# Fileset

[s41467-025-59961-2.pdf](https://mdr.nims.go.jp/filesets/00cf7a30-3c76-4bd0-bd42-13e0787a3d28/download)

## Creator

Deyi Fu, Jiawei Liu, Fuchen Hou, Xiao Chang, Tingyu Qu, Johan Félisaz, Gunasheel Kauwtilyaa Krishnaswamy, Sergey Grebenchuk, Yuang Jie, [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), Vitor M. Pereira, Kostya S. Novoselov, Maciej Koperski, Nikolai L. Yakovlev, Anjan Soumyanarayanan, Ahmet Avsar, Oleg V. Yazyev, Junhao Lin, Barbaros Özyilmaz

## Rights

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

## Other metadata

[Electric field-tunable ferromagnetism in a van der Waals semiconductor up to room temperature](https://mdr.nims.go.jp/datasets/5b142264-977d-4d6f-ba34-c1c83de86403)

## Fulltext

Electric field-tunable ferromagnetism in a van der Waals semiconductor up to room temperatureArticle https://doi.org/10.1038/s41467-025-59961-2Electric field-tunable ferromagnetism in avan der Waals semiconductor up to roomtemperatureDeyi Fu 1,2,13,14, Jiawei Liu2,3,14, Fuchen Hou4,5,14, Xiao Chang6,14,Tingyu Qu 1,7,14 , Johan Félisaz8, Gunasheel Kauwtilyaa Krishnaswamy 1,Sergey Grebenchuk3, Yuang Jie6, Kenji Watanabe 9, Takashi Taniguchi 10,Vitor M. Pereira 11, Kostya S. Novoselov 2,3, Maciej Koperski 3,6,Nikolai L. Yakovlev 1, Anjan Soumyanarayanan 1,12, Ahmet Avsar 1,2,6,Oleg V. Yazyev 8, Junhao Lin 4,5 & Barbaros Özyilmaz 1,2,3,6,7Ferromagnetic semiconductors, coupling charge transport andmagnetism viaelectrical means, show great promise for spin-based logic devices. Despitedecades of efforts to achieve such co-functionality,maintaining ferromagneticorder at room temperature remains elusive. Here, we address this long-standing challenge by implanting dilute Co atoms into few-layer black phos-phorus through atomically-thin boron nitride diffusion barrier. Our Co-dopedblack phosphorus-based devices exhibit ferromagnetism up to room tem-perature while preserving its high mobility (~1000cm2V�1s�1) and semi-conducting characteristics. By incorporating ferromagnetic Co-doped blackphosphorus into magnetic tunnel junction devices, we demonstrate a largetunnelling magnetoresistance that extends up to room temperature. Thisstudy presents a new approach to engineering ferromagnetic ordering inotherwise nonmagnetic materials, thereby expanding the repertoire andapplications of magnetic semiconductors envisioned thus far.The quest for ferromagnetic-and-semiconducting systems that elec-trically function at room temperature poses significant challenges1.The recent 2Dmagnets such as CrI32–4 andCrSBr5 can combine electric-field tunable charge transport and spin configuration, whichmanifestsas a magnetic transistor, yet the Curie temperature (TC) is still below160K. Conventional dilutemagnetic semiconductors (DMSs) based onsubstitutional magnetic doping of nonmagnetic semiconductors6–8(e.g. II-VI CdTe6, III-V GaAs9,10, InAs11,12 or GaSb13,14, oxides like (TiO215,Received: 14 June 2024Accepted: 8 May 2025Check for updates1Department of Physics, National University of Singapore, Singapore, Singapore. 2Centre for Advanced 2D Materials, National University of Singapore,Singapore, Singapore. 3Institute for Functional Intelligent Materials (I-FIM), National University of Singapore, Singapore, Singapore. 4Department of Physics,State key laboratory of quantum functional materials, and Guangdong Basic Research Center of Excellence for Quantum Science, Southern University ofScience and Technology (SUSTech), Shenzhen, China. 5Quantum Science Center of Guangdong-Hong Kong-Macao Greater Bay Area (Guangdong),Shenzhen, PR China. 6Department of Materials Science and Engineering, National University of Singapore, Singapore, Singapore. 7NUS Graduate School,Integrative Sciences andEngineeringProgramme,National University of Singapore, Singapore, Singapore. 8Institute of Physics, Ecole Polytechnique Fédéralede Lausanne (EPFL), Lausanne, Switzerland. 9Research Center for Electronic and Optical Materials, National Institute for Materials Science, 1-1 Namiki,Tsukuba, Japan. 10ResearchCenter forMaterials Nanoarchitectonics, National Institute forMaterials Science, 1-1 Namiki, Tsukuba, Japan. 11Centro deFisica dasUniversidades doMinho e do Porto, LaPMET, Departamento de Fisica e Astronomia, Faculdade de Ciências, Universidade do Porto, Porto, Portugal. 12InstituteofMaterials Research&Engineering (IMRE), Agency for Science, Technology andResearch (A*STAR), Singapore, Singapore. 13Present address: Department ofPhysics, Xiamen University, Xiamen, China. 14These authors contributed equally: Deyi Fu, Jiawei Liu, Fuchen Hou, Xiao Chang, Tingyu Qu.e-mail: ty.qu@nus.edu.sg; linjh@sustech.edu.cn; barbaros@nus.edu.sgNature Communications |        (2025) 16:10197 11234567890():,;1234567890():,;http://orcid.org/0000-0003-1365-8963http://orcid.org/0000-0003-1365-8963http://orcid.org/0000-0003-1365-8963http://orcid.org/0000-0003-1365-8963http://orcid.org/0000-0003-1365-8963http://orcid.org/0000-0002-4409-3072http://orcid.org/0000-0002-4409-3072http://orcid.org/0000-0002-4409-3072http://orcid.org/0000-0002-4409-3072http://orcid.org/0000-0002-4409-3072http://orcid.org/0000-0001-7371-8484http://orcid.org/0000-0001-7371-8484http://orcid.org/0000-0001-7371-8484http://orcid.org/0000-0001-7371-8484http://orcid.org/0000-0001-7371-8484http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-3462-524Xhttp://orcid.org/0000-0002-3462-524Xhttp://orcid.org/0000-0002-3462-524Xhttp://orcid.org/0000-0002-3462-524Xhttp://orcid.org/0000-0002-3462-524Xhttp://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0003-4972-5371http://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-8301-914Xhttp://orcid.org/0000-0002-5611-9604http://orcid.org/0000-0002-5611-9604http://orcid.org/0000-0002-5611-9604http://orcid.org/0000-0002-5611-9604http://orcid.org/0000-0002-5611-9604http://orcid.org/0000-0003-2680-6005http://orcid.org/0000-0003-2680-6005http://orcid.org/0000-0003-2680-6005http://orcid.org/0000-0003-2680-6005http://orcid.org/0000-0003-2680-6005http://orcid.org/0000-0003-0173-7242http://orcid.org/0000-0003-0173-7242http://orcid.org/0000-0003-0173-7242http://orcid.org/0000-0003-0173-7242http://orcid.org/0000-0003-0173-7242http://orcid.org/0000-0001-7281-3199http://orcid.org/0000-0001-7281-3199http://orcid.org/0000-0001-7281-3199http://orcid.org/0000-0001-7281-3199http://orcid.org/0000-0001-7281-3199http://orcid.org/0000-0002-2195-2823http://orcid.org/0000-0002-2195-2823http://orcid.org/0000-0002-2195-2823http://orcid.org/0000-0002-2195-2823http://orcid.org/0000-0002-2195-2823http://orcid.org/0000-0001-7665-4578http://orcid.org/0000-0001-7665-4578http://orcid.org/0000-0001-7665-4578http://orcid.org/0000-0001-7665-4578http://orcid.org/0000-0001-7665-4578http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-59961-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-59961-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-59961-2&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-59961-2&domain=pdfmailto:ty.qu@nus.edu.sgmailto:linjh@sustech.edu.cnmailto:barbaros@nus.edu.sgwww.nature.com/naturecommunicationsZnO16,17), SiGe18 and 2D transition-metal dichalcogenides19–22) providealternative solutions for high TC up to room temperature. However,magnetic doping typically degrades the mobility of DMSs to a merefew cm2V�1s�123,24 or limits gate-tunability18,22. A room-temperatureferromagnet that simultaneously exhibits critical semiconductingcharacters such as high mobility, a high ON/OFF ratio, and a low sub-threshold swing remains elusive.Doping of semiconducting van der Waals (vdW) materials can beminimally invasive to the electronic structure and thus well suited toincorporate magnetism in layered semiconductors25. Here blackphosphorous (BP) could be employed as amodel 2D semiconductor todemonstrate such doping26 owing to its unique semiconducting fea-tures such as a moderate band gap27, record electron mobilities28–30,room-temperature micron-scale spin relaxation lengths31, high spinanisotropy32 and suitability for cobalt (Co) doping to yield ferromag-netic (FM) order33. Previous intercalation studies on BP with non-magnetic adatoms show that the electronic mobilities of intercalatedBP can even increase compared to few-layer pristine BP26. Thesecharacteristics render BP an attractive platform for intercalation withmagnetic adatoms, presenting an opportunity to realize gate-tunableroom-temperature magnetism while preserving its intrinsic semi-conductor properties.In this paper, we study the magneto-transport properties of theresulting Co-doped BP devices, fabricated in both the lateral geometryand the vertical geometry. In the lateral configuration, we observedgate-tunable anomalous Hall effect and hysteretic in-plane magne-toresistance (MR); in the vertical junction, we observed bias-and-gatetunable tunnellingmagnetoresistance (TMR) up to room temperature.Our results establish dilute Co-intercalated BP as a gate-tunable mag-netic semiconductor operating up to room temperature, showcasingits potential for versatile applications in spintronics and magneto-electronics.ResultsOur density-functional theory (DFT) suggests that BP intercalated witha few percent of Co (See schematics in Fig. 1a) can be FM33–35 (Sup-plementary Notes 1-2). Here, the hybridization with native p-derivedbands of BP is minor, so the intercalated Co atoms induce little chargetransfer to BP, preserving its ambipolar gate-tunability. We define“majority” and “minority” spin states as the population imbalancebetween spin-up and spin-down states in Co-BP relative to Co elec-trode. Figure 1b presents the corresponding DOS schematic for Fig. 1c.In addition, in the conduction band, the population of spin-up is equalto that of spin-down (quenched magnetic moments). But near theconduction band edge, there exists a localized magnetic state. Thisstate can be probed with tunnelling experiments. Inside the gap, thereis no accessible state. On the other hand, in the valence band, there isan imbalance of spin-up and spin-down states. This gives rise toalternating behaviour of majority and minority carries as a function ofEF. We note that the electric-field tunability in BP also depends on Codoing concentration (See Supplementary Fig. S1).First, we confirm the feasibility of implantingCo atoms in BPusingplanar-view atomic-resolution scanning transmission electron micro-scopy (STEM) (See Fig. 2a, b, Supplementary Fig. S2 andMethods). Weachieve average Co doping concentration and Co separation of ~3%and ~2 nm, respectively, and do not detect signatures of Co clusteringor other defects (confirmed with four separate samples). To confirmthe semiconducting nature of our Co-BP (See the optical micrographand measurement set-up in Fig. 2c), we measure current under con-stant bias voltage (V SD = −1 V) and back-gate voltage Vg varying from−75 V to +100V. A clear OFF state is visible for Vg range between 0Vand 36 V. Beyond this range, our device shows bipolar behaviour withhole mobility up to 1000 cm2V�1s�1 and an ON/OFF ratio exceeding5000 (Top panel in Fig. 2d). This suggests that our Co-BP remainscharge neutral (or slightly p-type) and thus can be tuned into eithern-type or p-type by the back gate.Next, we investigate the magnetic nature of our Co-BP by per-forming AHE measurements (Device A, See Methods). We sweep anout-of-plane magnetic field (B?) and show the representative Rxy �B? curves in hole and electron regimes (Fig. 2e). In the hole regime,the field dependence can be described by a Langevin-type functionRAHExy =RAHE0 tanh BBC� �36, where RAHE0 is proportional to the saturationmagnetization and BC is the saturation field (See the fitting inFig. 2e). This suggests magnetism in Co-BP (See the extracted BC inbottom panel of Fig. 2d and details in Supplementary Fig. S3). Incontrast, we do not observe any AHE with a positive back gate,possibly because the lateral transport cannot probe the expectedmagnetic order near the conduction band minimum due to lowconductance. This regime will be examined in the vertical tunnellingDevices C andDbelow, which is less sensitive to the low conductivityat the conduction band edge.To corroborate the FM order in Co-BP, we compare its in-planeMR with a Co-dust film alone of the same evaporated thickness(~1.5 nm). Both geometries are shown in Fig. 2f (DeviceB, SeeMethods)and are capped by a Ti layer, since Co dust alone is non-conducting.We sweep the in-plane field (Bjj) along the direction of the appliedFig. 1 | Structural and electronic properties of Co-doped BP. a Atomic structureof the Co-intercalated BP illustrating magnetic doping in this system. Spin densityisosurfaces of the Co atoms support an FM order in Co-BP. b Spin-polarized bandstructure of the same supercell after intercalation of a Co atom (3%) between thephosphorene layers; blue (red) represents the spin-up and spin-down bands. Thegrey curves represent the bands of pristine bulk BP. c Calculated density of states(DOS) of Co-BP near the Fermi level (EF(0)) according to the bandstructure of thecharge neutral configuration shown in (b). Blue and red bands represent spin-upand spin-down, respectively. Dashed line is the EF. Light colours represent the totalDOS, while dark colours show only the d orbital projected DOS.Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 2www.nature.com/naturecommunicationscurrent along the zig-zag direction (See Supplementary Fig. S4). Theabsence of MR in “Co dust + Ti” electrode (probed with contactslabelled by “3” and “4”) strongly suggests the lack of FM long-rangeorder in Co dust itself (curve labelled “Co dust” in Fig. 2g). In contrast,the Co-doped BP region (probed with contacts labelled by “1” and “2”)displays very large and hysteretic MR, as reflected in the longitudinalfield sweeps at various temperatures up to 300K shown in Fig. 2f. Thisis characteristic of a FM channel undergoingmagnetization reversal asa function of a collinear magnetic field sweep37. Both BC and ΔR arefinite, indicating that Co-BP remains FM beyond room temperature.To further characterize the magnetism in Co-BP, we performmagneto-optic Kerr effect (MOKE) measurements at room tempera-ture. We note that the Co dusts are also evaporated on adjacent Aupads to compare the differences inMOKE signals. In Co-BP, the out-of-plane polarMOKE signal is large, negative, and saturates at 700mT. Asexpected, the saturation field at room temperature is smaller than thatmeasuredby theAHE at 1.6 K. Since the sample is slightly tilted, there isa small in-plane magnetic field component, allowing us to also recordsimultaneously the in-planeMR (SeeMethods).Weobserve an in-planeMR signal with a switching field that matches that of the MOKE signaloriginating from the initial in-plane reversal of themagnetization (blueregion in Fig. 3a). It is also comparable to that observed in the in-planeMRmeasurements shown in Fig. 2g. In contrast, MOKEmeasurementsof Co dust on Au electrodes exhibit none of these features. Instead,only a small kink near zero field indicative of disconnected Co clustersis observed (Fig. 3b).We also performed magnetic force microscopy (MFM) measure-ments (Methods & Supplementary Information). Here, we evaporateCo dusts onto the entire substrate such that we can compare theMFMsignals from Co-BP and Co-SiO2 (See the labelled region in themicrograph from in Fig. 3c). We record the frequency shift (Δf) as afunction of position (Fig. 3c) at 1.7 K. A large Δf is due to the attractiveforcebetween themagnetic cantilever and amagnetic substrate. Inourexperiments, we only observed such amagnetic signature for Co-BP atnegative gate voltages and absent on the SiO2 substrate. Here, wediscuss representative data at Vg =�40V. At zero magnetic field, nomagnetic signal is observed. As the magnetic field increases, Δfincreases in the Co-BP defined by the darker regionwith a well-definededge (top part of the scan). On the other hand, Δf on the SiO2 regionremains unchanged. At magnetic fields above 1.5 T, similar to the AHEmeasurements in Fig. 2c, themagnetic contrast saturates. Additionally,we show representative line scans of Δf in Fig. 3d. The difference in Δfbetween the scans at 0 T and 1.5 T is clearly visible in the Co-BP region(~0.1 Hz) but marginal on SiO2.Fig. 2 | Anomalous Hall effect and anisotropic magnetoresistance of Co-BP.a Planar-view atomic-resolution HAADF image of a Co-doped multilayer BP flake.The image intensity is directly related to the atomic weight of the imaged species.The brightest spots in the planar view HAADF image (marked with red circles) areisolated Co atoms on the BP surface while the dimmer spots (yellow circles) showisolatedCo atoms intercalated in deeper interlayer spaces.b Intensity histogramofthe mapped atomic columns in (a), indicating the adsorbed and intercalated indi-vidual Co atoms with P atomic columns. c Optical micrograph and measurementschematics of the lateral Hall bar device (Device A) of Co-BP on 285-nm SiO2/Sisubstrate. The ultra-thin BN acts as an encapsulation/diffusion barrier that protectsBP and allows dilute Co intercalation. Pairs of Ti/Au (2/85 nm) contacts labelled by#1 and #2 and by #1 and 3 are used for probing the Hall resistance (Rxy) and thechannel resistance (Rxx), respectively under a constant current from source (S) todrain (D). Scale bar: 4μm. d Top panel: Transconductance of the lateral Hall bardevice at V SD = −1 V. Bottom panel: The magnetization field (BC) as a function ofback-gate voltages. The error bars are extracted by fittingwith a confidence intervalof 99.7%. e Two representative anomalous Hall effect curves at 1.6 K in the holeregime (red dots, fitted by the black line) and electron regime (blue dots). fOpticalmicrograph and schematics of Device B for in-plane MR measurement. The MR isprobed by the voltage between the contacts labelled by #1 and #2. The injectionelectrode ismadeofCo/Ti (1.5 nm/35 nm),whereCowasdeliberatelydeposited asadiscontinuous dust layer of 1.5-nm nominal thickness. Scale bar: 3μm.g Temperature dependence of MR at Vg = − 50V and Ibias = −40μA. For compar-ison, the bottommost curve (black) refers to the metallic electrode (Co/Ti) itself atT = 2.5 K, confirming it hasnomagnetoresistance. The in-planefield (Bjj) is collinearwith the current. The zoomed-in details of the MR at 300K are shown in thetop panel.Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 3www.nature.com/naturecommunicationsFrom an application perspective, an important question is whetherCo-BP can be easily integrated into ready-to-use spintronic devices withgate tunability. Here, our proof-of-concept device is a magnetic tunneljunction (MTJ), which uses Co-doped BP layer and a 35-nm thick Costripe act as the ferromagnetic layers with BN as the tunnelling barrier(Devices C and D in Fig. 4a) (See Methods and Supplementary Fig. S5).Intriguingly, we observe hysteretic switching between two distinctresistance states in the TMRwhen sweepingBjj along the easy axis of theCo contact. A representative TMR measurement at T = 2.5 K shown inFig. 4b displays well defined, large resistance steps (ΔR = 800 Ω, upperpanel). In addition, the minor loop (lower panel) displays the square-shaped hysteretic profile expected for an FM–insulator–FM MTJs, thusconfirming the existence of an effective FMcontact on theBP side of thejunction with a well-defined coercive field.We then demonstrate the gate-tunability of the Co-BP DMS(Fig. 4c). The measurements reveal a sequence of four “operationalstates”, namely, an absenceof TMR in n-typeBP (Vg = + 60V), a positiveTMR near the edge of conduction band (Vg = + 30V), an absence ofTMR inside the gap (Vg = 0V ), a negative TMR p-type BP (Vg = � 50V).Specifically, we labelled these four states as I–IV with reference to theFig. 3 | MOKE andMFM characterizations of the Co-BP device. a Kerr signal andMR signal. Blue region indicates the in-plane switching and the orange regionindicates the out-of-plane saturation during the field scan. b A comparison of Kerrsignals on Co-BP and Co on Au pads. Inset: Optical micrograph showing the Co-BPregion. cMFM characterization of the sample at 1.7 K and under Vg = − 40 V. At, asB? increases, a darker region (the edge is denoted by the dashed line) begins toappear in the Co-BP region. This corresponds to a negative frequency shift (Δf) thatrepresents an attractive force between the cantilever and sample surface, indicat-ing Co-BP becomes magnetic. d MFM profiles of the line scans for frequency shift(Δf) from c. The curves are shifted vertically for better presentation.Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 4www.nature.com/naturecommunicationscorresponding energy ranges highlighted in the DOS schematic of Co-BP shown in Fig. 4a. Here the position of EF in Co-BP determines whe-ther the local moments μCo are finite or zero, allowing gate control oflocal moment formation. Specifically, in device state I, Co-BP is a n-typesemiconductor with EF in the conduction band and quenchedmoments, explaining the featurelessMR curve at Vg ≥ + 50 V. In devicestate II, EF lies near the band edge and μCo is finite, so that Co-BP is FMand the heterostructure operates as a tunnelling spin valve that probesthe localized magnetic moments from Co. In comparison, this state isnot visible in lateral transport (i.e., Device A) mostly likely due to themarginal conductance at the conduction band edge38. Device state III isobtained upon further reducing EF into the gap, in which case, theitinerant carrier density may now be insufficient to couple FM orderamong them. Further reducing Vg brings EF to the valence band with asubstantial increase in carrier density. In device state IV, unlike in theconduction bandminimum, Co-BP is reinstated as a FM contact and thespin-valve signal re-emerges but is inverted. The opposite TMR sign instates II and IV is in good agreement with DFT which also sees aninversion ofmajority andminority spins in BP. The tunnelling out of thebulk Co electrode injects minority spin polarization into Co-BP; con-sequently, a positive (negative) TMR step will only arise when minority(majority) states are dominant on the Co-BP side. The electric fieldeffect of BP not only allows for turning ON/OFF magnetism but also aswitch between positive and negative TMR signals.Importantly, we examinewhether our Co-BPDMS in a TMRdevicecan function up to room temperature. Remarkably, the TMR signal isfinite and retains sharp steps even at 300K39 (Fig. 4d, see also). Asexpected, theTMRvaluedecreaseswith increasing temperaturedue tospin-flip scattering and activation of inelastic tunnelling channels40,especially at high bias41,42 (Inset in Fig. 4d). The observation of TMR at300K (Fig. 4d) provides a lower bound for TC. Interestingly, this is inline with the indirect exchange coupling estimate (SupplementaryNote 1). The TMR value also depends on the current bias and is max-imum for �60 µA (See Supplementary Figs. S6 and 7).Finally, we investigate the doping profile of our Co-doped BP of theCo/BN/BP heterostructures using scanning transmission electronmicroscopy (STEM). As anticipated from the penetration of depositedmetals into vdW materials43 and also the BN dissolution-precipitationprocess44, we observe that the topmost BN layers in contact with the Coelectrode is amorphized and mixed with Co. However, a crystalline BNmonolayer remains adjacent to BP, critical for the MTJ performancediscussed above. This is also supported by non-linear I � V curves of thetunnelling junction (Supplementary Fig. S8). As mentioned earlier EELSmapping shows that Co is also incorporated into BP under the BN layer.Fromtheclose-upviewsofmultiple regions in Fig. 5a (Seepanels i-iii),wedo not detect any Co clusters inside BP layers. Figure 5b shows that theSTEM-extracted vdW gap expands by about 10% at the BN/BP interfaceand gradually recovers the bulk value outside the Co penetration depth.Fig. 4 | Tunnelling characteristics of vertical Co/BN/BP devices. a Left panel:Schematicdrawingof the verticalCo/BN/BPMTJdevices studied in thiswork,whichare grounded by a bottom-contacted graphene electrode. Right Panel: Schematicspin-resolved density of states of bulk Co (left) and of Co-BP (right) qualitativelyreflects theDFT +U band structures. For Co-BP, the light-shaded red and blue areasrepresent the phosphorus p-derived bands, only slightly changed relative to pris-tine BP; the darker, peaked traces represent the narrow bands associated with theCo d orbitals. The energy ranges labelled I–IV refer to the corresponding deviceoperation states discussed in the text. b Major (upper panel) and minor (lowerpanel) loop scans of the tunnelling magnetoresistance (TMR) of TMJ (Device C)under Vg = + 20 V at 2.5 K, with magnetic field parallel to the top Co electrode.c TMR of Device D at different gate voltages (Vg), measured at 2.5 K withIbias = − 60μA (DC) + 5 µA (AC). From top to bottom: Co-BPbecomes increasingly p-type with the positive-to-negative Vg variation; background resistances are 1.5, 3.5,7.6 and 2.7 kΩ, respectively (at Bjj = −60mT). d TMR under Ibias= − 60A (DC) + 5 µA(AC) inDevice C at 300K. Inset: The temperature dependenceof TMR. The data areextracted from Supplementary Fig. S6.Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 5www.nature.com/naturecommunicationsTo further characterize the Co penetration, we did electronenergy loss spectroscopy (EELS) along Co/BN/BP cross-sections, andobtained the elemental profiles shown in Fig. 5c. The depth profile ofCo shows a penetration depth ~ 5 nm, with an average doping ofaround 3%. This can be controlled by the initial BN thickness and post-annealing process under high vacuum condition (See SupplementaryTable S1 and Supplementary Fig. S9). Although the presenceof cluster-free Co-BP is thus consistent with our transport data, it may not be theonly possible explanation. The Co diffusion process may inevitablycreate certain atomic defects in the BN layers and the topmost BPlayer, which could also contribute to the TMR switching and requiresfuture experimental and theoretical investigation.The AHE in p-type BP, the large hysteretic in-plane MR seen inplanar transport and spin-valve switching behaviour observed in theMTJs are well-established hallmarks of robust FM order in Co-BP. ItsvdW nature and electronic two-dimensionality are decisive in allowingnon-detrimental magnetic doping and potentially large FM coupling.The coupled FM and semiconducting functionalities with high chargemobility up to room temperature is distinct from the types of electricaltunability commonly encountered in magnetic systems45 (See thecomparison with other magnetic semiconductors in SupplementaryTable S2). Given its electronic versatility30, its long spin-diffusionlength31, and recent advances in large-scale growth and integration ofBP46, our work opens the prospect of employing Co-BP as an activesemiconducting FM element in potential spintronics applicationswithout suffering conductivity mismatch problem. This methodologycould also be applied beyond BP (Supplementary Fig. S10), therebyexpanding the library of low-dimensional magnetic materials. Devel-oping encapsulation techniques to ensure the practical use of our Co-BP devices will be a subject for further investigation in future.MethodsDevice fabricationDevice A: We mechanically exfoliated black phosphorus (BP) crystalsonto a silicon substrate (covered by a 285 nm SiO2) and identify anideal flake (typically 10 × 5 µm2, thickness ~6~7 nm) as the channel. Anultra-thin BN flake (2~4 layers) is then transferred following the semi-dry transfer method47 to fully encapsulate the BP flake in inert condi-tions (N2 gas-filled glovebox with oxygen and moisture level below 1ppm). This BN is used as a dispersive atom diffusion barrier, whichreduces clustering while also preserving the structural integrity of BPduring the subsequent processing (Supplementary Note 3). The BN/BPstack is annealed at 200 °C for 6 h under high vacuum (10�6 Torr) toremove any bubbles formed during the transfer processes. This resultsin cleaner interfaces between 2D layers and hence improves thebonding of BP with BN layers. A standard electron beam lithographyFig. 5 | STEM characterization of the Co/BN/BP heterostructure. a Low magni-fication bright-field image of the Co/BN/BP heterostructure’s cross-sectionalinterface (scale bar 50nm). Labels i–iii refer to magnified atomic-resolution bright-field images of the denoted regions, showing that the BP vdW structure andstacking is intact over 100’s of nm. b Evolution of the vdW interlayer gap (red) andmonolayer thickness (blue), extracted from STEM cross-sections (inset), as afunction of distance to the BN/BP interface (in number of BP monolayers). Scalebar: 1.5 nm. The lower inset illustrates our definition of the two plotted distances.c Line cuts of the elemental abundance perpendicular to the Co/BN/BP interface,extracted from electron energy loss spectra (EELS). The four coloured panels areenergy dispersive spectrum (EDS) maps for the elements Co, P, N and B.Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 6www.nature.com/naturecommunications(EBL) technique is employed to pattern Ti/Au contacts (2 nm/85 nm). Asecond EBL is employed to pattern the Co-doping region and followedby the deposition of Co dust layers (~1.5 nm) under ultra-high vacuumconditions (5 × 10�8 Torr), which ensures Co does not form a con-tinuous layer. The deposition rate for the Co dust is 0.3 A/s.Throughout deposition, the high melting point of evaporated Cointroduces defects in the BN layer, both kinetically and chemically, asobserved in metal-semiconductor interfaces48. The device is finallyannealed at 200 °C for 6 h under high vacuum (10�6 Torr) for a moreuniform doping and improvement of the contact conductance beforemeasurements.Device B: The sample used for the in-plane MR studies followedthe same fabrication procedure as Device A except for depositing Co/Ti (1.5 nm/35 nm). The encapsulation of Co dust by light element Ti isused as a contact for current injection.Devices C and D: We mechanically exfoliated graphene onto asilicon substrate (covered by a 285 nm SiO2) as the bottom contact. ABP slab (thickness ~6~10 nm) was firstly exfoliated onto a PDMSstamp49, then aligned with the bottom graphene and finally brought incontact using amicroscope. By gently peeling off the PDMS stamp, theBP slab stays on the bottom contact. An ultra-thin BN flake (2~4 layers)is then transferred following the semi-dry transfer method47 to fullyencapsulate the BP/graphene stack in inert conditions (N2 gas-filledgloveboxwithoxygen andmoisture level below 1ppm). This BN isusedas a dispersive atom diffusion barrier, which reduces clustering whilealso preserving the structural integrity of BP during the subsequentprocessing (Supplementary Note 3). The final stack (BN/BP/Graphene)is annealed at 200 °C for 6 h under high vacuum (10�6 Torr) toremove any bubbles formed during the transfer processes. This resultsin cleaner interfaces between 2D layers and hence improves thebonding of BP with BN layers. A standard electron beam lithographytechnique is employed to pattern electrodes and followed bythe deposition of Co/Ti (35 nm/7.5 nm) layers under ultra-highvacuum conditions (5× 10�8 Torr). The deposition rate for the Coand Ti layers is 0.3A/s and 1 A/s, respectively. The vertical tunnellingdevices have a typical area of 1 × 1μm2 in each tunnelling junction(Supplementary Fig. S5).STEM characterizationThe cross-sectional STEM specimens were prepared using a Cryo-focused Ion Beam (FIB) in ultra-high vacuum (<10�6 mbar) in a liquid-nitrogen temperature environment. STEM imaging, EDS and EELSanalysis of vertical heterojunctions of Co/BN/X (X = BP, graphite,MoS2, NbSe2) were performed on a FEI Titan Themis with a X-FEGelectron gun and a DCOR aberration corrector operating at 300 kV.The inner and outer collection angles for the STEM images (β1 and β2)were 48 and 200mrad, respectively, with a convergence semi-angle of25mrad. The beam current was about 100 pA for imaging and spec-trum collection. All STEM experiments were performed at room tem-perature. In the elemental analysis, the Co concentration (nCo inFig. 5b) is defined as the ratio of detected Co atoms to the total atomsof all the detected elements. This ratio and its error bars were calcu-lated using a quantitative analysis model of EDS and EELS spectrumwith commercial codes that are standard and routinely used in thistype of elemental assessment.Magnetotransport measurementsThe magnetotransport measurements were performed in a home-made low-temperature electromagnetic system where the sample wassealed in vacuum and cooled down to a base temperature of 2.5 K. Astandard low-frequency (13Hz) lock-in technique was applied for sig-nal acquisition with AC current amplitudes in the range 2~5 µA.Meanwhile, a DC source was driven through to the devices to probetheir TMR or AMR. The differential resistance was measured by asecond lock-in amplifier. The magnetoresistance was recorded as afunction of an external magnetic field parallel to the easy axis of Coelectrodes.MOKE characterizationsMOKEmeasurements were performedwith a red laser of 5-mWpower,focused to a beam size of around 100-µm in length and 40-µm inheight, covering the entire flake and portion of the nonmagneticcontacts. The incidence angle was approximately 16-degree, providinglarge polar MOKE signal (along out-of-plane field direction) and smalllongitudinal MOKE (parallel to beam direction). The sample is slightlytilted by about 2-degree with respect to the magnetic field, allowing asmall in-plane reversal for the AMR signals at low magnetic fields.MFM characterizationsMFM was conducted using an attocube attoDRY 2100 closed-cyclecryogenic system with attoAFM I probe, which operates at a base tem-perature of 1.7 K. The system is equipped with a superconducting mag-net capable of applying out-of-plane field up to 9T. Silicon probes withmagnetic CoCr coating (MFMR by NanoWorld with a spring constantranging from 2.5 to 5N/m) were employed. Prior usage the probes weremagnetized in out-of-plane direction using a neodymium magnet. Forthe measurements with gate voltage applied, MFM was performed withphase-locked loop and amplitude control to keep the sameQ-factor. Themagnetic contrast was observed in the frequency shift Δf of cantilever.Lift was kept in the range of 100–200nm. For the room temperaturetests the sample space was pumped and filled with He gas below 1mbarto keep high Q-factor of the cantilever for better sensitivity.First-principles calculationsFirst-principles calculations were performed on a 5 ×4×2 supercell ofblack phosphorus with a single cobalt atom between the two layers.The VASP software package was used to relax the structure and com-pute the electronic structure. A 7 × 7×9 Monkhorst-Pack reciprocalspace grid and a plane wave energy cutoff of 500 eV were used. Inorder to account for the correlation effects intrinsic to transitionmetalcompounds, we used the DFT +U method of Dudarev et al.50 withUeff = 1 eV. This value was chosen to reproduce the splitting of the twoin-gap bands of the hybrid functional calculations of phosphorenereported in Seixas et al.33.Data availabilityThe source data generated in this study have been deposited in theFigshare under accession code51.References1. Lee, Y. H. Is it possible to create magnetic semiconductors thatfunction at room temperature? Science 382, eadl0823 (2023).2. Huang, B. et al. Electrical control of 2D magnetism in bilayer CrI3.Nat. Nanotechnol. 13, 544 (2018).3. Jiang, S., Li, L., Wang, Z., Mak, F. K. & Shan, J. Controlling mag-netism in 2D CrI3 by electrostatic doping. Nat. Nanotechnol. 13,549 (2018).4. Klein, D. R. et al. Probingmagnetism in 2D van derWaals crystallineinsulators via electron tunneling. Science 360, 1218 (2018).5. Telford, E. et al. Coupling between magnetic order and chargetransport in a two-dimensional magnetic semiconductor. Nat.Mater. 21, 754 (2022).6. Furdyna, J. K. Diluted magnetic semiconductors. J. Appl. Phys. 64,R29 (1988).7. Ohno, H. Making nonmagnetic semiconductors ferromagnetic.Science 281, 951 (1998).8. Dietl, T. A ten-year perspective on dilute magnetic semiconductorsand oxides. Nat. Mater. 9, 965 (2010).9. Ohno, H. et al. (Ga,Mn)As: a new diluted magnetic semiconductorbased on GaAs. Appl. Phys. Lett. 69, 363 (1996).Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 7www.nature.com/naturecommunications10. Chiba, D. et al. Magnetization vectormanipulation by electric fields.Nature 455, 515 (2008).11. Ohno,H. et al. Electric-field control of ferromagnetism.Nature408,944 (2000).12. Chiba, D., Yamanouchi, M., Matsukura, F. & Ohno, H. Electricalmanipulation of magnetization reversal in a ferromagnetic semi-conductor. Science 301, 943 (2003).13. Tu, N. T., Hai, P. N., Anh, L. D. & Tanaka, M. Magnetic properties andintrinsic ferromagnetism in (Ga,Fe)Sb ferromagnetic semi-conductors. Phys. Rev. B 92, 144403 (2015).14. Goel, S., Tu, N. T., Ohya, S. & Tanaka, M. In-plane to perpendicularmagnetic anisotropy switching in heavily-Fe-doped ferromagneticsemiconductor (Ga,Fe)Sb with high Curie temperature. Phys. Rev.Mater. 3, 084417 (2019).15. Matsumoto, Y. Room-temperature ferromagnetism in transpar-ent transition metal-doped titanium dioxide. Science 291, 854(2001).16. Ueda, K., Tabata, H. & Kawai, T. Magnetic and electric properties oftransition-metal-doped ZnO films. Appl. Phys. Lett. 79, 988 (2001).17. Chen, R. et al. Tunable room-temperature ferromagnetism in Co-doped two-dimensional van der Waals ZnO. Nat. Commun. 12,3952 (2021).18. Wang, H. et al. High Curie temperature ferromagnetism and highhole mobility in tensile strained Mn-doped SiGe thin films. Adv.Funct. Mater. 30, 2002513 (2020).19. Zhang, F. et al. Monolayer vanadium‐doped tungsten disulfide: aroom‐temperature dilute magnetic semiconductor. Adv. Sci. 7,2001174 (2020).20. Fu, S. et al. Enabling room temperature ferromagnetism in mono-layer MoS2 via in situ iron-doping. Nat. Commun. 11, 2034 (2020).21. Yun, S. J. et al. Ferromagnetic order at room temperature inmonolayer WSe2 semiconductor via vanadium dopant. Adv. Sci. 7,1903076 (2020).22. Song, B. et al. Evidence of itinerant holes for long-range magneticorder in the tungsten diselenide semiconductor with vanadiumdopants. Phys. Rev. B 103, 094432 (2021).23. Li, B. et al. A two-dimensional Fe-doped SnS2 magnetic semi-conductor. Nat. Commun. 8, 1958 (2017).24. Limmer, W. et al. Coupled plasmon-LO-phonon modes in Ga1-xMnxAs. Phys. Rev. B 66, 205209 (2002).25. Rajapakse, M. et al. Intercalation as a versatile tool for fabrication,property tuning, and phase transitions in 2D materials. npj 2. D.Mater. Appl. 5, 30 (2021).26. Wang, C. et al. Monolayer atomic crystal molecular superlattices.Nature 555, 231 (2018).27. Li, L. et al. Direct observation of the layer-dependent electronicstructure in phosphorene. Nat. Nanotechnol. 12, 21 (2016).28. Koenig, S. P. et al. Electric field effect in ultrathin black phosphorus.Appl. Phys. Lett. 104, 103106 (2014).29. Li, L. et al. Black phosphorus field-effect transistors. Nat. Nano-technol. 9, 372 (2014).30. Long,G. et al. Achievingultrahigh carriermobility in two-dimensionalhole gas of black phosphorus. Nano Lett. 16, 7768 (2016).31. Avsar, A. et al. Gate-tunable black phosphorus spin valve withnanosecond spin lifetimes. Nat. Phys. 13, 888 (2017).32. Cording, L. et al. Highly anisotropic spin transport in ultrathin blackphosphorus. Nat. Mater. 23, 479–485 (2024).33. Seixas, L., Carvalho, A. & Castro Neto, A. H. Atomically thin dilutemagnetism in Co-doped phosphorene. Phys. Rev. B 91, 155138(2015).34. Kulish, V. V., Malyi, O. I., Persson, C. & Wu, P. Adsorption of metaladatoms on single-layer phosphorene. Phys. Chem.Chem. Phys. 17,992–1000 (2015).35. Hu, T. & Hong, J. First-principles study of metal adatom adsorptionon black phosphorene. J. Phys. Chem. C. 119, 8199–8207 (2015).36. Gunkel, F. et al. Defect control of conventional and anomalouselectron transport at complex oxide interfaces. Phys. Rev. X 6,031035 (2016).37. Leven, B. & Dumpich, G. Resistance behavior and magnetizationreversal analysis of individual Co nanowires. Phys. Rev. B 71,064411 (2005).38. Kang, J. et al. Probing out-of-plane charge transport in blackphosphorus with graphene-contacted vertical field-effect transis-tors. Nano Lett. 16, 2580 (2016).39. Min, K.H. et al. Tunable spin injection anddetection across a vanderWaals interface. Nat. Mater. 21, 1144 (2022).40. Meservey, R. & Tedrow, P. M. Spin-polarized electron tunneling.Phys. Rep. 238, 173 (1994).41. Åkerman, J. J., Roshchin, I. V., Slaughter, J. M., Dave, R. W. &Schuller, I. K. Origin of temperature dependence in tunnelingmagnetoresistance. Europhys. Lett. 63, 104 (2003).42. Evans, R. F. L., Atxitia, U. & Chantrell, R. W. Quantitative simulationof temperature-dependent magnetization dynamics and equili-brium properties of elemental ferromagnets. Phys. Rev. B 91,144425 (2015).43. Wu, R. J. et al. Visualizing the metal-MoS2 contacts in two-dimensional field-effect transistors with atomic resolution. Phys.Rev. Mater. 3, 111001 (2019).44. Suzuki, S., Pallares, R. M. & Hibino, H. Growth of atomically thinhexagonal boron nitride films by diffusion through ametal film andprecipitation. J. Phys. D. Appl. Phys. 45, 385304 (2012).45. Matsukura, F., Tokura, Y. & Ohno, H. Control of magnetism byelectric fields. Nat. Nano. 10, 209 (2015).46. Wu, Z. et al. Large-scale growth of few-layer two-dimensional blackphosphorus. Nat. Mater. 20, 1203 (2021).47. Dean, C. R. et al. Boron nitride substrates for high-quality grapheneelectronics. Nat. Mater. 5, 722 (2010).48. Liu, Y. et al. Approaching the Schottky–Mott limit in van der Waalsmetal–semiconductor junctions. Nature 557, 696 (2018).49. Castellanos-Gomez, A. et al. Deterministic transfer of two-dimensional materials by all-dry viscoelastic stamping. 2D Mater. 1,011002 (2014).50. Dudarev, S. L., Botton, G. A., Savrasov, S. Y., Humphreys, C. J. &Sutton, A. P. Electron-energy-loss spectra and the structural stabi-lity of nickel oxide: an LSDA+U study. Phys. Rev. B 57, 1505 (1998).51. Fu, D. et al. Electricfield-tunable ferromagnetism in aVanderWaalssemi-conductor up to room temperature. Figshare.Dataset, https://doi.org/10.6084/m9.figshare.28813250 (2025).AcknowledgementsWe thank Pietro Gambardella and Thiti Taychatanapat for the fruitfuldiscussions. B.Ö. acknowledges the support from the Singapore NRFInvestigatorship (Grant No. NRF-NRFI2018-8), Competitive ResearchProgramme (Grant No. NRF-CRP22-2019-8), andMOE-AcRF-Tier 2 (GrantNo. MOE-T2EP50220-0017). D.F. acknowledges the support fromNational Natural Science Foundation of China (Grant No. 62174143). J.Linacknowledges the support from National Natural Science Foundation(Nos. 12304019, 52473302 and 12461160252), Guangdong Innovativeand Entrepreneurial Research Team Program (Grant No.2019ZT08C044), Guangdong Basic Science Foundation, China(2023B1515120039), Quantum Science Strategic Special Project (No.GDZX2301006), ShenzhenMunicipal Funding Co-construction ProgramProject (No. SZZX2301004) from the Quantum Science Center ofGuangdong-Hong Kong-Macao Greater Bay Area, China, and also theassistance of SUSTech Core Research Facilities, especially technicalsupport from Cryo-EM Centre and Pico-Centre that receives supportfrom Presidential fund and Development and Reform Commission ofShenzhen Municipality. J.F. and O.V.Y. acknowledge support by theSwiss National Science Foundation (grant No. 204254). First-principlescalculations have been performed at the Swiss NationalArticle https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 8https://doi.org/10.6084/m9.figshare.28813250https://doi.org/10.6084/m9.figshare.28813250www.nature.com/naturecommunicationsSupercomputing Centre (CSCS) under Project No. s1299 and the facil-ities of the Scientific IT and Application Support Center of EPFL. A.A.acknowledges support by the National Research Foundation, PrimeMinister’s Office, Singapore (NRFF14-2022-0083). K.S.N. acknowledgessupport from the Ministry of Education, Singapore (Research Centre ofExcellence award to the Institute for Functional Intelligent Materials, I-FIM, project No. EDUNC-33−18-279-V12), the National Research Foun-dation, Singapore under its AI Singapore Programme (AISG Award No:AISG3-RP-2022-028) and from the Royal Society (UK, Grant No. RSRP\R\190000). K.W. and T.T. acknowledge support from the JSPS KAKENHI(Grant Numbers 21H05233 and 23H02052), the CREST (JPMJCR24A5),JST and World Premier International Research Center Initiative (WPI),MEXT, Japan. A.S. acknowledges funding from the SingaporeMinistry ofEducation Academic Research Fund (AcRF NUS Grant No. 23−1072-A0001), and A*STAR’s Individual Research Grant scheme (MTC IRGGrant No. M23M6c0112).Author contributionsB.Ö. initiated, coordinated, and supervised the work. D.F., T.Q., X.C., J.Liu and Y.J. performed the device fabrication and transport measure-ments. D.F., T.Q., X.C., J.F., V.P., A.A. andB.Ö. analysed thedata. F.H. andJ. Lin performed the DF-TEM, HR-TEM and STEM characterizations.G.K.K., N.L.Y. andA.S. performedMOKEcharacterizations. S.G.,M.K. andK.S.N. performed MFM characterizations. J.F., O.V.Y. and V.M.P. pro-vided theorywork. K.W. and T.T. grew the BNcrystals. T.Q., A.A. and B.Ö.co-wrote the manuscript.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-025-59961-2.Correspondence and requests for materials should be addressed toTingyu Qu, Junhao Lin or Barbaros. Özyilmaz.Peer review information Nature Communications thanks the anon-ymous, reviewers for their contribution to the peer review of this work. Apeer 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-NonCommercial-NoDerivatives 4.0 International License,which permits any non-commercial use, sharing, distribution andreproduction in any medium or format, as long as you give appropriatecredit to the original author(s) and the source, provide a link to theCreative Commons licence, and indicate if you modified the licensedmaterial. Youdonot havepermissionunder this licence toshare adaptedmaterial derived from this article or parts of it. The images or other thirdparty material in this article are included in the article’s CreativeCommons licence, unless indicated otherwise in a credit line to thematerial. If material is not included in the article’s Creative Commonslicence and your intended use is not permitted by statutory regulation orexceeds the permitted use, you will need to obtain permission directlyfrom the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.© The Author(s) 2025Article https://doi.org/10.1038/s41467-025-59961-2Nature Communications |        (2025) 16:10197 9https://doi.org/10.1038/s41467-025-59961-2http://www.nature.com/reprintshttp://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/www.nature.com/naturecommunications Electric field-tunable ferromagnetism in a van der Waals semiconductor up to room temperature Results Methods Device fabrication STEM characterization Magnetotransport measurements MOKE characterizations MFM characterizations First-principles calculations Data availability References Acknowledgements Author contributions Competing interests Additional information