Programming sp³ Quantum Defects along Carbon Nanotubes with Halogenated DNA

Abstract

Atomic defect color centers in solid-state systems hold immense potential to advance various quantum technologies. However, fabricating high-quality, densely packed defects presents a significant challenge. Herein, we introduce a DNA-programable photochemical approach for creating organic color-center quantum defects on semiconducting single-walled carbon nanotubes (SWCNTs). Key to this precision defect chemistry is the strategic substitution of thymine with halogenated uracil in DNA strands that orderly wrap around the nanotube. Photochemical activation of the reactive uracil initiates the formation of sp³ defects along the nanotube as deep exciton traps, with a pronounced photoluminescence shift from the nanotube band gap emission (by 191 meV for (6,5)-SWCNTs). Furthermore, by altering the DNA spacers, we achieve systematic control over the defect placements along the nanotube. This method, bridging advanced molecular chemistry with quantum material science, marks a crucial step in crafting quantum defects for critical applications in quantum information science, imaging, and sensing.

Graphical Abstract

Quantum technologies hold the promise of transformative advances across diverse fields, including sensing, imaging, and information processing. Central to these innovations is the need for qubit arrays, in which each quantum bit, or qubit, can be individually addressed, consistently produced, and seamlessly integrated into large-scale structures.^1–2 Despite significant progress with ion traps,^3–4 superconducting qubits,^5–6 and entangled photons,⁷ it remains a challenge to pack qubits into dense arrays. A promising direction for this challenge is the exploration of organic color centers (OCCs)—sp³ quantum defects synthesized on the sp² carbon lattice of semiconducting single-wall carbon nanotubes (SWCNTs)^8–11—through advanced defect chemistry and DNA nanotechnologies.¹²

OCCs represent a promising frontier in quantum technologies due to their ability to create deep exciton traps. These molecularly tunable traps emit bright photoluminescence (PL) in the short-wave infrared encompassing telecom wavelengths.^{8, 13–14} Not only do these exciton traps emit single photons at room temperature,¹⁴ but they can also stabilize trions^15–16 and electron spin¹⁷ and support both optical and electrical pumping,^18–19 offering the potential for engineering photon qubits and efficient quantum interfacing. However, a significant challenge arises in precisely fabricating OCC arrays on SWCNTs. Although recent advances have enabled lithographic patterning²⁰ and atomic-level control of individual defects,^21–22 the majority of current methods—largely based on radical reactions^{8–11, 13, 21}—fall short in placing these defects into dense arrays.

Addressing this challenge requires the precise positioning of reactant molecules on predetermined SWCNT sites prior to reaction initiation—a task unattainable under conventional reaction conditions where reactants are free to move and react randomly in solution. In contrast, synthetic polymers and biopolymers, particularly DNA, have demonstrated their potential in forming ordered structures on SWCNTs, offering a pathway to achieving this positioning control.^23–26 Weisman,^27–29 Zheng,³⁰ and our team³¹ have previously explored using DNA to program defect arrays on SWCNTs. In a notable recent study, Zheng and co-workers utilized singlet oxygen to drive covalent bonding of guanine nucleobases to SWCNTs, achieving the synthesis of sp² defect pairs in ordered arrays.^29–30 However, these guanine-based defects only yield shallow exciton traps with spectrally broad emissions that significantly overlap with the nanotube E₁₁ emission, limiting their effectiveness for quantum applications.^{12, 27, 30}

In this work, we introduce a DNA-programmable photochemical approach of creating sp³ quantum defects along the length of SWCNTs. By incorporating halogenated uracil into DNA, we can photochemically trigger the halogenated nucleobase to achieve the synthesis of sp³ defects with deep exciton traps that function as OCCs. These defects exhibit bright photoluminescence with a significant spectral shift from that of the nanotube host, notably by 191 meV for (6,5)-SWCNTs. This defect emission is substantially more redshifted from the E₁₁ emission than those induced by guanine-induced sp² defects,^{28, 30} closely aligning with the shifts observed in sp³ defects produced by aryl diazonium⁸ and aryl halide chemistries^{13, 32} desirable for high purity single-photon generation with the added benefit of being spatially programmable with the DNA.

To program OCCs along a nanotube, we dispersed individual SWCNTs in water using single-stranded DNA (ssDNA) that had been incorporated with 5-iodo-uracil (5-I-dU), which is a halogenated and demethylated form of thymine (T). We chose (6,5)-SWCNTs that were wrapped with (5-I-dU)(GT)₁₅, hereafter referred to as (6,5)-SWCNT@(5-I-dU)(GT)₁₅, as our model system. While the fundamental nucleobases of DNA (G, A, T, and C) generally remain unreactive towards SWCNTs, the C-I bond in 5-I-dU can be cleaved under UV irradiation to produce uracil radical intermediates.^33–34 As we demonstrate in this work, these radicals, once formed, readily react with SWCNTs, leading to the creation of uracil OCCs (Figure 1A). Because uracil and thymine are structurally similar, we used (6,5)-SWCNTs wrapped with T(GT)₁₅ (labeled as (6,5)-SWCNT@T(GT)₁₅) as a control. Given this structural similarity, the 5-I-dU modified DNA strand wraps around the nanotube, largely retaining the ordered wrapping structure of the T(GT)₁₅ control, even after its halogenated uracil nucleobase covalently bonds to the nanotube to produce the uracil OCC-tailored nanotube, (6,5)-SWCNT-dU(GT)₁₅, as depicted by molecular modeling in Figure 1B.

Figure 1. Encoding the reactive sites into a DNA sequence for creating OCCs on SWCNTs.

(A) Schematic illustrating the use of engineered DNA sequences to program OCCs along a nanotube. The reactive sites (highlighted in red) are incorporated by replacing thymine (T) with halogenated uracil (5-I-dU), which is UV-cleavable to form a reactive radical with the nanotube. (B) Molecular models showing the retention of DNA ordered wrapping after covalent attachment. (C) Absorption spectra of the T(GT)₁₅ and (5-I-dU)(GT)₁₅ ssDNA-dispersed (6,5)-SWCNT samples. For clarity, the spectrum is offset. Inset shows the corresponding SWCNT solutions prior to a 33-fold dilution for the spectral measurements. (D) Cross-sectional view of the (6,5)-SWCNT@(5-I-dU)(GT)₁₅ before and after covalent bonding the dU nucleobase to the nanotube at the iodine leaving site. The bonding atoms and dU are highlighted by ball-sticks.

We first experimentally verified that the 5-I-dU modified DNA can effectively wrap around and disperse SWCNTs just like the unmodified DNA. To fabricate the samples, we sonicated a raw CoMoCAT SWCNT powder in a water solution containing (5-I-dU)(GT)₁₅ or T(GT)₁₅, and then centrifuged the solutions to remove nanotube bundles and metal catalysts (see details in the Supporting Information). Both the (5-I-dU)(GT)₁₅ and T(GT)₁₅ dispersed samples attained similar dispersion efficiencies for (6,5)-SWCNTs, as evidenced by their closely matched absorption spectra (Figure 1C). The absorption spectra of both samples present two pronounced peaks at 991 nm and 565 nm, corresponding to the E₁₁ and E₂₂ electronic transitions of the (6,5)-SWCNT, respectively, which is the predominant structure in the dispersions. Because the E₁₁ peak position is sensitive to the local dielectric environment, which can be affected by a slight change in the DNA morphology on the nanotube surface,³⁵ the identical E₁₁ peak positions indicate the 5-I-dU incorporation minimally perturbs the DNA wrapping structure on the nanotubes. This observation is further supported by AFM imaging that confirms the orderly wrapping of the 5-I-dU modified DNA on the SWCNTs (Figure S1).

Upon exposure to 254 nm UV light, a photochemical reaction is initiated between 5-I-dU and (6,5)-SWCNT, as illustrated by molecular models shown in Figure 1D. The successful introduction of OCCs is evident from the rise of the defect PL at 1,168 nm, which redshifts from the nanotube E₁₁ PL by 178 nm (191 meV) (Figure 2A). This observation is consistent with sp³ quantum defects created by diazonium chemistry.^8–9 Accompanying this new defect emission is an increase in the Raman D/G ratio from 0.053±0.002 to 0.157±0.004 after irradiation, which confirms the covalent nature of these defects (Figure 2B). We find that TEMPO, a known radical quencher, also deactivates the OCC creation (Figure S2). Additionally, when the (6,5)-SWCNT@T(GT)₁₅ sample was exposed to the UV light, there was no evidence of new defect PL (Figure S3A). These control experiments consistently suggest that the creation of the sp³ defects results from a radical-driven reaction between the 5-I-dU and the SWCNTs.

Figure 2. Photochemically triggered creation of OCCs on (5-I-dU)(GT)₁₅ ssDNA-wrapped (6,5)-SWCNTs.

(A) PL excitation-emission maps of the sample before (top) and after (bottom) exposure to 254 nm UV light. (B) Raman spectra before (red) and after 3-hours of UV-irradiation (blue), measured using 532 nm excitation. (C) Photoluminescence spectra within the first 3 hours of the UV irradiation, recorded at 565 nm excitation.

The defect PL intensity grows with increased dU sites (Figure 4B,C) and prolonged exposure to UV light, as shown in Figure 2C. The reaction, signified by rising defect emission over time, is considerably slower for 5-I-dU compared to free aryl halide molecules,^{13, 32} taking hours rather than minutes. This slowdown can be attributed to DNA-induced spatial constraints. Molecular modeling shows that uracil must rotate from a π-π stacking to a perpendicular orientation against the nanotube surface for covalent bonding, a process slowed by DNA wrapping (Figure S4). Further, bond length and angle analyses of the OCC supports the sp³ hybridization nature, aligning with what we observed experimentally (Figure S5).

Figure 4. Atomic force microscopy imaging of DNA-programmed quantum defects on individual nanotubes.

AFM images of (A) control: SWCNT@(5-I-dU)(GT)₁₅, and (B, C) (6,5)-SWCNT-dU(GT)₁₅ after removal of free DNA. The DNA is visible only in the SWCNT-dU(GT)₁₅ sample where the DNA strands are covalently bonded to the nanotube surface and cannot be removed by DOC. (D) Height profile along the length of the nanotube shown in (C), which is a zoomed-in view corresponding to the marked area in (B).

We then sought to control the spacing between the OCCs on the nanotube by using T(GT)₁₅ ssDNA as inert spacers. We first dispersed SWCNTs with both (5-I-dU)(GT)₁₅ and T(GT)₁₅ mixed at concentration ratios varying from 6:0 to 0:6, and subsequently initiated the OCC generation with UV light. We observed a decline in the defect PL intensity as the proportion of the DNA spacers increased (Figure 3A). When normalized to the E₁₁ value, the defect PL intensity scaled almost linearly with the (5-I-dU)(GT)₁₅ to T(GT)₁₅ ratio (Figure 3B). Corroborating this observation, single nanotube PL imaging shows that nanotubes with a high ratio of (5-I-dU)(GT)15 to T(GT)15 (6:0) displayed uniform OCC emission along their entire length (Figure 3C), although statistical analysis also revealed some heterogeneity (Figures S6 and S7). In contrast, nanotubes with a low, 1:5 ratio of (5-I-dU)(GT)15 to T(GT)15 exhibited sporadic OCC emission sites (Figure 3D), demonstrating effective OCC spacing control.

Figure 3. Controlling the defect spacing using inert DNA spacers.

(A) The PL emission-excitation maps and (B) PL spectra at 565 nm excitation of the DNA/SWCNT samples with different 5-I-dU(GT)₁₅:T(GT)₁₅ ratios after 1 hour reaction. The unmodified DNA sequence, T(GT)₁₅, serves as a spacer (depicted in gray in insets), while the modified sequence, 5-I-dU(GT)₁₅, is in red. (C, D) OCC PL images of individual SWCNTs reacted with 5-I-dU(GT)₁₅:T(GT)₁₅ at ratios of (C) 6:0 and (D) 1:5. Note these images display only OCCs, whereas the nanotube E₁₁ PL is filtered using a long pass filter.

We further conducted AFM imaging on heavily functionalized SWCNTs in the dry state to reveal the detailed spatial arrangement of DNA-programmed OCCs, which were identifiable at sites of covalent bonding along the nanotube length. After removing free and non-covalently bound DNA (Supporting Information), AFM shows that the (6,5)-SWCNT@(5-I-dU)(GT)₁₅ control exhibits a uniform height profile consistent with bare nanotubes, signifying the removal of DNA (Figure 4A). In contrast, heavily functionalized SWCNTs display bright spots indicative of covalently attached ssDNA (Figure 4B, C), with heights alternating between 1.8 and 1.2 nm corresponding to DNA-covered and exposed nanotube sections (Figure 4D). Notably, the observed minimum defect separation (10.2 nm) closely matches the DNA extension (~10.0 nm) derived from molecular modeling (Figure 1B).

The covalently bonded DNA exhibits markedly improved stability, as shown by surfactant displacement dynamics. Strong surfactants, like sodium deoxycholate (DOC), effectively remove non-covalently bound DNA from nanotubes.³⁶ As these DNA strands are displaced by surfactant molecules, the changing environment surrounding the nanotube induces spectral shifts, which we can experimentally detect using absorption and PL spectroscopy. For DOC-dispersed (6,5)-SWCNTs, the E₁₁ absorption peak is at 981 nm (Figure S8A), while for (6,5)-SWCNT@(5-I-dU)(GT)₁₅, it shifts to 991 nm (Figure S8B). Introducing 1 wt.% DOC to the latter solution causes the E₁₁ band to blue-shift from 991 nm to 981 nm within 1 minute, suggesting rapid DOC replacement of non-covalently bound DNA. However, after the covalent bonding, adding DOC only partially shifts the E₁₁ band to 986 nm (Figure S8C), highlighting the covalently bonded DNA’s resistance to surfactant displacement and thereby its improved stability on nanotubes.

In conclusion, we have developed a DNA-programmable photochemical method for creating 1D arrays of sp³ defect color centers in semiconducting SWCNTs by programing reactive uracil sites in DNA sequences that orderly wrap around the nanotube hosts. This uracil chemistry leads to the formation of deep exciton traps desirable for quantum applications, exhibiting substantial photoluminescence spectral shifts from the nanotube PL, notably by 191 meV in (6,5)-SWCNTs, compared to the 27 meV reported with sp² defects.³⁰ The versatility of DNA as a programmable biopolymer, combined with the potential to extend this strategy to other biomolecules, represents a key step in the chemical engineering of quantum defects. Our work lays the groundwork for crafting quantum defects for a range of applications—from quantum information processing, where the programmable deep exciton traps of OCCs are a promising system for achieving precision color center arrays,^{12, 14} to chemical sensing and bioimaging,^37–38 bolstered by the biocompatibility and stability conferred by the covalent DNA wrapping.