Humid Evolution of Haze in the Atmosphere of Super-Earths in the Habitable Zone

Analogs of exoplanetary hazes are produced without water under conditions that simulate a temperate atmospheric photochemical synthesis based on CH₄ and N₂, which is in agreement with Titan's atmospheric composition. Haze analogs are produced with the PAMPRE plasma experiment following a well-established protocol detailed in previous publications (Szopa et al., 2006). The reactor is composed of a stainless steel cylindrical reactor within which a RF-Capacitively Coupled Plasma discharge is triggered by an RF 13.56 MHz frequency generator. A gas mixture of 95% N₂ and 5% CH₄ is injected into the chamber as a continuous flow through polarized electrodes, and it is further extracted by a primary vacuum pump to ensure that gases are homogeneously distributed. The plasma discharge is maintained at a pressure of 0.9 ± 0.1 mbar and at room temperature (21.0 ± 0.1°C). Haze analogs are recovered as a brown powder after 1 day and stored in glass vials.

After synthesis, the samples are stored in an enclosed dark place. The humid aging process is performed by storing the vials in a box at 20°C, where hygrometry is kept stable at 70%. The box is protected from light and is additionally covered with an alumina sheet to enhance UV protection. The aging is performed by exposing the samples to atmospheric humidity at a hygrometry of 70%. A contribution of oxidation with molecular oxygen (O₂) is possible, but it has been shown to be negligible compared to the reaction with ambient humidity (see discussion). The samples are aged for periods that range from 1 month to 3.5 years in the absence of light. Five aging time-durations have been defined as follows: 1 month, 4.5 months, 1.8 years, 2.2 years, and 3.5 years after production.

The analyses are performed on a Fourier transform ion cyclotron resonance mass spectrometer equipped with a 21 tesla superconducting magnet (Hendrickson et al., 2015; Smith et al., 2018).

Ions are generated at atmospheric pressure via micro electrospray (Emmett et al., 1998). This ionization mode is required to recover a fraction of the sample extracted in methanol. For this purpose, 4 mg of the haze analog is dissolved in 1 mL of methanol (CH₃CH₂OH) in a vial. The vial is vigorously stirred for 3 min to solubilize the methanol-soluble species. The brown mixture is then filtered by a 0.2 μm polytetrafluoroethylene (PTFE) membrane filter on a filter holder. The filtered solution is transferred into a vial then dried with nitrogen and stored until analysis. Samples are dissolved in methanol to produce a stock solution of 500 μg/mL. All experiments are recorded in electrospray (ESI) positive ionization mode.

Ions are initially accumulated in an external multipole ion guide (21 T: 1–5 ms) and released m/z-dependently by a decrease of an auxiliary radio frequency potential between the multipole rods and the end-cap electrode (Kaiser et al., 2014). Ions are excited to m/z-dependent radius to maximize the dynamic range and number of observed mass spectral peaks (m/z 150–1500; 21 T: 32–64%) (Kaiser et al., 2013), and excitation and detection are performed on the same pair of electrodes (Chen et al., 2014). The dynamically harmonized ICR cell is operated with a 6 V trapping potential (Boldin and Nikolaev, 2011; Kaiser et al., 2011). Time-domain transients of 3.1 s are acquired with the Predator data station, with 100–500 time-domain acquisitions averaged for all experiments (Blakney et al., 2011). Mass spectra are internally calibrated with a high-abundance homologous series based on the “walking” calibration method (Savory et al., 2011).

The need for ultrahigh resolving power only achievable by FT-ICR mass spectrometry is explained by the chemical complexity of the sample. Figure 1 shows a mass spectrum in positive-ion ESI for the methanol-soluble fraction of the initial sample before aging. Analytes are distributed across a wide m/z range (from 180 < m/z < 1200) in a log-normal distribution with more than 30,000 peaks detected with signal magnitude greater than 6 times the baseline rms noise level, and the repetition patterns observed previously are recovered (Somogyi et al., 2005; Pernot et al., 2010). In Fig. 1, a mass-scale zoom insert between m/z 558.24 and m/z 558.40 highlights the immense compositional complexity, with more than 100 peaks detected within a 0.16 Da mass window. A second insert displays the data over a 0.04 Da m/z range with several fully resolved mass spectral peaks that correspond to compounds that differ in mass by roughly the mass of one electron (me⁻= 548 μDa).

Experimentally measured masses are converted from the International Union of Pure and Applied Chemistry (IUPAC) mass scale to the Kendrick mass scale (Kendrick, 1963) for rapid identification of homologous series for each heteroatom class (i.e., species with the same C_cH_hN_nO_oS_s content, differing only by the degree of alkylation) (Hughey et al., 2001). For each elemental composition, C_cH_hN_nO_oS_s, the heteroatom class, the double bond equivalent DBE (number of rings plus double bonds to carbon) (McLafferty and Turecek, 1993), and carbon number, c, are tabulated for subsequent generation of heteroatom class relative abundance distributions and graphical relative-abundance weighted DBE versus carbon number images. Peaks with signal magnitude greater than 6 times the baseline root-mean-square (rms) noise at m/z 500 are exported to peak lists, and molecular formula assignments and data visualization are performed with PetroOrg software (Corilo, 2014). Molecular formula assignments with an error >0.5 ppm are discarded, and only chemical classes with a combined relative abundance of ≥0.15% of the total are considered.

Infrared transmission spectroscopy is used to obtain information on the chemical composition of the sample (Chatain et al., 2020). Samples are conditioned for IR measurements as a thin deposit on a grid. A total of 1 mg of haze analogs is spread on a thin stainless steel grid of 1.3 cm in diameter. The grid is composed of threads of 25 μm large and meshes of 38 μm that lead to a transmission of 20% in IR wavelengths. The samples are placed inside a glass compartment under vacuum with CaF₂ windows, which is inserted into the sample compartment of a Nicolet 6700 FTIR from Thermo. The glass compartment and the sample are then degassed under vacuum for 30 min to remove water traces adsorbed on the sample. The spectra are acquired by a DTGS detector between 900 and 4000 cm⁻¹, this range being limited at 900 cm⁻¹ by the CaF₂ windows used on the FTIR and the reactor, respectively. We chose a spectral resolution of 2 cm⁻¹ and an average of 250 scans. The width of the IR beam at the focal point is a few millimeters. The spectra are corrected for H₂O and CO₂ atmospheric contributions with the OMNIC software. Baseline is corrected with an adjusted 2nd-order polynomial function using two reference wavelength ranges where the samples do not absorb: 1820–1880 cm⁻¹ and 3790–4000 cm⁻¹.

An elemental composition analysis of the bulk sample is carried out with a FLASH 2000 instrument from Thermo to measure the C, H, N, and O mass percentages in the sample under dark aging through time. The instrument is supplied by using a double reactor configuration arranged for simultaneous CHN and oxygen analysis. Oxygen determination is achieved with an oxygen-specific pyrolysis reactor heated at a temperature slightly above 1060°C. For C, H, and N, the sample is burnt in the presence of oxygen at 1800°C for a few seconds. At this high temperature, organic and inorganic substances are converted into elemental gases, which, after further reduction, are separated in a chromatographic column and finally detected by a highly sensitive thermal conductivity detector.

To further highlight the oxidative processes at work in the presence of humidity, we process the mass spectral data by representing them in several formats.

First, the sum of the relative abundance or signal magnitude for oxygen-containing compounds is calculated and compared to the overall sum of the relative abundances for each sample. Figure 3a provides a rapid visualization of trends in the abundance of oxygen compounds detected with time, with all samples measured in triplicate. The relative abundance of oxygenated species increases logarithmically with time and rapidly in the first year from ∼30% after 1 month to ∼60%; then it plateaus at approximately 70% of the total relative abundance.

Then a sorting between the detected elemental families is performed to better understand the incorporation of oxygen into the samples. Five compound classes are defined: compounds that do not contain oxygen and O₁–O₄ classes (respectively containing 1 to 4 oxygen atoms). The sum of the relative abundances of each compound class is calculated and plotted in Fig. 3b. As observed in Fig. 3a, the O₀ families follow a logarithmic decrease. However, the other families reveal different trends. The O₁ family grows during the first year and then decreases while the O₂, O₃, and O₄ families seem to increase as a function of time. This observation suggests that species oxidize over time to incorporate more oxygen into their structure. In addition, this illustrates that the chemical aging process is complex and often proceeds through multiple generations of oxidation before the products reach greater stability that limits further oxidation.

To confirm the results obtained and facilitate the interpretation, graphical treatments are carried out. We use van Krevelen diagrams (van Krevelen, 1950). This type of graph was originally developed to assess the origin and maturity of kerogen and petroleum through the use of the H/C and O/C ratio (Kim et al., 2003; Wu et al., 2004). It is very suitable for the study of oxidation with humidity. Figure 4a, 4b, and 4c show the van Krevelen diagrams for the three samples aged for 1 month, 1.8 years, and 3.5 years, respectively. Here, each point is a single species present in the mass spectrum, and all species are represented. First of all, we observe that many points are located on the y-axis. Those correspond to O-free species. The other species that spread in both dimensions are oxygenated species. By comparing the three diagrams, we observe that the number of points is much larger in the third sample than in the first one. This confirms the increase in the number of oxygenated species over time observed in Fig. 2. Oxygenated species are more intense as a function of time if we compare the three graphs. The H/C ratio changes very slowly, with a progressive spread of the distribution toward larger H/C ratios.

Figure 4d, 4e, and 4f shows the reconstructed mass spectra of the family of species that contain 8 nitrogen atoms. This family is chosen as an example because it is the most abundant one in the sample. It is divided into 5 subfamilies according to the number of oxygen atoms present. Thus, we observe that the O-free family (blue) is the major one in the first sample but is no longer in the oldest one (3.5 years). In addition, we observe an increase in oxygenated families following the trend observed in Fig. 3. This confirms the previous results concerning the amount of oxygen incorporated in the sample over time.

Elemental analysis was carried out on two samples that were aged for 1 month and 2.2 years, respectively. Obtained percentages of each element composing one fresher sample and the 2.2-year-old one are given in Table 1. As observed, the amount of oxygen of the oldest sample is twice that of the earliest one. The H/C ratio also increases through time, in agreement with the spread toward larger H/C ratios observed in Fig. 4. This global analysis validates quantitatively what is observed in detail for each molecule composing the material with mass spectrometry analyses.

Infrared spectroscopy enables further study of the chemistry behind this evolution. The scope is to gain knowledge on the oxygen position and on the chemical groups formed. Transmission infrared spectroscopy was performed on different-aged analog samples. The aging process ranges from 1 month to 3.5 years to observe differences induced by the incorporation of oxygen. To focus on the evolution of IR absorption bands, IR spectra are normalized according to the maximum peak in each wavenumber range studied. Figure 5 focuses on a wavenumber between 1000 and 1900 cm⁻¹. Panels 5a and 5c show the stacking of IR spectra for the samples. Panels 5b and 5d highlight the evolution of spectra by subtracting the spectrum of the freshest sample. As observed, a band at 1700 cm⁻¹ increased with aging. This band is attributed to the elongation of the double-bond C = O between an oxygen and a carbon, which is consistent with carbonyl (-C(R) = O), amide (-C(NH) = O), and carboxylic acid (-C(OH) = O) groups (Long, 2004). Furthermore, the band of alcohol usually present between 1280 and 1420 cm⁻¹ is absent in the reported IR spectrum (Long, 2004). Therefore, both absences involve a preferential fixation of oxygen in C = O form rather than alcohol or ether. Panels 5c and 5d show the elongation of –CH. As observed, there is a diminution of usual –CH bands at 2800–2900 cm⁻¹ and an increase of the –CH bonded to the aldehyde (-C(H) = O) band at 2600–2800 cm⁻¹. These observations highlight the presence of aldehyde functions in our samples.

Aerosols were exposed to humidity-controlled ambient air and may evolve through a possible contribution on the observed chemical evolution by both humidity (H₂O) and molecular oxygen (O₂). The evolution of the H/C ratios allows us to unequivocally distinguish between the two contributions. Indeed, oxidation with O₂ would lead to a decrease of the H/C ratio, with a change in the amount of the unsaturation, whereas hydrolysis with ambient humidity would lead to an increase of the H/C ratio. Therefore, the spread observed in Fig. 4 toward larger H/C ratios, reinforced by the global elemental analysis reported in Table 1, clearly points to a hydrolysis process with ambient humidity. If present, the oxidation with O₂ has negligible effect at the time scale of the experiment. The observed evolution is the result of a hydrolysis mechanism limited by the adsorption of water vapor at the surface of the aerosol. The observation by infrared spectroscopy of the appearance and increase through time of C = O chemical bonds in the material (Fig. 5) is also consistent with a progressive hydrolysis of the nitrile (-C≡N) and imine (-C = N-) chemical bonds present in the pristine aerosols.

In the present work, we considered the case of photochemical aerosols produced in dry and oxygen-free conditions. This extreme case was chosen to artificially distinguish in the laboratory the oxidation process when aerosols are exposed to humidity from the direct oxygen incorporation process during the aerosol's synthesis. However, in the atmospheres of early Earth and other worlds in habitable zones, water vapor may also participate in the direct synthesis of aerosols if the production occurs in humid tropospheric conditions. To our knowledge, no previous experimental simulations mimicked the formation of aerosols in the case of a N₂/CH₄-based atmosphere in the presence of H₂O.

Yet the contribution of oxygen in the aerosol synthesis has actually been addressed, when studying experimentally the effect of CO in the case of Titan's and Pluto's atmospheres (Hörst and Tolbert, 2014; Jovanović et al., 2020) or CO₂ in the case of the early Earth atmosphere (Trainer, 2013). The additional presence of CO or CO₂ in the initial reactive mixtures led to two main results compared to the N₂/CH₄ reference scenario: a decrease of the total aerosol production yield and the incorporation of oxygen in the aerosol chemical composition. Molecules with up to 2 or 3 oxygen atoms were detected, among them amino acids and nucleotide bases. In planetary conditions, oxygen is, therefore, expected to be incorporated in the aerosols both during their formation and through time by interaction of the aerosols with water vapor.

Even if the mechanisms are not yet totally unraveled, we can still discuss the relative effects of the presence of CO or CO₂ during aerosol synthesis on their final chemical composition as compared to their evolution via exposure to humidity. When CO₂ is present in a N₂/CH₄ reactive gas mixture, carbon incorporated into the aerosols can come from both CO₂ and CH₄ (Hicks et al., 2013). These results suggest that the direct incorporation of oxygen in the aerosols during the synthesis in the presence of CO₂, and possibly CO, would also be coupled with commensurate incorporation of carbon from the CO or CO₂. Under plausible planetary conditions, the effects of both processes on the final aerosol C/O ratio would, therefore, differ and require separate consideration to predict the composition of hazes in the atmospheres of planets within the habitable zone.

The experimental conditions chosen here focus on how aerosols exposed to humidity in the tropospheres of habitable zone exoplanets evolve. It does not take into account all the possible mechanisms that aerosols encounter during their time in the troposphere once they have left the relatively dry conditions of the upper troposphere. Moreover, for this first investigation, humidity was fixed at 70%. The kinetics of how the aerosols change under other humidity values will be an important parameter to be further investigated for addressing various planetary atmospheric conditions.

Nevertheless, we can already roughly discuss the order of magnitude of the timescale of the process studied here to question its viability in exoplanetary atmospheric conditions. The atmospheric humid aging process investigated here is shown to occur at a timescale of the order of a month. After 4 weeks, samples already contain about 35% of oxygenated molecules. Fast reactivity is observed during the first month, which means that the hazes are very prone to chemical evolution in the presence of humidity. This timescale can be compared to the estimated residence time of aerosols in the atmospheres of exoplanets in the habitable zone. The easiest estimation is provided by Earth's case, where the residence time of aerosols in the troposphere is well constrained. In Earth's troposphere, dry and wet deposition removal processes limit the residence time of the aerosols in the troposphere, similarly as it possibly would occur in the lower atmosphere of exoplanets in the habitable zone. The residence time, therefore, varies according to the altitude in the atmosphere, the latitude, and the meteorological conditions (Koch et al., 1996; Papastefanou, 2008). On Earth, it is found to range between 15 days and 1 month, sometimes longer. The timescale of the humid oxidation process has roughly the same order of magnitude as the residence time of aerosols in the troposphere of our habitable planet. The atmospheric evolution process studied here is, therefore, likely to impact significantly the composition of the aerosols before they reach the planetary surface.

Concerning their structure, the production of C = O chemical bonds is highlighted. These chemical bonds are consistent with the formation of carbonyl (-C(R) = O), amide (-C(NH) = O), and carboxylic acid (-C(OH) = O) functions after hydrolysis of the nitrile (-C≡N) and imine (-C = N-) groups present in the pristine aerosols. All these chemical bonds are important for prebiotic chemistry. Carbonyl bonds are precursors of amide bonds, and amide bonds are the bases of peptidic structures and sugars. Carboxylic acids, when adjacent to an amine group (-NH-) provide amino acid functions. As discussed previously, this chemical evolution results in both an increase of the O/C ratio and the H/C ratio. This trend is illustrated by using the graphic representation proposed in Fig. 6. In these, van Krevelen areas were added according to previous works reported in the literature (Rivas-Ubach et al., 2018; Ruf et al., 2018; Moran et al., 2020). As observed, each region is enriched within the oxidation process, leading to a strong intensity of oxidized nitrogen-containing polycyclic aromatic hydrocarbons (N-PAH) and an increase of the intensity of lipids and amino-sugar species. A temporal conversion trend can, therefore, be hypothesized from oxidized N-PAH to lipids and amino-sugar species. This work reveals a humid oxidation process of exoplanetary haze, modifying their chemical composition through time with impact on habitability.