THE ARECIBO ARP 220 SPECTRAL CENSUS. I. DISCOVERY OF THE PRE-BIOTIC MOLECULE METHANIMINE AND NEW CM-WAVELENGTH TRANSITIONS OF OTHER MOLECULES

The broad emission feature covering all six 1₁₀–1₁₁ transitions of the C-band multiplet of methanimine is presented in Figure 1. In Figure 1(a), the rest frequencies of the transitions in the frame of Arp 220 were derived using a recessional velocity of 5373 km s⁻¹, as found for the western component of the OH megamaser emission of Arp 220, while the heliocentric velocity axis of Figure 1(b) is for the transition expected to be the strongest in the methanimine multiplet, that at a rest frequency of 5289.813 MHz (Godfrey et al. 1973). The total velocity width (FWHM) of this emission line is 270 km s⁻¹; we note that a large velocity width is observed for almost all molecular spectra in Arp 220 (e.g., Takano et al. 2005). Using the angular size of 0.27'' × 0.21'' measured for the strongest component of the formaldehyde (H₂CO) emission (Baan & Haschick 1995) as an upper limit for the angular size of the methanimine emission region, we derive a lower limit to the brightness temperature of ∼2800 K. Taking into account the six lines in the multiplet and their relative intensities, we calculate a reduced value of this lower limit to be ∼1000 K for the strongest component. This is similar to the methanimine decomposition temperature of 1300 K (Nguyen et al. 1996). If the solid angle of the emission is smaller than the above, for instance if the emission comes from a number of very compact components, then the brightness temperature could be much higher. We conclude that, as for formaldehyde (Araya et al. 2004), methanimine in Arp 220 is likely to be showing weak maser emission for this transition. We have been awarded time with the MERLIN array to observe this source at higher spatial resolution in order to determine the brightness temperature more accurately and to discover where exactly in the galaxy the emission is located.

Figure 2 shows the energy diagram for the v₂ = 1 direct l-type transitions of hydrogen cyanide (HCN) in the J = 1 to J = 6 vibrational levels. Our spectra of the J = 2, 4, 5, and 6 HCN transitions are presented in Figure 3. The J = 4, 5, and 6 transitions are detected in absorption against the continuum emission of Arp 220, and we place an upper limit at the 3σ level of 0.0025 for the optical depth of the J = 2 line. The first detailed study of this type of HCN transition was carried out for the Galactic proto-planetary nebula, CRL 618, by Thorwirth et al. (2003). These observers detected a number of the higher vibrational levels in this HCN ladder (J = 8–14). However, we note that the lower-energy transitions presented here seem not to have been previously detected in any celestial source. We also note that these lines represent a high excitation energy above the HCN ground state (e.g., 1067 K for the J = 4 line.)

Zoom In Zoom Out Reset image size

HCN is a well-known indicator of high gas density. Gao & Solomon (2004) demonstrated that a very strong linear correlation exists between the luminosities L_IR and L_HCN for mm-wave transitions of HCN, extending over a wide range of L_IR from normal galaxies to ULIRGs. They found the similar L_IR–L_CO correlation to have a less linear form, marking out L_HCN as the best tracer of dense molecular gas mass in galaxies, and hence of active star formation. However, Graciá-Carpio et al. (2008) present evidence that L_IR/L_HCN is systematically higher in (U)LIRGs than in normal star-forming galaxies. The relative line integrals from the J = 4 and 6 lines suggest an approximate excitation temperature of ∼150 K. These transitions represent high excitation, and given the complicated scenarios now emerging for the central region of Arp 220, it would be of great interest to ascertain the precise circumstances under which these absorptions arise.

Unlike the relatively "smooth" line profiles seen for the excited-OH lines (e.g. Figure 5), the HCN lines each show evidence for the presence of a number of discrete components. In fact, the central, strongest HCN component becomes increasingly dominant from J = 4 to 6. The absence of the J = 2 line requires explanation, as the predicted absorption line is expected to be an order of magnitude stronger than the 3σ limit we place above. It is highly improbable that the fraction of the continuum emission "covered" by the clouds producing the HCN absorption could have decreased by such a large amount between 4488 and 1347 MHz that the J = 2 line is rendered unobservable. Much more likely is that free–free absorption in the foreground ionized screen of this starburst galaxy greatly attenuates the J = 2 line. For this to be the case, an opacity of ≳2.25 at 1347 MHz (in the rest frame of Arp 220) would be required. This would imply an optical depth of ≳1.5 at 1630 MHz in the galaxy, as found for the spectra of a number of supernova remnants (SNRs) near the twin nuclei of Arp 220 by Parra et al. (2007). These authors attributed this high opacity to the combined presence of "a patchy FFA (free–free absorption) ISM with a median opacity of less than 1," plus absorption in the regions of ionized circumstellar mass-loss envelopes surrounding the SN progenitors. The HCN lines we see are expected to arise from star-forming regions of high gas density, and a detailed study of all possible lines in the v₂ = 1 direct l-type transitions of the HCN ladder would provide useful evidence concerning the properties of the foreground gas screen to these regions. Clearly, the spectrum of the HCN J = 3 line (with a rest frequency of 2693.3 MHz) will be crucial to such a study, and we will be observing this line during our 2008 campaign to complete the present observations. It should be noted that for the implied optical depths at 1347 MHz, even the J = 4 line will be reduced through FFA by ≳20%, which would reduce the above-derived excitation temperature to ∼120 K.

The energy levels for the various OH transitions are shown in Figure 4. In Figures 5–8, we present the λ6, 5, and 4 cm Λ-doublet transitions of the OH radical seen in absorption against the continuum emission of Arp 220. The relevant quantum numbers, rest frequencies, and other measured line parameters are presented in Table 2. The λ6 and 5 cm lines have been previously detected by Henkel et al. (1987, 1986) respectively.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Contrary to the findings of Henkel et al. (1987), we find that the 4660, 4751, and 4766 MHz lines have intensity ratios closely in agreement with their expected local thermodynamic equilibrium (LTE) values of 1:2:1.

The λ5 cm main lines presented here have a considerably higher S/N than the measurements of Henkel et al. (1986), but show a similar form. The expected LTE intensity ratios for the 6017, 6031, 6035, and 6049 MHz lines are 1:14:20:1, respectively. However, we measure ratios of 1:13:13:<0.4. The two main lines of the λ5 cm transition (6031 and 6035 MHz) are blended due to their large velocity widths and give the appearance of being "saturated," as suggested by their similar optical depths. This may be due to their having high optical depths, resulting in saturation of the lines against a continuum component containing of order 10% of the total flux density of Arp 220. However, we note that there is a low-velocity component to (presumably) the 6031 MHz absorption line, seen in Figure 6 at 6026 MHz, which if also present for the 6035 MHz line would contribute significantly to the frequency of the 6031 MHz line. The satellite lines are seen at a much lower S/N level. However, the greater strength of the 6017 MHz line relative to that at 6049 MHz is interesting. A similar effect was also found by Gardner & Martín-Pintado (1983) for four compact HII regions in our own Galaxy. While they saw the 6017 MHz lines enhanced above the LTE ratio to the main lines, the 6049 MHz line was weaker than predicted, and may even sometimes have appeared in weak emission.

Zoom In Zoom Out Reset image size

The two λ4 cm OH lines at 7761 and 7820 MHz (Figure 7) are detected for the first time in Arp 220, and show velocity widths similar to most other molecular species in the galaxy. The expected ratio for these 2Π_1/2, J = 3/2 main lines (7761:7820 MHz) is 1:1.8 in thermal equilibrium, very close to the derived ratio of the peak and integrated brightnesses for the lines of 1:1.89 (see Table 2). The satellite lines in this multiplet are expected to be present only at the 1σ level, and indeed are not detected. Considering the absorption lines in the 2Π_1/2 ladder of Figure 4 at 4750 and 7820 MHz, in thermal equilibrium the ratio of their integrated optical depths would imply an excitation temperature of about 88 K. In a detailed Large Velocity Gradient (LVG) study of their CO observations, Greve et al. (2006) find that in Arp 220, the spectra of low-density tracers such as CO (and OH) can only be fitted with a two-phase molecular ISM with the kinetic temperature of one being >30 K, while their mm-wave spectra for the high-density tracer molecules, HCN and CS, indicate kinetic temperatures of ∼50–70 K.

A very tentative detection of the 2Π_1/2, J = 5/2, F = 2–2 line of OH at 8135 MHz is shown in Figure 8. If confirmed by the addition of the remaining data to be acquired in this project, this would be at least twice as strong as expected for an excitation temperature of 88 K. Further, this line is expected to be only 0.7 times as strong as for the other 2Π_1/2,  J = 5/2 main line at 8189 MHz. While the 8189 MHz line is also possibly detected, contrary to expectations this would be at an even lower S/N level than for the 8135 MHz transition.

In Figure 9, we present a high S/N detection of an absorption line from our L-band spectrum. Despite the presence of nearby RFI caused by Glonass emissions (at ∼1605 MHz), the reality of this absorption line has been verified by its presence with similar optical depth in each of 13 individual spectra of Arp 220, but in none of the spectra of the bandpass calibrator. As demonstrated by the horizontal lines in Figure 9, there is an ambiguity as to the species responsible for this absorption, which could be either the 1639.5 MHz main line of ¹⁸OH or the 1638.8 MHz line of formic acid (HCOOH). Given the prevalence of OH in Arp 220, it would perhaps not be unreasonable to detect the presence of ¹⁸OH as well. However, since formic acid is relevant to the chemical origin of life, it would be most interesting to resolve this ambiguity.

As is seen in Figure 9, as well as the strong absorption line detected near 1611 MHz, an apparently weaker absorption is also seen near 1609 MHz, close to where the second, 1637.6 MHz, main line of ¹⁸OH should be found. If this feature were indeed to be real, the ratio of the peak intensity of the higher-frequency absorption to this would be 3.1:1. This is higher than the expected ratio of 1.8:1 were the pair to be the main lines of ¹⁸OH (Barrett & Rogers 1964) and in local thermal equilibrium. The frequencies of the peak depths of the two features are separated by an amount that would correspond to a rest frequency separation of about 2.05 MHz were they both to be due to absorption in Arp 220. The laboratory separation of the ¹⁸OH main lines is 1.939 ± 0.003 MHz (Lovas 1986), and hence given the weakness of the feature near 1609 MHz, the agreement is considered to be satisfactory.

If the absorption near 1611 MHz is indeed the higher-frequency component of the ¹⁸OH main lines, then its radial velocity is closer to that of the peak velocity found for other molecules in Arp 220 than would be the case were the detection to be of formic acid (see Table 2 and the horizontal bars displayed in Figure 9.) However, even for ¹⁸OH, this velocity of ∼5265 km s⁻¹ is significantly lower than that of the normal molecular peak velocity in Arp 220. In addition, its width of about 190 km s⁻¹ is narrower than for any other species that we detect. Were this absorbing gas to represent ¹⁸OH, we note that a velocity of 5265 km s⁻¹ is "allowed" in Arp 220 from the HI absorption study of Mundell et al. (2001). However, a peak velocity of 5135 km s⁻¹, appropriate for the formic acid transition, has little associated HI absorption. Nevertheless, if the absorption were to be due to ¹⁸OH at 5265 km s⁻¹ this would imply a remarkably high isotopic ratio of ¹⁸OH:¹⁶OH for a cloud at this velocity, especially so as this is well away from the peak velocity range of the OH megamaser line as seen at low angular resolution (Baan et al. 1982).

Given the proximity of the Glonass RFI, it is difficult to establish the origins of this absorption feature. However, ¹⁸OH would seem a more likely identification than formic acid. Confirmation of the reality of the second, weaker, absorption component near 1609 MHz would effectively establish this identification. The possibility that the feature represents absorption in our Galaxy from the 1612.231 MHz satellite OH line has to be very small given the high Galactic latitude of the line of sight (b = 53°.0), the implied large radial velocity (∼+185 km s⁻¹), and large line width (∼190 km s⁻¹).

In Figure 10, we present a possible detection of the 5₁–6₀ A⁺ methanol line in absorption. Although this line is apparently detected with an S/N of almost 6:1, and the entire 100 MHz band in which it is observed is basically RFI-free, the quality of the baseline in this case is rather poor. We await our remaining observations for confirming the reality of this line. We note that while the excited-OH absorption line at 7761.7 MHz (Figure 7: upper) has a somewhat lower S/N than this possible methanol detection, the combination of a better overall baseline, and the 10.5σ level detection of the associated line at 7820.1 MHz (Figure 7: lower) yielding the expected LTE intensity ratio (see Section 4.3), makes for a more solid detection in that case. We can certainly conclude that at the 1 mJy level (5σ) no methanol maser emission, such as is commonly detected for this transition from regions of massive-star formation in our own Galaxy, is seen in Arp 220.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Assuming the reality of the methanol absorption line, an excitation temperature for the transition of T_ex = 20 K, and an Einstein coefficient of A = 0.1532 × 10⁻⁸ s⁻¹ (Cragg et al. 1993), and a covering factor of unity, in LTE the column density of the lower energy level, N_6,0, is

$$\begin{equation} N_{6,0} = 1.34 \times 10^{15} \int \tau \delta v\ {\rm cm}^{-2}, \end{equation} \tag{ 1 }$$

where the integral ∫τδv is expressed in units of km s⁻¹.

An approximation for the total column density of methanol can be obtained assuming a Boltzmann distribution for the methanol energy-level population. If the relative abundances of the A and E species of this molecule are taken to be 1:1 based on their having a relatively small ground-state energy difference, then the total column density of methanol molecules is

$$\begin{equation} N = N_{6,0} 2/13 Q(T_{\rm ex}) \,e^{\frac{E_{6,0}}{kT_{\rm ex}}}, \end{equation} \tag{ 2 }$$

where Q(T_ex) is the partition function, taken to be 39.8 (D. Cragg, quoted in Houghton & Whiteoak 1995), and E_6,0 is the energy above the ground state of the 6₀A⁺ level. Thus, the presence of a 6668 MHz methanol absorption line in Arp 220 would imply a total column density for this molecule of $N_{\rm CH_{3}OH} \sim 2.5 \times 10^{17}$ cm⁻².