Interface Chemistry of Graphene/Cu Grafted By 3,4,5-Tri-Methoxyphenyl

Chemical reaction with diazonium molecules has revealed to be a powerful method for the surface chemical modification of graphite, carbon nanotubes and recently also of graphene. Graphene electronic structure modification using diazonium molecules is strongly influenced by graphene growth and by the supporting materials. Here, carrying on a detailed study of core levels and valence band photoemission measurements, we are able to reconstruct the interface chemistry of trimethoxybenzenediazonium-based molecules electrochemically grafted on graphene on copper. The band energy alignment at the molecule-graphene interface has been traced revealing the energy position of the HOMO band with respect to the Fermi level.


and (b) 3,5bis-tert-butylbenzenediazonium (TBD).
Electrochemical surface modification of graphene was carried out during cyclic voltammetry. During the reduction process the diazonium molecules release the N 2 and they graft the graphene surface as 3,4,5-trimethoxyphenyl (TMeOP) and 3,5-bis-tert-butylphenyl (TBP) units. Figure S2 shows cyclic voltammogram of treated graphene on copper, of 5 mM and 1 mM TMeOP-grafted units, and of 5 mM TBP-grafted units. In Figure S3, the wide core level photoemission spectra collected on Gr/Cu foil, treated Gr/Cu foil and on the two Gr/Cu substrates grafted with TMeOP-grafted on Gr/Cu foil and Gr/Cu(111) and TBPgrafted on Gr/Cu foil are shown. Three peaks ascribed to photoemission from Cu 3d, Cu 3p and Cu 3s levels of the copper substrate are detectable, along with the C 1s and O 1s core levels. The binding energies (BE) of Cu 3s, Cu 3p and Cu 3d lines are 122.6 eV, 75.1 eV and 3.1 eV, respectively. For C 1s BE= 284.6 eV while for O 1s BE= 530.6 eV. To evaluate the contribution of graphene (G), as first we estimate the contribution of the grafted molecule Q mol from the integrated intensity of the methoxy group (O-CH 3 ) contribution divided by six, being six the number of C atoms related to the oxygen in in the molecule (i.e. 3 C atoms for the O-CH 3 group and 3 C in the ring). Q mol then corresponds to the "#$ % weight. Therefore, we have subtracted from the C sp 2 signal ( "#$ & ) the sp 2 contribution coming from the benzene ring of the molecules. For the TMeOP units, this contribution is obtained multiplying by 3 the signal of Q mol , as 3 out of 6 C atoms in the ring are sp 2 bonded.
Then, to estimate the contribution of the number of graphene unit cells, we have summed the value of G and "#$ % in order to obtain the contribution of all C atoms in the graphene layer. Then, as the number of C atoms in the unit cell of graphene is 2, the number of graphene cells (N cell ) is: Finally, the molecular density (i.e. the number of TMeOP units per unit cell) can be estimated as the ratio between N cells , and Q mol resulting in one molecule per few (3 ± 1) graphene unit cell for the case of the 5 mM solution.
This result is also consistent (within the approximation of the calculation) with the estimation obtained from the intensity ratio between the D and G bands in the Raman spectra 1 . In the high density (HD) condition 2 , it is possible to estimate the inter defect distance L D as: where E L is the laser photon energy (for He-Ne, E L =1.96 eV) and I D / I G the intensity ratio between the D and G Raman bands.
Considering that, as reported in the manuscript, after the electrochemical process the I D /I G becomes 0.92 for the treated sample and 1.4 for TMeOP-grafted one, we estimate L D = (17.6 ± 2.3) nm for the treated in the low density regime and L D = (1.07 ± 0.1) nm in the high density condition for the grafted sample.
The latter (L D =1.07) yields a density of 1 molecule per 4.5 ± 0.5 graphene unit cell. This result is consistent with the molecular density estimated by the photoemission measurements. The assumption to use two different regimes for the treated and grafted sample was supported from the STM measurements carried out on the treated sample ( Figure S4) where no defect seems to be present in the 11.5 x 11.5 nm 2 STM image. As reported in the manuscript, the C 1s has been interpolated by considering the C sp 2 , the O-C=O contamination, the p-p* shake up and for TMeOP -grafted graphene the contribution of the methoxy group (O-CH 3 ). To prove that the feature at 533.2 eV is due to the methoxy groups and not to oxygen contamination, in Figure S5 the C 1s and the O 1s XPS spectra for two different concentrations of the molecules (1 mM and 5 mM) in the electrochemical solution are shown. In both cases, the O-CH 3 contribution (purple peak in C 1s and orange peak in O 1s) increases with the molecule concentration, confirming that this feature could be taken as a fingerprint of molecule grafting.

Figure S5: Core level spectra of TMeOP-grafted on Gr/Cu at different concentration. (a) C 1s and (c) O 1s spectra of 1 mM TMeOP-grafted on Gr/Cu(111). (b) C 1s and (d) O 1s spectra of 5 mM TMeOP-grafted on Gr/Cu foil.
To understand the origin of the O 1s in the pristine graphene on copper, we consider the XPS spectra in the region of the O 1s XPS core level peak, at about 530 eV (Figure S6 a). The two structures observed in the spectrum are at 530.5 eV, ascribed to the CuO x layer between graphene and the copper substrate, and at 532.5 and 533.9 eV, due to the air contamination (C-O) species and H 2 O, respectively 3-6 . The origin of this peak has been proved by collecting the O 1s XPS spectrum also at 30° off-ormal emission, thereby increasing the surface sensitivity of the XPS probe. As expected, the ratio between the oxygen peaks at BE=532.5 eV and BE=533.9 eV (C-O and H 2 O) and the component ascribed to CuO x , duplicates with respect to the spectrum collected at normal emission (Figure S6 b). Figure S7 shows the O 1s core levels collected on HOPG. All the samples on HOPG show a single structure. As expected, the O 1s intensity is negligible for all the samples except for TMeOP-grafted on HOPG where a clear O 1s core level is present at about 533 eV 7 due to the methoxy groups (or O-CH 3 groups). We note that the contribution at 530.5 eV, present on the spectra collected on Gr/Cu is completely missing in HOPG samples, confirming that this oxygen component can be related to oxygen atoms placed between graphene and Cu. By comparing this value with the ones reported in literature 8 where the O 1s in CuO and Cu 2 O is at 530.5 eV, in Cu(OH) 2 + OHat 532 eV and in H2O at 533.0 eV, we can assume that the contribution at 530.5 eV comes from the Cu-Ox layer at the interface between graphene and copper 9 .   Figure S8 shows the ARPES data acquired at the K point of the first Brillouin zone (BZ) of the pristine graphene on copper foil and of the sample grafted with 1 mM and 5 mM of TMeOP-grafted molecules on Gr/Cu(111) and Gr/Cu foil, respectively. In the pristine Gr/Cu foil, a Dirac cone appears downshifted with respect to the Fermi level, confirming that the CuO x layer decouples the graphene from the substrate. Due to the high coverage, the ARPES image on 5mM of TMeOP-grafted Gr/Cu foil is completely dominated by the HOMO and HOMO-1 bands. We note that for this study, both Gr/Cu foil and Gr/Cu(111) have been used as a substrate for the grafting procedure even if, in principle, the two interfaces could be different. As previously discussed, the presence of Cu-Ox layer at the interface between graphene and copper, mainly due to the gentle annealing treatment, decouples the copper substrate from graphene, making the latter more free-standing and comparable the two interfaces. To further confirm this point, in Figure S9, the ARPES data collected at the G and the K-point of the first Brillouin zone are shown. At the G point, the Cu d band and the graphene s band dispersion are at 2-4 eV and 4 eV, respectively, while the graphene Dirac cone appears at the K-point at the Fermi level as for a free-standing (undoped) graphene.

Figure S10: Raman spectrum of pristine Gr/Cu(111).
Moreover, in Figure S10 the Raman spectrum collected on pristine Gr/Cu(111) is shown. The D/G band intensity ratio (I D /I G = 0.07 ± 0.04) is similar to that of Gr/Cu foil (I D /I G = 0.105 ± 0.06 as shown in Figure 2 of the manuscript) confirming that the quantity of defects and the degree of disorder of the graphene layers grown on both interfaces are comparable.