A combined laser-based ARPES and 2PPES study ofTd-WTe2

Petra Hein1, Stephan Jauernik1, Hermann Erk1, Lexian Yang2,Yanpeng Qi3,4, Yan Sun4, Claudia Felser4 and Michael Bauer1

1 Institute of Experimental and Applied Physics, University of Kiel, Leibnizstr. 19, D-24118Kiel, Germany2 State Key Laboratory of Low Dimensional Quantum Physics, Collaborative Innovation Centerof Quantum Matter and Department of Physics, Tsinghua University, Beijing 100084, China3 School of Physical Science and Technology, ShanghaiTech University, Shanghai 201210, China4 Max Planck Institute for Chemical Physics of Solids, Nothnitzer Str. 40, D-01187 Dresden,Germany

E-mail: hein@physik.uni-kiel.de

Abstract. Laser-based angle-resolved photoemission spectroscopy (ARPES) and two-photonphotoemission spectroscopy (2PPES) are employed to study the valence electronic structure ofthe Weyl semimetal candidate Td-WTe2 along two high symmetry directions and for bindingenergies between ≈ −1 eV and 5 eV. The experimental data show a good agreement with bandstructure calculations. Polarization dependent measurements provide furthermore informationon initial and intermediate state symmetry properties with respect to the mirror plane of the Tdstructure of WTe2.

Submitted to: J. Phys.: Condens. Matter

arX

iv:2

002.

0758

1v1

[co

nd-m

at.m

trl-

sci]

18

Feb

2020

A combined laser-based ARPES and 2PPES study of Td-WTe2 2

1. Introduction

As a potential type-II Weyl semimetal [1], the tran-sition metal dichalcogenide WTe2 attracted consid-erable interest in recent years. Even though char-acteristic Weyl points and Fermi arcs in the elec-tronic structure could not directly be observed sofar [2, 3, 4, 5, 6, 7, 8], several experimental studiesclearly evidenced the topologically nontrivial phase inthis material [9, 10, 11, 12]. WTe2 crystallizes in a dis-torted 1T structure (Td structure) with an orthorhom-bic unit cell as shown in Fig. 1(a) [13]. It consists ofcovalently bonded Te-W-Te trilayers that are held to-gether via weak van der Waals forces resulting in anatural cleavage plane perpendicular to the layers in[001] direction. The bulk Brillouin zone (BZ) and its(001) surface projection are shown in Fig. 1(b). Re-sults of a band structure calculation along three se-lected high symmetry directions of the bulk Brillouinzone are shown in Fig. 1(c). Further band structuredata are provided in Fig. S1 of the supplemental ma-terial.

In this work, we report on a combined laser-based ARPES and 2PPES study of the occupied andunoccupied electronic structure of Td -WTe2. Thecombination of these complementary photoemissiontechniques enables us to access a binding energy rangebetween ≈ −1 eV and 5 eV with respect to the Fermienergy EF. Analysis of the polarization dependenceof ARPES and 2PPES spectra reveal information onthe symmetry of the probed occupied and unoccupiedstates with respect to the mirror plane of the Tdstructure. The data additionally provide insights intothe final state electronic structure above the vacuumenergy EV. Overall, band energies, band dispersions,and symmetry properties as determined from theexperimental data show a good agreement with orbital-resolved band structure calculations.

2. Experimental details and theoreticalmethod

High-quality Td -WTe2 crystals were grown using achemical vapour transport technique. Stoichiometrictungsten powder (99.9 %) and tellurium powder(99.99 %) were ground together and loaded into a

WTe

(a)

ab

c

(b)

X

Y

Z T

ΓU

X

ΓY

2pb

2pa

2pc

x y

z

7

6

5

4

3

2

1

0

-1

EÅ-

ÅEF

(eV

)

X Y G

1

2

4

5

6eo

o

oe

Z

(c)

G

cÅ=Å14.018ÅÅ

bÅ=Å6.249ÅÅÅ

aÅ=Å3.477ÅÅ

3 e

Figure 1. Crystal and electronic structure of Td-WTe2. (a)Orthorhombic unit cell with experimental lattice constants atT = 113 K [14]. The mirror plane (b-c plane) of the Td structureis indicated. (b) 3D Brillouin zone and its projection ontothe (001) surface. High symmetry planes ΓXY and ZUT areindicated in red and gray, respectively. (c) Calculated bulk bandstructure along selected high symmetry directions. The bandshighlighted by blue dashed lines are observed in the bichromatic2PPES spectra. The bands highlighted by blue and green solidlines are observed in the monochromatic 2PPES spectra. Even(e) and odd (o) symmetry of these bands with respect to themirror plane of the Td structure of WTe2 is indicated. Thenumbering of the bands is introduced for comparison with theexperimental results shown in Fig. 5 and Fig. 6. Detailsregarding the three markers and the shaded area in the bandstructure data along Γ-Z are given in the text.

quartz tube with a small amount of the transportagent TeBr4. All weighing and mixing was carriedout in a glove box. The tube was sealed undervacuum and placed in a two-zone furnace. Thehot zone and the cold zone were maintained forone week at a constant temperature of 800 ◦C and700 ◦C, respectively. Prior to the photoemissionexperiments, the Td -WTe2 samples were cleaved atroom temperature under ultrahigh vacuum (UHV)conditions (3 × 10−10 mbar). From the low-energycutoff of the ARPES spectra, we determined a valueof Φ ≈ 4.9 eV for the work function of the samples.

A scheme of the experimental setup, consistingof a tunable femtosecond laser system and anUHV chamber, is shown in Fig. 2(a) [15].

A combined laser-based ARPES and 2PPES study of Td-WTe2 3

ARPES and 2PPES measurements were performedwith the second harmonic output of a non-collinearoptical parametric amplifier (NOPA) (ORPHEUS-N-3H, Light Conversion) that is pumped by the thirdharmonic of a chirped pulse amplifier (PHAROS, LightConversion). For monochromatic 2PPES experiments,the photon energy hν was varied between 3.0 eV(413 nm) and 4.7 eV (264 nm) in steps of 0.1 eV. Forthe ARPES experiments, we used 5.9 eV (210 nm)light pulses that were produced by second harmonicgeneration (SHG) of the 2.95 eV (420 nm) output ofthe NOPA system. The 5.9 eV pulses were also usedas probe pulses in bichromatic 2PPES experiments.30 fs near infrared (NIR) / visible (VIS) pump pulseswere generated by an additional NOPA (ORPHEUS-N-2H, Light Conversion) that is pumped by the secondharmonic of the chirped pulse amplifier. Pump photonenergies were varied between 1.5 eV (827 nm) and1.9 eV (653 nm). The maximum incident pump fluenceon the sample surface was ≈ 110 µJ cm−2. Thepolarization of the laser light could be switched froms- to p-polarization using tunable zero-order half-waveplates. Here, s- (p-) polarization denotes the case thatthe electric field vector lies perpendicular to the planeof incidence (in the plane of incidence) held by thesample surface normal and the direction of incidenceof the laser light.

Photoelectron spectra were recorded with aSPECS Phoibos 100 analyzer. For the ARPES andbichromatic 2PPES experiments, the total energyresolution was 40 meV. For the monochromatic 2PPESexperiments, the total energy resolution was 74 meVto 90 meV, depending on the photon energy. Thequality of sample cleavage and the crystallographicorientation of the sample surface was checked by low-energy electron diffraction (LEED) (BDL600IR MCP2,OCI Vacuum). For the photoemission experiments, theTd -WTe2 samples were aligned either in Γ-X or in Γ-Y direction of the (001) surface Brillouin zone withrespect to the analyzer entrance slit [see LEED imagein Fig. 2(b)]. All measurements were performed atroom temperature.

The electronic band structure was calculated byab-initio calculation based on density functional theory(DFT) with Projector augmented-wave (PAW) method[16] as implemented in the Vienna Ab-initio SimulationPackage (VASP) [17]. The exchange and correlationenergies were considered on the level of the generalizedgradient approximation (GGA) with a Perdew-Burke-Ernzerhof (PBE) functional [18]. The energy cutoffwas set to be 350 eV for the plane wave basis. Theexperimental lattice constants from Ref. [14] were usedin all the calculations [see Fig. 1(a)].

k y(Å

-1)

-3

-2

-1

0

1

2

3

-3 -2 -1 0 1 2 3

kx (Å-1

)

Ekin = 175 eVT = 300 K

Г

Y

X

(b)max

min

LEE

D intensity

PHAROS

chirpedLpulse

amplifier

L

SHG

wDLelectron

energyLanalyzer

DualLMCP

LEED

e-

-9LW7L-9v9Lnm7L

N99LkHz7Lww9Lfs

v69LeVL666LU67LeV

kU-vLnmL666Lw8ULnm.

w69NLeVkUw9Lnm.

N69LeVkw-9Lnm.

half(LwaveLplate

2PPES ARPES

delaystage

SHLOutput

ORPHEUS(N(wH

NOPA

NIR/VIS

pump

-6NLeVL666L-69LeVk8w7LnmL666L8NvLnm.

half(LwaveLplate

ORPHEUS(N(vH

NOPA

SHLOutput

(a)

Figure 2. (a) Scheme of the combined laser-based ARPES and2PPES setup. Details are described in the text. (b) LEED imageof the (001) surface of Td-WTe2 with the surface Brillouin zoneindicated.

3. Results and discussion

3.1. ARPES results

Figure 3(a) shows ARPES intensity maps of Td -WTe2 along Γ-X and Γ-Y recorded using s-polarizedlight. Overall, the spectra conform with resultsof other ARPES studies that used similar photonenergies of hν ≈ 6 eV [2, 6, 19]. For comparison,the experimental data are overlaid with results ofthe band structure calculations along Γ-X and Γ-Y. Band energies and band dispersions predictedby the calculations are reproduced reasonably well.Due to matrix element effects, some of the bandsare not visible in the spectra [3]. As expectedfor a measurement in s-polarization geometry andwithin the free electron final state approximation, weobserve for most bands a considerable reduction in the

A combined laser-based ARPES and 2PPES study of Td-WTe2 4

Tevpx

Tevpy

Tevpz

-0.2 -0.1 0.0 0.1 0.2

ky (Å-1

)

Г → YY ←

s-pol max

min

-0.2 -0.1 0.0 0.1 0.2

ky (Å-1

)

-0.8

-0.6

-0.4

-0.2

0.0

Ev-

vEF

(eV

)

(b)

PE

vintensity

Г → YY ←

s-pol p-pol

-0.2 -0.1 0.0 0.1 0.2

kx (Å-1

)

-0.8

-0.6

-0.4

-0.2

0.0

Ev-

vEF

(eV

)

(a)Г → XX ←

s-polevenodd

-0.8

-0.6

-0.4

-0.2

0.0

Ev-

vEF

(eV

)

Wvdx2-vy

2

Wvdxy

Wvdxz

Wvdyz

Wvdz2

YYГ Г

(c)

Figure 3. Laser-based ARPES data of Td-WTe2 in comparison to band structure calculations. (a) ARPES intensity maps (bindingenergy E − EF vs. surface parallel momentum kx and ky, respectively) along Γ-X and Γ-Y in comparison to calculated bandstructures along Γ-X and Γ-Y. Spectra were recorded with s-polarized light at hν =5.9 eV. (b) Calculated Tellurium and Tungstencontributions to the bands along Γ-Y in the energy range covered by the ARPES data. The size of the data points indicates therespective orbital contribution to the individual bands. In Fig. S2 of the supplemental material, the orbital-resolved data are shownseparately. (c) Comparison of ARPES intensity maps along Γ-Y recorded with s-polarized light (left) and p-polarized light (right)at hν =5.9 eV. Results of the band structure calculations along Γ-Y are added with solid lines indicating bands of even symmetryand dashed lines indicating bands of odd symmetry with respect to the mirror plane of the Td structure.

photoemission yield at Γ, i.e., for normal emission [20].However, in the binding energy range between −0.2 eVand −0.4 eV along Γ-X, the signal becomes maximumfor normal emission. We suspect final state effectsbeing responsible for this observation: The resultsof the band structure calculations indicate that weobserve here a resonant transition to bulk final statesat E − EF ≈ 5.6 eV indicated by the green marker inFig. 1(c). These states show only a weak dispersionin normal direction, i.e., along Γ-Z, implying that thistransition is not well described by the free electron finalstate approximation.

The ARPES data along the Γ-Y direction wastaken with the UV light incident in the mirror planeof the Td structure of WTe2 so that the analysisof the polarization dependence allows for conclusionson the symmetry of the probed initial states withrespect to the mirror plane [21, 22]. ARPES spectrarecorded with p-polarized (s-polarized) light probeinitial state wave functions with even (odd) symmetry.The comparison with orbital-resolved band structuredata shown in Fig. 3(b) enables us to identifythe orbital contributions probed with the differentpolarizations: dyz, dx2−y2 , dz2 , py, and pz are even

with respect to reflection in the mirror plane of the Tdstructure, whereas dxy, dxz, and px orbitals are oddwith respect to reflection in the mirror plane of theTd structure. Fig. 3(c) directly compares ARPESintensity maps along Γ-Y recorded with s-polarizedlight and p-polarized light, respectively. The distinctdifferences in the spectra result to large extent from thedipole selection rules. The parabolic band observedclose to EF with s-polarized light conforms with thepredominant Te-px and W-dxz character of the hole-like bands in this energy range [see Fig. 3(b)]. Thesecond band visible in the s-polarized data can beassigned to the hole-like band at a binding energy ofEB ≈ −0.6 eV, which also shows a predominant Te-pxand W-dxz character. The weak photoemission signalobserved in the p-polarized data in this energy rangeconforms with the Te-pz and W-dz2 orbital character ofthe next higher lying band along Γ-Y. Note that bothbands are downward dispersing along Γ-Z and crosseach other halfway between Γ and Z [see blue markerin Fig. 1(c)]. In general, an ARPES experimentprobes a momentum cut at a finite value of the wavevector perpendicular to the surface projected BZ thatdepends on photon energy, crystal inner potential,

A combined laser-based ARPES and 2PPES study of Td-WTe2 5

-0.2 0.0 0.2

0.8

0.6

0.4

0.2

0.0

-0.2

min max

Eb-

bEF

5eV

)

Г → XX ←

PEbintensityb5a)

Dtb=b-1bps Dtb=b0bfs

kxb5Å-1)b

pos0

-0.2 0.0 0.2

1.6

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0.0

Eb-

bEF

5eV

)

Г → XX ←

diff.bPEbintensity5b)

kxb5Å-1)b

hnb=b1.7beV

hnb=b1.9beV

5c)

PE

bInte

nsity

b5ar

b.bu

nits

)

hnb=b1.5beV

hnb=b1.7beV1.0

0.5

0.0

1.0

0.5

0.0

1.0

0.5

0.02.42.01.61.2

Ekin 5eV)

5d)

E

EF

EVE-EFb=bb4.6beV

1.5b

eV

1.7b

eV

1.9b

eV

5.9b

eV

Figure 4. Bichromatic 2PPES data of Td-WTe2. (a) Comparison of bichromatic 2PPES intensity maps (binding energy E − EF

vs. surface parallel momentum kx) along Γ-X recorded before the excitation with the pump pulse (∆t = −1 ps) and at temporaloverlap between pump and probe pulse (∆t = 0 fs). The data were recorded with p-polarized pump pulses (hν = 1.7 eV) ands-polarized probe pulses (hν = 5.9 eV). (b) Difference 2PPES intensity map along Γ-X compiled from subtraction of data recordedat ∆t = −1 ps and ∆t = 0 fs. The data cover an excited energy range up to a binding energy of E − EF = 1.6 eV. Results of bandstructure calculations along Γ-X are included as white lines. The red dashed line shows results of the band structure calculationsalong Γ-X at E − EF ≈ 4.6 eV for comparison with the electron-like parabola visible in the difference 2PPES intensity map. Notethe different energy scales in (a) and (b). (c) Difference EDCs deduced from difference 2PPES intensity maps for different pumpphoton energies. The blue markers indicate the shift in energy of the minimum of the electron-like parabola. 2PPES intensities were

integrated over a momentum window of ±0.025 �A−1around Γ. (d) Schematic illustration of the 2PPE process with the NIR / VIS

pulses acting as probe pulses.

electron momentum parallel to the surface, and finalstate effective mass [21]. The appearance of the twobands at similar binding energies and shifted to higherbinding energies in comparison to the results of theband structure calculations along Γ-Y may thereforeindicate that in our experiment we probe the bandstructure at a finite momentum value in the surfacenormal direction. The data recorded with p-polarizedlight is finally dominated by a broad spectral featureextending from E − EF ≈ −0.2 eV to −0.4 eV. Theband structure calculations predict in this energy rangetwo bands with even symmetry formed from Te-pz, W-dz2 , and W-dyz orbitals, which merge at Γ. Thesebands exhibit a strong dispersion along Γ-Z [see redmarker in Fig. 1(c)] covering an overall energy rangeof ≈ 0.25 eV. We assume that the observed spectralbroadening results from a combination of the Γ-Zdispersion and a limited momentum resolution of theARPES experiment normal to the sample surface.

3.2. Bichromatic 2PPES results

Figure 4(a) compares bichromatic 2PPES data alongΓ-X recorded at temporal overlap of a p-polarized1.7 eV-pump pulse and a s-polarized UV-probe pulse(temporal delay ∆t = 0 fs) with data recorded wellbefore the excitation, i.e., at ∆t = −1 ps. The

excitation by the intense pump pulse transientlypopulates states above the Fermi energy EF andallows, therefore, addressing part of the unoccupiedelectronic structure. The transient spectral signal inthe excited state regime, which is barely visible inthe raw data, becomes substantially enhanced in adifference intensity representation as shown in Fig.4(b). The graph covers the excited state regime up tobinding energies of E−EF = 1.6 eV and was compiledby subtracting 2PPES intensity maps recorded at∆t = −1 ps from 2PPES intensity maps recorded at∆t = 0 fs. In the data, we resolve a hole-like band

crossing EF at kx ≈ ±0.2�A−1

and an electron-likeband with an energy minimum at E − EF ≈ 0.4 eV.The hole-like band agrees well with the results of theband structure calculations [see solid white lines inFig. 4(b)] and was observed in a time-resolved ARPESstudy of Td -WTe2 before [19]. However, the bandstructure calculations do not yield any indication foran electron-like band in the relevant energy-momentumrange. Remarkably, a change in the photon energy ofthe pump pulse by ∆hν results in a corresponding shiftof the kinetic energy at which the electron-like parabolaappears in the spectra [see peak in the difference energydistribution curves (EDCs) at Γ indicated by the bluemarkers in Fig. 4(c)]. This observation implies that

A combined laser-based ARPES and 2PPES study of Td-WTe2 6

Figure 5. Monochromatic 2PPES data of Td-WTe2. (a) 2PPES intensity maps (kinetic energy Ekin vs. surface parallel momentumkx) along Γ-X recorded with p-polarized light at hν = 4.7 eV. EDCs derived from this intensity map (solid line) as well as froman intensity map recorded with s-polarized light (dashed line) are shown to the right. 2PPES intensities were integrated over a

momentum window of ±0.025 �A−1around Γ. The 2PPES data are overlaid with results of the band structure calculations along

Γ-X. Only bands are shown that were clearly identified in the p-polarized experimental data. Red lines indicate initial states, bluelines indicate intermediate states, green lines indicate final states. The color-coded markers in the EDCs indicate the correspondingspectral peaks. (b) Comparison of 2PPES intensity maps along Γ-X recorded at different photon energies. Data were recorded withp-polarized light. The color-coded lines indicate the kinetic energy shifts expected for initial states (red line), intermediate states(blue line) and final states (green line). (c) Peak kinetic energies at Γ as a function of hν deduced from Gaussian multi-peak fits ofp-polarized and s-polarized monochromatic 2PPES data along Γ-X. The numbering of the bands allows for better comparison withthe band structure data in Fig. 1 and the experimental results shown in Fig. 6.

for the detection of the electron-like band the role ofpump and probe pulses are inverted and that the finalphotoemission process determining the kinetic energyof the photoelectrons is now due to the NIR / VISpulse [see Fig. 4(d)]. The reanalysis of the dataunder consideration of the inversion of pump and probeprocess results in a corrected value of E−EF ≈ 4.6 eVfor the energy minimum of the electron-like band. Thedashed red line in Fig. 4(b) shows for comparison theresults of the band structure calculations along Γ-Xin the corresponding energy range, which match theexperimental data reasonably well.

3.3. Monochromatic 2PPES results

Figure 5(a) shows monochromatic 2PPES intensitymaps along Γ-X recorded with p-polarized light athν = 4.7 eV. To the right, the graph additionallyincludes EDCs at Γ extracted from the 2PPESintensity map shown to the left as well as fromdata recorded with s-polarized light. We observe amultitude of bands, which vary in their intensity andrelative peak position as the photon energy is changed[see Fig. 5(b)].

Spectral peaks of monochromatic 2PPES data canbe due to initial states with binding energies E < EF,intermediate states with EF < E < EV, or final stateswith E > EV. An assignment is possible by analyzingthe shift in the peak kinetic energy as a function of

photon energy hν [23]. For monochromatic 2PPESdata, the observation of an energy shift by 2 · ∆hν(∆hν) is indicative for an initial state (intermediatestate). Here, ∆hν denotes the change in the photonenergy. In contrast, final states yield no energy shift inthe peak kinetic energy at all. Characteristic kineticenergy shifts are exemplarily indicated by the color-coded straight lines in Fig. 5(b) for selected spectralfeatures that are due to initial, intermediate, andfinal states, respectively. For the evaluation of the2PPES data, we performed Gaussian multi-peak fitsto the EDCs around Γ. Figure 5(c) summarizes peakkinetic energies as a function of hν as deduced fromEDCs recorded with p-polarized and s-polarized laserlight along Γ-X. The different slopes associated withinitial, intermediate, and final states are well resolvedin the data and are color-coded correspondingly. Thediscrimination of the state character allows for anevaluation of the state binding energies and finallyan assignment to the results of the band structurecalculation, partly under consideration of the banddispersion along Γ-X. Results of this analysis areincluded in Fig. 5(a), with an initial state shown as redline, intermediate states shown as blue lines, and a finalstate shown as green line. Note that only calculatedbands are included which could uniquely be identifiedin the p-polarized experimental data.

By analogy with the ARPES data, the polariza-tion dependence of the monochromatic 2PPES signal in

A combined laser-based ARPES and 2PPES study of Td-WTe2 7

normal emission along Γ-Y contains information aboutthe intermediate state symmetry with respect to themirror plane of the Td structure of WTe2. Figure6(a) compares monochromatic 2PPES data recordedalong Γ-Y at a photon energy of hν = 4.5 eV withs- and p-polarization, respectively. We selected thisdata set for the symmetry analysis as for a photonenergy of hν = 4.5 eV all six bands associated withbands in the intermediate state energy range are vis-ible in the spectra [see Fig. 5(c)]. The intensity mapis overlaid with the intermediate state band structurealong Γ-Y with the bands highlighted that could beassigned based on the kinetic energy analysis of the2PPES data. Figure 6(b) shows in comparison thecalculated orbital-resolved band structure data. Thelowest energy band at E −EF ≈ 1.3 eV is visible withs-polarized light and lies at Γ close to a final state res-onance. Still, at finite momentum we can clearly iden-tify the downward dispersing character of the band aspredicted by theory. Near Γ, the band is predomi-nantly formed from Te-px and W-dxy orbitals and ex-hibits therefore an odd symmetry with respect to themirror plane in accordance with the observation withs-polarized light. The same orbital character is alsofound for the band visible in the s-polarized data atE − EF ≈ 2.5 eV. Finally, the odd symmetry of theband seen at E − EF ≈ 3.5 eV arises from a Te-px or-bital character. In the p-polarized data set, we identifytwo unoccupied bands at E − EF ≈ 2 eV and a fur-ther unoccupied band at E−EF ≈ 4.2 eV. The orbitalcharacter of the two low energy bands is predominantlyTe-py and W-dyz (with an increasing W-dz2 contribu-tion at high momentum for the lower band). The evensymmetry of the high energy band mainly arises fromTe-py orbitals and some contributions from Te-s andW-s orbitals. Overall, the accordance between the po-larization dependence of the 2PPES spectra along Γ-Yand the orbital character predicted by the band struc-ture calculations confirms the band assignment thatwe made based on the kinetic energy and dispersionanalysis of the experimental data.

Finally, two final state resonances can be identifiedin the monochromatic 2PPES data [see Fig. 5(c)].The observed kinetic energies correspond to bindingenergies of E − EF ≈ 6.0 eV and E − EF ≈ 6.9 eV.A convincing assignment of the latter state to resultsof the band structure calculations is possible [cf. Fig.1(c)] and is correspondingly included in the 2PPESspectra shown in Fig. 5(a). However, the bandstructure calculations do not yield any indication for aband near E − EF ≈ 6.0 eV. The resonance is insteadlocated within a surface projected band gap at Γ [seegreen shaded area in Fig. 1(c)] indicating that weprobe here a surface located resonance in the final stateenergy range.

0.0-0.2

5

4

3

2

1

E6-

6EF

(eV

)

s-pol

0.2

p-pol

Г → YY ←

Г Y Г Y Г Y

Te6px

Te6py

Te6pz

W6dx2-y2

W6dxy

W6dxz

W6dyz

W6dz2

W6sTe6s

oddeven

ky6(Å-1)

(a) (b)

Figure 6. Symmetry properties of intermediate states. (a)Comparison of monochromatic 2PPES intensity maps (bindingenergy E − EF vs. surface parallel momentum ky) recorded

at hν = 4.5 eV along Γ-Y with s- and p-polarized light. Theexperimental data are overlaid with the results of band structurecalculations in the corresponding intermediate state energyrange. Bands that could be assigned based on the kinetic energyanalysis of the 2PPES data are indicated in blue, with solid linesindicating bands of even symmetry and dashed lines indicatingbands of odd symmetry with respect to the mirror plane of theTd structure. The numbering of the bands allows for bettercomparison with the band structure data in Fig. 1 and theexperimental results shown in Fig. 5. (b) Calculated Telluriumand Tungsten contributions to the bands along Γ-Y in the energyrange covered by the monochromatic 2PPES data. The size ofthe data points indicates the respective orbital contribution tothe individual bands. In Fig. S2 of the supplemental material,the orbital-resolved data are shown separately.

4. Conclusion

In combining laser-based ARPES and 2PPES, westudied in this work the valence electronic structureof the Weyl semimetal candidate Td -WTe2 along twohigh symmetry directions of the crystalline structure.The low photon energy provided by the laser systemconsiderably limits the accessible momentum rangeto values close to Γ, however, the bichromatic andmonochromatic 2PPES data enabled us to access theunoccupied electronic structure up to binding energiesof E − EF ≈ 5 eV. Furthermore, the high flexibility intuning the polarization state of the laser light enabledus to experimentally gain information on initial andintermediate state symmetries under consideration ofthe dipole selection rules in photoemission. Basedon these results and in comparison with bandstructure calculations, it was possible to identify thepredominant orbital contributions to the ARPES and2PPES signal for photoemission within the mirrorplane of the Td structure of WTe2.

5. Acknowledgments

This work was supported by the German ResearchFoundation (DFG) through projects INST 257/419-1and INST 257/442-1

A combined laser-based ARPES and 2PPES study of Td-WTe2 8

References

[1] Soluyanov A A, Gresch D, Wang Z, Wu Q, Troyer M, DaiX and Bernevig B A 2015 Nature 527 495–498

[2] Bruno F Y, Tamai A, Wu Q S, Cucchi I, Barreteau C,de la Torre A, McKeown Walker S, Ricco S, Wang Z,Kim T K, Hoesch M, Shi M, Plumb N C, Giannini E,Soluyanov A A and Baumberger F 2016 Phys. Rev. B94 121112(R)

[3] Wang C, Zhang Y, Huang J, Nie S, Liu G, Liang A, ZhangY, Shen B, Liu J, Hu C, Ding Y, Liu D, Hu Y, He S,Zhao L, Yu L, Hu J, Wei J, Mao Z, Shi Y, Jia X, ZhangF, Zhang S, Yang F, Wang Z, Peng Q, Weng H, Dai X,Fang Z, Xu Z, Chen C and Zhou X J 2016 Phys. Rev. B94 241119(R)

[4] Snchez-Barriga J, Vergniory M G, Evtushinsky D, AguileraI, Varykhalov A, Blgel S and Rader O 2016 Phys. Rev.B 94 161401(R)

[5] Feng B, Chan Y H, Feng Y, Liu R Y, Chou M Y, KurodaK, Yaji K, Harasawa A, Moras P, Barinov A, Malaeb W,Bareille C, Kondo T, Shin S, Komori F, Chiang T C, ShiY and Matsuda I 2016 Phys. Rev. B 94 195134

[6] Wu Y, Mou D, Jo N H, Sun K, Huang L, Bud’ko S L,Canfield P C and Kaminski A 2016 Phys. Rev. B 94121113(R)

[7] Di Sante D, Das P K, Bigi C, Ergonenc Z, Gurtler N,Krieger J A, Schmitt T, Ali M N, Rossi G, Thomale R,Franchini C, Picozzi S, Fujii J, Strocov V N, SangiovanniG, Vobornik I, Cava R J and Panaccione G 2017 Phys.Rev. Lett. 119 026403

[8] Zhang Q, Liu Z, Sun Y, Yang H, Jiang J, Mo S K, HussainZ, Qian X, Fu L, Yao S, Lu M, Felser C, Yan B, Chen Yand Yang L 2017 Phys. Status Solidi - Rapid Res. Lett.11 1700209

[9] Li P, Wen Y, He X, Zhang Q, Xia C, Yu Z M, Yang S A, ZhuZ, Alshareef H N and Zhang X X 2017 Nat. Commun. 82150

[10] Lin C L, Arafune R, Liu R Y, Yoshimura M, Feng B,Kawahara K, Ni Z, Minamitani E, Watanabe S, Shi Y,Kawai M, Chiang T C, Matsuda I and Takagi N 2017ACS Nano 11 11459–11465

[11] Lv Y Y, Li X, Zhang B B, Deng W Y, Yao S H, Chen Y B,Zhou J, Zhang S T, Lu M H, Zhang L, Tian M, Sheng Land Chen Y F 2017 Phys. Rev. Lett. 118 096603

[12] Li P, Zhang C, Wen Y, Cheng L, Nichols G, Cory D G, MiaoG X and Zhang X X 2019 Phys. Rev. B 100 205128

[13] Brown B E 1966 Acta Cryst. 20 268–274[14] Mar A, Jobic S and Ibers J A 1992 J. Am. Chem. Soc. 114

8963–8971[15] Jauernik S, Hein P, Gurgel M, Falke J and Bauer M 2018

Phys. Rev. B 97 125413[16] Blochl P E 1994 Phys. Rev. B 50 17953–17979[17] Kresse G and Furthmuller J 1996 Phys. Rev. B 54 11169–

11186[18] Perdew J P, Burke K and Ernzerhof M 1996 Phys. Rev.

Lett. 77 3865–3868[19] Caputo M, Khalil L, Papalazarou E, Nilforoushan N,

Perfetti L, Taleb-Ibrahimi A, Gibson Q D, Cava R J andMarsi M 2018 Phys. Rev. B 97 115115

[20] Moser S 2017 J. Electron Spectrosc. Relat. Phenomena 21429–52

[21] Hufner S 2003 Photoelectron Spectroscopy - Principles andApplications 3rd ed (Berlin Heidelberg: Springer) ISBN978-3540418023

[22] Mulazzi M, Hochstrasser M, Corso M, Vobornik I, Fujii J,Osterwalder J, Henk J and Rossi G 2006 Phys. Rev. B74 035118

[23] Ueba H and Gumhalter B 2007 Prog. Surf. Sci. 82 193–223

Supplemental Material for:

”A combined ARPES and two-photon photoemission study of

Td-WTe2”

Petra Hein,1, ∗ Stephan Jauernik,1 Hermann Erk,1 Lexian Yang,2

Yanpeng Qi,3, 4 Yan Sun,4 Claudia Felser,4 and Michael Bauer1

1Institute of Experimental and Applied Physics,

University of Kiel, Leibnizstr. 19, D-24118 Kiel, Germany

2State Key Laboratory of Low Dimensional Quantum Physics,

Collaborative Innovation Center of Quantum Matter and Department of Physics,

Tsinghua University, Beijing 100084, China

3School of Physical Science and Technology,

ShanghaiTech University, Shanghai 201210, China

4Max Planck Institute for Chemical Physics of Solids,

Nothnitzer Str. 40, D-01187 Dresden, Germany

(Dated: February 19, 2020)

∗ hein@physik.uni-kiel.de

arX

iv:2

002.

0758

1v1

[co

nd-m

at.m

trl-

sci]

18

Feb

2020

I. BAND STRUCTURE OF TD-WTE2

Γ X Z U Γ Y Z T Γ Z

8

6

4

2

0

-2

E -

EF

(eV

)

FIG. S1. Calculated bulk band structure of Td -WTe2 along high symmetry directions of the

Brillouin zone.

II. ORBITAL-RESOLVED BAND STRUCTURES

FIG. S2. Calculated orbital-resolved band structures along Γ-Y for the initial and intermediate

state energy range covered by the ARPES and 2PPES data. The size of the data points indicates

the respective orbital contribution to the individual bands.