Characterization of interstellar carbon dust analogues synthesized by dielectric barrier discharge and evolution after irradiation with 3 MeV H⁺

ABSTRACT

‘Fluffy’ hydrogenated amorphous carbon (a-C:H) was synthesized using a dielectric barrier discharge plasma, driven by nanosecond high voltage pulses at 1 kHz frequency in a helium–butane mixture. The a-C:H samples were characterized by scanning and transmission electron microscopy, laser-assisted and secondary ion mass spectrometry, and Raman and Fourier-transform infrared spectroscopy. We find that a-C:H samples exhibit infrared absorption features in good agreement with those observed for carbonaceous dust in IRAS 08572 + 3915 galaxy. We discuss their nano- to microscale structure and derive their hydrogen to carbon (H/C) ratios from the results obtained by three distinct experimental characterization techniques. Relying on the average H/C value determined by mass spectrometry and Raman spectroscopy, we can then constrain the absorption strengths values to those best corresponding to our dust analogue, and calculate the H/C ratio from the infrared spectra. Altogether, we find that our dust analogue consists of a dominant hydrogen-rich aliphatic network, with small, isolated, aromatic regions. The a-C:H dust analogue was then irradiated with 3 MeV H⁺ and subsequently analysed ex situ. Morphological and chemical changes, including the evolution of H/C, CH₂/CH₃, and sp²/sp³ ratios, were observed with increasing proton fluence, indicating dehydrogenation and graphitization. Proton bombardment shifted the initial location of a-C:H in the hydrocarbon ternary phase diagram toward the central region defined by IRAS 08572 + 3915 observations. The decay of the 3.4 |$\mu$|m band with proton fluence was used to calculate CH destruction cross-sections, results consistent with a direct effect of cosmic rays on the disappearance of the 3.4 |$\mu$|m band.

astrochemistry, plasmas, dust, extinction, infrared: ISM

1 INTRODUCTION

Carbon-rich asymptotic giant branch stars can replenish the interstellar medium (ISM) with stellar ejecta taking the form of carbon dust grains. The grains are subjected to many interactions with weathering agents, including ultraviolet photons, hydrogen atoms and protons, and are partially transformed into isolated or core–shell structures made of hydrogenated amorphous carbon (a-C:H) and polycyclic aromatic hydrocarbons (PAH) (Pascoli & Polleux 2000; Pendleton & Allamandola 2002; Carpentier et al. 2012; Chiar et al. 2013; Contreras & Salama 2013). Such energetic interactions can thus drive morphological and dimensional modifications in carbon dust grains and ices (Bennett, Pirim & Orlando 2013). Heating processes in the ISM (photoelectric, UV radiation or X-rays, cosmic rays, gas-grain chemical reactions) can also induce modifications of atomic bonding types and aromatization of a-C:H materials (Gadallah, Mutschke & Jäger 2013; Martínez et al. 2020). Understanding the mechanisms of carbon dust grain formation in the ISM or in circumstellar regions, along with their subsequent energetic or thermal processing as they evolve within their complex environment, is needed to better interpret observational information and apprehend the underlying physical processes at play (Martínez et al. 2020; Santoro et al. 2020). Theoretical and experimental studies designed to support the interpretation of scientific observations of such processes are still the strongest approaches in astrophysics, astrochemistry, and astrobiology. In this respect, continuous development of laboratory astrophysics studies is essential, although reproducing all processes occurring in astrophysical environments in the laboratory is deemed difficult (van Dishoeck 2019; Salama 2019; Tielens 2022).

A few decades ago, many laboratory works bolstered the hypothesis that a-C:H was the carbon architecture responsible for the near- and mid-infrared absorption bands observed in the ISM, particularly for the 3.4 |$\mu$|m (aliphatic CH stretching), 6.85 |$\mu$|m, and 7.27 |$\mu$|m absorption bands (aliphatic CH bending) (Duley & Williams 1983). However, a unique molecular structure for a-C:H has not yet been identified and often depends on the production method. In fact, under the single label a-C:H can be found a large group of carbonaceous solids (polymer-like films, diamond-like carbon, graphite-like films and tetrahedrally bonded a-C:H), with additionally a relatively large variation of hydrogen to carbon (H/C) ratio and sp¹, sp², sp³ contents (Jones, Duley & Williams 1990; Ristein et al. 1998; Schultrich 2018). Infrared signatures may not always be sufficient to unambiguously differentiate the hydrogenated carbonaceous solids synthesized under laboratory conditions, as exemplified by the 3.4 |$\mu$|m non-specific absorption feature which is common to most laboratory a-C:H (Shinohara et al. 2008, 2018; Manis-Levy et al. 2014; Asnaz et al. 2022). In space, the 3.4 |$\mu$|m absorption feature can also appear non-specific as it is observed in various environments, such as in the ISM – whether it is in the Milky Way or in other galaxies (Imanishi 2002; Pendleton & Allamandola 2002; Dartois et al. 2007), in the protoplanetary nebula CRL 618 (Chiar et al. 1998), or in smaller objects, such as interplanetary and cometary dust particles or meteorites (Ehrenfreund et al. 1991; Matrajt et al. 2005; Sandford et al. 2006; Muñoz Caro, Dartois & Nakamura-Messenger 2008; Lantz et al. 2015).

The main techniques utilized in the laboratory to produce interstellar carbon dust analogues, in the form of thin films with variable density or porosity and compact, powder-like, solid particles are of two types: plasma enhanced chemical vapour deposition (Furton, Laiho & Witt 1999; Kovačević et al. 2005; Contreras & Salama 2013; Maté et al. 2014; Molpeceres et al. 2017; Günay et al. 2018; Gavilan Marin et al. 2020; Sciamma-O’Brien & Salama 2020) and laser ablation (Scott & Duley 1996; Mennella et al. 1999; Reynaud et al. 2001; Llamas-Jansa et al. 2007; Biennier et al. 2009; Fulvio et al. 2017), sometimes coupled with gas-phase condensation (Mennella, Brucato & Colangeli 2001a; Jäger et al. 2008, 2009; Gadallah, Mutschke & Jäger 2011). However, many other techniques have been described in the literature, such as combustion flames (Pino et al. 2008; Carpentier et al. 2012), pyrolysis (Reynaud et al. 2001; Biennier et al. 2009; Fulvio et al. 2017), mechanochemical milling (Dartois et al. 2020), sublimation and quenching (Pendleton & Allamandola 2002), and UV photolysis (Dartois et al. 2005; Gadallah et al. 2011). At the nanoscale, disordered molecular structures are characteristic to most interstellar dust analogues and various values of the aromatic to aliphatic ratio are reported depending on the deposition technique. Operational parameters such as the temperature or energy level, precursor molecules, atoms or targets, and pressure are all key factors involved in the experimental design of carbonaceous interstellar dust analogues syntheses and processing.

Plasma-based deposition methods offer some unique advantages in interstellar carbon dust analogues production because of the possibility to balance surface and volume reactions, control the ratio of two- versus three-body processes and the density of radicals and reactants. Most plasma devices employed for carbon dust deposition are low pressure plasma sources, based on capacitively or inductively coupled radio frequency (RF) discharges, laser ablation plumes, pulsed discharge nozzles, spark and arc discharges, or magnetron discharges. Recently, we demonstrated that an atmospheric pressure plasma source, i.e. a dielectric barrier discharge (DBD), could also be successfully utilized as a synthesis method for ‘fluffy’ interstellar carbon dust analogues under low temperature conditions (Hodoroaba et al. 2018). Along with the differences in plasma techniques, a variety of hydrocarbons, e.g. alkanes, alkenes, alkynes, or PAHs, can also be utilized as precursors to carbon dust synthesis (see for example table 1 in Contreras & Salama 2013). This diversity contributes to synthetizing hydrogenated carbonaceous materials exhibiting distinct properties. Table 1 gathers examples of plasma syntheses and further shows the relationship between plasma operating parameters, the nature of the precursor, and the subsequent properties of the a-C:H materials produced. One parameter characterizing interstellar carbon dust analogues that can be compared to observational data is the H/C ratio. For interstellar dust, this ratio can be assessed from the observed infrared absorption spectra of luminous mid-infrared galaxies, such as IRAS 08572 + 3915, the latter exhibiting one of the strongest known 3.4 |$\mu$|m absorption features (Wright et al. 1996; Mason et al. 2004; Imanishi, Dudley & Maloney 2006). Previous H/C estimates for IRAS 08572 + 3915 were found in the 0.29–0.69 range (Dartois et al. 2007). Nevertheless, these H/C ratios should be treated with caution due to the uncertainty in absolute infrared band strength values used in the calculations (Herrero et al. 2022). Table 1 shows that H/C ratios for most interstellar carbon dust analogues fall in the same range or close to the range assessed for IRAS 08572 + 3915. One can observe a trend for increasing H/C ratio towards values equal to 1 and a trend for decreasing density towards values close to 1 g cm|$^{-3}$|⁠. Note that in the laboratory, complementary analytical techniques can be used to evaluate H/C ratios, which can help constrain the uncertainty of absolute band strength values within such complex hydrogenated carbonaceous materials.

Table 1.

Typical properties of hydrogenated carbonaceous materials deposited using plasma techniques for the various studies on interstellar matter that exhibits 3.4, 6.85, and 7.27 |$\mu$|m absorption bands.

Plasma generation method	Precursor	Pressure (mbar)	Substrate	Product synth.	Thickness (⁠|$\mu$|m)	Density (g cm|$^{-3}$|⁠)	H/C	Reference
DC discharge	Methane	0.26	NaCl or fused silica	HAC film	0.3	1.5	0.5	Furton et al. (1999)
Microwave discharge	Methane	–	NaCl or CaF2	Quenched carbonaceous Composite	–	–	–	Goto et al. (2000)
Arc discharge or laser ablation	Carbon in hydrogen	10	KBr	Hydrogenated carbon grains	0.05, 0.07	1.5	0.1–0.72	Mennella et al. (2002, 2003)
Laser ablation	Graphite in helium and hydrogen	3.3–26.7	CaF2 or KBr	Soot particles	0.17	1.55	0.14–0.57	Jäger et al. (2008)
RF plasma enhanced chemical vapour deposition and burning propylene with dioxygen	Methane and butadiene and propylene	0.01 and 70	KBr or KCl and KBr	a-C:H & soot	7–10 & 5	1.2 & 1.8	1 & 0.01	Godard et al. (2011)
Laser ablation	Graphite in helium and hydrogen	4.5	CaF2 or KBr	HAC particles	–	–	0.14–0.82	Gadallah et al. (2012)
ICP RF discharge	Methane in helium	0.3	Silicon	HAC film	3	–	–	Maté et al. (2014)
ICP RF discharge	Methane in helium	0.32	Silicon	a-C:H film	0.42–0.52	1.2	1	Maté et al. (2016)
ICP RF discharge	Methane in helium	0.3	Silicon or ZnSe	HAC film	–	–	1	Molpeceres et al. (2017)
RF plasma enhanced chemical vapour deposition	Methane	0.01	ZnSe	a-C:H	<10	1.2	1	Dartois et al. (2017)
ICP RF discharge	Methane in helium	0.3	Silicon	a-C:H film	0.38–1.54	1.1	1	Peláez et al. (2018)
Pulsed discharge nozzle	Acetylene or isoprene in argon	–	Petri dish	–	–	–	1.32–1.67	Günay et al. (2018)
Dielectric barrier discharge	Methane, ethane, propane, butane in helium	800	Graphite or silicon, quartz, and NaCl	a-C:H ‘fluffy’ dust	100–200	–	1	Hodoroaba et al. (2018)
DC magnetron discharge	Graphite in argon, than acetylene exposure	0.1	SiOx or carbon grids or KBr	–	–	–	–	Santoro et al. (2020)
Dielectric barrier discharge	Methane	800	Graphite	a-C:H ‘fluffy’ dust	100–200	0.95	0.62–0.79	This work

Evolution of carbon dust in various energetic astrophysical environments (i.e. under exposure to vacuum ultraviolet photons or cosmic rays) was in the spotlight of many experimental studies. Some were devoted to the calculation of destruction cross-sections of the 3.4 |$\mu$|m band by means of energetic processing using (V)UV photons (Mennella et al. 2001b; Alata et al. 2014), ions (Mennella et al. 2003; Godard et al. 2011), or electrons (Maté et al. 2016). The most recent results of Godard et al. (2011) and Maté et al. (2016) suggest that cosmic rays cannot be solely responsible for the disappearance of the 3.4 |$\mu$|m band (marker of aliphatic C-H bonds) in dense clouds (Herrero et al. 2022).

In this paper, we thoroughly characterize the physico-chemical properties of interstellar dust analogues produced in low temperature DBD fed with 15 per cent butane. We focus our analysis on our analogues’ microscopic features, molecular structure, and carbon hybridization, using complementary analytical techniques addressing the particles’ surface and/or bulk composition. To do so, we utilized electron and optical microscopy, two different techniques of mass spectrometry, Raman and Fourier transform infrared (FTIR) spectroscopies. FTIR and electron microscopy were also used to characterize the interstellar dust analogues after their exposure to 3 MeV H⁺ irradiation experiments performed to simulate energetic processing. The analyses are discussed in an astrophysical context, with an emphasis on carbon structure and destruction rates of aliphatic C-H bonds and their relevance to dust evolution in diffuse and dense ISM.

2 EXPERIMENTAL METHODS

2.1 Dielectric barrier discharge deposition

The hydrogenated carbonaceous ‘fluffy’ dust was produced employing DBD to generate a pulsed plasma in helium (grade 4.6) and 15 per cent butane (grade 2.5), with a total initial pressure of 600 Torr in a closed stainless-steel chamber. Flexible graphite (GoodFellow Co.) was used as substrate in the form of 0.5 mm |$\times$| 30 mm rectangular strips, homogeneously distributed over the ground electrode. A total of 40 substrate strips were used during each deposition experiment. Further information relative to plasma generation, diagnosis, and monitoring methods are given in a previous work (Hodoroaba et al. 2018). High voltage pulses (1 kHz, 6 kV, 500 ns, positive polarity) were used to periodically drive ionization of the gas mixture between the planar electrodes covered with glass. The available discharge gap between the dielectric layers was 5 mm. Plasma filaments were occasionally visible, sustained by the enhanced local electric field around sharp substrate edges. The average gas temperature was close to room temperature and high-energy plasma ions were not major contributors to surface or volume processes. During plasma pulses, electron impact collisions and Penning ionization processes led to the generation of hydrocarbon radicals, which further triggered rich volume and surface chemistries between plasma pulses. As a result, after 6 h of total deposition time, a black ‘fluffy’ dust was accumulated on all graphite strips, with an average weight gain of 0.6 mg per strip. Macroscopic monitoring of a-C:H growth during the deposition was performed with a camera (Optika C-B5). Dust analogues samples were transferred immediately after deposition to an acrylic storage box equipped with indicating silica gel to minimize exposure to moisture. All characterization techniques were performed ex situ on pristine samples duplicates. In addition, the apparent density (⁠|$\rho$|⁠) of the DBD-produced carbon dust samples was evaluated using the buoyancy method in non-polar solvents.

2.2 Microscopy

A detailed analysis of the DBD-produced carbon dust morphology was performed with a Quanta FEI 250 scanning electron microscope (SEM). No metal coating was applied to preserve sample morphology. The conductive properties of the DBD-produced carbon dust allowed image acquisition with the electron gun operating at voltages up to 30 kV with no noticeable local charging or image artefacts. In contrast, irradiated samples (after 3 MeV H⁺ exposure) did exhibit slight local charging. For this reason, general surveys of the irradiated samples were performed at 1.5 kV and only higher magnification images were acquired at 30 kV.

In addition, high-resolution microscopy was carried out with an FEI Titan electron microscope. The microscope was operated in the scanning transmission electron microscope (STEM) mode, the signal was collected from the backscattered electrons achieving sub-Å resolution. The DBD-produced carbon dust was gingerly scraped off its substrate and transferred to a copper grid (holey carbon film). STEM images were then recorded at different magnifications.

A 4K high accuracy digital microscope (Keyence VHX 6000) was additionally utilized to reconstruct sample surface topography from multiheight focal planes recombination (real-time depth composition).

2.3 Two-step laser mass spectrometry and time of flight secondary ion mass spectroscopy

The molecular content of the DBD-produced carbon dust was studied using a high-resolution two-step laser mass spectrometer (HR-L2MS, Fasmatech S&T), combining ion cooling, RF guiding, and a reflectron orthogonal time of flight (Re-oToF) analyser, with a maximum mass resolution of about |$m/\triangle m \approx 15 000$| at |$m/z = 200$|⁠. The sample, placed under vacuum (⁠|$10^{-8 }$| mbar residual pressure), was irradiated normally by a frequency doubled Nd:YAG laser beam (Quantel Brilliant EaZy, |$\lambda$| = 532 nm, 4 ns pulse duration, |$\sim$| 120 mJ cm⁻² fluence). The instrument was used in ablation mode and thus the desorbed molecules stemmed from both the sample surface and sample bulk. The desorbed compounds formed a gas plume expanding into the vacuum normally to the sample surface, and were ionized by an orthogonal UV laser beam (Quantel Q-smart 850, |$\lambda$| = 266 nm, 5 ns pulse duration, 300 mJ cm⁻² fluence). The generated ions were then RF-guided to a He collision cell for thermalization and subsequently mass analysed. A total of 10 mass spectra was acquired in positive polarity across the sample surface. The mass spectra were calibrated using peaks corresponding known hydrogenated species (C₃H₃, C₄H₂, C₅H, C₆H₄, and C₇H₇) to provide confident assignments. Only peaks higher than three times the standard deviation of the background noise were included in the data reduction. Complementary analyses were performed using a ToF-SIMS 5 (secondary ion mass spectroscopy) instrument from ION-ToF GmbH. The DBD-produced carbon dust samples were sputtered with Bi₃ ⁺ ions and the generated secondary ions accelerated and analysed with a ToF mass spectrometer with maximum mass resolution |$m/\triangle m \approx 10 000$|⁠. The estimated ion dose of |$10^{11}$| ions cm⁻² was below the threshold of ToF-SIMS static mode, which means that only the uttermost surface layer was analysed. Mass spectra in positive polarity were recorded at 50 scans/acquisition on a 500 |$\times$| 500 |$\mu$|m|$^2$| surface with an image resolution of 128 pixels |$\times$| 128 pixels. Acquisitions were performed on three-to-five different regions of interest on each sample. The collected mass spectra were aligned, calibrated, and normalized by the total relevant ion count. Only peaks higher than three times the standard deviation of the background noise were included in the data reduction. Mass defect analysis was performed on mass spectra (for both HR-L2MS and ToF-SIMS) to assign a molecular formula to the recorded accurate mass (Sleno 2012; Duca et al. 2019). When mass defect is plotted against nominal mass, species that line up contain a repeating unit, which can simplify the visualization and the interpretation of complex sets of mass spectra (for instance, aliphatic and aromatic hydrocarbons are aligned on different slopes). The global H/C ratio can be calculated from the identified ions in the mass spectra:

$$\begin{eqnarray} \frac{H}{C}=\frac{N_H}{N_C},\ \ N_X=\sum _iN_{X,i}w_i,\ \ \sum _i w_i=1,\ \ X=H,C \end{eqnarray}$$

(1)

where N_X,i is the number of atoms X and w_i the normalized intensity for the corresponding molecular formula (Dobbins, Fletcher & Chang 1998). The caveats surrounding this method are: (i) identified ions are the only ones included in the calculation, regardless of the mass spectrometry method utilized, (ii) it is known to slightly overestimate the H/C ratio in ToF-SIMS spectra (Faccinetto et al. 2020), and (iii) the RIE is not taken into account in either methods.

2.4 Fourier transformed infrared spectroscopy

Analyses were performed on two instruments using three different modes to access complementary information. On the first instrument (Bruker Vertex 70), micro-FTIR analyses were carried out on specific DBD-produced carbon dust surface locations in order to reveal the vibrational signatures corresponding to the uppermost molecular sample layers. The spectra were recorded in transmission–reflection mode between 4000 and 550 cm⁻¹ with a resolution of 4 cm⁻¹ using the Bruker Vertex 70 FTIR spectrometer coupled to a Hyperion 1000 microscope equipped with |$15\times$| (N.A. 0.4) Cassegrain and |$4\times$| (N.A. 0.1) objectives. The FTIR spectrometer includes a KBr/Ge beam splitter and the microscope a liquid N₂-cooled narrow band HgCdTe photoconductor detector. Bare gold mirror backgrounds were recorded from 4000 to 550 cm⁻¹ at atmospheric pressure prior to sample analysis and were used as references in processing sample spectra. The IR beam diameter size was adjusted to 10 |$\mu$|m using blades mounted within the microscope and focus was adjusted optically to target the sample surface. The same FTIR instrument was later utilized in transmission mode, whereby the DBD-produced carbon dust samples were this time scraped off their substrate and transformed into KBr pellets and then placed within the instrument’s transmission cell. This second mode, where the infrared beam (1 cm beam diameter) passes through the whole pellet, was used to probe sample bulk, in contrast to the previous micro-FTIR analysis where surface features were preponderant. Spectra were recorded as transmittance and then were converted into absorbance.

Further analyses of the DBD-produced carbon dust samples were performed using a second instrument (Jasco FT/IR-4700) equipped with an attenuated total reflectance FTIR (ATR-FTIR) accessory (ATR-Pro One). In this configuration, the sample strip was directly placed face down onto a 2.5 mm-diameter germanium crystal so that the IR interacts with the carbon dust at a 45|$^\circ$| incident angle to the crystal/sample interface. Sample contact with the crystal was maintained using a dedicated screw on the sample mount. Average spectra of all measurements will be discussed in the results section. While ATR-FTIR spectra were recorded as transmittance, they were subsequently converted into absorbance in selected spectral ranges for the calculation of CH₂/CH₃, sp²/sp³, and H/C ratios, and further converted to optical depth values for visualization and comparison purposes with astronomical data from IRAS 08572 + 3915.

Processing of ATR spectra was further performed to retrieve quantitative information. Specifically, a two-point method was used for baseline removal, followed by a multiple peak fitting procedure using Gaussian components. CH₂/CH₃, sp²/sp³, and H/C ratios were calculated using the integrated absorbance of the relevant spectral features (Chiar et al. 2013; Molpeceres et al. 2017), ignoring any diamond network and small aromatic domains. However, several corrections have first to be applied. The first correction stems from the ATR method itself, in which the penetration depth (⁠|$p_d$|⁠) of evanescent waves into the samples varies with the wavelength of incident infrared radiations. Using an average refractive index value in the infrared range of interest equal to 2.0 for a-C:H materials (Jones 2012), one can estimate an average penetration depth of |$\sim$|300 and |$\sim$|600 nm for the 3.4 and 6.0 |$\mu$|m bands, respectively. Accordingly, a reduction factor is used to account for the different volumes sampled at different wavelengths and their consequence on band areas.

A second correction needs to be considered when calculating CH₂/CH₃ and sp²/sp³ ratios. In fact, band areas (A_band) corresponding to the CH₃ groups, CH₂ groups, aromatic CH groups, and olefinic C = C groups must be normalized by the vibrational modes’ absorption strength values (Dartois et al. 2007; Chiar et al. 2013). Band areas corresponding to A_C = C and the sum (A_CH2  + A_CH3) were used for calculation of sp² and sp³ fractions. Choosing absorption strength values best corresponding to a specific dust analogue is key because the formers are influenced by the density and local electric field, the temperature, local molecular structure, impurities, and average chain length, which makes them known only for a limited number of aliphatic or aromatic molecular structures. Accordingly, only few experimental works were devoted to assessing the absorption strength values at 3.4 |$\mu$|m for interstellar carbon dust analogues. This was done for instance for interstellar carbon dust analogues produced by laser ablation of graphite (Duley et al. 1998) and by a pulsed discharge nozzle method using isoprene and acetylene as precursor gases (Günay et al. 2018). The values determined for these materials were found to be less than half those calculated for small molecules, suggesting that a misuse of absorption strengths describing solids produced in different conditions might lead to deviations from actual CH₂/CH₃ and sp²/sp³ values. It is clear from these experiments that absorption strengths are to be chosen wisely, either after in situ determination within the same experimental conditions, from literature data obtained in very similar conditions, or when H/C ratios can be assessed by complementary experiments or simulated by theoretical models (Molpeceres et al. 2017). Our strategy is then to use H/C ratios determined from complementary spectroscopic approaches (e.g. FTIR and Raman) and ultrahigh sensitivity mass spectrometry techniques (e.g. HR-L2MS and ToF-SIMS) to deduce the absorption strength values best adapted to our DBD-produced carbon dust sample among the values available in the literature (Dartois et al. 2007; Chiar et al. 2013). This procedure will ultimately allow us to calculate representative CH₂/CH₃ and sp²/sp³ ratios.

A complementary method to calculate the H/C ratio from FTIR data is to find the positive solution of the following second-order polynomial equation, for H/C ratios comprised between 0.3 and 1 (Mennella et al. 2002):

$$\begin{eqnarray} b\left(\frac{H}{C}\right) +\left(a+12b-\frac{1}{m_Hk}\right)\left(\frac{H}{C}\right)+12a=0 \end{eqnarray}$$

(2)

The integrated mass-absorption coefficient for the 3.4 |$\mu$|m band, i.e. k in equation (2), is given by the integral of the ratio |$K^{\prime }\left(\tilde{\nu } \right)/ \tilde{\nu }$|⁠, with |$\tilde{\nu }$|  = 2956 cm⁻¹ and |$m_H$| is the mass of the hydrogen atom. Values of constants in equation (2) are |$a = (1.4 \pm 0.1) \times 10^{21}$| cm² and |$b = (-1.3 \pm 0.2) \times 10^{21}$| cm² (Mennella et al. 2002). The average mass-absorption coefficient |$K^{\prime }\left(\tilde{\nu } \right)$| was calculated using equation (3), where |$\rho$| is the apparent density of the sample determined experimentally using the buoyancy method in non-polar solvents, |$p_d$| is the penetration depth of evanescent waves into the samples at a given |$\tilde{\nu }$|⁠, and |$\%T$| is the per cent transmission:

$$\begin{eqnarray} K^{\prime }\left(\tilde{\nu } \right) = \frac{1}{\rho p_d} \ln {\left(\frac{100}{\%T}\right)} \end{eqnarray}$$

(3)

Note that all spectral analyses are performed on a series of 5–10 different samples to ensure both the reproducibility of the result, as well as the possibility to identify and exclude any possible outliers (i.e. samples with different material quantities in investigated areas) from the numerical calculations.

2.5 Raman spectroscopy

Analyses were performed with an inVia Reflex spectrometer (Renishaw) equipped with an Olympus microscope (BXFM) (Chazallon et al. 2014). The spectra presented here were obtained by irradiation with a 514.5 nm laser. The laser power was measured at sample and reduced to 21 |$\mu$|W to avoid thermal sample degradation. Using a 20|$\times$| magnification microscope objective (N.A. 0.4), the laser was focused on the sample surface with a spot diameter of |$\sim$|2 |$\mu$|m. The spectrometer was calibrated using the Stokes Raman signal of pure Si (520.5 cm⁻¹). Spectra were recorded in extended scan (500–2200 cm⁻¹) and their spectral resolution was |$\sim$|4 cm⁻¹ using the 1800 grooves mm⁻¹ diffraction grating. A total of 63 spectra were acquired across the DBD-produced dust surfaces. Over 63 spectra, 49 spectra (best signal-to-noise ratio) served as a basis for H atomic per cent calculation. Raman spectra of amorphous carbon materials are commonly characterized by the D (disordered, 1300–1400 cm⁻¹) and G (graphitic, 1550–1650 cm⁻¹) bands (Tuinstra & Koenig 1970; Lespade et al. 1984; Wang, Alsmeyer & McCreery 1990; Ferrari & Robertson 2000; Beyssac et al. 2003; Sadezky et al. 2005). The D band emerges in carbon materials exhibiting polyaromatic organic matter with finite-sized in-plane crystallites (i.e. aromatic domain diameter) and/or disrupted crystal symmetry (edges, defects, and vacancies) and corresponds to the breathing mode in aromatic rings. The G band characterizes the simultaneous in-plane carbon–carbon stretching mode (E_2g) in both chains and rings. Both Casiraghi, Ferrari & Robertson (2005) and Buijnsters et al. (2009) quantified the hydrogen content in a-C:H films using an empirical formula involving the photoluminescent background (i.e. slope in |$\mu$|m, denoted with m) and the maximum intensity of the Raman G band [I(G)], as described in equations (4) and (5), respectively, and for hydrogen contents varying approximately between 20 per cent and 50 per cent. Both formulas will be used to assess the H atomic per cent (⁠|$H[\mathrm{ at}.~{{\ \rm per\ cent}}]$|⁠) in our samples

$$\begin{eqnarray} H\left[{\rm at.}~{{\ \rm per\ cent}}\right]_{{\rm Casiraghi}~05} = 21.7 + 16.6\log \left(\frac{m}{I(G)}\right) \end{eqnarray}$$

(4)

$$\begin{eqnarray} H\left[{\rm at.}~{{\ \rm per\ cent}}\right]_{{\rm Buijnsters}~09} = 30.0 + 4.6\log \left(\frac{m}{I(G)}\right) \end{eqnarray}$$

(5)

2.6 Ion irradiation

The DBD-produced carbon dust samples were processed on the 3 MV Tandetron accelerator system located at the ‘Horia Hulubei’ National Institute for Physics and Nuclear Engineering – IFIN-HH, Magurele, Romania (Burducea et al. 2015). The ion implantation beam line was used to irradiate the DBD-produced carbon dust samples with 3 MeV protons while placed inside a vacuum chamber (10⁻⁵ mbar residual pressure). The average beam current, measured with a Faraday cup, was 100 nA. An X–Y beam sweep system was used to raster scan the samples and a four corner Faraday cup assembly was used to define the total scanned area and to measure the implanted ion dose. Carbon tape was used to fix the samples to the wafer target holder, positioned behind the corner Faraday cups. Different fluences, ranging between |$1\times 10^{14}$| and |$1\times 10^{16}$| protons cm⁻² were used and the sample holder temperature was monitored during the irradiation process. The maximum value of temperature measured was 390 K for |$1\times 10^{16}$| protons cm⁻² fluence, to avoid any thermal processing of interstellar dust analogues. The mean projected range of 3 MeV protons impinging a compact and homogenous carbonaceous material exhibiting the same characteristics as our a-C:H interstellar dust analogues was estimated at around 176 |$\mu$|m using Stopping and Range of Ions in Matter package (srim, Ziegler, Ziegler & Biersack 2010), which corresponds to a linear energy transfer equal to 19.4 eV nm⁻¹. The energy deposited per unit mass sample was in the range |$2\times 10^{13}$| to |$2\times 10^{15}$| MeV mg⁻¹. After exposure to high-energy protons, the DBD-produced carbon dust samples were quickly sealed in small plastic containers, which were themselves placed on an indicating silica gel bed and further sealed in a larger plastic container to minimize air exposure before ex situ analyses.

3 RESULTS

3.1 DBD-produced carbon dust description: from the macro to the nanoscale

Macroscopic monitoring of the DBD-produced carbon dust evidences the formation of conglomerate-like particles free roaming on the substrate surface and slowly increasing in size before settling in what we refer to here as islands. Consequently, the macroscale distribution of the deposited carbon dust is not uniform across the substrate; the material is preferentially found at the extremities and in cluster form, suggesting a 3D carbon island growth. This type of growth can be explained considering the transport of adsorbed seed radicals on 2D atomic arrays, where two limiting cases for molecular materials growth may be envisaged when taking into account the diffusion rate (D) and the deposition flux (F) (Barth, Costantini & Kern 2005): for large D/F ratios, diffusion is favoured and the molecular system reaches a minimum energy configuration which can induce a growth regime close to equilibrium conditions, whereas for low D/F values (which apply here), diffusion is prevented and the growth is essentially determined by kinetics, under non-equilibrium conditions. The apparent density (⁠|$\rho$|⁠) of the DBD-produced carbon dust is measured to be about 0.95 g cm|$^{-3}$|⁠, which lies at the lower end of density values commonly reported for plasma-produced a-C:H materials (1–1.5 g cm|$^{-3}$|⁠, details in Table 1).

SEM reveals two different morphologies at the mesoscale. First, we observe insular structures composed of aggregated a-C:H, their size and abundance increasing towards substrate corners (Fig. 1). This observation strongly correlates with the preferential formation and stabilization of discharge streamers in these regions. Second, the material shows a more compact central structure, mainly globular, in the regions where streamers are quite active. This can be attributed to streamer-induced local heating and subsequent transformation of a-C:H conglomerates. Thus, this mesoscopic ordering observed in SEM imaging is strongly influenced by intermolecular interactions and local electric field distribution during deposition.

Figure 1.

SEM micrographs at various magnifications and in various locations on the substrate of the as-deposited DBD-produced carbon dust. Capital letters indicate the regions explored to collect the high magnification images (⁠|$\times$|1k, |$\times$|5k, and |$\times$|10k).

Fig. 2 shows the morphological differences that appear in SEM micrographs when the DBD-produced carbon dust samples have been subjected to 3 MeV H⁺ with different irradiation fluences. Even at the lowest irradiation dose (⁠|$1\times 10^{14}$| protons cm⁻²), the flakes look damaged and compacted. Also, the flakes now appear etched, the process leaving massive holes that can reach all the way down to the substrate when the flakes are exposed to the highest proton doses. Proton irradiation makes the flakes adhere stronger to each other and gives the previous insular growth a coral-like structure.

STEM analyses show that some parts of the DBD-produced carbon dust consist of a significant number of ‘sheets’ with numerous creases (dark striae in Figs 3a and b). Note that the low contrast made only possible to image the edges of graphitic sheets, hereby concealing the amorphous phase. The observed ‘sheets’ are composed of several layers of carbon (10–15 layers), showing a graphite-like structure. The layers can be easily observed in the regions where the sheets are folding and have an approximate thickness of 4–5 nm. The profile of a crease was examined, showing that in Fig. 3 contains 13 carbon layers with an average interplanar distance of 3.2 Å. This value is in good agreement with values reported in the literature for graphite as well as the one obtained from electron diffraction patterns, i.e. 3.34 Å (Czigány & Hultman 2010).

Figure 2.

SEM micrographs of DBD-produced carbon dust samples after 3 MeV H⁺ irradiation with increasing fluences. Column 1 corresponds to |$1\times 10^{14}$| protons cm⁻², column 2 to |$1\times 10^{15}$| protons cm⁻², and column 3 to |$1\times 10^{16}$| protons cm⁻².

Figure 3.

A typical crease in STEM images and the corresponding intensity distribution of a cross-section profile in grey-level units, shown as arrow in the left image.

The graphitic structure of the observed sheets was observed in high-resolution STEM images, Figs 4(a)–(c). STEM images not only contain information about the lattice in the focal plane of the instrument, but also additional signals originating from out-of-focus layers and amorphous carbon. In order to highlight the structure of graphite sheets within the DBD-produced carbon dust, we subjected the STEM image to an image processing routine in Fig. 4(d). The first processing step was to apply a 2D fast Fourier transform (FFT), which revealed 6 high-intensity regions coming from the periodic signal. Then, we applied a bandpass filter, and finally, for visualization purposes, we calculated the inverse FFT in order to reconstruct the image associated with the selected frequency range of the filter (Fig. 4d). This processing effectively removes the noise and leaves only the information corresponding to the periodical lattice. The processed image (Fig. 4d) clearly shows the honeycomb-like structure of the graphite sheet with the distance between carbon atoms close to the one reported in the literature (140 pm, see the inset in Fig. 4d). It was shown previously that the high-resolution transmission electron microscopy images can be used to identify the degree of crystallinity of carbon materials (Endo et al. 1997) and FFT processing was found to be useful in assessing the extent of crystallinity (Lehman et al. 2011).

Figure 4.

(a)–(c) High-resolution STEM images of the interstellar dust analogues and (d) the result of inverse FFT calculated after application of the bandpass filter. The inset in (d) is a close-up of the observed patterns.

3.2 Determination of H/C ratios

The H/C ratio of the DBD-produced carbon dust is obtained from mass spectrometry data. Fig. 5 compares the mass spectra (left column) and the corresponding mass defect plot (right column) obtained independently from HR-L2MS (top row) and ToF-SIMS (bottom row) measurements.

Figure 5.

As-deposited DBD-produced carbon dust. (a) HR-L2MS normalized synthesis mass spectrum constructed by eliminating species also belonging to the virgin graphite substrate and (b) corresponding mass defects plot. The unidentified species appear in light grey. (c) ToF-SIMS normalized mass spectrum and (d) corresponding mass defect plot. In the mass defect plots, C_mH_n ⁺  ions are found aligned on different positive slopes (dashed lines). The area of the data points is proportional to the peak intensity.

After mass calibration, 62 per cent of the detected peak signals are assigned, which corresponds to about 87  per cent of the total ion count. Peak signals, detected up to around 300 |$m/z$|⁠, are identified by mass defect analysis following the protocol proposed in Irimiea et al. (2019). The vast majority of the identified species are assigned to C_mH_n ⁺  ions that span from small fragments to large polyaromatics, and are highly consistent between the two analyses. No nitrogen-bearing species were detected, while only few oxygen containing fragments were identified.

In general, peaks in the mass spectra recur in well-separated groups. The peaks in the same group share the same number of carbon atoms but exhibit a progressively increasing number of hydrogen atoms. Each group is found on a positive slope in the mass defect plot (dashed lines in Figs 5b and d). Neighbouring groups are separated by gaps that become progressively smaller as |$m/z$| increases, and completely disappear above 250 |$m/z$|⁠.

At low |$m/z$|⁠, the detected ions include CH_1-3 ⁺ , C₂H_3-5 ⁺ , C₃H_3-7 ⁺ , C₄H_3-8 ⁺ , C₅H_3-9 ⁺ , C₆H_4-7 ⁺ , C₇H_5-7 ⁺ , C₈H_5-7 ⁺ , and C₉H_5-7 ⁺ . Here, hydrogen-rich ions are found mixed with hydrogen-poor ions generated from the fragmentation of larger species. At high |$m/z$|⁠, the most intense peaks in each group are consistent with the patterns expected for PAHs. The main peaks show up along with less intense peaks at [M−1] ⁺  or [M−2] ⁺  attributed to hydrogen elimination, and to [M + 1] ⁺  and [M + 2] ⁺  attributed to the carbon isotopic ions. To each even numbers of carbon atoms corresponds an even |$m/z$| value characteristic of benzenoid PAHs (e.g. |$m/z$| C₁₀H₈ ⁺ , C₁₂H₈ ⁺ , C₁₄H₁₀ ⁺ , C₁₆H₁₀ ⁺ , C₁₈H₁₀ ⁺ , C₂₀H₁₂ ⁺, and C₂₂H₁₂ ⁺). In a similar fashion, to each odd carbon numbers corresponds an odd |$m/z$| value characteristic of resonance-stabilized cation radicals, markers for instance of PAHs with a five-member ring (e.g. |$m/z$| C₉H₇ ⁺ , C₁₁H₉ ⁺ , C₁₃H₉ ⁺ , C₁₅H₉ ⁺ , C₁₇H₁₁ ⁺ , C₁₉H₁₁ ⁺ , and C₂₁H₁₁ ⁺).

The H/C ratios are calculated using equation (1) and are found to be in reasonably good agreement, with 0.62 |$\pm$| 0.04 and 0.78 |$\pm$| 0.08 from HR-L2MS and ToF-SIMS, respectively. HR-L2MS and ToF-SIMS rely on different desorption and ionization methods. HR-L2MS can desorb molecular species down to several |$\mu$|m depth, and thus can provide information about both the surface and bulk molecular compositions. In addition, HR-L2MS is optimized for the detection of large polyaromatic hydrocarbons like PAHs that are ionized with a favourable resonant two photon ionization scheme at a wavelength of 266 nm, thereby producing mass spectra featuring reduced fragmentation. In contrast, ToF-SIMS uses high-energy Bi₃ ⁺  ions at low dose (static mode) that allow the sputtering of species stemming from the first atomic layers of the sample (1–3 nm). Although the relative ionization efficiencies (RIE) of the various molecular species are not taken into account for H/C calculations in either HR-L2MS or ToF-SIMS, both of them give H/C ratios in reasonable agreements with other methods giving OC/TC ratios (Delhaye et al. 2017) or H/C ratios (Dobbins et al. 1998; Faccinetto et al. 2020). This is made possible by averaging an extended |$m/z$| range which tends to cancel out any RIE variations. Fig. 5 shows that HR-L2MS and ToF-SIMS mass spectra feature different distributions of peak intensities. This is due to the different ionization methods, which lead to distinct fragmentation processes. However, equation (1) shows that the number of atoms of a given element X is conserved for H/C calculations, and thus the total number of atoms remains the same regardless of the fragmentation process. Consequently, the H/C ratios can be seen as independent on the method.

Raman spectra of the pristine graphite substrate before DBD exposure were first recorded as references and are shown in Fig. 6. The Raman spectrum of graphite is characterized by the D (disordered, 1300–1400 cm⁻¹) and G (graphitic, 1550–1650 cm⁻¹) bands (Tuinstra & Koenig 1970; Lespade et al. 1984; Wang et al. 1990; Ferrari & Robertson 2000; Beyssac et al. 2003; Sadezky et al. 2005). Parts of the substrate devoid of carbon dust after DBD exposure feature now a light surface coating of brownish colour at near-grazing light incidence. The corresponding Raman spectra exhibit a steep fluorescence background almost obscuring graphite’s G-band Raman signature (Fig. 6b). Such fluorescence background may originate from covalently bond clusters of PAHs (Ferrari & Robertson 2001) possibly formed via surface hydrocarbon radicals’ chemistry or may be triggered by the insertion of heteroatoms in the carbon lattice (Pal 2015). The previous HR-L2MS and ToF-SIMS analyses indicated the presence of heteroatoms, but only at trace levels. Further analyses of carbon particles discussed below will help shed some light on the fluorescence origin. The DBD-produced carbon dust was first observed through a digital microscope to reconstruct the sample’s surface topography after multiheight focal planes recombination (Fig. 7a). The topography shows ‘fluffy’ aggregated particles at the macroscale and corroborates previous SEM observations. The 3D height profile shows that when focused on the particles no contribution from the substrate is expected in Raman spectra, since the sample height (> |$\sim$| 100 |$\mu$|m) exceeds the theoretical depth of analysis (⁠|$\sim$| 5 |$\mu$|m, at 514.5 nm, using a 20|$\times$| objective with a numerical aperture of 0.4, and a refractive index of 2.0 for the carbon dust, Jones 2012).

Figure 6.

Raman spectra acquired across the graphite sheet substrate (a) before DBD deposition and (b) after DBD deposition in an area optically assessed to be devoid of particle but featuring a DBD-induced coating. Laser power is 21 |$\mu$|mW in both cases and intensity is in arbitrary units. Inset in Fig. 6(a) is an optical image of the pristine graphite sheet substrate taken with the 20|$\times$| objective.

Figure 7.

DBD carbon dust (a) 3D surface reconstruction showing its ‘fluffy’ character at the mesoscale and (b) corresponding Raman spectrum showing the G band and a steep background photoluminescence slope.

Akin to what was observed on graphite after DBD exposure, Raman spectra from the DBD-produced carbon dust exhibit fluorescence (Fig. 7b). This suggests the existence of aromatics such as condensable PAHs, whose presence was detected in both HR-L2MS and ToF-SIMS. Further studies have been carried out to evaluate their sensibility to thermal degradation or photobleaching. This was done by increasing the laser fluence in a stepwise manner when analysing the same spot. Specifically, once selected, a particle spot was subjected to consecutive Raman excitation laser exposures and Raman spectra, all acquired at the same lowest laser power, were subsequently compared (Fig. 8). This experiment shows that an increase in laser power from 21 to 121 |$\mu$|W was sufficient to kill most of the fluorescence while preserving Raman information, hereby suggesting a fragile or volatile character of the species generating the observed photoluminescence. In addition, these observations support a surface occurrence, i.e. the presence of these aromatic fluorophores within the uppermost molecular layers of the DBD-produced carbon dust, as confirmed by mass spectrometry.

Figure 8.

Sequence of three Raman spectra displaying the same DBD carbon dust spot subjected to different laser powers. Spectrum 1 was acquired with a laser power of 21 |$\mu$|W and is followed by spectrum 2 showing the same area exposed afterwards to 121 |$\mu$|W. Spectrum 3 is obtained exposing the same area again to the lowest laser power of 21 |$\mu$|W. Spectra 1 and 3 are directly comparable and illustrate the effect of laser power on fluorescence.

Most of the time, the fluorescence is not strong enough to cover Raman signatures, but the ‘fluffy’ nature of the particles can challenge getting proper Raman signal because some sampling spots are off laser focus. In fact, the small laser beam spot diameter (⁠|$\sim$|2 |$\mu$|m) and the relatively shallow Raman sampling depth make the resulting Raman spectrum quite sensitive to local roughness. To circumvent this obstacle and obtain meaningful statistics, 63 spectra were acquired across the particles’ surfaces. Hydrogen atomic per cents were calculated using equations (4) and (5) and are shown in Table 1. While the first value lies outside the 20–50 H [at.  per cent] range used to obtain the analytical form of equation (4) (Casiraghi et al. 2005), one might expect the linearity to remain valid nearby the interval (Adamopoulos et al. 2004). In addition, if one were to make the assumption that carbon atoms total up the atomic per cent to 100, then one can derive an average H/C ratio (Table 2). The H/C ratio of |$0.67\pm 0.01$| is more in line with the ratios earlier derived from mass spectrometry analyses (Table 2).

Table 2.

Atomic per cent of hydrogen H [at. per cent] and H/C ratio determined from Raman, ToF-SIMS, and HR-L2MS analyses.

	H [at.  per cent]	H/C
Raman – equation (4) (Casiraghi et al. 2005)	|$57.9\pm 2.5$|	|$1.38\pm 0.06$|
Raman – equation (5) (Buijnsters et al. 2009)	|$40.0\pm 0.7$|	|$0.67\pm 0.01$|
ToF-SIMS	–	|$0.78\pm 0.08$|
HR-L2MS	–	|$0.62\pm 0.04$|

FTIR experiments tailored to probe either sample surface (micro-FTIR) or sample bulk (macro FTIR with sample embedded in KBr pellet and analysed in transmission mode) further showed that the spectral features expected from the vibration of such aromatic structures exhibited weak signatures when analysed in micro-FTIR but did not show in transmission mode (Fig. 9). This is illustrated by the presence of the aromatic C-H stretching vibration (between 3000 and 3100 cm⁻¹ in PAHs (Leger & Puget 1984; Allamandola, Tielens & Barker 1985)) which only emerges in micro-FTIR and, of note, mostly on the edges of the ‘fluffy’ islands of the DBD-produced carbon dust (Fig. 9).

Figure 9.

FTIR spectra acquired in micro-FTIR mode (top) and in macro FTIR transmission mode with, for the latter, the sample embedded within a KBr pellet (bottom). The aromatic C-H stretches (3000–3100 cm⁻¹) emerge only in the micro-FTIR spectrum. The aliphatic CH_x stretches are all present in both spectra: CH₃ asymmetric (⁠|$\sim$|2960 cm⁻¹), CH₂ asymmetric (⁠|$\sim$|2929 cm⁻¹), CH₃ symmetric (⁠|$\sim$|2871 cm⁻¹), and CH₂ symmetric (⁠|$\sim$|2852 cm⁻¹). Interferences from moisture appear as a broad band between 3200–36001 cm⁻¹ in both spectra.

The DBD-produced carbon dust samples exhibit in their ATR-FTIR spectra absorption bands of interest with respect to astrophysical data (Fig. 10). Among them are the bands emerging between 3000 and 900 cm⁻¹, with maxima at 2956 cm⁻¹ (3.38 |$\mu$|m), 1455 cm⁻¹ (6.87 |$\mu$|m), and 1375 cm⁻¹ (7.27 |$\mu$|m). Other absorption bands are detailed in our previous works (Hodoroaba et al. 2018; Gerber et al. 2019). The ATR-FTIR spectra show no spectral features assigned to vibration modes expected from aromatic structures or sp² hybridized carbon (Ristein et al. 1998; Reynaud et al. 2001; Dartois et al. 2005; Pino et al. 2008; Carpentier et al. 2012; Molpeceres et al. 2017). Specifically, the following bands are not observable: 3000–3100 cm⁻¹ (or 3.3 |$\mu$|m, aromatic CH stretching), 1560–1660 cm⁻¹ (or 6.2 |$\mu$|m, aromatic C = C stretching), 1160 cm⁻¹ (or 8.6 |$\mu$|m, aromatic C = C–H in plane bending), 885–750 cm⁻¹ (or 11.3–13.3 |$\mu$|m solo, duo, trio, and quartet modes of aromatic CH in plane bending). The 3.4 |$\mu$|m band is assigned to aliphatic –CH stretching modes and is described with the highest accuracy only after a peak fitting procedure is applied, decomposing the signal into five distinct Gaussian components of variable central wavenumbers and full width at half-maximum (FWHM) (Fig. 10). The corresponding assignments are compiled in Table 3.

Figure 10.

Peak fitting of the 3.4 and 6.0 |$\mu$|m bands in ATR-FTIR spectra. Comparison between the DBD-produced carbon dust sample (top row) and IRAS 08572 + 3915 (bottom row), the latter taken from Brown et al. (2014).

Table 3.

Signal decomposition of the DBD-produced carbon dust ATR-FTIR spectrum and corresponding assignments in the 3000–2700 cm⁻¹ spectral range.

Band position (cm−1)	FWHM (cm−1)	Assignment
2960–2956	35–19	Asymmetric CH3
2935–2925	34–24	Asymmetric CH2
2911–2898	34–26	Second harmonic bending CH2
2872–2868	33–25	Symmetric CH3
2853–2842	32–27	Symmetric CH2

Our spectra are best fitted only including the contribution from the Fermi resonance (2911–2898 cm⁻¹) which corresponds to the second harmonic of the aliphatic CH₂ scissoring and that is expected for alkyl chains at around 2 |$\times$| 1455 cm⁻¹. It was previously emphasized that the analysis of spectral features from complex organic materials cannot be simply performed by considering independent vibrations and the analysis of asymmetric CH₂ mode should include a Fermi resonance peak to account for the red wing contribution around 2900 cm⁻¹ (Dartois et al. 2005, 2007). The olefinic C = C stretch was found in the 1680–1648 cm⁻¹ range (FWHM  = 121–58 cm⁻¹), together with a contribution of carbonyl (R₂C = O) groups, centred at 1710 cm⁻¹. Note that the oxygen content of the interstellar carbon dust analogues probed by mass spectrometry techniques is very low, while in FTIR spectra the C=O stretching mode is present. These carbonyl groups formed from traces in the plasma working gas mixture or during exposure of samples to air, even in very small amounts, are detected in FTIR spectra due to the significant difference in the intrinsic strength of C = O stretches, as compared to C-H stretches (Pendleton & Allamandola 2002; Kovačević et al. 2005; Jäger et al. 2008; Carpentier et al. 2012; Gadallah et al. 2012; Fulvio et al. 2017; Günay et al. 2018; Hodoroaba et al. 2018). The spectra of this DBD-produced dust are in good agreement with the astronomical observations recorded in the 3.4 and 6.0 |$\mu$|m band ranges, as shown in Fig. 10.

From the ATR-FTIR spectra of the DBD-produced carbon dust, we calculated using equation (2) an H/C ratio of |$0.79\pm 0.20$|⁠. This value is in line with the previous values calculated from the results given by mass spectrometry analyses (considering only the values of H/C < 1): |$0.78\pm 0.08$| from ToF-SIMS data, |$0.62\pm 0.04$| from HR-L2MS data, and |$0.67\pm 0.01$| from Raman spectra using equation (5), which gives an average H/C value of |$0.72\pm 0.20$|⁠. Now that the H/C ratio is known, it is possible to choose the absorption strength values from the literature (Dartois et al. 2007; Chiar et al. 2013) best representing the DBD-produced carbon dust samples. In fact, using the absorption strength values from Dartois et al. (2007) (⁠|$12.5\times 10^{-18}$| cm group⁻¹ for CH₃ groups, |$8.4\times 10^{-18}$| cm group⁻¹ for CH₂ groups plus Fermi resonance, |$0.4\times 10^{-18}$| cm atoms⁻¹ for C=C groups) and the integrated absorbance of Gaussian components obtained from peak fitting, we obtain H/C  = |$0.72\pm 0.04$|⁠, consistent with the average H/C ratio from the other techniques. Without this a priori knowledge derived from a multitechnique analysis, the H/C values derived from the ATR-FTIR spectra would range from 0.33 to 1.37. This result shows that a proper selection and cross-checks of absorption strength values for relevant carbon-containing chemical groups is a prerequisite before accurate CH₂/CH₃, H/C, and sp²/sp³ ratios calculations can be derived from FTIR laboratory data of astrophysical relevance. The absorption strength values chosen for the DBD-produced carbon dust will serve as a basis for the FTIR study of the evolution of its carbon structure upon 3 MeV H⁺ irradiation.

4 EVOLUTION AFTER IRRADIATION WITH 3 MEV H ⁺ : ASTROPHYSICAL IMPLICATIONS

4.1 Carbon structure

After proton irradiation, we observe a decrease of absorbance values in FTIR spectra, together with the modification of relative band ratios (Table 4). It can be observed that depending on fluence values the sp²/sp³ ratio increases, this being usually attributed to a graphitization process, while the H/C ratio decreases due to dehydrogenation by proton bombardment.

Table 4.

Evolution of integrated absorbance, integrated mass absorption coefficient of the 3.4 |$\mu$|m band k, and CH₂/CH₃, H/C, and sp²/sp³ ratios, after 3 MeV H⁺ irradiation of the DBD-produced carbon dust samples. A 10 per cent uncertainty is estimated for calculated CH₂/CH₃, H/C, and sp²/sp³ ratios.

Fluence (protons cm−2)	tirr (h)	Integrated absorbance (cm−1)			k (cm2 g−1)	CH2/CH3	H/C	sp2/sp3
		CH3 as.	CH2 as.  + Fermi. res.	C = C
0	0	0.89	1.18	0.19	88.46	2.0	0.72	2.23
|$1.0\times 10^{14}$|	0.2	0.68	0.63	0.09	58.69	1.4	0.91	1.67
|$3.0\times 10^{14}$|	0.5	0.36	0.48	0.12	43.77	2.0	0.53	3.4
|$6.0\times 10^{14}$|	0.9	0.36	0.5	0.09	27.80	2.1	0.68	2.43
|$1.0\times 10^{15}$|	1.5	0.26	0.44	0.09	40.55	2.5	0.58	2.94
|$4.0\times 10^{15}$|	6.1	0.19	0.3	0.07	38.73	2.3	0.50	3.61
|$1.0\times 10^{16}$|	16.1	0.11	0.23	0.06	25.65	3.1	0.45	3.99

The various forms of carbon-based materials can be described structurally taking into account the hydrogen, sp², and sp³ contents. These three quantities, normalized to 1, are usually plotted as a ternary diagram as shown in Fig. 11, and also proposed for discussions on the physical properties of interstellar carbon dust analogues (Dartois et al. 2007; Chiar et al. 2013; Peláez et al. 2018). Some distinct regions can be clearly identified in the ternary diagram: (1) the ‘no network’ region characterized by a high hydrogen content; (2) the polymers region, characterized by appropriate values of H, sp², and sp³ content for polymers to form; and (3) the regions nearby sp² and sp³ apexes, corresponding to materials close to exhibiting pure graphite and pure diamond structures, respectively. The visual analysis of the ternary diagram and the scattering of various candidates for interstellar dust analogues emphasize the grouping into two families: the polymer-like materials and the a-C:H materials. In the specific case of IRAS 08572 + 3915, the H content between 0.19–0.40 derived from observations induces constraints and highlights a specific region of the ternary diagram (Dartois et al. 2007). There is a good agreement with the properties of many laboratory-produced a-C:H dust analogues, with structural models based on dominant aliphatic network, with small, isolated, aromatic regions or analogues based on polyaromatic networks, with significant aliphatic content as endings. One can note the data clustering in the ternary diagram from many reported interstellar dust analogues at the intersection of the polymers area with that of IRAS 08572 + 3915.

Figure 11.

Ternary phase diagrams for materials containing hydrogen (H), C (sp²), and C (sp³), studied as interstellar dust analogues. Source of data points and constraints: Dartois et al. (2007), Jäger et al. (2008), Gadallah et al. (2011), Chiar et al. (2013), Brown et al. (2014), Molpeceres et al. (2017), Hodoroaba et al. (2018), and Peláez et al. (2018).

It is possible to lock the DBD-produced carbon dust as a-C:H exhibiting a branched aliphatic structure with contributions from multiple C-C bonds. Further processing, using various energetic particles or radiation, usually leads to dehydrogenation and graphitization of interstellar dust analogues. Under the conditions studied in this work, the 3 MeV H⁺ irradiation of the DBD-produced carbon dust induces limited changes in H and sp² contents but a marked left shift in the location of the processed dust inside the region delimited by the constraints from IRAS 08572 + 3915 astronomical observations.

4.2 Destruction rates of aliphatic CH bonds

After irradiation with 3 MeV protons and ex situ analyses, we have observed a gradual decrease of the 3.4 |$\mu$|m band integrated area as a function of proton fluence with a saturation tendency for high fluences (Fig. 12). The behaviour is qualitatively similar to that described in other in situ energetic processing of a-C:H samples using 30 keV He ⁺  (Mennella et al. 2003), 0.2 MeV and 10 MeV H ⁺ , 50 MeV C^5 + , 91 MeV C^6 + , 85 MeV Si^7 + , 100 MeV Ni^9 + , 160 MeV I^21 +  (Godard et al. 2011) or 5 keV electrons (Maté et al. 2016). The plot of 3.4 |$\mu$|m band intensity decay versus protons fluence is presented in Fig. 12. The use of previously proposed data analysis methods, i.e. the exponential fit and the hydrogen recombination model (Godard et al. 2011; Maté et al. 2016) returns the following values for destruction cross-section (⁠|$\sigma _d$|⁠): |$1.69\times 10^{-15}$| cm² from the recombination model and |$4.40\times 10^{-15}$| cm² from the exponential fit. The value of asymptotic relative band intensity (I₀/I_f) is close to 0.3 using both data fitting methods, leading to a value of 43 ų for the characteristic recombination volume within the solid, value typical for ion and electron irradiation experiments (Godard et al. 2011; Maté et al. 2016). It is worth to stress that the values of destruction cross-sections were obtained in ex situ studies on ‘fluffy’ a-C:H interstellar dust analogues, experimental data of this type being unavailable in literature.

Figure 12.

Ex situ analysis of the 3.4 |$\mu$|m band area decay after 3 MeV proton bombardment, using a phenomenological exponential decay and the hydrogen recombination model.

Ion bombardment was carried out at near room temperature. The macroscopic thickness of ‘fluffy’ samples exposed to irradiation ranges between 100 and 200 |$\mu$|m (Hodoroaba et al. 2018). Although this is comparable to the simulated ion penetration depth of 176 |$\mu$|m for a compact material with similar density and stoichiometry, an accurate determination of the samples’ thickness is not possible due to their morphology. While ion processing can generally be considered uniform, it is possible that certain regions of the sample may not receive enough radiation due to variations in morphology. The reduction of the 3.4 |$\mu$|m band intensity after ion irradiation was up to 70  per cent, reaching a saturation point beyond which no further reduction occurs. Mennella et al. (2003) reported a similar finding at 300 K and a lower sample temperature during ion processing (i.e. 12 K) was leading to a further reduction of up to 85  per cent, while complete reduction was not achieved.

The characteristic times for the destruction of aliphatic CH bonds, |$\tau _{d,\mathrm{ CR}}$|⁠, under the action of cosmic rays can be calculated as the inverse of the cosmic rays destruction rate, |$R_{d,{\rm CR}}$|⁠. This last physical quantity might be described, using a monoenergetic (1 MeV) proton beam (Mennella et al. 2003), as the product of destruction cross-section |$\sigma _d$|(1 MeV) and the effective flux of protons |$\phi _p(1{\rm MeV})$|⁠. Using the hypothesis of direct proportionality between the destruction cross-section (⁠|$\sigma _d$|⁠) and the stopping cross-section (⁠|$S_p$|⁠), we can write:

$$\begin{eqnarray} \tau _{d,{\rm CR}}=\left[\frac{S_p(1{\rm MeV})}{S_p(3{\rm MeV})}\sigma _d(3{\rm MeV})\phi _p(1{\rm MeV})\right]^{-1} \end{eqnarray}$$

(6)

Using the values of destruction cross-section (⁠|$\sigma _d$|⁠), from both the recombination model and the exponential fit method, the values of characteristic times for destruction of aliphatic CH bonds are in the range 1–15 millions of years (Myr), i.e. within the typical lifetime of diffuse regions (100 Myr) and dense clouds (30 Myr) (Jones et al. 1994). The following numerical values were used for calculations, returned by srim code (Ziegler et al. 2010) applied for physical characteristics of our interstellar dust analogues (⁠|$\rho _{a-C:H}=0.95$| g cm⁻³, H/C  = 1): S_p(3 MeV)  = 0.119 MeV mg⁻¹ cm², S_p(1 MeV)  = 0.259 MeV mg⁻¹ cm². The analysis of available data from previous studies (i.e. table 5 in Maté et al. 2016 and section 4.1 in Dartois et al. 2017) is allowing to group the results into two families (Table 5): data that support the scenario of unlikely aliphatic CH bonds destruction by cosmic rays (Godard et al. 2011; Maté et al. 2016), and data that support a probable aliphatic CH bonds destruction by cosmic rays (this work, Dartois et al. 2017; Mennella et al. 2003). The relationship between the CH destruction cross-section (⁠|$\sigma _d$|⁠) and the stopping cross-section (⁠|$S_p$|⁠) was previously investigated (i.e. fig. 9 in Dartois et al. 2017). However, establishing a direct correlation between the available |$\sigma _d$| data at high and low |$S_p$| values remains challenging. Additionally, Dartois et al. (2017) suggested that smaller particles may have a larger destruction cross-section than predicted by the aforementioned proposed correlation. The analogue synthesized and tested in this study, generically named ‘fluffy’ dust analogue, consists at nanoscale of an agglomeration of numerous individual flakes (Fig. 3). The calculated |$\sigma _d$| is an order of magnitude larger than the value predicted by the power law used by Dartois et al. (2017) to fit the ion cross-section data. One possible explanation for this discrepancy is the morphological characteristics of the analogue used in this study. In this context, the present study improves the existing literature by providing new data in the low |$S_p$| range, thereby improving the overall understanding of this dependence.

Table 5.

Comparison of available data for destruction rates by cosmic rays for aliphatic CH bonds and destruction time in diffuse regions or dense clouds.

Aliphatic CH bonds	Aliphatic CH bonds	Destruction time (Myr)		Reference
destruction by cosmic rays	destruction rates (yr−1)	Diffuse -regions	Dense clouds
Unlikely	|$(0.95-10.4)\times 10^{-9}$|	100–1000	100–1000	Godard et al. (2011)
	|$(0.59-2.08)\times 10^{-9}$|	200–500	300–900	Maté et al. (2016)
Probable	|$2.52\times 10^{-8}$|	40		Dartois et al. (2017)
	|$0.54\times 10^{-7}$|	20	30	Mennella et al. (2003)
	|$(0.53-1.39)\times 10^{-7}$|	4–10	7–20	This work

5 CONCLUSIONS

Using a DBD in He/C₄H₁₀ mixtures we produced ‘fluffy’ a-C:H samples exhibiting features in the infrared absorption spectra (e.g. the 3.0 |$\mu$|m band) in good agreement with those of carbonaceous dust from IRAS 08572 + 3915. These interstellar dust analogues were characterized using a variety of microscopy, mass spectrometry, and vibrational spectroscopy techniques in order to gain some insights into their structure, from the macro to the nanoscale, and derive their H/C ratio from multiple experimental methods. The average H/C value served then as reference to choose the absorption strength values that allowed the retrieval of the closest matching H/C value from FTIR spectra. In fact, calculation methods based on band intensities measurements of known absorption strengths were debated recently as they can return results with significant deviations if the absorption strengths available in the literature are not adapted to the studied solid dust analogue. We show here that a thorough multitechnique analysis from which can be derived H/C ratios can support the choice of absorption strengths values best adapted to the studied carbon dust analogue. Upon irradiation with 3 MeV H ⁺ , the DBD-produced carbon dust shows morphological and chemical changes. The microscale structure appears modified and indications of ion etching are visible on all investigated samples. The H/C, CH₂CH₃, and sp²/sp³ ratios show an evolution with the proton fluence, whereby the H/C ratio tend to decrease due to the proton bombardment-induced dehydrogenation, whereas the sp²/sp³ ratio increases upon graphitization of the sample. The location of the DBD-produced carbon dust in the ternary phase diagram falls close to many other reported data for dust analogues and is shown to shift further in the IRAS 08572 + 3915 constrained region upon proton bombardment. Based on the ex situ study of the DBD-produced carbon dust, we were able to measure the intensity decay of the 3.4 |$\mu$|m band as a function of proton fluence and calculate the CH destruction cross-sections. Comparison with the lifetimes of diffuse and dense regions in astrophysical environments showed that the observed intensity decay is in good agreement with the basic hydrogen recombination model. Finally, the results are discussed in the light of the relevant literature.

ACKNOWLEDGEMENTS

Ion beam experiments measurements have been performed at 3 MV Tandetron accelerator from ‘Horia Hulubei’ National Institute for Physics and Nuclear Engineering (IFIN-HH) and were supported by the Romanian Government Programme through the National Programme for Installations of National Interest (IOSIN) and all three authors from IFIN-HH (MS, DI, and RFA) were supported from the Nucleus programme (PN 19 06 02 01 and PN 19 06 02 02). We acknowledge the use of FEI TITAN Themis for TEM images at Plateforme de Microscopie Électronique de Lille (PMEL) of Université de Lille, hosted by Institut Michel-Eugène Chevreul (CNRS FR2638). The authors also acknowledge the CaPPA project (Chemical and Physical Properties of the Atmosphere) funded by the French National Research Agency (ANR) through the PIA (Programme d’Investissement d’Avenir) under contract |$\ll$|ANR-11-LABX-0005-01|$\gg$| for their financial support. IT also acknowledges the CaPPA project and the Faculty of Sciences and Technologies of the University of Lille for their visiting scientist fellowship programs. This research was also partially funded by the French National Research Agency (ANR) under the project ≪ANR-23-CE30-0023-03≫ INTERSTELLAR. The authors would also like to acknowledge the contribution of the Centre d’ Etudes et de Recherches Lasers et Applications (CERLA) platform for the materials, equipment, and support staff.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding authors.