Thin Solid Films 540 (2013) 194–201

Contents lists available at SciVerse ScienceDirect

Thin Solid Films

j ourna l homepage: www.e lsev ie r .com/ locate / ts f

Uniaxial stress influence on electrical conductivity of thin epitaxiallanthanum-strontium manganite films

V. Stankevič a,b,⁎, Č. Šimkevičius a,b, S. Balevičius a,b, N. Žurauskienė a,b, P. Cimmperman a,A. Abrutis c, V. Plaušinaitienė a,c

a Center for Physical Sciences and Technology, Semiconductor Physics Institute, A.Gostauto 11, Vilnius, Lithuaniab Vilnius Gediminas Technical University, Sauletekio 11, Vilnius, Lithuaniac Vilnius University, Dept. of General and Inorganic Chemistry, Naugarduko 24, Vilnius, Lithuania

⁎ Corresponding author at: Center for Physical Sciencductor Physics Institute, A.Gostauto 11, Vilnius, Lithuan

E-mail address: wstan@pfi.lt (V. Stankevič).

0040-6090/$ – see front matter © 2013 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.tsf.2013.05.127

a b s t r a c t

a r t i c l e i n f o

Article history:Received 25 June 2012Received in revised form 3 May 2013Accepted 31 May 2013Available online 14 June 2013

Keywords:Colossal magnetoresistanceThin filmsStrain effects

This is a study of the influence of external uniaxial mechanical strains on the transport properties of thin ep-itaxial La0.83Sr0.17MnO3 (LSMO) films. Our measurements were carried out using standard isoscelestriangle-shaped cantilever. Films which were tensed in-plane or compressed or were subjected to both ten-sion and compression strains were grown onto SrTiO3 (STO), LaAlO3 (LAO) and (001) NdGaO3 (NGO) sub-strates, respectively. It was found that for thin films (less than 100 nm), the uniaxial compression of suchfilms which were initially tensed in-plane (grown onto STO substrates) produces a decrease of their resis-tance, whereas the compression of initially compressed films (on LAO substrates) produces an increase ofthe films' resistance. The same results were obtained for LSMO films grown onto (001) NGO substrateswhen they were compressed along the [010] and [100] directions, respectively. For thicker films (morethan 100 nm), the resistance behavior after uniaxial compression was found to be identical to that producedby hydrostatic compression, namely, the resistance decreases irrespective of the substrate. These experi-ments also reveal an increase of resistance and a shift of metal–insulator transition temperature Tm tolower temperatures corresponding to a decrease of the film thickness. The occurrence of this effect is also in-dependent of the kind of substrate used. Thus it was concluded that the influence of film thickness on its re-sistance as well as on the behavior of such films while under external uniaxial compression cannot beexplained fully by only the presence of residual stress in these films. A possible reason is that the inhomoge-neous distribution of the mechanical stresses in the films can lead to the appearance of two conductivityphases, each having a different mechanism. The results which were obtained when these films weresubjected to hydrostatic compression were also explained by this model.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

The discovery of colossal magnetoresistance (CMR) in epitaxial pe-rovskite manganite thin films has led to an increasing interest in theuse of these materials both for fundamental science and for other appli-cations. Since the main applications of perovskite manganites are in thedesign of integrated devices, the properties of thin films are of great im-portance. The experimental and theoretical studies performed thus farhave shown that the biaxial strain due to the lattice mismatch betweenthe substrate and the thin film has a strong effect on the changes of thestructural, magnetic and transport properties of these thin manganitefilms. In particular, variations of their resistivity (ρ), their Curie temper-ature (TC) and their metal–insulator transition temperature (Tm) wereobtained when the thicknesses of these films were decreased [1–3].

es and Technology, Semicon-ia.

rights reserved.

Thus far, these changes have been attributed in most cases tosubstrate-induced strains and the disorder present in these thin films.It has been assumed that the compressive strains usually reduce the re-sistivity of thesemanganite films and increase their TC and Tm [4,5]. Thislattice strain effect has been interpreted using the double-exchangemodel. Thus it has been predicted that compressive strain could inducean increase of the TC by increasing the electron transfer due to thecompression of the Mn\O bond lengths. Millis et al. [6] proposed ananalytical model to describe the effects of the biaxial strain (εxx andεyy) on the transport properties of CMR manganites. According to thismodel, the TC depends on two parameters: a) the bulk compressionεB = 1/3(εxx + εyy + εzz), and b) the biaxial distortion ε⁎ =1/2(εxx − εyy). This model indicates the very high sensitivity of the ma-terial to such strains: a 1% biaxial strain can cause a 10% shift of the TC.Such behavior can be explained by a decrease of the electron transferdue to the stretching of the Mn\O bonds. This predicted dependencein general was confirmed experimentally using films which weregrown on different substrates [3–5,7]. However, recent detailed studies

Fig. 1. In-plane lattice sketches with pseudocubic lattice constants for bulk LSMO crys-tals (dashed line) and for different substrates (continuous line).

195V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

have shown that many of these properties in thin films cannot beexplained as due only to the presence ofmechanical stresses. During de-position of the thin films, a lattice mismatch between the film and thesubstrate can cause not only biaxial mechanical stresses, but also distor-tions of the crystallographic structure [8,9] and chemical composition[10]. Thus was introduced the concept of the coexistence of two (ormore) phases in these thin manganite films, which have different con-ductivity mechanisms (unstressed ferromagnetic (FM) metallic andstressed charge-ordered (CO) insulating). This phase separation is dueto structural inhomogeneity caused by non-uniform distribution of thestrains within these thin films [11–13]. According to this conception, itis assumed that the shift of the metal–insulator transition temperature(Tm) in their resistivity vs. temperature dependence can be explainedby changing the quantity of the FM or CO insulating phase fractions[14]. One possible reason for the increase of this paramagnetic phasein stressed films is that this may be due to an increase of the oxygen va-cancies [15], which decreases the lattice parameters and in this waycompensates for the lattice mismatch between film and substrate [16].Moreover, the existence of a so-called “dead layer” at the interface be-tween the film and the substrate [17] as well as the presence of a layerwhich is formed due to the diffusion of components from the substrateinto thin film has to be taken into consideration [18]. As a result, thesemay cause an increase in the resistivity of the film and a shift of the Tmtowards lower temperatures.

In order to investigate the influence of mechanical strain on theelectrical properties of manganite films, hydrostatic pressure was ap-plied [19,20]. Also, inner pressure was produced by changing the av-erage sizes of the divalent alkaline-earth ions (Sr, Ca, etc.). It is knownthat hydrostatic pressure changes the TC and the resistivity in thinfilms in a manner that is similar to that of epitaxial strain in thinfilms. However, as these methods induce only average stresses inthin epitaxial films (hydrostatic pressure can be treated as a uniaxialpressure applied in three perpendicular directions), thus they cannotgive a clear enough answer as to the exact influence of biaxial strainson the resistivity behavior and the magnetoresistance phenomenonin manganites.

There have been only a few reported investigations in which theinfluence of the uniaxial external strain on the electrical and mag-netic properties of manganite films has been analyzed [21–25].The gauge factor G of polycrystalline La0.67Sr0.33MnO3 (LSMO) andLa0.67Ca0.33MnO3 (LCMO) thin films deposited on oxidized Si hasbeen measured at different temperatures [21]. The measuredvalue of G, defined as the ratio of the relative resistance change tothe strain ε, was found to be about 50 for LCMO films and 70 forLSMO films at temperatures of maximum resistance. It decreasedto values of about 10 at room temperature. Very high gauge factorvalues (about 105–106) were also obtained for epitaxial (La1 −

yPry)1 − xCaxMnO3 films at temperatures of maximum resistance[23]. It has also been proposed that strains could be generated inmanganite films by using piezoelectric substrates and applyingelectric fields to them [22,24]. The influence of thesesubstrate-induced strains was investigated in La0.75Ca0.25MnO3 ep-itaxial thin films [24] and La0.7(Ca,Sr)0.3MnO3 polycrystalline films[22] grown onto piezoelectric substrates. Strain-induced resistancechanges in La0.7Sr0.3MnO3 film grown onto (001) NdGaO3 sub-strates were also investigated previously by our group [25]. It wasfound that the shift of the metal-insulator transition temperatureand the resulting increased resistivity in thin film with respect tothicker films cannot be explained only by the strains induced bylattice mismatches. Therefore, the investigation of the influence ofthese uniaxial external strains on the properties of epitaxial man-ganite thin films remained of continuing interest.

In this paper, we present a comprehensive study of the influenceof uniaxial external mechanical strains on the transport propertiesof thin epitaxial La0.83.Sr0.17MnO3 films grown onto SrTiO3 (STO),LaAlO3 (LAO) and (001) NdGaO3 (NGO) substrates.

2. Experimental details

A series of thin La0.83Sr0.17MnO3 films were grown onto (001)NdGaO3, (001) SrTiO3 and (001) LaAlO3 substrates using a verticalhot wall Injection Chemical Vapour Deposition reactor. Details ofthe epitaxial growth of these films are described in [26]. All the sam-ples were prepared under the same conditions, while their thick-nesses were varied from 8 nm to 140 nm. The metal–insulatortransition temperature Tm of these films was determined in a rangebetween 240 K and 340 K, respectively. The structural characteriza-tion of these films and the orientation of their substrates were deter-mined using an X-ray diffractometer. {100} and {010} NdGaO3

reflections were used to identify the orientations of the substrates.The analysis of the surface morphology of these films was conductedusing an atomic force microscope.

It is known that the LAO substrate has a pseudo-cubic crystallo-graphic structure with lattice constant a = 0.3788 nm, the STO sub-strate has a cubic structure with a = 0.3905 nm, and NGO substratehas an orthorhombic structure with a = 0.5426 nm, b= 0.5499 nmand c = 0.7706 nm. On the other hand, bulk LSMO has a pseudocubicperovskite structure with lattice parameter ap = 0.387 nm [27]. As aresult, the coherent LSMO films deposited onto STO substrates aretensile, while films deposited on LAO substrates are compressed.In-plane lattice sketches with the pseudo-cubic lattice constants forbulk crystals of LSMO (dashed line) and three different substrates(continuous line) are illustrated in Fig. 1. The LSMO films grownonto (001) NGO substrates having orthorhombic structures with dif-ferent a and b parameters in the (001) plane were of particular inter-est. In their case, the crystallographic structure of LSMO can berepresented as an orthorhombic structure with lattice parametersa ≈ b = 0.547 nm, c = 0.771 nm [27–29] rotated at a 45° anglewith respect to the cubic lattice. For this reason, thin LSMO filmsgrown onto (001) NdGaO3 substrates have different lattice mismatchsigns for different crystallographic directions in the film plane [30].Thus it can be assumed that these films are compressed in the a[100] direction and are tensile in the b [010] direction. Therefore,films coherently grown onto NGO substrates having (001) crystallo-graphic orientation are anisotropically strained. In our case, the rela-tive lattice mismatch between the film and NGO substrate iscalculated according to the equation ε = (asub − abulk)/abulk, whereasub and abulk are the lattice parameters of the substrate materialand the bulk manganite, respectively. Thus they are equal to −0.8%in direction a [100] and 0.53% in direction b [010]. The lattice

196 V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

mismatch between the film and the LAO substrate is thus equal to −2.0%, while between the film and STO substrate is equal to 0.9%.

Our investigations of the structure of these films conducted usingGrazing Incidence X-Ray Diffraction (Beamline I811, MaxLab-II, Swe-den) demonstrated that the LSMO/NGO films are anisotropicallystrained in-plane. Fig. 2 shows the high-resolution reciprocal spacemaps of the reciprocal lattice units (r.l.u.) around the asymmetricalBragg reflections (404) and (0–44) of 20 nm LSMO films grownonto NGO (001) substrates. As it can be seen, the maps show thatthe in-plane component of the film peak is equal to that of the sub-strate (Qx and Qy) and indicate coherent growth.

A standard isosceles triangle-shaped horizontal cantilever wasused to induce uniaxial compressive (or tensile) strains in the films.The wide end of cantilever was champed, then either a downwardor upward force was applied to the free narrow end of the cantileverto induce a longitudinal tensile (ε > 0) or a compressive (ε b 0)strain in the sample. This isosceles triangle-shaped cantilever wasused, because the strain induced using this method remains uniformalong the length of the sample [31] and is not dependent on the lon-gitudinal distance from the base of the cantilever. The samples wereglued to the cantilever and were then compressed (or tensed) bydeflecting the cantilever. The axial strain induced was measured byusing a constantan strain gauge. This strain gauge was glued ontothe substrate “satellite” which was positioned close to the samplesbeing investigated. The maximal value of the strain applied wasabout 0.003 (0.3%). The resistance measurements were carried outusing the “four-point-probe” method along a temperature range of150–340 K while applying a dc current of 0.1 mA to the films. Theseprobes were aligned in parallel to the axis of the cantilever (for longi-tudinal measurements) and perpendicular (for transversal measure-ments). Samples were glued in such a way that the [100] or [010]directions of the NGO (LAO, STO) substrates were kept parallel tothe axis of cantilever. In order to minimize plastic deformation ofthe glue due to changes of temperature, the deflection of the cantile-ver and the measurements of the resistance were conducted onlyafter the whole apparatus being used had reached the assigned tem-perature. After that the cantilever with the sample was returned to itsinitial position.

The effect of applied hydrostatic pressure on the resistivity wasstudied at room temperature using a standard piston gauge produc-ing a hydrostatic pressure of up to 200 MPa.

Fig. 2. Reciprocal space maps in reciprocal lattice units around (404) and (0–44) withasymmetric Bragg reflections for 20 nm LSMO film deposited on NGO substrates.

3. Results

3.1. LSMO/NGO films under uniaxial stress

The temperature dependence of the resistivity of LSMO films onvarious film thicknesses is shown in Fig. 3. As it can be seen, themetal–insulator transition temperature Tm decreases with a decreaseof the film thickness. For 8 nm thin film, the Tm is 220 K. For filmsthicker than 20 nm, the Tm is shifted to 290 K. However, any furtherincrease of the film thickness causes only small changes of the Tm.The dependence of the transport behavior on thickness in LSMOthin films is in good agreement with previous reports [32].

The resistance change (Rstr − R)/R of films grown onto NGO sub-strates when subjected to external uniaxial compression and tensionalong the [100] direction and measured at 290 K temperature isshown in Fig. 4. Here Rstr is the resistance of the film under stress; Ris the resistance when it was not subjected to external stress (theprobe was aligned in parallel to the cantilever axis). The resistancechange exhibited linear dependence on the strain in all the investigat-ed directions for both the compressive and tensile strains. Thesechanges have the reversible character in the range of the strainsthat were investigated (up to 0.3%) The sign of the resistance change(the increase or decrease) depends on the film thickness. In filmswith thicknesses of 140 nm, uniaxial compression causes a decreasein its resistance (in a manner similar to that caused by hydrostaticpressure), while in films with thicknesses of 20 and 8 nm, the resis-tance change has an opposite sign (it increases). The same behaviorthe resistance change was obtained along the whole measured tem-perature range. Only slight differences were observed between theslopes of the straight lines of films with different thicknesses. Thusthe same linear dependence and reversible character of the resistancechange were observed for all investigated films when subjected com-pressive and tensile strains. The discussion of the results obtainedwhen the films were under compressive strains is given in more de-tail below.

Fig. 5a, b, c presents the dependences of this resistance change((Rcomp − R)/R vs. temperature) of LSMO films deposited onto(001) NGO substrates. Here Rcomp is the resistance of these filmswhen under compressive strain. The results are shown for variousfilm thicknesses, when an external uniaxial compressive strain ε =0.1% was applied to these films along two different crystallographicdirections: [010] and [100]. This resistance was measured when theelectrical current was parallel (longitudinal effect) and perpendicular(transversal effect) to the direction of the strain. All the curves

Fig. 3. The temperature dependence of the resistivity of LSMO/NGO films with differentthicknesses. The dashed line is the calculated resistivity vs. temperature dependence of8 nm film obtained by extrapolation of the uniaxial compression along direction [010]up to the value of the strain −0.53%.

Fig. 4. Relative resistance changes (Rstr − R)/R vs. uniaxial strain applied along the[100] direction to LSMO/NGO films having different thicknesses.

Fig. 5. The relative resistance change (Rcomp − R)/R vs. temperature at a compressivestrain of 0.1% applied along directions [010] and [100] to LSMO/NGO films having dif-ferent thicknesses: 8 nm (a), 20 nm (b) and 140 nm (c). The filled symbols show theresistance that was measured in parallel to strain direction; the open symbols showthe resistance that was measured perpendicularly to the strain.

197V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

exhibited peaks at a certain temperature which were shifted to lowertemperatures relative to Tm.

For 8 nm thickness films (Fig. 5a), the resistance change (Rcomp −R)/R vs. temperature depends very strongly on the direction of thestrain applied. When the film is compressed along the [010] direction,the resistance decreases along the whole measured temperate rangeindependent of the measurement direction. A special feature of thisresistance change is an increase of the resistance when the film iscompressed along [100] direction. As was already mentioned above,the film was initially compressed along the [100] direction due tothe mismatch of the film and substrate lattice constants (see Fig. 1).Such behavior (the resistance increase) is opposite to the resistancechange (decrease) which is obtained when the film is compressedalong the [010] direction, in which the film was initially tensile. Mea-surements of the transversal resistance (open circles and triangles)demonstrate similar results (the same behavior and slightly lowervalues) as in the case of longitudinal measurements.

The results obtained for films of 20 nm thicknesses (Fig. 5b)showed a similar behavior of the (Rcomp − R)/R dependence vs. tem-perature as was observed in the case of films with 8 nm thicknesses.However, the anisotropy of the resistance change for these sampleswas less pronounced. The absolute value of the resistance change inthe case of films compressed along the [100] direction was nearlytwo times smaller than that along the [010] direction.

In the case of films of 140 nm thicknesses (Fig. 5c), the resistancechange when compressed along the [100] direction changes its sign(shows a decrease of resistance) and is similar to that which isobtained when compressed along the [010] direction. It needs to benoted that such a decrease of resistance under compression is usuallyalso observed in thick relaxed films when they are subjected to hy-drostatic pressure. This then is an indication that such films are al-most or at least partially relaxed. The measurements performed inthe perpendicular direction (transverse resistance) showed the oppo-site sign (an increase of resistance) and were similar to thinnerstrained films. Thus we concluded that the 140 nm films are notcompletely relaxed.

3.2. LSMO/LAO films under uniaxial stress

Similar investigations were performed on LSMO films depositedonto LAO substrates which induce compression of the film due tothe mismatch of the film/substrate lattice constants. Fig. 6a showsthe electrical resistivity of the films vs. temperature dependence intheir case. As can be seen, for the thin 8 nm films, the Tm is alsoshifted to lower temperatures with respect to the Tm of thick films.The resistance changes of LSMO films deposited onto LAO substrates

when under uniaxial compression (Fig. 6b) show the same behavioras in the case of the thin films (8, 20 nm) deposited onto (001)NGO substrates: the resistance increases when subjected to uniaxialcompression in the [100] direction. Film with 20 nm thicknessesshowed a very small change of the resistance under uniaxial com-pression, while thicker film (140 nm) showed a decrease of resis-tance. The transverse resistance change measurements (not shown

Fig. 6. (a) Temperature dependence of the resistivity of LSMO/LAO films having differ-ent thicknesses. Dashed line shows the calculated resistivity vs. temperature depen-dence of 8 nm film obtained by extrapolation of the uniaxial tension along the [100]direction up to the value of the strain 2%. (b) The longitudinal resistance change vs.temperature at a compressive strain of 0.1% applied along [100] direction to LSMO/LAOfilms having different thicknesses.

Fig. 7. (a) Temperature dependence of the resistivity of LSMO/STO films having differ-ent thicknesses. The dashed line shows the calculated resistivity vs. temperature de-pendence of 8 nm film obtained by extrapolation of the uniaxial compression alongthe [100] direction up to the value of the strain −0.9%. (b) The longitudinal resistancechange vs. temperature at a compressive strain of 0.1% applied along the [100] direc-tion to LSMO/STO films having different thicknesses.

198 V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

in the figure) reveal the same behavior as the longitudinal ones. Themaximal values of the resistance change were observed at tempera-tures T b Tm.

3.3. LSMO/STO films subjected to uniaxial stresses

Fig. 7a presents the resistivity vs. temperature dependence forLSMO/STO films of various thicknesses. These results are similar tothose obtained for films deposited onto LAO substrates. Films withthicknesses of 8 nm had the lowest Tm and the highest resistivity.Contrary to the previous cases, the resistance changes under uniaxialstrain of films deposited onto STO substrates, which induce tension ofthe film because of mismatches of their film/substrate lattice con-stants, showed different behavior. When subjected to compressivestrain, the resistance of such LSMO/STO films when subjected to com-pressive strains decreased in all the measured thicknesses of the films(Fig. 7b). Moreover, a maximum of the resistance change wasobtained at almost the same temperature for all these thicknesses. Ithas to be noted that in this case, the external uniaxial compressionis opposite to the initially tensile film due to the mismatch of thefilm/substrate lattices constants. These results correspond to the re-sistance change under compression in the [010] direction of films

grown onto NGO substrates which are also initially tensile in thisdirection.

The measurement results of the resistance change when theprobes were aligned perpendicular (transverse) to the stress direc-tion for films on LAO and STO substrates had the same sign as the re-sults when the probes were aligned parallel to the stress, but themaximum values were lower approximately by a factor of two.

3.4. Hydrostatic pressure

The influence of hydrostatic pressure on the resistance of the filmsat room temperature is shown in Table 1. The sensitivity to hydrostat-ic pressure of the resistance is expressed as the resistance change perpressure magnitude SH = (Rp − R)/(R⋅P). Here Rp is the resistance atapplied hydrostatic pressure P, and R is the resistance without any ex-ternal pressure being applied.

In spite of this difference of the resistance behavior of initiallycompressed and tensile films when under uniaxial compression, allthese films showed a decrease of resistance when under hydrostaticpressure which was in good agreement with results obtained byother investigators [8,33]. Our results demonstrate that in all cases,the SH almost does not differ in all the samples tested except in the8 nm LSMO/NGO film which is strained in-plane non-uniformly, be-cause of the different lattice mismatch signs running in different

Table 1Hydrostatic pressure and uniaxial compressive stress sensitivities of films at T = 290 K.When calculating the values of the sensitivities Sl,t according to equation Sl,t = (Rcomp − R)/(R⋅ε⋅Y),the elasticity moduli of LSMO Y[100] = 95 GPa [36] were used.

Substrate h, nm Sl [100],%(108 Pa)−1

Sl [010],%(108 Pa)−1

St [100],%(108 Pa)−1

St [010],%(108 Pa)−1

SH, %(108 Pa)−1

Measured Calculated[100]

Calculated[010]

NGO 8 −0.59 0.95 −0.59 0.95 −4.5 1.77 −2.8520 −0.63 3.17 −0.81 2.08 −8.1 2.25 −7.33

140 0.36 2.30 −0.68 1.58 −10.0 1.00 −5.46LAO 8 −1.22 −1.18 −0.59 −0.59 −9.9 2.40 2.38

20 0.05 0.05 0.02 0.02 −9.2 −0.09 −0.09140 2.31 2.04 1.18 1.15 −8.7 −4.66 −4.34

STO 8 2.72 2.85 1.31 1.22 −11.1 −5.34 −5.2920 6.83 6.88 3.12 3.66 −11.0 −13.1 −14.2

140 3.49 3.44 1.81 1.83 −11.3 −7.11 −7.10

199V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

directions. It should be taken into account that in all samples (exceptfor the 8 nm LSMO/NGO and 8 nm LSMO/LAO films) the temperatureTm is higher than the temperature at which the hydrostatic pressuremeasurements were performed. However, according to the resultsof other authors, the resistance change when under hydrostatic pres-sure is negative both in their paramagnetic and ferromagnetic phases.The appreciable increase of the negative value of SH is observed onlyin the temperature region of the phase transition [34]. At a tempera-ture of 290 K, the 8 nm thick LSMO/NGO films are in a paramagneticstate that is far from the phase transition temperature and thus has asmaller value of SH.

4. Discussion

Themeasurement results of the electric transport properties of theinvestigated films show that the resistivity maximum and metal–in-sulator transition temperature Tm depend on the film thicknessesfor all films deposited onto NGO, STO and LAO substrates. Thin filmshave higher resistivity and lower Tm, compared to the thicker ones.The presence of the initial stresses in these films due to lattice mis-matches between the film and the substrate can be one of the reasonsfor this dependence on the thickness of the films.

As was already mentioned, due to the lattice mismatch betweenthe film and the substrate, the LSMO/LAO films are compressed inthe (001) plane, while the LSMO/STO films are tensile in the (001)plane. Moreover, the LSMO films grown onto (001) NGO substratesare compressed in the [100] direction and are tensile in the [010] di-rection. Our investigations show that the reduction of these strains byexternal mechanical uniaxial stress leads to a decrease of the electri-cal resistance of these films. Therefore, the uniaxial compression ofthe thin (less than 140 nm) films in the direction in which they areinitially tensile (LSMO/STO and LSMO/NGO in the [010] direction)leads to a decrease of their resistance. On the other hand, uniaxialcompression of such films in the direction in which they were initiallycompressed (LSMO/LAO and LSMO/NGO in the [100] direction) leadsto an increase of their resistance. For thick films, in which thesestrains are partially or completely relaxed, the external compressionproduces a decrease of their resistances. Thus from the obtained re-sults, it can be concluded that the magnitude and sign of the resis-tance change vs. external stresses depend on the internal stresses inthe films. The observed changes of these resistances when uniaxialstress is applied to these films can be explained by the reduction ofthe Mn\O bonds at the metal–insulator transition which originatefrom the formation of metallic bonding in their ferromagnetic phaseand from the significant decrease of Jahn–Teller distortions of theMnO6 octahedra around the Mn3+ ions [18–20]. This decrease of dis-tortions and/or reduction of the neighboring-site hopping distancesleads to an increase of the charge-carrier mobility and as a result, todecreases of their resistances. This effect is especially pronounced inthe phase separation region. Phase separation theory predicts the

coexistence of two phases, namely the presence of ferromagnetic me-tallic and paramagnetic insulator phases in the temperature rangeclose to the phase transition temperature [37]. In such cases, the elec-trical transport properties can be explained by percolative transportthrough the FM regions, whose electrical properties (resistance inour case) are controlled by the external uniaxial stresses being ap-plied, i.e. by the additional increase or decrease of the stress valuesof initially stressed films due to their film/substrate lattice mismatch.Therefore, the compression (tensing) of initially tensed (compressed)films reduces (increases) the tensed (compressed) in-plane latticeparameters, decreases the amount of strained regions, and increasesthe amount of the FM metallic phase. It was shown that at tempera-tures close to the Tm in the percolative regime of coexisting metallicand insolating clusters, these small changes in the system may leadto considerable changes in their resistivity: for instance, the changesof the metallic fraction by only 5% may result in a change of the resis-tivity by two orders of magnitude [37]. Thus the changes of the FMphase fraction in the phase-separation region in our investigatedthin films after the application of external uniaxial stress lead tolarge changes in its resistivity. As a result, the peak of such resistancechange is observed at a temperature where the derivative of resis-tance dR/dT is at its maximum. Similar results were obtained also inother works [21,23,34]. This resistance change due to uniaxial stressvs. temperature is similar to the resistance change in its magneticfield (the colossal magnetoresistance effect) which reveals itself atmaximum also at temperatures corresponding to the highest deriva-tive dR/dT.

Longitudinal and transverse resistance changes after the applica-tion of external uniaxial compression demonstrate the dependencesof these changes on the type of substrates used and the thicknessesof the films. Moreover, this resistance change (sensitivity) has to beconsidered in three perpendicular directions with respect to the ap-plied stress: along the stress direction (Sl) and in two perpendiculardirections St1 and St2. St1 is the transversal resistance change in thefilm plane and St2 is the transversal sensitivity in the direction per-pendicular to the film plane. The relationships between these coeffi-cients depend on the crystal structure and the band structure of theinvestigated material. For example, the cubic crystals Sl and St havedifferent values, but do not depend on strain in the direction {100}.Our results show that these sensitivities are mostly determined bythe presence of internal strain in the films. The results obtainedfrom hydrostatic pressure experiments have partially confirmed thisassumption. It is known that the resistance change (sensitivity) SHvs. hydrostatic pressure for cubic crystals is equal to the sum of thelongitudinal and twofold transversal uniaxial sensitivities obtainedin {100} directions with opposite signs: SH = −(Sl + 2⋅St) [35]. As-suming that for the investigated films the two transversal coefficientsare equal, the sensitivity of the films to hydrostatic pressure was cal-culated by using results obtained from uniaxial compression of thesefilms. The results of measured and calculated SH values as well as their

200 V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

measured sensitivities to uniaxial stress applied in the [100] and[010] directions at a temperature of 290 K are presented in Table 1.One can see that this assumption is approximately valid only forfilms with 140 nm and 20 nm thicknesses grown onto STO sub-strates. For films of 8 nm thickness grown onto STO substrates andfor 140 nm thick films grown onto LAO substrates, the calculated sen-sitivities are two times lower than those measured. For 20 nm and8 nm thick films grown onto LAO substrates and for all films grownonto NGO substrates (if sensitivities in the direction [100] are used),this difference is much higher. It can thus be assumed that onlyfilms grown onto STO substrates and which have thicknesses ofmore than 20 nm are relaxed and behave as films with cubic struc-tures. Resistance vs. temperature dependences also confirm this as-sumption (see Fig. 7a). The other films have residual strains whichdistort the crystal structure and thus change the sensitivities Sl andSt. Furthermore, it should be noted that for all films grown onto STOand LAO substrates, four-fold symmetry ratios for their longitudinaland transversal sensitivities are observed: Sl[100] = Sl[010] and St[100] =St[010]. This symmetry is typical for crystals with cubic structures andalso for crystals with tetragonal structures which can be obtained by ap-plying uniform in-plane strains to the films along the [100] and [010] di-rections. It also has to be noted that for all films grown onto STO and LAOsubstrates, their longitudinal sensitivity is approximately twice higherthan that of the transversal one (Sl ≈ 2St) at all measured temperatures.

For films grown onto NGO substrates, the relation between the hy-drostatic and uniaxial (longitudinal and transversal) sensitivities ismore complex due to the orthorhombic deformation of these films,caused by the lattice mismatch between the film and the substrate.The results show that even for 140 nm thick films grown onto thesesubstrates, the sensitivities Sl and St are different from the sensitivi-ties of relaxed crystals with cubic structures. They have differentvalues and signs and have two-fold symmetry ratios for their longitu-dinal and transversal sensitivities, which are typical for orthorhombicdistortion. When we compared the results of calculations performedfor these films in directions [100] and [010], we concluded that withany increase of the film's thickness, the relaxation of stresses in the[010] direction takes place earlier than in the [100] direction. Thusit needs to be noted that the calculated SH for films thicker than20 nm, when compared with the data obtained in the [010] direction,is similar to the measured values. This confirms the assumption thatin this direction, the films are more relaxed.

In order to qualitatively evaluate the possible influence of thesemechanical stresses on the electrical properties of thin films, weperformed simple calculations of this resistance change, assumingthat the applied external uniaxial stresses are equal to the value andare opposite in sign to the initial strains existing in the film (due tomismatches of the film/substrate lattice constants). Using the resultspresented in Fig. 4, it was assumed that the resistance change vs.strain has linear dependence along the whole measurement range.This linear behavior was also observed in films grown onto STO andLAO substrates. Using this assumption, the resistance changes ofthin films were extrapolated up to the values of the compression ortension at which the mechanical stresses, due to the mismatch ofthe lattices, become equal to zero. We then used the data obtainedfrom thin films at which these strains have maximum values. Thedashed line in Fig. 3 shows the calculated resistivity vs. temperaturedependence of such 8 nm LSMO/NGO films when obtained by extrap-olation of the uniaxial compression along the direction [010] up to thevalue of the strains equal to−0.53%, because the film in this case wasinitially tensile by +0.53% due to lattice mismatch. The dashed line inFig. 6a shows the similar calculations of the resistance dependenceobtained by extrapolation of the uniaxial tension of 8 nm LSMO/LAOfilm up to the value of the strain +2%. In this case, the film is com-pressed due to lattice mismatch by −2%. Similar calculations werealso made by extrapolation of the uniaxial compression in films de-posited onto STO substrates (see Fig. 7a dashed line) up to the value

of the strain −0.9%. As can be seen in the case of the LSMO/NGOfilms, the maximum of the resistance dependence on temperature isshifted only slightly to the position of this maximum in the thickfilms. Therefore, the shift of the resistance maximum temperatureTm and the increase of resistance because of any increase of the filmthickness in LSMO/NGO films cannot be explained by taking into ac-count only the biaxial strains caused by lattice mismatches betweenthe film and the substrate. In the case of LSMO/LAO films, the calculat-ed curve is shifted more to the position of the thicker films, thus theshift of the Tm for this film can be explained by the strain effect. Italso has to be noted that the shape of the calculated curves is farfrom that which is obtained for thick films. Therefore, it is evidentthat when assuming only the mechanical strain effects, it is not possi-ble to explain such a strong dependence of the resistance change andthe Tm shift on the film thickness.

As was mentioned in the introduction, compressive strains inLa1 − xSrxMnO3 films should decrease the film resistance. However,we obtained such results only for the thick films in which the strainsdue to the lattice mismatch between the film and the substrate aremost likely relaxed. Our experiments show that the residual mechan-ical stresses occurring due to a mismatch of the lattice constants inthin films cause an increase of the film resistivity independently onthe sign of the stress.

A possible reason for these results can be the inhomogeneous dis-tribution of mechanical stress in the films which can lead to the ap-pearance of two phases with different conductivity mechanisms. Itwas shown that when films are grown under compressive strain,this can readily lead to an island growth mode which can produce anon-uniform distribution of the strains [12]. The tops of the islandsremain relatively strain free. These regions are ferromagnetic, butare separated by the strained insulating regions at the peripheries ofthese islands. These strains hamper the transition from paramagneticto the ferromagnetic phase in the film and, as a result, decrease thetransition temperature Tm. In such cases, the total resistivity can becalculated by using a model of a random-resistor-network whichmimics the prominent FM–CO mixtures [14,32]. A shift of Tm in theresistivity vs. temperature dependence was obtained by changingthe quantity of the fractions FM or the CO phase [14]. In our earlierwork [32], it was shown that La0.83Sr0.17MnO3 thin films grown ontoNGO substrates have insertions of a high resistivity inter-grainphase with orthorhombic face centered structures and a lattice con-stant of a = 0.406 nm in plane and c = 0.460 nm out of plane. Itwas found that the relative volume of this phase decreases with an in-crease of the thickness of the films. Simulations performed in [32]have demonstrated that the obtained large decrease in the Tm whenthe film thickness is decreased can be due to this two-phase natureof thin film.

This model can also explain qualitatively the changes of resistancewhen the films are subjected to external uniaxial compression. Thispresupposes that the residual mechanical stresses are concentratedmainly in this inter-grain phase and that this stress causes the in-crease of resistivity of films independently on the sign of the stress.When the film is compressed externally, this causes compression ofboth these phases. This external compression of the grain phase al-ways leads to a decrease of resistance. At the same time, any furthercompression of the initially compressed inter-grain phase leads toan increase of these stresses and to an increase of the electrical resis-tance (in the case of LSMO/LAO and LSMO/NGO in the [100] direc-tion). In the opposite case, when the inter-grain phase is initiallytensed, this compression leads to a decrease of the stress and to a de-crease of the resistance (in the case of LSMO/STO and LSMO/NGOfilms in the [010] direction). According the percolation model, thetotal resistance of thin film is defined mostly by this inter-grainphase, thus producing high resistivity. Therefore, the resistancechange of the grain phase affects the change of the total resistanceonly insignificantly. When the thickness of the films is increased,

201V. Stankevič et al. / Thin Solid Films 540 (2013) 194–201

the influence of the inter-grain phase decreases and behavior of thefilm resistance when compressed externally is defined by the grainphase.

5. Conclusions

In conclusion, the application of external uniaxial mechanicalstrain to thin epitaxial La0.83.Sr0.17MnO3 films grown on differenttypes of substrates has revealed the following features: In the caseof thin films (8–20 nm), the uniaxial compression of initially tensedfilm because of the mismatch of the film and substrate lattice con-stants leads to a decrease of the stress in the film and to a decreaseof its resistance, while uniaxial compression of initially compressedfilm leads to an increase of the stress in the film, and, as a result, toan increase of its resistance. In the case of thicker films (>140 nm),their resistance in both cases decreases. The peculiarities of this resis-tance change in such films when subjected to uniaxial strains can beexplained by the interaction of two phases, namely the grain andinter-grain phases which appear during the growth process. In thinfilms, the total resistance is determined mostly by inter-grain phaseand produces high resistivity. When the thickness of the film in-creases, the influence of this inter-grain phase decreases and behaviorof the film resistance when the film is subjected to external compres-sion is defined by the main grain phase. Under hydrostatic compres-sion, the resistivity of all the samples decreases by nearly equalvalues except in very thin (8 nm) LSMO/NGO films which arestrained in-plane non-uniformly and show much smaller decreasesof resistance. It was therefore concluded, that the increase of thefilm resistance and the shift of the metal–insulator transition temper-ature Tm to lower temperatures with a decrease of the film thicknesscannot be explained by taking into account only the biaxial straincaused by the lattices mismatch between the film and the substrate.To a large extent, the existence of the inter-grain phase determinesthe transport properties of thin films.

Acknowledgments

The authors acknowledge the financial support from the ResearchCouncil of Lithuania through the project No. MIP-062/2012. The au-thors also thank Tomas Stankevic for structural measurements usingGrazing Incidence X-Ray Diffraction.

References

[1] Bing Li, Fa Liang Tian, Lei Yang, Xiao PingWang, Hong Zhu, Tamio Endo, Thin SolidFilms 519 (2011) 2381.

[2] Y. Lu, J. Klein, F. Herbstritt, J.B. Philipp, A. Marx, R. Gross, Phys. Rev. B 73 (2006)184406.

[3] T. Dhakal, J. Tosado, A. Biswas, Phys. Rev. B 75 (2007) 092404.[4] Yuan-Chang Liang, Yung-Ching Liang, J. Cryst. Growth 304 (2007) 275.[5] C. Thiele, K. Dörr, O. Bilani, J. Rödel, L. Schultz, Phys. Rev. B 75 (2007) 054408.[6] A.J. Millis, T. Darling, A. Migliori, J. Appl. Phys. 83 (1998) 1588.[7] Q. Qian, T.A. Tyson, M. Deleon, C.-C. Cao, J. Bai, A.I. Frenkel, J. Phys. Chem. Solids 68

(2007) 458.[8] Y. Ogimoto, M. Izumi, T. Manako, T. Kimura, Y. Tomioka, M. Kawasaki, Y. Tokura,

Appl. Phys. Lett. 78 (2001) 3505.[9] Y.H. Li, K.A. Thomas, P.S.I.P.N. de Silva, L.F. Cohen, A. Goyal, M. Rajeswari, N.D.

Mathur, M.G. Blamire, J.E. Evetts, T. Venkatesan, J.L. MacManus-Driscoll,J. Mater. Res. 13 (1998) 2161.

[10] W. Prellier, A. Biswas, M. Rajeswari, T. Venkatesan, R.L. Greene, Appl. Phys. Lett.75 (1999) 397.

[11] A. Weisse, J. Loos, H. Fehske, Phys. Rev. B 64 (2001) 104413.[12] A. Biswas, M. Rajeswari, R.C. Srivastava, Y.H. Li, T. Venkatesan, R.L. Greene, Phys.

Rev. B 61 (2000) 9665.[13] K.A. Thomas, P.S.I.P.N. de Silva, L.F. Cohen, M. Rajeswari, T. Venkatesan, R. Hiskes,

J.L. MacManus-Driscoll, J. Appl. Phys. 84 (1998) 3939.[14] J. Burgy, M. Mayr, V. Martin-Mayor, A. Moreo, E. Dagotto, Phys. Rev. Lett. 87

(2001) 277202.[15] S.I. Khartsev, P. Johnson, A.M. Grishin, J. Appl. Phys. 87 (2000) 2394.[16] L. Malavasi, M.C. Mozzati, C.B. Azzoni, G. Chiodelli, G. Flor, Solid State Commun.

123 (2002) 321.[17] J.Z. Sun, D.W. Abraham, R.A. Rao, C.B. Eom, Appl. Phys. Lett. 74 (1999) 3017.[18] Hsiung Chou, S.G. Hsu, C.B. Lin, C.B. Wu, Appl. Phys. Lett. 90 (2007) 062501.[19] F.S. Razavi, G.V.S. Rao, H. Jalili, H.U. Habermeier, Appl. Phys. Lett. 88 (2006)

174103.[20] G.M. Gross, R.B. Praus, S. Yaghoubzadeh, F.S. Razavi, H.-U. Habermeier, J. Alloy

Compd. 317 (2001) 141.[21] J. Sarkar, A.K. Raychaudhuri, J. Nanosci. Nanotechnol. 7 (2007) 2058.[22] R.B. Gangineni, J.W. Kim, K. Nenkov, L. Schultz, J. Magn. Magn. Mater. 324 (2012)

1153.[23] J. Tosado, T. Dhakal, A. Biswas, J. Phys. Condens. Matter 21 (2009) 192203.[24] R.K. Zheng, Y. Wang, J. Wang, K.S. Wong, H.L. Chan, C.L. Choi, H.S. Luo, Phys. Rev. B

74 (2006) 094427.[25] C. Simkevicius, V. Stankevic, S. Balevicius, A. Abrutis, V. Plausinaitiene, L. Dapkus,

B. Vengalis, N. Zurauskiene, J. Phys. IV France 11 (2001) Pr11.[26] A. Abrutis, V. Plausinaitiene, V. Kubilius, A. Teiserskis, Z. Saltyte, R. Butkute, J.P.

Senateur, Thin Solid Films 413 (2002) 32.[27] Y. Suzuki, H.Y. Hwang, S.-W. Cheong, R.B. van Dover, Appl. Phys. Lett. 71 (1997)

140.[28] P. Dai, J. Zhang, H.A. Mook, S.-H. Liou, P.A. Dowben, E.W. Plummer, Phys. Rev. B 54

(1996) R3694.[29] T.M. Perekalina, I.E. Lipinski, V.A. Timofeyeva, S.A. Cherkezyan, Sov. Phys. Solid

State 32 (1990) 1827.[30] H. Boschker, M. Mathews, E.P. Houwman, H. Nishikawa, A. Vailionis, G. Koster, G.

Rijnders, D.H.A. Blank, Phys. Rev. B 79 (2009) 214425.[31] R.M. McClintock, Rev. Sci. Instrum. 30 (1959) 715.[32] S. Balevičius, P. Cimmperman, V. Petrauskas, V. Stankevič, E.E. Tornau, N.

Žurauskienė, A. Abrutis, V. Plaušinaitienė, M. Sawicki, T. Dietl, M. Aleszkiewicz,Thin Solid Films 515 (2006) 691.

[33] Y. Moritomo, A. Asamitsu, Y. Tokura, Phys. Rev. B 51 (1995) 16491.[34] T. Roch, S. Yaghoubzadeh, F.S. Razavi, B. Leibold, R. Praus, H.-U. Habermeier, Appl.

Phys. A 67 (1998) 723.[35] C.S. Smith, Phys. Rev. 94 (1954) 42.[36] N.G. Bebenin, R.I. Zainullina, V.V. Ustinov, J. Magn. Magn. Mater. 321 (2009) 682.[37] E. Dagotto, T. Hotta, A. Moreo, Phys. Rep. 344 (2001) 1.