Infrared Detection of Abundant CS in the Hot Core AFGL 2591 at High Spectral Resolution with SOFIA/EXES^*

CS is a linear molecule and, like CO, it has a simple rovibrational spectrum with approximately equidistant lines separated by 2.0 cm⁻¹. We have detected 18 CS absorption lines with excitation energies ranging from 14 to 1317 K. For ¹³CO, 16 lines were detected with an energy range of 5–719 K and 8 lines of the ¹²CO v = 1–2 band are detected spanning an energy range of 3200–4234 K. Missing transitions coincide with strong telluric absorption lines or are blended with other hot core lines.

Figure 1 compares molecular line profiles between the sub-mm and IR in AFGL 2591. The CS lines reveal the presence of a single component at approximately −10 km s⁻¹. We have fitted those observed lines with a Gaussian line profile, leaving the width, peak position, and integrated strength as free parameters.

Zoom In Zoom Out Reset image size

The ¹²CO v = 0–1 rovibrational lines are saturated up until J = 9, and thus we have elected to focus here on the optically thin ¹³CO and ¹²CO v = 1–2 transitions. All components in the ¹²CO v = 0–1 line blend into one saturated line except for a broad high-velocity component indicative of a high-velocity outflow. ¹²CO v = 1–2 line profiles also show a single component at −10 km s⁻¹. The line parameters derived from the Gaussian fits are summarized in Table 1. The line widths have been de-convolved with the instrumental resolution.

Table 1.  CS and CO Line Parameters

Species	Transition	Wavelength ($\mu {\rm{m}}$)	El (K)	gl	Aij (s−1)	vlsr (km s−1)	${\rm{\Delta }}v$ (km s−1)	τ0	Nl (×1014 cm−2)
CS v = 0–1	R(3)	7.8212	14.1	7	7.2	−11.9±0.5	8.3 ± 1.5	0.031 ± 0.004	1.9 ± 0.4
	R(4)a	7.8116	23.5	9	7.4	−9.4 ± 0.5	6.1 ± 1.3	0.032  ± 0.004	1.6 ± 0.4
	R(5)	7.8020	35.3	11	7.6	−10.1 ± 0.5	7.7 ± 1.3	0.039  ± 0.004	2.3 ± 0.5
	R(7)a	7.7833	65.8	15	7.8	−10.4 ± 0.2	6.4 ± 0.6	0.056  ± 0.003	3.0 ± 0.3
	P(8)	7.9443	84.6	17	8.1	−10.9 ± 0.5	9.6 ± 1.5	0.055 ± 0.006	4.5 ± 0.9
	R(9)	7.7649	105.8	19	8.0	−9.5 ± 0.6	11.2 ± 1.8	0.051 ± 0.005	4.1 ± 0.5
	R(10)	7.7559	129.2	21	8.0	−11.1 ± 0.3	10.4 ± 0.9	0.061  ± 0.004	4.7 ± 0.5
	R(11)	7.7469	155.1	23	8.1	−12.1 ± 0.3	9.9 ± 0.8	0.060  ± 0.002	4.4 ± 0.4
	R(18)	7.6868	401.8	37	8.5	−12.0 ± 0.3	11.0 ± 0.8	0.063  ± 0.003	5.0 ± 0.6
	R(22)a	7.6545	594.5	45	8.7	−12.7 ± 0.4	10.7 ± 1.2	0.080  ± 0.004	6.1 ± 2.4
	R(23)	7.6467	648.5	47	8.8	−10.3 ± 0.9	8.4 ± 3.8	0.016  ± 0.003	4.2 ± 1.1
	R(24)a	7.6389	704.7	49	8.8	−5.2  ± 0.7	15.0 ± 1.8	0.060  ± 0.003	5.9 ± 1.9
	R(26)	7.6237	824.4	53	8.9	−8.4 ± 0.6	7.9 ± 2.0	0.048 ± 0.007	3.0 ± 0.7
	R(27)a	7.6162	887.8	55	9.0	−11.2 ± 0.6	4.9 ± 1.2	0.076 ± 0.006	3.6 ± 0.8
	R(28)	7.6088	953.4	57	9.0	−10.0 ± 0.4	6.8 ± 1.0	0.055  ± 0.008	3.1 ± 0.7
	R(29)	7.6015	1021.5	59	9.1	−10.9 ± 0.4	6.7 ± 1.2	0.078 ± 0.008	4.4 ± 0.8
	R(31)a	7.5871	1164.5	63	9.2	−8.0 ± 0.9	11.4 ± 4.1	0.024 ± 0.005	1.8 ± 0.8
	R(33)	7.5731	1316.8	67	9.2	−9.0 ± 0.6	5.2 ± 1.5	0.051  ± 0.009	2.5 ± 0.7
13CO v = 0–1	P(1)	4.7792	5.2	6	32.4	−9.4 ± 0.2	4.1	0.50 ± 0.04	72.4 ± 7.3
	P(2)	4.7877	15.8	10	21.5	−9.3 ± 0.1	4.1	0.57 ± 0.02	77.5 ± 3.6
	R(2)	4.7463	15.8	10	14.2	−9.7 ± 0.3	4.1	0.68 ± 0.07	67.0 ± 8.0
	R(3)	4.7383	31.7	14	14.8	−9.3 ± 0.1	4.1	0.70 ± 0.02	68.6 ± 2.7
	P(3)	4.7963	31.7	14	19.3	−9.3 ± 0.2	4.1	0.61 ± 0.04	79.0 ± 6.2
	P(4)	4.8050	52.8	18	18.2	−9.3 ± 0.2	4.1	0.57 ± 0.03	72.4 ± 5.2
	R(5)	4.7227	79.3	22	15.5	−8.3 ± 0.2	4.1	0.44 ± 0.06	38.6 ± 5.7
	R(6)	4.7150	111.0	26	15.8	−9.1 ± 0.3	4.1	0.31 ± 0.06	29.3 ± 5.9
	R(9)	4.6927	237.9	38	16.4	−12.1 ± 0.7	11.2	0.14 ± 0.01	23.6 ± 3.1
	P(9)	4.8501	237.9	38	16.4	−13.2 ± 0.8	11.2	0.10 ± 0.01	18.3 ± 2.6
	R(10)	4.6853	290.8	42	16.5	−10.5 ± 0.6	11.2	0.10 ± 0.01	17.5 ± 2.5
	R(11)	4.6782	349.0	46	16.7	−12.0 ± 0.5	11.2	0.11 ± 0.01	18.9 ± 2.3
	P(11)	4.8689	349.0	46	16.1	−13.0 ± 0.6	11.2	0.05 ± 0.01	10.0 ± 1.8
	R(12)	4.6711	412.3	50	16.8	−12.0 ± 0.6	11.2	0.09 ± 0.01	16.9 ± 1.9
	R(13)	4.6641	481.1	54	16.9	−12.0 ± 0.4	11.2	0.09 ± 0.01	15.9 ± 1.4
	R(16)	4.6437	718.8	66	17.3	−10.8 ± 0.5	11.2	0.10 ± 0.01	16.6 ± 1.8
12CO v = 1–2	R(6)	4.6675	3199.5	13	33.2	−9.6 ± 0.6	17.9 ± 1.4	0.102 ± 0.006	13.1 ± 1.6
	R(7)	4.6598	3237.9	15	33.7	−8.2 ± 0.7	19.9 ± 1.9	0.108 ± 0.008	15.6 ± 2.6
	R(8)	4.6522	3281.8	17	34.1	−9.1 ± 0.4	18.3 ± 0.9	0.124 ± 0.006	16.4 ± 1.4
	R(9)	4.6448	3331.0	19	34.4	−8.6 ± 0.4	17.9 ± 0.9	0.113 ± 0.005	14.8 ± 1.2
	R(17)	4.5887	3922.5	35	36.6	−8.5 ± 0.3	16.6 ± 0.9	0.104 ± 0.005	12.7 ± 1.1
	P(18)	4.8948	4020.9	37	31.7	−7.1 ± 0.3	13.9 ± 0.9	0.083 ± 0.004	9.4 ± 0.9
	P(19)	4.9054	4124.8	39	31.4	−9.9 ± 0.8	14.6 ± 3.8	0.010 ± 0.009	12.0 ± 1.9
	R(20)	4.5694	4234.4	41	37.2	−6.1 ± 0.5	16.0 ± 1.2	0.088 ± 0.005	10.3 ± 1.5

Notes. ${\rm{\Delta }}v$ is the line FWHM, τ₀ is the optical depth at line center, and N_l is the column density in the lower energy level of the transition, with energy E_l. A_ij is the Einstein A coefficient for the transition and g_l is the statical weight.

^aThese lines suffer from systematic errors such as poor baselines, therefore there is a larger uncertainty in the continuum placement. Line data were taken from the high-resolution transmission molecular absorption (HITRAN) database (Gordon et al. 2017).

Download table as:  ASCIITypeset image

The ¹³CO lines show complex profiles with multiple velocity components (Figure 1), one of which coincides in velocity with the CS lines and ¹²CO v = 1–2 transitions. Two temperature components (hot and cold; see Figure 2) are observed at this velocity. For the hot component (high J), we adopted the CS line width, but the line width for the cold component (low J) had to be reduced (to 4.1 km s⁻¹). Absorption by the narrower cold component overwhelms the contribution by the warm component at the lower J-levels.

Zoom In Zoom Out Reset image size

The detected, unblended lines of CS and CO are presented in Figure 3 in the Appendix. There is a spread in centroid velocity for a given species. The line widths for CS and hot ¹³CO are in agreement for equivalent energy level.

The rotation diagrams of CS, ¹³CO, and ¹²CO v = 1–2 are shown in Figure 2. In all cases the rotation diagrams are straight lines, indicating that local thermodynamic equilibrium (LTE) is a good approximation. We have calculated the vibrational excitation temperature of ¹²CO from the v = 1–2 transitions and the warm ¹³CO column density assuming a ¹²CO/¹³CO abundance ratio of 60 (Wilson & Rood 1994). The derived vibrational excitation temperature of 625 ± 56 K agrees well with a similar estimate by Mitchell et al. (1989). The similar values for the rotational and vibrational temperatures suggest vibrational LTE.

The excitation temperature and velocity for all three species (CS, ¹³CO high J, and ¹²CO v = 1–2) are very similar. This supports an origin in the same region. The physical conditions for the detected species are summarized in Table 2. An estimate of the column density of hot ¹²CO from the lines of vibrationally excited ¹²CO gives 1.5 ± 0.6 × 10¹⁸ cm⁻². We derive an upper limit on CS in the −33 km s⁻¹ velocity component of <8 × 10¹⁴ cm⁻².

From our observations we derive a different systemic velocity than from sub-mm studies of AFGL 2591 (Figure 1) in CS, ¹³CO, and ¹²CO v = 1–2 (−10 km s⁻¹ compared to −5.5 km s⁻¹; Bally & Lada 1983). van der Tak et al. (1999) studied AFGL 2591 at sub-mm, radio, and IR wavelengths, at lower resolution, and derived a centroid velocity for ¹³CO absorption of −5.5 km s⁻¹, in agreement with the systemic velocity of rotational emission lines, including CS (Boonman et al. 2001; Benz et al. 2007; Kaźmierczak-Barthel et al. 2014). However, their IR transitions show a spread in velocity of −3.5 to −12.4 km s⁻¹ which they attribute to atmospheric interference, picking only a few lines to derive their value of −5.5 km s⁻¹.

The intrinsic line widths that we observed are broader than for sub-mm studies (Figure 1). As an example, the CS J = 10–9 transition has a width of 3.5 km s⁻¹, whereas our CS rovibrational transition v = 0–1 R(10) has a width of 10.4 km s⁻¹. The H₂O 3(1, 2)−3(0, 3) line has a width of 3.1 km s⁻¹ in the sub-mm compared to 16.5 km s⁻¹ for the ν₂ = 0–1 2(2, 1)–3(1, 2) IR line.

We suggest that our higher-temperature, higher-velocity IR observations trace more turbulent gas closer to the central source, probing deep into the base of the outflow, whereas sub-mm observations trace more extended gas at the velocity of the quiescent envelope.

The CS, ¹³CO v = 0–1, and ¹²CO v = 1–2 temperatures are very similar and, given the difference in critical densities, suggest gas in LTE. The critical densities for J = 7–6 are 2 × 10⁵ cm⁻³ (Yang et al. 2010) and 3 × 10⁷ cm⁻³ (Turner et al. 1992) for CO and CS, respectively. Therefore, we estimate that the density must exceed 10⁷ cm⁻³ in order to maintain LTE. The critical density of ¹²CO v = 1–2 cannot be used due to the high optical depth of the ¹²CO v = 0–1 transitions, which may lead to line trapping and tends to increase the vibrational excitation temperature.

Assuming a ¹²CO/H₂ abundance of 2 × 10⁻⁴ (Lacy et al. 1994) and constant density of 10⁷ cm⁻³, we derive a physical size for the absorbing region of <130 au. At a distance of 3.3 kpc (Rygl et al. 2012) this corresponds to <0.04''.

The velocity structure of the inner region (down to 500 au) of the hot core in AFGL 2951 has been well resolved and modeled by Wang et al. (2012). A high-velocity gradient is found for SO₂ and the velocity field of the blueshifted component is observed to have the highest negative velocities toward the center on the source. However, the highest velocity that is reached is ∼−7 km s⁻¹, at the very center. This further indicates that CS at −10 km s⁻¹ would not be observed at sub-mm wavelengths. Combined with the fact that SO₂ is a molecular tracer for shocks or outflows (Schilke et al. 1997), Wang et al. (2012) concluded that the SO₂ emission originates in an interactive layer between the upper parts of a disk-like structure and the outflow.

van der Tak et al. (1999) concluded that the outflow of AFGL 2591 points toward us along the line of sight. In order to explain CS (7–6) emission, Bruderer et al. (2009) proposed that their observations trace dense walls of the outflow, at a velocity of around −5.5 km s⁻¹. Given that the extreme velocities originate in the center of the source, and that our EXES observations show an average velocity around −10 km s⁻¹, we propose that we are looking into the outflow, probing deeply toward the protostar to the base of the outflow. Also the broad line profiles of CS are indicative of a shocked region. The high temperature implies that the gas lies close to the protostar, further supporting the proposed origin as the base of the outflow.

We derive a CS/¹²CO abundance of 8 × 10⁻³ and CS/H₂ abundance of 2 × 10⁻⁶, assuming a CO/H₂ ratio of 2 × 10⁻⁴. This abundance is two orders of magnitude higher than sub-mm observations of the hot core and envelope (van der Tak et al. 2003; Jiménez-Serra et al. 2012). This, once again, illustrates that the IR observations probe a very different region in this source. Consequently, hot CS contains 6 ± 0.8% of the cosmic sulfur budget, assuming an S/H ratio of 1.3 × 10⁻⁵. Compared to the Orion Hot Core, which has a CS/CO abundance of 4.1 × 10⁻⁴ derived from sub-mm observations (Tercero et al. 2010), the CS abundance in the gas traced by the IR observations of the hot core in AFGL 2591 is much higher.

Chemical models based on sub-mm observations (Charnley 1997; Doty et al. 2002) derive a CS abundance of the order 1 × 10⁻⁸ with respect to H₂. A two phase, time-dependent gas-grain model (Viti et al. 2004) has been used to study the sulfur chemistry in the Orion Hot Core, and predicts the abundance of CS as a function of hot core age (Esplugues et al. 2014). The first and second phase simulate depletion onto, and sublimation from, grain surfaces, respectively. For a hot core with a mass of 10 M_⊙, solar sulfur abundance and a density of 10⁷ cm⁻³, a CS/CO abundance of around 5 × 10⁻³ is achieved after 6 × 10⁴ yr.

The abundances of CS, H₂CS, and SO₂ increase such that these species become the most abundant sulfur-bearing molecules for an evolved hot core (Esplugues et al. 2014). The main chemical pathway that is responsible for the production of CS in this model is CH₂ + S → CS + H₂. Chemical models predict large amounts of atomic sulfur at high temperatures (Doty et al. 2002). This arises due to abstraction of H₂S, which is efficient at high temperatures. At the same time, small hydrocarbons are formed by the breakdown of CO via cosmic ray ionization.

Therefore, we propose two scenarios to interpret the CS abundance. The first is that AFGL 2591 is a more evolved hot core in which all sulfur is converted to H₂S on grain surfaces, and then converted back to S in the gas-phase after ice mantle sublimation. Then, at long timescales, enough CS is produced in the hot inner region of the hot core to explain our observations (Esplugues et al. 2014). These models are not optimized to AFGL 2591, which would likely have an effect on the CS abundance, therefore a more in-depth study using chemical models would be necessary to clarify the proposed timescales. We note that H₂S ice has not been observed in absorption toward massive protostars (Smith 1991) at abundance upper limits a factor of 7 lower than that of CS, putting H₂S as a source of the gas phase S into question. Deeper searches for H₂S would be very helpful to asses the sulfur budget in interstellar ices.

The alternative scenario is that our observations trace a disk-wind interaction zone very close to the protostar. In this case, the cosmic ray ionization rate would be high, which would favor the breaking of C out of CO, which could lead to the enhancement of CS production. Again, atomic sulfur is produced via abstraction of H₂S. May et al. (2000) found that grain sputtering becomes important in shocks around 15 km s⁻¹, therefore sulfur could also be released from grain mantles in the presence of shocks.

The conditions and chemical history of AFGL 2591 are clearly favorable for the production of CS. A high CS/CO abundance of 7 ± 0.4 × 10⁻³ has also been observed at MIR wavelengths in NGC 7538 IRS 1 (Knez et al. 2013) with TEXES. Nevertheless, the enhanced CS abundance is still not enough to appoint this molecule as the main reservoir of sulfur in hot cores. A high abundance of warm SO₂ has also been observed with EXES in the hot core Mon R2 IRS 3. (Dungee et al. 2018). This also is not observed in the sub-mm, suggesting that a large amount of sulfur is visible only at IR wavelengths.