Fluorescence excitation and dispersed fluorescence spectra of iso-quinolinyl radicals 4-, 5-, and 8-iso-HC₉H₇N isolated in solid para-hydrogen

Abstract

Nitrogen-substituted polycyclic aromatic hydrocarbons and their derivatives have been proposed as possible carriers of the unidentified infrared emission feature near 6.2 μm in astronomical observations and might also contribute to the diffuse interstellar bands (DIB). We present the previously unobserved dispersed fluorescence and fluorescence excitation spectra of three isomers of hydrogenated iso-quinoline, iso-quinolinyl radicals (iso-HC₉H₇N), isolated in solid para-hydrogen (para-H₂). The three observed progressions, originating at 18 492, 19 430, and 18 688 cm⁻¹ and exhibiting distinct fluorescence lifetimes, are assigned to the D₁–D₀ electronic transitions of the 4-, 8, and 5-iso-HC₉H₇N isomers, respectively. Assignments are supported by Franck–Condon Herzberg–Teller simulations based on (TD-)uB3PW91/6-311++G(2d,2p) optimized geometries and scaled harmonic vibrational frequencies. The dispersed fluorescence and fluorescence excitation spectra predominantly involve vibrational normal modes of a′ symmetry. We compare our spectra to previously reported IR spectra and discuss how the position of hydrogenation influences vibronic structure. The potential contribution of 4-, 8, and 5-iso-HC₉H₇N to the DIB spectrum is discussed.

---

1. Introduction

Since the discovery of the first diffuse interstellar bands (DIB) by Heger in 1922,¹ over 500 DIB, i.e. absorption bands of interstellar origin observed spanning the spectral range from the near infrared (NIR) to the ultraviolet (UV), have been catalogued.^2–4 Yet, the carriers responsible for these bands remain largely unidentified, due to the limited availability of laboratory reference spectra suitable for comparison to astronomical observations. To date, only the buckminsterfullerene cation C₆₀⁺ has been conclusively identified as the carrier of five DIB in the NIR.^5,6 Polycyclic aromatic hydrocarbons (PAH) and their protonated, hydrogenated, and cationic derivatives are considered promising candidates for the carriers of DIB.^7,8 They are also generally accepted to contribute to the unidentified infrared (UIR) emission bands at 3.3, 6.2, 7.7, 8.6, 11.2, and 12.7 μm.^9–11 Variations in the line shape of the 6.2-μm band observed, for instance, in the Spitzer/IRS spectra of starburst dominated galaxies¹² suggest a potential role for nitrogen-substituted PAH (PANH) and their derivatives.^13,14 PANH have previously been detected in samples from the Murchison meteorite.^15,16 If PANH contribute to the UIR bands and are present in meteorites, they might also be DIB carriers. However, electronic absorption spectra of PANH and their derivatives at low temperatures have scarcely been recorded, and, to the best of our knowledge, no such spectra have been reported for hydrogenated PANH to date.

The low vapor pressures of large PAH and PANH pose significant challenges for obtaining their gas-phase electronic spectra. Unlike their protonated counterparts, hydrogenated PAH and PANH cannot be stored in cryogenic ion-traps, and their low-temperature gas-phase spectra have rarely been reported. Matrix isolation techniques provide an alternative, as they require lower sample concentrations and allow the matrix to be maintained at temperatures relevant to interstellar conditions. Depending on the matrix host, the preparation of radicals or ions in matrices can be relatively straightforward. Interactions between the matrix host and the isolated guest molecules, however, can induce substantial spectral shifts: matrix shifts up to several hundred wavenumbers have been reported for electronic transitions of molecules isolated in solid rare gases.^17,18 In contrast, the quantum solid para-hydrogen (para-H₂) is soft and interacts with guest molecules only weakly. Protonated molecules can be conveniently prepared by proton-transfer reaction from H₃⁺, which can be generated in situ by electron bombardment of para-H₂. Corresponding hydrogenated molecules are coproducts in this process through either neutralization of protonated species or reactions of the parent molecule with H atoms, a side product of H₃⁺ generation. Infrared (IR) absorption spectra of PAH, PANH, and their protonated and mono-hydrogenated derivatives, produced from electron bombardment of para-H₂ matrices containing trace PAH or PANH, mostly exhibit narrow line widths and small matrix shifts, typically <1%, induced by the solid para-H₂ environment.^19–21 Despite these advantages, para-H₂ matrix-isolation spectroscopy has rarely been applied to study electronic transitions. In recent years, we have recorded dispersed fluorescence and fluorescence excitation spectra of several PAH and their protonated and hydrogenated derivatives containing ≤42 carbon atoms in para-H₂ matrices and found less-divergent and consistent red shifts relative to the gas-phase, compared to Ne and Ar matrices.^22–25 PANH and their derivatives, however, have not yet been examined under similar conditions in our studies.

Quinoline (C₉H₇N) and iso-quinoline (iso-C₉H₇N), the smallest members of the PANH family, are depicted in Fig. 1. As core structural motifs in iso-quinoline alkaloids such as morphine and DTC (d-tubercurarine chloride), iso-quinoline has long been of interest in organic, bioorganic, and pharmaceutical synthesis and analysis.²⁶ Recently, with the growing interest in molecular hydrogen as a clean energy carrier and an alternative to fossil fuels, iso-C₉H₇N, C₉H₇N, and their substituted derivatives have attracted increasing attention as potential liquid organic hydrogen carriers (LOHC). In LOHC, molecular hydrogen is stored and released through catalytic hydrogenation and dehydrogenation of H₂-lean and H₂-rich molecular pairs. The efficiency of hydrogen storage and retrieval depends on the thermodynamic properties of these molecular systems and the availability of suitable catalysts, resulting in an increased interest in the properties of partially and per-hydrogenated PANH.^27–29 Despite this relevance, spectral studies of these molecules remain limited.

Fig. 1 Molecular structures and skeleton-atom numbering of the two smallest PANH quinoline (C₉H₇N (a)) and iso-quinoline (iso-C₉H₇N (b)).

To the best our knowledge, the only study of mono-hydrogenated iso-quinoline, iso-HC₉H₇N, is from our laboratory. In two types of experiments, Joshi et al.³⁰ prepared various isomers of iso-HC₉H₇N in solid para-H₂ either by electron bombardment or by reacting iso-C₉H₇N with H atoms generated from the reaction of Cl atoms with vibrationally excited H₂. The recorded IR absorption features were grouped according to their behaviour upon secondary photo-irradiation at various wavelengths between 365 and 553 nm and assigned to eight isomers of iso-HC₉H₇N according to comparison with density functional theory (DFT) calculations. Hydrogenation on the N atom and on all accessible carbon atoms except two at the bridged sites 4a and 8a (Fig. 1) was observed. According to the potential energy scheme (PES) computed at the CCSD(T) level, iso-C₉H₇NH (or 2-iso-HC₉H₇N, hydrogenation at the N-site) has the lowest relative energy. Isomers 5-, 4-, 8, and 1-iso-HC₉H₇N lie within 6 kJ mol⁻¹, whereas 3- and 6-iso-HC₉H₇N have much larger energies, 22–24 kJ mol⁻¹ above iso-C₉H₇NH.

We extend our study of the electronic spectra of PAH and their derivatives isolated in solid para-H₂ to hydrogenated iso-quinoline and present dispersed fluorescence and fluorescence excitation spectra of three isomers, 4-iso-HC₉H₇N, 5-iso-HC₉H₇N, and 8-iso-HC₉H₇N, characterized by three distinct progressions originating at 18 492, 18 688, and 19 430 cm⁻¹, respectively, with fluorescence lifetimes of ∼6, <50, and ∼105 ns. Spectral assignments are derived from a comparison to vibrationally resolved electronic absorption and emission spectra simulated using a Franck–Condon Herzberg–Teller approach on the basis of quantum-chemically optimized geometries and scaled harmonic vibrational frequencies. In general, the spectra are mostly composed of vibrational normal modes of a′ symmetry. The dispersed fluorescence spectra are compared with the IR absorption spectra of the same isomers in solid para-H₂ reported by Joshi et al.,³⁰ and spectral differences are discussed. A brief comparison to astronomical DIB data is provided.

2. Experimental and computational methods

The para-H₂/laser-induced fluorescence experiment has been described in detail elsewhere;^23,31 therefore, only a brief overview is given here. The substrate, a nickel-coated copper flat mounted onto the second stage of a closed-cycle helium refrigerator, is cooled to ∼3.5 K and serves as the reflective surface for spectroscopy.

Para-H₂ was prepared by passing normal-H₂ (Chiah-Lung, purity 99.9999%) over a Fe(III) oxide catalyst cooled to ∼13 K; the minimum mole fraction of ortho-H₂ achievable at this temperature is ∼100 ppm. Due to the low vapour pressure of iso-C₉H₇N, preparing pre-mixed C₉H₇N/para-H₂ samples was not feasible. Instead, a small flow of para-H₂ (flow rate ∼15 STP L min⁻¹, in which STP stands for standard temperature at 273 K and pressure at 760 torr) was passed over the liquid sample, which was purified through three freeze–pump–thaw cycles prior to use, maintained at room temperature. To produce hydrogenated iso-HC₉H₇N, the sample mixture was bombarded during deposition with an electron beam at 270 eV and 8 μA, generated by an electron gun (Kimball Physics, EPG-7). A typical deposition lasted 5–6 h. Matrix deposition was monitored with a Fourier-transform infrared spectrometer (FTIR, Bruker, iFS66v) equipped with a KBr beam splitter and a Hg–Cd–Te (MCT) detector cooled to 77 K. IR spectra were typically acquired over 300 scans and covered the range 500–4500 cm⁻¹ at a resolution of 0.25 cm⁻¹.

To record dispersed fluorescence spectra, we irradiated the matrix with the output of an optical parametric oscillator (OPO, EKSPA NT340) laser, pumped by the output of a frequency-tripled Nd:YAG laser (EKSPLA, NT300) operating at 10 Hz. The OPO laser beam was expanded to a diameter of ∼1.5 cm with a telescope to maximize the overlap with the sample; the laser energy at the entrance window was ∼1 mJ. Emission from the matrix was collected with a convex lens and a collimator, coupled into an optical fiber, and transmitted to the spectrograph consisting of a monochromator (Andor Shamrock SR500i, focal length 0.5 m) and an intensified charge-coupled device (iCCD, Andor iStar DH320T-18U-73, 1024 × 225 pixels, pixel size 26 μm × 26 μm). We employed a 600 grooves per mm grating blazed at 500 nm and, for the recording of dispersed fluorescence spectra, a monochromator entrance slit width of 22 μm, resulting in a linear dispersion of 3.33 nm mm⁻¹; in this configuration, each pixel of the iCCD corresponds to 0.086 nm. Dispersed fluorescence spectra were corrected for wavelength-dependent variations in grating efficiency and iCCD photocathode quantum efficiently. Wavelengths were calibrated with low-pressure Hg, Ne, and Kr lamps. Fluorescence excitation spectra were acquired by stepping the excitation laser wavelength in increments of 0.1 nm, corresponding to 4.9–3.3 cm⁻¹ across the range 450–550 nm, while probing fluorescence emission in a specific wavelength range. In the spectral range covered, the spectral linewidth of the OPO output is <4 cm⁻¹. Laser-induced fluorescence (LIF) experiments were typically conducted over 2–3 days and no significant changes in signal intensities in the recorded dispersed fluorescence spectra were observed over this period.

Quantum-chemical calculations were performed with the Gaussian 16 program package, Revision B.01.³² We computed optimized geometries and harmonic vibrational frequencies of the electronic ground state (D₀) and electronically excited states with the uB3PW91/6-311++G(2d,2p) and TD-uB3PW91/6-311++G(2d,2p) methods, respectively. To simulate vibrationally resolved electronic absorption and emission spectra, we employed a Franck–Condon Herzberg–Teller approach;³³ the computed stick spectra were convoluted with Gaussian line shapes with full-width at half-maximum (FWHM) selected to match with the experimental spectra. For 1-iso-HC₉H₇N, the vibrationally-resolved absorption and emission spectra were calculated with a local modified version of Gaussian to achieve a higher progression by using internal coordinates and reduced dimensionality schemes.^34,35 Harmonic vibrational wavenumbers were scaled by 0.978 to account for systematic deviations in the calculations. This scaling factor was derived from a comparison between vibrational frequencies computed at this level of theory and the IR spectra of selected PAH isolated in solid para-H₂ in previous studies.²³ Due to the lack of independent experimental vibrational frequencies for electronically excited states, the determination of a separate scaling factor for excited-state vibrational frequencies was not possible. We therefore applied the same scaling factors to harmonic wavenumbers of both ground and excited states. Our previous studies support this approach: for peri-HBC,²² the mean absolute deviations in peak positions inferred from fluorescence excitation spectra and theoretical predictions at the B3PW91/6-311++G(2d,2p) level was 8 ± 6 cm⁻¹.

3. Results and discussion

3.1. Quantum-chemical calculations

Hydrogenation of iso-C₉H₇N can proceed at ten distinct atomic sites, as indicated in Fig. 1. Among these ten possible isomers, iso-C₉H₇NH, formed via H-atom addition to the N atom, is the most stable. In contrast, the two isomers 4a- and 8a-iso-HC₉H₇N, resulting from H-atom addition to the bridging C atoms, exhibit the highest relative energies of +100 and +91 kJ mol⁻¹, respectively.³⁰ Oscillator strength (f) and vertical absorption and emission energies corresponding to the transition between the electronic ground state (D₀) and the first electronically excited state (D₁) of the eight most stable isomers calculated with the TD-uB3PW91/6-311++G(2d,2p) method are summarized in Table 1; their simulated D₁ → D₀ emission spectra except 1-iso-HC₉H₇N, are compared in Fig. 2(d)–(j).

Table 1 Relative energies of various iso-HC₉H₇N isomers, vertical absorption and emission wavelengths, and vertical absorption oscillator strengths f of their D₁–D₀ transitions and D₁ spin contamination 〈S²〉

	Rel. energy^a/kJ mol^−1	TD-uB3PW91/6-311++G(2d,2p)
		ΔE^vertabs/nm	ΔE^vertemi/nm	f	〈S^2〉
a Relative energies reported by Joshi et al.,^30 computed at the CCSD(T)/6-311++G(d,p)//B3LYP/6-311++G(d,p) level.
1	6	525.30	1179.79	0.001	0.842
2	0	839.58	2268.43	0.001	0.837
3	22	514.06	664.25	0.004	0.895
4	3	447.47	522.88	0.025	0.907
5	2	448.55	528.04	0.006	0.890
6	24	498.47	592.81	0.005	1.025
7	24	539.75	703.92	0.007	0.951
8	3	432.48	483.62	0.006	0.880

---

Fig. 2 Comparison of observed dispersed fluorescence spectra with simulated D₁ → D₀ emission spectra of various isomers of hydrogenated iso-quinoline. (a) Experimental system A, (b) Experimental system B, and (c) Experimental system C (partial). Simulated emission spectra of various isomers of iso-HC₉H₇N are shown in (d)–(j). To facilitate direct comparison, all spectra were shifted to align their 0⁰₀ bands and the computed Franck–Condon Herzberg–Teller stick spectra were convoluted with a Gaussian line shape function with a full-width half-maximun (FWHM) of 35 cm⁻¹. Relative energies in kJ mol⁻¹ are given in parentheses. The origin bands of the experimental spectra are located at 18 492 cm⁻¹ (system A), 19 430 cm⁻¹ (system B), and 18 688 cm⁻¹ (system C, not shown).

According to the simulations, the most intense peak in the depicted spectra corresponds to the vibrationless origin. Vibronic bands are less intense, with maximum intensities <45% relative to the origin band. In the D₀ state, the ring-skeleton is planar, and, for most isomers except 1-iso-HC₉H₇N, this planarity is preserved upon excitation to the D₁ state. Consequently, their vibronic spectra are composed mainly of a′-symmetric vibrational normal modes and their combination bands. For 2-iso-HC₉H₇N (iso-C₉H₇NH), our calculations predict a slight out-of-plane deformation of the ring-skeleton, as indicated by planarity indices^22,36 of 2.2 and 0.2 for the N-substituted and all-C-ring moieties, respectively, resulting in lower intensities in the vibronic spectrum and vibrational progressions extending up to ∼5000 cm⁻¹.

In contrast, for 1-iso-HC₉H₇N, TD-DFT calculations predict substantial deformation of the ring-skeleton of the optimized D₁ geometry, as shown in Fig. S1 (SI). The N-substituted ring has a planarity index of 28.2, whereas the all-C ring has a planarity index of 3.9. Due to the pronounced geometric disparity between D₁ and D₀, Franck–Condon factors for fundamental vibrational normal modes and first-order overtone and combination bands are small, and the Franck–Condon Herzberg–Teller algorithm, implemented in Gaussian 16, Revision B.01,³² fails to converge. Consequently, no simulated D₁ → D₀ emission spectrum of 1-HC₉H₇N is depicted in Fig. 2. With new definitions of internal coordinates and a reasonable reduced dimensionality, we achieved a final spectral progression of 78% (original calculation yielded 0.01%) for the D₁ → D₀ emission spectrum of 1-iso-HC₉H₇N and obtained the spectrum depicted in Fig. S2 (SI). The high periodicity of the spectral structure and the substantially reduced line density, when additional six vibrational normal modes are omitted, reflect the significant change in geometry predicted for the D₁–D₀ transition at this level of theory; the dominance of long progressions of overtone and combination bands was observed.

Although the vibronic patterns in Fig. 2(d)–(j) vary considerably depending on the position of H addition, several notable trends emerge. Contributions of vibrational normal mode in the observed transitions are discussed in detail in Section SA (SI).

3.2. Dispersed fluorescence of 4-iso-HC₉H₇N and 8-iso-HC₉H₇N

Upon excitation at 497.0 (20 120 cm⁻¹) and 475.0 nm (21 053 cm⁻¹) of a matrix of iso-C₉H₇N in solid para-H₂ deposited under electron bombardment, two distinct emission patterns were observed with most intense features at 18 492 cm⁻¹ (540.8 nm; system A) and 19 430 cm⁻¹ (514.7 nm; system B), as shown in Fig. 2a and b, and also in Fig. S5 (SI). To determine their lifetimes, we probed temporal profiles of fluorescence emission and fitted the decay curves with a mono-exponential decay function y(t) = y₀ + A × exp(−τ/t), in which τ denotes the excited-state lifetime (Fig. S6, SI). System A exhibits a fluorescence lifetime of ∼6 ns, whereas system B yields a longer lifetime of 104 ± 3 ns, indicative of two distinct electronic transitions.

The observed dispersed fluorescence spectra are compared with simulated D₁ → D₀ emission spectra of various hydrogenated iso-quinoline isomers in Fig. 2. To facilitate comparison, all simulated vibronic stick spectra were shifted to align their 0⁰₀ bands and convoluted with a Gaussian lineshape (FWHM of 35 cm⁻¹). Characteristic of system A are two intense and partially unresolved features at 1507 and 1577 cm⁻¹, with relative intensities of 45 and 41% compared to the 0⁰₀ band, respectively. A moderately intense feature at 615 cm⁻¹ with a relative intensity of ∼25% also appeared, alongside five weaker bands (<20%) located between 615 and 1500 cm⁻¹. System B displays two characteristic bands ∼1175 and 1570 cm⁻¹, both with relative intensities of ∼24%, accompanied by three weaker features below 1000 cm⁻¹ with relative intensities under 10%.

The simulated D₁ → D₀ emission spectra of 3-, 4-, and 7-iso-HC₉H₇N all exhibit a characteristic double-peak structure in region 1500–1600 cm⁻¹, as observed in system A. However, only the simulated spectrum of 4-iso-HC₉H₇N reproduces the relative intensities of these two peaks in the experimental spectrum. Furthermore, in the 500–1500 cm⁻¹ region, the most intense peak in the simulated spectrum of 4-iso-HC₉H₇N appears near 617 cm⁻¹ (rel. int. ∼25%), consistent with experiments, and the remaining spectral features are also consistent with our experimental data. In contrast, the simulated spectra of 3- and 7-iso-HC₉H₇N both show features near 495 and 350 cm⁻¹ with relative intensities of 35 and 20%, but lack any prominent feature near 615 cm⁻¹. We therefore assign the dispersed fluorescence spectrum originating at 18 492 cm⁻¹ (system A) to the D₁ → D₀ transition of 4-iso-HC₉H₇N in solid para-H₂.

Detailed vibrational assignments for the dispersed fluorescence spectrum of 4-iso-HC₉H₇N isolated in solid para-H₂, derived on comparison with simulations, are presented in Fig. 3 and Table 2. The relative intensities and peak positions show satisfactory agreement, with an average absolute deviation of 8 ± 6 cm⁻¹, comparable to our results for the 1-hydronaphthyl radical (1-C₁₀H₉) in solid para-H₂.²⁴ Most features are assigned to vibrational normal modes of a′ symmetry, consistent with the preservation of ring-skeleton planarity and the symmetric deformation of the HCH angle and C–H bond lengths in the CH₂ moiety upon excitation to the D₁ state. The predicted weak contributions from overtone and combination bands, either involving a′ vibrations or two quanta of a′′ vibrations (ν₄₅–ν₄₈), are consistent with the allowed transitions governed by symmetry operations of the point group C_s.

Fig. 3 Comparison of observed and simulated dispersed fluorescence spectra of 4-iso-HC₉H₇N. (a) Dispersed fluorescence spectrum of 4-iso-HC₉H₇N isolated in solid para-H₂ recorded upon excitation at 20 120 cm⁻¹ (497.0 nm). The bottom axis denotes the displacement from the orgin band, observed at 18 492 cm⁻¹ (540.8 nm), while the top axis represents the wavelength of the observed spectrum. (b) The D₁ → D₀ emission spectrum simulated by convolution the Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian line shape of FWHM 40 cm⁻¹; vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

Table 2 Assignments of peaks in the dispersed fluorescence spectrum of 4-iso-HC₉H₇N isolated in solid para-H₂

para-H2		uB3PW91			Assignment	Sym.
IR^a/cm^−1	LIF^b/cm^−1	FCHT^c/cm^−1	Rel. int.^d/%	Scaled^e/cm^−1
a Peak positions in the IR absorption spectrum of 4-iso-HC9H7N isolated in solid para-H2 reported by Joshi et al.^30 Values in parentheses represent modes predicted by the Frank–Condon Herzberg–Teller (FCHT) simulations with a relative intensity of <1.5% in the simulated LIF spectrum. b Vibrational spacing relative to the origin band at 18 492 cm^−1 (540.8 nm). c Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. d Intensities relative to the most intense peak (0^00 band). Only vibrational modes with relative intensities ≥1.5% are listed. The weak features corresponding to IR absorption bands reported by Joshi et al.^30 are listed in parentheses. e Harmonic vibrational wavenumbers scaled by a factor of 0.978. Values given in parentheses for combination and overtone bands are the sum of the fundamentals.
	0	0	100	0	0^00	a′
		113	5.9	(112)	48^02	a′
	344	336	5.1	(331)	47^02	a′
			3.3	344	32^01	a′
		399	5.6	(398)	46^0147^01	a′
	486	466	3.0	(465)	46^02	a′
	615	610	17.3	612	29^01	a′
			1.5	(642)	44^0147^01	a′
			2.2	(653)	45^0146^01	a′
(748.1)			(<0.1)	742	40^01	a′′
	786	778	7.0	780	27^01	a′
	945	943	3.5	940	25^01	a
			2.0	(953)	44^02	a′
	1036	1037	4.2	1034	24^01	a′
	1188	1181	2.4	1113	23^01	a′
			2.6	1151	21^01	a′
			5.4	1183	20^01	a′
			1.9	1217	19^01	a′
			2.2	(1223)	29^02	a′
(1245.9)			(<0.1)	1247	18^01	a′
1269.4	1304	1318	1.6	1277	17^01	a′
			4.0	1320	16^01	a′
		1389	1.8	(1391)	27^0129^01	a′
(1483.7)			(0.8)	1486	11^01	a′
	1507	1514	21.8	1513	10^01	a′
	1577	1584	10.7	1576	9^01	a′
			8.3	1602	8^01	a′
(3087.9)			(<0.1)	3132	1^01	a′

---

The IR spectrum of 4-iso-HC₉H₇N in solid para-H₂ has previously been reported by Joshi et al.³⁰ Consistent with the low molecular symmetry of the radical (point group C_s), all vibrational normal modes are symmetry allowed and may contribute to both IR and LIF spectra, although their intensities might differ depending on the governing rules for electronic and vibronic transitions. In electronic spectra, the intensities of individual vibrational modes are determined by their Franck–Condon factors, which are more pronounced for vibrational modes reflecting geometry changes induced by electronic transition. Since the planarity of the ring-skeleton is retained upon electronic excitation, intense features observed in the dispersed fluorescence spectra are predominantly associated with vibrational normal modes of a′ symmetry. Conversely, peak intensities in IR spectra are governed by the change in dipole moment induced by the vibrational motion, and, in consequence, the most intense IR bands (ν₄₀) are associated with vibrational modes of a′′ symmetry.

Joshi et al.³⁰ identified five IR lines of 4-iso-HC₉H₇N in solid para-H₂ and assigned them to ν₁ (a′), ν₁₁ (a′), ν₁₇ (a′), ν₁₈ (a′), and ν₄₀ (a′′); other IR lines predicted to have significant intensities were interfered by either iso-C₉H₇N or other isomers of iso-HC₉H₇N. Among these modes, our simulations predict that only ν₁₇ (a′) and ν₁₁ (a′) contribute weakly to the electronic emission spectrum, with intensities 1.6 and 0.8% of the most intense feature 0⁰₀, respectively (Table 2). The band at 1304 cm⁻¹ in the dispersed fluorescence spectrum is predicted to be a superposition of ν₁₇ (1277 cm⁻¹, a′) and ν₁₆ (1320 cm⁻¹, a′), with relative intensities 1.6 and 4.0%, respectively, resulting in an effective peak position at 1308 cm⁻¹. The corresponding peak in the IR spectrum of 4-iso-HC₉H₇N in solid para-H₂ at 1269.4 cm⁻¹ is assigned to only ν₁₇, with no contribution from ν₁₆, explaining the large deviation between the two spectra.

For system B, among all simulated spectra, only the simulated D₁ → D₀ emission spectrum of 8-iso-HC₉H₇N exhibits two pes of comparable intensity (∼22%) near 1178 and 1567 cm⁻¹, similar to the experimental spectrum. A third, weaker feature (rel. intensity ∼15%) is predicted near 620 cm⁻¹ and also observed in the experimental spectrum, albeit with slightly lower intensity (∼9%); other predicted features are generally in agreement with the experimental data. The only other isomer whose simulated D₁ → D₀ spectrum displays characteristic features near 1180 and 1580 cm⁻¹ is iso-C₉H₇NH, the lowest-energy isomer. In this case, however, the two bands are predicted with much higher relative intensities (∼63 and ∼40%, respectively), and the predicted bands near 355, 490, and 1420 cm⁻¹ were not observed in our experiments.

Therefore, we assign system B, with an origin at 19 430 cm⁻¹, to the D₁ → D₀ transition of 8-iso-HC₉H₇N in solid para-H₂. A contribution from protonated iso-C₉H₇N to either spectrum is unlikely because our calculations predict the S₁–S₀ transition of iso-C₉H₇NH⁺, the lowest-energy protonated isomer and the only one observed previously in IR absorption in solid para-H₂,³⁰ to occur at 376.6 nm with an oscillator strength f = 0.442. Joshi et al. differentiated protonated and hydrogenated iso-C₉H₇N based on their behavior over time: recombination of iso-C₉H₇NH⁺ with solvated electrons in the para-H₂ matrix resulted in a decrease in iso-C₉H₇NH⁺ and an increase in iso-HC₉H₇N. No decrease in intensities in the dispersed fluorescence spectra were observed over the experimental runtime of 2–3 days, supporting that iso-C₉H₇NH⁺ is unlikely to contribute to the observed spectra.

Detailed vibrational assignments for the dispersed fluorescence spectrum of 8-iso-HC₉H₇N isolated in solid para-H₂, derived on comparison with the simulations, are presented in Fig. 4 and Table 3. The simulated and observed peak positions and relative intensities are in satisfactory agreement, with an absolute average deviation 6 ± 4 cm⁻¹. As discussed above for 4-iso-HC₉H₇N, most of the observed peaks are assigned to vibrational normal modes of a′ symmetry.

Fig. 4 Comparison of observed and simulated dispersed fluorescence spectra of 8-iso-HC₉H₇N. (a) Dispersed fluorescence spectrum of 8-iso-HC₉H₇N isolated in solid para-H₂ recorded upon excitation at 21 053 cm⁻¹ (475.0 nm). The bottom axis denotes the displacement from the orgin band, observed at 19 430 cm⁻¹ (514.7 nm), while the top axis represents the wavelength of the observed spectrum. (b) The D₁ → D₀ emission spectrum simulated by convolution the Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian line shape of FWHM 75 cm⁻¹; vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

Table 3 Assignments of peaks in the dispersed fluorescence spectrum of 8-iso-HC₉H₇N isolated in solid para-H₂

para-H2	uB3PW91				Assignment	Sym.
IR^a/cm^−1	LIF^b/cm^−1	FCHT^c/cm^−1	Rel. int.^d/%	Scaled^e/cm^−1
a Peak positions in the IR absorption spectrum of 8-iso-HC9H7N isolated in solid para-H2 reported by Joshi et al.^30 Values in parentheses represent modes predicted by FCHT simulations with a relative intensity of <1.5% in the simulated spectrum. b Vibrational spacing relative to the origin band at 19 430 cm^−1 (514.7 nm). c Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. d Intensities relative to the most intense peak (0^00 band). Only vibrational modes with relative intensities ≥1.5% are listed. The weak features corresponding to IR absorption bands reported by Joshi et al.^30 are listed in parentheses. e Harmonic vibrational wavenumbers scaled by a ftor 0.978. Values given in parentheses for combination and overtone bands are the sum of the fundamentals.
	0	0	100	0	0^00	a′
	474	473	2.6	472	31^01
	618	618	13.6	616	29^01	a′
(653.7)			(<0.1)	648	42^01	a′′
	743	778	2.6	784	27^01	a′
(830.5)			(<0.1)	831	39^01	a′′
	936	919	1.5	910	26^01	a′
941.6			1.7	936	25^01	a′
			3.2	1056	24^01	a′
			1.2	(1091)	43^02	a′
	1174	1166	6.7	1150	22^01	a′
			16.2	1173	21^01	a′
			1.2	(1232)	29^02	a′
(1245.5)			(<0.1)	1264	18^01	a′
			1.4	1404	14^01	a′
(1432.2)			(<0.1)	1431	12^01	a′
(1467.6)			(<0.1)	1473	11^01	a′
	1560	1553	10.7	1528	10^01	a′
			11.0	1558	9^01	a′
1576.2			7.8	1590	8^01	a′
	1792	1784	2.0	(1789)	21^0129^01	a′
		2175	1.7	(2144)	10^0129^01	a′
			2.4	(2174)	9^0129^01	a′
			1.2	(2206)	8^0129^01	a′
	2311	2332	1.0	(2322)	21^0122^01	a′
			1.3	(2345)	21^02	a′
	2737	2723	1.3	(2700)	10^0121^01	a′
			1.5	(2731)	9^0121^01	a′
			1.2	(2762)	8^0121^01	a′
	3124	3106	1.4	(3086)	9^0110^01	a′
			1.9	(3116)	9^02	a′
(3076.6)			(<0.1)	3119	2^01	a′
(3083.9)			(<0.1)	3130	1^01	a′

---

For 8-iso-HC₉H₇N, Joshi et al.³⁰ assigned nine IR vibrational lines to two C–H stretching modes ν₁ (a′) and ν₂ (a′), five C–C and C–N stretching modes ν₈ (a′), ν₁₁ (a′) and ν₁₂ (a′), ν₁₈ (a′), ν₂₅ (a′), and two out-of-plane vibrations ν₃₉ (a′′) and ν₄₂ (a′′). Of these, only ν₂₅ (a′) and ν₈ (a′) are predicted to contribute to the D₁ → D₀ emission spectrum, with relative intensities of 1.7 and 7.8%, respectively.

A notable distinction between the dispersed fluorescence spectra of 4- and 8-iso-HC₉H₇N lies in the relative intensities of the three a′ symmetric C–C and C–N stretching modes ν₈, ν₉, and ν₁₀. In the simulated D₁–D₀ emission spectrum of 4-iso-HC₉H₇N, their relative intensities follow the order ν₁₀ (21.8%) ≫ ν₉ (10.7%) > ν₈ (8.3%), whilst in the simulated D₁–D₀ emission spectrum of 8-iso-HC₉H₇N, ν₁₀ and ν₉ exhibit comparable relative intensities – 10.7 and 11.0%, respectively – both exceeding that of ν₈ (7.9%). Atomic displacement vectors induced by these modes are illustrated in Fig. 5. In the 4-iso-HC₉H₇N radical, ν₁₀ primarily affects the C–N bond, and, to a lesser extent, the C(8a)–C(1) bond; the C-skeleton of the all-C ring is largely unaffected. On the contrary, ν₉ and ν₈ mostly distort the C-skeleton of the all-C ring, with the largest impact on C(6)-C(7) and C(4a)–C(8a), and C(4a)–C(5) and C(7)–C(8), respectively.

Fig. 5 Displacement vectors induced by the three a′ C–C stretching modes of 4-iso-HC₉H₇N (a) and 8-HC₉H₇N (b) in their electronic ground states calculated at the uB3PW91/6-311++G(2d,2p) level.

As illustrated in Fig. S4b (SI), the greatest change in bond length upon transition between the D₁ and D₀ in 4-iso-HC₉H₇N occurs at the C–N bond. Among the three pairs of C–C and C–N bonds mostly impacted by the three normal modes ν₈, ν₉, and ν₁₀, the pair C(4a)–C(5) and C(7)–C(8) corresponding to ν₈ is least affected by the electronic transition, consistent with the low intensity of ν₈ predicted in the D₁–D₀ emission spectrum. Changing the hydrogenation site from C(4) to C(8) also changes the displacements associated with the three vibrational normal modes ν₈, ν₉, and ν₁₀: in 8-iso-HC₉H₇N, ν₁₀ mainly affects the C(6)–C(7) bond and, to a lesser extent, C(4a)–C(5); ν₉ mainly impacts the C–N and C(4a)–C(8a) bonds, and ν₈ mainly impacts the C(8a)–C(1) and C(3)–C(4) bonds. Similar to the 4-iso-HC₉H₇N radical, the change in bond lengths of 8-iso-HC₉H₇N associated with the D₁–D₀ transition is smallest for the pair of C–C bonds characteristic of ν₈, and the intensity predicted for ν₈ in the emission spectrum is comparably small. The changes in bond lengths of the C(6)–C(7) and the C–N bonds (+3.1 and +3.3%, respectively) in 8-iso-HC₉H₇N are similar, consistent with the comparable intensities of ν₁₀ and ν₉ in the simulated D₁–D₀ emission spectrum.

3.3. Fluorescence excitation spectrum of 4-iso-HC₉H₇N

Probing fluorescence emission in the ranges 18 399–18 601 cm⁻¹ (537.6–543.5 nm) and 16 787–17 030 cm⁻¹ (587.2–595.7 nm), while varying the excitation wavelength at 0.1 nm increments from 490 to 516 nm and 516 to 545 nm, respectively, we obtained the fluorescence excitation spectrum depicted in Fig. 6a. This spectrum consists of a very intense 0⁰₀ band at 18 495 cm⁻¹ (540.7 nm) along with several weaker vibrational bands with relative intensities <25%. The 0⁰₀ band closely matches the 0⁰₀ band at 18 492 cm⁻¹ in the dispersed fluorescence spectrum of 4-iso-HC₉H₇N, shown in Fig. 3a. The probed range 16 787–17 030 cm⁻¹ covers the weaker bands of 8⁰₁, 9⁰₁, and 10⁰₁. in the dispersed fluorescence spectrum, whereas the probed range 18 399–18 601 cm⁻¹ covers the intense 0⁰₀ band. Consequently, the signal-to-noise ratio is superior below 516 nm. According to the overlay of dispersed fluorescence spectra of the three iso-HC₉H₇N isomers recorded in this study and the probed regions shaded in gray (Fig. S7, SI), interferences due to emission of 8-iso-HC₉H₇N and 5-iso-HC₉H₇N are expected to be negligible.

Fig. 6 Comparison of observed and simulated excitation of 4-iso-HC₉H₇N. (a) Fluorescence excitation spectrum of 4-iso-HC₉H₇N isolated in solid para-H₂ obtained by probing emission in regions 16 787–17 030 cm⁻¹ (<900 cm⁻¹) and 18 399–18 601 cm⁻¹ (>900 cm⁻¹). The bottom axis denotes the displacement from the orgin band, observed at 18 495 cm⁻¹ (540.7 nm), while the top axis represents the wavelength of the observed spectrum. (b) The D₁ ← D₀ absorption spectrum simulated by convolution the Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian line shape of FWHM 25 cm⁻¹; vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

The experimental spectrum is compared with the simulated D₁ ← D₀ absorption spectrum of 4-iso-HC₉H₇N in Fig. 6. Detailed assignments for the observed features are provided in Fig. 6b and Table 4. In general, the experimental and theoretical results are in satfactory agreement, with an average absolute deviation in peak positions of 8 ± 7 cm⁻¹. We also compare this spectrum to the simulated D₁ ← D₀ absorption spectra of various iso-HC₉H₇N isomers in Fig. S8 (SI). The observed excitation spectrum (trace a) agrees with that predicted for 4-iso-HC₉H₇N, but not with predictions for other isomers of iso-HC₉H₇N.

Table 4 Assignments of peaks in the fluorescence excitation spectrum of 4-iso-HC₉H₇N isolated in solid para-H₂

para-H2	uB3PW91			Assignment	Sym.
LIF^a/cm^−1	FCHT^b/cm^−1	Rel. int.^c/%	Scaled^d/cm^−1
a Vibrational spacing relative to the origin band at 18 495 cm^−1 (540.7 nm). b Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. c Intensities relative to the most intense peak (0^00 band). Only vibrational modes contributing with a relative intensity ≥1.5% are listed. d Harmonic vibrational wavenumbers scaled by a factor 0.978. Values given in parentheses for combination and overtone bands are the sum of the fundamentals.
	0	100	0	0^00	a′
48	51	29.6	(51)	48^20	a′
	102	13.1	(102)	48^40	a′
	153	6.5	(153)	48^60	a′
215	203	3.3	(204)	48^80	a′
		2.2	(228)	47^20	a′
		1.8	(255)	48^100	a′
338	336	1.8	336	32^10	a′
	528	2.0	(530)	43^1047^10	a′
593	602	20.1	604	29^10	a′
655	653	5.9	(655)	29^1048^20	a′
703	704	2.6	(706)	29^1048^40	a′
751	755	8.1	755	27^10	a′
	806	2.4	(806)	27^1048^20	a′
957	951	3.7	950	25^10	a′
1022	1025	5.4	1026	23^10	a′
1152	1150	3.2	1125	22^10	a′
		11.1	1151	20^10	a′
	1205	3.3	(1202)	20^1048^20	a′
1250	1256	3.9	1357	18^10	a′
1304	1307	3.8	1299	17^10	a′
		3.7	1313	16^10	a′
		2.6	1335	15^10	a′
1410	1428	14.5	1428	11^10	a′
1498	1483	4.3	(1479)	11^1048^20	a′
		1.8	1492	10^10	a′
	1534	1.9	(1530)	11^1048^40	a′
		2.6	1532	9^10	a′
1620	1623	31.5	1624	8^10	a′
1700	1675	9.3	(1675)	8^1048^20	a′
	1725	4.1	(1726)	8^1048^40	a′
	1772	1.8	(1755)	20^1029^10	a′
		2.0	(1777)	8^1048^60	a′

---

Although both dispersed fluorescence and fluorescence excitation spectra of 4-iso-HC₉H₇N are primarily composed of vibrational normal modes of a′ symmetry, they differ in relative intensities and in the contribution of combination and overtone bands: overtones containing 2n quanta of ν₄₈ (a′′, 26 cm⁻¹) up to n = 5 and their combination bands with intense modes of a′ symmetry are predicted to contribute only to the fluorescence excitation spectrum.

3.4. Fluorescence excitation spectrum of 8-iso-HC₉H₇N

To obtain the fluorescence excitation spectrum of 8-iso-HC₉H₇N, we monitored fluorescence emission in the range 18 103–18 312 cm⁻¹ (546.1–552.4 nm), shaded in light red in Fig. S7 (SI), corresponding to a superposition of bands 22⁰₁ and 21⁰₁ in the dispersed fluorescence spectrum. The signal-to-noise ratio of this excitation spectrum is worse than that of 4-iso-HC₉H₇N, likely due to the oscillator strength predicted for the D₁–D₀ transition of 8-iso-HC₉H₇N being about 1/4 that of 4-iso-HC₉H₇N. This spectrum, depicted in Fig. 7a, consists of an intense 0⁰₀ band at 19 433 cm⁻¹ (514.6 nm), closely matching the 0⁰₀ band at 19 430 cm⁻¹ in the dispersed fluorescence spectrum of 8-iso-HC₉H₇N; characteristic vibrational features appeared at 700, 1169, 1532, and 1620 cm⁻¹.

Fig. 7 Comparison of observed and simulated excitation of 8-iso-HC₉H₇N. (a) Fluorescence excitation spectrum of 8-iso-HC₉H₇N isolated in solid para-H₂ obtained by probing emission at 18 103–18 312 cm⁻¹. The bottom axis denotes the displacement from the orgin band, observed at 19 433 cm⁻¹ (514.6 nm), whereas the top axis represents the wavelength of the observed spectrum. (b) The D₁ ← D₀ absorption spectrum simulated by convolution the Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian line shape of FWHM 20 cm⁻¹; vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

Peak-by-peak assignments, derived via comparison with the simulated D₁ ← D₀ spectrum of 8-iso-HC₉H₇N, are listed in Table 5 and indicated in Fig. 7b. Similar to the dispersed fluorescence spectrum, the fluorescence excitation spectrum of 8-iso-HC₉H₇N is also dominated by vibrational normal modes of a′ symmetry. The average absolute deviation in peak positions between experiment and simulation, 8 ± 6 cm⁻¹, is comparable to that derived for 4-iso-HC₉H₇N. We also compare this spectrum with the simulated D₁ ← D₀ absorption spectra of various iso-HC₉H₇N isomers in Fig. S8 (SI). The observed excitation spectrum (trace b) agrees with that predicted for 8-iso-HC₉H₇N, but not with other isomers of iso-HC₉H₇N.

Table 5 Assignments of peaks in the fluorescence excitation spectrum of 8-iso-HC₉H₇N isolated in solid para-H₂

para-H2	uB3PW91			Assignment	Sym.
LIF^a/cm^−1	FCHT^b/cm^−1	Rel. int.^c/%	Scaled^d/cm^−1
a Vibrational spacing relative to the origin band at 19 433 cm^−1 (514.6 nm). b Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. c Intensities relative to the most intense peak (0^00 band). Only vibrational modes contributing with a relative intensity ≥1.5% are listed. d Harmonic vibrational wavenumbers scaled by a factor 0.978. Numbers given in parentheses for combination and overtone bands are the sum of the fundamentals.
	0	100	0	0^00	a′
252	246	1.4	(245)	47^1048^10	a′
	591	2.0	587	29^10	a′
		1.9	(597)	450^2	a′
700	689	12.0	690	28^10	a′
	751	2.1	752	27^10	a′
992	986	9.8	986	24^10	a′
1055	1064	5.6	1065	22^10	a′
1148	1146	4.7	1143	21^10	a′
1169	1170	2.1	1169	20^10	a′
		7.9	1172	19^10	a′
1242	1232	4.0	1233	18^10	a′
1462	1463	6.2	1464	10^10	a′
1532	1537	9.6	1539	9^10	a′
1620	1604	19.7	1604	8^10	a′
	1674	0.7	(1676)	24^1028^10	a′

---

Unlike 4-iso-HC₉H₇N, which shows overtones involving up to 10 quanta of ν₄₈ (26 cm⁻¹, a′′) in Fig. 6a, no significant contribution from higher-order overtones or their combination bands is predicted for the D₁ ← D₀ absorption spectrum of 8-iso-HC₉H₇N. Differences between harmonic vibrational frequencies of a specific normal mode in different electronic states serve as indicators of changes in the curvature of the potential energy surface along this normal coordinate near the equilibrium geometry. In Fig. S9 (SI), quotients ν_n(D₀)/ν_n(D₁) for n = 1–48 are plotted for 4- and 8-iso-HC₉H₇N; ν_n(D₀) and ν_n(D₁) are vibrational wavenumbers of the nth vibrational mode in the D₀ and D₁ state, respectively. The significant reduction of ν₄₈ from 56 to 26 cm⁻¹ upon excitation from D₀ to D₁ of 4-iso-HC₉H₇N indicates a shallower potential energy surface for ν₄₈ in the D₁ state, explaining the appearance of ν₄₈ overtones and their combination bands in fluorescence excitation spectrum of 4-iso-HC₉H₇N.

3.5. Dispersed fluorescence and fluorescence excitation spectra of 5-iso-HC₉H₇N

Upon excitation at 495.5 nm (20 182 cm⁻¹) of a matrix of iso-C₉H₇N in solid para-H₂ deposited under electron bombardment, we observed additional emission features (system C) that could not be assigned to either 4- or 8-iso-HC₉H₇N, as illustrated in Fig. 2c and Fig. S5 (SI). Fluorescence emission probed in the range 580.9–587.6 nm (17 018–17 215 cm⁻¹) upon excitation at 18 688 cm⁻¹ decays to <10% within 50 ns, cf. Fig. S10 (SI), suggesting an emission lifetime between those of 4- and 8-iso-HC₉H₇N, and, hence, the presence of a third distinct species.

Varying the excitation wavelength in increments of 0.1 nm while probing emission in the range 580.9–587.6 nm (17 018–17 215 cm⁻¹), as shown in Fig. S7 (SI), we obtained the fluorescence excitation spectrum depicted in Fig. 8a. The most intense feature is located at 18 688 cm⁻¹ (535.1 nm), accompanied by sharp bands at 1502 (relative intensity 35%), 1161 (27%), 1063 (17%), 950 (17%), 847 (13%), 591 (20%), and 469 cm⁻¹ (17%) above the most intense band. We compare this spectrum with the experimental fluorescence excitation spectra of 4- and 8-iso-HC₉H₇N isolated in solid para-H₂ and the simulated D₁ ← D₀ absorption spectra of various iso-HC₉H₇N isomers in Fig. S8 (SI). The second most intense feature (relative intensity ∼35%) in the simulated D₁ ← D₀ spectrum of 5-iso-HC₉H₇N, located near 1530 cm⁻¹, is consistent with the single intense experimental band near 1500 cm⁻¹. In contrast, in the simulated absorption spectra of all other isomers, this band is either shifted further to higher energies or accompanied by additional bands of comparable intensity. Furthermore, the observed spectral pattern in the range 400–1000 cm⁻¹ is satisfactorily reproduced by simulation, with the exception of the lowest-energy feature at ∼335 cm⁻¹, which is predicted but only tentatively identified in the experimentally spectrum due to its small intensity.

Fig. 8 Comparison of observed and simulated excitation spectra of 5-iso-HC₉H₇N. (a) Fluorescence excitation spectrum of 5-iso-HC₉H₇N isolated in solid para-H₂ obtained on probing emission in region 580.9–587.6 nm (17 018–17 215 cm⁻¹). The bottom axis denotes the displacement from the origin band, observed at 18 688 cm⁻¹ (535.1 nm), whereas the top axis represents the wavelength of the observed spectrum. (b) The D₁ ← D₀ absorption spectrum simulated on convolution the simulated Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian function with FWHM = 15 cm⁻¹; harmonic vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

Vibrational assignments derived from a comparison to the simulated absorption spectrum of 5-iso-HC₉H₇N are indicated in Fig. 8b and listed in Table 6; the agreement between experiment and FCHT calculation is satisfactory, with an average absolute deviation of 13 ± 11 cm⁻¹. This deviation is slightly larger than those of 4- and 8-iso-HC₉H₇N, presumably because of the lower signal-to-noise ratio due to small fluorescence emission intensity and the likely interference due to emission from other isomers as indicated in Fig. S7 (SI).

Table 6 Assignments of peaks in the fluorescence excitation spectrum of 5-iso-HC₉H₇N isolated in solid para-H₂

para-H2	uB3PW91			Assignment	Sym.
LIF^a/cm^−1	FCHT^b/cm^−1	Rel. int.^c/%	Scaled^d/cm^−1
a Vibrational spacings relative to the origin band at 18 688 cm^−1 (535.1 nm). Tentatively assigned peaks are marked with?. b Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. c Intensities relative to the most intense peak (0^00 band). Only vibrational modes contributing with a relative intensity ≥1.5% are listed. d Harmonic vibrational wavenumbers scaled by a factor 0.978. Values given in parentheses for combination and overtone bands are the sum of the fundamentals.
	0	100.0	0	0^00	a′
341?	333	11.9	334	32^10	a′
469	462	8.1	462	31^10	a′
591	599	2.3	574	45^20	a′
		18.5	599	29^10	a′
711	712	4.9	711	28^10	a′
760	767	8.7	766	27^10	a′
847?	810	6.1	811	26^10	a′
950	927	2.2	923	25^10	a′
		1.9	933	29^1032^10	a′
	966	2.3	966	24^10	a′
1063	1048	3.0	1050	23^10	a′
1090?	1099	2.9	1098	22^10	a′
	1142	5.7	1142	21^10	a′
1161	1166	10.7	1167	20^10	a′
	1342	2.5	1340	15^10	a′
	1377	1.6	1365	27^1029^10	a′
		3.8	1378	13^10	a′
	1459	6.8	1458	11^10	a′
1502	1522	33.6	1522	9^10	a′
	1584	4.2	1585	8^10	a′

---

Following the above discussion, we assign the fluorescence excitation spectrum depicted in Fig. 8a to the D₁ ← D₀ absorption of 5-iso-HC₉H₇N. Similar to the fluorescence excitation spectrum of 8-iso-HC₉H₇N, the fluorescence excitation spectrum of 5-iso-HC₉H₇N is composed nearly exclusively of fundamental vibrational modes of a′ symmetry.

As demonstrated in Fig. S11 (SI), except for the origin band, the fluorescence excitation spectrum of 5-iso-HC₉H₇N overlaps significantly with those of 4- and 8-iso-HC₉H₇N. Due to these similarities, identifying an excitation band other than the origin band that yields an interference-free dispersed fluorescence spectrum of 5-iso-HC₉H₇N is particularly challenging. The pure dispersed fluorescence spectrum of 5-iso-HC₉H₇N isolated in solid para-H₂, depicted in Fig. 9a, was acquired upon excitation at the 0⁰₀ band (18 688 cm⁻¹). Consequently, the spectrum close to the 0⁰₀ band is strongly interfered by the excitation laser, and the experimental spectrum depicted in Fig. 9, therefore, only covers the energy region well below the excitation energy. Nevertheless, a comparison of the spectral structure above 1000 cm⁻¹ with the simulated D₁ → D₀ emission spectra, Fig. 2 and 9b strongly supports our assignment of the recorded spectrum to the 5-iso-HC₉H₇N radical.

Fig. 9 Comparison of observed and simulated dispersed fluorescence spectra of 5-iso-HC₉H₇N. (a) Partial dispersed fluorescence spectrum of 5-iso-HC₉H₇N isolated in solid para-H₂ recorded upon excitation at 18 688 cm⁻¹ (535.1 nm). The bottom axis denotes the displacement from the origin band, observed at 18 688 cm⁻¹, whereas the top axis represents the wavelength of the observed spectrum. And (b) The D₁ → D₀ emission spectrum simulated by convolution the Franck–Condon Herzberg–Teller stick spectrum (red) with a Gaussian line shape of FWHM 30 cm⁻¹; vibrational wavenumbers were scaled by 0.978. Vibrational mode assignments are indicated.

Peak positions inferred from the partial dispersed fluorescence spectrum in Fig. 9a and their assignments derived by comparison to the simulated D₁ → D₀ emission spectrum of 5-iso-HC₉H₇N are presented in Table 7, together with the IR absorption lines reported by Joshi et al.³⁰ The most intense feature at 1571 cm⁻¹ is assigned to the 9⁰₁ transition, with partial overlap from weaker bands of 10⁰₁ and 8⁰₁. Five additional features below 1500 cm⁻¹ were identified; these show satisfactory agreement with FCHT predictions. Three weak features above 1900 cm⁻¹ are tentatively assigned to combination bands involving ν₉ and ν₂₇, ν₂₉, or ν₃₂ because of their small intensities. Within the spectral range covered, simulated and experimental peak positions for fundamental vibrational modes exhibit an average absolute deviation of 6 ± 4 cm⁻¹. When the weak combination bands are included, the average absolute deviation increases to 14 ± 15 cm⁻¹.

Table 7 Assignments of peaks in the dispersed fluorescence spectrum of 5-iso-HC₉H₇N isolated in solid para-H₂

para-H2		uB3PW91			Assignment	Sym.
IR^a/cm^−1	LIF^b/cm^−1	FCHT^c/cm^−1	Rel. int.^d/%	Scaled^e/cm^−1
a Peak positions in the IR absorption spectrum of 5-iso-HC9H7N isolated in solid para-H2 reported by Joshi et al.^30 Lines in parentheses represent modes predicted by FCHT simulations with a relative intensity of <1.5%. b Vibrational spacing relative to the origin band at 18 668 cm^−1 (535.1 nm) in the fluorescence excitation spectrum of 5-iso-HC9H7N. Tentatively assigned peaks are marked with ?. c Vibrational spacings from the convoluted simulated stick spectrum calculated with the FCHT method. d Intensities relative to the most intense peak (0^00 band). Only vibrational modes with relative intensities ≥1.5% are listed. The weak features corresponding to IR absorption bands reported by Joshi et al.^30 are listed in parentheses. e Harmonic vibrational wavenumbers scaled by a factor 0.978. Values given in parentheses for combination and overtone bands are the sum of the fundamentals.
		0	100.0	0	0^00	a′
		344	11.3	345	32^01	a′
		493	3.4	471	31^01	a′
			4.2	499	30^01	a′
		618	15.1	617	29^01	a′
(652.3)			(<0.1)	645	42^01	a′′
(708.6)			(<0.1)	712	41^01	a′′
		786	8.4	787	27^01	a′
(821.9)			(<0.1)	821	39^01	a′′
		943	3.0	906	26^01	a′
			2.4	938	25^01	a′
			2.3	(962)	29^0132^01	a′
	1054	1060	2.5	1052	24^01	a′
	1158	1158	1.6	(1132)	27^0132^01	a′
			4.7	1148	22^01	a′
			5.5	1167	21^01	a′
1201.3	1211	1205	2.6	1198	20^01	a′
			1.6	1213	19^01	a′
	1314	1314	8.2	1314	16^01	a′
	1415	1404	2.8	1404	14^01	a′
			1.8	(1404)	27^0129^01	a′
(1402.0)			(1.3)	1406	13^01	a′
	1571	1561	1.9	1522	10^01	a′
			42.3	1561	9^01	a′
			4.1	1587	8^01	a′
	1934?	1905	4.2	(1906)	9^0132^01	a′
	2210?	2179	5.7	(2177)	9^0129^01	a′
	2403?	2347	3.6	(2348)	9^0127^01	a′

---

Joshi et al.³⁰ reported five IR absorption lines below 2500 cm⁻¹ for 5-iso-HC₉H₇N isolated in solid para-H₂: three a′′ symmetric vibrations ν₄₂, ν₄₁, and ν₃₉, which are not expected to contribute to the dispersed fluorescence spectrum, and the two a′ symmetric vibrations ν₂₀ and ν₁₃, which are predicted to appear in the emission spectrum with relative intensities of 2.6 and 1.3%, respectively. The ν₂₀ band, which overlaps with ν₁₉, was observed in the dispersed fluorescence spectrum, whereas the ν₁₃ band was too weak to be positively identified.

For the three C–N and C–C stretching modes ν₁₀, ν₉, and ν₈, as previously discussed for 4- and 8-iso-HC₉H₇N, calculations predict relative intensities of 1.9, 42.3, and 4.1%, respectively. Atomic displacement vectors associated with these vibrations and the geometry changes induced by the electronic transition between D₁ and D₀ are illustrated in Fig. S12 (SI). The electronic transition strongly affects the bond lengths of the C–N and the central C(4a)–C(8a) bond, consistent with the large Franck–Condon factor calculated for mode ν₉. In contrast, modes ν₈ and ν₁₀ exert only minor impact on these two bonds.

Of the eight iso-HC₉H₇N isomers previously identified in solid para-H₂ by IR spectroscopy, only the dispersed fluorescence and fluorescence excitation spectra of the three low-energy isomers 4-, 5-, and 8-iso-HC₉H₇N were observed. The lowest-energy isomer, iso-C₉H₇NH, although reported as the most abundant in the study by Joshi et al.,³⁰ was not observed. TD-DFT calculations predict the lowest-energy electronic transition of iso-C₉H₇NH at 839.58 nm (11 911 cm⁻¹, vertical excitation, Table 1), which is beyond the detection range of our iCCD. Moreover, with a predicted oscillator strength of 0.001 and a vertical emission wavelength of 2268 nm, detecting the weak fluorescence emission of iso-C₉H₇NH is expected to be challenging. Another low-lying isomer, 1-iso-HC₉H₇N with relative energy +6 kJ mol⁻¹, was also not observed, likely because the transition is predicted to have a low fluorescence yield (f = 0.001). Furthermore, the D₁–D₀ electronic transition induces a significant out-of-plane distortion of the molecule (Fig. S1, SI), resulting in a small Franck–Condon overlap between these two states and a vertical emission wavelength of 1180 nm, which is beyond our detection limit. Of the three high-energy isomers, 3-, 6-, and 7-iso-HC₉H₇N, Joshi et al.³⁰ only observed 6-iso-HC₉H₇N in experiments on iso-C₉H₇N/para-H₂ matrices deposited under electron bombardment. For this isomer, our TD-DFT calculations predict a vertical emission energy of 592.8 nm and oscillator strength of 0.005. The small oscillator strength and expected low concentration of this high energy isomer make the detection unlikely.

The positions of the vibrationless transitions inferred from the fluorescence spectra of the three experimentally observed iso-HC₉H₇N isomers differ substantially from the predicted vertical excitation energies listed in Table 1. For hybrid functionals such as B3LYP and B3PW91, TD-DFT benchmark studies typically indicate mean absolute deviations of 0.3–0.5 eV in excitation energies, either between TD-DFT and experimental data or between TD-DFT and higher-level ab initio calculations.^37,38

These benchmark studies, however, primarily concern only singlet and triplet transitions of closed-shell molecules and rarely address doublet–doublet transitions of open-shell species. As doublet states are often subject to spin-contamination and exhibit significant multi-reference character, which are in most cases not captured adequately by TD-DFT, uncertainties in excitation energies for doublet–doublet transitions are expected to be larger. The mean absolute deviation of 0.46 eV between computed vertical excitation energies and experimentally determined 0⁰₀ band positions for the three iso-HC₉H₇N isomers isolated in solid para-H₂ therefore appears reasonable. According to our experimental results, the D₁–D₀ energy gap increases in the order 4-iso-HC₉H₇N < 5-iso-HC₉H₇N < 8-iso-HC₉H₇N. TD-DFT calculations predict the highest D₁–D₀ excitation energy for 8-iso-HC₉H₇N, in agreement with our experimental findings, but an inverted order for 4- and 5-iso-HC₉H₇N with a minor difference of only ∼1 nm from calculations.

3.6 Hydrogenated iso-quinolines as potential DIB carriers

The most prominent features in the fluorescence excitation spectra of 4-, 5-, and 8-iso-HC₉H₇N isolated in solid para-H₂ are located at 540.8, 535.1, and 514.7 nm, respectively; other features are comparatively weak. Within this wavelength range, Hobbs et al.³⁹ reported a DIB density of ∼17 DIB per 20 nm from observations of HD 20 4827. Unfortunately, electronic absorption spectra of hydrogenated PAH and PANH in the gas phase have rarely been reported, and matrix shifts for electronic transitions of these open-shell species isolated in solid para-H₂ are not fully understood. For 1-HC₁₀H₈, a red-shift of 86 cm⁻¹ was observed for the D₁–D₀ transition located at 18 881 cm⁻¹ in solid para-H₂,²⁴ aligning with the average matrix shift of 70 ± 28 cm⁻¹ derived from available data for PAH containing up to 42 carbon atoms and their protonated and hydrogenated derivative isolated in solid para-H₂.²⁵ Adopting this average matrix shift, we estimate positions of the D₁–D₀ 0⁰₀ bands of gaseous 4-, 5-, and 8-iso-HC₉H₇N to be 539.5–537.9 nm, 533.9–532.4 nm, and 513.6–512.1 nm, respectively. A detailed comparison of the experimental fluorescence excitation spectra with the overall DIB spectrum reported by Jennsikens and Désert⁴⁰ is shown in Fig. S13 (SI).

In their DIB catalogue based on observations of six stars characterised by varying degrees of reddening, Jenniskens and Désert⁴⁰ list four “certain” DIB in the range 510–545 nm: at 510.970, 536.214, 536.360, and 540.456 nm. The first three lie clearly outside the range of the transitions observed in this work; a matrix red shift of ≥140 cm⁻¹ would be required to map our experimental data onto these line positions, substantially larger than the average shift typically observed for PAH in solid para-H₂. The final DIB at 540.456 nm (18 503.89 cm⁻¹) is located ∼10 cm⁻¹ higher in energy than the observed D₁–D₀ transition of 4-iso-HC₉H₇N in solid para-H₂ at 18 492 cm⁻¹. However, after applying the average matrix shift of 70 ± 28 cm⁻¹ to our experimental transition wavelength, the corrected gas-phase transition for 4-iso-HC₉H₇N becomes ∼60 cm⁻¹ higher in energy than the DIB band, rendering a contribution of 4-iso-HC₉H₇N to the DIB spectrum unlikely. This is consistent with astrochemical models predicting that small PAH (containing less than 40 C atoms) are not stable under the conditions of the ISM. Pursuing electronic absorption spectra of larger PAH/PANH is hence desirable.

4. Conclusions

This study reports the previously unobserved dispersed fluorescence and fluorescence excitation spectra of three hydrogenated iso-quinoline isomers: 4-, 5-, and 8-iso-HC₉H₇N. Upon deposition of a mixture of iso-C₉H₇N and para-H₂ under electron bombardment and subsequent irradiation at wavelengths <540 nm, we observed three distinct fluorescence emission progressions originating at 18 492, 18 688, and 19 430 cm⁻¹, with characteristic emission lifetimes of ∼6, <50, and ∼105 ns, respectively. These transitions were assigned to the D₁ → D₀ transitions of 4-, 5-, and 8-iso-HC₉H₇N, based on comparison of the recorded spectra with spectra simulated by a Franck–Condon Herzberg–Teller approach. The experimental excitation spectra of these three isomers also agree satisfactorily with the corresponding simulated absorption spectra. Dispersed fluorescence and fluorescence excitation spectra associated with the D₁–D₀ transition of all three isomers are dominated by vibrational normal modes of a′ symmetry, with the 0⁰₀ band as the most intense feature. Vibrational modes inferred from the dispersed fluorescence spectra are distinctively different from those observed in the IR absorption spectra previously reported by Joshi et al.³⁰ Differences in relative intensities and the contribution of overtone and combination bands are discussed in relation to changes in geometries and in the vibrational potential energy surfaces between the D₀ and D₁ states. Based on the observed spectral positions and expected matrix shifts, a contribution of electronic bands of 4-, 5- and 8-iso-HC₉H₇N to the DIB appears unlikely.

Conflicts of interest

There are no conflicts to declare.

Data availability

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information: variation of vibronic pattern as function of the position of H-addition. Tables: Scaled harmonic vibrational wavenumbers of the D₀ and D₁ state of the eight most stable isomers of iso-HC₉H₇N. Figures: Optimized geometry of the D₁ state of 1-iso-HC₉H₇N. Simulated D₁ → D₀ emission spectrum of 1-iso-HC₉H₇N. Characteristic displacements induced by modes ν₂₉, ν₃₀, ν₃₁, and ν₃₂. Geometry changes induced by the D₁–D₀ transition for n-iso-HC₉H₇N (n = 3–8). Partial dispersed fluorescence spectrum upon excitation at 497.0, 475.0, and 495.5 nm. Temporal behaviour of fluorescence emission of 4-, 8-, and 5-iso-HC₉H₇N. Relative changes of bond length and angles upon electronic transition between D₀ and D₁ for 4- and 8-iso-HC₉H₇N. Normalised dispersed fluorescence and fluorescence excitation spectra of 4-, 5-, and 8-iso-HC₉H₇N isolated in solid para-H₂. Quotients of vibrational wavenumbers of the D₀ and D₁ state for 4- and 8-iso-HC₉H₇N. Comparison of observed fluorescence excitation spectra with simulated D₁ ← D₀ spectra of various iso-HC₉H₇N isomers. Changes in optimized geometries of 5-iso-HC₉H₇N upon transition between D₀ and D₁ and atom displacements induced by modes ν₈, ν₉, and ν₁₀. Comparison of the fluorescence excitation spectra of 4-, 5-, and 8-iso-HC₉H₇N to the DIB spectrum. See DOI: https://doi.org/10.1039/d5cp03455h.

Acknowledgements

This work was supported by National Science and Technology Council, Taiwan (grants NSTC113-2639-M-A49-002-ASP), Center for Emergent Functional Matter Science of National Yang Ming Chiao Tung University from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan, and the EU's Horizon Europe program MSCA (ActBio, No. 101107213) and COST Action (CA21101 “COSY”). The National Center for High-Performance Computing (NCHC) of National Applied Research Laboratories (NARLabs) in Taiwan and the computer facilities of the Scuola Normale Superiore, Italy, provided computer time. We thank Prof. Julien Bloino for fruitful discussions and access to the Gaussian development version.

References

1. M. L. Heger, Lick Obs. Bull., 1922, 10, 141–145.
2. R. Vogrinčič, J. Kos, T. Zwitter, G. Traven, K. L. Beeson, K. Čotar, U. Munari, S. Buder, S. L. Martell and G. F. Lewis, et al. , Mon. Not. R. Astron. Soc., 2023, 521, 3727–3748.
3. H. Fan, L. M. Hobbs, J. A. Dahlstrom, D. E. Welty, D. G. York, B. Rachford, T. P. Snow, P. Sonnentrucker, N. Baskes and G. Zhao, Astrophys. J., 2019, 878, 151.
4. P. Sonnentrucker, B. York, L. M. Hobbs, D. E. Welty, S. D. Friedman, J. Dahlstrom, T. P. Snow and D. G. York, Astrophys. J., Suppl. Ser., 2018, 237, 40.
5. E. K. Campbell, M. Holz, D. Gerlich and J. P. Maier, Nature, 2015, 523, 322–323.
6. E. K. Campbell, M. Holz, J. P. Maier, D. Gerlich, G. A. H. Walker and D. Bohlender, Astrophys. J., 2016, 822, 17.
7. N. L. J. Cox, EAS Publ. Ser., 2011, 46, 349–354.
8. M. K. Crawford, A. G. G. M. Tielens and L. J. Allamandola, Astrophys. J., 1985, 293, L45.
9. J. L. Puget, A. Leger and F. Boulanger, Astron. Astrophys., 1985, 142, L19–L22.
10. L. J. Allamandola, A. G. G. M. Tielens and J. R. Barker, Astrophys. J., 1985, 290, L25–L28.
11. A. G. G. M. Tielens, Annu. Rev. Astron. Astrophys., 2008, 46, 289–337.
12. C. M. Canelo, A. C. S. Friaça, D. A. Sales, M. G. Pastoriza and D. Ruschel-Dutra, Mon. Not. R. Astron. Soc., 2018, 475, 3746–3763.
13. D. M. Hudgins, J. C. W. Bauschlicher and L. J. Allamandola, Astrophys. J., 2005, 632, 316–332.
14. A. Ricca, C. Boersma and E. Peeters, Astrophys. J., 2021, 923, 202.
15. P. G. Stoks and A. W. Schwartz, Geochim. Cosmochim. Acta, 1982, 46, 309–315.
16. P. G. Stoks and A. W. Schwartz, Geochim. Cosmochim. Acta, 1981, 45, 563–569.
17. M. E. Jacox, J. Mol. Struct., 1987, 157, 43–59.
18. C. Crépin and A. Tramer, Chem. Phys., 2001, 272, 227–241.
19. M. Bahou, P. Das, Y.-F. Lee, Y.-J. Wu and Y.-P. Lee, Phys. Chem. Chem. Phys., 2014, 16, 2200–2210.
20. M. Bahou, C.-W. Huang, Y.-L. Huang, J. Glatthaar and Y.-P. Lee, J. Chin. Chem. Soc., 2010, 57, 771–782.
21. M. Tsuge, C.-Y. Tseng and Y.-P. Lee, Phys. Chem. Chem. Phys., 2018, 20, 5344–5358.
22. I. Weber and Y.-P. Lee, J. Phys. Chem. A, 2024, 128, 4984–4991.
23. I. Weber, M. Tsuge, P. Sundararajan, M. Baba, H. Sakurai and Y.-P. Lee, J. Phys. Chem. A, 2022, 126, 5283–5293.
24. I. Weber, C.-W. Wang, S.-C. Huang, C.-Y. Zhu and Y.-P. Lee, J. Phys. Chem. A, 2022, 126, 8423–8433.
25. I. Weber, P. R. Joshi, D. T. Anderson and Y.-P. Lee, J. Phys. Chem. Lett., 2024, 15, 11361–11373.
26. E. Plazas, M. C. Avila M, D. R. Muñoz and L. E. Cuca S, Pharmacol. Res., 2022, 177, 106126.
27. S. P. Verevkin, S. P. Safronov, A. A. Samarov and S. V. Vostrikov, Appl. Sci., 2021, 11, 11758.
28. S. P. Safronov, S. V. Vostrikov, A. A. Samarov and S. P. Verevkin, Fuel, 2022, 317, 123501.
29. H. Harb, S. N. Elliott, L. Ward, I. T. Foster, S. J. Klippenstein, L. A. Curtiss and R. S. Assary, Digital Discovery, 2023, 2, 1813–1830.
30. P. R. Joshi, M. Tsuge, C.-Y. Tseng and Y.-P. Lee, Phys. Chem. Chem. Phys., 2023, 25, 11934–11950.
31. S.-C. Huang and Y.-P. Lee, Chem. Phys. Lett., 2016, 665, 53–58.
32. M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, G. A. Petersson and H. Nakatsuji, et al., Gaussian 16 Rev. B.01, Gaussian Inc., Wallingford CT, 2016.
33. A. Baiardi, J. Bloino and V. Barone, J. Chem. Theory Comput., 2013, 9, 4097–4115.
34. J. Bloino, A. Baiardi and M. Biczysko, Int. J. Quantum Chem., 2016, 116, 1543–1574.
35. Q. Yang, M. Fusè, J. Bloino and V. Barone, Spectrochim. Acta, Part A, 2021, 254, 119631.
36. J. C. Dobrowolski and P. F. J. Lipiński, RSC Adv., 2016, 6, 23900–23904.
37. I. Knysh, F. Lipparini, A. Blondel, I. Duchemin, X. Blase, P.-F. Loos and D. Jacquemin, J. Chem. Theory Comput., 2024, 20, 8152–8174.
38. J. Liang, X. Feng, D. Hait and M. Head-Gordon, J. Chem. Theory Comput., 2022, 18, 3460–3473.
39. L. M. Hobbs, D. G. York, T. P. Snow, T. Oka, J. A. Thorburn, M. Bishof, S. D. Friedman, B. J. McCall, B. Rachford and P. Sonnentrucker, et al. , Astrophys. J., 2008, 680, 1256–1270.
40. P. Jenniskens and F.-X. Désert, Astron. Astrophys., Suppl. Ser., 1994, 106, 39–78.

---