Visualizing the elongated vortices in γ-Ga nanostrips Hui-Min Zhang,1,2 Zi-Xiang Li,3 Jun-Ping Peng,2 Can-Li Song,1,4,∗ Jia-Qi Guan,2 Zhi Li,2 Lili Wang,1,4 Ke He,1,4 Shuai-Hua Ji,1,4 Xi Chen,1,4 Hong Yao,3,4,† Xu-Cun Ma,1,2,4,‡ and Qi-Kun Xue1,4 1State Key Laboratory of Low-Dimensional Quantum Physics, Department of Physics, Tsinghua University, Beijing 100084, China 2Institute of Physics, Chinese Academy of Sciences, Beijing 100190, China 3Institute for Advanced Study, Tsinghua University, Beijing 100084, China 4Collaborative Innovation Center of Quantum Matter, Beijing 100084, China (Dated: April 5, 2016) 6 Westudythemagneticresponseofsuperconductingγ-Gavialowtemperaturescanningtunneling 1 microscopyandspectroscopy. ThemagneticvortexcoresrelysubstantiallyontheGageometry,and 0 exhibit an unexpectedly-large axial elongation with aspect ratio up to 40 in rectangular Ga nano- 2 strips(widthl<100nm). Thisisinstarkcontrastwiththeisotropiccircularvortexcoreinalarger r round-shapedGaisland. WesuggestthattheunusualelongatedvorticesinGananostripsoriginate p fromgeometricconfinementeffectprobablyviathestrongrepulsiveinteractionbetweenthevortices A andMeissnerscreeningcurrentsatthesampleedge. Ourfindingprovidesnovelconceptualinsights intothegeometricalconfinementeffectonmagneticvorticesandformsthebasisforthetechnological 3 applications of superconductors. ] n PACSnumbers: 68.37.Ef,74.25.Op,74.78.Na,74.25.Ha o c - Magnetic vortices in mesoscopic superconductors, in- the model applies for the extremely narrow supercon- r p cluding symmetric disks, triangles, squares and rectan- ducting nanostrips remains unjustified. u gles [1–4], have recently stimulated tremendous interest Herein we report the direct visualization of unexpect- s due to their fundamental importance in controlling su- edly elongated vortex cores in superconducting γ-Ga . t perconductingenergydissipation,asignificantfeaturefor nanostrips by STM, and show that they are caused by a m superconductor-based nanotechnologies. When a super- thestrongrepulsiveinteractionbetweenthevorticesand conductor is shrunk to nanoscale, exotic vortex shapes Meissner screening currents at the sample edge. All ex- - d (i.e. giant vortex and antivortex-vortex molecule) and periments were conducted in a Unisoku ultrahigh vac- n configurationsmayoccur,butsofarmostlyinferredfrom uum STM system equipped with molecular beam epi- o indirectmeasurements[5]. Directrealspacevisualization taxy (MBE) for in-situ sample preparation. The MBE c of vortices in nanostructured superconductors remains growth of superconducting Ga nanoislands has been de- [ challenging and rarely explored. A few scanning tunnel- scribed in detail elsewhere [1], and in the Supplemental 3 ing microscopy (STM) studies present the preliminary Material [17]. The measurements were carried out at 4.4 v evidence of giant vortex in Pb islands [6–8], but the ill- K unless otherwise noted. A magnetic field up to 7 T 7 6 definedgeometryofthesamplesinvestigatedrendersthe can be applied perpendicular to the sample surface. The 0 issue quite complicatedand difficult fora direct compar- differential conductance dI/dV spectra and maps were 5 ison with theory. measured by disrupting the feedback circuit at the set- 0 Amongallthesesuperconductingnanostructures,rect- point of V = 10 mV and I = 100 pA, sweeping the . s 1 angular nanostrips or nanowires are especially appealing sample voltage V , and extracting the conductance using s 0 forsuperconductingelectricalcircuitandquantuminter- a standard lock-in technique with a bias modulation of 5 ferencedevice. Regularoscillationsofresistanceandcrit- 0.3 mV at 987.5 Hz. 1 ical current as a function of magnetic field have been re- Figure 1(a) shows the constant-current topographic : v spectivelyreportedinsuperconductingInOandAlstrips image of a rectangular Ga nanostrip, which has a length i X [9,10],whichcanbeaccountedforbytheso-calledWeber l∼440nmandawidthw ∼64nm. Notethatthenanos- r blockadetheory[11]. Hereeachoscillationineithermag- tripexhibitsanalmostperfectrectangle,whichcaneasily a netoresistance or critical current corresponds to adding be modeled in theory. The atomically resolved STM im- asinglevortexintothestrip, inanalogywithsingleelec- age on the nanostrip, shown in Fig. 1(b), reveals defect- tron transport through quantum dots in the Coulomb free surface. The unit cell, marked by the blue rhombus, blockade regime. Furthermore, the penetrating vortices consists of two bright spots with a periodicity of a = 0 areanticipatedtoconfigureintoaone-dimensionalchain 0.86 nm and an intersection angle of 76o, which are con- at the center of strip [10–12], because of the lower vor- sistent with γ-Ga(001) plane [Fig. 1(c)] [18]. Figure 1(d) tex potential energy there. Upon increasing field, the plots a series of differential conductance dI/dV spectra vortex chain may split into two or more buckled parallel acquired on the Ga nanostrip at various temperatures, rows [13–15]. However, so far as we know, such vortex normalized to the normal-state one above T (10 K). A c chains have little been directly visualized, and whether superconducting gap with clear coherence peaks is visi- 2 (a) (b) (d) 5 7.0 K diately noticed. The first phenomenon, which is also the 6.8 K 4 6.6 K more interesting and unexpected one, is that the mag- malized dI/dV23 45566.....25804 KKKKK n(naeatnaiocxsivtsro)irpat,enxrdecssophreoecrsttieevrxepl(yba,nardexgiasan)rddsilcdeosemssopofrfettshhseeafirloeecnldtga.ntThgehuillsaorncogGnear- Nor 3.0 K sequentlyleadstotheanisotropicvortexcoreswithgreat 1 2.5 K 1 nm axial elongation. The elongation ratio η is estimated to 6.36nm 0 (c) -10 -5 0 5 10 be15at0.435T,andincreasesfurtherwiththemagnetic (e) Sample bias (mV) 1.5 field. At1.0T,manyvorticesmergetogetherandthesu- perconductivity is almost completely killed, pushing the 0 ba 30 nm meV)1 Data whole island into the normal state. Figure 2(b) reveals m)6 76o Δ (0.5 BCS theory the field-dependent ZBC profiles across the vortex cores. Height (n0240 4P0ositio8n0 (nm12)0 160 bca 00 Tc=2T6.e8m0 Kpera4ture (K)6 8 AevleornygZtBheCbparxofiisl,etehxeceluxdisetsenrecaesoofnoanbllyy othnee pexotsrseibmiluitmy ionf vortex chain for a single individual yellow stripe. On the FIG. 1. (color online) (a) STM topography of a rectangular other hand, ZBC oscillates along the a axis, with each Ga nanostrip with a height of 5.85 nm (45 ML) (V = 3.5 V, s peaked region representing single or multiple (e.g. giant I = 50 pA, 483 nm × 161 nm). Line profile taken along the vortex) of the magnetic flux quantum Φ . The number red curve characterizes the height and width w of Ga nanos- 0 trip. Bothlabelsa andb denoterespectivelytheorientations of peaks invariably appears smaller than, but tends to of the short and long side of the rectangular strips through- close the applied flux Φ/Φ0 as the field increases. Above out this paper. Dotted rectangle marks the region where all 0.7 T (Φ/Φ = 10.1), the ratio between them becomes 0 ZBC maps are taken in Fig. 2(a). (b) Atomic resolution im- greaterthan0.5,rulingoutthepossibilityofgiantvortex. age (V = 20 mV, I = 50 pA, 10 nm × 10 nm) of the Ga s We therefore suggest that each peak (or yellow stripe) nanostrip. The bright spots correspond to the Ga atoms in actually denotes a single flux quantum Φ , namely the thetoplayer. (c)Schematiccrystalstructureofγ-Ga. Along 0 vorticity L = 1 [Fig. S1]. the c(001) axis, each unit cell consists of four atomic layers. For clarity, the spheres in each layer are coded in different Secondly, we find that the magnetic vortex first pen- opacity. The brightest Ga spheres at the top layer form a etrates into the Ga nanostrip in a rather high filed, i.e. dimmer-likestructure,consistentwiththeSTMimagein(b). 0.435 T in Fig. 2(a). This is caused by the finite en- (d) Temperature-dependent dI/dV spectra on Ga nanostrip ergy barrier for vortex entry, consisting of the geometric (blackdots)andtheirbestfitstotheDynes’sexpression(red [28]andBean-Livingston(BL)surfacebarrier[29]. Both curves), with the broadening parameter Γ of 0.1 ∼ 0.2 meV. barriers conspire to preclude the penetration of vortices (e)Theextractedgapmagnitude∆versustemperature(black at low field. The first vortex penetration occurs only squares). The red line shows the fit to BCS gap function. when the surface superconducting currents exceed the pair-breakingcurrent,locallyweakeningsuperconductiv- ble at 2.5 K. At elevated temperature, the gap gradually ity and then allowing the nucleation of a vortex. reduces and both the coherence peaks are suppressed, as The anisotropic internal vortex structure constitutes expected. Thesuperconductinggapscanbefittedbyus- the major finding in this study: their large elonga- ing Dynes’s expression [19], as illustrated in Fig. 1(d). tion and the mechanism behind are unprecedented. For Figure 1(e) plots the extracted superconducting energy anisotropic vortex cores, such as sixfold star-shape [20], gap ∆ at various temperatures. Fitting the data to BCS fourfold [23–25] and twofold symmetry [26, 27, 30], to gap function yields T = 6.80 K and ∆(0) = 1.28 meV. name a few, seen previously in certain unconventional c Having identified the lattice structure and supercon- superconductors, the anisotropy in either the Fermi sur- ductivity, we then focus on the magnetic response of faceorthesuperconductinggaphasbeensuggestedtobe γ-Ga nanostrip. To visualize the magnetic vortices in the primary cause. However, this is not the case here: in real space, we map the spatial variation of zero bias con- γ-Ga the superconducting gap can be well fitted by the ductance (ZBC) on the dotted rectangle in Fig. 1(a), a isotropic BCS theory [Fig. 1(e)], and no gap anisotropy common approach for STM imaging of vortices [6–8, 20– isinvolved. Toshedmoreinsightintothepossiblemech- 27]. ThistechniquetakesadvantageoftheZBCcontrast anism behind, we have investigated the vortices in more within and outside the magnetic vortex cores, and has a Ga nanostructures, and find that the vortices invariably higher spatial resolution (∼ξ). Here, the suppressed su- elongate onceif thewidth ofGa nanostrips is reducedto perconductivitywithinthevortexcoresleadstoahigher below∼100nm. Incontrast,usualisotropiccircularvor- ZBC value, leaving superconductivity only outside the ticesdevelopinaroughlyround-shapedGaislandwitha vortex cores with lower ZBC. Thus the regions with en- largerlateraldimension[Fig.3(a)],whichalthoughshows hancedZBCindicatetheemergenceofmagneticvortices. the same lattice structure and superconductivity as Ga Figure2(a)illustratessuchmapsatvariousfieldsranging nanostrips. ShowninFigs.3(b-i)aretheisotropicvortex from0.3Tto1T.Twointriguingphenomenaareimme- cores, which bear strong similarities with those observed 3 (a) 0.3 T 0.435 T 0.44 T 0.45 T 0.5 T 0.6 T 0.7 T 0.8 T 0.9 T 1.0 T (b) a b L Φ/Φ0 14.4 8 13 7 11.6 6 10.1 1.0 4 8.7 C B Z 3 7.2 2 6.5 0.42 2 6.4 1 6.3 0 4.4 ba 0 10 20 30 40 50 0 100 200 300 Position (nm) Position (nm) FIG. 2. (color online) (a) Normalized ZBC maps at various magnetic fields, acquired in a field of view of 380 nm × 51 nm on theGananostripofFig.1(a). Thenormalizationwasperformedbydividingthesemapsbythenormal-stateZBCvalueatthe vortex core. Yellow regions with enhanced ZBC correspond to the vortex cores, all showing enormous elongation along the b axis. The small ZBC enhancement on the right side Ga nanostrip at 0.435 T probably indicates that we are probing a critical state justly before the penetration for the second vortex. (b) ZBC profiles across vortex centers in (a). The ZBC presents oscillationsalongthea axis,whileitshowsasingleextremumalongtheb axis. HereΦ/Φ andLdenotetheappliedmagnetic 0 flux and total vorticity in the strip, respectively. (a) (b) (c) (d) (e) (f) 70 nm 0.12 T 0.25 T 0.3 T 0.35 T 0.4 T (g) (h) (i) (j) 1 (k) 0.8 l/w Normalized σ000...864 σ(r)=σ0+(1−σ0)(1−tanh(r/2ξ)) hanced ZBC (arb. units)000...246 6281...915 n 0.45 T 0.5 T 0.8 T E 0 0 20 40 60 80 100 0 2 4 6 8 0.3 1.0 Radial distance r (nm) Voticity L FIG. 3. (color online) (a) STM topography (V = 3.0 V, I = 50 pA) of a larger Ga island (350 nm × 280 nm) with a height s of3.39nm(26ML).ThedottedsquaremarkstheregionoverwhichallZBCmapsin(b-i)areacquired. (b-i)MagneticField dependence of ZBC maps, acquired in a field of view of 140 nm × 140 nm in (a). The isotropic vortex cores are unchanged respective of the fields. (b) ZBC map (140 nm × 140 nm) taken on a round shaped Ga island with a height of 3.39 nm (26 ML), showing a highly isotropic vortex core. (j) Radial dependence of σ(r) across the isotropic vortex, which are color-coded to match ZBC map in (b). Here σ(r) has been normalized to optimize their fit to the inserted expression by minimizing the residual, as shown by the red curve. (k) Magnetic vortex-induced ZBC enhancement on various Ga nanostructures. in conventional superconductors such as Pb [22]. This expression for the superconducting order parameter ∆ completelyexcludestheanisotropyinintrinsicelectronic near the interface between a superconductor and a nor- structure of γ-Ga as a possible cause for the elongated mal metal, the radial ZBC profile across the vortex core vortices. Instead, the vortex elongates mainly via a geo- should obey [21] metrical confinement effect. √ σ(r)=σ +(1−σ )(1−tanh(−r/ 2ξ)), (1) ∞ ∞ Figure 3(j) plots the radial dependence of vortex- induced ZBC [or σ(r)]. Based on the Ginzburg-Landau 4 (a) (b) core cannot be ignored. Moreover, the nanostrip thick- Je Je 40 ness (several nanometers) is considerably smaller than B the penetration depth λ, vortices are of the Pearl rather Js minEinmeirzgayt ion Js pect ratio η2300 l/8w.5 tthhaenMAeibsrsinkeorsosvcretyenpieng[32cu].rrTenhtesvaotrttehxe isnatmerpalcetieodngsew(Jiteh) As10 62..91 becomes long ranged and prominent. In terms of the 1 counterpropagationbetweenJ andthescreeningcurrent 0 e b J encirclingthevortex(leftpanelinFig.4(a)),theinter- a 0.3 0.4 0.5 0.6 0.7 0.8 s B (T) actions are dominantly repulsive and pronounced along the a axis. To minimize this repulsive interaction, the FIG. 4. (color online) (a) Schematic of vortex axial elonga- vortex current J will deform or more specifically elon- tion. J and J denote the Meissner screening currents at s e s gate along the b axis until they are fully exploring the the sample edge and around the vortex, respectively. Black thickarrowslabeltherepulsiveforcebetweenthevortexand sample geometry (right panel in Fig. 4(a), less energeti- Meissner screening currents Je. (b) Magnetic field and γ-Ga cally costly), where the Je-Js repulsive interaction along geometrydependenceofvortexelongationη. Theextractedη the b axis gets stronger as well. Consequently, an equi- for l/w = 1.0, 2.1, 6.9 and 8.5 are coded magenta, red, black librium with the strong deformation in J and thus ZBC s and blue, respectively. or vortex cores is reached, as observed. FromconsideringJ -J repulsiveinteractions,onemay e s where σ(r) is the normalized ZBC away from the vortex further argue that vortex elongation η could enhance as core and r is the radial distance from the vortex center. theratiol/w andmagneticfieldincreases,becauseJ in- e A best fit of ZBC profile σ(r) to Eq. (1) yields a super- creaseswithincreasingfield. Inaddition,onceifmultiple conducting coherence length ξ = 24.6 ± 1.1 nm at 4.4 vortices enter into the nanostrips, the strong interaction K. Figure 3(k) further qualifies the vortex-induced ZBC between vortices will not only compress a certain vortex enhancementasafunctionofthevorticityLintheround along the a axis as well and contributes to enhance η, Ga island and rectangular Ga nanostrips, seen more vi- but also configure all vortices parallel to the b axis, as sually in Fig. S2. The linear relationship between them, observed. Figure 4(b) summarizes the vortex elongation guided by the dashed line, indicates the equivalence (ex- η as function of the field B and l/w of the γ-Ga nanos- ceptfortheshape)amongallobservedmagneticvortices, tructrues. Clearly, η increases abruptly with B and l/w. regardless of the geometry. For example, in the Ga nanostrip with the aspect ratio We now consider possible explanations for the elon- l/w of 8.5, η is as high as 40.5 at 0.7 T. Therefore, even gatedvorticesintheGananostrips. Firstly,asthewidth though further detailed theoretical investigations will be w of Ga nanostrips is comparable to their Fermi wave needed to understand the vortex elongation in a quanti- length λ , quantum confinement will become important tativeway,ourexperimentalfindingssupportthatthere- F and might lead to sizable spatial modulation in ∆ across pulsiveinteractionbetweenthevorticesandsampleedge the width of nanostrips[3], forming a single or a multi- screening current is a reasonable explanation of the vor- ple of weak superconducting thin slices normal to the a tex elongation in Ga nanostrips. axis. As a result, vortices, as locally quenched supercon- ductivity, will most probably reside inside the weak su- To summarize, our real space visualization of unusu- perconducting slices (more energetically favorable), and ally elongated vortices in γ-Ga nanostrips sheds light on may enormously elongate along the slices [27, 31]. How- Pearl vortex dynamics in the extremely thin and narrow ever, our Bogoliubov de-Gennues (BdG) self-consistent superconductingnanostrips(w <4ξ),wherethegeomet- calculations show that this explanation from the quan- rical confinement effect may deform the internal vortex tum confinement effect is applicable only if the Fermi core structure and then configure vortices in a brand- wave length λF is larger than about eight times of the new way. The elongated vortices observed here provide lattice constant. In a usual metal, however λF is com- anovelplatformtostudyexoticvorticesinnanoscalesu- monlycomparabletoitslatticeconstant(e.g.λF ∼0.4a0 perconductor. Fromtheviewpointofsuperconductorap- inγ-Ga)[1],asindicatedbyconsideringfreeelectronap- plications, the strong geometrical confinement could im- proximation. Therefore, the quantum confinement effect posedeeppotentialwellforthevortextransversemotion might not play the primary role in the vortex elongation andimmobilizethevortices,providingapromisingstrat- observed here. egytodesignnanoscalesuperconductingdevicesworking Alternatively,anotherpossibleoriginofelongatedvor- at high magnetic field, a long-standing dream of super- tices is from the geometric confinement effect imposed conducting research. by narrowness of the rectangular Ga nanostrips. In the Ga nanostrips studied here whose widths are of the or- ThisworkwassupportedbyNationalScienceFounda- der of ξ (e.g. w ∼2.5ξ in Fig. 2), the finite size of vortex tion and Ministry of Science and Technology of China. 5 [26] C. L. Song, Y. L. Wang, P. Cheng, Y. P. Jiang, W. Li, T.Zhang,Z.Li,K.He,L.L.Wang,J.F.Jia,H.H.Hung, C. J. Wu, X. C. Ma, X. Chen, and Q. K. Xue, Science ∗ [email protected] 332, 1410 (2011). † [email protected] [27] C. L. Song, Y. L. Wang, Y. P. Jiang, L. Wang, K. He, ‡ [email protected] X.Chen,J.E.Hoffman,X.C.Ma, andQ.K.Xue,Phys. [1] B. J. Baelus and F. M. Peeters, Phys. Rev. B 65, 104515 Rev. Lett. 109, 137004 (2012). (2002). [28] E. H. Brandt, Phys. Rev. B 59, 3369 (1999). [2] R. Geurts, M. V. Miloˇsevi´c, and F. M. Peeters, Phys. [29] C.P.BeanandJ.D.Livingston,Phys.Rev.Lett.12,14 Rev. Lett. 97, 137002 (2006). (1964). [3] L.F.Zhang,L.Covaci,M.V.Miloˇsevi´c,G.R.Berdiyorov, [30] Z. Y. Du, D. L. Fang, Z. Y. Wang, Y. F. Li, G. Du, and F. M. Peeters, Phys. Rev. Lett. 109, 107001 (2012). H. Yang, X. Y. Zhu, and H. H. Wen, Nat. Commun. 5, [4] G. Teniers, L. F. Chibotaru, A. Ceulemans, and V. V. 9408 (2015). Moshchalkov, Europhys. Lett. 63, 296 (2003). [31] C.Y.Liu,G.R.Berdiyorov, andM.V.Miloˇsevi´c,Phys. [5] R. C´ordoba, T. I. Baturina, J. Ses´e, A. Y. Mironov, Rev. B 83, 104524 (2011). J. M. De Teresa, M. R. Ibarra, D. A. Nasimov, A. K. [32] J. Pearl, App. Phys. Lett. 5, 65 (1964). Gutakovskii, A. V. Latyshev, I. Guillam´on, H. Suderow, S. Vieira, M. R. Baklanov, J. J. Palacios, and V. M. Vi- nokur, Nat. Commun. 4, 1437 (2013). [6] T. Cren, D. Fokin, F. Debontridder, V. Dubost, and SUPPLEMENTAL MATERIAL D. Roditchev, Phys. Rev. Lett. 102, 127005 (2009). [7] T. Cren, L. Serrier-Garcia, F. Debontridder, and Commercially purchased Si(111) wafers with a resis- D. Roditchev, Phys. Rev. Lett. 107, 097202 (2011). [8] T.Tominaga,T.Sakamoto,H.Kim,T.Nishio,T.Eguchi, tivity of 0.01 Ω·cm were used as substrate, and the clean and Y. Hasegawa, Phys. Rev. B 87, 195434 (2013). Si(111)-7 × 7 surfaces were obtained by flash annealing [9] A.Johansson,G.Sambandamurthy,D.Shahar,N.Jacob- to ∼ 1200oC while keeping the vacuum better than 1 × son, and R. Tenne, Phys. Rev. Lett. 95, 116805 (2005). 10−9 Torr. We then prepared Ga nanoislands by evap- [10] T.Morgan-Wall,B.Leith,N.Hartman,A.Rahman, and orating Ga (99.999%) sources from a standard Knudsen N. Markovi´c, Phys. Rev. Lett. 114, 077002 (2015). cell at a nominal beam flux of approximately 0.4 mono- [11] D. Pekker, G. Refael, and P. M. Goldbart, Phys. Rev. layer (ML, ∼ 0.13 nm)/min, during which the Si(111)- Lett. 107, 017002 (2011). [12] P. Sa´nchez-Lotero and J. J. Palacios, Phys. Rev. B 75, 7 × 7 substrate was kept at 110 K. Subsequent post- 214505 (2007). annealing at a slightly higher temperature of 150 K led [13] E. Bronson, M. P. Gelfand, and S. B. Field, Phys. Rev. to many Ga nanostructures with varying geometries and B 73, 144501 (2006). the lateral dimensions [1]. Analogous to MBE growth of [14] J. J. Palacios, Phys. Rev. B 57, 10873 (1998). Pb on Si(111)-7 × 7 [2], the substrate just beneath Ga [15] R. B. G. Kramer, G. W. Ataklti, V. V. Moshchalkov, islands consists of a slightly disordered Ga wetting layer and A. V. Silhanek, Phys. Rev. B 81, 144508 (2010). (1∼2 ML), which exhibits no superconductivity down to [1] H. M. Zhang, J. P. Peng, J. Q. Guan, Z. Li, C. L. Song, L. Wang, K. He, X. C. Ma, and Q. K. Xue, Sci. China the minimum measurement temperature of 2.2 K. 58, 107402 (2015). [17] Materials and Method are available as supporting mate- rial. 8 A=48847 nm2, Fig. 3(a) A=45491 nm2, nanostrip [18] L. Bosio, H. Curien, M. Dupont, and A. Rimsky, Acta 7 A=35052 nm2, Fig. 1(a) Crystal. Sect. B 29, 367 (1973). 6 [19] R.C.Dynes,V.Narayanamurti, andJ.P.Garno,Phys. L y 5 [20]RNev..HLaeytat.sh4i1,,M15.0I9ch(i1o9k7a8,).and K. Machida, Phys. Rev. orticit 4 V 3 Lett. 77, 4074 (1996). 2 [21] N.Bergeal,V.Dubost,Y.Noat,W.Sacks,D.Roditchev, N. Emery, C. H´erold, J.-F. Marˆech´e, P. Lagrange, and 1 G. Loupias, Phys. Rev. Lett. 97, 077003 (2006). 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 [22] Y.X.Ning,C.L.Song,Z.L.Guan,X.C.Ma,X.Chen, B (T) J. F. Jia, and Q. K. Xue, Europhys. Lett. 85, 27004 (2009). FIG. S1. Vorticity L plotted as a function of the applied [23] H. Nishimori, K. Uchiyama, S. Kaneko, A. Tokura, fieldB forthreeGananaoislandswithvaryinggeometryand H.Takeya,K.Hirata, andN.Nishida,J.Phys.Soc.Jpn. lateral area A. 73, 3247 (2004). [24] T.Hanaguri,K.Kitagawa,K.Matsubayashi,Y.Mazaki, FigureS1depictstheevolutionofvorticityL(thenum- Y. Uwatoko, and H. Takagi, Phys. Rev. B 85, 214505 (2012). ber of individual isolated stripy or round yellow features [25] B. B. Zhou, S. Misra, E. H. da Silva Neto, P. Aynajian, in ZBC maps) with varying magnetic field B. Above a R. E. Baumbach, J. D. Thompson, E. D. Bauer, and threshold field B , depending on A or the width W of m A. Yazdani, Nat. Phys. 9, 474 (2013). superconducting strips, L increases almost linearly with 6 B, as anticipated for the gradual penetration of mag- neticvorticesintoGaislands[3]. Byextractingtheslope (dashedlines)andislandareaA,weestimatethateachof ∗ [email protected] such features accommodate the magnetic flux of 1.18Φ 0 † [email protected] (left triangle), 1.25Φ0 (upper triangle) and 1.47Φ0 (cir- ‡ [email protected] cle),respectively,closetoasinglemagneticfluxquantum [1] H. M. Zhang, J. P. Peng, J. Q. Guan, Z. Li, C. L. Song, Φ and never exceeding 2Φ . This strongly supports a L. Wang, K. He, X. C. Ma, and Q. K. Xue, Sci. China 0 0 single magnetic vortex nature of the observed stripy or 58, 107402 (2015). round features in Figs. 2 and 3. The small discrepancy [2] J. F. Jia, S. C. Li, Y. F. Zhang, andQ. K. Xue, J. Phys. Soci. Jpn. 76, 082001 (2007). from Φ most possibly originates from an overestimate 0 [3] G.Stan,S.B.Field, andJ.M.Martinis,Phys.Rev.Lett. of the effective island area A for nanoscale superconduc- 92, 097003 (2004). tors. Compared to Nb strips previously reported [3], B m appears larger in Ga strips we studied here, due to their extremely small dimensions. (a) (b) 35 35 30 30 dI/dV (arb. units)221505 dI/dV (arb. units)221505 10 10 5 5 0 0 -10 -5 0 5 10 -10 -5 0 5 10 Sample bias (mV) Sample bias (mV) FIG. S2. Differential conductance dI/dV spectra straddling (a) a round magnetic vortex core (Fig. 3, 0.15 T) and (b) an elongatedvortexcore(Fig.2,0.435T),respectively. Inserted are the two vortices, with the dashed lines indicting where the dI/dV spectra are acquired. As anticipated, the super- conductivity is significantly suppressed with enhanced ZBC when approaching the vortex cores, regardless of their inter- nal core structure.The small E -near conductance reduction F in (b) is primarily caused by a large (∼ 12 nm) spatial in- terval between the neighboring dI/dV spectra, which fails to acquire the spectrum justly at the vortex core center.