Ana
Santidrián
*ab,
José M.
González-Domínguez
bc,
Valentin
Diez-Cabanes
d,
Javier
Hernández-Ferrer
b,
Wolfgang K.
Maser
b,
Ana M.
Benito
b,
Alejandro
Anśon-Casaos
b,
Jérôme
Cornil
d,
Tatiana
Da Ros
d and
Martin
Kalbáč
*a
aJ. Heyrovský Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Dolejskova 3, 18223 Prague 8, Czech Republic. E-mail: martin.kalbac@jh-inst.cas.cz
bInstituto de Carboquímica ICB-CSIC, Miguel Luesma Castan 4, 50018 Zaragoza, Spain. E-mail: asantidrian@icb.csic.es
cINSTM Unit of Trieste, Department of Chemical and Pharmaceutical Sciences, University of Trieste, Via Licio Giorgieri 1, 34127 Trieste, Italy
dUniversity of Mons, Place du Parc 20, 7000 Mons, Belgium
First published on 23rd January 2019
The effect of doping on the electronic properties in bulk single-walled carbon nanotube (SWCNT) samples is studied for the first time using a new in situ Raman spectroelectrochemical method, and further verified by DFT calculations and photoresponse. We use p-/n-doped SWCNTs prepared by diazonium reactions as a versatile chemical strategy to control the SWCNT behavior. The measured and calculated data testify an acceptor effect of 4-aminobenzenesulfonic acid (p-doping), and a donor effect (n-doping) in the case of benzyl alcohol. In addition, pristine and covalently functionalized SWCNTs were used for the preparation of photoactive film electrodes. The photocathodic current in the photoelectrochemical cell is consistently modulated by the doping group. These results validate the in situ Raman spectroelectrochemistry as a unique tool box for predicting the electronic properties of functionalized SWCNTs in the form of thin films and their operational functionality in thin film devices for future optoelectronic applications.
To fully exploit the potential of SWCNTs, physicochemical methods have been developed to process pristine SWCNTs and adjust their optoelectronic properties. Among them, covalent functionalization is one of the most efficient strategies. In this way, SWCNTs have shown enough chemical versatility to undergo reactions not only with electron-deficient species, such as radicals, carbene or nitrene compounds, but also with electron-rich reagents such as alkali metals.6–8 Most of the chemical functionalization procedures are sensitive to the electronic properties of the SWCNTs and can also lead to significant changes in their electronic structure. Previous works have discussed how SWCNTs can be selectively doped through functionalization, in which electrons or positive holes are introduced into their structure by interaction with either electron donor or acceptor groups.9–12
If there were one particular kind of covalent chemistry applied to SWCNTs as the most trusted methodology to achieve chemical doping, it would undoubtedly be the diazonium-based reactions.9,13 Briefly, these consist of the thermally or electrochemically induced reaction of a diazo compound that evolves into an aryl radical capable of creating a cascade of radical reactions between SWCNTs and aryl moieties, ending up in stable C–C bonds between both units.14 This scheme has been widely employed not only for altering the SWCNT electronic structure but also for grafting functional moieties with multiple anchoring possibilities.15 More recently, the diazonium chemistry of SWCNTs has been used for the creation of bright, fluorescent quantum defects. Changes in SWCNT photoluminescence through covalent sidewall functionalization have proven enormously versatile. Some studies have reported that several photoluminescence features (such as emission energy and maximum brightness) can be chemically tuned using withdrawing/donating substituents on the aryl functional group.16,17 It has even been shown that SWCNTs can be photoexcited to induce an acceleration in diazonium functionalization.18
Even if the intended alteration of the SWCNTs’ electronic structure is a fact, by means of diazonium chemistry or any other approach, on many occasions there is a need to unravel the nature of doping (negative, n- or positive, p-) and its extension. Transport measurements19,20 or thermoelectric power measurements,21,22 that require a sophisticated system, can be performed to investigate the Fermi level changes. However, these techniques are not usually adapted to distinguish between n- or p-doping. Optical techniques, such as near-infrared fluorescence and photoluminescence excitation, albeit very powerful, are exclusively limited to individual SWCNTs, which is often unviable in many samples and procedures.23,24
Raman spectroscopy is much more accessible and powerful as it is a non-destructive, contactless and quick technique that requires relatively simple or no preparation, and is greatly sensitive to changes in the physical and chemical properties of SWCNTs.25 To characterize doping, Raman spectroscopy primarily relies on G-band shifts.26 However, these changes in the G-band are sometimes masked by other effects, such as strain. In addition, the G-band frequency is only weakly dependent on the tubes’ diameter, and therefore, in nanotube bundles one cannot extract shifts for individual tubes. Despite a few successful attempts reported in the literature,27–29 an assignment of the up/down shift to a given doping process is still not straightforward. Therefore, alternative methods to determine the qualitative differentiation between n- and p-doping induced by functionalization would be beneficial. Ideally, the method should be suitable to tailor the electronic effects upon functionalization by various functional groups. A prospective method to address this task is in situ Raman spectroelectrochemistry, as it allows changes in the Raman spectra of SWCNTs during doping to be followed in a wide potential range in a precise and reproducible way.30–33
In the present work, we refer to a previously described functionalization scheme, based on diazonium chemistry. We controlled the experimental variables to end up in a high level of sidewall decoration with one or several functional groups, which can be further derivatized, and minimizing the oligomerization of reactive species.34 This approach was adapted to obtain functionalized SWCNTs with two different grafted moieties, achieving the dual effect of high functionalization level and the tuning of their electronic structure. In situ Raman spectroelectrochemistry is used to investigate the radial breathing mode (RBM) and to assess the electronic structure of covalently functionalized SWCNT bundles. The proposed methodology allows unambiguous distinguishing between chemical p- and n-doping. The approach has been tested in bulk samples of SWCNTs, which represent the majority of applicable cases. The experimental results were supported by density functional theory (DFT) calculations. Furthermore, changes in the photoresponse of SWCNT films were recorded, as a proof of principle, confirming the effect of n-/p-doping when a specific functional group is attached to the SWCNTs.
The ultraviolet/visible (UV/vis) spectroscopic measurements were done using a Shimadzu UV-2401PC spectrophotometer in quartz cuvettes with a path length of 1 cm.
The three-electrode electrochemical cell for in situ Raman spectroelectrochemistry was assembled in a glove box. The working electrode was prepared by drop casting SWCNTs on a Pt wire undergoing a sonicated dispersion of SWCNTs in methanol. Another Pt wire was used as the counter electrode and a silver (Ag) wire as the reference electrode. Calibration for the Ag pseudoreference electrode was performed via cyclic voltammetry (CV) and the ferrocenium/ferrocene (Fc+/Fc) redox couple was used as an internal standard because of its ideal reversible behaviour, being the potential of the Fc+/Fc couple versus an Ag electrode:
EFc/Fc+(Pt) − EAg = 540 mV | (1) |
EAg − EAg/AgCl = −126 mV | (2) |
The electronic properties have been calculated by performing DFT calculations using the CRYSTAL09 package41 on the previously optimized structures, using the PBE hybrid functional (PBE0)42 and a 6-21G* basis set; it should be noted that there are no major differences in the studied properties when optimizing the structures with the CRYSTAL program at the PBE0 level (see Fig. S2, ESI†); it should also be noted that SIESTA has demonstrated its ability to provide reliable optimized structures for large systems, such as the functionalized SWCNTs studied in this work, in a reasonable computational time; however, it is not able to calculate 1-D periodic properties and use hybrid functionals, thus explaining the choice of CRYSTAL. According to previous works, the PBE0 hybrid functional proved to describe the electronic properties in SWCNTs properly.43 The threshold on the self-consistent field (SCF) method energy was set to 10−7 Ha. The level of accuracy when evaluating the Coulomb and Hartree–Fock exchange integrals is controlled by tolerance values of 8 8 8 8 16. The k-sampling in the Monkhorst–Pack scheme used for calculating the electronic properties was the same as for the optimization step. According to the experimental data (as discussed later in the Results section), the ratio of carbon atoms functionalized vs. the total number of carbon atoms is 1/166 and 1/180, depending on the attached functional group. Thus, in our model structures, we defined the length of the repeat unit substituents that have been attached periodically to a top C atom of the nanotube. Following previous work,44 the Fermi level is 21.3 Å for the SWCNTs (9,0) and a = 12.78 Å for the SWCNTs featuring a single substituent oriented along the z-axis to match the experimental functionalization coverage, namely a = (15,0). The resulting structures lead to a ratio of 1/180 for both families. Fig. 1 shows the model functionalized structures. The substituents have been attached periodically to a top C atom of the nanotube. Following previous works,43 the Fermi level of the SWCNT has been calculated as the Dirac energy, which is defined at the center of the small energy gap.
Fig. 1 View of the unit cell of SWCNTs (top), A-SWCNTs (middle) and B-SWCNTs (bottom) for the (9,0) SWCNT family. |
(3) |
In addition, we used another formula to estimate the functionalization degree in terms of the number of functional groups per X carbon atoms in SWCNTs (1/X):
(4) |
A-SWCNTs | B-SWCNTs | |
---|---|---|
μmol g−1 | 407 | 457 |
1/X | 184 | 166 |
The functionalization reactions may strongly affect some of the SWCNTs’ optoelectronic properties, such as their UV/vis spectrum (Fig. S3 in the ESI†). In fact, the diazonium reactions of SWCNTs with 4-aminobenzenesulfonic acid and with benzyl alcohol derivatives, resulting in A-SWCNTs and B-SWCNTs, respectively, cause a total bleaching of the absorption bands of SWCNTs.
The RBM, in which all carbon atoms of the SWCNT vibrate radially in phase, is observed at low Raman shifts between 100 and 350 cm−1. The energy of the RBM depends inversely on the SWCNT diameter.25 In Fig. 2b and c inset, changes in the RBM intensity and frequency are observed due to chemical functionalization. All the RBM bands of functionalized SWCNTs showed either bleaching or amplification compared with SWCNTs. However, nanotubes with mainly metallic character present large modifications in the RBM intensity (EM11 yellow region) upon functionalization, in good agreement with the previous literature.9,34
The tangential modes are the most intense in SWCNTs and form the G-band, which is related to in-plane C–C bond stretching, at around 1590 cm−1.25 The line shape of the G-band provides additional information about the metallic and semiconducting SWCNTs in resonance. With the 2.33 eV laser line, the G-band is highly asymmetric, which indicates that the band is narrower, indicating that mostly semiconducting SWCNTs are in resonance.45 The intensity and the position of the G-band change with functionalization (see Table 2). This could indicate changes in the electronic structure, in doping or in the resonance conditions of SWCNTs contributing to the G-band. This result suggested that the functional groups attached to the SWCNTs lead to a change in the doping level of SWCNTs.28 In addition, the SWCNT spectra presented in Fig. 2 reveal information about structural defects. The disorder-induced mode (the D-band) is observed at 1329 and 1312 cm−1 using 2.33 and 1.96 eV excitation energies, respectively.25 The D-band intensity depends on the laser energy and also correlates with the degree of functionalization.34,46 The D-band increases with functionalization due to the formation of sp3 defects on the nanotube surface where the functional group is covalently attached to the SWCNT wall (Table 2).
2.33 eV | 1.96 eV | |||
---|---|---|---|---|
G-Band shift (cm−1) | I D/IG | G-Band shift (cm−1) | I D/IG | |
SWCNTs | 1593 | 0.07 | 1593 | 0.09 |
A-SWCNTs | 1585 | 0.32 | 1585 | 0.20 |
B-SWCNTs | 1590 | 0.29 | 1588 | 0.28 |
The potential-dependent Raman spectra for the RBM region of the SWCNTs in the acetonitrile electrolyte solution are shown, as an example, in Fig. 3 (the spectra of A- and B-SWCNTs are shown in Fig. S4 and S5 in the ESI†). Sections (a) and (b) show the spectra obtained using 2.33 and 1.96 eV laser excitation energies, respectively. The positions of the most intense RBM bands are studied and the diameters of SWCNTs in resonance are estimated:49 183 (dt = 1.33 nm), 268 cm−1 (dt = 0.89 nm) in the case of laser energy Elaser = 2.33 eV, and 193 (dt = 1.24 nm) and 257 cm−1 (dt = 0.93 nm) at 1.96 eV laser excitation energy. When applying positive potentials, from 0 V to +1.5 V, in steps of 0.1 V, the Fermi level (EF) is downshifted (introducing holes into the π-band). The opposite occurs when negative potentials are applied, from 0 V to –1.5 V; the electron density increases due to the upshift in the EF value (introducing electrons into the π-band). In this way, when the EF value reaches a van Hove singularity (vHs), the corresponding electronic transition is blocked, and the Raman signal is bleached.
For a more detailed analysis of the SWCNT metallic/semiconducting character and the doping level, the dependence of the RBM intensity on the electrode potential at the selected laser lines was studied (see Fig. S6 in the ESI†). A difference in the profiles of the Raman intensity (IRBM) vs. applied potential (Eapp) of metallic and semiconducting SWCNTs has been previously reported.47,48 For semiconducting SWCNTs in resonance at 2.33 eV, the profile IRBMvs. Eapp exhibits a plateau close to 0 V. The IRBM profile is not significantly attenuated close to 0 V and starts changing only after the electrode potential reaches the first vHs. In contrast, the intensity of the band for metallic SWCNTs is attenuated at potentials close to 0 V. A similar effect occurs for the SWCNT in resonance at the laser excitation energy of 1.96 eV. Therefore, for semiconducting SWCNTs, the profile of IRBMvs. Eapp at around 0 V is a plateau and for the metallic SWCNT, it is sharp and shows a maximum. So one can distinguish between metallic and semiconducting SWCNTs.47 These assignments obtained using the IRBMvs. Eapp profiles are in agreement with the electronic transitions in the Kataura plot.49–51
The IRBMvs. Eapp profiles of the SWCNT, A- and B-SWCNT samples are shown in Fig. 4. The IRBM profiles of the bands in resonance via the EM11 transition reach a maximum at a given optimum potential. We found that maximum IRBM for the metallic SWCNT vs. Eapp is shifted for the functionalized samples in comparison with SWCNTs (Fig. 4b and c). In the case of semiconducting SWCNTs, there is no noticeable shift because of the occurrence of a plateau instead of a maximum. For both laser excitation energies, the profiles of IRBM in resonance via EM11 are downshifted for A-SWCNTs and upshifted for B-SWCNTs, in comparison to SWCNTs. This indicates that doping is induced on the SWCNTs by the functional groups attached to their walls. The benzenesulfonic acid moiety has a large electron affinity due to its structure composed of a benzene ring attached in the para position to a sulfur(VI) atom in a central part of a SO3H group. With three electronegative oxygen atoms, the SO3H group is strongly electron-withdrawing. It is necessary to apply an additional negative potential to reach the maximum peak of Raman intensity, pointing out that the A-SWCNTs are p-doped and the EF value is downshifted. In this way the valence band vHs is depleted and it is necessary to apply an extra negative potential to reach the neutral point. On the other hand, in the case of B-SWCNTs the maximum peak is shifted to positive potentials. This means that the EM11 transition becomes occupied (n-doped, filling) because EF is upshifted, as benzyl alcohol gives electrons to the SWCNT electronic structure.
Fig. 5 DOS of the SWCNTs (dashed lines), A-SWCNTs (red lines) and B-SWCNTs (blue lines) for the two families studied. |
Theoretical calculations support that there is a shift in the Fermi level and the vHs due to the functionalization. As observed experimentally, the A-SWCNT functionalization yields a downward shift of the SWCNT Fermi level. On the other hand, the B-SWCNT functionalization takes place at the origin of an upward shift of the Fermi level with respect to the SWCNTs. The calculated shifts of the Fermi level are given in Table 3.
E F (eV) | SWCNTs | A-SWCNTs | B-SWCNTs |
---|---|---|---|
(9,0) | −4.65 | −4.69 | −4.51 |
(15,0) | −4.73 | −4.85 | −4.62 |
Dark voltammograms for SWCNT, A-SWCNT and B-SWCNT film electrodes show the features that have been previously observed in the literature for similar systems, i.e. a ‘bow tie’ shape,54,55 with the narrowest part indicating the position of the Fermi level (see Fig. S7 in the ESI†).
Fig. 6 summarizes the changes observed by CV under illumination. In particular, the current in the cathodic branch of the scan is modified when the lamp is alternately switched on and off (Fig. 6a). Thus, the photocurrent can be calculated as the difference between consecutive on and off currents (Ilight − Idark) at various Eapp values (Fig. 6b). Photoanodic activity, typically less than 1 μA, was much lower than the photocathodic one (see Fig. S8, ESI†). When the three SWCNT electrodes are irradiated, a negative photocurrent appears at potentials lower than −0.4 V. In previous literature, photoelectroactivity for SWCNTs has been observed in another electrochemical reaction.56 In our work, the photocathodic process has to be observed at low scan rates (2 mV s−1), and the values observed for the photocurrent reached the noticeable value of ca. 20 μA. In these cases, the effects of the SWCNT functionalization are observed: A-SWCNTs show higher photocathodic currents than SWCNTs, indicating p-doping, while the contrary effect is observed for B-SWCNTs, indicating n-doping. Differences in photoanodic current between SWCNT, A-SWCNT and B-SWCNT films are not significant (see Fig. S9, ESI†). However, in the case of photocathodic current, the differences between SWCNT, A-SWCNT and B-SWCNT films are above 5 μA. These results are fully consistent with those described in the preceding sections, and indicate that the photoelectrochemical properties of SWCNTs can be tuned by chemical functionalization. Moreover, the chemically modified photoresponse can be predicted by using spectroelectrochemical techniques in a reliable manner, opening new possibilities in fields such as solar energy conversion and photosensors.
Footnote |
† Electronic supplementary information (ESI) available: Further details of the purification of SWCNTs, reagents and solvents used; deconvolution of the D-band and the G-band; DOS of the SWCNTs for the structure optimized with SIESTA and CRYSTAL; UV-vis absorbance spectra of SWCNTs, A-SWCNTs and B-SWCNTs; the RBM region of Raman spectra at different Eapp values for A-SWCNTs and B-SWCNTs; dark voltammograms for SWCNTs, A-SWCNTs and B-SWCNTs. See DOI: 10.1039/c8cp06961a |
This journal is © the Owner Societies 2019 |