(n,m) Assignments of Metallic Single-Walled Carbon Nanotubes by

Nov 16, 2016 - Therefore, hereafter we mainly use accurate Mii's as well as the Mii splitting from ωERS to make (n,m) assignments, with some aids of ...
1 downloads 10 Views 1MB Size
(n,m) Assignments of Metallic Single-Walled Carbon Nanotubes by Raman Spectroscopy: The Importance of Electronic Raman Scattering Daqi Zhang, Juan Yang,* Meihui Li, and Yan Li* Beijing National Laboratory for Molecular Sciences, Key Laboratory for the Physics and Chemistry of Nanodevices, State Key Laboratory of Rare Earth Materials Chemistry and Applications, College of Chemistry and Molecular Engineering, Peking University, Beijing 100871, China S Supporting Information *

ABSTRACT: In this work, we report an accurate and convenient method that can be used to assign the chirality of all metallic single-walled carbon nanotubes (M-SWNTs). This method is designed based on the electronic Raman scattering (ERS) features, which are resonantly enhanced at the corresponding excitonic transition energies (Mii+ and Mii−). Using this method, we are able to accurately determine the electronic property Mii with the resolution of a vibrational Raman spectroscopy (∼0.3 meV), which is significantly higher than that of the electronic spectroscopies (∼3 meV). We use the Mii splitting value, which is found insensitive to environmental changes, as a universal criteria for (n,m) assignments in various environments. As an illustrative example, simply using a commercialized Raman spectrometer with two laser lines (1.959 and 2.330 eV), we are able to unambiguously assign 18 metallic chiralities with M11 in the 1.6−2.3 eV range in our samples. This method provides an accurate database of Mii’s in a similar way as photoluminescence excitation spectroscopy does for Sii’s. It can facilitate further systematic studies on the properties of MSWNTs with defined chirality. KEYWORDS: single-walled carbon nanotubes, electronic Raman scattering, resonant Raman spectroscopy, chirality assignments, Mii splitting

T

information for (n,m) assignments, such as dt from the Raman shift of the radial breathing mode (RBM).31−37 However, obtaining Eii from the resonance Raman excitation profile requires an expensive and labor-intensive tunable laser system.38−41 A compromise is to use a commercial microRaman spectrometer equipped with several discrete laser lines (EL) for (n,m) assignments by combining the RBM position (ωRBM) and the so-called resonance window.35 A resonance condition of |EL − Eii| ≤ 100 meV is empirically assumed when the RBM feature can be observed experimentally.17 This large uncertainty in Eii will lead to ambiguous assignments. On the other hand, the methods for (n,m) assignments of metallic (M)-SWNTs are less than that of semiconducting (S)SWNTs since the PL spectroscopy is not applicable for metallic tubes. Recently, a feature termed the electronic Raman scattering (ERS) is reported by Farhat et al. and found to

he chiral index (n,m) determines the geometric structure (diameter dt and chiral angle θ) and electronic structure of a single-walled carbon nanotube (SWNT). Therefore, accurate and reliable (n,m) assignments are important in nanotube-related science and technology, such as chirality-controlled SWNT growth,1−4 fundamental comprehension of chirality-dependent properties,5−8 as well as applications of SWNTs in nanoelectronics9−12 and bioimaging.13−16 Spectroscopic methods are developed and widely used for (n,m) assignments due to their capability in obtaining the chirality-dependent excitonic transition energies (Eii).17,18 The optically allowed transitions are resonantly enhanced at the corresponding Eii’s of a particular SWNT and can thus be detected by spectroscopic instruments. The commonly used spectroscopic methods in SWNT characterization include absorption,19−21 photoluminescence (PL),22−25 Rayleigh,5,26,27 and Raman scattering28−30 spectroscopies. Compared to other spectroscopic methods that can only provide information on Eii, Raman spectroscopy is a versatile method capable of giving additional structural © 2016 American Chemical Society

Received: July 5, 2016 Accepted: November 16, 2016 Published: November 16, 2016 10789

DOI: 10.1021/acsnano.6b04453 ACS Nano 2016, 10, 10789−10797

Article

www.acsnano.org

Article

ACS Nano appear exclusively in the Raman spectra of M-SWNTs.42 The ERS is a resonant enhancement process that occurs when the energy of photons inelastically scattered by the low-lying electron-hole pairs near the Fermi level matches the corresponding excitonic transition energy (Mii) of the MSWNTs.42 Later, we report the simultaneous observation of both ERS+ and ERS− features at the corresponding Mii+ and Mii− split by the trigonal warping effect in nonarmchair MSWNTs.43 For armchairs, one ERS at the corresponding Mii is theoretically expected.44 Due to the resonant nature of the ERS, we are able to determine multiple Mii’s by fitting the ERS bands. The resulting uncertainty in Mii can be in the order of ±1 meV for a single measurement and still in the order of ±10 meV when the environmental effects are taken into account. This uncertainty is significantly better than the empirical ±100 meV uncertainty from the RBM resonance conditions. Based on the accurate Mii’s from the ERS features, in principle, unambiguous and accurate (n,m) assignments for all M-SWNTs can be realized. Here, we observe the ERS features in both armchair and nonarmchair M-SWNTs. We show how unambiguous (n,m) assignments of M-SWNTs can be made based on accurate Mii’s from the ERS positions (ωERS), with some aids of dt from ωRBM. The experimental uncertainties in both Mii and ωRBM are discussed in detail with the statistical data of the near-armchair chirality (12,9). A reference Kataura plot as well as an ωRBM−dt relation for suspended M-SWNTs is also provided.

Figure 1. (a) Raman spectra of two Y-shaped bundles of an individual (12,9) bundling to other tube(s) at 1.959 eV excitation. Spectral intensities are normalized with respect to the 2D band. Blue dashed lines show the Lorentzian fitting of the ERS bands. (b) Statistical data of the M11+ and M11− values obtained from the ERS bands of 11 different (12,9) samples. Uncertainties in ωRBM are marked by error bars. (c) Statistical data of the M11 splitting of 11 different (12,9) samples. The error bars in M11 splitting are introduced by spectral resolution, random noise, and error in spectral fitting. Blue dashed line shows the average M11 splitting value of 66 meV for (12,9).

Rayleigh scattering26,27 and transmission spectroscopies,52 to be determined directly from the vibrational Raman spectroscopy with much higher resolution. Therefore, the Mii’s can be determined much more accurately. Compared to Rayleigh scattering5,53 and transmission spectroscopies52 as well as resonance Raman excitation profile38,54,55 with a resolution ranging from 3 to 10 meV and a bandwidth of ∼100 meV, our Raman measurements provide a spectral resolution of ∼0.3 meV (∼2 cm−1) and an ERS bandwidth of ∼50 meV with a commonly used 600 grooves/mm grating. Therefore, the ERS features in Raman spectra allow Mii’s to be accurately determined down to meV. Taking the Y-shaped bundle 2 as an example, we estimate that the error in determining M11 at the individual part is ±0.7 meV, considering the spectral resolution of 0.3 meV, the error introduced by random noise of 0.2 meV, and the spectral fitting error of 0.9 meV. The spectral fitting is carried out by several independent runs of fitting with different initial parameters, and the details are provided in the Supporting Information. We estimate that the corresponding error at the bundled part is ±1 meV with a larger spectral fitting error of 1.5 meV due to the decreased ERS intensity. Owing to the easy identification of (12,9) by its characteristic double-hump ERS bands, we observe with confidence altogether 11 different samples for this particular chirality at room temperature. From these 11 statistical data (Figure 1b), we obtain 1.866 ± 0.013 eV with a range of 36 meV for M11+ and 1.800 ± 0.010 eV with a range of 26 meV for M11−. Heating (12,9) at different laser powers (Figure S2) shows 4 meV variation in ωERS, indicating negligible influence of Mii with respect to moderate temperature changes. In Figures 1b and S2, we observe similar trends in M11+ and M11− in various environments, implying a constant M11 splitting value (M11+ − M11−) for a particular (n,m), insensitive to the environmental changes. Figure 1c plots the M11 splitting of the 11 different (12,9) samples with respect to their ωRBM, and a

RESULTS AND DISCUSSION It is well-known that the local environments of SWNTs, such as amorphous carbon coating,45 adsorbed gas molecules (e.g., O2 and H2O),46,47 bundling with other nanotubes,43,48 and temperature variations,43,49,50 will affect both Eii and ωRBM. In order to make accurate (n,m) assignments, we first focus on obtaining statistical data to show how accurately we can determine Eii and ωRBM for the very same chirality in various environments. Here, we intentionally choose the near-armchair chirality (12,9) for this study because of its easy identification by the characteristic double-hump ERS bands at 1.959 eV excitation, given that M11− = 1.78 eV and M11+ = 1.84 eV from ref 5. In this work, most of our samples are suspended SWNTs across open slits grown by chemical vapor deposition (CVD) method. Both individual and bundled tubes as well as Y-shaped bundles are available. Figure 1a shows the Raman spectra of two Y-shaped bundles, labeled as 1 and 2, of an individual (12,9) bundling to other tube(s). Spectra 1i/2i are acquired at the individual part and 1b/2b at the bundled part. By fitting the spectra, two ERS bands (blue dashed lines) can be resolved for both parts. At the bundled part, additional RBM and 2D bands from the other tube(s) appear and additional G bands overlap with that of (12,9). For (12,9), ωRBM upshifts slightly (