Reactive Desorption and Radiative Association as Possible Drivers of Complex Molecule Formation in the Cold Interstellar Medium
aa r X i v : . [ a s t r o - ph . GA ] A p r Reactive desorption and radiative association as possible driversof complex molecule formation in the cold interstellar medium
A.I. Vasyunin
Department of Chemistry, The University of Virginia, Charlottesville, Virginia, USA [email protected]
Eric Herbst
Departments of Chemistry, Astronomy, and Physics, The University of Virginia,Charlottesville, Virginia, USA [email protected]
ABSTRACT
The recent discovery of terrestrial-type organic species such as methyl formateand dimethyl ether in the cold interstellar gas has proved that the formation oforganic matter in the Galaxy begins at a much earlier stage of star formationthan was thought before. This discovery represents a challenge for astrochem-ical modelers. The abundances of these molecules cannot be explained by thepreviously developed “warm-up” scenario, in which organic molecules are formedvia diffusive chemistry on surfaces of interstellar grains starting at 30 K, andthen released to the gas at higher temperatures during later stages of star for-mation. In this article, we investigate an alternative scenario in which complexorganic species are formed via a sequence of gas-phase reactions between precur-sor species formed on grain surfaces and then ejected into the gas via efficientreactive desorption, a process in which non-thermal desorption occurs as a resultof conversion of the exothermicity of chemical reactions into the ejection of prod-ucts from the surface. The proposed scenario leads to reasonable if somewhatmixed results at temperatures as low as 10 K and may be considered as a steptowards the explanation of abundances of terrestrial-like organic species observedduring the earliest stages of star formation.
Subject headings: astrochemistry – ISM – molecular processes 2 –
1. Introduction
Although it has been understood for some time that a combination of ion-moleculeand neutral-neutral reactions can explain much of the exotic “carbon-chain” chemistry thatoccurs in the gas phase of cold interstellar cores, both starless and prestellar, the chem-istry of hot cores, in which gaseous terrestrial (partially saturated) organic species of sixor more atoms, known as “complex organic molecules”, or COMs for short, are formedat relatively high abundance, is not well understood (Herbst & van Dishoeck 2009). Well-known examples of COMs include methanol (CH OH), dimethyl ether (CH OCH ), ethylcyanide (C H CN), and methyl formate (HCOOCH ). Early theories of the formation ofthese molecules were based on a warm gas-phase chemistry during the hot core stage, pos-sibly preceded by a cold ice-mantle chemistry during the prior cold core stage (Brown et al.1988; Charnley et al. 1992; Hasegawa et al. 1992; Caselli et al. 1993; Horn et al. 2004). Thecold ice chemistry is thought to be dominated by reactions involving weakly-bound atomssuch as hydrogen, which can diffuse at low temperatures. More recently, the chemistry oc-curring during the warm-up leading to the formation of a hot core has been emphasized.In this approach, the chemical synthesis of hot core molecules begins on the ice mantles ofgrains in cold cores, where species as complex as methanol are synthesized (Garrod & Herbst2006). As the cold core collapses isothermally into a pre-stellar core and then begins to heatup as a protostellar core is formed, the collapsing gas and dust also increase in temperature.Starting at temperatures of approximately 30 K, the heavy species formed on the 10 K icebegin to diffuse and collide with one another, although they are mainly unreactive. Photonsor energetic particles strike the dust particles and dissociate some of these heavy species intoradicals, which are likely to be quite reactive. For example, methanol can be dissociatedto produce the methoxy radical (CH O) while formaldehyde can form the formyl radical(HCO). These two radicals can possibly undergo what is known as a recombination reactionto form methyl formate (HCOOCH ).Many other COMs are likely formed via recombination of radicals on ices during thewarm-up period, as discussed by Garrod & Herbst (2006) and Garrod et al. (2008), whoshowed that large abundances of COMs in the gas can be achieved by the stage at which thetemperature of the collapsing material has risen to 100 - 300 K, where thermal evaporationtakes place. At temperatures above 30 K but below 100 K, smaller abundances of complexspecies formed on grains can also be produced via non-thermal desorption as long as theradical-radical recombination reactions can occur. Indeed, observations in various sourceshave shown that small abundances of gaseous COMs can be found in cooler sources than hotcores such as the galactic center or along the line of sight to protostars ( ¨Oberg et al. 2010).But, below about 25 K, it would appear that the diffusion of radicals on ice surfaces canno longer occur sufficiently rapidly to react to form COMs (Garrod & Herbst 2006). Thus, 3 –based solely on the warm-up radical mechanism, one would not expect to find any moleculesof this type in the gas of cold cores such as TMC-1.Two recent observations of a number of terrestrial-type complex organic moleculesin the cold prestellar cores L1689b and B1-b have excited the astrochemical community(Bacmann et al. 2012; Cernicharo et al. 2012) although some more limited work had reachedsimilar conclusions for a variety of sources including TMC-1 regarding methanol and ac-etaldehyde (CH CHO). New observations of other cold cores (Bacmann, private communica-tion(2013)) also show the presence of COMs. The COMs seen in L1689b and/or B1-b includemethanol (CH OH), acetaldehyde (CH CHO), methyl mercaptan (CH SH), dimethyl ether(CH OCH ), and methyl formate (HCOOCH ), while smaller species include the methoxyradical (CH O), detected in B1-b for the first time, the formyl radical (HCO), formic acid(HCOOH), ketene (CH CO), and formaldehyde (H CO). Of the complex species, only theproduction of methanol is well understood; it is formed via non-thermal desorption followingthe successive hydrogenation by atomic hydrogen on ice mantles of CO that is accreted fromthe gas (Watanabe & Kouchi 2002). In this paper, we propose and investigate a scenario forthe formation of the complex terrestrial-type organic molecules methyl formate, dimethylether, and acetaldehyde in cold cores. We also report our results for the gas-phase abun-dances of ketene, formaldehyde, methoxy, and methanol. The scenario involves an enhancedrate of non-thermal desorption of selected precursor molecules (e.g. methoxy, formaldehyde,methanol) from cold ice mantles by a process known as reactive desorption, in which theexothermicity of surface chemical reactions is at least partially channeled into kinetic energyneeded to break the bond of the product with the ice. The enhanced desorption rate isfollowed by gas-phase reactions leading to the COMs, among which are radiative associationprocesses with rate coefficients that are still poorly understood. Extension of this approachto other COMs is certainly feasible.The remainder of this paper is organized as follows. In Section 2 we describe our chemicalmodel, while in Section 3 modeling results are presented and compared with observationaldata. Section 4 contains a general discussion, while Section 5 is a summary of our work.
2. Chemical model
For this study, we utilized a gas-grain chemical model with a network of gas-phase andgrain-surface chemical reactions taken from that used by Vasyunin & Herbst (2013), whichin turn is based on the well-established OSU astrochemical database . The diffusive surface ∼ eric/research.html O) was detected in the cold core B1-b (Cernicharo et al. 2012), we introduced four gas-phase reactions for this species, which is produced during the hydrogenation of CO on grainmantles followed by desorption. In our network, the methoxy radical and its isomer, thehydroxymethyl radical (CH OH), are not distinguished. The first two reactions are simpledestruction reactions involving the abundant atoms H and O:CH O + H → CH + OH , (1)and CH O + O → H CO + OH . (2)Both of these reactions have been studied to some extent in the laboratory for methoxy andhydroxymethyl radicals, as can be seen in the NIST Chemical Kinetics Database . For thereaction with atomic hydrogen, only the process involving CH OH has a listed and reviewedrate coefficient, which possesses a temperature independent value of 1.6 × − cm s − inthe range 300-2000 K. We assume that the rate coefficient remains constant at temperaturesdown to 10 K in our model. For the reaction with atomic oxygen, measured and reviewedvalues both exist at room temperature for CH OH only; we assume that the average result,1.0 × − cm s − , pertains down to 10 K in our model.The third and fourth reactions occur between the methoxy and methyl radicals:CH O + CH → H CO + CH , (3)and CH O + CH → CH OCH + hν. (4)The rate coefficient of reaction (3), also taken from the NIST Chemical Kinetics Database,has been measured to be 4.0 × − cm s − by several investigators at room temperatureand higher. We assume that this value pertains down to 10 K. Reaction (4) is a radiativeassociation, which has not been studied in the laboratory to the best of our knowledge. Thesystem has been studied at high density, where it essentially reaches the collisional limit atroom temperature. The radiative mechanism, which is the low density process relevant tothe interstellar medium, is likely to be much slower at room temperature, but to have adependence on temperature of T − (Herbst 1980) so that by 10 K, reaction (4) is also quite http://kinetics.nist.gov/kinetics/Search.jsp − (T /
300 K) − cm s − with a large uncertainty. For convenience, information on newand other important reactions is summarized in Table 1.The elemental abundances available for gas-grain chemistry in our simulations corre-spond to the EA1 set from Wakelam & Herbst (2008). These are so-called “low metal”abundances, which are lower by a factor of ∼
100 for heavy elements in comparison to solarelemental abundances in order to take into account elemental depletion on grains in coldinterstellar cores. The ionization rate ζ (s − ) due to the main ionizing agent — cosmic rays– has a standard value of 1.3 × − s − for molecular hydrogen. The initial abundances areall atomic except for hydrogen, which starts in the form of H . The gas-to-dust mass ratiois 0.01.To treat the diffusive surface chemistry, we assume that the ice mantle surroundinginterstellar grains can be represented as a two-dimensional lattice with a periodic potential.Wells of the potential are the binding sites for atomic and molecular adsorbates. Due to ther-mal hopping, species can move across the grain surface and react with each other. In thecase of atomic and molecular hydrogen, quantum tunneling through the potential barriers isalso considered as a source of surface mobility. The width of a potential barrier between twobinding sites is assumed to be 1 ˚A. The height E b of a barrier against diffusion for a speciesis assumed to be 1 / E D . In our simulations, we consider sphericalgrains of a single size of 10 − cm with ∼ binding sites. Note that our network is inheritedfrom previous studies in which complex organic molecules were formed at higher tempera-tures via diffusive surface chemistry. It contains surface radical-radical chemical reactionsthat lead to the formation of COMs. Since at 10 K, the mobility of molecular species on sur-faces is negligible, these reactions do not contribute to the formation of COMs in this study.The only efficient diffusive reactants on grain surfaces at 10 K are atoms such as hydrogen,oxygen and nitrogen, especially atomic hydrogen, which can be an efficient tunneler. Forexample, hydrogen atoms that accrete onto dust particle surfaces are able to hydrogenateCO ice all the way to methanol ice at 10 K (Charnley et al. 1997; Watanabe & Kouchi 2002):CO → HCO → H CO → H CO → CH OH . (5)The chemistry in the gas phase is connected to grain-surface chemistry via accretion anddesorption. The accretion probability for all neutral species is set to unity except for helium,for which it is taken to be zero because it is chemically inert in surface chemistry and hasa very low desorption energy, so that it would thermally desorb rapidly anyway. Molecularions are not allowed to stick on grains. There are four types of desorption included in ourmodel: thermal evaporation (Hasegawa et al. 1992), photodesorption (Fayolle et al. 2011),desorption due to grain heating by cosmic ray bombardment (Hasegawa & Herbst 1993), and 6 –reactive desorption (Garrod et al. 2007). In cold dark cloud cores, which are well shieldedfrom UV radiation and have very low temperatures of ≈
10 K, thermal evaporation is nilexcept for He, H, and H while photodesorption can only occur slowly via photons producedby energetic electrons. Desorption due to cosmic rays has limited efficiency and has only beenproven to be important on large timescales for species such as CO (Hasegawa & Herbst 1993).Constraints on the efficiency of reactive desorption, in which the products of an exothermicsurface reaction are immediately ejected into the gas, are poor. To the best of our knowledge,there are no experimental constraints on its efficiency except for the case of H (Katz et al.1999), and only a simple statistical theory is available for larger products (Garrod et al.2007). In this study, we consider three different efficiencies for reactive desorption: 0%, 1%,and 10% per exothermic reaction on a grain surface. The 1% efficiency is sufficient to explainobserved fractional abundances of methanol ( ≈ − ) in cold cores (Garrod et al. 2007). The10% value is equal to the highest efficiency considered by Garrod et al. (2007). Our modelswith these three different efficiencies for reactive desorption are labeled M0, M1, and M10.All chemical models are run for 10 yr, which is slightly longer than so-called “earlytime”, at which the ion-molecule chemistry best reproduces the carbon-chain species observedin cold cores. We utilize physical conditions typical for sources such as L1689b and B1-b,where COMs were found at low temperature: T=10 K, a proton density n H =10 cm − , anda visual extinction A V =10. In our analysis, we focus on the time span 10 —10 yr.
3. Results
Because the calculated ice mantle composition in all three models is quite similar, wepresent the composition for only one model – M10 – in Figure 1. The similarity of the icecomposition in all considered models arises because in this work reactive desorption affectsthe rates of all surface reactions by a relatively small factor of at most 10%. As such, itdoes not change considerably the balance between different surface chemical processes, andthe resulting ice composition remains almost unchanged. As expected, our chemical modelproduces a typical ice composition which is similar to that calculated in a number of previousstudies (Garrod & Pauly 2011; Vasyunin & Herbst 2013), and which is also generally similarto observed ice abundances towards protostars ( ¨Oberg et al. 2011). The major ice componentis water, which has a nearly constant fractional abundance of ∼ − after 10 yr. Theabundances of the carbon-bearing major ice components change with time more significantly.Carbon monoxide accretes from the gas phase and at 10 yr possesses an abundance on grainsurfaces similar to that of water. By 10 yr, its abundance has decreased by approximatelya factor of 3 as it is gradually converted to methanol. Correspondingly, the abundances of 7 –formaldehyde, methanol and methane rise with time by an order of magnitude or more toreach values of roughly ten to thirty percent of the abundance of water ice. Another majorice constituent observed in ices towards star-forming regions, carbon dioxide, has a very smallfractional abundance of ∼ × − in our model, despite a much larger observed abundance.This discrepancy arises because carbon dioxide is probably formed during other stages ofstar formation when dust grains have higher temperatures of ∼
20 K (e.g., Garrod & Pauly2011; Vasyunin & Herbst 2013).In Figure 2, abundances of selected gas-phase species are presented for models M0, M1,and M10, with formaldehyde, methanol, and methoxy in the left panels and methyl formate,ketene (CH CO), dimethyl ether, and acetaldehyde in the right panels. The upper row ofpanels represents results for M0 (no reactive desorption), the middle row results for M1(1% reactive desorption efficiency), and the lower row results for model M10 (10% reactivedesorption efficiency). The results for M1 are similar to those of the best-fit model in theearlier paper of Garrod et al. (2007). Fractional abundances of all these species observedin L1689b and B1-b are shown in the two rightmost columns of Table 2. We derived theobserved fractional abundances in these sources by dividing the observed column densitiesas presented in Bacmann et al. (2012) and Cernicharo et al. (2012) by the column densitiesof molecular hydrogen in L1689b and B1-b. The observed fractional abundances in L1689brange from approximately 10 − to 10 − whereas those in B1-b range from approximately10 − to more than 10 − . Given the uncertainties both in observed abundances and in ourastrochemical models, we consider a model to explain successfully the observed abundanceof a species if the modeled abundance differs from the observed abundance by not more thanone order of magnitude (Vasyunin et al. 2004, 2008; Wakelam et al. 2010). Formaldehyde is mainly produced by the gas-phase reaction O + CH → H CO + H,while ketene (CH CO) has two paths of formation. First, it is produced in the dissociativerecombination of C H O + , which in turn is created in the radiative association reactionbetween the second most abundant gas-phase molecule CO and the simple ion CH +3 :CO + CH +3 → C H O + + hν. (6)Secondly, ketene is produced in the neutral-neutral reactionO + C H → CH CO + H . (7)Other observed organic molecules in the model M0 have very low abundances, because theydo not have gas-phase routes of formation efficient enough at 10 K to compete with accretion. 8 – Our results change drastically when reactive desorption is enabled. For model M1, as canbe seen in the left middle row of Figure 2, formaldehyde and methanol reach peak abundancesin excess of 10 − , while the methoxy radical reaches a peak abundance of 10 − . As discussedearlier, methanol is not formed in the gas, but is formed by surface hydrogenation of COfollowed by desorption (Garrod et al. 2007). This is not only true for CH OH but also forintermediate products of the reaction chain in eq. (5), and other products of surface reactions.As such, reactive desorption becomes a major source of methanol and the methoxy radicalin the gas phase. The abundance of formaldehyde is affected to a lesser extent becauseit is already produced efficiently by gas-phase processes. The increase in the gas-phaseabundances of methoxy, methanol and formaldehyde after 3 × yr is caused by the increasedabundances of these species in the ice mantle, as can be seen in Figure 1.For the molecules in the right middle panel, model M1 also shows enhanced abundancesover M0. While the final abundance of ketene (CH CO) increases only by a factor of three,the abundance of acetaldehyde (CH CHO) is increased by four orders of magnitude, and theabundance of dimethyl ether (CH OCH ) is increased by more than ten orders of magnitude.The fractional abundances of both species with respect to hydrogen now exceed 10 − towards10 yr. The additional acetaldehyde is produced via the gas-phase reaction O + C H → CH CHO + H. The C H product is ejected into the gas-phase via reactive desorption as anintermediate product of the chain of surface reactions that hydrogenate C into C H :C → C H → C H → C H → C H → C H → C H . (8)Gas-phase dimethyl ether, on the other hand, is mainly produced via the gas-phase radiativeassociation reaction shown in eq. (4). Although inclusion of the “standard” reactive desorp-tion with an efficiency of 1% in the model allows us to produce complex organic moleculesin much higher abundances than with no reactive desorption at all, some of the predictedabundances are still quite low. For example, methyl formate (HCOOCH ) has an abundancein model M1 below 10 − in the entire time span 10 –10 yr, which is 3–4 orders of magnitudebelow the observed values.The results produced by the model M10, with an efficiency for reactive desorption of10%, are presented in the bottom row of Figure 2. In this model, the abundances of alldepicted species are enhanced in comparison with model M1 and especially with modelM0. In model M10, the contribution to gas-phase H CO from the chain of surface reactionsin reaction (5) via reactive desorption is an order of magnitude larger than the gas-phaseformation routes, which leads to an increase in the peak fractional abundance of H CO to6 × − . The abundances of methoxy and methanol are increased by an order of magnitude 9 –in comparison with model M1, and reach 8.7 × − and 3.4 × − , respectively, at theirmaximum.As can be seen on the right panel in the bottom row of Figure 2, the abundances ofketene and complex organic species are increased in model M10 in comparison with modelM1 by different factors. The smallest enhancement in abundance, a factor of 7 at 10 yr,is exhibited by ketene. The order of magnitude increase in the abundance of gas-phasemethanol throughout the time range shown leads to an increase in the abundance of CH +3 via the reaction H +3 + CH OH → CH +3 + H O + H . The methyl ion can then react withCO in reaction (6) to produce C H O + , and subsequently CH CO. An order-of-magnitudeincrease in the abundance of acetaldehyde is caused by the increased abundance of C H ,which is ejected into the gas via efficient reactive desorption from the surface reaction chainin reaction (8).The abundance of dimethyl ether in model M10 is increased by two orders of magnitudein comparison with model M1 because the abundances of both gaseous reactants (CH andCH O) in reaction (4) are increased in model M10 by an order of magnitude. The increase ingaseous CH O derives from the increase in efficiency of reactive desorption in model M10. Inturn, CH in model M10 is mainly produced in the gas-phase via reaction (1). Another routefor the formation of CH OCH is via the dissociative recombination of protonated dimethylether, which is produced in the reaction of methanol with protonated methanol:CH OH +2 + CH OH → CH OHCH +3 + H O (9)The strongest abundance enhancement in switching from model M1 to model M10 occursfor methyl formate, which increases from 4 × − to 1.8 × − at 10 yr. The most efficientgas-phase formation route for HCOOCH in our model is via dissociative recombinationof H COHOCH +2 . The principal formation route of this species occurs via the radiativeassociation reaction (Horn et al. 2004)H COH + + H CO → H COHOCH +2 + h ν. (10)The product is not normal protonated methyl formate, which has the structure HC(OH)OCH +3 .Unfortunately there is a transition state barrier between the product of reaction (10) andprotonated methyl formate (Horn et al. 2004). We assume, as did Horn et al. (2004), thatdissociative recombination of the product of reaction (10) can also form methyl formate +H despite the change in structure required. Reaction (10) is similar to reaction (9) in thesense that its rate depends quadratically on the abundance of gas-phase formaldehyde. Con-sequently, the gas-phase abundance of HCOOCH in the model M10 becomes comparablealthough still somewhat lower than the observed values at times exceeding 3 × yr. If the 10 –dissociative recombination reaction does not lead to methyl formate, another process thatcan produce protonated methyl formate is the radiative association reaction (Horn et al.2004) CH +3 + HCOOH → HC(OH)OCH +3 + h ν, (11)followed by dissociative recombination. This reaction, however, is not as efficient in produc-ing methyl formate in our current models.The general behavior of models with reactive desorption is similar to that described inGarrod et al. (2007). These authors found that at ≤ · yr, the abundances of the majorityof gas-phase species are not affected significantly by the inclusion of reactive desorption.The exceptions are several hydrogenated species such as methanol and formaldehyde. Inthis study, we observe a similar picture extended by the enhanced abundances of COMs,descendants of H CO and CH OH, which were not considered in Garrod et al. (2007).
It is interesting to investigate how well our models reproduce the observed abundancesof complex organic species and their precursors. To quantify the level of agreement, wecalculated individual confidence parameters κ i for each i -th species versus time in modelsM1 and M10 according to the prescription in Garrod et al. (2007) and then took the meanof the individual confidence parameters as the criterion of overall agreement. The formulafor each individual confidence parameter is κ i = erf c (cid:18) | log ( X i ) − log ( X obsi ) |√ σ (cid:19) , (12)where erf c is the complementary error function, X i and X obsi are the modeled and observedabundances of species i at time t , and the standard deviation σ is assumed to equal unity,which means that we estimate the uncertainty in observed abundances to be one orderof magnitude. With this definition, κ i is equal to unity when the observed and modelabundances are equal, and quickly goes towards a limit of zero with increasing discrepancybetween these two values. For example, when the difference is one order of magnitude, κ i =0.32, and when the difference is two orders of magnitude, κ i = 0.046. Model M0 is excludedfrom this comparison as its results are obviously too far from observational values.The molecules shown in Table 2 are used in the analysis. In addition to these molecules,several other simple species were observed towards B1-b and L1689b such as CO, HCO + ,N H + , NH , CS and HNCO/HCNO (Hirano et al. 1999; Bacmann et al. 2002; Marcelino et al.2009; ¨Oberg et al. 2010). We did not include them in the comparison set for two reasons. 11 –First, the abundances of these species come from different sets of observations, and theirinclusion will make the set heterogeneous. Secondly, the abundances of these species areonly slightly different in our models with and without reactive desorption. The resultingindividual and mean confidence parameters for L1689b and B1-b are shown in Figures 3 and4, respectively. For L1689b, model M1 reaches its best agreement at 1.3 × yr and modelM10 at 5.1 × yr (if we exclude for M10 a small cusp at much earlier time that has a similarextent of agreement). In the case of B1-b, the time of best agreement time for model M1 is1 . × yr, while for model M10 it is 2.6 × yr. At these times, the average agreementis better than one order of magnitude. The individual molecular abundances at these besttimes are shown in Table 2. Because there are two more organic species detected in B1-bthan in L1689b, we choose B1-b for detailed discussion.At the time of best agreement for M1, the modeled abundances for the two precur-sor species CH CO and H CO, and the two complex species CH OH and CH CHO, outof seven species, differ from the observational values by less than one order of magnitude.For one species, CH OCH , the difference between modeled and observed abundances justslightly exceeds one order of magnitude. At the same time, model M1 somewhat overpro-duces methoxy and significantly underproduces methyl formate. Model M10 exhibits its bestagreement with observations of B1-b significantly earlier. While it manages to reproduce theabundances of all complex organic species including HCOOCH , the modeled abundance ofH CO is two orders of magnitude higher than the observed value. The abundance of CH Ois also significantly higher than the observed value.A possible reason for the overly high calculated abundance of methoxy in both the M1and M10 models has been given by Cernicharo et al. (2012), who showed that that on grainsthe isomer CH OH is produced more efficiently than CH O and released into the gas phase.This isomer has different spectral properties, and was not observed in the gas phase. Theauthors proposed an alternative formation route for methoxy via the gas-phase reactionCH OH + OH → CH O + H O . (13)Although our chemical model currently does not discriminate between CH O and CH OH,we estimate that if methoxy is only produced by reaction (13) at its newly determinedrate coefficient, then the abundance of methoxy will be in much closer agreement withobservations, at least in case of B1-b. But, one must remember that since methoxy is aprecursor species to complex organic molecules, lessening its abundance will also lessen theabundance of the complex molecules formed from it, especially dimethyl ether.We show the organic abundances vs time for an altered model M10, in which the gas-phase route to methoxy is included and the reactive desorption route is removed, in Figure 5.The results can be compared with the abundances obtained with the original model M10, 12 –which are shown in the bottom panel of Figure 2. The time of the best agreement betweenobserved values for B1-b and results of the altered model M10 is the same as for the originalmodel, but at later times, the level of agreement remains at almost the same value until 10 yr.For B1-b, the new fractional abundance of CH O at the time of best agreement is 5 × − ,which is closer to the observed value, although still too large by an order of magnitude. Thisfact may imply that the rate coefficient of reaction (13) is somewhat overestimated, which issupported by older estimations. With the high value for the rate coefficient of reaction (13),the new abundance of CH OCH at the moment of best agreement for B1-b is 1 × − ,but it increases steadily, reaching a value of 2 × − at 10 yr.Finally, it is interesting to note that in B1-b the observed abundances of all organicspecies we consider are lower than in L1689b, including formaldehyde. If future observationsconfirm that this correlation is not caused by systematic errors between observational datasets, the lower abundances in B1-b argue in favor of our scenario for the formation of complexmolecules in cold clouds, because our scenario implies a strong correlation between theabundances of H CO and COMs.
4. Discussion
We have considered one approach to understanding how terrestrial-type organic moleculescan be synthesized in cold starless and prestellar cores. In our approach, reactive desorptionsucceeds in removing precursor molecules efficiently from the ice mantles of cold dust grainsinto the gas, where subsequent reactions produce the organic molecules in reasonable, ifnot perfect, agreement with the small observed abundances detected in the cold prestellarcores L1689b and B1-b. Radiative association reactions in the gas play an important rolein the synthesis of the complex organic molecules methyl formate and dimethyl ether. Bothreactive desorption and radiative association reactions are still poorly understood processesin general. Moreover, the product for the proposed radiative association reaction that leadsto the formation of protonated methyl formate most likely yields a different structure thandesired, which may or may not form methyl formate in its lowest conformer when the ionundergoes a dissociative recombination with electrons.Although we have discussed our model results mainly for observed COMs, our model canalso be tested by its predictions for the abundances of other complex species, such as ethylcyanide (C H CN), which is a well-known weed in hot cores. Under cold core conditions, ourmodel first produces ethyl cyanide in the ice mantle by hydrogenation of the CCCN radical(Hasegawa et al. 1992; Caselli et al. 1993). Reactive desorption then results in a considerableabundance of this species in the gas phase at selected times. For example, in Model M1, 13 –the fractional abundance of ethyl cyanide peaks at over 10 − at a time near 2 × yr,while for Model M10, the peak value approaches a fractional abundance of 10 − at the sametime. At the times of best agreement for models M1 and M10 with B1-b, the calculatedfractional abundances of ethyl cyanide are somewhat under 10 − and somewhat over 10 − respectively, while for L1689b, these numbers are somewhat over 10 − and near 10 − .There are perhaps other complementary approaches to the production of COMs incold cores, where the dust surfaces are typically too cold for the surface radical-radicalphotochemistry suggested by Garrod & Herbst (2006). One possibility is that we are missinga large number of thermal surface/ice reactions that can occur at 10 K, typically by thediffusion of atoms somewhat heavier than hydrogen. A figure showing the many species thatmight be synthesized in this way was produced by S. Charnley (2009, private communication)and can be found in Herbst & van Dishoeck (2009). Another possibility is that the dustparticles are not always at 10 K, and that a variety of effects cause a temporary increase intemperature high enough to allow surface radicals to diffuse and react with one another. It iseven possible that photodissociation or particle bombardment processes can cause temporarynon-thermal motions of species so as to promote the rates of non-thermal diffusive reactions(Occhiogrosso et al. 2011; Kim & Kaiser 2012). Finally, tunneling from one lattice site toanother for species other than atomic hydrogen will promote more rapid surface/ice reactionsat lower temperatures than if the only diffusive mechanism for species heaver than H is simplethermal hopping. We hope to investigate some of these possibilities in the near future.
5. Summary
In this study, we have proposed and examined a scenario of formation of complex organicmolecules recently found in the cold cores L1689b and B1-b with gas temperatures of 10 K.The proposed scenario is based on the assumption that complex organic species in the coldregions are formed via gas-phase ion-molecular and neutral-neutral chemistry from simplerprecursors such as formaldehyde and methanol. These precursors are formed on icy surfacesof interstellar grains, and then ejected into the gas via reactive desorption. Because of ourlack of knowledge of the efficiency of reactive desorption, we considered three models: M0,M1, and M10, in which the efficiency of reactive desorption per exothermic surface reactionis 0%, 1%, and 10%, respectively.We found that the proposed scenario gives somewhat mixed agreement with observationsof COMs and their precursors. Both models with non-zero reactive desorption exhibit muchbetter agreement with observations than the model without reactive desorption. The bestagreement for the whole set of species is exhibited by model M10, which possesses the 14 –highest efficiency of reactive desorption. On the other hand, this model tends to overestimategas-phase abundances of formaldehyde in comparison to observations by ∼ two orders ofmagnitude. Such a discrepancy can be partly but not fully explained by uncertainties inour chemical model, particularly in the rate coefficients of chemical reactions. This factmay imply that the proposed scenario of formation of complex organic species via gas-phasereactions driven by reactive desorption is only partially responsible for the formation ofthese molecules, and should be assisted by other routes such as processes that warm upgrains temporarily or that permit non-thermal ice reactive processes.The authors wish to thank the anonymous referee for valuable comments, which helpedus to improve the manuscript. E. H. acknowledges the support of the National ScienceFoundation for his astrochemistry program, and support from the NASA Exobiology andEvolutionary Biology program through a subcontract from Rensselaer Polytechnic Institute.This research has made use of NASA’s Astrophysics Data System. REFERENCES
Anicich, V., Aeronautics, U. S. N., Administration, S., & (U.S.), J. P. L. 2003, JPL Pub-lication, Vol. 03-19, An Index of the Literature for Bimolecular Gas Phase Cation-Molecule Reaction Kinetics (Jet Propulsion Laboratory, National Aeronautics andSpace Administration, Pasadena, CA)Bacmann, A., Lefloch, B., Ceccarelli, C., Castets, A., Steinacker, J., & Loinard, L. 2002,A&A, 389, L6Bacmann, A., Taquet, V., Faure, A., Kahane, C., & Ceccarelli, C. 2012, A&A, 541, L12Brown, P. D., Charnley, S. B., & Millar, T. J. 1988, MNRAS, 231, 409Caselli, P., Hasegawa, T. I., & Herbst, E. 1993, ApJ, 408, 548Cernicharo, J., Marcelino, N., Roueff, E., Gerin, M., Jim´enez-Escobar, A., & Mu˜noz Caro,G. M. 2012, ApJ, 759, L43Charnley, S. B., Tielens, A. G. G. M., & Millar, T. J. 1992, ApJ, 399, L71Charnley, S. B., Tielens, A. G. G. M., & Rodgers, S. D. 1997, ApJ, 482, L203Fayolle, E. C., Bertin, M., Romanzin, C., Michaut, X., ¨Oberg, K. I., Linnartz, H., & Fillion,J.-H. 2011, ApJ, 739, L36 15 –Garrod, R. T. & Herbst, E. 2006, A&A, 457, 927Garrod, R. T. & Pauly, T. 2011, ApJ, 735, 15Garrod, R. T., Wakelam, V., & Herbst, E. 2007, A&A, 467, 1103Garrod, R. T., Weaver, S. L. W., & Herbst, E. 2008, ApJ, 682, 283Gerlich, D. & Horning, S. 1992, Chemical Reviews, 92, 1509Hasegawa, T. I. & Herbst, E. 1993, MNRAS, 261, 83Hasegawa, T. I., Herbst, E., & Leung, C. M. 1992, ApJS, 82, 167Herbst, E. 1980, ApJ, 241, 197—. 1985, ApJ, 291, 226Herbst, E. & van Dishoeck, E. F. 2009, ARA&A, 47, 427Hirano, N., Kamazaki, T., Mikami, H., Ohashi, N., & Umemoto, T. 1999, in Star Formation1999, ed. T. Nakamoto (Nobeyama Radio Observatory), 181–182Horn, A., Møllendal, H., Sekiguchi, O., Uggerud, E., Roberts, H., Herbst, E., Viggiano,A. A., & Fridgen, T. D. 2004, ApJ, 611, 605Karpas, Z. & Meot-Ner, M. 1989, The Journal of Physical Chemistry, 93, 1859Katz, N., Furman, I., Biham, O., Pirronello, V., & Vidali, G. 1999, ApJ, 522, 305Kim, Y. S. & Kaiser, R. I. 2012, ApJ, 758, 37Marcelino, N., Cernicharo, J., Roueff, E., Gerin, M., & Mauersberger, R. 2005, ApJ, 620,308Marcelino, N., Cernicharo, J., Tercero, B., & Roueff, E. 2009, ApJ, 690, L27¨Oberg, K. I., Boogert, A. C. A., Pontoppidan, K. M., van den Broek, S., van Dishoeck, E. F.,Bottinelli, S., Blake, G. A., & Evans, II, N. J. 2011, ApJ, 740, 109¨Oberg, K. I., Bottinelli, S., Jørgensen, J. K., & van Dishoeck, E. F. 2010, ApJ, 716, 825Occhiogrosso, A., Viti, S., Modica, P., & Palumbo, M. E. 2011, MNRAS, 418, 1923Vasyunin, A. I. & Herbst, E. 2013, ApJ, 762, 86 16 –Vasyunin, A. I., Semenov, D., Henning, T., Wakelam, V., Herbst, E., & Sobolev, A. M. 2008,ApJ, 672, 629Vasyunin, A. I., Sobolev, A. M., Wiebe, D. S., & Semenov, D. A. 2004, Astronomy Letters,30, 566Wakelam, V. & Herbst, E. 2008, ApJ, 680, 371Wakelam, V., Herbst, E., Le Bourlot, J., Hersant, F., Selsis, F., & Guilloteau, S. 2010, A&A,517, A21Watanabe, N. & Kouchi, A. 2002, ApJ, 571, L173Woon, D. E. & Herbst, E. 2009, ApJS, 185, 273
This preprint was prepared with the AAS L A TEX macros v5.2.
Table 1: New and Other Important Gas-phase Chemical ReactionsNew Reaction α β γ
Temperature range (K) ReferenceCH O + H → CH + OH 1.6(-10) 0.0 0.0 300—2000 NISTCH O + O → H CO + OH 1.0(-10) 0.0 0.0 300—2500 NISTCH O + CH → H CO + CH O + CH → CH OCH + h ν α β γ Temperature range (K) ReferenceCH +3 + CO → C H O + + h ν H → CH CO + H 1.6(-10) 0.0 0.0 10—2500 NISTO + C H → CH CHO + H 1.33(-10) 0.0 0.0 10—2500 NISTH +3 + CH OH → CH +3 + H O + H OH +2 + CH OH → CH OHCH +3 + H O 1.0(-10) -1.0 0.0 100—300 Anicich et al. (2003)Karpas & Meot-Ner (1989)H COH + + H CO → H COHOCH +2 + h ν +3 + HCOOH → HC(OH)OCH +3 + h ν OH + OH → CH O + H O 4.0(-11) 0.0 0.0 10—100 Cernicharo et al. (2012)Notes: a(-b) stands for a × − b . α , β and γ are the coefficients in the modified Arrhenius expression forthe reaction rate coefficient: k = α · ( T / β · exp ( − γ/T ). URL for NIST Chemical Kinetics Database:http://kinetics.nist.gov/kinetics/index.jsp 18 –Table 2: Observed and best-fit modeled fractional abundances of organic species detected incold interstellar cloudsM1 M10 ObservationsSpecies L1689b B1-b L1689b B1-b L1689b B1-bHCOOCH OCH CHO 3.2(-11) 4.7(-12) 6.4(-11) 3.7(-11) 1.7(-10) B 1.0(-11) CCH CO 4.4(-11) 9.0(-12) 8.3(-11) 3.2(-11) 2.0(-10) B 1.3(-11) CCH O 5.3(-14) 8.5(-11) 8.5(-10) 1.5(-10) — 4.7(-12) CH CO 9.8(-10) 2.9(-09) 5.4(-08) 4.8(-08) 1.3(-09) B 4.0(-10) MCH OH 5.7(-12) 2.5(-09) 2.3(-08) 3.3(-09) - 3.1(-09) ONotes: a(-b) stands for a × − b . B refers to Bacmann et al. (2012), C refers toCernicharo et al. (2012), M refers to Marcelino et al. (2005), and O refers to ¨Oberg et al.(2010). Times of best fits: Model M1: 1.3 × yr with L1689b, 1.0 × yr with B1-b,Model M10: 5.1 × yr with L1689b, 2.6 × yr with B1-b. 19 – t, years10 -8 -7 -6 -5 -4 -3 -2 n ( X ) / n H grH OgrCOgrCO grH COgrCH OHgrCH Fig. 1.— Abundances vs. time of water and major carbon-bearing ice compounds in themodel M10. The prefix “gr” denotes species on and in ice mantles of interstellar grains. 20 – t, years10 -14 -12 -10 -8 -6 n ( X ) / n H H COCH OHCH O t, years10 -14 -12 -10 -8 -6 HCOOCH CH COCH OCH CH CHO t, years10 -14 -12 -10 -8 -6 n ( X ) / n H H COCH OHCH O t, years10 -14 -12 -10 -8 -6 HCOOCH CH COCH OCH CH CHO t, years10 -14 -12 -10 -8 -6 n ( X ) / n H H COCH OHCH O t, years10 -14 -12 -10 -8 -6 HCOOCH CH COCH OCH CH CHO