Spatially resolved photoluminescence brightening in individual single-walled carbon nanotubes

Single-walled carbon nanotubes (SWCNTs) are quasi-1D graphitic materials with unique photophysical properties resulting from quantum confinement effects around the nanotube (NT) circumference.^1–3 This one-dimensional confinement gives rise to strongly bound electron–hole pairs, or excitons, that dominate optical transitions at room temperature.^4,5 Stable and robust excitonic photoluminescence (PL) makes SWCNTs desirable for applications requiring near-infrared (NIR) emitters;^5–11 however, extremely low quantum yields (QYs) often limit their practicality.^6,12 The QY of SWCNTs suspended in an aqueous solution ranges from a mere 0.01% to 1%.^6,13–15 Although this value increases in organic solvents,^16–18 SWCNT PL is still significantly weaker than PL from other NIR-emissive nanoparticles, such as colloidal semiconductor nanocrystals that demonstrate band-edge luminescence with QYs as high as 80%.¹⁹ 

The poor intrinsic PL QY for SWCNTs has motivated recent efforts to brighten their PL.^20–23 For example, adding aryl, epoxide, ether, or oxygen defects to the SWCNT sidewall create lower-energy traps for the exciton that have been found to be highly emissive.^24–30 By adjusting the chemistry on the NT sidewall over several different NT chiralities, emission from the defect states can be tuned across the NIR.^27,31–34 The bright, robust, and tunable PL signatures of these sp³ quantum defects have driven recent advancements in the use of SWCNTs as room-temperature single-photon sources^27,35,36 or as high-contrast in vivo fluorescence imaging agents.²⁹ 

Alternatively, it was discovered that the PL QY from the S₁₁ exciton state of DNA- or sodium dodecyl-sulfate (SDS)-wrapped SWCNTs could be enhanced upon the addition of molecular reducing agents such as Trolox, dithiothreitol (DTT), and β-mercaptoethanol (BME).³⁷ After brightening, individual SWCNTs demonstrated QYs ranging from 15% to 40%. The removal of the reducing agents from the NT suspension decreased the QY to pre-brightening levels. Studies of isolated NTs over a range of aqueous surfactant-SWCNT suspensions suggested that changes in the local dielectric environment and in the surfactant coverage of the NT were important factors responsible for the PL enhancement.³⁸ Altogether, although SWCNTs suspended in water could be made highly emissive through the addition of certain reductants for particular surfactants, much was still unknown regarding the brightening mechanism: it was unclear how the reducing agents interacted with the SWCNT, and whether the brightening effect could be more generally applied.

A significant limitation of all ensemble-based studies is that the results represent an average over multiple (n,m) chiralities and individual SWCNTs with identical chiral indices in inhomogeneous environments. On the other hand, a single-molecule (SM) microscopy is a powerful technique⁷ that clearly identifies properties of individual SWCNTs within an ensemble.^7,24,39–41 In addition, SM microscopy can provide important complementary information to spectroscopy studies through spatially resolving PL changes along the NT axis.^42,43 Here, we use SM PL microscopy to image the PL brightening in individual SDS-SWCNTs upon the addition of DTT. The PL enhancement factor varies, with some NTs brightening by only 16% while others brighten by almost a factor of 7. Those SWCNTs that are initially dimmer generally exhibit a larger PL increase than their brighter counterparts, which is consistent with previous studies on DNA-SWCNTs.³⁷ For SWCNTs longer than the diffraction limit, we observe a single-step, uniform brightening along the NT axis upon the addition of DTT. This unexpected behavior differs from localized exciton quenching at single defect sites, which has been well studied.^44–46 Additionally, we find that observation of significant PL brightening is dependent on the relative mobility of surfactant molecules surrounding the NT. These results support the hypothesis that surfactant rearrangement is responsible for increasing PL QY values after DTT addition.

Suspensions of CoMoCAT-manufactured SWCNTs (Sigma-Aldrich) were prepared at a concentration of 0.1–0.3 mg/ml in a 1 wt. % SDS (Sigma-Aldrich) aqueous solution. To increase the presence of long nanotubes in the suspension, a short sonication time of 5 min (@ 140 W/cm²) was used. After sonication, the sample was centrifuged (Heraeus Biofuge Pico) for 10 min at 13 000 rpm (16 060 g), removing nanotube bundles and other insolubilized material. Roughly 80% of the supernatant was collected and used for future experimentation. SWCNTs with lengths greater than 1 μm were routinely observed.

Absorbance measurements were collected using a PerkinElmer Lambda 950 spectrometer with a 1-cm path length, microvolume quartz cuvette (Starna Cells, Inc.). PL spectra were obtained using a home-built fluorometer system containing a SpectraPro 300i spectrometer (Acton Corporation) and liquid N₂-cooled InGaAs photodiode detector (Electro-Optical Systems, Inc.). All spectra were collected with a 1 s integration time and were corrected for grating and detector efficiencies. A 695-nm long-pass filter was placed in the emission pathway to block scattered excitation light. PL QY measurements were performed using IR-140 dye (Exciton, Inc.) in ethanol (QY ∼16.7%)⁴⁷ as the reference. Absorbance and PL spectra for SDS-SWCNT suspensions are presented in Fig. S1 in the supplementary material.

It was critical to ensure that the same nanotubes were imaged before and after the addition of DTT. To accommodate this requirement, individual SDS-SWCNTs were immobilized on poly(diallyl dimethylammonium chloride) (PDDA) (Sigma-Aldrich, 20 wt. % in water) coated quartz coverslips through electrostatic interactions. A silicone perfusion chamber (Grace Bio-Labs, 9 mm diameter, 1.8 mm thick, one-sided adhesive surface) was adhered to a clean quartz coverslip and then heated on a hotplate for a few seconds to ensure a secure seal. After cooling to room temperature, the chamber was incubated with 1% PDDA aqueous solution for 30 min and then rinsed twice with nanopure water. A dilute SDS-SWCNT sample was added to the chamber and incubated for a certain amount of time (see Sec. III) on the coverslip. The chamber was rinsed twice with nanopure water, followed by the addition of 50 μl of 1% SDS aqueous solution to maintain the solution phase. A diagram of the as-prepared sample chamber is presented in Fig. S2 in the supplementary material.

Imaging was performed on a home-built optical setup (see Fig. S3 in the supplementary material) that featured an inverted microscope (Nikon, TE-2000U) with a 100× oil immersion objective (Nikon, NA 1.3) and 635-nm dichroic mirror (Semrock, Inc.). Samples were photoexcited with a defocused beam (diameter ∼10 μm) provided by either an optical parametric oscillator (Coherent, Inc.) or a 532-nm continuous-wave diode laser (Coherent, Inc.). Either a 530 ± 40 nm band-pass filter (Thorlabs, Inc.) or a 568 ± 10 nm bandpass filter (Chroma Technology Corporation) was placed in the excitation pathway. The average excitation power density was kept below 5 kW/cm² to avoid exciton–exciton annihilation effects that would shorten exciton lifetime and reduce exciton diffusion length.^44,48 Emission was collected by a liquid N₂-cooled CCD (Princeton Instruments) after passing through a SpectraPro 2500i spectrometer (Acton Corporation). An 800-nm long-pass filter (Newport Corporation) was placed in the emission pathway to block any scattered excitation light. After accounting for the actual pixel size of the CCD and magnification of the objective being used, the pixel size in the PL images was determined to be 200 × 200 nm².

PL brightening experiments were conducted by adding 1 μl of 1 M DTT solution to a 50 μl sample solution. PL intensity changes were monitored by videos recorded at frame rates ranging from 1/100 to 1/500 ms⁻¹. For long SWCNTs, a two-dimensional spatiotemporal map was constructed using individual time traces of the PL intensity at locations along the nanotube axis.

Ensemble PL measurements of the SDS-SWCNTs probe sonicated for 5 min were acquired to verify that its brightening characteristics were similar to suspensions consisting primarily of short SWCNTs. Figure 1 shows PL spectra of the as-prepared SDS-SWCNTs before and after the addition of DTT. Similar to previous results,³⁸ the NTs exhibited a significant PL brightening with an average threefold enhancement. The initial PL QY of the sample was ∼0.08%, which is slightly lower than the QY of samples prepared using longer sonication times (∼0.12%), likely due to an increase in NT bundling for the shorter sonicated sample.⁴⁹ After the addition of DTT, the PL QY increased to ∼0.24%, which was determined using the following equation:

where the integrated PL intensity was determined from spectra acquired using the microscope setup (Fig. S3 in the supplementary material) and the QY (before) is simply the QY measured on the standard fluorometer. This relationship is accurate under the condition that the addition of DTT does not cause absorbance changes at the excitation wavelength, which was confirmed directly.

For single NT PL imaging, individual SDS-SWCNTs were fixed to PDDA-coated coverslips to ensure that PL from the same NT could be continuously monitored in an aqueous environment throughout the brightening process. Figure 2 presents widefield PL images of individual NTs before and after the addition of DTT. Without DTT, SWCNTs appeared relatively dim in the images, with many SWCNTs exhibiting PL intensities only slightly higher than the background. After adding DTT, SWCNT PL brightened by up to ∼6.5 times and, in some cases, additional NTs that did not appear in initial PL images became visible.

Brightening statistics for 52 individual SDS-SWCNTs observed both before and after DTT addition using single NT PL imaging are presented in Table S1 in the supplementary material and summarized in Fig. 3. The photon count distributions displayed in Fig. 3 clearly show that, upon adding DTT, the average detected PL photon counts shift to higher values. Additionally, the presence of DTT leads to a broadening of the photon count distribution, which is a characteristic of an increasingly inhomogeneous environment due to the interaction between the SDS-SWCNTs and DTT molecules. On average, NTs displayed PL enhancement by a factor of 1.9, which is smaller than the enhancement factor calculated for ensemble measurements. This difference can be attributed to the presence of SWCNTs not immediately visible in PL images acquired before DTT addition. Indeed, for the weakly emitting SWCNTs that were initially indistinguishable from background noise, PL enhancement factors could not be accurately calculated; however, these NTs often displayed PL intensity increases of more than one hundred counts (see Table S1 in the supplementary material for comparison), which would undoubtedly yield enhancement factors greater than the average of 1.9.

It is evident from the brightening statistics that PL enhancement was inconsistent across the 52 individual SDS-SWCNTs. Some NTs brighten by a mere 16%, while others significantly brighten by up to 6.5 times (Table S1 in the supplementary material). Additionally, there appears to be no correlation between NT length and the corresponding PL enhancement factor. This inconsistent brightening behavior is in many respects expected because PL properties of SWCNTs are heavily dependent on extrinsic factors such as structural and chemical defects,^{24,25,50–52} dispersing surfactant and solvent,^53–58 or other variations in the local environment.⁷ The intersection between these uncontrolled factors and the added DTT molecules undoubtedly lead to inconsistencies in PL brightening. Also, we found that SDS-SWCNTs displaying a lower initial PL QY were more likely to exhibit a larger PL enhancement (Fig. S5 in the supplementary material), an observation that is consistent with previous studies of DNA-SWCNTs.³⁷ Indeed, SDS-SWCNTs demonstrated a smaller average PL enhancement upon addition of DTT than DNA-SWCNTs, the latter of which is known to exhibit smaller initial QY values.³⁷ This QY-dependent brightening behavior suggests that there may be some inherent PL QY value that exists for SWCNTs that is reliant on optimal surfactant coverage, and the addition of DTT brings the QYs of single NTs closer to that value.

SM microscopy of SDS-SWCNTs longer than the diffraction limit affords the opportunity to investigate the spatially resolved dynamics of PL brightening. Figures 4(a) and 4(b) show PL images of a ∼3 μm long SWCNT before and after the addition of DTT, respectively. The NT exhibits significant PL enhancement upon addition of DTT, with the center of the NT appearing brightest. A video of the brightening process was used to construct time traces [Fig. 4(c)] at discrete locations along the NT axis. These time traces were used to develop a two-dimensional spatiotemporal map [Fig. 4(d)] that displays PL intensity as a function of NT length and experimental time, which more clearly illustrates the brightening dynamics. For this specific experiment, 1 μl of 1M DTT was added to the SDS-SWCNT sample at ∼28 s and the brightening process began in one uniform, single step along with the entire NT at ∼40 s. A slightly delayed PL enhancement after DTT addition is consistent with observations at the ensemble level and could result from a finite time for diffusion of DTT molecules to the NT. Aside from the intensity jump due to PL brightening, it is important to note that other PL characteristics remain unchanged upon the addition of DTT: PL emission appears to remain stable with no major intensity fluctuations over extended periods of time.

The PL enhancement ratio for each NT segment in Fig. 4(c) with a corresponding time trace was calculated and is presented in Table S2 in the supplementary material. Remarkably, the PL enhancement ratio from the SWCNT is fairly consistent across the entire SWCNT (enhancement ratio ∼3), except at the NT ends where chemical defect sites presumably quench the PL.^59–61 Also unexpected was the observation of a uniform increase in PL intensity at the same time point for each segment along the SWCNT axis, which strongly suggests that the phenomenon responsible for the brightening process acts in a delocalized fashion along the entire SWCNT. An analogous analysis was performed for several long SWCNTs and similar brightening dynamics were observed for all. Occasionally, the addition of DTT immediately led to a decrease in PL intensity, which eventually recovered and then increased in a uniform, single step after a few seconds. These transient artifacts are attributed to solution turbulence upon addition of DTT and can be readily observed in Fig. S4 in the supplementary material.

Previous studies have reported stepwise PL quenching in SWCNTs after exposure to acid, base, or diazonium reactants,⁴⁴ which is attributed to the quenching of mobile excitons at localized sites of reversible or irreversible chemical reactions. The addition of AuCl₃ to individual NTs led to the formation of an electronic impurity level which could locally trap excitons and quench PL.⁴⁵ For SWCNTs doped with oxygen or diazonium salts, localized exciton quenching of S₁₁ excitons has been reported.^62–65 Our observations related to exciton brightening are quite contrary to the localized quenching observed at defect sites. If single point defects did play a major role in PL brightening, then we would expect to see spatially localized PL enhancement upon DTT addition since the SDS-SWCNTs studied here are longer than the exciton diffusion length (100 to 600 nm).^45,66–69 Instead, every NT that was investigated regardless of length demonstrated uniform, single-step PL brightening that was consistent across the entirety of the NT. Clearly, the physical phenomenon responsible for the PL brightening effect is delocalized along the entire NT.

The single-step, uniform PL brightening of the entire SWCNT is consistent with the hypothesis that the addition of DTT triggers a rearrangement of SDS molecules adsorbed to the SWCNT surface. Surfactant reorganization would lead to changes in the local dielectric constant and offer better isolation from the aqueous environment,³⁸ which can impact radiative and nonradiative decay rates⁷⁰ and lead to an increase in PL QY.^{57,58,71–73} Because the SDS molecules wrap the entire SWCNT in a consistent manner, the PL brightening images that result from a rearrangement of surfactant molecules should be uniform along the NT axis.

The hypothesis of surfactant reorganization was confirmed by additional experiments designed to limit the mobility of SDS molecules on the NT surface. We thoroughly mixed an SDS-SWCNT suspension with a melted agarose gel prepared according to the method from Cognet et al.⁴⁴ This mixture was deposited on coverslips and brightening experiments were performed. As expected, the addition of DTT caused no changes in PL intensity because of the relative rigidity of the SWCNTs in a gel environment, which ultimately hindered the reorganization of SDS molecules adsorbed to the SWCNT. Additionally, increasing the incubation time for aqueous SDS-SWCNT suspensions resulted in no change or a decrease in the PL intensity upon the addition of DTT (Fig. 5). For samples incubated for longer periods, more SDS-SWCNTs were observed in the initial PL images, suggesting that the SWCNTs slowly adhered to the quartz coverslip. The adhesion of the NTs to the coverslip surface is due to the electrostatic interaction between negatively charged SDS molecules and positively charged PDDA which coats the coverslip. After a sufficiently long incubation time, an equilibrium forms between the bound SDS and PDDA molecules, deterring the reorganization of SDS and relegating DTT to a small molecule, which simply quenches SWCNT PL as previous studies have demonstrated.^44,74–76

Altogether, these experiments imply that the local environment plays a large role in SWCNT PL brightening. Prior to DTT addition, factors such as defect concentration, surfactant coverage, and the local dielectric environment, all of which can vary from NT to NT, lead to inconsistencies in PL QY measurements.⁴ The addition of DTT counteracts the effects of such factors by optimizing surfactant organization on the NT sidewall, resulting in PL brightening behavior that appears to be dependent on initial PL QY values exhibited by individual SWCNTs. Those NTs most heavily impacted by factors that diminish or quench initial PL generally demonstrate larger PL enhancement factors relative to NTs that are originally brighter. Not only does this suggest the existence of some intrinsic PL QY value that may be attainable for particular surfactant-SWCNT systems in an aqueous solution, but it also implies that the addition of DTT provides a degree of control over the local environment, protecting individual NTs from the factors that are detrimental to PL QY. This idea, when coupled with the observation that PL brightening occurs as a uniform, single-step process independent of localized defect sites, indicates that SWCNT PL QY can be improved substantially, regardless of the pristine nature, or lack thereof, of the initial sample.

SM PL imaging was used to study exciton PL brightening of individual SWCNTs solubilized using SDS upon addition of the mild chemical reductant DTT. PL brightening experiments were conducted for dozens of individual SDS-SWCNTs, revealing heterogeneous PL enhancement behaviors, with some NTs brightening by only a fraction while others by almost a factor of seven. Interestingly, SWCNTs that displayed lower initial QY values generally exhibited the largest increase in PL intensity after the addition of DTT, suggesting that in aqueous suspensions SWCNT PL QY is currently limited by external factors such as the environment. SWCNTs longer than the diffraction limit revealed a uniform, single step, non-localized brightening along the entirety of the NT axis. This behavior was quite unexpected because when molecules interact with the SWCNT to quench PL, they do so in a stepwise fashion at specific, localized defect sites. Our observations of PL brightening suggest that single point defects do not play a major role in this process. Instead, these results provide evidence for the hypothesis that the addition of DTT leads to a rearrangement of the SDS on the NT surface that favors larger PL QYs. The idea of surfactant reorganization as the mechanism for PL brightening upon the addition of molecular reducing agents is not specific to the DTT/SDS system, as it has been shown to impact DNA-wrapped SWCNTs in the presence of Trolox, DTT, and BME.³⁷ The results of the current study offer an avenue to further control PL brightening in SWCNTs such that they become competitive with other NIR-emitting nanoparticles.

See the supplementary material for the ensemble absorbance and PL spectra for SDS-SWCNTs, a picture and diagram of the sample chamber used for SM PL imaging experiments, a schematic of the home-built optical setup used for SM PL imaging experiments, PL images and time traces showing transient artifacts upon the addition of DTT, PL brightening statistics for 52 individual SDS-SWCNTs, PL brightening statistics for individual segments of a long SDS-SWCNT, and a scatterplot displaying PL enhancement factor vs initial PL counts for individual SDS-SWCNTs.

This work was supported by the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Science, Office of Science, U.S. Department of Energy (DOE) through Grant No. DE-FG0206ER15821. This work was also supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1939268.

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