Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Spin injection by spin–charge coupling in proximity induced magnetic graphene

Spin injection by spin–charge coupling in proximity induced magnetic graphene Anyfurtherdistribution Withinthefieldofspintronicsmajoreffortsaredirectedtowardsdevelopingapplicationsfor ofthisworkmust maintainattributionto spin-basedtransportdevicesmadefullyoutoftwo-dimensionalmaterials.Inthisworkwepresent theauthor(s)andthetitle anexperimentalrealizationofaspin-valvedevicewherethegenerationofthespinsignalis ofthework,journal citationandDOI. exclusivelyattributedtothespin-dependentconductivityofthemagneticgrapheneresultingfrom theproximityofaninterlayerantiferromagnet,chromiumsulfidebromide(CrSBr).Weclearly demonstratethattheusageoftheconventionalair-sensitive3Dmagneticcontactscanbefully avoidedwhengraphene/CrSBrheterostructuresareemployed.Moreover,apartfromproviding exceptionallylongspinrelaxationlength,theusageofgrapheneforbothgenerationandtransport ofthespinallowstoautomaticallyavoidtheconductivitymismatchbetweenthesourceandthe channelcircuitsthathastobeconsideredwhenusingconventionallow-resistivecontacts.Our resultsaddressanecessarystepintheengineeringofspintroniccircuitryoutoflayeredmaterials andprecedefurtherdevelopmentsintheareaofcomplexspin-logicdevices.Moreover,we introduceafabricationprocedurewherewedesignedandimplementedarecipeforthe preparationofelectrodesviaadamage-freetechniquethatoffersanimmediateadvantageinthe fieldsofair-sensitiveanddelicateorganicmaterials. Giant magnetoresistance effect [1, 2] and spin- Yet graphene cannot offer the means for creation, transfer torque [3, 4] phenomena have already manipulation and detection of the spins. Other two- allowed for a breakthrough spin-based technology dimensional (2D) materials, however, can supple- within the area of memory-related applications. Yet, ment what graphene lacks: when it is combined in a the utilization of the spin degree of freedom within single heterostructure with an appropriately chosen thescopeofthesemiconductorindustryremainslim- companion it attains a spin-to-charge coupling via ited [5]. In order to progress further and make prac- the proximity effect that allows for an active gener- tical use of the spin transport functionality one has ation and control of spin. Such possibility to com- to advance substantially in every constituent of the bine the properties of different materials in a single spintransportdevices.Fortunately,layeredmaterials, structurehasrecentlydriventheboominginterestin beingbothratherversatileasafamily[6–8]andeasy the van der Waals heterostructures [12, 15–17]. Of to assemble into a heterostructure [9], offer a prom- direct relevance for the spintronics are the reports ising pathway to take in the view of both efficiency that experimentally demonstrate spin Hall [18], and size miniaturization and, thus, have become the Rashba–Edelstein [19–21], Zeeman spin Hall [22, mainmaterialchoiceforspintronicdevices[10–12]. 23], anomalous Hall [24–26] and spin-dependent Graphene is the prevailing host material for Seebeck [26] effects, spin-dependent conductivity spin signals that can withstand the relaxation pro- and other transport phenomena appeared/modified cesses for a record long time and be transferred due to the present spin–orbit and/or exchange inter- over tens of micrometer-long distance [13, 14]. actionsinducedingraphene[27–34]. ©2022TheAuthor(s). PublishedbyIOPPublishingLtd 2DMater.9(2022)045003 AAKaverzinetal Within this work we demonstrate the genera- tion of the spin current exclusively by the graphene itself, which is possible when it is placed on top of a layered magnetic material such as chromium sulfide bromide (CrSBr). The CrSBr is an inter- layerantiferromagnetwithaninplanemagneticeasy- axis [35, 36]. As it was demonstrated in [26], in a graphene/CrSBr heterostructure the large exchange shiftofthebandstructure(estimatedexperimentally to be ∼20meV, figure 2(a)) results in a considerable difference between the conductivities for the carriers ofoppositespinalignment,i.e. σ ̸= σ .Thisdirectly u d impliesafinitespinpolarizationofthegraphenecon- σ −σ u d ductivitydefinedasP = . Gr σ +σ u d The finite P grants graphene an active role in Gr thegenerationanddetectionofthespinsignal[37].A Figure1.(a)Thedevelopedprocessofthefabricationof chargecurrentI ,whenpassingthroughthegraphene c stand-alonecontactsisshownschematically.Top:Ti/Au contactsarefabricatedviaregularlithographyafter channel, generates an associated spin current I P c Gr exfoliationofhBNonSi/SiO /Electrasubstrate.Afterthis andviceversa.Theinitialexperimentsweresofarper- stepandbeforethestepshownbelowthesubstrateis formedondevicesthathadonlyspin-polarizedelec- spin-coatedwithPMMA.Middle:PMMAlayerwith embeddedinitTi/AucontactsandhBNflakeissuspended trodes which also inevitably contributed to the spin onwaterafterElectraisdissolved.Bottom:PMMAisplaced injection/detection[26].Incontrast,hereweprepare ontopofaPDMSstampandcanbetransferredlaterona targetedsubstrate/flakewiththehelpofatransferstage.In our stacks having both types of contacts (magnetic ourcaseitwasusedtobothpickupbilayergrapheneand andnon-magnetic)anddemonstrateadistinctgener- transfertheresultingstructureontoaCrSBrflake.Panel(b) ationofthespinsignalingraphenebyusingTiOx/Au givesthesampleimageafterthetransferisfinalized. Graphene(hBN)isoutlinedwiththered(black)dashed electrodesonly,withouthavingthemagneticcontacts line.(c)Opticalimageofthedeviceafterthedevelopment involvedinthespincurrentgenerationcircuit.Inthis andbeforethefinaldepositionoftheTiOx/Cocontacts. casethespingenerationtakesplaceexclusivelywithin thegraphene,whichbecomesa2Dsourceofthespin signal. structure is shown in figure 1(b). Graphene and For the preparation of the Au contacts we CrSBr flakes were exfoliated in advance on two developed a novel recipe that allows less mechanical separate Si/SiO substrates. Panel (c) in figure 1 stress exerted on graphene during the pick-up pro- gives the optical image of the sample taken just cedure. In addition, it avoids multiple direct spin- before depositing the magnetic electrodes made of coating of graphene with poly(methyl methacrylate) AlOxCo. (PMMA), which is known to introduce additional Overall, our observations confirm the robust- residues on the graphene surface. Our damage-free ness and consistency of the devices based on fabrication procedure has a great potential to be graphene/CrSBr stacks and offer an evident exper- particularly beneficial for the materials that cannot imental demonstration of the generation of the withstand regular lithography-based preparation of spin current distinctly within a 2D system without metallic contacts while having similar flexibility and the usage of conventional ferromagnetic contacts. resolution. The newly developed part of the sample Importantly, when graphene is used as both source preparation is summarized in the schematics and and transport channel for the spin signal, the con- sample images shown in figure 1 and is explained ductivity mismatch between the impedances of the in details in the section 3. In short, we start with sourceandtransportcircuitsisautomaticallyavoided the exfoliation of hBN and fabrication of the Ti/Au which is not the case for the commonly used metal- contacts and gates (by e-beam lithography) on a lic contacts. Moreover, the spin relaxation length water-soluble layer (Electra), as shown in panel (a). inmagneticgrapheneismuchlargerthanthatinthe By spin-coating a PMMA layer on top and dissolv- conventionalferromagneticmaterialsthussuggesting ing the Electra in water, we release the hBN–Ti– itasamoreefficientsourceofthespinsignal.Finally, Au–PMMA from the SiO substrate. The PMMA being atomically thin, graphene allows an effective layer floating on water is picked up and brought modulationofitsFermilevelwhichinturnisexpec- on top of a polydimethylsiloxane (PDMS) stamp ted to result in an active control of the spin valve and then can be transferred on any targeted flake action by the electric gate. All these facts promote on a separate substrate. In our case this hBN–Ti– graphene/CrSBr based devices as a very promising Au–PMMA–PDMS structure was used to pick up system for realizing spin functionality in a fully 2D a bilayer graphene flake and to transfer the res- systemwherethespinactioniscontrolledexclusively ulting stack on top of a CrSBr flake. Theresulting byelectricalmeans. 2 2DMater.9(2022)045003 AAKaverzinetal Figure2.(a)Exchangeshiftresultinginthespinpolarizationofthecarrierdensity.(b)Conventionalschematicsofthetwo channelmodelthatisusedtodescribethetransportingraphenewithspinpolarizedconductance.Currentisinjectedviaa non-magneticcontact,whereasthevoltageispickedupviaamagneticone.(c)Schematicsofthedevicetogetherwithcircuit geometry.(d)Spinvalvemeasurementcollectedinthegeometryschematicallyshowninthepanel(c)havingA5asinnercurrent injectorandC4asinnervoltagedetector.Black(gray)dotsaremeasuredwhensweepingB from0Tto0.4T(−0.4T)withprior alignmentdoneat −0.6T(0.6T).GreyanddarkblueverticalarrowsindicatethedirectionsofM andM .Horizontalarrows CSB C4 indicatethesweepingdirectionoftheappliedmagneticfield. 1.Resultsanddiscussion ofthecurrentflowsthroughthechannelwiththelow- est resistivity (ρ for the case shown in figure 2(a)). ThegrapheneconductivityintheproximityofCrSBr When the charge current is passed through the left becomesspinpolarizedduetotheinduced exchange part of the channel only, the created discontinuity of splitting, figure 2(a). The presence of the exchange thespincurrentattheinjectionpointgeneratesaspin interaction implies a shift in energy between elec- accumulation. From there, the spin accumulation tronstatespolarizedin‘up’and‘down’directions.We decays exponentially with the characteristic length consider these ‘up’ and ‘down’ electrons as two sep- scale λ.Consideringthecircuitshowninfigure2(b), arate carrier species of the current in our magnetic itisalsoapparentthattheinjectioncircuitisaWheat- graphene which are dissociated from each other but stonebridge.Passingthecurrentthroughitwillresult can still communicate via spin relaxation processes. in a voltage difference (spin accumulation) appear- Schematicallysuchtwo-channelmodelcanbedepic- ing between the ‘up’ and ‘down’ channels when the ted in an electrical circuit as shown in the panel (b) bridgeisunbalanced. of the figure 2, where ρ and ρ are corresponding The described above spin generation mechanism u d resistivities in each of the two channels, ρ (in units isverysimilartoaconventionalonewhenthecharge of Ω ·m)istheresistivitythatrepresentstheconnec- current is passed between a ferromagnetic material tion between the two channels via the present spin and a non-magnetic one [38], yet it does not require relaxationprocesses.Eachelectricalconnectiontothe an additional 3D ferromagnetic electrode. Similarity channel has to couple to both ‘up’ and ‘down’ chan- between these mechanisms is clear from our resist- nelsviathecorrespondingcontactresistances.Forthe ance model (figure 2(b)) and is also directly reflec- circuitshowninfigure2(b)detectionofthespinsig- ted in the appearance of the relevant terms in the nal (right side of the circuit) is realized with the use expressionfortheassociatednon-localresistanceR nl of the magnetic material (Co). This implies that the derivedin[26]: contactresistancesthatcouplethecontacttothetwo det det λR spinchannelsarenotequal,i.e.R ̸= R ,andthat sq −L/λ u d R = e (P − P )(P − P ). nl i Gr d Gr 2W(1 − P ) the spin polarization of this contact resistance P is Gr nonzero. Contrary to the common implementation (1) ofthespingenerationcircuit,fortheinjectionofthe currenthereweuseanon-magneticelectrode.Inthis Here W, R and λ are width, square resistance and sq casethespinpolarizationofthecontactresistancesP spin relaxation length of the channel, respectively. inj inj inj is exactly zero since R = R = R . Nonetheless, Theformulaisderivedforthecasewhenbothinject- sincethespinpolarizationofthechannelresistanceis inganddetectingcontactshaveafiniteP andP .The i d finite, the spin current is still generated by graphene firstpairoftheparenthesisrepresentsthetotalinjec- itself. tionefficiencyofthecircuitwhereP andP enterin gr i The mechanism of generation of the spin accu- averysimilarfashion. mulation in magnetic graphene can be understood The second parenthesis in equation (1) repres- as follows: in a homogeneous magnetic system the ents the reciprocal process in the spin circuitry, i.e. chargecurrentisspinpolarized,sincealargerportion spin detection. There are two components related to 3 2DMater.9(2022)045003 AAKaverzinetal thespinpolarizationofthemagneticgrapheneandto resistancethantheanti-parallelalignment.Therefore, thatofthemagneticdetectorelectrode.Thetwoasso- basedonthederivedrelationR ∝P (P − P )we nl Gr Gr d ciated detection mechanisms enter the equation in a conclude that P is of the opposite sign compared Gr similar way, which is also reflected by the resistance to P . Under the same assumptions as described in C4 circuit. [26] this may suggest that the graphene sample is In the experiment described here we inject the hole doped which is in agreement with the depend- current via a non-magnetic electrode (figure 2(c)). ence of graphene resistance on the applied V (see G1 Thus, P = 0, which modifies the formula for the SI,section2).However,sincewearenotabletotune measured non-local signal as R ∝ (P − P )(P − theFermilevelacrosstheDiracpointinthegraphene nl i Gr d P )| = P (P − P ) resulting in two contri- channel between contact A5 and C4 we cannot reli- Gr P =0 Gr Gr d butions.ThefirstoneisproportionaltoP andgives ably determine the position of the Fermi level and, Gr a positive spin-related background which does not thus,thesizeoftheinducedexchangesplitting.Nev- depend on the relative orientation of the magnetiza- ertheless, assuming that spin polarizations of both tionofthetopmostlayerofCrSBr,M ,withrespect grapheneconductivityandcobaltcontactsresistance CSB tothemagnetizationofthedetector.Thesecondterm are equal, we are able to estimate the spin relaxation is proportional to the product P ·P and results in lengthtobe ∼450nm(SIsection4). Gr d the two levels of the non-local resistance depending Furthermore, our assumpions result in a rough ontherelativealignmentofthemagnetizations.This estimate of the involved spin polarizations P = Gr is exactly confirmed with the experiment shown in −P ≈ 50%.Thepossibleuncertaintyintheestim- C4 figure 2(d) where the non-local spin valve is meas- ation of P can be due to several factors, e.g. Gr ured as a voltage difference between cobalt contacts inhomogeneity of the channel doping, non-equal C4 and C1 when the AC current is supplied between spin polarization of the contact and graphene res- the main gold injector A5 and a reference cobalt istances, etc. In particular, the uncertainty in the contact C6 (contacts are numbered in figure 1(c)). spin-relaxation length estimate is the most probable Themeasuredvoltagedifferenceisnormalizedbythe reasonfortheratherlargecalculatedspinpolarization applied current which gives the non-local resistance of graphene conductivity. Nevertheless, the obtained R . The measurement is performed as a function of number still suggests a consistently high efficiency nl the external magnetic field B applied along the easy of the spin signal generation within the magnetic magnetic axis of both cobalt electrodes and CrSBr graphene, thus, implying that in our spin transport flake(y-axis,alsoreferredtoascrystallographicb-axis circuit we can substitute the commonly used mag- of CrSBr [35, 36]). The value of the measured non- netic contacts with the regular non-magnetic ones local resistance is observed to have two clear levels andstillobtainasignalofasimilarmagnitude. as it is also expected from the derivation. In the case Inordertoexplorethepossibilitytotune/change of a regular graphene channel on a SiO substrate the sign of the spin polarization of graphene con- the same measurement would show no modulation ductivityweaddedtwotop-gatesG1andG2intothe ofthenon-localresistancesincetheinjectioncurrent design of the device (figure 1(c)). In the full possible viaagoldcontacthaszerospinpolarization.Yet,once rangeof [−10;10]VappliedtoG1weobservenosig- graphene is in the proximity to a ferromagnetic sub- nificantchangeinthesizeoftheswitchesyetthereis strate, the inherited exchange interaction results in a modulation of the background level (SI section 5, grapheneactingasasourceofthespinaccumulation figure4(d))whichislikelytoberelatedtothechange detectedbythemagneticcontact. inthechargerelatedbackground.Unfortunately,dur- Note that in a homogeneously proximitized ingthesweepofthegateG2,oneofthesideconnec- graphenechannel,whenonlynon-magneticcontacts tionstotheelectrodeA5waslostandtherewasasig- are used, the separation of the spin-associated sig- nificant change in the channel resistance covered by nal from the commonly present background (due the hBN (SI section 2). After the sample change the to a non-ideal Ohmic current) is not straightfor- samespinvalvegeometryasusedforfigure2(d)still ward.Inthatcasethenon-localspinsignalwouldbe showsacharacteristicspinvalvemeasurement,yetthe independent of the direction of the graphene mag- magnitudeisdecreasedfrom ∼30Ωto ∼6.6Ωasseen netization since both injection and detection cir- infigure3(b).Fromnowonwewillbediscussingthe cuits would share exactly the same magnetization sampleinthealteredstate.Notethattheestimationof determined by the underlying CrSBr. Therefore, we thespinrelaxationlengthisdoneforthechangedstate use spin-sensitive magnetic electrodes in order to of the sample since the distance dependence of the unambiguously identify the spin currents generated signal was measured only after the sample changed. when non-magnetic TiOx/Au contacts are employed Therefore, the calculation of the spin polarization of intheinjectioncircuit. grapheneconductivitydescribedearlierisdoneusing From figure 2(d) we observe that the parallel the spin valve measurement shown in figure 3(b) alignment of M and M gives a lower non-local wherethesizeoftheswitchis6.6Ω. CSB C4 4 2DMater.9(2022)045003 AAKaverzinetal Figure3.(a)Schematicsofthesamplecircuitforthe1stharmonicmeasurementsshowninpanels(b)and(c).Colorscale representsthevoltagedistributioninthechannel.(d)Schematicsofthesamplecircuitforthe2ndharmonicmeasurements showninpanels(e)and(f).Colorscalerepresentsthetemperaturedistributioninthechannel.Measurementsinspinvalve((b) and(e))andHanle((c)and(f))geometriescollectedat1st((a)and(c))and2nd((b)and(d))harmonicsofthelock-inamplifier. Blackandreddotsrepresentparallelandanti-parallelalignmentofM andM .Forpanels(b)and(e)themagneticfieldis CSB C4 sweptfrom0Tto0.4T(−0.4T)forpositive(negative)partsoftheaxisafteralignmentofallthemagnetizationsat −0.6T(0.6T). Furthermore, we noticed that the exact switch- perpendicular to the alignment of the injected spins ingbehaviorofM directionisnotalwaysreprodu- usuallyresultsinaHanleprecessionofthespinsthat CSB cibleanddoesdependonavalueofthemagneticfield takes place while they diffuse from the injector to used for the alignment. This in fact may be expected the detector contact. Yet, in magnetic graphene the considering that in our measurements we are mostly inducedexchangefieldB isstrongenoughtodes- exch sensitivetothemagnetizationofthetopmostlayerof troy all the spin components except those that are the bulk CrSBrwhereasthe full magnetization beha- (anti)parallel to the direction of B irrespective of exch viorisdeterminedbytheanti-ferromagneticinterac- relativelyweakappliedexternalfield[26].Underthis tion between the top most and its neighboring lay- conditionspinsarealwaysalignedwiththeexchange ers.Infigure3(b)weplotaspinvalvemeasuredwith field. The role of the external field is to change the thesameconnectionswhereforthepositiverangeof directions of both the exchange field (parallel to the applied magnetic field the value of the R cor- M ) and of the detecting electrode magnetization. nl CSB responds to an anti-parallel alignment. This implies Here and below we define the geometry where the that M switched back towards positive direction non-local signal is measured with the magnetic field CSB ofy-axiswhilethemagneticfieldwasbroughttozero applied along x-axis as ‘Hanle’ geometry yet this is afteralignmentat −0.6T.Asaconsequencewithfur- not a Hanle precession measurement as understood ther increase of the field only the switch of M is conventionally. C4 observedat44mTafterwhichM andM remain The magnetic detector is sensitive to the spin CSB C4 inaparallelalignment. components that are (anti)parallel to its magnet- Note that the switching field for the magnetiza- ization direction. Therefore, in this case the func- tionsoftheusedCoelectrodes(M ,M )isdeterm- tional dependence of the change observed in R C4 C3 nl ined by their shape anisotropy and is below 100mT. with applied B is proportional to the cosine of the Thisallowsustounambiguouslyidentifytheswitches angle between M and M . Cobalt contacts are CSB C4 in the R associated with the switches of M /M . much softer magnetically compared to CrSBr along nl C4 C3 Thus,foralloursampleswecanalwaysdeterminethe thex-axis.Basedonsuperconductingquantuminter- alignment of the magnetization of the used Co elec- ference device (SQUID) measurements [35], within trodeandthatofthetopmostlayerofCrSBr. the magnetic field range of B < 0.2T the direc- Tocompletetheinvestigationoftheinducedspin tion of M changes only by a few degrees while CSB accumulation we supplement spin valve measure- M becomes fully aligned with the field at B ≃ C4 x ments with the R dependence on the magnetic 0.2T. Therefore, irrespective of the initial alignment nl field applied perpendicular to the easy axis (along between M and M at B ≃ 0.2T both black C4 CSB x x-axis), figure 3(c). Black and red curves correspond and red dotted curves merge and continue jointly toparallelandanti-parallelalignmentsbetweenM at higher fields until the direction of M satur- CSB CSB and M , respectively. Applying the magnetic field ates along the y-axis at B ≃ 1.4T. Above this value C4 x 5 2DMater.9(2022)045003 AAKaverzinetal Figure4.First(a)andsecond(b)harmonicsspinvalvemeasuredatthesecondpaircontactsC3-C1.Onlylefthalfisshown. Star-shapeddatapointsareinterpretedasbeingmeasuredinthesamemagnetizationalignmentasotherredcolorpoints,butina differentresistancestateoftheparallelconductancechannel.(c)Non-localmeasurementinHanlegeometrycorrespondingto panel(a).Blackandreddotsrepresentparallelandanti-parallelalignmentofM andM .(d)Spinvalvemeasuredatpair CSB C3 C4-C1plottedfordifferentappliedgatevoltages.Additionalswitchat ±0.38Tassociatedwiththeparallelconductivechannelis suppressedatV = 10V.Arrowsindicatethedirectionofthemagneticfieldsweeping.Thesweepingisdoneinthefollowing G1 sequence:from0Tto0.6T;from0.6Tto −0.6T;from −0.6Tto0T. both cobalt contact magnetization and injected spin Thecorrespondingspinvalve(1stand2ndharmonic) direction coincide again and therefore the signal and the dependence on B (1st harmonic) are dis- recovers its initial value at B = 0T for the parallel playedinfigures4(a)–(c).Firstofall,thetwoswitches alignment. occurring at B = −32mT and B = −0.21T are y y Together with the first harmonic response of the clearly associated with spin-related signal as it fully lock-in amplifier we collected the second harmonic complies with the expected behavior. Namely, the that is commonly associated with the phenomena non-local resistance switches up at the moment of driven by the temperature gradient in the graphene the switch of the magnetization of contact C3 and it channel induced by Joule heating. In figures 3(e) switchesdowntogetherwiththeswitchofM .Fur- CSB and (f) both the second harmonic spin valve and thermore,whencomparingthepairsC4-C1andC3- dependenceofthenon-localsignalonB areshown, C1 we find that the size of the spin signal decreases measured at the same time as those given in panels substantially with the distance between the injector (b) and (c). Similar to the first harmonic response and detector electrodes. This is fully in accordance 2f there are two distinct levels of R depending on the withtheexpectedchangeofthespinsignalduetothe nl relative alignment of M and M , as expected. spinrelaxationprocesses. C4 CSB These observations clearly identify the spin origin Significantly, the magnitude of the additional of the measured signal and confirm its attribution switch is almost the same as for pair C4-C1 clearly to the spin-dependent Seebeck effect which results indicatingminimalscalingwiththedistance.Asseen, from the induced exchange interaction as in [26]. itssizeiscomparablewiththespin-associatedswitch Charge current generates a Joule heating which, due whichimpliesthatnotonlythespinvalvebutalsothe to the present spin-dependent Seebeck effect, results dependence on B should be largely affected by this inafinitespincurrentandspinaccumulationthatis additionalcontribution.Indeed,asseeninfigure4(c) sensed by the ferromagnetic detector circuit. In [26] the shape of the dependence is quite different from itwasconcludedthatthesignofthesecondharmonic what is expected and observed for example for the signal does not depend on the position of the Fermi pairC4-C1(figure3(c)).Specificallyonecouldpoint levelwhichmeansthattheswitchingbehaviorhasto out that at high enough field B ≳ 2T, when all the be identical throughout different samples irrespect- magnetizations are assumed to be aligned with the ive of the doping level. This is experimentally justi- field, the value of the non-local resistance does not fiedheresincethesignoftheswitchisthesameasthat saturate at the same level as R (B = 0). The dis- nl x reportedin[26],thus,furtherconfirmingtheconsist- tinction between the spin-related and the additional encyofourinterpretationoftheresults. contributions is further accentuated by the apparent In figure 3(b) next to the main ∼6.6Ω switch we absence of the former one in the second harmonic observe a much smaller one of ∼1Ω in size which spinvalve,figure3(b). occursatB = ±0.38T(notvisibleinthesecondhar- All these observations and particularly the monic signal). In order to understand the origin of absence of the signal scaling with the distance hint this additional switch, we compare the signal collec- at the origin of the additional contribution to be ted at the contacts pair C4-C1 with that collected at associated with the charge transport through a par- the pair C3-C1. Similar to C4, contact C3 is made allel channel. In fact, CrSBr is a semiconductor that out of cobalt but is placed further away at a distance can have a finite conductivity due to the residual of1.9µmfromtheAuinjectorA5(center-to-center). doping, although it is expected to be much lower 6 2DMater.9(2022)045003 AAKaverzinetal thanconductivityofgraphene[35](alsoseeSI).The graphene/CrSBr-based devices as a technologically dependence of CrSBr resistance on the applied mag- relevant block for building fully 2D spintronic/ netic field (also in [35] and SI) furtherindicatesthat spin-caloritronic devices. In addition, we have the behavior shown in figures 4(a) and (c) is likely developedandimplementedinthemeasureddevices to be a combination of the charge related contribu- anoveldamage-freerecipeforthepreparationofthe tion associated with the CrSBr magnetoresistance contacts separately from the studied flake/material. and the spin-related contribution originating from Suchrecipe is of a great value for air sensitive as well the spin transport in graphene. Charge transport in asfororganicmaterials. CrSBr is expected to be controlled by the position of the Fermi level that is tuned by applying a gate 3.Methodsandsamplefabrication voltage. Therefore, we studied how the spin valve measurementsoncontactspairC4-C1changeswhen Devices D1–D3 were prepared starting with exfoli- the voltage is applied on the local gate G1 within ation of the necessary layered components, namely the range [−10;10]V. In figure 4(d) the correspond- graphene, CrSBr and hBN. Graphene and CrSBr are ing spin valves are plotted with an offset for clarity. exfoliated on top of doped Si/SiO substrates. The Evidently the spin-associated signal does not change detailsofthegrowthofCrSBrbulkcrystalsaregiven significantlyunderapplyingthegatevoltagewhereas in [26, 35]. The flakes with appropriate thicknesses the additional contribution does get suppressed at are identified by their optical contrast with respect V =10V. This can be understood assuming that G1 to the SiO /Si substrate. For D1 (discussed in the by applying positive gate voltage we shift the Fermi manuscript)weusedbilayergrapheneand20–40nm level in CrSBr more into the band gap thus reducing thick CrSBr flakes. The hBN flake is exfoliated on its conductivity and blocking the unwanted paral- a separate Si/SiO substrate (SI, figure S4, step 1) lel conduction channel. Thus, we conclude that the thatispreliminarycoveredwithawatersolublecon- observed additional contribution to the measured ductive polymer Electra (AR-PC 5090.02, Allresist). signalislikelytobeduetothefiniteresistivityofthe Electra is spin-coated with the rate of 1000rpm and CrSBr, yet we show that it is possible to distinguish baked afterwards at 95 C for 1min on the hot plate. it from the spin-related component by studying the The thickness of the resulting Electra layer is about dependenceofitonthedistanceand/orgatevoltage. 200nm.WhenanappropriatehBNflakeof20–50nm thickness is selected (that is intended to be an insu- 2.Conclusions latorforthetopgate),a500nmPMMA(4%, 950K) layerisspin-coatedontopoftheSi/SiO /Electra/hBN We have performed the non-local measurements in substrate at a rate of 1000rpm (SI, figure S4, step graphene/CrSBr heterostructure in both first and 2). A freshly covered substrate is baked on the hot second harmonic and in both spin valve and Hanle plate at 180 C for 1min. By means of e-beam litho- geometries. We have demonstrated that by using graphyanappropriatelydesignedstructureisexposed exclusively a non-magnetic electrode we are able to in PMMA, including contacts and the top gate elec- createafinitespinaccumulationinsidethemagnetic trodes. After developing, the substrate is loaded into graphene. Moreover, the usage of graphene/CrSBr ane-beamevaporationsetupwhere0.5nmofTiand heterostructure for spin injection/transport offers 90nm of Au is deposited. Lift-off is done in acet- other immediate advantages compared to conven- one either at room or elevated temperature of 45 C tional3Dmagneticelectrodes.Firstly,theconductiv- and results in the structure schematically shown in ity mismatch between the spin source and transport figure1(a)(top).WeuseherethefactthatElectradoes channel is automatically avoided which simplifies notgetdissolvedineitheracetonenordevelopersolu- the optimization of the performance of the spin- tionandstaysintact. troniccircuit.Secondly,alargespinrelaxationlength Resulting Si/SiO /Electra/hBN/Ti/Au is again in graphene suggests higher efficiency of magnetic covered with PMMA using the same coating para- grapheneasasourceofthespinsignalincomparison meters as described earlier (SI, figure S4, step 4). to the conventional ferromagnetic contacts. Finally, In the next step the sample is attached to a scotch a large Seebeck coefficient of graphene ensures the tape that has a 7 by 7mm window centered at the presence of the spin-dependent Seebeck effect and hBN flake and then is immersed in water. Water dis- offers even richer functionality with coupling the solves Electra which leads to a gradual detachment spin and heat currents. Overall, our findings con- of the Si/SiO substrate that eventually sinks down firm the graphene/CrSBr heterostructure as a robust while hydrophobic PMMA film stays floating on platform for studying spin transport in a 2D mag- top of water together with the attached scotch tape, netic channel. The distinct generation of spin signal figure 1(a) (middle, the scotch tape is not shown within a 2D system when using nonmagnetic elec- in the schematics). The tape with the PMMA film trodes together with the potential tunability of the is taken from water, dried in air and later put on spin valve action by an electrical gate introduces top of a PDMS stamp with an Au/Ti/hBN structure 7 2DMater.9(2022)045003 AAKaverzinetal facing outwards, figure 1(a) (bottom). At this stage Acknowledgments the Ti layer is directly exposed to air and water and, thus, gets oxidized. Subsequently, the stamp with We would like to thank T J Schouten, H Adema, the PMMA film is used to pick up a chosen bilayer AJoshua, H de Vries and J G Holstein for technical grapheneflakefromaseparateSi/SiO substrate.The support. The presented research was funded by the pick-upsurfaceofthefullstampisflatsincebothhBN Dutch Foundation for Fundamental Research on layerandTiOx/Aucontactsarefullyimbeddedinthe Matter (FOM) as a part of the Netherlands Organ- PMMA film. Graphene/hBN/Ti(TiOx)/Au/PMMA isation for Scientific Research (NWO), the European heterostructure is thereafter placed onto a targeted Union’s Horizon 2020 research and innovation pro- CrSBr flake. The final stack is shown in figure 1(b) gram under Grant Agreement Nos. 785219 and where it is still covered with the PMMA layer. The 881603 (Graphene Flagship Core 2 and Core 3), samePMMAisusedlateronforthelithographyand NanoNed,theZernikeInstituteforAdvancedMater- depositionofAlOx/Cocontacts.PictureofdeviceD1 ials, and the Spinoza Prize awarded in 2016 to after development and just before the deposition of BJvanWees by NWO. Synthesis, structural charac- AlOx/Cocontactsisgiveninfigure1(c). terizationandmagneticmeasurementsreceivedsup- Ourrecipeoffersseveraladvantagesfromtheper- port as part of Programmable Quantum Materials, spective of fabrication of graphene-based samples an Energy Frontier Research Center funded by the that require different types of contacts simultan- U.S. Department of Energy (DOE), Office of Sci- eously. First of all, before the pick-up of graphene ence, Basic Energy Sciences (BES), under award DE- flake,thehBNflaketogetherwiththeTi/Aucontacts SC0019443. A D is supported by the NSF graduate on Si/SiO /Electra substrate is covered with a fully researchfellowshipprogram(DGE16-44869). relaxedfilmofPMMA.Thisassuresasmoothgap-less pick-up surface of the prepared mask. Second of all, ORCIDiD theTi/Aucontactsaremadeinadvanceonaseparate substrate, thus, fully avoiding the risks of any errors AlexeyAKaverzin https://orcid.org/0000-0001- occurring during this procedure. Third of all, the 5057-5038 graphene is covered by the polymer only once (dur- References ing the pick-up). Normally, during the preparation of the graphene-based structure with the two differ- [1] BaibichMN,BrotoJM,FertA,vanDauFN,PetroffF, enttypesofcontactsthesample(includinggraphene) EtienneP,CreuzetG,FriederichAandChazelasJ1988Giant magnetoresistanceof(001)Fe/(001)Crmagnetic is covered by the polymer three times: once during superlatticesPhys.Rev.Lett.612472 thestacking(whenitisdoneusingthepolycarbonate [2] BinaschG,GrünbergP,SaurenbachFandZinnW1989 film)andtwotimesforfabricatingseparatelythetwo Enhancedmagnetoresistanceinlayeredmagneticstructures differentcontacttypes.Reducingthenumberofsteps withantiferromagneticinterlayerexchangePhys.Rev.B wheregrapheneiscoveredbythepolymerisexpected [3] SlonczewskiJCetal1996Current-drivenexcitationof tolowertheamountofresiduesonitssurface. magneticmultilayersJ.Magn.Magn.Mater.159L1 Overallwehavemeasuredthreedeviceswherewe [4] MyersE,RalphD,KatineJ,LouieRandBuhrmanR1999 were able to see the switching behavior of the non- Current-inducedswitchingofdomainsinmagnetic multilayerdevicesScience285867–70 local resistance associated with the spin signal where [5] AwschalomDDandFlattéME2007Challengesfor non-magnetic contacts were used either as the cur- semiconductorspintronicsNat.Phys.3153–9 rent injector or as the voltage detector. Full charac- [6] GongCetal2017Discoveryofintrinsicferromagnetismin terization of the spin transport parameters was done two-dimensionalvanderWaalscrystalsNature546265–9 [7] BurchKS,MandrusDandParkJ-G2018Magnetismin only for the sample D1. The size of the observed two-dimensionalvanderWaalsmaterialsNature56347–52 spin signal was found to range from 50mΩ for the [8] GongCandZhangX2019Two-dimensionalmagnetic samples D2-D3 to 30 Ω for the sample D1 indicat- crystalsandemergentheterostructuredevicesScience ing a large variation in the relevant sample para- 363eaav4450 [9] GeimAKandGrigorievaIV2013VanderWaals meters (spin relaxation length, distance between the heterostructuresNature499419–25 injector/detectorelectrodes,spinpolarizationofcon- [10] CardosoC,SorianoD,García-MartínezNand ductivity of graphene and Co electrodes, etc). Fur- Fernández-RossierJ2018VanderWaalsspinvalvesPhys. ther details of the fabrication and measurements on Rev.Lett.121067701 [11] ZhaiX,XuZ,CuiQ,ZhuY,YangHandBlanterYM2021 the other samples are given in the supplementary ElectricallycontrollablevanderWaalsantiferromagnetic information. spinvalvePhys.Rev.Appl.16014032 [12] SierraJF,FabianJ,KawakamiRK,RocheSand ValenzuelaSO2021VanderWaalsheterostructuresfor Dataavailabilitystatement spintronicsandopto-spintronicsNat.Nanotechnol. 16856–68 The data that support the findings of this study are [13] HanW,KawakamiRK,GmitraMandFabianJ2014 availableuponreasonablerequestfromtheauthors. GraphenespintronicsNat.Nanotechnol.9794–807Review 8 2DMater.9(2022)045003 AAKaverzinetal [14] DrögelerM,FranzenC,VolmerF,PohlmannT,BanszerusL, spincurrentsbymagneticbilayergrapheneNat. WolterM,WatanabeK,TaniguchiT,StampferCand Nanotechnol.16788–94 BeschotenB2016Spinlifetimesexceeding12nsingraphene [27] GhiasiTS,Ingla-AynésJ,KaverzinAAandvanWeesBJ nonlocalspinvalvedevicesNanoLett.163533–9 2017Largeproximity-inducedspinlifetimeanisotropyin [15] GarciaJH,VilaM,CummingsAWandRocheS2018Spin transition-metaldichalcogenide/grapheneheterostructures transportingraphene/transitionmetaldichalcogenide NanoLett.177528–32 heterostructuresChem.Soc.Rev.473359–79 [28] BenítezLA,SierraJF,TorresWS,ArrighiA,BonellF, [16] ZollnerK,GmitraMandFabianJ2020Swappingexchange CostacheMVandValenzuelaSO2018Stronglyanisotropic andspin-orbitcouplingin2dvanderWaalsheterostructures spinrelaxationingraphene–transitionmetaldichalcogenide Phys.Rev.Lett.125196402 heterostructuresatroomtemperatureNat.Phys.14303 [17] KurebayashiH,GarciaJH,KhanS,SinovaJandRocheS [29] ZihlmannS,CummingsAW,GarciaJH,KedvesM, 2022Magnetism,symmetryandspintransportinvander WatanabeK,TaniguchiT,SchönenbergerCandMakkP WaalslayeredsystemsNat.Rev.Phys.4150–66 2018Largespinrelaxationanisotropyandvalley-Zeeman [18] SafeerC,Ingla-AynésJ,HerlingF,GarciaJH,VilaM, spin-orbitcouplinginWSe /Gr/hBNheterostructuresPhys. OntosoN,CalvoMR,RocheS,HuesoLEandCasanovaF Rev.B97075434 2019Room-temperaturespinHalleffectingraphene/MoS [30] Ingla-AynésJ,HerlingF,FabianJ,HuesoLEandCasanovaF vanderWaalsheterostructuresNanoLett.191074–82 2021Electricalcontrolofvalley-Zeeman [19] GhiasiTS,KaverzinAA,BlahPJandvanWeesBJ2019 spin-orbit-coupling-inducedspinprecessionatroom Charge-to-spinconversionbytheRashba–Edelsteineffectin temperaturePhys.Rev.Lett.127047202 two-dimensionalvanderWaalsheterostructuresuptoroom [31] LeutenantsmeyerJC,KaverzinAA,WojtaszekMandvan temperatureNanoLett.195959–66 WeesBJ2016Proximityinducedroomtemperature [20] BenítezLA,TorresWS,SierraJF,TimmermansM, ferromagnetismingrapheneprobedwithspincurrents2D GarciaJH,RocheS,CostacheMVandValenzuelaSO2020 Mater.4014001 Tunableroom-temperaturespingalvanicandspinHall [32] SinghS,KatochJ,ZhuT,MengK-Y,LiuT,BranghamJT, effectsinvanderWaalsheterostructuresNat.Mater. YangF,FlattéMEandKawakamiRK2017Strong 19170–75 modulationofspincurrentsinbilayergraphenebystaticand [21] KhokhriakovD,HoqueAM,KarpiakBandDashSP2020 fluctuatingproximityexchangefieldsPhys.Rev.Lett. Gate-tunablespin-galvaniceffectingraphene-topological 118187201 insulatorvanderWaalsheterostructuresatroom [33] KarpiakBetal2020MagneticproximityinavanderWaals temperatureNat.Commun.113657 heterostructureofmagneticinsulatorandgraphene2D [22] WeiPetal2016Stronginterfacialexchangefieldinthe Mater.7015026 graphene/EuSheterostructureNat.Mater.15711–6 [34] WuYetal2020Largeexchangesplittinginmonolayer [23] BeheraSK,BoraM,ChowdhurySSPandDebP2019 graphenemagnetizedbyanantiferromagnetNat.Electron. ProximityeffectsingrapheneandferromagneticCrBr van 3604–11 derWaalsheterostructuresPhys.Chem.Chem.Phys. [35] TelfordEJetal2020Layeredantiferromagnetisminduces 2125788–96 largenegativemagnetoresistanceinthevanderWaals [24] WangZ,TangC,SachsR,BarlasYandShiJ2015 semiconductorCrSBrAdv.Mater.322003240 Proximity-inducedferromagnetismingraphenerevealedby [36] LeeK,DismukesAH,TelfordEJ,WisconsRA,WangJ, theanomalousHalleffectPhys.Rev.Lett.114016603 XuX,NuckollsC,DeanCR,RoyXandZhuX2021 [25] TangC,ChengB,AldosaryM,WangZ,JiangZ,WatanabeK, Magneticorderandsymmetryinthe2Dsemiconductor TaniguchiT,BockrathMandShiJ2018Approaching CrSBrNanoLett.213511–7 quantumanomalousHalleffectinproximity-coupled [37] ZayetsV2012Spinandchargetransportinmaterialswith YIG/graphene/h-BNsandwichstructureAPLMaterials spin-dependentconductivityPhys.Rev.B86174415 6026401 [38] JohnsonMandSilsbeeRH1985Interfacialcharge-spin [26] GhiasiTS,KaverzinAA,DismukesAH,deWalDK,RoyX coupling:injectionanddetectionofspinmagnetizationin andvanWeesBJ2021Electricalandthermalgenerationof metalsPhys.Rev.Lett.551790 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png 2D Materials IOP Publishing

Spin injection by spin–charge coupling in proximity induced magnetic graphene

Download PDF
9 pages

Loading next page...
 
/lp/iop-publishing/spin-injection-by-spin-charge-coupling-in-proximity-induced-magnetic-b1UaH0rA3j

References (43)

Publisher
IOP Publishing
Copyright
© 2022 The Author(s). Published by IOP Publishing Ltd
eISSN
2053-1583
DOI
10.1088/2053-1583/ac7881
Publisher site
See Article on Publisher Site

Abstract

Anyfurtherdistribution Withinthefieldofspintronicsmajoreffortsaredirectedtowardsdevelopingapplicationsfor ofthisworkmust maintainattributionto spin-basedtransportdevicesmadefullyoutoftwo-dimensionalmaterials.Inthisworkwepresent theauthor(s)andthetitle anexperimentalrealizationofaspin-valvedevicewherethegenerationofthespinsignalis ofthework,journal citationandDOI. exclusivelyattributedtothespin-dependentconductivityofthemagneticgrapheneresultingfrom theproximityofaninterlayerantiferromagnet,chromiumsulfidebromide(CrSBr).Weclearly demonstratethattheusageoftheconventionalair-sensitive3Dmagneticcontactscanbefully avoidedwhengraphene/CrSBrheterostructuresareemployed.Moreover,apartfromproviding exceptionallylongspinrelaxationlength,theusageofgrapheneforbothgenerationandtransport ofthespinallowstoautomaticallyavoidtheconductivitymismatchbetweenthesourceandthe channelcircuitsthathastobeconsideredwhenusingconventionallow-resistivecontacts.Our resultsaddressanecessarystepintheengineeringofspintroniccircuitryoutoflayeredmaterials andprecedefurtherdevelopmentsintheareaofcomplexspin-logicdevices.Moreover,we introduceafabricationprocedurewherewedesignedandimplementedarecipeforthe preparationofelectrodesviaadamage-freetechniquethatoffersanimmediateadvantageinthe fieldsofair-sensitiveanddelicateorganicmaterials. Giant magnetoresistance effect [1, 2] and spin- Yet graphene cannot offer the means for creation, transfer torque [3, 4] phenomena have already manipulation and detection of the spins. Other two- allowed for a breakthrough spin-based technology dimensional (2D) materials, however, can supple- within the area of memory-related applications. Yet, ment what graphene lacks: when it is combined in a the utilization of the spin degree of freedom within single heterostructure with an appropriately chosen thescopeofthesemiconductorindustryremainslim- companion it attains a spin-to-charge coupling via ited [5]. In order to progress further and make prac- the proximity effect that allows for an active gener- tical use of the spin transport functionality one has ation and control of spin. Such possibility to com- to advance substantially in every constituent of the bine the properties of different materials in a single spintransportdevices.Fortunately,layeredmaterials, structurehasrecentlydriventheboominginterestin beingbothratherversatileasafamily[6–8]andeasy the van der Waals heterostructures [12, 15–17]. Of to assemble into a heterostructure [9], offer a prom- direct relevance for the spintronics are the reports ising pathway to take in the view of both efficiency that experimentally demonstrate spin Hall [18], and size miniaturization and, thus, have become the Rashba–Edelstein [19–21], Zeeman spin Hall [22, mainmaterialchoiceforspintronicdevices[10–12]. 23], anomalous Hall [24–26] and spin-dependent Graphene is the prevailing host material for Seebeck [26] effects, spin-dependent conductivity spin signals that can withstand the relaxation pro- and other transport phenomena appeared/modified cesses for a record long time and be transferred due to the present spin–orbit and/or exchange inter- over tens of micrometer-long distance [13, 14]. actionsinducedingraphene[27–34]. ©2022TheAuthor(s). PublishedbyIOPPublishingLtd 2DMater.9(2022)045003 AAKaverzinetal Within this work we demonstrate the genera- tion of the spin current exclusively by the graphene itself, which is possible when it is placed on top of a layered magnetic material such as chromium sulfide bromide (CrSBr). The CrSBr is an inter- layerantiferromagnetwithaninplanemagneticeasy- axis [35, 36]. As it was demonstrated in [26], in a graphene/CrSBr heterostructure the large exchange shiftofthebandstructure(estimatedexperimentally to be ∼20meV, figure 2(a)) results in a considerable difference between the conductivities for the carriers ofoppositespinalignment,i.e. σ ̸= σ .Thisdirectly u d impliesafinitespinpolarizationofthegraphenecon- σ −σ u d ductivitydefinedasP = . Gr σ +σ u d The finite P grants graphene an active role in Gr thegenerationanddetectionofthespinsignal[37].A Figure1.(a)Thedevelopedprocessofthefabricationof chargecurrentI ,whenpassingthroughthegraphene c stand-alonecontactsisshownschematically.Top:Ti/Au contactsarefabricatedviaregularlithographyafter channel, generates an associated spin current I P c Gr exfoliationofhBNonSi/SiO /Electrasubstrate.Afterthis andviceversa.Theinitialexperimentsweresofarper- stepandbeforethestepshownbelowthesubstrateis formedondevicesthathadonlyspin-polarizedelec- spin-coatedwithPMMA.Middle:PMMAlayerwith embeddedinitTi/AucontactsandhBNflakeissuspended trodes which also inevitably contributed to the spin onwaterafterElectraisdissolved.Bottom:PMMAisplaced injection/detection[26].Incontrast,hereweprepare ontopofaPDMSstampandcanbetransferredlaterona targetedsubstrate/flakewiththehelpofatransferstage.In our stacks having both types of contacts (magnetic ourcaseitwasusedtobothpickupbilayergrapheneand andnon-magnetic)anddemonstrateadistinctgener- transfertheresultingstructureontoaCrSBrflake.Panel(b) ationofthespinsignalingraphenebyusingTiOx/Au givesthesampleimageafterthetransferisfinalized. Graphene(hBN)isoutlinedwiththered(black)dashed electrodesonly,withouthavingthemagneticcontacts line.(c)Opticalimageofthedeviceafterthedevelopment involvedinthespincurrentgenerationcircuit.Inthis andbeforethefinaldepositionoftheTiOx/Cocontacts. casethespingenerationtakesplaceexclusivelywithin thegraphene,whichbecomesa2Dsourceofthespin signal. structure is shown in figure 1(b). Graphene and For the preparation of the Au contacts we CrSBr flakes were exfoliated in advance on two developed a novel recipe that allows less mechanical separate Si/SiO substrates. Panel (c) in figure 1 stress exerted on graphene during the pick-up pro- gives the optical image of the sample taken just cedure. In addition, it avoids multiple direct spin- before depositing the magnetic electrodes made of coating of graphene with poly(methyl methacrylate) AlOxCo. (PMMA), which is known to introduce additional Overall, our observations confirm the robust- residues on the graphene surface. Our damage-free ness and consistency of the devices based on fabrication procedure has a great potential to be graphene/CrSBr stacks and offer an evident exper- particularly beneficial for the materials that cannot imental demonstration of the generation of the withstand regular lithography-based preparation of spin current distinctly within a 2D system without metallic contacts while having similar flexibility and the usage of conventional ferromagnetic contacts. resolution. The newly developed part of the sample Importantly, when graphene is used as both source preparation is summarized in the schematics and and transport channel for the spin signal, the con- sample images shown in figure 1 and is explained ductivity mismatch between the impedances of the in details in the section 3. In short, we start with sourceandtransportcircuitsisautomaticallyavoided the exfoliation of hBN and fabrication of the Ti/Au which is not the case for the commonly used metal- contacts and gates (by e-beam lithography) on a lic contacts. Moreover, the spin relaxation length water-soluble layer (Electra), as shown in panel (a). inmagneticgrapheneismuchlargerthanthatinthe By spin-coating a PMMA layer on top and dissolv- conventionalferromagneticmaterialsthussuggesting ing the Electra in water, we release the hBN–Ti– itasamoreefficientsourceofthespinsignal.Finally, Au–PMMA from the SiO substrate. The PMMA being atomically thin, graphene allows an effective layer floating on water is picked up and brought modulationofitsFermilevelwhichinturnisexpec- on top of a polydimethylsiloxane (PDMS) stamp ted to result in an active control of the spin valve and then can be transferred on any targeted flake action by the electric gate. All these facts promote on a separate substrate. In our case this hBN–Ti– graphene/CrSBr based devices as a very promising Au–PMMA–PDMS structure was used to pick up system for realizing spin functionality in a fully 2D a bilayer graphene flake and to transfer the res- systemwherethespinactioniscontrolledexclusively ulting stack on top of a CrSBr flake. Theresulting byelectricalmeans. 2 2DMater.9(2022)045003 AAKaverzinetal Figure2.(a)Exchangeshiftresultinginthespinpolarizationofthecarrierdensity.(b)Conventionalschematicsofthetwo channelmodelthatisusedtodescribethetransportingraphenewithspinpolarizedconductance.Currentisinjectedviaa non-magneticcontact,whereasthevoltageispickedupviaamagneticone.(c)Schematicsofthedevicetogetherwithcircuit geometry.(d)Spinvalvemeasurementcollectedinthegeometryschematicallyshowninthepanel(c)havingA5asinnercurrent injectorandC4asinnervoltagedetector.Black(gray)dotsaremeasuredwhensweepingB from0Tto0.4T(−0.4T)withprior alignmentdoneat −0.6T(0.6T).GreyanddarkblueverticalarrowsindicatethedirectionsofM andM .Horizontalarrows CSB C4 indicatethesweepingdirectionoftheappliedmagneticfield. 1.Resultsanddiscussion ofthecurrentflowsthroughthechannelwiththelow- est resistivity (ρ for the case shown in figure 2(a)). ThegrapheneconductivityintheproximityofCrSBr When the charge current is passed through the left becomesspinpolarizedduetotheinduced exchange part of the channel only, the created discontinuity of splitting, figure 2(a). The presence of the exchange thespincurrentattheinjectionpointgeneratesaspin interaction implies a shift in energy between elec- accumulation. From there, the spin accumulation tronstatespolarizedin‘up’and‘down’directions.We decays exponentially with the characteristic length consider these ‘up’ and ‘down’ electrons as two sep- scale λ.Consideringthecircuitshowninfigure2(b), arate carrier species of the current in our magnetic itisalsoapparentthattheinjectioncircuitisaWheat- graphene which are dissociated from each other but stonebridge.Passingthecurrentthroughitwillresult can still communicate via spin relaxation processes. in a voltage difference (spin accumulation) appear- Schematicallysuchtwo-channelmodelcanbedepic- ing between the ‘up’ and ‘down’ channels when the ted in an electrical circuit as shown in the panel (b) bridgeisunbalanced. of the figure 2, where ρ and ρ are corresponding The described above spin generation mechanism u d resistivities in each of the two channels, ρ (in units isverysimilartoaconventionalonewhenthecharge of Ω ·m)istheresistivitythatrepresentstheconnec- current is passed between a ferromagnetic material tion between the two channels via the present spin and a non-magnetic one [38], yet it does not require relaxationprocesses.Eachelectricalconnectiontothe an additional 3D ferromagnetic electrode. Similarity channel has to couple to both ‘up’ and ‘down’ chan- between these mechanisms is clear from our resist- nelsviathecorrespondingcontactresistances.Forthe ance model (figure 2(b)) and is also directly reflec- circuitshowninfigure2(b)detectionofthespinsig- ted in the appearance of the relevant terms in the nal (right side of the circuit) is realized with the use expressionfortheassociatednon-localresistanceR nl of the magnetic material (Co). This implies that the derivedin[26]: contactresistancesthatcouplethecontacttothetwo det det λR spinchannelsarenotequal,i.e.R ̸= R ,andthat sq −L/λ u d R = e (P − P )(P − P ). nl i Gr d Gr 2W(1 − P ) the spin polarization of this contact resistance P is Gr nonzero. Contrary to the common implementation (1) ofthespingenerationcircuit,fortheinjectionofthe currenthereweuseanon-magneticelectrode.Inthis Here W, R and λ are width, square resistance and sq casethespinpolarizationofthecontactresistancesP spin relaxation length of the channel, respectively. inj inj inj is exactly zero since R = R = R . Nonetheless, Theformulaisderivedforthecasewhenbothinject- sincethespinpolarizationofthechannelresistanceis inganddetectingcontactshaveafiniteP andP .The i d finite, the spin current is still generated by graphene firstpairoftheparenthesisrepresentsthetotalinjec- itself. tionefficiencyofthecircuitwhereP andP enterin gr i The mechanism of generation of the spin accu- averysimilarfashion. mulation in magnetic graphene can be understood The second parenthesis in equation (1) repres- as follows: in a homogeneous magnetic system the ents the reciprocal process in the spin circuitry, i.e. chargecurrentisspinpolarized,sincealargerportion spin detection. There are two components related to 3 2DMater.9(2022)045003 AAKaverzinetal thespinpolarizationofthemagneticgrapheneandto resistancethantheanti-parallelalignment.Therefore, thatofthemagneticdetectorelectrode.Thetwoasso- basedonthederivedrelationR ∝P (P − P )we nl Gr Gr d ciated detection mechanisms enter the equation in a conclude that P is of the opposite sign compared Gr similar way, which is also reflected by the resistance to P . Under the same assumptions as described in C4 circuit. [26] this may suggest that the graphene sample is In the experiment described here we inject the hole doped which is in agreement with the depend- current via a non-magnetic electrode (figure 2(c)). ence of graphene resistance on the applied V (see G1 Thus, P = 0, which modifies the formula for the SI,section2).However,sincewearenotabletotune measured non-local signal as R ∝ (P − P )(P − theFermilevelacrosstheDiracpointinthegraphene nl i Gr d P )| = P (P − P ) resulting in two contri- channel between contact A5 and C4 we cannot reli- Gr P =0 Gr Gr d butions.ThefirstoneisproportionaltoP andgives ably determine the position of the Fermi level and, Gr a positive spin-related background which does not thus,thesizeoftheinducedexchangesplitting.Nev- depend on the relative orientation of the magnetiza- ertheless, assuming that spin polarizations of both tionofthetopmostlayerofCrSBr,M ,withrespect grapheneconductivityandcobaltcontactsresistance CSB tothemagnetizationofthedetector.Thesecondterm are equal, we are able to estimate the spin relaxation is proportional to the product P ·P and results in lengthtobe ∼450nm(SIsection4). Gr d the two levels of the non-local resistance depending Furthermore, our assumpions result in a rough ontherelativealignmentofthemagnetizations.This estimate of the involved spin polarizations P = Gr is exactly confirmed with the experiment shown in −P ≈ 50%.Thepossibleuncertaintyintheestim- C4 figure 2(d) where the non-local spin valve is meas- ation of P can be due to several factors, e.g. Gr ured as a voltage difference between cobalt contacts inhomogeneity of the channel doping, non-equal C4 and C1 when the AC current is supplied between spin polarization of the contact and graphene res- the main gold injector A5 and a reference cobalt istances, etc. In particular, the uncertainty in the contact C6 (contacts are numbered in figure 1(c)). spin-relaxation length estimate is the most probable Themeasuredvoltagedifferenceisnormalizedbythe reasonfortheratherlargecalculatedspinpolarization applied current which gives the non-local resistance of graphene conductivity. Nevertheless, the obtained R . The measurement is performed as a function of number still suggests a consistently high efficiency nl the external magnetic field B applied along the easy of the spin signal generation within the magnetic magnetic axis of both cobalt electrodes and CrSBr graphene, thus, implying that in our spin transport flake(y-axis,alsoreferredtoascrystallographicb-axis circuit we can substitute the commonly used mag- of CrSBr [35, 36]). The value of the measured non- netic contacts with the regular non-magnetic ones local resistance is observed to have two clear levels andstillobtainasignalofasimilarmagnitude. as it is also expected from the derivation. In the case Inordertoexplorethepossibilitytotune/change of a regular graphene channel on a SiO substrate the sign of the spin polarization of graphene con- the same measurement would show no modulation ductivityweaddedtwotop-gatesG1andG2intothe ofthenon-localresistancesincetheinjectioncurrent design of the device (figure 1(c)). In the full possible viaagoldcontacthaszerospinpolarization.Yet,once rangeof [−10;10]VappliedtoG1weobservenosig- graphene is in the proximity to a ferromagnetic sub- nificantchangeinthesizeoftheswitchesyetthereis strate, the inherited exchange interaction results in a modulation of the background level (SI section 5, grapheneactingasasourceofthespinaccumulation figure4(d))whichislikelytoberelatedtothechange detectedbythemagneticcontact. inthechargerelatedbackground.Unfortunately,dur- Note that in a homogeneously proximitized ingthesweepofthegateG2,oneofthesideconnec- graphenechannel,whenonlynon-magneticcontacts tionstotheelectrodeA5waslostandtherewasasig- are used, the separation of the spin-associated sig- nificant change in the channel resistance covered by nal from the commonly present background (due the hBN (SI section 2). After the sample change the to a non-ideal Ohmic current) is not straightfor- samespinvalvegeometryasusedforfigure2(d)still ward.Inthatcasethenon-localspinsignalwouldbe showsacharacteristicspinvalvemeasurement,yetthe independent of the direction of the graphene mag- magnitudeisdecreasedfrom ∼30Ωto ∼6.6Ωasseen netization since both injection and detection cir- infigure3(b).Fromnowonwewillbediscussingthe cuits would share exactly the same magnetization sampleinthealteredstate.Notethattheestimationof determined by the underlying CrSBr. Therefore, we thespinrelaxationlengthisdoneforthechangedstate use spin-sensitive magnetic electrodes in order to of the sample since the distance dependence of the unambiguously identify the spin currents generated signal was measured only after the sample changed. when non-magnetic TiOx/Au contacts are employed Therefore, the calculation of the spin polarization of intheinjectioncircuit. grapheneconductivitydescribedearlierisdoneusing From figure 2(d) we observe that the parallel the spin valve measurement shown in figure 3(b) alignment of M and M gives a lower non-local wherethesizeoftheswitchis6.6Ω. CSB C4 4 2DMater.9(2022)045003 AAKaverzinetal Figure3.(a)Schematicsofthesamplecircuitforthe1stharmonicmeasurementsshowninpanels(b)and(c).Colorscale representsthevoltagedistributioninthechannel.(d)Schematicsofthesamplecircuitforthe2ndharmonicmeasurements showninpanels(e)and(f).Colorscalerepresentsthetemperaturedistributioninthechannel.Measurementsinspinvalve((b) and(e))andHanle((c)and(f))geometriescollectedat1st((a)and(c))and2nd((b)and(d))harmonicsofthelock-inamplifier. Blackandreddotsrepresentparallelandanti-parallelalignmentofM andM .Forpanels(b)and(e)themagneticfieldis CSB C4 sweptfrom0Tto0.4T(−0.4T)forpositive(negative)partsoftheaxisafteralignmentofallthemagnetizationsat −0.6T(0.6T). Furthermore, we noticed that the exact switch- perpendicular to the alignment of the injected spins ingbehaviorofM directionisnotalwaysreprodu- usuallyresultsinaHanleprecessionofthespinsthat CSB cibleanddoesdependonavalueofthemagneticfield takes place while they diffuse from the injector to used for the alignment. This in fact may be expected the detector contact. Yet, in magnetic graphene the considering that in our measurements we are mostly inducedexchangefieldB isstrongenoughtodes- exch sensitivetothemagnetizationofthetopmostlayerof troy all the spin components except those that are the bulk CrSBrwhereasthe full magnetization beha- (anti)parallel to the direction of B irrespective of exch viorisdeterminedbytheanti-ferromagneticinterac- relativelyweakappliedexternalfield[26].Underthis tion between the top most and its neighboring lay- conditionspinsarealwaysalignedwiththeexchange ers.Infigure3(b)weplotaspinvalvemeasuredwith field. The role of the external field is to change the thesameconnectionswhereforthepositiverangeof directions of both the exchange field (parallel to the applied magnetic field the value of the R cor- M ) and of the detecting electrode magnetization. nl CSB responds to an anti-parallel alignment. This implies Here and below we define the geometry where the that M switched back towards positive direction non-local signal is measured with the magnetic field CSB ofy-axiswhilethemagneticfieldwasbroughttozero applied along x-axis as ‘Hanle’ geometry yet this is afteralignmentat −0.6T.Asaconsequencewithfur- not a Hanle precession measurement as understood ther increase of the field only the switch of M is conventionally. C4 observedat44mTafterwhichM andM remain The magnetic detector is sensitive to the spin CSB C4 inaparallelalignment. components that are (anti)parallel to its magnet- Note that the switching field for the magnetiza- ization direction. Therefore, in this case the func- tionsoftheusedCoelectrodes(M ,M )isdeterm- tional dependence of the change observed in R C4 C3 nl ined by their shape anisotropy and is below 100mT. with applied B is proportional to the cosine of the Thisallowsustounambiguouslyidentifytheswitches angle between M and M . Cobalt contacts are CSB C4 in the R associated with the switches of M /M . much softer magnetically compared to CrSBr along nl C4 C3 Thus,foralloursampleswecanalwaysdeterminethe thex-axis.Basedonsuperconductingquantuminter- alignment of the magnetization of the used Co elec- ference device (SQUID) measurements [35], within trodeandthatofthetopmostlayerofCrSBr. the magnetic field range of B < 0.2T the direc- Tocompletetheinvestigationoftheinducedspin tion of M changes only by a few degrees while CSB accumulation we supplement spin valve measure- M becomes fully aligned with the field at B ≃ C4 x ments with the R dependence on the magnetic 0.2T. Therefore, irrespective of the initial alignment nl field applied perpendicular to the easy axis (along between M and M at B ≃ 0.2T both black C4 CSB x x-axis), figure 3(c). Black and red curves correspond and red dotted curves merge and continue jointly toparallelandanti-parallelalignmentsbetweenM at higher fields until the direction of M satur- CSB CSB and M , respectively. Applying the magnetic field ates along the y-axis at B ≃ 1.4T. Above this value C4 x 5 2DMater.9(2022)045003 AAKaverzinetal Figure4.First(a)andsecond(b)harmonicsspinvalvemeasuredatthesecondpaircontactsC3-C1.Onlylefthalfisshown. Star-shapeddatapointsareinterpretedasbeingmeasuredinthesamemagnetizationalignmentasotherredcolorpoints,butina differentresistancestateoftheparallelconductancechannel.(c)Non-localmeasurementinHanlegeometrycorrespondingto panel(a).Blackandreddotsrepresentparallelandanti-parallelalignmentofM andM .(d)Spinvalvemeasuredatpair CSB C3 C4-C1plottedfordifferentappliedgatevoltages.Additionalswitchat ±0.38Tassociatedwiththeparallelconductivechannelis suppressedatV = 10V.Arrowsindicatethedirectionofthemagneticfieldsweeping.Thesweepingisdoneinthefollowing G1 sequence:from0Tto0.6T;from0.6Tto −0.6T;from −0.6Tto0T. both cobalt contact magnetization and injected spin Thecorrespondingspinvalve(1stand2ndharmonic) direction coincide again and therefore the signal and the dependence on B (1st harmonic) are dis- recovers its initial value at B = 0T for the parallel playedinfigures4(a)–(c).Firstofall,thetwoswitches alignment. occurring at B = −32mT and B = −0.21T are y y Together with the first harmonic response of the clearly associated with spin-related signal as it fully lock-in amplifier we collected the second harmonic complies with the expected behavior. Namely, the that is commonly associated with the phenomena non-local resistance switches up at the moment of driven by the temperature gradient in the graphene the switch of the magnetization of contact C3 and it channel induced by Joule heating. In figures 3(e) switchesdowntogetherwiththeswitchofM .Fur- CSB and (f) both the second harmonic spin valve and thermore,whencomparingthepairsC4-C1andC3- dependenceofthenon-localsignalonB areshown, C1 we find that the size of the spin signal decreases measured at the same time as those given in panels substantially with the distance between the injector (b) and (c). Similar to the first harmonic response and detector electrodes. This is fully in accordance 2f there are two distinct levels of R depending on the withtheexpectedchangeofthespinsignalduetothe nl relative alignment of M and M , as expected. spinrelaxationprocesses. C4 CSB These observations clearly identify the spin origin Significantly, the magnitude of the additional of the measured signal and confirm its attribution switch is almost the same as for pair C4-C1 clearly to the spin-dependent Seebeck effect which results indicatingminimalscalingwiththedistance.Asseen, from the induced exchange interaction as in [26]. itssizeiscomparablewiththespin-associatedswitch Charge current generates a Joule heating which, due whichimpliesthatnotonlythespinvalvebutalsothe to the present spin-dependent Seebeck effect, results dependence on B should be largely affected by this inafinitespincurrentandspinaccumulationthatis additionalcontribution.Indeed,asseeninfigure4(c) sensed by the ferromagnetic detector circuit. In [26] the shape of the dependence is quite different from itwasconcludedthatthesignofthesecondharmonic what is expected and observed for example for the signal does not depend on the position of the Fermi pairC4-C1(figure3(c)).Specificallyonecouldpoint levelwhichmeansthattheswitchingbehaviorhasto out that at high enough field B ≳ 2T, when all the be identical throughout different samples irrespect- magnetizations are assumed to be aligned with the ive of the doping level. This is experimentally justi- field, the value of the non-local resistance does not fiedheresincethesignoftheswitchisthesameasthat saturate at the same level as R (B = 0). The dis- nl x reportedin[26],thus,furtherconfirmingtheconsist- tinction between the spin-related and the additional encyofourinterpretationoftheresults. contributions is further accentuated by the apparent In figure 3(b) next to the main ∼6.6Ω switch we absence of the former one in the second harmonic observe a much smaller one of ∼1Ω in size which spinvalve,figure3(b). occursatB = ±0.38T(notvisibleinthesecondhar- All these observations and particularly the monic signal). In order to understand the origin of absence of the signal scaling with the distance hint this additional switch, we compare the signal collec- at the origin of the additional contribution to be ted at the contacts pair C4-C1 with that collected at associated with the charge transport through a par- the pair C3-C1. Similar to C4, contact C3 is made allel channel. In fact, CrSBr is a semiconductor that out of cobalt but is placed further away at a distance can have a finite conductivity due to the residual of1.9µmfromtheAuinjectorA5(center-to-center). doping, although it is expected to be much lower 6 2DMater.9(2022)045003 AAKaverzinetal thanconductivityofgraphene[35](alsoseeSI).The graphene/CrSBr-based devices as a technologically dependence of CrSBr resistance on the applied mag- relevant block for building fully 2D spintronic/ netic field (also in [35] and SI) furtherindicatesthat spin-caloritronic devices. In addition, we have the behavior shown in figures 4(a) and (c) is likely developedandimplementedinthemeasureddevices to be a combination of the charge related contribu- anoveldamage-freerecipeforthepreparationofthe tion associated with the CrSBr magnetoresistance contacts separately from the studied flake/material. and the spin-related contribution originating from Suchrecipe is of a great value for air sensitive as well the spin transport in graphene. Charge transport in asfororganicmaterials. CrSBr is expected to be controlled by the position of the Fermi level that is tuned by applying a gate 3.Methodsandsamplefabrication voltage. Therefore, we studied how the spin valve measurementsoncontactspairC4-C1changeswhen Devices D1–D3 were prepared starting with exfoli- the voltage is applied on the local gate G1 within ation of the necessary layered components, namely the range [−10;10]V. In figure 4(d) the correspond- graphene, CrSBr and hBN. Graphene and CrSBr are ing spin valves are plotted with an offset for clarity. exfoliated on top of doped Si/SiO substrates. The Evidently the spin-associated signal does not change detailsofthegrowthofCrSBrbulkcrystalsaregiven significantlyunderapplyingthegatevoltagewhereas in [26, 35]. The flakes with appropriate thicknesses the additional contribution does get suppressed at are identified by their optical contrast with respect V =10V. This can be understood assuming that G1 to the SiO /Si substrate. For D1 (discussed in the by applying positive gate voltage we shift the Fermi manuscript)weusedbilayergrapheneand20–40nm level in CrSBr more into the band gap thus reducing thick CrSBr flakes. The hBN flake is exfoliated on its conductivity and blocking the unwanted paral- a separate Si/SiO substrate (SI, figure S4, step 1) lel conduction channel. Thus, we conclude that the thatispreliminarycoveredwithawatersolublecon- observed additional contribution to the measured ductive polymer Electra (AR-PC 5090.02, Allresist). signalislikelytobeduetothefiniteresistivityofthe Electra is spin-coated with the rate of 1000rpm and CrSBr, yet we show that it is possible to distinguish baked afterwards at 95 C for 1min on the hot plate. it from the spin-related component by studying the The thickness of the resulting Electra layer is about dependenceofitonthedistanceand/orgatevoltage. 200nm.WhenanappropriatehBNflakeof20–50nm thickness is selected (that is intended to be an insu- 2.Conclusions latorforthetopgate),a500nmPMMA(4%, 950K) layerisspin-coatedontopoftheSi/SiO /Electra/hBN We have performed the non-local measurements in substrate at a rate of 1000rpm (SI, figure S4, step graphene/CrSBr heterostructure in both first and 2). A freshly covered substrate is baked on the hot second harmonic and in both spin valve and Hanle plate at 180 C for 1min. By means of e-beam litho- geometries. We have demonstrated that by using graphyanappropriatelydesignedstructureisexposed exclusively a non-magnetic electrode we are able to in PMMA, including contacts and the top gate elec- createafinitespinaccumulationinsidethemagnetic trodes. After developing, the substrate is loaded into graphene. Moreover, the usage of graphene/CrSBr ane-beamevaporationsetupwhere0.5nmofTiand heterostructure for spin injection/transport offers 90nm of Au is deposited. Lift-off is done in acet- other immediate advantages compared to conven- one either at room or elevated temperature of 45 C tional3Dmagneticelectrodes.Firstly,theconductiv- and results in the structure schematically shown in ity mismatch between the spin source and transport figure1(a)(top).WeuseherethefactthatElectradoes channel is automatically avoided which simplifies notgetdissolvedineitheracetonenordevelopersolu- the optimization of the performance of the spin- tionandstaysintact. troniccircuit.Secondly,alargespinrelaxationlength Resulting Si/SiO /Electra/hBN/Ti/Au is again in graphene suggests higher efficiency of magnetic covered with PMMA using the same coating para- grapheneasasourceofthespinsignalincomparison meters as described earlier (SI, figure S4, step 4). to the conventional ferromagnetic contacts. Finally, In the next step the sample is attached to a scotch a large Seebeck coefficient of graphene ensures the tape that has a 7 by 7mm window centered at the presence of the spin-dependent Seebeck effect and hBN flake and then is immersed in water. Water dis- offers even richer functionality with coupling the solves Electra which leads to a gradual detachment spin and heat currents. Overall, our findings con- of the Si/SiO substrate that eventually sinks down firm the graphene/CrSBr heterostructure as a robust while hydrophobic PMMA film stays floating on platform for studying spin transport in a 2D mag- top of water together with the attached scotch tape, netic channel. The distinct generation of spin signal figure 1(a) (middle, the scotch tape is not shown within a 2D system when using nonmagnetic elec- in the schematics). The tape with the PMMA film trodes together with the potential tunability of the is taken from water, dried in air and later put on spin valve action by an electrical gate introduces top of a PDMS stamp with an Au/Ti/hBN structure 7 2DMater.9(2022)045003 AAKaverzinetal facing outwards, figure 1(a) (bottom). At this stage Acknowledgments the Ti layer is directly exposed to air and water and, thus, gets oxidized. Subsequently, the stamp with We would like to thank T J Schouten, H Adema, the PMMA film is used to pick up a chosen bilayer AJoshua, H de Vries and J G Holstein for technical grapheneflakefromaseparateSi/SiO substrate.The support. The presented research was funded by the pick-upsurfaceofthefullstampisflatsincebothhBN Dutch Foundation for Fundamental Research on layerandTiOx/Aucontactsarefullyimbeddedinthe Matter (FOM) as a part of the Netherlands Organ- PMMA film. Graphene/hBN/Ti(TiOx)/Au/PMMA isation for Scientific Research (NWO), the European heterostructure is thereafter placed onto a targeted Union’s Horizon 2020 research and innovation pro- CrSBr flake. The final stack is shown in figure 1(b) gram under Grant Agreement Nos. 785219 and where it is still covered with the PMMA layer. The 881603 (Graphene Flagship Core 2 and Core 3), samePMMAisusedlateronforthelithographyand NanoNed,theZernikeInstituteforAdvancedMater- depositionofAlOx/Cocontacts.PictureofdeviceD1 ials, and the Spinoza Prize awarded in 2016 to after development and just before the deposition of BJvanWees by NWO. Synthesis, structural charac- AlOx/Cocontactsisgiveninfigure1(c). terizationandmagneticmeasurementsreceivedsup- Ourrecipeoffersseveraladvantagesfromtheper- port as part of Programmable Quantum Materials, spective of fabrication of graphene-based samples an Energy Frontier Research Center funded by the that require different types of contacts simultan- U.S. Department of Energy (DOE), Office of Sci- eously. First of all, before the pick-up of graphene ence, Basic Energy Sciences (BES), under award DE- flake,thehBNflaketogetherwiththeTi/Aucontacts SC0019443. A D is supported by the NSF graduate on Si/SiO /Electra substrate is covered with a fully researchfellowshipprogram(DGE16-44869). relaxedfilmofPMMA.Thisassuresasmoothgap-less pick-up surface of the prepared mask. Second of all, ORCIDiD theTi/Aucontactsaremadeinadvanceonaseparate substrate, thus, fully avoiding the risks of any errors AlexeyAKaverzin https://orcid.org/0000-0001- occurring during this procedure. Third of all, the 5057-5038 graphene is covered by the polymer only once (dur- References ing the pick-up). Normally, during the preparation of the graphene-based structure with the two differ- [1] BaibichMN,BrotoJM,FertA,vanDauFN,PetroffF, enttypesofcontactsthesample(includinggraphene) EtienneP,CreuzetG,FriederichAandChazelasJ1988Giant magnetoresistanceof(001)Fe/(001)Crmagnetic is covered by the polymer three times: once during superlatticesPhys.Rev.Lett.612472 thestacking(whenitisdoneusingthepolycarbonate [2] BinaschG,GrünbergP,SaurenbachFandZinnW1989 film)andtwotimesforfabricatingseparatelythetwo Enhancedmagnetoresistanceinlayeredmagneticstructures differentcontacttypes.Reducingthenumberofsteps withantiferromagneticinterlayerexchangePhys.Rev.B wheregrapheneiscoveredbythepolymerisexpected [3] SlonczewskiJCetal1996Current-drivenexcitationof tolowertheamountofresiduesonitssurface. magneticmultilayersJ.Magn.Magn.Mater.159L1 Overallwehavemeasuredthreedeviceswherewe [4] MyersE,RalphD,KatineJ,LouieRandBuhrmanR1999 were able to see the switching behavior of the non- Current-inducedswitchingofdomainsinmagnetic multilayerdevicesScience285867–70 local resistance associated with the spin signal where [5] AwschalomDDandFlattéME2007Challengesfor non-magnetic contacts were used either as the cur- semiconductorspintronicsNat.Phys.3153–9 rent injector or as the voltage detector. Full charac- [6] GongCetal2017Discoveryofintrinsicferromagnetismin terization of the spin transport parameters was done two-dimensionalvanderWaalscrystalsNature546265–9 [7] BurchKS,MandrusDandParkJ-G2018Magnetismin only for the sample D1. The size of the observed two-dimensionalvanderWaalsmaterialsNature56347–52 spin signal was found to range from 50mΩ for the [8] GongCandZhangX2019Two-dimensionalmagnetic samples D2-D3 to 30 Ω for the sample D1 indicat- crystalsandemergentheterostructuredevicesScience ing a large variation in the relevant sample para- 363eaav4450 [9] GeimAKandGrigorievaIV2013VanderWaals meters (spin relaxation length, distance between the heterostructuresNature499419–25 injector/detectorelectrodes,spinpolarizationofcon- [10] CardosoC,SorianoD,García-MartínezNand ductivity of graphene and Co electrodes, etc). Fur- Fernández-RossierJ2018VanderWaalsspinvalvesPhys. ther details of the fabrication and measurements on Rev.Lett.121067701 [11] ZhaiX,XuZ,CuiQ,ZhuY,YangHandBlanterYM2021 the other samples are given in the supplementary ElectricallycontrollablevanderWaalsantiferromagnetic information. spinvalvePhys.Rev.Appl.16014032 [12] SierraJF,FabianJ,KawakamiRK,RocheSand ValenzuelaSO2021VanderWaalsheterostructuresfor Dataavailabilitystatement spintronicsandopto-spintronicsNat.Nanotechnol. 16856–68 The data that support the findings of this study are [13] HanW,KawakamiRK,GmitraMandFabianJ2014 availableuponreasonablerequestfromtheauthors. GraphenespintronicsNat.Nanotechnol.9794–807Review 8 2DMater.9(2022)045003 AAKaverzinetal [14] DrögelerM,FranzenC,VolmerF,PohlmannT,BanszerusL, spincurrentsbymagneticbilayergrapheneNat. WolterM,WatanabeK,TaniguchiT,StampferCand Nanotechnol.16788–94 BeschotenB2016Spinlifetimesexceeding12nsingraphene [27] GhiasiTS,Ingla-AynésJ,KaverzinAAandvanWeesBJ nonlocalspinvalvedevicesNanoLett.163533–9 2017Largeproximity-inducedspinlifetimeanisotropyin [15] GarciaJH,VilaM,CummingsAWandRocheS2018Spin transition-metaldichalcogenide/grapheneheterostructures transportingraphene/transitionmetaldichalcogenide NanoLett.177528–32 heterostructuresChem.Soc.Rev.473359–79 [28] BenítezLA,SierraJF,TorresWS,ArrighiA,BonellF, [16] ZollnerK,GmitraMandFabianJ2020Swappingexchange CostacheMVandValenzuelaSO2018Stronglyanisotropic andspin-orbitcouplingin2dvanderWaalsheterostructures spinrelaxationingraphene–transitionmetaldichalcogenide Phys.Rev.Lett.125196402 heterostructuresatroomtemperatureNat.Phys.14303 [17] KurebayashiH,GarciaJH,KhanS,SinovaJandRocheS [29] ZihlmannS,CummingsAW,GarciaJH,KedvesM, 2022Magnetism,symmetryandspintransportinvander WatanabeK,TaniguchiT,SchönenbergerCandMakkP WaalslayeredsystemsNat.Rev.Phys.4150–66 2018Largespinrelaxationanisotropyandvalley-Zeeman [18] SafeerC,Ingla-AynésJ,HerlingF,GarciaJH,VilaM, spin-orbitcouplinginWSe /Gr/hBNheterostructuresPhys. OntosoN,CalvoMR,RocheS,HuesoLEandCasanovaF Rev.B97075434 2019Room-temperaturespinHalleffectingraphene/MoS [30] Ingla-AynésJ,HerlingF,FabianJ,HuesoLEandCasanovaF vanderWaalsheterostructuresNanoLett.191074–82 2021Electricalcontrolofvalley-Zeeman [19] GhiasiTS,KaverzinAA,BlahPJandvanWeesBJ2019 spin-orbit-coupling-inducedspinprecessionatroom Charge-to-spinconversionbytheRashba–Edelsteineffectin temperaturePhys.Rev.Lett.127047202 two-dimensionalvanderWaalsheterostructuresuptoroom [31] LeutenantsmeyerJC,KaverzinAA,WojtaszekMandvan temperatureNanoLett.195959–66 WeesBJ2016Proximityinducedroomtemperature [20] BenítezLA,TorresWS,SierraJF,TimmermansM, ferromagnetismingrapheneprobedwithspincurrents2D GarciaJH,RocheS,CostacheMVandValenzuelaSO2020 Mater.4014001 Tunableroom-temperaturespingalvanicandspinHall [32] SinghS,KatochJ,ZhuT,MengK-Y,LiuT,BranghamJT, effectsinvanderWaalsheterostructuresNat.Mater. YangF,FlattéMEandKawakamiRK2017Strong 19170–75 modulationofspincurrentsinbilayergraphenebystaticand [21] KhokhriakovD,HoqueAM,KarpiakBandDashSP2020 fluctuatingproximityexchangefieldsPhys.Rev.Lett. Gate-tunablespin-galvaniceffectingraphene-topological 118187201 insulatorvanderWaalsheterostructuresatroom [33] KarpiakBetal2020MagneticproximityinavanderWaals temperatureNat.Commun.113657 heterostructureofmagneticinsulatorandgraphene2D [22] WeiPetal2016Stronginterfacialexchangefieldinthe Mater.7015026 graphene/EuSheterostructureNat.Mater.15711–6 [34] WuYetal2020Largeexchangesplittinginmonolayer [23] BeheraSK,BoraM,ChowdhurySSPandDebP2019 graphenemagnetizedbyanantiferromagnetNat.Electron. ProximityeffectsingrapheneandferromagneticCrBr van 3604–11 derWaalsheterostructuresPhys.Chem.Chem.Phys. [35] TelfordEJetal2020Layeredantiferromagnetisminduces 2125788–96 largenegativemagnetoresistanceinthevanderWaals [24] WangZ,TangC,SachsR,BarlasYandShiJ2015 semiconductorCrSBrAdv.Mater.322003240 Proximity-inducedferromagnetismingraphenerevealedby [36] LeeK,DismukesAH,TelfordEJ,WisconsRA,WangJ, theanomalousHalleffectPhys.Rev.Lett.114016603 XuX,NuckollsC,DeanCR,RoyXandZhuX2021 [25] TangC,ChengB,AldosaryM,WangZ,JiangZ,WatanabeK, Magneticorderandsymmetryinthe2Dsemiconductor TaniguchiT,BockrathMandShiJ2018Approaching CrSBrNanoLett.213511–7 quantumanomalousHalleffectinproximity-coupled [37] ZayetsV2012Spinandchargetransportinmaterialswith YIG/graphene/h-BNsandwichstructureAPLMaterials spin-dependentconductivityPhys.Rev.B86174415 6026401 [38] JohnsonMandSilsbeeRH1985Interfacialcharge-spin [26] GhiasiTS,KaverzinAA,DismukesAH,deWalDK,RoyX coupling:injectionanddetectionofspinmagnetizationin andvanWeesBJ2021Electricalandthermalgenerationof metalsPhys.Rev.Lett.551790

Journal

2D Materials – IOP Publishing

Published: Oct 1, 2022

Keywords: magnetic graphene; spin-dependent conductivity; two-dimensional spin valve; spintronics; spin-dependent Seebeck effect

There are no references for this article.