betaFeedback

WorksPhysics and Astronomy

Out-of-plane interference effects in the angle-resolved photoemission of epitaxial graphene

Abstract

The epitaxial growth of monolayer graphene by Si sublimation of 6H-SiC(0001) is one of the most effective ways of producing large-scale monocrystalline graphene samples of high quality. However, graphene is left interacting with a reconstructed carbon buffer layer known as zerolayer graphene (ZLG) and the SiC substrate underneath. Beyond the mere creation of Dirac cone replicas, the effect of this interaction on photoemission measurements is poorly understood. We utilized synchrotron radiation for photon-energy-dependent angle-resolved photoemission spectroscopy (ARPES) to investigate the complex final-state interactions of photoelectrons within the heterostructure of monolayer graphene, ZLG, and SiC. We identify multiple resonances in the photoemission intensity of the Dirac cone as a function of photon energy that originate from interlayer interference, confirmed with a multiple scattering model. We also demonstrate for the first time that the relative intensity of the replica bands induced by the (63×63)R30∘(6\sqrt{3} \times 6\sqrt{3})\text{R}30^\circ reconstruction scales with the electron kinetic energy. Finally, we resolved four nondispersing electronic states, which originate from the ZLG. Our findings show that the three-dimensional (3D) structure of SiC and ZLG manifests itself in the photoemission from two-dimensional (2D) epitaxial graphene. By revealing how 3D substrate architecture dictates the photoemission of a 2D layer, this work establishes a crucial framework for investigating the electronic structure of complex heterostructures via ARPES.

Introduction

The emergence of two-dimensional (2D) materials, especially graphene, has revolutionized the field of condensed matter physics and opened new possibilities for tunable materials by stacking to form heterostructures [1, 2]. Arguably the best platform for producing large-scale, high-quality graphene samples is by thermal decomposition of silicon carbide (SiC) [3, 4]. This method can produce monolayer graphene (MLG) at a wafer scale. However, underneath the MLG lies another, (63×63)R30∘(6\sqrt{3}\times6\sqrt{3})R30^\circ reconstructed, carbon layer. This layer has the typical honeycomb structure of graphene, but is in part covalently bonded to the SiC. Therefore it does not possess graphene’s typical π\pi-band structure and is called zerolayer graphene (ZLG). The ZLG can interact with and strain the MLG [5]. Together the MLG, ZLG, and SiC form a complex three-dimensional (3D) heterostructure. To understand this system, it is important to measure the electron energy dispersion.

In angle-resolved photoelectron spectroscopy (ARPES), understanding the matrix element effect on measurements is of the utmost importance to effectively study a material [6]. It is well understood how the matrix element depends on properties of the material such as orbital symmetry, atomic locations, and material thickness as well as the choice of photon energy hνh\nu, polarization, and angle of incidence to the sample [7, 8, 9]. The best known example of this effect in graphene is the so-called dark corridor effect, in which pseudospin suppresses the photoemission intensity in parts of the Fermi surface [10, 11]. Work was done to leverage the matrix element by varying polarization to allow for the full visibility of the Fermi surface [12, 13]. Furthermore, recent studies have demonstrated that hνh\nu tuning is critical for modulating the intensity distribution around the K‾\overline{\mathrm{K}} and Γ‾\overline{\mathrm{\Gamma}} points, providing a pathway to investigate the final-state effects in 2D systems [13, 14, 15]. However, while these models effectively capture the intrinsic symmetry of the graphene lattice, they typically ignore how the presence of the substrate will affect the photoemission process. A significant knowledge gap remains in how the 3D structure of the MLG/ZLG/SiC system fundamentally changes the matrix element of the final-state electron. Additionally, previous hνh\nu-dependent studies [13, 14, 15] have generally been restricted to limited energy ranges, which may fail to resolve high-energy resonances or the full evolution of many-body features.

In this paper, we address these gaps by performing systematic ARPES measurements over a wide photon energy range of 20 eV to 200 eV allowing a better understanding of 3D substrate-driven final-state effects. By combining ARPES results with multiple scattering (MS) simulations we aim to isolate the intrinsic matrix element effects of MLG from 3D scattering. First, we establish photoelectron intensity resonances due to constructive interferences within the substrate scattering path and relate them to out-of-plane layer distances. Second, we investigate the dependence of the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ replica band intensity on hνh\nu and demonstrate that a simple explanation related to the inelastic mean free path (IMFP) is insufficient. Third, we identify the optimal photon energy for resolving electron-phonon coupling in MLG. Finally, we report four interfacial states arising from the bonding of the ZLG to SiC. Our findings reinforce that an understanding of final-state effects is a prerequisite for the accurate determination of electronic properties of supported 2D materials.

Experiment and numerical methods

Sample preparation

Single-crystalline 6H-SiC(0001) wafer pieces were used for the epitaxial growth of MLG. The substrate was first cleaned using a wet etching process. The sample was then placed in an induction heated graphite susceptor where it was brought to a temperature of 1400 °C under a flow of hydrogen for 20 min. This process results in a clean substrate of unit-cell-high steps with terrace widths on the order of 1 µm [16]. The substrate was then heated to 1420 °C for 15 min under a flow of argon to induce graphitization via Si sublimation up to MLG coverage [3]. Figure 1(a) shows a schematic side view of MLG which rests on top of the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ reconstructed ZLG which is weakly bonded to the underlying SiC. Due to the covalent bonding between the ZLG and the underlying SiC, the ZLG exhibits a characteristic buckling. Ab initio calculations by Varchon et al. [17] and Kim et al. [18] predict vertical displacements between 0.34 Å and 1.2 Å, while grazing incidence fast atom diffraction measurements by Zugarramurdi et al. [19] report an average corrugation of 0.27 Å. A value of 0.8 Å was used in Figure 1(a) for visual clarity.

Figure 1. (a) Schematic ball-and-stick model (side view) of the sample crystal structure showing the MLG and ZLG layers and the SiC substrate. The ZLG exhibits vertical buckling due to partial covalently bonding to the substrate and the MLG is interacting with the ZLG via van der Waals forces. The nominal layer separation distances are set to dSiC−ZLG=2.00d_{\mathrm{SiC-ZLG}} = 2.00 Å and dZLG−MLG=3.24d_{\mathrm{ZLG-MLG}} = 3.24 Å [20]. (b) AFM image of a 20×2020 \times20 µm2^2 region of the sample showing uniform terraces. (c) LEED image taken at 80 eV with reciprocal lattice vectors for graphene (red), SiC (white) indicated and a diamond highlighting typical spots of the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ reconstructed ZLG (orange).

The sample was first checked with atomic force microscopy (AFM) [Fig. 1(b)] to determine the quality of the growth process. AFM shows the typical step-bunching observed in the growth process and reveals terraces of 2 to 4 µm width. Further characterization was carried out after transferring the sample into an ultrahigh vacuum (UHV) environment and degassing with a resistive heater at 750 °C for 30 min. When observed with low-energy electron diffraction (LEED) [Fig. 1(c)], the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ surface reconstruction of the ZLG can be seen as well as the first order diffraction spots of the MLG and SiC. LEED data were taken with the sample normal to the incident electron beam, using a beam energy of 80 eV.

Experimental procedure

The sample was transferred in air to beamline 5-2 (BL5-2) at the Stanford Synchrotron Radiation Light-source (SSRL) where it was loaded into UHV and degassed at 300 °C for 1 h. ARPES energy-momentum cuts were taken through the Dirac point at K, perpendicular to the KΓ‾\overline{\mathrm{K}\Gamma} direction [Fig. 2(a)] using a Scienta Omicron DFS30 electron analyzer with a vertical entrance slit. The sample was held at 150 K and a pressure was less than 3×10−113 \times10^{-11} torr for the duration of the measurements. Measurements were made using pp polarized light with photon energies in the range from 20 eV to 200 eV in 2 eV steps. As sketched in Figure 2(b), the sample was rotated from θ≈55∘\theta\approx55^\circ to 11∘11^\circ relative to the spectrometer axis to remain at the K‾\overline{\mathrm{K}} point as the photon energy was changed. While the low photon-energy light was almost normal to the sample, the angle of incidence was gradually increasing as the sample was rotated towards the analyzer with increasing hνh\nu. For constant energy plots, an electron deflector was used as opposed to rotating the sample away from K‾\overline{\mathrm{K}}.

Figure 2. (a) Surface Brillouin zone of graphene in which the orientation of the kk-space cuts for photon-energy dependent ARPES is indicated by the red line. (b) Sample orientation relative to the electron analyzer and the incident light. The analyzer slit direction is perpendicular to the plane of light incidence.

Computational Methods

Final-state plane wave and trLEED

To simulate ARPES photoemission intensity, the electron dipole transition matrix elements were first computed in the nearly free electron or final-state plane wave (FSPW) approximation. As shown in Ref. [7], this model uses a single plane wave to model the final state of a photoelectron and is effective at capturing orbital characteristics and the geometric phase interference of electrons emitted from different atomic positions within the crystal lattice. The initial states were determined from a tight-binding model of unstrained MLG with first nearest neighbor hopping parameters reported in [21], alongside a monolayer work function of 4.16 eV [22]. Computations were performed using the chinook python package by Day et al. [23]. The photoemission intensity was calculated using pp-polarized light with a photon-energy step size of 1 eV.

In Day’s implementation of the FSPW model, the dipole operator is implemented in the length gauge, and the matrix element for each orbital index aa, MaFSPW(k⃗,ω)M_a^{\mathrm{FSPW}}(\vec{k},\omega), is evaluated as a function of the electron wave vector k⃗\vec{k} and incident photon energy ω\omega using a spherical-harmonic expansion of the atomic orbitals [23]. Because the final state is approximated as a plane wave eik⃗⋅r⃗e^{i\vec{k}\cdot\vec{r}}, the dipole matrix element reduces to a Fourier transform of the initial state atomic orbitals. Since the Fourier transform of the spherical harmonics is known, the software uses a table of precalculated values to avoid the computationally expensive real-space integrals [23]. The total matrix element is expressed as a weighted sum of the precalculated terms which allows for the program to be very computationally efficient.

To approximate a more realistic final state while preserving computational efficiency, we included a simple model of time-reversed LEED (trLEED) states [24] that captures reflection of the photoelectron from the surface potential of the MLG. The trLEED final state inside a material has two components: a transmitted wave (kz>0k_z > 0) and a reflected wave (kz<0k_z < 0). To bypass full real-space zz-integration, we approximate this state using a complex coefficient β(E)\beta(E) that accounts for the resulting constructive and destructive interference. By matching the internal wave function with the external vacuum solution at the surface boundary (z=0z = 0) for an electron propagating through an effective monolayer depth z0z_0, we define

β(E)=eikzz0+αe−ikz′z0,(1)\beta(E) = e^{ik_z z_0} + \alpha e^{-ik'_z z_0}, \tag*{(1)}

where

α(E)=kz−kz′kz+kz′.(2)\alpha(E) = \frac{k_z - k'_z}{k_z + k'_z}. \tag*{(2)}

Here, the reflection amplitude α\alpha, Eq. (2), is modeled as reflection from a step potential with complex inner potential V=−V0+iViV = -V_0 + iV_i and electron kinetic energy EE [25, 26], where kz=1ℏ2me(Ecos⁡2θ+V0)k_z = \frac{1}{\hbar}\sqrt{2m_e(E\cos^2\theta+ V_0)} represents the wave in the material and kz′=1ℏ2meEcos⁡2θk'_z = \frac{1}{\hbar}\sqrt{2m_e E\cos^2\theta} represents the free wave in vacuum which behaves like eikz′ze^{ik'_z z}. For our calculations V0V_0 and ViV_i were treated as modeling parameter with initial values set to 16.4 eV [27] and 0 eV respectively. At z0=0z_0 = 0, this simplifies to the original FSPW model up to the energy-dependent reflection factor 1+α(E)1 + \alpha(E).

This complex amplitude β(E)\beta(E) acts as an overall phase and attenuation factor for the final state wave function. Since this trLEED modification is implemented by multiplying the plane-wave final state by β(E)\beta(E), it simply scales each term in chinook’s precalculated sum. Expressing the kinetic energy as E(ω)=ℏω−ϕ−EBE(\omega) = \hbar\omega- \phi- E_B, the modified matrix element is written as

MatrLEED(k⃗,ω)=β(E(ω))MaFSPW(k⃗,ω),(3)M_a^{\mathrm{trLEED}}(\vec{k},\omega) = \beta(E(\omega))M_a^{\mathrm{FSPW}}(\vec{k},\omega), \tag*{(3)}

where ϕ\phi is the material’s work function and EBE_B is the binding energy.

This approach successfully captures out-of-plane interference and phase-shifting effects without introducing real-space numerical integrals over zz, completely retaining the original computational speed of the FSPW approximation. The resulting photoemission intensity is evaluated as

I(k⃗,ω)∝∣∑aMatrLEED(k⃗,ω)∣2.(4)I(\vec{k},\omega) \propto\left|\sum_a M_a^{\mathrm{trLEED}}(\vec{k},\omega)\right|^2. \tag*{(4)}

Real-space multiple scattering

In a second approach, the ARPES spectra were computed in the one-step model of photoemission with trLEED boundary conditions, using the real-space multiple scattering (MS) method by Krüger et al. [28]. Unlike the FSPW and the simple trLEED approximations used in the previous section, this model describes the photoelectron wave as the sum of a plane wave and spherical waves incoming at each atomic site and relies on a finite-size cluster of atoms in contrast to the periodic in-plane boundary conditions used in chinook.

A free-standing MLG sheet was modeled with a disk-shaped cluster of 348 carbon atoms and 1185 empty cells. These empty cells are used for describing the spatial extension of C ss and pp orbitals beyond the radius of the C atomic spheres and are needed to properly represent the anisotropic graphene potential within the muffin-tin approximation [29]. The potentials of the atomic carbon and empty spheres were determined for the infinite graphene layer using the linear-muffin-tin orbital software in the atomic sphere and local density approximation [30].

The matrix elements of the transition dipole operator are evaluated in acceleration gauge [26]. In this form the dipole operator is approximately ε⃗⋅∇V\vec{\varepsilon}\cdot\nabla V, where ε⃗\vec{\varepsilon} is the polarization vector of the incoming photon and VV is the crystal potential. This is a convenient choice in the muffin-tin approximation as inside the atomic spheres ∇V=(dV/dr)r^\nabla V=(dV/dr)\hat{r} and ∇V=0\nabla V=0 in the interstitial region [31]. Because of this, the interstitial region does not contribute to the transition matrix. This computational approach was shown to be very reliable for ARPES intensity calculations of freestanding MLG [15, 29]. The photoemission intensity was calculated as a spherical tensor of the polarization vector, which were then used to reconstruct the experimental p-polarization. The calculations

were performed with a photon-energy step size of 1 eV.

Results and Discussion

Out-of-plane effects on intensity resonances

To probe substrate interactions, we utilized photon-energy scanned ARPES to resolve final-state scattering enhancements across the MLG/ZLG/SiC heterostructure. The orientation of our measurements relative to the MLG Brillouin zone is shown with dashed lines in the constant energy map of Figure 3(e) and in Figure 2(a) with a solid red line through the K point. Figure 3(a) displays the corresponding EE vs. kxk_x dispersion at hν=40h\nu=40 eV, where the Dirac bands of the MLG are resolved. The horizontal dashed lines in Figure 3(a) indicate three representative binding energies at which the intensity was tracked as a function of hνh\nu as presented in Figure 3(b)-(Figure 3d). To isolate the spectral weight of the π\pi bands, a numerical mask was used (Appendix A). This captures the total cross section of the dispersing bands, excluding the background without distorting the lineshape or intensity. The intensity values shown in Figure 3(b)-(Figure 3d) represent the kxk_x-integrated sum of the masked signal. This extraction was completed systematically across ten binding energies, three of which are highlighted here.

Figure 3. (a) Characteristic ARPES scan at hν=40h\nu= 40 eV through the K point with dashed lines at E−EF=−0.25E-E_{\mathrm{F}}=-0.25, −1.25-1.25, and −2.25-2.25 eV, indicating the binding energies used for the intensity plots in (b)-(d). (b)-(d) Photoemission intensity as a function of photon energy for experimental data, FSPW, trLEED, and MS simulations. For the experimental data, background intensity was removed as outlined in Appendix A. Multiple Gaussian distributions were fitted to determine resonant peak positions for all but the FSPW model where a Lorentzian distribution was used. (e) ARPES constant energy surface taken at E−EF=−1.25E-E_{\mathrm{F}}=-1.25 eV, with a red line indicating the cut direction in (a) and dashed white lines indicating the boundaries of the first Brillouin zone. (f) Average photon energy position of the resonant peaks, determined from ten binding energy cuts. The error bars represent the standard deviation of the peak positions, with a minimum floor set to 1 eV.

The key result of this paper is the multiple matrix element modulations observed in the plot of photoemission intensity with respect to hνh\nu, as shown by the blue curves in Figure 3(b)-(Figure 3d). These photoelectron intensity modulations are different from the so-called scattering resonances originating from initial state effects that do not depend on photon energy [32, 33, 34]. The precision of the extracted resonant energy positions from the experimental data in Figure 3(b)-(Figure 3d) is limited by the 2 eV photon-energy step size. This may result in slight shifts in the extracted peak positions; however, we do not expect these to significantly affect the conclusions drawn from their locations.

To determine the physical mechanism that is responsible for these enhancements, we first implemented a FSPW simulation of photoemission from a freestanding monolayer of graphene [black dashed line in Figure 3(b)-(Figure 3d)], as schematized in Figure 4(a). As can be seen in Figure 3(f) this model replicates the first peak. This method also correctly captures orbital characteristics such as the dark corridor [15], but fails to capture any other hνh\nu-dependent features. The failures of this model are consistent with past results [14, 15, 35, 36].

Figure 4. Cartoons of the (a) final-state plane wave approximation, (b) time-reversed LEED state, (c) real-space multiple scattering, and (d) interlayer interference. The red circles represent the C atoms, and the black wavy arrow represents the emitted photoelectron. In (a), we see an electron that does not interact before reaching the detector. (b) shows an electron and its time-reversed LEED state which can interfere before reaching the detector. The gray region represents the inelastic mean free path of the electron within the MLG. In (c), the electron can scatter (elastically) off neighboring muffin-tin potentials (gray circle) as implemented in the real-space MS model [28]. (d) shows two electrons emitted from the same location scatter off two different layers of the system. This results in interlayer interference that is sensitive to the kinetic energy of the electrons.

Next, we tested a simple trLEED extension to the model, Figure 4(b), which includes a wave component traveling into the monolayer. By modifying the FSPW model to include this reflected wave component, the interference between the incoming and outgoing waves is captured with a single phase factor. The results of these calculations are shown in orange in Figure 3(b)-(Figure 3d). The trLEED model is able to qualitatively reproduce the first three peaks when using an effective emission depth of z0=6z_0 = 6 Å and an inner potential of V0=16.5V_0 = 16.5 eV [20]. Within this model the locations of the second and third peaks can be shifted to higher or lower hνh\nu by varying the inner potential. The number of visible peaks can also be increased by increasing the effective emission depth. However, the additional peaks are periodic which does not reflect the experimental results. Furthermore, the model requires a large effective emission depth of 6 Å to even qualitatively match the experimental results, which is physically inconsistent with the thickness of a freestanding graphene monolayer (3.35 Å) [37]. Ultimately this model fails to reproduce the experimental results but shows that the addition of a reflected wave and finite z0z_0 component of electron propagation can introduce intensity oscillations.

Since the simple models fail to capture many of the features observed in the experiment, a more sophisticated approach is necessary. A MS model [28] is implemented for freestanding monolayer graphene. In this model, Figure 4(c), the emitted electrons are scattered off neighboring atomic potentials. The resulting phase shifts allow for new constructive interference conditions that produce additional hνh\nu-dependent intensity peaks. The MS calculation, as presented in the red curves of Figure 3(b)-(Figure 3d), accurately predicts the location of many of the peaks found in the experiment data [Figure 3(f)], indicating that in-plane scattering of photoelectrons within the graphene lattice [Figure 4(c)] is the primary perturbation of the matrix element in this energy range. The polarization dependence of the calculated intensity resonances is also investigated using the MS simulations, with comparisons for ss-polarized and circularly polarized light provided in Appendix D. Figure 3(f) shows that the MS model effectively tracks the experimental peak positions with a few notable disagreements: there are three experimentally observed peaks at hν=27±1h\nu= 27 \pm1, 38 ±\pm 1, and 100 ±\pm 1 eV that are not present in the MS data. The difference between the experimental data, the MS model data and the large emission distance (z0=6z_0 = 6 Å) needed for our trLEED model strongly indicates that the source of these additional final-state scattering enhancements must originate from out-of-plane interactions that are not captured by the freestanding graphene models.

With freestanding MLG simulations unable to reproduce all observed resonances, we turn to a structural analysis of the heterostructure. To determine if these additional peaks come from interlayer photoelectron interference [Figure 4(d)], we map the experimental data from hνh\nu to the electron wave vector k\mathbf{k}. This is done using the standard free-electron final state approximation accounting for the inner potential V0V_0, binding energy EBE_B, and work function ϕ=4.16\phi= 4.16 eV [22]. Here the wavevectors ki\mathbf{k}_i inside the material and ko\mathbf{k}_o outside the material are given by

ki=2me(hν−ϕ−EB+V0)ℏk_i = \frac{\sqrt{2m_e(h\nu-\phi-E_B+V_0)}}{\hbar}

and,

ko=2me(hν−ϕ−EB)ℏk_o = \frac{\sqrt{2m_e(h\nu-\phi-E_B)}}{\hbar}

with the corresponding resonance condition expressed as

2kix1−kox2=2πn,2k_i x_1-k_o x_2=2\pi n,

where x1=d/cos⁡θix_1=d/\cos\theta_i and x2=2dsin⁡θotan⁡θix_2=2d\sin\theta_o\tan\theta_i. Here, dd is the vertical separation between two lattice planes and θi\theta_i and θo\theta_o represent the angle of the electron propagating relative to the surface normal inside and outside of the material. This can then be further simplified to,

2dkicos⁡θi=2πn.(5)2dk_i\cos\theta_i=2\pi n. \tag*{(5)}

The derivation of this expression is detailed in Appendix § 12.

For our calculations, V0V_0 was optimized to 15.95 eV, which is close to the expected value for MLG [20]. Now applying the above condition, we find that the three experimental peaks at hν=27±1h\nu=27\pm1, 38 ±1\pm1, and 100 ±1\pm1 eV correspond to vertical distances of d=3.49±0.06d=3.49\pm0.06, 1.97 ±0.03\pm0.03, and 0.610 ±0.003\pm0.003 Å, respectively. These values correspond to interference conditions for the MLG/ZLG separation, the ZLG/SiC interfacial distance, and the short internal C-Si spacing within the SiC bilayers. Our results are in excellent agreement with previous structural determinations by Razado-Colambo et al. [20]. Using synchrotron-radiation photoelectron diffraction they found that the corresponding distances were d=3.48d=3.48, 2.0, and 0.63 Å. This provides strong evidence that these features originate from the 3D geometry of the multilayer stack rather than the 2D structure of graphene.

If, instead, V0V_0 is extracted using the known distances reported by Razado-Colambo et al. [20], the inner potential for each successive scattering event would be approximately V0≈16.2V_0\approx16.2, 14.8, and 10.0 eV. This trend indicates an effective inner potential that decreases monotonically when moving spatially deeper from the MLG surface, through the ZLG interface, and into the SiC substrate. While 16.2 eV and 14.8 eV are well within the expected range for MLG [20] and the MLG/ZLG/SiC interface, the lowest value (10.0 eV) is smaller than typically expected for a bulk semiconductor [38, 39, 40].

The present interference model could be extended to other 2D materials and heterostructures, including intercalated quasi-free-standing graphene (QFMLG), hexagonal boron nitride, and transition-metal dichalcogenides (TMDs). Beyond the determination of interlayer separations, the significance of this approach lies in understanding how out-of-plane structural interactions influence the photoemission response of electronic states. For example, applying this framework to intercalated QFMLG or other vdW heterostructures could help distinguish changes in the intrinsic electronic structure associated with intensity modulations arising from final-state scattering. Extending the model to incorporate element-dependent scattering phase shifts and realistic interface electronic states would be an important direction for future studies, particularly for materials containing heavier elements such as TMDs.

Regarding the resonance observed at hν=100h\nu=100 eV, an experimental artifact from the beamline diffraction grating causes the Si 2p2p core level to appear in the background of the ARPES scan. While this partially enhances the intensity observed at hν=100h\nu=100 eV, we note that the resonance persists at band and photon energies isolated from this contribution. A detailed analysis of this artifact is presented in Appendix B.

Photon-energy dependence of replica bands

Figure 5(a)–(Figure 5d) shows constant energy maps around the K‾\overline{\mathrm{K}} point of the MLG Brillouin zone 1.00 eV below the Fermi level. Here the characteristic dark corridor, where matrix-element effects suppress photoemission in certain directions, can be seen. Surrounding the primary Dirac cone contour are the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ replica bands. The physical origin of these replicas has been attributed to two competing models. These are photoelectron final-state scattering [41] and initial-state modulation because of the real-space moiré periodicity coming from the ZLG and SiC substrate [42]. Our results show a scaling of the replica-band intensity that is dependent on hνh\nu. To our knowledge, this scaling behavior has not been previously reported and strongly favors the final-state scattering interpretation.

Figure 5. (a)–(d) Constant energy maps taken at E−EF=−1.00E-E_F=-1.00 eV, for hν=40h\nu=40, 68, 100, and 180 eV. (e) Comparison of IπI_\pi and IRI_R using the average of the intensities for five different energies E−EFE-E_F at each photon energy. (f) Ratio of the replica-band intensity (IRI_R) to the π\pi-band intensity (IπI_\pi) as a function of photon energy for cuts through the K‾\overline{\mathrm{K}} point perpendicular to the ΓK‾\overline{\Gamma\mathrm{K}} direction at various E−EFE-E_F. Intensities at each photon energy were normalized using the synchrotron beam flux and momentum-space maps were acquired using an electrostatic electron deflector.

If the replica-bands were purely an initial-state effect, then we would expect that their intensity is coupled to that of the π\pi band, maintaining a constant intensity ratio regardless of hνh\nu. Figure 5(e) shows that the absolute magnitude of both the replica-band intensity (IRI_{\mathrm{R}}) and the π\pi band intensity (IπI_{\pi}), averaged across five binding energies, do follow the same overall trend as the intensities in Figure 3(b)–(Figure 3d). However, as seen in Figure 5(f), the ratio of the replica band intensity to the π\pi band intensity (IR/IπI_{\mathrm{R}}/I_{\pi}) increases over the hνh\nu energy range from 27 eV to 180 eV. This shows that the replicas experience a less steep decrease in relative intensity than the primary π\pi band. We note the outlying data at E−EF=−1.25E-E_{\mathrm{F}}=-1.25 eV, hν=100h\nu=100 eV in Figure 5(f). This comes from photoemission of the Si 2p2p core level that was discussed previously, cf. Appendix B. Here the increased background intensity artificially increases the replica band signal.

To determine the source of the energy dependence of the intensity ratio IR/IπI_{\mathrm{R}}/I_{\pi}, we first consider the contribution of the IMFP and the polarization dependence of the matrix element. Historically, the IMFP was treated as a universal curve, described by Seah and Dench [43], with a minimum near E=50E=50 eV. For MLG, modern determinations of the IMFP range from approximately constant for energies up to 100 eV, to monotonically decreasing for energies up to 25 eV [44, 45]. Neither trend is consistent with the monotonic increase in the ratio observed here, indicating that the origin is not rooted in simple attenuation of the electron propagating through the heterostructure. Secondly, we can distinguish this behavior from the polarization-dependence of the matrix element. In our experimental geometry, the sample is rotated as hνh\nu is varied, which results in a change in the angle between the light’s polarization vector and the sample surface normal. While small differences in momentum-space position and orbital character may result in changes to the polarization-dependence of the photoemission matrix elements, making it impossible to entirely exclude these effects as a contribution to the observed intensity ratio, we expect their impact to be relatively minor given similar momentum-space locations and orbital character of the primary and replica bands. Instead, the observed behavior may depend on how the scattering cross section changes in the experiment at different hνh\nu.

For electrons scattered by atomic potentials, the scattering cross section is dependent on both the kinetic energy of the electron and the scattering angle θs\theta_s, where θs=180∘−2θi\theta_s=180^\circ-2\theta_i. As shown by Heinz, the magnitude of the atomic scattering factor ∣t(k,θs)∣\lvert t(k,\theta_s)\rvert (where k∝Ek\propto\sqrt{E}) increases with EE [46]. Further, ∣t(k,θs)∣\lvert t(k,\theta_s)\rvert for C atoms exhibits maximum values in the forward (θs=0∘\theta_s=0^\circ) and backward (θs=180∘\theta_s=180^\circ) scattering directions and a minimum near θs≈90∘\theta_s\approx90^\circ [46]. Crucially, ∣t(k,θs)∣\lvert t(k,\theta_s)\rvert for C increases monotonically with θs\theta_s while moving from the minimum toward the backscattering direction [46].

In our experiment, the sample angle is adjusted with hνh\nu to align photoemission from the K‾\overline{\mathrm{K}} point with the analyzer axis. At lower photon energies (27 eV), emission occurs at higher off-normal angles, corresponding to a ZLG scattering angle of θs≈108∘\theta_s\approx108^\circ near the minimum of the final-state scattering cross section. As hνh\nu increases, the emission direction rotates closer to normal incidence, driving θs\theta_s toward 180∘180^\circ (perfect backscattering). Consequently, the overall scattering rate increases at higher hνh\nu due to both the increase in EE and the angular trajectory shift toward stronger backscattering. While these results cannot completely exclude contributions from the initial state, they do indicate that IR/IπI_{\mathrm{R}}/I_{\pi} (and hence the presence of the replica bands) is predominantly driven by final-state backscattering of the photoelectrons.

ZLG interfacial states and π\pi-band electron phonon coupling

To disentangle the diverse electronic signatures within the MLG photoemission spectra, we now examine the summed energy distribution curves for the π\pi bands and the background signal across all 81 hνh\nu scans. By separating these components, we identify localized electronic states at the ZLG/SiC interface within the background signal, alongside a distinct feature in the π\pi band signal that coincides with the known E2gE_{2g} electron-phonon coupling (EPC) signature [48] [47]. Figure 6(a)–(Figure 6d) shows the MLG π\pi bands at hν=30h\nu=30, 58, 100, and 138 eV. To better distinguish the dispersing bands from the background, we utilized the masking procedure outlined in Appendix A to extract the momentum-integrated intensity for the π\pi bands (IπI_{\pi}) and the background (IBI_{\mathrm{B}}) [Figure 6(e)–(Figure 6h)].

Figure 6. (a)–(d) ARPES scans of the Dirac cone taken at hν=30h\nu= 30, 58, 100, and 138 eV. (e)–(h) Integrated intensity as a function of E−EFE - E_{\mathrm{F}} for the π\pi bands (IπI_{\pi}, red) and the background signal (IBI_{\mathrm{B}}, blue) corresponding to panels (a)–(d). Solid black curves are fits to the data. (i) Tracking of the energy levels (E−EFE - E_{\mathrm{F}}) for the peak near the E2gE_{2g} EPC kink and interfacial states g1g_{1}, l1l_{1}, g2g_{2}, and l2l_{2} across the measured photon energy range. Dashed lines represent the mean binding energy for each state. (j) Photoemission intensity as a function of photon energy for the four interfacial states.

Turning first to the extracted π\pi-band intensity (IπI_{\pi}) shown in Figure 6(e)–(Figure 6h), we observe a notable maximum near the energy of the known E2gE_{2g} EPC kink [48] [47]. The peak location was determined by fitting the data with a model consisting of a primary Lorentzian peak and a Gaussian component (accounting for the intensity increase at higher binding energies), multiplied by a Fermi-Dirac distribution, plus a constant background. Figure 6(i) shows that the location of the peak remains constant across all measured photon energies. The average value of E−EF≈−0.2 eVE - E_{\mathrm{F}} \approx-0.2\,\mathrm{eV} is in excellent agreement with previous theoretical and experimental results [48] [47]. We note that the relative intensity of the peak is sensitive to the momentum-space alignment of the ARPES scan. The high intensity shown in Figure 6(e) and (f), and the low intensity in (h) can be attributed to a small misalignment of the emission angle θ\theta relative to K‾\overline{\mathrm{K}} for these photon energies.

Shifting our focus to the isolated background signal (IBI_{\mathrm{B}}), our scans resolve four distinct electronic states localized at the interface, labeled g1g_{1}, l1l_{1}, g2g_{2}, and l2l_{2} in Figure 6(i). These interface features were fitted at each photon energy using four Lorentzian distributions alongside two background shaping components. A broad Gaussian centered near E−EF≈−0.3 eVE - E_{\mathrm{F}} \approx-0.3\,\mathrm{eV} was used to capture intensity beyond EFE_{\mathrm{F}}, which likely originates from higher-harmonic leakage through the monochromator, and a sigmoidal step function near E−EF≈−2.9 eVE - E_{\mathrm{F}} \approx-2.9\,\mathrm{eV} to model the sharp drop in intensity at the edge of the detector energy window. A constant offset term was also included in the fit. As indicated by the dashed lines in Figure 6(i), these features remain essentially fixed in binding energy across the entire measured photon energy range. The lack of kzk_z dispersion confirms their origin as intrinsic initial states of the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})R30^\circ reconstruction.

The states g1g_{1} and g2g_{2}, located at E−EF≈−0.6 eVE - E_{\mathrm{F}} \approx-0.6\,\mathrm{eV} and ≈−1.7 eV\approx-1.7\,\mathrm{eV}, are consistent with previously reported surface states associated with partially covalent C-Si bonding [49, 50, 51, 52, 53].

The two additional states, which we have named l1l_{1} and l2l_{2}, are located at E−EF≈−1.2 eVE - E_{\mathrm{F}} \approx-1.2\,\mathrm{eV} and ≈−2.3 eV\approx-2.3\,\mathrm{eV}, respectively. Sensitivity of these binding energies to fitting constraints and Si 2p2p core-level contamination near hν=100 eVh\nu= 100\,\mathrm{eV} introduced no observable change within our experimental resolution. These additional nondispersing states likely originate from the same ZLG/SiC interface. Density functional theory (DFT) calculations of the ZLG using the meta-generalized gradient approximation report multiple flat bands in the energy range of E−EF=−1.5 eVE-E_{\mathrm{F}} = -1.5\,\mathrm{eV} to 0 eV0\,\mathrm{eV} [54], while separate DFT calculations show a peak in the density of states at E−EF≈−2.2 eVE-E_{\mathrm{F}} \approx-2.2\,\mathrm{eV} [55], which is in good agreement with our observed l2l_{2} state.

After establishing their binding energies, we tracked the photoemission intensity of g1g_{1}, l1l_{1}, g2g_{2}, and l2l_{2} as a function of photon energy hνh\nu Figure 6(j) to deduce their orbital character. Although the intensity modulation features described in § 4 A are present, they do not produce strong resonances here, manifesting instead as minor modulations superimposed on a monotonic decay. The states l1l_{1}, g2g_{2}, and l2l_{2} display a sharp intensity drop from hν≈20 eVh\nu\approx20\,\mathrm{eV} to 80 eV80\,\mathrm{eV}, closely mirroring the atomic photoionization cross section of C 2p2p orbitals [56]. From this behavior, we conclude, in agreement with Emtsev et al. [49], that g1g_{1} and g2g_{2} originate predominantly from C 2pz2p_z orbitals, and that this orbital character extends to l1l_{1} and l2l_2 as well. Above hν≈130 eVh\nu\approx130\,\mathrm{eV}, the intensity of l2l_2 subtly overtakes that of g2g_2 [inset of Figure 6(j)]. This intensity crossover aligns with the photoionization cross section of Si 3s3s and 3p3p orbitals, which surpass C 2p2p in this photon energy region [56], suggesting partial hybridization with the underlying Si substrate.

Conclusions

In summary, we have comprehensively studied the electronic and structural properties of epitaxial graphene on SiC by means of photon-energy-dependent ARPES. By utilizing the large photon energy range provided by a synchrotron light source, we have resolved multiple interactions between the MLG and the substrate. First, we observed multiple hνh\nu-dependent resonances in the photoemission spectrum of the graphene π\pi bands and successfully reproduced most of these features with a MS model of freestanding MLG. We then demonstrated that electron final-state scattering from the substrate interfaces produces additional photon-energy resonances at 27, 38, and 100 eV. Secondly, we provided evidence that the (63×63)R30∘(6\sqrt{3} \times6\sqrt{3})\mathrm{R}30^\circ replica bands show intensity scaling governed by the final-state scattering dependence on hνh\nu. Lastly, we report the observation of four interfacial electronic states (g1g_1, l1l_1, g2g_2, l2l_2). Our study demonstrates that the out-of-plane structure of a heterostructure can strongly influence the measured photoemission intensity of its electronic states through final-state scattering. Extending this framework to other 2D materials and heterostructures could provide a means of disentangling intrinsic electronic-structure modifications from substrate-induced photoemission effects, facilitating a more reliable interpretation of ARPES measurements in complex layered systems.

Acknowledgments

This work is supported by the MonArk NSF Quantum Foundry from the National Science Foundation under Award No. DMR-1906383, and μ\mu-ATOMS Energy Frontier Research Center by the U.S. Department of Energy under Award No. DE-SC0023412. The graphene preparation was supported by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) within the Research Unit FOR5242 through Projects No. Ku4228/1-1 and No. Sta315/13-1. Use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-76SF00515. E.F. and S.W. are supported via a fellowship by the NSF under the Award No. 2244274.

Data Availability

All raw data supporting the findings of this work are available from the corresponding authors upon reasonable request.

Numerical Masking Procedure

Figure 7 provides an example of the masking procedure used to isolate the electronic band dispersions from the nondispersing background signal. This approach was selected over derivative-based methods, such as the one by Zhang et al. [57], which fail to maintain intensity information upon isolating dispersive features.

Figure 7. (a) Raw data taken at hν=100 eVh\nu= 100\,\mathrm{eV}. (b) Mask generated using the band dispersion. (c) Data with background removed using the mask. (d) Data with background only. (e) Data with kxk_x-averaged background subtracted.

The mask was made in four steps. The first was to identify the two peak positions on either side of the Dirac point in the momentum distribution curve (MDC) for each energy E−EF<0E-E_F < 0. If no peaks greater than four times the signal-to-noise ratio were found, then that MDC was skipped. Next, the peak positions were used to fit two symmetric lines of the form E−EF=±v(kx−k0)+EDE-E_F = \pm v(k_x-k_0)+E_D, where (k0,ED)(k_0,E_D) represents the coordinates of the Dirac point. Then, the mask was defined by setting all points within a total momentum width of 0.065 A˚−10.065\,\mathrm{\AA}^{-1} (i.e., ±0.0325 A˚−1\pm0.0325\,\mathrm{\AA}^{-1} centered on the fitted lines) to 1 and 0 elsewhere [Figure 7(b)]. The width of the region was chosen to avoid any clipping of the bands [Figure 7(c)]. Finally, this mask was used to determine regions of background intensity [Figure 7(d)] which were kxk_x-averaged for each energy and subtracted from the raw data [Figure 7(e)]. A mask width of 0.065 A˚−10.065\,\mathrm{\AA}^{-1} was applied to all data evaluated in § 4 C for tracking the ZLG surface states, whereas a wider mask width of 0.1 A˚−10.1\,\mathrm{\AA}^{-1} was used for all datasets in § 4 A to ensure no leakage of background intensity across the photon-energy-dependent analysis.

Diffraction Grating Artifacts

When using synchrotron radiation for ARPES, a diffraction grating is used to monochromatize the photon source which allows for tunable photon energies. An unavoidable consequence of this is the leakage of higher-order diffraction peaks that will overlap the chosen hνh\nu. An example of this would be setting the monochromator to hν=100h\nu= 100 eV which will also pass the higher harmonics, such as hν=200h\nu= 200 eV. The result of this leakage will typically be an energy shifted replica of the ARPES signal that appears at higher kinetic energies not detected during a measurement. In this study, however, the second harmonic of hν=100h\nu= 100 eV excites the Si 2p2p core level which appears in the background of the valence-band ARPES scan. An example of this can be found in Figure 6(g) and Figure 7(a) and Figure 7(d). Below we provide x-ray photoemission spectroscopy (XPS) data which confirms the location of the Si 2p2p core level.

Figure 8. XPS data of the Si 2p2p level taken at hν=200h\nu= 200 eV and normal emission. The solid black lines indicate the position of the intensity peak relative to the hν=100h\nu= 100 eV ARPES energy window where the right line is the Fermi edge and the left is E−EF=−3E-E_{\mathrm F}=-3 eV. The blue dashed line indicates the intensity maximum of the peak.

Figure 8 presents XPS data of the Si 2p2p core level taken at hν=200h\nu= 200 eV. Here we observed two Si peaks coming from chemical bonding to oxygen (E−EF≈103.7E-E_{\mathrm F}\approx103.7 eV) and carbon (E−EF≈101.3E-E_{\mathrm F}\approx101.3 eV). The oxygen peak indicates partial undergrowth on some regions of the sample. These values are consistent with results from Refs. [58, 59]. When these core levels are excited by the second harmonic of hν=100h\nu= 100 eV, the primary Si-C peak will appear at E−EF≈−1.4E-E_{\mathrm F}\approx-1.4 eV which corresponds to the background peak in Figure 6(g).

The intensity curves in Figure 9 demonstrate the efficacy

Figure 9. (a)–(c) Intensity as a function of E−EFE-E_{\mathrm F} for the π\pi bands (IπI_{\pi}, red) and the background signal (IBI_{\mathrm B}, blue) at hν=94h\nu= 94, 100, and 106 eV. Intensities have been normalized to show average counts per pixel of the detector.

of the masking procedure outlined in Appendix A. While a distinct background artifact (IBI_{\mathrm B}) is present at E−EF≈−1.4E-E_{\mathrm F}\approx-1.4 eV for hν=100h\nu= 100 eV [Figure 9(b)], the extracted π\pi band intensity profile (IπI_{\pi}) preserves its line shape across all measured photon energies [Figure 9(a)–(Figure 9c)]. This confirms that the masking method successfully removes the Si 2p2p background artifact and that the peak in Figure 3(c) and (Figure 3d) at hν=100h\nu= 100 eV can be attributed to interlayer scattering and not just an artifact of the second harmonic leakage through the monochromator.

Constructive Interference Condition for Interlayer Scattering

In photoemission, by changing the angle of a sample relative to the detector one can select the in-plane momentum k∥k_{\parallel} of the electrons that are measured. The angle θo\theta_o required to select a particular location in momentum space can be calculated within the free electron model in terms of the kinetic energy EE, mass mem_e of the electron, and the magnitude of k∥k_{\parallel},

θo=sin⁡−1(ℏk∥2meE).(6)\theta_o = \sin^{-1}\left(\frac{\hbar k_{\parallel}}{\sqrt{2m_eE}}\right). \tag*{(6)}

In a similar manner, the angle inside the material (θi\theta_i) can be defined by conservation of energy and k∥k_{\parallel} and the inner potential V0V_0 of the material:

θi=sin⁡−1(ℏk∥2me(E+V0)).(7)\theta_i = \sin^{-1}\left(\frac{\hbar k_{\parallel}}{\sqrt{2m_e(E+V_0)}}\right). \tag*{(7)}

Because the crystal surface potential is periodic in the plane, parallel momentum is conserved across the surface interface (k∥,i=k∥,o=k∥k_{\parallel,i}=k_{\parallel,o}=k_{\parallel}). Expressing these parallel components in terms of the total wavevector magnitudes inside (kik_i) and outside (kok_o) the material gives

kisin⁡θi=kosin⁡θo.(8)k_i\sin\theta_i=k_o\sin\theta_o. \tag*{(8)}

In Figure 10, a schematic of interlayer scattering of photoelectrons is shown. The path kk represents direct emission of the photoelectron from the material to the detector at an angle θo\theta_o. On the other hand, k′k' represents a path where the electron diffracts from a lattice plane before traveling to the detector. For both paths to represent the same initial state, θo\theta_o and the final kinetic energy EE must be the same. Since energy is conserved in this interaction, we know that the electron inside the material must have the same angle θi\theta_i before and after diffracting from the lattice plane. This restricts the electron to a single path inside the material where the distance traveled from the emitting layer to the scattering plane is x1x_1. The distance x1x_1 can be shown to be x1=d1/cos⁡θix_1=d_1/\cos\theta_i, yielding a total path length inside the material of 2x12x_1. The segment x2=2d1sin⁡θotan⁡θix_2=2d_1\sin\theta_o\tan\theta_i is the projection of the lateral offset between the two surface exit points onto the direct wavevector kk to construct a plane of constant phase.

Diagram of photoelectron scattering paths

Figure 10. Diagram of photoelectron scattering paths. The path kk represents direct emission from the k′′k^{\prime\prime} surface layer, shown in red, to the detector while k′k^{\prime} and k′′k^{\prime\prime} represent the components scattered by lattice planes separated by a distance d1d_1 and d2d_2, shown in green. The wavevectors k′k^{\prime} and k′′k^{\prime\prime} refract at the surface due to the change in inner potential V0V_0 at the material/vacuum interface as illustrated by the internal and external emission angles θi\theta_i and θo\theta_o. The lengths xix_i indicate different components of the paths traveled by the electron.

With the path lengths defined, we now consider the phase accumulation for both the direct and scattered paths. The wave vectors for these paths inside the material and in the vacuum are given by ki=2me(E+V0)ℏk_i=\frac{\sqrt{2m_e(E+V_0)}}{\hbar} and ko=2meEℏk_o=\frac{\sqrt{2m_eE}}{\hbar}. The phase difference Δϕ\Delta\phi between the two paths is then determined by the difference in path lengths times the corresponding wave vectors. The total phase difference is expressed as

Δϕ=2x1ki−x2ko=2(d1cos⁡θi)ki−(2d1tan⁡θisin⁡θo)ko=2d1kicos⁡θi−2d1tan⁡θi(kosin⁡θo)=2d1kicos⁡θi−2d1kisin⁡2θicos⁡θi=2d1kicos⁡θi(1−sin⁡2θi)=2d1kicos⁡θi.(9)\begin{aligned} \Delta\phi&= 2x_1k_i-x_2k_o\\ &=2\left(\frac{d_1}{\cos\theta_i}\right)k_i-(2d_1\tan\theta_i\sin\theta_o)k_o\\ &=\frac{2d_1k_i}{\cos\theta_i}-2d_1\tan\theta_i(k_o\sin\theta_o)\\ &=\frac{2d_1k_i}{\cos\theta_i}-2d_1k_i\frac{\sin^2\theta_i}{\cos\theta_i}\\ &=\frac{2d_1k_i}{\cos\theta_i}(1-\sin^2\theta_i)\\ &=2d_1k_i\cos\theta_i. \tag*{(9)} \end{aligned}

and the final constructive interference condition for a general layer separation dd is

2πn=2dkicos⁡θi.(10)2\pi n=2dk_i\cos\theta_i. \tag*{(10)}

This process can be repeated for the paths k′k' and k′′k'' to determine the constructive interference condition for electrons that scatter from two different lattice planes instead of one taking a direct path to the detector. In this case the phase difference can be expressed as in Eq. (10) and depends on the separation d2d_2 between the planes.

It is instructive to compare this result with the photoemission diffraction framework of Fadley [60]. In the single scattering cluster approximation, a photoelectron emitted from a substrate atom and scattered by a neighboring atom at distance rjr_j at an angle θj\theta_j accumulates an overall path length difference between the direct and scattered paths of ΔL=rj(1−cos⁡θj)\Delta L=r_j(1-\cos\theta_j). The total phase shift is

Δϕ=krj(1−cos⁡θj)+ψj(θj),\Delta\phi=kr_j(1-\cos\theta_j)+\psi_j(\theta_j),

where ψj(θj)\psi_j(\theta_j) is the phase shift associated with the atomic scattering event [60].

In our model of planar interlayer scattering, the distance to the scatterer is rj=d/cos⁡θir_j=d/\cos\theta_i and the scattering angle is θj=π−2θi\theta_j=\pi-2\theta_i. Substituting these parameters into the path difference term and setting k=kik=k_i yields

kirj(1−cos⁡θj)=ki(dcos⁡θi)[1−cos⁡(π−2θi)]=ki(dcos⁡θi)(1+cos⁡2θi)=ki(dcos⁡θi)(2cos⁡2θi)=2dkicos⁡θi.\begin{aligned} k_ir_j(1-\cos\theta_j)&=k_i\left(\frac{d}{\cos\theta_i}\right)\left[1-\cos(\pi-2\theta_i)\right]\\ &=k_i\left(\frac{d}{\cos\theta_i}\right)(1+\cos2\theta_i)\\ &=k_i\left(\frac{d}{\cos\theta_i}\right)(2\cos^2\theta_i)\\ &=2dk_i\cos\theta_i. \end{aligned}

Thus, our simple geometric model recovers the exact spatial phase term 2dkicos⁡θi2dk_i\cos\theta_i of the full scattering theory, differing only by the atomic scattering phase shift ψj(θj)\psi_j(\theta_j). Because ψj(θj)\psi_j(\theta_j) scales with the atomic number ZZ of the scattering species, we approximate ψj(θj)≈0\psi_j(\theta_j)\approx0 given the light elements composing our sample (ZC=6Z_{\mathrm{C}}=6, ZSi=14Z_{\mathrm{Si}}=14).

Polarization Dependence of Intensity Resonances

It is well established that photoemission intensity in ARPES depends on the dipole selection rules governing the photoemission matrix element, which can strongly influence the measured spectral intensity [7]. To investigate the influence of light polarization on the photon-energy-dependent intensity resonances, MS simulations were performed using pp-, ss-, and circularly polarized light. Figure 11 compares the resulting photon-energy-dependent intensities for the three polarization configurations at three binding energies: −0.25, −1.25, and −2.25 eV. For each binding energy, the intensities are normalized to the maximum pp-polarized intensity to facilitate comparison of the relative intensities and resonance positions. Although circular dichroism has been observed in materials such as graphite [29], in our experimental geometry, the opposite dichroic contributions from the two mirror-symmetry related branches (kxk_x and −kx-k_x) of the Dirac cone cancel when their intensities are summed.

Figure 11. Photon-energy-dependent photoemission intensity from MS simulations at binding energies of (a) −0.25 eV, (b) −1.25 eV, and (c) −2.25 eV. Each panel compares pp-polarized (MS-P), ss-polarized (MS-S), and circularly polarized (MS-C) light. Intensities are normalized to the maximum pp-polarized intensity at each binding energy.

The photon-energy-dependent intensity resonances observed with pp-polarized light [Fig. 3(b)-(d)] are also present with both ss- and circularly polarized light. However, their relative intensities and resonance positions exhibit some dependence on polarization. The ss-polarized intensities are substantially lower than the pp-polarized intensities. Additionally, the resonance locations generally shift toward lower photon energies for ss-polarized light. For circularly polarized light, the resulting intensity is dominated by the much larger pp-polarized contribution, producing a resonance profile that closely resembles that obtained with pp-polarized light. These results demonstrate that, although the photoemission matrix elements influence the relative intensities and positions of the resonances, the photon-energy-dependent intensity modulations persist for all three polarization configurations.

References

  1. [1]K. S. Novoselov, V. I. Fal’ko, L. Colombo, P. R. Gellert, M. G. Schwab, and K. Kim, A roadmap for graphene, Nature 490, 192 (2012).DOI
  2. [2]J. Li, X. Yang, Z. Zhang, W. Yang, X. Duan, and X. Duan, Towards the scalable synthesis of two-dimensional heterostructures and superlattices beyond exfoliation and restacking, Nat. Mater. 23, 1326 (2024).DOI
  3. [3]K. V. Emtsev, A. Bostwick, K. Horn, J. Jobst, G. L. Kellogg, L. Ley, J. L. McChesney, T. Ohta, S. A. Reshanov, J. Röhrl, E. Rotenberg, A. K. Schmid, D. Waldmann, H. B. Weber, and T. Seyller, Towards wafer-size graphene layers by atmospheric pressure graphitization of silicon carbide, Nat. Mater. 8, 203 (2009).DOI
  4. [4]M. Kusunoki, W. Norimatsu, J. Bao, K. Morita, and U. Starke, Growth and features of epitaxial graphene on SiC, J. Phys. Soc. Jpn. 84, 121014 (2015).DOI
  5. [5]T. Schumann, M. Dubslaf, M. H. Oliveira Jr, M. Hanke, J. M. J. Lopes, and H. Riechert, Effect of buffer layer coupling on the lattice parameter of epitaxial graphene on SiC(0001), Phys. Rev. B 90, 041403(R) (2014).
  6. [6]A. Damascelli, Probing the electronic structure of complex systems by ARPES, Phys. Scr. 2004, 61 (2004).DOI
  7. [7]S. Moser, An experimentalist’s guide to the matrix element in angle resolved photoemission, J. Electron Spectrosc. Relat. Phenom. 214, 29 (2017).DOI
  8. [8]A. Bansil and M. Lindroos, Importance of matrix elements in the ARPES spectra of BISCO, Phys. Rev. Lett. 83, 5154 (1999).DOI
  9. [9]T. Ohta, A. Bostwick, J. L. McChesney, T. Seyller, K. Horn, and E. Rotenberg, Interlayer interaction and electronic screening in multilayer graphene investigated with angle-resolved photoemission spectroscopy, Phys. Rev. Lett. 98, 206802 (2007).pubmed.ncbi.nlm.nih.gov/17677726
  10. [10]E. L. Shirley, L. J. Terminello, A. Santoni, and F. J. Himpsel, Brillouin-zone-selection effects in graphite photoelectron angular distributions, Phys. Rev. B 51, 13614 (1995).pubmed.ncbi.nlm.nih.gov/9978164
  11. [11]H. Daimon, S. Imada, H. Nishimoto, and S. Suga, Structure factor in photoemission from valence band, J. Electron Spectrosc. Relat. Phenom. 76, 487 (1995).DOI
  12. [12]I. Gierz, J. Henk, H. Höchst, C. R. Ast, and K. Kern, Illuminating the dark corridor in graphene: Polarization dependence of angle-resolved photoemission spectroscopy on graphene, Phys. Rev. B 83, 121408 (2011).DOI
  13. [13]P. Avria, A. R. T. Nugraha, E. H. Hasdeo, T. R. Czank, S.-i. Tanaka, and R. Saito, Photon energy dependence of angle-resolved photoemission spectroscopy in graphene, Phys. Rev. B 92, 195148 (2015).
  14. [14]C. S. Kern, A. Haags, L. Egger, X. Yang, H. Kirschner, S. Wolff, T. Seyller, A. Gottwald, M. Richter, U. De Giovannini, A. Rubio, M. G. Ramsey, F. C. Bocquet, S. Soubatch, F. S. Tautz, P. Pusch­nig, and S. Moser, Sim- ple extension of the plane-wave final state in photoemission: Bringing understanding to the photon-energy dependence of two-dimensional materials, Phys. Rev. Res. 5, 033075 (2023).
  15. [15]M. Nozaki and P. Krüger, Computational method for angle-resolved photoemission spectra from repeated-slab band structure calculations, Phys. Rev. B 110, 195406 (2024).
  16. [16]S. Soubatch, S. E. Saddow, S. P. Rao, W. Lee, M. Konuma, and U. Starke, Structure and morphology of 4H-SiC wafer surfaces after H₂-etching, Mater. Sci. Forum 483, 761 (2005).
  17. [17]F. Varchon, P. Mallet, J.-Y. Veuillen, and L. Magaud, Ripples in epitaxial graphene on the Si-terminated SiC (0001) surface, Phys. Rev. B 77, 235412 (2008).DOI
  18. [18]S. Kim, J. Ihm, H. J. Choi, and Y.-W. Son, Origin of anomalous electronic structures of epitaxial graphene on silicon carbide, Phys. Rev. Lett. 100, 176802 (2008).pubmed.ncbi.nlm.nih.gov/18518318
  19. [19]A. Zugarramurdi, M. Debiossac, P. Lunca-Popa, A. Mayne, A. Momeni, A. Borisov, Z. Mu, P. Roncin, and H. Khemliche, Determination of the geometric corrugation of graphene on SiC(0001) by grazing incidence fast atom diffraction, Appl. Phys. Lett. 106, 101902 (2015).
  20. [20]I. Razado-Colambo, J. Avila, D. Vignaud, S. Godey, X. Wallart, D. Woodruff, and M. Asensio, Structural determination of bilayer graphene on SiC(0001) using synchrotron radiation photoelectron diffraction, Sci. Rep. 8, 10190 (2018).
  21. [21]H. Rezaei and A. Phirouznia, Modified spin–orbit couplings in uniaxially strained graphene, Eur. Phys. J. B 91, 295 (2018).DOI
  22. [22]S. Mammadov, J. Ristein, J. Krone, C. Raidel, M. Wanke, V. Wiesmann, F. Speck, and T. Seyller, Work function of graphene multilayers on SiC(0001), 2D Mater. 4, 015043 (2017).DOI
  23. [23]R. P. Day, B. Zwartsenberg, I. S. Elfimov, and A. Damascelli, Computational framework chinook for angle-resolved photoemission spectroscopy, npj Quantum Mater. 4, 54 (2019).DOI
  24. [24]P. J. Feibelman and D. E. Eastman, Photoemission spectroscopy-correspondence between quantum theory and experimental phenomenology, Phys. Rev. B 10, 4932 (1974).DOI
  25. [25]J. B. Pendry, in Interaction of Atoms and Molecules with Solid Surfaces, edited by V. Bortolani, N. H. March, and M. P. Tosi (Springer, New York, 1990) pp. 201–211.
  26. [26]J. Pendry, Theory of photoemission, Surf. Sci. 57, 679 (1976).
  27. [27]A. Grüneis, C. Attaccalite, T. Pichler, V. Zabolotnyy, H. Shiozawa, S. L. Molodtsov, D. Inosov, A. Koitzsch, M. Knupfer, J. Schiessling, R. Follath, R. Weber, P. Rudolf, L. Wirtz, and A. Rubio, Electron-electron correlation in graphite: A combined angle-resolved photoemission and first-principles study, Phys. Rev. Lett. 100, 037601 (2008).pubmed.ncbi.nlm.nih.gov/18233036
  28. [28]P. Krüger, F. Da Pieve, and J. Osterwalder, Real-space multiple scattering method for angle-resolved photoemission and valence-band photoelectron diffraction and its application to Cu(111), Phys. Rev. B 83, 115437 (2011).DOI
  29. [29]P. Krüger and F. Matsui, Observation and theory of strong circular dichroism in angle-resolved photoemission from graphite, J. Electron Spectrosc. Relat. Phenom. 258, 147219 (2022).DOI
  30. [30]O. K. Andersen, Linear methods in band theory, Phys.
  31. [31]P. Krüger, Photoelectron diffraction from valence states of oriented molecules, J. Phys. Soc. Jpn. 87, 061007 (2018).DOI
  32. [32]V. Nazarov, E. Krasovskii, and V. Silkin, Scattering resonances in two-dimensional crystals with application to graphene, Phys. Rev. B 87, 041405 (2013).DOI
  33. [33]M. Krivenkov, D. Marchenko, J. Sánchez-Barriga, O. Rader, and A. Varykhalov, Suppression of electron scattering resonances in graphene by quantum dots, Appl. Phys. Lett. 111 (2017).DOI
  34. [34]B. Matta, P. Rosenzweig, O. Bolkenbaas, K. Küster, and U. Starke, Momentum microscopy of Pb-intercalated graphene on SiC: Charge neutrality and electronic structure of interfacial Pb, Phys. Rev. Res. 4, 023250 (2022).DOI
  35. [35]M. Dauth, M. Graus, I. Schelter, M. Wießner, A. Schöll, F. Reinert, and S. Kümmel, Perpendicular emission, dichroism, and energy dependence in angle-resolved photoemission: The importance of the final state, Phys. Rev. Lett. 117, 183001 (2016).pubmed.ncbi.nlm.nih.gov/27834988
  36. [36]R. Ono and P. Krüger, A one-dimensional model for photoemission calculations from plane-wave band structure codes, e-J. Surf. Sci. Nanotechnol. 16, 49 (2018).DOI
  37. [37]C. J. Shearer, A. D. Slattery, A. J. Stapleton, J. G. Shapter, and C. T. Gibson, Accurate thickness measurement of graphene, Nanotechnology 27, 125704 (2016).DOI
  38. [38]P. Kruse, A. Rosenauer, and D. Gerthsen, Determination of the mean inner potential in III-V semiconductors by electron holography, Ultramicroscopy 96, 11 (2003).pubmed.ncbi.nlm.nih.gov/12623168
  39. [39]P. Kruse, M. Schowalter, D. Lamoen, A. Rosenauer, and D. Gerthsen, Determination of the mean inner potential in III-V semiconductors, Si and Ge by density functional theory and electron holography, Ultramicroscopy 106, 105 (2006).
  40. [40]R. S. Pennington, J. J. Mortensen, T. Kasama, C. B. Boothroyd, and R. E. Dunin-Borkowski, Theoretical and experimental factors affecting measurements of semiconductor mean inner potentials, J. Phys.: Conf. Ser. 209, 012030 (2010).DOI
  41. [41]C. Polley, L. I. Johansson, H. Fedderwitz, T. Balasubramanian, M. Leandersson, J. Adell, R. Yakimova, and C. Jacobi, Origin of the π-band replicas in the electronic structure of graphene grown on 4H-SiC(0001), Phys. Rev. B 99, 115404 (2019).DOI
  42. [42]L. Huang, Y. Wu, M. Hershberger, D. Mou, B. Schrunk, M. C. Tringides, M. Hupalo, and A. Kaminski, Effects of moiré lattice structure on electronic properties of graphene, Phys. Rev. B 96, 035411 (2017).DOI
  43. [43]M. P. Seah and W. A. Dench, Quantitative electron spectroscopy of surfaces: A standard data base for electron inelastic mean free paths in solids, Surf. Interface Anal. 1, 2 (1979).DOI
  44. [44]H. T. Nguyen-Truong, B. Da, L. Yang, Z. Ding, H. Yoshikawa, and S. Tanuma, Low-energy electron inelastic mean free path for monolayer graphene, Appl. Phys. Lett 117 (2020).DOI
  45. [45]D. Geelen, J. Jobst, E. Krasovskii, S. J. van der Molen, and R. M. Tromp, Nonuniversal transverse electron mean free path through few-layer graphene, Phys. Rev. Lett. 123, 086802 (2019).DOI
  46. [46]K. Heinz, Electron based methods: 3.2.1 low-energy electron diffraction (LEED), in Surface and Interface Science, edited by K. Wandelt (John Wiley & Sons, Ltd, 2013) pp. 93–150. A. Bostwick, T. Ohta, T. Seyller, K. Horn, and E. Rotenberg, Quasiparticle dynamics in graphene, Nat. Phys. 3, 36 (2007).
  47. [48]M. Calandra and F. Mauri, Electron-phonon coupling and electron self-energy in electron-doped graphene: Calculation of angular-resolved photoemission spectra, Phys. Rev. B. 76, 205411 (2007).DOI
  48. [49]K. Emtsev, F. Speck, T. Seyller, L. Ley, and J. D. Riley, Interaction, growth, and ordering of epitaxial graphene on SiC{0001} surfaces: A comparative photoelectron spectroscopy study, Phys. Rev. B 77, 155303 (2008).DOI
  49. [50]Y. Qi, S. Rhim, G. Sun, M. Weinert, and L. Li, Epitaxial graphene on SiC(0001): More than just honeycombs, Phys. Rev. Lett. 105, 085502 (2010).pubmed.ncbi.nlm.nih.gov/20868110
  50. [51]M. Nevius, M. Conrad, F. Wang, A. Celis, M. Nair, A. Taleb-Ibrahimi, A. Tejeda, and E. Conrad, Semiconducting graphene from highly ordered substrate interactions, Phys. Rev. Lett. 115, 136802 (2015).DOI
  51. [52]J. Hwang and C. Hwang, Study on the high spectral intensity at the Dirac energy of single-layer graphene on an SiC substrate, New J. Phys. 18, 043005 (2016).DOI
  52. [53]M. N. Nair, I. Palacio, A. Celis, A. Zobelli, A. Gloter, S. Kubsky, J.-P. Turmaud, M. Conrad, C. Berger, W. de Heer, E. H. Conrad, A. Taleb-Ibrahimi, and A. Tejeda, Band gap opening induced by the structural periodicity in epitaxial graphene buffer layer, Nano Lett. 17, 2681 (2017).DOI
  53. [54]E. Mohebbi, E. Pavoni, P. Stipa, M. Petroselli, L. Pierantoni, E. Laudadio, and D. Mencarelli, Reliable prediction of the band gap properties of graphene buffer layer on SiC using meta-GGA approximation, Carbon Trends 21, 100575 (2025).DOI
  54. [55]T. Cavalucci and V. Tozzini, Intrinsic structural and electronic properties of the buffer layer on silicon carbide unraveled by density functional theory, Sci. Rep. 8, 13097 (2018).
  55. [56]J. J. Yeh and I. Lindau, Atomic subshell photoionization cross sections and asymmetry parameters: 1 ≤ Z ≤ 103, At. Data Nucl. Data Tables 32, 1 (1985).DOI
  56. [57]P. Zhang, P. Richard, T. Qian, Y.-M. Xu, X. Dai, and H. Ding, A precise method for visualizing dispersive features in image plots, Rev. Sci. Instrum. 82, 043712 (2011).DOI
  57. [58]J. Chastain and J. F. Moulder, Handbook of X-ray Photoelectron Spectroscopy: A Reference Book of Standard Spectra for Identification and Interpretation of XPS (Perkin-Elmer Corporation, Eden Prairie, MN, 1992).
  58. [59]Z. H. Lu, M. J. Graham, D. T. Jiang, and K. H. Tan, SiO₂/Si(100) interface studied by Al Kα x-ray and synchrotron radiation photoelectron spectroscopy, Appl. Phys. Lett. 63, 2941 (1993).
  59. [60]C. S. Fadley, Angle-resolved x-ray photoelectron spectroscopy, Prog. Surf. Sci. 16, 275 (1984).DOI

Paper details

vlnc:W_9tpdx5gc
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents