CONSTRAINTS ON OH MEGAMASER EXCITATION FROM A SURVEY OF OH SATELLITE LINES

For purposes of comparing detections and non-detections of lines that may appear in absorption or emission, we report velocity integrated line fluxes (Jy km s⁻¹), rather than isotropic line luminosities. Line luminosities are useful for comparing the same line between different sources, while integrated line fluxes make comparisons of different lines in the same source more direct. Our method for defining upper limits on integrated line fluxes, denoted by F, for the 1612 MHz, 1665 MHz, and 1720 MHz lines is analogous to, but slightly less conservative than, that used in Darling & Giovanelli (2000, 2001, 2002) with

$$\begin{equation} F = \sigma \; \Delta \nu _{1667}. \end{equation} \tag{ 1 }$$

Here, σ represents the rms error in the spectrum at the expected location of the line. Δν₁₆₆₇ is a measure of the width of the 1667 MHz line, defined as the frequency width in which 75% of the line flux is contained. It is so defined to provide a more flexible measure of line width than the FWHM for the complicated profiles often seen in OHMs, but gives the same answer for the case of a Gaussian profile. Effectively, this upper limit is a factor of 1.5 smaller than the Darling & Giovanelli (2000) definition, and allows for the incorporation of information from the 1667 MHz detection by determining the upper limit on satellite line emission.

IRAS F01417+1651 (III Zw 35). The 1720 MHz detection agrees reasonably well with that published in BHH89, so we do not reproduce it here. RFI plagues the 1612 MHz spectrum, which prevents a detection, and results in relatively uninteresting upper limits.

IRAS F02524+2046. Darling & Giovanelli (2002) detected this OHM, noting multiple matched components in the 1665 MHz and 1667 MHz lines. They provided hyperfine ratios for individual narrow features in the spectrum, with values R_H = 1.4, 5.63, 1.88 from high velocity to low. In our observations, shown in Figure 1, it is only possible to distinguish two peaks in the 1665 MHz emission that align with peaks in the 1667 MHz emission. The highest velocity feature is quite unusual, in that the 1665 MHz feature is so strong. The peak fluxes of the two lines are equal, and the ratio we find for the integrated flux at velocities 54,260–54,360 km s⁻¹ is R_H = 1.3, consistent with what Darling & Giovanelli (2002) found. The hyperfine ratio is even smaller on the blue side of the line, as the 1665 MHz is moderately narrower than the 1667 MHz line and centered at a lower velocity.

Zoom In Zoom Out Reset image size

The observed hyperfine ratio in this velocity range is consistent with the range of 1–1.8 that is expected for thermal emission. To explore this possibility, we follow the example of Baan et al. (1982) in considering the required number of OH molecules to produce the observed line strength if it is optically thin thermal emission. Adjusting the equation they used for our definition of Stokes I, the total number of OH molecules is

$$\begin{equation} N = 1.3 \times 10^{58} \alpha \left(\frac{D}{{\rm Mpc}}\right)^2 \left(\frac{F}{{\rm Jy\;}{\rm km\; s}^{-1}}\right), \end{equation} \tag{ 2 }$$

where α takes the values of 1 and 1.8 for 1667 MHz and 1665 MHz emission, respectively, D is the distance in Mpc, and F is the flux of the line integrated over velocity in units of Jy km s⁻¹. The required number of OH molecules is roughly N = 10⁶⁴ for each line, or a mass in OH of ∼1.5 × 10⁸ M_☉. For an OH abundance n(OH)/n(H₂) of the order of 10⁻⁶, this implies unrealistic quantities of molecular gas. Instead, the emission must include a significant non-thermal contribution, despite the hyperfine ratio that is consistent with thermal emission.

The modeling of LE08 found that 1665 MHz maser emission was roughly as strong as 1667 MHz emission for line widths ΔV ≃ 2 km s⁻¹, and the 1665 MHz line was stronger than the 1667 MHz line for narrower line widths. This was used to explain the observation that the 1667 MHz line is always stronger in OHMs, while the 1665 MHz line is typically stronger in Galactic OH masers, which have narrower lines. The total line width of the feature is considerably broader than that, and with this mechanism would require many narrow features of comparable strengths that have blended together.

This OHM is notable in another respect, as it is among the most luminous known OHMs. The two most luminous OHMs, IRAS F14070+0525 (Baan et al. 1992a) and IRAS F12032+1707 (Darling & Giovanelli 2001), both have such broad profiles that it is not possible to distinguish the two main lines. Baan et al. (1992a) nevertheless highlighted two pairs of features in the blended spectrum of IRAS F14070+0525 that corresponded to the frequency separation of the two main lines, which would be consistent with strong 1665 MHz emission. The next most luminous OHM is IRAS F20100–4156 (Staveley-Smith et al. 1989), for which only an upper limit on 1665 MHz emission could be placed. IRAS F02524+2046 follows in this list of most luminous OHMs, and displays the unusual hyperfine ratio in part of its spectrum that we have discussed. The limited evidence regarding the most luminous OHMs is thus mixed, but hints at differences from the less luminous OHM population.

IRAS F10173+0829. To the best of our knowledge, we present the first detection of 1720 MHz emission in this OHM. Baan et al. (1992b) noted that Mirabel & Sanders (1987) had previously looked for satellite lines in IRAS F10173+0829, but discussion of this OHM in that paper is limited to the 1667/1665 MHz lines. Our detection of the 1720 MHz line is shown in Figure 2, along with the moderately stronger 1665 MHz line. The 1720 MHz emission is produced at velocities 14,600–14,700 km s⁻¹, corresponding to the weaker of the two peaks in the 1667 MHz emission. There is negligible 1720 MHz emission, if any, at the same velocity as the stronger of the 1667 MHz peaks. The 1667:1665:1720 line ratio at the lower velocity peak is roughly 30:5:2. In this region, the 1720 MHz emission is considerably stronger than would be expected for equal excitation temperatures in the lines. In the higher velocity 1667 MHz peak, the hyperfine ratio is R_H ∼ 25, and no 1720 MHz emission is visible. The limit on 1720 MHz in the higher velocity range does not rule out equal excitation temperatures of the lines in this region. When viewed over the entire range of velocities, the line ratios are consistent with equal excitation temperatures.

Zoom In Zoom Out Reset image size

IRAS F14043+0624. Darling & Giovanelli (2002) discovered this OHM and noted that it had an anomalous hyperfine ratio, with R_H = 1.4, while pointing out weak absorption at the edge of the 1665 MHz line. They suggested that a stable source of RFI could be present and produce the anomalous ratio. While we did see a strong, narrow spike of RFI near 1500 MHz, there was no evidence for RFI at the redshifted locations of the main lines between 1496–1499 MHz. RFI is often strongly polarized, but there is no structure in the Stokes Q, U, or V spectra for IRAS F14043+0624 at the frequency of the OH lines. Another line of evidence disfavoring RFI as the cause of the feature is that none of the spectra of other sources at this frequency during our observations showed any evidence of RFI. Nevertheless, the weak absorption that Darling & Giovanelli (2002) observed at the edge of the 1665 MHz feature in their spectrum was not visible in our observations. We cannot rule out some form of RFI near the expected location of the 1665 MHz line of this galaxy, even if none of our tests for RFI within our data uncovered evidence for RFI.

Putting potential RFI issues aside, the hyperfine ratio we observed is R_H = 1.1, even smaller than that which Darling & Giovanelli (2002) reported. Figure 3 shows our spectra of the main lines of IRAS F14043+0624, and reveals line shapes that are different, with the centers of the two lines offset by approximately 100 km s⁻¹. These differences suggest that the hyperfine ratio may not actually be a meaningful measurement. The observations of the satellite lines do not help resolve the mystery, as spectra since both lines are affected by RFI.

Zoom In Zoom Out Reset image size

IRAS F15107+0724. This OHM was reported in Baan et al. (1987, hereafter BHH87), and is among the least intrinsically luminous OHMs, with an isotropic luminosity of ∼15 L_☉. In our observations, all four 18 cm OH lines are detected (Figure 4). Emission in the 1612 MHz line falls predominantly between 3910 and 3980 km s⁻¹, and at its peak it is nearly as strong as the 1665 MHz line at the same velocity. Over this same velocity range, the 1720 MHz line appears in absorption, with a similar line shape, but at a weaker level than the 1612 MHz emission. The areas of conjugate line shape represent competition between the 1612 MHz and 1720 MHz transitions for pumping photons.

Zoom In Zoom Out Reset image size

The mechanism for producing conjugate emission is asymmetry in pumping that results from quantum mechanical selection rules in the OH rotational transition ladder, shown in Figure 5. The 18 cm OH maser lines result from hyperfine transitions in the ²Π_3/2(J = 3/2) level, which is the ground state. The 1720 MHz line is produced by a transition from an F = 2 to F = 1 state, while the 1612 MHz line is a transition from F = 1 to F = 2. Transitions between rotational levels are permitted when |ΔF| = 0, 1. The ²Π_3/2(J = 5/2) (119 μ above the ground state) has hyperfine levels with F = 2, 3, and thus will preferentially populate F = 2 levels in the ground state. The result is 1720 MHz inversion, and anti-inversion of the 1612 MHz line. A cascade from ²Π_1/2(J = 1/2) (79 μ above the ground state), which has hyperfine levels with F = 0, 1, will overpopulate the F = 1 levels in the OH ground state, producing 1612 MHz inversion and 1720 MHz anti-inversion.

Zoom In Zoom Out Reset image size

For either of these pumping mechanisms, the FIR transition must be optically thick. When both FIR lines are optically thick, the 1612 MHz inversion dominates (Elitzur 1992). To produce 1720 MHz inversion and conjugate 1612 MHz absorption requires the ²Π_5/2(J = 3/2) transition to be optically thick and the ²Π_1/2(J = 1/2) transition to be optically thin. The result is that for a column density per velocity interval just below 10¹⁵ cm⁻² km⁻¹ s, the 1720 MHz line is inverted and the 1612 MHz line anti-inverted. The reverse behavior occurs just above 10¹⁵ cm⁻² km⁻¹ s (van Langevelde et al. 1995).

In the LE08 model, 53 μm radiative pumping of the main lines begins to occur at similar column densities per velocity interval at which the conjugate satellite lines are produced. At velocities of 3900–4000 km s⁻¹, the ratio of line to continuum flux for the 1667 MHz line gives an optical depth of −0.3, while the ratio of the 1667 MHz to 1665 MHz line gives an optical depth of −0.8. These are both consistent with weak amplification of the OH lines. While the conditions to produce the main lines and the conjugate satellite lines appear similar, the line overlap needed to produce the inversion of the main lines is not compatible with simultaneously producing conjugate satellite lines. Thus, within the same range of velocities in IRAS F15107+0724, there must be two separate OH inversion mechanisms acting.

Conjugate OH satellite lines have been observed before in a diverse group of galaxies. Main line OH masing and absorption is seen in the starburst galaxies Messier 82 (Seaquist et al. 1997) and NGC 253 (Frayer et al. 2007); in Centaurus A (van Langevelde et al. 1995), the 18 cm OH main lines appear in absorption; in the distant radio galaxy PMN J0134–0931 (Kanekar et al. 2005), the main lines also appear in absorption; and in another radio galaxy, PKS 1413+135 (Darling 2004; Kanekar et al. 2004), the main lines were not detected. Our detection represents the first such example of conjugate emission in what otherwise appears to be a typical OHM, albeit one at the low end of the luminosity distribution. Both M 82 and NGC 253, which have lower FIR luminosities than IRAS F15107+0724, are examples of kilomasers. HW90 argued that kilomasers represent a transition between powerful OHMs in LIRGs, and the OH absorbers found in more typical star-forming galaxies. The observations of IRAS F15107+0724 suggest that the transition from powerful OH masing to a mixture of masing and absorption may also feature increased satellite line strengths. The gain for each of the lines is low though, meaning moderately bright radio continuum emission is also required.

IRAS F15327+2340 (Arp 220). The 1612 MHz and 1720 MHz lines were first reported in BHH87, and the line ratios in different components and regions were discussed in detail. They concluded that there were differences in the excitation of the lines in each of the regions. Our 1612 MHz spectrum included features that could be astrophysical emission, but strong RFI produced serious structure in the bandpass that could not be removed. Parts of the line structure look quite similar to that published in BHH87, but, given the superior quality of their spectrum, we do not provide our spectrum here. The 1720 MHz emission we see is consistent with that previously reported in BHH87.

IRAS F17207−0014. Our redetection of 1720 MHz emission agrees nicely with that published in BHH89, so it is not shown again here.

IRAS F20550+1655. BHH89 noted a tentative detection of the 1612 MHz line with a strength 10 mJy at a velocity 55 km s⁻¹ above that of the 1667 MHz line. We confidently rule out a line of this strength, as our spectrum has an RMS error ∼3 mJy (∼6 mJy in the "classical" definition), and no hint of a line is seen at the velocity given in BHH89.

The explanation of OHM emission provided by HW90 assumed that the excitation temperatures of the OH 18 cm main lines are roughly the same. Multiple lines of evidence were provided for this assumption, including observations of other OH lines in the rotational ladder and earlier radiative transfer modeling by Henkel et al. (1987). Equal excitation temperatures of the 18 cm OH lines was a result of the LE08 model calculations, which they noted occurred because of the line overlap produced by intrinsic line widths of ΔV ⩾ 10 km s⁻¹. Observations of 18 cm OH satellite lines in OHMs provide an important test of equivalent excitation temperatures, as the lines should each be amplified according to their LTE line ratios. Thus, the observed line ratios should relate to the optical depth of the 1667 MHz line, τ, as

$$\begin{eqnarray} R_{\rm H}& = \frac{e^{-\tau } - 1}{e^{-\tau / 1.8} - 1}, \end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray} R_{\rm 1612}& = \frac{e^{-\tau } - 1}{e^{-\tau / 9} - 1}, {\rm \; and} \end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray} R_{\rm 1720}& = \frac{e^{-\tau } - 1}{e^{-\tau / 9} - 1}. \end{eqnarray} \tag{ 5 }$$

LE08 plotted R_H and R₁₇₂₀ (in their Figure 6) for the four OHMs then known with detected 1665, 1667, and 1720 MHz emission, calling it a "color–color" diagram for OH maser lines. They found very nice agreement between the observed ratios and the expectation for equivalent excitation temperatures. We reproduce this plot, including all OHMs in the literature, our detections, and upper limits on the non-1667 MHz lines (meaning lower limits on the line ratios), for both the 1720 MHz line (Figure 6(a)) and the 1612 MHz line (Figure 6(b)).

Zoom In Zoom Out Reset image size

The large number of non-detections makes a detailed test of equal line excitation impossible. Even so, some fraction of the lower limits on the ratios of 1667 MHz emission to satellite line emission are physically interesting. While IRAS F15107+0724 has 1612 MHz emission that is considerably stronger than would be expected for equal line excitation temperatures, it is the only such example in the survey, and a few of the other sources have lower limits on the 1667/1612 ratio that preclude stronger-than-expected 1612 MHz emission of the nature seen in IRAS F15107+0724.

For the 1720 MHz line, the sources detected in emission all lie close to the line of equal excitation temperature, within error. IRAS F15107+0724 is again the exception, as its 1720 MHz line appears in absorption, and for that reason is left out of Figure 6(a) altogether. In the observed sample, there is no example of an OHM with 1720 MHz emission that is significantly stronger over the entire line profile than that which would be expected from equal excitation temperatures in the lines. The relative strength of the main lines and satellite lines does vary over different parts of the spectrum. For example, in IRAS F10173+0829, the 1720 MHz emission is only visible in the region where 1667 MHz emission is weakest, and is absent over the region where 1667 MHz emission is strongest. Baan & Haschick (1987) observed similar variation within the spectrum of Arp 220, and from that concluded that excitation conditions of the lines were different from one another within the same region.

Overall, the relative weakness of satellite line emission as compared to 1667 MHz emission in OHMs is consistent with the results of HW90 and the modeling of LE08, in which all four of the 18 cm OH lines have roughly equal excitation temperatures. In sources with detected satellite lines, the lines are generally seen only within sub-regions of the 1667 MHz emission, and appear to be moderately stronger than expected for equivalent excitation temperatures. This suggests that secondary pumping mechanisms may occasionally contribute within OHMs, but 53 μm radiative pumping is the dominant pumping mechanism in the OHM population.