Pyrazine-embedded N^N^N-chelating tetracoordinate organoboron polycyclic heteroaromatics with low-lying LUMO levels and near-infrared luminescence

Abstract

Pyrazine-based acceptor engineering offers an effective approach to lowering molecular orbital energies and developing near-infrared (NIR) luminophores, but remains less developed in tridentate-type tetracoordinate organoboron systems. Herein, we report a series of N^N^N-chelating tetracoordinate organoboron-fused heteroarenes constructed by combining an electron-deficient pyrazine core with tunable skeleton fusion. This molecular design effectively modulates the molecular geometry and electronic structure, leading to stabilized lowest unoccupied molecular orbital (LUMO) levels, narrowed optical band gaps, and distinct π-electron localization. As a result, the double boron-fused compounds exhibit markedly red-shifted absorption with molar extinction coefficients exceeding 1.0 × 10⁴ M⁻¹ cm⁻¹. Notably, BN-6 and 2BN-4–2BN-6 exhibit broad NIR emission with peaks extending to 1096 nm and show good thermal and photostability. In addition, these compounds can function as hybrid interfacial materials to improve the performance of organic solar cells. This work establishes pyrazine-enabled B ← N coordination and skeleton fusion as a useful strategy for constructing narrow-bandgap organoboron luminophores, offering useful molecular design strategies for NIR-I/NIR-II optoelectronic materials.

Pyrazine-embedded tetracoordinate organoboron emitters integrate N^N^N chelation, B ← N coordination and boron-driven skeletal fusion enabling precise electronic-structure modulation and tunable NIR-I/NIR-II luminescence extending to 1096 nm.

Introduction

Organic near-infrared (NIR) emitters, particularly those emitting within the first (NIR-I, 700–1000 nm) and second (NIR-II, 1000–1700 nm) near-infrared windows,¹ have emerged as pivotal materials for deep-tissue bioimaging,^2–5 optical communication,⁶ and advanced optoelectronic applications.^7–10 Compared with visible-light emitters, NIR luminescence can effectively reduce light scattering and autofluorescence interference in biological tissues, thereby significantly enhancing imaging penetration depth and signal transmission efficiency.² However, pushing emission toward longer wavelengths imposes intrinsic physical constraints. According to the energy-gap law,¹¹ narrowing the energy difference between the first singlet excited state (S₁) and the ground state (S₀) exponentially enhances non-radiative decay mediated by strong vibronic coupling, while simultaneously reducing the radiative transition rate. This fundamental competition complicates the molecular design of NIR emitters, making it highly challenging to simultaneously achieve strong NIR absorption, high efficiency, facile structural tunability, and NIR-II coverage.¹²

A variety of molecular design strategies have been developed to effectively modulate the energy gap of organic small-molecule NIR emitters, including π-conjugation extension,¹³ donor–acceptor (D–A) design to induce intramolecular charge transfer (CT),^14,15 structural rigidification,¹⁶ and heteroatom incorporation.¹⁷ These approaches have enabled the development of some of the most widely used NIR fluorophores, including cyanine,¹⁸ polymethine,¹⁹ xanthene,²⁰ and boron-dipyrromethene (BODIPY)-based fluorophores.²¹ Recently, B/N-doped polycyclic aromatic hydrocarbons (PAHs),^22–24 especially those with multiple-resonance (MR) characteristics,²⁵ have emerged as attractive molecular platforms because their distinctive electronic structures enable fine regulation of excited-state properties while maintaining narrow-band emission and high color purity. Nevertheless, most MR-B/N derivatives rely on short-range CT within triangulene-like frameworks, which inherently limits their emission to the visible or short-wavelength NIR-I region (<800 nm), and their extension into the NIR-II region remains largely unexplored.^26–33

To circumvent these limitations and further expand the structural diversity of NIR organic emitters, tetracoordinate organoboron complexes have attracted increasing interest as an alternative molecular platform owing to their distinctive electronic structures and unique configurational characteristics.^34,35 In contrast to tricoordinate systems, embedding boron atoms into nitrogen- or oxygen-containing chelating ligands to form rigid, fused-ring architecture not only improves thermal and photochemical stability but also enables effective tuning of the molecular orbitals and energy levels.^36–40 The electron-accepting character associated with the tetracoordinate boron centers, together with the polar B–N/B–O bonds and the B ← N/B ← O dative-bonding environment,^41–44 allows for a more pronounced lowering of the lowest unoccupied molecular orbital (LUMO), thereby narrowing the energy gap between the LUMO and the highest occupied molecular orbital (HOMO) and accessing a lower-lying S₁ state favorable for red/NIR emission.^{39,41,45–47} Within this context, the pyrazine unit is a particularly potent building block due to its strong electron-withdrawing capability and multidentate coordination sites.^48–52 Compared with previously reported N^N^N-chelated tetracoordinate boron emitters mainly based on pyridine-type acceptor units and O^N^O-chelated tetracoordinate boron-fused double helicenes (Fig. 1a), the incorporation of a pyrazine core provides a more electron-deficient and structurally versatile N^N^N-chelating framework. This unit not only deepens the LUMO level and narrows the orbital energy gap, but also serves as a key structural motif for constructing mono- and diboron-embedded heteroaromatic architecture, in which the number and orientation of B ← N interactions can modulate the orbital distributions and band gaps, thereby extending the emission from the visible to the NIR region.

Fig. 1. (a) Representative N^N^N- and O^N^O-chelated tetracoordinate boron compounds. (b) The molecular design strategy developed in this work to achieve NIR emission through pyrazine incorporation and skeleton fusion. Absorption and emission wavelengths were measured in solution.

Inspired by the structural tunability and intriguing optoelectronic properties of four-coordinate boron derivatives, herein we adopt a pyrazine-embedded N^N^N-chelating strategy to construct a series of B–N-doped tetracoordinate organoboron derivatives with good thermal and photostability (Fig. 1b). The synthesized double boron-centered 2BN-5 and 2BN-6 exhibit distinct saddle-shaped conformations, which effectively suppress close π–π stacking. Concurrently, the 2BN-4–2BN-6 exhibit low-lying LUMO levels (−3.66 to −3.81 eV) and narrow HOMO–LUMO gaps (1.41–1.58 eV). Consistent with this electronic modulation, their lowest-energy absorption bands in solution extend to 714–914 nm, with molar extinction coefficients exceeding 1.0 × 10⁴ M⁻¹ cm⁻¹. Notably, these emitters display broadband NIR emission with peak wavelengths ranging from 817 to 1096 nm and emission tails extending into the NIR-II window. Beyond their promising photophysical properties, the potent electron-accepting ability of these pyrazine-based diboron complexes renders them effective as hybrid interfacial electron-transporting layers in organic solar cells (OSCs), suggesting their potential applications in organic optoelectronic devices.

Results and discussion

Synthesis and characterization

A series of N^N^N-chelating tetracoordinate organoboron compounds was synthesized by fusing the pyrazine core with either 3,6-di-tert-butylcarbazole (TCz) or indole units. As shown in Scheme 1, the precursor ligands were readily synthesized in a single step via Pd-catalyzed Suzuki–Miyaura coupling in high yields (77–95%). Boron insertion was subsequently performed using PhBCl₂ as the borylation reagent and NEti-Pr₂ as the base in dry PhCl at 120 °C, efficiently producing monoboron (BN-4–BN-6) and diboron (2BN-4–2BN-6) derivatives in 80–90% yields. It is worth emphasizing that previous studies on multi-boron-bridged conjugated systems have revealed the inevitable formation of both cis and trans isomers arising from the phenyl substituents on the boron centers, which greatly complicates synthesis and purification.⁵³ By contrast, under the present reaction conditions, no additional byproducts were observed during purification, and single-crystal X-ray diffraction analysis verified the exclusive formation of the cis isomers (vide infra). Furthermore, the rigidified analogues BN-6 and 2BN-6 were obtained via intramolecular C–C oxidative cyclization of BN-5 and 2BN-5 in yields of 46% and 66%, respectively. All products were rigorously characterized by NMR and HRMS spectroscopy (see the SI). Specifically, ¹¹B NMR spectra in CD₂Cl₂ at 298 K displayed sharp singlet resonances (5.90–7.48 ppm), confirming the successful formation of tetracoordinate boron centers. In addition, these compounds exhibited good stability under ambient conditions and thermal robustness, with decomposition temperatures (5% weight loss) ranging from 326 to 438 °C (Fig. S1 and S2).

Scheme 1. The synthetic routes to the mono-boron series BN-4–BN-6 and the di-boron series 2BN-4–2BN-6.

Single-crystal structure analysis

As shown in Fig. 2, the single-crystal structures of BN-6, 2BN-5, and 2BN-6 unambiguously confirm the successful construction of the desired molecular structures. In BN-6, the central boron atom adopts a tetracoordinate geometry within the nearly planar N^N^N-chelated framework, featuring both B ← N coordination and B–N covalent bonds, and the phenyl unit attached to the boron center is oriented nearly perpendicular to the fused backbone. The B ← N coordination bond length is 1.64 Å, comparable to that previously reported for the BN3 emitter.⁵⁴ However, this bond is notably longer than those reported for other C^N^C-chelated (1.572–1.617 Å) and O^N^O-chelated (1.579 Å) tetracoordinate boron complexes (Fig. 2a).^55–57 In addition, the two B–N_TCz bonds between the boron center and the nitrogen atoms of the TCz units are 1.50–1.51 Å, consistent with typical σ-covalent bonding. In contrast, 2BN-5 exhibits a more pronounced multi-helical twisted conformation, resulting in a saddle-shaped geometry, with lateral spans of 9.94 and 11.87 Å for the two upturned sides, and corresponding local depression depths of 3.71 and 2.63 Å, respectively. Notably, the larger span of 11.87 Å is observed on the side where the two boron-bound phenyl substituents adopt a cis arrangement, reflecting the steric hindrance imposed by their cofacial orientation (Fig. 2b). Taking the conformer shown in Fig. 2b as an example, the aromatic segments along the molecular backbone are arranged in an alternating P/M helical pattern, and the overall configuration can be described as (P, M, P, M). The local helical dihedral angles of the P-helical segments, θ (A, H) and θ (D, E), range from 55.61° to 56.52°. In contrast, those of the M-helical segments, θ (B, C) and θ (F, G), range from 69.38° to 69.39°. Upon further annulation of the A/H and D/E ring pairs, the molecular framework becomes more conformationally constrained, leading to markedly decreased θ (B, C) and θ (F, G) values of 45.76° and 59.28°, respectively, in 2BN-6 (Fig. 2c). The two B ← N coordination bonds in 2BN-5 and 2BN-6 are 1.64–1.65 Å, whereas the B–N_TCz covalent bonds are considerably shorter, measuring 1.50–1.51 Å in 2BN-6.

Fig. 2. X-ray crystallographic structures of BN-6 (CCDC 2550662), 2BN-5 (CCDC 2550661), and 2BN-6 (CCDC 2550730). Top and side views of BN-6, 2BN-5, 2BN-6, respectively, with selected bond lengths and helical dihedral angles indicated. (a) Packing structure of BN-6 viewed from the side; (b) side view of the packing motif of 2BN-5, showing alternating P/M/P/M and M/P/M/P helicities. (c) Crystal packing of 2BN-6 viewed along the c axis, showing interlaced (M, M) and (P, P) helical conformers in the unit cell.

Crystal packing analysis further reveals that these three molecules adopt distinctly different assembly modes in the solid state. For BN-6, two molecules adopt an antiparallel face-to-face arrangement, forming a relatively ordered layered packing motif. The average intermolecular distance, d₁, between two adjacent molecules is 3.48 Å, indicating effective π–π interactions. By comparison, the longer average intermolecular distance d₂ within the unit cell is 9.07 Å, reflecting pronounced layered separation in the packing motif (Fig. 2a, S3 and S4). By contrast, 2BN-5 exists in the crystal as a pair of enantiomeric conformers, namely (P, M, P, M) and (M, P, M, P), arranged alternately within the unit cell (Fig. 2b and S5). Upon oxidative cyclization of 2BN-5, additional C–C bonds form between the A/H and D/E rings, further locking the distorted π-framework and affording 2BN-6 with a more constrained saddle-shaped conformation. In contrast to the packing motif of 2BN-5, 2BN-6 crystallizes as a pair of enantiomeric conformers with homochiral helical segments, namely (M, M) and (P, P), which are alternately arranged within the unit cell (Fig. 2c and S6). Owing to their pronounced helical distortion and nonplanar framework, adjacent 2BN-5 and 2BN-6 molecules do not adopt a simple fully overlapped face-to-face packing mode. Instead, they interlock in a spatially offset manner, giving rise to unique packing arrangements, while the nearly perpendicular phenyl substituents at the boron centers act as steric spacers that further separate neighboring π-frameworks.

Theoretical modelling

To gain deeper insight into how the incorporation of B–N covalent bonds and B ← N coordination bonds affects the electronic structure and aromaticity distribution of the molecules, we performed nucleus independent chemical shift (NICS), localized orbital locator-π (LOL-π), anisotropy of the induced current density (ACID), and two-dimensional isotropic chemical shielding surface (2D-ICSS) calculations on BN-4–BN-6 and 2BN-4–2BN-6 (Fig. 3).⁵⁸ The averaged NICS(1)_ZZ values, calculated at points 1.0 Å above and below the ring centers, reveal pronounced differences in the magnetic response among the various rings within each molecule. The peripheral fused benzene rings exhibit strongly negative values, indicating that these regions retain strong localized aromaticity. In contrast, the magnetic shielding effect of the central pyrazine N₂C₄ ring is significantly weakened and is highly sensitive to skeletal modification. (i) In the monoboron series, the averaged NICS(1)_ZZ values of the N₂C₄ ring are more negative than those in the double boron-centered analogues; for example, the corresponding values for BN-4 and 2BN-4 are −11.44 and −3.72 (Fig. 3a), respectively. This suggests that the introduction of the second boron center reconfigures the electronic distribution and local ring current within the central framework, thereby weakening the local aromaticity of the central N₂C₄ ring. (ii) Upon further construction of the seven-membered ring through oxidative C–C coupling, the averaged NICS(1)_ZZ value of the N₂C₄ ring increases further; notably, in the diboron system, it becomes positive (4.09), indicating that this ring evolves from weak aromaticity to a distinctly antiaromatic character. In addition, the NICS(1)_ZZ results show that the six-membered heterocycles containing N–B ← N units are weakly antiaromatic. In contrast, the corresponding seven-membered heterocycles exhibit more pronounced positive values, indicating enhanced antiaromaticity upon formation of the seven-membered rings.

Fig. 3. (a) Calculated NICS(1)_ZZ values (ppm) of BN-4–BN-6 and 2BN-4–2BN-6 at the B3LYP/6-311G(2d,p) level (the tert-butyl groups have been omitted). (b) LOL-π isosurfaces plotted at an isovalue of 0.5. (c) ACID plots shown with an isovalue of 0.02; the red arrows denote diatropic ring current arising exclusively from π electrons. (d) 2D-ICSS distributions mapped at 1.0 Å above the XY plane.

The LOL-π isosurfaces further elucidate the π-electron delocalization pattern in these molecules. As shown in Fig. 3b, the π electrons are mainly distributed over the peripheral fused benzenoid rings, and the central BN-containing heterocyclic region exhibits reduced or even disrupted delocalization, suggesting that the π-conjugation in these molecules is not uniformly distributed across the entire framework. Moreover, the ACID plots provide visual evidence for the extent and pathway of π-electron delocalization, allowing direct assessment of local/global ring currents and thus the aromatic or antiaromatic character of different ring units (Fig. 3c). Overall, the induced current is mainly distributed along the peripheral fused benzenoid rings. The central BN-containing heterocyclic region shows weakened and partially disrupted current pathways, suggesting relatively limited delocalization and a non-uniform global π-electronic structure. Notably, the region highlighted by the purple circle exhibits a sparser ACID isosurface and poorer current connectivity (Fig. 3c), indicative of further weakened local π delocalization. Consistently, the 2D-ICSS maps at 1.0 Å above the molecular plane also show a strongly inhomogeneous distribution of shielding and deshielding regions (Fig. 3d), further confirming the localized aromatic/antiaromatic nature of these systems.

Next, to obtain a reliable description of the excited-state energies, we employed the spin-component-scaled second-order algebraic diagrammatic construction (SCS-ADC(2)) method, which includes double-excitation effects to some extent through its second-order treatment.^59,60 For computational efficiency, the tert-butyl substituents on the peripheral carbazole units of BN-5–BN-6 and 2BN-5–2BN-6 were omitted. As these groups are electronically inert and do not participate in the conjugated framework, this simplification is expected to have only a minor influence on the frontier orbital distributions and excited-state properties. Based on SCS-ADC(2)/def2-TZVP calculations, the predicted ΔE_ST values for BN-4, BN-5, BN-6, 2BN-4, 2BN-5, and 2BN-6 are 0.25, 0.11, 0.02, 0.42, 0.37, and 0.16 eV, respectively (Fig. 4). Notably, the values for BN-4 and BN-5 align well with the experimentally determined singlet–triplet energy gaps (vide infra). Electron density difference plots further elucidate the electronic reorganization upon excitation. For BN-4–BN-6, the density changes in the S₁ and T₁ states are mainly localized on the pyrazine core and its adjacent π-framework, displaying an alternating pattern of increased and decreased density at neighboring atomic sites, indicative of localized and short-range CT character in their lowest excited states. In contrast, for 2BN-4–2BN-6, the density distributions extend across the saddle-shaped skeleton anchored by the two boron centers, suggesting enhanced π-electron delocalization and more pronounced intramolecular CT characters in the low-lying excited states. The relatively large ΔE_ST values of 2BN-4 and 2BN-5 can be attributed to substantial electron–hole overlap in their low-lying excited states, which gives rise to stronger exchange interaction (Fig. S7–S12). In particular, the seven-membered-ring fusion in BN-6 and 2BN-6 expands the spatial extent of electronic reorganization, thereby lowering the S₁ excitation energy and oscillator strength (f), which signifies enhanced CT character. These results are consistent with the markedly stabilized LUMO levels, narrowed HOMO–LUMO gaps (Fig. S13), and reduced vertical excitation energies, demonstrating that the diboron fusion strategy effectively stabilizes the low-lying excited states and enables access to the NIR region.

Fig. 4. HOMO and LUMO orbital energies (red), together with the corresponding vertical excitation energies and oscillator strengths (f), calculated at the B3LYP/6-31G(d,p) level; excited state energies (black), calculated at the SCS-ADC(2)/def2-TZVP level; and hole (skyblue)/electron (yellow) difference density plots for the S₁ and T₁ states of each emitter, plotted with an isovalue of 0.001. All calculations were performed in the gas phase.

Electrochemistry

The electrochemical properties of BN-4–BN-6 and 2BN-4–2BN-6 were investigated by cyclic voltammetry (CV) and differential pulse voltammetry (DPV) in DCM. The corresponding voltammograms are shown in Fig. 5, with key redox data summarized in Table 1. All compounds exhibit only one reversible reduction wave at a scan rate of 0.02 V s⁻¹. Upon increasing the scan rate to 0.25 V s⁻¹, the reduction peak currents increased proportionally while maintaining reversibility, indicating good stability of the reduced species under the electrochemical measurement conditions (Fig. S14). This reversible reduction behavior is closely associated with the electron-accepting nature of the electron-deficient pyrazine core. Relative to the Fc/Fc⁺ couple, the reduction potentials (E_red) of the monoboron species BN-4–BN-6 were −1.60, −1.63, and −1.55 V, corresponding to LUMO energy levels (E_LUMO) of −3.20, −3.17, and −3.26 eV, respectively. In contrast, the E_red values of the diboron analogs 2BN-4–2BN-6 shifted markedly toward more positive potentials, reaching −0.99, −1.13, and −1.14 V, with corresponding E_LUMO values of −3.81, −3.67, and −3.66 eV, respectively. The low-lying LUMO levels indicate that the peripheral electron-donating indole and TCz units exert negligible influence on the reduction process. Conversely, the transition from the monoboron to the diboron framework induces a pronounced anodic shift, suggesting that dual B ← N coordination further reinforces the electron-deficient nature of the whole molecular skeleton and leads to substantial stabilization of the LUMO.

Fig. 5. (a–c) The electrochemical characteristics of the BN compounds were investigated by CV and DPV in anhydrous CH₂Cl₂, using 0.1 M nBu₄NPF₆ as the supporting electrolyte and ferrocene/ferrocenium (Fc/Fc⁺) as the internal reference standard. (d) The HOMO/LUMO energy levels and the corresponding band gaps of the BN compounds were then estimated experimentally.

Table 1. Summary of the photophysical properties of BN-4–BN-5 and 2BN-4–2BN-5 in toluene.

Compound	λ abs a [nm] (ε [×104 M−1 cm−1])	λ em a [nm]	Stokes shiftb [cm−1]	E opt c [eV]	Φ PL d [%]	E red e [V]	First Eoxde [V]
BN-4	321 (0.91), 387 (0.51), 497 (1.20)	587/586	3085	2.22	0.20	−1.60	0.91
BN-5	323 (2.24), 404 (0.80), 529 (1.43)	623/815	2852	2.07	0.25	−1.63	0.70
BN-6	344 (1.81), 395 (2.39), 486 (0.92), 631 (0.33)	817/798	3608	1.66	—	−1.55	0.42
2BN-4	446 (1.47), 581 (2.16), 714 (1.63)	874/938	2564	1.44	—	−0.99	0.59
2BN-5	497 (1.57), 618 (3.19), 799 (2.39)	937/952	1843	1.38	—	−1.13	0.37
2BN-6	384 (3.06), 470 (1.46), 682 (2.61), 914 (1.05)	1096/1066	1816	1.15	—	−1.14	0.27

^aUV-vis absorption (toluene solution) and emission maxima in solution (left) and neat films (right).

^bStokes shift, Δν (cm⁻¹) = 10⁷/λ_max,abs −10⁷/λ_em.

^cOptical band gaps estimated from the absorption onsets in solution.

^dAbsolute photoluminescence quantum efficiency measured using an integrating sphere.

^eElectrochemical measurements in CH₂Cl₂ containing 0.1 M n-Bu₄NPF₆ at 298 K, values derived from the first reduction and oxidation peak potentials in the DPV curves.

Regarding oxidation, BN-4 and 2BN-4 exhibit a one-electron irreversible oxidation process (E_oxd = +0.91 and +0.59 V, respectively, Fig. 5a), whereas the remaining compounds display two well-separated oxidation peaks. For BN-5 (Fig. 5b), the E_oxd values (+0.70 and +0.93 V) are shifted anodically by 330 and 360 mV relative to 2BN-5 (+0.37 and +0.57 V). A similar trend is observed for BN-6 compared with 2BN-6 (Fig. 5c). These results suggest that the double boron-containing framework renders the compounds more readily oxidized, consistent with their higher-lying HOMO levels. Moreover, both series exhibit a progressive cathodic shift in oxidation potentials as the donor strength and conjugation increase, thereby reducing the band gaps (E_gap). Specifically, from BN-4 to BN-6, the HOMO level derived from the first oxidation onset shifts upward from −5.71 to −5.22 eV, and together with the slight lowering of the LUMO energy level, the corresponding E_gap gradually decreases from 2.51 to 1.96 eV. Among the diboron series, the HOMO reaches −5.07 eV in 2BN-6, leading to a remarkably narrow E_gap of 1.41 eV (Fig. 5d). Consequently, while peripheral donors exert a minor influence on the reductive processes, fusion expansion into a double boron framework via B ← N coordination significantly enhances the electron affinity.

Photophysical properties

The photophysical properties of these emitters were systematically investigated to elucidate the change in their electronic structures. In dilute toluene solution, BN-4–BN-6 exhibit absorption in the visible region. Their low-energy absorption bands (λ_max,abs) exhibit a gradual bathochromic shift from 497 to 631 nm as the π-conjugation of the molecular framework extends (Fig. 6a). In contrast, the double-boron analogs 2BN-4–2BN-6 display significantly more pronounced redshifted profiles, resulting in intense UV-vis-NIR absorption with maxima at 714, 799, and 914 nm, respectively. Notably, the emergence of prominent NIR-I absorption features suggests that the second boron center further stabilizes the pyrazine-centered LUMO and extends π-electron delocalization of the molecular framework, thereby strengthening the low-energy transitions. Specifically, 2BN-4 and 2BN-5 possess high molar absorption coefficients (ε) in this region (1.63 × 10⁴ and 2.39 × 10⁴ M⁻¹ cm⁻¹, Fig. S15 and S16), surpassing that of 2BN-6 (1.05 × 10⁴ M⁻¹ cm⁻¹). Consistently, the experimental absorption spectra and optical band gaps estimated from absorption onsets follow the same trend as the simulated spectra and calculated vertical excitation energies (Fig. S13, S17 and S18). As shown in Fig. 6b, the absorption spectra in the thin-film state mirror the solution-phase trends, suggesting that the intrinsic electronic characteristics are preserved in the condensed state. Furthermore, the film colors shown in the inset visually reflect the strong broadband absorption characteristics of these dyes. Because many long-wavelength NIR dyes suffer from limited photostability, especially when red-shifted through extended π-conjugation or donor–acceptor band-gap narrowing,⁶¹ the photostability of 2BN-4, 2BN-5, and 2BN-6 in toluene (1.0 × 10⁻⁵ M) was evaluated in comparison with a commercial naphthalocyanine dye (NP, λ_max,abs = 863 nm in toluene, CAS: 105528-25-4), which absorbs in a similar NIR region.¹⁸ Upon irradiation with a 75 W halogen lamp for 45 min, the absorbance of NP decreased by approximately 98%, while that of 2BN-4, 2BN-5, and 2BN-6 decreased by only 10–20% (Fig. S19), demonstrating the good photostability of these double boron-centered compounds.

Fig. 6. (a and b) UV-vis absorption spectra of the BN compounds in toluene solution (1.0 × 10⁻⁵ M) and in neat spin-coated films, respectively, with the insets showing photographs of the corresponding solutions and films under ambient light. (c) PL spectra of BN compounds recorded in toluene solution (1.0 × 10⁻⁵ M), λ_ex =460 nm for BN-4 and BN-5; λ_ex =660 nm for BN-6 and 2BN-4–2BN-6. (d) Transient PL decay profiles of BN-4 and BN-5 in degassed toluene solutions.

The solution photoluminescence (PL) spectra (Fig. 6c) reveal that BN-4 and BN-5 emit orange-red and red light, with emission maxima at 587 and 623 nm and PL quantum yields (Φ_PLs) of 20% and 25%, respectively. Increasing the solvent polarity from n-hexane to dichloromethane induces a ∼34 nm redshift in their PL maxima; such solvatochromism underscores the intramolecular CT character of their excited states, whereas the other molecules show only weak solvent-polarity dependence (Fig. S20). Transient PL measurements confirm that BN-4 and BN-5 exhibit thermally activated delayed fluorescence (TADF), with prompt fluorescence lifetimes of 6.96 and 8.34 ns and delayed lifetimes of 0.62 and 1.52 µs, respectively (Fig. 6d and S21). Compared with that under a nitrogen atmosphere, the PL intensity is noticeably decreased under oxygen, suggesting the involvement of oxygen-sensitive triplet excitons in the emission process. To further clarify the emissive mechanism, low-temperature (77 K) fluorescence and phosphorescence spectra were recorded in toluene. The estimated ΔE_ST values of 0.23 and 0.12 eV, respectively, provide strong support for a TADF mechanism (Fig. S21 and Tables S1–S2). Additionally, BN-6 and 2BN-4–2BN-6 exhibit broad NIR emission bands in solution, with maxima at 817, 874, 937, and 1096 nm, respectively, covering both the NIR-I and NIR-II regions. Notably, the emission tails of 2BN-6 extend beyond 1600 nm. Among them, BN-6, featuring a seven-membered ring, shows an exceptionally large Stokes shift of 3608 cm⁻¹. This behavior is likely attributable to the planarization of the N^N^N-chelating skeleton, which could facilitate intermolecular aggregation. Compared with those in solution, the neat-film PL spectra further confirm their solid-state NIR emission, with emission maxima ranging from 798 to 1066 nm (Fig. S22). Owing to the limited detection range of our transient PL setup, no reliable prompt or delayed fluorescence signals could be detected for BN-6 and the diboron compounds. In addition, their NIR fluorescence quantum yields could not be reliably determined due to instrumental limitations. Given that emission brightness (γ) is governed by both the Φ_PL and ε, i.e., γ = Φ_PL × ε, the relatively high ε values of 2BN-4 and 2BN-5 nevertheless suggest their potential for greater NIR emission brightness.^4,12

Organic solar cells (OSCs)

Given the relatively low-lying LUMO levels of 2BN-4, 2BN-5, and 2BN-6 (−3.81, −3.67, and −3.66 eV, respectively), their enhanced electron affinity is expected to facilitate electron extraction and transport at the cathode interfacial layer (CIL) while suppressing interfacial charge recombination. We therefore further explored their potential applications in binary OSCs by incorporating them into PDINN-based hybrid interfacial electron-transporting layers (Fig. 7a).⁶² The corresponding devices were fabricated with the architecture of ITO/2PACz/active layer (PM6:L8-BO)/CIL (PDINN containing 5 wt% 2BN-4–2BN-6)/Ag (150 nm).^62,63 The chemical structures of the materials used in the functional layers are shown in Fig. S23. As summarized in Table 2, incorporation of these tetra-coordinated organoboron compounds leads to improved device performance. Compared with the control device, the devices with B–N-modified PDINN interfacial layers exhibit simultaneously enhanced short-circuit current (J_SC) and fill factor (FF) values, affording power conversion efficiencies (PCEs) of 19.04%, 19.06%, and 19.60% for 2BN-4-, 2BN-5-, and 2BN-6-based devices, respectively, all higher than that of the pristine PDINN-based device (18.35%). In particular, the 2BN-6-modified device delivers the best overall performance, with an open-circuit voltage (V_OC) of 0.885 V, a J_SC of 27.14 mA cm⁻², and an FF of 81.61% (Table 2). Moreover, the EQE spectra of the devices with B–N-hybrid PDINN interfacial layers are consistently higher than that of the pristine PDINN-based device across a broad spectral range, particularly in the 350–850 nm region, indicating more efficient photocurrent generation and charge collection (Fig. 7b). These results demonstrate that tetracoordinate organoboron compounds with low-lying LUMO levels can serve as components of a hybrid interfacial layer to optimize cathode-side charge extraction and transport in high-performance OSCs.

Fig. 7. (a) J–V characteristics and (b) EQE curves of OSCs with pristine PDINN or PDINN: 5 wt% 2BN-x (x = 4, 5, 6) as the hybrid interfacial layer.

Table 2. Photovoltaic parameters of binary OSCs^a.

Active layers	Hybrid interfacial ETL	V OC b [V]	J SC c [mA cm−2]	FFd [%]	PCEe [%]
PM6: L8-BO	PDINN	0.877	26.10 (25.75)	80.17	18.35
	PDINN: 5 wt% 2BN-4	0.880	26.85 (26.02)	80.60	19.04
	PDINN: 5 wt% 2BN-5	0.883	26.92 (26.09)	80.20	19.06
	PDINN: 5 wt% 2BN-6	0.885	27.14 (26.15)	81.61	19.60

^aThe PDINN electron-transporting layer either without or with 5 wt% B–N compound.

^bOpen-circuit voltage (V_OC).

^cThe measured J_SC values from J–V curves and the integrated J_SC values from the EQE spectra.

^dThe fill factor (FF).

^eThe power conversion efficiency (PCE) was calculated as PCE = (V_OC × J_SC × FF)/P_in, where P_in is the incident light power density, typically set to 100 mW cm⁻² under standard AM 1.5 G illumination conditions.

Conclusions

In summary, this work demonstrates that pyrazine-enabled B ← N coordination, stepwise boron incorporation, and selected skeletal fusion provide a facile strategy for tuning tridentate tetracoordinate organoboron π-systems toward narrow-bandgap and NIR-active materials. Beyond simply extending π-conjugation, this strategy modulates the interplay among molecular distortion, aromaticity/antiaromaticity distribution, and frontier-orbital localization, thereby providing an effective means to access narrow-bandgap organoboron frameworks. The resulting mono- and diboron systems reveal how the number and spatial arrangement of boron centers can strongly influence LUMO stabilization, band-gap narrowing, and NIR optical responses. In particular, the double-boron architecture highlights the potential of multi-boron incorporation for constructing stable NIR-active heteroarenes with both photophysical and interfacial electronic functions. The ability of these compounds to improve OSCs' performance further suggests that tetracoordinate organoboron materials can serve not only as NIR luminophores but also as functional electronic materials. This study provides a molecular design platform for developing boron-based materials with tunable narrow-bandgap characteristics, and future work may further explore their applications in NIR organic light-emitting diodes and other organic optoelectronic devices.

Author contributions

G. M. conceived the idea for the study and designed the experiments. S. Z. synthesized and characterized the compounds. S. X. fabricated the single-carrier devices. X. C. and X. G. performed crystallographic analysis and carried out the theoretical calculations. H. W. and Z. Z. assisted in data analysis. G. M. and J. D. reviewed and edited the manuscript. All authors contributed to the manuscript and participated in the discussion of the results.

Conflicts of interest

There are no conflicts to declare.

Data availability

CCDC 2550661 (2BN-5), 2550662 (BN-6) and 2550730 (2BN-6) contain the supplementary crystallographic data for this paper.^64a–c

The data supporting this article have been included as part of the supplementary information (SI). Supplementary information is available. See DOI: https://doi.org/10.1039/d6sc04042j.