Context. Gas phase Elemental abundances in molecular CloudS (GEMS) is an IRAM 30-m Large Program aimed at determining the elemental abundances of carbon (C), oxygen (O), nitrogen (N), and sulfur (S) in a selected set of prototypical star-forming filaments. In particular, the elemental abundance of S remains uncertain by several orders of magnitude, and its determination is one of the most challenging goals of this program. Aims. This paper aims to constrain the sulfur elemental abundance in Taurus, Perseus, and Orion A based on the GEMS molecular database. The selected regions are prototypes of low-mass, intermediate-mass, and high-mass star-forming regions, respectively, providing useful templates for the study of interstellar chemistry. Methods. We have carried out an extensive chemical modeling of the fractional abundances of CO, HCO+, HCN, HNC, CS, SO, H2S, OCS, and HCS+ to determine the sulfur depletion toward the 244 positions in the GEMS database. These positions sample visual extinctions from AV ∼ 3 mag to >50 mag, molecular hydrogen densities ranging from a few × 103 cm3 to 3 × 106 cm3, and Tk ∼ 10-35 K. We investigate the possible relationship between sulfur depletion and the grain charge distribution in different environments. Results. Most of the positions in Taurus and Perseus are best fitted assuming early-time chemistry, t = 0.1 Myr, ζH2 ∼ (0.51) × 1016 s1, and [S/H] ∼ 1.5 × 106. On the contrary, most of the positions in Orion are fitted with t = 1 Myr and ζH2 ∼ 1017 s1. Moreover, ∼40% of the positions in Orion are best fitted assuming the undepleted sulfur abundance, [S/H] ∼ 1.5 × 105. We find a tentative trend of sulfur depletion increasing with density. Conclusions. Our results suggest that sulfur depletion depends on the environment. While the abundances of sulfur-bearing species are consistent with undepleted sulfur in Orion, a depletion factor of ∼20 is required to explain those observed in Taurus and Perseus. We propose that differences in the grain charge distribution might explain these variations. Grains become negatively charged at a visual extinction of AV ∼ 3.5 mag in Taurus and Perseus. At this low visual extinction, the S+ abundance is high, X(S+) > 106, and the electrostatic attraction between S+ and negatively charged grains could contribute to enhance sulfur depletion. In Orion, the net charge of grains remains approximately zero until higher visual extinctions (AV ∼ 5.5 mag), where the abundance of S+ is already low because of the higher densities, thus reducing sulfur accretion. The shocks associated with past and ongoing star formation could also contribute to enhance [S/H].

@en

Context. Sulfur is a biogenic element used as a tracer of the evolution of interstellar clouds to stellar systems. However, most of the expected sulfur in molecular clouds remains undetected. Sulfur disappears from the gas phase in two steps. The first depletion occurs during the translucent phase, reducing the gas-phase sulfur by 7-40 times, while the following freeze-out step occurs in molecular clouds, reducing it by another order of magnitude. This long-standing question awaits an explanation. Aims. The aim of this study is to understand under what form the missing sulfur is hiding in molecular clouds. The possibility that sulfur is depleted onto dust grains is considered. Methods. Experimental simulations mimicking HS ice UV photoprocessing in molecular clouds were conducted at 8 K under ultra-high vacuum. The ice was subsequently warmed up to room temperature. The ice was monitored using infrared spectroscopy, and the desorbing molecules were measured by quadrupole mass spectrometry in the gas phase. Theoretical Monte Carlo simulations were performed for interpretation of the experimental results and extrapolation to the astrophysical and planetary conditions. Results. HS formation was observed during irradiation at 8 K. Molecules HS x with x > 2 were also identified and found to desorb during warm-up, along with S to S 4 species. Larger S x molecules up to S 8 are refractory at room temperature and remained on the substrate forming a residue. Monte Carlo simulations were able to reproduce the molecules desorbing during warming up, and found that residues are chains of sulfur consisting of 6-7 atoms. Conclusions. Based on the interpretation of the experimental results using our theoretical model, it is proposed that S + in translucent clouds contributes notoriously to S depletion in denser regions by forming long S chains on dust grains in a few times 10 4 yr. We suggest that the S to S 4 molecules observed in comets are not produced by fragmentation of these large chains. Instead, they probably come either from UV photoprocessing of HS-bearing ice produced in molecular clouds or from short S chains formed during the translucent cloud phase.

@en

Gas phase Elemental abundances in Molecular CloudS (GEMS) is an IRAM 30 m Large Program designed to provide estimates of the S, C, N, and O depletions and gas ionization degree, X(e-), in a selected set of star-forming filaments of Taurus, Perseus, and Orion. Our immediate goal is to build up a complete and large database of molecular abundances that can serve as an observational basis for estimating X(e-) and the C, O, N, and S depletions through chemical modeling. We observed and derived the abundances of 14 species (13CO, C18O, HCO+, H13CO+, HC18O+, HCN, H13CN, HNC, HCS+, CS, SO, 34SO, H2S, and OCS) in 244 positions, covering the AV ~3 to ~100 mag, n(H2) ~ a few 103 to 106 cm-3, and Tk ~10 to ~30 K ranges in these clouds, and avoiding protostars, HII regions, and bipolar outflows. A statistical analysis is carried out in order to identify general trends between different species and with physical parameters. Relations between molecules reveal strong linear correlations which define three different families of species: (1) 13CO and C18O isotopologs; (2) H13CO+, HC18O+, H13 CN, and HNC; and (3) the S-bearing molecules. The abundances of the CO isotopologs increase with the gas kinetic temperature until TK ~ 15 K. For higher temperatures, the abundance remains constant with a scatter of a factor of ~3. The abundances of H13 CO+, HC18 O+, H13 CN, and HNC are well correlated with each other, and all of them decrease with molecular hydrogen density, following the law ∝ n(H2)-0.8  ±  0.2. The abundances of S-bearing species also decrease with molecular hydrogen density at a rate of (S-bearing/H)gas ∝ n(H2)-0.6  ±  0.1. The abundances of molecules belonging to groups 2 and 3 do not present any clear trend with gas temperature. At scales of molecular clouds, the C18O abundance is the quantity that better correlates with the cloud mass. We discuss the utility of the 13CO/C18O, HCO+/H13CO+, and H13 CO+/H13CN abundance ratios as chemical diagnostics of star formation in external galaxies.

@en

Context. The chemical and physical evolution of starless and pre-stellar cores are of paramount importance to understanding the process of star formation. The Taurus Molecular Cloud cores TMC 1-C and TMC 1-CP share similar initial conditions and provide an excellent opportunity to understand the evolution of the pre-stellar core phase. Aims. We investigated the evolutionary stage of starless cores based on observations towards the prototypical dark cores TMC 1-C and TMC 1-CP. Methods. We mapped the prototypical dark cores TMC 1-C and TMC 1-CP in the CS 3 → 2, C34S 3 → 2, 13CS 2 → 1, DCN 1 → 0, DCN 2 → 1, DNC 1 → 0, DNC 2 → 1, DN13C 1 → 0, DN13C 2 → 1, N2H+ 1 → 0, and N2D+ 1 → 0 transitions. We performed a multi-transitional study of CS and its isotopologs, DCN, and DNC lines to characterize the physical and chemical properties of these cores. We studied their chemistry using the state-of-the-art gas-grain chemical code NAUTILUS and pseudo time-dependent models to determine their evolutionary stage. Results. The central nH volume density, the N2H+ column density, and the abundances of deuterated species are higher in TMC 1-C than in TMC 1-CP, yielding a higher N2H+ deuterium fraction in TMC 1-C, thus indicating a later evolutionary stage for TMC 1-C. The chemical modeling with pseudo time-dependent models and their radiative transfer are in agreement with this statement, allowing us to estimate a collapse timescale of ~1 Myr for TMC 1-C. Models with a younger collapse scenario or a collapse slowed down by a magnetic support are found to more closely reproduce the observations towards TMC 1-CP. Conclusions. Observational diagnostics seem to indicate that TMC 1-C is in a later evolutionary stage than TMC 1-CP, with a chemical age ~1 Myr. TMC 1-C shows signs of being an evolved core at the onset of star formation, while TMC 1-CP appears to be in an earlier evolutionary stage due to a more recent formation or, alternatively, a collapse slowed down by a magnetic support.

@en

Context. The origin of the reservoirs of water on Earth is debated. The Earth's crust may contain at least three times more water than the oceans. This crust water is found in the form of phyllosilicates, whose origin probably differs from that of the oceans. Aims. We test the possibility to form phyllosilicates in protoplanetary disks, which can be the building blocks of terrestrial planets. Methods. We developed an exploratory rate-based warm surface chemistry model where water from the gas-phase can chemisorb on dust grain surfaces and subsequently diffuse into the silicate cores. We applied the phyllosilicate formation to a zero-dimensional chemical model and to a 2D protoplanetary disk model (PRODIMO). The disk model includes in addition to the cold and warm surface chemistry continuum and line radiative transfer, photoprocesses (photodissociation, photoionisation, and photodesorption), gas-phase cold and warm chemistry including three-body reactions, and detailed thermal balance. Results. Despite the high energy barrier for water chemisorption on silicate grain surfaces and for diffusion into the core, the chemisorption sites at the surfaces can be occupied by a hydroxyl bond (-OH) at all gas and dust temperatures from 80 to 700 K for a gas density of 2 × 104 cm-3. The chemisorption sites in the silicate cores are occupied at temperatures between 250 and 700 K. At higher temperatures thermal desorption of chemisorbed water occurs. The occupation efficiency is only limited by the maximum water uptake of the silicate. The timescales for complete hydration are at most 105 yr for 1 mm radius grains at a gas density of 108 cm-3. Conclusions. Phyllosilicates can be formed on dust grains at the dust coagulation stage in protoplanetary disks within 1 Myr. It is however not clear whether the amount of phyllosilicate formed by warm surface chemistry is sufficient compared to that found in Solar System objects.

@en

Context. Molecular hydrogen (H2) is the main constituent of the gas in the planet-forming disks that surround many pre-main-sequence stars. H2 can be incorporated in the atmosphere of the nascent giant planets in disks. Deuterium hydride (HD) has been detected in a few disks and can be considered the most reliable tracer of H2, provided that its abundance throughout the disks with respect to H2 is well understood. Aims. We wish to form H2 and HD efficiently for the varied conditions encountered in protoplanetary disks: the densities vary from 104 to 1016 cm-3; the dust temperatures range from 5 to 1500 K, the gas temperatures go from 5 to a few 1000 Kelvin, and the ultraviolet radiation field can be 107 stronger than the standard interstellar field. Methods. We implemented a comprehensive model of H2 and HD formation on cold and warm grain surfaces and via hydrogenated polycyclic aromatic hydrocarbons in the physico-chemical code PROtoplanetary DIsk MOdel. The H2 and HD formation on dust grains can proceed via the Langmuir-Hinshelwood and Eley-Ridel mechanisms for physisorbed or chemisorbed H (D) atoms. H2 and HD also form by H (D) abstraction from hydrogenated neutral and ionised PAHs and via gas phase reactions. Results. H2 and HD are formed efficiently on dust grain surfaces from 10 to ∼700 K. All the deuterium is converted into HD in UV shielded regions as soon as H2 is formed by gas-phase D abstraction reactions. The detailed model compares well with standard analytical prescriptions for H2 (HD) formation. At low temperature, H2 is formed from the encounter of two physisorbed atoms. HD molecules form on the grain surfaces and in the gas-phase. At temperatures greater than 20 K, the encounter between a weakly bound H-(or D-) atom or a gas-phase H (D) atom and a chemisorbed atom is the most efficient H2 formation route. H2 formation through hydrogenated PAHs alone is efficient above 80 K. However, the contribution of hydrogenated PAHs to the overall H2 and HD formation is relatively low if chemisorption on silicate is taken into account and if a small hydrogen abstraction cross-section is used. The H2 and HD warm grain surface network is a first step in the construction of a network of high-temperature surface reactions.

@en

When studying chemistry of photodissociation regions (PDRs), time dependence becomes important as visual extinction increases, since certain chemical time-scales are comparable to the cloud lifetime. Dust temperature is also a key factor, since it significantly influences gas temperature and mobility on dust grains, determining the chemistry occurring on grain surfaces. We present a study of the dust temperature impact and time effects on the chemistry of different PDRs, using an updated version of theMeijerink PDR code and combining it with the time-dependent code Nahoon. We find the largest temperature effects in the inner regions of highG0 PDRs,where high dust temperatures favour the formation of simple oxygen-bearing molecules (especially that of O₂), while the formation of complex organic molecules is much more efficient at low dust temperatures. We also find that time-dependent effects strongly depend on the PDR type, since long time-scales promote the destruction of oxygen-bearing molecules in the inner parts of low G0 PDRs, while favouring their formation and that of carbon-bearing molecules in high G0 PDRs. From the chemical evolution, we also conclude that, in dense PDRs, CO₂ is a late-forming ice compared to water ice, and confirm a layered ice structure on dust grains, with H₂O in lower layers than CO₂. Regarding steady state, the PDR edge reaches chemical equilibrium at early times (≲10⁵ yr). This time is even shorter (<10⁴ yr) for high G0 PDRs. By contrast, inner regions reach equilibrium much later, especially low G0 PDRs, where steady state is reached at ∼10⁶-10⁷ yr.

@en

GEMS is an IRAM 30 m Large Program whose aim is determining the elemental depletions and the ionization fraction in a set of prototypical star-forming regions. This paper presents the first results from the prototypical dark cloud Taurus molecular cloud (TMC) 1. Extensive millimeter observations have been carried out with the IRAM 30 m telescope (3 and 2 mm) and the 40 m Yebes telescope (1.3 cm and 7 mm) to determine the fractional abundances of CO, HCO ⁺ , HCN, CS, SO, HCS ⁺ , and N ₂ H ⁺ in three cuts which intersect the dense filament at the well-known positions TMC 1-CP, TMC 1-NH3, and TMC 1-C, covering a visual extinction range from A _V ~ 3 to ~20 mag. Two phases with differentiated chemistry can be distinguished: (i) the translucent envelope with molecular hydrogen densities of 1-5 × 10 ³ cm ^-3 ; and (ii) the dense phase, located at A _V > 10 mag, with molecular hydrogen densities >10 ⁴ cm ^-3 . Observations and modeling show that the gas phase abundances of C and O progressively decrease along the C ⁺ /C/CO transition zone (A _V ~ 3 mag) where C/H ~ 8 × 10 ^-5 and C/O ~ 0.8-1, until the beginning of the dense phase at A _V ~ 10 mag. This is consistent with the grain temperatures being below the CO evaporation temperature in this region. In the case of sulfur, a strong depletion should occur before the translucent phase where we estimate an S/ H ~ (0.4-2.2) × 10 ^-6 , an abundance ~7-40 times lower than the solar value. A second strong depletion must be present during the formation of the thick icy mantles to achieve the values of S/H measured in the dense cold cores (S H ~ 8 × 10 ^-8 ). Based on our chemical modeling, we constrain the value of ζ _{H 2} to ~(0.5-1.8) × 10 ^-16 s ^-1 in the translucent cloud.

@en

Context. Spectroscopic studies of ices in nearby star-forming regions indicate that ice mantles form on dust grains in two distinct steps, starting with polar ice formation (H2O rich) and switching to apolar ice (CO rich). Aims. We test how well the picture applies to more diffuse and quiescent clouds where the formation of the first layers of ice mantles can be witnessed.
Methods. Medium-resolution near-infrared spectra are obtained toward background field stars behind the Pipe Nebula. Results. The water ice absorption is positively detected at 3.0 µm in seven lines of sight out of 21 sources for which observed spectra are successfully reduced. The peak optical depth of the water ice is significantly lower than those in Taurus with the same AV . The
source with the highest water-ice optical depth shows CO ice absorption at 4.7 µm as well. The fractional abundance of CO ice with respect to water ice is 16+7
−6 %, and about half as much as the values typically seen in low-mass star-forming regions.
Conclusions. A small fractional abundance of CO ice is consistent with some of the existing simulations. Observations of CO2 ice in the early diffuse phase of a cloud play a decisive role in understanding the switching mechanism between polar and apolar ice formation.@en

The temperature of interstellar dust particles is of great importance to astronomers. It plays a crucial role in the thermodynamics of interstellar clouds, because of the gas-dust collisional coupling. It is also a key parameter in astrochemical studies that governs the rate at which molecules form on dust. In 3D (magneto)hydrodynamic simulations often a simple expression for the dust temperature is adopted, because of computational constraints, while astrochemical modelers tend to keep the dust temperature constant over a large range of parameter space. Our aim is to provide an easy-to-use parametric expression for the dust temperature as a function of visual extinction (A_V) and to shed light on the critical dependencies of the dust temperature on the grain composition. We obtain an expression for the dust temperature by semi-analytically solving the dust thermal balance for different types of grains and compare to a collection of recent observational measurements. We also explore the effect of ices on the dust temperature. Our results show that a mixed carbonaceous-silicate type dust with a high carbon volume fraction matches the observations best. We find that ice formation allows the dust to be warmer by up to 15% at high optical depths (A_V> 20 mag) in the interstellar medium. Our parametric expression for the dust temperature is presented as T_d = [11 + 5.7 × tanh(0.61 - log ₁₀(A_V)]χ_uv ^1/5.9, where χ_uv is in units of the Draine (1978, ApJS, 36, 595) UV field.

@en

In the earliest phases of star-forming clouds, stable molecular species, such as CO, are important coolants in the gas phase. Depletion of these molecules on dust surfaces affects the thermal balance of molecular clouds and with that their whole evolution. For the first time, we study the effect of grain surface chemistry (GSC) on star formation and its impact on the initial mass function (IMF). We follow a contracting translucent cloud in which we treat the gas-grain chemical interplay in detail, including the process of freeze-out. We perform 3D hydrodynamical simulations under three different conditions, a pure gas-phase model, a freeze-out model, and a complete chemistry model. The models display different thermal evolution during cloud collapse as also indicated in Hocuk, Cazaux & Spaans, but to a lesser degree because of a different dust temperature treatment, which is more accurate for cloud cores. The equation of state (EOS) of the gas becomes softer with CO freeze-out and the results show that at the onset of star formation, the cloud retains its evolution history such that the number of formed stars differ (by 7 per cent) between the three models. While the stellar mass distribution results in a different IMF when we consider pure freeze-out, with the complete treatment of the GSC, the divergence from a pure gas-phase model is minimal. We find that the impact of freeze-out is balanced by the non-thermal processes; chemical and photodesorption. We also find an average filament width of 0.12 pc (±0.03 pc), and speculate that this may be a result from the changes in the EOS caused by the gas-dust thermal coupling. We conclude that GSC plays a big role in the chemical composition of molecular clouds and that surface processes are needed to accurately interpret observations, however, that GSC does not have a significant impact as far as star formation and the IMF is concerned.

@en

Context. The presence of dust can strongly affect the chemical composition of the interstellar medium. We model the chemistry in photodissociation regions (PDRs) using both gas-phase and dust-phase chemical reactions. Aims. Our aim is to determine the chemical compositions of the interstellar medium (gas/dust/ice) in regions with distinct (molecular) gas densities that are exposed to radiation fields with different intensities. Methods. We have significantly improved the Meijerink PDR code by including 3050 new gas-phase chemical reactions and also by implementing surface chemistry. In particular, we have included 117 chemical reactions occurring on grain surfaces covering different processes, such as adsorption, thermal desorption, chemical desorption, two-body reactions, photo processes, and cosmic-ray processes on dust grains. Results. We obtain abundances for different gas and solid species as a function of visual extinction, depending on the density and radiation field. We also analyse the rates of the formation of CO₂ and H₂O ices in different environments. In addition, we study how chemistry is affected by the presence/absence of ice mantles (bare dust or icy dust) and the impact of considering different desorption probabilities. Conclusions. The type of substrate (bare dust or icy dust) and the probability of desorption can significantly alter the chemistry occurring on grain surfaces, leading to differences of several orders of magnitude in the abundances of gas-phase species, such as CO, H₂CO, and CH₃OH. The type of substrate, together with the density and intensity of the radiation field, also determine the threshold extinction to form ices of CO₂ and H₂O. We also conclude that H₂CO and CH₃OH are mainly released into the gas phase of low, far-ultraviolet illuminated PDRs through chemical desorption upon two-body surface reactions, rather than through photodesorption.

@en

Observations of star-forming environments revealed that the abundances of some deuterated interstellar molecules are markedly larger than the cosmic D/H ratio of 10^-5. Possible reasons for this pointed to grain surface chemistry. However, organic molecules and water, which are both ice constituents, do not enjoy the same deuteration. For example, deuterated formaldehyde is very abundant in comets and star-forming regions, while deuterated water rarely is. In this paper, we explain this selective deuteration by following the formation of ices (using the rate equation method) in translucent clouds, as well as their evolution as the cloud collapses to form a star. Ices start with the deposition of gas-phase CO and O onto dust grains. While reaction of oxygen with atoms (H or D) or molecules (H₂) yields H₂O (HDO), CO only reacts with atoms (H and D) to form H ₂CO (HDCO, D₂CO). As a result, the deuteration of formaldehyde is sensitive to the gas D/H ratio as the cloud undergoes gravitational collapse, while the deuteration of water strongly depends on the dust temperature at the time of ice formation. These results reproduce well the deuterium fractionation of formaldehyde observed in comets and star-forming regions and can explain the wide spread of deuterium fractionation of water observed in these environments.

@en

Context. Water and O₂ are important gas phase ingredients for cooling dense gas when forming stars. On dust grains, H₂O is an important constituent of the icy mantle in which a complex chemistry is taking place, as revealed by hot core observations. The formation of water can occur on dust grain surfaces, and can impact gas phase composition. Aims. The formation of molecules such as OH, H₂O, HO₂ and H₂O ₂, as well as their deuterated forms and O₂ and O ₃ is studied to assess how the chemistry varies in different astrophysical environments, and how the gas phase is affected by grain surface chemistry. Methods. We use Monte Carlo simulations to follow the formation of molecules on bare grains as well as the fraction of molecules released into the gas phase. We consider a surface reaction network, based on gas phase reactions, as well as UV photo-dissociation of the chemical species. Results. We show that grain surface chemistry has a strong impact on gas phase chemistry, and that this chemistry is very different for different dust grain temperatures. Low temperatures favor hydrogenation, while higher temperatures favor oxygenation. Also, UV photons dissociate the molecules on the surface, which can subsequently reform. The formation-destruction cycle increases the amount of species released into the gas phase. We also determine the timescales to form ices in diffuse and dense clouds, and show that ices are formed only in shielded environments, as supported by observations.

@en

Aims. The production of molecular hydrogen and its deuterated forms onto carbonaceous dust grains is investigated in detail. The goal of this study is to estimate the importance of the chemistry occuring on grain surfaces for the deuteration of H. Furthermore, we aim to find a robust and general surface chemical model that can be used in different astrophysical environments.Methods. Surface processes are described for the cases of graphitic and amorphous-carbon grains, where laboratory work is available. Langmuir-Hinshelwood, as well as Eley-Rideal surface chemistries are included in the model and their relative contributions highlighted. Analytic expressions are derived for H, HD, and D formation efficiencies for both types of grains. Rate equations are tested against stochastic methods.Results. As expected, rate equations and stochastic methods diverge for grain sizes below a critical value . For these sizes, D formation decreases to favour HD formation. The formation efficiencies of H 2 and D2 can be calculated by adding a correction factor to the rate equations methods (this factor is a simple exponential factor that becomes unity when a ≥ acrit. We find that, because of the presence of chemisorbed sites, which can store atoms to form molecules up to high grain temperatures, the formation efficiency of HD and D is very high compared to models where only physisorption sites are taken into account. When considering a realistic distribution of dust grains, we find that the formation rates of and HD are enhanced by an order of magnitude if small grains are taken into account. The formation of D, on the other hand, comes from the contribution of small (≤ 100 Å) and big (100 ≥Å) grains, depending on the D/H ratio, the grain temperature, and the volume density. The processes described in this paper, which allow a strong enhancement of the deuterated forms of molecular hydrogen, could explain the high degree of deuterium fractionation observed in protostellar environments.@en

Molecular hydrogen is the most abundant molecule in the Universe and dominates the mass budget of the gas, particularly in regions of star formation. H2 is also an important chemical intermediate in the formation of larger species and can be an important gas coolant when the medium lacks metals. Because of the inefficiency of gas-phase reactions to form H2, this molecule is generally thought to form on grain surfaces. Observations of H 2 in a wide variety of objects showed that this molecule could form efficiently over a wide range of physical conditions. To understand the mechanism responsible for such an efficient formation, we developed a model for molecular hydrogen formation on grain surfaces. This model considers the interaction between atom and surface as beeing either weak (Van der Waals interaction - physisorption) or strong (covalent bound - chemisorption), as well as the mobility of the atom on a surface due to quantum mechanical diffusion and thermal hopping. This model solves the time-dependent kinetic rate equation for the formation of molecular hydrogen and its deuterated forms. Our results have been benchmarked with laboratory experiments on silicates, carbonaceous and graphitic surfaces. This comparison allowed us to derive some characteristics of the considered surfaces. An extension of our model to astrophysical conditions gives an estimate of H2 formation efficiency for a wide range of physical conditions. One of our main results is the efficient formation of molecular hydrogen for gas and grain temperatures up to several hundreds of kelvins. We also compared our predictions to observations in astrophysical objects such as photodissociation regions (PDRs). The addition of deuterium in our model for the formation of HD and D2 molecules is also discussed.@en