Infrared Spectroscopy of Jet-cooled “GrandPAHs” in the 3–100 μm Region

Polycyclic aromatic hydrocarbons (PAHs) are by now generally accepted to be responsible for the IR emission referred to as aromatic infrared bands (AIBs) observed from gaseous and dusty regions in the interstellar medium (ISM; Gillett et al. 1973; Allamondola et al. 1985; Tielens 2008). Much effort has thus been put into analyzing the IR emission spectra and determining the size distribution and structure of the PAHs in the ISM (Tielens 2013). Although in general IR spectroscopy is an effective tool for structure elucidation (Bakels et al. 2020), when considering the conditions of the ISM combined with that of the heterogeneity of the mixture, it is challenging—and in fact as yet not possible—to identify specific species in the ISM. A more detailed insight into the IR spectroscopic properties of directly relevant PAHs is imperative to interpret the interstellar IR emission.

Interstellar PAHs absorb UV radiation from a nearby star, resulting in the excitation of rovibronic levels of electronically excited states. Internal conversion to the electronic ground state populates highly excited vibrational levels of the ground state that subsequently relax to the vibrational ground state by emitting IR photons through a vibrational cascade (Allamandola et al. 1989; Mackie et al. 2018). As it is not trivial to produce the UV-pumped IR emission spectra of the PAHs of interest under the relevant conditions (Cook et al. 1998; Kim & Saykally 2002, 2003), current state-of-the-art analyses rely on theoretical models (Bauschlicher et al. 2018; Mackie et al. 2018). Such analyses have led to the conclusion that the astronomically relevant PAHs typically have 50–150 carbon atoms organized in a compact structure (Ricca et al. 2012). PAHs that meet these characteristics have been designated as grandPAHs (Andrews et al. 2015). Currently used models are, however, only validated with smaller PAHs and may require optimization for larger molecules (Candian & Mackie 2017).

These theoretical models should include accurate frequency and intensity predictions and be validated by laboratory experiments preferably carried out under cold and gas-phase conditions, e.g., in a molecular beam experiment. Difficulties producing a sufficient amount of gas-phase molecules to seed into a molecular beam by resistively heating the sample have so far restricted cold, gas-phase spectroscopy to smaller PAHs (Maltseva et al. 2015, 2016, 2018). To overcome these problems, we make use here of laser desorption (Meijer et al.1990) to bring a sufficient amount of molecules into the gas phase and subsequently entrain and cool them in a molecular beam. This approach thus enables us to record high-resolution, cold, gas-phase IR spectra of much larger PAHs than have been reported so far.

In the present study we report on the IR absorption spectra of the large and compact PAHs coronene (C₂₄H₁₂), peropyrene (C₂₆H₁₄), ovalene (C₃₂H₁₄), and hexa(peri)benzocoronene (C₄₂H₁₈). Experimental IR spectra in the range between 3 and 100 μm are obtained under cold and isolated conditions and compared to, where available, state-of-the-art anharmonic calculations. As a result of its accessibility with ground-based telescopes, the CH-stretch 3 μm region is a well-studied region. Our recent studies on smaller PAHs have shown that this region is dominated by Fermi resonances, which considerably complicate its analysis (Maltseva et al. 2015, 2016, 2018). The CC-stretch region between 5 and 10 μm (1000–2000 cm⁻¹) shows most activity from cationic species, while neutral large PAHs dominate the 15–10 μm (667–1000 cm⁻¹) CH out-of-plane range (Tielens 2008). This range is most diagnostic for the structure of PAHs. As shifts of only a few wavenumbers are observed for different astronomical objects, a correct interpretation requires accurate reference spectra (Hony et al. 2001; Matsuura et al. 2014). Longer-wavelength, far-IR regions contain weak bands, but since these bands are associated with unique global modes with molecule-specific IR frequencies, they provide clear fingerprints of the size and global structure of the PAHs from which they originate (Mulas et al. 2006). Laboratory spectroscopic studies of large PAHs in this region are therefore important to support the interpretation of new observational searches (Ricca et al. 2012).

Experiments were performed using a laser desorption molecular beam apparatus (Rijs & Oomens 2015) located at the FELIX laboratory in the Netherlands (Oepts et al. 1995). Coronene (97%) was purchased from Sigma-Aldrich, ovalene (99.6%) and peropyrene (99.5%) were purchased from Kentax, while hexa(peri)benzocoronene was synthesized using the procedure described by Liu et al. (2011). The compounds were, without further purification, mixed with carbon black in a 1:1 ratio, after which a graphite sample bar was pressed into the mixture. A slightly focused 1064 nm laser beam (10 Hz, Polaris Pulsed Nd:YAG Laser, NewWave Research) with a typical energy of 1 mJ per pulse was used to desorb the molecules while translating the sample bar, thereby providing a fresh sample at every shot. The sample bar was placed in front of the nozzle of a pulsed valve (Jordan Co.) covering about half the orifice, and the desorbed molecules were picked up by an argon jet (10 Hz, 8 bar backing pressure). The expanding gas pulse was skimmed before being crossed with three laser beams to perform IR–UV ion dip spectroscopy: one IR laser to probe the ground-state vibrational frequencies, and two UV/VIS lasers perpendicular to the molecular beam to perform 1 + 1′ resonance-enhanced multiphoton ionization (REMPI). IR light was provided by the free electron laser FELIX for spectroscopy between 100 and 2000 cm⁻¹ counterpropagating the molecular beam or by a high-resolution OPO laser (Laser Vision) covering the 2950–3150 cm⁻¹ region perpendicular to the molecular beam. The bandwidth of FELIX is 0.5%–1% of the IR frequency, while the bandwidth of the OPO is approximately 0.1 cm⁻¹. The idler frequency of the OPO was calibrated continuously during measurements using a HighFinesse WS-7 wavelength meter. Excitation and ionization laser beams were provided by an Nd:YAG laser-pumped dye laser (Innolas Spitlight, Lioptec) and a 193 nm ArF excimer laser (Neweks), respectively. The ions were detected in a reflectron time-of-flight mass spectrometer (Jordan Co.). In our experiments we used excitation frequencies of 24,187.7 cm⁻¹ for coronene (; Kunishige et al. 2017), 21,455.9 cm⁻¹ for ovalene (; Amirav 1981), and 23,074.7 cm⁻¹ for hexa(peri)benzocoronene (feature a; see Kokkin et al. 2008). The REMPI excitation spectrum of peropyrene has not been measured in previous studies and is reported in the Appendix (Figure A1). From this spectrum it can be concluded that the transition occurs at 23,166.2 cm⁻¹, which has been the frequency used for the IR–UV ion dip experiments on this compound. For each of these electronic transitions a typical FWHM of about 1 cm⁻¹ was observed. The IR lasers were operated at half the repetition rate of the REMPI lasers in order to collect alternating IR-on/IR-off ion yields.

Results of harmonic calculations on coronene, ovalene, and hexa(peri)benzocoronene were retrieved from the NASA Ames PAH IR spectroscopic database (PAHdb; Bauschlicher et al. 2018). The anharmonically calculated spectrum of coronene was taken from the study of Mulas et al. (2018), while the anharmonic spectra of ovalene, hexa(peri)benzocoronene, and peropyrene were calculated using GVPT2 (Frisch et al. 2016) as incorporated into the Gaussian16 suite of programs, employing the B3LYP/Jun-cc-pVDZ or B3LYP/N07D level of theory with a superfine grid and very tight optimization (Frisch et al. 2016). For ovalene we found that the standard anharmonic treatment resulted in unrealistic frequency shifts and intensity changes of fundamental bands in the CH out-of-plane region. To resolve this, we recomputed the anharmonic treatment but skipped modes 55, 58, and 62 and simply applied a scaling factor of 0.96 to these particular modes. This is discussed in more detail in Section 3.7. All displayed calculated spectra have been convoluted with a 1 cm⁻¹ FWHM Gaussian in the 3 μm region and a Gaussian with an FWHM of 1% of the frequency in the 5–100 μm region, roughly corresponding to the FELIX laser bandwidth.

The IR absorption spectra of jet-cooled coronene, ovalene, hexa(peri)benzocoronene, and peropyrene in the 3 μm (OPO, 2950–3150 cm⁻¹) and the 5–100 μm (FELIX, 100–2000 cm⁻¹) regions are shown in Figure 1. Since the width of the peaks in the FELIX region is determined by the bandwidth of FELIX, narrower line widths are observed in the far-IR than in the mid-IR region. The bandwidth of the OPO laser, on the other hand, is approximately 0.1 cm⁻¹, meaning that the observed line width of about 1 cm⁻¹ is a result of the population distribution over rotational states, which is in line with the observed band widths of the electronic transitions. A rough estimate of the rotational temperature leads to a ballpark figure of 2 K, indicating that even with such large molecules effective cooling is possible. A list of the observed transitions is provided in the Appendix (Table A1).

Zoom In Zoom Out Reset image size

3.1. General Comparison to Harmonic Calculations

Figure 1 compares experimentally recorded IR spectra with spectra predicted by harmonic and anharmonic calculations. This figure shows that the most intense bands in the FELIX region associated with the CH out-of-plane modes at approximately 800 cm⁻¹ are well predicted using the harmonic approximation. However, as shown in Figure 2, the combination bands observed in the experimental spectrum between 1600 and 2000 cm⁻¹ are not reproduced, as they are per definition absent at this level of theory. For this reason Boersma et al. analyzed the astronomical PAH emission bands that fall in this region using a combination of laboratory measurements and anharmonic computations (Boersma et al. 2009). On average, the region below 1000 cm⁻¹ is better predicted than the region between 1000 and 1600 cm⁻¹ (difference of 0.6% and 1.3%, respectively). Similar to what was observed for smaller PAHs, the 3 μm region of these large PAHs is dominated by Fermi resonances and combination bands. These lead to the observation of many more bands than predicted by harmonic theory and cause a large discrepancy between experiment and theory. Since the 3 μm and the 5–6 μm regions are much more influenced by anharmonicity, it is clear that the interpretation of the IR emission in these regions should be based on models that incorporate resonances and are able to predict combination bands.

Zoom In Zoom Out Reset image size

3.2. General Comparison to Anharmonic Calculations

3.3. Coronene

3.4. Hexabenzocoronene

3.5. Ovalene

3.6. Peropyrene

3.7. Anharmonic Calculations on Large PAHs

To understand why the anharmonic treatment of the larger PAHs ovalene and peropyrene leads to large and unrealistic changes in their IR absorption spectra, we performed the same procedure on other archetypal small and large PAHs. This study includes naphthalene, anthracene, phenanthrene, acenaphthene, pentacene, benzo[ghi]perylene, isoviolanthrene, anti-dipyrenylene, and syn-dipyrenylene, as well as the nitrogen-containing PAHs quinoline and isoquinoline. Since the fundamental transitions in the 100–1600 cm⁻¹ region—and especially the more intense bands—are well predicted using the harmonic approximation, we have plotted in Figures 3(a) and (b) the anharmonically calculated frequencies as a function of the harmonic frequencies for smaller PAHs and larger PAHs, respectively, taking 22 carbon atoms as dividing small and large.

Zoom In Zoom Out Reset image size

For small PAHs we observe a linear relationship between harmonic and anharmonic frequencies, in line with the generally accepted use of a scaling factor to correct harmonic frequencies. No significant (>1%) deviation from the linear relation is observed. Phenanthrene is in this respect an exception, but the affected modes are IR-inactive. In general, the larger PAHs show a similar linear relationship, but large deviations are observed for modes with frequencies between 750 and 1000 cm⁻¹. We find that the unrealistically large frequency shift that was observed in the anharmonic calculations on ovalene and peropyrene is also present in the calculations on other large PAHs, but absent in anharmonic calculations on smaller PAHs. Concurrently with these large frequency changes, also large changes in intensity occur, e.g., the mode at 879 cm⁻¹ of ovalene loses almost all of its intensity after the anharmonic treatment. These modes correspond for all investigated PAHs with C and H out-of-plane bending vibrations. Modes in the 3 μm region, on the other hand, behave regularly, with almost all fundamental anharmonic frequencies being within a 1% deviation from a linear relationship with the harmonic frequencies.

In order to overcome these clear inadequacies of the anharmonic treatment, we have removed the anharmonic contributions of selected modes and instead simply scale the frequencies of these modes. In the case of ovalene these concern modes 55, 58, and 62, with modes 55 and 62 corresponding to the bands with the highest intensity in the region up to 2000 cm⁻¹. Mode 58 is the most affected mode in this region in terms of frequency. For peropyrene we similarly find that the modes with the highest intensity (modes 39 and 42) need to be removed from the anharmonic treatment. In the harmonic framework, these modes include out-of-plane displacements of both C and H atoms. The results of this alternative treatment are shown in Figure 4. Zoom-ins of the 1000–2000 cm⁻¹ region with the corrected anharmonic treatment are displayed in Figure 2. For ovalene we observe that—besides the modes that are removed from the anharmonic treatment—also the bands between 1000 and 2000 cm⁻¹ are slightly affected. In contrast, the 3 μm region is not affected by removal of the three modes in the anharmonic analysis. The zoom-in of the 1000–2000 cm⁻¹ region of ovalene shows a large improvement with respect to the harmonic approximation. In the case of peropyrene, our alternative anharmonic treatment only minimally affects the 1000–2000 cm⁻¹ region and the 3 μm region (see Figure 4). A significant improvement is, however, observed in the bands in the 600–750 cm⁻¹ region, directly to the red of modes 39 and 42. Similar to what was found for the 3 μm and mid-IR regions of small PAHs (Maltseva et al. 2015, 2016, 2018), we thus conclude that for large PAHs an anharmonic treatment is a suitable and necessary route for improving predicted IR spectra, with an important caveat that one needs to eliminate unrealistic frequency and intensity changes in the 750–1000 cm⁻¹ region.

Zoom In Zoom Out Reset image size

It is interesting to notice that other recent studies (Bauschlicher2016; Karadakov 2016; Fortenberry et al. 2017; Lee & Fortenberry 2020) also report on problems with calculating frequencies of out-of-plane bending modes of systems that include C=C bonds. In these studies, harmonic calculations gave rise to imaginary frequencies for geometries for which these modes should have real frequencies. In our harmonic calculations on large PAHs we do not encounter such imaginary frequencies and also find that the low-frequency peaks in the far-IR are relatively well predicted using harmonic theory. Instead, the unrealistically large intensity changes and frequency shifts of the out-of-plane bending modes only arise after anharmonic analysis. It is nevertheless remarkable that the modes for which this occurs are the same kind of modes discussed by Lee & Fortenberry (2020) for benzene-like molecules, namely, modes where both C and H atoms move out of the plane, suggesting that there is link between these two issues.

The present study has investigated the 3–100 μm (100–3150 cm⁻¹) IR spectra of coronene, ovalene, hexa(peri)benzocoronene, and peropyrene under astronomically relevant conditions, that is, at low temperatures and under isolated conditions. These are larger PAHs with different geometries and edge topologies that come close to the size of PAHs concluded to be dominantly responsible for the AIBs (Ricca et al. 2012; Andrews et al. 2015). Previously, gas-phase high-resolution IR spectra of small PAHs have been used to validate anharmonic calculations, and the incorporation of anharmonicity was proven to be essential. However, larger PAHs have so far remained out of reach both from an experimental point of view and from the application of high-end anharmonic theoretical methods. The present study thus provides the first direct lead to assessing the suitability of theoretical methods to predict IR spectra of large PAHs and their application in astronomical models.

First, we find that the experimental IR spectra of larger PAHs confirm the conclusion drawn from previous experiments on small PAHs that far-IR modes are not by default anharmonic but can be calculated harmonically with a comparable accuracy as with more expensive anharmonic treatments. In quantitative terms we show that the frequencies of the main CH out-of-plane features in the region below about 1000 cm⁻¹ are computed on the harmonic level of theory within 0.6% of their experimental values. This is an important conclusion since this region is also the spectral region in the interstellar IR emission where most characteristic bands are attributed to neutral PAHs.

Second, the bands that are located above about 1000 cm⁻¹ are largely affected by anharmonicity. In the combination band region (1600–2000 cm⁻¹), where no intensity is predicted using the harmonic approximation, and especially in the 3 μm region (2950–3150 cm⁻¹), the spectra are dominated by anharmonicity. The combination band region is relatively accurately predicted for coronene by Mulas et al., and also in the 3 μm region a qualitative improvement can be observed for ovalene and peropyrene. For all large PAHs studied here we find that anharmonicity results in an average redshift of the CH-stretch peaks and that its inclusion is necessary to come to a realistic comparison with the experimental spectrum.

Third, the symmetry of coronene and hexabenzocoronene impedes the anharmonic treatment in Gaussian16. Quite unexpectedly, we find that an anharmonic treatment of PAHs with a lower symmetry such as ovalene and peropyrene results in unrealistically large frequency shifts and intensity changes of C and H out-of-plane bending modes between about 700 and 1000 cm⁻¹. This renders the default GVPT2 calculations in Gaussian16 unreliable. Our studies show that an alternative treatment in which the anharmonic contributions from the most affected modes are not taken into account avoids such problems. Interestingly, we find that removal of these modes leaves the anharmonicity-dominated 3 μm region almost unaffected. Although incorporating anharmonicity is thus key for a proper analysis of AIBs, our studies show that this should be done with much care when considering larger PAHs.

Laser desorption opens up the possibility of bringing large PAHs in the gas phase and studying them under cold and isolated conditions. Using IR–UV ion dip spectroscopy, mass-selective IR spectra of neutral PAHs have been obtained that can be used to validate theoretical methods that are crucial for the modeling and subsequent interpretation of astronomical studies of interstellar IR emission. Previous studies have focused on small PAHs and mainly on their CH-stretch region. In contrast, the present work gives access to PAHs with astronomically relevant sizes and covering the 3–100 μm region. We have found that the anharmonic treatment that is beneficial for smaller PAHs results in unrealistic, large frequency shifts and intensity changes of C and H out-of-plane bending modes of large PAHs between about 10 and 13 μm. These bands are, on the other hand, relatively well predicted using the harmonic approximation. In the far-IR region global modes are located that are unique for individual PAH species and that may allow for discrimination of specific PAHs. These modes appear to be well described within the harmonic approximation, and their anharmonic treatment does not lead to significant improvements. Anharmonicity remains, however, also for larger PAHs the dominating factor that absolutely needs to be taken into account to come to a proper description of the combination band (5–6 μm) and CH-stretch region (3 μm).

This work was supported by the Netherlands Organization for Scientific Research (NWO) and is part of the Dutch Astrochemistry Network (DAN) II (project No. 648.000.029). The authors thank the FELIX laboratory team for their experimental assistance and scientific support and also acknowledge the Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO) for the support of the FELIX Laboratory and SURFsara for their computational resources. Last but not least, we would like to thank Bernard Postema for his technical assistance during the experiments, Alessandra Candian for fruitful discussions regarding anharmonic calculations, and Hans Sanders for the synthesis of hexabenzocoronene.