Introduction

Over the last decades, the interstellar sulfur depletion problem, i.e., astronomical observations of lower-than-expected sulfur abundances in dense molecular clouds, represents a fundamental, unresolved dilemma in the fields of astronomy and astrochemistry1,2. Although the exact degree of sulfur depletion in interstellar regions is still a matter of vigorous debate3,4, in molecular clouds such as the Taurus Molecular Cloud (TMC-1), sulfur is observed at fractional abundances some three orders of magnitude less in comparison to cosmic abundances2,5,6. However, in the diffuse regions of space, sulfur is only mildly depleted by ~30% such as toward GX 5-1, a low mass X-ray binary near the galactic center, or the observed amounts of sulfur match cosmic abundances, such as those observed in the local part of our Galaxy3,5,7,8,9. Astronomers and astrochemists theorize that this missing gas-phase sulfur is likely deposited onto interstellar nanoparticles, also referred to as carbonaceous and silicate dust particles (or simply “interstellar grains”)8,10, but both observational information and experimental evidence are lacking.

Since the detection of the very first interstellar sulfur-containing molecule, carbon monosulfide (CS) in 197111, ~40 sulfur-containing molecules have been identified in the interstellar medium (ISM) (Fig. 1)12. In the gas phase, these molecules range from simple triatomics such as hydrogen sulfide (H2S) to thiols (RSH), thioaldehydes (HCSR), thioketenes (RCCS), thioacids (HSOCR), and thioamides (RCSNR’) with R and R’ being organic side chains. Recent mid-infrared observations with the James Webb Space Telescope (JWST) provided fundamental compositional insights of the icy grains in dense molecular clouds, such as in pristine cloud ices toward two background stars NIR38 and J11062113, where carbonyl sulfide (OCS) and possibly sulfur dioxide (SO2) have been detected. Water is ubiquitously found in the interstellar ices amongst the other major ice species: carbon monoxide (CO), carbon dioxide (CO2) and ammonia (NH3)14. Although the simplest sulfur-containing molecule: hydrogen sulfide (H2S) has not been observed yet in interstellar ices via infrared observation13,14,15, chemical models predict that hydrogen sulfide is present in these ices16. Since infrared interstellar ice analyses displayed evidence for compounds more volatile than hydrogen sulfide (such as methane) still remaining on the ices13, non-detection of hydrogen sulfide is likely due to the weak infrared features. Since interstellar dust grains and cold molecular clouds provide essentially the raw material for protoplanetary disks and resulting solar systems17, the astronomy community inferred that interstellar grains and sulfur-carrying molecules are at least partially incorporated into comets and other planetary bodies as the molecular clouds transform into star-forming regions10,18,19.

Fig. 1: Sulfur-bearing molecules identified in the interstellar medium.
figure 1

Atoms are color coded: hydrogen: white, carbon: gray, nitrogen: blue, oxygen: red, sulfur: yellow and silicon: teal.

Full size image

Therefore, the interstellar sulfur reservoir has been linked to observations of comets where sulfur-containing molecules such as hydrogen sulfide (H2S), sulfur monoxide (SO), sulfur dioxide (SO2), carbonyl sulfide (OCS), and carbon disulfide (CS2) have been detected20,21,22,23. Hydrogen sulfide is the most common sulfur species, with an abundance of 1.5% with respect to water identified in the comet C/1995 O1 (Hale-Bopp)20 and at a level of 0.5% in comet C/2014 Q2 (Lovejoy)23. Hydrogen sulfide (H2S) was identified in the nitrogen-rich asteroid (101955) Bennu as well24. Diatomic sulfur (S2) was detected in, e.g., comet C/1983 H1 (IRAS-Araki-Alcock)25 and most recently in 67 P/Churyumov–Gerasimenko26. In 67 P, signal associated with trisulfur (S3; m/z = 96), tetrasulfur (S4; m/z = 128), methanethiol (CH3SH; m/z = 48), and ethanethiol and/or dimethyl sulfide (C2H6S; m/z = 62) were also identified utilizing the Double Focusing Mass Spectrometer (DFMS) of ROSINA (Rosetta Orbiter Spectrometer for Ion and Neutral Analysis) instrument onboard the Rosetta mission26,27. In addition, sulfur molecules (S6, S7, S8) were recently identified in the hexane extracts of the carbonaceous asteroid (162173) Ryugu28. Furthermore, Murchison, Tagish Lake, and Allende carbonaceous chondrites reveal the presence of complex organo-sulfur compounds including alkylsulfonic acids (CnH2n+1SO3H; n = 1–14), inorganic sulfur oxides (SO3−, SxO3−; x = 2–4), and oxyacids (HSxO6−; x = 2–7) suggesting an unresolved complexity of synthetic pathways in deep space29.

To make inroads in solving the interstellar sulfur depletion problem, gas-grain astrochemical models have been developed in an attempt to replicate sulfur depletion in cold molecular clouds21,22. These models speculate that sulfur is incorporated as organo-sulfur species16 or as octasulfur (S8) formed through cosmic ray or ultraviolet-driven processing of interstellar ices30. Laboratory simulation experiments exploiting radiation processing of hydrogen sulfide (H2S) ices provide a key to substantiate these astronomical observations and astrochemical modeling efforts through the detection of S–H stretching feature in the infrared spectra of the processed model ices31,32. In addition to pure H2S ices, mixed ice experiments have been reported incorporating the major species detected in the interstellar ices. Analog ices of H2S or H2S/H2O via UV irradiation31,32,33, X-ray processing34, energetic electron irradiation35, protons36, or hydrogen atom interactions37 conducted so far have identified disulfane (H2S2). Processing of H2S/CO ices is also found to form H2S219,38,39. In these processed H2S and mixed ices, sulfanes up to H2S7 were identified via their subliming mass fragments32,40,41. A recent study by Carrascosa, H. et al., displayed the sulfanes (H2S4 and H2S5) leading to molecular sulfur fragments (S3+ and S4+)32. Laboratory experimental studies of pristine H2S or mixed ices also observed allotropes of sulfur (Sn: S2, S3 and S4)30,40,42,43. However, widely applied infrared spectroscopy (IR) and quadrupole mass spectroscopy (QMS) utilized as tools to identify the products of radiation processing have limitations with extensive fragmentation of the products in the electron-impact ionizer (QMS) and overlapping group frequencies (FTIR), thus, limiting the identification of individual molecules44,45. Therefore, previous laboratory simulation experiments note that a significant fraction of the sulfur species produced remains unidentified35.

Here, by exploiting tunable photoionization reflectron time-of-flight mass spectroscopy (PI-ReToF-MS), we reveal the inventory of sulfur and hydrogen bearing molecules synthesized in astrophysically relevant model ices of hydrogen sulfide exposed to proxies of Galactic Cosmic Rays (GCRs)46 over typical lifetimes of cold molecular clouds from 106 to a few 107 years at temperatures of 5 K47,48. The transition from the cold molecular cloud to the star-forming region is simulated in the temperature-programmed desorption (TPD) phase from 5 K to 330 K. The interstellar analog ices are characterized utilizing FTIR spectroscopy during the radiation processing to probe the consumption of the hydrogen sulfide reactants, whereas subliming species are identified in the gas phase by PI-ReToF-MS. Soft, single-photon photoionization with isomer selectivity via PI-ReToF-MS, coupled with isotopic studies, provided unique evidence in identifying the complex reservoir of sulfur-containing molecules synthesized in these processed ices. These studies provide experimental testimony on two key sinks of sulfur-carrying molecules: polysulfanes with up to eleven sulfur atoms (H2Sn; n = 2–11) and octasulfur (S8), exclusively in the crown conformation, with up to 33% of the hydrogen sulfide reactants converted over molecular cloud lifetimes of 107 years. These observations provide fundamental knowledge as to why sulfur may be depleted on interstellar nanoparticles in cold, dense molecular clouds. Such results direct the astronomy community toward the search for polysulfanes in the gas phase of star-forming regions, where these molecules should be observable as a result of fractionated sublimation from dust grains. Our study further affords persuasive evidence on the molecular level of the source of cyclic octasulfur (S8) detected recently in the carbonaceous asteroid (162173) Ryugu, thus documenting for the first time the link between interstellar matter and planetary bodies in solar systems, including our own.

Results and discussion

Infrared spectroscopy

The deposited and processed ices were characterized using FTIR spectroscopy to identify functional groups. In the neat ices, a combination of symmetric and antisymmetric stretching modes of S–H bonds (v1 and v3) gives a strong characteristic peak of these hydrogen sulfide ices centered at 2552 cm–1. The bending mode (v2) emerges as a peak of weak intensity at 1169 cm−131,49. After irradiation, new absorption features were observed at 2482 and 2443 cm–1. These were attributed as S–H stretching modes of dihydrogen disulfide (H2S2; v5) and higher order sulfanes (H2Sn, n > 2)40,50. A comparison of spectra before and after irradiation is compiled in Fig. 2 with three doses to simulate distinct lifetimes in cold molecular clouds: Fig. 2B) 106 years: 3.6 ± 0.5 eV molecule–1, 2C) 107 years: 36 ± 1 eV molecule–1, and 2D) 5 × 107 years: 180 ± 11 eV molecule–1. Supplementary Table 1 summarizes the IR absorption features of the unirradiated pristine ice and the novel peaks emerging during the irradiation.

Fig. 2: Infrared spectra of hydrogen sulfide (H2S) ices at 5 K.
figure 2

Pristine ice after deposition (A) and after irradiation at a dose of (B) 3.6 ± 0.5 eV molecule–1, (C) 36 ± 1 eV molecule–1, and (D) 180 ± 11 eV molecule–1. The original spectrum (gray) is deconvoluted, showing the peaks assigned to hydrogen sulfide (pink). New peaks from the irradiation (cyan) generate the peak-fitted spectrum (red-dashed).

Full size image

The changes in the column densities of the hydrogen sulfide (H2S) reactant and the dihydrogen disulfide (H2S2) products were traced over the irradiation period to identify the kinetic relationships between the decay of the reactant and the formation or decay of the products. A band strength (A’) of v1/ v3 of 1.12 × 10–17 cm molecule–1 was considered for the amorphous hydrogen sulfide ice51. Column density of H2S2 was computed with an average of two band strengths from literature, 9.9 × 10–17 cm molecule–137 and 2.4 × 10−17 cm molecule−1 52. Supplementary Fig. 1 compares the temporal profiles of column densities of H2S and H2S2 for three irradiation doses. These temporal developments in the column densities as a function of irradiation time were fitted with a kinetic reaction scheme (Supplementary Fig. 2). A second-order decay was observed for H2S, forming dihydrogen disulfide (H2S2) molecules at all doses. The extracted rate constants are compiled in Supplementary Table 2. These data can be exploited to analyze the sulfur budget (Supplementary Fig. 3). More sulfur is obviously consumed through the destruction of hydrogen sulfide than that which can be accounted for in H2S2 alone. Therefore, additional sulfur species, among them potentially infrared-inactive molecules, must have been formed; as the dose increased from 3.6 ± 0.5 to 180 ± 11 eV molecule–1, the fraction of sulfur sequestered in species other than dihydrogen disulfide (H2S2) rapidly increased from 79% to 87%. Sulfur-sulfur stretching and bending modes are not identifiable in mid-IR region, and thereby these molecules were unidentifiable via FTIR spectroscopy. Note that hydrogen sulfide sublimed at ~90 K, and the new absorption features at wavenumbers 2482 and 2443 cm−1 returned to baseline at 320 K. No infrared absorption features were detected on the residues at 330 K.

Photoionization reflectron time-of-flight mass spectrometry

Tunable vacuum ultraviolet (VUV) photoionization (PI) coupled with reflectron time-of-flight mass spectrometry (ReToF-MS) was employed to identify the subliming products isomer selectively; the exploitation of photon energies higher and lower than the adiabatic ionization energies (IEs) of distinct structural isomers affords a selective identification44,45,53. No products were observed in the blank experiments confirming that the products detected are the result of the radiation exposure. At 10.49 eV, sulfanes and sulfur clusters can be ionized (Fig. 3); the TPD analysis revealed both sulfanes (H2Sn, n = 2 − 11) and elemental octasulfur (S8). Overlapping TPD profiles of ions at different mass-to-charge ratios indicate lower-mass ions fragment from the parent molecular ion and are indicated by colored bands in Figs. 3 and 4. Sulfane (H2Sn) fragmentation to H2S and an Sn-1+ elemental sulfur fragment was observed for all higher-order sulfanes (H2Sn, n > 2). Experiments with deuterium sulfide (D2S) ice (Fig. 4) confirmed the assigned molecular formula at each mass-to-charge ratio (m/z).

Fig. 3: Temperature-programmed desorption (TPD) profiles for the indicated masses observed in the PI-ReToF mass spectrometer at 10.49 eV photons for processed hydrogen sulfide (H2S) ice.
figure 3

The colored lines connect the sublimation events that take place at the same temperature. Each colored line begins at the molecular ion of the parent peak and links to the observed fragmentation of it. Arranged in order of low to high sublimation temperatures; Red: H2S2, dark blue: H2S3, dark green: H2S4, peach: H2S5, brown: H2S6, purple: H2S7, light blue: H2S9, pink: S8. Fragments along the light green colored line are projected to be fragments of H2S10. The panels of single sulfur species show the sublimation of H2S centered around 90 K.

Full size image
Fig. 4: Temperature-programmed desorption (TPD) profiles for the indicated masses observed in the PI-ReToF mass spectrometer at 10.49 eV photons for processed deuterium sulfide (D2S) ice.
figure 4

The colored lines connect the sublimation events that take place at the same temperature. Each colored line begins at the molecular ion of the parent peak and links to the observed fragmentation of it. Arranged in order of low to high sublimation temperatures; Red: D2S2, dark blue: D2S3, dark green: D2S4, peach: D2S5, brown: D2S6, purple: D2S7, light blue: D2S9, pink: S8. Fragments along the light green colored line are projected to be fragments of D2S10. The panels of single sulfur species show the sublimation of D2S centered ~90 K.

Full size image

Only ionization energies for dihydrogen disulfide (H2S2) of 9.06 ± 0.2 eV and dihydrogen trisulfide (H2S3) of 9.09 eV as an upper limit are available from the literature54,55. Therefore, for larger sulfanes, the adiabatic ionization energies (IEs) were calculated for H2Sn (n = 2 − 8) molecules at the F12-TZ//B3LYP/aVTZ level of theory including the zero-point vibrational energy (ZPVE) corrections. To verify the methods, higher-quality, fully-optimized CCSD(T)-F12b/cc-pVTZ (F12-TZ) was used for the calculation of the IEs and relative energies of H2Sn (n = 2 - 4). The ionization energies differ only by 0.01 eV between the F12-TZ//B3LYP/aug-cc-pVTZ approach and the F12-TZ calculations, and the relative energies by less than 1.0 kJ mol−1. Geometries of the B3LYP/aug-cc-pVTZ optimized neutral molecules are provided as a Supplementary Data File. The IEs are corrected for the thermal and Stark effect by 0.03 eV, and the combined error limits of ± 0.05 eV (Supplementary Data File). Sulfanes up to H2S11 were identified and confirmed in the deuterated ice experiment by a mass shift of 2 amu. In addition, the natural isotopic distribution of signal was observed in mass channels differing by 2 amu for 32S versus 34S for sulfanes and deuterated sulfanes, where the respective H234S32Sn-1 or D234S32Sn-1 are expected (Figs. 3 and 4). Photon energies of 9.34 eV, 9.08 eV, 8.81 eV, 8.34 eV, and 8.17 eV (Fig. 5) were exploited to distinguish between isomers. Figure 6A–F compile the resulting TPD profiles observed at these photon energies for H2S2 (m/z = 66), H2S3 (m/z = 98), H2S4 (m/z = 130), H2S5 (m/z = 162), H2S6 (m/z = 194), and H2S7 (m/z = 226).

Fig. 5: Computed adiabatic ionization energies (IE) of isomers of sulfanes H2Sx; x = 2–7 (top), H2S8 and S8 (bottom) along with the different photon energies for the ReTOF-MS experiments.
figure 5

Ranges of IEs incorporated with thermal and Stark effects (gray) and calculated IEs of isomers/conformers (black lines). Colored dashed lines indicate the VUV photon energies used in experiments to ionize subliming molecules in the gas phase.

Full size image
Fig. 6: PI-ReToF-MS signals detected for sulfanes, H2Sn (n = 2–7), at different VUV photon energies as a function of temperature during the TPD for hydrogen sulfide ices.
figure 6

Sublimating profiles for A H2S2 at m/z = 66, B H2S3 at m/z = 98, C H2S4 at m/z = 130, D H2S5 at m/z = 162, E H2S6 at m/z = 194, and F) H2S7 at m/z = 226 are compared at different photoionization energies. Signals observed with photon energies higher than the calculated IEs of molecules.

Full size image

H2S2

A sublimation peak at 150 K at m/z = 66 is observed at 9.34 eV; this sublimation event diminishes to a few counts at 9.08 eV and is absent at 8.81 eV (Fig. 6A). The H2SS isomer (IE = 8.68–8.78 eV) was not identified; therefore, ion counts at m/z = 66 are assigned to HSSH (C2h; IE = 8.97–9.07 eV).

H2S3

A signal at m/z = 98 is observed at 9.34 eV and 9.08 eV, peaking at 179 K; these ion counts are absent at 8.81 eV (Fig. 6B). The HSSSH isomer (IE = 8.91–9.02 eV) is the most probable assignment as SSH2S (IE = 9.14–9.24 eV) would not produce a signal at 9.08 eV and SHSHS (IE = 8.67–8.77 eV) should be observed at 8.81 eV if present.

H2S4

A sublimation peak at 197 K was observed for m/z = 130 at 9.08 eV (Fig. 6C), whereas no signal is evident at 8.17 eV. This eliminates the S-SH-SH-S(a) isomer, but the overlap of ionization energies of the remaining isomers prevents a more specific identification.

H2S5

Signal at m/z = 162 was observed at 8.75 eV showing a broad sublimation peak at 225 K that disappears at 8.34 eV (Fig. 6D).

H2S6

The adiabatic ionization energies of 15 conformers of H2S6 were calculated and they range between 8.47 and 8.87 eV. A sublimation peak was observed at 240 K for m/z = 194 at 9.08 eV, but not at 8.34 eV (Fig. 6E) thus confirming the identification of H2S6.

H2S7

The profile at m/z = 226 showed a sublimation peak at 268 K (Fig. 6F) at 8.81 eV, but no signal at 8.34 eV; this is consistent with the ionization range (IE = 8.40–8.75 eV) of the 23 calculated conformers of H2S7.

H2S8

Further isotopic analysis was required to identify the presence of H232S8 at m/z = 258; this is shared with 32S734S molecules and m/z = 260 in the deuterated ice experiment, comprised of both D232S8 and 32S634S2. The overlapping signals of H2S8 and S8 are discussed below.

H2S9

Ionization energies were not calculated for H2S9 structures due to the overwhelming number of isomers and conformers possible for the molecular formula. A sublimation peak was observed for m/z = 290 at 280 K in the 10.49 eV photoionization experiment which disappears when photon energy is lowered to 8.34 eV.

H2S10 & H2S11

The parent ions for H2S9, H2S10, and H2S11 were not observed, but their formation is justified using their fragments. Supplementary Fig. 4 overlays the TPD sublimation plots of sulfanes (H2Sn, n = 2–10) and molecular sulfur ions (Sn+, n = 2–10) at 10.49 eV (Supplementary Fig. 4: left) in comparison to the respective deuterated sulfanes (D2Sn, n = 2–10) and the Sn+ ions in the deuterated ice experiment (Supplementary Fig. 4: right). At 10.49 eV, fragmentation of the parent ion is less intense in small sulfanes up to H2S4. However, larger sulfanes exhibit fragmentation in which the fragment ions show greater intensity than their parent signals. Using the peak sublimation temperature of the S9+ fragment, the sublimation temperature of its predicted parent sulfane, H2S10, was determined to be 291 K. Likewise, exploiting the sublimation peak of S10+ fragment, the sublimation temperature of H2S11 was determined to be 297 K. Regression curves were utilized to identify the relationships between sublimation temperatures for directly observed sulfanes (H2S2 to H2S9) and their Sn-1+ mass fragments (Supplementary Fig. 5). These curves were extended to the predicted higher order sulfanes, H2S10 and H2S11, and show good agreement between the fragment sublimation temperature. In the irradiated deuterium sulfide ice (D2S), sublimation peaks for D2S2 (m/z = 68), D2S3 (m/z = 100), D2S4 (m/z = 132), D2S5 (m/z = 164), D2S6 (m/z = 196), and D2S7 (m/z = 228) were observed at 139 K, 173 K, 191 K, 218 K, 237 K, and 253 K, slightly below the peak sublimation temperatures observed for the corresponding sulfanes (H2Sn, n = 2–7).

The F12-TZ/B3LYP/aVTZ adiabatic ionization energies and relative energies of six conformers and isomers of S8 report errors of ±0.05 eV (Supplementary Table 3). The crown, which results in a twisted crown conformer upon ionization, the chair, and boat conformers represent eight-membered closed ring structures. An isomer labeled chain is an open-ended sulfur chain (Supplementary Table 3). The S7S isomer represents a seven-membered sulfur ring with an exocyclic sulfur attached to the ring. Figure 7A displays the signal observed for m/z = 256 at each photon energy utilized. At 9.08 eV, two sublimation peaks are present, indicating the possibility of S8 conformers/isomers. Upon lowering the photon energy to 8.75 eV, which is below the adiabatic ionization energy of the crown conformer, no ion signal was observed. Therefore, only the crown isomer was formed and identified in the processed ices. In the deuterated ice experiment, m/z = 256 showed two-sublimation peaks arranged the same as in the non-deuterated ice. No other elemental sulfur molecules were identified, as signals for lower mass Sx species, i.e., S2+ (m/z = 64), S3+ (m/z = 96), S4+ (m/z = 128), S5+ (m/z = 160), S6+ (m/z = 192), and S7+ (m/z = 224), are fragments of higher order sulfanes (Figs. 3 and 4).

Fig. 7: Identification of signal for octasulfane (H2S8) and octasulfur (S8) in the shared mass channels.
figure 7

PI-ReToF-MS signals detected for A) octasulfur (S8) at different VUV photon energies as a function of temperature during the TPD for hydrogen sulfide ices. B) Identification of H2S8 utilizing isotopic distribution of S8 in their shared mass channels. Sublimating profiles for S8 at m/z = 256 at different photoionization energies are compared. Signals observed with photon energies higher than the calculated IEs of molecules. H–K Photoionized ion signals recorded via ReToF-MS as a function of temperature at m/z = 257 is compared with m/z = 260 considering the natural isotopic abundance to identify the composition of the signal observed at m/z = 256.

Full size image

To investigate the presence of H2S8 in m/z = 258, which is shared with 32S734S, the natural isotopic distribution for S8 signal was considered. Natural isotopic distribution of sulfur is well known in literature56. For S8, the signal distribution expected in mass channels m/z = 256 to m/z = 262 can be calculated utilizing the natural isotope probabilities and the combinations of isotopes, resulting in each mass. (Supplementary Information) For m/z = 257, only contributor is 32S733S, a 4.25% distribution of S8 signal. Whereas m/z = 260, isotopologous channel of S832,S634S2 is only a minor contributor with only 3.7%. Signal observed at m/z = 260 can also contain signal for H232S734S in addition to the contribution of S8: 32S634S2. The signal expected at m/z = 260 for 32S634S2 was calculated by utilizing the m/z = 257 signal for 32S733S and correcting for the isotopic distribution for 32S634S2. Figure 7B compiles the signal observed for m/z = 257 (black), m/z = 260 in (red), and the calculated signal for 32S634S2 (m/z = 260; blue). This calculated 32S634S2 signal (blue) was subtracted from the observed m/z = 260 signal (red) to reveal the contribution from H232S734S (yellow).

These profiles are displayed in Fig. 7B for photon energies of 10.49 eV (H), 9.34 eV (I), 9.08 eV (J), and 8.81 eV (K). At 10.49 eV, a higher degree of H2S8 fragmentation is observed, and S8+ fragments from H2S9 may also contribute to this signal. Thus, an H2S8 contribution to m/z = 260 could not be observed at 10.49 eV. However, H2S8 contribution is observed at 9.34 eV and 9.08 eV for the 270 K peak but not for the 310 K peak. At 8.81 eV, the 310 K peak is absent, suggesting that this peak is linked only to S8 sublimation because this photon energy falls into the ionization energy range (8.77–8.87 eV) of the crown isomer of S8. The peak at 270 K shows no contribution from H2S8, indicating that this peak is solely due to the S8+ fragment from H2S9.

Quantification of products

Having provided compelling evidence on the identification of sulfanes (H2Sn; n = 2–11) and octasulfur (S8) molecules in interstellar analog ices composed of hydrogen sulfide via PI-ReToF-MS, we now focus on the quantification of these observed irradiation product molecules. Comparing the infrared spectra of the ice before and after the electron irradiation, 1.0 ± 0.2 × 10–18 molecules cm–2 of hydrogen sulfide was found to be decayed off at the irradiation dose of 7 ± 1 eV molecule–1. The yields for sulfanes from H2S2 to H2S8 and S8 molecules were quantified with the help of the photoionization cross sections at both 10.82 eV and 10.49 eV. In the 10.82 eV photoionization experiment, a yield of (3.5 ± 0.2) × 10–3 molecules eV–1 was observed for the simplest product: H2S2. A yield of (1.2 ± 0.2) × 10–3 molecules eV–1 of H2S8 and a (7.0 ± 1.0) × 10–4 molecules eV–1 of S8 were observed. Consistently, calculations using the 10.49 eV photoionization data, yields (4.6 ± 0.2) × 10–3 molecules eV–1 H2S2, (1.1 ± 0.5) × 10–3 molecules eV–1 H2S8, and (6.8 ± 0.3) × 10–4 molecules eV–1 of S8. Table 1 summarizes the calculated photoionization cross sections, yields, and the product percentages obtained from photoionization energies 10.82 eV and 10.49 eV. Supplementary Fig. 7 summarizes the percentages of products identified on the hydrogen sulfide ices. When comparing the average branching ratios, the smallest sulfane H2S2 shows the highest yield of 21.9 ± 0.3%, whereas the yield drops as the number of sulfur atoms in the product molecule increases from two to eight; this finding suggests stepwise molecular mass growth processes during the radiation exposure. Overall, FTIR observed that 33 ± 4% of the deposited hydrogen sulfide molecules were converted to these products. Molecular sulfur and sulfane molecules (H2S2 to H2S8) account for 80 ± 6% of the sulfur from the decayed sulfur budget. The finding of octasulfur rings exclusively in crown conformers as the only molecular sulfur product indicates the directed reaction pathways to this molecule. A staggering 4 ± 1% of hydrogen sulfide was transformed into these S8 molecules at the irradiation dose of 7 ± 1 eV molecule–1 which simulates a typical lifetime of a few 107 years in the molecular clouds.

Table 1 Yields and branching ratios calculated for the irradiation products in the hydrogen sulfide ices at 10.82 eV and 10.49 eV photoionization
Full size table

Electron-impact quadrupole mass spectroscopy—residual gas analysis (EI-QMS/RGA)

The gas-phase analysis with QMS provided insights into the mass-to-charge ratios (m/z) of fragments of the subliming molecules in the TPD phase. Confirming the infrared and PI-ReToF-MS identifications, RGA data also identified sulfanes: H2S2 (m/z = 66), H2S3 (m/z = 98), H2S4 (m/z = 130), and H2S5 (m/z = 162) via their molecular ions, and signals in respective channels of natural isotopic distributions (Supplementary Fig. 7). Evidence for S2+ (m/z = 64), S3+ (m/z = 96), S4+ (m/z = 128), S5+ (m/z = 160) and S6+ (m/z = 192) fragments were observed as well. RGA data confirmed that the signal for Sn+ observed in the mass channels are fragments of the respective H2Sn+1. RGA recorded signals agreed well with identifications made via PI-ReToF-MS data. The ability to reduce fragmentation with the tunable PI and the higher sensitivity provided by the PI-ReToF-MS over EI-QMS/RGA results in less ambiguity in identification. The sublimation temperatures of the sulfanes observed via RGA were also plotted against the sublimation temperatures identified with the ReToF data for the sulfanes H2S2 to H2S11. (Supplementary Fig. 5B). RGA observed fragments S5+ (m/z = 160) and S6+ (m/z = 192) were extrapolated to be fragmented from H2S6 and H2S7, respectively. However, beyond the sublimation of H2S7 at 270 K, the identification of higher-order sulfanes utilizing RGA data becomes difficult. The trendlines created using the sublimation temperatures of sulfanes as determined by ReToF and RGA data overlap in their predicted error ranges.

The sulfur loss in the dense molecular clouds described as ‘the sulfur depletion problem’ has represented an enigma for astronomers and astrochemists. Laboratory simulation studies so far have failed to provide a comprehensive explanation, resorting to limitations in detection techniques utilized. The present laboratory study focuses on the identification of astrochemical changes and transformation of the hydrogen sulfide (H2S) ice deposited on interstellar dust grains in the cold molecular cloud conditions upon the radiation exposure of GCR proxies in the order of 107 years via condensed phase FTIR and tunable VUV PI-ReToF-MS to identify the products isomer selectively during TPD in the gas phase. GCR proxy exposure of hydrogen sulfide ices identified a homologous series of sulfanes in the form of H2Sn, ranging from two to eleven sulfur atoms, and molecular sulfur, only in one specific ring conformer of S8, the thermodynamically most stable crown conformer57, as irradiation products. Signals observed at mass channels corresponding to molecular sulfur (Sn) are observed to be fragments of the higher order sulfane (H2Sn+1), extending the observations by Carrascosa, H. et al. to all sulfanes observed32. Isomer-specific identifications of HSSH (m/z = 66), HSSSH (m/z = 98) were possible for H2S2 and H2S by fine-tuning the photoionization energy at the TPD phase. Overlapping ionization energies prevented the isomeric identifications of larger sulfanes. No previous study experimentally observed octasulfur in the processed ices, even if octasulfur was assumed to form on these model ices that identified molecular sulfurs (Sn)30,40,41,42,43. These sulfanes and octasulfur (S8) may hold the key to the missing sulfur of the dense clouds; polysulfanes undergo fractionated sublimation from 150 K to 297 K; octasulfur sublimed at 310 K in the warm-up phase, which simulates the molecules subliming into the gas phase once the molecular cloud transits to star-forming regions.

Proxies for the effect of GCRs on the icy grains over a typical lifespan of cold molecular clouds are replicated in these experiments at astrophysically relevant temperatures and pressures. The secondary electron cascades formed in the ices are replicated with energetic electron irradiation. These electrons deposit the energy that drives photochemical reactions on the reagents in the ice. Having identified the products formed on processed hydrogen sulfide ices resulting from energetic electron radiation, potential formation mechanisms are investigated to link the reactants to the identified products. During the irradiation of the deposited ices to a typical depth of 350 nm, depositing a dose of 7 eV molecule–1, the energetic electrons can easily break the S–H bonding of hydrogen sulfide (H2S) generating suprathermal sulfur and hydrogen atoms58,59. Endoergic homolytic bond cleavage of hydrogen sulfide (H2S) can form sulfanyl (HS) and hydrogen (H) radicals, which is endoergic by 376 kJ mol–1, or sulfur in its first electronically excited state (S(1D)) and molecular hydrogen; this process is endoergic by 405 kJ mol–1 (reactions (1) and (2))60,61.

$${{{\rm{H}}}}_{2}{{\rm{S}}}\to {{\rm{HS}}}+{{\rm{H}}}$$
(1)
$${{{\rm{H}}}}_{2}{{\rm{S}}}\to {{\rm{S}}}({}^{1}{{\rm{D}}})+{{{\rm{H}}}}_{2}$$
(2)

The formation of the simplest sulfane molecule, disulfane (H2S2) can proceed subsequently via the barrierless recombination of sulfanyl radicals (HS) or via insertion of sulfur atoms into a hydrogen-sulfur bond of the hydrogen sulfide molecules through reactions (3) and (4).

$${{\rm{HS}}}+{{\rm{HS}}}\to {{{\rm{H}}}}_{2}{{{\rm{S}}}}_{2}$$
(3)
$${{{\rm{H}}}}_{2}{{\rm{S}}}+{{\rm{S}}}({}^{1}{{\rm{D}}})\to {{{\rm{H}}}}_{2}{{{\rm{S}}}}_{2}$$
(4)

Sulfane molecules formed on the ice can also undergo further homolytic bond cleavage reactions with the necessary energies obtained from the energy deposited on the ice during the irradiation. Radiolytic decomposition of disulfane via a hydrogen loss is endoergic by 117 kJ mol–1 (reaction (5))62. Formation of higher-order sulfane molecules (H2Sn; n > 2) can be explained via one of the two processes described below. Here, atomic sulfur insertion into already formed sulfane molecules via reaction (6), and secondly, via barrierless radical recombination processes described in reaction (7).

$${{{\rm{H}}}}_{2}{{{\rm{S}}}}_{2}\to {{\rm{H}}}+{{{\rm{HS}}}}_{2}$$
(5)
$${{\rm{S}}}+{{{\rm{H}}}}_{2}{{{\rm{S}}}}_{{{\rm{X}}}}\to {{{\rm{H}}}}_{2}{{{\rm{S}}}}_{({{\rm{X}}}+1)}$$
(6)
$${{{\rm{HS}}}}_{{{\rm{X}}}}+{{{\rm{HS}}}}_{{{\rm{Y}}}}\to {{{\rm{H}}}}_{2}{{{\rm{S}}}}_{({{\rm{X}}}+{{\rm{Y}}})}$$
(7)

The observed yields drop as the number of sulfur atoms in the product molecule increase from two to eight complements this conceptual molecular mass growth route forwarded.

In the case of elemental sulfur molecules of cyclic octasulfur (S8), these can be formed by sulfur atoms preferentially showing catenation, leading to the most stable elemental sulfur form57. The drastic drop in the observed product yield for octasulfane (H2S8) poses a possibility of the formation of octasulfur via molecular hydrogen loss. The presence of other components in these interstellar model ices, such as water, can affect the reactions forwarded via parallel competing reactions. If the energetics of the steps are favorable, these radiation initiated reactions would proceed in these ices, regardless of the size or complexity of product molecules63,64. In pure hydrogen sulfide ices, in the abundance of hydrogen atoms, sulfanes seem the most favorable product in comparison to molecular sulfurs (Sn), except for octasulfur.

Pristine hydrogen sulfide sublimes at a relatively low temperature of 90 K, but in the presence of less volatile irradiation products or other components, such as water, entrapping hydrogen sulfide in the matrix, hydrogen sulfide can co-sublime at warmer temperatures. In these laboratory simulation experiments, an ice thickness greater than the penetration depth of energetic electrons is utilized to avoid any reactions at the interface of the ice with the substrate (silver wafer). Thus, the FTIR measure of 33% of reactants converted to products cannot be compared with the interstellar ice composition of hydrogen sulfide. However, if we consider the conversion rate of ice processed by impinging electrons, 80% of the destroyed hydrogen sulfide was identified as converted to sulfanes and octasulfur. Nevertheless, efficient conversion of hydrogen sulfide upon ionizing radiation does not imply an absence of hydrogen sulfide in interstellar ices. The observation of about 1% hydrogen sulfide (with respect to water) in several comets21,65 cannot be quantified with the present study either.

Even though the analog ices studied here can be utilized to model the transition of ices coated on dust from denser clouds to star-forming regions, a sulfur anomaly is reported towards planetary nebulae as well66. Circumstellar sulfur deficits in oxygen-rich objects are predicted to be in small molecules of SO, SO2, H2S, or CS, and in carbon-rich objects, via magnesium or iron sulfides, but underestimation of ionized sulfur species in the gas phase is seen in many current model studies67,68. Current models on protoplanetary disks favor interpreting the sulfur composition of interstellar dust available to incorporate in planet formation predominantly of FeS and other sulfide minerals at higher temperatures relying on the sublimation temperatures of water or smaller sulfur chains (Sn) subliming below 200 K69, the sublimation temperatures of higher order sulfanes and octasulfur identified in these hydrogen sulfide ices span up to 330 K. Simulations, or models can utilize these novel volatiles and possible reaction pathways to update the inventories, eventually reaching to realistic predictions about these regions of ISM. Visible residues left on the substrates at 330 K did not show any characteristic features in the mid-IR region, thus reiterating the functional limitations of FTIR in interstellar characterizations. Therefore, it should be noted that interstellar observations relying on a single technique, such as infrared, heavily undermine the present-day knowledge of realistic ice compositions. The identified products: sulfanes and octasulfur, should be observable subliming to the gas phase in warmer regions of the ISM above their respective sublimation temperatures, via radio or far infrared (400—200 cm–1) measurements70.

Sulfur is the tenth most abundant element in the universe and the fifth most abundant on Earth. Considering the planets in our Solar System, octasulfur is confirmed on the Earth and recently on the surface of Mars by NASA’s Curiosity Mars rover71. Samples from near–Earth Asteroid (162173) Ryugu identified octasulfur rings28. Additionally, Galilean moons with volcanic activity are assumed to contain octasulfur as well72,73. Even though molecular sulfur deposits on the Earth are believed to be of volcanic and bacterial origins74 and the history of volcanism for Mars75, octasulfur on comets and meteorites might originate from interstellar icy mantles. Although the conversion rate of hydrogen sulfide to cyclic octasulfur of about 4% as derived in our ice experiments, might appear low, this number can be placed in an astronomical context. For the TMC with an age of 1–2 × 106 years76,77 and the S/H ratio of 1.5 × 10–578, about 2.1 × 1027 kg of octasulfur (S8) can be produced on the icy grains (Supplementary Information). This amount equals 350 times the weight of Earth. Trapped amongst refractory or less volatile matrices, only a minor fraction of elemental sulfur synthesized in the molecular cloud from which our Solar System originated had to be eventually delivered to proto-Earth and other scattered objects, thus rationalizing the contemporary geochemical ties between the interstellar clouds and objects in circumstellar disks, providing insights to the interstellar sulfur puzzle.

Methods

The experiments were conducted in a hydrocarbon–free stainless steel ultrahigh–vacuum chamber at pressures of a few 10−11 Torr generated by magnetically suspended turbomolecular pumps (Osaka, TG420MCAB and TG1300MUCWB) backed by an oil-free scroll pump (Edwards, GVSP30). Inside the chamber, a highly-polished silver wafer is placed on a freely-rotatable cold finger, which can be cooled down to 5.2 ± 0.2 K using a two–stage, closed–cycle helium compressor (Sumimoto Heavy Industries, RDK-415E). Hydrogen sulfide gas (H2S, Sigma Aldrich, ≥99.5%) was introduced into the chamber via a glass capillary array and condensed on the silver wafer at a pressure of 4 × 10−8 Torr. The deposition of the hydrogen sulfide ice on the silver wafer was monitored in situ using a HeNe laser (MellesGriot, 25LHP-230, λ = 632.8 nm). A refractive index (nice) of 1.41 and a band strength A’(S–H) of 1.12 × 10−17 cm molecule−1 were employed for the amorphous hydrogen sulfide ice51,79. The refractive index (n) was utilized in Eq. (8) to determine the thickness (d) of the deposited ices using the interference pattern of the HeNe laser at an angle of incidence θ = 4⁰;

$$d=\frac{m \lambda } {\root 2 \of {{n}_{{ice}}^{2}-{sin }^{2} \theta }}$$
(8)

The fringes of the laser interference pattern (m) were monitored while depositing. Depositing 4 interference fringes of ice, the average thickness of these deposited ices was calculated to be 1300 ± 300 nm. To determine the column densities of the deposited ice (NS-H), Eqs. 9 and 10 were utilized:

$${{N}_{S-H}}=\frac{{{\mathrm{ln}}}(10){{\int }_{\bar{\nu }_{1}}^{\bar{\nu }_{2}}}A\left({\bar{\nu}} \right)d{\bar{\nu }}\cos (\beta )}{2{A}^ {{\prime} }(S-H)}$$
(9)

where, \({\int }_{\bar{\nu }_{1}}^{\bar{\nu }_{2}}A\left({\bar{\nu}} \right)d\bar{\nu }\) is the integrated peak area in wavenumbers (cm–1), infrared band strength A’, a factor of 2 accounting for the incoming and outgoing beams, and the angle of light passing through the ice relative to the normal surface (β). Applying Snell’s law on the incoming beam of light in vacuum (nν = 1), to the refractive index of hydrogen sulfide ice (nice), and angle of incidence of the infrared beam α = 43⁰;

$$\beta={\sin }^{-1}\frac{{n}_{\nu }}{{n}_{{ice}}}\sin (\alpha )$$
(10)

To simulate the effects of galactic cosmic rays and the secondary electrons they produce on extraterrestrial ices, hydrogen sulfide ices were processed with 5 keV energetic electrons isothermally at 5 K. Four different doses were utilized to irradiate the ices for 60 minutes: two low dose experiments with 3.6 ± 0.5 eV molecule–1 and 7 ± 1 eV molecule–1, a medium dose up to 36 ± 1 eV molecule–1 and a high dose up to 180 ± 11 eV molecule–1. Energetic electrons were generated with an electron gun (Specs PU-EQ 22). Calculated irradiation doses and the parameters used can be found in Supplementary Table 4. The average penetration depth for the irradiated energetic electrons was found to be 350 ± 30 nm using CASINO simulations and 90% of the energy is deposited in the first 420 ± 30 nm of these ices80.

After the irradiation, the ices were warmed up from 5 K to 330 K using TPD at 1 K min–1 by a temperature programmable controller (Lake Shore 336), simulating the transformation of cold molecular clouds to star-forming regions. A Fourier transform infrared spectrometer (FTIR, Nicolet 6700) with a mercury-cadmium-telluride (Thermo, MCT-B) detector was used in the range of 6000–500 cm–1 and a spectral resolution of 4 cm–1 to monitor the functional groups of the condensed ices in situ after deposition. FTIR Spectra were measured continuously and averaged every two minutes during irradiation and TPD to identify the transformation of the functional groups of the ices while being irradiated.

Photoionization experiments that utilize the four-wave mixing schemes to generate vacuum ultraviolet (VUV) light at energies 10.82, 10.49, 9.34, 9.08, 8.81, 8.75, 8.34 and 8.17 eV were carried out at the W.M. Keck Research Laboratory in Astrochemistry. Reflectron time-of-flight mass spectrometry (Jordan TOF Products, Inc.) was used to analyze the molecules subliming off the silver wafer after vacuum ultraviolet (VUV) photoionization (PI-ReToF-MS) at different energies. Coherent VUV light was generated at 30 Hz via resonant or non-resonant four-wave mixing processes. The third harmonic of a pulsed neodymium yttrium-aluminum garnet (Nd:YAG, Spectra-Physics, Quanta Ray PRO 250-30) laser was utilized to generate the 10.49 eV photons via frequency tripling in xenon (99.999%) as the non-linear medium. The third harmonic of another pulsed Nd:YAG laser (Spectra-Physics, Quanta Ray PRO 270, 30 Hz) was used to pump a dye laser (Sirah Lasertechnik, Cobra-Stretch) containing Stilbene-420 dye to obtain 425.112 nm, which undergoes second harmonic generation to produce 212.556 nm (ω1). The 9.34 eV energy was generated by spatially overlapping and time synchronizing colinear beams of ω1 (212.556 nm) with the third harmonic of the other pulsed Nd:YAG laser (ω2 = 355 nm) in an evacuated chamber pulsed with jets of krypton gas. To generate 8.75 eV vacuum ultraviolet (VUV) light, above-generated coherent beam 425.112 nm dye-laser output was overlapped with its second harmonic; 212.556 nm (ω1) which essentially utilized only one Nd:YAG laser. Likewise, in generating other wavelengths listed in the Supplementary Table 5, four-wave mixing processes were utilized with the corresponding dye solutions indicated.

The generated VUV photons (ωVUV) were spatially separated from the fundamental wavelengths by transmission through a biconvex lithium fluoride lens (ISP Optics) placed off-axis and passed into the main chamber to ionize the subliming molecules during TPD. The photoionized molecules were then mass-analyzed via ReToF-MS. In addition, an electron-impact (100 eV) quadrupole mass spectrometer (EI-QMS, QMG 420) in the RGA mode monitored the fragments of gas-phase sublimation products.

Theoretical methods

The relative energies and adiabatic excitation energies are computed from a combination of single-point explicitly correlated coupled-cluster singles, doubles, and perturbative triples (CCSD(T)-F12b) energies with the cc-pVTZ-F12 basis set81,82 from the MOLPRO 2024.1 program83,84,85,86,87 computed at optimized B3LYP/aug-cc-pVTZ geometries from Gaussian1688,89,90,91,92. The harmonic B3LYP zero-point vibrational energies (ZPVEs) are added to the electronic energies to provide more physically meaningful total energies. These total energies are defined as the F12-TZ//B3LYP/aVTZ energies. The differences in these ZPVE-corrected adiabatic energies between neutral isomers or between neutral and cation states of the same isomers produce the adiabatic relative energies and IEs, respectively. Again, the smaller species also employ only the F12-TZ energies and ZPVEs in order to benchmark that the B3LYP optimized geometries are sufficient for computing the F12-TZ relative energies and IEs for the larger molecules.

The conformers of S8 are modeled after those from ref. 93. However, the modern F12 methods used here do not find the twisted conformer as a minimum in the neutral form, only the cation and vice versa for the crown conformer. Additionally, the search for the H2Sn isomers for n > 4 involved scans of the dihedral angles. These were displaced by 30° steps in all combinations and allowed to optimize from the initial guesses as defined from the static scans. This resulted in repetitions of isomers being constructed, but these were discarded from the conformational sampling after their energies and geometries were found to be redundant with existing values in the dataset. For H2S8, for instance, several thousand conformers were produced from the nested loops of the dihedral scans for each dihedral angle present in the molecule. These initial structures were, then, optimized with B3LYP/aug-cc-pVTZ, but only 83 in H2S8 were found to be unique. At each, unique minimum, ZPVE-corrected F12-TZ//B3LYP/aVTZ energies were computed for both the neutral and the cation from the same starting geometry guess. These differences produce the total relative energy and IE values reported in Supplementary Table 3.

To calculate photoionization cross sections, for each of the H2Sn isomers (n = 1–8) and for S8, the B3LYP/aug-cc-pVTZ optimized geometry with the lowest energy was used to calculate Dyson orbitals at the EOM-IP-CCSD/aug-cc-pVTZ level of theory94,95,96,97,98. An exact method for treating the photoelectron wave function in the photoionization of molecules is not currently available, making photoionization cross-section calculations necessarily approximate. However, previous studies have indicated that, for molecules, treating the photoelectron wave function as a Coulomb wave with some partial (effective) charge Zeff between 0 and 1 gives cross sections that are in good agreement with the experimental values99. Therefore, to estimate the photoionization cross sections, we compute the cross sections of each isomer at six values of Zeff (0, 0.2, 0.4, 0.6, 0.8, 1.0) and take their average and standard deviation, with the latter used to reflect the sensitivity of the cross sections to the treatment of the photoelectron wave function and estimate the error in the calculations94,95. The cross sections and uncertainties were then used to compute each of the product yields and branching ratios. Dyson orbitals were computed with Q-Chem 6.0100, while photoionization cross sections were computed with ezDyson 5.0101. The calculated photoionization cross sections are included in the Supplementary Table 6.