Inverse photoemission spectroscopic studies on phase separated La 0.2 Sr 0.8 MnO 3
aa r X i v : . [ c ond - m a t . s t r- e l ] N ov Inverse photoemission spectroscopic studies on phase separated La . Sr . MnO Navneet Singh , M. Maniraj , J. Nayak , S.K. Pandey , and R. Bindu ∗ School of Basic Sciences, Indian Institute of Technology Mandi, Kamand, Himachal Pradesh- 175005, India UGC-DAE Consortium for Scientific Research, Khandwa Road, Indore- 452001 School of Engineering, Indian Institute of Technology Mandi, Kamand, Himachal Pradesh- 175005, India (Dated: July 16, 2018)We have studied the temperature evolution of the inverse photoemission spectra of phase sepa-rated La . Sr . MnO . To identify the features in the room temperature experimental spectra, bandstructure calculations using Korringa-Kohn-Rostoker Green’s function method were carried out. Wefind that the features generated by local moment disorder calculations give a better match with theexperimental spectrum. In the insulating phase, we observed unusually an increased intensity ataround the Fermi level. This puzzling behaviour is attributed to the shift in the chemical poten-tial towards the conduction band. The present results clearly show the importance of unoccupiedelectronic states in better understanding of the phase separated systems. PACS numbers: 71.27.+a,75.47.Lx,71.15.Mb
I. INTRODUCTION
In strongly correlated electron systems , the subtleinterplay between charge, spin, lattice and orbital de-grees of freedom stabilize emergent phenomena like phaseseparation, colossal magneto resistance, superconduc-tivity, multiferroicity, etc. The material under study,La . Sr . MnO is one such kind of system, which re-veals structural phase separation at the nanoscale . Atroom temperature, this material stabilizes in a simplestructure namely cubic. Like any other compounds inthe Sr doped series, the compound under study does notlie in any of the phase boundary region thereby avoidingany complications due to the coexistence of phases ly-ing in the phase boundary region . Hence it is expectedthat this material provides a better platform to unravelits properties. This material undergoes structural, elec-tronic and magnetic phase transitions; cubic to tetrago-nal, metal to insulator, paramagnetic to C-type antifer-romagnetic respectively, all at around Neel temperature(T N = 265 K) . Our results based on tempera-ture dependent transmission electron microscopy (TEM),synchrotron based x-ray diffraction (XRD), high resolu-tion photoemission spectroscopy revealed inhomogenitiesat the nanoscale. To characterize such inhomogenities,it is important to obtain the behaviour of the chemi-cal potential as a function of the thermodynamic vari-able.Photoemission spectroscopy is one such tool whereone can get the information of the behaviour of chemicalpotential shift .In the case of photoemission spectroscopy, (i) the in-formation of chemical potential shift is obtained basedon the shift in the binding energy. There are severalfactors which contribute to the shift in the binding en-ergy apart from the chemical potential shift . These arechemical shift, madelung potential, screening potential,relaxation energy; (ii) There occurs ambiguity in extract-ing the binding energy shift if more than one unresolvedfeatures contribute to the core level. Such unresolvedfeatures occur due to various screening channels. So es- sentially, the information obtained about the chemicalpotential shift is rather indirect.Inverse photoemission spectroscopy is one such tech-nique which gives information of the unoccupied part ofthe electronic structure. It also gives a better insightinto the behaviour of the chemical potential when com-pared with the results of the photoemission data, as nosuch complications will occur when extracting the infor-mation of the chemical potential shift. Until now, mostof the work carried out to extract chemical potentialshift are based on the core level binding energy shift. Inthe systems which show electronic phase separation, thefractions of the coexisting phases vary in such a way thatthe chemical potential of both the phases remains thesame. Keeping this in mind, it is important to extractthis parameter unambiguously. We show here that thisparameter can be extracted unambiguously using inversephotoemission spectroscopy. Apart from this, it is im-portant to bear in mind that band structure calculationsgive information about the ground states.But the infor-mation provided by inverse photoemission spectroscopyis about the excited states. It has always been chal-lenging to match the features generated by the excitedstates. The situation becomes more challenging, if thesystem is paramagnetic and belongs to the category ofstrongly correlated electron system. Here, we show thatlocal moment disorder calculation gives a better repre-sentation of the features in the experimental inverse pho-toemission spectra. This motivated us to carry out tem-perature dependent inverse photoemission spectroscopic(IPES) studies on phase separated La . Sr . MnO . II. EXPERIMENTAL DETAILS
Polycrystalline La . Sr . MnO was prepared by con-ventional solid state route. Details of the sample prepara-tion and characterisation are given elsewhere . To studythe unoccupied density of states inverse photoemissionexperiments (IPES) were carried out on La . Sr . MnO -0.5 0.0 0.5 1.0 (a) I n t e n s i t y ( a . u . ) Energy (eV) B C RT100 K A RT100 K (b) iiiiii μ i μ f A FIG. 1: (a) The inverse photoemission spectra collected atRT and 100 K. The inset shows the comparison of the spec-tra around the region of Fermi level, (b) Schematic displayingthe shift in the chemical potential when the material under-goes transition from metal (panel (i)) to insulator. The panel(ii) shows the chemical potential, µ i (black dot) in the in-sulating phase lying at the middle of the band gap. Thepanel(iii) shows the chemical potential, µ f (magenta dash) inthe insulating phase shifted towards the conduction band.Therectangular block represents the region up to which the den-sity of states contribute at the Fermi level due to instrumentalbroadening of 0.55 eV. in ultra high vacuum at pressure 10 − mbar. The spec-imen was scrapped in vacuum using a diamond file inorder to get rid of the surface contamination. An elec-tron gun of Stoffel Jhnson design and an acetonefilled band pass detector with CaF window is used forthe experiments. The experiments were carried outin the isochromat mode where the energy of the inci-dent electron was varied in steps of 0.1 eV and photonsof fixed energy (9.9 eV) were detected with an overallinstrumental resolution of 0.55 eV. The electron beamcurrent variation as a function of kinetic energy was ac-counted for by normalizing the measured counts by thesample current at each step, as in our previous studies. III. COMPUTATIONAL DETAILS
The non-magnetic, ferromagnetic and local momentdisorder (LMD) calculations for La . Sr . MnO werecarried out by using Korringa-Kohn-Rostoker (KKR)Green’s function method. In the compound understudy 80% Sr is doped at La site. These doping effectswere studied well under the coherent potential approxi- (a) (b)
O 2p Mn 3d T D O S ( s t a t es / e V / un i t ce ll ) (c) P D O S ( s t a t es / e V / a t o m ) La 4f Sr 5d
Energy (eV)
FIG. 2: (a) The total density of states (TDOS) obtained basedon non magnetic calculations, (b) The black dot and red linerepresent O 2 p and Mn 3 d partial density of states (PDOS),respectively and (c) The blue line and magenta dash dot rep-resent La 4 f and Sr 5 d PDOS, respectively. mation (CPA). As the crystal structure is cubic, latticeparameter used for calculation are a=7.18886 Bohr. Muf-fin tin radii for La/Sr, Mn and O 3.064 Bohr, 1.566 Bohrand 2.05 Bohr, respectively are used for the calculations.The exchange correlation function used for the calcula-tion were taken after Vosko, Wilk and Nusair. The selfconsistency was achieved by demanding convergence oftotal energy to be less than 10 − Ryd/cell.
IV. RESULTS AND DISCUSSIONS
Room temperature (RT) inverse photoemission spec-trum for La . Sr . MnO is shown in Fig. 1a.The Fermilevel ( ǫ F ) is marked as E = 0 eV. In this spectrum, thereare three clear features A, B and C observed at and above ε F . A small kink upto around 0.7 eV is labeled as Aand two broad features are labeled as B and C. Peak Bis centered around 2.3 eV and spread up to about 4.5eV, whereas the peak C is centered around 8.8 eV andspread upto 13 eV. To identify these features band struc-ture calculations were carried out.To begin with, we performed non magnetic calcula-tions to understand the paramagnetic phase displayedby this compound at RT. The results indicate three im-portant features in the total density of states (TDOS),Fig.2a. The feature 1 which is close to the ε F is centeredaround 0.1 eV; feature 2 is spread in the region 2 to 3 eVand the third one in the broad region from 5 to 11 eV.The partial density of states as displayed in Figs.2b & (a) 'a 'b (b) T D O S ( s t a t es / e V / un i t ce ll ) O 2 p Mn 3 d (c) P D O S ( s t a t es / e V / a t o m ) La 4 f Sr 5 d Energy (eV)
FIG. 3: (a) The total density of states (TDOS) obtained basedon ferromagnetic calculations, (b) The black dot and red linerepresent O 2 p and Mn 3 d partial density of states (PDOS),respectively and (c) The blue line and magenta dash dot rep-resent La 4 f and Sr 5 d PDOS, respectively. c indicate that the feature 1 corresponds mainly to Mn3 d states with negligible contribution from O 2 p states;the feature 2 corresponds to La 4 f and the feature 3 tomainly Sr 5 d states. On comparing the calculated DOSwith experimental spectrum we observe that only fea-tures 2 and 3 are fairly matching with the peaks B andC in the experimental spectrum. The discrepancy in thematching of feature 1 with peak A is mainly due to thenon-magnetic calculations which are based on the itiner-ant model of paramagnetism. In the itinerant model ofparamagnetism, no Hund’s like on-site exchange interac-tion at magnetic ion (Mn) site are taken into account.But for the material under study it is important to in-voke the contribution of on-site exchange interaction asthis belongs to strongly correlated electron systems wherethe Mn 3 d electrons are highly localised.As a next step; to capture the localized picture of3 d -electrons and to include the exchange interactions,we performed ferromagnetic calculations, Fig.3. In theTDOS, we observe three main features labeled as 2 ′ a ,2 ′ b and 3 ′ . The features 2 ′ a &2 ′ b are extended in the en-ergy range 0.5 to 4 eV and the feature 3 ′ is spread inthe region from 6 to 11 eV. The feature 2 ′ a has strongcontribution from Mn 3 d and weak contribution from O2 p states. The feature 2 ′ b has main contribution fromLa 4 f states with small contribution from Mn 3 d states.The feature 3 ′ is mainly contributed by Sr 5 d states andnegligible contribution from O 2 p states. On compar-ing the calculated DOS with the experimental spectrum (c)(b) P D O S ( s t a t es / e V / a t o m ) T D O S ( s t a t es / e V / un i t ce ll ) (a) O 2p Mn 3d
La 4f Sr 5d
Energy (eV)
FIG. 4: (a) Total density of states (TDOS) obtained based onLMD calculations, (b) The black dot and red line represent O2 p and Mn 3 d partial density of states (PDOS), respectively.The inset shows the contribution from O 2 p , Mn 3 d PDOSclose to the Fermi level and (c) The blue line and magentadash dot represent La 4 f and Sr 5 d PDOS, respectively. we observe, only peak B matches fairly good with fea-tures 2 ′ a and 2 ′ b and peak C with feature 3 ′ . No featurecorresponding to peak A has been generated in the calcu-lated DOS. This discrepancy could be due to the follow-ing reason. In the ferromagnetic calculations, there arecontributions from both on-site and inter-site exchangeinteractions. The former interaction gives rise to localmagnetic moment while the latter gives rise to long rangemagnetic order. The compound under study is paramag-netic at room temperature, so the missing of the featurecorresponding to peak A could be due to the long rangemagnetic ordering generated by the calculation.To capture the localized picture of d -electrons withoutinter-site exchange interactions, local moment disorder(LMD) calculations were carried out, Fig.4. We observefour features 1 ′′ , 2 ′′ a ,2 ′′ b and 3 ′′ in the calculated TDOS.The feature 1 ′′ is observed as small kink upto 0.5 eV,features 2 ′′ a &2 ′′ b are extended in the energy range 0.6to 5 eV and the feature 3 ′′ is spread in the region from6 to 11 eV. On comparing the calculated DOS with theexperimental spectrum, we attribute feature 1 ′′ to Mn and O hybridized states; the features 2 ′′ a and 2 ′′ b toMn and La states, respectively and the feature 3 ′′ mainly to Sr states. The features generated in thecalculations are matching fairly good with peaks A, Band C of experimental spectrum.Having identified the features in the room temperaturespectrum, we now discuss the results of the tempera-ture dependent spectra displayed in Fig.1a. On reducingthe temperature to 100 K, we find an increased inten-sity upto about 0.9 eV above ε F as shown in the inset ofthe figure. The studies on temperature dependent highresolution photoemission spectra have shown that hardgap is opened up only at 200 K even though metal toinsulator transition found from the resistivity measure-ments occurs at 265 K. In this situation, it is normallyexpected that for the inverse photoemission spectra, theintensity at around ε F should also decrease. But we ob-serve opposite behaviour which is puzzling. In the firstinstance, several possibilities come into the mind whenone observes such increase in the intensity. The pos-sibilities could be either due to (a) charging effects or(b) destruction of insulating state as a result of electronirradiation or (c) current induced phase transition atthe surface or (d) chemical potential shift towards theconduction band.After careful analysis of the data we rule out the firstpossibility due to the following reasons; (i)At 100 K, thecharging effect was not observed even in the X-ray pho-toemission spectra ; (ii)Had there been charging then onewould have expected the spectrum to be shifted towardshigher energy. So in this case one would expect depletionof intensity at the Fermi level which is not the case we ob-serve; (iii) Apart from this, we also observe the positionof the higher energy peaks at RT and 100 K i.e. around 3and 8 eV to be almost the same despite the material arein two different electronic phases namely room temper-ature metallic and low temperature (100 K) insulatingphase. .The second possibility can be ruled out based on ourearlier high resolution XRD and TEM results , . Thiscompound exhibits co-existence of charge ordered andtwinned phase in a wide temperature range (even at 300K) which becomes more prominent in the temperaturerange 260 K to 200 K. With decrease in temperature, itis found that the fractions of the twinned phase increaseat the cost of the charge ordered phase and at 100 K thereis about 10 % of charge ordered phase. In the light of theabove fact, there is no question of melting of charge or-dering at 100 K due to irradiation as the electron energyused in TEM is more than 100 times higher than thatused in the IPES experiments. Had there been meltingof charge ordering due to electron irradiation, we wouldnot have observed robust charge ordered phase in theTEM experiments. If we consider the third possibility,then the intensity at ε F for 300 K will be more than thatin 100 K which is opposite to the experimental observa-tion. Thus one can also rule out the third possibility.Regarding the fourth possibility, it is important torecollect our photoemission results . The signatureof above phase separation was also confirmed based onthe behaviour of the chemical potential shift calculatedfrom the binding energy shift of the core level spectra.Even though the metal to insulator transition occurredat around 265 K, the chemical potential shift towardshigher binding energy was observed below 200 K. Thepinning of the chemical potential shift in the tempera-ture range 300 to 200 K indicates electronic phase sepa- ration. In the region of phase separation, pseudo gap wasobserved and on further reducing the temperature hardgap was opened in the antiferromagnetic phase. The ob-served shift in the binding energy in the PES data hasbeen explained in terms of shift in the chemical potentialtowards the conduction band . The observed incrementin the intensity of peak A in the IPES data at 100 Kcan also be understood based on the shift in the chem-ical potential towards the conduction band as explainedbelow.The pseudogap was observed in the phase separatedregion, we assume the behaviour of the spectra closeto ε F (both in the occupied and unoccupied region) toa parabola. So as the compound enters the insulatingphase, states at ε F undergo depletion. In Fig. 1b, thestates contributing to the intensity at ε F due to the in-strumental broadening (0.55 eV) is represented by theregion marked by the rectangular block. The panel (i)shows finite density of states at ε F corresponding to RTmetallic phase. Here, the states within the rectangularblock are expected to contribute at ε F which leads tofinite intensity at ε F . As the material enters into the in-sulating phase, the hard gap is opened. The states in theregion marked by the rectangular block are expected tocontribute at ε F , panel (ii). So in the IPES data one ex-pects decrement in the intensity at ε F with the chemicalpotential lying in the middle of the gap. Such behaviouris not in line with the experimental result, inset of Fig1a. We now discuss the situation where the chemicalpotential is shifted towards the conduction band, panel(iii). When such shift is taken into account, the contri-bution of the states belonging to the rectangular blockis expected to be more at ε F . This leads to the increasein the intensity of peak A with decrease in temperature.It is important to point out that, in the ultraviolet pho-toemission spectra , it is essentially because of the highresolution (5 meV) of the photoemission set up, a cleargap was observed at low temperature. The band gap at130 K using PES was estimated to be of about 150 meV,under the assumption that there is no chemical potentialshift. But as we have observed chemical potential shiftat 100 K, the band gap is expected to lie between 75 to150 meV. V. CONCLUSIONS
Inverse photoemission spectra were collected onLa . Sr . MnO at 300 and 100 K. The features in theinverse photoemission spectra were identified using bandstructure calculations and it was found that local momentdisorder calculation gives a better match with the experi-mental spectra. As the compound becomes insulating, ataround ǫ F , an increase in the intensity was found. Thisbehaviour is in contrast with the results of the high res-olution photoemission spectra. After detailed analysis,we find that this unusual behaviour can be understoodbased on the shift in the chemical potential. The presentwork clearly suggests the importance of occupied and un-occupied electronic states to understand the behaviour ofsystems with electronic phase separation. VI. ACKNOWLEDGEMENT
The authors thank Dr. S.R. Barman, UGC-CSR, In-dore for fruitful discussions. ∗ Corresponding author: [email protected] J. B. Goodenough, Rep. Prog. Phys. , 1915 (2004);Y.Tokura, Phys. Today (2003) 50 ;E. Dagotto et al. ,Physics Reports (2001) 1 ;T. V.Ramakrishnan, Cur-rent Science (2008) 1284 ;K.F. Wangab, J.-M. Liuaband Z.F. Renc (2009) 321 R. Bindu, Sanjay Singh, Navneet Singh, Rajeev Ranjan,Kalobaran Maiti, A. H. Hill and S. R. Barman,Phys. Rev.B. (2012) 140104(R) J. Hemberger, A. Krimmel, T. Kurz, H. A. KrugvonNidda,V. Y.Ivanov, A. A. Mukhin, A. M. Balbashov, and A.Loidl, Phys. Rev.B (2002) 094410 . O. Chmaissem et al. , Phys. Rev. B. (2003) 094431 . R. Bindu, K. Maiti, R. Rawat, and S. Khalid, Appl.Phys.Lett. (2008) 121906 . R. Bindu, G. Adhikary, Nishaina Sahadev, N.P.Lalla andK. Maiti,Phys. Rev. B. (2011) 052407 . R. Bindu, G. Adhikary, S. K. Pandey, S. Patil, and K.Maiti,New J. Phys. (2010) 033003 . P. G. Steeneken, L. H. Tjeng, G. A. Sawatzky, A. Tanaka,O. Tjernberg, G. Ghiringhelli, N. B. Brookes, A. A. Nu-groho, and A. A. Menovsky, Phys. Rev. Lett. (2003)247005 ; A. Fujimori,K.Morikawa,A. Ino, andT.Mizokawa,J. Electron Spectrosc.Relat. Phenom. (1996) 31 ; K.Maiti, J. Fink, S. de Jong, M. Gorgoi, C. Lin, M. Raichle,V. Hinkov, M. Lambacher, A. Erb, and M. S. Golden,Phys. Rev. B (2009) 165132 ; N. Harima, J. Matsuno, A. Fujimori, Y. Onose, Y. Taguchi, and Y. Tokura, ibid. (2001) 220507(R) . R. Bindu, Eur. Phys. J. B (2004) 321 D.Funnemann and H.Merz,J. Phys. E (1985) 554 . M. Maniraj and S. R. Barman, Rev. Sci. Instrum. (2014) 033301 . S. Banik, A. K. Shukla, and S. R. Barman, Rev. Sci. In-strum. (2005) 066102 . A. K. Shukla, S. Banik, and S. R. Barman, Curr.Sci. (2006) 490 . M. Maniraj, A. Rai, S. R. Barman, M. Krajci, T. Lograsso,D. L. Schlagel and K. Horn, Phys. Rev. B (in press); S. K.Pandey, A. Kumar, S. Banik, A. K. Shukla, S. R. Barman,and A. V. Pimpale, Phys. Rev. B , 113104 (2008); R.Kundu, P. Mishra, B. R. Sekhar, M. Maniraj, and S. R.Barman, Physica B , 827 (2012); R. S. Dhaka, S. Banik,A. K. Shukla, V. Vyas, Aparna Chakrabarti, S. R. Barman,B. L. Ahuja, and B. K. Sharma, Phys. Rev. B , 073107(2008). H. Akai and P. H. Dederichs, Phys. Rev. B (1993) 8739. http://kkr.phys.sci.osaka-u.ac.jp/ S. H. Vosko, L. Wilk, and M. Nusair, Can. J. Phys. (1980) 1200. M. Hervieu, A. Barnabe, C. Martin and and B. Raveau,Phys. Rev. B60