Direct Mapping of Spin and Orbital Entangled Wavefucntion under Interband Spin-Orbit coupling of Rashba Spin-Split Surface States Ryo Noguchi,1 Kenta Kuroda,1,∗ K. Yaji,1 K. Kobayashi,2 M. Sakano,1 A. Harasawa,1 Takeshi Kondo,1 F. Komori,1 and S. Shin1 1Institute for Solid State Physics, University of Tokyo, Kashiwa, Chiba 277-8581, Japan 2Department of Physics, Ochanomizu University, Bunkyo-ku, Tokyo 112-8610, Japan (Dated: January 26, 2017) We use spin- and angle-resolved photoemission spectroscopy (SARPES) combined with polarization-variable laser and investigate the spin-orbit coupling effect under interband hybridiza- 7 tion of Rashba spin-split states for the surface alloys Bi/Ag(111) and Bi/Cu(111). In addition to 1 theconventionalbandmappingofphotoemissionforRashbaspin-splitting,thedifferentorbitaland 0 spin parts of the surface wavefucntion are directly imaged into energy-momentum space. It is un- 2 ambiguouslyrevealedthattheinterbandspin-orbitcouplingmodifiesthespinandorbitalcharacter of the Rashba surface states leading to the enriched spin-orbital entanglement and the pronounced n momentum dependence of the spin-polarization. The hybridization thus strongly deviates the spin a J and orbital characters from the standard Rashba model. The complex spin texture under inter- band spin-orbit hybridyzation proposed by first-principles calculation is experimentally unraveled 5 by SARPES with a combination of p- and s-polarized light. 2 PACSnumbers: ] i c s Arealizationoffunctionalcapabilitiestogeneratespin- systemBiCu /Cu(111)asshowninFig.1(b)[14,20,21]. - 2 l splitting of electronic states without any external mag- Whilethepresenceofthespin-polarizedelectronicbands r t netic field is a key subject in the research of spintron- has been demonstrated by spin and angle-resolved pho- m ics [1]. A promising strategy exploits the influence of toemission spectroscopy (SARPES) [21–23] and inverse- . t spin-orbit (SO) interaction that can give rise to the lift- SARPES [8], the SO entanglement and particularly the a ingofspindegeneracyunderbrokenspaceinversionsym- spintextureduetotheinterbandSOcouplingisstillun- m metry, the so-called Rashba effect [2]. In the conven- derdiscussion, becauseofthelackoftheorbitalselectiv- - d tional Rashba model, an eigenstate of the SO-induced ityinthepreviousexperiments. Uptonow, thecomplex n spin-splitting is treated with an assumption of a pure spintextureofthesesurfacestateswasonlyindirectlyde- o spin state fully chiral spin-polarized which protects elec- tected by quantum interference mapping through scan- c trons from backscattering [2–5]. However, in real mate- ning tunneling spectroscopy [9, 24, 25]. Thus, no con- [ rials, the assumption can be usually broken because the clusiveunderstandingofthephenomenonapartfromthe 1 SO coupling mixes different states with different orbitals conventional Rashba model has been achieved yet. v and orthogonal spinors in a quasiparticle eigenstate [6– 9 In this Rapid Communication, we directly investi- 8]. The SO entanglement can permit the spin-flip elec- 6 gate the SO entanglement in the Rashba surface states 2 tron backscattering [9] and moreover orbital mixing in of BiAg /Ag(111) surface alloy by using a combina- 2 7 theeigenstatecanplayasignificantroleinanemergence tion of polarization variable laser with SARPES (laser- 0 of the large spin-splitting [10–14]. Therefore, it is essen- SARPES) and compare to those of BiCu /Cu(111) as . 2 1 tially important to experimentally explore the SO cou- a simple case. In contrast to the previous experi- 0 pling not only in the lifting spin degeneracy but also in ments [8, 21–23], our laser-SARPES deconvolves the or- 7 the spin and orbital wavefunction as eigenstates. bital wavefucntion and the coupled spin, and the surface 1 : wavefunctionsaredirectlyimagedintomomentum-space v BeyondtheconventionalRashbamodel,awell-ordered throughorbital-selection rule. Itis shownthat theinter- i surface alloy BiAg grown on Ag(111) provides an ideal X 2 bandSOcouplingmodifiesthespinandorbitalcharacter case to study the SO entanglement in Rashba surface r of the Rashba surface states leading to the spin-orbital states. In the surface alloy, an occupied sp -like band a z entanglementandthek dependence. Theresultingspin and a mostly unoccupied p -like band show significant || xy texture thus shows a large deviation from the conven- Rashbaspinsplitting[15,16]andcrosseachotheratthe tional Rashba model. The full spin information is ex- specific k [17–19] as shown in Fig. 1 (b). In particular, || perimentally unraveled only by a combination of p- and densityfunctionaltheory(DFT)calculationsshowedthe s-polarized light in accordance with a view of the SO strong SO entanglement [8, 9] and predicted the com- entanglement. plex spin texture that is significantly different from the conventional Rashba model; the sp band switches spin- Thelaser-SARPESmeasurementwasperformedatthe z polarization at the crossing through SO-induced inter- Institute for Solid State Physics (ISSP), the Univesity of bandhybridization[19]whichisincontrasttothesimilar Tokyo, with a high-flux 6.994-eV laser and ScientaOmi- 2 (a) light Dete=cMtiiornro pr lpalnaene photoelectrons Surface Brillouin zone ptalel etneemrgpyera(atunrgeulawra)srekseopltutaiotn∼s 1o5f tKh.e Tsehteupinisstr2ummeenV- p-pol. s-pol. (0.3◦)and20meV(0.7◦)forspin-integratedARPESand Spin-up z 50 Γ Κ SARPES,respectively. TheBiAg2 andBiCu2 surfaceal- y x Crystal surface ky kx Spin-down lloityesrawtuerreeso[1b4ta,i1n5e]d. Lboywt-hene√eprgroyce√eldeuctrreosnpdrieffseranctteidoninmtehae- (b) BiAg pxy BiCu pxy surements confirmed the ( 3× 3)R30◦ reconstruction 2 2 of the surface alloys. E E E F F First-principles calculations were performed using the k sp sp VASP code [27]. The projector augmented wave (c) 0BiAg2/Ag(111) z pxy 0 BiCu2/Cu(111)z spz pxy method [28] is used in the plane-wave calculation. The generalized gradient approximation by Perdew, Burke, eV) -0.3 -0.2 High and Ernzerhof [29] is used for the exchange-correlation ( EF potential. The spin-orbit interaction is included. The E- -0.6 -0.4 spz atom positions of BiAg2 are optimized. Those of BiCu2 -0.9 -0.2 -0.1 0 0.1 0.2 -0.6-0.2 -0.1 0 0.1 0.2 0.3 Low are taken from the experimental data of Ref. [30]. kx (Å-1) kx (Å-1) Let us start with showing a brief overview of observed (d) Low High Low High Odd Even electronicstructureofBiAg andBiCu surfacealloysin sp p sp p sp p 2 2 z xy z xy z xy Fig. 1(c). The sp -derived bands and the higher-lying 0.0 z V) p bands disperse downwards in energy with a large (eEF-0.2 Rxayshba spin-splitting in the both materials. Compared E--0.4 withthesurfacebandsinBiAg ,themostpartofthesur- 2 -0.6 BiAg2 BiAg2 BiAg2 facebandsinBiCu2 isabovetheFermilevel(EF). These 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 results are in good agreement with previous works [14– (e) sp p sp p sp p 17]. z xy z xy z xy 0 Considering the orbital selection rule in the dipole ex- (eV)-0.2 citation[31],(cid:15)p and(cid:15)s enableustodrawdifferentorbital E-EF-0.4 spyomsemdeotfryB.i 6Ssinacnedt6hpeosrubriftaacles [s1t9a]t,e(cid:15)s nseealerctEivFelayrdeecteocmts- p BiCu BiCu BiCu -0.6 2 2 2 weight of even-parity orbital with respect to the mirror 0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3 kx (Å-1) kx (Å-1) kx (Å-1) plane, mainly from s, pz and px components, while (cid:15)s is sensitive to the odd-parity orbital mainly derived from FIG.1: (Coloronline)(a)Theexperimentalconfigurationfor p . y p- or s-polarization with an angle incidence of 50 ◦. A mir- Figure 1(d) summarizes the linear polarization depen- ror plane of the surface coincides with the plane of incidence dence in BiAg . For the result obtained by (cid:15) [see the (x-z plane). Surface Brillonin zone of the surface alloys [un- 2 p derlyingfcc(111)substrateofAg(111)andCu(111)]isshown left panel], we observe the strong intensity for the spz by solid (dashed) line. (b) Schematic of the energy disper- and inner p bands. The data particularly displays xy sion of Rashba spin-split bands in (left) BiAg2 and (right) the band crossing of the outer spz and inner pxy bands BiCu2 surface alloys. (c) ARPES intensity maps for BiAg2 aroundk =0.15˚A−1[Fig.1(d)],wherethespectralinten- andBiCu surfacealloyswithp-polarizationalonghighsym- || metry Γ¯-K¯2 line of the surface Brillouin zone. The dashed sityoftheouterspz isstronglysuppressed. Surprisingly, lines indicate the edge of the projected bulk bands. (d) and switching the light polarization (cid:15)p to (cid:15)s, the spectral in- (e) The magnified ARPES intensity maps with (left) p- and tensity is dramatically changed [see the middle panel in (middle) s-polarization, and (right) the differential intensity Fig. 1(d)]. The dispersion of the outer sp is clearly ob- z maps, which are obtained by Ip−Is where Ip and Is are the servedevenatlagerk >0.15˚A−1 togetherwiththeouter photoelectron intensity obtained by p- and s-polarized light || p band. Consequently, the overall parabolic dispersion without normalization, respectively. xy of the outer sp band is clearly seen, which was absent z in previous experiments [14, 15, 22]. The right panel of Fig. 1(d) shows the differential intensity map [see figure cron DA30L photoelectron analyzer [26]. The experi- captionofFig.1]. Thered-bluecolorcontrastreflectsthe mental configuration is shown in Fig. 1(a). The p- and contribution of the even- and odd-orbital components in s-polarizations ((cid:15) and (cid:15) , respectively) were used in the the surface wavefunction. It is immediately found that p s experiment. The photoelectrons were detected along Γ¯- theorbitalcharacterofspz bandchangestheorbitalchar- K¯ line of the surface Brillouin zone. The spectrometer acter at the band crossing. resolved the spin component along y, which is perpen- In contrast, the result for BiCu is found to be simple 2 dicular to the mirror plane of the surface. The sam- [see Fig. 1(e)]: the sp and p bands comprise mainly z xy 3 BiAg BiCu 2 2 (a) p-polarization (b) s-polarization (c) p-polarization (d) s-polarization kx (Å-1) kx (Å-1) nits) -0.03 nits) 0.07 b. u 0.00 b. u 0.10 ar 0.03 ar 0.13 y ( y ( sit 0.06 sit 0.16 n n nte 0.09 nte 0.19 d i 0.12 d i 0.22 e e v v ol 0.15 ol 0.25 s s e e n-r 0.18 n-r 0.28 pi pi S 0.21 S 0.31 0.24 0.34 -0.6 -0.4 -0.2 0 -0.6 -0.4 -0.2 0 -0.6 -0.4 -0.2 0 -0.6 -0.4 -0.2 0 E-E (eV) E-E (eV) E-E (eV) E-E (eV) F F F F (e) p-polarization (f) s-polarization (g) p-polarization (h) s-polarization Sy Sy V) 0 1 0 1 (e-0.2 V) E-EF-0.4 0 E(eF-0.2 0 E--0.4 -0.6 -1 -1 Low High Low High -0.6 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3 -1 -1 -1 -1 kx (Å ) kx (Å ) kx (Å ) kx (Å ) FIG. 2: (Color online) (a)-(d) SARPES spectra with spin quantum axis along y for BiAg and BiCu by using (cid:15) and (cid:15) . 2 2 p s Spin-upandspin-downspectraareplottedwithredandbluelines. Thepeakpositionsinthespin-resolvedspectraforsp and z p bands are indicated by open and closed triangles, respectively. (e)-(h) The corresponding spin-polarization and intensity xy maps with the two-dimensional color codes [32]. The dashed lines indicate the edge of the projected bulk bands. even- and odd-parity orbitals, respectively. These two (cid:15) and (cid:15) . Figures 2(a)-(d) show SARPES spectra ob- p s bands in BiCu are separated in momentum-space away tained by using (cid:15) and (cid:15) for BiAg and BiCu . The 2 p s 2 2 from the band crossing. Nevertheless, we see the reduc- corresponding spin-polarization and intensity maps [32] tion of the spectral intensity of the outer sp band when are shown in Figs. 2(e)-(h). For (cid:15) , the inner and outer z p itoverlapswiththeprojectedbulk-band,asobservedalso sp -bands around Γ¯ point in the both materials show z in the even-part of the outer sp band in BiAg [see the negative and positive spin-polarization, respectively, dis- z 2 left panel in Fig. 1(d)]. This common feature suggests playingaconventionalRashba-typespin-texture[22,23]. the interaction of the the outer sp band with the bulk Theobservedspinpolarizationisfoundtobelargeupto z sp projection from the substrate [14, 16] modifies the nearly 80 and 60 % for BiAg and BiCu , respectively. z 2 2 spectral weight of the orbital wavefunction particularly Most remarkably, we find that the sign of the spin for the even-orbitals. polarization sensitively depends on the linear polariza- Apparently, the significant k dependence of the or- tion [33]. This can be seen in the SARPES spectra par- || bital symmetry is unique for the outer spz of BiAg2. ticularly for kx=0.12 ˚A−1 in BiAg2. The spectral weight ThisresultindicatesthepresenceoftheinterbandSOhy- of the spin-up is considerably larger than the spin-down bridization that allows to mix the even- and odd-orbital for (cid:15) and achieves nearly +80 % spin-polarization. For p components in the surface wavefunction of the outer spz (cid:15)s,theintensityrelationturnstotheoppositeandthere- band. Previously,ithasbeenbelievedthatthehybridiza- sultingspin-polarizationisfoundtobe−70%. Sincethe tion associates with gap opening [17–19, 25]. However, linearpolarizationissensitivetodifferentorbitalsymme- in our laser-SARPES experiment, there is no clear gap try, our laser-SARPES umambiguously reveals that the observed around the crossing point. spin direction strongly depends on the orbital character. To get further insight into the influence of the inter- Figures 3(a) and (b) represent the calculated spin tex- band SO coupling, we carried out laser-SARPES mea- ture for the both materials, which is consistent with pre- surements collaborated with the orbital selection rule of vious theoretical results [14, 19, 20]. In BiAg , the hy- 2 4 (a) (c) (a) (b) Even Odd p-pol.+s-pol. p-pol.+s-pol. Spin-texture BiAg BiCu V) 0 0 0 2 0 2 Sy (e-0.3 -0.3 V) 1 F e-0.2 -0.2 E ( -E--00..96BiAg2 --00..96 BiAg2 E-EF-0.4 -0.4 0 0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3 -1 -1 -0.6 -1 kx (Å ) kx (Å ) Low High -0.6 0 0.1 0.2 0.3 0.1 0.2 0.3 (b) (d) -1 -1 Spin-texture Even Odd kx (Å ) kx (Å ) 0.8 0.8 V) 0.4 0.4 FIG. 4: (Color online) (a) and (b) Experimentally obtained e ( orbital-integrated spin-texture of the Rashba surface states F 0 0 -E-0.4 -0.4 in BiAg2 and BiCu2, which is in good agreement with that E theoreticallypredictedasshowninFigures3(a)and(b). The BiCu BiCu -0.8 2 -0.8 2 color arrangement indicates the total photoelecton intensity 0 0.2 0.4 0 0.2 0.4 0 0.2 0.4 and spin-polarization [32]. -1 -1 kx (Å ) kx (Å ) FIG. 3: (Color online) (Color online) (a) and (b) Calculated InBiAg [Figs.3(a)and(c)],thespin-polarizationcou- totalspintextureforBiAg andBiCu surfacealloys,respec- 2 2 2 tively. (c) and (d) The coupled spin textures decomposed pledtotheeven-orbitalcomponentpredominatesthespz into (left) even- and (right) odd-orbital characters. The size state around the Γ¯ point, and finally the opposite spin of the circles is proportional to the total spin polarizations. coupledtotheodd-orbitalcomponentbecomesdominant Theredandbluecolorindicatespin-upandspin-downquan- at higher k . This indicates that the weight of the even- || tized along y, respectively. and odd-orbital components in the surface wavefunction play a significant role in the total spin texture through the SO entanglement [Fig. 3(a)]. Our experiment indeed bridization of the sp and p bands is found in a gap z xy demonstratesthesignificantk -dependenceoftheorbital opening at the band crossing where the spin polariza- || wavefunction [Fig. 1(d)], which shows a good agreement tion changes its sign [Fig. 3(a)]. Due to the hybridiza- with DFT calculation [42]. Therefore, the hybridization tion, one may not assign whether the two branches at through the interband SO coupling modifies the orbital thegappedpointoriginatefromeitherthesp orthep z xy component and induces the SO entanglement in the sp , derived states. Nevertheless let us refer the sp and p z z xy which considerably deviates the spin texture from the bands,sinceourexperimentalresultshowsthehybridiza- conventional Rashba model. tion avoids the gap opening. Apparently, the mapping of the spin in our experi- The complex spin textures are decomposed into the ment [Fig. 2(e)-(h)] does not reflect the predicted spin even and odd orbital contributions in Figs. 3(c) and (d). texture [Figs. 3(a) and (b)]. This is because photomie- TheseresultsclearlyshowtheSOentanglementinwhich sionmeasurementbyusinglinearlypolarizedlightinour the different orbital components are coupled with op- experimental set-up selects the specific orbital symme- posite spin. The calculated SO entangled texture re- try [8]. Indeed, the orbital-dependent spin texture in produces our experimental results for the spin mapping theory [Figs. 3(c) and (d)] shows good agreement with of the surface wavefunction (see Fig. 2). By general our laser-SARPES results for (cid:15) and (cid:15) . group-theoretical analysis for the mirror symmetry [6], p s We now show that the total spin information can be the wavefucntion under SO coupling is generally repre- tracedbackonlybyusingacombinationofthespinmap- sented as: ping with (cid:15) and (cid:15) lasers. Owing to selective detection p s |Ψ >=|even,↑(↓)>+|odd,↓(↑)>, (1) of the pure orbital-symmetry in our experimental set-up ±i of Fig. 1(b), the orbital-dependence in the spin polar- wherethespinors|↑>,|↓>arequantizedalongy,which ization is eliminated by integrations of spin-polarization is perpendicular to the mirror plane, and the index ±i maps (P ) in Figs. 4(a) and (b) as follows: total is representation for the mirror symmetry. This explains not only that the even- and odd-parity orbitals couple (I +I )−(I +I ) P = ↑,p ↑,s ↓,p ↓,s , (2) with opposite spins but also that the SO entanglement total (I +I )+(I +I ) ↑,p ↑,s ↓,p ↓,s is a general consequence of the SO coupling. Indeed the similar SO-coupled states have been recently con- whereI (I )andI (I )indicatesthespin-upand ↑,p ↑,s ↓,p ↓,s firmed in surface states of topological insulators [34–39] spin-down intensities obtained by (cid:15) ((cid:15) ). The mapping p s and Rashba states in BiTeI [40, 41]. of P clearly demonstrates the complex spin texture total 5 of the sp band under interband hybridization, which [2] Y. A. Bychkov and E. I. Rashba, JETP Lett. 39, 78 z is obviously comparable to the theoretical predictions in (1984). Figs. 3(a) and (b). Since the SO entanglement is gen- [3] S. LaShell, B. A. McDougall and E. Jensen, Phys. Rev. Lett. 77, 3419 (1996). erally expected in materials as long as the SO coupling [4] G. Nicolay, F. Reinert, S. Hu¨fner and P. Blaha, Phys. plays a significant role, this technique therefore demon- Rev. B 65, 033407 (2001). strates a general advantage to investigate the unconven- [5] M.Hoesch,M.Muntwiler,V.N.Petrov,M.Hengsberger, tionalspintexturesinstrongSO-coupledstatesalthough L. Patthey, M. Shi, M. Falub, T. Greber and J. Oster- the results could be infuenced by the cross-section for (cid:15) walder, Phys. Rev. B 69, 241401 (2004). p and (cid:15) and require the photoemission calculation to con- [6] J.Henk,A.Ernst,andP.Bruno,Phys.Rev.B68,165416 s sider photon energy dependence [43, 44]. (2003). [7] K. Miyamoto, H. Wortelen, H. Mirhosseini, T. Okuda, The question remains as to why a hybridization gap is A. Kimura, H. Iwasawa, K. Shimada, J. Henk, and absent in our experiment while the calculation predicts M. Donath, Phys. Rev. B 93, 161403 (2016). thegap. Weattributetheabsenceofthegaptoaninter- [8] S.N.P.Wissing,A.B.Schmidt,H.Mirhosseini,J.Henk, action of the confined surface states with the electronic C.R.AstandM.Donath,Phys.Rev.Lett.113,116402 stateinthesubstrate. Itwasrecentlyshownthatthesize (2014). [9] S.Schirone,E.E.Krasovskii,G.Bihlmayer,R.Piquerel, ofthehybridizationgapsensitivelydependsonthethick- P. Gambardella and A. Mugarza, Phys. Rev. Lett. 114, nessofAg(111)quantumwellanddecreaseswithincreas- 166801 (2015). ing substrate thickness [17]. Hence, one can expect the [10] M.Nagano,A.Kodama,T.ShishidouandT.Oguchi,J. gapabsenceinthecaseofthebulksubstratewithcontin- Phys.: Condens. Matter 21, 064239 (2009). uum electronic states. The similar hybridization recon- [11] H. Ishida, Phys. Rev. B 90, 235422 (2014). structed by the bulk interaction is reported in image po- [12] S.-R.Park,C.-Y.Kim,J.ElectronSpectrosc.Rel.Phen. tential resonances [45]. This fact indicates that the pres- 201, 6 (2015). ence/absence of the gap does no give a direct evidence [13] I. Gierz, B. Stadtm´’uller, J. Vuorinen, M. Lindroos, F. Meier, J. H. Dil, K. Kern, and C. R. Ast, Phys. Rev. of the interband SO-coupling but the pronounced recon- B 81, 245430 (2010). struction of the surface wavefunction directly displays. [14] H. Bentmann, F. Forster, G. Bihlmayer, E. V. Chulkov, In particular, the knowledge of the interband SO cou- L.Moreschini,M.GrioniandF.Reinert,Europhys.Lett. pling is critically important for emergence of Dirac and 87, 37003 (2009). Weylfermionsinsemiconductorsandsemimetals[46–51], [15] C.R.Ast,J.Henk,A.Ernst,L.Moreschini,M.C.Falub, related to non-trivial band topology. D.Pacile,P.Bruno,K.Kern,andM.Grioni,Phys.Rev. Lett. 98,186807 (2007). In conclusion, we have deconvolved the spin and or- [16] K.He,T.Hirahara,T.Okuda,S.Hasegawa,A.Kakizaki, bital wavefuntion of the Rashba spin-split surface states and I. Matsuda, Phys. Rev. Lett. 101, 107604 (2008). for the Bi-based surface alloys, and directly mapped [17] H.Bentmann,S.Abdelouahed,M.Mulazzi,J.Henk,and these wavefucntions into momentum-space by combin- F. Reinert, Phys. Rev. Lett. 108, 196801 (2012). ing orbital-selective laser-SARPES and first principles [18] G. Bian, X. Wang, T. Miller and T. C. Chiang, Phys. calculations. The interband SO hybridization strongly Rev. B 88, 085427 (2013). [19] G. Bihlmayer, S. Blugel and E. V. Chulkov, Phys. Rev. influence the spin and orbital character in the surface B 75, 195414 (2007). wavefunctionleadingtothek dependenceoftheSOen- || [20] H.Mirhosseini,J.Henk,A.Ernst,S.Ostanin,C.-T.Chi- tanglement. Theresultingspintexturethusconsiderably ang,P.Yu,A.Winkelmann,andJ.Kirschner,Phys.Rev. deviates from the conventional Rashba model. Although B 79, 245428 (2009). themeasuredspintexturebyusing(cid:15) or(cid:15) doesnotgive [21] H. Bentmann, T. Kuzumaki, G. Bihlmayer, S. Blugel, p s thefullspininformationinthiscase,thefullspin-texture E. V. Chulkov, F. Reinert and K. Sakamoto, Phys. Rev. is experimentally unraveled by SARPES with a combi- B 84, 115426 (2011). [22] F. Meier, H. Dil, J. Lobo-Checa, L. Patthey and J. Os- nation of the both linear polarization. Our findings can terwalder, Phys. Rev. B 77, 165431 (2008). be widely applied for clarifying the complex spin infor- [23] K. He, Y. Takeichi, M. Ogawa, T. Okuda, P. Moras, mation in the SO-entangled surface states. D. Topwal, A. Harasawa, T. Hirahara, C. Carbone, We gratefully acknowledge funding JSPS Grant-in- A. Kakizaki, and I. Matsuda, Phys. Rev. Lett. 104, Aid for Scientific Research (B) through Projects No. 156805 (2010). [24] H. Hirayama, Y. Aoki, and C. Kato, Phys. Rev. Lett. 26287061 and for Young Scientists (B) through Projects 107, 027204 (2011). No. 15K17675. [25] L.E.El-Kareh,P.Sessi,T.Bathon,andM.Bode,Phys. Rev. Lett. 110, 176803 (2013). [26] K. Yaji, A. Harasawa, K. Kuroda, S. Toyohisa, M. Nakayama, Y. Ishida, A. Fukushima, S. Watanabe, C. Chen, F. Komori, S. Shin, Rev. Sci. Instrum. 87, ∗ [email protected] 053111 (2016). [1] A. Mabchon, H. C. Koo, J. Nitta, S. M. Frolov, and [27] G. Kresse and J. Furthm´’uller, Phys. Rev. B 54, 11169 R. A.Duine, Nature Mat. 20, 871 (2015). 6 (1996). Y. Ishida, S. Watanabe, C.-T. Chen, T. Kondo, F. Ko- [28] P. E. Bl´’ochl, Phys. Rev. B 50, 17953 (1994). mori, and S. Shin, Phys. Rev. B 94, 165162 (2016). [29] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. [40] L. Bawden, J. M. Riley, C. H. Kim, R. Sankar, Lett. 77, 3865 (1996). E. J. Monkman, D. E. Shai, H. I. Wei, E. B. Lochocki, [30] D. Kaminski, P. Poodt, E. Aret, N. Radenovic, and J. W. Wells, W. Meevasana, T. K. Kim, M. Hoesch, E. Vlieg, Surf. Sci. 575, 233 (2005). Y. Ohtsubo, P. L. Fevre, C. J. Fennie, K. M. Shen, [31] A. Damascelli, Z. Hussain, Z. X. Shen, Rev. Mod. Phys. F. Chou, P. D. C. King, Sci. Adv. 1, e1500495 (2015). 75, 473 (2003). [41] H.Maaß,H.Bentmann,C.Seibel,C.Tusche,S.V.Ere- [32] C. Tusche, A. Krasyuk, J. Kirschner, Ultramicroscopy, meev,T.R.F.Peixoto,O.E.Tereshchenko,K.A.Kokh, 159, 520 (2015). E. V. Chulkov, J. Kirschner and F. Reinert, Nat. Com- [33] The inner sp band in BiCu does not show the orbital- mun. 7, 1162 (2016). z 2 dependence in Figs. 2(g) and (h), which is conflict with [42] SeeSupplementalMaterialat[http://...]formoredetails. the theoretical result in Fig. 3(d). We expect that this [43] Ulrich Heinzmann and J. Hugo Dil, J. Physics: Cond. may indicate an interaction with the bulk projection Mat. 24, 173001 (2012). bands.However,itisstronglyrequiredfurthertheoretical [44] E. E. Krasovskii, J. Phys. Cond. Matt. 27, 49 (2015). supports to fully explain this behavior. [45] M.Winter,E.V.Chulkov,andU.Ho¨fer,Phys.Rev.Lett. [34] H. Zhang, C.-X. Liu, and S.-C. Zhang, Phys. Rev. Lett. 107, 236801 (2011). 111, 066801 (2013). [46] Z. H. Hasan, C. L. Kane, Rev. Mod. Phys. 82, 82, 3045 [35] Z.-H.Zhu,C.N.Veenstra,G.Levy,A.Ubaldini,P.Syers, (2010). N.P.Butch,J.Paglione,M.W.Haverkort,I.S. Elfimov, [47] Y. Ando, J. Phys. Soc. Jpn. 82, 102001 (2013). andA.Damascelli,Phys.Rev.Lett.110,216401(2013). [48] X. Wan, A. M. Turner, A Vishwanath, and [36] Y.Cao,J.A.Waugh,X.-W.Zhang,J.-W.Luo,Q.Wang, S. Y. Savrasov, Phys. Rev. B 83, 205101 (2011). T. J. Reber, S. K. Mo, Z. Xu, A. Yang, J. Schneeloch, [49] A. A. Burkov and L. Balents, Phys. Rev. Lett. 107, G. Gu, M. Brahlek, N. Bansal, S. Oh, A. Zunger, and 127205 (2011) D. S. Dessau, Nat. Phys. 9, 499 (2013). [50] S. Y. Xu, I. Belopolski, N. Alidoust, M. Neupane, [37] Z.-H. Zhu, C. N. Veenstra, S. Zhdanovich, M. P. Schnei- G. Bian, C. Zhang, R. Sankar, G. Chang, Z. Yuan, der, T. Okuda, K. Miyamoto, S.-Y. Zhu, H. Namatame, C.C.Lee,S.M.Huang,H.Zheng,J.Ma,D.S.Sanchez, M. Taniguchi, M. W. Haverkort, I. S. Elfimov, and B. Wang, A. Bansil, F. Chou, P. P. Shibayev, H. Lin, A. Damascelli, Phys. Rev. Lett. 112, 076802 (2014). S. Jia, M. Z. Hasan, Science, 349, 6428 (2015). [38] Zhuojin Xie, Shaolong He, Chaoyu Chen, Ya Feng, [51] G. Bian, T. R. Chang, R. Sankar, S. Y. Xu, H. Zheng, Hemian Yi, Aiji Liang, Lin Zhao, Daixiang Mou, Jun- T. Neupert, C. K. Chiu S. M. Huang, G. Chang, I. Be- fengHe,YingyingPeng,XuLiu,YanLiu,GuodongLiu, lopolski,D.S.Sanchez,M.Neupane,N.Alidoust,C.Liu, Xiaoli Dong, Li Yu, Jun Zhang, Shenjin Zhang, B. Wang, C. C. Lee, H. T. Jeng, C. Zhang, Z. Yuan, Zhimin Wang, Fengfeng Zhang, Feng Yang, Qin- S.Jia, A.Bansil, F.Chou, H.LinandM.Z.Hasan, Na- junPeng,XiaoyangWang,ChuangtianChen,ZuyanXu, ture Commun. 7, 10566 (2016). and X. J. Zhou, Nature Comm. 5, 3382 (2014). [39] K. Kuroda, K. Yaji, M. Nakayama, A. Harasawa,