Sheng
Hu
a,
Zachary P. L.
Laker
b,
Hannah S.
Leese
a,
Noelia
Rubio
a,
Martina
De Marco
a,
Heather
Au
a,
Mark S.
Skilbeck
b,
Neil R.
Wilson
b and
Milo S. P.
Shaffer
*a
aDepartment of Chemistry and London Centre for Nanotechnology, Imperial College London, London SW7 2AZ, UK. E-mail: m.shaffer@imperial.ac.uk
bDepartment of Physics, University of Warwick, Coventry CV4 7AL, UK
First published on 16th June 2017
Graphene and graphene nanoplatelets can be functionalised via a gas-phase thermochemical method; the approach is versatile, readily scalable, and avoids the introduction of additional defects by exploiting existing sites. Direct TEM imaging confirmed covalent modification of single layer graphene, without damaging the connectivity of the lattice, as supported by Raman spectrometry and AFM nano-indentation measurements of mechanical stiffness. The grafting methodology can also be applied to commercially-available bulk graphene nanoplatelets, as illustrated by the preparation of anionic, cationic, and non-ionic derivatives. Successful bulk functionalisation is evidenced by TGA, Raman, and XPS, as well as in dramatic changes in aqueous dispersability. Thermochemical functionalisation thus provides a facile approach to modify both graphene monolayers, and a wide range of graphene-related nanocarbons, using variants of simple CVD equipment.
Previously, we developed a self-limiting, scalable, thermochemical functionalisation method and demonstrated its applicability to multi-walled carbon nanotubes (MWCNTs),22,23 and carbon black nanoparticles (CB NPs),24 with a wide range of functionalizing moieties. The mechanism involves the thermal decomposition of pre-existing defect groups to generate surface radicals which are able to graft many different reagents; since only existing sites are activated, it is hypothesised that additional damage to the carbon framework is avoided. It is feasible to implement this process in an entirely solvent-free set-up, as long as the grafting reagent is sufficiently volatile to be introduced, and the excess removed, by vacuum. The approach typically introduces only a low degree of functionalisation, at pre-existing defect sites, and hence does not intrinsically degrade the original as-synthesised framework properties; nevertheless, these low levels of functionalization have been shown to be sufficient to adjust wetting and dispersibility as required. Herein, the approach was extended to single layer graphene (SLG) to confirm the generality of the approach and explore the effects on the graphene lattice in detail. By using graphene supported on a transmission electron microscopy (TEM) grid, it is possible to directly image the grafted functionalities, and explicitly measure the effects on mechanical performance. The use of a gas phase approach is particularly helpful for maintaining the cleanliness needed for high resolution imaging, as well as representing a generic, clean, route for functionalisation of CVD graphene on substrates. Having established efficacy on single monolayers, the approach was applied to the functionalisation of bulk quantities of GNPs, enabling different types of characterisation, and highlighting the versatility of the approach. The gas phase approach is not expected to exfoliate individual layers, rather modify the accessible surface of SLG/GNP materials.
AFM nano-indentation was used to measure the mechanical properties of the functionalised graphene sheet, using an established method.27 As shown in Table 1 (and ESI Fig. S2†), neither the thermochemical treatment (TTC-SLG) nor the functionalization (TFI-SLG) changed the Young's modulus, measurably, compared to the pristine graphene grid (AR-SLG).
Sample | Young's modulus (TPa) |
---|---|
AR-SLG | 1.2 ± 0.4 |
TTC-SLG | 1.4 ± 0.1 |
TFI-SLG | 1.3 ± 0.2 |
Whilst the very dilute modification of high quality monolayer graphene helped to identify covalent binding of the functional groups, the low surface concentration and absolute quantity preclude further chemical measurements. The functionalisation was, therefore, extended to typical bulk GNPs with high surface area, synthesised by a non-oxidising chemical growth process. SEM images of the initial GNP powders (Fig. 2a) showed stacks of flake structures, as further confirmed by TEM and AFM of GNPs deposited from dispersions (Fig. 2b and c). The lateral average size of the GNPs used here was estimated to be 2.4 ± 2.1 μm from image analysis (histogram available in ESI Fig. S5†).
![]() | ||
Fig. 2 (a) SEM, (b) TEM, (c) AFM images of the as-received GNPs, and (d) the height profile of AFM image. |
The typical thickness of the GNPs, as determined by AFM was approximately 4 nm (refer to Fig. 2d), indicating around 10 graphitic layers; bulk XRD data gave similar values, 5–6 nm (see ESI, Fig. S4 & Table S2†). The GNPs have a high defect concentration, as shown by their waviness, small size, large ID/IG ratio in Raman spectra (0.97 ± 0.27) and high oxygen content determined by XPS (refer to Table 2). A series of functionalities that have been proved suitable for thermochemical functionalisation reactions on MWCNTs and CB NPs23,24 were used again for the GNP grafting, including the polymerisable monomers methyl methacrylate (MMA), 4-vinyl pyridine (4VP) and poly(ethylene glycol)methyl acrylate (PEGMA), as well as 1-iodododecane (IDD) as a non-polymerisable reference. A thermally-treated control sample (TTC-GNPs) was also prepared as a baseline for the functionalised GNPs (f-GNPs).
Sample | M w | Grafting ratio (wt%) | Grafting concentration (μmol g−1) | Monomeric repeats | Atomic ratio (%) | Water dispersability (μg mL−1) | I D/IG ratio | I 2D/IG ratio | ||
---|---|---|---|---|---|---|---|---|---|---|
C | O | N | ||||||||
AR-GNPs | N/A | N/A | N/A | N/A | 93.2 | 6.3 | 0.5 | 191.5 | 0.97 ± 0.27 | 0.30 ± 0.25 |
TTC-GNPs | N/A | N/A | N/A | N/A | 97.4 | 2.2 | 0.4 | 145.2 | 1.04 ± 0.09 | 0.38 ± 0.03 |
IDD-GNPs | 169.3 | 1.7 | 102 | 1.0 | 95.3 | 4.2 | 0.5 | 113.6 | 1.04 ± 0.01 | 0.43 ± 0.01 |
P(MMA)-GNPs | 100.1 | 3.8 | 384 | 3.8 | 93.2 | 6.6 | 0.2 | 79.0 | 0.99 ± 0.19 | 0.42 ± 0.23 |
P(4VP)-GNPs | 105.1 | 4.4 | 417 | 4.1 | 94.7 | 1.5 | 3.7 | 523.4 | 0.96 ± 0.17 | 0.43 ± 0.17 |
P(PEGMA)-GNPs | 518.6 | 15.1 | 285 | 2.8 | 89.4 | 10.6 | 0.0 | 734.7 | 1.07 ± 0.07 | 0.41 ± 0.24 |
TGA was employed to quantify the grafting ratio, taking advantage of the differences in thermal stability between the grafted organic components and carbon framework. Mass spectrometry of the evolved gases confirmed the identity of the specific grafted species (Fig. 3b–e). The AR-GNPs showed a continuous weight loss during the TGA experiment, attributed to volatile impurities and the decomposition of oxygen-containing functional groups generated during the original synthesis, resulting in a 7.1 wt% loss at the final temperature of 850 °C. The TTC-GNPs displayed a smaller weight loss (2.4 wt%) than the AR-GNPs, as the volatile impurities are removed during the thermal treatment. XPS (see below) confirmed that the surface oxide groups decrease in concentration after thermal treatment. 1-Iodododecane (IDD) grafted GNPs (IDD-GNPs) showed a grafting ratio of 1.7 wt%, relative to TTC-GNPs; the grafted hydrocarbon chains (m/z 29 and m/z 85), but no iodide groups (m/z 127) or unreacted IDD (m/z 169), were observed by TGA-MS analysis. The grafting concentration (μmol of grafted functionalities per gram of GNPs), and hence the accessible radical site concentration, was therefore estimated to be 102 μmol g−1, assuming a 1:
1 reaction between the IDD and radicals. Previously, a range of nanotube materials have shown IDD grafting ratios in the range 1.4 wt% to 3.3 wt%, or 87 μmol g−1 to 202 μmol g−1, (refer to ESI Table S1†);23,24 the specific grafting density (normalised by specific surface area) was in a broadly similar range (around 0.4 μmol m−2), indicating a similar surface concentration of defects. By comparison, the GNPs show a similar grafting ratio (1.7 wt%) to the other nanocarbons, but a slightly lower surface concentration of 0.16 μmol m−2, which could be associated with relatively inaccessible slit-pore type sites at the flake edges. XRD of selected GNP samples confirmed that there was negligible change in the underlying platelet thickness; the lack of interlayer expansion is consistent with the expected functionalization of the accessible GNP surface (ESI Fig. S4 and Table S2†).28,29 The grafting ratios (refer to Table 2) of P(MMA), P(4VP) and P(PEGMA) ranged from 3.7 wt% to 13.1 wt%, mainly reflecting the molecular weight of the grafting monomer; the PEGMA grafting ratio was relatively high due to the larger mass of the monomer, as previously observed on nanotubes.23,24 Interestingly, the degree of oligomerisation is very similar (ca. 4 repeating units) across different types of nanocarbons (refer to ESI Table S1†). It seems that the polymerisation process is rather similar for different nanocarbons, leading to similar oligomer chain lengths. In these estimates, the termination of oligomerisation/polymerisation was assumed to occur readily at any point of the GNP surface rather than requiring another radical site; such an assumption is also consistent with the low degree of polymerisation. Preliminary data show that thicker GNPs from the same supplier (grade GM5) have a lower absolute degree of functionalization (ESI, Fig. S5 and S6†), as expected, in line with their lower accessible surface area.
![]() | ||
Fig. 3 TGA-MS profile for AR-, TTC- and various f-GNPs. Measurements were carried out under N2 atmosphere. |
The surface atomic ratios measured by X-ray photoelectron spectroscopy (XPS) analysis also confirmed successful grafting (Table 2). The heat treated control (TTC-GNPs) displays a reduced oxygen content (2.2%) compared to the original AR-GNPs (6.3%), due to the expected decomposition and rearrangement of the oxide groups; some groups are expected to be eliminated entirely,30 but in other cases, radicals formed at high temperature re-trap oxygen on exposure to atmosphere at room temperature to form oxygen-containing pyrones and other species.13,30,31 For the samples grafted with oxygen-containing polymers, P(MMA)- and P(PEGMA)-GNPs, the oxygen atomic ratio was significantly increased compared to AR- and TTC-GNPs, as expected. The clearest evidence of grafting can be seen in the P(4VP) samples, since they introduce a unique nitrogen signal. The nitrogen (N) atomic ratio was significantly increased (3.7%), compared to TTC-GNPs (0.4%) (further discussion in the ESI†).
The Raman spectra of the GNPs (Table 2 and Fig. S3†), show a broadly consistent ID/IG, within uncertainty, showing no evidence of damage during grafting. Although this ratio is not a very sensitive indicator when it is near unity,32 a constant defect concentration is consistent with the proposed reaction mechanism, taking place only on the pre-existing defects. Similarly consistent Raman data have been found across a range of nanocarbons.23,24 The slightly increased I2D/IG values indicate that the crystallinity of the GNPs was improved after thermal treatment, due to the removal of amorphous residue and/or some healing of the framework during the thermal activation treatment.
Dispersibility tests were carried out, by bath sonicating GNP samples in HPLC water for 5 minutes (an initial 1 mg mL−1 loading was previously found to be sufficient to ensure saturation). The dispersions were left to stand overnight and then the supernatants were carefully decanted in order to quantify the concentration by UV-Vis spectrometry. The AR-GNPs can be readily dispersed at 191.5 μg mL−1, due to the high volume of surface oxides produced during the synthesis. The f-GNPs can be stably dispersed in water at a wide range of concentrations, from 79.0 to 734.7 μg mL−1, driven by the various hydrophilicities of the grafted functionalities. The hydrophobic PMMA grafted sample was less dispersible than the AR-GNPs, as expected, with a concentration of 79.0 μg mL−1; but notably the P(PEGMA)-GNP with the highly hydrophilic PEG chain, increased the dispersibility very significantly, reaching over 700 μg mL−1.
For HR-TEM, a JEOL ARM 200F was used, operating at 80 kV, with CEOS probe and image aberration correction. The dose was measured through the screen and was adjusted to ≈100 e− Å−2 s−1 for high magnification imaging. AFM force curves were acquired with an Asylum Research MFP-3D-SA using methods outlined by Lee et al.27 Multislice simulations were performed using spherical aberration −1 μm and defocus −20 nm (as measured for imaging). Simulations were performed using clTEM, an open-source multislice simulation package (M. A. Dyson, clTEM: OpenCL TEM/STEM simulation code, 2014). The multislice technique involves sectioning the structure that is to be simulated into thin slices normal to the direction of incidence of the electron beam. Then, the interaction between the incident electron wavefunction and one such thin slice can be modelled by multiplying the wavefunction with the transmission function obtained using only those atoms present in the thin slice. The wavefunction arriving at the sample is recursively multiplied by thin-slice transmission functions and then propagated through the sample to the next thin slice using Fresnel propagator functions modelling diffraction in a vacuum. The wavefunction propagating from the rear of the sample is the exit-wave, and a simulated image of this may be formed through convolution with a transfer function modelling the objective lens in the TEM, including its associated aberrations such as defocus, astigmatism, and spherical aberration. Alternatively, simulated diffraction patterns can be obtained by propagating the exit-wave via Fresnel diffraction to the imaging plane.25,33–35
SEM was performed on a LEO Gemini 1525 machine. Bright field TEM imaging was carried out using a JEOL2000 TEM, operating at 200 kV. AFM images were taken on a Nanoscope IV Digital Instruments AFM intapping mode. TGA-MS was performed using a Mettler TGA/DSC 1 coupled with a Hiden HPR-20 QIC R&D mass spectrometer. 5.0 ± 0.5 mg sample was heated from 30 °C to 100 °C at a ramping rate of 30 °C min−1 under N2 atmosphere (with a continuous flow rate of 100 sccm). The specimen was then heated from 110 °C to 850 °C at a 10 °C min−1 ramping rate. For dispersibility tests, GNPs were bath sonicated for 15 minutes (45 kHz, 80 W) in HPLC water, with an initial loading of 1 mg mL−1. The dispersed samples were left to stand overnight to allow the aggregates to settle. The supernatant concentration was determined by UV-Vis spectroscopy. Raman maps were obtained on a Renishaw InVia Confocal Raman spectrometer, using a green excitation laser (wavelength 532 nm, intensity 5%, scan time 30 seconds) under StreamlineTM mode (1800 lines per mm diffraction grating, at least 100 spectra taken for each sample). XPS spectra were recorded using a K-Alpha+ instrument equipped with an MXR3 Al Kα monochromated X-ray source (E = 1486.6 eV). X-ray gun power was set to 72 W (6 mA and 12 kV). Charge compensation was achieved using the FG03 flood gun using a combination of low energy electrons and the ion flood source. Argon etching of the samples was applied using the standard EX06 Argon ion source using 500 V accelerating voltage and 1 μA ion gun current. Survey scans were acquired using 200 eV pass energy, 1 eV step size and 100 ms dwell times. All high resolution spectra were acquired using 20 eV pass energy, 0.1 eV step size and 1 s dwell times. Samples were prepared by pressing the sample onto carbon based double sided tape. The pressure during data acquisition was ≤10−8 mbar. XRD was carried out using a PANalytical X'Pert3 PRO diffractometer operating at 40 kV and 40 mA. Dried powder samples were mounted onto a zero-background Si sample holder (PANalytical Ltd., UK). The diffractogram was recorded at a scan rate of 0.09° s−1 with the Cu Kα (λ = 1.542 Å) line, in the 2θ range from 5° to 60°.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6sc05603b |
This journal is © The Royal Society of Chemistry 2017 |