Charge‐Transfer‐Controlled Growth of Organic Semiconductor Crystals on Graphene

Abstract Controlling the growth behavior of organic semiconductors (OSCs) is essential because it determines their optoelectronic properties. In order to accomplish this, graphene templates with electronic‐state tunability are used to affect the growth of OSCs by controlling the van der Waals interaction between OSC ad‐molecules and graphene. However, in many graphene‐molecule systems, the charge transfer between an ad‐molecule and a graphene template causes another important interaction. This charge‐transfer‐induced interaction is never considered in the growth scheme of OSCs. Here, the effects of charge transfer on the formation of graphene–OSC heterostructures are investigated, using fullerene (C60) as a model compound. By in situ electrical doping of a graphene template to suppress the charge transfer between C60 ad‐molecules and graphene, the layer‐by‐layer growth of a C60 film on graphene can be achieved. Under this condition, the graphene–C60 interface is free of Fermi‐level pinning; thus, barristors fabricated on the graphene–C60 interface show a nearly ideal Schottky–Mott limit with efficient modulation of the charge‐injection barrier. Moreover, the optimized C60 film exhibits a high field‐effect electron mobility of 2.5 cm2 V−1 s−1. These results provide an efficient route to engineering highly efficient optoelectronic graphene–OSC hybrid material applications.

. Schematic illustration of polymer-doped graphene for C 60 growth template.
Graphene characterizations. The doping effects of polymers on graphene have been described previously. [1] Briefly, the charge-carrier density n g in polymer-contact doped graphene was 1.7 × 10 12 cm -2 with P4VP, 7.2 × 10 11 cm -2 with P4VPh, ~0 with PS, -1 × 10 12 cm -2 with PMMA and -2.7 × 10 12 cm -2 with PVC. C 60 deposition. Graphene/SiO 2 /Si was annealed at 350 °C in H 2 atmosphere for 1 h with slow cooling to remove the polymeric residues before further use. An organic molecular beam deposition (OMBD) system was used to deposit C 60 (Aldrich Chemicals, 99.99% purity) in ultra-high vacuum (UHV, 10 -8 Torr). The deposition rate, film thickness, and substrate temperature mentioned in the main text were the values recorded by the monitor. Throughout the experiment, the substrate was kept at 303 K, and C 60 was deposited onto it at a rate of 5 × 10 -2 monolayer per second (ML/s) unless stated otherwise.
To achieve the charge transfer control during the preparation process, we had to apply gate voltage to graphene during C 60 deposition. To do that, we deposited an Au electrode pad on a graphene that had been transferred onto a 300-nm-thick SiO 2 / p-doped Si wafer. Heavilydoped silicon was used as a bottom gate electrode. Then a electrical potential difference between the Au electrode and Si was generated using power sources. The voltage was controlled by connecting power sources in series and was checked before each C 60 deposition.
To prevent possible charge accumulation in the sample, we used a Cu wire to connect the Au electrode pad to the conducting part of OMBD system. For safety reason, the applied voltage was kept < 100 V. Figure S2. Graphene properties after deposition of 1 ML C 60 . a, UPS measurement in the SEC region of graphene with different charge carrier densities before (open) and after (filled) C 60 deposition. b, Schematic illustration of graphene-C 60 system for the following experiments: KPFM (top) and Raman mapping (bottom).

Discussion
Qualitative estimation of charge transfer between graphene and C 60 . Charge transfer between graphene and C 60 was qualitatively observed by tracing the changes in graphene properties by using ultraviolet photoelectron spectroscopy (UPS, 4D-beamline, Pohang Accelerator Laboratory (PAL), Korea), Raman spectrometry (Alpha300R, WITec, λ = 532 nm), and Kelvin probe force microscopy (KPFM) ( Figure S2). UPS measurements in the secondary electron emission region (SEC) were performed on graphene/polymer samples before and after deposition of 1 ML C 60 ( Figure S2a). After C 60 deposition, E F of graphene/C 60 was pinned at the lowest unoccupied molecular orbitals (LUMO) level of C 60 . Such pinning can be detected by the n g -independent SEC-spectrum of graphene/C 60 . The p-doping effect of C 60 on graphene was clearly demonstrated by the right-shifts of the UPS spectra. The degree of shift increased as n g increased; this trend means that as the E F of graphene approached the LUMO of C 60 , the (E F -E F, c ) increased, so the doping effect increased.
For Raman mapping and KPFM measurement, 1 ML C 60 was deposited through a shadow mask onto freshly-grown graphene/copper which was n-doped. The C 60 /graphene/copper was transferred using the conventional transfer process, then flipped over onto a substrate so that the upper surface was clean graphene ( Figure S2b). This structure ensures that the collected results were from graphene only, not from the C 60 layer. A KPFM image clearly showed the difference of V CPD between the clean graphene area and the graphene area with the underlying C 60 layer; this change implies that graphene's work function was shifted as coupled with C 60 .
For Raman mapping, we optimized the scan conditions so that the spectra of C 60 and graphene did not mix. The scan was performed over the area that we had examined using KPFM. The intensity ratio (I 2D /I G ) of the 2D peak (I 2D ) and G peak (I G ) revealed the doping intensity of graphene. The clean graphene area showed a very high I 2D /I G ~ 2.5, which indicates that the graphene was undoped. In the area with the underlying C 60 , I 2D /I G was significantly reduced; this result confirms that C 60 had doped the graphene.  Quantitative estimation of charge transfer between graphene and C 60 . Charge transfer between C 60 and graphene was quantified by measureing changes in resistance. Graphene's resistance without contact effects was measured using transfer-length method (TLM); and the changes in graphene channel resistance before and after 3 s of C 60 deposition at a deposition rate 5 × 10 -2 ML/s were used to calculate the charge transfer number ( Figure 1 in the main text and Figure S3, S4).
To minimize contact resistance R C , annealed graphene/SiO 2 /p-Si was exposed to UV-O 3 for 5 s right before depositing 100 nm of gold (Au) with shadow masks. Au was evaporated at a low rate of 0.2 Å/s. The shadow mask consisted of five channel lengths: 30, 80, 130, 180, and 280 µm. A second shadow mask was used with reactive-ion etching to achieve graphene stripes with 30 µm width. The measurements were conducted under high vacuum (10 -6 Torr).
Before C 60 deposition, the peak of R C was located at the Dirac voltage of a graphene FET; this result indicates that ∆ , > ∆ , in the range of measured V G . This relationship probably occurred because E F of graphene under a metal electrode was pinned by the metal.
This effect simplifies the interpretation of V G -dependent R C after C 60 deposition. Graphene regions under the metal electrodes were not affected by C 60 deposition, so the change in R C upon C 60 deposition was solely attributed to the change in ∆ , , i.e., ∝ ∆ , . As a result, the R C -V G curves of graphene transistor before and after C 60 deposition showed the same trend as the R Ch -V G curves of graphene transistors ( Figure 1b in the main text); this similarity confirms our results.  (2) mode.
Raman spectra of C 60 thin films. C 60 thin films of 20-nm thickness were deposited onto graphene with Δn CT ~ 0 which corresponds to lack of charge transfer, and onto graphene with Δn CT >> 0 which corresponds to presence of charge transfer. The laser's wavelength was 532 nm and was focused onto C 60 layer so that graphene's signal was phased out ( Figure S5).
The spectra were collected with low accumulation to limit damage to the sample. Figure S6. Crystal structure of C 60 thin films grown on graphene. GIXD patterns of C 60 thin films, 2 nm and 100 nm thick, grown on polymer-contact doped graphene.
Crystal structure of C 60 films grown on graphene. Crystal structure of C 60 films was characterized using grazing incident x-ray diffraction (GIXD, 3C and 9A beamlines in PAL) and transmission electron microscopy (TEM, HR-FE-TEM-2200FS with Cs correction).
For the general observation, both electrical-gate doped and polymer-contact doped graphene templates were used separately for growth of C 60 crystals. The crystal structure of C 60 and the trend of results on both kinds of graphene templates were similar ( Figure 2 in the main text and Figure S6); this observation confirms the effects of charge transfer on the crystal structure of C 60 films.
The coherent size of (111) domain was calculated using the Scherrer equation , where K is the dimensionless Scherrer constant,  is the incident X-ray wavelength,  is the full-width at half-maximum of the corresponding reflection, and θ Bragg is its Bragg angle.  outside the C 60 /graphene area, and let it naturally diffuse to the Au area under the C 60 /graphene area. Once the Au layer was dissolved, DI water was immediately but tenderly dropped on the sample to rinse the etchant. Then, C 60 /graphene was scooped using by a piece of polyethylene terephthalate (PET) and transferred to DI water for further rinsing. We scooped the floating sample on water by using a TEM grid (Lacey Carbon, copper, Ted Pella).
Lastly, the sample was dried in ambient air and stored in UHV before measurement ( Figure 3 in the main text and Figure S7).
To check whether the gold etchant damaged the C 60 /graphene samples, we compared the morphology and crystal structure of 1.25-ML C 60 /graphene/300-nm SiO 2 /Si before and after dipping the sample in gold etchant solution for 3 min. We performed AFM and GIXD characterizations on this sample in each condition; the morphological image and crystal structure patterns of the sample were almost the same. In fact, the AFM images which were taken in the exact same area showed similar nucleation density, islands positions and islands heights. Moreover, the GIXD patterns clearly showed the same fcc crystal structure with the same (111) domain size. Therefore, we confirmed that any damage to C 60 thin film by gold etchant is insignificant ( Figure S8). Morphology of C 60 films grown on graphene. The morphology was examined using atomic-force microscopy (AFM, Bruker) in tapping mode. The height, surface coverage and nucleation density were analyzed using NanoScope Analysis software.
For the general observation, both electrical-gate doped and polymer-contact doped graphene templates were used separately for growth of C 60 crystals. The morphology of C 60 and the trend of results on both kinds of graphene templates were similar ( Figure 4 in the main text and Figure S9); this result confirms the effects of charge transfer between graphene and C 60 , but not the other substrate effects, on the growth behavior of C 60 . Figure S10. Plot of ΔE Nuc vs. Δn CT at 0.25 ML thick of C 60 thin films grown on graphene

Effects of charge transfer on C 60 crystal's growth.
To quantify the dependence of C 60 growth on charge transfer between the C 60 ad-molecules and the graphene template, numerous C 60 thin films with nominal thickness of 0.25 ML were grown on graphene templates in which E F had been finely controlled using either electrical gating or polymer doping. The plot of the nucleation density N i of these films against Δn CT reveals the correlations between nucleation of C 60 and the charge transfer between graphene and C 60 ( Figure 4c in the main text). Here we introduce the term ΔE Nuc to represent the difference in activation energy for a C 60 ad-molecule to nucleate between at a significant Δn CT and at Δn CT ~0. Clearly, Δn CT determined N i and the activation energy for C 60 nucleation E Nuc ( Figure S10). E Nuc can be extracted using where C is a pre-exponential factor, k B is the Boltzmann constant, and T is the substrate temperature.
At room temperature, as Δn CT increased, N i increased, so E Nuc also increased, with a linear slope of 1.2 × 10 -13 eV/cm 2 ( Figure S10). We also directly measured E Nuc by analyzing the dependence of N i on T at different Δn CT . E Nuc was obtained as the slope of a plot of ln(N i ) vs.
1/(k B T) (Figure 4d in the main text). These results means that E Nuc increased with a linear slope of 1.3 × 10 -13 eV/cm 2 as Δn CT increased ( Figure 4e in the main text). The similarity of these two slopes ( Figure 4e in the main text and Figure S10) confirms that the dependence of N i on Δn CT is attributable to the dependence of E Nuc on Δn CT . Figure S11. Schematic illustration of a C 60 island and a C 60 ad-molecule on graphene substrate along with calculation details.
Mechanism of C 60 growth on graphene. When electrons in graphene are transferred to a C 60 island or a C 60 ad-molecule, an electrostatic repulsion develops between the island and the negatively-charged ad-molecule ( Figure S11). Additional electrostatic attractions develop between negatively-charged ad-molecules or islands, and positively-charged graphene beneath them. Here, we tried to calculate the additional attachment barrier E B ' that these electrostatic interactions induce.
Assuming the C 60 island is a circular plate, the electrostatic potential energy that corresponds to repulsion can be calculated as ) and the electrostatic potential energy that corresponds to attraction can be calculated as where the parameters are related to the geometry of the island ( Figure S9), and the integrations are done over the surface of the C 60 island. Then the total electrostatic potential energy of the system is simply 2( + ). Using dimensionless parameters = ⁄ , = ⁄ and = ⁄ , can be expressed as The ∆ -dependent attachment barrier is the limit of as approaches 1, The method to treat SiO 2 /Si substrate with ODTS has been described elsewhere. [4] The key of this transfer method is that PDMS contacts only the C 60 film and not the bottom graphene, so that the PDMS piece could lift ~50-nm-thick C 60 film up and place it on the target substrate. The rest of the C 60 film remained on graphene on the initial substrate. The whole transfer process was conducted in a glove box under inert N 2 atmosphere to avoid oxidation of C 60 . Although perfect transfer large-area C 60 films by the stamping method was not easy, we could efficiently obtain small (~1 cm 2 ) crack-free C 60 films. Then we fabricated channels (1000 µm × 50 µm) by depositing 100-nm-thick Al source and drain electrodes on these intact transferred C 60 films.
The measurements were conducted using a Keithley 2400 in vacuum (10 -6 Torr) to avoid oxidation.
The electron mobility µ e of the C 60 -OFET was extracted in the saturation region (V DS = 100 V) by using the equation: where I DS is the drain current, C i is the capacitance of the oxide, V G is the gate voltage and V T is the threshold voltage.
To fabricate a barristor, a 150-nm-thick layer of C 60 was deposited on gate-biased graphene on SiO 2 /Si substrate through a shadow mask. Then 100-nm Au was deposited onto graphene as a source electrode, and 100-nm Al was deposited on the C 60 layer as a drain electrode using shadow masks. The devices were characterized inside the vacuum probe station. To avoid affecting the delicate morphology of C 60 film, the measurement temperature was varied from 223 to 303 K ( Figure S12). Interfacial states of C 60 thin film confirmed by photocurrent measurements. To measure the charge de-trapping effects, we fabricated graphene/C 60 transistors ( Figure S13a).
The thickness of C 60 layers was 20 nm, grown with or without charge-transfer effect. The channel size was 10 µm × 1000 µm. The measurement was performed in a vacuum chamber (10 -6 Torr) equipped with an optical fiber connecting to a monochromator. The beam energy was 0.62 eV, V G was 80 V, V DS was 0.05 V. Bottom: x-z plan view). DOS of C 60 dimers with c, structure of (C 60 ) 2 and d, 2C 60 .
Density function state (DFT) calculations of C 60 molecules. Initial single C 60 structure was adopted from the fcc crystal C 60 . Then 16-Å vacuum slaps in all Cartesian coordinates were set for the C 60 dimer before relaxation to avoid interaction of the periodic lattice. To determine the configuration of bonded C 60 dimer, each C 60 was initially positioned with a distance of 3.3 Å and fully relaxed to its lowest energy. To compare the crystal and electronic structure of C 60 dimer before and after bonding, we also calculated the structure of isolated C 60 dimer; a 9 Å distance was set for isolated C 60 . We abbreviate the bonded dimer as (C 60 ) 2 and the isolated 2 C 60 as 2(C 60 ).
The relaxed structure and the corresponding density of states (DOS) were calculated using Quantum Espresso 6.3.0 ver., [5] and Plain Augment Waves, Optimized Norm-Conserving Vanderbilt pseudopotentials from the PAL library.7. [6] The cut-off energies and k-grid for relaxed structure and DOS were set to ecutwfc = 40 Ry for Gamma point and 60 Ry for 3 × 3 × 1. The convergence threshold on force was set to 1.0D-6 for calculation of optimized structures. A small scissors-operator of 0.6 eV was applied.