Prevalence of non-aromatic carbonaceous molecules in the inner regions of circumstellar envelopes

Abstract

Evolved stars are a foundry of chemical complexity, gas and dust that provides the building blocks of planets and life, and dust nucleation first occurs in their photosphere. Despite their importance, the circumstellar regions enveloping these stars remain hidden to many observations, thus dust formation processes are still poorly understood. Laboratory astrophysics provides complementary routes to unveil these chemical processes, but most experiments rely on combustion or plasma decomposition of molecular precursors under physical conditions far removed from those in space. We have built an ultra-high vacuum machine combining atomic gas aggregation with advanced in-situ characterization techniques to reproduce and characterize the bottom-up dust formation process. We show that carbonaceous dust analogues formed from low-pressure gas-phase condensation of C atoms in a hydrogen atmosphere, in a C/H₂ ratio similar to that reported for evolved stars, leads to the formation of amorphous C nanograins and aliphatic C-clusters. Aromatic species or fullerenes do not form effectively under these conditions, raising implications for the revision of the chemical mechanisms taking place in circumstellar envelopes.

Stars, like our Sun, in their final stages of evolution reach the asymptotic giant branch (AGB) accompanied by a massive ejection of matter containing the vast chemical complexity that provides the building blocks of planets and the main ingredients necessary for the emergence of life^1–3. The advent of powerful radio-telescopes like ALMA, with improved radial resolution and sensitivity, opens the possibility to outline the basic physical and chemical conditions in which stardust is formed in the circumstellar envelopes (CSE)^4–6. This dust, which consists of silicates or carbonaceous material depending on the C/O elemental abundance ratio of the AGB star, will then be ejected into the interstellar medium (ISM) where, after processing, it becomes involved in the formation of new stars and planets. However, the chemical nature of the dust seeds and carbonaceous molecules produced by C-rich evolved stars, as well as the chemical routes leading to them, is still controversial.

The dust seeds are first generated in the vicinity of the photosphere, at about 1-2 stellar radii⁷. In these regions, carbon nucleates into nanometer-sized grains in an equilibrium process with a low degree of ionization due to the absence of energetic radiation. The nucleation zone of an AGB star can be regarded as a homogeneous dense atomic cloud in the photosphere that, contrary to the outer layers, is devoid of UV photons. Physical parameters in nucleation zones remain uncertain because they are difficult to derive from observations⁴. Therefore, understanding the formation and nature of carbon stardust is complex and cannot be addressed solely by astronomical observations; dedicated laboratory simulations are required ^8–15. However, most experiments are unable to approach the relevant physical conditions, since they rely on energetic plasmas, combustion or decomposition of molecular precursors under conditions very unlike those in the dust formation regions of CSE⁸.

A related issue is the formation of large carbonaceous molecules, polycyclic aromatic hydrocarbons (PAHs) and fullerenes. PAHs, identified by the emission of aromatic infrared bands (AIBs)¹⁶, are widespread in regions of massive star formation and in C-rich protoplanetary and planetary nebulae. Buckminsterfullerene C₆₀ has been detected in some of these environments^6,17,18,19. A robust explanation of how these large molecules form is still lacking, especially in evolved stars where AIBs are only detected at late stages when the star evolves from red giant to white dwarf emitting ultraviolet (UV) radiation²⁰.

In this work, using a specifically designed ultra-high vacuum (UHV) experimental setup²¹ we investigate the basic carbonaceous chemistry taking place in the dust formation region of CSEs. Our approach produces carbon dust seeds using exclusively gas-phase C atoms and molecular hydrogen in a ratio close to that in the atmospheres of AGB stars without the requirement for introducing ad hoc C-bearing molecular precursors, such as acetylene or methane. Under these conditions, we show that nanosized carbon particles, pure carbon clusters and aliphatic carbon species are efficiently formed, whereas aromatics are only present at trace levels and no fullerenes are detected. Our results reproduce the relative abundances of simple molecules detected in CSEs, such as C₂H₂ and C₂H₄, which are directly generated by bottom-up chemistry and are simulated by a simple chemical kinetics model. Our data suggest that the pressures in circumstellar condensation environments are not sufficiently high for initiating cyclization mechanisms²² and the formation of aromatic species must occur via other processes. We suggest that one pathway could be through thermal processing of aliphatic material on the surface of dust.

Results

Ultra-high vacuum gas aggregation sources for mimicking the dust formation region of CSEs

Our experimental set-up, called Stardust²¹, is an UHV machine specifically conceived to simulate, with a high level of control, the complex conditions of stardust formation and processing in the environment of evolved stars. It uses UHV technology to avoid the high reactivity expected to atmospheric and residual gases of the formed carbon nanostructures²³. A more detailed technical description is provided in Supplementary Information 1.

Many different techniques have been proposed to make dust analogues, each with advantages and limitations. In Supplementary table 1 we present a detailed critical compilation of the most commonly used techniques. Among them, laser ablation is one of the best suited ¹⁰, for instance, to simulate the reformation mechanisms of cosmic grains in the ISM¹⁵. In Stardust, we have chosen a sputter gas aggregation source (SGAS) because it is able to generate small clusters of nano-sized particles by gas-phase aggregation of individual atoms in a weakly ionized environment²⁴. It is based on the vaporization of C atoms from a graphite target using Ar as the sputtering gas. Contrary to conventional standard magnetron sputtering ^26,27, in SGAS nanoparticles and clusters form from gas phase reactions, as the sputtered material is trapped in the aggregation zone of the machine. Gas temperature during aggregation can be estimated to be < 1000 K²⁴, in the range of the condensation temperatures of solids in CSEs (500-2000 K)^7,25. Thus, the chemistry proceeds via atom aggregation under conditions in which most of the reactions that occur are neutral-neutral, bimolecular and termolecular, closely resembling what happens in the dust formation region of CSEs.

In this work we unveil the formation of carbonaceous dust grains starting with a gas mixture containing the two most relevant species: carbon and molecular hydrogen. In carbon-rich CSEs, oxygen is locked into CO and SiO and therefore is not available to form dust ^28,29. Fig. 1 shows a pictorial representation of the CSE where typical carbon and hydrogen densities are provided. Supplementary table 2 further compares the conditions achieved in the gas aggregation zone with those in the dust formation zone of CSEs. Fig. 1b presents a scheme of Stardust in the configuration used for these experiments, which start with the vaporization of C atoms, followed by the injection of H₂. The C atoms react with H₂ and are subsequently dragged through a nozzle by differential pressure into the next chambers, where they are either analysed in gas phase or collected on inert surfaces for further analysis. We have used different chemically-inert substrates to collect the carbonaceous stardust analogues for further studies, including Au, SiO₂ or graphite. The results we show hereafter are unaffected by the chosen substrate. Once the formed species pass the nozzle of the first chamber, no further growth takes place in the expansion.

Fig. 1. Production of stardust analogues.

Upper part: pictorial representation of the CSE of a C rich AGB star, with typical carbon and hydrogen densities at given distances provided according to reference²⁸. Lower part: Schematic of the Stardust machine configuration used in these experiments, considering typical C and H₂ densities. From left to right: there are three different vacuum environments: aggregation zone, where OES measurements are performed and H₂ introduced; expansion area, where the quadrupole mass spectrometer (QMS) is placed and we collect samples for ex situ studies (sample 1 position), and analysis area (sample 2), where thermal programmed desorption (TPD) is performed.

The initial gas-phase vaporized C atoms may interact with residual or injected molecular hydrogen at different densities, and there are no other species involved except Ar (the residual density before magnetron operation, excluding Ar and H₂, is about 10⁷ cm^-3). The vaporized C atoms have a density of about 2.5x10¹⁰ atoms·cm^-3, estimated from the sputtering energy (200 eV), the total drag current (200 mA) and the yield of emitted C atoms per impinging Ar⁺ ion³⁰. Under these conditions, sputtering of individual C atoms is preferential over C₂ and C₃³¹. Although the estimated value is some orders of magnitude higher than that accepted for the dust nucleation zones²⁸, it provides an effective method to accelerate the chemistry, which takes years in a real AGB envelope. Setting higher C densities together with an Ar overpressure, which accelerates the number of three-body reactions, are the two strategies adopted to overcome the exposure time limitation. Regarding hydrogen, we consider cases of low and high densities. In the first case, H₂ molecules and C atoms are in equal proportion, which can be achieved either using residual hydrogen or by introducing very low H₂ densities estimated in the range of 1.5×10¹⁰mol·cm^-3, a value typically accepted for the dust nucleation zone at distances from the star of 2-3 R* ²⁸. In the second case, a high H₂ density is achieved by introducing extra hydrogen at a concentration estimated at 1.5×10¹² mol·cm^-3, two orders of magnitude higher than that of vaporized C atoms. This situation provides a more appropriate concentration to simulate the chemistry in CSE in terms of H₂/C abundance (≈1000)³².

Structure of the carbonaceous stardust analogues produced

Fig. 2a-b shows typical AFM images obtained from the C-dust collected on a surface. Two different morphological features can be identified. Firstly, carbonaceous nanosized grains (encircled in white in Fig. 2a) are efficiently produced as the most abundant constituents^10,15. These spherical grains exhibit an extremely narrow size distribution with an average size of 9 nm (see Fig. 2a inset). Although larger grains (typically 100 nm) are thought to be present in space, in Stardust the nanoparticle size is conditioned by the geometry and experimental parameters and can be considered to represent the initial stages of carbonaceous dust growth. While the images of Fig. 2 were recorded for low hydrogen concentration, similar structures can be found for high H₂ densities.

Fig. 2. Production and structure of the stardust analogues.

a-b, AFM images recorded ex situ on a SiO_x surface at the position labelled “sample 1” in Fig. 1, showing the presence of C-nanoparticles and small clusters forming an almost continuous layer. A single nanoparticle is encircled in fig. 2a. Inset of fig. 2a shows the size distribution in a bar-histogram performed over several images. The blue line represents the log-normal fitting of the size distribution c-d, in situ recorded STM images of a C-cluster film deposited on a Au surface at “sample 2” position in Fig. 1. Scanned area: 25x25 nm², recorded at 50 pA, 1400 mV. Inset: Zoom of the rectangular area 5x3 nm², I= 10 pA, V=1400 mV. Two small molecules are encircled and zoomed in the inset of fig. 2d. The white lines showing the enlarged area in fig. 2b are for clarity and do not correspond to the exact area where measurements were taken. All images were recorded for the case of low- H₂ concentration.

Besides the carbonaceous nanograins, the other principal constituent of the C-dust analogues can be observed on the “flat” regions of Fig. 2b, where a thin mantle of material is observed, for instance in the area marked by a white square (see Supplementary Information 3). Structural resolution of these zones cannot be achieved using standard AFM, and more powerful in situ microscopies are required. Fig. 2c-d shows a scanning tunnelling microscope (STM) images obtained from the C analogues deposited on a Au surface. A STM-based detailed characterization of the imaged carbon products is complicated due to the large number of different species generated and their sub-nanometric size. Fig. 2c shows a wide plethora of individual nanometer-sized structures at the surface, with apparent heights ranging from 1 to 5 Å that correspond to small amorphous C-clusters, but are difficult to ascribe to a particular cluster type. Some of these small molecules can self-assemble on the surface forming 2D ordered islands (see Fig. 2d). This process is surface-mediated suggesting electrostatic interactions between the molecules upon adsorption on surfaces. The structures forming the ordered regions are smaller than 8 Å and with a maximum apparent height of 0.6 Å (see inset). The small size suggests that they correspond to individual molecules containing only 3 to 5 C atoms. Unambiguous assignment to a precise chemical structure is difficult due to the fact that STM images show a topography convoluted with the electronic structure of the inspected element.

Aliphatic nature of the dust analogues

Swan C₂ lines are observed by Optical Emission Spectroscopy (OES) in the presence of low H₂ density, Fig. 3a, but are not found in the case of higher H₂ density. On the contrary, the H_α emission line is only seen at high H₂ density. These observations suggest that excited H atoms are generated during the chemical evolution, involving C₂ + H₂ → CCH + H. The absence of the C₂ lines at high H₂ density indicates that most of the excited C₂ formed is consumed in this reaction. Interestingly, C₂ and H species are only detected in the regions close to the magnetron, which compares very well to observations from the photosphere of evolved stars where atomic hydrogen has been found and C₂ bands are prominent³³.

Fig. 3. Spectroscopic characterization of the molecular species.

a, Optical emission spectra recorded at position OES in Fig. 1b for low and high H₂ densities. b, In situ gas-phase mass spectroscopy detection at position QMS of most important masses for low and high H₂ densities. Vertical dashed lines indicate the time period where the magnetron was switched on and off. All masses have been offset for clarity.

Fig. 3b presents the evolution of the masses of some relevant species during the experiment recorded at position QMS of Fig. 1b. The ON region in the figure corresponds to the time when the magnetron is switched on and the background values are obtained before and after this point. Although a direct assignment of the masses to their parent molecules is not straightforward due to the electron ionization that induces cracking patterns, we discuss the most important contribution for each mass observed. In the case of high H₂ densities, the predominant mass formed is 26 amu, mainly corresponding to C₂H₂ along with fragments of C₂H₄, C₂H₆ and larger aliphatic molecules. An increase of mass 16 amu is also observed, corresponding to methane. In the case of high H₂ densities, the formation of these species is more abundant and an increase is also detected in the main ionization fragments of ethane (C₂H₆), ethylene (C₂H₄), propane (C₃H₈) and larger alkanes, which later contribute to most of the masses. Importantly, in this work these species were formed directly from C atoms and molecular hydrogen, without introducing them as precursors, using ratios and aggregation processes similar to those reported for dust formation zones of CSEs¹⁴. Fig. 3b shows a decrease in the partial pressure of H₂ when C-evaporation is initiated, suggesting that in Stardust H₂ is partially consumed by chemical reactions. Neither benzene (78 amu) nor toluene (91 amu) are detected, indicating that they are below the detection limit (3×10³ mol cm^-3). Interestingly, masses 18 and 32, corresponding to water and oxygen respectively, are unchanged with respect to their background pressure, indicating that, they are not involved in the reactions.

We have performed ex-situ laser desorption ionization mass-spectrometry (LDI-MS) in the so-called AROMA set-up ³⁴, a machine with high sensitivity for PAH detection. Fig. 4 provides the ion distribution obtained by desorption/ionization of representative samples grown with both H₂ densities, showing that the molecular species generated have < 20 carbon atoms. C-clusters with a low degree of hydrogenation are observed for species containing 4 to 9 C atoms. Pure carbon clusters are observed from C₈⁺ to C₁₉⁺, and these are poorly hydrogenated above C₉⁺, which can be understood by the formation of ring structures presenting lower reactivity³⁵. Above 300 amu the signal completely vanishes, indicating the absence of fullerenes (see Supplementary Figure 4 for full mass range spectra).

Fig. 4. Ex situ laser desorption/ionization mass-spectrometry (LDI-MS).

from samples grown with a low (upper panel) and high (lower panel) H₂ densities. Data are normalized to the highest peak. Insets: Stacked bar graphs summarizing the family compositions by double bond equivalent (DBE) analysis on the full spectra (see Methods). The error bars correspond to the relative standard deviation obtained for the total ion signal over a mass range for each family of compounds.

The results from LDI-MS are summarized in the insets of Fig. 4 where the species are classified in families (see Supplementary Information 5 and methods). The analysis suggests a higher abundance of C_nH_m clusters for low H₂ density, indicating the formation of more saturated aliphatic species at higher H₂ density, which are not found by LDI-MS but evidenced by QMS (Fig. 3). Benzene or toluene, as key aromatic species, are marginally detected in both LDI-MS and QMS. With LDI-MS, a few aromatic species are evidenced in very low abundance (total of 3% for both H₂ densities), the largest being C₁₆H₁₀ (Fig. 4). Thus, it appears that the chemistry involved in the CSE does not favour the formation of PAHs, which can account for up to 18% of the total carbon species in the ISM³⁶.

Discussion

Kinetic and energetic modelling

We have constructed a time-dependent chemical kinetics model that computes the evolution of the chemical composition of a gas mixture. The model follows the formation of carbon-containing molecules from the precursor C + H₂, and aims to predict the principal molecules that can be formed and to understand the underlying chemical reactions behind their formation.

We use a gas-phase chemical network consisting of about 100 neutral and ionic pure carbon clusters and hydrocarbons containing up to 10 carbon atoms³⁷. The initial conditions correspond to those of the high H₂ concentration experiment, and details are given in methods section.

In Fig. 5 we show the results of the calculations for some relevant species. It is seen that after several tens of seconds atomic carbon starts to be processed into different carbon-containing molecules. Pure carbon clusters such as C₃, unsaturated hydrocarbons like C₂H₂ and C₃H₂, and saturated hydrocarbons such as CH₄, C₂H₆, and C₃H₈ are abundantly formed. These results qualitatively agree with the molecules detected in the gas-phase (Fig. 3) in the Stardust machine, which indicates the disappearance of C₂ and the emergence of CH₄ and C₂H₂. According to the model, the molecular synthesis takes place in various steps.

Fig. 5. Computed evolution of the molecular species formed.

Left: Evolution of the chemical composition computed with the chemical kinetics model as the gas flows in the aggregation zone, calculated using the values for high hydrogen densities (see text) and at T=500K. Right: Chemical scheme of synthesis of hydrocarbons from atomic carbon given by the chemical kinetics model. Calculations performed for the case of high hydrogen density.

The initial step, in which carbon is transformed from atoms to molecules, proceeds via three-body reactions with Ar as the third body. In particular,

$$\text{C}+\text{C}+\text{Ar}-->{\text{C}}_2+\text{Ar}$$

$$\text{C}+{\text{H}}_2+\text{Ar}-->\text{C}{\text{H}}_2+\text{Ar}$$

are key the reactions that constitute a bottleneck to the whole process of formation of C-containing molecules. For the pressure and temperature of the model, the time scale of these reactions is of the order of tens of seconds. The bimolecular reaction of atomic carbon with H₂ to produce CH + H is endothermic by about 100 kJ/mol and thus is not efficient at the considered temperature of 500 K.

Once the aforementioned starting reactions have formed the first molecular species, the growth of molecules is driven by bimolecular reactions and three-body reactions involving neutral species. Routes involving ions are less efficient under our experimental conditions. Moreover, reactions of hydrocarbons with Ar⁺ are fast and result in fragmentation³⁸, which tends to produce a delay in molecular synthesis.

According to the model, carbon growth mainly occurs via insertion reactions of carbon atoms. For example, the reactions

$$\begin{array}{l}{\text{C}}_2{\text{H}}_2+\text{C}-->{\text{C}}_3+{\text{H}}_2\\ -->{\text{C}}_3\text{H}+\text{H}\end{array}$$

are fast³⁹ and drive the transformation from two carbon atoms to molecules with three C atoms. Analogous reactions of polyacetylenes with atomic carbon allow the progressive processing of carbon into aliphatic hydrocarbons of increasing complexity.

Apart from the increase in the number of carbon atoms, hydrogenation is also an important part of the chemical processing in the Stardust experiment. Hydrogenation essentially occurs via three-body reactions involving H atoms of the type

$${\text{C}}_{\text{n}}{\text{H}}_{\text{x}}+\text{H}+\text{Ar}-->{\text{C}}_{\text{n}}{\text{H}}_{\text{x}+1}+\text{Ar}$$

in which H atoms are successively added. This way, CH₄ is formed from CH₂, and C₃H₈ from less hydrogenated species like C₃ and C₃H. Under our experimental conditions, H atoms are abundantly formed through various reactions involving ions like

$${\text{Ar}}^{+}+{\text{H}}_2-->\text{Ar}{\text{H}}^{+}+\text{H}$$

There are some hydrogenation steps that mostly follow bimolecular reactions involving H₂ rather than three-body reactions involving H atoms. This is the case of the conversion of C₂ into C₂H, which occurs via the reaction

$${\text{C}}_2+{\text{H}}_2-->{\text{C}}_2\text{H}+\text{H}$$

which could be rapid⁴⁰ and more efficient than the three-body reaction, C₂ + H + Ar.

The synthesis of hydrocarbons from atomic carbon is summarized in the scheme shown in Fig. 5. Carbon growth by C insertion and hydrogenation by three-body reactions with H can be extrapolated to larger species. However, the kinetics of reactions involving larger hydrocarbons is not so well constrained and it is likely that other routes, e.g., association of hydrocarbon fragments of different size, will also be efficient.

The calculations undertaken at different temperatures indicate that below 500 K (down to 300 K) the results are qualitatively similar to those obtained at 500 K, with little variation in the abundances caused by the dependence of the rate coefficients of the reactions involved in the molecular synthesis. At temperatures higher than 500 K the behavior gradually changes, with unsaturated hydrocarbons like polyacetylenic chains becoming more abundant at the expense of saturated hydrocarbons, like alkanes. At temperatures above 1000 K, the model predicts that alkanes should have negligible abundance. This suggests that there could be a turning point for temperatures around 800-1000 K. Below these temperatures, hydrocarbons with high H/C ratios would form efficiently whilst at higher temperatures C-bearing molecules with low H/C ratios would dominate. This may have implications for the formation of carbonaceous material in C-rich CSEs. At thermochemical equilibrium, graphite is predicted to condense at temperatures as high as 1500-2000 K⁴¹. However, the actual temperature at which carbonaceous dust forms in C-rich CSEs could be significantly lower because solid carbon may not be graphitic but rather in the form of less ordered materials like amorphous carbon⁴² and because the process of condensation is probably controlled by chemical kinetics rather than by thermochemical equilibrium.

Thermal induced formation of aromatics on surfaces

Fig. 6 shows a representative mass spectrum single scan at 430K from an in situ TPD experiment. Although precise assignment of masses is difficult, families of C_nH_m hydrocarbons can be identified. The most abundant masses of 26 amu, 39 amu and 41 amu all belong to known fragments of larger aliphatic hydrocarbons⁴³. Interestingly, benzene and naphthalene were observed exclusively after annealing the deposited carbon above 355K. These aromatic species are most likely formed after a catalytic recombination of small hydrocarbons on the gold surface, a process typically reported in surface chemistry^44,45. Note that in the LDI-MS analysis, which is highly sensitive to aromatics, benzene was marginally observed whilst naphthalene was not detected. The fact that aromatics could be formed after annealing of carbonaceous aliphatics deposits on a C-rich surface needs to be further explored as a possible mechanism for the formation of these species in warm CSE environments. Indeed AIBs, the signature for PAHs, are not convincingly detected in AGBs but are observed at later stages when the star emits ultraviolet (UV) radiation that leads to photo-processing of the carbon-dust²⁰.

Fig. 6. Representative mass spectrum from a thermal desorption experiment.

on C-analogues formed using high H₂ density on a Au surface. Temperature of 430 K. (see full TPD spectra in Supplementary Figure 6).

Astrophysical implications

There are several models to account for the presence of aromatic species in the ejecta of evolved stars; one involves the polymerization of acetylene to benzene⁵ and further growth by the hydrogen abstraction acetylene addition (HACA) mechanism^22,46. Other scenarios involve the processing of SiC dust grains⁴⁷ or the catalytic activity of C₂H₂ on silicate surfaces⁴⁸. Although the HACA mechanism has been invoked to unravel the synthesis of PAHs in the outflows of carbon-rich AGB stars^22,46, our results suggest an alternative scenario. We produce acetylene in a ratio C₂H₂/H₂ of 2 × 10^-2, compared to the 10^-4 estimated for the inner layers of AGBs⁴. This is due to the fact that we have accelerated the chemistry by introducing higher C densities. Importantly, even under these favourable conditions we do not observe an efficient formation of PAHs. Earlier models of the formation of PAHs in CSE have shown that there is a narrow temperature window, between 900 and 1,100K, in which PAHs could form^22,49. However, the conversion of benzene to larger PAHs was found to be a bottleneck in the reaction scheme. The role of shocked regions close to the star photosphere in which temperatures higher than 1,700 K could be reached have therefore been emphasized⁵⁰.

Our results also suggest the formation of a large amount of amorphous nanosized carbon grains that can be considered as the main component of the carbon stardust (see Fig. 3). Aliphatic species deposited on dust grains are susceptible to further reaction, between themselves in a catalytically activated process, leading to larger molecules or aromatic species upon thermal activation (see Fig. 6). Such a temperature rise could be achieved in the highly UV-irradiated environments of protoplanetary nebulae where polymerization of gas species and dust chemistry could also lead to the formation of PAHs^5,32.

Finally, we are able to qualitatively reproduce the abundances of the C₂H₂ and C₂H₄ found in CSEs around AGB stars, indicating that SGAS, a technique not previously used in laboratory astrophysics, can be a very valuable tool to gain information on the chemistry operating in CSEs. When compared to other techniques, SGAS more closely resemble the delicate conditions of these cosmic environments (see Supplementary table 2). Most of the previously used techniques lead to the formation of aromatics, in addition to grains and clusters, either because of the use of much higher temperatures or because of the use of very high pressures of acetylene.

Conclusions

The combination of laboratory astrophysics, surface science and astronomical observations can unveil the chemical routes that operate in the inner layers of the envelope of evolved stars, providing new insights into the chemistry of carbonaceous stardust seed formation and fostering new observations. Our results suggest that PAHs might not be efficiently form during gas-phase growth in the CSE, which we show to generate aliphatic material and C-clusters together with carbonaceous nano-sized grains. However, the species that are deposited on dust grains are susceptible to further processing and dust can catalyse the formation of aromatic species upon thermal activation, which could take place as a result of the significant rise in dust temperature that occurs in highly UV-irradiated environments.

Methods

Location of the experimental techniques in the Stardust machine

The composition of the small gas-phase species was analysed in situ by mass spectrometry (MS) and optical emission spectroscopy (OES). The material generated can be collected on a surface at position “sample-1” for ex situ studies, such as laser desorption ionization mass spectroscopy (LDI-MS) or atomic force microscopy (AFM). Finally, the deposit (at position sample 2 in Fig.1) is characterized in situ in a third chamber by thermal programmed desorption (TPD).

Fabrication of the dust analogues

The carbon dust analogues were fabricated using a scaled-up multiple ion cluster source (MICS)²¹, a type of sputtering gas aggregation source (SGAS), working under UHV conditions (base pressure 1×10^-9 mbar) from Oxford Applied Research Ltd.. The magnetrons were loaded with a graphite target of 99.95% purity. The Ar flux used for the experiments was 150 sccm. The typical power applied to the magnetron was 100 W and the working aggregation length (distance between the magnetron and the exit nozzle) was 374 mm.

Extra-pure H₂ (99.99% purity) was injected during fabrication through one of the lateral entrances of the aggregation zone of the MICS (see Supplementary Figure 1), using gas-dosing valves with flux-regulator to inject into the UHV system. The doses used for these experiments were: 0, 4×10^-4 and 1 sccm, with an Ar flux maintained at 150 sccm. The substrates employed for collecting the NPs are boron-doped Si(100) with its native oxide for AFM; polycrystalline Au for AROMA experiments; Au(111) and graphite for TPD analysis.

Atomic Force Microscopy (AFM)

measurements were performed in the dynamic mode using a Cervantes AFM System equipped with the Dulcinea electronics from Nanotec Electronica S.L. All images were analyzed using WSxM software⁵¹. AFM images were recorded ex situ.

Scanning tunnelling microscopy (STM)

Samples for scanning tunnelling microscopy (STM) were prepared in Stardust and transferred without exposing the samples to the atmosphere to the STM chamber, located in a different laboratory, using a UHV suitcase with an ion-getter pump (base pressure <1×10^-9 mbar) to avoid chemical reaction with environmental gases²³. This procedure can be considered as in situ since the surface has never reached a pressure higher than 10^-9 mbar. Samples are deposited on an atomically flat single crystal Au(111) surface pre-cleaned by repeated sputtering/annealing cycles, which consisted of sputtering for 10 min using Ar⁺ ions at an energy of 1.25 kV and subsequent annealing at 800 K for 5 min. Clean Au(111) is later exposed to the carbon analogues for 5 minutes keeping the sample at room temperature, then immediately transferred to the UHV suitcase and to the STM chamber. The total time between sample growth and insertion in the STM cryostat was limited only by the pumping of the load lock volume. STM measurements were performed at 4K in a LT-STM (Scienta Omicron) with a base pressure <10^-11 mbar.

Optical emission spectroscopy (OES)

The light emitted by the plasma generated in front of the magnetron was collected through a fused silica window and an optical fibre of fused silica, and analysed by OES (see position in Fig. 1a) using a 193 mm focal length, motorized Czerny-Turner spectrograph (Andor, model Shamrock SR-193-i-A) equipped with a CCD camera (iDus DU420A-BVF). Two diffraction gratings with 1200 grooves/mm and 1800 grooves/mm, installed in a movable turret, provide spectral ranges of 300-1200 nm and 200-950 nm, respectively, and nominal spectral resolutions of 0.22 nm and 0.15 nm, respectively (for an input slit width of 20 μm). The relative spectral efficiencies of all the spectroscopic equipment was quantified for both diffraction gratings with a calibrated tungsten lamp.

In situ mass spectrometry (MS)

The gaseous species produced during the fabrication of the C-dust analogues were detected in situ with a quadrupole mass spectrometer (QMS) (0-100 amu) Prisma plus spectrometer from Pfeiffer. Lowest detectable partial pressure 10^-10 mbar. The system includes a continuous secondary electron multiplier.

Laser desorption/ionization mass-spectrometry (LDI-MS): the AROMA setup

AROMA (Aromatic Research of Organics with Molecular Analyzer) is dedicated to the analysis, with micron-scale resolution, of the carbonaceous molecular content of cosmic dust analogues. The experimental setup consists of a microprobe laser desorption ionization chamber and a segmented linear quadrupole ion trap connected to an orthogonal time of flight mass spectrometer (LQIT-oTOF). For each detected m/z peak with a signal-to-noise ratio (S/N) greater than 10, a chemical formula was assigned using the mMass software⁵², an open source mass spectrometry tool. The double bond equivalent (DBE) for each molecular formula was then calculated using the following equation for a given chemical formula of hydrocarbon species C_cH_h: DBE = c-(h/2)+1. This method allows sorting of the detected ions into families of compounds⁵³. This is done using empirical factors that delineate DBE boundaries in complex natural organic matter⁵⁴. Hydrocarbons with 0.5 < DBE/C < 0.9 are considered to be aromatics. The line DBE/C ~1 corresponds to C clusters and the species falling in the zone 0.9≤DBE/C<1 are referred to as HC clusters, which includes C_cH₂ polyynes. Once all the detected species were sorted, the ion signal percentage was calculated for each family by dividing the sum of peak intensities detected in each family by the total ion intensity. The reported error bars in DBE analysis correspond to the relative standard deviation obtained for the total ion signal over a mass range of a family of compounds when the measurement was repeated several times (7 to 10) on natural samples. These error bars, of around 25%, are conservative since they have been derived from a complex matrix.

Thermal Programmed Desorption (TPD)

Experiments were performed in situ to determine the mass spectra of the carbon analogues deposited on low reactivity surfaces annealed at increasing temperatures. TPD was performed using a Pfeiffer HiQuad QMG 700 with QMA 400 mass spectrometer with a CP 400 ion counter preamplifier. The instrumentation allows measurement of masses between 0 and 512 amu with a detection limit of 10^-15 mbar. After carbon analogue deposition the samples were placed in front of the quadrupole at a short distance (>1 cm) ensuring that the majority of the collected gas species derive from the surface of the sample surface. Atomically flat Au(111) and graphite substrates were used in the experiments, yielding comparable TPD spectra. Au(111) surfaces were cleaned in situ by the standard methodology of sputtering/annealing. Graphite was mechanically cleaved ex situ and consecutively introduced into the vacuum system, which ensures an atomically fresh graphite surface prior to deposition. The carbon analogues were deposited during 100 minutes, ensuring that there is sufficient carbon material to obtain well-defined signals in the TPD experiments. A temperature ramp of 12.5°K·min^-1 was applied from room temperature to the maximum temperature examined (700K).

Kinetic model

We use a gas-phase chemical network consisting of about 100 neutral and ionic pure carbon clusters and hydrocarbons containing up to 10 carbon atoms. These species are linked by about 2000 chemical reactions, whose rate constants are extracted from the literature on combustion chemistry, atmospheric chemistry, astrochemistry, as well as from specific chemical kinetics studies³⁷. The starting composition of the gas consists of a mixture of Ar/Ar⁺/H₂/H/C with mole fractions 1.0/10^-4/1.6×10^-6/10^-7/2.6×10^-8. We consider that the gas is trapped in the aggregation zone at a constant pressure of 66 mbar (measured value) and a constant temperature of 500 K, which can be considered as an averaged representative value. In a second stage, the gas is neutralized and expands reaching a pressure of 0.1 mbar and a temperature of 300 K at a distance of 10 cm from the magnetron sources. At this stage, the sharp decrease in pressure makes the chemical reactions too slow in comparison with the dynamic time scale associated with the gas expansion, and the chemical composition remains quenched. Therefore, molecular synthesis takes place only during the first stage in the aggregation zone.