Surface Versus Bulk Dirac States Tuning in a Three-Dimensional Topological Dirac Semimetal
Madhab Neupane, Su-Yang Xu, Nasser Alidoust, Raman Sankar, Ilya Belopolski, Guang Bian, Daniel S. Sanchez, Chang Liu, Tay-Rong Chang, Horng-Tay Jeng, BaoKai Wang, Guoqing Chang, Hsin Lin, Arun Bansil, Fangcheng Chou, M. Zahid Hasan
SSurface Versus Bulk Dirac States Tuning in a Three-Dimensional Topological DiracSemimetal
Madhab Neupane, Su-Yang Xu, Nasser Alidoust, Raman Sankar, Ilya Belopolski, GuangBian, Daniel S. Sanchez, Chang Liu, Tay-Rong Chang, Horng-Tay Jeng,
3, 4
BaoKai Wang,
5, 6, 7
Guoqing Chang,
5, 6
Hsin Lin,
5, 6
Arun Bansil, Fangcheng Chou, and M. Zahid Hasan
1, 8 Joseph Henry Laboratory, Department of Physics,Princeton University, Princeton, New Jersey 08544, USA Center for Condensed Matter Sciences, National Taiwan University, Taipei 10617, Taiwan Department of Physics, National Tsing Hua University, Hsinchu 30013, Taiwan Institute of Physics, Academia Sinica, Taipei 11529, Taiwan Centre for Advanced 2D Materials and Graphene Research Centre,National University of Singapore, 6 Science Drive 2, Singapore 117546 Department of Physics, National University of Singapore, 2 Science Drive 3, Singapore 117542 Department of Physics, Northeastern University, Boston, Massachusetts 02115, USA Princeton Center for Complex Materials, Princeton University, Princeton, New Jersey 08544, USA (Dated: November 5, 2018)Recently, crystalline-symmetry-protected three-dimensional (3D) bulk Dirac semimetal phase hasbeen experimentally identified in a stoichiometric high-mobility compound, Cd As . The Dirac stateobserved in Cd As has been attributed to originate mostly from the bulk state while calculationsshow that the bulk and surface states overlap over the entire Dirac dispersion energy range. In thisstudy, we unambiguously reveal doping induced evolution of the ground state of surface and bulkelectron dynamics in a 3D Dirac semimetal. We develop a systematic technique to isolate the surfaceand bulk states in Cd As , by simultaneously utilizing angle-resolved photoemission spectroscopy(ARPES) and in - situ surface deposition. Our experimental results provide a method for tuning thechemical potential as well as to observe surface states degenerate with bulk states, which will beuseful for future applications of 3D Dirac semimetal. Two-dimension (2D) Dirac electron systems often ex-hibit exotic quantum phenomena and, therefore, havebeen considered as one of the central topics in moderncondensed matter physics. The most promising exam-ples are the extensive studies on graphene and the sur-face states of topological insulators (TI) [1–11]. Diracsemimetals have gapless linear dispersions in three spa-tial dimensions, which can be regarded as a 3D analogof graphene. Recently, 3D Dirac semimetals have beentheoretically predicted [12, 13] and experimentally real-ized in Cd As [14–17] and Na Bi [18–20]. These Diracsemimetals possess two Dirac nodes which are protectedby the crystalline symmetry and are believed to holdthe key for realizing exotic Weyl physics [12–27]. It hasbeen theoretically understood that a 3D Dirac node canbe viewed as a composite of two sets of distinct Weylfermions [13, 21]. Moreover, if time-reversal or space in-version symmetry is broken, then the theory predicts thatthe 3D Dirac point (DP) can split into two Weyl nodesthat are separated in momentum space. Thus, a Weylsemimetal phase can be realized, which can be describedby Weyl nodes at the Fermi level for the bulk and thepresence of the exotic spin- polarized Fermi arcs on thesurfaces connecting the bulk nodes [13].The Dirac state observed in Cd As system has beenattributed to originate, mostly, from the bulk states [13–15] where the Dirac dispersion varies with photon en-ergy [14, 15]. Also, spin-resolved angle-resolved photoe-mission spectroscopy (SR-ARPES) measurements show negligible spin polarization [14] suggesting its bulk ori-gin. However, calculations show that the bulk and sur-face states overlap on the entire Dirac dispersion energyrange [13, 14]. Therefore, it is natural to ask a question:how can the surface state be resolved in the Cd As sys-tem? Furthermore, the overlapping of the surface andbulk bands may cause the hybridization between them,which makes it difficult to separate the contributions ofthe surface and bulk states. Hence, it is important to de-velop a technique to enhance the spectral weight of thesurface states in Cd As .In this Letter, we report the realization of the dop-ing induced evolution of the ground state of surfaceand bulk electron dynamics in a 3D Dirac semimetal.We develop a methodology to isolate the surface andbulk states in high-mobility stoichiometric 3D Diracsemimetal Cd As . Using high-resolution angle-resolvedphotoemission spectroscopy (ARPES) and in-situ alkalimetal surface deposition, we show that Cd As featuresa bulk Dirac cone located at the center of the (001) sur-face Brillouin zone (BZ), whose DP is tunable across thebinding energy. This surface doping induced tunabilityof the chemical potential reveals between the variationof surface versus bulk Dirac dispersion. The reportedobservation of a tunable bulk Dirac semimetal phase instoichiometric Cd As with a high Fermi velocity andelectron mobility opens the door to observe exotic 3DDirac relativistic physics in bulk materials and to realiz-ing the exciting Weyl semimetal topological phase with a r X i v : . [ c ond - m a t . m e s - h a ll ] J a n FIG. 1: Crystal structure. (a) Cd As crystalizes in a tetrag-onal body center structure with space group of I cd, whichhas 32 number of formula units in the unit cell. (b) The basicstructure unit is a 4 corner-sharing CdAs -trigonal pyramid.(c) The bulk Brillouin zone (BZ) and the projected surfaceBZ along the (001) direction. The red crossings locate at( k x , k y , k z ) = (0 , , . πc ∗ ) ( c ∗ = c/a ). (d) First-principlescalculation of the bulk electronic structure. Projected bulkband structure on to the (001) surface, where the shaded areashows the projection of the bulk bands. novel surface Fermi arcs.Single crystalline samples of Cd As used in this studywere grown using the standard method, which is de-scribed elsewhere [28]. ARPES measurements for thelow energy electronic structure were performed at thebeamlines 4.0.3 and 12.0.1 at the Advanced Light Source(ALS) in Berkeley California, equipped with high effi-ciency VG-Scienta R8000 and R200 electron analyzers.Samples were cleaved in situ and measured at 10 K in avacuum better than 1 × − torr. The energy resolu-tion was 10-30 meV and the angular resolution was bet-ter than 0.2 ◦ for all synchrotron-based ARPES measure-ments. Samples were observed to be stable and, for a typ-ical 20 hour measurement period, unsusceptible to degra-dation. Sodium deposition was performed at the beam-line 4.0.1 of the ALS from SAES getter source (SAESGetters USA, Inc.), which was thoroughly degassed be-fore starting the experiment. During deposition, pres-sure in the experimental chamber was < × − torr.First-principles calculations were based on the general-ized gradient approximation (GGA) [29] using the pro-jector augmented wave method [30] as implemented inthe VASP package [31]. The experimental crystal struc-ture was used [28]. The electronic structure calculationswere performed over 4 × × k − meshwith the spin-orbit coupling included self-consistently.We start our discussion by reviewing the crystal struc-ture and the Brillouin zone (BZ) of Cd As . The crys-tal structure of Cd As has a tetragonal unit cell with a = 12 .
67 ˚A, c = 25 .
48 ˚A, Z = 32, and the space groupof I cd (see Figs. 1a and b). In this structure, ar-senic ions are approximately cubic close-packed and Cdions are tetrahedrally coordinated. Each As ion is sur-rounded by Cd ions at six of the eight corners of a dis-torted cube (see Fig. 1b), with the two vacant sites lo-cated at the diagonally opposite corners of a cube face[28]. The corresponding BZ is shown in Fig. 1c, where the center of the BZ is the Γ point, the centers of the topand bottom square surfaces are the Z points, and otherhigh symmetry points are also noted. In electrical trans-port measurements, Cd As has long attracted much at-tention because of its very high electron mobility up to (cid:39) cm V − s − [32] and interesting magneto-transportproperties including a very small effective mass [33]. Intheoretical calculations, Cd As is of further interests be-cause it is believed to have an inverted band structure[34]. More interestingly, a very recent theoretical study[13] has shown that spin-orbit interaction in Cd As can-not open up a full energy gap between the inverted bulkconduction and valence bands due to the protection of anadditional crystallographic symmetry [22] (in the case ofCd As it is the C rotational symmetry along the k z di-rection [13]), which is in contrast to other band-invertedsystems such as Bi Se or HgTe [3]. Theory predicts [13]that the C rotational symmetry protects two bulk (3D)Dirac band touchings at two special k points along theΓ − Z momentum space cut-direction, as shown by thered crossings in Fig. 1c. At the (001) surface, theoreticalcalculations show the presence of one 3D Dirac cone atthe BZ center point (¯Γ) as shown in Fig. 1d. There-fore, Cd As offers a platform for an exotic space groupsymmetry-protected bulk Dirac semimetal (BDS) phase,which additionally features a very high carrier mobilityand an inverted band-structure.In order to convey the tunability of the chemical po-tential, we systematically study the electronic structureof Cd As at the cleaved (001) surface by successive Nadeposition. High-resolution ARPES dispersion measure-ments are performed in the close vicinity of the Fermilevel as shown in Fig. 2. The used photon energy of102 eV corresponds to the photon energy at which bulkDirac dispersion is observed. At this “magic” photon en-ergy, a linearly dispersive Dirac cone is observed at thesurface BZ center ¯Γ point, whose Dirac node is foundto be located at binding energy of E B (cid:39) . k points along theΓ − Z momentum space cut-direction, as shown by thered crossings in Fig. 1d. At the (001) surface, these two k points along the Γ − Z axis project on to the ¯Γ point ofthe (001) surface BZ (Fig. 1d). Therefore, at the (001)surface, theory predicts the existence of one 3D Diraccone at the BZ center ¯Γ point, as shown in Fig. 2a.More substantial changes are observed by the in situ surface deposition of alkali metal (sodium, Na) on thecleaved surfaces of Cd As . As a function of Na-deposition time, three stages can be identified. At firstfor moderate Na deposition ( ∼
60 seconds) the Dirac
FIG. 2: Fermi level tuning by surface Na deposition. (a) ARPES dispersion map with successive Na deposition. The time forNa deposition is noted on spectra. Dirac-cone-like intensity continuum is observed for no surface deposition spectrum. Long(short) blue dash line shows the binding energy position of the Dirac point without (with) surface deposition. And verticalarrow indicates the energy shift of the binding energy ( ∼
300 meV). These data were measured with photon energy of 102 eVand temperature of 10 K at ALS BL4. (b) Corresponding momentum distribution curves (MDCs). point (DP) moves to higher binding energy by electrondoping and a sharper state is observed to emerge at theedge of the bulk continuum. We can identify this stageat which surface states start to develop. For intermedi-ate Na deposition [ ∼
60 to 120 seconds, Fig. 2 middlepanel], the electron doping further increases and sharperboundary states appear. This stage corresponds withfully develop surface states. Finally, heavy Na deposition[beyond 180 seconds] leaves the surface states and DPunchanged, which indicates that the sample cannot befurther doped. These observations are further illustratedby the momentum distribution curves (MDCs) plotted inFig. 2b.In order to directly visualize the isolation of the surfaceand bulk states in Cd As , we compare MDCs extractedfrom 50 meV binding energy at different deposition time(see Fig. 3). It is worth noting that the topologicalDirac SM carries a nontrivial 2D Z invariant on eitherthe k z = 0 or the k z = π plane. Therefore, the Dirac semimetal turns into a strong Z topological insulator ifan external perturbation breaks the C symmetry butdoes not break time reversal symmetry [35]. Thus theprojected state on the (001) surface gives both bulk andsurface states, as shown schematically in Fig. 3a. TheMDCs at different deposition times are shown in Fig. 3c,whereas the white-dash line on the ARPES spectra inFig. 3b depicts the approximate energy position of theextracted MDCs. From Fig. 3c, it is evident that afterdeposition the double peaks emerge from a single peakof MDC before deposition. These two peaks correspondto the surface state of linear dispersion.In order to illustrate the surface state nature of thedispersion after deposition, we perform photon energydependent ARPES measurement as shown in Fig. 4. Forthis purpose, we use a relatively p -type sample where thechemical potential is located in the vicinity of the DP(see Fig. 4). A 3D Dirac semimetal is shown to featurenearly linear dispersions along all three momentum spacedirections close to the crossing point, even though Fermivelocity can vary significantly along different directions[14]. With the introduction of in - situ Na deposition of ∼
120 seconds, the chemical potential is shifted up by 300meV, leaving the upper cone is now visible. Furthermore,the boundaries of the upper dispersion maps seem to besharper. To identify their origin, we perform photon en-ergy dependent measurements from 96 to 110 eV with a2 eV energy increment (see Fig. 4). Upon varying thephoton energy, the resulting dispersion maps are foundto be unchanged, which is evidence supporting their 2Dnature (see Fig. 4). These measurements further supportour claim, which is that the precise control of the carrierdensity, and, therefore, the desired Fermi level can beachieved in the Cd As systemNow we discuss some interesting consequences of ourexperiment. First, the surface alkali metal deposition onCd As has revealed that the chemical potential of thebulk Dirac point can be tuned, at least near the sur-face since ARPES is relatively surface sensitive. The re-sulting measurements of this study show that the Diracpoint is shifted in binding energy by more than ∼ As can serve as a methodology to isolate the sur-face and bulk bands in this system. Third, our photonenergy dependent measurements reveal that the surfacedeposition provide a technique for tuning the bulk ver-sus surface state. By the addition of Na on the surfaceof Cd As , the ARPES signal becomes more surface likewhich may probably be due to decreasing the penetrationdepth of the photon flux.In conclusion, we experimentally show how to tunethe chemical potential of the three-dimensional Diracsemimetal compound Cd As by simultaneously execut-ing surface alkali metal deposition and ARPES measure-ments. In doing so, we develop a methodology to distin-guish between the surface and bulk states in this high-mobility compound. The observation of the tunable 3Dbulk Dirac phase and the surface doping induced surfaceversus bulk tunability in Cd As paves the way for realiz-ing a number of exotic topological phenomena and raisesits potential application for future devices. Acknowledgements
Work at Princeton University is supported by theUS National Science Foundation Grant, NSF-DMR-1006492. M. Z. H. acknowledges visiting-scientist sup-
FIG. 3: Surface vs bulk state tuning. (a) Schematic view ofthe surface and bulk BZ along (001) plane. The red x signshows the bulk Dirac cone and circle sign denote the topo-logical surface state. BS and TSS represent bulk Dirac stateand topological surface states. (b) ARPES measured surfacedispersion maps of Cd As . The deposition time is notedon the spectra. The white dash-line represents the approx-imate binding energy position for the extracted momentumdistribution curves (MDCs). (c) MDCs extracted from thespectra measured with different deposition timesz. The ar-row lines represent the peak position. The evolution of thedouble peaks from a single peak shows how the spectral weightmostly dominated by surface states arise from the bulk state.FIG. 4: Observation of surface state. ARPES dispersionmaps measured before and after Na deposition. The samplemeasured here is relatively p − type, which is prepared underslightly different growth conditions (temperature and growthtime). Photon energy dependent ARPES dispersion mapsafter 120 seconds of Na deposition. The measured photonenergies are noted on the spectra. port from Lawrence Berkeley National Laboratory andadditional partial support from the A. P. Sloan Foun-dation and NSF-DMR-0819860. The work at North-eastern University was supported by the US Depart-ment of Energy (DOE), Office of Science, Basic EnergySciences grant number DE-FG02-07ER46352, and ben-efited from Northeastern University’s Advanced Scien-tific Computation Center (ASCC) and the NERSC su-percomputing center through DOE grant number DE-AC02-05CH11231. H. L. acknowledges the SingaporeNational Research Foundation for support under NRFGrant No. NRF-NRFF2013-03. T.-R. C. and H.-T. J.are supported by the National Science Council, Taiwan.H.-T. J. also thanks NCHC, CINC-NTU, and NCTS, Tai-wan, for technical support. We thank J. D. Denlinger andA. V. Fedorov for beamline assistance at the AdvancedLight Source (ALS-LBNL) in Berkeley. M.N. acknowl-edges discussion with A. Alexandradinata and Z. Wangfrom Princeton Physics. [1] P. A. M. Dirac, Proc. R. Soc. Lond. A , 610 (1928).[2] A. K. Geim, and K. S. Novoselov, Nature Mat. , 183(2007).[3] M. Z. Hasan, & C. L. Kane, Rev. Mod. Phys, , 3045-3067 (2010).[4] X-L. Qi & S.-C. Zhang, Rev. Mod. Phys. , 1057-1110(2011).[5] D. Hsieh et al. , Nature , 970-974 (2008).[6] D. Hsieh et al. , Science , 919 (2009).[7] Y. Xia et al. , Nature Phys. , 398-402 (2009).[8] Y. Hor et al. , Phys. Rev. B , 195208 (2009).[9] Y. L. Chen et al. , Science , 178-181 (2009).[10] S. Y. Xu et al. , Science , 560-564 (2011).[11] M. Neupane et al. , Nat. Commun. , 2991 (2013).[12] Z. Wang et al. , Phys. Rev. B , 195320 (2012).[13] Z. Wang et.al. , Phys. Rev. B , 125427 (2013).[14] M. Neupane et al. , Nature Commun. , 3786 (2014).[15] S. Borisenko et al. , Phys. Rev. Lett. , 027603(2014).[16] Z. K. Liu et al. , Nat. Mat. , 677 (2014).[17] H. Yi et al. , Sci. Rep. , 6106(2014).[18] Z. K. Liu et al. , Science , 864 (2014).[19] S.Y. Xu et al. , arXiv:1312.7624 (2013).[20] S.Y. Xu et al. , Science (2014) [Science 10.1126/sci-ence.1256742 (2014)].[21] S. Murakami, New. J. Phys. , 356 (2007).[22] S. M. Young et al. , Phys. Rev. Lett. , 140405 (2012).[23] H. Weyl, Phys. , 330 (1929).[24] G. T. Volovik, JETP Lett. , 55 (2002).[25] Z. Fang, N. Nagaosa, K. S. Takahashi et.al. , Science ,92 (2003).[26] X. Wan et al. , Phys. Rev. B , 205101 (2011).[27] G. B. Halasz, L. Balents, Phys. Rev. B. ,1062 (1968.)[29] J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev.Lett. , 3865 (1996).[30] P. E. Blochl, Phys. Rev. B. , 17953 (1994); G. Kresseand J. Joubert, Phys. Rev. B. , 1758 (1999). [31] G. Kress and J. Hafner, Phys. Rev. B. , 13115 (1993);G. Kress and J. Furthmuller, Comput. Mater. Sci. , 15(1996); Phys. Rev. B. , 11169 (1996).[32] J.-P. Jay-Gerin, M.J. Aubin and L.G. Caron, Solid StateCommunications , 771 (1977).[33] W. Zdanowicz et al. , Thin Solid Films , 41 (1979).[34] B. D. Plenkiewicz and P. Plenkiewicz, Physica StatusSolidi(b) , K57 (2006).[35] B.-J. Yang et al., Nature Commun.5