Pyridine vs N-Hydrogenated Pyridine Moieties: Theoretical Study of Stability and Spectroscopy of Nitrogen-Contained Heterocyclic Aromatic Compounds and Graphene Nanoflakes.

Nitrogen is one of the most common heteroatom appearing in heterocyclic aromatic compounds (HACs) as well as the frequently applied dopant in graphene nanoflakes/nanoribbons. The pyridine moiety is an intuitive and stable common feature of these compounds; but interestingly, using density functional theory calculations, we found that the N-hydrogenated pyridine moiety could be even more stable in large HACs and in N-doped graphene nanoflakes considering their formation reaction energies. The hydrogenation reaction of the pyridine moiety was calculated to be exothermic for models of four and more fused aromatic rings with specific substitutional positions of nitrogen. This theoretical investigation provides energetic and spectroscopic hints to the existence of the N-hydrogenated pyridine moiety under proper conditions.

Introduction

Heterocyclic aromatic compounds (HACs) have been well known to chemists for 100 years and used as laboratory reagents, industrial materials, biological metabolites, environmental pollutants, etc. in our daily life. Nitrogen is one of the major heteroatoms that substitute aromatic carbon in these compounds. Starting from the simplest six-membered ring aromatic hydrocarbon and stepping toward fused-ring polycyclic aromatic hydrocarbons (PAHs), a single nitrogen substitution results in pyridine (C5H5N) from benzene, quinoline, and isoquinoline (C9H7N) from naphthalene, acridine, and benzoquinoline (C13H9N) from anthracene, and so on. In these quasi-one-dimensional (quasi-1D) HACs, one nitrogen atom replaces one C–H unit, forming the pyridine-type moiety where one electron in the perpendicular 2p orbital of nitrogen is incorporated into the π-conjugated electronic system, and one set of lone-pair electrons fills the in-plane sp2 nonbonding orbital outside the aromatic structure. Such an arrangement satisfies Hückel’s 4n + 2 rule of π-electrons and thus retains the aromaticity.[1,2]

What interests us is the role of the lone pair, which seems outsider of the system. It can play as a Lewis base and form a coordinate covalent bond to a Lewis acid, e.g., proton. This N-protonated pyridine moiety is also well known; the N–H+ σ bond is basically irrelevant to aromaticity. But what will happen, or could it happen, if nitrogen bonds with “hydrogen atom” instead of proton? Supposing it takes place, not only the N–H bond is formed but also one extra electron is injected into the compound. The extra electron is supposed to reside on the π* antibonding orbital, yielding a total number of 4n + 3 electrons, deteriorating the aromaticity, raising the total energy, and making the system unstable.

We temporarily switch to HAC anions with 4n + 3 electrons, which meet a similar one-extra-electron situation (but without the extra hydrogen). Early experiments showed that pyridine anion (C5H5N–) is unstable, readily getting dimerized or oxidized after formation.[3] Quinoline anion (C9H7N–) has a longer lifetime in the sub-ms range,[4] and, interestingly, quinoline itself has a small but noticeable electron affinity.[5] From the theoretical point of view, it is an intuitive scenario that an extra electron residing in the π* orbital, turning the original lowest unoccupied molecular orbital (LUMO) to singly occupied molecular orbital (SOMO), will decrease the stability of the HAC based on the simple Hückel’s theory. However, this scenario may be altered by several factors. First, elongating or enlarging the HAC with more fused rings will lower the original LUMO energy level. Second, counting in electron correlation (i.e., beyond Hückel’s theory) will further lower the SOMO energy. Moreover, in the case of N-hydrogenated HAC, the formation of N–H bond would also contribute to the overall stability of the system.

Whereas a pyridine anion or an N-hydrogenated pyridine requires harsh generation conditions like electron beam irradiation or strong reducing reagent,[3,6,7] a stable N-hydrogenated pyridinic HAC could be possible considering the above factors. Although this species has not been reported experimentally, it has been proposed in theoretical studies of X-ray photoelectron spectroscopy (XPS)[8] and electrochemical oxygen reduction reactions.[9−11] We shall further explore these N-hydrogenated pyridinic HACs by using density functional theory (DFT) calculations in this work. Calculations on the formation reaction energies and stability as well as analyses on frontier MOs and excitation configurations of small nitrogen-contained quasi-1D HACs and graphene nanoflakes (nanoribbon fragments) are given in the following sections.

Computational Methods

A “bottom-up” approach of increasing building blocks (aromatic rings) to construct model clusters was adopted. Quasi-1D PAH[n,1] model molecules (following Nakano’s notation)[12,13] were first constructed with n fused six-membered rings along the zigzag direction and one carbon atom on the edge position was replaced by nitrogen to generate N-hydrogenated pyridinic HACs. The normal pyridine-type HACs were also obtained by removing the hydrogen attached on nitrogen. Whereas the N-hydrogenated HACs were neutral radicals with doublet ground states, the quasi-1D pyridine-type ones were neutral closed-shell singlets. All these model compounds were geometrically optimized by DFT calculations, and their excitation configurations and UV–vis absorption spectra were simulated by time-dependent (TD) DFT. The calculations adopted the B3LYP functional and the 6-31G(d) basis set, which have been shown to be adequate in previous studies.[14,15] The ground-state vibrational frequencies were scaled by a factor of 0.960,[16] and peaks in vibrational spectra were broadened with a half-width at half-maximum (HWHM) of 5 cm–1. The UV–vis absorption spectrum of each model counted 50 excited states, and peaks were broadened with HWHM of 0.025 eV (200 cm–1). It was noticed that for open-shell/doublet systems, the spin contamination occurred in TD-DFT calculations,[17,18] and we have set threshold criteria to rule out unphysical excited states by considering the change of square of total spin, Δ⟨Ŝ2⟩, and the participation of SOMO upon excitation.[15] Calculations using other functionals including CAM-B3LYP, M06HF, and ωB97X-D were also carried out to verify the consistency of the DFT results.

The N-doped graphene nanoflake models were constructed accordingly with [n,m] fused-ring arrays where m denotes the number of rows along the armchair direction. It has been found that nitrogen substitution tends to occur on edge positions rather than inside the ribbon/flake according to calculated total energies.[15] The N-hydrogenated graphene nanoflakes were also neutral radicals with doublet ground states. On the other hand, most of the pyridine-type models, unlike the quasi-1D HAC cases, possessed open-shell singlet (oss) characteristics, where frontier MOs distribute on both zigzag edges.[12,19−21] Sample geometric structures of these HACs are illustrated in Figure , with the numbering of possible substitutional positions noted. We have applied the HuLiS calculator[22,23] for simple Hückel MO analysis, and the Gaussian 09 package[24] for all detailed calculations.

Figure 1

Sample models of nitrogen-contained HACs and numbering of substitutional positions. (a) HAC[5,1]NH@3a, one quasi-1D HAC of five fused rings with N-hydrogenated pyridine moiety at position 3a. (b) HAC[5,5]N@13b, one single-nitrogen-doped graphene nanoflake of 5 × 5 array with pure pyridine moiety at position 13b.

Results and Discussion

Formation Energy and Stability

Considering the single-nitrogen substitution reaction that converts PAH to HAC,where eq refers to the formation of simple pyridine moiety and eq refers to the formation of N-hydrogenated pyridine (which has the N–H bond) moiety from their parent PAHs. The energy of substitutional formation reaction can be calculated aswhere Es on the rhs of eq 2 denote the total energies of the corresponding species; subscripts HAC-N: and HAC-NH refer to the HACs with simple pyridine and N-hydrogenated pyridine moieties, respectively. Accordingly, the hydrogenation reaction that converts simple pyridine moiety to N-hydrogenated pyridine can be written asand the hydrogenation energy isThe formation reaction energies depending on the sizes of HACs and substitutional positions are plotted in Figure a, and the numerical data are given in Table S1 in the Supporting Information (SI). The size is represented by the array [n,1], meaning a model with n fused rings in one stripe, and the numbering of the substitutional position has been denoted in Figure a. It is clearly seen that the formation energies of the pyridine moiety are almost independent of either the model size or substitutional position; the values are distributed in a small range from 0.45 to 0.55 eV calculated according to eq . The formation energies of N-hydrogenated pyridine moieties, on the other hand, show a strong dependence on both size and position. For HACs of the same size, nitrogen substitution on the zigzag edge (position 1a, 2a, etc.) is always more stable than that on the short edge (position 1b), and the middle position of the zigzag edge gives the most stable form (e.g., HAC[5,1]NH@3a). As the fused-ring stripe elongates, the formation energy of N-hydrogenated pyridine moiety gets lower, even lower than the simple pyridine moiety.

Figure 2

(a) Formation reaction energies of simple pyridine (N:) and N-hydrogenated pyridine (NH) moieties and (b) hydrogenation reaction energies of pyridine moieties of quasi-1D HACs. The sizes of model species are indicated by HAC[n,1]N, and the substitutional positions of nitrogen (defined in Figure ) are noted along the horizontal axis.

The hydrogenation energies of pyridine moieties are plotted in Figure b, and the trend is nearly identical to the formation energies of N-hydrogenated pyridine moieties because the formation energies of simple pyridine moieties are quite uniform. The Gibbs free energies of the hydrogenation reactions have also been calculated at the standard conditions. They were found to be uniformly slightly higher than Eh by ∼0.25 eV, with identical trend of size- and position-dependence (see Figure S1 in the SI for comparison). It implies that when one obtains α-naphthoquinoline (C17H11N, in our notation HAC[4,1]N@2a), there may be the coexisting C17H11NH; when one synthesizes HAC[7,1]N compounds, there would be preference for C29H17NH rather than C29H17N.

The investigation then extended to 2D single-nitrogen-doped graphene nanoflakes with [n,m] arrays of fused rings as indicated in Figure b. The formation energies and hydrogenation energies of HAC[7,1]N to HAC[7,7]N models are demonstrated in Figure a. Different from quasi-1D HACs, the ground states of the multiple-row models with simple pyridine moiety were calculated as open-shell singlets. Despite the difference in spin multiplicity, the formation energies of these pyridine models were still rather uniform in the range from 0.4 to 0.6 eV. The N-hydrogenated moieties, on the other hand, showed that substitutions on zigzag edges (positions 21a, 31a, 41a) were generally more stable than those on armchair edges (13b, 14a). And the most stable species presented independence of the number of armchair rows, e.g., all HAC[7,m]NH@41a models had similar formation energies. The hydrogenation energies shown in Figure b followed the same trend.

Figure 3

(a) Formation energies of simple pyridine (N:) and N-hydrogenated pyridine (NH) moieties and (b) hydrogenation energies of pyridine moieties of single-nitrogen-doped graphene nanoflakes. The sizes of model species are indicated by HAC[n,m]N and the substitutional positions of nitrogen (defined in Figure ) are noted along the horizontal axis.

The following question arises: If the N-hydrogenated species could be energetically more stable than the simple pyridine moieties, why are they not observed or reported in experiments? One may first doubt the reliability of DFT calculation results. A possible concern relates to the zigzag edges of graphene nanoflakes, where the open-shell character may cause difficulty and artifact in frontier MOs by DFT calculations. This issue has in fact been demonstrated and interpreted in literatures that spin-unrestricted DFT is adequate for such systems.[20,21,25] Moreover, the extraordinary stability appears in not only graphene nanoflakes but also quasi-1D HACs with as small as 4–5 fused aromatic rings, where the MOs are quite definite (see following sections) and have nothing to do with the artifact. Calculations applying very different density functionals (Table S1 in the SI) also give consistent trends in the formation reaction energies, and hence the DFT results should be reliable.

We turn back to seek the experimental explanation of the nonexistence of N-hydrogenated pyridine species. A probable reason is the radical character of the N-hydrogenated moiety, which means a high reactivity in the synthesis process especially in a “crowded” environment such as solvent or high-pressure gases.[26−29] A second possibility is attributed to the hydrogenation mechanism, where a thermally stable species does not guarantee that it is kinetically preferable. The concept of “chemical hardness” may give some clues to this issue. The chemical hardness quantitatively describes the resistance of a compound to deformation or change, and to the zeroth-order approximation it is equal to the HOMO–LUMO energy gap.[30,31] Referring to Figure , the gap values of pyridine moieties are always larger than those of the corresponding N-hydrogenated species, probably implying the kinetic instability of the latter.

Figure 4

(a) Hückel and (b) DFT/B3LYP frontier MO energy levels of quasi-1D HACs.

Furthermore, the common knowledge about the aromaticity of pyridine may prevent people from thinking of its hydrogenated (not protonated) species. We suppose that under a proper condition like highly reductive gases, long-stripe HACs with N-hydrogenated pyridine moieties have a chance to be formed and detected.

Frontier MO Energy Levels

A comparison of Hückel MO energies of quasi-1D HACs with simple pyridine (N:) and N-hydrogenated pyridine (NH) moieties is depicted in Figure a, where the energy scale is represented by the Coulomb integral (a) and the bond integral (b). For pyridine moieties, HOMO and LUMO energies are close to their pristine PAH analogues, and the HOMO–LUMO gaps decrease smoothly when the numbers of fused rings increase just as expected. For N-hydrogenated moieties, the one extra electron could be regarded to be injected into the original LUMO of simple pyridine moiety (to form a new SOMO), and dragging this energy level downward to the original HOMO. This SOMO level gets lower when the HAC elongates, representing an overall trend toward stability.

This scenario is more clear and precise in the DFT-calculated MOs shown in Figure b. The extra electron of N-hydrogenated moiety creates a dopant-like SOMO level (here given the α-spin) in the original HOMO–LUMO gap of the pyridine moiety, and this level energy goes lower when the HAC elongates. In fact, this SOMO level energy is always negative even in the smallest hydrogenated pyridine (C5H5NH) model, implying a bond state and the stability of its existence. As the overall size-dependent trend of gap decreases and SOMO stabilization can be clearly seen in the figure, the position-dependent stability is also indicated. In models of the same size, nitrogen substitution near the center of zigzag edge gives the lowest HOMO energy for both pyridine and N-hydrogenated moieties.

Frontier MO Distributions, Excitation Configurations, and Spectroscopy

The DFT-calculated distributions of frontier MOs are plotted in Figure taking HAC[5,1]N@3a as an example (more detailed maps are given in Figure S2 in the SI). For the pyridine moiety, all frontier MOs have the π character with one nodal plane coinciding with the molecular plane, whereas the nonbonding orbital (lone pair) takes place as HOMO – 3. For the N-hydrogenated pyridine moiety, the distributions are almost identical to the pyridine moiety, and the extra electron (with the α-spin) resides in the π* SOMO, which is just like the original LUMO. The pyridinic nonbonding orbital now forms N–H σ bond and associates with other C–C σ bonds, and its energy goes down to SOMO – 8.

Figure 5

Electron density distributions of frontier MOs of pyridine-type HAC[5,1]N@3a (C21H13N, left) and its N-hydrogenated moiety (C21H13NH, right). The correspondence of most MOs between two models species is obvious, whereas the major difference occurs for nonbonding of pyridine moiety vs σ-bonding of N-hydrogenated moiety.

These MO distributions are closely related to their electronic excitation configurations and UV–vis absorption spectra, which are simulated in Figure . For pyridine moieties, it is clearly seen that the major peak positions and intensities (assigned as peak A) are independent of the substitutional position for same-sized HACs and are almost identical to their PAH analogues, which is attributed to the isoelectronic character of these aromatic compounds.[14,15] And these peaks red-shift as the model size increases, e.g., 7.56 eV for pyridine (HAC[1,1]N), 5.37 eV for acridine (HAC[3,1]N@2a), 4.38 eV for HAC[5,1]N@3a, and 3.82 eV for HAC[7,1]N@4a, due to the quantum confinement effect.[14,15,32] These peak As correlate with π–π* transitions with excitation configurations of HOMO – 2 → LUMO and HOMO → LUMO + 2, generating one more nodal plane in the destination MOs (cf. Figure ). There are small peaks of same transition configuration but different coefficients in the lower energy range, e.g., 3.26 eV for HAC[5,1]N@3a and 2.91 eV for HAC[7,1]N@4a, assigned as peak B. And the lowest transitions of HOMO → LUMO appear with the lowest energy in the spectra, e.g., 1.98 eV for HAC[5,1]N@3a and 1.25 eV for HAC[7,1]N@4a, assigned as peak C. All of the above peaks correspond to π–π* transitions, whereas the n–π* transitions are invisible because of symmetry forbiddenness. A comparison of representative peak positions is listed in Table .

Figure 6

Simulated UV–vis absorption spectra of selected quasi-1D HACs of different sizes and substitutional sites. The hindmost (gray) profiles correspond to pristine PAHs.

Table 1

Representative UV–Vis Absorption Peak Positions (in eV) and Transition Oscillator Strengths (Dimensionless, in Parentheses) of HACs with Simple Pyridine Moieties

peak	pyridine C5H5N	acridine C13H9N	HAC[5,1]N@3a C21H13N	HAC[7,1]N@4a C29H17N	configurationa
A	7.56 (0.44)	5.37 (1.71)	4.38 (3.05)	3.82 (4.11)	HOMO – 2 → LUMO
					HOMO → LUMO + 2
B	5.64 (0.03)	3.95 (0.05)	3.26 (0.10)	2.91 (0.18)	HOMO – 2 → LUMO
					HOMO → LUMO + 2
C	4.91 (0.00)	3.35 (0.05)	1.98 (0.04)	1.25 (0.03)	HOMO → LUMO
D	7.84 (0.00)	3.75 (0.00)	3.19 (0.00)	2.89 (0.00)	n → LUMO
E	8.26 (0.00)	5.24 (0.00)	4.94 (0.00)	4.34 (0.00)	n → LUMO + 1

MO numbering according to HAC[5,1]N@3a.

For N-hydrogenated moieties, the spectra show numerous small peaks, which correspond to various combinations of excitation configurations and depend on substitutional positions without apparent regulation. The only rough common feature is the major peaks, assigned as peak A′, with energies around peak A of the pyridine moieties. The peak A′ corresponds to the β HOMO → LUMO + 2 transition in HAC[5,1]NH@3a, where β MO energy levels are quite like the pyridine moiety (less perturbed by the extra electron of α-spin, cf. Figure b) and hence present roughly similar excitation energies. A similar situation applies to 2D graphene nanoflakes with single nitrogen dopants, whose absorption spectra are plotted in Figure S3 in the SI. The ground states of HAC[n,m]N with n > 3, m ≥ 3 are found to be open-shell singlet instead of closed-shell ones,[12,15,19] and this fact makes the spectra somewhat more complicated while keeping a similar overall trend. The pyridine moieties show a uniform size dependence and position independence, whereas the N-hydrogenated moieties present irregular profiles.

IR and Raman spectra could provide another aspect to distinguish simple pyridine and N-hydrogenated pyridine moieties. As shown in Figure , the unique peaks located around 3460 cm–1 corresponding to the N–H vibrational mode appear only for N-hydrogenated pyridine moieties. This peak could be distinguished from the N–H vibrational modes of protonated pyridine around 3410 cm–1, pyrrole moiety around 3520 cm–1, and amino groups around 3410 and 3500 cm–1.[33] (See Table S2 in the SI for details.) Although the intensity of this peak might be diluted as the compound grows larger, it would be clear and observable in small- to medium-sized HACs.

Figure 7

Comparison of simulated IR (upper panels) and Raman (lower panels) spectra of quasi-1D HACs and nitrogen-contained graphene nanoflakes. Whereas the C–H and C=C vibrational modes appear for all model species, the N–H modes are additionally found in the N-hydrogenated moiety.

In summary, HACs and N-doped graphene nanoflakes with pyridine moiety show a clear dependence on model sizes but hardly on substitutional positions of nitrogen in their UV–vis absorption spectra, whereas those with N-hydrogenated pyridine moiety show both dependency and fingerprint-like spectra. In addition, the unique N–H vibration mode in IR and Raman spectra belongs only to N-hydrogenated moiety. These give us clues to identify N-hydrogenated pyridinic HACs if they are synthesized.

Conclusions

In this work, we have considered geometric structures, energies, molecular orbitals, and electronic configurations of nitrogen-contained quasi-1D HACs and graphene nanoflakes with a “bottom-up” approach by increasing the number of building blocks. Many of these properties depend on the model sizes, i.e., numbers of fused aromatic rings, and are related to the positions of the substitutional nitrogen atoms. The most interesting finding is that whereas the aromatic pyridine moiety is known to be stable, the N-hydrogenated pyridine moiety could be even more stable given sufficiently large model compounds and proper substitutional sites. The hydrogenation reaction of the pyridine moiety is found to be exothermic for models of four and more fused aromatic rings with nitrogen substitution on the zigzag edges. The stability of the N-hydrogenated species could be interpreted as a multiple effect from the lowering of pyridinic LUMO energy in a large HAC, the correlation of the one extra electron with other electrons, and the formation of the N–H bond. However, the N-hydrogenated moiety is not yet reported in experiments probably due to its radical character and hence high reactivity.

We have also calculated and compared the UV–vis absorption spectra and IR/Raman spectra of these heterocyclic species. The pyridine moieties present π–π* transitions just like pristine PAHs, showing a strong size-dependence due to quantum confinement effect but a little position-dependence attributed to their isoelectronic characteristics. The N-hydrogenated moieties, on the other hand, possess irregular, fingerprint-like spectra case by case. Moreover, the pyridinic N–H vibrational modes are apparent and unique for the N-hydrogenated models. In conclusion, we have confirmed the energetic stability of the N-hydrogenated pyridine moiety in large HACs from the theoretical point of view and provided spectroscopic hints to detect them in the experiments.