Optical Excitation of Carbon Nanotubes Drives Localized Diazonium Reactions.

Covalent chemistries have been widely used to modify carbon nanomaterials; however, they typically lack the precision and efficiency required to directly engineer their optical and electronic properties. Here, we show, for the first time, that visible light which is tuned into resonance with carbon nanotubes can be used to drive their functionalization by aryldiazonium salts. The optical excitation accelerates the reaction rate 154-fold (±13) and makes it possible to significantly improve the efficiency of covalent bonding to the sp(2) carbon lattice. Control experiments suggest that the reaction is dominated by a localized photothermal effect. This light-driven reaction paves the way for precise nanochemistry that can directly tailor carbon nanomaterials at the optical and electronic levels.

Organic reactions have been widely used to covalently modify carbon nanomaterials such as single-walled carbon nanotubes (SWCNTs) and graphene by covalently attaching organic functional groups that improve solubility and chemical compatibility.[1−3] These new functionalities are typically achieved at the expense of the remarkable electronic and optical properties of SWCNTs, as with most other covalent chemistries. Our recent experiments[4] demonstrate that meticulous control of diazonium reactions unexpectedly leads to the creation of bright, fluorescent quantum defects. This discovery has motivated a series of new experiments[5−9] that are unraveling exciting new opportunities to directly tailor the electronic and optical properties of low dimensional carbon materials through organic chemistry. It is worth noting that covalent modification of SWCNT with diazonium chemistry in the low defect density regime minimally disrupts their transport properties.[9] However, diazonium reactions, and covalent chemistries in general, lack the precision and efficiency required to directly tailor carbon nanomaterials at the optical and electronic levels. Here, we show that by optically exciting the SWCNT, functionalization of the nanostructure by aryldiazonium salts is accelerated and enhanced significantly. To our knowledge, covalent chemistries, particularly diazonium reactions, have never been achieved by photoexcitation that directly targets SWCNTs at the electronic and optical levels.

Functionalization of SWCNTs using aryldiazonium salts typically results in low grafting yield[4,10] due to their susceptibility to self-polymerization and rapid decomposition, hindering precise control of the functional degree particularly in the low defect density regime. Although diazonium chemistry can be accelerated electrochemically or using UV light that excites the salt,[1,11] these techniques may result in uncontrolled excessive functionalization of the material. The reaction can also be halted by removing the reactants,[6] but ultimately, the reaction is intrinsically slow, inefficient, and rather imprecise. Optical methods to trigger covalent bonding chemistries on carbon nanomaterials are rarely reported[12,13] and in the case of SWCNT, the light targets the bonding moiety[11] generating reactive species which impede control of the reaction. Our optically triggered reaction is driven by light that directly targets the SWCNT, rather than the diazonium. This results in a reaction that is 154 times faster and 5 times more efficient in functionalizing SWCNT. Furthermore, functional degree strongly depends on photon energy and the extent to which it resonates with chirality-specific optical transitions of the SWCNT, opening possibilities for developing nanochemistry that directly targets the nanostructure at the electronic and optical levels.

It is particularly challenging to control the degree of covalent functionalization in the low functional degree regime due to lack of sensitive techniques to monitor reactions. This challenge is addressed here by following the rise of the unique defect PL (E11–), which occurs at a wavelength red-shifted from the native exciton PL (E11) signaling the covalent bonding of aryl groups to the sp2 lattice of SWCNTs.[4] This allows the progress of the reaction to be monitored in situ using the defect PL as an ultrasensitive optical fingerprint. Furthermore, defect PL is highly tunable in terms of both its intensity and position based on the chemical nature and concentration of the defect moiety,[4] making it a quantitative indicator of the reaction progress.

Figure a outlines the light-driven diazonium reaction, where the covalent bonding of an aryl group to the SWCNT occurs by irradiation with light that is tuned into resonance with the nanotube. The optical absorption spectra of the isolated starting materials in Figure b reveal that the SWCNT can be excited to the exclusion of the aryldiazonium salt at either the second, E22 (565 nm; 2.19 eV) or first, E11 (980 nm; 1.27 eV) optical transitions of the nanotube. In this study, (6,5)-SWCNT is used as a model system, and our concept can be readily applied to other nanotube chiralities. The nanotubes are stabilized by sodium dodecyl sulfate as individual particles in D2O under dilute conditions so that inner-filter effects are negligible for PL measurements (see Experimental Methods in Supporting Information). The reaction is initiated by adding freshly prepared p-nitrobenzenediazonium tetrafluoroborate (NO2–BDT) solution to the suspension of SWCNTs and irradiating the solution with light. In our first experiment we use light that is resonant with the E22 optical transition of the (6,5)-SWCNT, specifically 565 nm, sourced from a monochromator-selected (10 nm bandpass) output from a 450 W xenon arc lamp. The reaction is monitored in situ by following the evolution of defect PL at E11– (1140 nm, 1.09 eV in the case of nitroaryl defects) over time, with care taken to minimize light exposure for the controls. PL excitation–emission maps in Figure a demonstrate that functionalization results in defect PL at E11– and that the irradiation results in enhanced functional degree versus the control, as indicated by brighter defect PL. We note that the observed defect PL originates from covalent nitrobenzene defects, not from physically adsorbed diazonium or photo-oxidation of the SWCNT, as confirmed in our previous chemical composition studies by Raman scattering and X-ray photoelectron spectroscopy.[4] The position of the defect PL in the case of nitroaryl defects is 19 meV red-shifted from that of O-doped (6,5)-SWCNT (1140 nm versus 1120 nm, ref[14]).

Figure 1

Resonant excitation of a SWCNT with light drives covalent functionalization on the sp2 carbon lattice by diazonium salts. (a) The schematic shows a region of a (6,5)-SWCNT where a p-nitrobenzenediazonium tetrafluoroborate (NO2–BDT) molecule is converted to a covalently bonded nitroaryl functional group upon irradiation with light that resonates with the chirality-specific excitonic transitions of the SWCNT. This chemistry creates a fluorescent quantum defect whose defect photoluminescence (PL) at E11– is used to monitor the reaction. (b) Absorption spectra reveal that (6,5)-SWCNT (black line) can be excited to the exclusion of the NO2–BDT (blue line) at either its E22 or E11 optical transitions (red shaded peaks). Asterisk indicates minor (6,4)-SWCNT content. The concentration of NO2–BDT is 40 μM in D2O.

Figure 2

Irradiating SWCNTs with light significantly accelerates their functionalization with diazonium salts. (a) PL excitation–emission maps of (6,5)-SWCNTs which are pristine (top), functionalized under protection from ambient light (middle), or functionalized under irradiation (bottom). (b) Functionalization is inefficient in the absence of irradiation. Filled circles denote region where functional degree is positively correlated to defect PL. In all cases, PL peak intensities are reported that were obtained after spectroscopic monitoring indicated that defect PL had stabilized (approximately 3 and 250 h in the case of samples irradiated and protected from light, respectively). Spectral evolution of E11 (black trace) and E11– (red, blue traces) corresponding to PL maps in (a), which were (c) irradiated and (d) protected from ambient light. The molar ratio of reactants, [NO2–BDT]:[SWCNT carbon] is 1:400 unless otherwise noted and 565 nm light was used to drive the reaction.

Notably, our light-driven reaction produces much brighter defect PL than the dark control over the range of relative reactant concentrations tested (Figure b). In the cases of both irradiated and dark (control) conditions, the intensity of defect PL maximizes at a reactant molar ratio, [NO2–BDT]:[SWCNT carbon], of 2.5 × 10–3 (1:400) before rapidly decreasing at higher ratios of reactants. Corresponding figures describing the evolution of the native E11 PL, PL excitation–emission maps, and optical absorption spectra of these samples are provided in Figures S1–S3. We note that the intensity of defect PL is quantitatively correlated to the functional degree for cases where the relative concentrations of reactants are less than or equal to the maximum defect PL intensity (region is emphasized by filled circles in Figure b). Because of this limitation, the remainder of the experiments discussed in this work are carried out so that defect PL intensity is correlated quantitatively to functional degree.

Kinetic studies reveal a significant acceleration of the reaction rate by light. The temporal evolutions of the native E11 PL and defect PL at E11– are shown in Figure c and d in the cases of irradiation and protection from light, respectively. A strong first order increase in covalent functional degree, indicated by an increase in defect PL at E11–, is observed regardless of whether the solution was irradiated or protected from light. However, the reaction rate and final intensity of defect PL differ substantially. The rate constants for the temporal evolution of defect PL intensity, when irradiated, klight, and protected from it, kdark, were determined to be (247 ± 14) × 10–2 h–1 and (1.6 ± 0.1) × 10–2 h–1, respectively. These values correspond to a 154-fold (±13) increase in the rate of reaction when the samples are irradiated with light that resonates with the nanostructure but is out of resonance with the diazonium salt. Furthermore, in the case of the irradiated sample, defect PL is more than three times brighter than when protected from light. This effect is even more pronounced if the brief exposures to light that were required for PL measurements of the control sample in this kinetic study are avoided. In that case, the light-driven reaction produces defect PL that is more than five times brighter than samples that were not exposed to light (Figure b). These observations suggest that light not only significantly accelerates the rate of functionalization, but the difference in the final intensity of defect PL suggests that light drives the PL brightening nanochemistry, converting reactants to covalently bound moieties efficiently.

In contrast, E11 PL is insensitive to covalent functionalization. It is rapidly quenched due to p-doping upon the addition of the diazonium resulting from noncovalent bonding of the diazonium (or an intermediate thereof) to the SWCNT.[15] E11 only partially recovers before stabilizing throughout the defect PL evolution. We presume that both noncovalent doping and our PL-inducing defect have similar effects on E11, making it apparently unresponsive to covalent functionalization.

In the absence of light, the conversion of diazonium to covalently bonded aryl groups is inefficient even with increasing relative concentrations of diazonium salt (Figure b). However, we find that we can drive covalent bond formation long after the reaction stabilizes in the absence of light. Upon irradiation of a solution of SWCNT and NO2–BDT, which was aged for 6 weeks and protected from light, there is a 2-fold increase in defect PL intensity (Figure S4). This suggests that nonreactive adsorbates may be activated optically to drive covalent bonding. Although we have not identified these intermediate species, this type of species may find interesting applications in nanostructure lithography because their reactivity may be turned on through optical excitation of their nanostructure hosts allowing for direct application of conventional photolithography tools to pattern chemical functionalities on carbon nanostructures. Ongoing studies are aimed at this long sought-after goal.

In general, reactions of carbon nanotubes and diazonium salts proceed through a radical intermediate and are complicated by environmental conditions including pH and the presence of surfactants.[10] The first critical step of the radical chain reaction, initiation, consists of reduction of the diazonium species into an aryl radical, which may proceed via two possible paths.[10] The first pathway is extraction of an electron from the SWCNT to generate the radical.[10] The second possible pathway occurs when the diazonium reacts with a base (here, the solvent) to generate a diazonium anhydride intermediate species.[10] This species can be thermally decomposed by homolytic cleavage to generate aryl radicals that can then attach to the SWCNT sidewall via electrophilic addition.[10,16] This second pathway is dominate at the neutral pH used here. We hypothesize that light can accelerate the initiation step by locally promoting decomposition of the adsorbed aryldiazonium or the intermediate to generate the aryl radical that can react with the SWCNT. To unravel the mechanism by which light accelerates the functionalization of SWCNT, we have investigated the dependence on the photon energy used to drive the chemistry, the power density, and the temperature of reaction.

Most interestingly, the functional degree of SWCNT varies strongly with the energy of photons used to drive the reaction as shown in Figures a (corresponding PL spectra are provided in Figure S5). In fact, this wavelength dependence closely traces the optical absorption spectrum of (6,5)-SWCNT, ranging from the ultraviolet to the near-infrared. The maxima of the reaction efficiency match well with the excitonic transitions of the particular SWCNT chirality. In addition, when irradiated with light that is completely out of resonance with the SWCNT, for example, 400, 456, and 790 nm, the reaction is nearly as inefficient as the dark control. These observations unambiguously confirm that the SWCNT itself is targeted by light to drive the chemistry, which to our knowledge has never been shown before.

Figure 3

Functionalization of (6,5)-SWCNTs is best promoted by SWCNT-resonant light. (a) Upon varying the photon energy of the incident radiation, the defect PL intensity relationship closely traces the absorption spectrum of unfunctionalized SWCNTs (gray trace), with the maxima matching those of the SWCNT’s E33, E22, and E11 optical transitions. Defect PL intensity of the thermally controlled reaction (dark) is indicated by the blue dashed line for reference. The molar ratio of reactants is 1:400 for these experiments. (b) Energy level diagram depicts the relative positions of the redox potentials of NO2–BDT,[10] the nitroaryl radical,[17] as well as the valence (VB) and conduction (CB) bands of the (6,5)-SWCNT.[18]

We present an energy diagram in Figure b from redox potentials of the reactants, including both the nitroaryl radical (+0.05 V/SCE[17]) and NO2–BDT (+0.45 V/SCE[10]) for reference to evaluate the possibility of an electron transfer mechanism. Given the relatively low oxidation potential of the (6,5)-SWCNT valence band (VB; +0.541 V/SCE[18]), the reactivity of the semiconducting (6,5)-SWCNT is poor toward diazonium. If photoinduced electron transfer were to occur, light would enhance the ability of the SWCNT to reduce the aryldiazonium cation to an aryl radical by raising its oxidation potential. We note that this is in contrast to the electrophilic addition of the aryl radical species onto the SWCNT[10] that directly results in covalent bonding.

We observe that E11-resonant light does enhance the reaction versus the dark control, but it does so less significantly than E22- and E33-resonant light (Figure S7). The relatively lower efficiency of the E11-resonant light is attributed to the inability for that energy to efficiently promote electrons to the conduction band for subsequent reduction of NO2–BDT. Furthermore, the fluorescence lifetime of the first state (E11) in (6,5)-SWCNT is 30 ps[19] or about 3000 times less than the fluorescence lifetime of small molecule PAHs such as pyrene and naphthalene.[20] Given the short lifetime at the excited state, it is expected that the photochemical reactivity of SWCNT would be quite low. Though we cannot rule out some degree of photoinduced electron transfer, the primary energy transfer mechanism appears to be thermal. The wavelength-dependent behavior of the chemistry suggests that the directed energy transfer is localized to individual SWCNT in the solution. When incident power density normalizes this behavior, the reaction does not show a stronger dependence on photon flux than simply irradiation energy (Figure S6). This suggests that the photoexcitation does not directly induce electron transfer from the SWCNT to the diazonium.

At a specific photon energy, the chemistry is highly dependent on the incident power density (Figure a), which agrees with our theory of a localized photothermal reaction. When the intensity of the irradiation (565 nm) increases from 0 (dark conditions), to 8.1, and finally to 11.5 mW cm–2 with other conditions unchanged, the functional degree, as indicated by the final intensity of defect PL, increases in a linear fashion. The contribution of light to the reaction efficiency reaches ∼82%, even at a relatively low power density of 11.5 mW cm–2. We expect that by using a more intense radiation source, the efficiency and rate of reaction may be further improved.

Figure 4

Contribution from light is substantial for PL-brightening nanochemistry on SWCNTs. (a) The significant contribution that light plays in driving the chemistry on SWCNTs is revealed by the positive relationship between defect PL intensity and irradiation power density. (b) Temperature dependence studies reveal a (51 ± 7)% decrease in apparent activation energy upon optical irradiation. Error bars in (b) represent the uncertainty in calculation of the rate constants, k. The molar ratio of reactants is 1:500 and 565 nm light was used for all experiments.

We further rule out the possibility that the photoeffect is due to a bulk thermal enhancement by studying the effect of temperature on reaction rate. Reactions were carried out at bulk solution temperatures ranging from 23 to 50 °C and monitored spectroscopically (Figure b). It should be noted that although we recognize that PL measurements of SWCNT vary with temperature,[7] we only monitor the change in defect PL intensity for our evaluation here. The temporal evolutions of defect PL at various temperatures fit well per first order reaction kinetics. The Arrhenius equation is applied to derive the apparent activation energies for the bulk thermal (protected from light) and localized thermal (irradiated) processes. When the chemistry is performed at various temperatures, irradiation of samples reveals a substantially different trend in kinetics than when protected from light. These trends represent a (51 ± 7)% decrease in the apparent activation energy upon irradiation of the mixtures of SWCNT and NO2–BDT. Though we cannot comment on the local change in activation energy at the SWCNT surface of this localized chemistry, the substantial difference in these values, 57 ± 7 kJ mol–1, represents the major contribution of light to the functionalization of SWCNTs at low defect densities and confirms that the chemistry occurs locally.

In summary, we have shown that functionalization of semiconducting SWCNT using diazonium salts can be significantly accelerated, by 154-fold (±13), through optical excitation of the carbon nanotubes. This light-driven reaction leads to over five times more efficient functionalization, as indicated by the evolution of defect PL. The chemistry’s strong wavelength dependence suggests that the reaction progresses through directed energy transfer from the optically excited SWCNT to the adsorbed diazonium or intermediate. Though it does not appear that the primary medium of energy transfer is electronic, it is abundantly clear that the chemistry is driven by local heating, as opposed to bulk thermal means. With this light-driven chemistry that directly targets the SWCNT it becomes possible to tailor the optical and electronic properties of these materials through fast and highly controllable covalent modification. Our method may also allow for chemical targeting of a specific chirality within a SWCNT mixture, a long sought after goal,[2] and optical lithography of low dimensional carbon materials.