Journal of Molecular Spectroscopy 238 (2006) 158–167 www.elsevier.com/locate/jms Infrared spectroscopic investigation of higher diamondoids Jos Oomens a,*, Nick Polfer a, Olivier Pirali a, Yuko Ueno b,c, Roha Maboudian b, Paul W. May d, Jacob Filik d, Jeremy E. Dahl e, Shenggao Liu e, Robert M.K. Carlson a e FOM Institute for Plasma Physics ‘‘Rijnhuizen’’, Edisonbaan 14, 3439 MN Nieuwegein, The Netherlands b Department of Chemical Engineering, University of California Berkeley, Berkeley, CA 94720, USA c NTT Microsystem Integration Laboratories, Atsugi, Kanagawa 243-0198, Japan d School of Chemistry, University of Bristol, Bristol BS8 1TS, UK e Molecular Diamond Technology, Chevron Technology Ventures, Richmond, CA 94802, USA Received 11 April 2006 Available online 5 May 2006 Abstract Attenuated total reflection Fourier transform infrared spectra are recorded for a number of diamond molecules (or higher diamondoids) in the spectral range between 650 and 3000 cm 1. Molecules studied are diamantane, triamantane, [121]tetramantane, [123]tetramantane, [1(2,3)4]pentamantane, [12(3)4]pentamantane, 3-methyl-[1(2,3)4]pentamantane, and [12312]hexamantane (cyclohexamantane). Spectral trends are clearly recognized throughout the spectra revealing the general fingerprint of this class of molecules. In general, the spectra show good agreement with density functional theory calculations at the B3LYP/D95(d,p) level of theory. Ó 2006 Elsevier Inc. All rights reserved. Keywords: Diamondoids; Infrared; Spectroscopy 1. Introduction Diamondoid molecules consist of a diamond-like carbon cage, where all carbon atoms are sp3 hybridised, and dangling bonds at the edges of the systems are terminated with hydrogen atoms. The smallest member of the family, adamantane (C10H16) is made up of the central cage of a single diamond unit cell and was first synthesized by carbocation equilibration in 1957 by von Rague Schleyer [1]. The next members of the family are diamantane and triamantane possessing two and three diamond cages, respectively. For higher diamondoids, starting with tetramantane (four cages), structural isomers can be formed. Diamondoids show remarkable rigidity, strength, and thermodynamic stability, as well as interesting electronic properties [2], which may be of use in chemical, polymer, and pharmaceutical applications, as well as in nanotechnology. * Corresponding author. Fax: +31 30 603 12 04. E-mail address: joso@huygens.rijnh.nl (J. Oomens). 0022-2852/$ - see front matter Ó 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.jms.2006.05.001 Moreover, there has been substantial interest in the spectroscopic properties of this class of molecules from an astrophysical view-point. Not only have nanometersized ‘diamond-like’ molecules been found in meteorites [3], the occurrence of these compounds as isolated gasphase molecules in the interstellar medium, has also been suggested, based on the observation of infrared absorption [4] and emission [5,6] bands around 3.5 lm (2880 cm 1) in the spectra of protostars as well as the post-AGB object HR 4049 [6]. This band has been assigned to the tertiary sp3 carbon (i.e., carbon bound to three carbon atoms) C–H stretching mode [7], but other explanations have also been put forward [8,9]. Further evidence for the existence of interstellar diamondoid molecules could be obtained by comparing the complete vibrational spectra of such diamondoids with astronomical data [10]. Although the very smallest members of the diamondoid family can now be readily synthesized, this is not the case for species larger than triamantane. Hence, in spite of the wide interest in diamondoids, relatively few studies have been reported on the characterization of these molecules. J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 A recent breakthrough in diamondoid research was the discovery that higher diamondoids can be isolated from petroleum, which was shown for species up to the size undecamantane (11 cages) [11]. As a result, various analytical techniques including GC/MS and NMR have been applied to identify and characterize these novel species [12]. From a spectroscopic point of view, only Raman spectra of a selection of diamondoids with a size ranging from adamantane to [121321]heptamantane have been published very recently [13]. In addition, theoretical methods have been applied to predict the vibrational spectra of some of these molecules. An extensive computational study of adamantane and some of its ionic and dehydrogenated Adamantane C10H16 Diamantane C14H20 Triamantane C18H24 159 radical species was reported by Yan et al. [14]. For the adamantyl closed-shell cation, an experimental infrared spectrum was also obtained [15]. The Raman spectrum of crystalline cyclohexamantane was interpreted with density functional theory (DFT) calculations by Richardson et al. [16]. An infrared spectroscopic study on deposited diamondoid nanoparticles (5–350 nm in size) focusing on the 3 lm spectral range was successfully interpreted with DFT computations on individual molecular diamondoid structures, giving insight into the size dependence of the spectra [17]. To our knowledge, no infrared spectra have to date been reported for individual higher diamondoid molecules. Moreover, no direct comparison between theoretical and experimental spectra for these molecules has been reported. In this contribution, we present the first infrared spectra for a number of higher diamondoids, and present an analysis based on DFT computations of the vibrational spectra. The systems studied are diamantane, triamantane, [121]tetramantane, [123]tetramantane, [1(2,3)4]pentamantane, [12(3)4]pentamantane, and [12312]hexamantane (cyclohexamantane), which are shown in Fig. 1. (Note that we follow the nomenclature introduced in ref. [18].) In addition, we present the infrared spectrum for the [1(2,3)4]pentamantane molecule methylated at the position of the arrow in Fig. 1 (3-methyl-[1(2,3)4]pentamantane). 2. Methods [121] P[123] Tetramantanes C22H28 [1(23)4] [12(3)4] Pentamantanes C26H32 [12312]Hexamantane C26H30 Fig. 1. Higher diamondoid structures considered in this work. Hydrogen atoms have been omitted for clarity; all structures are fully saturated. The smallest member of the family, adamantane, is shown for comparison. The arrow indicates the carbon atom where by the substitution of a hydrogen atom for a methyl group, the 3-methyl-[1(2,3)4]pentamantane is formed. 2.1. Experimental Higher diamondoid species (size going up to undecamantane) have been isolated from petroleum oil by vacuum distillation, high-temperature pyrolytic destruction of non-diamondoid species combined with chromatographic techniques [11]. The size and shape separations are performed using high performance liquid chromatography in ultrahigh purity. The molecules studied here are neutrals, fully sp3 hybridized molecules with completely hydrogenated surfaces plus one mono-methylated species. The attenuated total reflection Fourier transform infrared (ATR-FTIR) spectra of the diamondoid molecules were recorded with the use of a Thermo Nicolet Avatar 370 FT-IR spectrometer and an ATR accessory with a ZnSe ATR crystal. The spectra of [12(3)4]pentamantane and 3-methyl-[1(2,3)4]pentamantane were recorded in Bristol (UK) at slightly higher resolution, using a Perkin-Elmer ‘Spectrum One’ FTIR spectrometer. The higher resolution was required to better resolve the bands in the congested spectra of these molecules. All spectra are obtained from pure, powdered diamondoid samples, and measurements were performed at room temperature. The intensity of the ATR absorbance spectra are corrected for the penetration depth of the evanescent wave, which is proportional to the wavelength. One can notice that two enantiomers (P and M) exist for the molecule [123]tetramantane. Since the light source used 160 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 to obtain the ATR-FTIR spectra is not polarized, the spectrum recorded is not dependent on the enantiomer studied. The structure of the P[123]tetramantane is given in the Fig. 1, but the recorded spectrum corresponds to a racemic mixture of the P and M enantiomers. 2.2. Computational Geometry optimization and computation of the harmonic vibrational frequencies were performed using the Becke3LYP hybrid density functional [19] and Dunning’s double zeta basis set with polarization functions, D95(d,p). This method has been shown to give reliable results for numerous organic compounds, such as a large variety of polycyclic aromatic hydrocarbons [20]. The computations are carried out at an IBM cluster1600 system located at the SARA Supercomputer center in Amsterdam using the Gaussian98 program package [21]. The high symmetry of many of the systems was conveniently used to reduce the computational cost of the calcuTable 1 Empirical scaling factors for DFT computed frequencies Formula Diamantane Triamantane [121]Tetramantane [123]Tetramantane [1(2,3)4]Pentamantane [12(3)4]Pentamantane [12312]Hexamantane 3-Methyl-[1(2,3)4] pentamantane C14H20 C18H24 C22H28 C22H28 C26H32 C26H32 C26H30 C27H34 Sym D3d C2v C2h C2 Td Cs D3d C3v Frequency scaling Mid-IR 3 lm 0.9916 0.986 0.986 0.988 0.975 0.985 0.990 0.976 0.949 0.947 0.947 0.948 0.947 0.941 0.948 0.945 lations. All calculations converged and yielded no negative frequencies. For comparison with the experimental spectra, the calculated spectra have been convoluted with a 5 cm 1 FWHM Gaussian line profile (except for the spectra recorded at higher resolution in Bristol, where a 2.5 cm 1 FWHM Gaussian convolution was used in the mid-infrared range). Due to the generally substantial differences in anharmonicity for the CH stretching modes in the 3 lm spectral range as compared to the modes in the mid-infrared range, empirical scaling factors are commonly determined for the two spectral ranges separately. In the 3 lm region, a scaling of the computed frequencies was applied so that the most intense peak matches exactly. In the mid-infrared range, the scaling was (arbitrarily) chosen such that the strong peak occurring around 1050–1150 cm 1 shows exact agreement. Thus, average scaling factors of 0.986 and 0.947 are found for the mid-infrared and CH stretching regions (see Table 1), respectively, which appears to be nicely in accord with the overall value of 0.961 reported for the B3LYP functional [22]. Moreover, the 3 lm scaling factor of 0.947 is very close to that of 0.946 reported by Chen et al. [17] for computed harmonic frequencies of large diamondoids using the same functional but with the 6-31G(d) basis set. 3. Results and discussion The spectra are displayed in Figs. 2–9, compared ‘‘back-to-back’’ with the scaled computed spectra. As expected, the spectra of the more symmetric species are much simplified relative to those of the less symmetric ones, even if they are bigger in size; a striking example DIAMANTANE 60 3 30 0 0 -30 -3 -60 600 800 1000 1200 1400 1600 2700 wavenumber (cm-1) Fig. 2. Infrared spectrum of diamantane. 2800 2900 3000 161 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 TRIAMANTANE 2 30 0 0 -30 -2 600 800 1000 1200 1400 1600 2700 2800 2900 3000 2800 2900 3000 wavenumber (cm-1) Fig. 3. Infrared spectrum of triamantane. [121]TETRAMANTANE 4 50 2 0 0 -2 -50 -4 600 800 1000 1200 1400 1600 2700 wavenumber (cm-1) Fig. 4. Infrared spectrum of [121]tetramantane. is observed by comparing the spectra of [123]tetramantane (C2) with those of [1(2,3)4]pentamantane (Td) or [12312]hexamantane (D3d). On the other hand, one can also notice that the spectra of two different molecules belonging to the same symmetry group can be quite different. Compare for example the spectra of adamantane (see ref. [23]) and [1(2,3)4]pentamantane (both Td symmetry) or diamantane and [12312]hexamantane (both D3d symmetry). However, despite the dense spectrum for the lower symmetry species, the experimental- versus-theoretical match in the mid-infrared frequency range is usually acceptable. For instance, the spectra of [123]tetramantane (Fig. 5) and [12(3)4]pentamantane (Fig. 7), recorded at higher resolution because of the highly congested mid-infrared spectrum, show good agreement with theory. At this point we could discuss the spectrum of each molecule individually, however, we have chosen to discuss the cyclohexamantane spectrum in somewhat more detail and then give a more general discussion of the spectra of the 162 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 [123]TETRAMANTANE 40 2 20 1 0 0 -1 -20 -2 -40 600 800 1000 1200 1400 1600 2700 2800 2900 3000 2800 2900 3000 wavenumber (cm-1) Fig. 5. Infrared spectrum of [123]tetramantane. [1(2,3)4]PENTAMANTANE 1 10 0 0 -1 600 -10 800 1000 1200 1400 1600 2700 wavenumber (cm-1) Fig. 6. Infrared spectrum of [1(2,3)4]pentamantane. remaining molecules. Experimental peak listings and computed optimized structures for all molecules can be found in the Supplementary material. The choice for hexamantane is not completely arbitrary since due to the high symmetry of the molecule, it represents a relatively simple and clean spectrum. Moreover, this particular higher diamondoid seems to have received more attention than others recently [12,16]. It is the largest species in the series investigated here. Finally, we briefly investigate the effect of methylation on the spectrum by discussing the spectrum of 3-methyl-[1(2,3)4]pentamantane in somewhat more detail. 3.1. IR spectrum of cyclohexamantane [12312]Hexamantane or cyclohexamantane belongs to the D3d symmetry group and this high symmetry simplifies the infrared spectrum considerably. Of the 162 vibrational modes, only those of A2u and Eu symmetry are infrared allowed. The Eu modes are doubly degenerate. From the 163 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 [12(3)4]PENTAMANTANE 40 2 20 0 0 -20 -2 -40 600 800 1000 1200 1400 1600 2700 2800 2900 3000 wavenumber (cm-1) Fig. 7. Infrared spectrum of [12(3)4]pentamantane. The resolution of the calculated spectrum is 2.5 cm 1. [12312]HEXAMANTANE 3 50 0 0 -50 -3 600 800 1000 1200 1400 1600 2700 2800 2900 3000 wavenumber (cm-1) Fig. 8. Infrared spectrum of [12312]hexamantane. comparison of the experimental and theoretical spectra shown in Fig. 8, it is quite straightforward to obtain an assignment for the majority of the bands. Some difficulties are encountered in the 1250–1400 cm 1 range, where more bands, and with more intensity, are observed than calculated. In the CH stretching range around 2800–2900 cm 1, the proximity of several very intense modes causes severe congestion in the experimental spectrum, which makes a secure assignment very difficult. The proposed assignments are listed in Table 2, where those in the above mentioned wavenumber ranges should be regarded as tentative at best. Table 2 shows that using the scaling factors of Table 1 gives a deviation of less than 5 cm 1 for all bands except for the bands in the 1250–1400 cm 1 range. In addition, for the band observed at 1445 cm 1, assigned securely as a CH2 scissoring mode, the deviation of more than 20 cm 1 appears to be anomalously large. In fact, two closely spaced scissoring modes are predicted, one of Eu and one of A2u 164 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 4 3-METHYL-[1(2,3)4]PENTAMANTANE 40 2 20 2700 2800 2900 3000 0 0 -20 -2 -40 -4 600 800 1000 1200 1400 1600 2700 2800 2900 3000 wavenumber (cm-1) Fig. 9. Infrared spectrum of 3-methyl-[1(2,3)4]pentamantane. A comparison with the 3-lm spectrum of its non-methylated counterpart is shown in the inset. Table 2 Infrared active modes of cyclohexamantane, experimental versus calculated Calc freq (cm 1) Intensity (km/mol) Sym Exp freq (cm 1) Description 668.2 758.9 774.4 867.1 873.5 929.9 943.0 952.1 1027.9 1063.9 1073.0 1084.0 1118.2 1181.8 1210.3 1280.8 1285.8 1326.1 1366.8 1368.1 1477.7 1478.7 2834.6 2841.5 2853.2 2858.5 2860.4 2870.1 2877.9 2904.4 2905.3 0.2 0.2 4 0.1 0.02 2 2 1 1 4 19 0.1 0.1 0.7 1 0.1 2 6 2 0.3 13 13 8 150 17 2 36 293 418 62 176 Eu A2u Eu A2u Eu Eu Eu A2u Eu Eu A2u Eu Eu A2u Eu Eu Eu A2u Eu A2u Eu A2u Eu A2u Eu A2u Eu A2u Eu Eu A2u 665 760 775 865 874 931 939 — 1026 1058 1074 1088 — 1178 — 1276 1297 1311 1337 1352 1445 — — 2847 — — 2858 — 2874 — 2906 Deform Deform Deform CH2 rock Deform/CH2 rock Deform/CH2 rock CC str/CH bend CC str/CH bend CC str/CH bend CH2 wag/CH bend CH2 rock/CH bend CH bend CH2 twist/CH bend CC str/CH bend CH bend CH bend/CH2 twist CH bend CH bend CH bend CH bend CH2 scissor CH2 scissor CH str CH str CH str CH str CH str CH str CH str CH str CH str character. The experimental band at 1445 cm 1 indeed appears to be somewhat asymmetrically shaped suggesting the presence of two unresolved bands. When the molecule is viewed along the (1 0 0) crystal lattice plane, the modes correspond to asymmetric scissoring of the CH2 moieties on diagonally opposing corners of the molecule. J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 As impressive as the match for the frequencies is, the intensity predictions are not all that excellent. Particularly the relative intensities of the modes in the 1250–1400 cm 1 range are underestimated. Moreover, the intensity ratio between the CH stretching range and the rest of the spectrum is not so well reproduced. Analogous to for instance olefins and acetylenes, the strongest CH stretching bands roughly correspond to collective asymmetric stretching modes, though now along one of the crystal lattice planes. For instance in cyclohexamantane, the band calculated at 2877.9 cm 1 corresponds to the asymmetric CH stretching band in the (1 1 0) lattice plane, the band at 2905.3 cm 1 oscillates asymmetrically in the (1 1 1) plane, and the 2870.1 and 2841.5 cm 1 bands in the (1 0 0) lattice plane. 3.2. General discussion of diamondoid spectra Upon global inspection of all spectra, one clearly observes large similarities between them. Many parallels with the cyclohexamantane spectrum can be observed, although the lower symmetry of most other species yield a richer vibrational pattern. The general appearance of the spectra could be considered as the spectral fingerprint of the diamondoid class of molecules. Inspecting the spectra more closely, four distinct groups of bands can be recognized in all diamondoid spectra. Visualization of the calculated vibrational displacements allows one to roughly assign a general mode description pertinent to the four groups (see Table 3). The modes around 2900 cm 1 (Group A) are of course the CH stretching modes. Due to the very strong and overlapping bands and the broader line widths in the 3 lm range, it is harder to recognize the quality of the match. In some cases, it appears that the experimental bands are more widely spaced than their computed counterparts. A possible explanation could be strong anharmonic interactions between closely spaced modes of equal symmetry, which all have very similar CH stretching character. A very recognizable group of modes in all spectra, mostly appearing as a doublet around 1450 cm 1 is due to two or more CH2 scissoring modes. These bands are clearly identifiable throughout the spectra, mostly appearing as a doublet. In [123]tetramantane, the low symmetry causes further splitting into about four bands. In [12312]hexamantane, where only a single band is observed, the computation predicts three bands with non-vanishing intensity Table 3 General assignment pertinent to all molecules studied in four distinct spectral ranges Group Group Group Group A B C D Region (cm 1) General description 2900 1450 1000–1400 61000 CH stretch CH2 scissor CH2 rock, wag, twist Skeletal deformation 165 within 1 cm 1. In all cases the bands are well reproduced by theory, except for a consistent blue shift of the calculated frequencies of around 35 cm 1. A slightly smaller scaling factor of about 0.966 on average appears to be more appropriate for these modes. Compared to the average scaling factor used for the mid-infrared range of 0.986 (see Table 1) this amounts to a discrepancy of 2 %. Despite this discrepancy, the assignment appears to be fairly secure on account of the consistency of the shift and the recurring pattern of these bands throughout the spectra (see Figs. 2–9). The bands falling roughly between 1000 and 1400 cm 1 have mostly CH bending and CH2 rocking, wagging, and twisting character. Toward the red end of this group, some CC stretching character is found. Finally, in the long wavelength part of the spectra, below 1000 cm 1, mostly skeletal deformation modes are found and these modes have been classified as belonging to Group D in Table 3. Comparing the bands in Groups C and D to the DFT calculation, a striking observation is the underestimation of their intensities (except some around 1050 cm 1) by the DFT calculations. The reason for this discrepancy, that appears to be a general feature in all diamondoid spectra, is at present unclear. A possible influence of the basis set used was investigated for the [121]tetramantane molecule (see Fig. 10). However, the underestimation of intensities in Groups C and D is consistently reproduced upon using basis sets ranging in size from 3-21G to cc-pVDZ. As to the computed frequencies, one notices in Fig. 10 that the frequency scaling factor becomes closer to unity as the basis set is increased, which is a well-known effect. Interestingly, however, the CH2-scissoring modes (Group B) appear to shift more than the other bands in the spectrum, such that the anomalous scaling factor mentioned for these modes above, is almost completely corrected for when using the cc-pVDZ basis set. In conclusion, a global inspection of the spectra shows that the match between theoretical and experimental frequencies is reasonable for most of the bands in the mid-infrared range. This does, however, not apply to the relative intensities, which are not so well reproduced. Although intensities were corrected for the wavelength dependency of ATR, there remains a significant discrepancy (a factor of 3 approximatively) between the experimental and computed intensity ratios for the mid-IR versus the 3 lm spectral ranges. Judging from the increased linewidth and the more Lorentzian-like lineshape, it seems that this discrepancy results at least partly from saturation in the 3 lm range. It is, however, also known that while DFT usually provides a good prediction of the frequencies, intensities are not so reliably predicted. In addition, interaction of the molecules with their solid environment may affect the intensities differently for different vibrational modes. In fact, for a number of polyaromatic molecules, Joblin et al. [24] found that the CH stretching modes were attenuated by a factor of 3 in the solid-phase with respect to the gas-phase. Although such an effect would indeed explain 166 J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 3-21G 6-31G 6-31G(d,p) D95(d,p) cc-pVDZ 4 30 EXP 20 2 0 600 10 800 1000 1200 1400 0 1600 2700 2900 3100 Wavenumbers (cm-1) Fig. 10. Infrared spectrum of [121]tetramantane compared to (unscaled) calculated spectra using different basis sets. here the observed discrepancies, only when gas-phase diamondoid spectra become available can such a conclusion be truly substantiated. 3.3. Methyl substitution at the surface of the cage In addition to the pure diamondoid molecules, the infrared spectrum has been recorded for a methyl substituted diamondoid: 3-methyl-[1(2,3)4]pentamantane (see Fig. 9), where the H atom at the position indicated with an arrow in Fig. 1 is substituted by a CH3 group (thereby causing a reduction in symmetry from Td to C3v). With respect to the general appearance of diamondoid spectra as described above, significant differences are found. In the CH stretching region around 3 lm, the calculation produces CH stretching modes localized on the methyl group at 2930.1 and 2860.5 cm 1, corresponding to the CH3 stretching modes parallel and perpendicular to the threefold symmetry axis, respectively. These modes can be well observed in the experimental spectrum, particularly when it is compared with the experimental spectrum of the non-methylated [1(2,3)4]pentamantane (see inset in Fig. 9). Though weak, additional bands are observed for the methylated species at 2857 cm 1 and around 2945 cm 1 (we cannot explain the observation of what appears to be a doublet at present). At the red end of the CH stretching manifold, there appears to be another additional band at about 2820 cm 1, but the calculation indicates that this band is due to the three degenerate CH stretching modes in the (1 1 1) lattice plane, J. Oomens et al. / Journal of Molecular Spectroscopy 238 (2006) 158–167 which are red-shifted by about 10 cm 1 in the methylated molecule with respect to the non-methylated one. In the remainder of the spectrum, it is not so easy to pinpoint the direct influence of the methyl group on the spectrum. Comparison of the spectrum with that of unsubstituted [1(2,3)4]pentamantane (see Fig. 6) clearly shows that additional bands appear, but inspection of the corresponding normal modes indicates that they are due to non-localized modes. It therefore appears that these differences are mainly induced by the reduction of symmetry from Td for [1(2,3)4]pentamantane to C3v for 3-methyl[1(2,3)4]pentamantane. Some local modes of the methyl group are computed at 1369.7 cm 1 (CH3 umbrella) and 1457.0 cm 1 (CH3 scissor), but they are not easily recognized in the experimental spectrum, probably due to overlapping diamondoid bands. 4. Conclusion We report in this paper the first infrared spectroscopic characterization of several higher diamondoid molecules. These experimental spectra, supported with density functional theory calculations provide the spectroscopic fingerprint for this class of molecules. Among others, this may strengthen a possible detection of this family of compounds in different astrophysical objects, which has thus far been based only on the observation of CH stretching bands around 2850 cm 1. In general, a good agreement between experiments and DFT computations was found, which allowed us to identify several groups of bands corresponding to certain classes of molecular vibrations. Finally, methyl substitution on the diamondoid cage induced significant changes particularly in the 3 lm range of the spectrum, which may help to evaluate their astrophysical relevance. Appendix A. Supplementary data Supplementary material. Optimized structures at the B3LYP/D95(d,p) level for all molecules studied here as well as experimental frequencies. Supplementary data for this article are available on ScienceDirect (www.sciencedirect.com) and as part of the Ohio State University Molecular Spectroscopy Archives (http://msa.lib.ohio-state.edu/jmsa_hp.htm). References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] Acknowledgments We gratefully acknowledge the SARA Supercomputer center in Amsterdam, The Netherlands, for providing the CPU time. This work is part of the research program of FOM (Project No. 03PR2218), which is financially supported by the ‘Nederlandse Organisatie voor Wetenschappelijk Onderzoek (NWO)’. 167 [19] [20] [21] [22] [23] [24] P.V.R. Schleyer, J. Am. Chem. Soc. 79 (1957) 3292. D.V. Korolkov, O.V. Sizova, Int. J. Quantum Chem. 88 (2002) 606. E. Anders, E. Zinner, Meteorites 28 (1993) 490. L.J. Allamandola, S.A. Sandford, A.G.G.M. Tielens, T.M. Herbst, Astrophys. J. 399 (1992) 134. O. Guillois, G. Ledoux, C. Renaud, Astrophys. J. 521 (1999) L133. C. van Kerckhoven, A.G.G.M. Tielens, C. Waelkens, Astron. Astrophys. 384 (2002) 568–584. H.C. Chang, J.C. Lin, J.Y. Wu, K.H. Chen, J. Phys. Chem. 99 (1995) 11081–11088. J.C. Blades, D.C.B. Whittet, Mon. Not. R. Astron. Soc. 191 (1980) 701. W.A. Schutte, A.G.G.M. Tielens, J.L. Allamandola, M. Cohen, D.H. Wooden, Astrophys. J. 360 (1990) 577. K. Sellgren, Spectrochim. Acta A 57 (2001) 627. J.E. Dahl, S.G. Liu, R.M.K. Carlson, Science 299 (2003) 96. J.E.P. Dahl, J.M. Moldowan, T.M. Peakman, J.C. Clardy, E. Lobkovsky, M.M. Olmstead, P.W. May, T.J. Davis, J.W. Steeds, K.E. Peters, A. Pepper, A. Ekuan, R.M.K. Carlson, Angew. Chem., Int. Ed. 42 (2003) 2040–2044. J. Filik, J.N. Harvey, N.L. Allan, P.W. May, J.E.P. Dahl, S. Liu, R.M.K. Carlson, Spectrochim. Acta A 64 (2006) 681–692. G. Yan, N.R. Brinkmann, H.F. Schaefer, J. Phys. Chem. A 107 (2003) 9479–9485. N. Polfer, B.G. Sartakov, J. Oomens, Chem. Phys. Lett. 400 (2004) 201–205. S.L. Richardson, T. Baruah, M.J. Mehl, M.R. Pederson, Chem. Phys. Lett. 403 (2005) 83–88. Y.R. Chen, H.C. Chang, C.L. Cheng, C.C. Wang, J.C. Jiang, J. Chem. Phys. 119 (2003) 10626–10632. A.T. Balaban, P. von Rague Schleyer, Tetrahedron 34 (1978) 3599–3609. A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652. C.W. Bauschlicher, S.R. Langhoff, Spectrochim. Acta A 53 (1997) 1225–1240. M. J. Frisch et al., Gaussian 98 Revision A.11.4 Gaussian, Inc., Pittsburgh PA. A.P. Scott, L. Radom, J. Phys. Chem. 100 (1996) 16502–16513. http://webbook.nist.gov. C. Joblin, L. d’Hendecourt, A. Léger, D. Défourneau, Astron. Astrophys. 281 (1994) 923–936.