Probing charge transfer between molecular semiconductors and graphene

The unique density of states and exceptionally low electrical noise allow graphene-based field effect devices to be utilized as extremely sensitive potentiometers for probing charge transfer with adsorbed species. On the other hand, molecular level alignment at the interface with electrodes can strongly influence the performance of organic-based devices. For this reason, interfacial band engineering is crucial for potential applications of graphene/organic semiconductor heterostructures. Here, we demonstrate charge transfer between graphene and two molecular semiconductors, parahexaphenyl and buckminsterfullerene C60. Through in-situ measurements, we directly probe the charge transfer as the interfacial dipoles are formed. It is found that the adsorbed molecules do not affect electron scattering rates in graphene, indicating that charge transfer is the main mechanism governing the level alignment. From the amount of transferred charge and the molecular coverage of the grown films, the amount of charge transferred per adsorbed molecule is estimated, indicating very weak interaction.

. During the annealing, the back-gate electrode was grounded and V SD was fixed to 10 mV. (c) and (d) respectively show AFM topography images of the rim of the graphene channel prior and after the annealing (also indicating rms roughness σ of graphene), while (e) shows the overlayed height histograms of (c) and (d).
Figure S1 presents the effects of the annealing in high vacuum on one of the graphene devices used in this study. As can be seen from Fig. S1a, the amount of unintentional doping is reduced allowing the CNP to be reached within the safe scanning range of V SG . However, the device still remains slightly p-doped (E F = -(88 ±5) meV). A similar trend of Fermi level shifts was observed for most of the samples. Furthermore, the height of the graphene/SiO 2 step edge was found to decrease by (0.3-0.6) nm upon annealing. The root mean square surface roughness of graphene also decreased upon annealing by ∼10%, both indicating the removal of the surface contamination layer.
Unfortunately, the annealing did not always result in more stable or less doped devices. After the annealing, some of the devices were even further doped, or had unstable transfer curves (not repeating subsequent transfer curves or even forward and backward sweeps). These devices were not used for the growth experiments. The negative effect of the annealing in some cases is most likely related to releasing of surface contaminations that were localized near the device, which then contaminated the surface of the channel.

Influence of the surface contaminations and annealing on the growth morphologies of 6P.
In order to highlight the impact that the residues remaining on the surface of the devices have on the growth of molecular semiconductors, 6P was grown at T D = 365 K with a surface coverage of ∼0.6 ML. This chosen T D is higher than the ones used in the main manuscript (300 K), resulting in the growth of larger crystallites and thus emphasizes the impact of surface contaminations on nucleation and growth. As shown in Fig. S2, three different graphene surfaces were used: an as-fabricated graphene device with top surface contaminated with UV photolithography residues, a device annealed in high vacuum (the same device is also shown in Fig. S1), and a pristine as-exfoliated graphene flake which serves as a representative for the uncontaminated surface. In all cases, 6P molecules lay flat on graphene, forming needle like crystallites with their herringbone layers perpendicular to the surface of graphene. The striking difference in the nucleation density of the crystallites depending on the surface treatments clearly reveals the necessity of the annealing in order to ensure good contact between the molecules and graphene and to minimize the possibility that the charge transfer is mediated by contaminations.
As can be seen from Fig. S2, the annealing in high vacuum can not recover the pristine surface of as exfoliated graphene, however, 6P molecules grown on the annealed devices tend to form significantly longer needle-like crystallites with clear tendency to follow crystallographic directions dictated by the symmetry of the underlaying graphene lattice. Longer annealing times, or higher annealing temperatures, or annealing in H 2 /Ar atmosphere could help to restore pristine surface of graphene -yet risking to further dope the devices -and can result in unstable electrical properties due to formation of the defects with the graphene lattice and covalently bonded species.
Morphology of 6P and C 60 films grown on graphene FETs. Ex-situ AFM topography images of the samples have been used to determine the total volume of the deposited molecules on the active surfaces of graphene FETs. Considering the most likely bulk structures for both, 6P and C 60 for the chosen growth conditions, the total number of molecules has been estimated. Unfortunately, under the chosen growth conditions neither 6P nor C 60 experiences an ideal layer-by-layer growth on the surface of graphene FETs. Figure S3 presents two typical cases of the resulting growth morphologies of 6P and C 60 layers grown on graphene FETs (T D = 300 K).
In the case of C 60 (Fig. S3a,b), a rather uniform film thickness was obtained across the entire surface of graphene, while on the surface of SiO 2 3D structures were found. The thickness of the grown films was expressed in the equivalent volume of the complete mono-layers considering that each layer of C 60 should add ∼0.8 nm of the total height and that the initial height of graphene was measured prior to the growth. In the case shown in Fig. S3b, two complete mono-layers have been grown on the surface of the device, with an incomplete third and fourth layer on top. This sums up to ∼3 ML covering the entire  device. However, edge decorations and several higher structures have also been observed, and considering their total volume gives (4.3 ±0.9) ML of C 60 coverage for the particular case presented in Fig. S3a,b (also in Fig. 4a-c of the main manuscript). While C 60 can be considered as a zero-dimensional building block, 6P on the other hand will form different crystallites depending on the orientation of the molecule with the respect to the substrate plane. If the molecules are flat-lying then 3D needle-like crystallites are formed. If the molecules are nearly up-right standing (∼2.6 nm), island-like crystallites grow. Both types of crystallites have the same monoclinic β -phase structure. The total volume of 6P has been considered for each sample and has been expressed as the equivalent of the complete covered mono-layers of up-right standing molecules. Figure S3c-e shows an example of 6P grown on the sample presented in Fig. 1d,e and Fig. 2a-c of the main text. In this case, 6P islands cover 25% of the device surface. However, needle-like crystallites also contribute significantly to the total volume of the deposited 6P. If these are also taken into account, the volume totally deposited would be equal to (0.8±0.2) ML. Surrounding SiO 2 had a coverage of (0.6±0.1) ML of 6P (Fig. S3e). This is in a good agreement with the total volume of 6P deposited on graphene considering that the sticking coefficient of the molecules is not the same in these two cases and that for needle-like crystallites -found on graphene -there is a larger overestimation of the volume due to the tip-sample dilation.

The calculation of the position of the Fermi level in graphene
The position of the Fermi level in graphene was calculated considering a parallel plate capacitor between graphene and SiO 2 , and linear approximation of the dispersion relation of graphene. The area specific capacitance of the gate oxide was considered as C = ε 0 ε r /d. d stands for the thickness of the dry thermal SiO 2 (ε r = 3.9). The thickness of the oxide was determined by ellipsometry measurements and was found to be d = (80 ± 2) nm. The expected variation of the oxide thickness introduces an uncertainty of less than 2% for the determined position of the Fermi level. Thus it can be ignored, since a several times larger uncertainty is introduced by the measurements of the transfer curves (I SD (V SG )). The density of the electrons in the capacitor is obtained as n = CV SG /e. On the other hand, considering a linear approximation for the dispersion relation of graphene: E(k) =hv F k, where v F stands for he Fermi velocity of electrons in graphene (10 6 m/s), the density of electrons can be expressed as: n = E F 2 /πh 2 v F 2 . This gives the relation between the position of the Fermi level and the applied voltage at the gate as: E F =hv F ((πε 0 ε r /e) · (V CNP /d)) 1/2 , where V CNP stands for the V SG which is needed to reach the minimum current in the transfer curves. Considering the known thickness of SiO 2 and the Fermi velocity in graphene yields a rather simple estimation for the position of the Fermi level as: E F = ±60.5 · (V CNP [V ]) 1/2 meV .