Complex organic molecules in strongly UV-irradiated gas

Abstract

We investigate the presence of complex organic molecules (COMs) in strongly UV-irradiated interstellar molecular gas. We have carried out a complete millimetre (mm) line survey using the IRAM 30 m telescope towards the edge of the Orion Bar photodissociation region (PDR), close to the H₂ dissociation front, a position irradiated by a very intense far-UV (FUV) radiation field. These observations have been complemented with 8.5″ resolution maps of the H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 and C¹⁸O J = 3 → 2 emission at 0.9 mm. Despite being a harsh environment, we detect more than 250 lines from COMs and related precursors: H₂CO, CH₃OH, HCO, H₂CCO, CH₃CHO, H₂CS, HCOOH, CH₃CN, CH₂NH, HNCO, ${\text{H}}_2^{13}\text{CO},$ and HC₃N (in decreasing order of abundance). For each species, the large number of detected lines allowed us to accurately constrain their rotational temperatures (T_rot) and column densities (N). Owing to subthermal excitation and intricate spectroscopy of some COMs (symmetric- and asymmetric-top molecules such as CH₃CN and H₂CO, respectively), a correct determination of N and T_rot requires building rotational population diagrams of their rotational ladders separately. The inferred column densities are in the 10¹¹ – 10¹³cm⁻² range. We also provide accurate upper limit abundances for chemically related molecules that might have been expected, but are not conclusively detected at the edge of the PDR (HDCO, CH₃O, CH₃NC, CH₃CCH, CH₃OCH₃, HCOOCH₃, CH₃CH₂OH, CH₃CH₂CN, and CH₂CHCN). A non-thermodynamic equilibrium excitation analysis for molecules with known collisional rate coefficients suggests that some COMs arise from different PDR layers but we cannot resolve them spatially. In particular, H₂CO and CH₃CN survive in the extended gas directly exposed to the strong FUV flux (T_k = 150 – 250 K and T_d ≳ 60 K), whereas CH₃OH only arises from denser and cooler gas clumps in the more shielded PDR interior (T_k = 40 – 50 K). The non-detection of HDCO towards the PDR edge is consistent with the minor role of pure gas-phase deuteration at very high temperatures. We find a HCO/H₂CO/CH₃OH ≃ 1/5/3 abundance ratio. These ratios are different from those inferred in hot cores and shocks. Taking into account the elevated gas and dust temperatures at the edge of the Bar (mostly mantle-free grains), we suggest the following scenarios for the formation of COMs: (i) hot gas-phase reactions not included in current models; (ii) warm grain-surface chemistry; or (iii) the PDR dynamics is such that COMs or precursors formed in cold icy grains deeper inside the molecular cloud desorb and advect into the PDR.

1. Introduction

Almost 200 molecules have been detected in the interstellar medium (ISM) and circumstellar shells. A very large fraction of them are complex organic molecules (COMs). COMs are traditionally defined as carbon-based molecular species with more than six atoms in their structure (Herbst & van Dishoeck 2009; Caselli & Ceccarelli 2012). To date, the largest organic molecules detected in the ISM (excluding PAHs and fullerenes) are propyl cyanide (C₃H₇CN; Belloche et al. 2009, 2014) and benzene (c-C₆H₆; Cernicharo et al. 2001).

COMs have been detected in the ISM since the 1970s (e.g. CH₃CHO, Gottlieb 1973; or HCOOCH₃, Brown et al. 1975). Most of the detections have been reported towards hot cores, that is, dense gas surrounding high-mass protostars in massive star-forming regions such as Sgr B2 and Orion KL (e.g. Bisschop et al. 2007; Ziurys & McGonagle 1993; Tercero et al. 2013, 2015; Ikeda et al. 2001), and towards hot corinos, that is, the low-mass analogs of hot cores such as IRAS 16293-2422 (Cazaux et al. 2003) and NGC 1333 IRAS 4A (Bottinelli et al. 2004). The high degree of chemical complexity (e.g. C₂H₅CN, CH₃CCH, HCOOCH₃, CH₃OCH₃…) found in these protostellar sources is thought to result from thermal desorption of the ice mantles coating dust grains. Herbst & van Dishoeck (2009) have reviewed the subject. They classify the observed COMs towards protostars in three different generations depending on the time they are produced. The zeroth-generation species, such as H₂CO and CH₃OH, are formed through grain-surface reactions in a previous cold interstellar phase in which icy mantles built up around granular cores of carbon and silicate grains. The first-generation species, such as HCOOCH₃, are formed during the passive warm-up of the inner envelope of the protostar. During this phase, zeroth-generation species are photodissociated producing radicals such as HCO and CH₃O, which can associate through surface reactions to form larger molecules. Finally, the second-generation species are formed once the core has become a hot core or hot corino. The dust temperature at this stage is high enough (T_d ≳ 100 K) to sublimate the mantles completely, and new molecules are formed through warm gas-phase reactions (ion molecule and endothermic neutral-neutral reactions).

When protostellar outflows impact the ambient envelope material, grains can be eroded, and because of the high temperatures reached in shocks, icy mantles can sublimate. In sufficiently high-velocity shocks, ices can also be sputtered and directly injected into the gas phase (e.g. υ_s > 20 − 25 km s⁻¹ for water ice, Draine et al. 1983). Several observations of the L1157 outflow have revealed the existence of H₂CO, H₂CS, CH₃OH, HC₃N, HNCO, NH₂CHO, and CH₃CHO in shocked gas (Bachiller & Pérez Gutiérrez 1997; Mendoza et al. 2014; Codella et al. 2015). In addition, Requena-Torres et al. (2006) have shown that the galactic centre contains dense clouds that are rich in CH₃OH, HCOOCH₃, and CH₃OCH₃. Widespread shocks have been invoked to explain their abundance and extended spatial distribution. The relative importance of the gas-phase reactions immediately after ice mantle sublimation compared to a formation of COMs on the grain surfaces is however far from being understood.

Recent observations towards UV-shielded cold cores (e.g. TMC 1, L1689B, or Barnard 1-b) have revealed molecules once considered to be present only in hot molecular cores (e.g. Remijan et al. 2006; Bacmann et al. 2012; Cernicharo et al. 2012). In these cold environments, COMs are thought to form on the surface of grains and to be released through non-thermal desorption processes, chemical desorption, direct desorption by cosmic ray impacts, or secondary photon induced processes (Cernicharo et al. 2012). Finally, a number of organic species have been identified in circumstellar envelopes around evolved stars (e.g. IRC+10216, Cernicharo et al. 2000), towards extragalactic sources (e.g. Meier & Turner 2005, 2012; Aladro et al. 2011), meteorites (e.g. Cronin & Chang 1993), and comets (e.g. Bockelée-Morvan et al. 2004).

Studies of environments permeated by stellar far-UV photons (FUV, 6.0 eV < hν < 13.6 eV) are more scarce. Guzmán et al. (2014) and Gratier et al. (2013) presented the unexpected detection of HCOOH, CH₂CO, CH₃CN, CH₃OH, CH₃CHO, and CH₃CCH in the Horsehead photodissociation region (PDR; a relatively low-FUV-flux dense PDR, χ ≈ 60 times the mean interstellar field in Draine units), finding enhanced abundances compared to a nearby cold and dense core shielded from external FUV radiation. Guzmán et al. (2014) proposed that owing to the cold grain temperatures, ice-mantle photodesorption processes dominate the formation of COMs in the Horsehead. In lower-density translucent clouds (χ ≈ 1), only H₂CO has been unambiguously detected (Liszt et al. 2006). These observations might suggest that, in FUV-irradiated environments, the presence of COMs diminishes as the χ/n_H ratio increases. Thus, COMs might not be present in strongly FUV-irradiated gas.

In this work we test the above hypothesis and investigate the presence and abundances of COMs at the high FUV-illuminated edge of the Orion Bar, with an impinging radiation field of a few 10⁴ times the mean interstellar field (Marconi et al. 1998). Because of its proximity (414 ± 7 pc, Menten et al. 2007) and nearly edge-on orientation, the Orion Bar provides an excellent template to determine the chemical content and also to investigate the structure and dynamics of strongly FUV-irradiated molecular gas (e.g. Tielens et al. 1993; Hogerheijde et al. 1995; Cuadrado et al. 2015, 2016; Goicoechea et al. 2016).

Multi-wavelength observations towards different positions of the Orion Bar have been historically used in the development of PDR models (e.g. Tielens & Hollenbach 1985a,b) and today they are still used as a template to understand the unresolved emission from sources as different as the nuclei of distant star-burst galaxies (e.g. Fuente et al. 2008) or the illuminated surfaces of protoplanetary disks (e.g. Agúndez et al. 2008). The transition from ionised to neutral gas in the Orion Bar has been extensively mapped, generally at low angular resolution, in various atomic and molecular tracers (see e.g. Tielens et al. 1993; Hogerheijde et al. 1995; van der Werf et al. 1996; Walmsley et al. 2000; Leurini et al. 2010; Ossenkopf et al. 2013). The detailed analysis of these observations suggested an inhomogeneous density distribution. The most commonly accepted scenario is that an extended gas component, with mean gas densities of 10⁴ − 10⁵ cm⁻³, causes the chemical stratification seen perpendicular to the dissociation front as the FUV field is attenuated (Hogerheijde et al. 1995; Jansen et al. 1995; Simon et al. 1997; van der Wiel et al. 2009; Habart et al. 2010; van der Tak et al. 2013). In addition, another component of clumpy material with higher densities (≳ 10⁶ cm⁻³) and more shielded from FUV radiation is embedded in the interclump gas (Burton et al. 1990; Parmar et al. 1991; Stoerzer et al. 1995; Young Owl et al. 2000; Lis & Schilke 2003; Batrla & Wilson 2003; Andree-Labsch et al. 2017). Previous observations of H₂CO and CH₃OH in the Orion Bar have shown that the two molecules trace these two different environments: CH₃OH, the denser and cooler clumps seen deeper inside the Bar, and H₂CO, the warmer interclump medium directly exposed to the strong FUV-field (Leurini et al. 2006, 2010).

We have performed a complete millimetre line survey using the IRAM 30 m telescope towards the edge of Orion Bar, a high FUV-illuminated position close to what Stoerzer et al. (1995) call the “CO⁺ peak”, near the H₂ dissociation front (see Fig. 1). This position shows a distinctive chemistry that can only be understood due to the presence of a strong flux of FUV photons (e.g. compared to that in the more shielded clumps deeper inside the Bar): enhanced abundances of simple reactive ions (e.g. CH⁺, CO⁺, and HOC⁺) and small hydrocarbon ions (e.g. l-C₃H⁺) that are only abundant in the presence of C⁺; vibrationally excited H₂; and high gas temperatures (e.g. Nagy et al. 2013; Guzmán et al. 2015; Cuadrado et al. 2015). Indeed, high-angular resolution ALMA-ACA images do show that reactive ions such as SH⁺ or HOC⁺ do not emit from the more shielded clumps (Goicoechea et al. 2017). Therefore, their chemistry is different to that of the PDR edge observed in this work. Our survey covers ~220 GHz of bandwidth, between 80 GHz and 360 GHz. These observations have been complemented with several 8.5″ resolution maps of different molecules at 0.9 mm to put our line survey position in the context of the Bar large-scale emission. In this paper we focus on the detection of rotational lines from COMs and related precursors.

Fig. 1.

C¹⁸O J = 3 → 2 (left panel) and H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 (right panel) line integrated intensity maps (W in K km s⁻¹) in the 10 − 11 km s⁻¹ velocity interval observed with the IRAM 30 m telescope at ~329 GHz and ~351 GHz, respectively (colour scale). Black contours are the ${\text{H}}_{\text{2}}^{\text{*}}$ υ = 1 → 0 S (1) emission delineating the H₂ dissociation front (DF, Walmsley et al. 2000). The red contours represent the O i fluorescent line at 1.32 µm (Walmsley et al. 2000) marking the position of the ionisation front (IF) that separates the H ii region and the neutral cloud. The blue contours represent the H¹³CN J = 1 → 0 emission tracing dense molecular clumps inside the Bar (Lis & Schilke 2003). The target position of the Orion Bar survey, close to the DF, is indicated with a blue circle. The IRAM 30 m beam at 1 mm is plotted in the bottom left corner (black circle).

The paper is organised as follows. In Sect. 2 we describe the line survey and the mapping observations. In Sect. 3 we report the observational features of the detected organic molecules, while in Sect. 4 we present the C¹⁸O and H₂CO integrated line intensity maps at 0.9 mm. The data analysis is explained in Sect. 5. In Sect. 6 we discuss the results, and finally in Sect. 7 we summarise the main conclusions.

2. Observations and data reduction

2.1. Line survey

We performed a complete millimetre line survey towards the edge of Orion Bar using the IRAM 30 m telescope. The target position is at α₂₀₀₀ = 05^h 35^m 20.8^s, δ₂₀₀₀ = − 05°25′17.0″, close to the dissociation front. This position is at (Δα, Δδ) = (3″, −3″) from the “CO⁺ peak” of Stoerzer et al. (1995).

Our observations cover a total of 217 GHz along 3, 2, 1, and 0.9 mm bands of the EMIR receivers using the WILMA (2 MHz spectral resolution) and FTS (200 kHz spectral resolution) backends. The observing procedure was position switching (PSW) with the reference position located at an offset (−600″, 0″) to avoid the extended molecular emission from the Orion Molecular Cloud (OMC-1). The antenna temperature, $T_{\text{A}}^{*},$ was converted to the main beam temperature, T_MB , through the $T_{\text{MB}}=T_{\text{A}}^{\text{*}}/{\eta }_{\text{MB}}$ relation, where η_MB is the antenna efficiency. All intensities in tables and figures are in main beam temperature. A local standard of rest (LSR) of 10.7 km s⁻¹ has been used in the line survey target position in the Orion Bar dissociation front. Table 1 shows an overview of the frequency ranges observed with each backend, the spectral resolution in velocity units (δυ) in the observed frequency ranges, as well as the variation in the telescope efficiencies, η_MB , and the half power beam width (HPBW) across the covered frequency range.

Table 1.

Observed frequency ranges and telescope parameters.

Rec.a	Backend	Freq.b [GHz]	δυc [km s-1]	ηMB d	HPBWe [arcsec]
E0	FFTS	80–117	0.75–0.51	0.87–0.82	31–21
E1	WILMA	128–176	4.7–3.4	0.80–0.74	19–14
E2	FFTS	202–275	0.30–0.22	0.70–0.56	12–9
E3	FFTS	275–305	0.22–0.20	0.56–0.50	9–8
	FFTS	328–359	0.18–0.17	0.46–0.40	8–7

Notes.

^(a)Emir receiver.

^(b)Observed frequency range.

^(c)Spectral resolution in velocity units (δυ) in the observed frequency ranges.

^(d)Antenna efficiencies.

^(e)The half power beam width can be well fitted by HPBW[arcsec]≈2460/Frequency[GHz].

In Appendix A we study the possible contamination of bright molecular line emission from the Orion BN/KL region (located at a distance of ~20′) through the beam side lobes (several arcmin, see Greve et al. 1998). Although the contribution to the detected power ranges from ~12% (3 mm) to ~30% (0.8 mm), most of the emission from the hot core region arises at different velocities (~8 km s⁻¹) than those of the Orion Bar (~10 − 11 km s⁻¹) and thus can be easily separated.

Data reduction and spectral analysis were done using the CLASS software of the GILDAS package¹ developed by IRAM. Weighted spectra were averaged and calibrated, and a polynomial baseline of low order (typically second or third order) was subtracted from each ~200 MHz wide spectrum. Finally, Gaussian profiles were fitted to all the detected lines (Appendix B). The rms noise of our observations obtained after integration during ~4 h ranges between 4 mK and 20 mK per resolution channel.

2.2. Maps

The 0.9 mm line emission from different molecules was mapped with the IRAM 30 m telescope in January 2014 under excellent winter conditions (<1 mm of precipitable water vapour). The E3 receiver and the FTS backend at 200 kHz spectral resolution were used. On-the-fly (OTF) scans were obtained along and perpendicular to the Bar over a 170″ × 170″ region, with an OFF position at (−600″,0″) relative to the map centre at α₂₀₀₀ = 05^h 35^m 20.1^s, δ₂₀₀₀ = − 05°25′07.0″, which is slightly different from that of the line survey (see above). The HPBW at this frequency is ~7″. Data processing consisted in a linear baseline subtraction in each observed spectra. The resulting spectra were gridded to a data cube through convolution with a Gaussian kernel providing a final resolution of 8.5″. The total integration time was approximately 6 h, and the achieved rms noise is ~1 K per ~0.2 km s⁻¹ channel. Figure 1 shows the C¹⁸O J = 3 → 2 and H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 integrated line intensity maps in the range υ_LSR = 10 − 11 km s⁻¹ in which the Orion Bar shows prominent emission. We note that van der Wiel et al. (2009) also presented a smaller H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 map at lower angular resolution (a factor of ~2).

3. Results: complex organic molecule detections

We have identified more than 250 lines from COMs and related organic precursors. The detected lines are attributed to ten different molecules with up to seven atoms: HCO, HNCO, H₂CO, H₂CS, CH₂NH, H₂CCO, HC₃N, CH₃OH, CH₃CN, and CH₃CHO. We also detect several lines of the isotopologue ${\text{H}}_{\text{2}}^{\text{13}}\text{CO}.$ We note that among the organic molecules with up to seven atoms, we also identified several lines of formic acid, trans- and cis-HCOOH, in the Orion Bar PDR. Cuadrado et al. (2016) have recently reported the first detection of the cis conformer of HCOOH in the interstellar medium as well as a detailed analysis of the detected lines of both HCOOH conformers in the 3 mm spectral band. Line profiles peak at υ_LSR = 10 − 11 km s⁻¹, the velocity of the Bar, thus confirming their PDR origin. The detected organic molecules and their dipole moments are listed in Table 2.

Table 2.

Dipole moments (μ), number of detected lines of complex organic molecules and related organic precursors, and their corresponding Figure and Table numbers.

Molecule	No. of lines	μ [Debyes]		Fig.	Table
HCO	16	μa=1.363	μb=0.700a	2	B.1
H2CO	15	μa=2.332b		3	B.2
${\text{H}}_2^{13}\text{CO}$	8	μa=2.332b		4	B.3
H2CS	26	μa=1.649b		5	B.4
HNCO	6	μa=1.602	μb=1.350c	6	B.5
CH2NH	6	μa=1.340	μb=1.446d	7	B.6
H2CCO	30	μa=1.422b		8	B.7
HC3N	11	μa=3.732e		9	B.8
CH3CN	44	μa=3.922f		11	B.9
CH3OH	55	μa=0.896	μb=1.412g	12	B.10
CH3CHO	36	μa=2.423	μb=1.260h	13	B.11

References.

^(a)Blake et al. (1984)

^(b)Fabricant et al. (1977)

^(c)Hocking et al. (1975)

^(d)Allegrini et al. (1979)

^(e)DeLeon & Muenter (1985)

^(f)Gadhi et al. (1995)

^(g)Sastry et al. (1981)

^(h)Kleiner et al. (1996)

Previous studies of the Orion Bar have already reported the presence of HCO, H₂CO, and CH₃OH in several positions close to the “CO+ peak”: (i) Schilke et al. (2001) detected one hyperfine component of HCO N = 1 → 0 using the IRAM 30 m telescope; (ii) Hogerheijde et al. (1995) detected nine lines of H₂CO and three lines of CH₃OH in several positions along the Bar using the JCMT, IRAM 30 m, and CSO telescopes; (iii) Nagy et al. (2017) detected ten submm H₂CO lines with Herschel/HIFI. In addition, H₂CO, ${\text{H}}_{\text{2}}^{\text{13}}\text{CO},$ HDCO, H₂CS, and CH₃OH have been detected towards the colder, denser, and more FUV-shielded clumps: (iv) Leurini et al. (2006) detected seven lines of H₂CO, four lines of H₂CS, and ten lines of CH₃OH using APEX between 279 GHz and 361.5 GHz towards clump #1 of Lis & Schilke (2003); (v) Parise et al. (2009) also detected two isotopologues, one line of ${\text{H}}_{\text{2}}^{\text{13}}\text{CO}$ and one line of HDCO, towards clump #3 of Lis & Schilke (2003) using the IRAM 30 m telescope. To our knowledge, our work presents the first detection of CH₂NH, H₂CCO, HC₃N, CH₃CN, and CH₃CHO in the Orion Bar PDR.

Line assignment was carried out using J. Cernicharo’s spectral catalogue (MADEX, Cernicharo 2012), and the JPL (Pickett et al. 1998)² and CDMS (Müller et al. 2001, 2005)³ public databases. Although some lines have peak intensities lower than 3σ noise level, we included them in our results as tentative detections because other transitions from the same molecule are well detected in this survey, and these 3σ limits are in agreement with our excitation models. Given their intricate spectroscopy, below we summarise the main spectroscopic information of the detected species.

3.1. Formyl radical: HCO

The formyl radical, HCO, is a light asymmetric rotor with one unpaired electron and one non-zero nuclear spin. The quantum numbers designating the energy levels are N, J, and F. Spin doubling (J = N + S ) is produced by the coupling between the rotational angular momentum, N, and the unpaired electron spin, S , while the hyperfine structure (F = J + I) is due to the coupling of the angular momentum, J, and the spin of the hydrogen nucleus, I. Each rotational level N_Ka,Kc is split into a doublet by the spin-rotation interaction, the levels of which are further split into doublets by magnetic hyperfine interactions, with the final energy levels labelled by the quantum number F = J ± 1/2 (e.g. Bowater et al. 1971; Saito 1972; Austin et al. 1974; Blake et al. 1984). a-type (ΔK_a = 0, ±2...) and b-type (ΔK_a = ±1, ±3...) transitions are allowed, with a stronger dipole moment for the a-type transitions (see Table 2).

We identified a total of 16 lines of HCO. They consist of four sets of rotational transitions corresponding to the hyperfine splitting of the N = 1 → 0 to 4 → 3 transitions. All transitions observed here are a-type. The quantum numbers of the detected HCO transitions, their spectroscopic parameters, and the results from fitting the line profiles with Gaussians are listed in Table B.1. In Fig. 2 we present the spectra of the N = 1 → 0 to 4 → 3 rotational lines.

Fig. 2.

Detected HCO hyperfine structure (HFS) lines of the N = 1 → 0, 2 → 1, 3 → 2, and 4 → 3 rotational transitions (black histogram spectra). A single excitation temperature model is shown overlaid in red (see Sect. 5.4). HFS lines are indicated by the blue arrows. The other spectral features appearing in the selected windows are labelled with their corresponding identification.

3.2. Formaldehyde: H₂CO

Formaldehyde, H₂CO, was the first polyatomic organic molecule found in the ISM (Snyder et al. 1969). It is a slightly asymmetric prolate rotor molecule with K-type doublets and an ortho-para symmetry because of the two indiscernible off-axis hydrogen atoms. The ortho states are those with K_a odd, and the para states with K_a even. The nuclear spin weights are 3 and 1 for o-H₂CO and p-H₂CO, respectively. The K-type doublets are the result of the slight asymmetry in the H₂CO produced by the light H atoms. There are no line doublets in the K_a = 0 para state (e.g. Bocquet et al. 1996; Brünken et al. 2003; Eliet et al. 2012).

We detected nine lines of the ortho species (in the K_a = 1 and 3 ladders) with E_u/k ≤ 125.8 K, and six lines for the para species (in the K_a = 0 and 2 ladders) with E_u/k ≤ 82.1 K. Line profiles are shown in Fig. 3. Table B.2 gives the observed line parameters.

Fig. 3.

Example of ortho- and para-H₂CO lines (black histogram spectra). A non-LTE LVG model (T_k ≃ 200 K, n(H₂) ≃ 1 × 10⁶ cm⁻³, N(o-H₂CO) = 4.0 × 10¹³ cm⁻², and N(p-H₂CO) = 1.8 × 10¹³ cm⁻²) is shown overlaid in red (see Sect. 5.2). The dashed lines indicate the LSR velocity (10.7 km s⁻¹) of the Orion Bar PDR.

We also detected eight lines of ${\text{H}}_{\text{2}}^{\text{13}}\text{CO},$ consisting of six lines of ortho species and two lines of the para species, in the K_a = 0 and 1 ladders, respectively (E_u/k ≤ 34.0 K). Line profiles are shown in Fig. 4. Table B.3 gives the observed line parameters. No lines of deuterated formaldehyde were detected.

Fig. 4.

Example of ortho- and para-${\text{H}}_{\text{2}}^{\text{13}}\text{CO}$ lines (black histogram spectra). A non-LTE LVG model (T_k ≃ 200 K, n(H₂) ≃ 1 × 10⁶ cm⁻³, $N{(\text{o}-\text{H}}_{\text{2}}^{\text{13}}\text{CO})=7.3\times {\text{10}}^{\text{11}}{\text{cm}}^{-2},$ and $N{(\text{p}-\text{H}}_{\text{2}}^{\text{13}}\text{CO})=2.3\times {\text{10}}^{\text{11}}{\text{cm}}^{-2})$ is shown overlaid in red (see Sect. 5.2).

3.3. Thioformaldehyde: H₂CS

Thioformaldehyde, H₂CS, has a formaldehyde-like structure. It is a slightly asymmetric prolate rotor, with only a-type transitions and ortho-para symmetry (e.g. Johnson et al. 1971; Beers et al. 1972; Maeda et al. 2008).

We detected 15 lines of ortho species (in the K_a = 1 and 3 ladders) with E_u/k ≤ 149.8 K, and 11 lines of para species (in the K_a = 0 and 2 ladders) with E_u/k ≤ 112.0 K. Figure 5 shows a selection of ortho and para detected lines. Table B.4 gives the observed line parameters.

Fig. 5.

Example of ortho- and para-H₂CS lines (black histogram spectra). Single excitation temperature models are shown overlaid in red (see Sect. 5.4).

3.4. Isocyanic acid: HNCO

Isocyanic acid, H-N=C=O, is the simplest molecule containing carbon, hydrogen, oxygen, and nitrogen. It is a slightly asymmetric prolate rotor with a- and b-type transitions. Its rotational levels are designated as J_Ka,Kc (e.g. Kukolich et al. 1971; Hocking et al. 1975; Lapinov et al. 2007).

We detect six a-type transitions inside the K_a = 0 rotational ladder (E_u/k ≤ 96.0 K). The observed lines do not show evidence for hyperfine splittings. Line profiles are shown in Fig. 6. Table B.5 gives the observed line parameters.

Fig. 6.

Example of HNCO detected lines (black histogram spectra). A single excitation temperature model is shown overlaid in red (see Sect. 5.4).

3.5. Methanimine: CH₂NH

Methanimine is the simplest molecule containing a carbon-nitrogen double bond (H₂C=NH). It is a planar molecule and a nearly prolate asymmetric rotor with components of the electric dipole moment along both the a and b molecular axes (e.g. Kirchhoff et al. 1973; Dore et al. 2010, 2012).

We detected six lines of CH₂NH consisting of (i) three a-type transitions (in the K_a = 0 and 1 ladders) and (ii) three b-type transitions, with E_u/k ≤ 64.1 K. The observed lines do not show evidence for hyperfine splittings. The line profiles are shown in Fig. 7. Table B.6 gives the observed line parameters.

Fig. 7.

CH₂NH detected lines (black histogram spectra). A single excitation temperature model is shown overlaid in red (see Sect. 5.4).

3.6. Ketene: H₂CCO

Ketene, H₂C=C=O, is a slightly asymmetric prolate rotor with only a-type transitions and ortho-para symmetry (e.g. Johnson & Strandberg 1952; Fabricant et al. 1977; Brown et al. 1990; Johns et al. 1992). We detected 20 lines of ortho species (in the K_a = 1 and 3 ladders) with E_u/k ≤ 205.2 K, and ten lines for the para species (in the K_a = 0 and 2 ladders) with E_u/k ≤ 154 K. Examples of the line profiles of ortho and para lines are shown in Fig. 8. Table B.7 gives the observed line parameters.

Fig. 8.

Example of ortho- and para-H₂CCO lines (black histogram spectra). Single excitation temperature models are shown overlaid in red (see Sect. 5.4).

3.7. Cyanoacetylene: HC₃N

Cyanoacetylene, H–C≡C–C≡N, is a linear molecule with a large dipole moment of μ_a = 3.732 D (DeLeon & Muenter 1985). It is the simplest example of cyanopolyynes, HC_nN (with n = 3, 5, 7…; see e.g. Turner 1971; Avery et al. 1976; Cernicharo et al. 1986). The rotational transitions show hyperfine splittings owing to the interaction of the electric-quadrupole moment of the nitrogen nuclear spin with the electronic-charge distribution (e.g. de Zafra 1971; Mbosei et al. 2000; Chen et al. 1991; Yamada et al. 1995; Thorwirth et al. 2000). This hyperfine splitting is only fully resolved in the lowest rotational transitions (e.g. Turner 1971; Dickinson 1972), therefore, HC₃N is described here by a simple rotational spectrum, with J+1 → J transitions.

We detected 11 lines of HC₃N, from J = 9 → 8 to 24 → 23, with E_u/k ≤ 131.0 K. Figure 9 shows a selection of detected lines and Table B.8 gives the observed line parameters.

Fig. 9.

Example of HC₃N detected lines (black histogram spectra). A non-LTE LVG model (T_k ≃ 80 K, n(H₂) ≃ 1 × 10⁶ cm⁻³, N(HC₃N) = 4.0 × 10¹¹ cm⁻²) is shown overlaid in red (see Sect. 5.2).

3.8. Methyl cyanide: CH₃CN

Methyl cyanide, CH₃CN, is a strongly prolate symmetric rotor, therefore, the permitted radiative transitions are all ΔK = 0. The internal rotation of the methyl group gives rise to two non-interacting torsional substates, denoted A and E. The A state levels are described by K = 3n, and those of E state by K = 3n ± 1, with n ≥ 0 (e.g. Kukolich et al. 1973; Kukolich 1982; Boucher et al. 1977; Anttila et al. 1993; Šimečková et al. 2004; Cazzoli & Puzzarini 2006; Müller et al. 2009). Transitions with different K but the same J occur in narrow frequency regions but they have quite different energies (see Fig. 10).

Fig. 10.

Observed emission lines from the J = 5 → 4 transitions of CH₃CN. Transitions with different K but the same J occur in narrow frequency regions despite they have different level energies.

We detected 18 lines of A-CH₃CN (in the K = 0 and 3 ladders) with E_u/k ≤ 184.4 K, 23 lines for the E-CH₃CN (in the K = 1, 2, and 4 ladders) with E_u/k ≤ 199.0 K, and three lines corresponding to several fully overlapped A-E transitions. Examples of line profiles of A- and E-CH₃CN lines are shown in Fig. 11. Table B.9 gives the observed line parameters. We did not detect the CH₃NC isomer (see Sect. 5.6).

Fig. 11.

Example of A- and E-CH₃CN lines (black histogram spectra). A non-LTE LVG model (T_k ≃ 150 K, n(H₂) ≃ 1 × 10⁶ cm⁻³, N(A-CH₃CN) = 5 × 10¹¹ cm⁻², and N(E-CH₃CN) = 6 × 10¹¹ cm⁻²) is shown overlaid in red (see Sect. 5.2).

3.9. Methanol: CH₃OH

Methanol, CH₃OH, is the simplest alcohol molecule. It is a slightly asymmetric rotor showing hindered internal rotation. Its energy levels are classified as a symmetric rotor with quantum number J_K and a threefold barrier that causes two states in the molecule, A and E; the latter being doubly degenerate (e.g. Lees & Baker 1968; Lees et al. 1973; Xu et al. 2008).

We detected 20 lines of A-CH₃OH (in the |K| = 0 and 1 ladders) with E_u/k ≤ 104.4 K, 33 lines for the E-CH₃OH (in the |K| = 1, 2, 3, and 4 ladders) with E_u/k ≤ 114.8 K, and two lines corresponding to several fully overlapped A-E transitions. Examples of line profiles of A- and E-CH₃OH lines are shown in Fig. 12. Table B.10 gives the observed line parameters.

Fig. 12.

Example of A- and E-CH₃OH lines (black histogram spectra). A non-LTE LVG model (T_k ≃ 40 K, n(H₂) ≃ 3 × 10⁷ cm⁻³, N(A-CH₃OH) = 1.2 × 10¹³ cm⁻², and N(E-CH₃OH) = 1.9 × 10¹³ cm⁻²) is shown overlaid in red (see Sect. 5.2).

3.10. Acetaldehyde: CH₃CHO

Acetaldehyde, CH₃CHO, is an asymmetric-top molecule with A and E symmetry states and a- and b-type transitions (e.g. Kleiner et al. 1996, and references therein).

We detected 18 lines of A-CH₃CHO, 15 lines of E-CH₃CHO, and three lines corresponding to a fully overlapped A-E transition (a-type transitions with E_u/k ≤ 109.7 K). Figure 13 shows a selection of detected lines and Table B.11 gives the observed line parameters.

Fig. 13.

Examples of A- and E-CH₃CHO lines (black histogram spectra). Single excitation temperature models are shown overlaid in red (see Sect. 5.4).

4. Spatial distribution of C¹⁸O and H₂CO

Figure 1 shows the spatial distribution of the rotationally excited C¹⁸O J = 3 → 2 (E_u/k ≃32 K, A_ul ≃ 2 × 10⁻⁶ s⁻¹) and H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 (E_u/k ≃ 47 K, A_ul ≃ 1 × 10⁻³s⁻¹) lines along the Bar. The molecular dissociation front (DF) is traced by the vibrationally excited H₂ ν = 1 → 0 S (1) line emission (${\text{H}}_{\text{2}}^{\text{*}};$ black contours in Fig. 1; Walmsley et al. 2000). The red contours represent the O i fluorescent line at 1.32 µm (Walmsley et al. 2000) marking the position of the ionisation front (IF) that separates the H ii region and the neutral cloud. The blue contours represent the H¹³CN J = 1 → 0 emission tracing dense molecular clumps inside the Bar (Lis & Schilke 2003).

The emission from the Orion Bar can be distinguished from the extended OMC-1 cloud component by the emission LSR velocity. While OMC-1 is brighter in the 8 − 10 km s⁻¹ velocity range, the Orion Bar emits predominantly in the 10 − 11 km s⁻¹ range. The C¹⁸O and H₂CO line emissions shown in Fig. 1 are integrated in this latter interval. In these maps, H₂CO peaks at the position of the dense (n(H₂) ≃ 6 × 10⁶ cm⁻³) and lukewarm (T_k ≃ 50 K) clumps inside the Bar (Lis & Schilke 2003). In addition, our maps reveal extended H₂CO emission along the PDR and close to the DF. This conclusion was initially inferred by Leurini et al. (2010) from ~11″ resolution observations of lower-excitation H₂CO lines.

To investigate the chemical stratification as a function of FUV flux attenuation, we built averaged line intensity crosscuts perpendicular to the Bar. The maps were rotated by 50° to bring the incident FUV radiation from left, and then, line intensities in the mapped area were averaged parallel to the IF. In Fig. 14 we plot the resulting normalised intensity cuts as a function of distance to the IF position. Distances are in arcseconds and increase with depth into the molecular cloud. Although these cuts do not represent true abundance variations (different lines have different excitation conditions), to first order they can be used to study the chemical stratification in the PDR. The ${\text{H}}_{\text{2}}^{\text{*}}$ emission peak is at ~15″ (vertical dot-dashed line) from the IF, and shows a narrow emission profile. The H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 emission peak appears at ~27″, followed by C¹⁸O J = 3 → 2 that peaks at ~35″.

Fig. 14.

Normalised intensity crosscuts for C¹⁸O J = 3 → 2 (blue curve), H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 (red curve), and ${\text{H}}_{\text{2}}^{\text{*}}$ (black curve) as a function of distance from the IF (in arcsec). The vertical dot-dashed line indicates the H₂ dissociation front. FUV radiation comes from left.

We note that the different position of the emission peaks of the C¹⁸O J = 3 → 2 and H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 lines shown in Fig. 14 could be related to a sharp thermal gradient in the region, with the gas temperature increasing from ~50 K inside the dense clumps seen in the PDR interior (Lis & Schilke 2003), to ~300 K near the DF. The H₂CO J_Ka,Kc = 5_1,5 → 4_1,4 line requires higher excitation conditions (higher temperature/density) and thus it would naturally peak closer to the DF than the C¹⁸O J = 3 → 2 line, even for constant H₂CO and C¹⁸O abundances throughout the PDR. Still, the main point is that observations show that H₂CO is present in the warm interclump gas and close to the PDR edge where the FUV flux is strong.

5. Analysis

5.1. Line parameter fitting procedure

Detected lines show simple Gaussian line profiles centred at the Orion Bar LSR velocity and thus Gaussian profiles were fitted to all detected lines. For the resolved and unblended COM lines detected with significant S/N, the mean line width is Δυ = 1.8 km s⁻¹, with a standard deviation of σ = 0.6 km s⁻¹.

The spectroscopic and observational line parameters of the detected lines are given in Tables B.1-B.11. We provide line frequencies (in MHz), energy of the upper level of each transition (E_u/k in K), Einstein coefficient for spontaneous emission (A_ul in s⁻¹), intrinsic line strength (S _ul), and the level degeneracy (g_u) from the MADEX spectral catalogue, and the JPL and CDMS molecular databases. The velocity-integrated intensity (∫ T_MB dυ in mK km s⁻¹), LSR velocity (v_LSR in km s⁻¹), FWHM (full width at half maximum) line width (Δυ in km s⁻¹), and the line peak temperature (T_MB in mK) were obtained from Gaussian fits. Parentheses indicate the uncertainty. When two or more transitions were found to be overlapping, the total profile was fitted. Fully overlapping transitions are marked with connecting symbols in the tables.

5.2. Non-LTE LVG excitation analysis: physical conditions

We estimate the beam-averaged physical conditions towards the line survey position from non-LTE (local thermodynamic equilibrium) excitation models for molecules with known inelastic collisional rate coefficients (H₂CO, HC₃N, CH₃CN, and CH₃OH)⁴. In particular, we run a large grid of non-LTE excitation and radiative transfer models in which the statistical equilibrium equations are explicitly solved in the LVG (large velocity gradient) approximation using MADEX (Cernicharo 2012, see also Appendix A in Cuadrado et al. 2015).

We tested a broad range of column density (N), H₂ density (n(H₂) = 10³ − 10⁹ cm⁻³), and gas temperature (T_k = 10 − 1000 K) values. We adopted Δυ = 2 km s⁻¹ line widths. In order to constrain the physical conditions that reproduce the observed line intensities and line profiles towards the observed position, we compared the observed H₂CO, CH₃OH, CH₃CN, and HC₃N lines to synthetic lines obtained from the grid of LVG models. The best fit model (assuming uniform beam filling emission) was obtained by finding the minimum root mean square (rms); we refer to Cuadrado et al. (2015) for a detailed explanation. The range of column densities that we used as an input in the models is around the value derived from the rotational diagrams analysis (see Sect. 5.4). In fact, the best fit models have column densities within a factor of 2 (see Table 3) of the inferred value from the rotational diagram analysis (we note that in these models the observed lines are optically thin). Figure 15 represents the rms_min/rms ratio as a function of T_k and n(H₂) for a grid of excitation models trying to fit the observed lines towards the PDR position.

Table 3.

Best fit LVG model parameters and comparison with the column density obtained with the rotational diagram (RD) analysis.

	LVG calculations			RD analysis*
	Tk [K]	n(H2) [cm−3]	N(X)LVG [cm−2]	N(X)RD [cm−2]
o-H2CO	~200	~106	4.0 × 1013	(4.4 ± 0.6) × 1013
p-H2CO	~200	~106	1.8 × 1013	(1.6 ± 0.1) × 1013
HC3N	~80	~106	4.0 × 1011	(4.2 ± 0.3) × 1011
A-CH3CN	~150	~106	5.0 × 1011	(5.8 ± 0.8) × 1011
E-CH3CN	~150	~106	6.0 × 1011	(5.7 ± 0.7) × 1011
E-CH3OH	~40	~3×107	1.9 × 1013	(1.9 ± 0.2) × 1013

Notes.

^*Rotational diagram analysis assuming uniform beam filling (see Sect. 5.4).

Fig. 15.

Grid of LVG excitation and radiative transfer models for CH₃CN, H₂CO, CH₃OH, and HC₃N lines towards the PDR position assuming uniform beam filling and using column densities shown in Table 3. The best fits are those with the higher rms_min / rms value.

From the excitation analysis it seems that not all COMs arise from the same PDR layers (i.e. same cloud depth). The set of beam-averaged physical conditions that best fit the CH₃CN and H₂CO lines lie within T_k = 150 − 250 K and n(H₂) = (1 − 3) × 10⁶ cm⁻³, similar (although slightly denser gas) to that obtained for C₂H in Cuadrado et al. (2015). Figures 3 and 11 show the best fit models overlaid over the observed lines. The HC₃N detected lines (see Fig. 9) fit within T_k = 70 − 90 K and n(H₂) > 1 × 10⁶ cm⁻³. The physical conditions that fit the CH₃OH detected lines (see Fig. 12) are colder and denser (T_k = 40 − 50 K and n(H₂) ≃ 5 × 10⁷ cm⁻³). The best-fit parameters (n(H₂), T_k, and N) of the LVG models fitting the lines in Fig. 3, 9, 11, and 12 are shown in Table 3. We note, however, that the best fit density value depends on the assumed emission beam filling factor⁵ for each molecule. Lower densities (by a factor of up to ~10) would still be consistent with observations if one assumes a semi-extended emission source (i.e. more compact than the beam-size at each observed frequency; see Sect. 5.4).

Higher-angular-resolution maps are needed to spatially resolve the structures producing the molecular emission (Goicoechea et al. 2016). Either way, CH₃OH and HC₃N seem to arise from a different, cooler component where the FUV flux has been attenuated. This scenario agrees with the different H₂CO and CH₃OH spatial distributions observed by Leurini et al. (2010) with the IRAM 30 m telescope and the PdB interferometer. In particular, Leurini et al. found that both species show different spatial distributions, with CH₃OH only tracing the denser and cooler clumps seen deeper inside the Bar, whereas H₂CO also traces the warmer extended (interclump) gas directly exposed to the strong FUV-field. Consistent with previous works (Goicoechea et al. 2011, 2016; Nagy et al. 2013), we derive very high thermal gas pressures at the PDR edge (P_th/k = n_H T_K ≳ 10⁸ K cm⁻³).

5.3. Subthermal excitation effects on the rotational diagrams of symmetric- and asymmetric-top molecules

In this Section we show how subthermal excitation effects modify the resulting rotational population diagram of symmetric- and asymmetric-top molecules such as CH₃CN and H₂CO, respectively. In particular we run a grid of LVG models for varying gas densities, but keeping the same gas temperature and column density. Figures 16 and 17 show model results in the form of population diagrams for A-CH₃CN transitions in the K_a = 0 and 3 rotational ladders (T_k = 150 K and N_tot = 5 × 10¹¹ cm⁻²), and for p-H₂CO in the K_a = 0 and 2 rotational ladders (T_k = 200 K and N_tot = 1.8 × 10¹³ cm⁻²). Only at very high gas densities, higher than the critical density for collisional excitation (n » n_cr), do inelastic collisions drive the level populations close to LTE (T_rot ≃ T_k). In a rotational diagram, the K_a ladders merge in a single straight line with a slope equal to −1/T_k. Owing to the high gas densities, this characteristic straight diagram is typically seen towards hot cores (e.g. Bisschop et al. 2007). As the gas density decreases, the excitation becomes subthermal (T_rot < T_k), and the rotational population diagram starts to show separate rotational ladders for each set of transitions with the same K_a quantum number. For very polar molecules such as CH₃CN and H₂CO (thus high n_cr), subthermal excitation happens at relatively high densities. Therefore, accurate column densities and rotational temperatures from rotational diagrams can only be obtained if the individual column densities for each rotational ladder are computed independently. This likely explains the discrepancy with the high T_rot obtained by Nagy et al. (2017) in the rotational diagram of their submm o-H₂CO lines observed with Herschel/HIFI towards the Bar.

Fig. 16.

Rotational population diagrams for A-CH₃CN computed with a non-LTE excitation code. All models adopt the same gas temperature (T_k = 150 K) and column density (N_tot = 5 × 10¹¹ cm⁻²), but three different n(H₂) values. For simplicity, only rotational transitions in the K_a = 0 (filled symbols) and 3 (empty symbols) ladders are shown.

Fig. 17.

Rotational population diagrams for p-H₂CO computed with a non-LTE excitation code. All models adopt the same gas temperature (T_k = 200 K) and column density (N_tot = 1.8 × 10¹³ cm⁻²), but three different n(H₂) values. For simplicity, only rotational transitions in the K_a = 0 (filled symbols) and 2 (empty symbols) ladders are shown. We note that at densities <10⁸ cm⁻³, the set of transitions with the K_a = 2 number are also split into two components: (i) transitions with K_c = J − 1 (empty symbols connected with solid lines) and (ii) transitions with K_c = J − 2 (empty symbols connected with dashed lines).

5.4. Rotational diagrams and column-density determination

We determined the beam-averaged column density (N) and the rotational temperature (T_rot) for each molecule found by building rotational diagrams (e.g. Goldsmith & Langer 1999; Cuadrado et al. 2015). The large number of detected lines allows us to accurately determine rotational temperatures and column densities. Optically thin emission is assumed (consistent with our best LVG models of H₂CO, HC₃N, CH₃CN, and CH₃OH). We note that owing to their specific spectroscopy and prevailing excitation conditions at the edge of the Bar (subthermal excitation), a correct determination of the rotational temperatures and column densities for species such as H₂CO, H₂CS, H₂CCO, CH₃OH, and CH₃CN requires building specific diagrams for each rotational ladder (see Sect. 5.3). For the molecules with known collisional rates, we determine N_LVG ≈ N_RD (within a factor of 2). This comparison shows that the N_RD computed for the other molecules are accurate.

The rotational diagrams were built considering two limiting cases: (i) That the detected emission is extended (i.e. uniform beam filling), with η_bf = 1; and (ii) that the emission is semi-extended, assuming that θ_s = 9″ (the typical beam of the IRAM 30 m telescope at ~1 mm)⁶. We only considered lines not blended with other transitions and detected above a 3σ level. The resulting rotational diagrams are shown in Fig. 18. Rotational temperatures and column densities obtained by linear least squares fits are listed in Table 4. The uncertainties shown in Table 4 indicate the uncertainty obtained from the least squares fit to the rotational diagram. The uncertainty obtained in the determination of the line parameters with the Gaussian fitting programme are included in the uncertainty bars at each point of the rotational diagram.

Fig. 18.

Rotational diagrams of the detected complex molecules in the Orion Bar PDR (assuming uniform beam filling). Fitted values of the rotational temperature, T_rot, column density, N, and their respective uncertainties are also indicated for each molecule.

Table 4.

Rotational temperatures (T_rot), column densities (N), and abundances with respect to hydrogen nuclei inferred in the Orion Bar PDR.

	Uniform beam filling		Semi-extended source
	Trot [K]	N(X) [cm−2]	Trot [K]	N(X) [cm−2]	Abundance *	Notes
HCO	15 ± 2	(1.1 ± 0.3) × 1013	9 ± 2	(5.3 ± 3.5) × 1013	(1.7 − 8.4) × 10−10	a
o-H2CO Ka = 1	17 ± 2	(3.7 ± 0.6) × 1013	12 ± 1	(1.1 ± 0.3) × 1014		a
o-H2CO Ka = 3	17	(6.9 ± 1.0) × 1012	12	(1.1 ± 0.2) × 1013		b,c
p-H2CO Ka = 0	16 ± 2	(9.8 ± 1.4) × 1012	9 ± 1	(3.9 ± 0.6) × 1013		a,c
p-H2CO Ka = 2	18 ± 1	(6.2 ± 0.2) × 1012	13 ± 1	(1.3 ± 0.1) × 1013		a,c
[(o+p)-H2CO]	—	(6.0 ± 0.6) × 1013	—	(1.7 ± 0.3) × 1014	(0.9 ± 2.7) × 10−9	d
o-${\text{H}}_2^{13}$CO	15 ± 3	(7.3 ± 2.4) × 1011	10 ± 2	(3.0 ± 1.1) × 1012		a
p-${\text{H}}_2^{13}$CO	9 ± 1	(2.3 ± 0.4) × 1011	7 ± 1	(8.0 ± 1.2) × 1011		a,c
[(o+p)-${\mathbf{\text{H}}}_2^{13}$CO]	—	(9.6 ± 2.4) × 1011	—	(3.8 ± 1.1) × 1012	(1.5 − 6.0) × 10−11	d
o-H2CS Ka = 1	30 ± 3	(3.1 ± 0.5) × 1012	19 ± 1	(1.3 ± 0.2) × 1013		a
o-H2CS Ka = 3	30	(6.7 ± 1.9) × 1011	19	(2.7 ± 0.4) × 1012		e,c
p-H2CS Ka = 0	29 ± 6	(5.0 ± 1.4) × 1011	16 ± 2	(2.6 ± 1.0) × 1012		a
p-H2CS Ka = 2	35 ± 8	(7.8 ± 4.4) × 1011	15 ± 3	(2.6 ± 1.9) × 1012		a
[(o+p)-H2CS]	—	(5.0 ± 0.7) × 1012	—	(2.1 ± 0.3) × 1013	(0.8 − 3.3) × 10−10	d
HNCO	51 ± 7	(1.0 ± 0.1) × 1012	26 ± 3	(5.6 ± 1.3) × 1012	(1.6 ± 8.9) × 10−11	a
CH2NH	28 ± 7	(1.1 ± 0.4) × 1012	27 ± 7	(2.4 ± 1.0) × 1012	(1.7 − 3.8) × 10−11	a
o-H2CCO Ka = 1	55 ± 2	(3.0 ± 0.1) × 1012	30 ± 2	(1.3 ± 0.2) × 1013		a
o-H2CCO Ka = 3	64 ± 6	(1.3 ± 0.3) × 1012	34 ± 1	(4.2 ± 0.7) × 1012		a
p-H2CCO Ka = 0	57 ± 4	(4.8 ± 0.6) × 1011	42 ± 3	(1.3 ± 0.2) × 1012		a
p-H2CCO Ka = 2	54 ± 3	(1.1 ± 0.1) × 1012	29 ± 2	(6.1 ± 1.8) × 1012		a
[(o+p)-H2CCO]	—	(5.9 ± 0.3) × 1012	—	(2.5 ± 0.3) × 1013	(0.9 − 4.0) × 10−10	d
HC3N	43 ± 2	(4.2 ± 0.3) × 1011	27 ± 1	(3.2 ± 0.3) × 1012	(0.7 − 5.1) × 10−11	a
cis-HCOOH	23 ± 4	(4.6 ± 0.7) × 1011	21 ± 4	(4.2 ± 0.6) × 1012		f
trans-HCOOH Ka = 0	12 ± 2	(3.5 ± 0.5) × 1011	6 ± 1	(4.1 ± 0.6) × 1012		f
trans-HCOOH Ka = 1	12 ± 3	(3.3 ± 1.3) × 1011	6 ± 1	(3.6 ± 2.1) × 1012		f
trans-HCOOH Ka = 2	13 ± 3	(6.3 ± 2.8) × 1011	7 ± 1	(5.0 ± 2.4) × 1012		f
[(cis+trans)-HCOOH]	—	(1.8 ± 0.3) × 1012	—	(1.7 ± 0.3) × 1013	(0.3 − 2.7) × 10−10	g
A-CH3CN Ka = 0	26 ± 2	(2.8 ± 0.5) × 1011	19 ± 1	(1.9 ± 0.6) × 1012		a
A-CH3CN Ka = 3	30 ± 1	(3.0 ± 0.6) × 1011	21 ± 1	(1.5 ± 0.6) × 1012		a
E-CH3CN Ka = 1	28 ± 2	(2.9 ± 0.5) × 1011	20 ± 1	(1.8 ± 0.4) × 1012		a
E-CH3CN Ka = 2	31 ± 1	(1.7 ± 0.2) × 1011	22 ± 1	(1.0 ± 0.3) × 1012		a
E-CH3CN Ka = 4	26 ± 1	(1.1 ± 0.4) × 1011	18 ± 1	(6.7 ± 3.2) × 1011		a
[(A+E)-CH3CN]	—	(1.2 ± 0.1) × 1012	—	(6.9 ± 1.0) × 1012	(0.2 − 1.1) × 10−10	h
A-CH3OH Ka = 0	34 ± 2	(8.0 ± 0.6) × 1012	18 ± 2	(5.3 ± 1.1) × 1013		a
A-CH3OH Ka = ± 1	26 ± 2	(3.5 ± 0.9) × 1012	14 ± 1	(2.8 ± 0.7) × 1013		a
E-CH3OH	36 ± 3	(1.9 ± 0.2) × 1013	26 ± 3	(6.5 ± 1.4) × 1013		a
[(A+E)-CH3OH]	—	(3.1 ± 0.2) × 1013	—	(1.5 ± 0.2) − 1014	(0.5 − 2.4) × 10−9	h
A-CH3CHO	31 ± 2	(2.3 ± 0.3) × 1012	20 ± 1	(1.2 ± 0.2) × 1013		a
E-CH3CHO	33 ± 3	(2.6 ± 0.4) × 1012	21 ± 1	(1.2 ± 0.2) × 1013		a
[(A+E)-CH3CHO]	—	(4.9 ± 0.5) × 1012	—	(2.4 ± 0.3) × 1013	(0.8 − 3.8) × 10−10	h

Notes.

^*The abundance of each species with respect to H nuclei is given by $\frac{N(X)}{N_H}=\frac{N(X)}{N(H)+2N(H_2)},$ with N(H₂) ≃ 3 × 10²² cm⁻² and N(H) ≃ 3 × 10²¹ cm⁻² (van der Werf et al. 2013).

^(a)T_rot and N from rotational diagram analysis.

^(b)Only two lines detected with the same E_u/k. N calculated assuming the same T_rot that o-H₂CO K_a = 1.

^(c)ΔN estimated assuming a 15% of the calculated N.

^(d)Total N calculated as the sum of the ortho and para species.

^(e)Only two lines detected with upper level energies of 138.3 K and 149.8 K. They are too similar for an accurate estimation of T_rot and N from a rotational diagram analysis. N calculated assuming the same T_rot that o-H₂CS K_a = 1.

^(f)From Cuadrado et al. (2016).

^(g)Total N calculated as the sum of the cis and trans species.

^(h)Total N calculated as the sum of the A and E species.

For H₂CO, ${\text{H}}_{\text{2}}^{\text{13}}\text{CO},$ H₂CS, H₂CCO, A-CH₃OH, A-CH₃CN, and E-CH₃CN we have built specific rotational diagrams for different sets of transitions with the same K_a quantum number value. The total column density of the molecule is obtained by adding the column density of each rotational ladder. For the other molecules, it is possible to fit the observed transitions by a single T_rot and N, independently of the K_a value. The cis- and trans-HCOOH rotational diagrams are shown in Cuadrado et al. (2016).

For molecules with hyperfine structure, like that of HCO, each rotational transition was described without splits, only with a single rotational number N. For this purpose, the integrated intensity, W, level degeneracy, g, and line strength, S , of each transition was calculated as the sum of all allowed hyperfine components of each N+1 → N transition. The characteristic frequency, ν, was determined using the weighted average with the relative strength of each line as weight, and the Einstein coefficient, A, was calculated using the weighted values in the usual relation.

Comparing the resulting rotational diagrams, we see that: (i) H₂CO and CH₃OH are, by far, the most abundant organic molecules (by a factor of ≳5), followed by HCO, H₂CCO, CH₃CHO, H₂CS, HCOOH, CH₃CN, CH₂NH, HNCO, ${\text{H}}_2^{13}\text{CO},$ and HC₃N (in decreasing order of abundance); (ii) the ortho-to-para (OTP) ratios obtained from the H₂CO, ${\text{H}}_{\text{2}}^{\text{13}}\text{CO},$ H₂CS, H₂CCO column densities are ~3, as expected at high gas temperature; (iii) we obtain similar rotational temperatures for ortho-para and A-E species; (iv) and the inferred $[{\text{H}}_2\text{CO}]/[{\text{H}}_2^{13}\text{CO}]$ ratio is 63 ± 3, similar to the ¹²C/¹³C ≈ 67 isotopic ratio in Orion (Langer & Penzias 1990). Hence, this confirms that even lines from the most abundant organic species studied in this work, are optically thin.

In order to confirm the correct line identification and possible line blendings, we have modelled the spectrum of each molecule using the MADEX radiative transfer model assuming a Boltzmann distribution and a single excitation temperature for the rotational levels population (we recall that for the observed set of lines/ladders of a given molecule, we infer a single temperature component from the rotational diagrams). Figures 2, 5, 6, 7, 8, and 13 show the observational spectra (black histograms) and the modelled spectra (red lines) of each molecule. The obtained fits agree very well with the observations and rotational diagrams.

5.5. Abundances

In order to determine molecular abundances with respect to hydrogen nuclei⁶ towards the line survey position, we derived the beam-averaged H₂ column density from our observations of the optically thin C¹⁸O lines (J = 1 → 0, 2 → 1, and 3 → 2 transitions) assuming ¹⁶O/¹⁸O ≈ 500 (Wilson & Rood 1994) and a [CO]/[H₂] abundance of ~10⁻⁴ (lower than the canonical value due to photodissociation). The resulting H₂ column density, N(H₂) ≃ 3 × 10²² cm⁻², is in good agreement with previous estimations of N(H₂) close to the dissociation front (see e.g. Hogerheijde et al. 1995). A hydrogen atom column density, N(H), of ~3 × 10²¹ cm⁻² has been inferred from H_I observations towards the edge of the Orion Bar (van der Werf et al. 2013). Molecular abundances with respect to hydrogen nuclei are listed in Table 4. The abundances of the detected species range from 10⁻⁹ to 10⁻¹¹.

5.6. Undetected complex organic molecules and precursors

The broadband frequency coverage of the survey allowed us to obtain upper limits for other chemically interesting organic molecules that have not been detected towards the edge of the PDR, but could have been expected. In particular, we searched for HDCO, CH₃O, CH₃NC, CH₃CCH, CH₃OCH₃, HCOOCH₃, CH₃CH₂OH, CH₂CHCN, and CH₃CH₂CN, because they have been detected in other PDRs and star-forming regions (e.g. Horsehead, Orion KL, Barnard 1-b, or IRAS 16293-2422; see Table 6 for references). The CH₃O radical is thought to be an important intermediary for the H₂CO and CH₃OH chemistry but so far, has only been detected in cold and dense gas (Cernicharo et al. 2012; Bacmann & Faure 2016). HDCO has been detected by Parise et al. (2009) towards the lukewarm, dense and more FUV-shielded clump #3 of Lis & Schilke (2003), but it is not detected towards the warmer and high FUV-illuminated line survey position, near the dissociation front.

Table 6.

Abundances relative to H₂ in different environments (in units of 10⁻¹⁰).

Molecule	Orion Bar1*	Horsehead		Orion KL		L1157-B17	IRAS 16293-24228	B1-b9
	PDR	PDR2*	Core3*	HC4	CR5	Outflow	Hot corino	Dense core
HCO	1.7	8.4b	< 0.8b	—	< 0.36,g	—	< 2.0	0.2n
HNCO	0.2	—	—	780h	—	605k	1.7	0.7p
H2CO	9.0	2.9c	2.0c	1200h	440h	4000l	7.0	4.6q
H2CS	0.8	—	—	150h	74h	1100l	—	0.9q
t-HCOOH†	0.2a	0.5d	0.1d	800i	—	—	< 3.0	0.1n
CH2NH	0.2	—	—	42h	—	—	< 5.0	—
H2CCO	0.9	1.5d	0.5d	—	51h	—	1.8	0.2n
HC3N	0.07	0.06e	0.08e	81h	—	100l	0.3	3.1r
CH3OH	5.0	1.2f	2.3f	22000h	12000h	115000l	44	31r,s
CH3CN	0.2	2.5e	0.08e	300h	120h	—	1.5	0.4r
CH3CHO	0.8	0.7d	0.2d	9.5i	—	250m	1.0	0.1n
CH3CCH	< 1.0	4.4d	3.0d	24i	1336,i	—	6.5	5.0r

Notes.

^*Abundances with respect to total hydrogen nuclei.

^†To date, the cis conformer of HCOOH has only been detected towards the Orion Bar PDR (Cuadrado et al. 2016), therefore, we only provide the abundances of the trans conformer.

⁽¹⁾This work. Abundances calculated assuming uniform beam filling.

Ref.:

^(a)Cuadrado et al. (2016).

⁽²⁾Low-UV field Horsehead PDR (χ ≈ 60; N_H = 3.8 × 10²² cm⁻²)

⁽³⁾condensation shielded from the UV field (dense core, N_H = 6.4 × 10²² cm⁻²) behind the Horsehead PDR edge.

Ref.:

^(b)Gerin et al. (2009)

^(c)Guzmán et al. (2011)

^(d)Guzmán et al. (2014)

^(e)Gratier et al. (2013)

^(f)Guzmán et al. (2013).

⁽⁴⁾Orion KL Hot Core (HC)

⁽⁵⁾Orion KL Compact Ridge (CR).

⁽⁶⁾Abundances calculated for the Extended Ridge.

Ref.:

^(g)Blake et al. (1987)

^(h)Crockett et al. (2014)

⁽ⁱ⁾B. Tercero, private communication.

⁽⁷⁾Peak B1 in the blue lobe of the L1157.

Ref.:

^(k)Mendoza et al. (2014)

^(l)Bachiller & Pérez Gutiérrez (1997)

^(m)Codella et al. (2015).

⁽⁸⁾Low-mass protostar (hot corino) IRAS 16293-2422. The abundances are computed for a N(H₂) = 2 × 10²³ cm⁻². Abundances averaged over a ~20″ beam (van Dishoeck et al. 1995).

⁽⁹⁾Quiescent dark core Barnard 1-b (B1-b), N(H₂) = 1.3 × 10²³ cm⁻² (see e.g. Hirano et al. 1999; Lis et al. 2002).

Ref.:

⁽ⁿ⁾Cernicharo et al. (2012)

^(p)Marcelino et al. (2009)

^(q)Marcelino et al. (2005)

^(r)N. Marcelino, private communication

^(s)Öberg et al. (2010).

First, we estimated 3σ line intensities using the relation

$${\displaystyle \int {\text{T}}_{\text{MB}}\text{dv}=3\sigma \sqrt{\text{2}\delta \upsilon \Delta \upsilon }}\left[{\text{K}\text{kms}}^{-1}\right],$$ (1)

where σ is the rms of the observations per resolution channel [K], δυ is the velocity spectral resolution [km s⁻¹], and Δυ is the assumed line widths (~2 km s⁻¹). Second, we used MADEX to create synthetic models that simulate the line emission to constrain their column densities. The column densities and 3σ upper limit abundances assuming T_rot = 20 − 30 K are listed in Table 5. We also provide the following abundance ratio upper limits: [HDCO]/[H₂CO] < 0.003 and [CH₃NC]/[CH₃CN] < 0.1. Parise et al. (2009) estimated an abundance ratio of [HDCO]/[H₂CO] > 0.006 towards clump #3, which suggests that the deuteration diminishes from the lukewarm and shielded clumps to the warmer and more FUV-irradiated cloud edge.

Table 5.

Upper limits for undetected COMs.

Molecule	N(X) [cm−2]	Abundance a
HDCO b	(1.3 - 1.5) × 1011	(2.1 - 2.4) × 10−12
CH3O	(2.0 - 4.0) × 1012	(3.2 - 6.3) × 10−11
CH3NC	(5.0 - 8.0) × 1010	(0.8 - 1.3) × 10−12
CH3CCH	(5.0 - 8.0) × 1012	(0.8 - 1.3) × 10−10
CH3OCH3	(7.0 - 9.0) × 1012	(1.1 - 1.4) × 10−10
HCOOCH3	(9.0 - 9.5) × 1012	(1.4 - 1.5) × 10−10
CH3CH2OH	(1.0 - 1.2) × 1012	(1.6 - 1.9) × 10−11
CH2CHCN	(9.0 - 9.5) × 1012	(1.4 - 1.5) × 10−10
CH3CH2CN	(7.0 - 5.0) × 1011	(1.1 - 0.8) × 10−11

Notes.

^(a)The abundance of each species with respect to H nuclei (see Sect. 5.5).

^(b)Detected by Parise et al. (2009) towards clump #3 (Lis & Schilke 2003) of the Orion Bar.

6. Discussion

In order to shed more light on the possible chemical formation routes of COMs in strongly FUV-irradiated gas from an observational perspective, we compare the observed abundances of several COMs and precursors in different environments (see Table 6). In addition to the Orion Bar, we consider the Horsehead PDR (a low-FUV-flux PDR) and a nearby cold core also in the Horsehead nebula, the Orion KL hot core and compact ridge (warm dense gas at roughly the same distance to the Bar), the L1157 outflow (shocked gas), the low-mass protostar IRAS 16293-2422 (hot corino), and the quiescent dark cloud Barnard 1-b (B1-b).

6.1. COMs in different environments

Figure 19 shows a comparison between the abundances of 12 organic molecules detected in the above sources. As expected for widespread interstellar molecules, abundant H₂CO and CH₃OH are found in all sources (N(X)/N(H₂) > 10⁻¹⁰). In diffuse and translucent clouds (n(H₂) ≃ 10² − 10³ cm⁻³), only H₂CO has been detected so far (Liszt et al. 2006), suggesting that the formation of COMs is more efficient in dense gas.

Fig. 19.

Molecular abundances with respect to H₂ in several sources. In the Orion Bar PDR, blue points represent the molecular abundance assuming uniform beam filling.

In general, the abundances of COMs, and CH₃OH in particular, are much higher towards Orion KL (hot core and compact ridge) and towards the L1157-B1 outflow. This translates into low [X_COM]/[CH₃OH] ≪ 1 abundance ratios in those environments. The Horsehead PDR, however, shows [X_COM]/[CH₃OH] > 1 ratios for X_COM = HCO, H₂CO, H₂CCO, CH₃CN, and CH₃CCH abundances. The Orion KL hot core and compact ridge show enhanced COM abundances. Owing to the very high dust temperatures (T_d > 100 K), ice sublimation and subsequent warm gas-phase chemistry dominates (e.g. Blake et al. 1987). On the other hand, the abundances inferred in the hot corino IRAS 16293-2422 (lower dust temperatures) are more similar to those in the Orion Bar PDR and Horsehead. Interestingly, the L1157 outflow shows the highest abundances of CH₃OH, H₂CO, H₂CS, HC₃N, and CH₃CHO. This suggests that the combined effects of ice mantle sublimation, grain sputtering, and hot gas-phase chemistry in shocks results in a very efficient COM formation.

Among the studied species, only the HCO radical and cis-HCOOH (Cuadrado et al. 2016) are more abundant in PDRs. This enhancement is thus a characteristic feature of FUV-illuminated gas. Figure 20 shows normalised abundances with respect to the abundances in B1-b (cold and gas shielded from strong FUV radiation). HCO, CH₃CHO, H₂CCO, H₂CO, and t-HCOOH are a factor of ~2 − 10 more abundant in the PDRs than in the cold core.

Fig. 20.

Comparison of the abundance ratios with respect to B1-b in several sources.

Guzmán et al. (2014) observed two positions of the Horsehead, the moderately warm PDR (T_k ≃ 60 K) and a cold core behind the PDR and shielded from stellar FUV field. Interestingly, they found enhanced COM abundances in the PDR compared to the core. Given the low-FUV field in the Horsehead, dust grains are relatively cold even at the PDR edge (T_d ≲ 30 K, Goicoechea et al. 2009). These dust temperatures are significantly below the sublimation temperatures of abundant interstellar ices (Gibb et al. 2000) such as H₂O (~100 K), CH₃OH (~100 K), and even H₂CO (~40 K). Hence, grains must be coated with mantles, even in the Horsehead PDR, and photodesorption can be efficient, either desorbing specific COMs or their gas-phase precursors (Guzmán et al. 2014). Indeed, PDR models adapted to the Horsehead and including grain surface reactions and ice photodesorption were invoked to explain the observed gas-phase CH₃OH and H₂CO abundances (Guzmán et al. 2013). In addition, the low H₂CO ortho-to-para ratio of two in the PDR (Guzmán et al. 2011) might support a cold surface origin for H₂CO. However, those Horsehead models required relatively high photodesorption yields (several 10⁻⁴ to 10⁻³ molecules per photon) that are apparently not supported by recent laboratory experiments of methanol photodesorption. In fact, little CH₃OH is actually seen to desorb in CH₃OH ice irradiation experiments (Bertin et al. 2016; Cruz-Diaz et al. 2016).

Also in the Horsehead, CH₃CN is 30 times more abundant in the PDR than in the shielded core (Gratier et al. 2013). CH₃CN is also an order of magnitude more abundant in the Horsehead than in the Orion Bar PDR. Given the high binding energy of CH₃CN ice (similar to CH₃OH, Collings et al. 2004) this result is quite surprising (methanol is more abundant than CH₃CN in the Bar). In addition, Gratier et al. (2013) detected CH₃NC towards the Horsehead PDR, with an isomeric ratio of [CH₃NC]/[CH₃CN] = 0.15, higher than our upper limit. These authors suggested that photodesorption triggers the CH₃CN abundance in the Horsehead. Therefore, photodesorption seems to enhance the gas-phase CH₃CN and CH₃NC abundances, but not that of CH₃OH. This suggests that photodesorption is a very selective process and thus, should be studied molecule by molecule in representative ISM ice analogs. In addition, gas phase reactions may allow the enhancement of CH₃CN with respect to CH₃OH.

In the Orion Bar, a much more strongly FUV-irradiated PDR, gas and dust grains are warmer (from T_d ≃ 40 to 80 K; Arab et al. 2012). Thus, ice mantle abundances must be small close to the cloud edge. Interestingly, we find higher H₂CO and CH₃OH abundances in the Bar than in the Horsehead PDR (but not HCO). In addition, H₂CO is detected close to the irradiated edge of the Orion Bar and we determine a H₂CO ortho-to-para ratio of approximately three (consistent with the high temperature limit). Finally, we determine a HCO/H₂CO/CH₃OH ≃ 1/5/3 abundance ratio, whereas a ~7/2/1 ratio is found in the Horsehead PDR (Gerin et al. 2009; Guzmán et al. 2013). The ratios typically inferred in cold dense gas are ~1/10/10 (Bacmann & Faure 2016). Although HCO is approximately five times more abundant in the Horsehead, both PDRs show [CH₃OH]/[H₂CO] < 1 abundance ratios. We note that H₂CO is also a product of photodissociation of CH₃OH. In warm but FUV-shielded environments (hot cores and shocks), CH₃OH is much more abundant than in PDRs and cold cores.

6.2. Limits to steady-state PDR gas-phase chemistry

In order to explore the gas-phase production of HCO, H₂CO, and CH₃OH, we show PDR model results adapted to the illumination conditions in the Bar using the Meudon PDR code (e.g. Le Petit et al. 2006; Goicoechea & Le Bourlot 2007). This is a model of a high gas-pressure isobaric PDR with χ = 2 × 10⁴ and P_th/k = 10⁸ K cm⁻³. Only H₂O, CO, and atoms are allowed to deplete from the gas following their adsorption energies (e.g. Hollenbach et al. 2009). However, we do not include surface chemistry (e.g. water ice only forms through water vapour freeze-out and desorbs thermally or by photodesorption; we refer to Cuadrado et al. 2015 for details). Figure 21 shows the abundance profiles as a function of cloud depth. Although the dust temperature is never low enough to allow CO ice-mantle formation, water ice starts to be very abundant at A_V ≳ 5 (a similar behaviour can be expected for CH₃OH ice). We note that the typical gas temperature inferred from our observations towards the line survey position (T_k ≈ 150 K) corresponds to A_V ≈ 1.5 − 2.0 mag and T_d ≈ 60 K in this model. Hence, we expect mostly bare grains at the edge of the PDR, only coated with a few monolayers of very polar ices such as water (with high adsorption energies).

Fig. 21.

Isobaric PDR model for the Orion Bar with P_th/k = 10⁸ K cm⁻³ and χ = 2 × 10⁴. The thick horizontal bars show the inferred abundances of HCO, H₂CO, and CH₃OH towards the PDR edge. In this model, these molecules are formed only through gas-phase reactions.

The model shown in Fig. 21 roughly reproduces the HCO abundance enhancement at the PDR edge (driven by reaction O + CH₂ → HCO + H), but underpredicts the observed H₂CO and CH₃OH abundances. In these models, the H₂CO formation at A_V ≈ 1.5 − 2.0 mag is largely dominated by reaction O + CH₃ → H₂CO + H. This comparison suggests either the gas-phase model misses important formation routes (or rates) in hot molecular gas, or that (nearly bare) grain surface formation and subsequent desorption is important. Esplugues et al. (2016) have shown PDR models including surface chemistry on grains with different ice content. In their models, chemical/reactive desorption (the surface reaction exothermicity is used to break the adsorbate-surface bond, e.g. Herbst 2015; Minissale et al. 2016), can dominate over photodesorption for some species (e.g. methanol). Nevertheless, the H₂CO and CH₃OH abundances predicted by Esplugues et al. (2016) in a PDR with high FUV fluxes are still lower than our observed values. Hence, at present, no model seems to reproduce the inferred abundances of molecules such as H₂CO and CH₃OH towards the Bar.

All these models, however, simulate a static PDR in which steady-state has been reached. A real PDR is likely to be more dynamic (Goicoechea et al. 2016). Time-dependent flows of molecular gas and icy grains might advect from inside the cold molecular cloud to the warm PDR edge. There they can be reprocessed by the combination of high temperatures and the presence of a strong FUV photon flux for some time before photodissociation. Thus, time-dependent desorption and advection of COMs (or gas-phase precursors) from the molecular cloud interior to the PDR may contribute to enhance their abundances.

The non-detection of HDCO at the illuminated edge of the Bar ([HDCO]/[H₂CO] < 3 × 10⁻³), and also in the Horsehead PDR (Guzmán et al. 2011), shows that pure gas-phase H₂CO deuteration is not efficient at high gas temperatures. However, HDCO has been detected towards the lukewarm (T_k ≲ 70 K),dense and more FUV-shielded clump #3 by Parise et al. (2009). This agrees with specific chemical models in which pure gas-phase deuteration is efficient in gas below Tk ≲ 70 K (Roueff et al. 2007). Time-dependent gas-phase models show that the [HDCO]/[H₂CO] abundance ratio is particularly low at early cloud times (Treviño-Morales et al. 2014). In hot cores where high abundances of COMs have been detected, the gas deuteration is very high as well. Both effects are related to the sublimation of ice mantles formed and deuterated in a previous cold cloud stage. In the context of a dynamic PDR, deuterated species could desorb from grain surfaces and advect to the PDR edge as well. The very low [HDCO]/[H₂CO] upper limit abundance towards the edge of the Bar suggests that this mechanism is not efficient for deuterated molecules, and that the gas temperature is too high to enhance the deuterium fractionation by pure gas-phase reactions alone.

6.3. CH₃CCH non detection

CH₃CCH is a widespread hydrocarbon present in the Horsehead PDR (~4.4 × 10⁻¹⁰) and cold core (~3 × 10⁻¹⁰), the extended ridge of Orion KL (~1 × 10⁻⁸), the hot corino IRAS 16293-2422 (~7 × 10⁻¹⁰), the Monoceros R2 ultra-compact H II region (~2 × 10⁻⁹; Ginard et al. 2012), and even the nucleus of the starburst galaxy M82 (~1 × 10⁻⁸; Fuente et al. 2005; Aladro et al. 2011). Unexpectedly, CH₃CCH lines are not bright towards the Bar edge. A few 2σ line features seen in the 1 mm range coincide with the expected frequencies of CH₃CCH. However, the detection cannot be confirmed with confidence at the sensitivity level of our line survey. Based on the tentative 2σ features, we estimate an upper limit abundance of [CH₃CCH] < (0.8 − 1.3) × 10⁻¹⁰. The unattenuated CH₃CCH photodissociation rate is high (several 10⁻⁹ s⁻¹ for χ = 1) and CH₃CCH reacts with C⁺ ions relatively fast (e.g. Wakelam et al. 2012). Both the FUV photon flux and the C⁺ column density are particularly high towards the observed position at the edge of the Bar (Ossenkopf et al. 2013; Goicoechea et al. 2015). This combination likely explains the reduced CH₃CCH abundance. Indeed, in our PDR models, CH₃CCH only reaches detectable abundances at A_V > 8 (Cuadrado et al. 2015).

7. Summary and conclusions

We have investigated the presence and abundance of complex organic molecules in the strongly FUV-irradiated edge of the Orion Bar PDR, near the H₂ dissociation front. We used the IRAM 30 m telescope to carry out a millimetre line survey, complemented with 8.5″ resolution maps of the molecular line emission at 0.9 mm. Despite being a very harsh environment, our observations show a relatively rich spectrum, with more than 250 lines arising from COMs and related organic precursors with up to seven atoms: H₂CO, CH₃OH, HCO, H₂CCO, CH₃CHO, H₂CS, HCOOH, CH₃CN, CH₂NH, HNCO, ${\text{H}}_2^{13}\text{CO},$ and HC₃N (in decreasing order of abundance). In particular, we obtained the following results:

• –The abundance of the detected species range from 10⁻⁹ to 10⁻¹¹; H₂CO being the most abundant. The inferred rotational temperatures range from ~10 to ~55 K, significantly lower than the gas temperature and thus consistent with subthermal excitation. ${\text{H}}_2^{13}\text{CO}$ and H₂CCO have the lowest and the highest rotational temperatures, respectively. We obtain similar rotational temperatures for the ortho-para and A-E species, and ortho-to-para ratios of approximately three.
• –For some molecules, we constrain the beam-averaged physical conditions from LVG excitation models. As shown by previous works (e.g. Leurini et al. 2006, 2010), we conclude that not all COMs arise from the same PDR layer. In particular, while CH₃OH only arises from dense gas in the more shielded PDR interior (T_k = 40 − 50 K), CH₃CN and H₂CO also trace warmer gas layers (T_k = 150 − 250 K), more exposed to a strong FUV-field.
• –We determine a HCO/H₂CO/CH₃OH ≃ 1/5/3 abundance ratio. Such relative abundances are not inferred in environments dominated by ice-mantle thermal desorption or grain sputtering. Overall, pure gas-phase models have difficulty reproducing the observed H₂CO and especially CH₃OH abundances in PDRs. Taking into account the elevated gas and dust temperatures at the edge of the Bar (T_d ≃ 60 K), we suggest the following scenarios for the formation of COMs: (i) hot gas reactions not included in current models; (ii) COMs are produced in the warm grain surfaces of nearly bare grains; or (iii) the PDR dynamics is such that COMs or specific precursors formed in cold icy grains deeper inside the molecular cloud desorb and advect into the PDR.

The presence of COMs in the interstellar medium is more widespread that initially expected. It includes very harsh environments such as shocked gas and now strongly FUV-irradiated gas. COM formation reflects the complicated interplay between gas and grain surface chemistry in different environments. However, the specific formation pathways are not fully clear and may even not be the same in different environments. More laboratory experiments to study different grain surface processes and to investigate the products and rates of different desorption mechanisms (photodesorption, sublimation, chemical desorption, etc.) are needed to distinguish between the different possible scenarios. The census of increasingly complex organic molecule detections will obviously increase in the years to come.