Mapping Structure-Property Relationships of Organic Color Centers

SUMMARY

Organic color centers are an emergent class of quantum emitters that hold vast potential for applications in bioimaging, chemical sensing, and quantum information processing. Here, we show that these synthetic color centers follow interesting structure-property relationships through comparative spectral studies of 14 purified single-walled carbon nanotube chiralities and 30 different functional groups that vary in electron-withdrawing capability and bonding configurations. The defect emission is tunable by as much as 400 meV in the near-infrared as a function of host structure and the chemical nature of the color centers. However, the emission energy is nearly free from chiral angle and family patterns of the nanotube host (although this strongly depends on the nanotube diameter), suggesting that a trapped exciton at the organic color centers to some degree electronically decouples from the one-dimensional semiconductor host. Our findings provide important insights for designing and controlling this new family of synthetic color centers.

Organic color centers are an emergent class of quantum emitters that hold vast potential for photonic, sensing, and optoelectronic applications. Covalently bonding functional groups to semiconducting single-walled carbon nanotubes creates molecularly tunable color centers where mobile excitons can be trapped and fluoresce brightly. Comparative studies of a series of 14 high-purity nanotube hosts and 30 organic color centers reveal new insights on the structure-property relationships that may be used to guide the synthesis of these centers featuring tailored light-emitting properties.

INTRODUCTION

Organic color centers are an emerging class of synthetic defect emitters that can be chemically incorporated into the sidewall of a single-walled carbon nanotube (SWCNT) through covalently bonding functional groups to the semiconductor host.^1–4 The introduced chemical defect locally modulates the electronic structure of the SWCNT host to enable trapping of excitons at the defect site, where they burst brightly as single photons.^1,5 The defect-induced emission (E₁₁ ⁻) can be significantly brighter than the native photoluminescence (PL) (E₁₁) of the nanotube and emits at the near-infrared (NIR),¹ suggesting vast potential for applications in imaging, sensing, and creating single photon sources at telecom bands.^3,6–8

The electronic and optical properties of organic color centers vary substantially with the structure of the SWCNT host, each of which is assigned a specific (n,m) chirality that describes its construction as a rolled-up graphene sheet. On the other hand, there are increasingly large numbers of functional groups that are being identified capable of creating organic color centers, suggesting virtually unlimited opportunities in broadly applying organic chemistry in this emergent field.^1,2,4,9–13 The unfunctionalized SWCNT hosts are quasi-one-dimensional nanostructures with highly predictable electronic structure and optical properties.¹⁴ Particularly, the excitonic transition energies of semiconducting SWCNTs show a strong dependence on both diameter and chiral angle, as summarized in correlation plots known as the Kataura plots.^15,16 In contrast, systematic analysis of the structure-dependence of organic color centers has yet to be established. It is important to understand how the defect state is related to the nanotube family pattern and chirality, and how the one-dimensional SWCNT couples to the zero-dimensional quantum state of the organic color centers, in order to provide a predicative understanding that will enable the design and synthesis of this family of quantum emitters for bioimaging, biosensing, quantum computing, and nanophotonics, as well as probing the rapidly unraveled fundamentally new phenomena of defect chemistry and physics. Even simply knowing the energies of these defect states for specific (n,m) structures and organic color centers would be important not only for fundamental photophysics of defect-trapped excitons but also for applications in infrared imaging, chemical sensing, and tailored design and synthesis of quantum materials.

Herein, we aim at establishing this structure-property relationship for organic color centers through controlled synthesis and comparative spectral studies of 30 chemically distinct organic color centers and 14 purified nanotube hosts. We observed distinct defect-induced PL features for 14 semiconducting SWCNT species (ranging from 0.62 to 0.94 nm in diameter and from 0° to 27.5° in chiral angle) using 30 different functional groups. Based on the spectrofluorometric measurements of perfluorohexyl defect-tailored (n,m)-SWCNTs, we analyzed the E₁₁ ⁻ as empirically fitted functions of nanotube diameter. Our results show that the measured E₁₁ ⁻ PL energy is tunable by 400 meV in the NIR as a function of SWCNT diameter. However, the emission energy of defect-trapped excitons is nearly free from chiral angle and family patterns, suggesting that an exciton at an organic color center to some degree behaves independently from the nanotube host but this color center-host coupling is chemically tunable depending on the chemical nature of the defect in terms of the group’s electron-withdrawing ability and bonding configuration. These findings provide a comprehensive picture of the structure-property relationships of organic color centers that is required to guide controlled and tailored synthesis of this new family of quantum emitters.

RESULTS AND DISCUSSION

Incorporation of Organic Color Centers in SWCNTs

We covalently attached functional groups into a series of (n,m)-SWCNTs to study the correlation of the photon energy emitted from the resulting organic color centers with the structure of the SWCNT host (Figure 1). In this work, we sorted 14 types of semiconducting SWCNT species into samples highly enriched in single chiralities using polymer aqueous two-phase (ATP) extraction¹⁷ or gel chromatography¹⁸ (Figure S1). The sorted SWCNTs were stabilized in 1 wt/v% sodium dodecyl sulfate (SDS) in D₂O for subsequent functionalization (see Experimental Procedures for detailed protocols). The semiconducting SWCNT structures studied in this work ranged from 0.62 to 0.94 nm in diameter and from 0° to 27.5° in chiral angle (Table S1). We then used diazonium¹ or alkyl/aryl halide^2,4 chemistry to covalently attach 30 different functional groups to the selected SWCNT chiralities (Table S2). The attached functional groups vary in electron-withdrawing capability and bonding configurations, enabling us to systematically modify the energy level of the defect state relative to the native electronic structure of the nanotube. The functionalized SWCNTs were characterized by UV-visible (vis)-NIR absorption and PL spectroscopy. The E₁₁ and E₁₁⁻ wavelengths were determined by fitting the PL spectrum at E₂₂ excitation using Voigt profiles.

Figure 1. Organic Color Centers in Semiconducting SWCNT Hosts.

(A) Schematic representation of organic color centers incorporated as perfluorohexyl defects into a series of (n,m)-SWCNTs. An individual color center creates a host-structure-dependent potential well where a mobile exciton can be trapped and fluoresce brightly.

(B) UV-vis-NIR absorption spectra of purified (5,4), (6,5), and (10,3) SWCNT solutions dispersed in 1 wt/v % SDS in D₂O.

(C) Excitation-emission PL maps of perfluorohexyl functionalized (5,4), (6,5), and (10,3) SWCNTs, from top to bottom.

Figure 1 displays the absorption spectra and emission-excitation PL maps of three exemplary nanotube species: (5,4), (6,5), and (10,3). Incorporation of perfluorohexyl organic color centers produces a new E₁₁⁻ PL peak at a redshifted wavelength from the native E₁₁ emission of the nanotube host. The E₁₁ ⁻ peak originates from the radiative recombination of trapped excitons from fluorescent defect sites,¹ as shown by the fact that both E₁₁ and E₁₁ ⁻ PL peaks resonate with the E₂₂ excitation of the SWCNT in excitation-emission maps of the functionalized samples (Figure 1). Although the attached functional groups (−C₆F₁₃) are the same, the emission wavelength of E₁₁⁻ was found to vary dramatically with the nanotube species, which implies a correlation with the host structures.

We tabulated the emission wavelengths of E₁₁ and E₁₁⁻ PL for the different nanotube chiralities studied and the energy difference between these two PL peaks (optical energy gap, ΔE = E₁₁ – E₁₁ ⁻) in Tables 1 and S3. The results provided the basis for an empirically determined energy plot of defect PL versus (n,m) chirality for semiconducting SWCNTs (Figures 2 and 3). Among the species investigated here, the longest wavelength of defect PL appears at 1,487 nm for perfluorohexyl functionalized (11,1) nanotube, (11,1)-SWCNT-C₆F₁₃. The shortest wavelength of defect PL occurs at 1,000 nm for (5,4)-SWCNT-3,5-C₆H₃(NO₂)₂. The largest ΔE value was 282 meV for (5,4)-SWCNT-C₆H₄NO₂, while the smallest was 92 meV for (7,6)-SWCNT-C₆H₁₃. This wide emission wavelength range covers the biological transparency window¹⁹ and most of the telecommunication regime,^3,8 and can be exploited for chemical sensing, bioimaging, quantum information, and a host of optoelectronic applications that require bright, high-quality light sources in the NIR wavelengths.^3,6–8

Table 1.

Optical Characteristics of Selected Combinations of Nanotube Chirality and Organic Color Centers

	1,052 nm	1,027 nm	-	1,000 nm	-	1,025 nm
	282 meV	265 meV		238 meV		275 meV
	211 meV	199 meV		178 meV		206 meV
	1,066 nm	1,082 nm	1,049 nm	1,084 nm	1,053 nm	1,097 nm
	247 meV	264 meV	228 meV	265 meV	232 meV	277 meV
	185 meV	198 meV	171 meV	225 meV*	174 meV	208 meV
	1,160 nm	1,190 nm	1,150 nm	1,177 nm	-	1,211 nm
	172 meV	198 meV	156 meV	184 meV		206 meV
	129 meV	148 meV	117 meV	138 meV		154 meV
	1,115 nm	1,128 nm	1,086 nm	1,169 nm	-	-
	240 meV	241 meV	202 meV	287 meV
	180 meV	181 meV	151 meV	215 meV
	1,145 nm	1,152 nm	1,105 nm	1,160 nm	1,132 nm	1,168 nm
	178 meV	190 meV	137 meV	190 meV	161 meV	200 meV
	133 meV	142 meV	103 meV	143 meV*	121 meV	150 meV
	1,159 nm	1,169 nm	1,124 nm	1,171 nm	1,127 nm	-
	229 meV	238 meV	195 meV	237 meV	200 meV
	171 meV	178 meV	146 meV	160 meV*	150 meV
	1,310 nm	-	-	-	-	-
	137 meV
	123 meV
	1,189 nm	1,206 nm	1,174 nm	1,192 nm	1,164 nm	-
	167 meV	173 meV	151 meV	162 meV	136 meV
	152 meV	156 meV	130 meV	146 meV	122 meV
	1,276 nm	1,284 nm	1,228 nm	1,282 nm	1,242 nm	-
	141 meV	149 meV	102 meV	146 meV	114 meV
	127 meV	134 meV	92 meV	132 meV*	103 meV
	-	-	-	1,225 nm	-	-
				160 meV
				144 meV
	1,281 nm	1,291 nm	1,228 nm	1,281 nm	1,227 nm	-
	135 meV	134 meV	92 meV	124 meV	92 meV
	121 meV	120 meV	83 meV	112 meV	83 meV
	-	1,270 nm	-	-	-	-
		137 meV
		123 meV
	1,486 nm		-	-	-	-
	137 meV	137 meV
	123 meV	123 meV
	1,455 nm	1,445 nm	-	-	-	-
	127 meV	126 meV
	114 meV	113 meV

Listed are the emitting wavelength (E₁₁ ⁻, top), the optical shift from that of the native exciton (ΔE, middle), and the thermal trapping potential of the E₁₁ ⁻ exciton (E_trap, bottom). Note that the PL spectra were obtained from SWCNTs in 1% SDS-D₂O. Changes in the dispersion media will cause deviations in the emission wavelengths from the reported data due to dielectric screening effects.²⁰ Also note that some E_trap values are derived from linear extrapolation of experimentally measured series (asterisks).

Figure 2. Tunable Defect Photoluminescence as a Function of Host Chirality.

Experimentally determined E₁₁ ⁻ wavelength of (A) (n,m)-SWCNT-C₆H₄NOM₂, (B) (n,m)-SWCNT-C₆F₁₃, and (C) (n,m)-SWCNT-C₆H₃(NO₂)₂ on a graphene sheet showing (n,m) lattice points. The color bar shows the emission wavelength of the defect PL. A lattice with diagonal stripes represents non-emitting metallic SWCNTs, in which (n - m)/3 is equal to an integer. A filled lattice indicates a semiconducting SWCNT chirality studied in this work. The blue arrow and θ in (A) represent the chiral vector and chiral angle of (6,2)-SWCNTs as a demonstration of SWCNT chirality. The molecular structure of the organic color center is specified in each part of the figure.

Figure 3. Semiconductor Host-Color Center Relationship.

Closed and open circles indicate mod 1 and mod 2, respectively.

(A) Positive correlation between native (E₁₁) and defect-induced PL (E₁₁ ⁻) wavelengths. The solid line is a quadratic function drawn to guide the eye.

(B) Diameter dependence of ΔE. Thesolid line is an empirical fitting to the inverse second order of the diameter (Equation 1).

(C) Curvature chirality effect on ΔE. The dashed line indicates metallic armchair SWCNTs with θ = 30°. Each point represents different (n,m) species that are covalently functionalized with perfluorohexyl defects.

Structure-Property Relationships of the Defect PL Emission

Each SWCNT structure can be uniquely indexed by a pair of (n,m) integers to indicate the chirality of the nanotube based on the roll-up vector of a graphene sheet (Figure 2). The direction of rolling determines not only the physical structure, such as diameter and chiral angle,²¹ but also the electronic band structure of the nanotube.^14,15,22 Specifically, since an SWCNT imposes boundary conditions on the electron wave function in the direction of rolling, for semi-conducting (n,m) SWCNTs, dividing (n – m) by 3 leaves a remainder of 1 or 2, which is classified as mod (n – m, 3) = 1 or mod (n – m, 3) = 2, respectively. If (n – m) is evenly divisible by 3, the SWCNT is metallic and therefore does not emit PL.

Covalent functionalization may break the intrinsic symmetry of the SWCNT by converting sp² carbon to sp³, which modifies the energy level of the functionalized nanotube locally at the site of the defect.^1,2 At this local defect state, the optical properties, including the emission energy and the size of the exciton,²³ are different from those of the intrinsic E₁₁ excitons. Hence, we analyzed the structure-dependent properties of defect-trapped excitons in (n,m)-SWCNTs with perfluorohexyl, 4-nitroaryl, and 3,5-dinitroaryl defects, and mapped out the structure dependence of the E₁₁ ⁻ emission wavelength onto the graphene lattice (Figures 2 and S2). The results show in general that the E₁₁ ⁻ for larger SWCNT diameters emits at longer wavelength, but it also depends on the chiral angle of the nanotube.

We observed a positive correlation between E₁₁ and E₁₁ ⁻ emission energies for the studied chiralities with the same perfluorohexyl functional groups (Figure 3A), indicating that the defect state is closely related to the E₁₁ excitonic state. The energy difference between E₁₁ and E₁₁ ⁻ (ΔE values) were examined for dependence on diameter (d), chiral angle (θ), and mod (n – m, 3). We explored the diameter dependence by fitting ΔE with the inverse second-order equation of diameter (Equation 1):

$$\Delta \mathrm{E}∕\mathrm{meV}=11.3{\mathrm{d}}^{-2}+253{\mathrm{d}}^{-1}-158,$$ (Equation 1)

in which d is the SWCNT diameter in nanometers. Figure 3B shows in general an inverse correlation between ΔE and diameter. However, we also observe 28% deviation of ΔE on average from Equation 1 for the range of diameters studied, suggesting diameter alone cannot account for ΔE variation, thus the chirality effect should be considered. Although the ΔE versus chirality plot displays no obvious pattern (Figure 3C), we found that the deviation of ΔE is mod dependent with higher ΔE values for structures of mod (n – m, 3) = 2 than for mod (n – m, 3) = 1. This mod-dependent deviation of ΔE is reminiscent of what was reported for the E₁₁ energy and diameter correlation.^14,16 The mod-dependent deviation represents the degree of electronic decoupling and is discussed in the next section. Even though the empirical fitting provides an estimate of the defect PL energy on the ensemble level, the PL energy of an individual quantum defect can vary due to the bonding configuration, as predicted by Kwon et al.² and Gifford et al.²⁴

Exciton Trapping Potential at Organic Color Centers

The exciton trapping potential (E_trap) is the minimum energy required to detrap the E₁₁ ⁻ exciton from the defect trap to restore as a free E₁₁ exciton. It is an important parameter to understand the structure-related properties of excitons, as well as to predict the PL stability. As established in our previous work,²⁵ the trapping potential of a defect state can be experimentally determined by monitoring the E₁₁ and E₁₁ ⁻ PL as a function of temperature or calculated from the difference between the optical energy gap (ΔE) and the reorganization energy (λ).

Reorganization occurs due to deformation of the nanotube geometry upon exciton trapping at the defect site²⁵ and is related to the exciton localization at the defect site,^26,27 with greater localization presumably leading to a larger reorganization energy. When the density of defects increases, the exciton wave function may be delocalized across multiple defects, leading to weaker localization of excitons at the defect site and smaller reorganization energy. For the series of perfluorohexyl defect-tailored SWCNTs, we also experimentally derived a larger λ in smaller diameter SWCNTs (d < 0.84 nm, Figure S3). Such greater spatial localization of the wave function effectively increases the amount of exciton-phonon coupling and thus increases reorganization energy in small-diameter SWCNTs. By linear extrapolation from the experimentally determined series (Table 1), we derived that the trapping potential of perfluorohexyl defect-trapped excitons ranges from 113 to 200 meV, as shown in Figure 4A.

Figure 4. Trapping Potential of E₁₁ ⁻ Exciton at Organic Color Centers.

(A) Diameter dependence of the trapping potential of E₁₁ ⁻ excitons in (n,m)-SWCNT-C₆F₁₃.

(B) Diameter dependence of E₁₁ (gray) and E₁₁ ⁻ (blue) emission energies. The labels categorize the (2n + m) families. Filled and open circles mark mod 1 and 2 SWCNT structures, respectively.

Our results suggest that the trapping potential is related to the size of the trapped exciton (electron-hole separation). Fora larger trapping potential, the wave function of the exciton, which can be computed using density functional theory,²⁵ is more spatially localized at the defect site.^23,25,28 This is congruent with our previous theoretical prediction that an E₁₁ exciton of a (6,5)-SWCNT is squeezed by 17% in size when trapped at an aryl defect.²⁵ The inverse correlation between E_trap and diameter (regardless of λ values) implies that, in smaller diameter SWCNTs, a defect trap can effectively reduce the exciton size and increase the oscillator strength of the defect state.²⁸ Strong localization at a deep trap may enhance the binding energy of the E₁₁ ⁻ exciton, and improves the stability of the defect-trapped excitons. This may be related to the diameter-dependent quantum yield enhancement¹ and PL stability of E₁₁ ⁻ excitons in single photon emission.^8,29 Although the photon conversion efficiency is an important parameter for many potential applications of organic color centers, it remains a challenging and labor-intensive task,¹ which warrants the development of more efficient techniques to quantify this value. Extrapolating beyond d > 0.94 nm (the largest diameter studied here), our empirical fitting of E_trap (solid line in Figure 4A) predicts that sp³ defects created by covalent functionalization create shallow traps in large-diameter SWCNTs (e.g., few millielectronvolts for d > 1.3 nm), making trapping and radiative recombination of E₁₁ ⁻ excitons less efficient.

Another interesting finding is that only a weak chirality dependence was observed for the E₁₁ ⁻ fitting. Figure 4B shows the empirically fitted emission energy of the E₁₁ and E₁₁ ⁻ exciton as a function of diameter for 12 SWCNT chiralities that are tailored with perfluorohexyl organic color centers. There is a clear inverse correlation between nanotube diameter and exciton emission energy. The deviation from the diameter fitting is due to the chiral angle dependence. This dependence becomes more apparent if we group the same nanotube families (2n + m) with a solid line, indicating the set of SWCNTs with similar diameters but different chiral angles.^15,16,22 It is also clear that the curvature effects in E₁₁ ⁻ are less significant than those in E₁₁. These two trends may be understood by examining the electronic structure of the nanotube host and the molecular nature of the defects. The E₁₁ energy levels influence the energy levels of the defect states because the defect state originates from the splitting of the doubly degenerate frontier orbitals of the SWCNT host^1,23 and thus exhibits some degree of chiral angle dependence that arises from the trigonal wrapping effect inherent in the host.²² Meanwhile, once a mobile exciton is trapped at an sp³ defect, the trapped exciton manifests photophysics that is different from the E₁₁ exciton, which is governed by the sp² symmetry of the SWCNT lattice. Thus, our data suggest that, as the defect trap becomes deeper, the exciton localization is stronger, and the optical and electronic properties of the trapped excitons become closer to an isolated zero-dimensional system with weaker dependence on the chiral angle of the nanotube host.

Lastly, Figure 5 demonstrates that the trapping potential and emission energies of defect-trapped excitons are highly tunable depending on the chemical nature of the defects. Here we specifically studied (6,5)-SWCNT because it is easier to prepare high-purity samples compared with other chiralities. The spectral characterization of all defect-tailored (6,5)-SWCNTs studied in this work is available in Figure S4. We note that this correlation between the chemical natures of the defect and ΔE is observed for other chiralities as well (Table 1). Due to the wide choice of defects (30 different functional groups), the defect PL of (6,5)-SWCNTs can be broadly tuned over the emission wavelength of 1,094–1,164 nm. We found that, in general, E_trap, as well as the E₁₁ ⁻ wavelength and ΔE, increased as the electron-withdrawing ability of the defects became stronger. The electron-withdrawing ability can be quantified using Hammett constants for aryl defects³⁰ and Taft constants for alkyl defects.³¹ From the study presented here, we show that the chemical nature of defects has a significant impact on the organic color centers, as manifested in E_trap, which follows a linear correlation with both Hammett (σ) (Figure 5A) and Taft constants (σ*) (Figure 5B). Combining the results of Figures 4 and 5, we derived empirical prediction models of ΔE for monovalent aryl-and alkyl-defect-tailored SWCNTs. For aryl-defect functionalized SWCNTs,

$$\frac{\Delta \mathrm{E}}{\mathrm{meV}}=\frac{17.3}{{(\mathrm{d}∕\mathrm{nm})}^2}+\frac{235.5}{\mathrm{d}∕\mathrm{nm}}-146.3+\Delta \sigma \left(613.2-\frac{559}{\mathrm{d}∕\mathrm{nm}}\right).$$ (Equation 2)

For alkyl-defect functionalized SWCNTs,

$$\frac{\Delta \mathrm{E}}{\mathrm{meV}}=\frac{11.3}{{(\mathrm{d}∕\mathrm{nm})}^2}+\frac{253}{\mathrm{d}∕\mathrm{nm}}-158+\Delta {\sigma }^{\ast }\left(205.1-\frac{248.6}{\mathrm{d}∕\mathrm{nm}}\right),$$ (Equation 3)

in which Δσ is σ – 0.788 and Δσ* is σ* – 4.87. Due to the chiral angle dependence of the E₁₁ and E₁₁ ^— PL, the chiral angle dependence generates, on average, 9% deviation from the experimentally determined values from the diameter fitting.

Figure 5. Trapping Potential of E₁₁ ⁻ Excitons at Organic Color Centers in (6,5)-SWCNT Host.

(A) The inductive and resonance effects of terminating aryl moieties on E_trap.

(B) Inductive effects of alkyl chains on E_trap. The solid lines in (A) and (B) are linear fitting of the correlations.

(C) Divalent methyl (square), aminoaryl (circle), aryl (triangle), and perfluoromethyl (diamond) defects have larger E_trap compared with their monovalent counterparts. The dashed line is E_trap (monovalent) = E_trap (divalent) drawn to guide the eye.

Lastly, we note that bonding configuration also influences E_trap and E₁₁ ⁻ PL (Figure 5C). Divalent defects, including >CF₂, >CH₂, >C₆H₄, and >C₆H₃NH₂, tend to create deeper defect potentials and increase the exciton trapping potentials compared with their monovalent counterparts (-CF₃, -CH₃, -C₆H₅, and -C₆H₄NH₂).

In conclusion, we have synthesized a series of organic color centers in semiconducting SWCNTs and determined the diameter and chirality dependence of PL emission wavelengths and the thermal trapping potential for defect-trapped excitons. Similar to the native exciton (E₁₁) in unfunctionalized SWCNTs, the E₁₁ ⁻ PL is strongly correlated with nanotube diameter. The emission energy of defect-trapped excitons, however, is largely free from chiral angle and family patterns of the semiconductor host, suggesting that an exciton at an organic color center to some degree decouples from the one-dimensional nanotube host. Our work establishes the structure-property relationships for organic color centers, which may help guide the controlled and tailored synthesis of this new family of quantum emitters for applications in bioimaging, quantum information, and optoelectronics in general.

EXPERIMENTAL PROCEDURES

A detailed description of the synthesis protocols can be found in the Supplemental Experimental Procedures.

High-Purity SWCNT Sorting

We sorted single-chirality SWCNTs, including (11,1), (10,3), (11,0), (7,5), (6,5), (9,1), (6,4), and (5,4), from CoMoCAT SG65i or SG76 SWCNTs (Southwest Nanotechnologies) using polymer ATP separation.^5,17 For the ATP separation process using single-stranded DNA,¹⁷ the sorted SWCNTs were precipitated from DNA solution by sodium thiocyanate (Sigma-Aldrich, 98%) and re-dispersed in 1 wt/v % sodium deoxycholate (DOC, Sigma-Aldrich, ≥97%). In other ATP experiments,⁵ we pressure filtrated the SWCNT-polymer mixture (Amicon, number 5123, using 100 kDa ULTRAcel regenerated cellulose filter membranes) with 1 wt/v % DOC to dilute the polymer concentration by a factor of 10³. Alternatively, gel chromatography was used to sort (7,6) + (8,4), (8,3) + (8,4), (8,3) + (7,3), and (6,5)-SWCNT enriched samples from HiPco SWCNTs (Rice University, batch #194.3) as described previously by Liu et al.¹⁸ The purified SWCNTs were then suspended in 1 wt/v % SDS (Sigma-Aldrich, ≥98.5%) in D₂O (Cambridge Isotope Laboratories, 99.8%) for subsequent functionalization.

Covalent Functionalization of Organic Color Centers

The optical densities of the SWCNT solutions were adjusted to 0.03–0.12 at the E₁₁ transition for subsequent optical studies. Monovalent aryl defects were introduced by diazonium chemistry.^1,11 Forsome aryl defects that cannot be incorporated using diazonium reactions, such as divalent aryl and aminoaryl defects, light activated arylation was used.^2,4 A series of alkyl defects were introduced using alkyl halides as reported by Kwon et al.² Covalent functionalization was monitored by defect PL evolution at E₂₂ excitation light, as described in the following section.

Spectroscopic Characterization

The SWCNT PL was characterized with a NanoLog spectrofluorometer (Horiba Jobin Yvon) using a liquid-N₂-cooled InGaAs array. The SWCNTs were excited with monochromator-selected light (10 nm slit width) from a 450 W xenon arc lamp. The excitation power was lower than 10 mW with an integration time of 1–10 s. The spectral resolution was 10 nm for the emission detection channel. UV-vis-NIR absorption spectra were obtained with a spectrophotometer equipped with a broadband InGaAs detector (Lambda 1050, PerkinElmer). The path length of absorption measurements was 10 mm.

HIGHLIGHTS.

Bonding functional groups to carbon semiconductors creates organic color centers

Organic color centers produce tunable near-infrared emission by as much as 400 meV

The electronic coupling between a color center and its host can be chemically tailored

The Bigger Picture.

Color centers, such as nitrogen vacancies in diamond, possess intriguing optical properties that have been intensively studied for imaging, sensing, lasing, and quantum information. However, most color centers occur as native defects, making tailored synthesis difficult. By covalently bonding functional groups to carbon nanotube semiconductors, we can synthetically create a whole new class of color centers that are organic and can be chemically tailored in highly predictable manners. These synthetic quantum emitters allow systematic tuning of the emission color in the near-infrared, a spectral regime that is important to biochemical imaging and quantum communications but had been largely unexplored. Unlike nitrogen vacancy centers in diamond, which is an electrical insulator, organic color centers are synthetically created in a semiconductor and electronically coupled to the host, opening the possibility for electrical addressing and chemical tailoring with molecular precision.