First hot corino detected around an isolated intermediate-mass protostar: Cep E-mm

Context . Intermediate-mass (IM) protostars provide a bridge between the low- and high-mass protostars. Despite their relevance, little is known about their chemical diversity. Aims . We want to investigate the molecular richness towards the envelope of I-M protostars and to compare their properties with those of low- and high-mass sources. Methods . We have selected the isolated IM Class 0 protostar CepE-mm to carry out an unbiased molecular survey with the IRAM 30m telescope between 72 and 350GHz with an angular resolution lying in the range 7–34 (cid:48)(cid:48) . Our goal is to obtain a cen-sus of the chemical content of the protostellar envelope. These data were complemented with NOEMA observations of the spectral bands 85.9–89.6GHz and 216.8–220.4GHz at angular resolutions of 2.3 (cid:48)(cid:48) and 1.4 (cid:48)(cid:48) , respectively. Results . The 30m spectra show bright emission of O- and N-bearing complex organic molecules (COMs): CH 3 OH and its rare isotopologues CH DOH and CH OH, CH CHO, CH OCH CH HCOOH, HCOOCH H 2 CCO, NH 2 CHO, CH 3 CN, C 2 H 3 CN, C 2 H 5 CN, HNCO and H 2 CO. We identify up to three components in the spectral signature of COMs: an extremely broad line (eBL) component associated with the outﬂowing gas ( FWHM > 7kms − 1 ), a narrow line (NL) component ( FWHM < 3kms − 1 ) associated with the cold envelope, and a broad line (BL) component ( FWHM (cid:39) 5 . 5kms − 1 ) which traces the signature of a hot corino. The eBL and NL components are detected only in molecular transitions of low excitation and dominate the emission of CH 3 OH. The BL component is detected in highly excited gas ( E up > 100K). The NOEMA observations reveal CepE-mm as a binary protostellar system, whose components, CepE-A and CepE-B, are separated by ≈ 1 . 7 (cid:48)(cid:48) . CepE-A dominates the core continuum emission and powers the long-studied, well-known, high-velocity jet associated with HH377. The lower ﬂux source CepE-B powers another high-velocity molecular jet, reaching velocities of ≈ 80kms − 1 , which propagates in a direction close to perpendicular with respect to the CepE-A jet. Our interferometric maps show that the emission of COMs arises from a region of ≈ 0 . 7 (cid:48)(cid:48) size around CepE-A, and corresponds to the BL component detected with the IRAM 30m telescope. On the contrary, no COM emission is detected towards CepE-B. We have determined the rotational temperature ( T rot ) and the molecular gas column densities from a simple population diagram analysis or assuming a given excitation temperature. Rotational temperatures of COMs emission were found to lie in the range 20 − 40K with column densities ranging from a few times 10 15 cm − 2 for O-bearing species, down to a few times 10 14 cm − 2 for N-bearing species. Molecular abundances are similar to those measured towards other low- and intermediate-mass protostars. Ketene (H 2 CCO) appears as an exception, as it is found signiﬁcantly more abundant towards CepE-A. High-mass hot cores are signiﬁcantly less abundant in methanol and N-bearing species are more abundant by two to three orders of magnitude. Conclusions . CepE-mm reveals itself as a binary protostellar system with a strong chemical di ﬀ erentiation between both cores. Only the brightest component of the binary is associated with a hot corino. Its properties are similar to those of low-mass hot corinos.


Introduction
The chemical composition of protostellar envelopes and their properties along the evolutionary stage of protostars is an important topic in astrochemistry. Since the pioneering work by Cazaux et al. (2003) and Sakai et al. (2008), systematic chemical studies of solar-type protostars (see Ceccarelli et al. 2007;Caselli & Ceccarelli 2012 for a review; also Lefloch et al. 2018) have identified two classes of objects. The first class corresponds to the so-called "hot corinos", that is, sources which display a rich content in complex organic molecules (COMs) in the central inner regions of the protostellar envelope (see Ceccarelli et al. 2007 for a review; also Taquet et al. 2015). Only a few hot corinos have been identified so far either with single dish or interferometric observations: IRAS16293-2422 (Cazaux et al. 2003;Bottinelli et al. 2004b;Jørgensen et al. 2011Jørgensen et al. , 2016, IRAS2, IRAS4B (Bottinelli et al. 2007), IRAS4A (Bottinelli et al. 2004a;Taquet et al. 2015), HH212 , L483 (Oya et al. 2017), B335 (Imai et al. 2016), SVS13A (Bianchi et al. 2017), Serpens SMM1, and SMM4 (Öberg et al. 2011). We note that very few sources were investigated in a systematic manner meaning that the COM budget in hot corino sources is very inhomogeneous, making a general picture difficult to come by. Hot corinos share some similarities with the hot cores observed around high-mass stars but they are not scaled-down versions of these. Bottinelli et al. (2007) showed that the abundances of Obearing species scaled to methanol are higher than those measured in hot cores by one to two orders of magnitude or more. The second chemical class of protostars corresponds to the so-called Warm Carbon Chain Chemistry (WCCC) sources, which have a rich content in C-chains but are poor in COMs. A recent survey of a sample of 36 Class 0/I protostars of the Perseus molecular cloud complex by Higuchi et al. (2018) shows that the majority of the sources observed have intermediate characters between these two distinct chemistry types.
In comparison, very little is known on intermediate-mass (IM) protostars. The first systematic study was carried out by Crimier et al. (2010) and Alonso-Albi et al. (2010), who both investigated the physical and chemical properties of a sample of five Class 0 IM protostars (CB3-mm, Cep E-mm, IC1396 N BIMA 2, NGC 7129 FIRS 2, and Serpens FIRS 1). Crimier et al. (2010) derived the dust and gas temperature and density profiles of the protostellar envelopes from modelling their spectral energy distribution (SED) with the one-dimensional (1D) radiative transfer code DUSTY. Comparing the physical parameters of the envelopes (density profile, size, mass) with those of the envelopes of low-and high-mass protostars led them to conclude that the transition between the three groups appears smooth, and that the formation processes and triggers do not substantially differ. Alonso-Albi et al. (2010) studied the CO depletion and the N 2 H + deuteration in the same source sample (and additionally towards L1641 S3 MMS1 and OMC2-FIR4). They found hints of CO underabundance by a factor of two with respect to the canonical value in the inner protostellar regions; they pointed out that high-angular-resolution observations are needed to conclude on the origin of such a deficit, and the possible role of outflows and the UV radiation from the star. The chemical properties of the source sample (molecular species and abundances) remain to be fully characterised. A few individual sources have been studied in detail Neri et al. 2007;Hogerheijde et al. 1999;Schreyer et al. 2002). In particular, Fuente et al. (2005) reported the presence of a hot core in NGC 7129 FIRS 2.
Intermediate-mass Class 0 protostellar clusters are important too as they also provide the transition between the low-density groups of TTauri stars and the high-density clusters around massive stars. OMC2-FIR4 is one of the (if not the) best-studied prototypes of this class. OMC-2 FIR4 is itself a young proto-cluster that harbours several embedded low-and intermediate-mass protostars (Shimajiri et al. 2008;López-Sepulcre et al. 2013). Analysis of the CH 3 OH emission lines observed with Herschel Space Observatory (Pilbratt et al. 2010) in the framework of the CHESS Large Program  led Kama et al. (2010) to suggest the presence of a hot core with kinetic temperatures above 100 K and around 400 au in size towards OMC-2FIR4. This is supported by the similarity of the molecular line spectrum of OMC-2 FIR4 and the hot corino source NGC 1333-IRAS4A, as observed with ASAI 1 (Lefloch et al. 2018). Observations with the Northern Extended Millimeter Array (NOEMA) at 5 resolution by López-Sepulcre et al. (2013) revealed a relatively complex structure with the presence of several components each of one or several solar masses, with chemical differentiation and an ionized HII region just next to the OMC-2 dust envelope. The chemical properties of IM protostars therefore remain to be characterized. The chemical diversity and its similarities and 1 http://www.oan.es/asai/ differences with low-and high-mass sources remain to be established.
As part of an observational effort to address these questions, we have conducted a detailed molecular line survey of the isolated IM Class 0 protostar Cep E-mm (IRAS23011+6126) with the IRAM 30 m telescope, complemented with observations with the IRAM interferometer NOEMA. Cep E-mm is located in the Cepheus E molecular cloud at a nearby distance of d = 730 pc (Sargent 1977). Its luminosity is ∼100 L and its envelope mass is 35 M (Lefloch et al. 1996;Crimier et al. 2010). Since its early discovery by Wouterloot & Walmsley (1986) and Palla et al. (1993), subsequent studies have confirmed the source to be an isolated intermediate-mass protostar in the Class 0 stage (Lefloch et al. 1996;Moro-Martín et al. 2001). The source drives a very luminous molecular outflow and jet (Lefloch et al. 2011(Lefloch et al. , 2015, terminated by the bright Herbig-Haro object HH377 (Ayala et al. 2000).
The article is organised as follows. In Sect. 2, we present the observations carried out with the IRAM 30 m telescope and the NOEMA interferometer. In Sect. 3, we present the results on the source multiplicity obtained from the thermal dust emission, as mapped with NOEMA, and the status of the dust components. In Sect. 4, we present the results of a systematic search for COMs, chemically related species with the IRAM instruments, which shows the presence of a hot corino associated with one of the dust components. In Sect. 5, we discuss the physical structure of Cep E-mm and the evidence for chemical differentiation in the core, and we compare our results with those obtained towards a few typical objects, from low-to high-mass. Our conclusions are summarised in Sect. 6.

Single-dish survey
Observations on Cep E-mm were obtained with the IRAM 30 m telescope near Pico Veleta (Spain) on coordinates α(2000) = 23 h 03 m 12 s .8, δ(2000) = 61 • 42 26 . Unbiased spectral coverage was carried out at 3 mm (80-116 GHz), 2 mm (129-173 GHz), 1.3 mm (200-300 GHz) and 0.9 mm (330-350 GHz) in Winter 2010 & 2015 and in Summer 2016 using the Eight MIxer Receiver (EMIR). The Fourier Transform Spectrometer (FTS) units were connected to the receivers on the 3 and 2 mm bands, providing a resolution of 195 kHz (∆V ∼ 0.6 and 0.4 km s −1 respectively). The Wideband Line Multiple Autocorrelator (WILMA) was connected to the higher-frequency bands, providing a resolution of 2 MHz (∆V ∼ 2.5 and 1.8 km s −1 for the 1.3 and 0.9 mm bands respectively). The observations were performed in wobbler switching mode with a throw of 180 (90 for the 2 mm) in order to ensure a flat baseline. The telescope beam size ranges from 34 at 72 GHz to 7 at 350 GHz. Calibration uncertainties in the 3, 2, 1.3 and 0.9 mm bands are typically 10, 15, 20 and 30%, respectively. The data were reduced using the GILDAS-CLASS software 2 . Intensities are expressed in units of antenna temperature corrected for the atmospheric absorption T * A . The rms noise per velocity interval of 1 km s −1 expressed in units of T * A lies in the range 2-5 mK in the 3 mm band, 3-7 mK in the 2 mm and 1.3 mm band, and 15-20 mK in the 0.9 mm band. The spectral bands and observations properties are summarised in Table 1.  .4 ) and (+4.5 , +20.8 ) with respect to the protostar, in order to map the emission of the protostar and the high-velocity outflow. The Wideband Express (WideX) backend was connected to the receivers providing a resolution of 2 MHz (∆V ∼ 2.7 km s −1 at 1.3 mm). Narrow spectral bands with lines of interest were observed in parallel with a resolution of 256 kHz. Observations properties are summarised on Table 2 for both the Wide Correlator (Widex) and the Narrow correlator backends. The antenna baselines sampled in our observations range from 19 to 176 m at 1.3 mm (from 21 to 443 m at 3 mm), allowing us to recover emission on scales from 14.7 to 1.5 at the 1.3 mm (from 34 to 1.6 at 3 mm). At 1.3 mm, for both C and D tracks, phase was stable (rms ≤ 45 • ) and the precipitable water vapor (pwv) was 1-3 mm with system temperatures ∼100−200, leading to less than 2% flagging in the dataset. At 3 mm, phase rms was ≤50 • pwv was 1-10 mm, and system temperatures were ∼60−200 K, leading to less than 1% flagging in the dataset.
Calibration and data analysis were carried out following standard procedures using the GILDAS software 3 . The continuum emission at 1.3 mm and 3 mm was obtained by selecting spectral windows free of molecular line emission. Continuum-free molecular maps were subsequently produced by subtraction of the continuum contribution.

Source multiplicity
Map of the 1.3 mm continuum emission is displayed in the top panel of Fig. 1. It shows a core of dimensions 1.56 × 1.08 (1141×788 au; beam deconvolved), with a PA of 45 • . The peak of the flux distribution (99.4 mJy/1.4 -beam) is located at the nominal position of Cep E-mm, and drives the well-known, bright, high-velocity CO outflow (Lefloch et al. 2015). Integrating over the region defined by the 3σ contour level (3.5 mJy/1.4 -beam), the total flux of the condensation is then S 1.3 mm = 233 mJy.
The location of the peak flux is shifted with respect to the center of the continuum core, suggesting the presence of a secondary, fainter source located at about 1.7 southwest of the peak flux.
In what follows, we will refer to the first (bright) and secondary (faint) components as Cep E-A and Cep E-B, respectively (Fig. 1).
We determined the position, size and intensity of the two continuum sources from a fit to the visibility table, assuming two 1D Gaussian distributions. We note that the uncertainties correspond to the statistical errors derived from the minimisation procedure. We find that the bright source Cep E-A has a diameter of 1.20 ± 0.01 (875 au) (beam-deconvolved) and a peak flux of 97.1 mJy/1.4-beam and a total flux of 168.8±0.7 mJy. The fainter source Cep E-B is located at the offset position (−0.85 , −1.52 ), at 1250 au from the Cep E-A. It has a size of 0.92 ± 0.02 (670 au). We estimate a peak flux of 26.2 mJy/1.4-beam and a total flux 40.3 ± 0.6 mJy.
In order to elucidate the nature of both condensations, we searched for evidence of outflowing gas motions associated with Cep E-A and Cep E-B using emission maps of the SiO J = 2−1 Fig. 1. Top: Cep E-mm 1.3 mm continuum map. Base contour and contour spacing are 3.6 (3σ) and 7.2 mJy beam −1 (6σ). The peak intensity is 99.2 mJy/1.42 -beam. The blue and red stars mark the position of Cep E-A and B components, respectively. Bottom: continuum emission of components A (left) and B (right) deduced from a two-component fit to the visibilities (see text). Base contour and contour spacing are 10.8 (9σ) and 7.2 mJy beam −1 (6σ). Peak intensities are 97.1 and 26.2 mJy/1.42beam, respectively. and 5-4 transitions present in the observed NOEMA bands, a good tracer of young protostellar outflows (Lefloch et al. 1998).
The left hand panel in Fig. 2 shows the SiO J = 5−4 emission between −135 and −110 km s −1 (blue) and +40 and +80 km s −1 (red); SiO traces the previously known, high-velocity jet oriented in the southwest-northeast direction (Lefloch et al. 1996(Lefloch et al. , 2011(Lefloch et al. , 2015. Both lobes slightly overlap at the location of Cep E-A, confirming the association with this protostar. The right hand panel of Fig. 2 displays the SiO J = 5−4 emission integrated between −80 and −40 km s −1 (blue) and +50 and +70 km s −1 (red). We note that the redshifted emission is slightly contaminated by the emission from the high-velocity jet from Cep E-A. Overall, the red-and blueshifted emissions draw a collimated, jet-like structure oriented in the eastern-western direction with a PA of −8 • . This jet emission is asymmetric, as it appears to be more extended along the blueshifted lobe. Both lobes overlap at about 1.5 south of Cep E-A. This definitely excludes a possible association with Cep E-A. Instead, the lobes appear to overlap at the position of Cep E-B, supporting a physical association with the secondary dust core.

Physical properties
The large-scale physical structure of Cep E-mm dust envelope was studied by Crimier et al. (2010). The authors derived the dust density and temperature profiles from a 1D modelling of the dust and between +40 and +80 km s −1 , respectively. Base contour and contour spacing of both jet emission are 10% of their maximum intensity. Right: the blue and red contours represent the integrated line emission between −80 and −40 km s −1 and between +50 and +70 km s −1 , respectively. Base contour and contour spacing of both jet emissions are 5% and 10% of their maximum intensity, respectively. . Continuum (real part of the visibilities, black squares) and corresponding best-fit power-law model (red line) averaged over baseline bins of 1 m, as a function of baseline length. Residuals are shown as a blue line. The zero-spacing 790 mJy flux density is extrapolated from single-dish observations at 1.3 mm by Lefloch et al. (1996). continuum emission at 24, 70, 450, 850 and 1300 µm by SCUBA and Spitzer. Fluxes were measured with an angular resolution in the range 7.5−14.8 , thereby probing mainly the outer regions (>8 ) of the envelope. They constrained a power-law index on the density profile n ∝ r −p with p = 1.88, and obtained selfconsistently the temperature profile T ∝ r −q with q = 1.10 in the inner region (r < 500 au). In the Rayleigh-Jeans approximation and for optically thin dust emission, the emergent dust continuum emission also has a simple power-law form I(r) ∝ r −(p+q−1) . For interferometric observations, the visibility distribution is then , and depends only on the density and temperature power-law index.
As can be seen in Fig. 3, a simple fit of the form V(b) ∝ b (−0.47±0.01) reproduces well our interferometric continuum observations and the single-dish flux of Lefloch et al. (1996). We obtain a (p + q) value of 2.53 ± 0.01 lower than that of Crimier et al. ( 2.98). Such a lower value is consistent with a steeper density profile in the inner regions probed by the interferometer, and could indicate collapse of the inner layers. In order to estimate the physical parameters of the dust condensation and the protostellar cores Cep E-A and B, we have adopted as dust temperature T dust the value estimated by Crimier et al. (2010) at the core radius (55 K at r = 480 au), and we have adopted a dust mass opacity κ 1.3 mm = 0.0089 cm 2 g −1 calculated by Ossenkopf & Henning (1994), specifically their OH5 dust model which refers to grains coated by ice. Under these assumptions, we obtain a beam-averaged gas column density N(H 2 )= (5.7 ± 1.0) × 10 23 cm −2 in a beam of 1.4 . From this gas column density estimate and the size of the condensation, we obtain the average gas density (nH 2 ) = (4.0 ± 0.8) × 10 7 cm −3 for the condensation. This value is in agreement with the gas density derived by Crimier model 4.3 × 10 7 cm −3 at the same radius of 480 au. From the integrated flux (233 mJy), we derive the mass of the condensation M = 0.83 ± 0.17 M .
Our NOEMA maps show that the millimeter flux is dominated by Cep E-A. Under the assumption that this source dominates the SED, we can use the Crimier model to estimate the physical parameters of Cep E-A. Proceeding in the same way as above gives T dust 60 K at the fitted core radius r = 440 au. We obtain N(H 2 ) = (5.2 ± 1.0) × 10 23 cm −2 and n(H 2 ) = (4.0 ± 0.8) × 10 7 cm −3 and a total dust mass M = 0.5 ± 0.1 M .
For Cep E-B, we constrain an upper temperature limit T dust,max = 60 K similar to that of Cep E-A. A lower limit to T dust is provided adopting the temperature derived in the Crimier model at the radial distance between components A and B (1.7 ): T dust = 25 K; hence, we estimate the gas column density N(H 2 ) = (1.3−4.0) × 10 23 cm −2 the gas density n(H 2 ) = (1.3−4.0) × 10 7 cm −3 and the mass M = (0.13−0.38) M .
The physical properties of both protostars are summarised in Table 3.

Complex organic molecules
The first hint on the presence of COMs in Cep E-mm was provided by López-Sepulcre et al. (2015) who reported the presence of formamide (NH 2 CHO). Following this study, we have carried out a systematic search for COM emission in the IRAM 30 m survey and the NOEMA observations. We searched for emission lines from the species identified in the solar-type protostar IRAS16293-2422 (Cazaux et al. 2003;Jaber et al. 2014) and the protostellar shock L1157-B1 (Lefloch et al. 2017). Line identification was carried out using the WEEDS package in GILDAS (Maret et al. 2011) and the public databases CDMS (Müller et al. 2005) and JPL (Pickett et al. 1998).

Content
The following complex O-bearing and N-bearing organic species are detected towards the Cep E-mm protostellar envelope: methanol (CH 3 OH) and its rare isotopologues 13 CH 3 OH and CH 2 DOH, acetaldehyde (CH 3 CHO), dimethyl ether (CH 3 OCH 3 ), methyl formate (HCOOCH 3 ), Acetone (CH 3 COCH 3 ), formamide (NH 2 CHO), methyl cyanide (CH 3 CN) and ethyl cyanide (C 2 H 5 CN). The number of lines detected with both instruments and the range of upper-level energies E up (K) are given in Table 4. A montage of illustrative transitions from each species observed with the IRAM 30 m telescope is displayed in Fig. 4. In this work, we also report the detection of the following species, chemically related to COMs: formaldehyde (H 2 CO), ketene (H 2 CCO), formic acid (HCOOH), isocyanic acid (HNCO).
Because of the limited spectral coverage of NOEMA observations and the lower sensitivity of the IRAM 30 m survey, there is only a partial match between the sets of detected lines detected with both instruments. We note that, apart from CH 3 CN (not observed with NOEMA), Propanone (CH 3 COCH 3 ) and Ethyl Cyanide (C 2 H 5 CN) were not detected by the IRAM 30 m telescope; and Vinyl cyanide (C 2 H 3 CN) was not detected by NOEMA. Our NOEMA observations show bright emission from highly excited molecular transitions, like HCOOCH 3 20 0,20 -19 0,17 (E up = 111.5 K) or CH 3 OH-A 6 1,−1 -7 2,−1 (E up = 373.9 K) which are not detected with the IRAM 30 m. This effect results from the dilution of the emission in the telescope main beam.
In order to estimate the amount of flux filtered out by NOEMA, we compared the fluxes of molecular transitions detected at 5σ. The results obtained are summarised in Table 5. We found that all the flux is recovered for lines from high E up , like CH 3 OH 8 0,0 −7 1,0 (E up = 88.7 K). For the other transitions of lower excitation, the interferometer recovered from 20 to 60% of the flux collected with the IRAM 30 m. This is illustrated in the case of H 2 CO J = 3 0,3 -2 0,2 (E up = 21 K), CH 3 OH J = 4 2,0 -3 1,0 (E up = 45.5 K) and HNCO 10 0,10 -9 0,9 (E up = 58.0 K) in Fig. 5. Figure 4 shows that a large variety of profiles is observed, depending on the molecular species and the line excitation conditions. For instance, narrow line components ( 1−3 km s −1 ) peaking at the cloud systemic velocity (ν lsr = −10.9 km s −1 ) are detected in HCOOH and CH 3 CHO. Broad velocity components extending over 20 km s −1 are also detected in the profiles of H 2 CCO or CH 3 CHO. We have identified three types of spectral line features, depending on the line width, the velocity range, and the emission peak velocity: a narrow line component (NL), a broad line component (BL), and the extremely broad line component (eBL).

Spectral signatures
The three components are detected in the low-excitation transitions of methanol, with E up < 50 K. The eBL and BL components are both found in transitions up to E up < 100 K, whereas only the BL component is detected in higher excitation transitions with E up > 100 K. This is shown in Fig. 6 in which we report the line width of each component as a function of the upper energy level E up for all methanol transitions, as derived from a Gauss fitting to the line profiles. Figure 7 shows three methanol transitions with E up = 28 K, 70 K, and 110 K, respectively, observed in the 2 mm band at approximately the same angular resolution (15.8 ). It turns out that the low-and mid-excitation line methanol emission (E up = 28 K and 70 K) is dominated by the contribution of the extremely broad line component, whereas the high-excitation line emission is dominated by the BL component.
We have applied the same decomposition to the line profiles of the other detected molecular species that are listed in Table 4. We describe their properties in more detail hereafter.

The extremely broad line (eBL) component
Methanol spectra show an eBL with a line width (FWHM) ∆ν 7 km s −1 (Fig. 6) and peaking at velocities up to ±10 km s −1 with respect to the source (ν lsr = −10.9 km s −1 ). The line emission can be reasonably well fitted by a Gaussian centred at a velocity shifted by 1-3 km s −1 from the source (Fig. 7, top and middle panel; also Fig. 4). It is detected in the transitions with E up ≤ 100 K, only. This component is detected with NOEMA and traces the cavity walls of the high-velocity outflow driven by Cep E-A. We estimate a typical size of 20 (see Lefloch et al. 1996Lefloch et al. , 2015. This component is also detected in the lines of formaldehyde, methyl cyanide, and acetaldehyde (see Table 4).

The narrow line (NL) component
An NL ∆ν 3 km s −1 peaking at the systemic velocity of the cloud, is detected only in the profiles of the low-excitation CH 3 OH lines (E up ≤ 50 K). This component is present in the line spectra of many COMs (CH 3 CHO, HCOOH, CH 3 CN, and H 2 CCO) with E up ≤ 60 K, when observed with the IRAM 30 m. The flux filtered by the interferometer is about 60-80% for these transitions (see Table 5). We note that the low velocity dispersion and the low excitation conditions of this gas component are consistent with an origin in the cold and quiescent outer envelope.

The broad line (BL) component
The BL is seen in all the methanol transitions detected with IRAM 30 m. This component differs from the bipolar outflow as (a) its emission peaks at the source systemic velocity, (b) its linewidth varies little with E up , and (c) it is detected in the highexcitation lines of methanol, with E up in the range 100-500 K (see Tables A.1 and A.2). We find a very good match between the highexcitation line profiles of methanol observed with the NOEMA and those observed with the IRAM 30 m. This confirms that we are actually detecting the same component. Interestingly, the line profiles of COMs, as observed with NOEMA, can be well fitted by a Gaussian with a line width of 5 km s −1 , similar to that of the high-excitation CH 3 OH lines.

Spatial distribution
We have produced emission maps of selected molecular transitions of COMs and chemically related species from the calibrated UV table adopting a natural weight. A montage of such maps is shown in Fig. 8. The emission of the various COMs appears to be compact and centred on the dust emission peak of Cep E-A. Interestingly, there is no molecular emission centered on Cep E-B, the second core component.
We have estimated the size of the COM emission region around Cep E-A for the following bright molecular lines: CH 3 OH 5 1,0 −4 2,0 and 6 1,5 −7 2,6 , CH 2 DOH 5 1,5 −4 1,4 and HCOOCH 3 20 0,20 −19 0,19 . We performed 1D Gaussian fits to the visibilities in the UV plane around the Cep E-A. The result of the procedure does not appear to depend significantly on the E up of the transition, which ranges between 36 and 112 K. We obtained a typical size of 0.7 ± 0.1 (FWHP) for all the transitions, that is, the emission region is marginally resolved.
As can be seen in Fig. 8, the emission of H 2 CO 3 2,2 -2 2,1 (E up = 68 K) appears slightly more extended than that of CH 2 DOH and other COMs. A fit to the H 2 CO 3 2,2 -2 2,1 distribution yields a size of 1.2 . For COMs like CH 3 OCH 3 ,  219 798 10 0,10 −9 0,9 58.0 50 CH 3 COCH 3 , and H 2 CCO 3 0,3 -2 0,2 , the intensity of the detected lines is much lower and the results of a UV visibility fitting procedure are too uncertain. In such cases, we estimated the size of the emitting region directly from a simple Gaussian fit to integrated intensity maps, deconvolved from the synthesized beam (1.4 ). The results display relatively little scatter, and we find sizes ranging between 0.6 and 0.8 . We note that the emission of the low E up transitions of CH 3 OH, H 2 CO, and HNCO also trace an extended component associated with the cavity walls of the outflow system (Lefloch et al. 2015).
To summarise, our NOEMA observations provide direct evidence for a hot corino region around Cep E-A. Estimates of the emitting regions of various COMs yield a typical size of about 0.7 , with little scatter. This size is close to the half power beam width of the synthetic beam, meaning that the hot corino region is only marginally resolved.

Molecular abundances in Cep E-A
In this section, we discuss the physical properties (excitation temperature, column densities) of the COMs and the chemically related species detected towards the hot corino of Cep E-A. The hot corino properties were obtained from a population diagram analysis of their spectral line energy distribution (SLED) or under the assumption of local thermodynamical equilibrium (LTE) when only one line was detected, as is the case for CH 3 CHO, CH 3 COCH 3 , HCOOH, NH 2 CHO, C 2 H 5 CN and HNCO. The source size, as determined in Sect. 4.2, was taken into account and the derived column densities are source averaged. In our population diagram analysis, line opacities are taken into account following the approach of Goldsmith & Langer (1999). Under the LTE assumption, in the absence of constraints on the excitation temperature, calculations were done for the following excitation temperatures T ex : 30 K and 60 K. The first value is similar to the determinations of T rot of COMs obtained when a population diagram analysis was feasible (see below; also Table 6), while the second value is consistent with the gas kinetic temperature predicted at the hot corino radius by Crimier et al. (2010; see also below Sect. 5.1). In that way, we obtain a plausible range of values for molecular column densities. For the molecular species detected with NOEMA, the SLED was directly measured from the WideX spectra at the location of Cep E-A. For the IRAM 30 m data, we first applied the spectral decomposition discussed in Sect. 4.2 in order to obtain the flux of the three physical components (bipolar outflow, cold envelope, hot corino) and then built the SLED of the hot corino. The list of all the identified transitions, along with their spectroscopic and observational properties are given in Appendix A. We have adopted a hot corino size of 0.7 for all the COMs and of 1.2 for H 13 2 CO, in agreement with our NOEMA observations.  Only the eBL (red) and BL (blue) components are detected. Bottom: E-CH 3 OH J = 9 0,0 -9 −1,0 (E up = 109.6 K) transition. Only the BL (blue) component is detected.
The rotational diagrams for E-/A-CH 3 OH, 13 CH 3 OH, CH 2 DOH, H 13 2 CO, CH 3 OCH 3 , and CH 3 CN are presented in Fig. 9. We have superimposed the flux obtained with NOEMA. Molecular abundances were derived adopting a total H 2 column density N(H 2 ) = 4.0 × 10 23 cm −2 based on our continuum A145, page 8 of 20 emission analysis (see Sect. 5.1). The results of our analysis (rotational temperature, source-averaged column density, and molecular abundances relative to H 2 ) are presented in Table 6.

Methanol
The population diagrams for A-/E-CH 3 OH are displayed in Fig. 9; they deviate strongly from linearity meaning that the SLED cannot be fitted by a single rotational temperature. The presence of excitation gradients in the source, line opacity effects, and radiative pumping (see e.g. Leurini et al. 2016) could account for this behaviour. We estimated the opacity of a few CH 3 OH transitions detected in the IRAM 30 m survey. We compared the main-beam temperatures of two sets of CH 3 OH and 13 CH 3 OH transitions of similar excitation, 3 0,+ -2 0,+ (E up = 14 K) and 5 2,− -4 2,− (E up = 73 K) adopting a typical 12 C/ 13 C abundance ratio of 70 (Langer 1992) and under the standard approximation that both molecules share a similar excitation temperature. We find τ 12 3 (20) for the 3 0,+ -2 0,+ (5 2,− -4 2,− ) transition. We conclude that the CH 3 OH emission is optically thick over a wide range of excitation conditions. Unlike the main isotopologue, the rotational diagrams of CH 2 DOH and 13 CH 3 OH are well fitted by a single temperature component.
The population diagram analysis of 13 CH 3 OH yields N( 13 CH 3 OH) = (7.0 ± 2.2) × 10 15 cm −2 and T rot = (27.3 ± 5.3) K. The opacities of the detected 13 CH 3 OH transitions were computed under LTE conditions, and were all found to be τ 13 0.2, therefore optically thin. As a consequence, we used the optically thin 13 CH 3 OH emission to determine the total methanol column density. Assuming a 12 C/ 13 C abundance ratio of 70, we obtain the total methanol column density N(CH 3 OH) = (4.9 ± 1.5) × 10 17 cm −2 .
Six and eight transitions of A-and E-CH 3 OH, respectively, with E up > 200 K were detected in the IRAM survey. The flux of these transitions cannot be accounted for by the component detected in the 13 CH 3 OH lines, and requires higher excitation conditions. A simple population diagram analysis in the range E up = 150-500 K, yields T rot = 440 ± 210 K, N = (1.3 ± 0.4) × 10 16 cm −2 and T rot = 460 ± 220 K, N = (4.2 ± 1.1) × 10 16 cm −2 for A-and E-species, respectively. One might note a discrepancy between the column densities of E-and A-CH 3 OH. However, taking into account the uncertainties on the derived column densities, both estimates are consistent within the 2σ level. The total CH 3 OH column density is N ∼ 5 × 10 16 cm −2 , which is one order of magnitude less than the amount of CH 3 OH traced by the rare isotopologues. Observations at high angular resolution and better sensitivity would help to clarify the nature of this component.

Other COMs and chemically related species
The population diagrams for the other COMs and some chemically related species (H 13 2 CO, H 2 CCO, CH 3 OCHO,CH 3 OCH 3 , CH 3 CN and HC 3 N) are presented in Fig. 9. As for 13 CH 3 OH, the population diagrams are reasonably well fitted by a straight line.
As can be seen in Table 6, all COMs display rather similar rotational temperatures in the range T rot = 20−40 K. No trend is observed as a function of the rotational dipole moment, which is not unusual (see e.g. Lefloch et al. 2017). The H 2 CO physical properties were estimated from its 13 C isotopologue following the same procedure as for CH 3 OH. We obtain the total formaldehyde column density N(H 2 CO)= (3.5 ± 1.4) × 10 16 cm −2 .
Because of the limited spectral coverage of the NOEMA observations, we could only detect one transition from a few COMs (see Table 4). As explained above, the column density and the molecular abundance were estimated for T ex = 30 K and T ex = 60 K. Whereas the first value is similar to the T rot values of COMs obtained from a population diagram analysis (see below; also Table 6), the second value is consistent with the gas kinetic temperature predicted at the hot corino radius by Crimier et al. (2010; see also below Sect. 5.1). This procedure allows us to  derive a range of column densities along with their uncertainties. As for HCOOH, CH 3 CHO, HCNO, and NH 2 CHO, the results do not vary significantly with the adopted value of T ex , and the differences lie within the statistical uncertainties. For these species, in Table 8 we have reported, as abundance, the average results between the 30 and 60 K values. On the contrary, in the cases of CH 3 COCH 3 and C 2 H 5 CN, large differences, up to a factor of three, are found when T ex varies from 30 to 60 K. For these species, we have reported the range of obtained values in Table 8. We have also reported in Table 9 the molecular abundances of COMs relative to that of HCOOCH 3 , a molecule which is often detected in young stellar objects (see e.g. Lefloch et al. 2018). The molecular column densities of O-bearing species are found to be relatively similar, of the order of a few times 10 15 cm −2 , except for HCOOH which is less abundant by one order of magnitude. By comparison, N-bearing species (CH 3 CN, C 2 H 5 CN) are less abundant with column densities lower than 10 15 cm −2 . We note that NH 2 CHO and HNCO, which both display a peptide bond NH-C=O, harbour abundances similar to N-bearing species.
To summarise, a simple population diagram analysis towards Cep E-A reveals a rich and abundant content in COMs, typical of hot corinos. This is the first detection of a hot corino around an isolated intermediate-mass protostar.

Physical structure of Cep E-mm
Our observations with NOEMA have shown evidence for a system of two protostars in the dust condensation associated with Cep E-mm. This is consistent with the work by Moro-Martín et al. (2001), who reported hints of multiplicity in Cep E-mm. These authors detected two sources in the dust core observed at 222 GHz, whose coordinates are in good agreement with those of Cep E-A and Cep E-B. They derived the physical parameters of both sources assuming a lower dust temperature (18 K) and dust mass opacity (κ = 0.005 cm −2 g −1 ) than the values adopted here. This results in a higher envelope mass 13.6 M and higher cores masses of 2.5 and 1.8 M for Cep E-A and Cep E-B, respectively. Taking into account their hypothesis on T dust and κ, our mass estimates are in good agreement with theirs.
As discussed in Sect. 5.1, we adopted a dust temperature T dust = 55 K and dust mass opacity κ 1.3 mm = 0.0089 cm −2 g −1 A145, page 10 of 20 in a region of 950 au (1.3 ), corresponding to the mean size of the Cep E-mm condensation. We derived a mean gas density n(H 2 ) = (4.0 ± 0.8) × 10 7 cm −3 which is in excellent agreement with the value predicted by Crimier et al. (2010). Our continuum observations are therefore consistent with the presence of warm dust and gas at 0.7 scale around Cep E-A. However, NOEMA reveals CepE-mm to be a binary system and the spherical symmetry hypothesis is no longer valid at 1.7 scale, the separation between both components (Fig. 1).

Chemical differentiation
Our COM emission maps show that the flux distributions are strongly peaked towards Cep E-A, while there is barely any molecular emission detected towards Cep E-B (Fig. 8). Many of the COM transitions detected towards Cep E-A are rather weak and are sometimes marginally detected (see Table A.5). Since component B displays a smaller size and a lower gas column density, it is worth investigating whether the apparent chemical differentiation could be biased by the sensitivity of the data. We have applied a simple scaling to the intensities of the COM transitions detected towards Cep E-A, taking into account the difference of the continuum source size and peak flux. This provides a reasonable approximation to the line fluxes towards Cep E-B assuming that the rotational temperature and the physical conditions are the same. We find that five transitions of methanol and two of HNCO should be measured above 5σ noise level but only the brightest methanol transition is detected towards Cep E-B. For all other COMs, the expected intensity then falls under the 3σ detection limit. Only the lack of CH 3 OH towards Cep E-B pro-    vides evidence for a different chemical composition in COMs.
Recently, very sensitive observations of Cep E-mm have been obtained as part of the NOEMA Large Program "Seeds Of Life In Space" (SOLIS; Ceccarelli et al. 2017) and confirm the chemical differentiation observed between protostars Cep E-A and B (Lefloch in prep.). Such a chemical differentiation has been reported in other multiple systems, like the low-mass protostars IRAS16293-2422IRAS16293- (2004a and IRAS4A (Santangelo et al. 2015;López-Sepulcre et al. 2017), and, recently, towards the intermediate-mass protostars NGC 2264 CMM3 . With four examples at hand, we speculate that it could be a general feature of multiple protostellar systems and not a "pathological anomaly". There is no systematic trend between the millimetre thermal dust and molecular line emissions. Towards IRAS16293-2422 and IRAS4A, a rich content in COMs is observed towards the source with the less massive continuum source. Towards NGC 2264 CMM3, it is the most massive continuum component that displays a rich molecular content. The case of Cep E-mm appears similar to the latter one. High-mass star forming regions (HMSFRs) also present a rich chemical diversity. One of the best known examples is provided by Orion-KL. This source harbours: (i) a dichotomy between the spatial distribution of complex O-bearing and complex N-bearing species, with the latter species probing the hotter gas (see, e.g. Guélin et al. 2008;Favre et al. 2011;Friedel & Widicus Weaver 2012;Peng et al. 2013;Brouillet et al. 2013;Crockett et al. 2014Crockett et al. , 2015 but also (ii) differences between supposed chemically related species (see, e.g. Favre et al. 2017;Pagani et al. 2017). Other examples are provided by W3(OH) (Qin et al. 2015;Nishimura et al. 2017) and SgrB2 (Belloche et al. 2008(Belloche et al. , 2013. In an observational study of four HMSFRs (Orion KL, G29.96, IRAS 23151+5912,   Fig. 10. Molecular abundances with respect to methyl formate towards low-mass hot corinos (yellow): IRAS2A (Taquet et al. 2015) and IRAS4A2 in NGC 1333 (Taquet et al. 2015;López-Sepulcre et al. 2017), IRAS16293B Coutens et al. 2016;Lykke et al. 2017;Calcutt et al. 2018); the intermediate-mass protostars (red): Cep E-A (this work) and NGC 7129 (Fuente et al. 2014); the high-mass hot cores (blue): G29.96 (Beuther et al. 2009), Orion KL Beuther et al. 2009) and SgrB2(N1) (Belloche et al. 2013). Sources are ordered from left to right by increasing luminosity. and IRAS 05358+3543), Beuther et al. (2009) showed that the properties of CH 3 OH can be easily accounted for by the physical conditions (temperature) in the cores, whereas the N-bearing species appear to be more selective as they are detected only towards the sources at the (evolved) hot core stage. Recently, in an ALMA study of the filamentary HMSFR G35.20, Sánchez-Monge et al. (2014) found that only three out of the six continuum cores of the filament display COM emission typical of hot cores. Several hypotheses have been proposed to account for the observed chemical differentiation. López-Sepulcre et al. (2017) proposed that the COM-rich protostar is either more massive and/or subject to a higher accretion rate, resulting in a lower envelope mass. Watanabe et al. (2017) suggest that the less massive protostar is related to a younger evolutionary stage in which the hot corino (hot core) is not yet developed, meaning that its dimensions are still very small. In Cep E-mm, the presence of high-velocity SiO jets provides evidence of active mass ejection around both protostars. The short dynamical timescales (500-1000 yr) also indicate that these ejections began recently, meaning that both sources are still in an early evolutionary stage. Incidentally, Lykke et al. (2015) found an apparent correlation between the source luminosities and the relative abundance of complex organic molecules in a sample of sources including high-mass protostars. The authors have suggested that this could be the result of the timescale and the temperature experienced by a source during its evolution. The sample of sources with evidence for chemical differentiation should be increased in order to confirm this observational trend.

Comparison with hot cores and corinos
In this section, we compare the molecular abundances of COMs and chemically related species in Cep E-A with those obtained towards hot cores and corinos, relative to H 2 and methyl formate. The latter is commonly observed in star-forming regions and it does not suffer the drawbacks that affect methanol. Firstly, methanol line emission at low energy levels (E up < 100 K) is often strongly contaminated by the outflow and the envelope, as shown in our single-dish analysis of CH 3 OH emission (Sect. 4.2). Secondly, in the case of interferometric observations, the amount of filtered flux can vary significantly, thereby introducing biases in the derivation of molecular abundances. Finally, methyl formate likely suffers less from opacity issues in comparison to methanol. In order to minimise any possible bias in source comparison, we have selected targets in which the emission of hot cores and corinos could be isolated from the other components contributing to the emission (outflow, envelope). Also, the selected sources have been investigated in a systematic manner so that a large dataset is available: the IM protostar NGC 7129 (Fuente et al. 2014), the LM protostars IRAS2A (Taquet et al. 2015) and IRAS4A2 Taquet et al. 2015) in NGC 1333 and IRAS16293B Coutens et al. 2016;Lykke et al. 2017;Persson et al. 2018;Calcutt et al. 2018), and the hot cores G29.96 (Beuther et al. 2009), Orion KL (Beuther et al. 2009;Pagani et al. 2017) and SgrB2(N1) (Belloche et al. 2013). For all the sources but SgrB2(N1), we made use of interferometric observations. The molecular content of SgrB2(N1) was obtained by Belloche et al. (2013) from a careful multi-component analysis of a line survey with the IRAM 30 m telescope, which allowed them to disentangle the contributions of the two hot cores N1 and N2 of SgrB2(N). In the case of IRAS4A2, the molecular abundances of CH 3 OH, H 2 CO, and H 2 CCO were taken from a previous study with the Plateau de Bure Interferometer by Taquet et al. (2015), as those species were not covered by the observations of López-Sepulcre et al. (2017). The angular resolution of their observations (2 ) did not allow Taquet et al. (2015) to disentangle the contributions of protostars A1 and A2. They estimated the total gas column density from thermal dust continuum emission, which is dominated by the protostar A1, whereas the molecular emission arises mainly from the protostar A2. Therefore, the abundances of CH 3 OH, H 2 CO, and H 2 CCO relative to H 2 as determined by Taquet et al. (2015), are underestimated. On the contrary, the molecular abundances relative to methyl formate are not affected by this bias. The physical properties of the parental cores (distance, luminosity, size) are summarised in Table 7.
The molecular abundances measured towards Cep E-A are summarised in Tables 8 and 9. In order to facilitate the readibility of Tables 8 and 9, we report only the molecular abundances. Abundance uncertainties are indicated by a thick black line in Fig. 10, which displays a graphical representation of the abundances of COMs with respect to methyl formate. We note that Fuente et al.
(2014), Beuther et al. (2009) and Belloche et al. (2013) did not provide any uncertainty for molecular abundances in NGC 7129, Orion KL, G29.96 and SgrB2(N1). Beuther et al. (2009) pointed out that the uncertainties on the source temperature and sizes imply that the derived column densities must be taken with caution. Therefore, we have adopted a typical relative uncertainty of ∼35% for all column density determinations in these sources, similar to our own determinations in CepE-A, but we warn that they could be larger for the high-mass sources.
Overall, the molecular abundances derived in Cep E-A agree within a factor of four at most with those measured towards the IM protostar NGC 7129 by Fuente et al. (2005Fuente et al. ( ,2014. Most of the O-and N-bearing species of our study (CH 3 CN, C 2 H 5 CN and HNCO) have very similar abundances. The only exceptions are CH 3 COCH 3 and HCOOH. The CH 3 COCH 3 abundance is higher towards Cep E-A by one order of magnitude. We note that this abundance determination also suffers from large uncertainties (see Sect. 4.4.2). On the contrary, HCOOH appears more abundant towards NGC 7129 by a factor 30.
Molecular abundances of the LM hot corinos IRAS2A and IRAS4A2 are also similar to those of Cep E-A. These conclusions are unchanged when comparing molecular abundances relative to methyl formate, as illustrated by the cases of H 2 CO, HCOOCH 3 or CH 3 OCH 3 (Fig. 10). Our results also agree with the tight correlation between CH 3 OCH 3 and HCOOCH 3 previously reported by Jaber et al. (2014). The molecular abundances of O-bearing species are ≈1% of the CH 3 OH abundance, whereas those of Nbearing species are lower by one order of magnitude, ∼0.1%.
The molecular abundances of O-bearing species appear rather similar towards LM and IM sources, independently of the luminosity. Towards HM sources, a large scatter is observed in the relative abundances of HCOOH and CH 3 CHO. Comparison of molecular abundances relative to methyl formate does not reveal any specific trend as a function of the source luminosity for Obearing species, from LM to HM. On the contrary, the abundances of N-bearing species display marked differences. As can be seen in Table 8, the abundances in G29.96, Orion KL, and Cep E-A are rather similar, whereas SgrB2(N) display higher abundances. Comparison of the molecular abundances relative to methyl formate reveals a trend, which is best illustrated by C 2 H 5 CN in Fig. 10, namely the relative molecular abundance increases with the source luminosity, from LM to HM. The same trend seems to be present in HNCO.
In summary, we find molecular abundances to be similar between Cep E-A , LM and IM hot corinos. The relative composition of O-bearing species with respect to methyl formate seems relatively independent of the source luminosity, contrary to C 2 H 5 CN, and perhaps other N-bearing species, which increases as a function of source luminosity.

Conclusions
We performed an unbiased spectral survey from 3 to 0.9 mm with the IRAM 30 m telescope. It was complemented with interferometric observations in the 3 and 1.3 mm bands with NOEMA at an angular resolution of 2.4 and 1.4 , respectively. We report bright emission of COMs and chemically related species towards the isolated IM protostellar core Cep E-mm. Our main conclusions are as follows.
Cep E-mm appears to be a protostellar binary system with components Cepe E-A and Cepe E-B separated by 1.7 (1250 au). Cep E-A dominates the bulk of the continuum emission and powers the long-studied high-velocity jet associated with HH377. Cep E-B powers another high-velocity jet which propagates in a direction almost perpendicular to the Cep E-A jet.
We found evidence for a hot corino in a region of 0.7 around Cep E-A. Cep E-B seems devoid of molecular emission at the sensitivity of our observations. We successfully identified three components in the molecular spectral signatures: (a) an extremely broad line (eBL) component associated with the outflowing gas; (b) a narrow line (NL) component associated with the cold outer envelope of Cep E-mm; (c) a broad line (BL) component which traces the signature of the hot corino.
Methanol emission is dominated by the outflowing gas in transitions up to E up < 100 K. At higher E up , the hot corino dominates over the outflow.
Overall, COM molecular abundances in the Cep E-A hot corino are similar to those measured towards other low-and intermediate-mass protostars. N-bearing species are one order of magnitude less abundant that O-bearing species.
Relative to methyl formate, molecular abundances of O-bearing species are rather similar between protostars, independently of the source luminosity. On the contrary, a good correlation is observed between the relative C 2 H 5 CN abundance and the source luminosity.  (21)   Notes. Intensities are expressed in units of T * A .