Spontaneous Intercalation of Graphene on Sapphire

ABSTRACT

The direct synthesis of graphene on dielectric substrates has attracted growing interest due to its potential for scalable, transfer‐free integration in electronic and photonic applications. However, graphene grown on dielectrics typically exhibits lower carrier mobility compared to copper‐grown counterparts, limiting its performance. Here, we report the synthesis of large‐area graphene on Al‐rich reconstructed c‐plane sapphire (0001) via chemical vapor deposition (CVD) and reveal that, over time and under ambient storage conditions, a spontaneous decoupling occurs at the graphene–sapphire interface. Raman spectroscopy reveals a reduction in both strain and doping in the aged samples, consistent with electrical transport measurements showing a twofold increase in carrier mobility. X‐ray photoelectron spectroscopy (XPS) and cross‐sectional transmission electron microscopy (cross‐sectional TEM) identify the intercalation of oxygen‐containing species at the interface as the mechanism responsible for the decoupling. These findings uncover a previously unrecognized pathway to enhance the electronic performance of directly grown graphene on sapphire, reinforcing the viability of this platform for future scalable graphene‐based technologies.

When left at room temperature, even in depressurized atmosphere, foreign molecular species can intercalate in between graphene and the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction, effectively decoupling the graphene from its substrate: $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$(0001). This is observed in the spectroscopic fingerprints, as well as in a large enhancement of the transport figures.

1. Introduction

Graphene has ignited significant interest in both fundamental scientific research and its potential for widespread technological applications [1, 2, 3, 4, 5]. While the promise of scalable technology is on the horizon, one formidable challenge remains: how to produce devices with performance levels comparable to those achieved by exfoliated graphene, which can be straightforwardly integrated into CMOS technology. Chemical vapor deposition (CVD) on copper has emerged as a leading method for large‐scale graphene production, boasting impressive electronic properties [6, 7]. However, this approach might induce a level of metal contamination which is not acceptable for back‐end‐of‐line (BEOL) integration [8, 9, 10]. Furthermore, the needed transfer step might induce inhomogeneities, tears, and wrinkles in the transferred atomically thin material. Steps forward have been done to address these challenges, yet the development of synthesis approaches on dielectrics yielding high‐mobility graphene represents a promising alternative. CVD graphene growth on sapphire is well‐suited for integration into CMOS lines and for specific applications eliminates the need for a transfer step [8, 11, 12, 13, 14, 15, 16, 17], despite maintaining the need for the fabrication steps for the definition of top‐gating structures. Indeed, graphene grown on sapphire has demonstrated great potential in diverse applications, such as light‐emitting diodes (LEDs) [18], biosensors [19], and as a gate dielectric [20, 21]. Additionally, it serves as a substrate for the growth of third‐generation semiconductors, including GaN and AlN, where graphene acts as a buffer layer, effectively mitigating lattice mismatch issues [18, 22, 23].

Despite its promise, continuous graphene films grown on sapphire exhibit good, but not exceptional transport properties. A key limiting factor is the lack of control over the interface between graphene and its substrate, which in turn influences the interaction and hence its properties, such as electronic and of adhesion. Indeed, a growing number of articles are now focused toward the control of the interface for an ever wider range of 2D materials and substrates [6, 24]. This limitation could be attributed to the difficulty in controlling the growth mechanism – as a consequence of reconstructing the sapphire surface in a homogeneous manner, which is why in some cases researchers are starting to terminate the c‐plane of sapphire with a chalcogen, when it comes to the growth of transition metal dichalcogenides [25]. The detailed mechanism by which graphene grows on sapphire via methane decomposition is not yet fully understood, and despite a few attempts [15, 26], there is still no clear demonstration of large‐scale single‐crystal graphene on $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$ that avoids the use of metal catalysis.

To this end, an appealing strategy could be that of adopting an Al‐rich reconstruction with ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 periodicity that develops upon H‐etching the $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$ prior to growth [8]. Graphene grows on top of that reconstruction, the atomic registry of which is described in Ref. [27, 28], while retaining an epitaxial alignment. However, the highest graphene mobilities measured for this system were limited to $\sim$2000 ${\mathrm{cm}}^2$ V⁻¹s⁻¹ [8].

In this work, we reveal that the properties of CVD graphene grown on H‐etched sapphire significantly change overtime, with an overall beneficial effect on graphene mobility. Indeed, even when left in depressurized atmosphere (0.1 mbar), we observe a modification in its spectroscopic signatures and morphology, and doubling of the charge carrier mobility. When analyzing the aged samples with cross‐sectional transmission electron microscopy (Cross‐sectional TEM), we clearly observe a lifting of the graphene layer from the substrate. Our investigations indicate that oxygen‐rich molecular species spontaneously intercalate at the heterointerface between graphene and the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction, effectively decoupling graphene from the substrate. Microscopic studies indicate that the intercalation process induces morphological modifications at the graphene surface. This work points to interface engineering as a viable strategy for tuning the electronic properties of graphene on sapphire, potentially enabling significant progress in the development of scalable graphene‐based technologies.

While the present study focuses on graphene grown on Al‐rich reconstructed sapphire, aging‐induced interfacial spontaneous intercalation and partial decoupling have been observed on other systems [29] and are likely to occur on other interfaces, representing a more general phenomenon for graphene on oxide and dielectric substrates with dynamic surface chemistry and hydrophilic character, with sapphire constituting a particularly illustrative case.

2. Results and Discussion

Graphene was synthesized on c‐plane sapphire previously subjected to H‐etching, a process that not only reveals the terrace morphology of the substrate but also induces an Al‐rich ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction on $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$(0001) [8] (see Experimental Section for further details). This synthesis approach is particularly appealing to obtain large‐scale single‐crystal graphene as the grown honeycomb lattice retains an epitaxial registry with the reconstructed sapphire substrate (see sketch in Figure 1a). The grown material was first characterized via Raman spectroscopy, a widely used technique to assess the quality and thickness of graphene [30]. In Figure 1b, we report the average spectrum recorded on graphene on sapphire, right after the growth (light grey) and after 1 year in depressurized atmosphere (dark gray).

FIGURE 1.

(a) Sketch of sample making and of the aging process. (b) characteristic Raman spectra of ${\mathrm{Gr}}_{\mathrm{fresh}}$ and ${\mathrm{Gr}}_{\mathrm{aged}}$. in the inset: 2D peak FWHM. (c) Position of the 2D peak (Pos(2D)) as a function of Pos(G). (d) Transport figures of ${\mathrm{Gr}}_{\mathrm{fresh}}$ and ${\mathrm{Gr}}_{\mathrm{aged}}$ samples.

A quick comparison already reveals that the level of strain inhomogeneity of graphene improves overtime, as the width of the 2D peak decreases. Such reduction can be clearly appreciated by looking at the histogram in the inset, where we compare the FWHM of the 2D peak of the as‐grown sample, versus the aged sample, on which the FWHM is almost halved. Also, the increase of the 2D/G ratio and the shift of the G peak position to lower wavenumbers are indicative of a concurrent doping reduction upon aging. Further insight can be gained by looking at the so‐called correlation plot in panel c, where we display the position of the 2D peak as a function of the position of the ${\mathrm{E}}_{2g}$ or G mode at each measured point. According to the model reported in [31], this plot provides an estimation of the level of doping and strain, following two main trajectories, indicated by the dashed lines labeled with $\sigma$ and $\epsilon$, respectively. The plot confirms that ${\mathrm{Gr}}_{\mathrm{fresh}}$ has a higher level of doping than ${\mathrm{Gr}}_{\mathrm{aged}}$. In addition, the spread along the strain line suggest the presence of a distribution of strains in the as‐grown graphene. The ${\mathrm{Gr}}_{\mathrm{aged}}$ instead, appears to collocate itself much closer to the nominal values for neutral and strain‐free graphene, indicated by the black cross (although the value is strictly valid only for graphene on ${\mathrm{SiO}}_2$/Si). Thus, Raman analyses indicate that the crystalline registry of the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction induces a certain degree of strain in the as‐grown graphene, which is released upon aging, as seen in Figure 1c. In addition, the Al‐rich nature of this reconstruction likely contributes to initial doping of the graphene [8, 14, 15, 26], an effect that also diminishes with aging.

However, it should be noted that in aged samples we do not only observe promising concurrent doping and strain reduction but also the occasional appearance of a small D peak (as reported in Figure 1b, see also Figure S1 for additional spectra), which is indicative of the emergence of local defects that we identify later in the manuscript.

The low electronic interaction between sapphire and graphene suggested by the Raman investigation is corroborated by transport measurements, carried out in van der Pauw geometry and reported in panel d. It is apparent that the ${\mathrm{Gr}}_{\mathrm{aged}}$ samples exhibit a level of charge doping about an order of magnitude lower than the ${\mathrm{Gr}}_{\mathrm{fresh}}$ samples, i.e., reaching values as low as 2.5 × 10 ${\mathrm{cm}}^{-2}$ starting from doping levels of $\sim$2.5 × 10 ${\mathrm{cm}}^{-2}$. Correspondingly, the mobility on the ${\mathrm{Gr}}_{\mathrm{aged}}$ samples is found to be $\sim$5000 ${\mathrm{cm}}^2$ V⁻¹s⁻¹, about twice as high as in the as‐grown samples (cf. also Figures S2 and S3).

Angle‐resolved photoemission spectroscopy (ARPES) data (Figure S5) qualitatively reveal the same trend, attributing half the doping level to the ${\mathrm{Gr}}_{\mathrm{aged}}$ sample. The quantitative discrepancy can be attributed to the measurement being carried out in UHV condition, i.e., in absence of adsorbed water molecule, which can strongly $p$‐type dope the material. The indications derived from Raman and transport analysis speak for an initial material subject to doping and strain. Aging appears to reduce those values in such a way to allow the mobility in graphene to roughly double. To gain a deeper insight on the detailed phenomenology of the aging, we investigated the samples by several surface science analysis methods. XPS measurements have been carried out on three different types of samples: the bare substrate, the as‐grown Gr/Sapphire and the aged Gr/Sapphire after a storage period in depressurized atmosphere. Figure 2 shows how the O 1s and Al 2p core levels evolve during these different stages. The bare substrate shows a prominent single oxygen component, which is assigned to the bulk $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$, and a small high energy component, which is referred to adsorbates on the surface. Al 2p is described by a single component as well (panel b). The process for attaining graphene on the (0001) plane, modifies the interface by terminating the $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$ with an Al‐rich ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction. This is visible in both O 1s, where a new oxygen component emerges at about ‐1.4 eV from the bulk component, describing the different coordination of the oxygen atoms at the interface, and also in Al 2p, where we see now a low‐oxidation state component associated with the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 surface termination having essentially the same energy shift, as also discussed in Ref. [32]. On aged samples, the bulk component in the O 1s core level becomes extremely weak, while in the case of the Al 2p signal, we see only the signal coming from the interface component. We model this phenomenological finding by assuming the interface to become much thicker than it was in the as‐grown case and draw a sketch for each stage in the inset of each panel. In addition, the morphology of graphene becomes irregular and corrugated (cf. Figure S10), contributing to a strong decrease of any signal coming from the substrate and the interface. The presence of additional interfacial atomic and molecular species is supported also by the fact that the interfacial Al 2p component moves toward higher binding energies by about 380 meV, testifying a partial re‐oxidation of the Al‐rich termination of the (0001) plane. XPS data, together with TDS measurements (cf. Figures S7a and S8) and KPFM (cf. Figure 4) strongly suggest that the majority of the molecules intercalated at the interface are air, with a dominance of water vapor.

FIGURE 2.

XPS data of (top to bottom) bare sapphire, as‐grown Gr/Sapphire and aged Gr/Sapphire. Intensity of Al 2p and O 1s become weaker with time due a layer between graphene and sapphire.

FIGURE 4.

Kelvin probe atomic force microscopy (KPFM) analysis. (a,b) Topography image taken of ${\mathrm{Gr}}_{\mathrm{fresh}}$ and compared with ${\mathrm{Gr}}_{\mathrm{aged}}$ (1 year), (c,d) Corresponding surface potential measured of ${\mathrm{Gr}}_{\mathrm{fresh}}$ and ${\mathrm{Gr}}_{\mathrm{aged}}$. (e) Comparison of work function values obtained via Equation (1) of Experimental Section.

The effect of partial re‐oxidation of the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction can be observed as well via LEED, as we show in Figure S6, where the diffraction spots stemming from the crystalline order of the interface in aged samples is replaced by strongly diffused intensity.

In our XPS analysis, we did not observe the emergence of signal from elements other than C, O, and Al at any stage. Such a fact invites us to conclude that the most likely scenario is that oxygen‐rich molecules penetrate over time at the heterointerface between the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction and graphene.

We can try to verify this by heating the sample up to a temperature at which those molecules start to be desorbed. We carried out such an experiment in a thermal desorption spectroscopy (TDS) fashion. In Figures S7 and S8 of the Supporting Information, we show how heating the sample above 550 allows oxygen and oxygen‐rich molecules to leave the interface, in some cases by burning the graphene, hence creating holes in the graphene's lattice.

The most compelling method for verifying the modification of the interface is to observe it using cross‐sectional TEM at high spatial resolution. Figure 3 presents ABF‐STEM images of the cross‐sections of four different sample conditions, each prepared according to its respective processing step. The zone axis of the imaging was defined with respect to the sapphire substrate, and detailed atomic models corresponding to the STEM images are provided in Figure S9. Figure 3a shows an image of a bare substrate prior to both graphene growth and H‐etching. Figure 3b displays an image of an H‐etched substrate before graphene growth, where a distinct contrast is observed at the topmost surface of the sapphire, indicative of a reconstructed surface structure induced by H‐etching. Figure 3c presents a cross‐sectional image obtained immediately after graphene growth. A single layer of graphene was formed on the reconstructed surface, and the average distance between them was measured to be 0.38 nm. Figure 3d,e show cross‐sectional images acquired from different regions of the same aged sample. In numerous areas of the heterointerface between graphene and the sapphire substrate, aging‐induced decoupling is clearly observed. This decoupling results in a separation of a few nanometers, i.e., about an order of magnitude greater than that in as‐grown graphene/sapphire samples. Figure 3d shows that due to the lower interaction with the substrate, rippling of graphene appears, particularly in correspondence of step edges. AFM data reported in Figure S10 also confirm the topographical modification at nanometer scale, with evident ridges arising upon aging of the sample, a plausible cause for the increased D peak revealed via Raman spectroscopy, together with the fact that those lifted defective sites can become more chemically active. We performed EELS analysis to obtain the fine structural information on the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstructed surface formed by H‐etching (Figure 3f). Al ${\mathrm{L}}_{2,3}$ energy‐loss near‐edge structure (ELNES) spectra were acquired from the locations marked by boxes in Figure 3a–b, corresponding to the surface before and after H‐etching, as well as from an interior region away from the surface (bulk). The ELNES spectra of the pristine surface and the bulk exhibited very similar features. In contrast, the ELNES of the H‐etched surface showed the formation of a shoulder preceding the peak observed at 79.5 eV in the other two conditions, along with an overall reduction in intensity. $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$ is theoretically known to have Al atoms coordinated to six surrounding O atoms [33]. According to the literature, when Al is coordinated to four O atoms in a tetrahedral configuration, a new peak appears at an energy‐loss value approximately 1.5–2.0 eV lower than the peak associated with octahedral coordination to six O atoms [34, 35, 36]. Therefore, our ELNES results suggest that, in the ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstructed surface, the Al coordination number undergoes a structural change from octahedral to tetrahedral. This trend is consistent with recently reported coordination changes accompanying structural transitions at reconstructed $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$ surfaces [28].

FIGURE 3.

Cross‐sectional annular bright‐field (ABF) STEM images of (a) bare substrate, (b) H‐etched substrate, (c) as‐grown graphene/sapphire, and (d, e) aged graphene/sapphire. All images were acquired from sapphire along the [2$\overline{\mathrm{1}}$ $\overline{\mathrm{1}}$0] zone axis. (f) Electron energy loss spectra of pristine surface of sapphire (acquiring from the area of the blue box on (a), H‐etched surface (the green box on b), and bulk region (the black box on b). The atomic models on the right show the tetrahedral structure when Al is coordinated with four O atoms and the octahedral structure when coordinated with six O atoms. The scale bars are 2 nm on (a–c) and 5 nm on (d,e), respectively.

The gap observed in cross‐sectional TEM, together with the analysis of the TDS data returns a density of the interfacial layer of the order of 10 ${\mathrm{cm}}^{-3}$, i.e. close to the density of a gas in SPT conditions.

We can get a first insight into the kinetics of the decoupling process by looking at the scanning Kelvin probe microscopy measurements, which we report in Figure 4. For reference, we report to topographic signal in panels a and b, while in c and d we display the corresponding surface potential.

Two different surface voltage populations are observed on the as‐grown samples (Figure 4c and reported as a red plot in panel e. Darker areas, corresponding to $\mathrm{Wf}\approx$4.65 eV in (e), are found mostly in the presence of sapphire step edges and graphene grain boundaries (Figure 4c). Conversely, in correspondence of flat terrace areas, where the adhesion to the substrate can be assumed as stronger, the peaks that can be addressed to the graphene layer appear to be centered around Wf = 4.55 eV. It is worth noting that this value is in accordance with the theoretical ones reported for intrinsic graphene. Aged samples instead, show an enhanced degree of homogeneity (Figure 4d), with the surface potential peaks convoluted around a common value of $\approx$ 4.75 eV (black trace in panel e). The measurements have been carried out in an Argon saturated atmosphere in order to minimize spurious voltage contributions given by atmospheric contaminants. From these measurements (cf. also Figure S11), we can infer that the evolution of the intercalation process, starts at those sites where the interaction with the substrate is already reduced, i.e. substrate's step edges and thermal stress ridges. Thus, the intercalated surface grows with time resulting in an overall free‐standing graphene sheet. The 200 meV shift observed in SKPFM is also compatible with the $\approx$150 meV observed in ARPES (see Figure S5 of SI).

The combined spectroscopic and microscopic analyses reported in this work clearly indicate that the interaction between graphene and sapphire is spontaneously released at room temperature thanks to a spontaneous intercalation process taking place upon aging of the sample. Since controlling this effect on shorter timescales would be ideal, we attempted to reproduce the spontaneous decoupling of graphene from sapphire by either immersing the samples in deionized (DI) water for a few hours or exposing them to a hydrogen atmosphere. Figures S11 and S12 report the Raman analysis performed on samples treated using these two approaches. The spectroscopic data indicate a promising effect, although the decoupling is not as effective as in naturally aged samples.

3. Conclusion

In this work, we demonstrate a promising pathway to engineer and improve the properties of graphene epitaxially grown on sapphire. Using Raman spectroscopy and transport measurements, we show that when graphene is grown on the the Al‐rich ($\sqrt{\mathbf{31}}$ $\times$ $\sqrt{\mathbf{31}}$)R9 reconstruction of $\alpha {\mathrm{Al}}_2{\mathrm{O}}_3$(0001), a surface previously identified as a promising template for epitaxially aligned graphene on sapphire [8] the interaction with the metallic reconstruction comes at a cost, manifested in terms of strain, doping, and reduced carrier mobility. Upon natural ageing of the sample, both Raman scattering and transport measurements reveal a strong improvement in the defining figures of graphene and indicate a decoupling from the growth substrate. The carrier mobility is at least doubled, while the charge carrier density is lowered by one order of magnitude and the strain relieved.

To gain deeper insight into this process, we employed various surface analysis techniques such as XPS, Cross‐sectional TEM, STM, and KPFM which indicate that aged samples undergo a spontaneous intercalation with oxygen‐rich species lifting the graphene–sapphire interface. As a result, the graphene layer is lifted by a few nanometers, effectively weakening its interaction with the substrate. We have tested alterative approaches such as immersion in DI water or hydrogen intercalation to promote decoupling. Preliminary results are encouraging, suggesting additional viable paths to accelerate the spontaneous decoupling. By demonstrating carrier mobilities as high as 5800${\mathrm{cm}}^2$ V⁻¹s⁻¹, charge carrier densities in the low 10 ${\mathrm{cm}}^{-2}$ range, and nearly strain‐free conditions, the natural aging of graphene on Al‐reconstructed sapphire opens new perspectives for engineering substrate–graphene interactions and optimizing material performance for next‐generation electronic and photonic applications.

4. Experimental Section

4.1. Synthesis

Monolayer graphene was grown on sapphire, following the procedure described in our earlier publication [8]. $\alpha$‐${\mathrm{Al}}_2\mathrm{O}$ (0001) $\pm 0.{25}^{\circ }$, one side polished substrates were supplied by Silian optoelectronics (China). Prior to the growth, the substrates were cleaned with acetone, isopropanol (IPA), and rinsed in deionized (DI) water in an ultra‐sonicator bath. In order to remove organic contaminants, the samples were treated in piranha solution (3:15, ${\mathrm{H}}_2\mathrm{O}$ :${\mathrm{H}}_2\mathrm{SO}$ ) for 15 min. They were finally washed and sonicated in DI water and dried with ${\mathrm{N}}_2$ gas. The H‐etching was performed in a cold wall vertical reactor at 1180 for 5 min in an atmosphere of molecular hydrogen [37], with the aim of regularizing the sapphire surface into an ordered step–terrace structure and simultaneously chemically reducing the surface, which terminates with an Al‐rich reconstruction [8]. The H‐etched substrates were then annealed at 1130 for 10 min in 8 mbar of Ar atmosphere at 1000 sccm. The synthesis of graphene was performed in the same reactor by introducing 150 sccm of ${\mathrm{H}}_2$ and 5 sccm of ${\mathrm{CH}}_4$, while keeping the 1000 sccm of Ar flow for 13 min, at 750 mbar of total backround pressure. After the graphene growth step, H and CH gas flows were terminated and the heater switched off. The sample was cooled down to 200 under Ar flow (1000 sccm). Below 200, the system was allowed to cool‐down naturally. The sample was unloaded from the chamber once the temperature decreased below 100. In this work, “large‐area” denotes uniform graphene coverage over the full growth substrate, independent of the chip size; although the present measurements were performed on $\sim 1\times$1${\mathrm{cm}}^2$ samples, the approach is scalable to much larger substrates (up to 6''), as previously demonstrated.

The samples were kept in a desiccator (10 Pa) for a period of 1 year and measured from time to time to check the changes in graphene properties. Aged graphene (${\mathrm{Gr}}_{aged}$) was compared with as‐grown graphene (${\mathrm{Gr}}_{fresh}$). We point out that we didn't choose a low‐pressure atmosphere on purpose and keeping the samples at atmospheric pressure might even speed up the intercalation process.

To assess the reproducibility of the aging‐induced intercalation process, more than 15 graphene samples ($\sim$1${\mathrm{cm}}^2$) from different growth batches, all prepared using the same growth recipe and stored under identical conditions, were analyzed, all exhibiting consistent aging behavior.

4.2. Fabrication

The graphene transport properties were measured fabricating Hall bar devices directly on sapphire substrate through electron beam lithography (EBL), reactive ion etching (RIE) and thermal metal evaporation. EBL was performed with positive resist PMMA (poly‐methyl methacrylate), baked at a temperature of 120 for 5 min. The electron beam was accelerated at 20 kV in a Zeiss Ultra‐Plus SEM system. The exposure dose was set to be 280 $\mu \mathrm{C}$ ${\mathrm{cm}}^{-2}$. The first lithography step consists in the definition of the metallic contacts on the graphene: EBL was used to define the contacts and sequential metal evaporation in a commercial Sistec thermal evaporator (base pressure 10 mbar) of 50 nm of gold over 5 nm of chromium, used as an adhesion layer. In a second lithographic step, we defined the pads and metallic contacts; to better assure adhesion on the substrate, we first etch the graphene below the contacts and then evaporate gold over chromium, using the same lithography mask. The last lithographic step consisted in the definition of the graphene by etching the graphene around the contact area and around the metallic pads and connections, through 1 min of RIE plasma (5/80 sccm Ar/${\mathrm{O}}_2$), at 35 W of power, with chamber pressure of 2.8 e‐1 mbar.

4.3. Characterization Techniques

4.3.1. Raman Spectroscopy

Raman measurements were performed with a Renishaw Invia system with a 532 nm laser at a fluence of about 10 mJ $\mu \mathrm{m}$ . The graphene quality on sapphire were assessed in different regions of the sample. The Raman data presented in Figure 1 and Figure S1 are based on maps collected over a region of 28$\times$28 $\mu \mathrm{m}$ with a 1 $\mu \mathrm{m}$ grid step, corresponding to 784 spectra.

4.3.2. Atomic Force Microscopy (AFM)

Atomic force microscopy (AFM) measurements were carried out with a Bruker Dimension Icon operated in tapping mode in air and the Gwyddion software package was used to analyze the collected micrographs.

4.3.3. Low‐Energy Electron Diffraction (LEED)

Low‐energy electron diffraction (LEED) measurements were performed at room temperature with a SPECS ErLEED 150 instrument.

4.3.4. X‐Ray Photoelectron Spectroscopy (XPS)

X‐ray photoelectron spectroscopy analyses of the graphene/sapphire were carried out on a Thermo Fisher Scientific Escalab 250 xi, equipped with a monochromatic Al‐$\mathrm{K}\alpha$ anode (1486.61 eV).

4.3.5. Angle‐Resolved Photoemission Spectroscopy (ARPES)

Angle‐resolved photoemission spectroscopy analyses of the graphene/sapphire were carried out on a Specs Flex system, equipped with a micro‐Sirius He plasma UV source and an Astraios 190 with 2D CMOS‐CCD detector.

4.3.6. Transport

Electrical characterization had been performed using a Keithley 2450, in a four‐probe configuration. The contact on the device had been made through 20 $\mu \mathrm{m}$ tungsten tips, mounted on MPI MP‐40 micro‐positioner, in a home‐made probe‐station setup. The mechanical stability of the system was guaranteed by positioning the setup on a stabilized optical table. The IV measurements had been performed sweeping the applied current between source and drain contacts, measuring the potential drops into adjacent sense contacts. The Hall measurements had been made at constant source/drain current of 1 $\mu \mathrm{A}$, measuring the potential variation between two front‐facing contacts while varying the applied magnetic field through an electromagnet (used as the sample stage) calibrated to act between $\pm$32.2 mT with an applied current of $\pm$1 A.

4.3.7. Kelvin‐Probe Force Microscopy (KPFM)

We carried out Amplitude‐Modulation Scanning Kelvin Probe Force Microscopy (AM‐SKPFM) measurements either in ambient air or in Ar‐saturated conditions using an XE‐100 Park atomic force microscope and NSC14 Cr/Au‐coated conductive cantilevers (MikroMasch). In our SKPFM measurements, we operated in dual‐frequency mode, simultaneously capturing both the surface height and potential profiles during scanning. The potential signal was concurrently demodulated using an external Stanford Research System SR830 DSP Lock‐in amplifier with an AC reference signal with a frequency of 18 kHz and amplitude ranging between 1.8 and 1.9 V. The SKPFM technique revolves around measuring the contact potential difference between a sample and a conductive AFM tip brought into close proximity to it [1, 8, 37]. The work function difference between the two materials translates into an electrostatic force component, perturbing the oscillation condition of the tip as it scans the surface in non‐contact mode. An internal voltage feedback loop nullifies the electrostatic contribution point‐wise by applying an opposite voltage interpreted as $V_{surf}$. The latter can be expressed in terms of the work function of the surface being analyzed (${\varphi }_{surf}$) and that of the conductive tip (${\varphi }_{tip}$) as:

$$V_{surf}=\frac{{\varphi }_{tip}-{\varphi }_{surf}}{e}$$ (1)

The ${\varphi }_{tip}$ is initially obtained using Equation (1) by calibrating the Au‐coated tip on a freshly cleaved highly ordered pyrolytic graphite (HOPG) surface, which is regarded as a stable measurement standard with ${\varphi }_{HOPG}$ = 4.65 eV [2]. In the context of the proposed experiments, typical value for ${\varphi }_{tip}=4.87\pm 0.03$ eV.

4.3.8. Scanning Tunneling Microscopy (STM)

Scanning tunneling microscopy (STM) measurements were carried out at room temperature on a Omicron LT‐STM.

4.3.9. Residual Gas Analysis (RGA)

Residual gas analysis (RGA) spectra were recorded with a Stanford Research RGA350 quadrupole.

4.3.10. Cross‐Sectional TEM specimen preparation and STEM/EELS Analysis

For cross‐sectional transmission electron microscopy (TEM) analysis, a TEM lamella was prepared using a focused ion beam (FIB; Thermo Fisher Scientific, Helios NanoLab 450). Cross‐sectional scanning TEM (STEM) imaging was performed using an aberration‐corrected TEM (FEI Titan G2 60–300) operated at an accelerating voltage of 200 kV. Atomic‐resolution images were obtained through the annular bright‐field (ABF) STEM imaging under conditions of a convergence semi‐angle of 26.6 mrad, a collection semi‐angle of 8–32 mrad, and a probe current of approximately 60 pA. Electron energy‐loss spectroscopy (EELS) was conducted at the same accelerating voltage (200 kV) using a sub‐nanometer electron probe in STEM mode for localized analysis. EEL spectra were collected from more than ten locations at equivalent depths from the surface, and the average results are presented.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smtd70545‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.