Over the past few years, semiconducting single-walled carbon nanotubes (SWCNTs) have emerged as a potential room-temperature single photon source (SPS) - making them a viable candidate as an emitter with photonic chip integration capabilities for quantum information science applications (QIS) in the (near-IR) telecommunications band [1-3]. Specifically, SWCNTs with the semiconducting (6,5) chirality functionalized with sp3 chemical defects have shown to create localized exciton traps. Additionally, NTs exposed to the ambient environment including oxygen and other surface adsorbates can ultimately lead to a change in the electronic structure and create localized exciton traps along the tube [4-6]. The Krauss group has been studying how surface charge from the heterogeneously surfactant (sodium cholate) wrapped NTs (roughly 3-5 um long with 0.78 nm diameter) affect its photophysics and may lead to localized exciton traps deep enough (tens of meV) for single photon emission at room temperature. Our group is interested in exploring the feasibility of using SWCNTs as SPS at ambient conditions by measuring their single photon purity and indistinguishability - metrics ultimately characterizing their effectiveness for any quantum information/computing applications. Correlative studies that show how topological mechanical defects (bends/kinks/strain) and aggregates of surfactant charge in the environment lead to localized excitons/single emitter behavior through confocal PL microscopy will provide key insights for determining the effectiveness of SWCNTs for QIS. To study this robustly, a substrate compatible with correlative AFM/EFM/STM, PL microscopy of individual tubes, and other quantum optics experiments across various physical setups is needed. In the following sections, I will describe the process for a patterned substrate and how various NTs samples were characterized under the SEM, TEM, and AFM. Although correlative optical measurements with the samples were not performed as of yet, an interesting discrepancy in the nanotube size distribution was observed between the SEM and AFM measurements.

2. Sample Preparation

2.1 Patterned Substrate

For future measurements of individual SWCNTs and their respective optical characteristics, a gold pattern shadow mask was deposited onto an indium tin oxide (ITO) coated quartz microscope coverslip. The nanotubes would then be deposited on top of this coverslip either via drop-casting or spin-coating. The goal of this patterned substrate was to allow for correlation between specific individual tubes as measured using atomic force microscopy (AFM) and its corresponding confocal photoluminescence spectrum with an inverted microscope. The fixed gold grid pattern on the substrate would provide a coordinate system or a reference to take AFM measurements on one setup and take optical measurements of the same spot on a different setup. For inverted microscopy, we would investigate the shadow regions in between gold squares where the bare ITO would provide transparency in the visible and near-IR range. Conductive ITO could also enable future correlative electron force microscopy (EFM) and scanning tunneling microscopy (STM) measurements. The substrate was patterned by placing a bare copper mesh TEM grid onto the coverslip as a mask and gold was sputtered with approximate thickness of 20 nm. Initially, the TEM grids had a carbon-formvar support film which was dissolved away providing transparency for the gold to deposit through. Figure 1 below shows both the light microscope (bright-field) and SEM secondary electron image of the patterned substrate with dimensions. The brighter regions correspond to 20 nm tall square gold islands while the surrounding darker rectangular channels are the bare ITO regions. The overall grid pattern is roughly circular with 2 mm in diameter. We note that the right regions of the image appear to be out of focus - this is due to the pattern being diffuse as the thin TEM mask grid was not perfectly flat during gold deposition.

Figure 1 - Gold deposited (20 nm) onto ITO quartz coverslip using a TEM grid shadow mask. (left) Bright-field light microscope image (right) SEM micrograph

2.2 SEM Samples

A SWCNTs sample deposited onto the patterned ITO substrate described above and a carbon-support film copper TEM grid substrate were used to image the nanotube surface morphology via the SEM. Initially, 15 uL of concentrated (6,5) chirality (semiconducting) SWCNT solution with ionic sodium cholate surfactant was spin-coated onto the ITO coverslip at 3000 rpm for 60 seconds. The nanotube solution is shear-force mixed with surfactant and water to solubilize individual tubes and this will likely add topological defects in the tube structure due to violent forces present during mixing. Initially, tubes were not observed under the SEM perhaps due to low concentration or the tubes not sticking to the surface under the first spin coating. So, another round of spin coating was done and few uL of solution was drop-casted onto the grided pattern and left to dry overnight to achieve a high concentration of tubes on the surface for imaging. For the carbon support film TEM grid sample, few uL of solution was similarly drop-casted and air-dried overnight. Both SEM samples were left uncoated (in gold) but were fixed onto the SEM sample holder stub with small amount of liquid graphene to provide electrical connectivity for the ITO and the copper grid.

2.3 TEM Samples

Similar to SEM samples, few uL of nanotube solution was drop-casted onto a TEM grid with lacey carbon support film and left to dry overnight. However, no tubes were imaged with this sample as the contrast from the tubes was likely on the order of the signals generated by the support film. Instead, a bare TEM grid was used with no support film to deposit the sample similarly. The goal was to image the nanotubes hanging off the edge of the wired grid to provide maximal contrast.

3. Electron Flight Simulations

The Electron Flight Simulator (EFS) tool was used to model the electron-beam interaction volume of the nanotube samples for the SEM. Specifically, the geometry of the model included two thin layers of 5 nm of carbon followed by 12 nm of ITO on top of the quartz substrate tilted at 30 degrees relatie to the electron beam. Three levels of accelerating voltages (1, 5, 10 kV) were simulated using 16,000 electron trajectories. The 5 nm of carbon layer was used simulate the approximate height of the tube. As seen from Fig 1, the lateral and depth of the interaction area increase significantly as the beam accelerating voltage increases. The surface interaction width is about 24 nm for the 1 kV and gradually tapers down into the ITO layer. However, for the 5 and 10 kv, the interaction volume is largely spherical and most of the SE signal will arise from the quartz substrate - potentially making tubes more difficult to see. These results suggest imaging at lower accelerating voltages to avoid noisy signal from the underlying substrate which comes at the expense of not being able to focus the beam well as compared to higher voltages. Ultimately as observed in the below section, 5 kV was optimal choice for the ITO sample and 0.5 - 0.6 kV was used for the carbon film sample.

Figure 2 - Electron interaction volume simulation for ITO on quartz sample with varying beam accelerating voltages (left) 1 kV (center) 5 kV (right) 10 kV

4. SEM Micrographs

4.1 Gold-Patterned ITO Quartz Sample

Figure 3 below shows some micrographs of the individual and bundled nanotube structure on the bare ITO coated substrate - as this is the transparent region accessible for inverted confocal microscopy. For all samples, raster-contamination squares were an issue due to the electron beam interactions with the hydrocarbons from the surfactant. Therefore, only a 10 um aperture was used with 5 kV voltage providing adequate resolution without significant sample degradation. Both long (5+ um) bundled nanotubes and shorter tubes were observed on top of the grainy ITO metal coating. The sample was tilted at about 30 degrees to provide topographic relief using the In-Lens secondary electron detector at a short working distance to maximize SNR. Small clumps of surfactant material surround the nanotube environment and clear bends/kinks/splitting can be seen on various tubes which maybe sites of topological defects in the graphene structure and of high-interest to characterize optically.

Figure 3 - SEM micrographs of NTs deposited (drop-casted) onto ITO quartz. Both long bundled and short surfactant wrapped tubes are observed with various morphologies

4.2 Carbon-Support Film Sample

Figure 4 below shows some micrographs from the carbon support film sample and the resolution and SNR are clearly degraded compared to the ITO sample. Significant sample degradation was observed at even 1 kV and the tubes appeared to be transparent. Lowering the beam voltage to 0.5-0.6 kV reduced any raster scan artifacts but the beam was very difficult to focus at such a low voltage. This sample was also tilted at about 30 degrees relative to the beam to provide some topographic perspective. Compared to the ITO sample micrographs, we see contrast inversion as the tubes appear bright with dark background - perhaps indicating a smaller interaction volume with most of the signal arising from the tube surface. Figure on the right shows a crack in the surfactant and/or carbon film with a tube appearing to get stretched over the trench. This maybe a single tube or perhaps the surfactant on the tube getting extended as the sample was drying. Nominal individual tube diameters are expected to be around 3-5 um for surfactant covered tubes, however these SEM micrographs clearly show larger sizes - which will be analyzed in the AFM section.

Figure 4 - SEM micrographs of NTs deposited (drop-casted) onto carbon support film (TEM grid) taken at low 0.5 kV.

5. TEM & EDS

For the bare copper TEM grid sample, it was apparent that there were small islands of surfactant film adhering to the edges of the wired mesh which provided some images of the nanotubes as seen in Figure 5. In the top row, we see long diffraction patterns from the sidewalls of the bundled tubes as measured about 12 nm in diameter compared to the nominal 0.78 nm diameter for a single-walled (6,5) chirality tube. In the bottom row, we observed some interesting tube surface morphology which would be great to correlate with optical measurements - a sharp bend in the tube measuring about 1.72 nm and two-bundled tubes splitting into two branches. Although these tubes are larger than the 0.78 nm, some of these may still be single walled tubes covered in surfactant - making them appear larger. By under-focusing slightly, we can observe more contrast as seen on the image in upper right - the darker regions may be areas of heavier surfactant concentration causing greater energy loss of the electron beam. To confirm that indeed the images were carbon nanotubes, energy dispersive X-ray spectroscopy (EDS) analysis was done on a small portion of the bundled tube as highlighted by Figure 6 (left). The sample holder was also tilted towards the X-ray detector to provide greater X-ray counts. As seen on the spectrum on the right, a large carbon peak was observed along with some sodium from the surfactant. Platinum, cobalt, and iron were also observed which may be leftover catalysts from the nanotube synthesis process. The copper signal may also be arising from the nearby copper mesh grid. In the future, ultra-washed tubes containing no surfactant can be imaged to confirm the single walled nature and 0.78 nm diameter.

Figure 5 - TEM images of NTs deposited onto bare Cu TEM grid. Larger bundled tubes are seen with surfactant as well as smaller tubes (~2nm in diameter)

Figure 6 - (left) Focused electron beam spot on the tube for EDS measurements (right). Large signal for carbon along with surfactant and catalysts is confirmed.

6. AFM & Height Analysis

After observing the patterned ITO sample with deposited tubes on the SEM, AFM was used to accurately determine the tube diameters by measuring their heights. AFM was operated in tapping or semi-contact mode with approximate scan size of 8 um x 8 um as seen on Figure 7. Although, the regions observed by the SEM and AFM do not exactly correlate - bundle like structures were also observed similar to SEM with the AFM topographic images. In total, four (8 um x 8 um) images were taken and heights of 40 individual tubes were measured by taking linecuts approximately perpendicular to the tube long axis as seen on Figure X (right). The small background grainy structure on the order of few nanometers is likely the ITO metal coating, while the larger and taller spherical-like shapes are surfactant clumps. We note that the lateral AFM resolution is limited to around 25-30 nm which is limited by the tip radius and stage controls.

Figure 7 - (left) Topographic AFM scan in tapping mode of the ITO sample over a 8x8 um square region. (right) Height profile along blue line on left image indicates tube height of 7-8 nm.

Figure 8 below summarizes the tube diameter distributions measured by both SEM for two samples (50 tubes for the ITO sample & 45 tubes for the carbon film) and from the AFM as described above. For the SEM samples, micrographs were imported/calibrated, and tube widths were manually measured using the line tool in ImageJ. We see a clear separation in the size distribution as the AFM measures on average about 6.7 nm diameter while the SEM micrographs indicate tube diameters of about 23 and 53 nm. It is important to note that for the carbon film sample, micrographs were only taken at lower 0.5 to 0.6 kV as contamination from the beam interacting with surface and charging was readily observed - which could further make tubes appear larger. The ITO sample micrographs were taken at 5 kV which has a narrower beam spot size - potentially improving resolution to provide more accurate measurements.

Figure 8 - NT diameter distribution across the SEM and AFM measurements. SEM measures larger apparent tube diameters possibly due to limited resolution and beam interactions with the sample.
AFM is widely accepted as the more accurate tool for measuring small nanotube diameters.

Although these measurements are not correlative (i.e. not of the same tubes), I believe the sample size is great enough to indicate that characterizing small diameter SWCNTs under SEM may significantly overestimate tube diameter compared to the gold-standard AFM. We note that typical surfactant tubes that are just spin-coated measure about 3-5 nm, so our average of 6.7 nm with AFM is in reasonable agreement considering more bundling or surfactant buildup may have occurred from drop-casting. Other groups have similarly measured larger apparent pristine tube (as grown on substrate) diameters under the SEM compared to AFM at varying operating conditions and substrate type [7]. The electron beam interacting with local electric fields from the tube may also be responsible for the diameters to appear larger [8]. Regardless, SEM can still be used as a powerful tool to characterize the overall areal density of the tubes on the patterned substrate for future correlative studies of individual SWCNTs. Near-term measurements will involve direct surface correlation from AFM of NTs using the patterned substrate with photoluminescence spectroscopy.

Acknowledgments

I would like to thank the Krauss group (Team Tubes!) for providing the nanotube samples and assistance with sample preparations using the spin coater. I would also like to greatly thank Brian McIntyre for his mentorship with all the instrument training (i.e. SEM/TEM/AFM/sputter coater) as well as teaching the fundamentals of electron microscopy.

References

1. He, X., et al. "Carbon Nanotubes as Emerging Quantum-Light Sources." Nature Materials, vol. 17, no. 8, 2018, pp. 663-670., doi:10.1038/s41563-018-0109-2. 2. Ma, Xuedan, et al. "Room-Temperature Single-Photon Generation from Solitary Dopants of Carbon Nanotubes." Nature Nanotechnology, vol. 10, no. 8, 2015, pp. 671-675., doi:10.1038/nnano.2015.136. 3. "Pure and Efficient Single-Photon Sources by Shortening and Functionalizing Air-Suspended Carbon Nanotubes." doi:10.1021/acsanm.9b02209.s001. 4. He, Xiaowei, et al. "Tunable Room-Temperature Single-Photon Emission at Telecom Wavelengths from sp3 Defects in Carbon Nanotubes." 2017 International Conference on Optical MEMS and Nanophotonics (OMN), 2017, doi:10.1109/omn.2017.8051442. 5. Lin, Ching-Wei, et al. "Creating Fluorescent Quantum Defects in Carbon Nanotubes Using Hypochlorite and Light." Nature Communications, vol. 10, no. 1, 2019, doi:10.1038/s41467-019-10917-3. 6. Hofmann, Matthias S., et al. "Ubiquity of Exciton Localization in Cryogenic Carbon Nanotubes." Nano Letters, vol. 16, no. 5, 2016, pp. 2958-2962., doi:10.1021/acs.nanolett.5b04901. 7. Li, Dongqi, et al. "Scanning Electron Microscopy Imaging of Single-Walled Carbon Nanotubes on Substrates." Nano Research, vol. 10, no. 5, 2017, pp. 1804-1818., doi:10.1007/s12274-017-1505-7. 8. Brintlinger, T., et al. "Rapid Imaging of Nanotubes on Insulating Substrates." Applied Physics Letters, vol. 81, no. 13, 2002, pp. 2454-2456., doi:10.1063/1.1509113.

Surface Characterization of Carbon Nanotubes

Mitesh Amin