Selective emergence of photoluminescence at telecommunication wavelengths from cyclic perfluoroalkylated carbon nanotubes

Single-walled carbon nanotubes (SWNTs) exhibit structure-dependent electronic and optical properties, including near-infrared (NIR) photoluminescence (PL) from semiconducting chiralities. While covalent functionalisation can introduce local band-gap modulations that generate red-shifted defect-state PL beyond the intrinsic E11 emission, achieving selective PL at telecommunication wavelengths (>1300 nm) from common chiralities such as (6,5) has been difficult. Prior work indicates three main levers for tuning PL: binding configuration of addends, electronic effects of substituents, and local strain at binding sites. This study addresses whether combining a cyclic addition motif (two reactive sites) with strong electron-withdrawing perfluoroalkyl groups can selectively produce longest-wavelength PL (≥1300 nm) in (6,5) SWNTs and whether this approach generalises across different chiral angles. The purpose is to expand NIR light sources for applications in bioimaging and optical communications by controlling local band gaps via rational chemical design.

Three primary strategies for SWNT PL tuning via functionalisation have been reported: (1) control of binding configurations, which dominates local band-gap modulation and can yield multiple defect-state PL peaks (e.g., aryl addends giving 1,2- and 1,4-isomers with distinct PL); (2) electronic substituent effects, where electron-withdrawing or donating groups on aryls or alkyls shift defect PL (e.g., para-substituent effects and fluorination tuning PL from ~1096–1155 nm); and (3) local strain at binding sites (e.g., cycloaddition reagents with two reactive sites producing selective PL near 1215–1231 nm). Prior studies showed arylation can produce multiple sharp PL peaks between 1000–1350 nm at low temperature; alkylation versus perfluoroalkylation shifts PL systematically; and dibromoalkanes can enforce specific addition geometries leading to selective defect PL. However, a bulk-scale, selective method to produce PL >1300 nm from (6,5) SWNTs had not been established before this work.

Synthesis/functionalisation: SWNTs (SG65i) were reduced with sodium naphthalenide (naphthalene 300 mg, sodium 156 mg in 100 mL anhydrous THF; argon; 1 h) and then reacted with 1,4-diiodooctafluorobutane I(CF2)4I (0.260 mL, 1.405 mmol). The mixture was sonicated (1 h pre-reduction, then 30 min after addend addition), quenched with dry ethanol (4 mL), filtered (0.1 μm PTFE), and thoroughly washed with THF, acetone, methanol, and water; solids were vacuum-dried. For comparative studies, mono-iodoalkanes/iodofluoroalkanes and 1,4-diiodo-2,2,3,3-tetrafluorobutane were used to prepare SWNT-R, SWNT−CH2(CF2)2CH2, and SWNT−(CF2)4 adducts. Separation: Dispersions were prepared in aqueous surfactants (e.g., 1.0 wt% sodium cholate (SC), bath sonication 3 h; ultracentrifugation 140,000×g for 1 h). Before HPLC, 2.0 wt% SDS was added. Chiral-index-dependent gel chromatography used a Sephacryl S-200 column (10×200 mm) at 23 °C with an eluent of 0.5 wt% SC + 0.5 wt% SDS + X wt% DOC (X: 0→1 gradient), flow 2.0 mL min⁻1, fraction collection every 5 mL. Fractions were reassessed in 1 wt% SC D2O for optical measurements. Optical characterization: UV–vis–NIR absorption (JASCO V-670), Raman (λ=561 nm; LabRAM HR-800; spectra normalized to G-band), and PL excitation–emission mapping (HORIBA Nanolog; 450 W lamp; Symphony-II CCD; excitation 400–1000/1100 nm, 1 nm steps; 10 nm slits; emission collected at 90°; intensity normalized to integration time) were performed. Circular dichroism (JASCO J-820) assessed optical resolution/helicity sorting. XPS confirmed fluorine incorporation. Computational studies: One-unit-cell models of (6,4), (7,3), (6,5), (8,3), and (7,5) SWNTs with terminal H-passivation were used. Two substituents were placed at 1,2- or 1,4-positions within a hexagon in six axial orientations (L++, L+, L− for 1,2 and 1,4). Geometries were optimized at B3LYP/6-31G; TD-DFT vertical transitions at B3LYP/3-21G. Spin densities for radicals (monofunctionalised SWNT radicals and alkyl/perfluoroalkyl radicals) were computed at UB3LYP/3-21G using 3-unit models for SWNT radicals. Gaussian 09 (Rev. E.01) was used. Relative energies (ΔE) among isomers and HOMO/LUMO analyses were evaluated to rationalize selectivity and band-gap modulation.

• Selective telecom-wavelength PL from (6,5) SWNTs: Cyclic perfluoroalkylation using I(CF2)4I produced a single predominant defect-state PL at 1318–1320 nm in bulk dispersions, surpassing previous (6,5) records.
• Fluorination and two-site cyclisation synergize: Reagents with two reactive sites (e.g., I(CF2)4I, ICH2(CF2)2CH2I, Br(CH2)4Br) exhibited higher selectivity (single E11* or E11** peak) than mono-iodoalkanes; increasing fluorine content systematically red-shifted E11* and E11**.
• Functionalisation degree trends: Raman D/G ratios after functionalisation decreased with increasing number of fluorine atoms (excepting some cases), indicating suppressed addition by fluorine near reactive sites. For SWNT-(CH2)3CH3, D/G=0.21 (iodo) vs 0.19 (bromo), indicating higher reactivity for iodoalkane.
• PL behavior by reagent: SWNT−(CH2)3CF3 (1104, 1240 nm) and SWNT−(CH2)2CF2CF3 (1118, 1243 nm) showed two PL peaks; SWNT−CH2(CF2)2CF3 (1119 nm), SWNT−(CF2)3CF3 (1152 nm), SWNT−CH2(CF2)2CH2 (1246 nm), SWNT−(CH2)4 (1230 nm), and SWNT−(CF2)4 (1318 nm) each showed a single dominant peak.
• Chiral generality: Post-separation, single dominant PL peaks persisted across multiple chiralities. For SWNT>(CF2)4, longest-wavelength PL among compared addends was observed for each chirality; e.g., (6,5): 1320 nm; (7,5): 1345 nm; (9,1): 1207 nm; (6,4): 1064/1222 nm; (7,3): 1373/1389 nm; (8,3): 1269 nm.
• Diameter dependence: The emission energy shift ΔPL (defect PL vs intrinsic PL) increased with reciprocal diameter (1/d^2), indicating stronger band-gap modulation for smaller-diameter SWNTs. Differences between alkyl and perfluoroalkyl addends separated substituent electronic effects from two-site cyclisation effects; SWNT>(CF2)4 showed the largest combined shift.
• DFT/TD-DFT insights: Calculated transition energies confirmed binding configuration as primary determinant of local gap, with consistent redshifts upon increasing fluorination for any configuration. HOMO–LUMO analysis showed lower LUMO for perfluoroalkylated vs alkylated adducts (electron-withdrawing effect). Relative energies indicated similar stability orderings to alkyl analogs; steric hindrance disfavors bis-perfluoroalkylation (e.g., 1,2-adducts of [(CF2)3CF3]2 are much less stable than 1,4), rationalizing single-emitter behavior for SWNT-(CF2)3CF3. Spin densities decreased with increasing fluorination for both radicals and SWNT radicals, explaining lower functionalisation degrees with iodofluoroalkanes.
• Selectivity exceptions: (6,4) and (7,3) SWNT>(CF2)4 occasionally showed two PL peaks, with chirality-dependent selectivity not yet fully understood.

The work demonstrates that combining a cyclic addition motif (two reactive sites enabling local cyclisation) with strongly electron-withdrawing perfluoroalkyl substituents effectively and selectively lowers local band gaps to produce longest-wavelength defect-state PL beyond 1300 nm from (6,5) SWNTs, a longstanding goal for telecom applications. Systematic experimental trends (Raman D/G, PL peak emergence and position, chiral-resolved PL after HPLC) align with computational predictions: binding configuration governs the defect-state energy, while substituent fluorination induces additional redshifts by lowering the LUMO. The observed linear-like increase of ΔPL with 1/d^2 across addends confirms that smaller diameters are more susceptible to local band-gap modulation, and that the magnitude of shifts from SWNT>(CF2)4 arises from both the cyclisation constraint (two reactive sites) and fluorination. The generality across several chiral angles suggests broad applicability for defect engineering of telecom-band emitters, although chirality-dependent selectivity anomalies (e.g., (6,4), (7,3)) indicate nuances in local structure–property relationships.

This study establishes a selective, bulk-accessible route to telecom-wavelength PL in SWNTs via cyclic perfluoroalkylation. Using I(CF2)4I, (6,5) SWNTs exhibit a single dominant PL peak at ~1320 nm, and similar longest-wavelength emissions are achieved across multiple chiralities. Mechanistic DFT/TD-DFT analyses attribute the large redshifts to the additive electronic effect of fluorination and the positional constraints from two-site cyclisation, with binding configuration as the primary determinant of defect-state energy. The results provide design rules for generating bright, long-wavelength quantum defects in SWNTs and broaden the parameter space for NIR emitters relevant to telecommunications and sensing. Future work should quantify PL quantum yields and stability, elucidate chirality-dependent selectivity (e.g., cases with dual peaks), optimize functionalisation degrees, and extend to other perfluoroalkyl lengths/configurations and host chiralities.

• Chirality-dependent selectivity is not fully resolved: some chiralities (e.g., (6,4), (7,3)) showed two PL peaks after perfluoroalkyl cyclisation, and the origin of this behavior remains unclear.
• Functionalisation degree is relatively low with iodofluoroalkanes due to reduced spin densities and steric effects, potentially limiting defect densities and brightness; quantitative PL quantum yields were not reported.
• While computational models capture trends, certain chiralities (e.g., (8,3)) deviate from general behaviors, indicating remaining gaps in theory–experiment correlation.
• The study focuses on spectral positions and selectivity; device-level performance (stability, integration, modulation) and long-term photostability were not evaluated.