Electroluminescence from 4-nitroaryl organic color centers in semiconducting single-wall carbon nanotubes

Organic color-centers (OCCs) are an emergent type of solid-state quantum emitter that can be synthesized in semiconducting single-wall carbon nanotubes (SWCNTs) by covalently bonding organic molecules to the nanotube sidewalls.^1,2 OCCs create potential wells that can trap mobile excitons generated by the photoexcited SWCNTs to form localized excitons with enhanced photoluminescence (PL) quantum yield compared to the native SWCNT emissions.³ In addition to neutral excitons, OCCs can trap SWCNT trions (i.e., a hole or electron combined with a neutral exciton), which feature high binding energy and non-zero spin.^4,5 The rich structural diversity of both the OCC functional groups and different chiralities of the SWCNT host results in tunable emission wavelengths in the near-infrared (NIR) region, including the telecommunication O band for quantum communication.^6–8 Furthermore, OCC-localized excitons are bright quantum light sources, even at room temperature.^8,9 As a result, OCCs in SWCNTs feature broad potential applications in imaging,¹⁰ sensing,^7,11–14 and quantum technologies.^8,15,16 However, OCC-localized exciton and trion luminescence have only been observed via photoexcitation using visible wavelength light resonant with one of the SWCNT Van Hove transitions.¹ 

Various applications of OCC-localized exciton and trion emission, particularly, relating to quantum information science, will require on-chip integration and access by electrical pumping. Additionally, electrical pumping can provide desirable control of the luminescence by gating the electron and hole charge carrier density,^17,18 which may provide further insight into the interaction between charge carriers and OCC-localized excitons and trions. Electroluminescence (EL) of pristine SWCNTs has been previously demonstrated by the recombination of electrons and holes injected by asymmetric contacts^16,19 and via impact excitation using unipolar carriers injected from symmetric contacts.^18–21 However, there is currently no report of EL from OCC-localized SWCNT excitons or trions.

In this work, we report evidence of EL from OCCs using highly pure and long (6,5)-SWCNTs functionalized with 4-nitroaryl OCCs via diazonium chemistry. The long SWCNTs are able to bridge the channel between symmetric gold contacts through which a current is applied to enable EL. Using a custom-built hyperspectral imaging microscope, both defect-induced PL and EL from the same OCC-functionalized SWCNTs can be mapped, revealing a direct correlation in the position that demonstrates the EL of OCC-localized excitons. Additionally, the exponential change in the current-dependent integrated EL intensity verifies that the EL is realized by the impact excitation mechanism. Such electrical pumping should be advantageous for next generation chip-integrated optoelectronics for display, telecommunication, sensor, and quantum science technologies.^{19,20,22–27}

To prepare long (6,5) chirality-enriched SWCNTs, CoMoCAT SG65i SWCNTs (lot no. SG65i-L39, 5 mg/ml) were individually dispersed in the 1 wt/v. % sodium deoxycholate (DOC, Sigma-Aldrich, >97%) aqueous solution by gentle stirring at room temperature for 18 months.²⁸ The gentle stirring helps preserve the length of the nanotubes in contrast to more aggressive dispersion methods, such as tip sonication, which mechanically cuts the nanotubes into shorter segments.²⁹ (6,5)-SWCNTs were isolated from this solution by aqueous two-phase extraction.²⁸ The polymer used in this purification process was then removed, and the solvent was exchanged to 1% wt/v DOC-D₂O (Cambridge Isotope Laboratories, Inc., 99.8%) by ultra-filtration (Amicon Ultra-15, 100 kDa cut-off). The final sample contained individualized >2 μm long, single-chirality pure (6,5)-SWCNTs. The purity and concentration of this (6,5)-SWCNT solution were characterized using a UV–vis–NIR absorption spectrometer (Lambda 1050, Horiba, Japan). DOC surfactant is known to cover the nanotube surface, which prevents subsequent covalent reaction.³⁰ Therefore, to enable the incorporation of OCC quantum defects, the SWCNT solution was diluted using 1% wt/v sodium dodecyl sulfate (SDS, Sigma-Aldrich, >99%)-D₂O to an optical density of 0.12 at the (6,5)-SWCNT E₁₁ transition (990 nm), which corresponds to a (6,5)-SWCNT concentration of 0.66 μg/ml.³¹ The final concentration of DOC was lower than 0.2% wt/v for the subsequent functionalization reaction.

Light-activated diazonium chemistry was used to incorporate 4-nitroaryl OCCs into (6,5)-SWCNTs.³² Briefly, SDS-dispersed (6,5)-SWCNTs were mixed with 4-nitrobenzenediazonium tetrafluoroborate (synthesized in advance from 4-nitroaniline and nitrous acid³) at a molar ratio of 500:1 (SWCNT carbon:diazonium). The solution was then irradiated with 565 nm light using a NanoLog spectrofluorometer (Horiba Jobin Yvon). The reaction progress was monitored over time by measuring the evolution of the new emission feature that results from OCC-localized excitons (E₁₁⁻) at 1150 nm using the spectrofluorometer.³ The reaction was terminated when the defect PL intensity was twice that of the native (6,5)-SWCNT emission (E₁₁ at 980 nm) by diluting the solution with 3.2% wt/v DOC–D₂O. The sample was stored in ambient for further use.

To align the nanotubes for the EL measurement, the 4-nitroaryl functionalized (6,5)-SWCNT solution [optical density at the (6,5) E₁₁ transition in the absorption spectrum = 0.1] was drop-cast onto a silica glass substrate patterned with Ti/Au electrodes (5-nm-thick Ti bottom layer and 50-nm-thick Au top layer; the distance between contacts was 5 μm and the width of each contact was 100 μm). The SWCNTs were aligned by applying an alternating voltage of 5 V at 1 MHz for 1 min to minimize the inter-nanotube contact, which can cause current and photon loss due to the cross-relaxation of energy between SWCNTs and nanotube bundles with different degrees of functionalization.^33,34 The surfactant was then washed away with deionized water to avoid current loss. The device was dried using a cotton swab and annealed at 60 °C for 0.5 h in vacuum. To protect the nanotubes from oxidation during the electrical pumping experiments, 1 mg/ml of poly(methyl methacrylate) (PMMA; analytical purity, Sigma-Aldrich) was dissolved in toluene and spun coat onto the device at 1750 rpm for 1 min. Finally, the sample was heated at 70 °C for 15 min.

The sample was mounted on the stage of a custom-built hyperspectral imaging system with both broadband and bandpass modes optimized for the NIR range.³⁵ Specifically, the imaging system was built on a Nikon Eclipse U inverted microscope with IR-optimized objectives and equipped with a 640 × 512 pixel InGaAs camera (Cougar-640, Xenics, Belgium) cooled to −190 °C by liquid nitrogen, as well as a volume Bragg grating and a 561 nm laser source (Jive Cobolt AB, Sweden). PL spectra were obtained by exciting the SWCNTs with the 561 nm laser using a 150× objective. For the EL measurement, a voltage (∼0–10 V) was applied to the source and drain electrodes in contact with the SWCNTs in the sample using an electrochemical workstation (BioLogic SP-150). EL measurements were taken at a loading electric field of 19 kV/cm across the channel. Both PL and EL spectra were collected from the same devices over the 900 to 1500 nm spectral range at an interval of 10 nm.

We functionalized the purified (6,5)-SWCNTs with 4-nitroaryl groups to produce (6,5)-SWCNT-C₆H₄NO₂ OCC quantum defects using diazonium chemistry [Fig. 1(a); see Sec. II for details]. Incorporation of the OCCs on the nanotube surface was confirmed by the excitation–emission PL spectra [Figs. 1(b)–1(d)].³⁶ The pristine (i.e., unfunctionalized) (6,5)-SWCNTs exhibited a single PL emission peak at ∼990 nm, originating from band edge E₁₁ excitons following laser excitation at 561 nm [Fig. 1(b)]. After functionalization, the resulting (6,5)-SWCNT-C₆H₄NO₂ sample featured two emission peaks at ∼990 nm and 1150 nm, originating from the E₁₁ excitons and OCC-trapped excitons (E₁₁⁻), respectively [Fig. 1(c)]. Compared with the E₁₁ excitons, the electron and hole of E₁₁⁻ excitons have a higher binding energy [Fig. 1(d)], which could cause localized band bending at the defect sites.³⁷ 

We used a symmetric Au-SWCNT-Au two terminal device design to investigate the possible defect EL [Fig. 2(a)]. Scanning electron microscopy (SEM) showed the aligned nanotubes are sufficiently long to cross the channel [Fig. 2(b)]. We then used a hyperspectral imaging system with a 561 nm excitation laser to characterize the PL and EL properties of the (6,5)-SWCNT-C₆H₄NO₂ from the same location. The resulting PL [Fig. 2(c)] and EL [Fig. 2(d)] images in the broadband mode from ∼900 to .1500 nm feature emissions of both the E₁₁ and E₁₁⁻ excitons of (6,5)-SWCNT-C₆H₄NO₂ at 990 and 1150 nm, respectively. However, while there are several bright spots indicating luminescence in the PL image [Fig. 2(c)], only some of these areas remain bright in the EL image [Fig. 2(d)], which is due to EL from some of the OCCs as to be confirmed in the following experiments.

To study the EL properties of the OCC-functionalized SWCNTs, the bandpass PL and EL images at 1150 nm were compared to isolate the OCC-localized E₁₁⁻ emission. The PL map shows more emission spots [Fig. 3(a)] than that of the EL image [Fig. 3(b)], which is consistent with the broadband emission results shown in Figs. 2(c) and 2(d), where PL can occur at all OCCs position, while the disturbing of charge transport by bundling and crosstalk between SWCNTs may hinder the EL from some OCCs. Figure 3(c) shows the PL and EL spectra from the same position as the (6,5)-SWCNT-C₆H₄NO₂ sample, as indicated by the red boxes in Figs. 3(a)–3(b). The EL features the same emission wavelength (1150 nm) as that of the defect-induced PL, suggesting the EL occurs through the same OCC-localized E₁₁⁻ excitonic state. Besides the E₁₁ and E₁₁⁻ emissions, another peak appears at ∼1325 nm in the EL map at different regions of the (6,5)-SWCNT-C₆H₄NO₂ sample, as shown in Fig. 3(d) [the same position in Figs. 3(a) and 3(b), as indicated by the black boxes]. The redshift of the E₁₁ and E₁₁⁻ emissions to 1050 and 1165 nm in EL compared to PL may be related to the bundling of SWCNTs, and the overlapping of the peak from E₁₁ and E₁₁⁻ emissions with the peak at ∼1325 nm. As this new emission at 1325 nm is not observed in the PL spectrum, the possible influence by SWCNTs of other chiralities can be excluded. Under electrical pumping, it is known that excess charge carriers are injected into the system, which could enable OCC-localized excitons to capture these free charge carriers to form trions.⁴ In contrast, such excess charge carriers are not generated under photoexcitation. Additionally, the energy shift (∼245 meV) between the E₁₁ emission and 1325 nm peak in the (6,5)-SWCNT-C₆H₄NO₂ EL spectrum is consistent with a previous observation of trion emission from functionalized (6,5)-SWCNTs,⁴ which further suggests that this new peak may be ascribed to OCC-localized trions [E_T, Fig. 3(d)]. The difference of the EL spectra in Figs. 3(c) and 3(d) may be due to different local densities of OCCs, with a higher density of OCCs having a greater chance of capturing mobile carriers to form trions.

To reveal the mechanism of EL from OCCs, we plotted the integrated EL intensity of the (6,5)-SWCNTs-C₆H₄NO₂ sample from ∼900 to 1500 nm as a function of the current [Fig. 4(a)]. The applied current and EL intensity display an exponential relationship, indicating that EL occurs in the OCC-functionalized SWCNT device via the impact excitation mechanism.³⁸ In this mechanism, high energy or electric field is needed to overcome the threshold energy barrier to form hot carriers.^21,39 In SWCNTs, the threshold electric field should be at least equal to the lowest exciton energy.²¹ However, momentum conservation typically increases the threshold to approximately 1.5E_g/eλ_ph, where E_g is the transition energy of electrons from the ground state to excited state of E₁₁ exciton energy level corresponding to the optical bandgap of ∼1.27 eV for (6,5)-SWCNTs⁴⁰ and λ_ph is the optical phonon scattering length (∼10–20 nm).^21,41,42 According to this equation, the threshold electric field to generate EL in a pristine (6,5)-SWCNT should be 0.95–1.9 MV/cm. In contrast, as shown in Fig. 4(b), the smallest voltage that we observe EL in the OCC-functionalized (6,5)-SWCNTs-C₆H₄NO₂ sample is ∼7.5 V (noise dominates below 7 V), corresponding to a threshold electric field of 0.015 MV/cm, which is two orders-of-magnitude smaller than the pristine (6,5)-SWCNTs. This lower threshold energy can be attributed to band bending of the SWCNT electronic structure,³⁷ which may be caused by the large potential difference between the non-functionalized and OCC-functionalized sites of the SWCNT.

Such band bending enables the impact excitation mechanism as it generates a much higher localized electric field compared to the average electric field across the channel, which can help accelerate mobile charge carriers. Due to the high work function of the gold electrodes (∼5.31–5.47 eV),⁴³ low HOMO level of the SWCNTs (≤5.05 eV),²⁰ and symmetric gold contact configuration, the majority of charge carriers injected into the nanotubes should be holes. Therefore, the high electric field generated by band bending at the OCCs should accelerate the injected holes into hot holes, which then transfer this added energy to valence electrons via collisions, forming free electrons [Fig. 4(c)]. Experimentally, we observed an exponential dependence of current on voltage [Fig. 4(d)]. This observation is an indication of band-to-band ionization characteristic for the generated charge carriers, as the band-to-band ionization will lead to an exponential increase of the concentration of charge carriers and the exponential increase of current as a function of the applied voltage.³⁸ The excited electrons can then interact with holes to form free excitons as well as localized neutral excitons and trions at OCCs, leading to the EL emission of the observed E₁₁, E₁₁⁻, and E_T peaks, respectively.

It should be noted that although holes are the major carrier in the SWCNTs due to the symmetric gold contact configuration and the relative work function of the electrodes, there is the possibility that electrons can still be injected into the SWCNTs due to the high local electric fields. However, the number of injected electrons should be very small. If there are large amount of injected electrons, they can recombine with injected holes directly to show a linear EL intensity–current relationship,²⁰ which was not the case in Fig. 4(a). Thus, the injection of electrons should not be a major source for the EL.

The electric-photon conversion efficiency is an important property for EL. For our case, however, it is difficult to quantify this parameter that warrants future studies. Although the EL emission intensity can be integrated from a single emission site, it is impossible to isolate the electrical current that passes through the corresponding SWCNTs from the thin film network spanning the entire channel width. There is also the possibility that inhomogeneous functionalization and the larger resistance caused by defects will cause the charge carriers to bypass the OCCs through other pristine SWCNTs.⁴³ The non-homogenously distributed SWCNTs may also induce noise and local shunts, which could disturb the electric field distribution. For these reasons, it is impossible to quantify the electron–photon conversion efficiency of the OCC-localized excitons. These issues may be resolved with single OCC single nanotube device in which all carriers are forced through the same path where the OCC stands.

In conclusion, we observe EL from both neutral excitons and charged trions at OCC quantum defects in SWCNTs. Hot holes are introduced as unipolar charge carriers through a symmetric Au-SWCNT-Au device configuration and contribute to the impact excitation of electrons in the valence band to form excited excitons and trions. Compared with the threshold electric field needed to generate EL in pristine SWCNTs, a much lower threshold electric field is needed to generate EL in (6,5)-SWCNTs-C₆H₄NO₂. This work suggests that quantum-defect-modified SWCNTs can be integrated as electrically pumped NIR light sources. Further work needs to be conducted using a single SWCNT-based device to quantify the EL efficiency. Significant improvement of EL efficiency also may be achieved by controlling the injection and recombination of charge carriers through the gating effect,²⁰ as well as by precise synthesis of OCCs with controlled trapping depth to tune the band bending potential.^36,44 Additionally, deterministic positioning of OCCs along the device and optimization of the device configuration and SWCNT-electrode contacts⁴⁵ may reduce the cross-relaxation of energy between nanotubes.^3,46 Although further studies are needed, the observation of localized EL from OCCs may open new opportunities to develop chip-integrated emitters^1,8 for quantum science and technologies.

This work is supported, in part, by the National Science Foundation through Award No. PHY1839165. The work also makes use of instrumentation components funded, in part, by NSF (No. CHE-1904488) and NIH/NIGMS (Grant No. R01GM114167). B.X. acknowledges the support from Fundamental Research Funds for the Central Universities and the Open Fund of the State Key Laboratory of Modern Optical Instrumentation of Zhejiang University. M.K. and P.W. acknowledge the Millard and Lee Alexander Fellowship in Chemistry from the University of Maryland.

The data that support the findings of this study are available from the corresponding author upon reasonable request.