Polaronic state and nanometer-scale phase separation in colossal magnetoresistive manganites
Sahana Roessler, S. Ernst, B. Padmanabhan, Suja Elizabeth, H. L. Bhat, F. Steglich, S. Wirth
aa r X i v : . [ c ond - m a t . s t r- e l ] M a y Polaronic state and nanometer-scale phase separation in colossal magnetoresistivemanganites
Sahana R¨oßler, S. Ernst, B. Padmanabhan, Suja Elizabeth, H. L. Bhat, F. Steglich, and S. Wirth Max Planck Institute for Chemical Physics of Solids, N¨othnitzer Straße 40, 01187 Dresden, Germany Department of Physics, Indian Institute of Science, Bangalore 560012, India (Dated: October 29, 2018)High resolution topographic images obtained by scanning tunneling microscope in the insulatingstate of Pr . Pb . MnO single crystals showed regular stripe-like or zigzag patterns on a widthscale of 0.4–0.5 nm confirming a high temperature polaronic state. Spectroscopic studies revealedinhomogeneous maps of zero-bias conductance with small patches of metallic clusters on lengthscale of 2–3 nm only within a narrow temperature range close to the metal-insulator transition. Theresults give a direct observation of polarons in the insulating state, phase separation of nanometer-scale metallic clusters in the paramagnetic metallic state, and a homogeneous ferromagnetic state. PACS numbers: 71.30.+h, 75.47.Lx, 68.37.Ef
There is an intense research going on to understandthe remarkable and complex properties shown by a wholefamily of strongly correlated electron systems. In thesematerials, a self-inflicted, spontaneous instability of theelectronic state and competing long-range interactionsmay result in the formation of nanometer-sized regionsof different phases. Such states have been considered ascharge and spin ordered stripes in under-doped cuprates[1], as polar domains in relaxor ferroelectrics [2], or asphase separation (PS) between insulating paramagnetic(pm) and conducting ferromagnetic regions [3] in mixedvalence manganites of perovskite type A MnO ( A –rareearth or doped divalent ion). In the latter case, evidencefor PS has been found by various experimental tech-niques, such as electron microscopy [4], scanning tunnel-ing microscopy/spectroscopy (STM/S) [5, 6], magneticforce microscopy [7] and photoelectron spectroscopy [8].These experiments showed inhomogeneities of randomshape on a length scale of several hundred nanometers.Further, the PS persisted deep into the metallic state insome of these manganites. However, computational stud-ies on models of manganites considering double exchange,Jahn-Teller (JT) interaction and long range Coulomb po-tential could show only regularly spaced nanometer-scalePS [9]. The random location and shape of the clusters ob-served experimentally [4, 5, 6, 7, 8] are conjectured to becaused by quenched disorder in the couplings induced bychemical substitution [10, 11]. Recent STS studies com-bined with transmission electron microscopy [12] on A -site ordered and disordered La . Ca . MnO thin filmsshowed that PS persists in the metallic state only in thedisordered film. But, this study does not address thequestion of PS at the metal-insulator transition tempera-ture, T MI . Thus, the origin of the PS, the length scale in-volved, the role of quenched disorder originating from therandom A -site substitution, and the temperature rangeat which PS occurs, remain all strongly debated.The polaron effect due to strong JT electron-phononcoupling is considered central to understand the remark- able transport properties, specifically the colossal mag-netoresistance (CMR), of manganites [13]. The hightemperature polaronic state in reciprocal space has beenprobed experimentally [14, 15] revealing complex polaroneffects such as polaron correlation, polaron ordering, andcharge localization. Using STM, polarons can be imageddirectly in the real space. Polaron confinement was re-cently observed in a layered manganite single crystal us-ing STM [16]. Charge ordering was reported for thinfilms (La / − x Pr x )Ca / MnO [short range charge ex-change (CE) type] in the pm state [17] and for highlydoped Bi . Ca . MnO single crystals [18].Spatially resolved STS measurements [5] on thin filmsof La . Ca . MnO on SrTiO showed coexistence ofregions with metallic, insulating as well as intermediateconductivities, extending over several hundred nanome-ters. However, these images were obtained at a ratherhigh fixed bias voltage of 3 V (much larger than the semi-conducting gap of 0.2–0.3 V in manganites) and may notreflect the ground state properties. On the other hand,in Ref. 6 the zero-bias conductance, G = dI/dV | V =0 , ofLa . Sr . MnO /MgO thin films was mapped as a func-tion of temperature T , and a threshold criterion was ap-plied to distinguish metallic and insulating regions. Sucha threshold criterion will not give an unambiguous evi-dence for the existence of PS because any statistical dis-tribution of conductance, whose average value shifts with T , will seem to show PS [19]. Further, in thin film sam-ples, miss-fit strain induced by the substrates seems toinfluence the electrical properties [20].To resolve some of these issues from experimental side,we carried out STM/S on Pr . Pb . MnO (PPMO)single crystals providing largely strain-free materials. Weaddress two important questions in the physics of man-ganites, namely, the high temperature polaronic stateand the nanometer-scale electronic PS. We present STMimages of polaronic Mn -sites and doped hole localiza-tion on Mn -sites with atomic resolution. In addition,we provide clear evidence for nanometer-scale PS and T (K) M agne t i z a t i on ( A m / K g ) T MI T C
150 200 250 300 350-100102030405060 ( d / d T ) / ( / T ) T (K) R e s i s t i v i t y ( c m ) T2T5
T 9 T FIG. 1: Temperature dependence of magnetization (left scale)and resistivity (right scale) of Pr . Pb . MnO single crys-tals measured at different magnetic fields. Inset: The loga-rithmic derivative ( dρ/dT ) / ( ρ/T ) as function of T at H = 0. percolation just below T MI by looking at the entire dis-tribution of G and its dependence on T . Information onthe length scale of the inhomogeneities and the T rangewithin which it appears is obtained. We discuss the roleof quenched disorder or doping and compare the resultswith macroscopic properties of the same single crystal.Single crystals of PPMO used for the present studywere taken from a batch of crystals, whose prepara-tion and properties were already reported in [21, 22].In Pr − x Pb x MnO , the Curie temperature T C and T MI do not coincide, and metal-like conductivity occurs ina pm state in parts of the phase diagram [21, 23], aphenomenon uncommon to mixed valence manganites.Fig. 1 shows the temperature dependence of magnetiza-tion ( M ), resistivity ( ρ ) and magnetoresistive propertiesof a PPMO sample. The magnetoresistance, [ ρ ( H ) − ρ (0)] /ρ (0), is found to be ∼
90% close to T MI under afield of 9 T. From the maximum change in slope of the M vs. T curve, T C ≈
210 K was estimated which isabout 45 K lower than the corresponding T MI ≈
255 K.Such an approach to estimate T C is supported by elab-orate investigations on a similar Pr . Pb . MnO singlecrystal in which the so-determined T C ∼
197 K agreeswell with results from detailed static magnetization scal-ing analysis [22] as well as heat capacity measurements[24] ( T MI ≈
235 K in this compound [21]). The scal-ing analysis embracing the critical temperature indicatedthat the underlying magnetic transition is a conventionalone, with short-range Heisenberg-like critical exponents.This study emphasizes on the presence of additional frus-trated couplings which intercepts the formation of longrange order. Deviation of the inverse susceptibility fromthe Curie-Weiss law above T C [21] and history-dependenttransport properties [23] suggest a presence of small mag-netic metallic clusters above T C that form percolatingmetallic paths upon reducing T in the pm metallic state. he i gh t ( n m ) distance (nm) I ( p A ) z (pm) b) c)
50 nm n m a ) FIG. 2: (a) Dependence of tunneling current I on relativetip-sample distance z on a semi-logarithmic scale. (b) Surfacetopography over an area 50 ×
50 nm . (c) Height profile alongthe white line drawn in inset (b). Note that evidence for the formation of localized ∼ T C in another mixed valentmanganite has earlier been found by small-angle neu-tron scattering measurements [25]. We also note thesharpness of the resistance transition which can be in-ferred from the logarithmic derivative of the resistance( dρ/dT ) / ( ρ/T ) plotted in the inset of Fig. 1. Such asharp metal-insulator transition is indicative of a strain-free sample [20] (the tolerance factor, t = 0.965, is closeto unity indicating good ionic size match).For the tunneling studies a STM (Omicron Nanotech-nology) under ultra high vacuum conditions ( p ≤ − mbar) was utilized at eleven fixed temperatures, 30 K ≤ T ≤
300 K, mostly in the vicinity of T C and T MI . Sincecrystals with perovskite structure do not cleave easily,preparation of a clean surface for the STM is a challenge.Just before inserting the crystal into the UHV chamber,we thoroughly cleaned the crystal surface in isopropanolusing an ultrasonic bath and then, inside isopropanol,scraped the surface to rip off some part of the surface.This preparation gave us clean surfaces on a length scaleof microns. STM was conducted using tungsten tips, andtypically 0.3 nA for the current set point and 0.8 V forthe bias voltage, V . This implies that we probed the un-occupied electronic DOS of PPMO. Fig. 2(a) shows thedependence of tunneling current I on relative tip-sampledistance z on a semi-logarithmic plot. The exponentialnature of I ( z ) confirms an excellent vacuum tunnel bar-rier (effective work function φ ∼ ×
50 nm ) is presented in inset (b). Terraces with unitcell height ( ∼ h i surface of the pseudocubic perovskite crystal.High resolution STM images taken in the insulatingregime (at 300 K) on the terraces indicate bright anddark regions forming stripe-like features spread over alength scale of 0.4–0.5 nm, as seen in Fig. 3(a), (b). Whileprobing the unoccupied electronic DOS the doped holes FIG. 3: (a), (b) Topography of different areas in the insu-lating regime ( T = 300 K). (c) and (d) show correspondingintensity profiles along the white lines drawn in (a) and (b),respectively. The bright and dark spots in the image are as-sociated with Mn and Mn ions, respectively. localized on Mn sites appear as bright spots in theSTM image, whereas electron tunneling from a conven-tional metallic tip into a polaronic state (e.g. electronslocalized on Mn ions) is difficult and produces darkspots [17]. However, these contrasts were seen only occa-sionally, an observation similar to what was reported inthe case of layered manganite La − x Sr x Mn O [16].This suggests the short-range stripe-like order of Mn and Mn ions. The extent of these features [Fig. 3(c)and (d)] is slightly larger than the typical atomic distanceof ∼ . Ca . MnO [26]. Short-range po-laron correlation and CE-type of charge ordering was ob-served in manganites using diffused x-ray and neutronscattering [14, 15]. Recent STM studies [17] probingsimultaneously the occupied and unoccupied states of(La / − x Pr x )Ca / MnO thin films also showed short-range CE-type charge ordered clusters in the pm state.To map the surface electronic state, we carried outthousands of STS measurements at different locations onthe sample surface spanning the T range from 28 – 300K. Typically, a surface area of 50 ×
50 nm with a lateralresolution of 1 nm (2500 pixels) was investigated. Tun-neling current and differential conductance, G = dI/dV ,were measured simultaneously while ramping V from − G - V curves taken at rep-resentative T = 30, 221, 260 and 300 K are shown inFig. 4(a). At 30 K, the G - V curve is metal-like with afinite value of G signifying a finite DOS at the Fermienergy. In contrast, at 300 K, i.e. T > T MI ≈
255 K, the G - V curve around V = 0 is typical of a semiconductinggap. In the pm metallic state at 221 K, the value of G is only slightly reduced compared to G ( T = 30 K), thusshowing an expected T dependence of the conductance. FIG. 4: (a) G - V curves averaged over an area of 50 ×
50 nm at 30, 221, 260 and 300 K indicating metallic ( T < T MI ≈ T > T MI ). Panels (b)–(e)present conductance maps (50 ×
50 nm ) taken at 30, 199,243 and 300 K through T C and T MI . The color scale shownleft of panel (b) encodes G . (g)–(j) Histograms of G at thesame T as presented in (b)–(e). (f) Histograms for the threedifferent T regimes compared on a semi-logarithmic scale. For quantifying the STS results and mapping the ho-mogeneity of the DOS laterally as well as its temperatureevolution, we plot G as conductance maps at selected T . In Fig. 4(b)–(e), the local G is presented color-coded(with a color scale covering 0 ≤ G ≤ T = 30,199, 243 and 300 K, respectively. Corresponding distri-butions for the frequency of the observed G values areshown in the histograms Fig. 4(g)–(j). A sharp distri-bution of G at 30 K confirms a homogeneous electronicphase at low temperature. Similarly, the conductancemap at 300 K (in the semiconducting region) is also ho-mogeneous [Fig. 4(e)], with most of the values of G veryclose to zero [Fig. 4(j)].On the other hand, as T is raised through T C and ap-proaches T MI ≈
255 K inhomogeneities start to developat a length scale of 2 – 3 nm, as seen in Fig. 4(c) and(d). The peak in the histograms [panels (h), (i)] shiftsto lower conductance values as T is increased and, im-portantly, an increasing weight at G → G at T = 221 K is clearlyvisible in Fig. 4(f), with two maxima in G frequencylocated at similar G values as for low and high T , re-spectively. The distributions near T MI [cf. Fig. 4(f)] aresignificantly broadened compared to both, low T (30 K)and high T = 300 K > T MI . The sharp distribution at T = 300 K clearly indicates that these broad distribu-tions of G at intermediate T reflect a sample propertyrather than an instrumental influence.The T dependences observed in STS arise not onlyfrom the Fermi function, but also the sample’s DOS it-self is T dependent. This change of electronic proper-ties can be explained by the release of lattice distortionsaround T MI , when the immobilized polaronic carriers be-come successively mobile producing inhomogeneous spa-tial conductance distributions and electronic transportthrough percolating metallic regions. Thus, the increas-ing weight at G → G ∼ T C . T < T MI in this compound. The drastic change in G and its dis-tribution with T at around the bulk T MI indicates thatour STS results are not mere surface effects.Our results are distinct from previous experimental re-sults where PS is seen on a micrometer scale and persistedwell within the metallic regime. It remains an open ques-tion, whether the particular properties of PPMO with ametallic pm state in the region T C < T MI are responsiblefor the clear observation of this nanometer-scale PS phe-nomenon and whether the result can be generalized toother mixed-valence manganites (as pointed out beforethe electrical transport in PPMO occurs via percolationof nanometer-scale metallic clusters for T C < T < T MI ).Further, the specific pattern of electronic inhomogeneitiyin the local surface DOS is certainly affected by disor-der, induced by random chemical substitutions and/orsurface effects. In addition, disorder effects due to sizedifferences between A -site Pr and Pb ions may playa role. However, the observed nanometer-scale PS is nota simple and fixed result of static chemical disorder, ascan be inferred from the homogeneity of the electronicproperties deep in the metallic state (low T ) as well asin the insulating one (300 K). Hence, in order to resolve the relevance of disorder effects on PS and the associ-ated length scale, similar spatially resolved STS studieson different manganites are called for.In summary, our high resolution STM images providedirect evidence for the high temperature polaronic statein perovskite manganite. Polarons are confined to onelattice cell. Stripe-like features seen occasionally in theseimages suggest a short-range ordering of these polaronsin the form of a lattice. Spatially resolved STS imagesshow nanometer-scale phase separation in the paramag-netic metallic state. However, the homogeneous low- T as well as high temperature STS images confirm thatthis phase separation is limited only to the temperaturesclose to the metal-insulator transition suggesting that itis related to the non-coincidence of T C and T MI .We are grateful to U. K. R¨oßler, Ch. Renner and G.Aeppli for valuable discussions. We acknowledge finan-cial support by the DFG, grant WI 1324/1-1, and theEuropean Commission through CoMePhS 517039. [1] S. A. Kivelson et al. , Rev. Mod. Phys. , 1201 (2003).[2] W. Eerenstein, N. D. Mathur, and J. F. Scott, Nature(London) , 759 (2006).[3] E. Dagotto “Nanoscale phase separation and colossalmagnetoresistance” (Spinger, Heidelberg, 2002).[4] M. Uehara, S. Mori, C. H. Chen, and S.-W. Cheong,Nature (London) , 560 (1999).[5] M. F¨ath, S. Freisem, A. A. Menovsky, Y. Tomioka, J.Aarts, and J. A. Mydosh, Science , 1540 (1999).[6] T. Becker et al. , Phys. Rev. Lett. , 237203 (2002).[7] L. Zhang, C. Israel, A. Biswas, R. L. Greene, and A. deLozanne, Science , 805 (2002).[8] D. D. Sarma et al. , Phys. Rev. Lett. , 097202 (2004).[9] A. L. Malvezzi, S. Yunoki, and E. Dagotto, Phys. Rev.B , 7033 (1999).[10] A. Moreo, S. Yunoki, and E. Dagotto, Science , 2034(1999).[11] A. Moreo et al. , Phys. Rev. Lett. , 5568 (2000).[12] V. Moshnyaga et al. , Phys. Rev. Lett. , 107205 (2006).[13] A. J. Millis, P. B. Littlewood, and B. I. Shraiman, Phys.Rev. Lett. , 5144 (1995).[14] L. Vasiliu-Doloc et al. , Phys. Rev. Lett. , 4393 (1999).[15] C. S. Nelson et al. , Phys. Rev. B , 174405 (2001).[16] H. M. Rønnow, Ch. Renner, G. Aeppli, T. Kimura, andY. Tokura, Nature (London) , 1025 (2006).[17] J. X. Ma, D. T. Gillaspie, E. W. Plummer, and J. Shen,Phys. Rev. Lett. , 237210 (2005).[18] Ch. Renner, G. Aeppli, B.-G. Kim, Y.-A. Soh, and S.-W.Cheong, Nature (London) , 518 (2002).[19] Ch. Renner and H. M. Rønnow, in Scanning Probe Mi-croscopy: Electrical and Electromechanical Phenomenaat the Nanoscale , ed. S. Kalinin (Spinger, Berlin, 2006).[20] M. Paranjape, A. K. Raychaudhuri, N. D. Mathur, andM. G. Blamire, Phys. Rev. B , 214415 (2003).[21] B. Padmanabhan et al. , J. Magn. Magn. Mater. , 288(2006).[22] B. Padmanabhan et al. , Phys. Rev. B , 024419 (2007).[23] R.-W. Li, X. Zhou, B.-G. Shen, and B. Hillebrands, Phys. Rev. B , 092407 (2005).[24] B. Padmanabhan, unpublished.[25] J. M. DeTeresa et al. , Nature (London) , 256 (1997). [26] S. Mori, C. H. Chen, and S. W. Cheong, Nature (London)392