Electronic Spectroscopy of Anthracene Cations and Protonated Anthracene in the Search for Carriers of Diffuse Interstellar Bands

Figure 1 displays the spectra of An⁺ and AnH⁺ measured from 300 nm up to the infrared range. The upper frame shows the spectrum recorded by monitoring the ion signal increase on the mass of the photoproducts An⁺ and AnH⁺, while the lower frame shows the depletion of the sum of ion signals recorded on the masses of several He_n An⁺ and (H₂)_n AnH⁺ complexes summed up. The spectra from the upper frame have a much better signal-to-noise ratio. Unfortunately, they are also characterized by a wider bandwidth and cannot be used for the direct comparison with astronomical observations as the increase in ion signal on the mass of parent molecule occurs due to dissociation of many different complexes. Therefore, they are well suited for fast measurements in the broad wavelength range and are used to get an overall understanding of molecular spectra. When we need to obtain the exact values of the absorption band positions of the species in the gas phase for comparison with astronomical observations, it is necessary to use the depletion spectra, which are mass selective.

Zoom In Zoom Out Reset image size

Our spectrum recorded via monitoring the An⁺ ion signal is overall similar to the spectrum of An⁺ isolated in an Ar matrix (Szczepanski et al. 1993). The main difference is a much smaller spectral shift caused by the weaker interaction of An⁺ with the He atoms compared to the interaction with the Ar matrix. As can be seen, An⁺ has two intense and relatively narrow absorption bands, best suited for detection in the observational spectra. These are the red D₂(0) ← D₀(0) absorption band near 708 nm and the UV D₆(0) ← D₀(0) absorption band near 348 nm. We can confirm that the two bands observed in the matrix measurements at 482.6 and 456.4 nm indeed do not belong to An⁺, as we did not observe any absorption at these wavelengths. The D₅(0) ← D₀(0) absorption band near 420 nm is relatively weak compared to the two transitions near 348 and 708 nm and therefore is not very suitable for the astronomical detection. It should be noted that the absolute band intensities were not derived in our measurements. Therefore, to obtain the absolute band intensities, we rely on matrix isolation data. The matrix isolation data shows that the extinction on the maximum of the 352.1 nm peak is about six times higher compared to that on the 722.4 nm peak (17.11/2.9 M⁻¹ cm⁻¹). However, the bandwidth of the red band is about 5 times higher compared to the UV band (460/90 cm⁻¹). Therefore, the oscillator strengths associated with both bands do not differ much. Our measurements show that the red band is actually narrower in the energy units compared to the UV band. This makes these two bands approximately equally well suited for detecting them in observational spectra.

To compare our laboratory measurements with the observational astrophysical spectra the parameters of absorption bands such as peak positions and bandwidths of bare An⁺ have to be obtained. We recorded the depletion spectra of He_n An⁺ with n = 1–50 with a longer integration time around the position of the most intense bands. The upper panels of Figure 2 show few such spectra measured for the two most intense absorption bands. The fitting of such spectra with Gaussian function allows obtaining the parameters of absorption bands as a function of the number of attached helium atoms. The bottom panels of Figure 2 show linear fits of the shift of these absorption band positions as a function of the number of attached helium atoms n. The extrapolation of these fits gives the positions of the absorption bands for n = 0. As it can be seen, for small n, the positions of the absorption bands change linearly with the number of attached helium atoms. This allows to obtain the band position (λ_max) and the same procedure was used to evaluate the FWHM of the absorption bands. A sharp change in the value of the shift going to small n has been detected for neutral An (Even et al. 2001). In our measurements we observed a notable change in the dependence for both bands after adding the seventh He atom. However, for the UV band, the fit over small n gives the same values as the fit over n = 1–40. For the red band however, a more pronounced change in the dependence is observed. Therefore, we used the values obtained only for the fit over n = 1–6. Due to much lower signal to noise level, the accuracy of the determination of the peak parameters is much lower for this band. We obtained λ_max = 3478.9 ± 1.8 Å, FWHM = 28.5(229 cm⁻¹) ± 5.2 Å and λ_max = 7068.9 ± 5.7 Å, FWHM = 81.6(164 cm⁻¹) ± 6.7 Å.

Zoom In Zoom Out Reset image size

Unfortunately, we were unable to perform the same procedure for finding the positions of the strongest bands of the AnH⁺ due to a much lower S/N for these species. It is explained by a lower intensity of each individual band, because AnH⁺ has several isomers, which differ in the position of the added proton (Garkusha et al. 2011). Therefore, the total intensity is expected to be shared between several bands. Moreover, the bands of AnH⁺ are much broader compared to the bands of An⁺. We found a FWHM of 135.5 and 256.7 Å for the 3768.1 and 4034.9 Å bands, respectively. All these also relate to the possible presence of the hydrogenated An⁺ in the ISM and likely hamper its optical detection. The weaker optical bands of AnH⁺ lie in the range of the highest sensitivity of our system. Therefore, we were able to derive the parameters of the most intense optical band, which are λ_max = 4903.9 ± 1.7 Å and FWHM = 22.4 ± 2.2 Å. However, the optical bands are about an order of magnitude weaker compared to the UV bands and therefore not very well suited for the observational detection.

As can be seen, the measured λ_max of the An⁺ D₂(0) ← D₀(0) (7068.9 Å) absorption band differs by ∼19 Å compared to the value of 7087.6 Å obtained by CRDS (Sukhorukov et al. 2004). The shift could be attributed to a different temperature of the An⁺ produced in the experiments. The shift of the band position to higher energies with reducing the temperature of the species can also be seen in Figure 2 of Sukhorukov et al. (2004). To better understand the nature of this temperature effect, we performed simulations of the spectra of An⁺ for the transition to the second excited electronic state. The D₂(0) ← D₀(0) band was calculated to lie at 6795 Å. For a better comparison with the experimental spectrum, the position of the band was shifted to the experimentally derived value of 7068.9 Å. The simulated spectrum in the range of this 0–0 transition is shown in Figure 3. As can be seen in the figure, the increase in the temperature of An⁺ indeed leads to the appearance of the redshifted peak. The shift is caused by a considerable change in the vibrational frequencies between ground and excited states of An⁺. An has many low frequency vibrations that can be populated at relatively low temperatures. Therefore, at T = 300 K the population of the ground electronic state is low and the main contribution to the band at the position of the 0–0 transition is caused by the transitions between the same vibrational states. For example, a large contribution to the shoulder appearing on the red side of the main band as seen in Figure 3, is produced by ${5}_{1}^{1}$ and ${6}_{1}^{1}$ transitions (numbering starts from the lowest frequency). The calculated shift is about 19 Å, which is completely consistent with the difference in peak positions found in the CRDS and our experiments (∼19 Å). The intensity of this redshifted shoulder is predicted to be smaller compared with the intensity of the main band. This can be understood considering the well-known problem of quantum chemistry in optimizing the geometry of the excited states and evaluation of the values of low-frequency vibrations (note the large error of 263 Å for the calculation of the D₂(0) ← D₀(0) band position). Therefore, such computations could be used to get a qualitative understanding, but not precise quantitative values. In this understanding, the match between computation and experiment is rather good. With a further temperature increase from 300 to 500 K, we no longer observe the shift of the band position. Instead, the simulation only shows an enhancement of the lower-energy shoulder. As the vibrational temperature of An⁺ in previous CRDS measurements is unknown, and could be quite high (Staicu et al. 2004; Sukhorukov et al. 2004), it becomes difficult to give the exact temperature dependence of the band parameters. We can clearly define that for cold media (T < 50 K), the peak maximum should be at 7068.9 Å. The temperature increase should cause band broadening mainly due to the enhancement of the lower-energy shoulder. The exact temperature at which the 7087.6 Å peak becomes dominant cannot be evaluated exactly as the computations likely underestimate its intensity.

Zoom In Zoom Out Reset image size

This is important for comparing laboratory spectra with astrophysical observations, as the peak position may change depending on the physical conditions along the specific line of sights. For example, a broad 7088.8 ± 2.0 Å DIB is detected in the line of sight of star Cernis 52 (BD+31640), nicely matching with the absorption band of An⁺ measured by CRDS (Iglesias-Groth et al. 2010) but not with the peak maximum found in the current study. However, if An⁺ is present not in interstellar but in a hotter circumstellar environment, its temperature could be more similar to the temperature achieved during the CRDS measurements. This can also apply for other PAHs as these molecules have many low-frequency modes associated with out-of-plane bending vibrations, which could be active at relatively low temperatures. Therefore, band profiles or even band positions could considerably vary between different objects depending on the physical conditions.

To test the presence of An⁺ in the ISM, we checked the observational spectra in the range of the two most intense absorption bands of An⁺ found in our measurements. For the comparison, we selected the lines of sight toward HD 183143 and HD 172028. The first line of sight is characterized by the presence of a relatively dense (A_V = 3.9 mag) interstellar cloud along that light path. The intensities of DIBs were found to correlate with the cloud density and were found to be particularly strong in this direction. Moreover, a number of new DIBs were also found in the observational spectra (Hobbs et al. 2009). However, the spectrum in this line of sight does not show the absorption features from C₂ molecules, whose abundances were found to correlate with the intensities of few DIBs, which were referred to as the C₂ DIBs. Thus, we also selected HD 172028 as the second line of sight. Although it has a less dense cloud (A_V = 2.3 mag), it is characterized by strong absorption bands of C₂ molecules (Gredel 1999). Figure 4 shows the comparison of the observational and experimental spectra in the UV and optical ranges. As can be seen in the image, there are no absorption bands in the observational spectra along either line of sight matching the measured laboratory spectrum of cold An⁺. In the UV range, the peak maximum of the An⁺ absorption coincides with the broad emission maximum. The broad absorption features can be seen on both sides from the position of An⁺ absorption along the HD 183143 line of sight, while the spectrum along the HD 172028 line of sight shows a much smaller number of absorption bands. The red part of the observational spectrum along this line of sight has been previously explored by Hobbs et al. (2009) for the identification of DIBs. All DIBs found in this study are marked with an asterisk in the figure. They are much narrower compared to the absorption bands of An⁺.

Zoom In Zoom Out Reset image size

As no absorption of An⁺ has been detected in observational spectra, we can only give an upper limit of the An⁺ column density. The upper limit of the equivalent bandwidth of the band that will escape the detection is given by the equation $\overline{\mathrm{lim}}W=3\sqrt{M}\tfrac{{\rm{\Delta }}\lambda }{{\rm{S}}/{\rm{N}}}$, where Δλ is the spectral width of a resolution element, M is the number of resolution elements over which the sought absorption band spreads, and S/N is the signal-to-noise ratio near the expected absorption feature (Gredel et al. 2011). Using the derived values of equivalent bandwidth and the oscillator strength from matrix isolation data f = 0.045, we can get the upper limit for the column density of An⁺ along the observed lines of sight. The optical range provides a slightly better sensitivity for the detection of An⁺ with values of 3.5 × 10¹² and 1.0 × 10¹² cm⁻² for HD 172028 and HD 183143, respectively. The detection based on the UV band provides a factor two lower sensitivity with the values of 3.2 × 10¹² and 2.2 × 10¹² cm⁻² for the same lines of sight. However, it should be noted that there are several narrow absorption bands in the observational spectra at the position of an expected absorption of An⁺. Therefore, the real upper limits for column densities could be slightly larger than the derived values. It would be reasonable to assume values at least twice as high. Therefore, based on the observational spectra along HD 183143, we define $\overline{\mathrm{lim}}N$ = 2.0 × 10¹² cm⁻².

The obtained values are very similar to the value of 2.8 × 10¹² cm⁻² found for the upper limit of column densities of neutral An molecules along the HD 183143 line of sight (Gredel et al. 2011). This also means similar limits for the fractional abundance of An⁺, i.e., N(An⁺)/N(H) <10⁻⁹. The estimation of the intensity of the IR emission bands assigned to PAHs led to the estimation that a few percent of the available interstellar carbon is locked in small PAHs with a minimum abundance of PAHs per H nuclei of about 3 × 10⁻⁸ (Allamandola et al. 1989). As An was found to be one of the smallest PAHs stable against UV irradiation in the neutral ISM and it has a thermodynamically favorable structure, it is expected to have a larger abundance in the ISM compared to other PAHs in the case of the bottom up formation. Therefore, the low upper limits for amounts of An and An⁺ in translucent molecular clouds can point toward the top-down formation route of the PAHs. Alternative explanations for a low amount of An in space could be its chemical instability. Translucent molecular clouds are characterized by the highest number density of neutral C atoms compared to other astrophysical environments. Small PAHs react barrierlessly with atomic carbon, and therefore should be efficiently destroyed by such reactions in a broad temperature range (Kaiser et al. 1999; Krasnokutski & Huisken 2014; Krasnokutski et al. 2017). The lifetime of small PAHs in most common astrophysical environments was found to be in the order of only a few years (Krasnokutski et al. 2017). The reaction with C atoms would lead to a large diversity of the PAH molecular structures, leading to a very low abundance of each specific PAH. The C atoms also react barrierlessly with fullerenes, which were found in abundance (Krasnokutski et al. 2016). However, in these reactions the fullerene cage remains intact with C atoms bound on top. Therefore, the products of such reactions can be dissociated by UV photons with the release of fullerenes. To test this, it would also be interesting to define the upper limits for the fractional abundance of phenanthrene cations. These were also found to be stable against UV radiation and have an even more thermodynamically favorable structure compared to An.