parent PAH and the cation features disappear. They are replaced with other absorptions superposed on the baseline, but there is no clear spectral signature which can be compared to literature data on possible photoproducts. In Paper I more clear signatures of photoproducts were found, which were tentatively assigned to fundamental vibrations of functional groups in the newly formed photoproduct species (PAH–X_n, with X being H, O, or OH). It is likely that similar photoproducts are formed in the experiments described here.

5. Astrophysical implications

As indicated by the rate constants in Table 3, PAH ionization is an efficient process in VUV irradiated PAH containing H₂O ice for the laboratory conditions studied here. This was already known for Py, but is found here to be the case for several other PAHs as well. The largest molecule in our sample, Cor, contains 24 carbon atoms, which is still small from an astrophysical viewpoint. However, the experimental study presented here indicates that the ionization rate is size independent in the range of species investigated. We extend our findings to the astrophysical case, in which larger PAH species (N_C ≥ 50) are thought to be most relevant. Although additional experiments are needed to verify that these rates are indeed size independent, we assume for the moment that the rates reported here provide a good estimate for the larger species. This is sufficient to derive a first order approximation on the photoprocessing rates of PAHs in an interstellar ice.

Fig. 5 Left panel: modeled depletion of gas phase PAHs on cold interstellar grains as a function of cloud density. Right panel: modeled time-dependent PAH loss and PAH photoproduct growth in a PAH:H2O ice mantle within a cloud experiencing AV = 3 mag of visual extinction. It is explicitly assumed that H2O ice photodesorption is not efficient enough to act rate limiting.

Here, we discuss the relevance of PAH:H₂O ice chemistry in clouds where A_V ≥ 3 mag, the magnitude found for the onset of the 3 μm interstellar H₂O ice absorption band (Whittet et al. 2001). In a previous paper (Bouwman et al. 2010) we have estimated the rate of photoprocessing of PAHs versus photodesorption of the main mantle species using the recent water photodesorption results obtained by Öberg et al. (2009b). We concluded there that for A_V = 3 mag the role of photoionization of PAHs in the mantle is of the same order as that of the photodesorption of the mantle species. Since interstellar extinction attenuates the hard UV required for photodesorption (λ ≤ 140 nm) far more than the radiation sufficient to ionize PAHs in H₂O ice (λ ≈ 310 nm (Gudipati 2004; Woon & Park 2004)), even for the typical interstellar radiation field (Habing 1968), PAH:H₂O photochemical processes are expected to take place. This effect increases in importance as the cloud density increases. In the following we put the measured reaction rates described in Sect. 4 into their astrochemical context.

In the diffuse interstellar medium (PDRs), PAHs are initially in the gas phase. The formation of PAH photoproducts on grain mantles therefore requires two steps: the neutral PAHs must first freeze out from the gas phase onto the grains where they can then participate in solid state reaction networks. The rate of accretion of a PAH species onto a grain, ${\mathit{R}}_{\mathrm{acc}}^{\mathrm{ISM}}$, is given by $\hspace{0.17em}\begin{array}{}\begin{array}{}\end{array}{\mathit{R}}_{\mathrm{acc}}^{\mathrm{ISM}}\mathrm{=}{\mathit{v}}_{\mathrm{PAH}}{\mathit{n}}_{\mathrm{grain}}\mathrm{(}\mathit{\pi }{\mathit{a}}^{\mathrm{2}}\mathrm{)}{\mathit{n}}_{\mathrm{PAH}}\begin{array}{}\end{array}\\ \begin{array}{}\end{array}\mathrm{=}\sqrt{\frac{\mathrm{8}\mathit{k}}{\mathit{\pi }}}\frac{{\mathit{n}}_{\mathrm{grain}}}{{\mathit{n}}_{\mathrm{H}}}\mathrm{(}\mathit{\pi }{\mathit{a}}^{\mathrm{2}}\mathrm{)}\sqrt{\frac{{\mathit{T}}_{\mathrm{gas}}}{\mathit{M}}}{\mathit{n}}_{\mathrm{H}}{\mathit{n}}_{\mathrm{PAH}}\\ \begin{array}{}\end{array}\mathrm{=}\mathrm{4.57}\mathrm{\times }{\mathrm{10}}^{-8}\sqrt{\frac{{\mathit{T}}_{\mathrm{gas}}}{\mathit{M}}}{\mathit{n}}_{\mathrm{H}}{\mathit{n}}_{\mathrm{PAH}}\\ \begin{array}{}\end{array}\mathrm{=}{\mathit{k}}_{\mathrm{acc}}{\mathit{n}}_{\mathrm{H}}{\mathit{n}}_{\mathrm{PAH}}\mathit{,}\end{array}$(3)with v_PAH the velocity of the PAH molecule in the gas phase, n_PAH the gas phase number density of PAHs, n_H the total number density of hydrogen, $\frac{{\mathit{n}}_{\mathrm{grain}}}{{\mathit{n}}_{\mathrm{H}}}$ the dust to gas number ratio (10^-12), a the standard grain radius (0.1 μm), T_gas the gas temperature, M the molecular mass of the PAH molecule (in amu), and k_acc the accretion rate (in cm³ s^-1). Additionally, all measured PAH reaction coefficients, k^lab, are scaled to the interstellar photon flux and the extinction of the cloud by ${\mathit{k}}^{\mathrm{ISM}}\mathrm{=}{\mathit{k}}^{\mathrm{lab}}\left(\frac{{\mathrm{\Psi }}_{\mathrm{ISM}}}{{\mathrm{\Psi }}_{\mathrm{lab}}}\exp \left[\mathrm{-}\mathit{\gamma }{\mathit{A}}_{\mathrm{V}}\right]\mathrm{+}\frac{{\mathrm{\Psi }}_{\mathrm{CR}}}{{\mathrm{\Psi }}_{\mathrm{lab}}}\right)\mathit{,}$(4)with k^lab the measured ionization rate constant (s^-1), Ψ_ISM the interstellar UV flux (10⁸ photons cm^-2 s^-1 from Shen et al. 2004), Ψ_lab the laboratory VUV flux, γ a measure of UV extinction relative to visual extinction (≈ 2) (Roberge et al. 1991), A_V the visual magnitude, and Ψ_CR the cosmic ray induced photon flux.

An initial total gas phase PAH abundance of 4% with respect to H₂O and an abundance of H₂O of 10^-4 with respect to n_H is assumed. We further use the largest PAH investigated in this study, Cor, as a prototype system, which results in M = 300 amu, leaving T_gas, n_H, and A_V as input parameters for the model.

The timescale at which PAHs freeze out onto the grains is investigated first. The left panel of Fig. 5 plots the gas phase PAH abundance for different initial densities n_H. The graph clearly shows that for a dense cloud with n_H = 10³ cm^-3, it takes more than 10⁷ years for the gas phase to become depleted of PAHs, far longer than the few million year lifetime of a typical molecular cloud. Densities of 10⁵ cm^-3 and higher need to be reached for PAH freeze out to occur on a reasonable timescale. The reason for this is that at low densities the frequency with which PAHs encounter a grain is very low, since the grain abundance directly scales with density. Furthermore, the accretion rate scales with the velocity of the species and inherently with its mass. PAH molecules are heavy molecules and thus move slowly through the cloud, thereby slowing down their depletion process. Once the PAH does encounter a cold grain, the sticking probability is high and, since PAHs are non-volatile molecules, they will remain in the ice as long as the ice matrix remains.

At such high densities, however, the interstellar radiation field is almost fully attenuated and only cosmic ray induced photons play a role. This is not enough to drive significant PAH photoprocessing within a reasonable time. However, in Bouwman et al. (2011) we have shown for the high-mass Young Stellar Object (YSO) W33A and the low-mass YSO RNO 91 that the ice mantle may contain between 2–3% of PAH photoproducts with respect to H₂O. In order for this to have occured, these high PAH photoproduct concentrations must have been formed under the influence of a strong radiation field local to the embedded protostar. The plot in the right panel in Fig. 5 shows the PAH ice chemistry as a function of time under the standard interstellar radiation field at 3 mag (right panel). The model starts with PAHs frozen out onto the grains. The neutral PAH and the photoproduct abundances are given with respect to the total PAH abundance. For A_V = 3 mag the formation of significant amounts of photoproducts is after about 10⁵ yrs.

Consequently, to explain the high abundances of frozen PAH photoproducts reported in Bouwman et al. (2011), grains first must be in a high density environment after which they are exposed to a high UV field. These conditions can be met with the following scenario for both the high- and low-mass YSOs W33A and RNO 91: the PAH:H₂O ices form in the pre-collapse, high density phase. Once the newly formed star starts radiating UV photons, ice covered grains in the vicinity of the star will be exposed to the UV field. In protoplanetary disks, a similar situation can occur due to vertical mixing. The ice covered dust grains form in dense regions around the mid-plane. Vertical mixing brings them closer to the warmer top layer of the disk and back down towards the cold mid-plane of the disk.

Due to a lack of laboratory studies on the frequency dependency of PAH ionization and water photodesorption, it is hard to put both processes in direct perspective. Water photodesorption was shown to be effective for Lyman-α radiation (Öberg et al. 2009b) and one may argue that water is photodesorbed before PAH photo-ionization starts to play a role. The number of PAH ionizing photons, however, is expected to be larger (as ionization also takes place for lower energies) and, in addition, PAH ionization energies are lowered in water ice (Gudipati & Allamandola 2004). Therefore, we expect that PAH ionization is a process that should not be neglected a priori. The data presented here, actually, offer a way to search for PAH⁺ signatures in the solid state, paving the way to answer the question to which extend PAH ionization in the solid state will be relevant. Indeed, optical depths for Py and Py⁺ in the near UV/VIS have been estimated to amount to 0.01 ≤ τ ≤ 0.06 towards MWC297 (Bouwman et al. 2009) and the information listed in Table 2 provides the means for an optical detection of PAH/PAHs⁺ in the solid state. The corresponding electronic transitions are more unique than the infrared signatures, and consequently overlapping spectra will not add up; in the end, individual contributions may turn out to be small.

6. Conclusions

The work presented here describes photolysis experiments on interstellar H₂O ice samples containing four different PAHs: Ant, Py, B_ghiP, and Cor. UV, visible and near-IR band positions for PAH and PAH derivatives have been determined in a H₂O ice matrix. Furthermore, oscillator strengths of the PAH⁺ species have been derived for all the PAH⁺ electronic transitions relative to those of the neutral parent PAH molecule. The temporal evolution of cation production is tracked for the four PAHs at the two temperatures, 25 and 125 K. Experimental data are fitted to a reaction network and the resulting rate constants are used in a first-order time-dependent astrochemical model, including PAH freeze-out and photoprocessing in ices. The main conclusions are:

• 1.

  All four PAHs behave similarly upon VUV photolysis in a H₂O ice. The cationic species are quickly and efficiently produced in low temperature ices. The cation number density reaches a maximum and then slowly subsides with photolysis time. The PAH photoioniziation efficiency decreases in high temperature H₂O ice. This behavior can be attributed to PAH-radical or PAH⁺-radical reactions being more important due to a larger mobility of radical species (H and OH) in the ice at these temperatures.

• 2.

  The PAH and PAH⁺ photochemical time evolution data have been fitted with a model based on a chemical reaction scheme involving PAH ionization and PAH reactions with radical species. All four PAHs exhibit similar reaction rates, allowing for the general conclusion that PAH photoreaction rates in H₂O ice at low temperatures are size-independent over the range of species studied here.

• 3.

  An astrochemical model is applied to estimate the time-dependent freeze-out of PAHs on cold interstellar grains in a dense molecular cloud. Because of their large mass and low number density, this model indicates that PAH depletion is not important on time scales less than 10⁶–10⁷ years for regions of clouds with densities less than 10⁵ cm^-3. However, at higher densities PAH freeze out becomes important.

• 4.

  The photochemistry of PAH:H₂O ices is modeled for translucent conditions. The model results indicate that, under translucent cloud conditions and in denser embedded regions exposed to the local radiation field of young stellar objects, PAH ionization processes may play a role. The PAH⁺ data presented here offer spectroscopic information needed to carry out astronomical searches to test this suggestion.

Acknowledgments

This work is financially supported by “Stichting voor Fundamenteel Onderzoek der Materie” (FOM), “the Netherlands Research School for Astronomy” (NOVA), and NASA’s Laboratory Astrophysics and Astrobiology Programs. The research leading to these results has received

funding from the [European Community’s] Seventh Framework Programme [FP7/2007-2013] under grant agreement n° [238258]. The authors would like to thank Ankan Das for his contribution to the data reduction.

References

All Tables

All Figures

	Fig. 1 The correlation between the amount of cation produced and the amount of neutral consumed, both relative to the total amount of deposited neutral precursor; Ant, Py, BghiP, or Cor.
In the text

Fig. 2 The 280 to 800 nm spectra of the PAHs anthracene a), pyrene b), benzo[ghi]perylene c), and coronene d) in H2O ice, photolyzed at 25 K. Negative features indicate that the corresponding band carrier is destroyed, positive bands indicate that a species is formed. Hydrogen emission lines (sharp negative signals) originating from the VUV lamp are labeled with an H. The mixing ratios are 1:700, 1:5.000, 1:2.500, and 1:4.000 (PAH:H2O) for anthracene, pyrene, benzo[ghi]perylene, and coronene, respectively. The molecular structures are indicated in the corresponding spectra. Note that the y-axes for the different spectra are adjusted to facilitate the display and consequently band intensities of different species cannot be directly compared.

In the text

	Fig. 3 Neutral PAH decay and the rise and fall of the corresponding cation signals for the four PAHs, anthracene a), pyrene b), benzo[ghi]perylene c), and coronene d) trapped in H2O ice for two different temperatures (left – 25 K and right – 125 K) as a function of VUV photolysis time. The molecular structure and fitted curves (derived from the model in Fig. 4) are indicated in the plots as well.
In the text

	Fig. 4 The reaction scheme used to fit the experimental time evolution of PAHs, their cations and other photoproducts upon VUV irradiation is indicated in the dotted box. The total reaction scheme including the accretion of PAH species from the gas phase into the ice is used for modeling the astrophysical case as described in Sect. 5.
In the text

	Fig. 5 Left panel: modeled depletion of gas phase PAHs on cold interstellar grains as a function of cloud density. Right panel: modeled time-dependent PAH loss and PAH photoproduct growth in a PAH:H2O ice mantle within a cloud experiencing AV = 3 mag of visual extinction. It is explicitly assumed that H2O ice photodesorption is not efficient enough to act rate limiting.
In the text