Astronomy&Astrophysicsmanuscriptno.spinstate˙benchmark (cid:13)c ESO2015 January21,2015 Benchmarking spin-state chemistry in starless core models O.Sipila¨1,2,P.Caselli1,andJ.Harju2,1 1 Max-Planck-InstituteforExtraterrestrialPhysics(MPE),Giessenbachstr.1,D-85748Garching,Germany e-mail:[email protected] 2 DepartmentofPhysics,POBox64,00014UniversityofHelsinki,Finland Received/Accepted 5 ABSTRACT 1 0 Aims.Weaimtopresent simulatedchemical abundance profilesfor avarietyof important species, withspecial attention givento 2 spin-statechemistry,inordertoprovidereferenceresultsagainstwhichpresentandfuturemodelscanbecompared. Methods.Weemploy gas-phaseandgas-grainmodelstoinvestigatechemical abundances inphysical conditions corresponding to n starlesscores.Tothisend,wehavedevelopednewchemicalreactionsetsforbothgas-phaseandgrain-surfacechemistry,including a J the deuterated forms of species with up to six atoms and the spin-state chemistry of light ions and of the species involved in the ammonia andwater formationnetworks. Thephysical model iskept simpleinorder tofacilitatestraightforwardbenchmarking of 0 othermodelsagainsttheresultsofthispaper. 2 Results.Wefindthattheortho/pararatiosofammoniaandwateraresimilarinbothgas-phaseandgas-grainmodels,atlatetimesin particular,implyingthattheratiosaredeterminedbygas-phaseprocesses.Furthermore,theratiosdonotexhibitstrongdependence ] A oncoredensity.Wederivelate-timeortho/pararatiosof∼0.5and∼1.6forammoniaandwater,respectively.Wefindthatincluding orexcludingdeuteriuminthecalculationshaslittleeffectontheabundancesofnon-deuteratedspeciesandontheortho/pararatios G of ammonia and water,especially ingas-phase modelswhere deuteration isnaturally hindered owing tothepresence of abundant . heavyelements.Althoughwestudyarathernarrowtemperaturerange(10−20K),wefindstrongtemperaturedependencein,e.g., h deuterationandnitrogenchemistry.Forexample,thedepletiontimescaleofammoniaissignificantlyreducedwhenthetemperature p isincreasedfrom10to20K;thisisbecausetheincreaseintemperaturetranslatesintoincreasedaccretionrates,whiletheveryhigh - o bindingenergyofammoniapreventsitfrombeingdesorbedat20K. r Keywords.ISM:abundances–ISM:clouds–ISM:molecules–astrochemistry t s a [ 1. Introduction Owingtoitsimportance,spin-statechemistryisnowwidely 1 adopted in numerical chemical models, and has been ap- v Theortho-to-pararatioofmolecularhydrogen,H ,playsalarge 2 plied to all the stages in the star formation process, i.e., dif- 5 roleinthedevelopmentofdeuteriumchemistryatthehighden- fuse clouds (Albertssonetal. 2014a), starless/prestellar cores 2 sityandlowtemperatureattributedtostarlesscores.Thedeuter- (Walmsleyetal. 2004; Floweretal. 2004; Paganietal. 2009; 8 ationsequencebeginswiththeexothermicreactionbetweenH+ 4 3 Sipila¨ etal. 2010, 2013; hereafter S13) and protostellar sys- andHD: 0 tems (Taquetetal. 2013; Taquetetal. 2014). However, spin- 1. H+3 +HD←→H2D++H2+232K. (1) sptraotbelecmhesmaidsotrpytinisguvseuraylldyiffdeirsecnutsspehdysiincatlhme ocdoenltse,xatnodfcsopmecpiafirc- 0 The H D+ ion can then donate its deuteron to other abundant ison between the results of different works can be difficult. In 5 2 1 species, such as CO and N2, yielding DCO+ and N2D+, re- thispaper,weaimtoremedythisbypresentingrelativelyeasily : spectively. However, spin-state effects complicate the reaction reproduciblereferenceresultspertainingtospin-statechemistry, v scheme. Chemical species with multiple protons, or deuterons, inphysicalconditionsappropriatetostarlesscores.Tothisend, i X can exist in different spin configurations– for example H has we have developeda state-of-the-artspin-state chemicalmodel 2 two distinct spin states where the nuclear spin wavefunction thatincludesthespinstatesoflighthydrogen-containingspecies r a is either symmetric (ortho-H ; hereafter oH ) or antisymmet- (H , H+...),and the spin states of the speciestaking partin the 2 2 2 3 ric (para-H ; hereafter pH ). The difference in energy between formationanddestructionnetworksofammoniaandwater.Our 2 2 thegroundstatesofthesetwospinstates, ∼ 170K(Hugoetal. modelalso includesthe deuteratedformsof species with up to 2009),canbeaveryimportantenergyreservoiratthelowtem- sixatoms.Toextendtheusabilityofourresults,wepresentre- peraturesofstarlesscores.Indeed,reaction(1)canproceedrel- sultsforbothgas-phaseandgas-grainmodels. ativelyeasilyinthebackwarddirectionwhentheorthoformsof bothH D+andH areinvolved.Consequently,theortho-to-para 2 2 (hereaftero/p)ratioofH isanimportantparametercontrolling 2 deuteriumchemistry(Floweretal.2006b).Inadditiontodeuter- Thepaperisorganizedasfollows.Sect.2describesourphys- ation,spin-stateeffectsplayalargeroleinotherareasofstarless icalandchemicalmodelsindetail.InSect.3,wepresentthere- corechemistryaswell.Forexample,theformationchainofam- sultsofourcalculations.InSect.4,wediscussourresultsandin moniadependscriticallyontheN++H reaction,whichrequires Sect.5 we presentour conclusions.AppendicesA to E include 2 the presence of oH to proceed efficiently at low temperatures complementarydiscussiononourmainresults,andalsopresent 2 (Dislaireetal.2012). additionalmodelingresults. 1 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels Table1.Adoptedvaluesofthevariousphysicalparameters(see However,therearesomeexceptions.Forphotodissociationreac- textfordefinitionsoftheparameters). tions,theratecoefficientisdefinedas k =αexp(−γ A ) , (3) Parameter Value photo 1 V T =T 10K gas dust where A is the visual extinction. For cosmic-ray-induceddis- ζ 1.3×10−17s−1 V sociation,theratecoefficientisdefinedas A 10mag V a 0.1µm ρg 3.0gcm−3 kCR =γ2ζ, (4) g n 1.5×1015cm−2 s whereζ isthecosmicrayionizationrate.InEqs.(3)and(4),γ E /E 0.77 1 d b andγ areefficiencyfactors. R 0.01 2 d Theabundanceofdustgrainsiscalculatedas Table2.Bindingenergies(correspondingtoawatericesurface) X = ng =R µmolmH , (5) ofselectedspecies. g nH d43πa3gρg Species Bindingenergy[K] whereµ isthemeanmolecularweightperHatom(1.4);ρ is mol g H 450 the grainmaterialdensity;R is the dust-to-gasmassratio.For d H2 500 gas-phase chemicalreactions involvinga grain and another re- C 800 actantwithdifferentelectriccharge,theratecoefficientismulti- N 800 pliedbythe“J-factor”takingintoaccountincreasedreactionef- N 1000 2 ficiencyduetopolarization(Draine&Sutin 1987;Paganietal. O 1000 2 2009;Sipila¨etal.2010). CO 1150 Gas-phasechemistryis linkedwith grain-surfacechemistry O 1390 NH 2378 through adsorption and desorption processes. The adsorption OH 2850 ratecoefficientofspeciesiisgivenby NH 3956 2 NH 5534 kads =υ Sσ, (6) 3 i i H O 5700 2 whereυ = 8k T /πm isthethermalspeedofspeciesi(k i p B gas i B istheBoltzmannconstantandm isthemassofspeciesi);S is i 2. Model the sticking coefficient, set to unity for all species; σ = πa2 is g thegraincrosssectionassumingsphericalgrainswithradiusa . g 2.1.Physicalparameters Theadopteddesorptionmechanismis cosmic-rayinduceddes- orption,withratecoefficientgivenby Inordertofacilitatestraightforwardcomparisonoffuturemod- elingworksagainstthe resultspresentedin this paper,we con- kdes = f kTD(70K)= f ν exp −E /70K , (7) siderhomogeneousmodels.Thatis,wefixthevaluesofallphys- i 70 i 70 0,i (cid:2) b,i (cid:3) ical parameters but the density of the medium. The model pa- where f = 3.16 × 10−19 is an efficiency factor; ν = rameters(introducedbelow)alongwiththeirassumedvaluesare 70 0,i 2n k E /π2m is the characteristic vibration frequency of presentedinTable1.Inallcalculations,wesetthegastempera- p s B b,i i species i (Hasegawaetal. 1992; Hasegawa&Herbst 1993). In tureT equaltothedusttemperatureT .Althoughwerestrict gas dust the above, n and E stand for the density of binding sites on thetemperaturetoT = 10Kinthemaintext,wepresentrefer- s b,i thegrainsurfaceandthebindingenergyofspeciesionthegrain ence results also for T = 15K and T = 20K in AppendixC surface, respectively. For simplicity, we consider only cosmic- (theseresultsarebrieflydiscussedinSect.4.5). rayinduceddesorptioninthiswork,andneglectalternativedes- The binding energies of the variousspecies, corresponding orptionmechanismssuchasphotodesorptionorreactivedesorp- to a water ice surface, are mainly taken from Garrod& Herbst tion. (2006;see alsoSipila¨ 2012). Table2presentsthe bindingener- The rate coefficient for a grain-surface reaction between giesofselectedspecies.Weassumethatthebindingenergyofa speciesiand jisgivenby deuteratedspeciesisequaltothatofthecorrespondingundeuter- atedspecies.Thisapproachhasbeenpreviouslyadoptedby,e.g., Cazauxetal.(2010),Taquetetal.(2013)andS13.Tocovertyp- kij =ακij(cid:16)Rdiiff +Rdjiff(cid:17)/ng (8) icaldensityvaluesassociatedwithstarlesscores,wepresentre- whereαisthebranchingratioofthereaction;theefficiencyfac- sultsatfourdifferentdensitiesrangingfrom103to106cm−3. torκ = exp(−E /T )orunityforexothermicreactionswith ij a dust orwithoutactivationenergy(E ),respectively;n isthenumber a g 2.2.Chemicalcode density of dustgrains(see below).The (thermal)diffusionrate Rdiff isgivenby We use the gas-grain chemical code discussed in Sipila¨ etal. (2010),Sipila¨ (2012)andS13,whichusestherateequationap- Rdiff = νi exp(−E /T ), (9) proachto calculate chemicalevolutionin the gasphase and on i N d,i dust s grainsurfaces.Thereactionratecoefficientis,forthemajorityof gas-phasereactions,definedwiththemodifiedArrheniusequa- where Ns = ns4πa2g isthenumberofbindingsitesonthegrain tionas andEd,iisthediffusionenergyofspeciesi.Thediffusionenergy isdeterminedbyassumingaconstantvalueforthediffusion-to- k=α(T/300K)βexp(−γ/T) . (2) binding-energyratioE /E . d b 2 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels Table3.Initialchemicalabundanceswithrespectton ,andthe with rate coefficient k, separates into three reactions when the H adoptedinitialH o/pratio. spinstatesofthespeciescontainingmultipleprotonsareconsid- 2 eredexplicitly: Species Initialabundance H 0.5 H+(o) + O−→k OH++H (o) 2 3 2 He 9.00×10−2 HD 1.60×10−5 H+(p) + O−21→k OH++H (o) (11) O 2.56×10−4 3 2 C+ 1.20×10−4 H+(p) + O−12→k OH++H (p). N 7.60×10−5 3 2 S+ 8.00×10−8 In S13, the branchingratio matrices for some reactingsystems Si+ 8.00×10−9 werederivedbyhandandappliedintheo/pseparationroutine. Na+ 2.00×10−9 However, for more complicated reactions involving many pro- Mg+ 7.00×10−9 tonsand/ormultipleproducts,customseparationruleswereap- Fe+ 3.00×10−9 plied(seeAppendixAinS13).Themotivationfordoingsowas P+ 2.00×10−10 threefold.Firstly,Oka’smethodmaynotbeapplicablewhenthe Cl+ 1.00×10−9 exothermicityofagivenreactionisverylow.Secondly,wedid H (o/p) 1.00×10−3 2 ini not want to follow the spin states of species heavier than H+ 3 (with the exceptionof its deuterated forms),so in any reaction Weassumethatthegasisinitiallyatomicwiththeexception wheresuchspecies(e.g.,H2S)werepresent,weassumedthatthe ofhydrogenanddeuteriumwhicharelockedinH andHD,re- speciesisinitsparaform.Thirdly,thebranchingratiomatrices 2 spectively (Table 3). We used the same assumptions about the were deduced by hand, which quickly becomes tedious when initial abundances as Semenovetal. (2010) except for HD for multiple protons and product species are present, even though whichweadoptedn(HD)/n(H) = 1.6×10−5 basedontheD/H theseparationmethoditselfisstraightforward. ratiomeasuredinthelocalISM(Linskyetal.1995).Theinitial Forthispaper,wehavedevelopedanewversionofthesepa- H ortho/pararatiois(arbitrarily)setto1×10−3. rationroutine.Inthenewversion,thebranchingratiosarecalcu- 2 Finally we notethatwe do notconsiderquantumtunneling latedautomaticallyusingOka’smethodsothatweeasilyobtain on grain surfacesin the main bodyof the text. However,refer- thebranchingratiomatrixofanyreaction.Consequently,every ence results including tunneling are presented in Appendix E, specieswithmultipleprotonsisseparatedintoitsorthoandpara wherewealsodiscussthemodificationstothegas-grainmodel (andmetawhenapplicable)states.However,inordertomaintain requiredwhentunnelingisincluded. relativesimplicityofthereactionnetwork,weonlykeepthespin statesofthelighthydrogen-bearingspecies,andofthosespecies includedin theformationnetworksofwater andammonia(see 2.3.Gas-phaseandgrain-surfacereactionsets Sect.2.3.2). When spin species other than those mentioned above are Thechemicalreactionsetsforgas-phaseandgrain-surfacereac- presentin a reaction,we recombinethereactionsoverthe “un- tions used in this work are reworked versions of the networks wanted”spinspecies.Forexample,considerthereaction presented in S13; the present gas-phase network is based on osu 01 20091 insteadofthemodifiedversionofosu 03 2008 (Semenovetal.2010)adoptedinS13.Inthepresentmodel,the H+3 +CH2−k→1 CH+3 +H2. (12) spin-state chemistry description of S13 is expanded by adding When separated using spin selection rules, reaction (12) thespinstatesofthespeciesinvolvedintheformationofwater branchesintothefollowingsetofreactions: andammonia.Also,thepresentmodelcontainsdeuteratedforms toafnstpsepceiceisewssituhchupastow6ataetro,mams,minocnluiadianngdombseethrvaantoiol.nWalleydimiscpuosrs- oH+3 +oCH2 −3670→k1 oCH+3 +oH2 thevariousupdatesandadditionsindetailinthefollowing. −650→k1 oCH++pH 3 2 2.3.1. Thespin-stateseparationroutine −1640→k1 pCH++oH 3 2 IanraSte13an,yarreoauctitnioenwinavsodlevvineglolpigedhttohyadurtoogmeant-ibceaallryinsgpisnp-esctaietes(sHep2-, −640→k1 pCH+3 +pH2 oHn+2parned-dHe+3te)r.mIninperdacbtircaen,cthhiengroruattiinoes.crIenatSe1s3n,ewtherebarcatniocnhsinbgasread- oH+3 +pCH2 −41→k1 oCH+3 +oH2 t(i2o0s0f4o)r, minowsthriecahcrteiopnresswenetraetidoendsuocfedtheusninugcltehaer smpeinthwodavoeffuOnkca- −41→k1 oCH+3 +pH2 trieoancstivoeftchoellvisairoionus.sTspheecrieessualtrienugsberdantochdienrgivreatsieolsecctoiorrnesrpuolensdfotor −42→k1 pCH+3 +oH2 tphuerenunculcelieaarrespcionmsptaletitsetliycaml iwxeeidghintsthuendreeracthtieona,ssaunmdpthtiuosntthheayt pH+3 +oCH2 −1640→k1 oCH+3 +oH2 a(Oreklaik2e0l0y4t)o.Fboermexoasmt applep,litchaebrleeatcotiohnighly exothermicreactions −1600→k1 oCH+3 +pH2 H+3 +O−→k OH++H2, (10) −2680→k1 pCH+3 +oH2 1 Seehttp://www.physics.ohio-state.edu/∼eric/ −680→k1 pCH+3 +pH2 3 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels pH++pCH −51→k1 oCH++oH adopted here (Semenovetal. 2010), the non-additionreactions 3 2 3 2 thatcreateH∗ (theasteriskdenotesasurfacespecies),forexam- 2 −52→k1 pCH++oH pleH∗+H2CO∗ −→HCO∗+H∗2,alreadypresenthighactivation 3 2 barriers.Ofcourse,oH∗canbecreatedby(cr-induced)photodis- 2 −52→k1 pCH++pH . sociations;weassumethattheseeventsaresufficientlyenergetic 3 2 sothatoH∗ andpH∗ areproducedinthestatisticalratio3:1. 2 2 TodeterminethebranchingratiosoftheoH++CH −→CH++ We notethattheconservationofnuclearspin maynothold 3 2 3 (o/p)H system,i.e.,neglectingthespinstatesofCH andCH+, for grain-surface reactions as there is experimental evidence 2 2 3 wesumovertheoH andpH productionpathwaysandaverage to the contrary(Fushitani&Momose2002; Hama&Watanabe 2 2 theratecoefficientsonthenumberofspinmodificationsofCH 2013).WediscussthisissuebrieflyinSect.4.4. 2 (two).AsimilartreatmentofthepH++CH −→CH++(o/p)H 3 2 3 2 systemfinallyresultsinfourreactions: 2.3.2. Spin-statechemistryofammoniaandwater oH++CH −45→k1 CH++oH (13) In this work, we have included the spin-state chemistry 3 2 3 2 of nitrogen-containing species involved in the ammonia for- −15→k1 CH++pH (14) mation chain, recently studied using the Oka method by 3 2 LeGaletal. (2014; see also Floweretal. 2006a; Dislaireetal. pH++CH −2103→k1 CH++oH (15) 2012;Faureetal.2013;Ristetal.2013).Wehavecheckedthat 3 2 3 2 ouro/pseparationroutineproducesthesamebranchingratiosas −270→k1 CH++pH . (16) given in Tables B.1. and B.2. in LeGaletal. (2014) (see also 3 2 Sect.4.2.) ThebranchingratiosinEqs.(13)to(16)arebasedontheas- Because of the high exothermicities of the hydrogen addi- sumptionthattheorthoandparaformsofCH areequallyabun- tionreactionsongrainsurfacescreatingNH∗andNH∗(∼4.2eV 2 2 3 dant. Using the statistical ortho/para CH ratio (3:1), the four and 4.7eV, respectively, as calculated from Eqs.(3) and (4) in 2 branchingratiosare33/40,7/40,27/40and13/40,respectively, Allen&Robinson1977andusingdatapresentedinAppendixC while assuming that CH is completely in the para state yields ofDuetal.2012),weassumethatthehigh-temperaturestatisti- 2 thebranchingratios3/4,1/4,3/5and2/5.Ifweassumethatthe calbranchingratiosfortheformationreactions, reactionproceedsthroughaprotonhopprocessinsteadofcom- pletescrambling,weobtainthebranchingratios1,0,1/2and1/2 NH∗+H∗ −3→/4 oNH∗ 2 (Oka2004,Table8b).Evidently,therearedifferencesintheout- comedependingonthe assumedortho/pararatioofCH orthe −1→/4 pNH∗ 2 2 assumed reaction mechanism (which is not known for the vast oNH∗+H∗ −2→/3 oNH∗ majorityofreactions).Wenote,however,thatheaviermolecules 2 3 havea negligibleeffectonthe H2 ortho/pararatioforexample, −1→/3 pNH∗ andourassumptionsseemjustifiedinthepresentcasewherewe 3 onlyconsiderexplicitlythespinstatesofalimitedsetofspecies. pNH∗+H∗ −→1 pNH∗, 2 3 It should be noted that the Oka method allows the forma- tion of oH from pH , simply because this is statistically pos- arevalid.Thesebranchingratiosare,ofcourse,givendirectlyby 2 2 sible.However,theformationofoH shouldbeunlikelyatlow theOkamethod. 2 temperatureswherethecollisionalenergiesaresmallcompared Wehavealsoincludedthespin-statechemistryofthespecies with the energy difference ∆E/k = 170K between the ground involved in the water formation network. The branching ratios statesofoH andpH ,unlessthereactionissufficientlyexother- for the most important reactions in the water network are pre- 2 2 mic.In S13,we assumedthatthe requiredexothermicityisnot sented in AppendixB. Similarly to the nitrogen chemistry, we reached in general, and hence the network favored pH forma- adoptstatisticalbranchingforwaterformationongrainsurfaces: 2 tion.Inthispaper,weslightlyrelaxtheassumptionsofS13,and apply the separation routine to all reactions other than charge- OH∗+H∗ −3→/4 oH O∗ 2 transfer reactions, where we assume that spin states are con- served. However, we add an activation energy of 170K to the −1→/4 pH2O∗. (17) γ coefficient in those reactions where oH is created by reac- 2 tantswhosespinstatesarenotexplicitlyconsidered(forexam- 2.3.3. Deuteration ple,CH++CH −→C H++H ;seealsoAlbertssonetal.2014a). 3 2 2 3 2 ThenewapproachmodifiestheresultsofS13onlyveryslightly; In S13, a deuterationroutine was applied to the OSU network, see Sect.3.1. With these changes, the resultant network should andspecieswithuptofouratomsweredeuterated.Inthepresent be better applicable at higher temperatures (Albertssonetal. work,thedeuterationroutineisextendedtohandlespecieswith 2014b;seealsoAlbertssonetal.2014a).Thecombinationofnu- uptosixatoms,sothatthedeuteratedformsofimportantspecies clearspinstatisticalweightsandactivationenergiesforprobably such as methanol and ammonia (whose formation depends on endoergicreactionsapproximatesthemethodHugoetal.(2009) NH+) can be included. The deuteration routine is based on 4 used to calculate the thermal rate coefficients for the H+ +H thatpresentedbyRodgers&Millar (1996),in whichdeuterons 3 2 isotopicsystem,butwenotethatadetailedstate-to-stateanaly- are substituted in place of protons in the various reactions and sis such as that performed by Hugoetal. (2009) is required to branching ratios are calculated assuming complete scrambling studyendothermicreactionsconsistently. (seealsoS13). We apply the same spin-state separation routine to grain- Thecalculationofthenuclearspin-statebranchingratiosfor surfacereactions.Noratecoefficientcorrectionshavebeenmade multiply-deuterated species is more complicated than for un- tothesurfacereactionsbecauseinthebasesurfacereactionset deuterated species. For species with three or more deuterium 4 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels nuclei Oka’s method, based on angular momentum algebra, is D++CH 29k2ex−p(→−170/T) CD++oH insufficient because there is no one-to-one correspondencebe- 2 2 tween angular momentum and symmetry representations. The −19→k2 CD++pH statistical branching ratios between the nuclear spin states of 2 species with up to five hydrogen or deuterium nuclei can be −96→k2 CH++HD obtained from Tables III and IV of Hugoetal. (2009). In the present paper, the spin-state chemistry in reactions between D++CHD −29→k2 CH++oD 2 species containing H and/or D only is included using these rules,andtheyaremostlyavailableintheliterature.Specifically, 91k2ex−p→(−86/T) CH++pD 2 we include the spin-state chemistry of the H+ + H react- ing system from Hugoetal. (2009), complem3ented 2by data −69→k2 CD++HD fboerarointhgerspreeaccietison(sH+3c,onHce2rDn+in,gHlDighetthc.y)drforogmen-Wanadl-mdesluetyereiutmal-. D++CD2 −32→k2 CD++oD2 (c2ia0t0iv4e)raencdomFlboiwnaetrioetnabl.et(w20e0en4)H.T+haendraittescdoeeuffitecriaetnedtsffoorrmdsisasnod- 31k2ex−p→(−86/T) CD++pD2. 3 electronsaretakenfromPaganietal.(2009). Reactions yielding pH or oD are assumed to have no activa- Thenuclearspinchemistryofmultiply-deuteratedmolecules 2 2 tionenergies,whereasthoseproducingpD areassumedtohave containing heavy elements is not considered in the present 2 anactivationenergyof86K,correspondingtotheenergysepa- model,with the exceptionthat it is necessaryto predictthe ra- ration between the ground states of pD and oD (Hugoetal. tioofoD andpD releasedinreactionsinvolvingheavyspecies 2 2 2 2 2009). As noted above, the branching ratios for the reactions toachieveaproperlyclosedreactionnetwork.Furthermore,oD 2 producing D are correct under the implicit assumption that andpD producedbythesereactionscanberelevanttothechem- 2 2 [oCD ] : [pCD ] = 2:1. However, as in the case of H , reac- istry of light molecules containing only H and/or D which are 2 2 2 tionsinvolvingheavyspecieshaveanegligibleeffectontheD treatedusingtheappropriatespinselectionrules(seebelow). 2 o/pratio.Note thatfortheH+ +CHD andD+ +CH reactions Thedeuterationprocedureusedheredirectlypropagatesthe 2 above,one would expectH formationin the o/p ratio 3:1, but hydrogenspin-statechemistryintothedeuteriumchemistry,for 2 in the example the ratio is instead 2:1 because the deuterated molecules with up to two deuterium nuclei. For hydrogenated reactionsadoptthebranchingratiosfromtheundeuteratedpar- species like CH , we assumed that the ortho and para forms 2 entreactions,inthiscasereactions(19),wherethethree-proton of CH are equally abundant(see Sect. 2.3.1).However,copy- 2 systemleadstoanH o/pratioof2:1. ingthereactionschemefromhydrogenatedtodoublydeuterated 2 We haveverifiedthroughtesting thatthe spin separationof moleculesisequivalenttofollowingthecorrectnuclearspinse- doubly-deuterated(heavy)speciesisoflittleconsequencetothe lectionrulesforspin1systemswiththeassumptionthatoD and 2 resultspresentedlaterinthispaper,andthereforetheseparation pD areinsteadpresentintheratio2:1(i.e.,thestatisticalvalue) 2 method used here seems reasonable for the present work. We incomplexeswithheavynuclei,e.g.,CD . 2 willdiscussthespinseparationofmultiply-deuteratedmolecules As an example of the spin-state separation of doubly- indepthinafuturededicatedpaper. deuteratedspecies,considerthereaction H++CH −k→2 CH++H . (18) 2.3.4. OtherupdateswithrespecttoS13 2 2 In addition to the updates to the o/p separation and deutera- Sinceourprogramperformsthespin-stateseparationofhydro- tion routines described above, the rate coefficients of several gen species prior to deuteration, the above reaction is first di- reactions have been updated. First of all, we now adopt the videdintothefollowingtwobranches: osu 01 2009 reaction set instead of (modified) osu 03 2008, so any updates of the base reaction set, including anion chem- 2 H++CH −→ CH++oH , k= k exp(−170/T), istry, are included in our new model. The modifications made 2 2 2 3 to osu 03 2008 by Semenovetal. (2010), already included in 1 −→ CH++pH , k= k . (19) S13,havebeenincorporatedtothenewmodelaswell. 2 2 3 Wehavealsoincludednewratecoefficientdatafromthelit- erature. We adopt the rate coefficients given in Tables 12 and Thebranchingratio(2/3oH ,1/3pH )followsfromthenuclear 2 2 13 in Albertssonetal. (2013) – except for reactions involving spinstatisticsundertheimplicitassumptionthatpCH andoCH 2 2 hydrogen and/or deuterium onlyfor which we use other data haveequalabundances,i.e.[pCH ]= [oCH ].Deuteratingboth 2 2 as detailed above. We have also included selected data from branchesleadsinourroutinetothefollowingreactions: LeGaletal.(2014),whichwesummarizeinTableB.2.We re- ferthereadertoAlbertssonetal.(2013)andLeGaletal.(2014) H++CHD 92k2ex−p(→−170/T) CD++oH fortheoriginaldatareferences. 2 −91→k2 CD++pH2 SectT.h3e.1.impact of the rate coefficient updates is analyzed in −96→k2 CH++HD H++CD −29→k2 CH++oD 3. Results 2 2 In this section, we present benchmarking results for spin-state 19k2ex−p→(−86/T) CH++pD chemistry for different values of density, using gas-phase and 2 gas-grainmodels.Alloftheresultspresentedbelowcorrespond −96→k2 CD++HD toT =10K.ResultsforT =15KandT =20Karepresentedin 5 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels 10-3 10-4 10-3 10-4 CO oH 10-5 H2O∗ 2 HD 10-5 10-6 10-4 nH 1100--76 H2O NH3 10-7 D NH3∗ / 10-8 10-5 nX 1100--98 10-9 e− HDO∗ 1100--1110 N2H+ HCO+ 1100--1110 oH2D+ N DD+CO+ 10-6 NH2D∗ 10-12 10-12 2 10-7 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 t [yr] t [yr] t [yr] 10-3 10-4 10-3 10-4 CO oH 10-5 H2O∗ 2 HD 10-5 10-6 10-4 nH 1100--76 H2O NH3 10-7 D NH3∗ / 10-8 10-5 nX 1100--98 10-9 e− HDO∗ 1100--1110 N2H+ HCO+ 1100--1110 oH2D+ N DD+CO+ 10-6 NH2D∗ 10-12 10-12 2 10-7 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 t [yr] t [yr] t [yr] 10-3 10-4 10-3 10-4 CO oH 10-5 H2O∗ 2 HD 10-5 10-6 10-4 nH 1100--76 H2O NH3 10-7 D NH3∗ / 10-8 10-5 nX 1100--98 10-9 e− HDO∗ 1100--1110 HCO+ N2H+ 1100--1110 oH2D+ NDDC+O+ 10-6 NH2D∗ 10-12 10-12 2 10-7 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 t [yr] t [yr] t [yr] Fig.1.Upperpanels:abundancesofselectednon-deuterated(left),deuterated(middle)andgrain-surface(bothnon-deuteratedand deuterated,right)speciesasfunctionsoftime,ascalculatedwithmodelT1(solidlines)ormodelT2(dashedlines).Middlepanels: as the upper row, but comparingmodels T1 (solid lines) and T3 (dashed lines).Lowerpanels:as the other rows, but comparing modelsT1(solidlines)andF1(dashedlines).Inallcalculations,thedensityofthemediumissetton =105cm−3.Theabundances H ofH O,NH andNH Drepresentsumsovertheirrespectiveorthoandparastates. 2 3 2 AppendixC,whileresultswithoutdeuteriumandincludingtun- impactonthe overallchemistry.In thefollowing,we studythe neling are presented in AppendicesD and E, respectively.The effect of the various updates, by successively introducing new data for the variousfigures presented below are available from updates into the S13 chemical model. The models introduced theauthorsuponrequest. belowarealsodescribedinTable4. First,weconstructedatestreactionsetbyspin-stateseparat- ingthe S13 base reactionset (modifiedosu 03 2008)with the 3.1.ComparisonofthenewmodelagainstS13 new routine presented here, but using the old S13 deuteration In our new model, there are two main sources of updates: 1) routine and including no new rate coefficient data (model T2). our new spin-state separation and deuteration routines and 2) TheupperrowinFig.1plotstheabundancesofselectedspecies the rate coefficient updates for various species from the litera- as functions of time calculated using model T2 (dashed lines) tureandfromthetransitionfromosu 03 2008toosu 01 2009. andusingtheoldmodelofS13(solidlines;modelT1).Inthese Qualitatively, the new spin-state separation routine may affect calculations,thedensityofthemediumissetton = 105cm−3, H somewhattheabundancesofthevariousspeciesbecauseofthe andtheotherphysicalparameterscorrespondtothevaluesgiven increasedemphasisonoH inthenewseparationroutine(albeit inTable1.Intherightpanel,theabundancesoftherespectiveor- 2 with a 170K barrier in most cases). However,the extensionof thoandparastatesofH O∗,NH∗andNH D∗havebeensummed 2 3 2 deuterationfromfourtosixatomsisnotexpectedtohavealarge over. It is observed that the new spin-state separation routine 6 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels oH oH+ oH D+ pD H+ 10-2 2 10-8 3 10-9 2 10-10 2 110034 10-9 10-10 10-11 nH 105 10-11 10-12 / 10-3 106 10-10 10-13 nX 10-12 10-14 10-11 10-13 10-15 10-4 10-12 10-14 10-16 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 N H+ N D+ oNH pNH 10-8 2 10-10 2 10-7 3 10-7 3 10-9 10-11 10-8 10-8 H10-10 n / 10-11 10-12 10-9 10-9 X n 10-12 10-13 10-10 10-10 10-13 10-14 10-14 10-11 10-11 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 oH O pH O HCO+ DCO+ 10-5 2 10-5 2 10-8 10-9 10-6 10-6 10-9 10-10 nH 10-11 / 10-7 10-7 10-10 nX 10-12 10-8 10-8 10-11 10-13 10-9 10-9 10-12 10-14 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 10-3 CO 10-5 D 10-4 HD 10-6 e− 10-7 H10-4 10-6 n / 10-5 10-8 X n 10-5 10-7 10-9 10-6 10-8 10-6 10-10 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 t [yr] t [yr] t [yr] t [yr] Fig.2. Abundances of selected species calculated with model F2 (i.e., surface reactions and adsorption/desorption excluded) as functionsoftimeatdifferentdensities,labeledintheupperleftpanel.Thelabelsindicatedensitiesofthemedium;forexample,103 standsforn =103cm−3. H modifiesonlyslightlytheabundancesofnon-deuteratedspecies modelatlongtimescalesbecauseoflessefficientHDdepletion. atearlytimesbothinthegasphaseandongrainsurfaces,while Inspectionofthereactionratesrevealsthatthiseffectiscaused the abundances of deuterated species are virtually identical in bythedissociationofHDO∗,the maingrain-surfacedeuterium both cases. Hence the new oH -creating reactions with 170K carrierinourmodel,whosedissociationproducedonlyOD∗+H∗ 2 barriers included here (as opposed to S13) are of little conse- in S13 owing to a bug in the deuterationroutine, but here pro- quence to the chemistry as a whole (althoughat high tempera- ducesalsoOH∗+D∗.Consequently,moreD∗isavailabletoform turesthesituationmightbedifferent). HD∗,replenishingthegas-phaseHDabundance. The middle row in Fig.1 plots the abundances of selected InthebottomrowofFig.1,weplotagaintheabundancesof species calculated with a test reaction set constructed by ap- selectedspecies,nowcalculatedwiththefullmodelpresentedin plying the new spin-state separation and deuteration routines this paper, i.e., including the rate coefficient updates discussed to the modified osu 03 2008 reaction set; as before, no new inSect.2.3.4(modelF1).Evidently,theratecoefficientupdates ratecoefficientdatais included(modelT3).Itisobservedthat, leadtosignificantdifferenceswithrespecttothepreviouscases, overall,theextensionofthedeuterationschemehaslittleeffect bothinthegasphaseandongrainsurfaces.Forexample,theCO on the chemistry.However,deuterationis enhancedin the new formationtimescaleismuchshorterwiththeupdatedratecoef- 7 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels oH oH+ oH D+ pD H+ 10-2 2 10-8 3 10-9 2 10-10 2 10-3 10-9 10-10 10-11 nH 10-12 / 10-4 103 10-10 10-11 nX 104 10-13 10-5 110056 10-11 10-12 10-14 10-6 10-12 10-13 10-15 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 N H+ N D+ oNH pNH 10-8 2 10-10 2 10-7 3 10-7 3 10-9 10-11 10-8 10-8 H10-10 n / 10-11 10-12 10-9 10-9 X n 10-12 10-13 10-10 10-10 10-13 10-14 10-14 10-11 10-11 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 oH O pH O HCO+ DCO+ 10-5 2 10-5 2 10-8 10-9 10-6 10-6 10-9 10-10 nH 10-7 10-7 10-11 / 10-8 10-8 10-10 nX 10-9 10-9 10-12 10-10 10-10 10-11 10-13 10-11 10-11 10-12 10-14 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 10-3 CO 10-5 D 10-4 HD 10-6 e− 10-4 10-7 H 10-6 10-5 n / 10-5 10-8 X n 10-7 10-6 10-6 10-9 10-7 10-8 10-7 10-10 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 103 104 105 106 107 t [yr] t [yr] t [yr] t [yr] Fig.3.AsFig.2,butcalculatedwithmodelF1.NotethattheabundancescalesareinsomecasesdifferentthaninFig.2. Table4.Themodelsdiscussedinthiswork. initially very rapidly in the new model. From an observational point of view, the most striking difference between the models Model Description istheabundanceofN H+,whichisoveranorderofmagnitude 2 T1 theSipila¨etal.(2013)model loweratlate timesin thenewmodel,resultingmainlyfromin- T2 osu 03 2008+newspin-stateseparationroutine; creased(factor2-3)electronDRratecoefficientsandalsofrom olddeuterationroutine;noratecoefficientupdates changes in neutral-neutral chemistry, with respect to the OSU T3 osu 03 2008+newspin-stateseparationroutine+ values.Thechangesis N H+ andHCO+ abundancespropagate newdeuterationroutine;noratecoefficientupdates 2 totheirdeuteratedcounterpartionsN D+ andDCO+.Also,the F1 osu 03 2008+newspin-stateseparationroutine+ 2 abundanceofammoniadecreasesbyafactorof2-3atlatetimes newdeuterationroutine+ratecoefficientupdates (fullmodel) inthenewmodelwithrespecttoS13. F2 modelF1withoutgas-graininteraction(i.e.,no adsorption/desorptionorsurfacereactions); Surfacechemistryisalsoaffectedbytheratecoefficientup- H formationfromKongetal.(2013) dates.Inthenewmodel,atomichydrogenisatearlytimespro- 2 ducedmainlybyneutral-neutralreactionsbetweenlighthydro- carbonsinsteadofelectronDRreactionsofvariousions(owing ficients, which is mainly due to numerous changes in electron todecreasedelectronabundance),andthereismoreatomichy- DRratecoefficients.Accordingly,theelectronabundancedrops drogenavailableinthegasphase.Thehydrogenisthenadsorbed 8 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels ontograinsurfaces,leadingtoefficientproductionof,e.g.,water 1.0 and ammonia which form mainly by hydrogenation.We stress 0.9 ggggaaaassss pppphhhhaaaasssseeee ggggaaaassss ggggrrrraaaaiiiinnnn thattheseeffectsarecausedbychangesinthegas-phasechem- −−−− −−−− 0.8 istry, as none of the surface reaction rates have been updated. o These results emphasize the importanceof constrainingthe re- ati 0.7 r actionratecoefficients,particularlyforkeyreactions. p 0.6 / Our chemical model predicts gas-phase and grain-surface o 0.5 3 abundances in physical conditions attributed to starless cores, H 0.4 N and our model results could in principle be used to interpret 0.3 ice observations. However, direct observations of grain-mantle 0.2 species toward starless cores are extremely hard owing to the 0.1 verylargeextinctions.Thus,wewouldneedtorelyon,e.g.,ob- 103 servationstowardembeddedClass0sources,assumingthatthe 2.6 104 icefeatureoriginatesinthesourceenvelope.Adetailedinvesti- atio 2.4 110056 gationofmodelediceabundancesversusobservationsisleftfor r 2.2 p futurework,andwefocusourattentionongas-phasespeciesin / thepresentpaper. Oo 2.0 In what follows, all results have been produced with the H21.8 newortho/paraseparationanddeuterationroutines,andinclude 1.6 the rate coefficient updates from the literature as described in Sect.2.3.4. 1.1403 104 105 106 103 104 105 106 107 t [yr] t [yr] 3.2.Gas-phasechemistry Fig.4.o/pratiosofNH (upperpanels)andH O(lowerpanels) 3 2 Figure 2 present the results of gas-phase modeling at different as functions of time at different densities, labeled in the upper densities. Here, we have notconsidered adsorptionand/ordes- left panel. The left panels correspond to the gas-phase model orptionofanyspecies,andtheformationofH2,HDandD2 on (F2), while the right panels correspond to the gas-grain model grainsurfacesisincludedin theformofgas-phasereactionsas (F1). inthemodelofKongetal.(2013). Ion abundances decrease significantly with increasing den- lar) nitrogen is relatively low (Table 2), so nitrogen chemistry sity,whichisaconsequenceoftheincreasedelectronDRratesat can still occur efficiently after other heavy elements have been highdensity,andofthehighabundancesofheavyneutralspecies depletedontograinsurfaces.Consequently,anappreciableabun- owingtotheabsenceofdepletion.Theabundancesofheavyions follow largely that of H+ which is primarily destroyed by re- danceofN2,theprecursormoleculeofbothNH3andN2H+(see, 3 e.g.,Fontanietal.2012),canbepresentinthegasatlatetimes. actions between abundantneutral species like CO and N , and electrons.OneexampleofthistendencyisHCO+, which2isthe This is demonstrated by two effects. Firstly, the abundance of NH isless thanan orderofmagnitudelowerinthe fullmodel most important reaction partner of free electrons at late times 3 thaninthegas-phasemodelathighdensity,eventhoughdeple- bothatlowandhighdensities.Thisionis formedina reaction between H+ and CO and destroyed mainly in electron recom- tion tends to decrease its abundance. Secondly, the abundance 3 ofN H+, dependentonH+, isordersofmagnitudehigherwith binationwhichreturnsCO.Athighdensities,theenhancedpro- 2 3 respecttothegas-phasemodelevenathighdensity(weassume ductionoffreeelectronsbythecosmicrayionizationofH leads 2 thationsdonotdepleteontograinsurfaces,butcanrecombine toanincreaseintheelectronDRrates(atconstanttemperature), andtheHCO+abundancedecreasesaccordingly.N H+islocked withnegativelychargedgrains). 2 inasimilarcycle(H++N −→N H+;N H++e− −→N +H), Deuterationincreasesstronglywhengas-graininteractionis dependenton the ab3undan2ce of H2+. Also2the N H+ abun2dance taken into account.However,a localpeak in deuterationis ob- 3 2 servedaroundthetimethatHDstartsdeplete,inlinewiththere- decreasesclearlywhenthedensityincreases. sultsofS13.HDdepletionissomewhatlesssevereinthepresent Deuteration is also suppressed in the absence of depletion, modelthaninS13becausewedonotconsiderquantumtunnel- becauseH+reactspreferentiallywiththeabundantheavyspecies 3 ing.Inagreementwithpreviousresultsintheliterature,ourre- ratherthanwithHD. sultsclearlyshowthatdepletionisneededtoproduceobservable amountsof,e.g.,oH D+ andN H+ athighdensity. 2 2 3.3.Gas-grainchemistry Figures2and3indicatethattheo/pratiosofammoniaand waterdependonlyslightlyondensity,bothinthegas-phaseand Figure 3 presents the results of gas-grain modeling at differ- gas-grainmodels.Wediscussthisissuefurtherbelow. ent densities, adopting the full model (F1) described above. Evidently,theinclusionofdepletionsignificantlydecreasesthe abundances of carbon and oxygen-containing species with re- 4. Discussion specttothegas-phasemodel.TheH+abundanceismuchhigher 3 4.1.Theo/pratiosofwaterandammonia atlatetimesinthefullmodelwithrespecttothegas-phasemodel because of the depletion of its main reaction partners (e.g., Figure4plotstheo/pratiosofNH andH Oatdifferentdensi- 3 2 CO) onto grain surfaces. The chemistry of nitrogen-containing tiesaccordingtogas-phaseandgas-grainmodels.Theo/pratio speciesisdifferentthanthatof,e.g.,carbonandoxygenbecause ofammoniaisverysimilarinbothmodelsupton =105cm−3. H nitrogen chemistry depends on slow neutral-neutral reactions. Theratioislargerat∼ 104−105 yearsatn = 106cm−3 inthe H On the other hand, the binding energy of (atomic and molecu- gas-phasemodel,inwhichoneofthemainproductionpathways 9 O.Sipila¨,P.Caselli,andJ.Harju:Benchmarkingspin-statechemistryinstarlesscoremodels forNH+att∼105yristheH O++NH −→NH++H Osystem Table5.Abundancesandabundanceratiospredictedbyourgas- 4 3 3 4 2 whichformsallthreespinvariantsofNH+.However,mNH+for phasemodel(F2)andthatofLeGaletal.(2014). 4 4 example only dissociates to oNH (see Table 2 in LeGaletal. 3 2014). The detailed behaviorof the spin-state chemistry is dif- Parameter Ourmodel LeGaletal.(2014) ferentinthegas-grainmodel,wherethereactionpathwayswith [NH] 5.9×10−9 1.1×10−8 H O+ are practically absentat t & 105yr because of water de- [NH ] 9.5×10−10 3.5×10−9 3 2 pletion.Neverthelessourresultsindicatethat,atlatetimes,o/p- [NH ] 3.2×10−8 1.4×10−8 3 NH3is∼0.4−0.5regardlessoftheconsideredmodelorassumed [NH2]/[NH] 0.16 0.3 density. [NH3]/[NH] 5.4 1.3 The o/p ratio of H2O behaves similarly to that of NH3 in [[ooNNHH2]]//(cid:2)ppNNHH2(cid:3) 01..48 02..65 thesensethatthelargestdifferencesbetweenthegas-phaseand 3 (cid:2) 3(cid:3) gas-grainmodelsare observedat n = 106cm−3. Theo/pratio H ofH OistiedtothatofH O+,whichevolvesdifferentlyinthe gas-g2rainmodelowingtom3 ultiplespin-statedetails(H O+ and and[S] = 8×10−8).LeGaletal.(2014)donotexplicitlystate 3 H2O are connectedmainlythroughH2O+ +H2 −→ H3O+ +H theinitialH2o/pratio;weassumedavalueof3. andtheH++H O−→H O++H system).Theo/p-H Oratiosat Table5summarizestheresultsofthesetests.Inourmodel, differentd3ensit2iesandin3differe2ntmodelsare,again,2veryclose therearemanycosmic-ray-inducedandphotodissociationreac- toeachotheratlatetimeswherewederiveo/p-H O∼1.5−1.7. tionswithlongtimescales,whichpreventthesystemfromreach- 2 Unlike the o/p ratio of H which is strongly influenced by ingatruesteady-state;thevaluesgiveninthetable correspond 2 grain-surfacechemistry,theo/pratiosofH O andNH arede- to late-time chemical evolution (t = 2.0×107yr), after which 2 3 terminednearly completelyby gas-phasechemistrybecause of temporal changes in chemical abundances are generally small. theirhighbindingenergies(seeTable2).Thefactthattheo/pra- The Le Gal et al. steady-state values have been read off their tiosoftherespectivespeciesareverysimilarinbothgas-phase Figs. 4 and 5. Evidently, our model gives lower NH and NH2 and gas-grain models supports this interpretation (Fig.4). The abundancesandahigherNH3 abundancethantheLeGaletal. late-timevaluesoftheo/pratioscanbejustifiedbyaratecoeffi- model,buttheortho/pararatiosofbothNH2andNH3aresimilar cientanalysisofthemainformationanddestructionpathsofam- inbothmodels. moniaandwater;suchananalysisispresentedinAppendixA. Wealsotestedaninitial[C]/[O]ratioof0.3,andfoundthat We note that the late-time H2O o/p ratios derived here inthiscasethelate-timeNHandNH2 abundancesarelowerby are much lower than those typically observed in the ISM one and two orders of magnitude, respectively, than those pre- (vanDishoecketal. 2013;see also the discussion in Ketoetal. dictedbyLeGaletal.(2014).However,theNH3abundanceand 2014).Thedisagreementisprobablyduetothefactthatobser- theNH2andNH3o/pratiosareagainsimilartotheLeGaletal. vationstowarddarkcloudswithintermediatetohighdensityare (2014)model. missing. To investigatethis issue, we ran test calculationswith Our test results indicate that there are minor differences ourgas-grainmodelatconditionssimulatingtranslucentclouds in fractional abundances between our model and that of (n = 102cm−3 andT = 10K)with two valuesof A , 10mag LeGaletal. (2014). The two models predict similar o/p ratios H V and1mag.We foundthatat AV = 10mag,theH2Oo/pratiois for NH2 and NH3, which is expected since we adopt the same ∼ 1.6,butat A = 1magtheratiois ∼ 3,suggestingthatpho- spin-statechemicaldescriptionforthesespeciesasLeGaletal. V tochemistryplaysalargeroleindeterminingtheo/pratioinre- (2014). gionswithlowvisualextinction.Thisresultunderlinestheneed foraccuratephysicalmodelswheninterpretingobservations. 4.3.H formationingas-phasemodels Ammonia is a useful probe of the gas temperature in dark 2 clouds (Tafallaetal. 2004; Juvelaetal. 2012), and simulated In the gas-phase results presented above, the formation of H 2 NH line emission profiles can be used to constrain the ki- (anditsisotopologsHDandD )isincludedintheformofgas- 3 2 netic temperature. However, a proper comparison of simulated phasereactions(seeKongetal.2013).FromFig.2itisevident line emission against a detection of, e.g., pNH requires an thatthisapproachleadsto, e.g.,an increase ofthe H o/pratio 3 2 estimate of the NH o/p ratio. Given that the NH o/p ratio at late times. Another approach to including the formation of 3 3 changes very little with density in our models, we deduce that H anditsisotopologsin a gas-phasemodelisto invokegrain- 2 o/p-NH ∼ 0.4−0.5 is a goodconservativeestimate ofthe ra- surfaceformationprocesses,butlettingonlyatomicHandDto 3 tioindarkclouds.Thisresultisconsistentwithobservationsby beadsorbedontothegrainsurfaces. Perssonetal.(2012)andthemodelsofLeGaletal.(2014)(see WehaveruntestcalculationsatT =10Kcomparingthetwo alsoSect.4.2). approaches. The results of these calculations are presented in Finally, we stress that we have not carried out a complete Fig.5,wheretheabundancesofselectedspeciesasfunctionsof parameter-spacestudyoftheammoniaandwatero/pratios,and timeareplottedatdifferentdensities,usingdifferentapproaches thatsuchastudyshouldbetakenoninthefuture. to H formation. Evidently, the Kongetal. (2013) approach is 2 practically equivalent to the grain-surface formation approach when “normal”bindingenergies(450K for H and D) are used 4.2.ComparisonagainstLeGaletal.(2014) (thedottedanddash-dottedlinesarepracticallysuperimposed). Recently,LeGaletal.(2014)presentedgas-phasemodelingre- Inthegas-grainmodelwithverylowHbindingenergy(100K), sults of nitrogen chemistry in dark clouds. They included in H desorbsveryfast,leadingtoinefficientH productiononthe 2 their modelthe spin-state chemistrypertainingto the ammonia grain surfaces regardlessof the density. Increasing the binding formation network, as we do here. We have run some test cal- energyincreasestheaveragetimespentbyHatomsonthegrains culations to compare our results against those of LeGaletal. andthusleadstomoreefficientH production. 2 (2014),adoptingthesamephysicalconditionsandinitialchem- The choice of the H formation efficiency has little effect 2 icalabundancesforHe,C,NandFe(wechoose[C]/[O] = 0.8 ontheabundanceofH D+ becausedeuteriumchemistryissup- 2 10