The evolution of polycyclic aromatic hydrocarbons under simulated inner asteroid conditions

Large polycyclic aromatic hydrocarbons (PAHs) are an important component of the interstellar medium. PAHs have been identified in the soluble and insoluble matter of carbonaceous chondrites (CCs). Here, we study the evolution of PAHs under conditions relevant to the interiors of asteroids and compare our results to PAHs observed in CCs. We have performed long‐term and short‐term hydrothermal experiments, in which we exposed PAH‐mineral mixture analogs of meteorites to temperature conditions representative of those predicted for asteroids interiors. Our results show that small PAHs with melting points within the aqueous alteration temperature of CCs form carbonaceous spherules in the presence of water. In this work, we describe the microstructure and morphology of these spherules. We discuss the similarities and differences compared to globules isolated from CCs.


INTRODUCTION
Polycyclic aromatic hydrocarbons (PAHs) comprise up to 20% of the carbon material in the interstellar medium (ISM) (Allamandola et al. 1985) with sizes of 50-100 carbon atoms (Tielens 2008). PAHs also represent a significant proportion of the soluble and insoluble organic fraction in carbonaceous chondrites (CCs) (Sephton et al. 1998;Botta and Bada 2002;Sephton 2002). The identified PAHs in CCs range in size from naphthalene (128 amu) to coronene (300 amu) (Becker and Bunch 1997). It has been hypothesized that PAHs from the ISM have been incorporated into solar system bodies, including carbonaceous asteroids during accretion (Plows et al. 2003). We do, however, see a large discrepancy between the size of PAHs detected in the ISM (up to 50-100 carbon atoms; Tielens 2008) and the size of the soluble organic PAHs in carbonaceous meteorites (up to 24 carbon atoms; Sephton 2002). CCs have experienced aqueous alteration, as indicated by the presence of the hydrous minerals like serpentine (Mg 2 Si2O 5 [OH] 4 ), which is the aqueous alteration product of olivine ([Mg,Fe] 2 SiO 4 ). Aqueous alteration affected both the mineralogical composition (e.g., Zolensky 2005;Brearley 2006) and the organic matter (OM) (Bunch and Chang 1980;Herd et al. 2011;Le Guillou et al. 2014;Alexander et al. 2017;Vinogradoff et al. 2018). Here, we study the evolution of PAHs in an aqueous environment under conditions relevant to meteoritic parent bodies, that also includes a possible catalytic effect of olivine serpentinization on OM alteration, as the aqueous alteration of olivine produces significant changes in pH and is known to affect redox conditions in terrestrial samples (Evans et al. 2013). We have conducted two sets of hydrothermal experiments focusing on naphthalene (C 10 H 8 ), fluoranthene (C 16 H 10 ), and coronene (C 24 H 12 ) with forsteritic olivine used to represent the dominant mineral phase within a meteorite matrix and temperatures ranging from 21 to 150°C.

EXPERIMENTS
We performed two sets of experiments, long-term hydrothermal experiments (LTHE) and short-term hydrothermal experiments (STHE). The LTHE took place at the Steinmann-Institute for Geology, Mineralogy, and Palaeontology at the University of Bonn and the STHE at the Department of Earth Sciences at Utrecht University. The LTHE aimed to study PAH reactivity in response to the reaction of olivine with water, whereas the STHE focused on the physical alteration of PAHs in an aqueous environment at different temperatures. The LTHE were carried out for 70 days at 150°C and pressure around 4.8 bar (saturation vapor pressure). Control samples were placed in the dark, at ambient laboratory conditions of 21°C and atmospheric pressure. The STHE were performed for 20-180 min at 40-150°C and pressure from 1 to approximately 4.8 bar. The chosen temperature ranges are within calculated aqueous alteration temperature ranges for CCs (Table 1), where 21°C presents a typical low and 150°C a typical high temperature for aqueous alteration in PAH-containing CCs (Alexander et al. 2017).

Materials and General Sample Preparation
Three PAHs were used in these experiments, naphthalene (C 10 H 8 , 99% purity Sigma-Aldrich), fluoranthene (C 16 H 10 , 98% purity Sigma-Aldrich), and coronene (C 24 H 12 , 97% purity Sigma-Aldrich). Due to the toxicity of PAHs, all steps involving PAHs were carried out under a fume hood. Olivine slabs (Fo90, San Carlos quarry, AZ, USA) were cut from olivine single crystals. Before settling on using olivine single crystals in our experiments, we investigated the use of olivine powder. However, as powder is challenging to analyze with SEM and Raman techniques, we did not gain conclusive results from these experiments. The experiments were performed with ultrapure anoxic water that was produced by bubbling argon gas through 500 mL of ultrapure distilled water for up to 12 h using a gas sparger. The transfer of the anoxic water into the reaction vessels was conducted in a glove bag under an argon atmosphere. An argon headspace was created in the reaction vessels by sealing them while still held under argon. Table 2 gives an overview of the physical properties of the PAHs and water that are relevant to our experiments. Care was taken to avoid any contamination by other organic materials during all stages of sample preparation and handling. The olivine slabs were cut into 4 mm 9 5 mm 9 2 mm pieces and were not polished. No organic containing solvents were used during cutting. After cutting, the olivine surfaces were cleaned with ethanol to remove olivine dust and potential organic contamination. All equipment was autoclaved and cleaned with ethanol before use. The polytetrafluoroethylene (PTFE) liners for the hydrothermal autoclaves and their lids ( Fig. 1) were cleaned with ethanol, followed by nitric acid solutions. Ethanol is a highly volatile solvent and thus is not expected to leave an organic residue on the equipment. Autoclaved, single-use pipette tips were chosen when pipettes were needed. Sample material was stored in sealed amber-colored glass vials that were placed in a closed box. For short-term storage of a few hours, the  Basile et al. (1984); Hahn et al. (1988); Zenobi et al. (1989); Buseck and Hua (1993); De Vries et al. (1993); Clemett et al. (1998);Sephton (2002); Elsila et al. (2005). samples were stored at room temperature. However, for long-term storage, the samples were placed in a refrigerator at 4°C.

Long-Term Hydrothermal Experiments
Three types of samples were used in the long-term experiments, PAH-olivine samples in water, PAH in water, and olivine in water. Table 3 gives an overview of the number of experiments per compound, the substrate, and the composition of the headspace. Two sample preparation methods were used for the PAHolivine samples. In the first method, 10 mg of PAH was dissolved in 1 mL toluene (C 6 H 3 CH 3 for liquid chromatography LiChrosolv Ò , purity >99%). Naphthalene and fluoranthene were dissolved fully, and coronene powder was used in suspension. The PAH-toluene solution was applied dropwise and uniformly onto the olivine slab surface, whereby each drop was allowed to dry before further application of fluid droplets. The drops were applied on the largest olivine surface in as small a quantity as possible to avoid the solution running off the slab. FT-IR analysis of dry PAH-toluene suspension showed no residue of toluene on the olivine after drying. The PAHcoated olivine slab was transferred into the 2 mL PTFEinlay, which was already inserted into the steel container, with the coated surface facing upward so that the largest area was exposed to the solution. Figure 1 shows the stainless steel, PTFE-lined reaction vessel. The reaction vessel with the coated olivine was then placed into a glove bag, and the glove bag was sealed. After filling the glove bag with argon to create an anoxic atmosphere, 1.5 mL of anoxic water was added to each of the reaction vessels, which were then sealed using a PTFE lid and steel cap. The resulting water-PAH ratio was 15:1 by mass. For the second method, the olivine slab was placed into a reaction vessel; 10 mg of PAH powder was placed on top of the olivine crystal followed by 1.5 mL of anoxic water. This mixture led to the same water-PAH ratio of 15:1 by mass and a water-rock ratio of 6:1 by mass with~56 mm 2 surface of the olivine exposed within the experiment. We chose a higher water-rock ratio than probably expected in a meteorite to simulate water-saturated conditions, reducing the possibility of the reaction being limited due to lack of water. However, our chosen water-rock ratio still lies within the limit of water to rock ratios described as closely matching the gross characteristics for the aqueous alteration of carbonaceous meteorites (Schulte and Shock 2004;Brearley 2006). After filling, the reaction vessels were  Kell (1975). d Auwers and Fr€ uhling (1921 divided into two groups. The first group was placed in an oven that was purpose-built in-house at the Steinmann-Institute for Geology, Mineralogy and Palaeontology of the University of Bonn and heated to 150°C for 70 days. The other group was placed in a room with a stable temperature of 21°C for the same period of time. After the experiments, the reaction vessels were allowed to cool for several hours before being placed into the glove bag and opened under an argon atmosphere. First, the liquid phase was removed with a manual microliter pipette and transferred to an amber-colored glass vial. Care was taken to remove the liquid; however, in some reaction vessels, not all the fluid could be removed. Then, the olivine slab and PAH residue were removed using tweezers or a spatula and placed in a second amber-colored glass vial. An ambercolored glass vial was selected to protect the samples from photochemical reactions.

Short-Term Hydrothermal Experiments
Short-term experiments of naphthalene, fluoranthene, and coronene crystals in anoxic water at temperatures up to 150°C were carried out in a conventional laboratory oven (Memmert UF110, universal drying and baking oven) at the Department of Earth Sciences at the Utrecht University. Each PAH was tested in a separate experiment series. Each experiment series consisted of 18 1.5 mL glass vials filled with 6.7 mg of the PAH being examined. These vials were separated into two sets, where 1 mL of ultrapure anoxic water was added to one set of the vials (water-PAH ratio 15:1 by mass). The anoxic water was prepared in the same way as the LTHE. Argon gas was then flushed into the headspace, and the vials were sealed with a septum. The filled vials were placed on aluminum foil, which lay loosely on a metal grid before being placed into the oven to avoid direct contact with the oven surface and to ensure homogeneous heating. Once all the vials were secured within the oven, the temperature of the oven was increased from room temperature (21°C) to 40°C. The start of the first experiment was taken as the time that the oven had stabilized its temperature. After 20 min at 40°C, one vial of PAH with water and one vial of PAH without water were carefully and quickly removed from the oven and immediately photographed. This resulted in very little change in the temperature within the oven. The oven was then heated to 60°C, and the next experiment initiated once the temperature had stabilized again. This procedure was repeated every 20°C between 60 and 150°C, removing two vials with each temperature step until all 18 vials had been removed from the oven. Vials that had been removed and photographed were allowed to cool beside the oven so that changes related to cooling could be monitored. Once all vials had cooled to room temperature, the liquid phase was removed using a manual microliter pipette and transferred to an ambercolored glass vial. The PAH residues were removed, when possible, using a spatula and placed in a glass vial. No glove bag was used in these experiments. After the experiments, all samples were stored using the same procedures as the LTHE.

Raman Spectroscopy
The Raman spectroscopic analyses took place at the different locations of the experiments, that is, University of Bonn and Utrecht University, to guarantee analysis of the samples as quickly as possible after the end of experiments. Raman spectra of LTHE samples were obtained with a high-resolution Horiba Scientific HR800 Raman spectrometer equipped with an Olympus BX41 microscope, He-Ne (632 nm) and near infrared (785 nm) laser, and an electron-multiplier chargecoupled device (CCD) detector. The acquisition time of each spectrum was 6-15 s with a maximum of 20 iterations. Measurements used a grating of 600 grooves per mm. The Horiba Scientific HR800 Raman spectrometer is located at the Steinmann-Institute for Geology, Mineralogy and Palaeontology of the University of Bonn. Raman spectra of STHE samples and the initial sample material were obtained with a WITec 300 alpha Raman spectrometer with a built-in light microscope, and equipped with Nd:YAG (532 nm) and near infrared (785 nm) lasers and CCD detectors optimized for the different wavelengths. The acquisition time was 0.05-0.5 s with a maximum of five iterations. All samples were measured with a grating of 600 grooves per mm. The WITec 300 alpha Raman spectrometer is located at the Department of Earth Sciences, Utrecht University.
All coronene samples were measured with a 785 nm laser with both the Horiba and WITec Raman spectrometers. Fluoranthene and naphthalene crystals were both measured with a 632 nm laser at the Horiba and 532 nm laser at the WITec Raman instrument, where fluoranthene samples were measured at reduced laser intensity to prevent sample burning. All Raman spectra were analyzed after baseline correction and by fitting the spectra to Lorentzian or Gaussian line shapes.

Focused Ion Beam-Scanning Electron Microscopy
Solid samples were mounted on an aluminum stub using double-sided, conductive copper tape (purchased from VWR) and coated with platinum-palladium alloy to a thickness of 10 nm. A selected suite of solid samples was investigated using a focused ion beam scanning electron microscope (FIB-SEM; FEI Helios Nanolab G3 UC) equipped with an energy-dispersive X-ray spectroscopy (EDX) system. Overview images of the material were taken using acceleration voltages between 0.1 and 5 kV to avoid charging effects and recrystallization of the organic material under the electron beam. Lower acceleration voltages allowed higher resolution images to be taken of the surface features as the interaction volume is minimized. The focused ion beam capability of the FIB-SEM was used to acquire a slice-and-view series for 3D volume reconstructions. Prior to all FIB operations, a platinum strip of 2 lm thickness was deposited to shield the rest of the sample from ion beam damage. For crosssectioning, a Ga-FIB was used at 30 kV and 0.43 nA. Images of the cross-sectioned surface were obtained using the backscattered electron mode at 15 kV and 0.4 nA with a voxel size of 8.33 9 8.33 9 20 nm. No alteration of the material's microstructure due to the electron beam was observed during continuous imaging at this kV. The FIB-SEM nanotomography volume was reconstructed and analyzed using the FEI Avizo 9 software. A detailed explanation of the FIB-SEM for milling materials can be found in Liu et al. (2016).
We performed calculations based on heat conduction models to assure that we have not altered our samples during their interaction with the ion beam in FIB-SEM investigations. Following Volkert and Minor (2011), the increase in temperature T for an incoming ion beam can be simplified to T = P/ (p 9 a 9 j), where a is the radius of the circular ion beam profile on the sample surface, j the thermal conductivity of an organic crystal, and P is the ion beam power in Watts. Using an ion beam acceleration voltage of 30 kV, a beam current of 0.43 nA, a sample thickness of 10 lm, and a very low thermal conductivity of 0.1 W m À1 9 K (adequate to describe j of PAHs; Ross et al. 1979), one obtains a temperature increase of 4 K. A more complex model published by Ishitani and Yaguchi (1996) results in a temperature increase of approximately 16 K. Both results are much lower than any melting temperature of the investigated PAHs. Moreover, scan durations during individual ion beam treatments were less than a few milliseconds. Thus, together with a minimal heating effect by the ion beam, we infer that the FIB-SEM conditions used are adequate for morphological investigations of PAHs.

Extracted Fluids
Fluids extracted from the hydrothermal experiments with naphthalene and fluoranthene at 150°C exhibit a yellow to brownish coloration. In contrast, fluids from experiments with coronene, pure and with olivine, and experiments with solely olivine (no PAH) at 150°C, as well as all experiments left at 21°C, did not exhibit a discoloration of the fluid.

Baseline Hydrothermal Experiments: PAHs in Water and Olivine Crystal in Water
Naphthalene, fluoranthene, and coronene crystals submerged in water at 21°C for 70 days did not show any macroscopic alteration and remained present at the bottom of the PTFE reactor. In contrast, after treatment at 150°C, only coronene remained visible within the PTFE reactor, whereas naphthalene and fluoranthene crystals could no longer be distinguished macroscopically.
In order to separately evaluate the reaction of olivine with water at 21 and 150°C, we conducted baseline experiments using slabs of olivine submerged in anoxic deionized water in the absence of PAHs. Olivine slabs from experiments at 21°C did not exhibit any macroscopic alteration or changes in color, whereas olivine slabs placed at 150°C show surface discoloration, changing from the original olive green color to light or dark brown.

PAHs with Olivine in Water
Hydrothermal PAH alteration experiments with olivine show that at 21°C naphthalene, fluoranthene and coronene crystals remain unaffected and do not exhibit any macroscopic alteration features. The added olivine crystals from experiments with PAHs at 21°C also do not show signs of discoloration suggesting an alteration.
At 150°C, macroscopically visible naphthalene and fluoranthene crystals disappear, and a dark colored layer largely covers the olivine crystal surfaces (see Figs. 2d and 2f). In contrast, in experiments with coronene, crystals could still be observed lying on the olivine surfaces without any visible signs of alteration. The olivine slabs themselves display an intense discoloration to a dark brown color.
The most notable feature of all naphthalene and fluoranthene experiments at 150°C is the occurrence of nm-to lm-sized spherules seen on top of the olivine substrate in the SEM (Figs. 3 and 4). These spherules are distributed across the entire olivine slab surface and display a variety of morphologies. However, in experiments with PAHs without olivine and coronene with olivine, no spherules were seen.

Carbonaceous Spherules
Spherules formed during naphthalene-olivine alteration experiments at 150°C range in diameter from 0.04 to 0.47 lm and are arranged to form 120°interface angles. The spherule size distribution is in agreement with a lognormal cumulative probability distribution (Fig. 5c) and differed only minimally between duplicate experiments (compare Figs. 3a and 3b). In some minor cases, spherule-spherule interactions led to the distortion of the spherical shape. Cross-sectional imaging of the spherules using FIB-SEM-assisted nano-milling reveals that the naphthalene-derived spherules are homogenously solid and do not exhibit any distinctive internal microstructures (Figs. 3c and 3d). This is in contrast to spherules formed during the fluoranthene-olivine experiments at 150°C (Figs. 4a, 4g, and 4i). Spherules formed during these experiments are substantially larger and range in size from 20 to 50 lm. However, several smaller spherules (0.5-5 lm) can be found surrounding the largest spherules (Fig. 4a). The size distribution of these spherules reflects an asymptotic size cumulative probability distribution (Fig. 5d). In addition, plateshaped organic material can be found to form either directly on the olivine surface or the surface of large spherules (Figs. 4a and 4d). FIB-SEM cross-sectioning revealed a variety of internal microstructures. The shape of the largest spherules arises due to a double layer structure that homogeneously surrounds a more complex internal structure. The outer layer has a thickness of around 0.03 lm and the inner layer from around 0.06 lm. The internal structure displays a complex arrangement of isolated pockets that exhibits precipitates, which resemble a dendritic structure (Figs. 4b,4c, and 4f). The dendrites vary in thickness and appear to start randomly from the inner rim. In contrast, the intermediatesized spherules with a size of~5 lm and those smaller than 5 lm are internally hollow (Figs. 4g, 4h, and 4i).

Chemical Analysis
The microscopic alteration behavior of naphthalene on an olivine slab at 21°C could not be investigated due to the quick evaporation of naphthalene in the SEM vacuum chamber. Although analysis of naphthalene from the LTHE at 21°C was not possible with the SEM, Raman spectra could be obtained from duplicate samples. The obtained spectra are identical to those of initial naphthalene crystals ( Fig. 2; Table 4).
All spherules formed at 150°C in LTHE with naphthalene appear to have a lower vapor pressure than the starting material. Naphthalene visibly evaporated within hours when exposed to air in the laboratory or within minutes in the SEM vacuum, whereas the newly formed spherules showed no change in appearance even up to a year after the experiment. Similarly, the original fluoranthene material evaporated within hours of exposure to air in the laboratory, but the spherules produced during the fluoranthene experiments did not exhibit any changes in size or external appearance. The EDX spectra of the spherules show only C, although H may also be present in the material as this is not detectable using SEM-EDX.
Olivine slabs from LTHE at 150°C in water with and without the presence of PAHs experienced a color change. SEM-EDX and Raman spectroscopic analysis indicate that the different color is caused by the presence of hematite (Fe 2 O 3 ; Fig. 2o). Figure 2 shows the Raman spectra of naphthalene, fluoranthene, and coronene in comparison to their initial spectra, and in Table 4, the determined band positions with full width half maximum values (FWHM) are summarized. Crystals from LTHE with coronene at 150°C in water with and without the presence of olivine indicate no signs of chemical alteration, which is in agreement with the chemical analysis of samples from LTHE experiments at 21°C ( Fig. 2i; Table 4). However, fluoranthene and naphthalene from LTHE at 150°C in water showed evidence for chemical changes in the Raman spectra (Figs. 2a and 2b; Table 4), linked to their transformation from crystals to spherules, both in the presence of olivine, but, also in the absence of olivine. Unfortunately, fluorescence was often a problem when analyzing the Fig. 2. Raman spectra and corresponding LTHE samples. a, b) Raman spectra of naphthalene and fluoranthene exposed to water and different temperatures (21°C or 150°C) in comparison with initial spectra. There are two spectra each of naphthalene and fluoranthene samples at 150°C. c) Olivine slice with naphthalene exposed to water at 21°C in LTHE. d) Olivine slice with naphthalene exposed to water at 150°C in LTHE. e) Olivine slice with fluoranthene exposed to water at 150°C in LTHE. f) Olivine slice with fluoranthene exposed to water at 150°C in LTHE. g) SEM image of globules formed on olivine slice in LTHE with fluoranthene (e). The numbers (1; 2) in the image correspond to Raman spectra of fluoranthene exposed to water at 150°C in LTHE (b). h) Microscopy image of globules formed on olivine slice in LTHE with fluoranthene (e). The numbers (1; 2) in the image correspond to Raman spectra of fluoranthene exposed to water at 150°C in LTHE (b). i) Raman spectra of coronene after exposed to water and different temperatures (21°C or 150°C) in comparison with initial spectra. j) Raman spectra of olivine after exposure to water and different temperatures (21 or 150°C) in comparison with initial spectra. k) Olivine slice with coronene exposed to water at 21°C in LTHE. l) Olivine slice with coronene exposed to water at 150°C in LTHE. m) Olivine slice to water at 21°C in LTHE. n) Olivine slice exposed to water at 150°C in LTHE. o) Zoom into olivine spectra exposed to water at 150°C in LTHE (j). The zoom-in shows a spectrum with a hematite peak. (Color figure can be viewed at wileyonlinelib rary.com.) spherules even with reduced laser power resulting in only a few useable spectra being obtained from these samples. The new Raman modes of the spherules measured in the naphthalene LTHE at 150°C correspond to C=C stretching (1586 cm À1 , 1596 cm À1 , 1799 cm À1 ) or C-C ring vibration (1324 cm À1 ), thus can still be assigned to an aromatic compound (Lin-Vien et al. 1991). This is consistent with an absence of elements such as O or N in the EDX spectra from the SEM, which would indicate the incorporation of a new functional group into the PAH due to aqueous alteration. The Raman bands 823 cm À1 and 855 cm À1 to 856 cm À1 belong to the underlying olivine sample. The FWHM mean values of the new bands from LTHE with naphthalene at 150°C are up to 24 times broader in comparison to the initial naphthalene crystals and the bands from LTHE at 21°C. The fluoranthene spectra still contain a Raman band that can be assigned to C-C stretching (1102 cm À1 ; Lin-Vien et al. 1991) of the initial fluoranthene spectrum. The FWHM mean value of the resolvable band from LTHE with fluoranthene at 150°C is two (150-2) to four (150-1) times broader in comparison to the initial fluoranthene spectrum. The Raman spectra do not directly correlate with specific, cataloged organic molecules, such as functionalized or other PAHs (e.g., Ram et al. 1983;Colangeli et al. 1992;Schaeffer et al. 1995;Shinohara et al. 1998;Alajtal et al. 2010;Tan et al. 2017). Therefore, we were not able to identify any specific reaction products based on the Raman spectra. The broadening of the Raman bands centered at the same wave number as the original PAH could indicate a loss of crystallinity (similar to, e.g., metamictization in zircons; Nasdala et al. 1995) consistent with the formation of different morphology and change in vapor pressure of the PAH solids in these experiments. Alternatively, the broad bands could be an agglomeration of many, smaller, sharper bands that lie at similar wave numbers representing a mixture of molecules with similar chemical groups. Figure 6 summarizes the macroscopic observations of all STHE. In the experiments with naphthalene powder, the majority of the powder was observed to float at the water-gas interface at 21°C. Only a small amount of crystals settled at the bottom of the glass vessel. Increasing the temperature to 70°C, just below the melting point (see Table 2) of naphthalene resulted in no visible change in the naphthalene powder or the aqueous phase. However, increasing the temperature to 90°C, 10°C above the melting point of naphthalene caused an apparent visible change in naphthalene's behavior. Naphthalene crystals were no longer observed at the air-water interface. Instead, an immiscible fluid was observed floating at the air-water interface, and aggregated 2-3 mm sized droplets were found within the fluid at the bottom of the experimental vessel. Increasing the temperature to 100°C, 20°C above the melting point, resulting in the occurrence of 1 mm droplets found floating at the water-gas interface and the overall solution becoming cloudy. Just a few droplets formed at lower temperatures were still observable at the bottom of the experimental vessel. Above 120°C, the size of droplets decreases with rising temperature until the droplets were no longer visible by eye, the concave meniscus disappeared, and only a cloudy emulsion remained. The cloudy emulsion persisted during the temperature increase. At 150°C, 70°C above the melting point of naphthalene, the fluid became clear again without any indication of other phases present within the vessel.

Macroscopic Observations
Short-term experiments conducted with fluoranthene showed similar behavior as the naphthalene experiments when the higher melting point of fluoranthene . LTHE with fluoranthene on solid olivine at 150°C. A variety of carbonaceous globules formed in LTHE with fluoranthene and olivine in water at 150°C. a, g, and i) represent three types of globules formed in one experiment. These three types differ in both size and surface area density. a) An example of the largest globules (15-50 lm). This type of globule has a distinctive internal structure, as shown in (b), (e), and (f). b) A three-dimensional reconstruction of the internal morphology of a globule, built up from slices as shown in (c), revealing the presence of carbonaceous dendrites and a carbonaceous skeleton. d) The area where a FIB cross section was made. e and f) Magnified views into the cross section showing similar internal structure as shown in (b). g and h) Two medium-sized (Ø 5 lm) hollow globules. i) The smallest globules with a deflated appearance. (Color figure can be viewed at wileyonlinelibrary.com.) 109°C) was exceeded (Table 2). Before the melting point, fluoranthene crystals did not change their appearance. At 120°C, 10°C above the melting point, an immiscible fluid was formed at the air-water interface as well as aggregated 2-3 mm sized spherules. At 140°C, 30°C above the melting point, no spherules were visible in the fluid floating at the fluid-air interface or the bottom of the experiment vessel. Instead, only the immiscible fluid at the fluid-air interface, forming a concave meniscus, was present and the rest of the fluid in the vessel appeared cloudy. At 150°C, 40°C above the melting point the fluid remained cloudy with the immiscible fluid floating at the air-water interface. However, after the solution cooled, the fluid becomes clear again, without any indication of other phases present within the fluid. The immiscible fluid at the air-water interface solidified into a very thin layer on top of the aqueous phase.
In water-absent experiments with naphthalene and fluoranthene under an argon atmosphere, PAHs were observed to form a melt immediately around the PAH melting point. Before melting, the naphthalene powder was present as an aggregated lump, whereas the fluoranthene remained as individual crystals. After the melting point of naphthalene and fluoranthene was exceeded, both PAHs formed a single fluid layer on the bottom of the glass vials.
In the coronene experiments both with and without anoxic water, no difference in the appearance of the coronene crystals or the water was observed within the temperature range tested (21-150°C).

Ex situ Observations
After the STHE, the remaining visible residues of naphthalene and fluoranthene were present as a floating phase and spherules located on the bottom of the experimental vessel. Upon quenching, the floating PAH (naphthalene or fluoranthene) liquid phase solidified and sank. As a result, we were unable to separate the PAH (naphthalene or fluoranthene) material from the air-water interface and the bottom of the glass reaction vessel. SEM images (Figs. 7 and 8) were obtained for a representative suite of the solid residues retrieved from the STHE with naphthalene and fluoranthene. In the STHE coronene experiments, coronene crystals sank to    Maddams and Royaud (1990). c Raman bands of olivine. nanosized spherules aligned to form a string-like network. In contrast to the highly evaporative nature of the naphthalene-derived material, STHE conducted with fluoranthene resulted in carbonaceous material that was stable when subjected to high vacuum conditions. Figure 8 displays a summary of fluoranthene-derived material from experiments ranging between 100 and 150°C. A few micron-sized cavities developed within the fluoranthene at 100°C (Fig. 8a), 10°C below the melting point of fluoranthene (Table 2). These cavities are distributed within the fluoranthene material, and the walls of these cavities occasionally show spherule-like structures. Around 120°C, the cavities are enlarged with a diameter around 1-3 lm and appear to have more spherules on the walls (Fig. 8b). At 140°C, 30°C above the fluoranthene melting point, ring-shaped organic structures develop within small cavities (Figs. 8c-e). Sample material of fluoranthene from STHE conducted at 150°C shows large cavities filled with elongated, rounded, pillar-like material (Fig. 8f).

Chemical Analysis
The residues from the STHE samples were analyzed with SEM-EDX and Raman spectroscopy. The SEM-EDX measurement of the naphthalene experiments was only possible for the tiny stable fraction that did not evaporate immediately in the SEM vacuum chamber.
For naphthalene, fluoranthene, and coronene samples from the STHE, no apparent chemical changes were observed. As for the LTHE, only carbon was measured in the spherules from the STHE with the SEM-EDX. H may also have been present as this is not detectable using EDX. The Raman spectroscopic investigation of naphthalene and coronene samples (see Figure S1 in supporting information) showed no noteworthy change in the Raman band position or FWHM values, indicating no chemical alteration. Fluoranthene samples of the STHE had to be measured by Raman spectroscopy with lower laser intensity to minimize fluorescence (see supporting information) and also showed no indication of alteration. Only samples of naphthalene and fluoranthene up to 120°C could be measured. Thus, in contrast to the LTHE samples at 150°C, PAHs from the STHE samples show no spectroscopic evidence for chemical alteration.

Aqueous Alteration of PAHs Under Asteroidal Conditions
Large PAHs from the ISM are expected to have been incorporated into asteroids and should be found in meteorites (Plows et al. 2003). However, only small Fig. 6. Macroscopic observations of short-term hydrothermal experiments of naphthalene and fluoranthene with water. Samples were exposed to a temperature profile ranging from room temperature (21°C) to LTHE temperature (150°C). Melting points of naphthalene (~80°C) and fluoranthene (~110°C) fall within this temperature range (Karcher 1988;Yalkowsky et al. 2010). Both naphthalene and fluoranthene develop a concave meniscus floating on the surface at around 10°C above their respective melting points. Upon further heating, these menisci shrink, and completely disappear around 30°C above the melting point for naphthalene, and visibly reduce in size in the fluoranthene experiments around 40°C above the melting point. At 70°C above the melting point, naphthalene is no longer visible with the naked eye, leaving a clear fluid.
PAHs have been found in CCs thus far. In part of our study, we also tested if olivine alteration has an effect on PAHs (e.g., catalyzed breakdown). Alteration of olivine in water can take place within a wide temperature range (Evans 2004), including the aqueous alteration temperatures expected in asteroids (Brearley 2006). It is known that the alteration of olivine supports the alteration of OM in water (Sleep et al. 2004). However, our experiments show that there is no clear link between the aqueous alteration of olivine and the alteration of the tested PAHs.
Our experiments show that, within the aqueous alteration conditions predicted for CCs (Brearley 2006), breakdown of large PAHs into small ones by interactions with water is also unlikely. However, we do observe a chemical change of PAHs near the melting point within the aqueous alteration temperatures predicted for CC. The chemical alteration that occurs via melting was possibly promoted by solvation in water followed by chemical reactions assisted by protonation. Further studies are required to identify the chemical routes and molecular intermediaries involved.
There may be other explanations for the absence of large PAHs in CCs. Large PAHs could have already experienced destruction at a very early stage of their accretion into the parent body (Anderson et al. 2017). Alternatively, large PAHs might still be present as insoluble matter (kerogen) in asteroids and meteorites, respectively (Vinogradoff et al. 2018). In either of these cases, there may be mechanisms for the generation of smaller PAHs that we have not yet discovered, which should be explored in future research.

Organic Spherule Formation by PAHs
PAH derivatives have been suggested to be important in the stabilization of membranous vesicles in aqueous solutions in meteorites (Deamer et al. 2002). Spherule formation from PAHs appears to occur independently of changing chemical environments due to mineral reactivity, modeled using olivine in these experiments. To our knowledge, no published investigation has demonstrated the production of organic spherules by PAHs alone in the presence of aqueous fluids. A successful synthesis of PAH globule structures was reported in the literature during plasma deposition of benzene (C 6 H 6 ) and anthracene (C 14 H 10 ) (Saito and Kimura 2009). Thus, the results of our high-temperature PAH alteration experiments also indicate that spherule formation can occur through the melting of PAHs in water depending on the system temperature.
Our short-term experiments show that spherule formation only occurs in water above specific PAHdependent temperatures, that is, 90°C for naphthalene and 120°C for fluoranthene. These temperatures correlate with the melting temperatures of the PAHs Fig. 7. SEM-FIB images of residual solid material in naphthalene STHEs from 80 to 120°C. a-d) Carbonaceous globules formed in STHE at 80°C, 90°C, 100°C, and 120°C, respectively. All four images show string-like material. The string-like material is composed of clusters of carbonaceous globules. The carbonaceous globules show consistent sizes at all investigated temperatures. (Chanyshev et al. 2017), 80-82°C for naphthalene and 107-110°C for fluoranthene (Table 2). This observation is also in agreement with the absence of spherule formation in the coronene experiments, as coronene has a melting temperature of 428-440°C (Chanyshev et al. 2017) (Table 2), far above the highest temperature tested during our experiments.
PAHs are only sparingly soluble in water at room temperature, but their solubility has been shown to increase with increasing temperature (Kus 2012). The PAHs in our experiments are not expected to dissolve significantly into aqueous solutions as the amount of PAH weighed into each sample vessel exceeds the solubility limits (Table 2), even at elevated temperatures (naphthalene: 0.244 g L À1 at 73°C and fluoranthene: 0.017 g L À1 at 60°C; Yalkowsky et al. 2010). An experimental study of benzo(a)pyrene in aqueous solution by Fang and Kozi nski (2001) demonstrates that even as a melt, PAHs remain insoluble in the aqueous solution until the PAH-water solution becomes supercritical (above the triple point of water). Our experiments are below the supercritical point of water (374°C; Sengers 2000); thus, the second fluid phase observed in the naphthalene and fluoranthene STHEs is Fig. 8. SEM-FIB images of residual solid material from fluoranthene STHEs, 100-140°C. a) Solid material formed at 100°C containing a small cavity at the right lower corner. b) Solid material formed at 120°C containing several cavities. c-f) Internal structures of cavities formed at 140 and 150°C, respectively, where the cavity from 150°C seems to be filled with elongated rounded pillar-like material (f). d and e) Close-up images of a wall within the cavities formed at 140°C. Several round shape structures are visible. interpreted as molten PAH. A fraction of naphthalene might also have been lost by sublimation at the air-fluid interface in the experiments above 120°C in STHE, as an increase in the partial pressure of naphthalene at 150°C is expected (Fowler et al. 1968;Mackay and Shiu 1977;Macknick and Prausnitz 1979).
The formation of an emulsion, that is, an immiscible PAH melt, in an aqueous solution, helps to explain the formation of the spherule shape in our experiments. For example, crystallization experiments of the organic compound ethylene glycol distearate (C 38 H 74 O 4 ; Khalil et al. 2011), which also has low aqueous solubility, form an emulsion with water when molten. These experiments also revealed that solid material often retains the shape of the melt droplets after cooling in an unstirred solution (Khalil et al. 2011). The microstructure produced by the crystallization process of ethylene glycol distearate has parallels with the internal structures observed in the organic spherules produced here. With the fastest cooling rate, ethylene glycol distearate crystallized as dendritic structures in large droplets. Thereby, nucleation occurred both on the interfacial surface between the melt and aqueous solution as well as throughout the droplet bulk. Within the largest droplets produced by fluoranthene in the LTHE (Fig. 4a) and sample material from STHE at 150°C (Fig. 8), we also observed complex dendrite-like internal structures, indicating a similar formation mechanism as the ethylene glycol distearate dendrites.
In contrast, in the ethylene glycol distearate experiments of Khalil et al. (2011), smaller droplets retained their original size and shape throughout the crystallization process with crystallization only occurring at the droplet edge. Nucleation and growth from the droplet edge in our experiments would explain the hollow and filled structures found in the smaller spherules from the fluoranthene LTHE (Figs. 4g and 4i) as well as the smaller spherules generated in the naphthalene LTHE (Fig. 3). The spherule sizes formed in the naphthalene LTHE at 150°C show a log-normal distribution (Fig. 5c) indicating homogenous growth (Kile et al. 2000). However, the spherules formed during the LTHE with fluoranthene at 150°C show an asymptotic size distribution (Fig. 5d). This distribution indicates heterogeneous growth, again consistent with the findings of Khalil et al. (2011) who demonstrate that growth occurs initially in larger droplets of ethylene glycol distearate, with a larger extent of undercooling (temperature below the melting point of ethylene glycol distearate) necessary to initiate crystallization in smaller droplets. The formation of spherules appears to be a common characteristic of insoluble OM, like PAHs, in an aqueous solution after heating above the melting temperature as seen in our results and experiments from Khalil et al. (2011).

Organic Globules in CCs
The first micrometer-sized spherical shapes discovered, hereafter referred to as globules, were found in the Orgueil and Ivuna meteorites and initially interpreted to have been microfossils indigenous to the meteorite, potentially formed by algae (Claus and Nagy 1961) or bacteria (Or o and Tornabene 1965). Today, globules have been found in materials from various environments in the solar system, including interplanetary dust particles (IDPs; Messenger et al. 2008;Busemann et al. 2009), cometary dust (Matrajt et al. 2008;De Gregorio et al. 2010), and as minor components of the insoluble organic fraction in several CCs (Vanlandingham et al. 1967;Rossignol-Strick and Barghoorn 1971;Pflug 1984;Buseck 2004, 2006;Nakamura-Messenger et al. 2006;Garvie et al. 2008;Messenger et al. 2008;Clemett et al. 2009;De Gregorio et al. 2010Herd et al. 2011;Hashiguchi et al. 2013;Jones 2016;Alexander et al. 2017;Changela et al. 2018).
The globules described in the literature have a diameter size ranging from as small as 50 nm Buseck 2004) up to 340 nm (Rossignol-Strick andBarghoorn 1971). This distribution is replicated in both the spherules from the LTHE and STHE (Figs. 3,4,7,and 8) where the largest spherule observed was 50 lm but with the majority of spherules being between 100 and 500 nm. A variety of structures have been observed in CC globules such as double walls (Claus and Nagy 1961;Vanlandingham et al. 1967;Rossignol-Strick and Barghoorn 1971;Pflug 1984) as well as solid or hollow nanospheres (Nakamura et al. 2002;Garvie and Buseck 2004;Nakamura-Messenger et al. 2006;Garvie et al. 2008;Hashiguchi et al. 2013). These structures are also mirrored in the spherules produced in our experiments, particularly in the LTHE with fluoranthene at 150°C, which produced hollow, filled, and complex structures depending on the spherule size (Figs. 4b,4c,and 4f). For globules with internal cavities, the wall thickness is reported to be between 2 nm (Pflug 1984) and 100-200 nm (Nakamura et al. 2002;Hashiguchi et al. 2013;Nakamura-Messenger et al. 2013) in CCs. In comparison, the spherule wall thickness produced in our experiments (200 nm on average) fits with the thicknesses found in natural materials, albeit at the smaller end of the size range. Thus, morphologically, the experimentally produced spherules reported here closely resemble the microstructural characteristics of many globules found in CCs indicating that melting of PAHs in water is a potential mechanism for the formation of some organic globules within asteroids.
CCs contain a variety of organic compounds (Sephton et al. 1998;Sephton 2002) spanning a range in solubility and melting temperatures. Here, we recall that the behavior of naphthalene and fluoranthene in our experiments is characteristic for all organic compounds with limited solubility that are heated above their melting temperature. We expect that such a complex mixture would give rise to a varied composition of the globules formed after heating. We also recognize that different regions of the meteorites may have been heated to different temperatures, and the resulting globules may have been transported in the aqueous environment. We surmise, therefore, that melting and droplet formation may well have played a major role in the formation of globules in the meteoritic parent body. Microspherules have also been produced from other starting materials, such as formaldehyde polymerization . Therefore, the contribution of melting versus the formation of micelles in aqueous solution, possibly stabilized by PAH derivatives (Deamer et al. 2002), will have to be investigated in future experiments.
The results of the LTHE show that the PAHs have been chemically altered during spherule formation under the chosen experimental conditions (T = 150°C, P = 4.8 bar). This is indicated by the lower vapor pressure of the globule material in air and vacuum in comparison to the starting PAH powders. The change of fluid color in the LTHE samples and the appearance of new Raman bands also indicate that a chemical reaction has taken place. The new Raman bands can still be assigned to aromatic modes, but could not be matched with specific organic molecules. The chemical change may reflect the high proton affinity of PAHs (Chen and Cooks 1995), as compared to H 2 O, and efficient formation of protonated PAHs. The charge may assist in polymerization reactions converting small PAHs into larger ones, as well as in cracking reactions and hydrogen transfer (Chanyshev et al. 2017). The results of the STHE do not show the same indications for chemical alteration during spherule formation, which might be explained due to the chosen experimental conditions, including much shorter experiment duration and by not being able to analyze the samples formed at temperatures above 120°C. However, the samples indicate similar higher stability (high vapor pressure) to that observed from the samples of the LTHE. Therefore, we cannot rule out that chemical changes took place in the STHE samples at 150°C.
Globule populations in CCs are dominated by organics with a complex mixture of aliphatic and aromatic chemistries. These globules also carry variations in D and 15 N isotopic enrichments (e.g., Nakamura et al. 2002;Buseck 2004, 2006;De Gregorio et al. 2010). Experiments with D-enriched ice have shown that PAHs can attain isotopic signatures comparable to those observed in meteorites (Sandford et al. 2000). Furthermore, variations of D-enriched PAHs have been detected in several regions of the galaxy, including signatures reflecting D-enriched PAHs in CCs (Peeters et al. 2004;Doney et al. 2016). Our hypothesis may help to explain how D enrichments are retained as the formation of a hydrophobic melt directly from a solid would limit D/H isotopic exchange between the OM and aqueous solution. Based on our hypothesis, PAHs with a 15 N anomaly could have also retained their isotopic signature during globule formation from sparingly soluble material, such as PAHs, via an emulsion in aqueous solution. So far, however, no single formation mechanism has been proposed to explain all of the observed characteristics of these globules, suggesting that their properties are a result of multiple processes occurring in a variety of environments (Alexander et al. 1998;De Gregorio et al. 2013). In this regard, we emphasize that our proposed mechanism for the formation of globules with a PAH chemistry is linked to aqueous environments and hence does not address globule formation in solar system materials that have not seen liquid water such as comets and IDPs.
A Potential Link Between Globules and PAH Solubility Nakamura-Messenger et al. (2006) proposed that aqueous alteration in the parent body of the Tagish Lake meteorite might be a possible mechanism for carbonaceous globule formation. It is evident from our experiments that PAHs in aqueous solution can form solid spherules when an emulsion is formed and that these spherules can replicate the size and internal structure of globules that have been found in CCs. The aqueous alteration would have occurred at elevated temperatures due to 26 Al decay (Urey 1955;Fuse and Anders 1969). Figure 9 shows that the estimated aqueous alteration conditions for different CC meteorite groups overlain with the melting temperatures of different PAH molecules based on the number of their C atoms. This plot demonstrates that all of the meteorite CC groups, with the exception of CO meteorites, are expected to have seen aqueous alteration above the melting temperatures of naphthalene and fluoranthene. However, for such a mechanism to be active on a meteorite parent body, one needs to evaluate whether there will be sufficient concentrations of PAHs to surpass the solubility limit under the conditions of aqueous alteration. Taking the Allende meteorite as an example, with a water content of 0.2 wt% (Jarosewich et al. 1987) and 0.055-2 ppm naphthalene (Zenobi et al. 1989), the naphthalene concentration ranges between 10 À5 and 10 À3 mg g À1 . This is below naphthalene's solubility in water (0.031 g L À1 ; Yalkowsky et al. 2010).
On Earth, PAHs are preferentially partitioned onto solid particles due to their low solubility (Wilcke 2000). Similarly, PAHs in meteorite parent bodies are expected to be situated on/at mineral surfaces/interfaces. This is consistent with observations that PAHs are not homogeneously distributed within meteorites, as indicated by Zenobi et al. (1989). Heating may also have been localized (De Gregorio et al. 2010) causing an uneven distribution of water within the parent body (Brearley 2006). Dissolution into smaller fluid reservoirs would result in higher PAH concentrations in solution. Given the very low solubilities of PAHs, accumulation of PAHs into smaller reservoirs should thus allow PAH aggregates to surpass their aqueous solubility thus facilitating melt formation within these reservoirs.
PAH solubility decreases with increasing salinity (Gold and Rodriguez 1989) but increases with increasing temperature (Yalkowsky et al. 2010). Thus, the maximum concentration required for PAHs to surpass their saturation would be dependent on the alteration conditions. Although the elevated temperatures during alteration would increase PAH solubility, the extensive mineral alteration products (e.g., salts) indicate that ions should be present in the aqueous solution, which could have a counter effect. The composition of the solution does not appear to influence PAH dissolution (Giese et al. 2018); hence, increasing ionic strength would thus compensate for increasing solubility at higher temperatures, again facilitating the formation of globules via melting. Therefore, it seems that globule formation by PAH melting under asteroidal conditions is likely and could contribute to the variety of globules found in CCs.
The main result of our study is that naphthalene and fluoranthene produced spherules that are strikingly similar in shape, internal structure, and size to many globules found in CCs. In contrast, in the experiments with coronene, no spherules were formed. We considered a correlation between spherule formation and melting temperature of the PAHs. Based on this correlation, we suggest that the formation of an emulsion of small PAHs and the aqueous solution is a globule-forming mechanism. This is consistent with the overlap of PAH melting temperatures with predicted aqueous alteration conditions for CCs. The melting points of naphthalene and fluoranthene lie within the aqueous alteration conditions of CCs, whereas that of coronene lies outside these conditions. Thus, larger PAHs would not be expected to melt and form spherules. Khalil et al. (2011) show in their experiments that the formation of spherules seems to be a common characteristic of organic material with very low solubility in an aqueous solution after heating above the melting temperature.
Further studies will have to be performed to address the characteristics of globules formed from complex organic mixtures. Our proposed hypothesis for globule formation could also be translated to other OM as long as they are above the solution saturation limit and have similar chemical and physical properties to PAHs. The effect of PAHs with altered or more complex chemistries, reflecting the composition of the observed globules in CCs, also needs to be explored as changes in solubility and melting point due to differences in chemistry will play an important role in the formation of globules through melting. It is not yet clear whether the melting mechanism is present in nature. Globule structures can be generated prior to the meteorite parent body formation, as proposed by Jones (2016); however, whether these globules can be retained is unclear. Similarly, spherules have also been produced experimentally in aqueous solution from formaldehyde polymerization . Spatially resolved, in situ investigations of PAH concentration variability in CCs will help to establish whether the PAHs can exceed their solubility limit during aqueous alteration on different parent bodies, or if they can be completely dissolved into the aqueous solution as suggested by Wing and Bada (1991). Our proposed mechanism for the formation of globules with a PAH chemistry may explain how D and 15 N-enrichments can be retained in globules if isotopic signatures are imprinted prior to parent body formation. However, the effect of melting on the isotopic signatures associated with the globules warrants further investigation.