# Fileset

[Sakuma_et_al-2026-Nature_Communications.pdf](https://mdr.nims.go.jp/filesets/28c96eae-8811-473a-91b5-e6271a6fac8d/download)

## Creator

[Yoshiki Sakuma](https://orcid.org/0000-0001-6804-7217), Keisuke Atsumi, [Takanobu Hiroto](https://orcid.org/0000-0002-6176-5782), [Jun Nara](https://orcid.org/0000-0002-0486-2981), [Akihiro Ohtake](https://orcid.org/0000-0002-3519-4613), Yuki Ono, Takashi Matsumoto, Yukihiro Muta, Kai Takeda, Emi Kano, Toshiki Yasuno, Xu Yang, Nobuyuki Ikarashi, Asato Suzuki, Michio Ikezawa, Shuhong Li, Tomonori Nishimura, Kaito Kanahashi, Kosuke Nagashio

## Rights

[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Self-aligned and self-limiting van der Waals epitaxy of monolayer MoS2 for scalable 2D electronics](https://mdr.nims.go.jp/datasets/c3b34bcc-856c-400a-986f-8bb917ed74ee)

## Fulltext

Self-aligned and self-limiting van der Waals epitaxy of monolayer MoS2 for scalable 2D electronicsArticle https://doi.org/10.1038/s41467-026-68320-8Self-aligned and self-limiting van der Waalsepitaxy of monolayer MoS2 for scalable 2DelectronicsYoshiki Sakuma 1 , Keisuke Atsumi2, Takanobu Hiroto 3, Jun Nara4,Akihiro Ohtake 1, Yuki Ono5, Takashi Matsumoto5, Yukihiro Muta5, Kai Takeda5,Emi Kano 6, Toshiki Yasuno6, Xu Yang6, Nobuyuki Ikarashi6, Asato Suzuki7,Michio Ikezawa7, Shuhong Li2, Tomonori Nishimura 2, Kaito Kanahashi2 &Kosuke Nagashio 2Unidirectional nucleation followed by seamless stitching has emerged as apromising strategy for the scalable epitaxial growth of single-crystallinemonolayer transition metal dichalcogenides on sapphire substrates, whichholds potential for post-silicon electronics. In contrast, here we present adifferent growth mechanism for single-crystalline MoS2 on c-plane sapphirevia metal-organic chemical vapor deposition (MOCVD). We show that theinitial nucleation generates not only 0° and antiparallel 60° domains but alsolow-angle twisted domains, consistent with the coincidence site lattice fra-mework. However, these rotationally misoriented domains are observed todeterministically self-align and merge into energetically preferred 0° domainduring coalescence, yielding a continuous, unidirectional single-crystal.Additionally, by employing MoO2Cl2 as a molybdenum precursor, wedemonstrate that the growth of MoS2 occurs in a self-limiting manner. Thisepitaxial strategy is substantiated by a carrier mobility of 66 cm2/Vs at roomtemperature and 749 cm2/Vs at low temperatures. Our approach offers apractical and reproducible scheme for MOCVD-based van der Waals epitaxyfor 2D electronics.To fully exploit two-dimensional (2D) transitionmetal dichalcogenides(TMDCs) for future logic semiconductors at sub-1 nm nodes1, owing totheir inherent immunity against short-channel effects, scalable epi-taxial growth techniques for single-crystalline monolayer TMDC filmsare in critical demand2,3. Scientifically, a thorough and in-depthunderstanding of pure van der Waals (vdW) epitaxy between 2Dmaterials is indispensable to explore new frontiers in the 2D researchfield. From an industrial perspective, however, realizing 2D materialson a wafer scale requires prioritized development of quasi-vdW epi-taxy on large 3D crystalline substrates such as sapphire (α-Al2O3),which exhibits superior crystalline quality, exceptional chemical,thermal stability, as well as proven scalability to 300mm in diameter4.From a purely symmetry-based perspective, heteroepitaxy of athreefold-symmetric monolayer MoS2 (D3h) on a threefold-symmetricc-plane sapphire (C3v) should, in principle, yield a single crystal underequilibriumconditions5. In practice, however, the growthoccursunderReceived: 30 May 2025Accepted: 2 January 2026Check for updates1Research Center for Electronic and Optical Materials, National Institute for Materials Science, Tsukuba, Japan. 2Department of Materials Engineering, TheUniversity of Tokyo, Tokyo, Japan. 3ResearchNetwork and Facility Services Division, National Institute forMaterials Science, Tsukuba, Japan. 4ResearchCenterfor Materials Nanoarchitectonics, National Institute for Materials Science, Tsukuba, Japan. 5Tokyo Electron Technology Solutions Limited, Hosaka-cho,Nirasaki, Japan. 6Institute of Materials and Systems for Sustainability, Nagoya University, Furo-cho, Chikusa-ku, Nagoya, Japan. 7Department of Physics,University of Tsukuba, Tsukuba, Japan. e-mail: sakuma.yoshiki@nims.go.jp; nagashio@material.t.u-tokyo.ac.jpNature Communications |          (2026) 17:602 11234567890():,;1234567890():,;http://orcid.org/0000-0001-6804-7217http://orcid.org/0000-0001-6804-7217http://orcid.org/0000-0001-6804-7217http://orcid.org/0000-0001-6804-7217http://orcid.org/0000-0001-6804-7217http://orcid.org/0000-0002-6176-5782http://orcid.org/0000-0002-6176-5782http://orcid.org/0000-0002-6176-5782http://orcid.org/0000-0002-6176-5782http://orcid.org/0000-0002-6176-5782http://orcid.org/0000-0002-3519-4613http://orcid.org/0000-0002-3519-4613http://orcid.org/0000-0002-3519-4613http://orcid.org/0000-0002-3519-4613http://orcid.org/0000-0002-3519-4613http://orcid.org/0000-0001-6134-4980http://orcid.org/0000-0001-6134-4980http://orcid.org/0000-0001-6134-4980http://orcid.org/0000-0001-6134-4980http://orcid.org/0000-0001-6134-4980http://orcid.org/0000-0002-8000-5164http://orcid.org/0000-0002-8000-5164http://orcid.org/0000-0002-8000-5164http://orcid.org/0000-0002-8000-5164http://orcid.org/0000-0002-8000-5164http://orcid.org/0000-0003-1181-8644http://orcid.org/0000-0003-1181-8644http://orcid.org/0000-0003-1181-8644http://orcid.org/0000-0003-1181-8644http://orcid.org/0000-0003-1181-8644http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-68320-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-68320-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-68320-8&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-026-68320-8&domain=pdfmailto:sakuma.yoshiki@nims.go.jpmailto:nagashio@material.t.u-tokyo.ac.jpwww.nature.com/naturecommunicationsnonequilibrium conditions, and early studies reported the formationof antiparallel domains6. Consequently, the control of antiparalleldomains has become a critical technical challenge to be addressed inrealizing single crystallinity. To overcome these challenges, two majorstrategies known as step-edge-guided and symmetry-guided epitaxyhave been proposed. Specifically, in the step-edge-guided mode7, thestep edges help to lower the surface symmetry to C1 and break thedegeneracy of nucleation energy between twined islands, guiding theirin-plane orientation along a single preferred direction. In contrast, thecrucial role of atomic symmetry at the terrace surface is highlighted inthe symmetry-guided epitaxy. A single-type O-Al atomic slab structurewith even number of surface steps on c-plane sapphire, which wasexposed by deliberately engineered miscut angles, strongly facilitatesunidirectional growth of MoS28. Consequently, some studies havesuccessfully demonstrated the unidirectionally aligned TMDC films onsapphire substrates by employing techniques such as powder-sourcechemical vapor deposition (CVD)7–13, metal-organic CVD (MOCVD)14,15and other techniques16.However, although both mechanisms are conceptually plausible,their underlying atomistic processes remain elusive. This lack of clarityarises from unresolved questions regarding the surfacereconstructions14,17 and chemical adsorbates on sapphire in growthenvironment, the atomic configurations of sapphire step edges7,8, andthe edge terminations of TMDC grains18,19. Moreover, it has beenreported that the sulfur to metal precursor supply ratio13,20–22, alongwith the introduction of H223, influences the in-plane alignment ofTMDC nuclei on sapphire. In fact, these growth conditions can beutilized to underscore and control the favored growth mode sepa-rately. Conversely, this implies that both mechanisms inherentlycoexist in the growth on the same vicinal sapphire surface.Moreover, the formation of various interfacial layers on sapphiresurfaces15,24,25 has also been identified as playing a critical role indetermining the in-plane crystallographic orientations of MoS2. Nota-bly, the underlyingmechanism responsible for the emergence ofMoS2with two distinct orthogonal configurations on c-plane sapphire (0°/60° and 30°/90°) under different experimental conditions remains tobe fully elucidated. These complexities impede a comprehensive andunified understanding of quasi-vdW growth mechanism at nucleationstage, particularly pronounced in MOCVD-based research due to lim-ited investigations. Nonetheless, it should be emphasized that all theabove-mentioned studies are based on a common approach, in whichaligning all the nuclei in the same in-plane orientation and subsequentseamless stitching are regarded as a guiding principle26, as depictedin Fig. 1.In this study, unlike the conventional strategy, we report a fun-damentally distinct self-aligned coalescence mechanism for wafer-scale single-crystalline MoS2 films in an MOCVD-type reactor usingMoO2Cl2 and H2S as precursors, as shown in Fig. 1. Besides antiparallelnuclei, isolated MoS2 domains initially nucleated with well-defined in-plane twisted angles elucidated by coincidence site lattice (or super-cell) model between MoS2 and sapphire. Subsequently, upon domainimpingement, both the low-angle twisted and antiparallel 60° domainsare self-aligned and merged into the most energetically favorable 0°domain during coalescence process. Consequently, only the 0°domain survives, culminating in a deterministic formation of con-tinuous, unidirectionally aligned single-crystalline MoS2 film. Anothernoticeable feature of MoS2 growth using MoO2Cl2 precursor is a self-limiting manner over a wide range of growth conditions. As a proof ofhigh-quality MoS2 films with negligible grain boundaries, carriertransport exhibited typical power law behavior, yielding high electronmobilities of 66 cm2/Vs at room temperature and approximately749 cm2/Vs at low temperatures.ResultsEpitaxial registry and self-aligned coalescenceAn industrially compatible c-plane sapphire substrate with an orien-tation flat plane of (�1�120) and a miscut angle of 0.15°–0.2° towardm-axis of [1�100] direction (notated as c/m: −0.15° or −0.2°), manu-factured by Orbray Co. Ltd., was utilized as a standard substrate in thisstudy. Prior to growth, the substrate was annealed in air for 1 hour at1150 °C to form parallel monolayer atomic steps along a-axis of [11�20]with a terrace width of approximately 60–80 nm27. Monolayer MoS2films were grown via MOCVD reactor, utilizing MoO2Cl2 and H2S asprecursors in N2 carrier gas. Growth temperatures ranged from800 °Cto 1050 °C,with amaximumgrowthduration of 2 hours (seeMethods).The detailed description of the MOCVD process steps can be seen inSupplementary Fig. 1. Notably, MoO2Cl2 precursor not only exhibitsadequate volatility comparable to that of conventional Mo(CO)6around room temperature, but also the advantage of eliminating car-bon contamination28. At a growth temperature of 950 °C and a dura-tion of 30min, isolatedMoS2 domainswere clearly observed, as shownin Fig. 2a. These domains comprised antiparallel triangular domains ofFig. 1 | Strategies forwafer-scale epitaxial growthof transitionmetal dichalcogenides (TMDCs) on sapphire substrates. Schematic illustration showing the challengefor single-crystallization of TMDCs and two kinds of growth strategies for single crystal formation.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 2www.nature.com/naturecommunications0° and 60°, approximately 150 nm in size, oriented along [1�100] and[�1100], respectively. In terms of the quantity ratio of their domains,approximately 55% exhibited 0°, with the remainder at 60°. Uponcloser inspection, however, ~11° and ~49° domainswere alsodiscerned,as shown by blue dotted lines. Extending the growth duration to60min resulted in a fully continuous monolayer MoS2 film. It is note-worthy that the terrace width of the sapphire surface increased afterthe growth significantly14 probably due to the highly reactive nature ofH2S gas at elevated temperatures exceeding 950 °C (SupplementaryFig. 2). For the fully-covered sample, the epitaxial relationshipof [11�20]MoS2//[11�20]α-Al2O3 was confirmed through in-plane X-ray diffraction(XRD) ϕ scans (Supplementary Fig. 3), which is consistent with theprevious literature for MoS2 growth on sapphire substrates byMOCVD17,29 and powder-source CVD6,21,30.We performed a detailed analysis of the orientational relation-ships between the MoS2 and sapphire substrate for isolated domainsgrown at 975 °C for 20min. As depicted in Supplementary Fig. 4, high-resolution in-plane XRD measurement revealed the presence of addi-tional subpeaks at azimuthal angles of approximately 10.87° (averageof doublet), 19°, 30°, 41°, and 49.13° between the principal peaks at 0°and 60°, which are equivalent to 0° ± 10.87°, 30°±10.87°. Althoughsome of these subpeaks were also reported previously14,31,32, earlierworks do not have quantitatively captured either the presence or thetrue areal fraction of twisteddomains due to the limited resolution andscan rates, compared with the present precise measurement at step of0.02°, scan speed of 0.3°/min (total time for 180° scan is 10 hours).Therefore, their origin and implications have yet to be apparentlyelucidated. The observed azimuthal angles exhibit excellent agree-ment with those derived from supercells between the commensurateMoS2 and c-plane sapphire substrate, as detailed in the table ofSupplementary Fig. 46. Moreover, the subpeak with 10.87° twistedanglewas resolved into a doublet at 10.42° and 11.32° (average: 10.87°),although the underlying mechanism remains unclear. Notably, theprominent peaks at 0° and 10.87° with large XRD intensities corre-spond to the two smallest supercell sizes in the table, suggesting thatthe epitaxial registry of tiny nuclei at the early stage of MoS2 growth islikely governed by coincidence site lattice model. The presence ofisolated twisted domains expected from this in-plane XRD appears tobe in good agreement with the AFM images of the early-stage sampleafter 4min of growth, as shown in Supplementary Fig. 5.The time evolution of MoS₂ growth at 975 °C was systematicallyexamined to elucidate the pathway to full surface coverage, as depic-ted in Supplementary Fig. 6. Figure 2b overlays the data correspondingto the 0° domain at different growth durations, highlighting the cor-relation between primary 0° peak and doublet 10.42°/11.32° subpeaks.Intriguingly, the intensity of these doublet peaks initially exhibited amonotonic increase with the growth duration, followed by a rapiddecrease around 50min, ultimately vanishing altogether. Con-currently, a broad tail structure at the foot of the 0° peak emerged atapproximately 40min, and disappeared by 60min. Throughout thisprocess, the area intensity of the primary 0° peak exhibited a con-tinuous increase, as shown in the inset of Fig. 2b. To attain the real-space observation of this coalescence processes, the technique usingdifferential dark-field transmission electron microscopy (DF-TEM)analysiswasdevelopedhere.While in-planeXRD lacks the sensitivity tounambiguously resolve antiparallel domains of 0° and 60°, electrondiffraction techniques are capable of their clear differentiation33.Although Friedel’s law describes that the intensities of hkl and -h-k-lreflections are equivalent even for polar crystals, this equivalence isviolated in monolayer TMDCs owing to the broken in-plane inversionFig. 2 | Annihilation of antiparallel & low-angle twisted domains during coa-lescence of MoS2 domains on sapphire substrates. a Atomic force microscope(AFM) images ofMoS2 grown on the sapphire substrate at 30 and 60min at 950 °C.b in-plane X-ray diffraction (XRD) intensity around the 0° peak for the ϕ scan ofMoS2 acquired at different growth durations at 975 °C. The inset shows the inten-sities of XRDpeaks at 0° and ~ 11° as a function of growth durations. The sumof theintensity of doublet subpeaks at 10.42° and 11.32° was plotted as ~11°. c False-colormaps obtained by time-resolved differential dark-field transmission electronmicroscopy (DF-TEM) analysis of MoS2 grown at 975 °C. The area ratio of eachdomain was calculated based on the absolute coverage. d four-dimensional-scanning transmission electron microscopy (4D-STEM) data of MoS2 grown at975 °C. e Second harmonic generation (SHG) intensity map and polarizationdependence of SHG at different 25 locations (see Supplementary Fig. 13) for MoS2grown at 975 °C.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 3www.nature.com/naturecommunicationssymmetry and multiple scattering effects, resulting in anomalous dif-fraction intensity contrasts34. As shown in Supplementary Fig. 7,therefore, by taking the difference between DF-TEM images obtainedfrom a pair of diffraction spots related by 180° inversion, where Frie-del’s law is violated, 0° and 60° antiparallel domains are unambigu-ously distinguished. Moreover, comparison with the correspondingbright-field (BF) TEM image allows for the discrimination of low-angletwisted domains in addition to the 0° and 60° antiparallel domains,thereby enabling the construction of comprehensive false-color mapsfor monolayer MoS2 grown at 975 °C for various growth durations, asshown in Fig. 2c. From 25 to 35min, nearly all domains retain theirtriangular morphology and a substantial fraction of both 60° anti-parallel and low-angle twisted domains clearly remains. Interestingly,from45 to 53min, 60° antiparallel domains aswell as low-angle twisteddomains, which was observed in in-plane XRD, are significantlyreduced due to coalescence and almost annihilated by 55min, result-ing in the fully covered single-crystalline MoS2 film. It should beemphasized that such systematic time evolution is uniquely accessiblethrough the highly reproducible nature of MOCVD growth, asdemonstrated in Supplementary Fig. 8, whereas acquiring suchdetailed and consistent data is exceedingly difficult with conventionalpowder-source CVD.These observations strongly indicate the following scenario. Inthe early stage of growth, small nuclei formed independently whilemaintaining their twisted angles elucidated by supercell model. Uponimpingement of the domains in lateral at approximately 40min, thelow-angle twisted and the antiparallel 60° domains started to mergeinto the energetically favorable 0° domain to form a single crystal viathe self-aligned seamless stitching. Finally, complete surface coveragewas attainedby60min. Crucially, differentialDF-TEM images obtainedat ten different locations across the 3-mmϕ TEM grid, as shown inSupplementary Fig. 9, clearly confirm the single-crystalline nature ofthe film with only minimal occurrences of antiparallel and twistedorientations. Here, it is important to note that, using the time-resolvedin-plane XRD, this time evolution was also observed at a reducedtemperature of 950 °C (Supplementary Fig. 10) and even with alter-native precursors of Mo(CO)6 and H2S at 950 °C (SupplementaryFig. 11), suggesting that the formation of single crystals via mergingand seamless stitching following the impingement is a generalizablephenomenon, not limited to the MoO2Cl2 precursor.To further gain profound insights into the coalescence process ofMoS2 grains in fully continuous films grown at different temperatures,four-dimensional-scanning transmission electron microscopy (4D-STEM)29,35 and polarization-resolved second harmonic generation(SHG)36 was utilized to visualize the orientation of grains at the mul-tiple spatial resolutions from nanometer to micrometer, as illustratedin Fig. 2d, e and Supplementary Figs. 13. As the growth temperatureincreases, the low-angle twisted domains and antiparallel domains,initially incorporated into the dominant 0° domains at 900 °C, weresignificantly reduced, eventually disappearing almost entirely at975 °C. These observations strongly suggest that a relatively hightemperature of 975 °C is critical for achieving complete single-crystalformation via self-aligned coalescence. Although self-aligned singlecrystallization at low temperature is theoretically achievable, it hasproven experimentally impractical, as shown in Figure 14. Moreover,SHG is highly sensitive to strain levels as low as 0.1%, which remainundetectable by Raman spectroscopy37. The polar plot of the parallelSHG component for MoS2 grown at 975 °C reveals a uniform six-foldsymmetry with maximum intensity along the armchair direction,confirming the absence of substantial residual strain in the film.Deterministic epitaxial alignment of MoS2/sapphireThe unique phenomenon of single-crystal formation via self-alignedcoalescence, as observed in Fig. 2c, suggests the existence of adeterministic crystallographic orientation of MoS2 on the sapphiresubstrate. Figure 3a presents a cross-sectional high-angle annular darkfield scanning TEM (HAADF-STEM) image of MoS2 grown on the sap-phire substrate at 975 °C, observed from the [11�20] direction. Ther(1�102), R(�1104), and m(1�100) faces of the sapphire substrate wereclearly discerned, corroborating the crystallographic asymmetry along+m[�1100] and -m[1�100], i.e., the threefold symmetry of sapphire,depicted in the inset. Previous studies havediscussed the presence of abuffer layer on the sapphire substrate13,15,22. In ourHAADF-STEM image,this buffer layer exhibited weaker contrast than in previous report,which likely facilitates the formation of an epitaxial relationshipbetween MoS2 and sapphire. Although a vdWMoS2/sapphire distancewas observed as ~0.7 nm for bottomS-topAl distance7,8,13,25, the precisein-plane atomic alignment ofMoS2 on the sapphire substrate remainedunclear.To clarify this, low-energy electron diffraction (LEED) measure-ments were performed for monolayer MoS2 grown on the standardsapphire substrate (c/m = −0.2°) at 975 °C at specific voltage of 190 eVacross eight positions, each spaced 1mmapart. As shown in Fig. 3b andSupplementary Fig. 15, all exhibited a consistent threefold symmetricdiffraction pattern of single-crystalline monolayer MoS27,13, indicatingthe uniform alignment of MoS2 on a wafer scale. Furthermore, theatomic configuration of monolayer MoS2 on the sapphire substratewas uniquely determined by analyzing the LEED intensity-voltage (I-V)profiles from 10 and 01 LEED spots. It was revealed that the MoS2structural model shown in Fig. 3c can be directly overlaid onto thesapphire substrate schematically drawn. Additionally,whenmonolayerMoS2 was grown on custom-ordered sapphire substrates with miscutangle of 0.2° along +m-direction (c/m= +0.2°) and a-direction (c/a = +0.15°), identical LEED patterns were obtained irrespective of themiscut direction, as shown in Supplementary Fig. 15. The confirmationof miscut angle of sapphire substrates can be found in SupplementaryFig. 16. Given the inherent threefold symmetry of the c-sapphire sur-face, variations in themiscut direction inevitably lead to differences inthe atomic configurations at the step edges. Therefore, it can beconcluded that the orientation of MoS2 during the MOCVD growth isgoverned not by the step edges but by its interaction with the surfacestructure of the sapphire substrate. This interpretation is furthersupported by the AFM images in Supplementary Fig. 5, which showthat the initial growth nuclei on sapphire are not influenced by thesurface steps. Furthermore, as shown in Supplementary Fig. 2, theterrace width increases dynamically during MoS2 growth. A recentstudy highlighted the role of step-edge reconstructions on sapphiresurfaces in guiding epitaxial alignment during the nucleation stage12. Incontrast, the present work emphasizes the subsequent coalescencestage, not the nucleation stage, which constitutes a critical distinctionbetween the two studies. Consequently, the growth process reportedhere does not require fine control of the miscut direction to tailor theatomic configuration of steps and terraces, thereby ensuring bothrobustness and reproducibility.Thus, the orientation of MoS2 on the sapphire substrate grown at975 °C is uniquely determined, with the MoS2 [11�20] and [�1100]directions aligning with the sapphire [11�20] and [�1100] directions,respectively. MoS2 with this orientation is indeed the 0° domaindepicted in Fig. 3c. It should be noted that this configuration matchesthat reported in previous growth studies (0°/60°)6, while differingfrom the orientation relationships observed in other reports (30°/90°)7,8,13,38, as illustrated in Supplementary Fig. 17. As mentioned in theintroduction, thismaybe related to specific interfacial andbuffer layerson sapphire surface caused by the precursor supply ratio (S/Mo)12,21,22.Subsequently, we aim to explore the mechanism by which low-angle twisted and antiparallel domains are annihilated. As illustrated inFig. 3d, two mechanisms can be considered: the rotation model andthe grain boundary (GB) migration model. In general, the rotation ofantiparallel grains in opposite orientation might be possible in vdWheteroepitaxy due to the weak interaction39. However, time-resolvedArticle https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 4www.nature.com/naturecommunicationsdifferential DF-TEM images in Fig. 2c clearly indicate that there isinadequate space to accommodate the rotation process when coa-lescence initiates ~40min after the commencement of growth. Incontrast, GBmigrationmodel involvesmass transfer between adjacentgrains across the boundary via atomic diffusion within the solid,without requiring additional spatial allowance. When low-angle andantiparallel domains are in physical contact with 0° domains, themovement ofMo and S atoms, facilitated by vacancies and interstitials,will drive GB migration. Therefore, a diffusion-coupled GB migrationmodel is the most plausible explanation.It is also important to discuss the apparent onset time of coales-cence, around 40min, as indicated in Fig. 2c. Notably, residual low-angle and antiparallel domains are consistently accompanied byungrown peripheral regions (black contrast). This correlation stronglysuggests that GBmigration is interrupted at ungrown regions. In otherwords, sustained GB migration toward a unidirectional single crystalproceeds only after individual misoriented domains become fullysurrounded by 0° domain, that is, when the GB network becomescontinuous and is no longer interrupted by ungrown regions.This picture implies that GB migration also contains tangentialmotion along the GB, leading to progressive shrinkage and eventualdisappearance of the enclosed misoriented domains. This processstarts locally as the coverage of MoS2 on the sapphire substrateincreases, and then propagates across the wafer. After 60min, whenMoS2 fully covers the substrate, the self-aligned coalescence is com-pleted over the entire wafer, resulting in wafer-scale single crystal-lization. This evolution likely explains why full coverage and wafer-scale single crystallization occur nearly simultaneously at around60min. Therefore, under our growth conditions, the local surfacecoverage required to form fully enclosed regions is first achieved atapproximately 40min. This time point can thus be regarded as theapparent “threshold” for the onset of self-aligned coalescence in ourexperiments. Additional experiments supporting these growthmechanisms are presented in Supplementary Fig. 18.The self-aligned growth observed here has not been reported inprevious studies. Nevertheless, twisted domains are not unique to oursamples; they have also been observed in MOCVD- and MBE-grownfilms from other groups14,31,32. A key distinction is that such films haveseldom been examined using sufficiently high-sensitivity character-ization. In our in-plane XRD measurements, for instance, we employthe conditions of slow scan speed so that the subpeaks from therotational domains can be observed with a sufficiently good signal-to-noise ratio. Moreover, many powder-CVD reports primarily presentAFM images of comparatively large domains (~10–100μm), withoutresolving the early nucleation regime. To date, no study has system-atically and quantitatively traced the full evolutionary pathway—fromthe emergence and coalescence of smaller, rotated early-stage nucleito the ultimate seamless stitching into large single-crystal domains—through a combined analysis using in-plane XRD, low-magnificationDF-TEM, and 4D-STEM. This gap stems in part from the experimentaldifficulty of conducting time-resolved evolution studies in powder-CVD. Consequently, the existing evidence remains inadequate tosupport a definitive conclusion.The orientation-selective growth process is classically recognizedas epitaxial grain growth (EGG) in polycrystalline thin films on singlecrystalline substrates40. EGG is driven by the minimization of crystal-lographically anisotropic free energies, such as surface and interfaceenergies. At the monolayer limit of 2D materials, as discussed in Sup-plementary Fig. 19, MoS2/sapphire interface energy is expected to playa dominant role. Therefore, to elucidate the driving force, densityfunctional theory (DFT) calculations were conducted to compute theadsorption energies between the MoS2 and sapphire substrate atFig. 3 | Determination of unique MoS2/sapphire orientation & growthmechanism. a Cross-sectional high-angle annular dark-field STEM (HAADF-STEM)image ofmonolayerMoS2 grownon the sapphire substrate at 975 °C.b Low-energyelectron diffraction (LEED) patterns of MoS2 grown on sapphire substrates withopposing miscut angles. c Definition of orientation for the sapphire substate andmonolayer MoS2. Note that the orientation of monolayer MoS2 is defined by thebottom layer of the unit cell, as specified in the standard crystallographic infor-mation (cif) file. d Illustration of two distinct growth mechanisms. e Schematicrepresentation of the energy relationship between the 0°, 10.9° and 60° domainswith adsorption energy differences calculated by density functional theory (DFT).GB: grain boundary, ΔEab: adsorption energy difference. Yellow arrow indicates thedirection of atomic diffusion.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 5www.nature.com/naturecommunicationsvarious twisted angles, assuming commensurate infinite monolayerMoS2 slabs, as shown in Supplementary Fig. 20. Figure 3e summarizesthe relative adsorption stabilities of the 0°, antiparallel, and low-angletwisted domains. In isolation from the substrate, all domains areenergetically equivalent. However, during coalescence, themerging of10.9° domains into 0° domains is driven by an adsorption energy dif-ference of ΔEad = −35meV/MoS2 unit via the GB migration. On theother hand, ΔEad between 0 and 60 antiparallel domains should existbut was negligibly small (~−0.1meV/MoS2 unit), indicating that thegrain boundary energy must be considered in further studies, as dis-cussed in Supplementary Fig. 20. Nevertheless, it is important toemphasize that even such small energy differences can induce GBmigration, as reported for adjacent graphene islands on Ir(111), wherethe driving forcewas on the order of0.1meVperC atom39. The presentenergy considerations are supported by the differential DF-TEM and4D-STEM observations (Fig. 2c, d and Supplementary Fig. 13).Wafer scale uniformity via self-limiting growthTo guarantee the reliable performance of electronic devices whoseproperties are highly sensitive to the precise number of 2D materiallayers, self-limited layer-controlled synthesis provides great technicaladvantages. While atomic layer deposition (ALD) is among the mostpromising techniques owing to its surface adsorbed growth mechan-ism, substantial progress is still required to enhance the crystallinity ofMoS2 films41,42. In the epitaxial growth of TMDC on the sapphire sub-strate via MOCVD, precise layer control remains a significantchallenge, primarily due to the nonuniform gas flow pattern acrosswafer surface and the consumption of precursors along the gas flowdirection in reactor24,43,44. Here, although the MoO2Cl2 and H2S pre-cursors were supplied simultaneously, unlike ALD, a self-limitedmonolayer growth of MoS2 was demonstrated as shown in Fig. 4aand Supplementary Fig. 21. The absorbance at C exciton measured forMoS2 films, grown on double-side polished sapphire substrates at850 °C, saturates beyond 60min, in contrast to the continuous growthtendency observed using Mo(CO)6 and H2S precursors. Moreover, theself-limiting growth is rather profound at an elevated temperature of975 °C under optimized growth conditions, as shown in Fig. 4a andSupplementary Fig. 22. This is further supported by AFM images inFig. 4b and Supplementary Fig. 23, where no voids and very limitedsecond layer growth of MoS2 are discernable even at the growthduration longer than 60min. Moreover, the tail of main in-plane XRDpeak at 0° in Fig. 2b and Supplementary Fig. 6 no longer broadens forMoO2Cl2, while broad tail behavior observed due to themosaic spreadcaused by the second epitaxial layer of MoS2 for Mo(CO)6 in Supple-mentary Fig. 11. Themechanismof this self-limiting is attributed to site-selective adsorption, wherein theMoO2Cl2 precursor, unlikeMo(CO)6,exhibits a low affinity for adsorption on theMoS2 surface. The detaileddiscussion can be found in Supplementary Fig. 22.This self-limited layer-controlled synthesis enables wafer-scaleuniformity in MoS2 growth, as shown by a 2-inch monolayer MoS2/c-sapphire substrate in Fig. 4c. The quality and uniformity of the as-grown monolayer MoS2 films are illustrated via multi-scaleFig. 4 | Wafer scale uniformity via self-limiting growth. a Absorbance of MoS2grown on sapphire substrates at 850 °C and 975 °C as a function of growth dura-tions using different precursors. In case of Mo(CO)6 precursor, only data at 850 °Cis shown. The horizontal dashed line indicates the absorbance of fully-coveredmonolayerMoS2. b AFM images ofMoS2 grown on sapphire substrates at 975 °C atvarious growth durations. c Photograph of a 2-inch monolayer MoS2/c-sapphiresubstrate.d In-planeXRDϕ scans taken atfivedifferent locations,which is shown inthe inset, for monolayer MoS2 grown on a 2-inch sapphire substrate at 950 °C.e Raman A1g intensity map for monolayer MoS2 grown on a 2-inch sapphire sub-strate at 1025 °C. f Photoluminescence (PL) line scans along the wafer for mono-layer MoS2 grown on a 2-inch sapphire substrate at 950 °C.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 6www.nature.com/naturecommunicationscharacterizations. High-resolution in-plane XRDϕ-scans, performed atfive different points on the wafer (Fig. 4d), macroscopically demon-strate the formation of a uniform film without low-angle twisteddomains throughout thewafer. The variation in RamanA1g peaks of themonolayerMoS2 isonly2 cm-1 throughout the 2-inchwafer, as shown inFig. 4e and Supplementary Fig. 24. Moreover, Fig. 4f shows photo-luminescence (PL) line scans across a 2-inch MoS2 wafer with 2mmstep. No obvious variations in peak position and line width wereobserved. These results collectively demonstrate the excellent uni-formity and high reproducibility of MOCVD-grown MoS2 films, span-ning from the sub-micrometer to centimeter scale.Transport properties of monolayer MoS2To evaluate the electronic performance of monolayer MoS2, Hall barand two-terminal field-effect transistor (FET) configurations wereprepared via a standard photolithography process after transferringMoS2 to a SiO2/Si substrate, as shown in Fig. 5a and SupplementaryFig. 25. Figure 5b exhibit transfer curves with limited variationobtained fromMoS2 grown at 1025 °C, showing theon/off current ratioexceeding ~107 and clear saturation. Figure 5c shows output char-acteristics obtained from MoS2 grown at 1025 °C, showing ohmicbehavior using Ni/Au contacts. Moreover, short-channel devices withchannel lengths ranging from 2 μm down to 51 nm were fabricatedusing electron beam lithography, exhibiting an increase in drain cur-rent with decreasing channel length, as shown in SupplementaryFig. 26. This trend highlights their relevance to the practical realizationof 2D nanoelectronics.The mobilities achieved via MOCVD so far are comparativelylower than those obtained using powder-source CVD2, rendering smallcomparisons of μ values at room temperature less meaningful. Instead, our primary focus is placed on the temperature dependence ofmobility (μ)45,46, rather than its absolute vale, as this provides clearerinsight into the underlying electron transport mechanisms. Thereports on the temperature dependence of mobility inMOCVD-grownMoS2 films have been considerably limited47,48, compared with MoS2synthesized via powder-based CVD methods. Figure 5d and Supple-mentary Fig. 27 summarize the μ values of MoS2 channel, obtained byfour terminal measurements, as a function of temperature for MoS2grown under different conditions. As the growth temperature increa-ses from 900 °C to 1025 °C, a marked transition from thermally acti-vated behavior to typical power law behavior (μ ~ T−γ) above 100K. Thetemperature coefficient (γ) in this power law is discussed based on theeffect of the substrate, as shown in Supplementary Fig. 28. Thetemperature-dependent μ for MoS2 grown at 1025 °C attains a highmobility, approximately 66 cm2/Vs at room temperature and 749 cm2/Vs at 20 K, for monolayer MoS2 synthesized by MOCVD. Furthermore,this temperature dependence approaches values similar to the bestdata achieved previously with powder-source CVD49, where largesingle-crystal triangles have been measured on the SiO2/Si substrate,strongly indicating the single crystal formation by the annihilation ofGBs. It is noteworthy that MoS2 properties were characterized on2 × 2 cm2 sapphire wafers, while devices were fabricated using 2-inchwafers. The 2-inchwafers require ~50 °Chigher substrate temperaturesdue to different MOCVD susceptors. The μ values of two-terminalFig. 5 | Transport properties of monolayer MoS2. a Photograph of monolayerMoS2 field-effect transistors (FETs) on a SiO2/Si substrate.bTransfer characteristicsof 48MoS2 FETsgrown at 1025 °C.VBG: back-gate voltage, Id: drain current,Vd: drainvoltage. c Output characteristics of MoS2 FET grown at 1025 °C. d Temperaturedependence of mobility for four-terminal MoS2 FETs grown at various tempera-tures. Coloreddashed curves are eye guides, while black circles represent referencedata for powder-source CVD49. e Statistical analysis of mobility for two-terminalMoS2 FETs grown at various temperatures, where the Y-function method56 wasapplied for evaluation. N represents the number of measured devices. The boxrepresents the interquartile range, the line inside the box indicates themedian, andthe whiskers show the spread of the data. f Benchmark ofmobility values for TMDCgrown by powder-source CVD andMOCVD. Refs. [A], [B], and [C] are adapted fromrefs. 44,57, and 13.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 7www.nature.com/naturecommunicationsMoS2 FETs in Fig. 5e show that the average room-temperature μincreases with increasing growth temperatures. Finally, Fig. 5f bench-marks the maximum and average μ values at room temperature, withthe current study reporting the highest value to date. The mobilitiesreported here remain lower than those obtained via powder-sourceCVD. This discrepancy suggests that a higher defect density may stilllimit carrier transport. The systematic defect characterization andsubsequent reduction will be essential to further enhance mobility,thereby bridging the gap with powder-source CVD.DiscussionIn this study, we have revealed a self-aligned and self-limiting vdWepitaxy for unidirectional single-crystalline MoS2 film on c-plane sap-phire, by employing high-precision, time-resolved in-plane XRD, dif-ferential DF-TEM and 4D-STEM. The epitaxial relationship betweenMoS2 and sapphire substratewasdeterministically elucidated, with theMoS2 [11�20] and [�1100] directions aligningwith the sapphire [11�20] and[�1100] directions, respectively. Our results demonstrate the super-iority of MOCVD technology in growing monolayer TMDCs, providinga major impetus for future industrialization.MethodsMOCVD growthThe MOCVD reactor employed in this study is horizontal-flow andcold-wall configuration. The graphite susceptor for single wafer isresistively heated from its backside and the maximum wafer size is 2inches (Supplementary Fig. 1). Molybdenum oxychloride (MoO2Cl2)and hydrogen sulfide (H2S) were utilized as the molybdenum andsulfur precursors, respectively. TheMoO2Cl2 canisterwas controlled atatmospheric pressure and bath temperature of 18 °C. The flow rate ofN2 passing through the canister was varied from 50 to 400 sccm toregulate the supply of MoO2Cl2. The H2S flow rate was changed withinthe range of 2 to 20 sccm. The total N2 flow rate in the reactor was keptat 2500 sccm. MoS2 film was grown at substrate temperatures rangingfrom 800 °C to 1050 °C, under a constant chamber pressure of50 Torr, with growth durations limited to amaximumof 2 h. All the gasflow rates and pressures were controlled by mass flow controllers andautomatic pressure controllers, respectively. Alternatively, precursorsof Mo(CO)6 and H2S were also utilized for MOCVD growth run forcomparison. The Mo(CO)6 bath temperature is 5 °C and the flow rateN2 carrier gas is fixed at 20 sccm. Calculating from the sublimationpressures of MoO2Cl2 and Mo(CO)6 described in SupplementaryFig. 1c, all the MoS2 growth was conducted under sulfur-rich condi-tions. Sapphire substrates cutting into 2 × 2 cm2 were usuallyemployed for the growth experiments, while 2-inch diameter waferswere utilized for the MoS2 growth used for device fabrication. The2-inch c-sapphire substrates were procured by Orbray Co. Ltd. Prior togrowth, thermal annealing process was performed in a muffle furnaceat 1150 °C in air for 1 h to prepare surface steps aligned along thea-axis [11�20].CharacterizationsTo confirm the crystal quality and thickness of films, Raman spectro-scopy, PL spectroscopy and AFM were employed. Raman and PLmeasurements were conducted with a 532-nm excitation laser with an~1μm spot and ~0.5mW. In-plane XRDmeasurements were performedusing a Rigaku SmartLab system to identify the presence of low-angledomains in monolayer MoS2 (Supplementary Fig. 3). Initially, the dif-fraction peak corresponding to the {11�20} plane of MoS2 was observedto alignwith the {11�20} plane of the sapphire substrate, confirmed via a2θχ/ϕ scan. Subsequently, a ϕ scan of {11�20} planes of MoS2 wasconductedwith a typical scan rate of 0.3°/min and step width of 0.02°.Differential DF-TEM, 4D-STEM and HAADF-STEM experiments werecarried out using a JEOL ARM 200 F TEM with acceleration voltages of80, 200, and 200 kV, respectively. 4D-STEM orientation maps weregenerated from electron diffraction data set using a proprietaryalgorithm based on the NanoMEGAS ASTAR. For HAADF-STEM, thespherical aberration coefficient was less than 1μm. The convergentangle was 17 mrad and the inner and outer angles of the ADF detectorwere 50 and 150 mrad, respectively. The SHG measurements wereconducted using a mode-locked Ti:sapphire laser (wavelength:810 nm, pulse width: ~100 fs, and repetition rate: 82MHz) in a home-built optical microscope configured for backscattering. The LEEDpatterns at room temperature were acquired using a conventionalLEED system (OCI, BDL600IR) under ultrahigh vacuum (UHV) condi-tions at 10−8Pa, with a 1 eV step in the energy range from 30 to 380 eV.The LEED I-V curves forfive non-equivalent beamswere extracted fromLEED patterns with the background subtracted. The total cumulativeenergy range covered was approximately 1455 eV.Theoretical calculations were performed by using the PHASE/0code50, which is based on DFT51 and pseudo-potential schemes52,53 withplane-wave basis sets. For the exchange-correlation term, the PBE formwasused54. For the vanderWaals interactions, the DFT-D2methodwasapplied55. The cut-off energies for the wavefunctions and charge den-sity were 56Ry and 506Ry, respectively. The number of k pointssampled in the Brillouin zone was more than 5 × 5 per the surface unitcell of c-plane sapphire. All the models were optimized to meet theforce criterion of 0.02 eV/Å. Sapphire slab employed here consists ofsixAl2O3 layers and the thickness of a vacuumregion is 0.9 nm. The top(bottom) layer of a sapphire slab is terminated with only one Al atomper surface unit, to make its electronic states semiconducting. Thecalculated lattice constant for the hexagonal MoS2 monolayer is0.318 nm, while that for sapphire is 0.4798 nm. In this study, thechange in the lattice mismatch of MoS2 grown on sapphire is quiteessential to see the stability of MoS2/sapphire heterostructures. Thus,the lattice constant for sapphire is set to 0.4789 nm to followthe experimental ratio of 1.506 for MoS2 and sapphire. This treatmentis justified, because there is no chemical bond betweenMoS2 `monolayer and sapphire slab, and the lattice constant of sap-phire slab is not affected by MoS2. Adsorption energy, Ead isdefined as Ead = (EMoS2/sapphire − (EMoS2 + Esapphire))/S, where EMoS2, Esapphire,and EMoS2/sapphire are the calculated total energies for MoS2 monolayer,sapphire, and MoS2/sapphire. S is the area of a superstructure.Device fabrication and transport characterizationTo evaluate the transport properties of monolayer MoS2 grown byMOCVD, it was transferred onto a 90nm SiO2/n+-Si substrate using astandard transfer technique (Supplementary Fig. 25). Polymethylmethacrylate (PMMA, MicroChem, 495k) was spin-coated on theMoS2/sapphire wafer and further supported by thermal release tape(TRT, Nitto, Revalpha). The TRT/PMMA/MoS2 stack was released fromthe sapphire wafer in a KOH solution (1mol/L) and rinsed with deio-nized water. Following this, the stack was transferred to the SiO2/Sisubstates, and the TRT/PMMAfilmwas removed in acetone. For devicefabrication, a maskless aligner μMLA (Heidelberg instrument) withdouble layer resists (PMGI SF5/AZ1500) and NMD-3 developer wasused to define channel and electrode patterns. Then, the shape ofMoS2 channels were defined by CF4 plasma etching. A 1-nm/30-nm Ni/Au electrode was deposited in an UHV chamber at a base pressure of~5×10-8Pa with a low deposition rate of 0.0014 Å/s for Ni and 0.15 Å/sfor Au. The sample temperature was maintained at 15 °C in the UHVchamber to avoid thermal damage. All electrical measurements in thisstudywere conducted using a vacuumproberwith a cryogenic system,employing a Keysight B1500 semiconductor parameter analyzer.Data availabilityRelevant data supporting the key findings of this study are availablewithin the article and the Supplementary Information file. All raw datagenerated during the current study are available from the corre-sponding authors upon request.Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 8www.nature.com/naturecommunicationsReferences1. Cao, W. et al. The future transistors. Nature 620, 501–515 (2023).2. Wu, P., Zhang, T., Zhu, J., Palacios, T. & Kong, J. 2D materials forlogic device scaling. Nat. Mater. 23, 23–25 (2024).3. Liu, C. et al. Understanding epitaxial growth of two-dimensionalmaterials and their homostructures. Nat. Nanotechnol. 19,907–918 (2024).4. Akselrod, M. S. & Bruni, F. J. Modern trends in crystal growth andnew applications of sapphire. J. Cryst. Growth 360, 134–145 (2012).5. Dong, J., Zhang, L., Dai, X. & Ding, F. The epitaxy of 2D materialsgrowth. Nat. Commun. 11, 5862 (2020).6. Dumcenco, D. et al. Large-area epitaxial monolayer MoS2. ACSNano 9, 4611–4620 (2015).7. Li, T. et al. Epitaxial growth of wafer-scale molybdenum disulfidesemiconductor single crystals on sapphire. Nat. Nanotechnol. 16,1201–1207 (2021).8. Fu, J. H. et al. Oriented lateral growth of two-dimensional materialson c-plane sapphire. Nat. Nanotechnol. 18, 1289–1294 (2023).9. Wang, J. et al. Dual-coupling-guided epitaxial growthofwafer-scalesingle-crystal WS2 monolayer on vicinal a-plane sapphire. Nat.Nanotechnol. 17, 33–38 (2022).10. Liu, L. et al. Uniform nucleation and epitaxy of bilayer molybdenumdisulfide on sapphire. Nature 605, 69–75 (2022).11. Chang, Y. et al. Atomically thin decoration layers for robust orien-tation control of 2D transition metal dichalcogenides. Adv. Funct.Mater. 34, 2311387 (2024).12. Zheng, P. et al. Universal epitaxy of non-centrosymmetric two-dimensional single-crystal metal dichalcogenides. Nat. Commun.14, 592 (2023).13. Li, L. et al. Epitaxy of wafer-scale single-crystal MoS2 monolayer viabuffer layer control. Nat. Commun. 15, 1825 (2024).14. Chubarov, M. et al. Wafer-scale epitaxial growth of unidirectionalWS2 monolayers on sapphire. ACS Nano 15, 2532–2541 (2021).15. Zhu, H. et al. Step engineering for nucleation and domain orienta-tion control inWSe2 epitaxy on c-plane sapphire.Nat. Nanotechnol.18, 1295–1302 (2023).16. Yang, P. et al. Highly reproducible epitaxial growth of wafer-scalesingle-crystal monolayer MoS2 on sapphire. Small Methods 7,2300165 (2023).17. Park, Y. et al. Critical role of surface termination of sapphire sub-strates in crystallographic epitaxial growth of MoS2 using inorganicmolecular precursors. ACS Nano 17, 1196–1205 (2023).18. Lauritsen, J. V. et al. Size-dependent structure of MoS2 nanocrys-tals. Nat. Nanotechnol. 2, 53–58 (2007).19. Hansen, L. P. et al. Atomic-scale edge structures on industrial-styleMoS2 nanocatalysts.Angew. Chem. - Int Ed. 50, 10153–10156 (2011).20. Aljarb, A. et al. Substrate lattice-guided seed formation controls theorientation of 2D transition-metal dichalcogenides. ACS Nano 11,9215–9222 (2017).21. Suenaga, K. et al. Surface-mediated aligned growth of monolayerMoS2 and in-plane heterostructures with graphene on sapphire.ACS Nano 12, 10032–10044 (2018).22. Li, C. et al. Revisiting the epitaxial growth mechanism of 2D TMDCsingle crystals. Adv. Mater. 36, 2404923 (2024).23. Ji, H. G. et al. Hydrogen-assisted epitaxial growth of monolayertungsten disulfide and seamless grain stitching. Chem. Mater. 30,403–411 (2018).24. Lin, Y. C. et al. Realizing large-scale, electronic-grade two-dimen-sional semiconductors. ACS Nano 12, 965–975 (2018).25. Xiang, Y. et al. MonolayerMoS2 on sapphire: an azimuthal reflectionhigh-energy electron diffraction perspective. 2D Mater. 8,025003 (2021).26. Zhang, L., Dong, J. & Ding, F. Strategies, status, and challenges inwafer scale single crystalline two-dimensional materials synthesis.Chem. Rev. 121, 6321–6372 (2021).27. Curiotto, S. & Chatain, D. Surface morphology and composition ofc-, a- andm-sapphire surfaces in O2 and H2 environments. Surf. Sci.603, 2688–2697 (2009).28. Yang,X., Li, S., Ikeda,N.,Ohtake, A. &Sakuma, Y. Scalablegrowthofatomically thin MoS2 layers in a conventional MOCVD system usingmolybdenum dichloride dioxide as the molybdenum source. ApplSurf. Sci. 636, 157756 (2023).29. Chen, C. et al. Effect of growth temperature on the microstructureand properties of epitaxial MoS2 monolayers grown by metalor-ganic chemical vapor deposition. J. Vac. Sci. Technol. A 42,022201 (2024).30. Yu, H. et al. Wafer-scale growth and transfer of highly-orientedmonolayer MoS2 continuous films. ACS Nano 11,12001–12007 (2017).31. Chubarov,M., Choudhury, T. H., Zhang, X. & Redwing, J. M. In-planex-ray diffraction for characterization of monolayer and few-layertransition metal dichalcogenide films. Nanotechnology 29,055706 (2018).32. Dosenovic, D. et al. Mapping domain junctions using 4D-STEM:toward controlled properties of epitaxially grown transition metaldichalcogenide monolayers. 2D Mater. 10, 045024 (2023).33. Van Der Zande, A. M. et al. Grains and grain boundaries in highlycrystalline monolayer molybdenum disulphide. Nat. Mater. 12,554–561 (2013).34. Deb, P. et al. Imaging polarity in two dimensional materials bybreaking Friedel’s law. Ultramicroscopy 215, 113019 (2020).35. Han, Y. et al. Strain mapping of two-dimensional heterostructureswith subpicometer precision. Nano Lett. 18, 3746–3751 (2018).36. Li, Y. et al. Probing symmetry properties of few-layerMoS2 and h-BNby optical second-harmonic generation. Nano Lett. 13,3329–3333 (2013).37. Mennel, L. et al. Optical imaging of strain in two-dimensionalcrystals. Nat. Commun. 9, 516 (2018).38. Ahn, H. et al. Integrated 1D epitaxial mirror twin boundaries forultrascaled 2D MoS2 field-effect transistors. Nat. Nanotechnol. 19,955–961 (2024).39. Rogge, P. C. et al. Real-time observation of epitaxial graphenedomain reorientation. Nat. Commun. 6, 6880 (2015).40. Thompson, C. V. Grain growth in thin films. Annu. Rev. Mater. Sci.20, 245–268 (1990).41. Tan, L. K. et al. Atomic layer deposition of a MoS2 film.Nanoscale 6,10584–10588 (2014).42. Jurca, T. et al. Low-temperature atomic layer deposition of MoS2films. Angew. Chem. 129, 5073–5077 (2017).43. Smets,Q.et al. Sourcesof variability in scaledMoS2 FETs.2020 IEEEInt. Electron Devices Meet. (IEDM) 2020, 3.1.1–3.1.4 (2020).44. Shi, Y. et al. Engineering wafer-scale epitaxial two-dimensionalmaterials through sapphire template screening for advanced high-performance nanoelectronics. ACS Nano 15, 9482–9494 (2021).45. Yu, Z. et al. Towards intrinsic charge transport in monolayermolybdenum disulfide by defect and interface engineering. Nat.Commun. 5, 1–7 (2014).46. Cui, X. et al.Multi-terminal transportmeasurements ofMoS2 using avan der Waals heterostructure device platform. Nat. Nanotechnol.10, 534–540 (2015).47. Song, S. et al. High current and carrier densities in 2D MoS2 /AlScNfield-effect transistors via ferroelectric gating and ohmic contacts.ACS Nano 19, 8985–8996 (2025).48. Shen, P. C. et al. Ultralow contact resistance between semimetaland monolayer semiconductors. Nature 593, 211–217 (2021).49. Kurabayashi, S. & Nagashio, K. Transport properties of the top andbottom surfaces in monolayer MoS2 grown by chemical vapordeposition. Nanoscale 9, 13264–13271 (2017).50. Yamasaki, T. et al. Multi-axis decomposition of density functionalprogram for strong scaling up to 82,944 nodes on the K computer:Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 9www.nature.com/naturecommunicationsCompactly folded 3D-FFT communicators in the 6D torus network.Comput Phys. Commun. 244, 264–276 (2019).51. Kohn, W. & Sham, L. J. Self-consistent equations includingexchange and correlation effects. Phys. Rev. 140,A1133–A1138 (1965).52. Troullier, N. & Martins, J. L. Efficient pseudopotentials for plane-wave calculations. Phys. Rev. B 43, 1993–2006 (1991).53. Vanderbilt, D. Soft self-consistent pseudopotentials in a general-ized eigenvalue formalism. Phys. Rev. B 41, 7892–7895 (1990).54. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradientapproximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).55. Grimme, S. Semiempirical GGA-type density functional con-structed with a long-range dispersion correction. J. Comput Chem.27, 1787–1799 (2006).56. Chang, H.-Y., Zhu, W. & Akinwande, D. On the mobility and contactresistance evaluation for transistors based on MoS2 or two-dimensional semiconducting atomic crystals. Appl Phys. Lett. 104,113504 (2014).57. Zhu, J. et al. Low-thermal-budget synthesis of monolayer molyb-denum disulfide for silicon back-end-of-line integration on a200mm platform. Nat. Nanotechnol. 18, 456–463 (2023).AcknowledgementsTheoretical calculations were performed by using the Numerical Mate-rials Simulator of NIMS and the Earth Simulator (ES) of JAMSTEC. Thisresearch was supported by the JSPS KAKENHI (Grant Numbers:JP17H03241 to Y.S., JP21H05237 to K.N., JP21H05232 to K.N.,JP22H04957 to K.N., JP22K04212 to T.N., JP23K13622 to K.K.,JP23K03272 to M.I., and JP23K04592 to A.O.), Iketani Science andTechnology Foundation to M.I., the NICT (Grant Number: 05901 to K.N.),the JST-Mirai Program (Grant Number: JPMJMI22708192 to K.N.) and JST-CREST (Grant Number: JPMJCR24A3).Author contributionsY.S. and K.N. designed the research. Y.S., Y.O., and T.M. contributed tothe MOCVD growth, as well as AFM and Raman spectroscopy mea-surements. Y.M. andK.T. performedAFMmeasurements. S.L. performedPL measurements. T.H. conducted XRD measurements, while A.O. per-formed LEED analysis. J.N. carried out DFT simulations. E.K., T.Y., X.Y.,and N.I. performed HAADF-STEM measurements. A.S. and M.I. con-ducted SHGmeasurements. K.A. fabricated the devices and conductedelectrical characterization with assistance from K.K. and T.N. All authorscontributed to thediscussion, andK.N. andY.S. co-wrote themanuscriptwith input from all authors.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-026-68320-8.Correspondence and requests for materials should be addressed toYoshiki Sakuma or Kosuke Nagashio.Peer review information Nature Communications thanks Yi Wan, TianyiZhang, and the other, anonymous, reviewer(s) for their contribution tothe peer review of this work. A peer review file is available.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jur-isdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, aslong as you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons licence, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article's Creative Commons licence, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article's Creative Commons licence and your intendeduse is not permitted by statutory regulation or exceeds the permitteduse, you will need to obtain permission directly from the copyrightholder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.© The Author(s) 2026Article https://doi.org/10.1038/s41467-026-68320-8Nature Communications |          (2026) 17:602 10https://doi.org/10.1038/s41467-026-68320-8http://www.nature.com/reprintshttp://creativecommons.org/licenses/by/4.0/http://creativecommons.org/licenses/by/4.0/www.nature.com/naturecommunications Self-aligned and self-limiting van der Waals epitaxy of monolayer MoS2 for scalable 2D electronics Results Epitaxial registry and self-aligned coalescence Deterministic epitaxial alignment of MoS2/sapphire Wafer scale uniformity via self-limiting growth Transport properties of monolayer MoS2 Discussion Methods MOCVD growth Characterizations Device fabrication and transport characterization Data availability References Acknowledgements Author contributions Competing interests Additional information