arXiv:1501.04701v1 [astro-ph.IM] 20 Jan 2015 r eesr rcroso uhpeitcmtra and material prebiotic (NH such hydroxylamine them of among molecules precursors N-bearing necessary material. like are con- prebiotic compounds be and Prebiotic organic acids can complex amino of that 2000). precursors astrochem- observed as between already al. sidered et been link (Wincel have a in molecules astrobiology goals find main and to the istry is of one today complexity, astronomy in growing and ber hi o rdce bnacs erhsfrhydrox- for Searches abundances. predicted low their i,IMadCM fbt yrxlmn n glycine and hydroxylamine me- both circum-stellar of or CSM) (NH inter- and the ISM in dia, (either detection ical gas-phase through than muchreactions. surfaces formed grain on be bulk-ice efficiently can more and glycine that grain-surface model work shows chemical which gas-phase, recent chemistry, three-phase by couples confirmed new fully is a that This using 2013) phase. acetic gas (Garrod with recent the from ions in produced a hydroxylamine-derived acid efficiently of in be reactions gas that Actually, could possibility the the glycine the dismiss of in (2012) reactions. precursors al. under happens surface et still Barrientos synthesis by work is a or chi- question such phase and the whether simplest However, debate second yield amino-acid). (the to ral alanine in acids amino-acid) protonated react propanoic non-chiral can simplest and and (the it acetic glycine form with protonated 2007) protonated theo- phase al. its gas and et in Snow the experimental 2003; that Both al. shown et have (Blagojevic candidates. studies retical best the of [email protected] rpittpstuigL using 2018 typeset 21, Preprint March version Draft ihmlclsdtce nsaeicesn nnum- in increasing space in detected molecules With urnl,temi rbe stelc fastronom- of lack the is problem main the Currently, 2 CH mn cdgyie lhuhntytdtce,NH detected, yet not Although glycine. acid amino mrhu iiaesraevateoiaino moi.Teexperim forma The NH the ammonia. of of of formation investigation oxidation the the of experimental simulation via an surface on silicate report amorphous We ALMA. with h us odtc rboi oeue nsae oal mn acids (NH amino Hydroxylamine notably atoms. space, in nitrogen involving molecules prebiotic chemistry detect the to quest The ciaineeg are,NH barrier, energy activation atr1 ihrsett NH to respect with 10 factor Keywords: 2 OH,adti sovosycreae to correlated obviously is this and COOH), OMTO FHDOYAIEO UTGAN I MOI OXIDA AMMONIA VIA GRAINS DUST ON HYDROXYLAMINE OF FORMATION 1. S:mlcls—IM tm S:audne S:ut extinct ISM:dust, — abundances ISM: — atoms ISM: — molecules ISM: aaadPoess astrochemistry Processes: and Data A etrfrRdohsc n pc eerh onl Univer Cornell Research, Space and Radiophysics for Center INTRODUCTION T E tl mltajv 12/16/11 v. emulateapj style X hsc eatet yaueUiest,Srcs,N 132 NY Syracuse, University, Syracuse Department, Physics 2 H per ob one be to appears OH) 3 2 ugsin fcniin o uueosrain r provided are observations future for conditions of Suggestions . Hi on ob omdwt naudneta ee al eo a below falls never that abundance an with formed be to found is OH 2 Hi es lu odtos nie t1 n ihamodest a with and K 14 at ices On conditions. cloud dense in OH ioH,Gafac Vidali Gianfranco He, Jiao ai bevtr,Prs France Paris, Observatory, Paris rf eso ac 1 2018 21, March version Draft enLusLemaire Jean-Louis oi .Garrod T. Robin ABSTRACT uhdnergos omcry(R oie helium, ionized (CR) cosmic-ray within Deep regions, He away. farther dense colder much such is material This ( warm o.Te r ein uruddb es aeil(10 material cori- dense by and objects surrounded cm cores regions hot are stellar They called young also nos. to protostars, high-mass Such claimed low-mass luminous and 2013). is (Garrod are nearby CSM lines bright sources emission towards or narrow B2(N). ALMA with Sgr ISM using sources in plausible the highly 2008) in be al. et detection (Belloche Glycine detected been has ae eeto ntelwms r-tla oeL1544 core pre-stellar low-mass 2014). the al. (Jim´enez-Serra et in regions glycine detection better of Water be detection may earli- cores, the their pre-stellar for in cold of regions, stages star-forming est low-mass emission glycine that re- of many show calculations More where transfer places observed. radiative general cently, are in molecules are organic corinos terrestrial-like and cores Hot tm htitrc ihaudn tm oyield to atoms H abundant with interact NH that atoms N ebe hto neselrie ihhg bnacsof H re- abundances hot-cores as high of with species ices composition hydrogenated interstellar the of that that noting sembles worth is It rcro fgyie mn ctntie(NH acetonitrile direct amino possible a un- glycine, is is of mate- that it species interstellar precursor but related chemically pristine 2009), A al. represents rial. et it Elsila whether sam- 2008; known the STARDUST-returned al. cometary et in using (Glavin of detected ples Glycine L1157-B1 been Peak, has toward Array. Kitt and origin Telescope on (2013) W3IRS5, Radio telescope Charness CARMA B2(OH), 12m by Sgr NRAO ob- and B2(N), the (2012) Sgr using Orion al. W51M S, et IRC+10216, Pulliam toward Orion sources by KL, bright done seven serving been have ylamine 2 Hi osdrdalkl agto detection of target likely a considered is OH + − 3 rasu h eysal N stable very the up breaks , 3 hc hnsik ocl ris(iln 2005). (Tielens grains cold to sticks then which , o o oe n 10 and cores hot for ∼ 5 n 0 ,rsetvl)coet h core. the to close respectively) K, 100 and K 250 2 H scniee rcro othe to precursor a considered is OH) iy taa Y183 USA 14853, NY Ithaca, sity, eursa nesadn of understanding an requires , na aaaete e noa into fed then are data ental ino yrxlmn nan on hydroxylamine of tion 4 USA 44, 2 6 n NH and O cm − o Physical — ion 3 2 o o oio)and corinos) hot for oeue liberating molecule, TION 3 . Tees2005). (Tielens 2 CH 2 CN), 8 2

(Caselli et al. 2012) indicates that a fraction of the grain age. Such a process occurs without external energetic mantles has desorbed into the gas phase. This makes the input but implies several successive steps of hydrogena- observation of glycine plausible, assuming it co-desorbs tion. It is worth mentioning a crossed molecular beam 1 with water. After the first detection of nitrogen hydrides study of the O( D)+NH3 reaction at 295 K (Shu et al. in Sgr B2 (Goicoechea et al. 2004), a complex region en- 2001) forming vibrationally excited NH2OH with a sub- compassing a variety of physical conditions and in par- sequent dissociation into two different reaction channels: ticular shocks, the new detection with Herschel/HIFI of OH+NH2 and NHOH/NH2O+H with 90% and 10% yield NH (and ND), NH2, and NH3 towards the external enve- respectively. This reaction has been also studied theo- lope of the protostar IRAS 16293-2422 (Hily-Blant et al. retically (Wang et al. 2004). Finally, very recently, hy- 2010; Bacmann et al. 2010) has not improved our knowl- droxylamine have been formed in a two-step mechanism edge of the interstellar chemistry of nitrogen hydrides. In through the reaction of ammonia with hydroxyl radicals a recent publication, Le Gal et al. (2014) develop a new in a Neon matrix at 4 K (Zins & Krim 2014). model that could explain the formation of these hydrides However, most of the objects mentioned above, if not in the gas phase while some of the modelers mentioned all, either in high-mass or low-mass star-forming regions above claim surface formation. Concerning oxygen, one as well as in proto-stellar cores, share the common char- of the conclusions of deep observations of O2 towards acteristic of having a large abundance of NH3 (and also NGC 1333 IRAS 4A protostar (Yildiz et al. 2013) is that H2O), in any case far higher than the NO abundance. the observed low molecular oxygen abundance is due to This latter molecule appears to be present in a restricted the freeze-out of atomic O onto grains. All these obser- number of objects relating to high-mass star-forming re- vations lead to the conclusion that on the grains in dense gions. The abundance of NH3 is asserted in observations clouds surrounding protostellar cores, there can be NH3 of low-mass young stellar objects (Bottinelli et al. 2010) and O present. and even dark clouds (as the Bok globule Barnard 68 On the experimental side, two important facts are of (Di Francesco et al. 2002)) and comets (as very recently interest. First, concerning NH3, it has been experimen- detected on 67P/C-G by the ROSINA instrument aboard tally demonstrated (Hidaka et al. 2011) that the succes- ROSETTA (Altwegg 2014)). Then the formation pro- sive hydrogenation of N atoms trapped in a solid N2 cess in star-forming regions of hydroxylamine we present matrix at 10 K leads to NH3 formation. Second, the in this paper is a very likely more direct (single step) formation of NH2OH has been observed after irradiat- and realistic (considering the species involved) mecha- ing an ammonia-water ice at 90-130 K with UV photons nism, compared to the one already known (Congiu et al. (Nishi et al. 1984), and at 10 and 50 K with 5 keV elec- 2012b). trons (Zheng & Kaiser 2010). The chemical processing Actually this process is well suited with respect to to of H2O-NH3 ices induced by heavy ions (Bordalo et al. hot-cores and protostellar cores. In addition to N result- 2013) leads also, by radiolysis, to the formation of ing from N2 dissociation under cosmic-rays irradiation, N2O, NO, NO2, and NH2OH. As mentioned above, atomic O is also present for the same reason. Both N and (Blagojevic et al. 2003) showed that the reaction of hy- O stick on the colder bare or ices covered grains, the ones + droxylamine ions, NH2OH , with acetic or propanoic deep into the cloud surrounding the core (Av ∼ 15 - 20). acid makes ionized glycine and alanine. Much earlier, A very simplistic scheme could then be the following: (i) from about 1960 to mid 1980s, the NH3+O reaction has Very cold grains are covered by atoms (H, O, and N), ei- been the subject of numerous experiments reviewed by ther on the surface or in bulk, (ii) when the temperature (Cohen 1987). They have been performed at moderate begins to increase (>5-10 K), H atoms become mobile temperature (450 to 850 K) in flowing and static sys- and form NH3 (and H2O) by hydrogenation surface re- tems (Perry 1984; Baulch et al. 1984) where ground state actions , (iii) as the temperature continues to increase O(3P) is obtained by laser photolysis, or at higher tem- (>14 K in interstellar space conditions - corresponding peratures (850 to 2200 K) in flames (Fenimore & Jones in laboratory experiments to >20-25 K)), O atoms be- 1961) or in shock tube experiments (Fujii et al. 1986). come mobile to finally react with NH3 (and H2O) already Even if such a reaction has never been studied at very low formed to to produce in particular NH2OH through an temperature, it has been proposed that excited NH3O in- insertion mechanism. termediate could rearrange to give stable hydroxylamine In this paper we present experimental evidence of the NH2OH (Baulch et al. 1984). The formation of complex formation process of hydroxylamine via oxidation of NH3 organic molecules involving atomic addition reactions on on dust grain analogs (Section 3). The experimental the cold surface of ices-covered interstellar grains in di- findings are then used in a 3-phase gas-grain model of verse environments, has been extensively reviewed re- a dark cloud (Section 4). A comparison of NH2OH for- cently (Herbst & van Dishoeck 2009). As NO has been mation mechanisms and a discussion about suitable envi- detected in the gas phase towards a few high-mass star- ronments where NH2OH could be observed are presented forming regions, it has been suggested that hydroxy- in Sections 5 and 6. lamine could be formed in the gas phase through non- energetic hydrogenation of NO ice under dark cloud con- 2. EXPERIMENTAL SETUP ditions (Charnley et al. 2001). But recently, an NO ice 2.1. Apparatus hydrogenation surface reaction at low temperature has shown that an efficient route to NH2OH formation was The apparatus was described elsewhere (He et al. 2011; also possible with interstellar relevant ices (Congiu et al. Jing et al. 2013); here we summarize the main features 2012a,b; Fedoseev et al. 2012). It takes place at 10 to that are important for this study. The experiments 15 K at both NO submonolayer and multilayer cover- were carried out in an ultra-high vacuum main cham- ber connected to an atomic/molecular beam line. The Formation of hydroxylamine 3 main chamber is pumped by a combination of a cryop- ump, turbomolecular pump, and ion pump and can reach 0.1 1.2 × 10−10 torr after a bake-out. At the center of the chamber there is a 1 µm thick amorphous silicate thin 0.05 2 film sample grown on a 1 cm gold plated copper disk by 5 10 Relative weight electron beam physical vapor deposition. The detailed 0 preparation and characterization of the sample can be 2800 3000 3200 3400 found in Jing et al. (2013). The sample can be cooled Desorption energy (K) to 8 K by a liquid helium-cooled sample holder and be

heated up to 450 K by a cartridge heater behind the sam- Desorption rate (counts/K) ple. At the beginning of each day, the sample is heated 4 up to 400 K to clean it before cooling it down. The cham- 10 ber pressure after cooling down is about 5 × 10−11 torr. During exposure of the sample to the atomic/molecular 80 100 120 140 160 180 − beam, the pressure increases to about 1 × 10 10 torr. Temperature (K) At this pressure, the amount of water that stick on the Figure 1. TPDs with different exposure of NH3 at 70 K. The sample from the chamber background gas is negligible. exposure times are, from bottom to top, 0.5, 1, 2, 4, and 8 min- A triple-pass quadrupole mass spectrometer (QMS) is utes, respectively. Desorption energy distribution of NH3 calcu- lated from the 2 minutes exposure curve (1 ML deposition) is in mounted on a rotary platform to record desorbed species the inset. from the sample surface or to measure the beam com- position. The QMS is placed in a differentially pumped Bolina et al. (2005) for NH3 desorption from a graphite enclosure and is fitted with a cone aimed at a sample. surface. With the current experimental settings, 1 ML The distance between the tip of the cone and the sample NH3 coverage is achieved by 2 minutes exposure with can be changed by repositioning the sample via a XYZ 0.3 sccm gas flow. Thus the NH3 flow is 0.5 ML/minute, sample holder manipulator. This arrangement prevents or equivalently, 5 × 1014 cm−2minute−1, assuming 1 ML products loosely adsorbed on sample holder parts from ∼ 1015 cm−2. Following the procedures as described in reaching the QMS, thus reducing the background level He & Vidali (2014), the desorption energy distribution and improving the signal/noise. A radio-frequency dis- of NH3 from amorphous silicate surface can be obtained sociation source is mounted in the first stage of the triple by direct inversion of the 1 ML trace in Figure 1. The differentially pumped beam line. A mass flow controller resulted distribution is shown in the inset of Figure 1. (Alicat MCS-5SCCM) is used to ensure stable gas flow. Additional TPD runs are performed at a surface tem- perature of 10 K, 30 K, and 50 K. They show almost 2.2. Beam flux calibration identical TPD curves as the one at 70 K. This suggests The beam fluxes are calibrated by TPD experiments. that the sticking of ammonia on the silicate surface at 70 K is unity, and the desorption rate at 70 K is negligible. With the the amorphous silicate sample kept at 70 K, ammonia gas is introduced from the molecular beam line. In the atomic oxygen exposure, the O2 gas flow pass- The gas flow is set to 0.3 sccm on the mass flow con- ing through the flow controller is set to 0.1 sccm. At this flow the dissociation rate of O is measured to be troller. After ammonia exposure, the sample is cooled 2 down to below 40 K and then heated up at 1 K/s to 42%. The calibration of oxygen flux needs to be done about 200 K to do a TPD. In both ammonia exposure differently from ammonia because O2 is volatile and the sticking is not necessarily unity. The direct beam stage and TPD stage, the gas phase molecules are mea- sured using the QMS. Since ammonia fragments in the intensities of O2 gas at both 0.3 sccm and 0.1 sccm QMS ionizer, both mass 17 amu (NH+) and 16 amu are measured. Since it is known that 0.3 sccm corre- 3 sponds to 0.5 ML/minute, the flow at 0.1 sccm can be (NH+) are measured. The fragmentation of ammonia 2 calculated correspondingly. The O2 flow at 0.1 sccm + + − into NH /N (mass 15/14 amu) is found to be negli- is about 0.34 ML/minute (3.4 × 1014 cm−2minute 1). gible by direct beam measurements. A series of TPD When the RF is on, the beam intensities of O and O2 − experimental runs with different ammonia doses are per- are 0.29 ML/minute (2.9 × 1014 cm−2minute 1) and formed. The TPD spectra are shown in Figure 1. We − 0.20 ML/minute (2.0×1014 cm−2minute 1), respectively. calibrate the NH3 coverage by analyzing the shape of the TPD profile. In the submonolayer region the TPD With the radio frequency (RF) power on and the oxygen peak temperature should shift to lower values, because sent into the dissociation source, beam contamination is in the low coverage region molecules occupy the deep checked. The main contaminant is NO (mass 30 amu) sites (higher desorption energies). Here a layer is actu- due the small leak of air into the dissociation source. ally an equivalent layer, since it is unclear whether am- The NO (mass 30) signal is less than 3% of the O sig- monia would form clusters or islands on the surface; little nal, which is trivial in the context of the experiments experimental evidence support the existence of clusters performed here. or islands. As the deep sites get filled up with increasing 2.3. exposure, the TPD profile and peak temperature shift Experimental procedures to lower temperatures (He & Vidali 2014). In the mul- The ammonia oxidation reactions were studied in am- tilayer region, the TPD spectra should show a common monia and atomic oxygen sequential exposure TPD ex- leading edge (Kolasinki 2008), which is a typical first or- periments. The surface was covered with ammonia be- der desorption behavior. The trend of TPD traces shown fore introducing atomic oxygen. Two ammonia coverages in Figure 1 is in agreement with the one obtained by were used, 2 ML and 1/4 ML, representative of multilayer 4

4 5 x 10 x 10 3 18 amu 0 min 16 17 amu 300 0.5 min 16 amu 2.5 1 min 14 32 amu x 30 + 20000 30 amu x 30 250 2 min 12 33 amu x 30 4 min temperature 2 6 min 10 200 8 min 16 min 8 150 1.5

6 Temperature (K) QMS signal (count/s) 100 4 1

2

50 mass 17 QMS signal (counts/K) 0.5 0 5 10 15 20 Time (min)

Figure 2. A typical TPD of a NH3 + O sequential exposure 80 90 100 110 120 130 140 150 160 experiment at 70 K in which the QMS records simultaneously mul- Temperature (K) tiple signals. The temperature ramp (black line) shows that after Figure 3. the exposure is terminated, the sample is cooled to 40 K before the Mass 17 amu QMS signal of the TPD from an ice heating is started. consisting of NH3 (4 min or 2 ML) followed by exposure to various doses of O at 70 K. coverages and submonolayer coverages. After prepara- neath. After about 6 minutes of O exposure, the first tion of the ammonia sample, the residual ammonia in peak almost disappears, indicating the top layer of NH3 the beam line and dissociation source was cleaned by is almost all converted to NH2OH or other products. flushing the beam line several times using oxygen. This In Figure 4 two desorption peaks, peak A and peak B, ensures that almost no ammonia was mixed with oxy- are visible. Peak A shows up at O exposure as low as gen. The sample temperature in the exposure stage was 0.5 minutes. This indicates that the ammonia oxidation chosen to be 70 K so that O/O2/O3 does not stick onto reaction is efficient. As the O exposure increases from 0 the surface while the sticking of NH3 is unity. After the to 4 minutes, the area of peak A increases at first, but exposure stage, the surface was cooled down to below 40 then it decreases from 4 minutes to 16 minutes of O ex- K and then heated up linearly at 1 K/s to above 320 K posure. Peak A is followed, at larger O exposures, by to desorb species from the surface. The QMS recorded peak B at ∼ 260 K. Notice that peak B starts to ap- simultaneously the signals of various masses during both pear at 4 minutes of O exposure time at the expense of exposure and TPD stages. Figure 2 shows the TPD spec- peak A. As mentioned above, peak A falls in the temper- tra of a typical experimental run. ature range of the TPD mass 33 amu peak obtained by Congiu et al. (2012a). Therefore we attribute peak A to 3. RESULTS AND ANALYSIS the desorption of NH2OH. Concerning peak B, however, 3.1. Multilayer NH3 + O in a subsequent paper on an NO2 hydrogenation experi- In the NH + O sequential exposure TPD experiments, ment (Ioppolo et al. 2014), a mass 33 peak, confirmed by 3 infrared measurement spectra, appears at a temperature mass 16, 17, 18, 30, 32, and 33 amu were measured by ∼ the QMS, see Figure 2. Mass 16 amu is due to the frag- T 245 K, close to the temperature of our second peak (peak B in our Figure 4). Nevertheless, oddly there is no mentation of NH3 in the QMS ionizer; mass 17 amu is due to NH and fragmentation of H O; mass 32 amu is mention in their paper of a peak A at 160 K. 3 2 To find out what peak B represents, we show in Figure due to O2 and fragments of O3. Mass 30 amu could be due to NO from the beam line (but the amount is small), 5 a comparison of TPD traces for various masses for 2 or fragments of NH OH; mass 33 amu could be due to ML of NH3 + 8 or 16 minutes of O exposure, see left and 2 right panels of Figure 5, respectively. Peak A and peak NH2OH. HO2 has the same mass but it is unlikely to be there because of the lack of detection of the prod- B are marked by vertical lines. In both panels of Figure ucts it fragments into. The mass 33 amu peak centered 5, peak A (see trace of mass 33 amu) is accompanied by at around 180 K is accompanied by mass 30 amu and desorption of mass 30 amu, 17 amu, and 16 amu, while very small signals with masses 16, 17 and 32 amu at the this is not true for peak B. This suggests that peak B is same temperature; this suggests that the mass 33 amu not due to NH2OH. It could be the product of a frag- mentation of a dimer (Del Bene 1972) or of an oxidation peak at 180 K is due to NH2OH. This agrees with the TPD peak attribution in Congiu et al. (2012a) (in the product of NH2OH. At about 280 K, there is a peak C range of 160-200 K, peaking at T∼ 190 K) obtained in showing up for mass 30 amu, 17 amu, and 16 amu, but an NO hydrogenation experiment and confirmed by in- only when the O dose is high (16 minutes of O expo- frared measurements (RAIRS). Mass 18 amu is due to sure), see right panel of Figure 5. It could be due to a water. yet unidentified product formed in a further oxidation of NH OH. Peak B and peak C differs in both position and Figure 3 shows the mass 17 amu (NH3) signal after 2 depositing various doses of O on top of 2 ML (4 min- shape, therefore they should be attributed to different species. utes exposure) NH3. As the O exposure time increases, the NH3 peak decreases and the peak position also shifts 3.2. to higher temperatures. This is because the top layer of Submonolayer NH3 + O NH3 is gradually converted to NH2OH or other products, Experiments were also carried out at submonolayer thus hindering the desorption of NH3 molecules under- NH3 coverage. We exposed the silicate sample to NH3 Formation of hydroxylamine 5

4000 NH3 is gradually converted to NH2OH or other products, 0 min O thus preventing NH3 molecules underneath from desorb- peak A 0.5 min O 3500 1 min O ing. After about 6 minutes of O exposure, the first peak peak B 2 min O almost disappears, indicating that the top layer of NH3 is 3000 4 min O 6 min O almost all converted to NH2OH or other products. Thus 8 min O it takes at most 6 min of O (1.74 ML) to convert 1 ML 2500 16 min O of NH3, suggestion an efficiency for the reaction of each oxygen atom of at least 0.575 (=1/1.74). 2000 Using this efficiency, an upper limit for the activation

1500 energy barrier, EA, for the O + NH3 reaction may be estimated. mass 33 amu QMS signal (counts/K) 1000 We firstly assume that the incoming oxygen atom is thermalized on the surface, before overcoming the acti- 140 160 180 200 220 240 260 280 300 320 Temperature (K) vation energy barrier to reaction. This thermalization is the most likely immediate outcome of the impact, ex- Figure 4. As in Figure 3 but for QMS mass 33 amu signal. cept in a small minority of ”direct-hit” trajectories, as for 0.5 minutes, or about 1/4 of a layer, with the same the oxygen atom dissipates energy as it is drawn into the procedures as the ones followed to obtain the data in Fig- multi-dimensional potential of the surface. The atom is ure 1. This was followed by exposure to O. Mass 16 amu nevertheless available for immediate reaction if the ac- tivation energy barrier can then be overcome; in such a was chosen to represent the NH3 amount because the mass 17 amu signal has a non-trivial contribution from case, the reaction may be classed as Eley-Rideal, as it is not mediated by a diffusion process. water fragmentation when the NH3 signal is weak. In Figure 6 we show the integrated mass 16 amu TPD trace Assuming that the main loss mechanism for an oxy- for different exposures to O. With O exposure from 0 to gen atom on a pure NH3 surface is either reaction with NH or thermal desorption, the reaction efficiency may 2 minutes, the amount of NH3 follows more or less an 3 exponential decay. This is because when the O amount be formulated thus: is small, oxidation dominates and the NH3 destruction rate is proportional to the NH amount. With a further 3 1/1.74 = exp[−EA/T ]/(exp[−EA/T ]+exp[−Edes(O)/T ]) increase in O exposure, the NH3 decay does not follow (2) a simple exponential decay anymore because of possible where Edes(O) is the desorption energy for an oxy- secondary reactions. In Figure 6, a straight line is fit- gen atom and T is the surface temperature. Adopting ted to the loge plot from 0 to 2 minutes. The slope is E (O) = 1500 K and T = 70 K, this expression pro- − ± −1 des 0.46 0.06 minute . In an exponential decay, the duces a value EA(O+NH ) = 1479 K. − 3 amount of NH3 on the surface should follow exp( σφt), This value constitutes an upper boundary of the reac- where σ and φ are the reaction cross-section area and tion energy barrier because we assume a sticking coeffi- the beam flux, respectively. From it, we obtain the cross − cient of unity, and thence that each oxygen atom that section σ = −slope/φ = (1.6 ± 0.2) × 10 15 cm2. reaches the surface participates in a reaction. If this is not the case, then the reaction energy barrier for the first 3.3. Reaction energy barrier of NH3 + O oxidation should be even smaller. In typical ISM conditions, it is unlikely that situations The above treatment does not consider the possibil- arise similar to the ones that led to the appearance of the ity that surface oxygen atoms could meet each other second TPD peak of mass 33 amu at high temperature and react, reducing the efficiency of the reaction. Un- attributed to a subsequent oxidation process. Therefore der conditions in which such reactions could be impor- we focus on the first oxidation: tant, then in order for the O + NH3 reaction to be ef- ficient and the O + O reaction to be minimized, the → NH3 + O NH2OH (1) reaction barrier for O + NH3 would need to be lower When the surface is fully covered with NH and the than the oxygen diffusion barrier. This quantity has not 3 been measured in the laboratory, but there are rules- surface temperature is 70 K, O could react with NH3 also via the Eley-Rideal or hot-atom mechanisms, i.e. of-thumb based on experimental data from other sys- tems. The ratio of the diffusion energy barrier to the reaction without complete thermal accommodation of ∼ the oxygen atom. Based on prior experiments (He et al. desorption energy is typically found to be 0.3 for crys- 2014), we assume that at 70 K the residence time of O talline surfaces (Bruch et al. 2007). For rough surfaces, on NH ice is negligible. We also assume that all the values between 0.5 and 0.8 have been used (Katz et al. 3 1999; Garrod & Herbst 2006), although Garrod & Pauly reactions take place during the exposure stage instead of ∼ the TPD stage, because at 70 K the mobility of oxygen (2011) found that values lower than 0.4 were most con- should be relatively high. The degree of conversion of sistent with observed interstellar extinction thresholds NH can be gauged by looking at what remains of the for H2O, CO and CO2 ices. Garrod (2013) adopted a 3 value of 0.35, which would yield an upper limit on the O NH3 layers after various doses of oxygen. Figure 3 shows + NH3 activation energy barrier as low as 525 K. the TPD mass 17 signal (NH3) for various doses of O. The NH3 dose is fixed at 4 minutes, which is equivalent 3.4. to about 2 ML. As the O exposure time increases, the Control experiments NH3 peak decreases, and the peak position also shifts Control experiments were carried out to verify whether to higher temperatures. This is because the top layer of ammonia react with molecular oxygen or ozone. We ex- 6

5 5 10 10 mass 18 mass 18

peak A peak A peak C mass 17 mass 17

4 4 10 10 mass 16 mass 16

peak B mass 30 peak B mass 30 QMS signal (counts/K) QMS signal (counts/K)

mass 33 mass 33 3 3 10 10

100 150 200 250 300 100 150 200 250 300 Temperature (K) Temperature (K)

Figure 5. QMS signal of the TPD from an ice of NH3 (4 min) + 8 (left) and 16 (right) minutes of O exposure.

12.5 Experimental 2 ML NH + 1.1 ML O @70 K 4000 3 Fitting 2 ML NH + 1.4 ML O @70 K 12 3 2 3500 1.3 ML O + 2 ML NH @50 K 3 3

11.5 3000

2500 11 slope=-0.46±0.06/min 2000

10.5 1500 (mass 16 TPD peak area) e

log 1000 mass 33 QMS signal (counts/K) 10 500

9.5 0 0 2 4 6 8 10 12 14 16 100 150 200 250 300 O exposure time (min) Temperature (K)

Figure 6. Integrated TPD signs mass 16 amu from a deposit of Figure 7. Mass 33 QMS signal in the TPD from NH3 + O2 and NH3 (0.5 minutes or 1/4 of a ML) + various exposures of O at 70 O3 + NH3 deposited at 70K compared with that of NH3 + O at K. 50 K. Curves are offset for clarity. posed the ammonia layers to O2 and found no mass 33 or 30 amu peaks in the TPD, signifying that there has all species, Ediff = 0.35 Edes, following (Garrod 2013). been no conversion of ammonia to hydroxylamine in the Due to the low temperatures at which grain-surface ice presence of molecular oxygen. We also checked the re- mantles are formed, only surface chemistry is switched on activity of ozone with ammonia. Ozone was prepared in this model, although the three-phase treatment con- on a clean silicate following the procedure described in siders separate grain/ice surface and bulk mantle popu- He et al. (2014); then, 2 ML of ammonia were deposited lations for all species (as well as gas-phase abundances). on it. Again, there was no mass 33 or 30 amu peaks in The three-phase model allows the composition of each the following TPD, see Figure 7. layer within the ice mantle to be traced as it is de- 4. posited during the chemical evolution of the cloud (see MODELING Garrod & Pauly (2011)). Only Langmuir-Hinshelwood In order to test the astrophysical importance of the (i.e. diffusional) processes are considered in this model, O + NH3 → NH2OH reaction investigated here, we in- as the Eley-Rideal mechanism tends to require surface corporate this surface mechanism into the recent three- coverages of close to unity (in models of astrophysical phase gas-grain chemical kinetics model MAGICKAL grains) to compete with L-H mediated reactions. (Garrod 2013). The model parameters used here are in- Reactions are also included in the network allowing tended to approximate those present under dark cloud NO molecules to be hydrogenated to NH2OH by the se- conditions, during which significant ice mantles are quential addition of atomic H. Following (Garrod 2013), formed on the surfaces of dust grains. The model uti- this includes a barrier-mediated reaction, H + HNO → lizes the chemical network and initial elemental abun- HNOH, as well as an alternative H-abstraction branch, dances used by Garrod (2013), assuming a generic, static H + HNO → H2 + NO, for which the barrier is assumed 4 −3 dark-cloud gas density of nH = 2 × 10 cm and vi- to be 1500 K in both cases. Following (Hasegawa et al. sual extinction AV = 10. A gas temperature of 10 K 1992), tunneling through activation energy barriers is is used in all model runs. We assume a binding energy treated using a rectangular barrier treatment, with a de- for atomic oxygen of 1500 K, based on an experiment fault width of 1 A.˚ The faster of tunneling and thermal that measured the desorption energy of O from compact penetration is used, according to the grain temperature. amorphous water ice (He & Vidali 2014). Diffusion bar- The model described above is used to produce a grid riers are set to a uniform fraction of binding energies for of models for a range of dust-grain temperatures from 10 Formation of hydroxylamine 7

– 20 K, and using activation energy barriers, EA, for the production. O + NH3 → NH2OH reaction ranging from 0 – 2500 K. Models were also run to compare the case where the A smaller set of models is also run to compare the effects hydrogenation of NO by H atoms is assumed to occur ei- of NH2OH produced by this mechanism, with cases in ther (i) without barriers or (ii) not to occur at all. Case which the barrier to the H + HNO → HNOH reaction is (i), removal of the barrier to H + HNO → HNOH (while altered. the alternative hydrogen-abstraction branch retains its 4.1. barrier of 1500 K), in fact results in no appreciable dif- Model results ference from the runs that assume a barrier of 1500 K for Figure 8, panel (a), shows the composition of each layer H + HNO (the zero-barrier results are thus not plotted in the grain mantle ice for the model run corresponding explicitly in the figure). NH2OH production through this to EA=0 and Tgrain = 10 K. Typically observed com- mechanism is therefore already at its highest efficiency. ponents are shown, along with NH2OH. In this case, The alternative regime, case (ii), in which the H + NH2OH production is relatively small, and all such pro- HNO → HNOH barrier is increased to arbitrary height, duction is a result of the hydrogenation of NO. At 10 such that the reaction does not occur, results in negligi- K, oxygen atoms are insufficiently mobile to be able to ble NH2OH production at 12 K and below, because all reach and react with NH3 before they are hydrogenated NH2OH formation is dependent on the O + NH3 reac- to produce water, even under the assumption of a zero tion, which is inefficient at these temperatures. Above 12 barrier for the O + NH3 reaction. Here, the majority K, NH2OH is formed efficiently via O + NH3, and there of NH2OH is produced via NO hydrogenation, which is is only a small difference between these results and those strongly peaked at late time/outer layers. in which the H + HNO reaction is allowed to occur at Panels (b) and (c) show the results at 14 K, for acti- maximum efficiency. vation energy barriers of EA=1000 and 2000 K, respec- 5. tively. At this temperature, oxygen atoms are sufficiently DISCUSSION mobile to meet ammonia molecules on the surface be- Experiments of oxygen exposure on submonolayer and fore being hydrogenated. In the EA=1000 K case, the multilayer ammonia ices show an efficient formation of reaction barrier is low enough (in comparison to the dif- NH2OH with a reaction energy barrier that can be as low fusion barrier for the oxygen atom to move away from as ∼ 525K and no higher than 1479K. The experiments the ammonia molecule) to allow reaction to occur with were done on samples at 70 K to avoid contamination of high efficiency. In this case, the majority of NH2OH is molecular oxygen and ozone. At this temperature NH3 formed through this reaction, and NH2OH abundance is stuck on the surface while O has a short residence closely follows that of NH3, never falling below a factor time. Experiments have been performed at both bilayer of ten lower than NH3 at any layer in the ice. NH2OH and submonolayer coverage. Simulations were done to maintains a fairly steady abundance throughout the ice, find what rates of hydroxylamine formation one would up to around 1 Myr (170 ML), but the NH2OH produc- obtain in simulated dense cloud conditions. According tion is enhanced at all times/depths. In the EA=2000 K to the chemical models, at dust temperatures around 14 case, the reaction barrier is too high to allow reaction be- K or higher, the O + NH3 reaction is found to be an fore hydrogenation of oxygen occurs, making the NH3 + efficient mechanism for the production of NH2OH. While O reaction inefficient; here NH2OH production is again temperatures this high are not the typically considered the result of NO hydrogenation. dark-cloud dust temperatures, it is certain that the dust Figure 9 shows the final (maximum) grain-mantle in dark clouds must pass through such temperatures on abundances of NH2OH as a function of total hydrogen their way to the 8 – 10 K that is more typically assumed. in the cloud, for model runs assuming a range of grain Furthermore, interstellar dust will again pass through the temperatures from 10 – 20 K, and activation energies for ∼14 K threshold at which this reaction becomes efficient the O + NH3 reaction ranging from 0 – 1500 K. Results as part of the evolution of hot cores. for the EA=500 K case are not shown, as they are identi- 6. ASTROPHYSICAL IMPLICATIONS cal to the EA = 0 case; both values are sufficiently below the atomic oxygen diffusion barrier assumed in the model NH3 and H2O have been detected in the gas phase (525 K), such that the reaction with NH3 occurs at an ef- around transient protostellar cores or inferred for O from ficiency close to unity, if the temperature is high enough the strong O2 depletion. These cores are also classi- to allow the oxygen atom to diffuse and reach an NH3 fied as hot cores or hot corinos in the case of mas- before it is hydrogenated and becomes immobile. This sive and medium/low mass nascent stars, respectively. is the case for models of temperature 14 K and above; These cores whose temperature can be as high as 100 all models at 12 K or less produce the same results, with K close to the protostellar source (and even 250 K for NH2OH produced by NO hydrogenation. the densest ones) are much colder far from the core due Assuming a barrier of 1000 K, the reaction is still rel- to the strong extinction provided by the dusty molec- atively efficient, falling by a factor ∼2 at grain tempera- ular clumps in which they are embedded. A possible tures of ∼16 K and above, compared to the EA =0/500 scenario for the coldest regions where NH3 and O are K results. However, with a 1500 K barrier, reaction is stuck on the grains and approaching the core is that in inefficient, and yet higher barriers produce the same re- a first stage (at temperature in the 14-50 K range) O is sults; in such cases, NH2OH is produced overwhelmingly released in the gas phase and is then able to collide with by NO hydrogenation. A run with the O + NH3 bar- NH3 covered grains, inducing chemical reactions that can rier set to 1250 K allows this reaction to become slightly synthesize new species. These would later desorb from dominant over the NO hydrogenation mechanism, pro- the grains when the core temperature increases. Alterna- ducing a small, though significant, increase in NH2OH tively, NH2OH may be produced at lower temperatures, 8

Figure 9. Final model abundances of ice-mantle NH2OH with respect to total hydrogen. The grain temperature is varied for a selection of activation-energy values for the NH3 + O reaction. The dashed line shows the results assuming no barrier for either the NH3 + → NH2OH or H + HNOH → HNOH reactions.

during the formation of the ice mantles, to be released at later times. (Models of such scenarios are beyond the scope of this primarily experimental study, but will be considered in future work). Consequently, the present laboratory astrophysics ex- periment on O interaction with NH3 should be useful to interpret the observational data. Then the question is of the non-observation of hydroxylamine while, at the same time, it has been demonstrated that it can be made in the laboratory in conditions almost similar to those in protostellar cores.The upper limit relative abundance of NH2OH with respect to H2, as deduced from observa- tions (Pulliam et al. 2012) is found to be in the 3 × 10−9 to 8 × 10−12 range, depending on the object. Interest- ingly, combining this result with the relative abundance of NH3 with respect to H2 in a high-mass star-forming region, 4.6 × 10−8 (Battersby et al. 2014) (a value of 2 × 10−8 is already reported in the case of a cold dense cloud like TMC-1 (Turner 2000)), we deduce an aver- age fractional abundance of NH2OH with respect to NH3 in the 0.7 × 10−1 to 2 × 10−4 range, close to the val- ues deduced from the proposed model. Several reasons may explain the non-observation: the yield of such re- actions could be quite low due to the narrow tempera- ture range in which they have to take place as well as a possible spatial confinement of dusty clouds where the required temperature conditions for formation are ful- filled. In addition, recent observations (Sakai et al. 2013, 2014a,b; Sanna et al. 2014) have shown that the orien- tation of the protostellar core in the plane of the sky could be also important, particularly if it is associated to infall-disk-outflow systems and rotation. It could be expected that the amount of chemical species resulting from the NH3+O interaction will be almost equal both in high-mass and medium- to low-mass stars because of a higher species density in the first case and a longer in- teraction time in the other. All these conditions could explain the difficulty to observe such species as hydrox- Figure 8. Fractional composition of ices within each layer as a ylamine that have, however, been made in the labora- function of depth into the ice, lower horizontal scale, or, equiva- tory. ALMA could provide the required sensitivity and lently, as a function of age of the cloud, top horizontal scale. Panel (a) The activation energy barrier for the O+NH3 reaction is 0 K and the temperature of the ice is 10 K. Panel (b) EA=1000 K and T=14 K. Panel (c) EA=2000 K and T=14 K. Formation of hydroxylamine 9 the sub-arcsecond angular resolution to observe them. Fedoseev, G., Ioppolo, S., Lamberts, T., et al. 2012, J. Chem. A recent paper (Cleeves et al. 2014) demonstrates, Phys., 137, 54714 based on a physical and chemical model of the proto- Fenimore, C. P., & Jones, G. W. 1961, J. Phys. Chem., 65, 298 Fujii, N., Chiba, K., Uchida, S., & Miyama, H. 1986, Chem. Phys. planetary disk of the solar system, that the relative Lett., 127, 141 deuterated water abundance observed on Earth cannot Garrod, R. T. 2013, Astrophys. J., 765, 60 be explained by its formation in the disk itself but should Garrod, R. T., & Herbst, E. 2006, Astron. Astrophys., 457, 927 results in part from a primeval enrichment by interstellar Garrod, R. T., & Pauly, T. 2011, Astrophys. J., 735, 15 ices. All species embedded in these ices may then have Glavin, D. P., Dworkin, J. P., & Sandford, S. A. 2008, Meteorit. survived and have been injected into the disk and later Planet. Sci., 43, 399 Goicoechea, J. R., Rodr´ıguez-Fern´andez, N. J., & Cernicharo, J. on in the planetary system. This discovery confirms the 2004, ApJ, 600, 214 paramount importance of observing prebiotic species like Hasegawa, T. I., Herbst, E., & Leung, C. M. 1992, Astrophys. J. glycine or even smaller building blocks of such molecules Suppl. Ser., 82, 167 like hydroxylamine, which would emphasize the possible He, J., Frank, P., & Vidali, G. 2011, Phys. Chem. Chem. Phys., 13, 15803 extraterrestrial origin of life. He, J., Jing, D., & Vidali, G. 2014, Phys. Chem. Chem. Phys., 16, 7. 3493 ACKNOWLEDGMENTS He, J., & Vidali, G. 2014, Faraday Discuss., 168 This work is supported by the NSF Astronomy and —. 2014, Astrophys. J., 788, 50 Astrophysics Division (Grant No.1311958 to GV). We Herbst, E., & van Dishoeck, E. F. 2009, Annu. Rev. Astron. Astrophys., 47, 427 thank Dr. J. Brucato of the Astrophysical Observatory Hidaka, H., Watanabe, M., Kouchi, A., & Watanabe, N. 2011, of Arcetri (Italy) for providing the sample used in these Phys. Chem. Chem. Phys., 13, 15798 experiments and Zhirou Zhang, Jianming Shi, and Tyler Hily-Blant, P., Maret, S., Bacmann, A., et al. 2010, A&A, 521, Hopkins for technical help. RTG acknowledges the sup- L52 Ioppolo, S., Fedoseev, G., Minissale, M., et al. 2014, Physical port of the NASA Astrophysics Theory Program (Grant Chemistry Chemical Physics (Incorporating Faraday No. NNX11AC38G). Transactions), 16, 8270 Jim´enez-Serra, I., Testi, L., Caselli, P., & Viti, S. 2014, ApJL, REFERENCES 787, L33 Jing, D., He, J., Bonini, M., Brucato, J. R., & Vidali, G. 2013, J. Altwegg, K. 2014, in ESA blog, Phys. Chem. A, 117, 3009 http://blogs.esa.int/rosetta/2014/10/23/the–perfume–of–67pc– Katz, N., Furman, I., Biham, O., Pirronello, V., & Vidali, G. g/ 1999, Astrophys. J., 522, 305 Bacmann, A., Caux, E., Hily-Blant, P., et al. 2010, A&A, 521, L42 Kolasinki, K. W. 2008, Surface science : Foundations of catalysis Barrientos, C., Redondo, P., Largo, L., Ray´on, V. M., & Largo, and nanoscience (Wiley: Hoboken, N.J.) A. 2012, Astrophys. J., 748, 99 Le Gal, R., Hily-Blant, P., Faure, A., et al. 2014, A&A, 562, A83 Battersby, C., Bally, J., Dunham, B., et al. 2014, ApJ, 786, 116 Nishi, N., Shinohara, H., & Okuyama, T. 1984, J. Chem. Phys., Baulch, D. L., Campbell, I. M., & Hainsworth, R. 1984, J. Chem. 80, 3898 Soc., Faraday Trans. 1, 80, 2525 Perry, R. A. 1984, Chem. Phys. Lett., 106, 223 Belloche, A., Menten, K. M., Comito, C., et al. 2008, A&A, 482, Pulliam, R. L., McGuire, B. a., & Remijan, A. J. 2012, 179 Astrophys. J., 751, 1 Blagojevic, V., Petrie, S., & Bohme, D. K. 2003, Mon. Not. R. Sakai, N., Oya, Y., Sakai, T., et al. 2014a, ApJ, 791, L38 Astron. Soc., 339, L7 Sakai, N., Sakai, T., Hirota, T., et al. 2014b, Nature, 507, 78 Bolina, A. S., Wolff, A. J., & Brown, W. A. 2005, J. Chem. Phys., Sakai, T., Sakai, N., Foster, J. B., et al. 2013, ApJ, 775, L31 122 Sanna, A., Cesaroni, R., Moscadelli, L., et al. 2014, A&A, 565, Bordalo, V., da Silveira, E. F., Lv, X. Y., et al. 2013, Astrophys. A34 J., 774, 105 Shu, J., Lin, J. J., Wang, C. C., Lee, Y. T., & Yang, X. 2001, J. Bottinelli, S., Boogert, A. C. A., Bouwman, J., et al. 2010, ApJ, Chem. Phys., 115, 842 718, 1100 Snow, J. L., Orlova, G., Blagojevic, V., & Bohme, D. K. 2007, J. Bruch, L., Diehl, R., & Venables, J. 2007, Rev. Mod. Phys., 79, Am. Chem. Soc., 129, 9910 1381 Tielens, A. G. G. M. 2005, The Physics and Chemistry of the Caselli, P., Keto, E., Bergin, E. A., et al. 2012, Astrophys. J., 759, Interstellar Medium L37 Turner, B. E. 2000, ApJ, 542, 837 Charness, C. 2013, Am. Astron. Soc. Meet. Abstr., 221 Wang, L., Mebel, A. M., Yang, X., & Wang, X. 2004, J. Phys. Charnley, S. B., Rodgers, S. D., & Ehrenfreund, P. 2001, A&A, Chem. A, 108, 11644 378, 1024 Wincel, H., Fokkens, R. H., & Nibbering, N. M. M. 2000, Rapid Cleeves, L. I., Bergin, E. A., Alexander, C. M. O., et al. 2014, Commun. Mass Spectrom., 14, 135 Science, 345, 1590 Yildiz, U. A., Acharyya, K., Goldsmith, P. F., et al. 2013, 558, Cohen, N. 1987, Int. J. Chem. Kinet., 19, 319 A58 Congiu, E., Chaabouni, H., Laffon, C., et al. 2012b, J. Chem. Zheng, W., & Kaiser, R. I. 2010, J. Phys. Chem. A, 114, 5251 Phys., 137, 054713 Zins, E. L., & Krim, L. 2014, in 69th International Symposium on Congiu, E., Fedoseev, G., Ioppolo, S., et al. 2012a, ApJL, 750, L12 Molecular Spectroscopy, in press Del Bene, J. E. 1972, J. Chem. Phys., 57 Di Francesco, J., Hogerheijde, M. R., Welch, W. J., & Bergin, E. A. 2002, AJ, 124, 2749 Elsila, J. E., Glavin, D. P., & Dworkin, J. P. 2009, Meteorit. Planet. Sci., 44, 1323