Charles C.
Chusuei
* and
Ram Chandra
Nepal
Chemistry Department, Middle Tennessee State University, 440 Friendship Street, Murfreesoboro, TN 37132, USA. E-mail: Charles.Chusuei@mtsu.edu; Tel: +1-615-898-2079
First published on 4th December 2023
Carbon nanomaterials (CNMs, carbon dots, carbon nanotubes, and graphene) have received much attention in recent decades for their technological roles as electrocatalysts for biosensing and fuel cell applications in aqueous solutions. Their complex form factor presents challenges for delineating structure–property relationships, namely the interplay of electroactive area surface defects and exposed graphene planar structure, for optimizing their electrocatalytic activity. Conflicting examples in the literature show higher defect density in the graphene structure with increased or decreased conduction of the material. The graphenyl sheet curvature, voltage range of the electrochemical redox reaction, dispersion of charged impurities affecting the charge mobility, and overall resistivity of the CNM materials should be considered to optimize the overall electrochemical activity, particularly as they relate to redox reactions taking place in the −0.2 to +0.3 V standard potential range.
However, some experiments show defects hampering rather than enhancing conduction.7 Why is it that in some cases, electrochemical current reading increases with higher CNM defect density while in others, they do not? The answer lies with the location of the Fermi level (EF) relative to the measurement made (vide infra). An interplay between defect density and reactant exposure to graphene sheets influences the optimization of these electrocatalyst materials. The presence of charge carriers within the CNMs also play a role in their conductive properties.
![]() | ||
Fig. 1 (top) TEM of MWCNTs after (a) 2 h, (b) 4 h, and (c) 8 h of sonication; (bottom) (A) Raman spectra of 1-through-8 h sonicated MWCNTs. (B) Integrated Raman D/G peak areas (left axis) and atomic % oxygen (right axis) vs. sonication time. Reprinted from ref. 9 and 10 (Copyright 2005 and 2006, respectively) with permission from the American Chemical Society. |
The defect density within the graphene sheets reached a plateau at the 2 h mark. The kinetic growth of the O-containing functional groups that precisely matched with sonication time and D/G band growth (and subsequent plateauing) corroborates this conclusion within the study. Since functionalization directly correlated with the growth of sp3 carbon in the Raman analysis, we conclude that defectsindirectly increase electrocatalytic activity on the CNM surface by enabling the tethering of Pt nanoparticles (NPs) in a way that minimizes aggregation and, hence, in the case of the direct methanol fuel cell reaction (vide infra), increases the surface area of adsorbed Pt for redox reactions.
The Raman D-band results from the radial breathing modes of six-atom rings and require defects (topological imperfections, such as vacancies, edges, pores, etc.) for its activation;11 the D′, 2D, and 2D′ peaks are overtones of the defects.12 The Raman G band denotes the graphene sp2-hybridized carbon. Raman can distinguish between a hard amorphous carbon, a metallic nanotube, and doped graphene.13 Density functional theory (DFT) calculations predict that defects within the 2D structures increase electron transfer at the aqueous solution-solid surface interface, resulting in greater conductivity since defects increase activity by serving as electron acceptors from adsorbed metal nanoparticles or from analyte materials as they undergo solution redox.14 But, this observation is not universal since increases in defect density have been correlated with decreased current response in some instances,7,14 including those examples that will be discussed in section 5 of this article.
A caveat that should be noted is that flattened graphene structures within multiwalled carbon nanotubes have been reported to have a marked D-band intensity without introducing defects by Picheau et al.15 where variances in graphene curvature contribute to observed D band intensity.16 The phenomenon is observed for carbon nanotubes with a single or a few walls. Nanotubes at a critical diameter (greater than 5 nm) spontaneously collapse and create folds, enhancing D band intensity. In this experiment, the end caps of the MWCNTs were cut, and the inner concentric nanotubes were removed, flattening the remaining structure. The defect density of the outer-most walls was constant, but the D band intensity increased (Fig. 2)!15 Indeed, characterizing exposed graphene defects becomes more nuanced with this discovery and invites questions regarding CNM composite fabrication for optimized electrochemical performance.
![]() | ||
Fig. 2 Raman spectra of flattened and unflattened MWCNTs. Reprinted from ref. 15 (Copyright 2021) with permission from the American Chemical Society. |
For single-walled CNTs where Raman signals of peaks of the overtones would be more visible, defect density and graphene sheet exposure can be monitored using the D, G, D′, and 2D bands that appear at ca. 1330, 1580, 1615, and 2675 cm−1.17,18 The increase in defect density can be exhibited by splitting the primary G band into the G band, along with a resolved D′ band, observed at ca. 1560 cm−1.5 The D′ band may be subsumed within the G-band in cases where this feature is not well-resolved; its intensity is convoluted within the G band intensity.
In a study by Slate et al.19 on the electrochemical performance of graphite and graphene paste electrodes, greater electrochemical activity was observed with later lateral flake sizes as observed in the TEM and correlating with 1 mM hexaammineruthenium(II) chloride as a redox probe, denoting sp2 graphene sheets. Smaller flake size improved heterogeneous electron transfer (HET) kinetics. The beneficial resonance was due to the increased number of edge plane-like sites on the electrode surface. Density functional theory (DFT) showed that coverage of the edge plane-like sites led to greater electrochemical performance. Insight from this work shows that the flattening of the MWCNTs in Picheau et al.'s15 study resulted in the D-band enhancement due to greater exposure of the defects for Raman detection (Fig. 2). This interpretation is consistent with defects acting as electron acceptors, giving rise to increased conductivity while simultaneously having greater exposure to the graphene sp2 carbon.
Noteworthy is that in cases with substantial changes in graphene sheet curvature, Picheau et al.'s15 findings also indicate that normalization of defect density measurements from Raman spectra by dividing the D-band integrated peak area intensity by that of the G-band, under these conditions, is unreliable. Raman scattering from folds in a single-layer graphene sheet has been known to arise from spatially inhomogeneous curvature around a fold within the graphene sheet, resulting in an enhanced D band.16 In addition, wrinkling and crumpling in twisted few and multilayer graphene have increased the integrated area under the D Raman band.20 In such cases where there are severe distortions in the curvature and/or crumpling of the graphene sheet structure, complementary experimental evidence should be used to corroborate defect density changes to increase conductivity. So long as the curvature of the CNM is not affected, we postulate that defects enhance conductivity for those electrochemical reactions occurring between −0.2 and +0.3 V.
One such example is that of a Prussian Blue (PB) electrodeposited onto a glassy carbon modified with zirconia-doped functionalized carbon nanotubes (PB/ZrO2-fCNTs/GC), giving rise to enhanced sensitivity for H2O2 detection,21 in which redox activity for H2O2 was observed at +0.2 V. With functionalization, the defect density of the PB/ZrO2-fCNTs/GC increased as observed in the Raman spectra, accompanying enhanced signal for detecting the H2O2 analyte.
The three-dimensional porous and redox-active Prussian Blue in graphene (PB@G) to aerogels with mass ratios of 2.5:
1 to 1
:
2.5 show enhanced redox activity for H2O2 detection accompanying increased defect density as observed by Raman spectroscopy.22 XPS core level shifts of the C 1s orbital showed an increased number of oxidation states, accompanying greater PB loading and, subsequently, larger D band populations. The Raman ID/IG ratios of this series of PB-loaded graphene aerogels (with the graphene-to-PB loadings) for the pristine graphene aerogel, 2.5
:
1 G
:
PB, 1
:
1 G
:
PB, and 1
:
1.25 G
:
PB were 2.16, 1.17, 2.17, 2.02, respectively. The steepest slope for the current increase was obtained for the electrode with the largest density of electrodeposited PB to detect H2O2. These redox peaks were within the −0.2 to +0.3 V range. The largest current observed for the reduction of H2O2 was observed with a detection limit of 5 nM for the analyte with the largest ID/IG band ratio.
In addition, PB tethered to graphene (PB/GE) showed an enhanced reduction of H2O2; within this system, PB introduced defects onto the graphene sheet for a four-electron oxygen reduction in acidic media for the oxygen reduction reaction (ORR), important for fuel cell reactions.23 A higher ID/IG band ratio (in parentheses), accompanying higher measure current using the cyclic voltammetry (CV) peak-to-peak height, was observed for PB/GE (1.95) than for graphene (GE) (1.92) deposited on the glassy carbon working electrode (GCE) and is consistent with the prediction by Kislenko et al.13 for the observed standard reduction voltage of +0.2 V (Fig. 3).
![]() | ||
Fig. 3 (A) Raman stack plot of the PB, GE, GO, and PB/GE; (B) CVs of the GCE, GE–GCE, PB/GE–GCE, and PB–GCE. Reprinted from ref. 23 (Copyright 2013) with permission from Elsevier. |
PB deposited on graphene oxide within a graphene oxide/PB/glucose oxidase/chitosan composite showed enhanced sensitivity for detecting H2O2 with a 1.2 × 10−7 M detection limit. Redox peak current intensities for voltages within the −0.2 to +0.3 V (with Raman ID/IG values in parentheses) were seen in the following descending order: GO/PB (2.137) > GO (1.799) > PB.24 (The bare PB had no D or G Raman bands.) The higher defect density electrocatalyst exhibited greater sensitivity to glucose. For PB introduced to reduced graphene oxide/multi-walled carbon nanotube (RGO/MWCNT) composites, Silva et al.25 reported increased current at a +0.2 V standard potential for the detection of ClO− and H2O2 for reduced PB/RGO/MWCNTs as compared to RGO/MWCNTs. As with the previous example, the increased sensitivity for analyte detection accompanied the increase of defects for redox activity within the Kislenko range. The ID/IG ratio for the composites with and without PB were 2.92 and 2.47, respectively (Fig. 4).
![]() | ||
Fig. 4 Raman stack plot of (a) GO, (b) GO/PB, and (c) PB. Reprinted from ref. 24 (Copyright 2011) with permission from Elsevier. |
In the case of PB adsorbed onto ZnO nanoparticles tethered to carboxylic acid functionalized carbon nanotubes (PB/ZnO/COOH-MWCNTs) in our laboratory, the introduction of defects accompanied the increase in electrochemical sensing current for the detection of H2O2 in phosphate buffer solution at pH 7.26Fig. 5A shows control CV experiment of 5 mM H2O2 in phosphate buffer solution exposed to PB/ZnO/COOH-MWCNTs and subsets of this composite (see figure caption for details). The CV peak-to-peak height showed a 2.7-fold increase in current density between ZnO/COOH-MWCNTs and PB/ZnO/COOH-MWCNTs. The corresponding ID/IG ratios for ZnO/COOH-MWCNTs and PB/ZnO/COOH-MWCNTs were 0.733 and 0.940 (Fig. 5B), respectively, corresponding with a 1.3-fold increase in defect sites (Fig. 5A). The composite with the larger defect density exhibited the highest current for H2O2 detection. The APAP CV signal increased with increasing sp3 carbon density. Regarding the predicted trend according to the Kislenko range, the oxidation current of ZnO/COOH-MWCNTs was slightly outside the prescribed range with respect to measured current with applied voltage. The oxidation peak shifted from −0.341 V (outside the range) to +0.0004 V (within the range) within the CV data. Yet, in this single instance where the oxidation potential was outside the prescribed range, the defect density population (sp3 carbon) still accurately predicted the current increase.
![]() | ||
Fig. 5 (A) CVs of 5 mM H2O2 in phosphate buffer solution (PBS) (pH 7) using (a) PB/ZnO/COOH-MWCNTs and (b) ZnO/COOH-MWNTs; (c) CV of PBS using PB/ZnO/COOH-MWCNTs, (d) CV of 5 mM H2O2 using PB, and (d) CV of 5 mM H2O2 in PBS on the bare GCE; (B) Raman stack plot of (a) ZnO/COOH-MWCNTs and (b) PB/ZnO/COOH-MWCNTs. Fig. 5A is reprinted from ref. 26 (Copyright 2019) with permission from the American Chemical Society. Fig. 5B was newly acquired by the authors of this Focus article. |
The example of Fe2O3 and PB incorporated into an MWCNT composite for the oxidation reaction of 500 μM H2O2 (demonstrating its electrochemical sensing capabilities) was observed at +0.1 V.27 The ∼15% increase in current was accompanied by a decrease in defect density from the PB nanoparticles in the Fe2O3/MWCNT@PB electrocatalyst support, as observed in the Raman data (Fig. 6). Comparing the Raman spectra in the stack plot between Fe2O3/MWCNTs with that of Fe2O3/MWCNTs@PB, when the PB was introduced, the D′ peak at ca. 1550 cm−1, denoting the increase in defect density on the graphene sheet, which accompanied the decrease in peak-to-peak height signal observed in the corresponding CVs of these composites as 500 μM H2O2 was detected when comparing the CV in Fig. 6b (right) with this same composite Fig. 6a (right) without H2O2. The ID/IG ratios of the MWCNT-Fe2O3@PB, MWCNT-Fe2O3, and MWCNT were 0.444, 0469, and 0.671, respectively (Fig. 6, left). The Raman shift at ca. 2150 cm−1 is an experimental artifact of the PB. The peak at 2154.79 cm−1 is an experiment artifact of the Fe(III)–CN–Fe(II) additive used.
![]() | ||
Fig. 6 (Left) Raman stack plot of (a) MWCNT, (b) MWCNT-Fe2O3, and (c) MWCNT-Fe2O3@PB; (right) CVs of (a) MWCNT-Fe2O3@PB, (b) a + H2O2, (c) MWCNT + H2O2, and (d) MWCNT-Fe2O3 + H2O2. Reprinted from ref. 27 (Copyright 2014) with permission from Elsevier. |
For PB incorporated into reduced graphene oxide tethered to MWCNTs,25 the composite with the greater exposed graphene (and fewer defects verified by Raman and SEM) exhibited the highest current for H2O2 detection, even for redox activity within the −0.2 to +0.3 V range in contrast to the previous examples. Resistance and charge transfer measurements using electrochemical impedance spectroscopy (EIS) showed resistances of 297, 34.8, 6.91, and 3.54 kΩ for indium tin oxide (ITO), PB, RGO/MWCNT/PB, and RGO/MWCNTs, respectively (Fig. 7, right). The composite with PB had a higher semiconducting character (hence more resistance) than the composite without it. Integrated peak area intensities revealed the RGO/MWCNTs to have more defects (from sp3 carbon) than RGO/MWCNT/PB. We postulate that impurities from the copper wiring used in the ITO surface played a role in which defects hampered charge mobility instead of serving as electron acceptors. The ID/IG ratios for RGO/MWCNTs and RGO/MWCNTs/PB were 2.92 and 2.47, respectively (Fig. 7, left).
![]() | ||
Fig. 7 Raman stack plot (a) RGO/MWCNT, (b) PB, and (c) RGO/MWCNT/PB; (b) EIS of (a) ITO, (b) PB, (c) RGO/MWCNT, and (d) RGO/MWCNT/PB. Reprinted from ref. 25 (Copyright 2020) with permission from Elsevier. |
![]() | ||
Fig. 8 TEM of refluxed ZnO/COOH-MWCNTs at (A) 60, (B) 120, (C) 150, and (D) 165 min of sonication; (E) potentiostat current of 10 mM uric acid in PBS (pH 7) from CVs (left axis) and Raman G band integrated peak areas (right axis). Reprinted from ref. 28 (Open Access 2018) with permission from MDPI. |
In a study by Zakrzewska et al.,29 PB was incorporated into a Pt NP-reduced graphene oxide (PB/Pt/RGO) composite, which increased the number of defects of the electrocatalyst for the oxygen reduction reaction in 0.5 M H2SO4, important for the proton exchange fuel cell reactions. The overall peak current, which had a maximum +0.6 V standard potential, was compared with the relatively defect-rich, low-exposed graphene with that of the Pt NP-reduced graphene oxide composite without PB (Pt/RGO). The current was greater for the ORR for the Pt/RGO than that had fewer defects and higher exposed graphene (ID/IG = 1.35) than for PB/Pt/RGO (ID/IG = 1.49), which is a trend consistent for standard redox voltages outside −0.2 to +0.3 V.
A similar phenomenon has been observed for methanol oxidation at a standard potential of +0.5 V for Pt NP-MWCNT composite. For electrochemical reactions pertaining to the direct methanol fuel cell, low-defect MWCNTs with tethered Pt nanoparticles, 2–5 nm in diameter, were uniformly dispersed onto the sidewalls and were found to be highly effective for the direct methanol fuel cell reaction.6 The lower the defect density, the higher the measured current observed. For the acid-treated MWCNT-Pt nanocomposite with a feeding ratio of Pt-to-MWCNTs of 1.5-to-1 (PtAM1.5), the ID/IG = 0.0538 (determined from the integrated Raman D and G band integrated peak area ratios in Fig. 9) led to a forward anodic peak current-to-reverse anodic peak current ratio, IF/IR = 0.83. In the case of the low-defect composite using the same feeding ratio (PtLM1.5), the measured ID/IG = 0.0524, resulting in IF/IR = 2.94. Hence, the greater exposed graphene for this electrochemical reaction had a greater than 3.5-fold increase in IF/IR current. The measured charge transfer resistance was 24.99 Ω with excellent electrocatalytic activity with a forward anodic peak current density of 47.37 mA cm−2.
![]() | ||
Fig. 9 (A) Raman stack plot of PtAM1.5 and PtLM1.5; CVs of PtAM1.5 and PtLM1.5. Reprinted from ref. 14 (Copyright 2011) with permission from the American Chemical Society. |
For a similar composite consisting of Pt NPs on low-defect 3D carbon nanotube/nitrogen-doped graphene hybrid aerogels (Pt/LDCNT-NG), methanol oxidation occurred at +0.4 and +0.7 V standard potentials.7 The electrocatalytic activity was quantified using CVs in N2-saturated 0.5 M H2SO4 solution. Pt/DCNT-NG had an ECSA = 132.4 m2 g−1, while the Pt/NG had an ECSA = 58.1 m2 g−1. Current readings for GO, graphene (G), n-doped graphene (NG), and Pt nanoparticles tethered to graphene (Pt/G) were compared. In comparing the Raman ID/IG ratios (values in parentheses) between GO (0.87), G (1.03), NG (0.16), Pt/G (1.07), and Pt/LDCNT-NG (0.26), the Pt/LDCNT-NG composite exhibited the largest current reading in this series of carbon nanotube aerogel composites. Of the Pt-containing composites, the graphene architecture with the lowest ID/IG ratio also had the highest specific electrochemical surface areas (ECSA).
Cobalt and nitrogen-doped on reduced graphene oxide (N/RGO) with Co nanoparticles (Co@Co-N/RGO) is a useful catalyst for the ORR in the direct methanol fuel cell reaction, took place at +0.848 V; its activity was comparable to standard Pt nanoparticles deposited on carbon.30 Accompanying the high voltage, Raman spectra showed that more defects correlated with a diminished current density, which correlated with increased exposed sp2 carbon. In comparative Raman intensity measurements, the ID/IG band ratios for the RGO, N/RGO, and Co@Co-N/RGO were 1.02, 1.05, and 0.95, respectively. The electrocatalytic activity was governed by the degree of exposed graphene, where increasing current density across the electroactive surface correlated with greater exposed graphene, denoted by the ID/IG ratio.
To explore whether the effects of current are universally proportional to graphene sheet exposure, we carried out experiments on an array of saccharide-based carbon dots, where the ID/IG ratio was varied based on the ‘sweet taste’ characteristic (which summarizes molar volume, solute/solute interactions, and intramolecular H-bonding) of the saccharide precursors.31 Acetaminophen (APAP) was used as the probe molecule for redox detection using galactose CDs (GalCDs), lactose CDs (LacCDs), and glucose CDs (GluCDs) (Fig. 10A). Noteworthy is the fact that since CDs are zero-dimensional in morphology, the effect of variations in graphene sheet curvature can be eliminated as a variable that could produce additional D band intensity as an experimental artifact. Higher sp2-to-sp3 carbon ratios (in parentheses) within this CD series correlated with higher sensitivity for the measurement of APAP according to the measured Raman IG/ID values: GalCDs (10.18) > LacCDs (9.30) > GluCDs (6.53) (Fig. 10B). Currents were measured at redox voltages greater than +0.5 V. GalCDs, having the greatest sensitivity for APAP, had an oxidation voltage of +0.549 V. Hence, the observation for the Kislenko trend was evident for CDs.
![]() | ||
Fig. 10 (A) CVs of GluCDs, LacCDs, and GalCDs; (B) Raman stack plot of GluCDs, LacCDs, and GalCDs. Reprinted from ref. 31 (Copyright 2021) with permission from Wiley-VCH. |
It should also be noted that charge mobility from alkali ions, e.g., K+ and Na+, ubiquitous in many electrochemical systems, would not be involved in current density reduction by defects within the graphene structure, as showdown by Jeong et al.37 In their experiments, direct intercalation of alkali-metal cations from K+ and Na+ were incorporated for the electrochemical redox reaction of Prussian Blue on a graphene surface, comparing them with controls where Na+ and K+ ions were not involved. CV data showed that PB films passivated with monolayer graphene still underwent electrochemical redox reactions with these alkali ions present despite their inability to penetrate the graphene and be incorporated into PB; they showed that a transparent graphene electrode covering PB can still be used as an effective H2O2 transducer. The graphene overlayers did not hamper ionic interactions of the alkali cations with FeIII ions in PB, indicating that defects within the graphene (because of its transparency to the alkali cations) do not affect the charge mobility of these cations and, therefore, would not result in a current decrease.
This journal is © The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2024 |