A versatile route to edge-specific modifications to pristine graphene by electrophilic aromatic substitution

Graphical Abstract Electrophilic aromatic substitution produces edge-specific modifications to CVD graphene and graphene nanoplatelets that are suitable for specific attachment of biomolecules Electronic supplementary material The online version of this article (10.1007/s10853-020-04662-y) contains supplementary material, which is available to authorized users.


S1. Synthesis of nitrographene and aminographene
Nitrographene (G-NO2, Scheme S1) was synthesised by adding 100 mg GNPs to an ice-cold mixture of 20 mL 70 % nitric acid and 28 mL concentrated sulfuric acid. The reaction mixture was stirred and left at room temperature for 21 h. The mixture was poured over ice and neutralised in 5 M aqueous NaOH solution to pH 7. The solids were isolated by vacuum filtration through a 0.45 µm HV membrane filter (EMD Millipore), washed with deionised water and freeze-dried.
Aminographene (G-NH2, Scheme S1) was synthesised by decanting a mixture of 50 mg of nitrographene and 3 ml of aqueous Raney nickel suspension into 40 mL methanol at room temperature. Then, 0.76 g of sodium borohydride was slowly added intro the mixture, keeping temperature between 30°C and 40°C. [1] The reaction stirred at room temperature for 24 h, vacuum filtered through a 0.45 µm HV membrane, washed with deionised water and freeze-dried. Scheme S1. Synthesis of nitrographene (G-NO2) by electrophilic aromatic substitution, and its subsequent reduction to aminographene (G-NH2).

S2.1 Fitting method
Raman spectra were fit in Origin based on the methods put forward by Puech and co-workers. [2] The D peak was fit to a double Voigt model (i.e., two Voigt peaks sharing a single centre point). The G and D′ peaks were fit to Voigt peak shapes. The separation between the G and D peaks was sufficient that no asymmetry needed to be included in the G peak shape.

S2.2 Equations used to estimate graphene domain size and defect distance
Puech and co-workers also provided a useful summary of equations used to estimate the domain size (La) in graphenic materials as well as the distance between point defects. [2] The classic equation from Tuinstra and Koenig [3] predicts the domain size based on the relative intensity of the D band: The more recent work from Cançado et al. [4] uses the relative area of the D band and accounts for the energy of the Raman excitation source (EL): The D peak's linewidth (the halfwidth at half maximum, HWHMD) also provides an estimate of the domain size: [5] La HWHM ≅ 300 nm cm   Table S1. Elemental concentrations (in at%) of pristine GNPs and their edge-modified analogues determined by XPS survey spectra ( Figure S3 and Figure S22a).  Table S2. Estimates of non-graphenic carbon from XPS sp 3 :sp 2 ratios in addition to those given in Table 2.

S4. XPS spectra and elemental analysis
Source spectra shown in Figure S3 and Figure S20.
Sample Source a XPS sp 3 :sp 2 atomic ratio 1.083 ± 0.092 G-NH 2 C750 9.482 ± 9.457 a NMP = graphite sonochemically exfoliated in N-methyl-2-pyrrolidone, C750, C300, M25 = GNP from XG Sciences. b No sp 2 carbon detected; (6.83 ± 2.29) at% of C 1s assigned as C-SO3.  Figure S4. Raman spectra of edge-modified C750 GNPs relative to the unmodified GNPs (λ=633 nm). Intensity is normalised to G peak intensity.    Peak potentials differences (∆Ep) were measured for each scan rate and droplet. Seven ∆Epresults were taken from varying the scan rate for each droplet. The resulting k 0 values are shown in Figure S11. There was a high apparent outlier in 8 of 9 droplets. Some scans recorded at 50 mV s −1 showed a smaller ∆Ep, but k 0 should be the same for all scan rates. [8] Therefore, the mean k 0 for each scan rate was calculated (Figure S12). A Grubb's test for outliers was run (α = 0.05) and the 50 mV s -1 results were disregarded ( Figure S13). The revised mean k 0 was (8.8 ± 7.1) × 10 −6 cm s −1 .      Figure S17. Representative thermogravimetric analysis traces from C750 GNPs and its edge-modified analogues recorded under N2 atmosphere. The white balance of the images was adjusted by setting the shadow to a neutral grey. Concentration: 1 g l -1 .   Figure S19. Comparison of the Raman spectra of GNPs produced by ultrasonic exfoliation in NMP versus commercially produced C750 GNPs. λ=633 nm (EL = 1.96 eV) for NMP samples and λ=514 nm (EL = 2.14 eV) for C750 samples. Table S5. Summary of Raman disorder metrics for the GNPs and their sulfur-functionalised analogues shown in Figure S4, Figure S5, Figure S19 and Figure S23. Arranged by modification type. Identical to Table  S6 but primarily classified by modification stage.    Figure S4, Figure S5, Figure S19 and Figure S23. Identical to Table S5 but primarily classified by modification type.

S-19
Figure S20. XPS survey scans (left column) and C 1s high-resolution scans (right column) of graphene produced by exfoliation of graphite in NMP and edge-modified analogues of them. The unmodified graphene sample has a fluorine contaminant with peaks consistent with poly(tetrafluoroethylene).

S11. Zeta potential measurements
Zeta potential measurements were run on the three water-dispersible forms of edge-modified C750 GNPs: G-SO3 -, G-NO2 and G-NH2. Figure S24 shows that these samples all had negative zeta potentials, ranging between -60 mV and -50 mV. A significant negative charge was expected for G-SO3 -, no significant charge was expected on G-NO2 and a positive charge was expected for G-NH2. The nitration step likely oxidised the graphene, consistent with deconvolution of the associated XPS C 1s spectra (Table S2), and produced charged groups similar to those found in GO. This effect was previously observed for HOPG treated in a strongly oxidising mixture of nitric acid and sulfuric acid. [9] The zeta potential for G-NH2 will also have contributions by residual nickel oxide from the reduction of G-NO2 with Raney nickel, consistent with measurements of the zeta potential of aqueous suspensions of nickel powders. [10]