Correlated vs. conventional insulating behavior in the Jeff=1/2 vs. 3/2 bands in the layered iridate Ba2IrO4
M. Uchida, Y. F. Nie, P. D. C. King, C. H. Kim, C. J. Fennie, D. G. Schlom, K. M. Shen
aa r X i v : . [ c ond - m a t . s t r- e l ] A ug Correlated vs. conventional insulating behavior in the J eff = 1 / vs. / bands in the layered iridate Ba IrO M. Uchida, ∗ Y.F. Nie,
1, 2
P.D.C. King,
1, 3
C.H. Kim, C.J. Fennie, D.G. Schlom,
2, 3 and K.M. Shen
1, 3, † Laboratory of Atomic and Solid State Physics, Department of Physics,Cornell University, Ithaca, New York 14853, USA Department of Materials Science and Engineering,Cornell University, Ithaca, New York 14853, USA Kavli Institute at Cornell for Nanoscale Science, Ithaca, New York 14853, USA Department of Applied Physics, Cornell University, Ithaca, New York 14853, USA (Dated: April 29, 2019)
Abstract
We employ molecular beam epitaxy to stabilize Ba IrO thin films and utilize in situ angle-resolved photoemission spectroscopy to investigate the evolution of its electronic structure throughthe N´eel temperature T N . Our measurements indicate that dispersions of the relativistic J eff =1 / J eff = 1 / T N ,while the nearly fully occupied J eff = 3 / PACS numbers: 71.30.+h, 74.25.Jb, 71.70.Ej, 71.20.-b . INTRODUCTION d transition metal oxides have recently attracted great interest due to the interplay be-tween spin-orbit and Coulomb interactions which can give rise to novel many-body quantumstates, including the theoretically proposed Weyl semimetal, topological Mott insulator, or high-temperature superconductivity upon carrier doping. These studies have beenlargely motivated by the experiments in a model compound Sr IrO , which suggest thatthe low-energy electronic states can be suitably represented by their effective total angu-lar momentum J eff , and argue that the Ir (5 d ) t orbitals are split into a fully filled J eff = 3 / J eff = 1 / It has been proposed that themodest Coulomb interactions of the 5 d electrons can then result in a further splitting of thehalf-filled J eff = 1 / in analogy to the conventional 3 d transition metal oxides, or as Slaterinsulators, where the insulating behavior is tied directly to long-range antiferromagneticorder. For example, recent scanning tunnel spectroscopy on Sr IrO has reported an onsetof the temperature dependent spectra below the antiferromagnetic ordering temperature T N ,consistent with dynamical mean field theory results, although the in-gap spectral weight isstill strongly suppressed even above T N , as also seen in other spectroscopy experiments. In order to definitively address these issues, it has become critical to systematically examineother related 5 d transition metal oxides, such as Ba IrO . This also opens new avenues torealize proposed exotic phases such as superconductivity which may exist at the intersectionbetween strong spin-orbit interactions and electron correlations.Ba IrO is an ideal candidate material for addressing many of the outstanding issues in thefield of the iridates. Ba IrO possesses a less distorted, simple quasi-two-dimensional crystalstructure ( I mmm in bulk), as shown in Fig. 1(a), with Ba replacing Sr, while it showsthe same insulating behavior accompanied with the basal plane antiferromagnetic orderbelow the N´eel temperature T N = 240 K. The undistorted crystal structure without thein-plane IrO octahedral rotation is expected to result in simplified band dispersions withoutfolded features, enabling us to follow the detailed temperature evolution of the relativistic J eff = 1 / IrO has a number of distinct advantages over Sr IrO , includingthe capacity for being metallized either through carrier doping (K or La substitution forBa) or through the application of hydrostatic pressure, making it an excellent platformfor searching for the possibility of exotic, iridate-based superconductivity. Unfortunately,layered Ba IrO is metastable in bulk and can only be formed under high pressure, makingthe synthesis of large bulk single crystals difficult. Here, we apply oxide molecular beamepitaxy (MBE) to stabilize thin films of the desired structure of Ba IrO , and utilize in situ angle-resolved photoemission spectroscopy (ARPES) to investigate the parent insulatingground state of Ba IrO and how it evolves through T N . Longitudinal in-plane resistivity inFig. 1(b) shows clear insulating characteristics of the Ba IrO film. We compare our ARPESmeasurements with calculations based on density functional theory incorporating electron-electron and spin-orbit interactions. While both the relativistic J eff = 1 / E F , we reveal a surprising contrast in their temperature dependence.Although the gap survives well into the paramagnetic state, only the J eff = 1 / T N reminiscent of some of the parent cuprate superconductors, suggesting the importanceof long-range antiferromagnetic order. II. EXPERIMENTAL AND CALCULATIONAL PROCEDURES
Thin films of (001) Ba IrO of thickness ∼
15 nm were deposited epitaxially on (001) p (where the subscript p denotes pseudocubic indices) PrScO single crystal substrates usinga Veeco GEN10 oxide MBE system. Absorption-controlled deposition was performed indistilled 100% O at a background pressure of 1 × − Torr. Under optimized adsorptioncontrolled conditions, Ba and Ir were supplied continuously with a flux of 6 × and7 × atoms / cm s from an effusion cell and an electron beam evaporator, respectively. Thesubstrate was heated at 800 ◦ C as measured by a pyrometer. Under these conditions the extraIr atoms form volatile IrO x and then evaporate from the surface, leaving stoichiometricBa IrO films.During growth films were monitored with reflection high-energy electron diffraction(RHEED) while rotating the substrate. After growth, samples were immediately ( < α ) andan instrumental energy resolution of 20 meV. The sample was measured at temperatures be-tween 100 and 300 K while maintaining a base pressure typically better than 8 × − Torr.Following ARPES measurements, films were characterized in situ by low-energy electrondiffraction (LEED). The phase purity and crystallinity of Ba IrO films were characterizedalso using ex situ four-circle x-ray diffraction (XRD) with Cu Kα radiation.The ab-initio density functional theory (DFT) calculations were performed using Wien2kcode including spin-orbit coupling and an on-site Coulomb repulsion, with the local densityapproximation and the Perdew-Burke-Ernzerhof exchange-correlation functional. A tetrag-onal structure was assumed, with the slightly strained lattice ( a = 4 .
021 ˚A and c = 13 .
34 ˚A)on the PrScO substrate, and no peaks corresponding to octahedral rotation were observedfrom bulk diffraction measurements. III. FILM CHARACTERIZATION
Figure 2 shows typical RHEED images of the PrScO substrate and Ba IrO film alongthe [110] ( p ) and [100] ( p ) azimuth. Ba IrO films exhibit prominent Kikuchi lines, indicatinghigh crystalline perfection, ensuring the quality of the photoemission spectra. The in-planelattice mismatch to the PrScO substrate ( a p = 4 .
021 ˚A) induces only 0.2% compressivestrain and the RHEED streaks do not show a discernible shift resulting from any latticerelaxation (Figs. 2(c) and (f)).Exemplary LEED patterns are shown in Figs. 3(a) and (b), taken at normal incidencewith beam energies of 150 and 200 eV. The sharp diffraction peaks indicate a well orderedsurface crystal structure. The assumed tetragonal (1 ×
1) diffraction peaks are indicatedin the pattern, and weak Bragg peaks observed at √ × √ ◦ relative to the 1 × IrO , given the absenceof such peaks in bulk x-ray measurements. The XRD θ –2 θ scan in Fig. 3(c) shows clearKiessig fringes, indicating good surface smoothness, and a Nelson-Riley analysis of the peakpositions gives an out-of-plane lattice constant of c = 13 .
34 ˚A, meaning films may be slightlyelongated along the c -axis compared with bulk polycrystals ( a = 4 .
030 ˚A and c = 13 . due to the compressive in-plane strain.4 V. RESULTS AND DISCUSSION
Figures 4(a)-(c) show valence band photoemission spectra at high-symmetry points forBa IrO epitaxial films. By comparison to calculations, the peaks near the Fermilevel E F can be assigned to the Ir 5 d t bands, while the spectral features between 2 and 7eV can be ascribed to dominantly O 2 p states. As shown in the energy distribution curves(EDCs) in Fig. 4(d), low-energy dispersive features with clearly defined peaks are observedclose to E F , where the lowest-energy feature is located ∼ π, π, π ) point only 0.4 eV below E F . The full width at halfmaximum of the EDC peaks were typically ∼
200 meV near the top of the valence band, witha lineshape well fit by a Gaussian. This indicates that the broad spectra are dominated by amanifold of states that involve multiple bosonic excitations, suggesting polaronic behavior,similar to that observed in the insulating parent cuprates, as well as in the sister compoundSr IrO . In Figure 5, we compare the near- E F ARPES spectra and the experimental dispersions(Σ –Σ ) extracted from EDC and momentum distribution curve (MDC) fits, to our DFTcalculations with spin-orbit coupling and an on-site Coulomb repulsion U . As shown in Fig.5(a), along the ( π, π, π ) or ( π/ , π/ π, π ) high symmetry directions, the half-filledΣ band exhibits clear back-bending, indicating that a full gap opens along the ( π, , π )diagonal line, consistent with ( π, π ) N´eel antiferromagnetic order which is experimentallyobserved. The lowest energy states are at ( π, IrO and Ba IrO single crystals. Comparisons to our DFT calculations with U of2.5 eV (comparable to values reported in the literature for Sr IrO ) give qualitativelygood agreement with the experimental dispersions, and indicate that the Σ states whichopen a gap near ( π,
0) are of predominately J eff = 1 / and Σ statesnear ( π, π ) are observed to be very close in energy to the ( π,
0) states, and in fact formthe lowest lying states in the predicted valence band maximum at ( π, π ), at odds with ex-periment, but also observed in previous works on Sr IrO . Comparisons with the DFTcalculations indicate that these states are of predominately J eff = 3 / J eff character of these states do not appear to depend strongly on U (Figs. 5(b) and (c)) oroctahedral rotation angle, consistent with earlier work on Sr IrO by Martins et al. . This5nderscores the importance of low-lying J eff = 3 / J eff = 1 / , band at ( π, π ), which is probably due tothe √ × √ IrO , in that the intensity of the reflected shadow bandaround (0 ,
0) is significantly weaker than at ( π, π ), whereas in Sr IrO the intensity of thesefeatures are comparable. This may also imply that the octahedral rotation angles observedat the surface of the Ba IrO films are substantially smaller than those in Sr IrO . Here it is meaningful to quantitatively compare the observed energy bands with thecalculated ones. The experimentally extracted bandwidths (separation from band minimumto maximum) are W Σ (0 , − ( π, = 0 . ± . W Σ (0 , − ( π, = 0 . ± . W Σ ( π, − ( π,π ) =0 . ± . W Σ ( π, − ( π,π ) = 1 . ± . U of 2.5 eV are 0 .
74 eV, 0 .
68 eV, 0 .
48 eV, and 1 . U to zero only increases these bandwidths slightly to 1 .
13 eV,0 .
93 eV, 0 .
54 eV, and 1 .
48 eV, but this weak renormalization reflects the relatively modesteffective Coulomb interaction of the 5 d electrons, in contrast to the much larger bandwidthrenormalization typically observed in the high- T c cuprates. The other remarkable point isthe energy difference between the Σ and Σ bands. Since this gap at the Γ point directlymeasures the spin-orbit coupling parameter as suggested by tight-binding models, thespin-orbit interaction in Ba IrO is experimentally determined to be 0 . ± . Figure 6 shows a comparison of the measured low-energy electronic structures of Ba IrO and Sr IrO . (001) Sr IrO films were epitaxially grown on LSAT substrates by MBE, withthe ARPES spectra consistent with previous measurements. As shown in Fig. 6(d), theBrillouin zone of Sr IrO is reduced by half in momentum space and the band structure isfolded along the ( π, , π ) and equivalent diagonal lines, reflecting the √ × √ E versus k plots shown in Figs. 6(c) and (f). For example, the band dispersion seen centered at ( π, π )in Ba IrO can be clearly observed at the equivalent reconstructed position (0 ,
0) in the caseof Sr IrO . We interpret the weaker reconstruction observed in Ba IrO as an indication6hat the amount of octahedral rotation due to the surface reconstruction is significantly lessthan in Sr IrO . In addition, we find that the total bandwidth of the t bands are about1.4 eV both in Ba IrO and Sr IrO . This suggests that the transfer integrals between thenearest-neighbor d xy ( d yz , d zx ) orbitals are roughly comparable.Given that the J eff = 1 / E F , we investigatedtheir roles in the N´eel transition by tracking their temperature dependence through T N .Figure 7(a) shows temperature-dependent EDCs at the different momentum positions A–Csketched in Fig. 7(c). The spectra change reproducibly both in warming and cooling, rulingout the possibility of sample surface degradation. The effects of electrostatic charging andfinite energy resolution were also experimentally precluded from affecting the data in anyappreciable way. While peaks A (at ( π/ , π/ π, J eff = 1 / T N ,peak C (at ( π, π )), primarily of J eff = 3 / J eff = 1 / T N , then abruptly stop shifting above T N , while C on the J eff = 3 / ∼ . IrO and the undopedcuprates. This unusual temperature dependence of the peak position suggests that onlythe J eff = 1 / J eff = 3 / IrO or Sr IrO with temperature. The clear persistence of the gap and insulating behavior above T N obviously indicates7hat the system does not behave as a simple Slater insulator. Furthermore, the nonmono-tonic change of the observed J eff = 1 / T N suggests the possibility that theeffective Coulomb interactions may be enhanced below T N accompanied by an ordering ofthe local moments. For example, in the antiferromagnetic Mott insulator LaTiO , a similarnonmonotonic change of the Mott gap through T N has been reported. In Fig. 8(a) we illustrate schematically the temperature evolution of the gapped J eff = 1 / IrO . While the charge gap continues to soften when approaching T N , it does not collapse above T N , reflecting the robust correlated insulating state. Onthe other hand, the magnitude of the gap itself is relatively small and the bandwidth isfound to be barely renormalized relative to the predictions from density functional theory,as schematically shown in Fig. 8(c). This is in stark contrast to conventional 3 d transitionmetal oxide Mott insulators which feature a much larger charge gap and a more stronglyrenormalized bandwidth (Fig. 8(d)), a clear distinction between the parent insulating stateof the layered iridates versus the cuprates. When the spin-orbit interaction is sufficientlystrong, even a modest Coulomb repulsion is sufficient to push the nearly fully occupied J eff = 3 / E F (blue), while splitting the half-filled J eff = 1 / V. CONCLUSION
In summary, we have investigated the electronic ground state and its temperature evo-lution in Ba IrO using a combination of reactive oxide molecular beam epitaxy and angle-resolved photoemission spectroscopy. A comparison between the experimental data andtheoretical calculations have demonstrated the importance of the J eff = 3 / J eff = 1 / J eff = 1 / T N , suggesting an important role of the antiferromag-netic ordering in the formation of the insulating phase, while the gap itself remains robustwell into the paramagnetic state. This surprising contrast in the behavior of the low-energy J eff = 1 / Acknowledgments
We gratefully acknowledge insightful discussions with J. W. Freeland, J. H. Lee, and T.Senthil. In particular, we would like to thank M. Grioni for helpful discussions regardingpreviously unpublished data in Ref. 24, which we became aware of during the preparationof this manuscript. This work was supported by the Air Force Office of Scientific Research(Grant No. FA9550-11-1-0033 and FA9550-12-1-0035), and the National Science Foundationthrough the MRSEC program (Cornell Center for Materials Research, DMR-1120296). M.U.acknowledges the support by JSPS Postdoctoral Fellowships for Research Abroad. ∗ Present address: Department of Applied Physics and Quantum-Phase Electronics Center(QPEC), University of Tokyo, Tokyo 113-8656, Japan † Author to whom correspondence should be addressed: [email protected] X. Wan, A.M. Turner, A. Vishwanath, and S.Y. Savrasov, Phys. Rev. B W. Witczak-Krempa and Y.B. Kim, Phys. Rev. B D. Pesin and L. Balents, Nat. Phys.
376 (2010). X. Zhang, H. Zhang, J. Wang, C. Felser, and S.-C. Zhang, Science F. Wang and T. Senthil, Phys. Rev. Lett. H. Watanabe, T. Shirakawa, and S. Yunoki, Phys. Rev. Lett. B. J. Kim, H. Jin, S. J. Moon, J. -Y. Kim, B. -G. Park, C. S. Leem, J. Yu, T. W. Noh, C. Kim,S. -J. Oh, J. -H. Park, V. Durairaj, G. Cao, and E. Rotenberg, Phys. Rev. Lett. B. J. Kim, H. Ohsumi, T. Komesu, S. Sakai, T. Morita, H. Takagi, and T. Arima, Science S. J. Moon, H. Jin, W. S. Choi, J. S. Lee, S. S. A. Seo, J. Yu, G. Cao, T. W. Noh, and Y. S.Lee, Phys. Rev. B D. Hsieh, F. Mahmood, D. H. Torchinsky, G. Cao, and N. Gedik, Phys. Rev. B R. Arita, J. Kuneˇs, A. V. Kozhevnikov, A. G. Eguiluz, and M. Imada, Phys. Rev. Lett. Q. Li, G. Cao, S. Okamoto, J. Yi, W. Lin, B. C. Sales, J. Yan, R. Arita, J. Kunes, A. V.Kozhevnikov, A. G. Eguiluz, M. Imada, Z. Gai, M. Pan, and D. G. Mandrus, Sci. Rep. H. Okabe, M. Isobe, E. Takayama-Muromachi, A. Koda, S. Takeshita, M. Hiraishi, M. Miyazaki,R. Kadono, Y. Miyake, and J. Akimitsu, Phys. Rev. B S. Boseggia, R. Springell, H. C. Walker, H. M. Rønnow, Ch. R¨uegg, H. Okabe, M. Isobe, R. S.Perry, S. P. Collins, and D. F. McMorrow, Phys. Rev. Lett. H. Okabe, M. Isobe, E. Takayama-Muromachi, N. Takeshita, and J. Akimitsu, Phys. Rev. B H. Okabe, N. Takeshita, M. Isobe, E. Takayama-Muromachi, T. Muranaka, and J. Akimitsu,Phys. Rev. B E. H. P. Cordfunke and G. Meyer, Recueil des Travaux Chimiques des Pays-Bas J. P. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. C. Martins, M. Aichhorn, L. Vaugier, and S. Biermann, Phys. Rev. Lett. H. Watanabe, T. Shirakawa, and S. Yunoki, Phys. Rev. Lett. K. M. Shen, F. Ronning, W. Meevasana, D. H. Lu, N. J. C. Ingle, F. Baumberger, W. S. Lee,L. L. Miller, Y. Kohsaka, M. Azuma, M. Takano, H. Takagi, and Z.-X. Shen, Phys. Rev. B ,075115 (2007). P. D. C. King, T. Takayama, A. Tamai, E. Rozbicki, S. M. Walker, M. Shi, L. Patthey, R. G.Moore, D. Lu, K. M. Shen, H. Takagi, and F. Baumberger, Phys. Rev. B S. Moser, L. Moreschini, A. Ebrahimi, B. D. Piazza, M. Isobe, H. Okabe, J. Akimitsu, V. V.Mazurenko, K. S. Kim, A. Bostwick, E. Rotenberg, J. Chang, H. M. Rønnow, and M. Grioni,New J. Phys. , 013008 (2014). M. K. Crawford, M. A. Subramanian, R. L. Harlow, J. A. Fernandez-Baca, Z. R. Wang, and D.C. Johnston Phys. Rev. B Q. Huang, J. L. Soubeyroux, O. Chmaissem, I. Natali Sora, A. Santoro, R. J. Cava, J. J. rajewski, and W. F. Peck Jr., J. Solid State Chem.
355 (1994). See Supplemental Material at [URL will be inserted by publisher] for more information regardingtight binding calculation. R. Comin, G. Levy, B. Ludbrook, Z.-H. Zhu, C. N. Veenstra, J. A. Rosen, Y. Singh, P. Gegen-wart, D. Stricker, J. N. Hancock, D. van der Marel, I. S. Elfimov, and A. Damascelli, Phys.Rev. Lett. Q. Wang, Y. Cao, J. A. Waugh, S. R. Park, T. F. Qi, O. B. Korneta, G. Cao, and D. S. Dessau,Phys. Rev. B P. Lunkenheimer, T. Rudolf, J. Hemberger, A. Pimenov, S. Tachos, F. Lichtenberg, and A.Loidl, Phys. Rev. B
Figures
FIG. 1: (Color online) (a) Structure view of Ba IrO . (b) Temperature dependence of the longi-tudinal in-plane resistivity of a 16 nm Ba IrO film grown on PrScO . IG. 2: (Color online) RHEED images of (a) the bare PrScO substrate and (b) after the growthof a 16 nm thick Ba IrO film, taken along the [110] ( p ) azimuthal direction. (c) RHEED intensitycurves integrated within each rectangle window. (d)-(f) The results for the [100] ( p ) direction. IG. 3: (Color online) LEED images of a Ba IrO film, taken along the [001] direction with anelectron energy of (a) 150 and (b) 200 eV. (c) XRD θ –2 θ scan of a 16 nm thick Ba IrO film grownon PrScO substrate. Substrate peaks are marked with an asterisk. IG. 4: (Color online) Valence band photoemission spectra at (a) (0 , π, π, π )points for Ba IrO , respectively, taken at 100 K to prevent charging, showing Ir 5 d t bands and O2 p bonding and nonbonding states at higher binding energies. (d) Magnified low-energy electronicstructures along (0 , π, π, π )–(0 ,
0) high-symmetry lines of the tetragonal Brillouin zone. IG. 5: (Color online) (a) Near- E F ARPES spectral intensity plot along high-symmetry directions,taken at 100 K. Extracted EDC and MDC peak positions are overlaid with circles with error bars.DFT calculations including the spin-orbit coupling with U of (b) 0 and (c) 2.5 eV. The J eff characteris calculated by projecting the eigenstates onto the J eff = 1 / IG. 6: (Color online) (a) Unsymmetrized isoenergy map at 0 . ± .
005 eV binding energy, (b)crystal structure projected on (001), and (c) E versus k spectra along high-symmetry lines forBa IrO . Dashed squares in (a) and (b) indicate the in-plane Brillouin zone and unit cell, re-spectively. (d)-(f) The corresponding data from a Sr IrO thin film grown on LSAT is shown forcomparison. IG. 7: (Color online) (a) Temperature-dependent EDCs at momentum positions A ∼ ( π/ , π/ ∼ ( π, ∼ ( π, π ) as sketched in (c). The spectra were measured by cycling samples from112 K to 299 K and then back to 103 K. Fitting results (one Gaussian curve and a background) for103 K data are also represented by black curves. (b) Temperature evolution of the peak position(upper panel) and width (lower panel) for A–C. (c) Symmetrized isoenergy map at 0.4 eV in thefull Brillouin zone, integrated within a ± IG. 8: (Color online) Schematic illustration of the temperature dependence of (a) the near- E F band dispersion and (b) the in-plane spin state in Ba IrO . Dashed curves represent the originalbranch without including the electron correlation. Energy dispersions affected by (c) weak or (d)strong correlation are also shown.. Dashed curves represent the originalbranch without including the electron correlation. Energy dispersions affected by (c) weak or (d)strong correlation are also shown.