A new study of the chemical structure of the Horsehead nebula: the influence of grain-surface chemistry
Romane Le Gal, Eric Herbst, Gwenaelle Dufour, Pierre Gratier, Maxime Ruaud, Thomas Vidal, Valentine Wakelam
AAstronomy & Astrophysics manuscript no. Horsehead-chemical-study_2columns_to_publish c (cid:13)
ESO 2017July 5, 2017
A new study of the chemical structure of the Horsehead nebula:the influence of grain-surface chemistry
R. Le Gal , E. Herbst , G. Dufour , , P. Gratier , M. Ruaud , , T. H. G. Vidal , V. Wakelam Departments of Chemistry and Astronomy, University of Virginia, McCormick Road, Charlottesville, VA 22904, USA, e-mail: [email protected] Laboratoire d’Astrophysique de Bordeaux, Univ. Bordeaux, CNRS, B18N, allee Geo ff roy Saint-Hilaire, 33615 Pessac, France NASA Goddard Space Flight Center (GSFC), Greenbelt, MD, USA NASA Ames Research Center, Mo ff ett Field, CA, USAReceived 12 April 2017; Accepted 29 May 2017 ABSTRACT
A wide variety of molecules have recently been detected in the Horsehead nebula photodissociation region (PDR) suggesting that: (i) gas-phase and grain chemistries should both contribute to the formation of organic molecules, and (ii) far-ultraviolet (FUV)photodesorption may explain the release into the gas phase of grain surface species. In order to tackle these specific problems and moregenerally in order to better constrain the chemical structure of these types of environments we present a study of the Horsehead nebulagas-grain chemistry. To do so we used the 1D astrochemical gas-grain code N autilus with an appropriate physical structure computedwith the M eudon
PDR C ode and compared our modeled outcomes with published observations and with previously modeled resultswhen available. The use of a large set of chemical reactions coupled with the time-dependent code N autilus allows us to reproducemost of the observations well, including those of the first detections in a PDR of the organic molecules HCOOH, CH CO, CH CHOand CH CCH, which are mostly associated with hot cores. We also provide some abundance predictions for other molecules ofinterest. Understanding the chemistry behind the detection of these organic molecules is crucial to better constrain the environmentsthese molecules can probe.
Key words.
Astrochemistry – ISM: abundances – ISM: clouds – ISM: molecules – ISM: individual objects: Horsehead – Submil-limeter: ISM
1. Introduction
The spectral line survey WHISPER ( Wideband High-resolutionIram-30m Survey at two Positions with Emir Receivers , PI: J.Pety) recently enabled the detection of about thirty molecules upto seven atoms in size (plus their isotopologues) in the Horse-head nebula, a photodissociation region (PDR) situated at thewest extremity of the L1630 molecular cloud and illuminatedby the σ -Orionis star (O9.5V) (Gerin et al. 2009). At a dis-tance d ≈
400 pc (Anthony-Twarog 1982), this nearby PDR,seen “edge-on”, constitutes one of the brightest filaments in themid-infrared detected in our galaxy (Abergel et al. 2003).Given its geometry and intensity, the Horsehead nebula rep-resents an ideal opportunity to study the physical and chemicalstructure of a PDR. For this purpose, this region has been stud-ied deeply since 2001, in particular with the IRAM-30 meterradiotelescope and the Plateau de Bure interferometer (PdBI).Three di ff erent positions have been observed: (i) “the PDR po-sition”, corresponding to the peak of the HCO line emission,typical of the warmer UV-illuminated gas situated at the topedge of the nebula with a visual extinction of ∼ (ii) “the Core”, a cold condensation shielded fromthe UV field, corresponding to the DCO + line emission, locatedjust after the PDR edge with a much higher visual extinction( A V ∼ −
20 mag) (Pety et al. 2007); and (iii) “the PAH po-sition”, observed with complementary observations performedwith the PdBI (Guzman et al. 2015), closer to the edge of the http: // / ∼ horsehead / Horsehead_Nebula / WHISPER.html
PDR ( A V ∼ .
05 mag), corresponding to the peak at 7.7 µ m ofpolycyclic aromatic hydrocarbon emission (Abergel et al. 2003).These three positions correspond to distinguishable regionsof the nebula characterized by their di ff erent physical conditions( i.e. temperature, density, UV penetration) which vary from oneregion to the other as function of A V . It is therefore of utmostimportance to have some constraints on the physical conditionsprevailing in each region since they strongly a ff ect the chem-istry. The density in the nebula has been determined by Habartet al. (2005) to vary from ∼ cm − in the UV-illuminatedouter edges of the PDR and reaching ∼ × cm − in less than10 (0.02 pc) toward the denser region. These authors also pro-vided a temperature profile as function of the visual extinctionvia thermal balance modeling. Afterwards, Pety et al. (2007)constrained the gas temperature in the Core region, where thedensity is ∼ cm − , to be ∼
20 K by the deuterium fractiona-tion ratio DCO + / HCO + . This was confirmed by following stud-ies of the Horsehead nebula, including the WHISPER survey,which enabled the determination of average densities and tem-peratures to be respectively n H = × cm − and T gas =
60 Kfor the PDR position, and n H = cm − and T gas =
25 K forthe Core (Gerin et al. 2009; Guzman et al. 2013; Gratier et al.2013). The dust grain temperature profile has been found to stayrelatively low, from ∼
12 K in the Core to almost 30 K at theouter layers of the PDR (Goicoechea et al. 2009a). This low dusttemperature profile can be explained as a result of the combina-tion of the moderately low UV-photon radiation field of 60 × thatof the ISRF (Mathis et al. 1983; Habart et al. 2005; Rimmer Article number, page 1 of 18 a r X i v : . [ a s t r o - ph . GA ] J u l & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish et al. 2012) impinging the nebula and the high densities prevail-ing ( ∼ − cm − ).Altogether, the approximately thirty molecules recently de-tected have confirmed the chemical complexity in the neb-ula, and generated our interest for the present study. Of thesespecies, we will focus our attention on the seventeen specieslisted by molecular families in Table 1. As can be seen in thistable, the WHISPER survey allowed the detection of some or-ganic molecules in the Horsehead nebula, such as formaldehyde(H CO) and methanol (CH OH), which constitute key speciesin the likely synthesis of more complex organic molecules suchas some prebiotic molecules (Bernstein et al. 2002; Muñoz Caroet al. 2002; Garrod et al. 2008). Because they are detected in awide variety of interstellar sources - in hot cores (Sutton et al.1995; Ceccarelli et al. 2000), dark clouds (Bergman et al. 2011),shocked regions ( e.g.
Sakai et al. 2012; Codella et al. 2012;Tafalla et al. 2010) and even in comets (Mumma & Charnley2011; Cordiner et al. 2015) - it is of prime importance to under-stand well how these precursor molecules form. H CO is com-monly thought to form both in the gas-phase and on grain sur-faces, while CH OH is believed to be only formed on grain sur-faces (Garrod et al. 2006; Geppert et al. 2006). Guzman et al.(2013) reported the observations of these two molecules towardthe Horsehead nebula in both the PDR and Core positions. Un-able to reproduce the observed abundances of either H CO orCH OH at the PDR position with only pure gas-phase models,they concluded that, for this region, both species are formed ongrain surfaces and then photodesorbed into the gas phase. Onthe other hand, at the Core position, a pure gas-phase model canreproduce the observed H CO abundance, while photodesorp-tion of ices is still needed to explain the observed abundance ofCH OH. Other organic molecules were reported in the Horse-head nebula as first detections in a PDR environment, includingHCOOH (formic acid), CH2CO (ketene), CH3CHO (acetalde-hyde), and CH3CCH (propyne) (Guzman et al. 2014). Theirabundances were found to be higher at the PDR position thanat the Core, revealing that complex organic chemistry is also oc-curring in UV-illuminated neutral gas (Guzman et al. 2014). Ofthese molecules, some - HCOOH, CH CO, and CH CHO - havenow also been detected in the Orion bar PDR (Cuadrado et al.2016, 2017).Nitriles have also been observed in the WHISPER survey,such as acetonitrile (CH CN), cyanoacetylene (HC N), and cya-noethynyl (C N). These species are indeed commonly observedin the interstellar medium, including star formation regions (Bot-tinelli et al. 2004; Araya et al. 2005; Purcell et al. 2006) and pro-toplanetary disks (Öberg et al. 2015). The Horsehead CH CNspectral lines have been found to be ∼
40 times brighter at thePDR position than at the denser Core one (Gratier et al. 2013),suggesting that a surface desorption process, such as photodes-orption, should be e ffi cient enough to release organic moleculesinto the gas phase in far-UV illuminated regions. On the con-trary, for HC N, Gratier et al. (2013) have found more intensivelines in the dense Core than in the PDR region.Simple hydrocarbons such as CCH, l − C H, c − C H,c − C H , and l − C H have also been observed in the WHIS-PER survey, including the first detection in the ISM of the cationl − C H + (Pety et al. 2012). Pety et al. (2012) pointed out that thehigh abundances found in the Horsehead nebula for these hydro-carbons cannot be reproduced by current pure gas-phase models(Fuente et al. 2003; Teyssier et al. 2004), and suggested a newmechanism of “photo-erosion” by UV radiation of the PAHs toproduce them (Pety et al. 2005). Rotational lines of the carbon fluorine cation (CF + ) havebeen detected with a high signal-to-noise ratio toward both thePDR and the Core positions of the Horsehead (Guzman et al.2012). Synthesized from the reaction between hydrogen fluo-rine (HF) and C + (Neufeld et al. 2005), this cation representsa unique probe, so far, for the fluorine elemental abundance, as-suming that CF + is mainly formed from the reaction HF + C + andthat most of the fluorine is locked in the form of HF. The modelsof Guzman et al. (2012) predicted that CF + contains 4-8% of allthe fluorine, thus being the second reservoir of fluorine after HF.CF + , observed where C + is abundant, is also interesting becauseit can be observed from the ground, while submillimeter satel-lites (such as Herschel ) are required to observe C + . Then CF + could also probe C + , itself probing the cooling gas with its finestructure lines at 157.8 µ m.With the present study, we aim to understand the chemicalcontent and distribution of the Horsehead nebula according tothese recent published observations from the WHISPER surveywith a pseudo-time dependent astrochemical model. As far as weknow, our model is the first to consider a time-dependent studyof the gas-grain chemistry for such environments.In Sect. 2, we present the astrochemical model developedfor this study. In Sect. 3, we compare the time-dependent modelresults with the observations and analyze for each group ofmolecules their simulated spatial dependence and the chemicaland physical processes responsible for the physical conditionsat which our model best reproduces them. Section 4 contains adiscussion of our results as well as a comparison with previousstudies of the Horsehead nebula chemistry. We also present somepredictions from our new model regarding ice abundances andother gas-phase molecular abundances expected in the Horse-head nebula. Finally we summarize our results and draw ourmain conclusions in Sect. 5.
2. A model of the physical and chemical structureof the Horsehead nebula
We modeled the physical and chemical structures of the Horse-head nebula by using a 1D astrochemical model. To do so, wefirst ran the M eudon
PDR C ode (Le Bourlot et al. 1993; Le Petitet al. 2006; Le Bourlot et al. 2012) to obtain an appropriate phys-ical structure. Then we used this structure as an input parameterin the 1D pseudo-time-dependent N autilus code (Ruaud et al.2016) to compute the chemical time evolution.
The 1D astrochemical M eudon
PDR C ode is based on a station-ary plane-parallel geometry of gas and dust illuminated by anultraviolet (UV) radiation field coming from one or both sides ofthe cloud. From starting physical and chemical parameters, thecode resolves at each point of the cloud the UV radiative trans-fer and computes the thermal balance at steady state. We usedthe M eudon
PDR C ode to compute the temperature and den-sity profiles of the physical structure of the Horsehead nebulaby assuming a cloud illuminated on one side with a moderateUV-photon flux of 60 × that of the ISRF (Mathis et al. 1983;Habart et al. 2005; Rimmer et al. 2012) and a fixed pressureof 4 × K cm − (Abergel et al. 2003) for the PDR part onlysince for the dense part we fixed the density to 2 × cm − .The results are shown in Fig. 1, which displays as functions ofthe visual extinction from right to left the distance from the UVsource and the density profile in the first panel, the molecular Article number, page 2 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula hydrogen fraction f (H ) profile computed by N autilus and thechemical portion of the M eudon PDR C ode in the second panel,and the grain- and gas-temperature profiles in the third panel.The range of dust temperature given by the M eudon
PDR C ode for di ff erent grain size values in the range 0.03 to 0.3 µ m is indi-cated on the figure. From the third panel in Fig. 1, it can be seenthat the temperature of the gas is lower, by a few Kelvin, thanthe temperature of the dust grains for A V & . A V ∼ . ∼
12 K, and for A V . . ∼
40 K at A V ∼ ∼
130 K at A V ∼ . ∼
18 K at A V ∼ . Once an appropriate physical structure for the Horsehead nebulawas obtained with the M eudon
PDR C ode , we implemented itin the 1D N autilus code to compute the time-dependent chem-istry. N autilus is a pseudo-time dependent code based on thetwo-phase gas-grain model of Hasegawa et al. (1992). First de-veloped to model dense cold gas chemistry, it has recently beenextended to a three-phase model by Ruaud et al. (2016), in whichthe surface and bulk mantle ice phases of a grain are distin-guished. For this study we used this most recent version of N au - tilus . From an input chemical network, the code builds a sys-tem of kinetic rate equations and solves them to obtain the abun-dances of the species as functions of time. The kinetic equationsthat determine the evolution of the density of each species ineach of the phases are shown in Ruaud et al. (2016). Fig. 1.
Physical structure of the Horsehead nebula with the density fixedfor the dense part of the PDR. All panels are represented as functionsof the visual extinction, A V . The first panel shows the density and posi-tion profiles, while the second panel represents the molecular hydrogenfraction f (H ) computed by both codes, with solid line indicating theN autilus code and dashed line the M eudon PDR C ode . The third panelrepresents the grain- and gas-temperature profiles. The green area cov-ers the range of dust temperature given by the M eudon
PDR C ode fordi ff erent grain size values in the range 0.03 to 0.3 µ m. In addition to the standard elements of a gas-grain code, the lat-est version of N autilus takes into account swapping processesbetween molecules in ice mantle surfaces and bulk ice mantles(Ruaud et al. 2016). The desorption processes are both thermal,which are ine ffi cient at cold dense cloud temperatures for all butthe lightest species, and non-thermal, which includes desorptioninduced by cosmic-rays (Hasegawa & Herbst 1993), photodes-orption, and chemical desorption (Garrod et al. 2007), due tothe exothermicity of reactions whith a single product occurringat the surface of the grains. In particular, we assume that, forexothermic reactions leading to one product, 1% of the prod-ucts desorb. The binding energies considered in this work canbe found in table 2 of Wakelam et al. (2017) and on the KIDAdatabase website (http: // kida.obs.u-bordeaux1.fr / species.html).Most of the grain-surface reactions taken into account pro-ceed through the di ff usive Langmuir-Hinshelwood mechanism.Even though in the current version of N autilus (Ruaud et al.2016) some reactions can proceed through the Eley-Ridealmechanism (Ruaud et al. 2015), we did not include this mecha-nism. The N autilus version used here comprises recent updates,such as the reaction-di ff usion competition process added by Ru-aud et al. (2016). This process allows reactions with chemical ac-tivation energy to have this activation energy e ff ectively reduced,and so allows faster di ff usion reactions for those with chemicalbarriers.Regarding the photodesorption processes, several experi-mental studies have been done during the last decade for somespecies of particular interest: CO (Öberg et al. 2007; MuñozCaro et al. 2010; Fayolle et al. 2011; Bertin et al. 2012, 2013),H O (Öberg et al. 2009a), N and O (Fayolle et al. 2013),N and CO (Öberg et al. 2009b) and CH OH (Öberg et al.2009a). Two types of processes depending upon whether the ex-cited electronic state is discrete or continuous seem to emerge.But even though experimental coe ffi cients (species desorbed perphoton) have been measured, the global processes are still poorlyunderstood (Ruaud et al. 2016), and moreover these processesare highly wavelength dependent (Fayolle et al. 2011; Bertinet al. 2012, 2013). Therefore we chose to not distinguish specialcases to be more consistent with respect to the unstudied species,and we followed the recommendations of Bertin et al. (2013) bysetting a constant coe ffi cient to 1 × − for all the photodesorp-tions (Ruaud et al. 2016). For this study, we used an updated version of the kida.uva.2014network (Wakelam et al. 2015) specifically including sulfurchemistry updates (Vidal et al. 2017) and recent updates con-cerning the chemistry of carbon-bearing species such as HCCO,H C O, and the C H x (Wakelam et al. 2015; Loison et al. 2016;Hickson et al. 2016; Hickson et al. 2016). We modified the ratecoe ffi cients for the H + photodissociation using the referencedvalues from the UMIST database, which are consistent with theupper value ( < × − s − ) given in van Dishoeck (1987). Thechemical network involves 1093 species, with 585 gas-phasespecies, and 254 grain mantle-surface and bulk-mantle species.These species are linked by a total of 12437 reactions including8581 gas-phase reactions, 2748 grain mantle-surface reactionsincluding physical processes such as the desorption of moleculesinto the gas phase and their accretion onto dust-grain mantle sur-faces, and 1108 bulk-mantle reactions. Article number, page 3 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Table 2.
Elemental abundances and initial species used in this work.
Species 0D model ( a ) He 1.00(-1)N 7.95(-5)O 3.02(-4)H 0.80H + + + + + + + + ( b ) Notes.
Numbers in parentheses are powers of 10. Elemental abundances of thegas and ice mantle with respect to the total proton abundance. The electron abun-dance is computed internally in the code in order to have a neutral gas. ( a ) El-emental abundances used in N autilus code from Goicoechea et al. (2006) andused in Goicoechea et al. (2009b); Pety et al. (2012); Guzman et al. (2012, 2015),except for fluorine. ( b ) Value from Snow et al. (2007), as described in Sect. 2.3.1.
First, we modeled the chemical evolution during a million yearsof a starless dense cloud with a high visual extinction. Thepseudo-time-dependent model was run for constant physicalconditions typical of starless clouds with grain and gas tempera-tures of 10 K, a gas density n H = n (H) + n (H ) = × cm − ,and shielded from the UV photons by a visual extinction of30 mag. The cosmic-ray ionization rate is set to the constantvalue of ζ = × − s − , as previously constrained for theHorsehead nebula by Goicoechea et al. (2006, 2009b), in or-der to compare our results with those already published (Petyet al. 2012; Guzman et al. 2012, 2015). A single grain radius of0.1 µ m has been adopted to compute the adsorption rates follow-ing Hasegawa et al. (1992).Assuming that the chemistry starts from di ff use cloud con-ditions, the chemical components are assumed to be initially inatomic form, except for hydrogen, assumed to be already 20%molecular as considered in previous M eudon PDR C ode simu-lations (Pety et al. 2012; Guzman et al. 2012, 2015). The ele-ments with an ionization potential lower than that of hydrogen(13.6 eV) are initially singly ionized, i.e.
C, S, Si, Fe, Na, Mg,P, and Cl. For the 0D model of a cold dense cloud we consid-ered the elemental abundances listed in Table 2, which are fromGoicoechea et al. (2006) except for fluorine, for which we useda more appropriate di ff use elemental abundance of 1 . × − (Snow et al. 2007), justified later in the text. We used these initialchemical conditions in order to better compare our results withthe previous astrochemical modeling of the Horsehead nebula(Pety et al. 2012; Guzman et al. 2012, 2015). After a period of 10 yr for a typical interstellar molecular cloudto be formed, we assume that an internal star is also formed,which results in an unshielded moderate UV-photon flux of60 × that of the ISRF, which then irradiates the portion of thecloud to produce a PDR. The chemical evolution of the “proto-PDR” starts from the output abundances of the first starlessmolecular cloud model, the 0D model. The chemistry evolves once again over a period of 10 yr for each spatial division ofthe implemented physical structure of the 1D Horsehead nebulaobtained from the M eudon PDR C ode . Hereafter our standardmodel refers to the combination of the 0D and 1D models de-scribed in this section.
3. Comparison with observations of gas-phasemolecules
In this section, we compare our standard model results with theobservations of the di ff erent species listed in Table 1. Figures 2- 5 show the modeled and observed abundances in the Horse-head nebula vs. visual extinction for the di ff erent categories ofspecies listed in Table 1. The observations are represented by thevertical error bars delimiting the uncertainties in abundance. Be-cause the uncertainties in visual extinction are not observation-ally well constrained, mainly due to the geometry of the Horse-head nebula seen edge-on, the vertical abundance error bars areplaced for clarity where our model reproduces the observations,as functions of A V , within their error bars and, in addition, withina factor of ten. The modeling results are represented by thesolid curves at so-called early and late times, both of which startfrom the beginning of the PDR phase. The late time is the same1 × yr for the organic molecules (Fig. 2), the nitrogen-bearingspecies (Fig. 3), the hydrocarbons of interest (Fig. 4), and CF + (Fig. 5). The early time is a constant 4 . × yr, except for thehydrocarbons, where an earlier time of 4 . × yr is used tobring out what can happen at very early times. In these figures,we distinguish two regions of the nebula already discussed - theCore and PDR regions, and for the hydrocarbons (Fig. 4) we alsoinclude the PAH region for values of the visual extinction under0.1 mag (Guzman et al. 2015). The grayish boxes indicate the ap-proximate A V ranges at which the observations were performedin each region, according to the corresponding published articleson WHISPER survey observations. Table 3, which contains allthe molecules listed in Table 1, gives the extinction ranges in theCore and PDR phases for which our calculated abundances liewithin the error bars of the observed values at each time repre-sented in Figs. 2 - 5. For the PDR region, we can see from Fig. 2 that the selectedtimes do not influence the model abundance results significantly.The observed abundances of H CO and CH OH can be repro-duced by the model at ∼ . − . ∼ . − . A V ∼ . − . CCHat the border between the PDR and the Core regions at A V in ∼ . − . CCO, CH CHO andHCOOH are slightly underestimated by our model by at leasta factor of approximately three at their theoretical abundancepeaks. But if we consider a standard error factor of ten on themodel abundances, these last model results are still consistentwith the observations.For the Core region, Fig. 2 shows that the agreement amongthe observed species and our calculated abundances is at its bestat 1 × yr (upper panel), with almost all species calculated tebe within their observed error bars for some range of extinction.For example, the observed HCO abundance is indeed well re-produced by our model for either A V &
13 mag or in the range ∼ . − . Article number, page 4 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula
Fig. 2.
Comparison between the observed and modeled abundances in the Horsehead nebula for HCO and the six organic molecules listed in thefirst column of Table 1, at two typical early and late times. The model results are represented by solid curves and the observations by vertical errorbars which refer to a 50% error in the abundances (Guzman et al. 2014), and horizontal lines, which are solid where our model can reproduce theobserved abundances within their error bars and dashed where the agreement is only within a standard factor of ten. The grayish boxes indicatethe approximate A V ranges at which the observations were performed in each region, according to the corresponding published articles. The ranges of extinction at which the six other species repre-sented Fig. 2 are predicted to lie within their observed error barsare shown in Table 3. At the earlier time of 4 . × yr (lowerpanel), the agreement is not as good in the sense that the abun-dances of most species are not fit within the shown error barsexcept for CH CCH and, over a very small range of extinctionsaround its peak calculated abundance, HCOOH. For the otherspecies only a “second order” agreement is obtained by consid-ering a standard factor of ten of error on the calculation.To summarize, our 0D /
1D model seems to be able toreproduce the di ff erent observed abundances for the organicmolecules of interest in the present study both in the Core and inthe PDR positions, even if at a lesser extent for the latter. This isin particular the case for the species first detected in such an envi-ronment: HCOOH, CH CO, CH CHO and CH CCH. Moreoverour model is in better agreement with the observations of theCore region at the latest time represented, 1 × yr, promotingthe requirement of a longer timescale for the denser region. In the PDR region, Fig. 3 shows that our model can reproduce theobservations of the three N-bearing species at assorted ranges ofextinction at both times depicted. It can also be noticed that theC N and HC N calculated abundances follow similar trends asfunctions of the visual extension and do not change much from4 . × yr to 1 × yr. Thus the model for these two speciesmatches the observations at both times for ranges from ∼ ∼ ∼ . ∼ CN modeledabundance also seems not to change much from 4 . × yr to1 × yr but matches the observations for higher A V , ∼ . − . A V range lies slightly higher than the value of ∼ A V = N andC N reproduce the observations, the model underestimates theCH CN abundance by about two orders of magnitude.
Article number, page 5 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Fig. 3.
Same as Fig. 2 except that the nitrile abundances are shown andthe error bars in the observations come from Gratier et al. (2013).
For the Core region, Fig. 3 shows that the modeled abun-dances of the three N-bearing species drop by about one order ofmagnitude for A V & . × yr to 1 × yr, thusgetting closer to the observational error bars at the later time.Nevertheless, the computed abundances still lie outside the ob-servational error bars, but only barely so. At earlier times, theresults are significantly worse at high extinction.One can also note from Fig. 3 that the abundance order, fromthe most abundant to the least abundant, of the three N-bearingspecies changes between the PDR and the Core regions: fromC N, HC N and CH CN in the PDR region to HC N, C N andCH CN above ∼ For the small hydrocarbons, we compare our model results withthe observed abundances reported in Guzman et al. (2015), forthe Core, PDR and PAH positions. This study took advantage offollow-up interferometric observations performed with the PdBIto further explore and resolve the PDR region layers condensedin a 5 spatially narrow region (Guzman et al. 2012). The au-thors have thus been able to better constrain the observation pro-files for the small hydrocarbons as functions of the extinction(see Fig. 3 in Guzman et al. 2015). Here, Fig. 4 represents theobserved values for the PAH postion, and for the PDR and Corepositions where the other species discussed in the present studywere also observed within the WHISPER survey. According tothe observations of Guzman et al. (2015) the best agreementfound with our model seems to be around A V − . × yr, where the calculated abundances are higher thanat 1 × yr by several orders of magnitude, the model still un-derestimates by at least two orders of magnitude the abundancesof the three species observed in this region: C H, c − C H andC H + . One plausible explanation for the discrepancies beteweenthe observations and our model results is that these small hydro-carbon molecules might be formed from fragmentation of PAHs,as suggested in Pety et al. (2005, 2012).For the PDR region, one can note that all the neutral three-carbon species considered here can be observed in the samerange of extinction, A V ∼ .
05 mag, and that C H and the C H + ion have their own specific behaviours di ff ering from one an-other for A V & . A V & . A V ∼ − H + abundance de-creases by more than three orders of magnitude in this visualextinction range at both times. The C H + presence is indeed ex-tremely localized and depends on several parameters fine-tunedtogether, including the H abundance and ortho-to-para ratio, aswell as the temperature. This can explain that despite the quan-tity of sources where C H + has been searched for (McGuire et al.2014), it has only been detected in the Horsehead nebula (Petyet al. 2012; Guzman et al. 2015) and in the Orion Bar (McGuireet al. 2014; Cuadrado et al. 2015). The temperature dependenceis caused mainly by the strong inverse temperature dependenceof the rate coe ffi cient of the C H + + H reaction (Savi´c & Ger-lich 2005). As seen in Fig. 1 the gas temperature decreases from ∼
40 to 12 K in the range A V ∼ − H + ion in this A V range.For the Core region, our model is only in agreement with theupper limit derived for the observations of C H + , independentlyof time. All the other calculated hydrocarbon abundances areoverestimated by at least an order of magnitude: a factor of ap-proximately twelve for c − C H, l − C H, and l − C H and approx-imately two orders of magnitude for c − C H at the earlier timeshown Fig. 4. We can also see that the modeled abundances ofl − C H, c − C H and l − C H increase with time, by factors fromtwo to four, while c − C H decreases by factors of between twoand five, depending on the A V considered, reflecting the chemi-cal formation and destruction pathways of these molecules dis-cussed below. Guzman et al. (2015) discussed the known gas-phase syntheticprocesses for the small hydrocarbons of interest here, up toC H + . This ion can form through di ff erent pathways such asC H + (C, H) C H + , C H (C + , H) C H + , and C H (C + , H) C followed by C + (H , H) C H + . With our present astrochemicalmodel, the third pathway dominates independently of the visualextinction until a few 10 yr and then the C H + C + pathwaydominates up to 10 yr (except for A V > . H + + C also dominates between ∼ × and ∼ × yr).Meanwhile the predominant destruction pathway of C H + is therecombination with electron independent of the time and the vi-sual extinction. C H + is supposed to be the main precursor forC H and C H via successive hydrogenations with H leading tothe ions C H + and C H + , followed by electronic dissociative re-combinations to form the neutral hydrocarbons (Maluendes et al.1993). Even if C H is related to C H + via its formation path-way, the fact that the two species do not have the same spatial Article number, page 6 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula
Fig. 4.
As Figs. 2 and 3 except that the small hydrocarbon abundances are plotted, the early time value is two orders of magnitude lower, the PAHregion is included, and the error bars in the observations are from the interferometric study of Guzman et al. (2015). distribution does not obviously mean that the gas-phase chem-istry alone is insu ffi cient to explain the observations (Guzmanet al. 2015). As a counter example, this e ff ect can arise becausethe destruction pathways di ff er from one molecule to the other,which might give information on the chemical timescale. De-pending on the A V , C H is mainly destroyed either by photons,or as the extinction become higher ( A V & . + for thecyclic form, and by H for the linear form. On the contrary, C H + is mainly destroyed by electrons independently of the extinction,which could explain the discrepancy found between the spatialdistributions of C H + and C H . Moreover, Fig. 4 indeed showsthat the chemical behaviors of the two species seem to divergeat later times, which might lead to di ff erent spatial distributions.Thus the discrepancy found between the C H + and C H spa-tial distributions (Guzman et al. 2015) could potentially indicatesome new plausible “chemical clock”. + In Fig. 5, we compare the observed and calculated abundancesof CF + in both the Core and the PDR positions at 4 . × yrand 1 × yr. Since our standard model, hereafter Model 1, isnot able to reproduce the CF + abundance well, we have run twoadditional models: Model 2, which considers a lower elementalabundance for fluorine of 6 . × − (Goicoechea et al. 2006),and Model 3, considering new values for rate coe ffi cients for thefollowing important reactions: – CF + + Photon −−→ F + C + (Guzman et al. 2012), – HF + C + −−→ H + CF + (Neufeld et al. 2005; Guzman et al.2012), – CF + + e – −−→ C + F (Novotny et al. 2005; Neufeld & Wolfire2009; Guzman et al. 2012), – F + H −−→ HF + H (Tizniti et al. 2014).Figure 5 shows that, for the PDR region, the observations arebest reproduced by Model 3 which contains the updated fluorinechemistry and a less depleted fluorine elemental abundance (seeTable 2 and Sect. 2.3.1). A comparison between the upper and
Article number, page 7 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Fig. 5.
Comparison between the observed Guzman et al. (2012) and cal-culated abundances of CF + in both the Core and the PDR positions ofthe Horsehead nebula at: 4 . × yr and 1 × yr. Three di ff erentmodels are considered here concerning the fluorine chemistry: Model1, corresponding to the one described in Sect. 2.3; Model 2, which con-tains a lower fluorine abundance (Goicoechea et al. 2006), and Model3, with improved reaction rate coe ffi cients discussed in the text. Themodel results are represented by the solid curves and the observationsby the upper limit and hatched box represented in pink. lower panels of Fig. 5 also shows that the CF + abundance is in-dependent of time for the PDR region, but, for A V > + abundance decreases with time.
4. Discussion of some calculated abundances andpredictions
Overall the calculated results obtained with our 0D /
1D model re-produce reasonably the observed abundances derived toward theHorsehead nebula and especially the time-dependent character-istic of our simulation allows us to argue in favor of late chemi-cal timescales to reproduce the Core region abundances. We dis-cuss in this section our results in the light of the recent progressmade in understanding the chemistry of the di ff erent familiesof species presented in this study. Quantitative results contain-ing abundances and comparisons with observations are found inTable 4, in which we present theoretical molecular abundancesobtained with our standard model at a timescale of 1 × yrfor three specific A V values corresponding to those derived fromthe observations of the PAH, PDR, and Core postions (Guzmanet al. 2015; Pety et al. 2012; Gerin et al. 2009). Figures 8 - 10represent the abundances of some non-observed and observedmolecules not yet discussed in this study, as functions of the vi-sual extinction for early and late times as done in Figs. 2 - 7 togive a global overview of the behavior of each species with theextinction. These abundance predictions are discussed Sect. 4.5. The organic molecules detected wih the WHISPER survey havebeen found more abundant at the PDR than at the Core posi-tion except for methanol for which the reverse is observed (Guz-man et al. 2014). Moreover, according to Guzman et al. (2014),methanol is less abundant in the Horsehead nebula than towardhot core sources where it has been mostly detected (Bisschopet al. 2007). On the other hand, other organic molecules detectedin the Horsehead nebula have similar abundances to those foundin hot cores (Bisschop et al. 2007) and prestellar cores (Bac-mann et al. 2012). This interesting result suggests that methanolis probably tracing a hotter gas where thermal desorption or evenexplosion of grain mantles (Coutens et al. 2017) can occur, whilethe other detected organic molecules ( i.e. CH CO, CH CHO,HCOOH and CH CCH) might trace colder regions. In thesecolder regions, the dominant desorption mechanism proceedsprobably mainly by chemical desorption, as with our presentmodel and the model developed by Esplugues et al. (2016), in-stead of photodesorption as previously suggested (Guzman et al.2014) due to overestimated photodesorption rates (Bertin et al.2016; Cruz-Diaz et al. 2016). To study the e ffi ciency of thephotodesorption mechanism in releasing the organic moleculesinto the gas phase with our model, we ran two similar mod-els: one without taking into account photodesorption and anotherwith a photodesorption yield of 1 × − . We obtained very sim-ilar abundances to those represented and described in Sect. 3.1.Thus, according to our model, the photodesorption is not a mainpathway to release molecules in the gas phase. In our prelimi-nary 0D model, the main process that desorbs the molecules ischemical desorption. Subsequently, in our 1D model, depend-ing on A V , the organic molecules are mainly released into thegas phase by chemical desorption for the denser parts and, forthe less extinguished regions, also by thermal desorption due tothe higher temperatures prevailing. Therefore the fact that theorganic molecule abundances, observed in the Horsehead, arehigher in the PDR region than in the Core might indicate thatgrain surface processes are sped up thanks to an enhanced mo-bility on grains caused by the warming up of their surfaces. Thiswarming could accelerate di ff usion processes (Vinogrado ff et al.2013; Mispelaer et al. 2013) and thus increase the production oforganic molecules on grain surfaces.The abundance peaks for most of the complex organicmolecules discussed in the present study are found to lie in be-tween the PDR and Core regions. This is indeed expected sinceat lower densities there is less depletion onto grain surfaces, re-sulting in fewer grain-surface reactions and thus a lowered pro-duction of organic molecules. At higher densities, while morespecies are depleted onto grains, there is less desorption occuringso there are fewer organic molecules in the gas phase (see Fig. 6).The high peak abundances might also be generated by inclusionof a competition between di ff usion and reaction over chemicalbarriers, occasionally used to analyze experimental data and con-tained in the model used (Ruaud et al. 2016). Such a competi-tive process is indeed more realistic and required to better repro-duce the gas-phase abundances for species produced by reactionswith barriers (Herbst & Millar 2008) since without it methanolis under-estimated by three orders of magnitude (Ruaud et al.2016).Moreover, even though Guzman et al. (2014) conclude thatthe WHISPER survey results seem in agreement with the factthat none of the typical “hot molecules”, such as CH OCH orHCOOCH , is detected in the Horsehead, we present here someabundance predictions for these complex organic molecules Article number, page 8 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula
Fig. 6.
As Figs. 2 - 5 but here representing the predicted gas-phase and ice abundances in the Horsehead nebula for CH O, CH OCH andHCOOCH , at the same typical early and late times: 4 . × yr and 1 × yr. The “Jspecies” designate the mantle-surface molecules and the“Kspecies” the bulk-mantle molecules. based on their recent surprising observations in cold prestel-lar core ( T ∼
10 K) (Bacmann et al. 2012). Figure 6 repre-sents the calculated abundance predictions as functions of thevisual extinction obtained with our 0D /
1D model for the com-plex organic molecules (COMs) dimethyl ether (CH OCH ) andmethyl formate (HCOOCH ), both in the gas phase and on ices,and also for the radical CH O, one of the main precursors of sat-urated COMs (Brown et al. 1988; Garrod et al. 2006; Herbst &van Dishoeck 2009). According to our model CH O should bedetectable in the gas phase with a reasonable fractional abun-dance of & − for A V ∼ − and CH OCH , their gas-phase fractionalabundances are low for all visual extinction values and only be-come higher than & − for A V ∼ − O is more abun-dant in the bulk mantle at both plotted times, and CH OCH isalso more abundant on ices, either inside the mantle or at its sur-face, while HCOOCH is more abundant in the gas-phase untilan A V of ∼ . × yr and ∼ × yr. We note that the KIDA chemical network used for the present studydoes not include yet the new gas-phase COM synthesis recentlysuggested by Balucani et al. (2015).Figure 7 shows the ice abundance predictions for the organicmolecules listed in the first column of Table 1 and for which wepresented the gas-phase abundances in Fig. 2. These predictionsare given for information only since today it might be di ffi cult toobserve the ice features in the Horeshead. However, they couldbe used in the future to indicate which ice molecules could beobservable if such observations appear to be possible, as withJWST. One fifth of the ∼
200 molecules detected up to the present be-long to the nitrile (R-CN) / isonitrile (R-NC) family. Acetoni-trile (CH CN) is a commonly observed nitrile in the ISM, and isthought to be a good temperature probe because it is a symmetricrotor presenting metastable levels (Guesten et al. 1985). The factthat CH CN was observed with a higher abundance in the PDR
Article number, page 9 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish than in the Core, as opposed to HC N, made Gratier et al. (2013)suggest that a surface desorption process, such as photodes-orption, should be occurring to release organic molecules intothe gas phase in far-UV illuminated regions. These authors de-veloped a simple steady-state gas-phase chemical model whichunderestimated the CH CN abundance in the PDR by approx-imately three orders of magnitude. Our model also underesti-mates the CH CN abundance by about two orders of magnitude,in the PDR region ( A V ∼ N and C N reproduce the observations. But ourmodel can reproduce the observed CH CN abundance for highervisual extinction suggesting that the species might not be foundat the same position. For the Core region, the agreement withthe observation is less good but still better for the later time of1 × yr, such as obtained for the organic molecules.So, our model is once again able to reproduce the observa-tions depending on the visual extinction. The high acetonitrilecan be explained by a higher photodesorption rate and a lowerice photolysis rate than those currently assumed in the models(Gratier et al. 2013; Ligterink et al. 2015). The Bernstein et al.(2004) results on a slower photolysis process for solid CH CNthan for other organic molecules argue in this sense as do thoseof Danger et al. (2011) concerning CH CN formation by UVphotolysis of ethylamine (CH CH NH ) in ices. Gratier et al.(2013) also highlighted the fact that the species has also beenobserved in shocks (Arce et al. 2008; Codella et al. 2009) indi-cating that acetonitrile might be subject to sputtering in those re-gions. Moreover, recent experimental studies focused on the spe-cific topic of high-energy irradiation bombardment of CH CNices conclude that this process could enhance the complex ni-trile abundances in the gas phase (Ribeiro et al. 2015).
The high values found for the small hydrocarbons in the Horse-head nebula PAH region might pinpoint that a “top down” chem-istry is occuring in those environments, in which small hydro-carbons form from large molecule fragmentation, such as PAHs,with a high UV flux (Fuente et al. 2003; Pety et al. 2005, 2012;Guzman et al. 2015). Nevertheless, our model is able to repro-duce the hydrocarbons observed toward the PDR region andeven overproduce these same hydrocarbons for the densest re-gion. However, we note that the observed abundances for thesmall hydrocarbons are particularly low in the Core region com-pared with typical dense cloud values such as the case of TMC-1(Agúndez & Wakelam 2013).On the other hand it is true that for the PAH region, corre-sponding to the outside edges of the PDR region, the hydrocar-bon abundances are underestimated and might require a new typeof formation pathway or might suggest that a higher ionizationrate has to be considered for the edges of the PDR. This last pointhas indeed been explored in Rimmer et al. (2012), who studiedthe influence of the cosmic-ray ionization rate and showed thathigh value of ζ – either a fixed ζ ∼ − s − or a “high range”column-dependent ionization rate ζ ( N H ) – is in better agreementwith observations for both the PDR and the Core regions. Onthe other hand, in the PAH region the hydrocarbon abundancesseem not to depend much on ζ , probably becauce in this region,the hydrocarbon chemistry is dominated by photon interactionsthan by cosmic-rays’. It should thus be worthwhile to further ex-plore the impact of the cosmic-ray ionization rate in our modelto see how it impacts the chemistry and if it helps to reproducebetter the small hydrocarbons in the PAH region, even thoughRimmer et al. (2012) found with their model that it does not. + case In Sect. 3.5 we presented the impact of updates on the CF + chem-istry. We confirmed that indeed the fluorine elemental abundanceof (0 . − . × − derived from the CF + observation in theHorsehead nebula (Guzman et al. 2012), consistent with the dif-fuse gas value (Snow et al. 2007), allows a better agreement be-tween observations and models than the previous lower value of6 . × − (Goicoechea et al. 2006). Moreover the updates con-cerning the CF + chemistry modeling also improved the agree-ment with the observations. It would be desirable if our calculated results could be com-pared with previous time-dependent gas-grain model calcula-tions. There is a previous time-dependent chemical model of theHorsehead nebula by Rimmer et al. (2012) but it is di ff erent fromours in three respects: (i) it uses a variable value for the cosmic-ray ionization rate ζ that diminishes as the cosmic rays enter thenebula from its edge, (ii) it does not contain grain-surface chem-istry, and (iii) it appears to utilize incorrect values for the vi-sual extinction in the plane of the nebula, which are significantlylarger than used here. So, to the best of our knowledge, the ma-jority of our results can only be compared with observationalabundances. However we present here the predictions given byour model for some of the molecules that Rimmer et al. (2012)presented.As highlighted by Rimmer et al. (2012) it might be worth-while to look at the HCN and HNC molecules, which accord-ing to our respective models should be present with su ffi cientamount to be observable mostly in the inner regions. As depictedin Fig. 8, our calculated HNC / HCN ratio seems to slightly de-crease with increasing visual extinction until A V ’ A V ’ / HCN ratio of ∼ . N. Ammonia should alsobe observable in reasonable amount according to our model. Wewould like also to emphasize that CN and NO might also beobservable with the particular interest of their being potentialtracers of the C / O elemental abundance ratio.Le Gal et al. (2014) indeed showed that the nitrogen chem-istry depends highly on the C / O gas-phase elemental ratio. Thisis mainly because carbon and oxygen are the most abundant el-ements after hydrogen and helium in the interstellar medium,and are found in similar quantities. Therefore the elemental C / Oratio can vary with values higher or below unity, entailing nonnegligible consequences on the interstellar chemistry. Moreover,except for the reference value of C / O ∼ .
55 derived toward theSun (Asplund et al. 2009), we still do not know precisely howthis ratio varies with environment. Astrochemical models showthat the CN / NO ratio can vary between 6 and 10 orders of mag-nitude for a C / O ranging from 0.3 to 1.5 for typical dense coldgas conditions (Le Gal et al. 2014). The CN / NO ratio can thus apriori appear as an ideal probe for the elemental C / O ratio.Regarding the oxygen-containing ions OH + , H O + andH O + , Fig. 9 shows that, independently of A V , H O + is moreabundant than H O + , which is itself more abundant than OH + at early time. At later time, these ionic abundances increasewith decreasing visual extinction and in the PAH region for A V . .
04 mag, the abundance order is completely reversed, i.e. with OH + more abundant than H O + itself more abundant than Article number, page 10 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula H O + . Overall, the ion H O + reaches its calculated peak abun-dance as a function of A V ( ∼ × − at a time of 1 × yr)at A V ∼ O + at A V ’ . A V ’ . A V ’
12 mag). So,as Rimmer et al. (2012), our model predicts that H O + mightbe observable in the inner regions of the Horsehead. Thus, inopposition to the Rimmer et al. (2012) conclusion about theseions, we predict that all the three ions might be detectable at theextreme edge of the PAH regions.Concerning the sulfur chemistry, CS, C S and HCS + havebeen observed in the Horsehead nebula (Goicoechea et al. 2006)and seem to indicate that the elemental sulfur depletion of sev-eral orders of magnitude is no longer required to explain theabundances of these molecules. Instead, depletion of a factora few with respect to the solar abundance is su ffi cient. TheRimmer et al. (2012) modeling study support this assessmentwith also better reproducing the observed sulfur-bearing speciesabundances when taking into account a higher elemental sulfurabundance. This is why we ran our models with the “high” sul-fur elemental abundance value recommended by these authorsof 3 . × − as shown in Table 2. But in order to verify that ourmodel produce sulfur-bearing molecular abundance consistentwith those of Goicoechea et al. (2006) and Rimmer et al. (2012)we also ran a “low” sulfur elemental abundance model, with aninitial elemental sulfur abundance of 8 . × − (Graedel et al.1982; Wakelam & Herbst 2008). The results of the two di ff erentmodels are compared in Fig. 10 which represents the observedand calculated sulfur-bearing species abundances for the highand low sulfur amount models. It can be seen that both the highand low sulfur elemental abundance models can reproduce theobservations and even that for the Core region the low sulfur el-emental abundance case reproduces the observations better. Thefact that our results seem di ff erent from those of Rimmer et al.(2012) is mainly because these authors directly compared theirmodel fractional abundances, relative to the total H nuclei, withthe observed values from Goicoechea et al. (2006) which are rel-ative to H . Due to this inconsistency, we chose to compare ourresults with the observed values relative to the total H nuclei, asreported in Guzman et al. (2014). A notable point in our modeling is that the temperature profilewe obtained with the M eudon
PDR C ode gives a gas temperaturelower than the grain temperature in the cloud interior, as repre-sented in Fig. 1 and mentioned in Sect. 2.1. This occurs becausein the public version of the M eudon
PDR C ode we used, thefreeze-out of species from the gas phase onto the grain surfacesis not included. Consequently, the gas and grain temperatures arenot rigorously self-consistent with the chemistry.If freeze-out were included in the M eudon
PDR C ode , itwould remove some gas coolants from the gas phase (such asCO) in the cloud interior and thus result in a higher gas temper-ature for the denser regions. In order to investigate the impactof this change in the temperature profile we ran some additionalmodels for which we changed the temperature profile by settingthe gas temperature greater or equal to the dust temperature. Theresults are not significantly a ff ected, probably because the dusttemperature is low ( ≈
12 K) in the cloud interior, so that even ifthe gas temperature reaches lower values it does not impact thechemistry significantly.
5. Summary and conclusions
In this study, we developed a gas-grain chemistry model withtime dependence to explore and understand the rich chemistryoccurring in the Horsehead nebula. The initial 0D astrochemi-cal model allows us to build a starless initial molecular cloudto represent the birth place for the star σ Ori. We used thecode N autilus to run this 0D model for a period of 10 yr, andthen assumed that σ Ori has been formed. Afterwards we rana subsequent 1D astrochemical model to take into account theimpact of the FUV-flux coming from the star, which impingesthe Horsehead nebula. The initial conditions for the 1D astro-chemical model are the output chemical abundances of the 0Dmodel run coupled to a physical structure, including the densityand temperature profiles, previously computed with the M eudon
PDR C ode . This 0D /
1D model is to the best of our knowledgethe first of its kind and allows us to investigate the chemistry andthe chemical timescale impact further. In particular the time de-pendence of our model allows us to study realistically the denserpart of a PDR where the usual steady-state PDR model will notbe appropriate (Esplugues et al. 2016). We emphasize two typi-cal early and late times and find that the late-time model betterreproduces the observations. The modeling results obtained haveshown that our gas-grain time-dependent model seems appro-priate for modeling the chemistry of such an environement andprovides some predictions for future observational campaigns,which might lead to a deeper understanding of the Horseheadnebula chemistry and elemental composition, such as for in-stance the elemental C / O ratio. In particular, our model supportsthe conclusion of Esplugues et al. (2016) that the presence inthe gas-phase of H CO and CH OH and other COMs is mainlydue to chemical desorption and not as previously thought to pho-todesorption (Guzman et al. 2014).The laboratory work on chemical reactive desorption byDulieu et al. (2013) and further explored in Minissale & Dulieu(2014), has recently been summarized in Minissale et al. (2016).In this last work, a semi-empirical theory was derived that de-scribes the dependence of the e ffi ciency of chemical reactivedesorption on the type and chemical composition of the sur-face, on the enthalpy of the surface reaction considered, and onthe binding energy of the desorbing molecule. This new formulawas used in a subsequent modeling study (Wakelam et al. 2017),which, when combined with new binding energy computations,greatly lowers the chemical desorption rate and seems to rule outthe reproduction of the cold core COM observations by grain-surface chemistry. It might thus be worthwhile to further explorethe impact of the chemical desorption probability for the denserpart of PDRs. Acknowledgements.
The authors would like to thank the anonymous refereefor valuable suggestions and comments. We would like to thank also Jean-Christophe Loison, Evelyne Roue ff and Viviana Guzman for useful discussions.EH wishes to thank the National Science Foundation for continuing to supportthe astrochemistry program at the University of Virginia. VW, MR, PG andGD thank the European Research Council (Starting Grant 3DICE, grant agree-ment 336474) for fundings. VW and PG are also grateful to the CNRS program“Physique et Chimie du Milieu Interstellaire” (PCMI) for partial funding of theirwork. References
Abergel, A., Teyssier, D., Bernard, J. P., et al. 2003, A&A, 410, 577Agúndez, M., Cernicharo, J., Guélin, M., et al. 2008, A&A, 478, L19Agúndez, M. & Wakelam, V. 2013, Chemical Reviews, 113, 8710Anthony-Twarog, B. J. 1982, Astron.J., 87, 1213Araya, E., Hofner, P., Kurtz, S., Bronfman, L., & DeDeo, S. 2005, Ap. J. Supp.,157, 279
Article number, page 11 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Arce, H. G., Santiago-García, J., Jørgensen, J. K., Tafalla, M., & Bachiller, R.2008, ApJ, 681, L21Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, Annu. Rev. Astron.Astrophys., 47, 481Bacmann, A., Taquet, V., Faure, A., Kahane, C., & Ceccarelli, C. 2012, A&A,541, L12Balucani, N., Ceccarelli, C., & Taquet, V. 2015, MNRAS, 449, L16Bergman, P., Parise, B., Liseau, R., & Larsson, B. 2011, A&A, 527, A39Bernstein, M. P., Ashbourn, S. F. M., Sandford, S. A., & Allamandola, L. J. 2004,ApJ, 601, 365Bernstein, M. P., Dworkin, J. P., Sandford, S. A., Cooper, G. W., & Allamandola,L. J. 2002, Nature, 416, 401Bertin, M., Fayolle, E. C., Romanzin, C., et al. 2012, PCCP (Incorporating Fara-day Transactions), 14, 9929Bertin, M., Fayolle, E. C., Romanzin, C., et al. 2013, ApJ, 779, 120Bertin, M., Romanzin, C., Doronin, M., et al. 2016, ApJ, 817, L12Bisschop, S. E., Jørgensen, J. K., van Dishoeck, E. F., & de Wachter, E. B. M.2007, A&A, 465, 913Bottinelli, S., Ceccarelli, C., Lefloch, B., et al. 2004, ApJ, 615, 354Brown, P. D., Charnley, S. B., & Millar, T. J. 1988, MNRAS, 231, 409Ceccarelli, C., Loinard, L., Castets, A., Tielens, A. G. G. M., & Caux, E. 2000,A&A, 357, L9Codella, C., Benedettini, M., Beltrán, M. T., et al. 2009, A&A, 507, L25Codella, C., Ceccarelli, C., Lefloch, B., et al. 2012, ApJ, 757, L9Cordiner, M. A., Charnley, S. B., Mumma, M. J., et al. 2015, Proceedings of theIAU, 11, 233Coutens, A., Rawlings, J. M. C., Viti, S., & Williams, D. A. 2017, MNRASCruz-Diaz, G. A., Martín-Doménech, R., Muñoz Caro, G. M., & Chen, Y.-J.2016, A&A, 592, A68Cuadrado, S., Goicoechea, J. R., Cernicharo, J., et al. 2017, ArXiv e-printsCuadrado, S., Goicoechea, J. R., Pilleri, P., et al. 2015, A&A, 575, A82Cuadrado, S., Goicoechea, J. R., Roncero, O., et al. 2016, A&A, 596, L1Danger, G., Bossa, J.-B., de Marcellus, P., et al. 2011, A&A, 525, A30Dulieu, F., Congiu, E., Noble, J., et al. 2013, Scientific Reports, 3, 1338Esplugues, G. B., Cazaux, S., Meijerink, R., Spaans, M., & Caselli, P. 2016,A&A, 591, A52Fayolle, E. C., Bertin, M., Romanzin, C., et al. 2011, ApJ, 739, L36Fayolle, E. C., Bertin, M., Romanzin, C., et al. 2013, A&A, 556, A122Fuente, A., Rodrıguez-Franco, A., Garcıa-Burillo, S., Martın-Pintado, J., &Black, J. H. 2003, A&A, 406, 899Garrod, R., Park, I. H., Caselli, P., & Herbst, E. 2006, Faraday Discussions, 133,51Garrod, R. T., Wakelam, V., & Herbst, E. 2007, A&A, 467, 1103Garrod, R. T., Widicus Weaver, S. L., & Herbst, E. 2008, ApJ, 682, 283Geppert, W. D., Hamberg, M., Thomas, R. D., et al. 2006, Faraday Discussions,133, 177Gerin, M., Goicoechea, J. R., Pety, J., & Hily-Blant, P. 2009, A&A, 494, 977Gerin, M., Pety, J., & Goicoechea, J. R. 2009, in Astronomical Society of thePacific Conference Series, Vol. 417, Submillimeter Astrophysics and Tech-nology: a Symposium Honoring Thomas G. Phillips, ed. D. C. Lis, J. E. Vail-lancourt, P. F. Goldsmith, T. A. Bell, N. Z. Scoville, & J. Zmuidzinas, 165Goicoechea, J. R., Compiègne, M., & Habart, E. 2009a, ApJ, 699, L165Goicoechea, J. R., Pety, J., Gerin, M., Hily-Blant, P., & Le Bourlot, J. 2009b,A&A, 498, 771Goicoechea, J. R., Pety, J., Gerin, M., et al. 2006, A&A, 456, 565Graedel, T. E., Langer, W. D., & Frerking, M. A. 1982, Ap. J. Supp., 48, 321Gratier, P., Pety, J., Guzman, V., et al. 2013, A&A, 557, A101Guesten, R., Walmsley, C. M., Ungerechts, H., & Churchwell, E. 1985, A&A,142, 381Guzman, V., Goicoechea, J. R., Pety, J., et al. 2013, A&A, 560, A73Guzman, V., Pety, J., Goicoechea, J. R., et al. 2015, ApJ, 800, L33Guzman, V., Pety, J., Gratier, P., et al. 2014, Faraday Discussions, 168, 103Guzman, V., Pety, J., Gratier, P., et al. 2012, A&A, 543, L1Habart, E., Abergel, A., Walmsley, C. M., Teyssier, D., & Pety, J. 2005, A&A,437, 177Hasegawa, T. I. & Herbst, E. 1993, MNRAS, 261, 83Hasegawa, T. I., Herbst, E., & Leung, C. M. 1992, Ap. J. Supp., 82, 167Herbst, E. & Millar, T. 2008,
The chemistry of cold interstellar cloud cores (edited by I.W.M. Smith, Imperial College Press, London), pp. 1–54Herbst, E. & van Dishoeck, E. F. 2009, Annu. Rev. Astron. Astrophys., 47, 427Hickson, K. M., Loison, J.-C., & Wakelam, V. 2016, ApJ, 659, 70Hickson, K. M., Wakelam, V., & Loison, J.-C. 2016, Molecular Astrophysics, 3,1Le Bourlot, J., Le Petit, F., Pinto, C., Roue ff , E., & Roy, F. 2012, A&A, 541, A76Le Bourlot, J., Pineau Des Forets, G., Roue ff , E., & Flower, D. R. 1993, A&A,267, 233Le Gal, R., Hily-Blant, P., Faure, A., et al. 2014, A&A, 562, A83Le Petit, F., Nehmé, C., Le Bourlot, J., & Roue ff , E. 2006, Ap. J. Supp., 164, 506Ligterink, N. F. W., Tenenbaum, E. D., & van Dishoeck, E. F. 2015, A&A, 576,A35 Loison, J.-C., Agúndez, M., Marcelino, N., et al. 2016, MNRAS, 456, 4101Maluendes, S. A., McLean, A. D., & Herbst, E. 1993, ApJ, 417, 181Mathis, J. S., Mezger, P. G., & Panagia, N. 1983, A&A, 128, 212McGuire, B. A., Carroll, P. B., Sanders, J. L., et al. 2014, MNRAS, 442, 2901Minissale, M. & Dulieu, F. 2014, J. Chem. Phys., 141, 014304Minissale, M., Dulieu, F., Cazaux, S., & Hocuk, S. 2016, A&A, 585, A24Mispelaer, F., Theulé, P., Aouididi, H., et al. 2013, A&A, 555, A13Muñoz Caro, G. M., Jiménez-Escobar, A., Martín-Gago, J. Á., et al. 2010, A&A,522, A108Mumma, M. J. & Charnley, S. B. 2011, Annu. Rev. Astron. Astrophys., 49, 471Muñoz Caro, G. M., Meierhenrich, U. J., Schutte, W. A., et al. 2002, Nature,416, 403Neufeld, D. A. & Wolfire, M. G. 2009, ApJ, 706, 1594Neufeld, D. A., Wolfire, M. G., & Schilke, P. 2005, ApJ, 628, 260Novotny, O., Mitchell, J. B. A., LeGarrec, J. L., et al. 2005, Journal of Physics BAtomic Molecular Physics, 38, 1471Öberg, K. I., Fuchs, G. W., Awad, Z., et al. 2007, ApJ, 662, L23Öberg, K. I., Guzmán, V. V., Furuya, K., et al. 2015, Nature, 520, 198Öberg, K. I., Linnartz, H., Visser, R., & van Dishoeck, E. F. 2009a, ApJ, 693,1209Öberg, K. I., van Dishoeck, E. F., & Linnartz, H. 2009b, A&A, 496, 281Pety, J., Goicoechea, J. R., Hily-Blant, P., Gerin, M., & Teyssier, D. 2007, A&A,464, L41Pety, J., Gratier, P., Guzmán, V., et al. 2012, A&A, 548, A68Pety, J., Teyssier, D., Fossé, D., et al. 2005, A&A, 435, 885Purcell, C. R., Balasubramanyam, R., Burton, M. G., et al. 2006, MNRAS, 367,553Ribeiro, F. d. A., Almeida, G. C., Garcia-Basabe, Y., et al. 2015, PCCP, 17, 27473Rimmer, P. B., Herbst, E., Morata, O., & Roue ff , E. 2012, A&A, 537, A7Ruaud, M., Loison, J. C., Hickson, K. M., et al. 2015, MNRAS, 447, 4004Ruaud, M., Wakelam, V., & Hersant, F. 2016, MNRAS, 459, 3756Sakai, N., Ceccarelli, C., Bottinelli, S., Sakai, T., & Yamamoto, S. 2012, ApJ,754, 70Savi´c, I. & Gerlich, D. 2005, PCCP (Incorporating Faraday Transactions), 7,1026Snow, T. P., Destree, J. D., & Jensen, A. G. 2007, Ap. J. Supp., 655, 285Sutton, E. C., Peng, R., Danchi, W. C., et al. 1995, Ap. J. Supp., 97, 455Tafalla, M., Santiago-García, J., Hacar, A., & Bachiller, R. 2010, A&A, 522,A91Teyssier, D., Fossé, D., Gerin, M., et al. 2004, A&A, 417, 135Tizniti, M., Le Picard, S. D., Lique, F., et al. 2014, Nat Chem, 6, 141van Dishoeck, E. F. 1987, in Astrochemistry, Vol. 120, 51–63Vidal, T. H. G., Loison, J.-C., Yassin Jaziri, A., et al. 2017, MNRAS, in pressVinogrado ff , V., Duvernay, F., Danger, G., et al. 2013, A&A, 549, A40Wakelam, V. & Herbst, E. 2008, Ap. J. Supp., 680, 371Wakelam, V., Loison, J.-C., Hickson, K. M., & Ruaud, M. 2015, MNRAS, 453,L48Wakelam, V., Loison, J.-C., Mereau, R., & Ruaud, M. 2017, Mol. Astrophys., inpress Article number, page 12 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula
Table 1.
Observed molecules considered in this work ( a ) Organic molecules and precursors Nitriles Small hydrocarbons F-bearing moleculesHCO CH CN CCH CF + H CO HC N l − C HCH OH C N c − C HHCOOH l − C H CH CO c − C H CH CHO l − C H + CH CCH
Notes. ( a ) As seen in the WHISPER survey.
Table 3.
Visual extinction ranges at which our model reproduces the observations within their error bars.
Species A V (mag)Core region ( A V & A V . . × yr 1 × yr 4 . × yr 1 × yrHCO [3 . − . & . ( ∗ ) & . ( ∗ ) − [5 . − . &
13] [0 . − . ( ∗ ) − [1 . − .
1] [0 . − . ( ∗ ) − [1 . − . CO [3 . − . ( ∗ ) & . ( ∗ ) − [5 . − . &
15] [0 . − . ( ∗ ) − [1 . − .
4] [0 . − . ( ∗ ) − [0 . − . OH & . ( ∗ ) & . ( ∗ ) − [ ∼ .
5; 6 . −
10] [0 . − . ( ∗ ) − [0 . − .
9] [0 . − . ( ∗ ) − [0 . − . CCH & . ( ∗ ) − & . & . ( ∗ ) − [3 . − .
8; 5 . − .
4] [2 . − . ( ∗ ) − [2 . − .
0] [2 . − . ( ∗ ) − [2 . − . CCO & ( ∗ ) [3 − .
3; 5 . − . ( ∗ ) [2 . − . ( ∗ ) [1 . − . ( ∗ ) CH CHO & . ( ∗ ) [3 . − .
4; 5 . − . ( ∗ ) − [4 . − .
1] [1 . − . ( ∗ ) [1 . − . ( ∗ ) HCOOH [3 . − .
8] - & . ( ∗ ) [3 . − . ( ∗ ) − [3 . − .
5; 4 . − .
9] [2 . − . ( ∗ ) [2 . − . ( ∗ ) CH CN – & . ( ∗ ) [2 . − . ( ∗ ) − [2 . − .
7] [2 . − . ( ∗ ) − [2 . − . N – [7 − ( ∗ ) [0 . − . ( ∗ ) − [1 . − .
0] [0 . − . ( ∗ ) − [1 . − . N – & . ( ∗ ) [0 . − . ( ∗ ) − [1 . − .
1] [0 . − . ( ∗ ) − [1 − . . × yr 1 × yr 4 . × yr 1 × yrC H & . ( ∗ ) [3 − . ( ∗ ) − & . ( ∗ ) [0 . − . ( ∗ ) − [0 . − .
38] [0 . − . ( ∗ ) − [0 . − .
6; 1 . − . − C H – – [0 . − .
1; 1 . − . ( ∗ ) − [0 . − .
20] [0 . − . ( ∗ ) − [0 . − − C H – – [0 . − .
1; 1 . − . ( ∗ ) − [0 . − .
20] [0 . − . ( ∗ ) − [0 . − − C H – – [0 . − .
2; 1 . − . ( ∗ ) − [0 . − .
26] [0 . − . ( ∗ ) − [0 . − . − C H – – [0 . − ( ∗ ) − [0 . − .
33] [0 . − . ( ∗ ) − [0 . − H + & & . . − . ( ∗ ) − [0 . − .
28; 1 . − .
8] [0 . − . ( ∗ ) − [0 . − .
6; 1 . − . + & & . − .
6] [0 . − . Notes : ( ∗ ) While these species are under- or over-estimated by a factor of a few in the regions of interest, if we assume an arbitrary error factor of ten on the modeledabundances, these results are still consistent with the observations.
Article number, page 13 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Table 4.
Observations and model results for fractional abundances at 10 yr PAH postion ( i.e.
IR-edge) PDR position ( i.e.
IR-peak) Core positionObs. This work Obs. This work Obs. This work A V (mag): ’ . ( a ) ’ . ( a ) ’ [8 − ( b ) – 5.0(-12) – 1.4(-9) – 2.6(-11)O – 1.0(-6) – 5.8(-17) – 1.7(-15)OH – 2.3(-9) – 1.7(-8) – 2.8(-9)H O – 9.9(-10) – 4.7(-8) – 1.2(-9)OH + – 2.0(-11) – 8.6(-15) – 3.5(-16)H O + – 2.1(-11) – 3.2(-14) – 3.1(-15)H O + – 2.6(-11) – 1.6(-12) – 3.6(-11)C – 5.9(-8) – 3.4(-6) – 5.6(-8)CO – 1.0(-9) 1.9(-7) ( c ) ( † ) – 1.3(-7)C + – 1.4(-4) – 7.6(-7) – 2.5(-10)CO + – 2.2(-12) ≤ . ( c ) + – 8.2(-12) 9.0(-10) ( c ) ( c ) + – 1.4(-11) 4.0(-12) ( c ) HCO – 2.4(-15) 8.4(-10) ( c ) < . ( c ) H CO – 3.4(-13) 2.9(-10) ( c ) ( c ) CH OH – 2.8(-13) 1.3(-10) ( c ) ( c ) HCOOH – 2.8(-15) 5.2(-11) ( c ) ( c ) CH CO – 2.0(-20) 1.5(-10) ( c ) ( c ) CH CHO – 1.7(-20) 2.4(-11) ( c ) ( c ) CH CCH – 3.5(-24) 4.4(-11) ( c ) ( c ) O – 1.1(-17) – 5.6(-12) – 2.0(-13)CH OCH – 1.4(-19) – 8.7(-17) – 1.7(-15)HCOOCH – 4.7(-27) – 2.4(-20) – 2.4(-15) C H [1 . − . ( a ) . − . ( a ) < . ( a ) c − C H – 2.7(-17) [0 . − . ( a ) . − . ( a ) l − C H – 5.7(-18) [0 . − . ( a ) . − . ( a ) c − C H (3 . ± . ( a ) . − . ( a ) . − . ( a ) l − C H – 2.1(-18) [0 . − . ( a ) . − . ( a ) l − C H + [1 . − . ( a ) . − . ( a ) < . ( a ) H 5.2(-9) ( d ) ( d ) ( d ) – 1.6(-13) – 3.5(-8) – 2.3(-7)C H – 4.2(-17) 2.2(-11) ( c ) + – 3.3(-12) – 1.7(-13) – 3.3(-15)CH + – 3.7(-11) – 1.2(-10) – 5.8(-11)C H + – 2.3(-17) – 6.2(-13) – 4.4(-11)N – 7.9(-5) – 2.5(-5) – 1.3(-8)CN – 1.4(-10) – 1.2(-7) – 1.8(-9)NO – 2.9(-12) – 3.7(-9) – 1.4(-10)NH – 1.3(-10) – 7.9(-9) – 1.2(-8)HCN – 4.3(-12) – 5.6(-9) – 3.4(-9)HNC – 4.2(-12) – 2.5(-9) – 2.7(-9) C N – 9.0(-20) 5.0(-11) ( e ) ( e ) HC N – 6.8(-21) 6.3(-12) ( c ) ( c ) N – 1.8(-20) – 4.3(-14) – 2.2(-11) CH CN – 5.5(-20) 2.5(-10) ( c ) ( c ) + – 3.5(-6) – 3.0(-6) – 3.6(-10)CS – 1.4(-14) 2.0(-9) ( c ) ( c ) + – 3.0(-15) 1.7(-11) ( c ) ( c ) CF + – 9.8(-11) 5.7(-10) ( c ) < . ( c ) Notes.
Numbers in parentheses are powers of 10. Molecules in boldface correspond to those discussed in Sect. 3. ( a ) Values from Guzman et al.(2015); ( b ) Values from Pety et al. (2012) and Gerin et al. (2009); ( c ) Observed values from Guzman et al. (2014) and references therein; ( † ) Notethat the CO modeled value is for the specific A V of 1.5 mag shown here, but our model also predicts a CO abundance between 10 − and 10 − inthe A V range 0 . − . ( d ) Observed values from Pety et al. (2005); ( e ) Observed values from Agúndez et al. (2008);Article number, page 14 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula
Fig. 7.
As Fig. 2 but here the abundances of the equivalent grain mantle-surface molecules (“Jspecies”) and grain bulk-mantle molecules(“Kspecies”) are depicted. Article number, page 15 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Fig. 8.
Same as Figs. 2 - 7 but here representing the predicted abundances in the Horsehead nebula obtained with our model for selected nitrogen-containing molecules at the same typical early and late times: 4 . × yr and 1 × yr.Article number, page 16 of 18. Le Gal et al.: A new study of the chemical structure of the Horsehead nebula Fig. 9.
As Fig. 8 but here representing the predicted abundances in the Horsehead nebula obtained with our model for selected oxygen-containingmolecules at the same typical early and late times. Article number, page 17 of 18 & A proofs: manuscript no. Horsehead-chemical-study_2columns_to_publish
Fig. 10.
Same as Figure 8 and 9 but here representing the predicted abundances for selected sulfur-bearing species for two di ff erent elementalamount of sulfur: a high elemental abundance of 3 . × − in the left panels and low elemental abundance of 8 . × −8