# Fileset

[FeSi2_TEM_ASS_revised.pdf](https://mdr.nims.go.jp/filesets/9b190c63-b252-4933-8f1e-013e38d6e2ff/download)

## Creator

[Keisuke Sagisaka](https://orcid.org/0000-0002-5089-4271), [Kazutaka Mitsuishi](https://orcid.org/0000-0002-9361-4057)

## Rights

[Creative Commons BY-NC-ND Attribution-NonCommercial-NoDerivs 4.0 International](https://creativecommons.org/licenses/by-nc-nd/4.0/)

## Other metadata

[Atomic and electronic structures of the <math display="inline">  <mi>β</mi></math>-FeSi2/Si(001) interface](https://mdr.nims.go.jp/datasets/9073ee02-4503-4116-a361-1f61eb949120)

## Fulltext

Graphical AbstractAtomic and electronic structures of the β-FeSi2/Si(001) interfaceKeisuke Sagisaka, Kazutaka MitsuishiHighlightsAtomic and electronic structures of the β-FeSi2/Si(001) interfaceKeisuke Sagisaka, Kazutaka Mitsuishi• The atomistic interface structure between β-FeSi2/Si(001) substratehas been determined via STEM observations and DFT calculations.• A high quality single crystalline β-FeSi2 film was epitaxialy grown onSi(001) substrate.• The β-FeSi2(100) plane is bonded to the Si(001) plane with two Sidimers per unit cell.• DFT band calculations suggest two dimensional metallic electronicstates at the the β-FeSi2/Si(001) interface.Atomic and electronic structures of the β-FeSi2/Si(001)interfaceKeisuke Sagisakaa,∗, Kazutaka MitsuishiaaCenter for Basic Research on Materials, National Institute for Materials Science, 1-2-1Sengen, Tsukuba, 305-0047, Ibaraki, JapanAbstractThe atomistic interface structure between β-FeSi2 and Si substrate was in-vestigated by cross-sectional scanning transmission electron microscopy anddensity functional theory calculations (DFT). A high quality single crys-talline β-FeSi2 film was epitaxialy grown on Si(001) substrate. We demon-strate that the β-FeSi2(100) plane is bonded to the Si(001) plane with two Sidimers per unit cell. With such interface structure, DFT band calculationsreveal that metallic electronic states are localized at the interface. This find-ing implies the presence of two dimensional electronic states at the interfaceof β-FeSi2/Si(001), which may have potential applications in silicon-basedelectronic devices.Keywords: Iron silicide, Interface structure, STEM, DFT, STM1. Introductionβ-iron silicide (β-FeSi2) is a promising semiconducting material for light-emitting diode (LED) applications integrated into silicon electronic devices.Since the light emitting property at wavelength of 1.5 µm, relevant to opticalfiber communication, was reported [1], significant efforts have been devotedto realizing infrared LEDs using this silicon-based material.[2, 3, 4, 5, 6, 7,8, 9, 10] For the fabrication of these devices, Si(001) or Si(111) substrateis typically employed to grow β-FeSi2 for integration with silicon electronicdevices. These substrates also serve as a beneficial source of silicon. It is∗Corresponding authorEmail address: SAGISAKA.Keisuke@nims.go.jp (Keisuke Sagisaka)Preprint submitted to Applied Surface Science October 11, 2024critical to grow a continuous epitaxial β-FeSi2 film due to its superior emissionperformance compared to β-FeSi2 nano-particles buried in Si. [11, 12]Si and β-FeSi2 exhibit compatible lattice constants, facilitating the epi-taxial growth of β-FeSi2 on a Si wafer. β-FeSi2 possesses an orthorhombicunit cell with dimensions of a = 0.986 nm, b = 0.779 nm, and c = 7.83 nm.[13] When using a Si(001) substrate, which is prevalent in the electronic de-vice industry, the (100) plane of β-FeSi2 is grown with azimuthal orientationsrotated by 45◦ relative to the substrate surface normal (Fig. 1). Despite thegood crystallographic match between β-FeSi2 and Si(001), small strains areanticipated along the b and c axis of β-FeSi2, amounting to 1.4% and 2.0%,respectively. While β-FeSi2 tends to form islands on the Si(001) substrateat high growth temperatures (at 700◦C or above), [14] a flat and continu-ous epitaxial film can be achieved by employing a lower growth temperature(∼470◦C). [15, 16, 17] This thin film has been utilized as a template for thesubsequent growth of thicker continuous films. [18, 19, 20]The performance of β-FeSi2 LEDs remains unsatisfactory for practicalapplications. One proposed reason for this is the low quantum efficienciesfor electroluminescence, which is attributed to non-radiative recombinationthrough defects at the β-FeSi2/Si interface. [12] Additionally, the interfacemorphology may affect photoluminescence. [21] Furthermore, the thin epi-taxial film of β-FeSi2 directly grown on the Si(001) substrate at low tem-perature (∼470◦C) exhibits a intriguing feature of an undulating surfacewithout steps. [16, 17] The mechanism of formation and stress relaxationof such a membrane are not well understood. Therefore, detailed knowledgeof the atomistic interface structure between β-FeSi2/Si(001) is essential foraddressing these issues, as well as for controlling epitaxial growth and man-aging strains induced in the film. However, the interface structure betweenβ-FeSi2/Si(001) has not yet been identified.In this paper, we elucidate the atomistic structure of the β-FeSi2/Si(001)interface based on observations made using cross-sectional scanning trans-mission electron microscopy (STEM), supported by density functional theory(DFT) calculations. The detailed characterization of the interface structurepromoted us to calculate its electronic structure. Consequently, we deter-mined that the electronic property of the interface is metallic, characterizedby two bands crossing the Fermi level. This finding suggests the realizationof a two-dimensional (2D) electronic state at the interface of β-FeSi2/Si(001).22. Methods2.1. ExperimentThe preparation of the thin β-FeSi2 was conducted in an ultrahigh vac-uum (UHV) chamber with a base pressure of 5 × 10−9 Pa. A Si(001) sub-strate (n-type, resistivity: ∼ 0.01 Ωcm) was flashed to 1100 ◦C for 10 secondsto obtain a clean surface after overnight degassing. A thin film of β-FeSi2was grown by depositing iron at a controlled deposition rate of 0.006 Å/sfrom a high-purity iron rod (99.995 %) onto the clean Si(001) surface sub-strate at room temperature, followed by annealing to 430 ◦C for 30 minutes.The crystallinity of the film surface was confirmed using scanning tunnelingmicroscopy (STM) (Unisoku USM-1200). Following STM observation, theβ-FeSi2 surface was capped with amorphous silicon (∼10 nm thick, deposi-tion rate: ∼0.014 Å/s) in the UHV chamber to protect the film from an airexposure during transport. Subsequently, the sample was cut into a smallpiece and thinned by a focused ion beam. The β-FeSi2/Si(100) interface wasexamined using STEM (JEOL JEM-ARM200F) operated at 200 kV.2.2. ComputationThe β-FeSi2/Si(001) interface was modeled by a slab consisting of threeunit layers of β-FeSi2 and 21 Si atomic layers (Fig. S1 in the supplemen-tary material). The top layer of β-FeSi2 and the bottom layer of siliconwere terminated with pseudo-hydrogen atoms (charge = 1.5) [22] and hy-drogen atoms (charge = 1.0), respectively. Prior to STEM simulation andband calculations, structural optimizations of the slab were conducted by theVASP code [23, 24] with planewave basis sets (cutoff energy: 500 eV) and thePerdew-Burke-Ernzerhof exchange correlation [25] until forces were reducedto less than 0.01 eV/Å. A 8 × 8 × 1 or 22 × 22 × 1 Monkhorst-Pack k-pointmesh sampling [26] was adopted for the structural optimizations or electroniccalculations, respectively. STEM image simulations were conducted by themultislice method using the xHREM software (HREM Research Inc.), basedon structure models obtained from the structural optimization of the slabmodel.3. RESULTS AND DISCUSSIONInitially, the quality of a β-FeSi2 film grown on a Si(001) substrate wasconfirmed by STM. The surface of the β-FeSi2 film exhibits undulations along3the [100] and [010] crystallographic directions, forming interconnected patchyfeatures with variations in the image contrast and dimension spanning sev-eral nanometers (Fig. 2). The height variation over the image is a root-meansquare (RMS) of 0.11 nm. Notably, this surface of the β-FeSi2 film does notexhibit stepped or polycrystalline characteristics, but rather forms a contin-uous surface. The surface is terminated with Si tetramers, which appear as ap(2 × 2) structure in the STM images, [16, 17] as shown in the inset of Fig.2. This superstructure covers the surface continuously and uninterruptedly,even over regions that appear to have different heights in STM [Fig. S3 inthe supplementary material]. These observations suggest that the β-FeSi2film cover the stepped Si substrate similarly to how a carpet covers stairs.After the STM observation, the β-FeSi2 film was capped by amorphousSi and its cross section was examined using STEM. In the low magnificationimage [Fig. 3(a)], the layer with the brightest contrast corresponds to theβ-FeSi2 film, which is situated between the Si substrate and Si capping layer.The STEM survey was conducted at multiple locations to confirm that theβ-FeSi2 film is a continuous single crystal. The film exhibits three differentthicknesses: 1.0, 1.5 and 2.0 nm, which respectively correspond to 1, 1.5 and2 unit layer of β-FeSi2 along the [001] crystal orientation (a = 0.986 nm).At higher magnification, atomic arrangements in both the β-FeSi2 filmand the Si substrate are resolved [Fig. 3(b)]. This observation revealed thatthe relative configuration between β-FeSi2 and Si substrate lattice remainsconstant across more than 30 lattice periods in all examined crystallographicorientations, including [110], [110], and [010] of Si [Fig. 3(b)]. These findingindicate that the commensurate epitaxial growth of the β-FeSi2 is achievedon the Si(001) surface within this narrow thickness range. To enable thiscommensurate growth, the lattice mismatches should be compensated byelastic strains of 1.4% in the b axis and 2.0% in the c axis of β-FeSi2. Suchcommensurate growth is typically observed in thin layers of semiconductorheteroepitaxy even in the presence of a lattice mismatch between the growthlayer and substrate. [27]The film thickness of the β-FeSi2 appears to be regulated at the inter-face with the silicon substrate, while maintaining a flat surface. Figure 3(c)demonstrates the film thickness varies from one unit cell thick (a) to 1.5times that at the interface with the substrate. If a step forms on the top-most surface, its height should be a multiple of half the length of a (∼0.5nm). However, the STM image in Fig. 2 exhibits a RMS height variation ofonly 0.11 nm, which is not due to the presence of steps. The STM profile in4Fig. 2 shows smooth height changes across the surface. Although the STMimage contrast reflects both the electronic and physical heights, it is cur-rently unknown which effect predominates in the STM image. In either case,it should be ascribed to strain included within the β-FeSi2 film. The surfaceheight variation of 0.1 to 0.15 nm in the STM images (Fig. 2 and Fig. S3)closely matches the interplanar spacing of Si (0.136 nm) along the [001] di-rection. This suggests that the observed surface ripples are likely induced bythe stepped structure of the substrate beneath the β-FeSi2 membrane. Wehypothesize that the patchy structure observed in the STM image resultsfrom alternating patches of single-crystal β-FeSi2, each patch rotated by 90◦,which together form a continuous film. The lattice constants of the β-FeSi2unit cell differ between along the b- and c-axes, but the unit cell contracts totwice the lattice constant of Si(001), making the crystal symmetry equivalentalong the b- and c- axes, except for the Fe1, Si1 and Si4 layers, as labeled inFig. 1. While minor rearrangement in the atomic positions is required withinthese layers, it is possible to connect adjacent patches rotated by 90◦ relativeto each other without breaking crystal symmetry. When two patches of β-FeSi2 oriented along b-axis and c-axis are connected, with two Si substratelattices aligned and atomic positions adjusted to ensure a smooth transitionbetween the respective layers, a net surface height variation of approximately0.1 nm is expected. This variation, caused by the relative shift of one atomiclayer in the crystal symmetry of Si substrate between the (110) and (110)planes, is consistent with the features observed in the STM images. More-over, while the connection of rotated patches likely induces strain within themembrane, the overall stain should be minimized compared to a structurewithout patch rotations. However, this scenario remains hypothetical, andfurther investigations are necessary to confirm the structural characteristicsof the thin β-FeSi2 film grown on Si(001).In contrast to our results, Takemoto et al. recently reported that β-FeSi2 grown on a Si(001) substrate with a low anneal temperature (450 ◦C),similar to our sample in the present study, exhibited minimal strain (0%and -0.3% in the b and c axis, respectively), based on the results of X-raydiffraction analyses. [28] They also observed a lattice mismatch betweenβ-FeSi2 and Si(001) substrate occurring every 20 periods of the Si lattice, aphenomenon is not observed in our study. To explain this small strain in theirstudy, they postulated the existence of an interface layer to accommodate aweak interaction between the β-FeSi2 and Si(001) substrate. To address theatomistic structure at the interface, we conducted higher resolution STEM5observations.The left panels in Fig. 4 show experimental STEM images of the β-FeSi2/Si(001) interface at three distinct cross-sectional Si planes : (110),(110), and (100). Within the (110) and (110) planes, the appearance of β-FeSi2 thin film, characterized by alternating vertical and horizontal ovals,is equivalent except the interface. In contrast, within the (100) plane, theatomic arrangement exhibits two discernible types cf contrast and a higherdegree of packing compared to the other orientations. These observed fea-tures can be elucidated by considering the bulk structure of β-FeSi2. Sim-ulated STEM images, illustrated in the middle panel of Fig. 4, utilizing aslab model of β-FeSi2/Si(001) (depicted in the right panel of Fig. 4), demon-strated a noteworthy agreement with the experimental counterparts in termsof both Fe and Si atom configuration and contrast. These observation re-sults are the evidence for the growth of the β-FeSi2 thin film. Note that theTEM image presented by Takemoto et al. [28] does not align with any of theaforementioned images of β-FeSi2 in Fig. 4, but rather looks similar to theTEM image of γ-FeSi2 reported by Nissen et al. [29]A further discussion focuses on the interface structure. We have exam-ined several different structural models for the layer denoted as Si0 in Fig. 4.The structural model provided in Fig. 4 yields the best match between ex-perimental and simulated images and also provides the lowest energy amongtested models. [30] The characteristic feature of this interface structure isthat β-FeSi2 is stacked on dimers of Si atoms (bond length = 2.51 Å) locatedin the Si0 layer. The detailed atom’s coordinates at the interface can befound in Fig. S2 and Table S1 in the supplementary material.Utilizing the interface model represented in Fig. 4, we calculated itselectronic structure. The bands indicated by red dots displayed in Fig. 5are predominantly localized over Fe atoms in the Fe1 interface layer. Thesebands originate from the 3d orbitals of the Fe atoms and the 2p orbitals of Sidimers (Fig. S7 in the supplementary material). The corresponding chargedensity maps are found in Fig. S5 in the supplementary material. Thesebands are dispersive within the plane parallel to the interface, with two ofthem crossing the Fermi energy. This observation yields an estimated carrierdensity of 1.46 × 1014 cm−2. The plot of these bands around the Fermi levelwithin the first Brillouin zone (BZ) reveals a 2D character (Fig. 6). Sincethese bands appear within the bulk band gaps of both β-FeSi2 and Si, carriertransport is expected to occur along the interface. Based on the analysis ofPDOS (Fig. S6 in the supplementary material), the states associated with6these bands decay to 46% in the Fe2 layer and further diminished to 6% in theFe3 layer. Therefore, even with the presence of a faceted interface geometry,with a height of approximately half unit cell (∼5 Å) as shown in Fig. 3(c), themetallic properties are expected to persist. On the other hand, the shapes ofthe bands in the first BZ near the Fermi energy vary considerably with energy.Consequently, fluctuations in the Fermi energy, due to doping conditions inthe substrate or the β-FeSi2 film and other factors, can significantly alter thetransport properties. We also conducted spin-polarized calculation, resultingin spin-polarized bands for interface (Fig. S9 in the supplementary material).However, since energy gain due to spin-polarized calculation is only 33 meV,spin-polarized transport may be not achieved by thermal smearing at roomtemperature.4. ConclusionWe unambiguously determined the atomistic interface structure of β-FeSi2/Si(001) epitaxially grown on a Si(001) substrate based on cross-sectionalSTEM observations and DFT calculations. The (001) planes of β-FeSi2 andSi bond to each other with two Si dimers per unit cell of β-FeSi2. For theβ-FeSi2 thicknesses used in this study, commensurate epitaxial growth ofβ-FeSi2 and defect-free interface were achieved. Additionally, surface undu-lations observed by STM indicate that the β-FeSi2 thin film is strained. Theformation mechanism of the carpet-like film remains an open question for fu-ture study. Furthermore, DFT band calculation revealed metallic electronicstates at the interface, suggesting that the β-FeSi2/Si(001) interface could bea new 2D material.2D materials, such as graphene, MXenes, transitional metal dichalco-genide monolayers, generally refers to single-layer material, require a het-erostructure to decouple from a substrate and environment, enabling the useof their intrinsic properties. Using an insulator such as hexagonal boron ni-trite can protect their 2D properties both electronically and chemically. The2D system realized at the β-FeSi2/Si(001) interface is inherently protectedby these materials. Experimental verification of the properties of this newmaterial through transport measurement is necessary.CRediT author statementKeisuke Sagisaka: Conceptualization, Project administration, Investi-gation, Formal analysis, Visualization, Writing ‒ Original draft, Writing ‒7Review and Editing. Kazutaka Mitsuishi: Investigation, Formal analysis,Writing ‒ Review and Editing.Declaration of competing interestThe authors declare that they have no known competing financial inter-ests or personal relationships that could have appeared to influence the workreported in this paper.AcknowledgementsWe gratefully acknowledge Tomoko Kusawake for providing expert tech-nical assistance with sample preparation. We also thank David Bowler andJun Nara for their valuable discussions on DFT computational techniques.The work was supported by JSPS KAKENHI (Grant Nos. 17K04985 and22K04860). Computational simulations were performed on the NIMS numer-ical simulator.5. Data availabilityThe data that support the findings of this study are available from thecorresponding authors upon reasonable request.References[1] D. Leong, M. Harry, K. J. Reeson, K. P. Homewood, A silicon/iron-disilicide light-emitting diode operating at a wavelength of 1.5 µm, Na-ture 387 (1997) 686.[2] T. Suemasu, Y. Negishi, K. Takakura, F. Hasegawa, Room Temperature1.6 µm Electroluminescence from a Si-Based Light Emitting Diode withβ-FeSi2 Active Region, J. J. Appl. Phys. 39 (10B) (2000) L1013.[3] T. Suemasu, Y. Negishi, K. Takakura, F. Hasegawa, T. Chikyow, In-fluence of Si growth temperature for embedding β-FeSi2 and resultantstrain in β-FeSi2 on light emission from p-Si/β-FeSi2 particles/n-Si light-emitting diodes, Appl. Phys. Lett. 79 (12) (2001) 1804–1806.8[4] T. Suemasu, K. Takakura, C. Li, Y. Ozawa, Y. Kumagai, F. Hasegawa,Epitaxial growth of semiconducting β-FeSi2 and its application to light-emitting diodes, Thin Solid Films 461 (1) (2004) 209–218.[5] S. Chu, T. Hirohada, H. Kan, T. Hiruma, Electroluminescence and Re-sponse Characterization of β-FeSi2-Based Light-Emitting Diodes, J. J.Appl. Phys. 43 (2A) (2004) L154.[6] T. Sunohara, C. Li, Y. Ozawa, T. Suemasu, F. Hasegawa, Growthand Characterization of Si-Based Light-Emitting Diode with β-FeSi2-Particles/Si Multilayered Active Region by Molecular Beam Epitaxy, J.J. Appl. Phys. 44 (6R) (2005) 3951.[7] M. Takauji, C. Li, T. Suemasu, F. Hasegawa, Fabrication of p-Si/β-FeSi2/n-Si Double-Heterostructure Light-Emitting Diode by MolecularBeam Epitaxy, J. J. Appl. Phys. 44 (4S) (2005) 2483.[8] K. Akiyama, S. Kaneko, Y. Hirabayashi, H. Funakubo, Photolumines-cence properties of Si/β-FeSi2/Si double heterostructure, Thin SolidFilms 508 (1) (2006) 380–384.[9] C. Li, T. Suemasu, F. Hasegawa, Temperature dependence of electro-luminescence from Si-based light emitting diodes with β-FeSi2 particlesactive region, J. Lumin. 118 (2) (2006) 330–334.[10] Y. Ugajin, M. Takauji, T. Suemasu, Annealing temperature dependenceof EL properties of Si/β-FeSi2/Si(111) double-heterostructures light-emitting diodes, Thin Solid Films 508 (1) (2006) 376–379.[11] S. Chu, T. Hirohada, K. Nakajima, H. Kan, T. Hiruma, Room-Temperature 1.56 µm Electroluminescence of Highly Oriented β-FeSi2/Si Single Heterojunction Prepared by Magnetron-Sputtering De-position, J. J. Appl. Phys. 41 (11A) (2002) L1200. .[12] D. Chi, Semiconducting beta-phase FeSi2 for light emitting diode ap-plications: Recent developments, challenges, and solutions, Thin SolidFilms 537 (2013) 1–22.[13] Y. Dusausoy, J. Protas, R. Wandji, B. Roques, Structure cristalline dudisiliciure de fer, FeSi2β, Acta Crystallogr. B 27 (6) (1971) 1209–1218.9[14] T. Suemasu, M. Tanaka, T. F. Hashimoto, Y. Kumagai, F. H. F.Hasegawa, Aggregation of Monocrystalline β-FeSi2 by Annealing andby Si Overlayer Growth, J. J. Appl. Phys. 36 (9A) (1997) L1225.[15] M. T. M. Tanaka, Y. K. Y. Kumagai, T. S. T. Suemasu, F. H. F.Hasegawa, Formation of β-FeSi2 Layers on Si(001) Substrates, J. J.Appl. Phys. 36 (6R) (1997) 3620.[16] O. Romanyuk, K. Hattori, M. Someta, H. Daimon, Surface structure andelectronic states of epitaxial β-FeSi2(100)/Si(001) thin films: Combinedquantitative LEED, ab initio DFT, and STM study, Phys. Rev. B 90(2014) 155305.[17] K. Sagisaka, T. Kusawake, D. Bowler, S. Ohno, Emergence of metallicsurface states and negative differential conductance in thin β-FeSi2 filmson Si(001), Journal of Physics: Condensed Matter 35 (13) (2023) 135001.[18] N. Hiroi, T. Suemasu, K. Takakura, N. Seki, F. Hasegawa, DirectGrowth of [100]-Oriented High-Quality β-FeSi2 Films on Si(001) Sub-strates by Molecular Beam Epitaxy, J. J. Appl. Phys. 40 (10A) (2001)L1008.[19] T. Sunohara, K. Kobayashi, T. Suemasu, Epitaxial growth and char-acterization of Si-based light-emitting Si/β-FeSi2 film/Si double het-erostructures on Si(001) substrates by molecular beam epitaxy, ThinSolid Films 508 (1) (2006) 371–375.[20] S. Murase, T. Sunohara, T. Suemasu, Epitaxial growth and lumi-nescence characterization of Si/β-FeSi2/Si multilayered structures bymolecular beam epitaxy, J. Cryst. Growth 301-302 (2007) 676–679.[21] Y. Maeda, Y. Terai, M. Itakura, N. Kuwano, Photoluminescence prop-erties of ion beam synthesized β-FeSi2, Thin Solid Films 461 (1) (2004)160–164.[22] The results of the examination of the top surface termination usingpseudo-hydrogen atoms with various charge quantities are provided inthe supplementary material.[23] G. Kresse, J. Hafner, Ab initio molecular dynamics for liquid metals,Phys. Rev. B 47 (1993) 558–561.10[24] G. Kresse, J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54 (1996)11169–11186.[25] J. P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approxima-tion made simple, Phys. Rev. Lett. 77 (1996) 3865–3868.[26] H. J. Monkhorst, J. D. Pack, Special points for brillouin-zone integra-tions, Phys. Rev. B 13 (1976) 5188–5192.[27] J. M. Hartmann, A. Abbadie, S. Favier, Critical thickness for plasticrelaxation of SiGe on Si(001) revisited, J. Appl. Phys. 110 (8) (2011)083529.[28] S. Takemoto, K. Hattori, M. Someta, A. N. Hattori, H. Tanaka, K. Ku-rushima, Y. D. H. Otsuka, Shape-fitting analyses of two-dimensionalX-ray diffraction spots for strain-distribution evaluation in a β-FeSi2nanofilm, J. Appl. Cryst. 52 (2019) 732–744.[29] H.-U. Nissen, E. Müller, H. R. Deller, H. Von Känel, TEM investigationof iron disilicide films on Si(001) grown by molecular beam epitaxy,Phys. Status Solidi A 150 (1) (1995) 395–406.[30] Representative slab models tested in this work, along with their energeticstability, can be found in Fig. S1 in the supplementary material.11Figure 1: Stick-and-ball models for the unit cell of β-FeSi2 and Si(001) substrate.12Figure 2: STM image of a β-FeSi2 film grown on a Si(001) substrate. Imaging parameters:sample bias of -1.5 V, tunneling current of 3pA, and sample temperature of 4.3 K, Theimage dimensions are 70 × 50 nm2, Inset: Sample bias of -1.0 V, tunneling current of 0.1nA, and image dimensions of 10 × 10 nm2, The bottom graph presents a cross-sectionalprofile along the indicated line in the STM image.13Figure 3: Cross-sectional STEM image of a β-FeSi2 film grown on a Si(001) substrate.Horizontal image dimensions: (a) 65.7 nm, (b) 13.5 nm, (c) 6.5 nm. Red lines drawn in(b) and (c) serve as a visual guide to identify the commensurate configuration betweenthe β-FeSi2 and Si lattices and the interface, respectively.14Figure 4: Experimental and simulated STEM images (Left and Middle) and structuralmodel (Right) of the β-FeSi2/Si(001) interface.15Figure 5: Band diagram projected onto the (001) plane and PDOS of the β-FeSi2/Si(001)interface. The bands represented by red, orange, and black dots originate from the interfaceof β-FeSi2/Si(001), bulk β-FeSi2, and bulk Si, respectively. The bands represented by bluedots corresponds to the top most Si atoms of the slab, which are terminated with pseudo-Hatoms. These assignments were performed by examining the charge density distributionof each eigenvalues within the slab. The bands depicted in black lines have not beenexamined for their origin. (Inset) The notation of the BZ is consistent with that used forthe Si(001) surface.16Figure 6: Contour maps for the β-FeSi2/Si(001) interface bands within the first Brillouinzone at various energies around the Fermi level. The red lines represents the bands for theinterface states, while the blue dashed circles denotes the bands for the topmost surfacesilicon atoms terminated by pseud-hydrogen atoms of the slab.17