Experimental Simulation of Titan's Stratospheric Photochemistry: Benzene (C6H6) Ices

We performed laboratory experiments to study the photochemical evolution induced by long‐UV irradiation of benzene ices in Titan's atmosphere. The aim of this study was to investigate whether photo‐processed benzene ices could lead to the formation of aerosols analogs to those observed in Titan's stratosphere. Prior to that, spectroscopic properties of amorphous and crystalline benzene ices were studied as a function of temperature, using infrared spectroscopy. UV photolysis experiments (λ > 230 nm) of benzene ices led to the formation of volatile photo‐products, among which fulvene is identified, and of a residue dominated by νCH IR features, demonstrating that pure aromatic‐based polymeric structures are not sufficient to explain the composition of Titan's stratospheric haze layer. However, we provide a characterization of long‐UV‐induced benzene‐containing aerosol analogs, which will contribute to Titan's surface organics layer. These data are of prime interest in the context of the future Dragonfly space mission.

tral Mass Spectrometer (INMS) (Waite et al., 2005) during the Cassini mission (Coustenis et al., 2007(Coustenis et al., , 2003Waite et al., 2007). Remote sensing measurements performed over a range of altitudes allowed the determination of the benzene mixing ratio (Coustenis et al., 2007;Cui et al., 2009;Koskinen et al., 2011). It decreases with altitude, indicating a high altitude production and a low altitude sink (Vuitton et al., 2008;Waite et al., 2007;Wilson & Atreya, 2004;Yoon et al., 2014) which is partially explained by a condensation process, as this is the main loss process of organics in Titan's atmosphere (Anderson et al., 2018(Anderson et al., , 2016De Kok et al., 2007;Griffith et al., 2006;Khanna et al., 1987;Lavvas et al., 2011;Mayo & Samuelson, 2005;Samuelson & Mayo, 1991;Samuelson et al., 1997Samuelson et al., , 2007. Indeed, vapor condensation of organics occurs in the lower part of the stratosphere, where the temperature diminishes, at an altitude which depends, for a given compound, on several parameters such as its saturation vapor pressure, abundance and the local temperature, resulting in the formation of microscopic icy particles or aerosols . Using cloud microphysics and radiative transfer models, different groups (Barth, 2017;Lavvas et al., 2011) have published the condensation profile and cloud formation of the different gaseous species present in the stratosphere. They both proposed the condensation of benzene at around 85 km for temperatures of about 130 K from their models, based on the temperature profile measured during the descent of Huygens in the atmosphere (Fulchignoni et al., 2005) and vapor abundances measured at northern latitudes (Coustenis et al., 2010;Vinatier et al., 2015). However, seasonal dynamics can have an impact on ice formation, for instance during vertical dramatic cooling episodes, like the one that occurred at the south pole simultaneously to the reversal of Titan's atmospheric circulation, two years after the northern spring equinox (Bampasidis et al., 2012;Coustenis et al., 2016;Teanby et al., 2012). It resulted in an increase of the southern mixing ratios of benzene and other species (Coustenis et al., 2016(Coustenis et al., , 2018. Near the south pole during the fall season in this hemisphere, Vinatier et al. (2018) detected benzene ice between high (∼300 km, which is around three/four times higher than its expected condensation altitude) and low (<100 km) stratospheric altitudes.
Benzene is a particularly important molecule because it is suspected to play a key role in the formation of polycyclic aromatic hydrocarbons (PAHs), polyphenyls or polycyclic aromatic heterocycles (PANHs) or even aerosols at high altitude (e.g., Delitsky & McKay, 2010;Gautier et al., 2017;Lebonnois, 2002;Mahjoub et al., 2016;Waite et al., 2007). In fact, Vuitton et al. (2008) have suggested that benzene is formed in the highest atmospheric layers (>900 km) by ion molecule reactions and diffuses downward where it induces-under photolysis-the formation of phenyl radicals (C 6 H 5 • ) known to be at the origin of a rich production of aromatics. A recent study has upgraded the benzene photochemical reaction network, highlighting the importance of neutral and ion reactions for the production of aromatic species such as toluene (C 6 H 5 CH 3 ) or ethylbenzene (C 6 H 5 C 2 H 5 ) (Loison et al., 2019). In addition, some studies have been performed to investigate the influence of C 6 H 6 in Titan's aerosol production, either theoretically (Delitsky & McKay, 2010;Lavvas et al., 2011;Lebonnois, 2005) or experimentally (Gautier et al., 2017;Sciamma-O'Brien et al., 2014Sebree et al., 2014;Trainer et al., 2013;Yoon et al., 2014) by the incorporation of benzene into a N 2 /CH 4 mixture using plasma discharges or far-UV radiations. However, to our knowledge, the role of gas phase benzene photolysis in Titan's atmosphere is well investigated while the role of benzene ice photolysis in the stratospheric production of aerosol particles has not yet been investigated experimentally (cf. Figure 1, Yoon et al., 2014). Previous experiments were performed under conditions (temperatures between 4 K and 19 K)/energy sources) relevant for the ISM, either to understate its fate under UV radiations (Rahul et al., 2020;Ruiterkamp et al., 2005) or ion bombardment (Callahan et al., 2013;Strazzulla & Baratta, 1991), or to identify its photo-degradation pathways with the help of matrix-isolated environment (Johnstone & Sodeau, 1991;Ruiterkamp et al., 2005). But none of these studies can be used as a reference for simulating Titan's low stratospheric environment. Previous experimental work conducted by Couturier-Tamburelli et al., 2018;Gudipati et al., 2013) focused on understanding FUV solar photon-induced aging process of nitrile ices formed under stratospheric-like conditions in the laboratory. The authors demonstrated the formation of polymeric residues after warming up irradiated ices whose infrared features correspond, for the most part, to those observed by the CIRS and VIMS (Visible and Infrared Mapping Spectrometer) instruments in the aerosol layer present in the stratosphere of Titan. Although recent experimental works focused on the fate of acetylene (Abplanalp et al., 2019) or N 2 /CH 4 ices (Vasconcelos et al., 2020), no study has been performed on the fate of benzene ices at such stratospher-ic altitudes. As benzene is expected to condense below 100 km, that is, at altitudes mainly impacted by solar UV photons λ > 230 nm (Gudipati et al., 2013;Lavvas et al., 2008;Wilson, 2004), understanding the evolution of pure benzene ices submitted to low-energy photons is essential in identifying its subsequent contribution to the formation of aerosols that will participate in the organic layer that covers Titan's surface. Therefore, in this study, we conducted laboratory experiments to perform the photolysis of pure benzene ices, with photons at λ > 230 nm, in order to help in understanding the composition of photo-produced aerosols in the stratosphere. In particular, we highlight the role of benzene ice photolysis in the production of stratospheric aerosols by comparing our spectroscopic data with those of the stratospheric aerosol layer acquired by the VIMS instrument, before the period of formation of benzene ices at high altitude following the spring equinox.

Experimental Details
Benzene (for HPLC, assay ≥ 99.9%, from Sigma-Aldrich) was used after purification by vacuum distillation in several freeze-pump-thaw steps with the help of several liquid nitrogen baths. The absence of impurities (H 2 O, CO 2 ) in each deposited benzene ice film was confirmed by IR spectroscopy. Benzene was deposited from a glass line onto a gold-plated surface located within a high vacuum chamber. The vacuum chamber pressure was 10 −9 mbar for deposition temperature below 20 and 10 −8 mbar for higher temperatures. The sample holder temperature was controlled from 16 to 300 K with a closed-cycle helium refrigerator and maintained using a model 21 CTI cold head cryostat, a resistive heater and a Lakeshore 331 temperature controller. Each deposited ice sample corresponded to a quantity of about 20 μmol of benzene. The quantity of sample deposited was chosen to obtain an intensity inferior to 1 a. u. on the IR spectra recorded. Samples were warmed up with a heating rate between 0.5 and 2 K.min −1 , this warm-up at 300 K resulted in a residue. IR spectra were recorded in reflection-absorption mode in the mid-infrared region between 4,000 and 4,600 cm −1 using a Bruker Tensor 27 Fourier transform infrared spectrometer with DTGS detector (Butscher et al., 2015). They were taken after deposition, after pumping the vacuum chamber back to base pressure, that is, without benzene vapor flowing in the chamber, and after rotating the sample holder. Each spectrum was averaged over 100 scans during the photolysis experiment and over 500 scans for the vibrational analysis of the residue, with a resolution of 1 cm −1 , except for the background which was averaged over 300 scans with the same resolution. After deposition, the vacuum chamber was brought back to base pressure (10 −9 mbar) before conducting photolysis experiments.
Photolysis experiments were performed using an Oriel 500 W high-pressure mercury lamp (λ = 200-2,500 nm) with discrete Hg lines in the UV-Vis region between 200 and 600 nm, where most of the photon flux of 2.34 × 10 16 photon.cm -2 .s −1 (∼13 mW.cm −2 ) resides. The photolysis time (48 h) was chosen to simulate a period of time in which benzene ices can sediment in the stratosphere (here, considering a solar flux of ∼10 14 photons.cm -2 .s −1 at 100 km on Titan, the photons flux of the UV source employed is therefore 234 times more important than the one that reaches Titan's lower stratosphere. Two days of photolysis in laboratory corresponds to 468 days or 1.3 years on Titan. However, considering Titan's exposure to the sun during the day and night, these 1.3 years becomes 2.6 years). Besides, the sedimentation time scale of a particle in the stratosphere depends on several parameters such as the particle radius or the pressure and the altitude (e.g., Barth & Toon, 2003), Figure 5 shows the fall velocities depending on particle size). In addition, two supplementary criteria, imperative in terms of experimental feasibility and chemical consideration, had to be taken into account for the choice of such photolysis duration: (1) due to the weak electronic absorption of benzene at these wavelengths, the duration needed to be sufficiently long to observe spectral modification using a low-sensitivity in situ technique, that is, IR spectroscopy, and (2) to obtain only the formation of primary products.
We performed two different experiments: the first one reproduces fairly well the condensation sequence of ices in Titan's stratosphere, that is, benzene vapor was deposited onto the sample holder kept at 130 K. The temperature was maintained at 130 K for a few hours in order to insure a stable crystalline phase, and then cooled down to 70 K to prevent any desorption of benzene. The second experiment consisted of depositing benzene vapor at 70 K, followed by an annealing to 130 K before cooling to 70 K. These two experiments were performed in order to see if any differences are observed in the infrared spectrum due to the crystallization process. In addition, to identify the transition in the benzene ice phases from amorphous to crystalline, amorphous benzene was deposited at 16 K and then warmed-up to 300 K with several heating rates ranging between 0.5 and 2 K.min −1 . The ultimate goal of this experiment was to determine desorption temperature of solid benzene, with a quite accurate margin of variation supplied by the different heating rate, in our experimental conditions.
Sublimation is expected to happen at any temperature in a high vacuum chamber because, when an ice is formed, it is in equilibrium with the vapor phase. Since the chamber has turbo pumps, that vapor can continuously be pumped away, causing the ice to sublimate more and more over time. Nevertheless, this is a slow process at lower temperatures and this phenomenon is negligible at 70 K (see Figure 5), a temperature corresponding to the lowest one reached in the stratosphere of Titan. Furthermore, the choice of this temperature was all the more judicious. Indeed, in future experiments we will want to study co-condensed ices with different molecules, that is, HCN, which starts sublimating at around 80 K under our experi-mental conditions. In both experiments, benzene is in the crystalline phase prior to the photolysis step, and the irradiation of benzene ice leads to the same photo-chemical activity of this molecule under our experimental conditions. Likewise, this temperature is sufficiently low to trap in the ice the photo-produced volatile fraction that will be characterized in a future work. Under these conditions of temperature (70 K), the subsequent benzene loss rate will only be attributed to photochemical processes, that is, avoiding any other competitive pathway such as sublimation. The evolution of the solid samples during irradiation was monitored by in situ infrared spectroscopy. After irradiation, ice samples are warmed up to 300 K with the cryogenic system switched-off in order to sublimate the photo-produced volatiles. The refractory organic residue is retained on the surface of the substrate.

Benzene Ice
As the infrared data are poorly documented in the literature, it is necessary to study the thermal evolution of benzene in our experimental conditions to determine the evolution of the shape of its spectral features and its desorption/condensation temperature.
X-ray analysis supplemented by Fourier transform methods has shown the crystalline structure of benzene to be orthorhombic, with four molecules per unit (Bacon et al., 1964;Cox et al., 1958), the crystallographic parameters of which have also been determined. In this configuration, benzene molecules pack together like sheets of six-toothed gear wheels (Cox, 1958;Cox et al., 1958). The first member of the aromatic ring family, benzene belongs to the D 6h group and possesses 30 normal vibrational modes. Only two of them, namely A 2u (nondegenerate) and E 1u (degenerate), are infrared-active. However, the physical state of benzene can induce the activation of other vibrational modes (Halford & Schaeffer, 1946;Hollenberg & Dows, 1962), in particular for crystalline benzene, as explained by the development of intramolecular forces.
The solid benzene ice infrared spectrum was first studied by Halford and Schaeffer (1946) in the amorphous phase and then by numerous teams (Hollenberg & Dows, 1962;Mair & Hornig, 1949;Strazzulla & Baratta, 1991) in the crystalline phase. As we intend, in this study, to contribute to the interpretation of the VIMS-measured infrared spectra of aerosols present in Titan's stratosphere, we have chosen to focus on the six most intense absorption bands, the others being assigned to combination modes and/or harmonics, already detailed elsewhere (Bertie & Keefe, 2004;Mair & Hornig, 1949). The justification for this choice is that the quantities of benzene ice deposited in our experiments (and thus the resultant intensities of the IR signal) are much greater than those detected in Titan's atmosphere. Hence, it can be excepted that only the most intense benzene IR features would be detectable in Titan's IR spectra. Here, we present a study of the pure solid benzene infrared spectrum in both the amorphous and crystalline phases along with the vibrational assignments using Herzberg's notation (Herzberg, 1945).

Amorphous Pure Benzene Ice at 16 K
The infrared spectrum of amorphous benzene deposited at 16 K is shown in Figure 1. Wavenumbers of amorphous/crystalline benzene fundamental and combination modes observed in this study, along with their assignments (Bertie & Keefe, 2004;Herzberg, 1945;Mair & Hornig, 1949), are listed in Table 1. Intense absorption bands are observed below 1,500 cm −1 . First, the most intense one at 1,478 cm −1 corresponding to the νC = C asymmetric stretching mode (ν 13 ), followed by the infrared features of two other fundamental vibrational modes at 1,035 cm −1 (ν 14 ) and at 674 cm −1 (ν 4 ), assigned to the δCH in-plane bending mode and the γCH out of plane bending mode, respectively. Another prominent absorption band characteristic of aromatic molecules, associated with the νCH stretching mode (ν 12 ), is located at 3,033 cm −1 . The most intense combination modes are attributed to ν 2 + ν 13 + ν 18 and ν 13 + ν 16 at 3,070 and 3,089 cm −1 , respectively.

Influence of Temperature on Pure Benzene Ices: Identification of Phase Transition and Desorption Processes
As the thermal evolution of solid benzene is rarely (Hollenberg & Dows, 1962;Ishii et al., 1996;Mair & Hornig, 1949) documented in the literature, we have studied the influence of temperature on the position and the shape of its infrared absorption bands. To monitor its phase transition by infrared spectroscopy, as previously described in the literature (Fraser et al., 2001;Koehler, 2001;Toumi et al., 2016), benzene ice was deposited at 16 K and then warmed up to 300 K with several heating rates (from 0.5 to 2 K.min −1 ). Figure 2 presents the evolution of the surface coverage of C 6 H 6 ν 13 and ν 14 modes with temperature and Figure 3 shows the infrared spectra of benzene as an amorphous ice and a crystalline one at various temperatures (cf. Table 1). These experiments were repeated three times and the variations observed on the surface coverage were less than 5% for a given temperature. The surface coverage Θ(T) of a vibrational mode corresponds to the value of the corresponding band peak area which has been normalized to the highest peak area value obtained from this band during the heat-up experiment. Benzene crystallization is observed around 55 K, which is consistent with works published earlier (Dawes et al., 2017;Ishii et al., 1996). Depending on the vibrational mode considered, during crystallization, the associated absorption band increases (ν 14 , ν 4 , ν 2 + ν 13 + ν 18 ) or decreases (ν 13 , ν 12 , ν 13 + ν 16 ) in intensity. No explanation of these trends has been found yet. Once the crystallization step has been initiated, C 6 H 6 pure ice spectra show significant changes involving mainly the shift and the split of most infrared absorption bands, which become narrower. Above 3,000 cm −1 , the absorption band corresponding to ν 12 is split into two bands as well as ν 13 + ν 16 modes are split into two bands whereas the band for the ν 2 + ν 13 + ν 18 mode keeps the same profile, becoming thinner with respect to the amorphous ice. The band of the ν 13 mode infrared signal is also split into two bands as well as the ν 14 mode infrared feature. Shape modifications of this final vibrational mode as a function of temperature has already been demonstrated (Swenson et al., 1959). The ν 4 mode is shifted to higher frequencies but its shape looks almost identical. From 70 K to 130 K, no significant frequency shifts are observed and slight band intensity variations can be noted for all these modes. We observed that the surface coverage starts to decrease significantly around 145-150 K. Above 165 K, all the molecules desorb. We did not obtain any residue from pure C 6 H 6 ice sublimation at room temperature.

UV Absorption of Benzene and Dissociative Pathways
Several electronic absorption spectra of benzene in both liquid and gaseous phase have been published in the 200-300 nm range (Pantos et al., 1978;Suto et al., 1992). Recently, Dawes et al. (2017) published both amorphous and crystalline solid benzene (deposited under vacuum, 10 -9 mbar) VUV spectra, compared with the gas phase VUV spectrum. Whatever the phase under consideration, the benzene electronic spectrum shows small absorption bands above 230 nm; however, when we performed irradiation at λ > 230 nm we irradiated through either singlet-singlet absorption ( 1 B 2u ← 1 A 1g ) or singlet-triplet absorptions ( 3 E 1u ← 1 A 1g, 3 B 2u ← 1 A 1g) (Dawes et al., 2017).
Most laboratory experiments interested in the photochemical behavior of benzene have been carried out using lasers, leading to the formation of a solid polymer (Nakashima & Yoshihara, 1982) and C 6 H 6 valence isomers (Griffith et al., 1975;Yokoyama et al., 1990). The latter have also been formed during the photolysis of benzene in cryogenic matrices at low temperature (Johnstone & Sodeau, 1991;Ruiterkamp et al., 2005;Toh et al., 2015). Fulvene, benzvalene, and Dewar benzene have been produced in argon or nitrogen matrix, when benzene was photolyzed at 253.7 nm or λ > 230 nm. Their theoretical infrared spectra have already been calculated with density functional theory calculations (Wheeless et al., 1995;Zhou & Liu, 1997).
The photolysis of adsorbed benzene at 105 K under KrF laser (λ = 248 nm) has demonstrated several dissociative pathways; absorption of one or two 248 nm photons induces the breaking of the CH bond causing the release of H° and C 6 H 5 • , only three-photon excitation triggers the aromatic ring opening, allowing the detection of C 2 H 4 and C 2 H 2 fragments by mass spectrometry (Varakin, 2018

Simulation of Benzene Photochemical Aging at Its Condensation Altitude
In order to understand the photochemical evolution undergone by stratospheric benzene condensates, pure benzene ices at 70 K were irradiated using a mercury lamp emitting at wavelengths λ > 230 nm, that is simulating solar photons that cross the stratosphere. As detailed in the experimental section, two types of deposition experiments were used to produce the benzene ice layer at 70 K. In a first experiment (case 1), the benzene ice was deposited at 70 K, then annealed to 130 K, then brought back down to 70 K. A second experiment ( tolysis of these two resulting ices, benzene absorption bands decreased in intensity, as displayed in Figure 4 (this experiment corresponds to case 1). The evolution of most benzene vibrational modes as a function of time cannot be fitted by any-order kinetic rate, because of new spectral features growing at their edge. However, the combination mode ν 2 +ν 13 +ν 18 peaking at 3,068 cm −1 provides an exception to these observations, and its evolution over time can be fitted by a first-order kinetic rate, as shown in Figure 5, giving a kinetic constant of (6.4 ± 0.5).10 −4 min −1 (case 1) or (7.7 ± 1.3).10 -4 min −1 (case 2). Hence, no significant variation on the kinetic constant value is observed from the results of both cases. As the photon flux of our lamp is about 2.34 × 10 16 photon.cm -2 .s −1 , the corresponding photo-dissociation cross-section is estimated to be 4.6 × 10 −22 photons.cm -2 (case 1) and 5.5 × 10 -22 photons.cm -2 (case 2). This value of the benzene photo-dissociation cross-section is probably a consequence of radical recombinations and slow diffusion processes, etc. Indeed, at low temperature, the mobility of molecules and heavy radicals is low while H° radicals can locally diffuse through the benzene ice's bulk.
After 2,880 min of irradiation, 145% of the C 6 H 6 is consumed. The infrared features that grow at the edge of the benzene absorption bands (observed in cases 1 and 2) can be assigned to two different mechanisms: amorphization or polymerization. Figure 6 compares the subtraction infrared spectrum at 70 K of the benzene ice (case 1) after-before UV photolysis with the infrared spectrum of amorphous benzene ice (50 K). The peaks marked with asterisks, on the left side of the figure, correspond to amorphous benzene features while the ones, marked with triangles, on the right side, are characteristic of the formation of polymeric material, as has already been highlighted by Couturier-Tamburelli et al. (2015, 2018, in the case of HC 5 N or HC 3 N photolysis. In addition to these two pathways, we observed the formation of new absorption bands, whose frequencies are listed in Table 2. The weak bands at 2,930 and 2,855 cm −1 indicate the formation of aliphatic compounds, corresponding, respectively, to a CH 2 asymmetric and symmetric stretching mode (Carpentier et al., 2012;Imanaka et al., 2004;Sciamma-O'Brien et al., 2017). Several absorption bands below 1,400 cm −1 are found at 1,341, 1,338, 1,078, 946, 898, 783, and 627 cm −1 (case 1). Except the band at 898 cm −1 , these others bands are also observed during the photolysis of the benzene ice obtained from the case 2 (cf. Figure 7). Because the 898 cm −1 peak is the least intense in that spectral region, it is possible that, for the case 2 experiment, its intensity was below the detection threshold of the IR spectrometer. In the literature, the photo-excitation of benzene is known to result in the formation of valence isomers. These photoproducts have been identified in particular during the photolysis of benzene isolated in cryogenic matrices. However, the IR spectra of the same species in an amorphous or crystalline phase are unknown to our knowledge. Referring to the study of Johnstone and Sodeau (1991), the IR features of fulvene (5-methylenecyclopenta −1 ,3-diene), one of benzene's valence isomers, could correspond to the ones obtained from the photolysis of the benzene ice. Indeed, in this study the benzene was isolated in an argon matrix and irradiated with UV λ > 230 nm, so we can consider frequency shifts due to differences in the chemical environment between the matrix and the ice experiments. In order to confirm this hypothesis, benzene isolated in a nitrogen matrix was irradiated (λ > 230 nm), then softly warmed up ( Figure S1). This last step induces the sublimation of MOUZAY ET AL.
10.1029/2020JE006566 9 of 18 (c) the matrix gas (N 2 ) around 35 K, resulting in the obtention of an amorphous benzene ice in which the photo-products formed in the matrix, and more precisely the fulvene, were trapped. A subsequent warm up to 70 K allowed for the experimental IR spectra of solid fulvene, to be measured. We have compared this spectrum with our experimental solid state data. The experimental frequencies are compared, in Table 2, to those of solid state and in situ matrix-produced fulvene ( Figure S1). The comparison of these new absorption band frequencies with those obtained after nitrogen matrix sublimation confirms the formation of fulvene in irradiated C 6 H 6 ices.

Infrared Characterization of Benzene Ice Residues
At the end of the photolysis experiment, we performed a warming-up to allow nonirradiated benzene monomers as well as volatile photoproducts, such as fulvene, to be released as gases. This step allows the in situ characterization of the refractory polymer produced by photochemistry. Under our experimental conditions, a small amount of nonvolatile residue was observed after the warm up. The residue obtained corresponds to an organic material which has a brownish color, reminiscent of Titan's haze, and is soluble in dichloromethane, as observed for laboratory analogs of high-altitude aerosols soluble in methanol, another polar solvent (Carrasco et al., 2009).
The infrared spectrum of the benzene residue is presented in Figure 8 and shows the presence of a broad and well-structured band in the 3,350-2,750 cm −1 region, along with a band at 1,404 cm −1 that could be assigned to semi-circle stretching of aromatics rings (Larkin, 2011). Higher intensities are observed for the peaks located above 3,000 cm −1 , corresponding to νCH sp 2 stretching modes, in comparison to the peaks below 3,000 cm −1 , assigned to νCH sp 3 stretching modes. As this cluster is the result of the overlap of several absorption bands, a deconvolution into Gaussian components was performed in order to characterize MOUZAY ET AL.

10.1029/2020JE006566
10 of 18 Figure 6. Comparison of the subtraction FTIR spectrum (after-before UV irradiation) of case 1 benzene ice at 70 K (top), with the infrared spectrum of amorphous benzene ice at 50 K (bottom), in different spectral regions (a) between 3,120 and 3,000 cm −1 , (b) between 1,500 and 1,000 cm −1 , (c) between 2,840 and 2,800 cm −1 ). Peaks marked with asterisks correspond to amorphous benzene features while features characteristic of polymerization are marked with triangles.
(a) (b) (c) Figure 5. Evolution of the surface coverage of the benzene ν 2 + ν 13 + ν 18 combination mode at 70 K for nonirradiated ice (squares), and during irradiation for ice deposited at 70 K, annealed to 130 K and cooling down to 70 K (circles, case 1) and ice deposited at 130 K and cooled to 70 K (triangles, case 2). The decrease is fitted by a single exponential decay.  Wheeless et al. (1995). e Johnstone & Sodeau, (1991). Figure 7. Comparison of the subtraction spectrum (after-before UV irradiation experiments) for both cases 1 and 2 benzene ice deposition conditions: (a) deposition 70 K followed by an annealing to 130 K and then cooling at 70 K (case 1) (b) deposition at 130 K followed by a cooling down to 70 K (case 2); *CO 2 (g): dashed lines photo-produced absorption bands common to both experiments. each of them, as presented in Figure 9. Their respective assignments are reported in Table 3. During the fitting process, we set positions as free parameters, as in previous studies (Carrasco et al., 2018). Taking into account previous works concerning both the influence of benzene in the chemical composition of tholins (Sciamma-O'Brien et al., 2017) and the spectroscopic characterization of hydrogenated amorphous carbon-based structures (Dartois et al., 2004), only two vibrational modes are exclusive to this study, the others being common to the studies cited above.

Table 2 Positions and Assignments of New Infrared Absorption Bands Obtained During the Photolysis of C6H6 Ice (λ > 230 nm) at 70 K, Compared with the Infrared Features of Fulvene in the Solid State and in Cryogenic Matrices
The Gaussian component at 3,213 cm −1 could be assigned to CH stretching modes from conjugated alkyne derivatives which could explain the frequency shift toward lower energies with respect to the one observed in alkyne functions (Larkin, 2011). The intense 3,144 cm −1 peak could result from a CH sp 2 stretching mode from a derivative/substituted polyphenyl compound (Ghanem et al., 1987). Aromatic and vinylic -CH 2 asymmetric stretching modes are centered at 3,086 cm −1 (Sciamma-O'Brien et al., 2017) while CH stretching modes from the aromatic/alkene family are found at 3,060, 3,031, and 2,999 cm −1 (Sciamma-O'Brien et al., 2017). By means of previous studies (Imanaka et al., 2004;Quirico et al., 2008;Sciamma-O'Brien et al., 2017), -CH 3 asymmetric and symmetric stretching modes are assigned to the Gaussian component at 2,964 and 2,863 cm −1 respectively against 2,930 and 2,829 cm −1 for -CH 2 group (Larkin, 2011). A last peak has been found at 2,894 cm −1 which could be assigned either to the sp 3 νCH stretching mode or to the -CH 2 Fermi resonance (Carpentier et al., 2012) or the asymmetric -CH 2 wing mode (Dartois et al., 2004). This last vibrational mode presents its symmetric counterpart at 2,800 cm −1 (Dartois et al., 2004).

Discussion
Vapor condensation of benzene in Titan's stratosphere, modeled in different works, leads to the formation of icy particles whose formation altitude depends on climatic conditions as well as abundances. Based on different models, the condensation of benzene is expected at 85 km for a temperature of about 130 K (Barth, 2017;Lavvas et al., 2011). At these altitudes, ices can evolve according to two main pathways; the first one consists of the adsorption of condensable species and the second one is triggered by photochemical reactions induced by long-UV solar photons. For benzene, which is a key molecule at high altitude in the formation of both the complex organics and aerosols that cover Titan's surface, none of these processes have been experimentally investigated in stratospheric-like conditions. Hence, the role played by benzene ices in this part of the atmosphere has not been clearly identified so far. In this study, we have shown the photo-MOUZAY ET AL.
10.1029/2020JE006566 12 of 18 chemical activity of benzene ice under long-UV irradiation (λ > 230 nm) which results in the formation of volatile photoproducts, in particular fulvene-detected by IR spectroscopy-and a refractory material, that can be considered as an analogue of aerosols formed in Titan's atmosphere. At 70 K, it takes 48 h to lose almost 145 % of the benzene in our experiment. However, as the solar flux reaching Titan's stratosphere is around 10 14 photon.cm −2 .s −1 , the same depletion rate for benzene will be achieved after 2.6 Earth years on Titan. Even though we present the photochemistry of benzene at a lower temperature (70 K) than the one at which it is expected to condense, our results demonstrate a low dissociation rate of benzene under stratospheric conditions. Hence, most of the benzene ice would not be modified by the solar UV photons that penetrate the stratosphere. With small increases in temperature induced, for example, by transportation to higher altitudes, all photo-produced volatiles species and benzene undergo thermal desorption observed between 160 and 190 K depending on whether the experiments were carried out under vacuum or at 10 -2 mbar leaving only the residue in the solid state, depending on the season in Titan's atmosphere. So, we can assume that in this condition the photoproducts formed in solid phase could be released in the stratosphere during climatic condition variations.
The formation of aerosols produced by the photochemistry of gas phase benzene under more energetic conditions, relevant for simulating the chemistry occurring at higher altitudes in Titan's atmosphere, has been experimentally investigated in other groups (Gautier et al., 2017;Sciamma-O'Brien et al., 2017;Trainer et al., 2013). In Figure 9, our experimental spectrum of the benzene residue is compared with the IR spectra of aerosols generated with a N 2 :C 6 H 6 gas mixture submitted to UV irradiation (Gautier et al., 2017 top, blue) or by a N 2 :CH 4 :C 6 H 6 gas mixture in a cold plasma discharge (Sciamma-O'Brien et al., 2017). Figure 8 also shows a comparison of our experimental spectrum to observational data of Titan's atmosphere at ∼200 km supplied by the VIMS spectrometer (lowest altitude, where a well-resolved spectrum is observed) (Bellucci et al., 2009;Carrasco et al., 2018;Kim et al., 2011). In their study, Kim et al. retrieved the detailed spectral feature using a radiative transfer program including absorption and scattering by haze particles. The spectral features of the haze retrieved from the VIMS data at various altitudes are similar to each other, indicating relatively uniform spectral properties of the haze with altitude. The observations were carried out on January 15, 2006 during the solar egress at 71°S (Bellucci et al., 2009).
As observed in Figure 9, it is in the spectral region characteristic of aromatic structures that we have found major discrepancies. Indeed, the most intense band observed at 3,144 cm −1 in our residue matches only with the very small VIMS-observed spectral signature of stratospheric aerosols, in terms of position but not in intensity. On the other hand, our deconvolved bands at 3,086, 3,060, 3,031, and 2,999 cm −1 are common to those observed in these other laboratory-produced aerosols and the VIMS data. Their relative intensities correspond rather well to the bands observed in the other aerosols. In particular, the greatest intensities observed for these bands mimic the ones observed in the aerosols produced from the N 2 :CH 4 :C 6 H 6 gas mixture (Sciamma-O'Brien et al., 2017) In addition, from 3,000 to 2,850 cm −1 , we found a similar spectral profile, indicating that similar aliphatic structures obtained in all the laboratory experiments are in good agreement with those of Titan's haze. That said, it is important to keep in mind that none of these selected experimental works leads to aerosol analogs which present IR features that are strictly identical to Titan's aerosols ones. However, the relative intensity of these aliphatic structures to the aromatics in our residue are low compared with other experimental data and the observation. Nevertheless, these relative proportions are close to those produced in the cold plasma Figure 9. Deconvolved infrared spectrum of irradiated benzene ice residue (bottom, blue line) produced by photochemistry (λ > 230 nm) and subsequently warmed up to room temperature compared with VIMS solar occultation spectrum at 203 km altitude (Bellucci et al., 2009;Kim et al., 2011) (orange solid line, processed by Carrasco et al. [2018] from Kim et al. [2011] between 3,050 and 2,825 cm −1 , and orange dashed line corresponds to Titan's haze spectra from Kim et al. [2011] above 3,050 cm −1 ). The experimental spectrum is also compared with the IR spectra of aerosols generated from a N 2 :C 6 H 6 gas mixture submitted to UV irradiation (Gautier et al., 2017) (top, blue) or from a N 2 :CH 4 :C 6 H 6 gas mixture in a cold plasma discharge (Sciamma-O'Brien et al., 2017). For the interpretation of the legend, the reader must refer to the web version of this article. discharge. As a consequence, the polymeric structure that has been produced by the exposure of a benzene ice to long-UV irradiation (70 K) may present similar aromatic/aliphatic groups-in terms of structures and proportions-to the aerosols generated by plasma discharge (initiated at 150 K) that simulate higher altitudes in the atmosphere. As these aerosols analogs from our study and from the Sciamma-O'Brien et al. (2017) and Gautier et al. (2017) studies have not been produced with the same temperature conditions, the analogs formation could be temperature dependent.
However, several deconvolved absorption bands of the benzene residue, namely at 3,213, 2,829 and 2,800 cm −1 , are not observed among the infrared features of Titan's haze observed in the stratosphere. In our study, the benzene residue formed in our experiments results from photo-processed ices without considering further aging processes such as that induced by the accretion of condensates at their surface. In addition, Courtin et al. (2015) demonstrated the coating of condensates below 300 km could impact the ratio between aromatics and aliphatics. In addition, experimental works performed by Couturier-Tamburelli et al. (2018) highlighted a possibility of photo-insertion of hydrogen atoms into a nitrile-based residue coated with HCN ice. Therefore, it will be of interest to investigate the coating effects on the spectral features assigned to -CH stretching modes of benzene residues. However, if the 3,144 cm −1 absorption band, characteristic of benzene photochemistry, is normalized to the intensity of the corresponding band from VIMS observations, we can assume that the contribution of benzene aerosols was very low in Titan's atmosphere at the time the measurement was performed. It demonstrates that other kinds of aerosols, present in the stratosphere, explain the high intensity of aliphatic CH contributions. In fact, in our experiment, the residue is produced by photo-processed hydrogen-bearing carbonaceous ices, while in Titan's stratosphere the haze probed by Cassini VIMS spectrometer results from the complex photochemistry of hydrocarbons and nitriles. Nevertheless, these preliminary laboratory experiments constitute a first step in understanding the role played by benzene ices in the formation of gaseous species and aerosols in the stratosphere that will subsequently be transported to the surface. It will thus allow the experimental simulation of a more complex benzene-containing cloud to better understand their photochemical evolution at these altitudes as well as their contribution to the surface organic layer.

Conclusion
To understand the fate of benzene ices in Titan's stratosphere, we turned to laboratory experiments to simulate its photochemical evolution triggered by long-UV irradiation (λ > 230 nm). Our results demonstrate a photochemical activity of benzene which leads to the formation of volatile photo-products, in particular MOUZAY ET AL.  Imanaka et al. (2004). b Sciamma-O'Brien et al. (2017). c Dartois et al. (2004). d Larkin, (2011). e Ghanem et al. (1987). f Carpentier et al. (2012).

Table 3
Deconvolved Infrared Absorption Bands of C6H6 Residue in the 3,150-2,750 cm −1 Region fulvene, as well as a solid residue, that is, a laboratory analog of benzene aerosols. Its spectroscopic characterization highlights a good correlation with VIMS observations for the positions of most of the deconvolved absorption bands, in particular for the features that correspond to aliphatic groups, which does not hold for the relative intensities of CH sp 2 /sp 3 ratio obtained in the residue when warming up the irradiated ice to 300 K. Therefore, the photochemistry of benzene ice might lead to the formation of organic materials that will contribute to the composition of the haze and to the organic material present on Titan's surface which will be investigated by the future Dragonfly space mission. Moreover, this study will also help to establish points of comparison with the changes that may be observed when benzene is co-condensed with another molecule like HCN.

Data Availability Statement
The data used from in this article are available on zenodo data repository http://doi.org/10.5281/ zenodo.4350657s