Nanoscale Laterally Modulated Properties of Oxide Ultrathin Films by

Article pubs.acs.org/cm

Nanoscale Laterally Modulated Properties of Oxide Ultrathin Films by Substrate Termination Replica through Layer-by-Layer Growth Carmen Ocal,†,* Romain Bachelet,† Luis Garzón,† Massimiliano Stengel,†,‡ Florencio Sánchez,†,* and Josep Fontcuberta†,* †

Institut de Ciència de Materials de Barcelona (ICMAB-CSIC), Campus de la UAB, 08193 Bellaterra, Barcelona, Spain ICREAInstitució Catalana de Recerca i Estudis Avançats, Barcelona 08010, Spain

‡

ABSTRACT: Modulation of oxide properties in a direction perpendicular to surfaces can be easily achieved by advanced thin film deposition tools. In contrast, simultaneous nanoscale modulation of properties in directions perpendicular and parallel to surfaces, as required for building three-dimensional (3D) nanometric devices, has remained elusive. Though bottom-up approaches allow obtaining controlled lateral growth, these techniques fail to achieve long-range order or do not allow simultaneous modulation of properties in the perpendicular direction. Here, we show that, by exploiting a dual strategy based on the spontaneous surface restructuration of some perovskite substrates and the layer-by-layer growth of nanometric heteroepitaxial perovskite layers, surfaces with laterally modulated surface relief, chemical termination, electrostatic potential, and electrical conductance can be obtained. As illustrative examples, conducting ferromagnetic manganite La2/3Sr1/3MnO3 and ferroelectric BaTiO3 ultrathin films have been grown on (001)SrTiO3 and (001)(La0.18Sr0.82)(Al0.59Ta0.41)O3 substrates with self-ordered chemical terminations. KEYWORDS: oxide surfaces, chemical terminations, oxide interfaces, Kelvin microscopy, ferroelectric films, epitaxy

■

was first achieved for the widely used SrTiO3,5,6 and it has been progressively extended to other substrates.7 Interestingly, it has been shown that the distinct terminations of as-received (001)ABO3 single-crystal substrates can self-separate when appropriately treated [e.g., in SrTiO3 (STO),8−10 DyScO3 (DSO),11 (La0.18Sr0.82)(Al0.59Ta0.41)O3 (LSAT)12]. Moreover, these terminations have been found to self-order, forming regular patterns of AO and BO2 stripes, separated in height by half unit cell (uc) (≈ 0.2 nm, the interplanar distance in perovskites) and running parallel to the regularly spaced substrate step edges. These stripes, having distinct atomic arrangements and compositions, present a clear opportunity to realize the aforementioned challenging goal of laterally engineer the functional properties. Indeed, in STO, LSAT, and DSO, these stripes allow confined and selective lateral growth of organic and inorganic materials and have been used to obtain either two-dimensional (2D)-ribbons or arrays of threedimensional (3D)-dots.8,10−12 Experiments on the selective growth of organic molecules also showed that tribologic and electrostatic properties of some of these surfaces can also be tailored ad-hoc at the nanoscale.10 The structure and stability of the AO and BO2 terminating layers of perovskite (001)ABO3 single crystals were studied in detail for STO. It has been shown that free surfaces of the nonpolar SrO and TiO2 terminations have different surface energies and ionization potentials and reconstruct differently by atomic displacements and covalent mixing.13−16 These differ-

INTRODUCTION Ultrathin magnetic and ferroelectric epitaxial oxide thin films and heterostructures are being intensively investigated due to the emerging properties they may exhibit and foreseen promises of their potential applications. Interface phenomena in oxides, including strain effects, bond reconfigurations, charge transfer, and local structural distortions are explored to create materials with properties either nonexisting in bulk or dramatically different from them. It has been shown, for instance, that (i) EuTiO3 becomes a magnetic ferroelectric when grown under suitable strain conditions;1 (ii) the spin density at ferromagnetic−ferroelectric interfaces can be controlled via the ferroelectric polarization;2 and (iii) a twodimensional electron gas (2DEG) forms at the interface between band insulators, with indications that superconductivity and magnetism might coexist in this system.3,4 These findings challenge our understanding and the subsequent control of interface phenomena in complex oxides. In the above examples, the focus was the modification of interface-related properties along the perpendicular direction in epitaxial heterostructures. Commonly, single-crystalline ABO3 perovskites having a single atomic termination are used as substrates. However, an additional benefit, of potentially stronger impact, would be gained if nanoscale tailoring of properties could be simultaneously achieved along directions parallel and perpendicular to the interface. As-received (001)ABO3 single crystals contain two distinct coexisting atomic terminations (AO and BO2) but appropriate tools have been developed to obtain single AO or BO2 terminated surfaces of some ABO3 substrates. Such a control © 2012 American Chemical Society

Received: August 1, 2012 Published: September 28, 2012 4177

dx.doi.org/10.1021/cm302444s | Chem. Mater. 2012, 24, 4177−4184

Chemistry of Materials

Article

ences determine the characteristics of the interfaces created upon growth on each termination and, therefore, may propagate into ultrathin films of oxides epitaxially grown on them, thus tailoring their final properties. It follows that nanostructuration by self-ordering of the chemical termination of ABO3 substrates, forming nanometric-wide AO and BO2 stripes, may constitute a convenient and versatile approach to create templates for engineered growth of functional materials. We propose and demonstrate here the use of chemical lateral self-patterning of the ABO3 substrate surfaces to create 2Dtemplates for subsequent growth of functional oxides thus forming 3D structures. We explore this approach in detail by analyzing chemical selfordering of nonpolar and polar (001) surfaces of STO and LSAT perovskites, respectively, their topographic characteristics and electrostatic properties, as well as their impact on the electrical properties of nanometric films of some functional oxides, namely, La2/3Sr1/3MnO3 (LSMO) and BaTiO3 (BTO) grown on them. To reach this goal, first, the surface of the substrates will be tailored to display regular patterns, some tens of nanometers wide, of AO and BO2 terminations12 that are distinguishable on topographic and friction measurements and differ on their surface electrostatic potential (SP) as measured by Kelvin probe force microscopy (KPFM) (i.e. SP(AO) ≠ SP(BO2)), providing bimodal SP patterns. Note that the AO and BO2 terminations in (001)LSAT are nominally charged, (La0.18Sr0.82O)+q and (Al0.59Ta0.41O2)−q (q = 0.18), whereas the corresponding SrO and TiO2 terminations in (001)STO are neutral. These differences are exploited to create substrate patterns with different surface potential. Subsequently, by taking advantage of a layer-by-layer growth mechanism, the AO and BO2 patterns of the substrates are reproduced on ultrathin coatings (few monolayers, ML) of metallic and magnetic (LSMO) and ferroelectric (BTO) oxides, thus leading to La2/3Sr1/3O and MnO2 or BaO and TiO2 chemical surface patterns, respectively. The surface potential pattern of these A′B′O3 thin films mimics that of the underlying substrate, showing that their electrostatic properties can be tailored laterally and vertically. It turns out that the electrical conductance, as measured by conducting scanning force microscopy (CSFM), is also laterally modulated. We finish by discussing the mechanisms of the electrostatic patterns replication in A′B′O3//ABO3 heterostructures.

Figure 1. Topography of the treated LSAT substrate obtained in contact mode (a), and RHEED pattern (inset), lateral force image (b) and topographic profile (c) along the line in (a). The RHEED specular spot intensity oscillations typical of a layer-by-layer (LBL) growth mode in (e) correspond to the deposition of 4−5 ML of LSMO on the substrate shown in (a). Topography of the deposited LSMO film obtained in dynamic mode (f) with RHEED pattern of the film in the inset, phase-shift image (g) and topographic profile (h) along the line in (f). The schematics at the bottom panels illustrate the two chemical terminations (AO, BO2 and A′O, B′O2) for each surface: (d) LSAT substrate and (i) LSMO//LSAT film. Note the n + 1/2 uc (n = 0 and 1) differences in height between adjacent descending terraces in each case.

deposition (PLD) expecting a layer-by-layer (LBL) growth. The features in the RHEED specular spot intensity (Figure 1e) before the first maximum reflects differences on the first stages of growth on each substrate termination, whereas the regularity of subsequent oscillations and the postgrowth RHEED pattern (inset in Figure 1f) are signatures of a LBL growth, in agreement with the presence of steps and terraces at the film surface (Figure 1f). The step-height differences (Figure 1h) are ∼(n + 1/2) uc (here, n = 0 or 1), indicating that an extra layer has developed on one of every two terraces; this is in accordance with an initially preferential growth on one surface termination before a steady LBL growth mode sets in. This description also agrees with the phase-shift contrast between neighboring terraces observed in amplitude modulated AFM (AM-AFM) (Figure 1g) that confirms their distinct chemical termination, namely (La2/3Sr1/3)O and MnO2, as illustrated in the sketch of Figure 1i. The described topographic and chemical features (checked on other LSMO//LSAT samples) indicate that the terminations of the LSMO (A′O and B′O2) replicate the spatial distribution of the LSAT terminations underneath (AO and BO2, respectively). Replication has also been confirmed by growing LSMO on (001)STO substrates with self-ordered AO and BO2 terminations. Few monolayers of LSMO were deposited subsequently by LBL as monitored by RHEED (Figure 2a) on a (SrO/TiO2)-patterned STO substrate prepared by thermal annealing at 1300 °C for 4 h in air.9 The resulting surface (Figure 2b) shows terraces and ∼(n + 1/2) uc high steps (Figure 2c). Two kinds of terraces can be recognized: some (labeled A) have rounded edges, and others (labeled B) have

■

RESULTS Figure 1a shows the topographic atomic force microscopy (AFM) image of a (001)LSAT single-crystal after a thermal treatment at 1300 °C for 2 h. The surface consists of a stepand-terrace morphology with height differences of 1/2 uc between alternating atomically flat terraces (see height profile in Figure 1c). In agreement with recent reports,12 this surface corresponds to a nanoscale self-patterning of AO and BO2 terminations; according to Ohnishi et al.,17 the majority surface termination is of BO2 type. Lateral force imaging (Figure 1b) shows a strong contrast between adjacent terraces, confirming their different chemical nature: AO and BO2. The reflection high energy electron diffraction (RHEED) pattern of this surface is typical of an atomically flat single-crystal surface (inset in Figure 1a). As inferred from these data, the thermally treated LSAT surface is sketched in Figure 1d as a bimodal landscape. The described surface constitutes the substrate where few ML of LSMO were grown by RHEED-assisted pulsed-laser 4178



Article

corresponding to the growth of 4 ML of BTO on the chemically patterned surface of the LSMO film described above. The resulting BTO surface (Figure 2e) consists of terraces and steps preserving the morphological details (as highlighted by wavy and straight lines with right angles to guide the eye) of the chemical patterns of the buried LSMO and STO surfaces. This preservation allows spatially locating the two types of chemical terminations (BaOA′′O and TiO2 B′′O2) for the BTO surface layer. In full agreement, the height profile in Figure 2f confirms the presence of terraces and steps ∼1/2 uc height. This result means that the SrO−TiO2 surface termination pattern of the STO substrate replicates throughout perovskite layers, that is, at the top surface of LSMO//STO and up to the BTO film surface of BTO/LSMO//STO by forming in all cases A′′O−B′′O2 modulated surfaces, as sketched in Figure 2g. One straightforward implication of the LBL-promoted AO and BO2 substrate pattern replication across the epitaxial heterostructures is that unprecedented modulation of their properties should arise due to the distinct nature of these terminations for each material. To confirm these forethoughts, the surface potential and the conducting responses of the different self-patterned surfaces have been measured and are described in the following. The well differentiated AO and BO2 regions of LSAT in the topographic image (Figure 3a) give rise to a noisy but

Figure 2. (a) RHEED specular oscillations during deposition of LSMO on a chemical termination patterned STO substrate with SrO (AO) and TiO2 (BO2) terminated regions. The black arrows indicate start and stop times of deposition. The oscillations are proof of layerby-layer growth of the ultrathin LSMO film on STO (LSMO//STO). (d) RHEED oscillations during deposition of BTO on the LSMO// STO surface (BTO//LSMO//STO). (b) and (e) are the topographic images of the LSMO//STO and BTO//LSMO//STO surfaces, respectively. In both topographic images, two kinds of terraces presenting wavy or straight edges are marked as A or B to signal that they correspond to chemical terminations AO and BO2 of the respective perovskite ultrathin film top layer. Some steps of a couple of the smoother B regions have been lined to illustrate that they lie parallel to the crystallographic ⟨100⟩ directions. The wavy lined ledge and the different surface roughness (see text) can be used to spatially locate A regions (e). The different composition of terraces is also supported by the relief profiles in (c) and (f). The measured n + 1/2 uc height of descending steps corresponds to that expected for alternate planes of these perovskites (i.e. AO and BO2). The sketch in (g) illustrates the BTO//LSMO//STO structure with patterned surface and interfaces.

Figure 3. Top panels: topographic image (a) and corresponding KPFM or contact potential difference, SP, map (b) of the treated LSAT substrate exhibiting AO (A) and BO2 (B) terminated terraces. An SP profile along the line in (b) crossing AO and BO2 terraces is presented in (c). The AO-terminated region delimited by hand drawn lines in the images is marked also in (c). Bottom panels: topographic image (d) and corresponding SP map (e) of the LSMO ultrathin film on LSAT (LSMO//LSAT) showing a clear modulation replica of the substrate characteristics. As before, one AO-terminated region is marked in the images and SP profile (f). Averaged hill to valley values are ∼65 mV (c) and ∼35 mV (f).

straight edges and 90° kinks (signaled by arrows). These distinct ledge shapes are similar to those ascribed to SrO- and TiO2-terminated terraces, respectively, of treated STO substrates.9,18,19 The height differences between terraces (Figure 2c) agree with their distinct chemical termination, indicating that growth of LSMO occurs on STO, as on LSAT, by successive stacking of complete perovskite unit cell layers whatever the starting interface is (AO or BO2). The resulting topmost surface of LSMO is formed by regions of (La2/3Sr1/3)OA′O and MnO2B′O2 composition grown on the corresponding self-ordered TiO2BO2 and SrOAO terminations of the STO substrate. Outstandingly, the patterned termination phenomenon fostered by the LBL growth propagates not only through the first deposited film (LSMO) but even throughout ulterior LBL growth of a second heteroepitaxial layer, here chosen to be BaTiO3. Figure 2d shows the RHEED intensity oscillations

modulated contrast in SP (Figure 3b), more clearly seen in the corresponding line profile (Figure 3c), which permits saying that on average SP(AO)−SP(BO2) ∼ 65 mV. Visual inspection of the topographic image of the LSMO//LSAT surface (Figure 3d) allows identifying higher stripes along the step edges, similar to those already seen in Figure 1g and ascribed accordingly to the A′O terminations.17 The well-defined modulated contrast of Figure 3e indicates that SP(A′O)− SP(B′O2) ∼ 35 mV (Figure 3f). In Figure 3(a−e), one AO(A′O) region has been delimited by fine hand-drawn lines and marked at the SP profiles to help visualizing the data. Once the correspondence between terminations and their surface potential differences is established, the local conducting character of the thin LSMO film surface was investigated at the very same surface locations (Figure 4), as confirmed with the help of the dashed rectangles in Figure 4a and c and some fine 4179



Article

Figure 5. Top panels: topographic image (a) and the corresponding SP map (b) of the treated STO substrate exhibiting AO and BO2 terminated terraces, appearing as bright and dark in (b), respectively. A SP profile crossing the different terminations is presented in (c). Bottom panels: topography, current, and SP images (d, e, and h, respectively) and the corresponding line profiles (g, f, and i, respectively). The 1/2 uc high steps separating AO and BO2 terraces evidence that they correspond to alternate planes of the perovskite LSMO film. Line scans across SP and CSFM images are connected with vertical arrows to indicate the spatial correspondence between them.

Figure 4. Topography (a, c) and the corresponding SP (b) and CSFM (taken at −800 mV) (d) maps of the LSMO//LSAT surface shown in Figure 2. Total color range equals absolute values of 70 mV (b) and 300pA (d). The dash lined areas and arrows illustrate that the same surface location was recorded in both cases. As shown in the (b) and (d) regions, A′O (A) terminated exhibit a higher (lower) surface potential (work function) than regions B′O2 (B) terminated, where lower current values are detected.

Finally, the 4 ML thick BTO film grown on LSMO grown on STO (BTO/LSMO//STO) was characterized. The topographic relief images of two locations (Figure 6a and e) and their corresponding SP and CSFM maps (Figure 6b and f, respectively) are shown. As already described in Figure 2, the ∼1/2 uc steps at the BTO surface (Figure 6c) indicate the existence of the BaO and TiO2 terminations. The contrast in the corresponding SP map (Figure 6b) is about 25 mV (Figure 6d) with SP(B″O2) > SP(A″O), slightly smaller in magnitude and opposite in sign to that obtained for the LSMO//STO underneath. Consistent with all data described above, in the conductance map (Figure 6f) the B″O2 regions have higher conductance. The measured current is extremely low (tens of pA for positive voltages of some volts) due to the fact that the BTO film adds an extra thin insulating barrier to the tip-LSMO junction. It is worth highlighting that higher current is measured on the B″O2 regions of BTO that are stacked with the less conducting regions B′O2 of the free LSMO surface. This indicates that conduction at the surface is determined by the local electrostatic barrier height at the oxide surface but depends on the interfacial BTO/LSMO properties. As a final remark, we note that in all cases, regions of high(low) SP are always regions of high(low) conductance, no matter to which termination they correspond, indicating the direct relation between these magnitudes.

details (signaled by arrows in all images). The current image (Figure 4d) is a two levels map that reproduces the SP pattern (Figure 4b), such that those regions with higher(lower) conductance correspond to regions of higher(lower) SP. We note that while the chemical differences may be at the origin of the different SP values, the fact of measuring nonhomogeneous conductivity is a striking observation provided the lateral continuity and conducting character of the LSMO film, which homogeneously covers the insulating substrate. Similar behavior has been found for a range of LSMO thicknesses between 2 and 10 ML. Analogous experiments have been performed on STO substrates. Comparison of the topographic image of the bare STO (Figure 5a) with the corresponding SP map and profile (Figures 5b, c) shows that the SrO regions (bright areas) have higher SP than the surrounding TiO2 regions; this result implies SP(AO) > SP(BO2), as observed for LSAT. Statistical estimation including data from different samples10 gives a difference of SP(SrO)−SP(TiO2) ∼ 45 mV (Figure 5c). In the topographic image of the LSMO//STO (Figure 5d), approximately equally spaced and nearly parallel distinct terraces separated by ∼1/2 uc steps are visible (see profile in Figure 5g), reflecting the existence of A′O and B′O2 stripes. The developed quasi-periodic pattern is also clearly visible in the CSFM and the SP maps (Figure 5e and h, respectively). Line scans across each image (Figure 5f, i) have been connected with vertical arrows to indicate the spatial correspondence. Maxima (minima) values in surface potential are accompanied by maxima (minima) values in conductance. The surface potential difference is SP(A′O)−SP(B′O2) ∼ 30 mV about the same that for LSMO//LSAT.

■

DISCUSSION It has been shown that chemical terminations existing at the asreceived LSAT and STO surfaces self-arrange after the appropriate thermal treatment to form a pattern of AO and BO2 terraces separated by 1/2 uc high steps. The width of the AO and BO2 regions, being of about some tens of nanometers (∼50−150 nm) on the crystals analyzed here, can be controlled by substrate miscut.8 The self-ordering of AO and BO2 4180



Article

Figure 7. Sketches (not scaled) illustrating the results of SP (a and b) and conductance profiles (c) across differently terminated regions (AO and BO2) for all surfaces studied here (left, LSAT and LSMO on LSAT; center, STO and LSMO on STO; right, STO and BTO/LSMO on STO). The charge state is also given to point out the nonpolar or polar character of the distinct terminations in each perovskite. (d) Typical current−voltage (I−V) curves measured on each termination of the corresponding surfaces.

Figure 6. Topographic images (a, e) with their corresponding simultaneously measured surface potential (b) and current (f) maps for two locations of the same surface of the ultrathin BTO//LSMO// STO surface. Line profiles obtained along the lines marked in (a) and (b) are depicted in panels (c) and (d) to illustrate the correspondence between surface termination (A′′O or B′′O2) with a lower or higher surface potential value. The same correspondence made for the CSFM data (taken at 4.5 V) indicates that the current measured on A′′O is rather small but still larger than that in the B terminated regions. Total color scales correspond to absolute values of 25 mV (b) and 55 pA (f).

all systems studied here. In selected cases, we shall also make contact with fundamental theory by performing first-principles calculations of representative model systems. First, we focus on the KPFM contrast (ΔSP) observed at the bare SrTiO3 surface. The TiO2 and SrO terminations of STO are nonpolar,26 and therefore, the observed SP corrugation is expected to be due to intrinsic properties of the oxide. In particular, each termination are characterized by distinct electronic27 and/or ionic28 relaxation patterns, which will produce (via changes in the surface dipole) a lateral modulation of the local IP. First-principles calculations indeed show a marked termination-dependent difference in the IP: IP(TiO2) = 6.41 eV and IP(SrO) = 4.07 eV.29 Accordingly, one expects SP(SrO) > SP(TiO2) in qualitative agreement with the SP profile shown (Figure 7a, center and right). However, the experimental observation of ΔSP ∼ 45 mV is by far lower than the calculated ΔIP ∼ 2.34 eV. To account for this discrepancy, one can think of different possible scenarios, which we can group into two main categories: (i) the surface terminations themselves physically differ from the calculated models; (ii) the measuring technique is inadequate to quantify the ionization potential differences in these nanostructured surfaces. Within (i), we note that even though all measurements are performed at low RH conditions, the sample is brought through air from the preparation chamber to the SFM equipment and, therefore, it cannot be excluded that its surface is covered with adsorbates, in particular with H2O. To check for the impact of such adsorbates on the IP, we performed first-principles calculations of SrO- and TiO2terminated STO surfaces at various H2O coverages, exploring both dissociative and physisorbed states, as described in ref 30. We find (full details of the calculations will be reported elsewhere) that water adsorption increases IP(SrO) by, at most, 0.47 eV [A1 model of ref 30, 1 × 1 coverage] and decreases IP(TiO2) by, at most, 0.43 eV [B2 model of ref 30, √2 × √2 R45 coverage]. Accordingly, adsorption of one full single monolayer of H2O could reduce the predicted ΔIP down to

terminations in STO results from an atomic surface diffusion process minimizing the surface energy, as described elsewhere;10 a similar process seems to be at work in LSAT.12 In addition, we have shown that LBL growth of few monolayers of other perovskites (LSMO and BTO) allows obtaining surfaces that mimic the substrate termination pattern and produces a lateral modulation of surface potential and conductivity. Understanding this modulation of properties deserves some detailed discussion. During the last years, long-range electrostatic forces measured by local microscopy have been used to assist chemical identification in systems consisting of clearly different materials (metals, insulating thin films, molecules, etc).20 Quantification of SP values is now well established for metals, where it is commonly associated to local work function variations. In the case of an insulating surface, it is tempting to interpret the measured signal as variations in the local ionization potential (IP, defined as the energy difference between the valence band maximum and the vacuum level), in close analogy to the metallic case (where the IP coincides with the work function). However, one must keep in mind that the nonconducting case is substantially more challenging: First, trapped charges or defects may contribute to the local electrostatic surface potential and therefore to the contrast in the KPFM images.21−25 Second, the absolute reference for energy alignment is ill-defined in poorly conducting or dielectric samples. Being conscious of such difficulties, our discussion will rely on the comparative analysis of our results and will benefit from the sketches of Figure 7, where we summarize the SP (Figure 7a and b), conductance profiles (Figure 7c), and current−voltage characteristics (Figure 7d) of 4181



Article

the “polar” nature of the individual LSMO layers (A′O+q and B′O2−q) that are in contact with nonpolar STO. One can speculate that these effects may be qualitatively similar but quantitatively different in the case of LSMO on LSAT (Figure 7b, left), in which both surface terminations are already polar, and this could explain the 5 meV difference in the surface potential corrugation. We now briefly address the conductance profiles of LSMO on STO and LSAT as sketched in Figure 7c (center and left panels, respectively). As already commented, in all cases, the conductance profiles mimic the SP modulation of the LSMO film. As higher SP corresponds in our experiments to lower electron barrier height (see experimental details), the higher conductance in those regions is in agreement with expectations. The current−voltage (I−V) characteristics measured on each termination (Figure 7d) of the corresponding LSMO surfaces agree with this local electronic property, which results in lateral modulation. The SP and conductance profiles at the surface of the BTO/ LSMO//STO are sketched in Figure 7b, c (right). The persistence of the STO bimodal signature (Figure 7a, right) up to the free topmost surface of BTO is remarkable. However, the BTO covered surface has a SP profile reversed to those of the bare STO and LSMO//STO surfaces, being SP(BaO) < SP(TiO2). Accordingly, the corresponding conductance profile (Figure 7c, right) is also reversed. Interestingly, the I−V curve obtained on the higher SP region (Figure 7d, right) does present a nearly perfect n-type rectifying behavior of the new junction. The overall observations deserve deeper investigation out of the scope of the present work, since it may result either from the charge transfer at the newly formed BTO//LSMO interface or from the ferroelectric nature and domain structure of BTO and/or the concomitant charge screening due to adsorbates at the two BTO surface terminations. As a closing remark, we note that the complete consistency of the KPFM and CSFM measurements (which are done in very different operation modes) gives strong support to the experimental procedures and data analysis and gives further evidence of the dramatic impact of interfacial properties via substrate termination design.

1.44 eV, still far from the experimentally measured ΔSP. It is likely that other species, such as ionized adsorbates, might produce more drastic effects. Additional factors altering the local IP include surface reconstructions and/or defects such as oxygen vacancies. To clarify all these issues, further experimental and theoretical investigations are needed. Concerning possibility (ii), we note that the regions of AO and BO2 are about 50−100 nm wide, at the limit of our KPFM lateral resolution. Therefore, a reduced ΔSP might result from the otherwise unavoidable averaging produced by the tipcantilever ensemble. Note in addition that we are implicitly assuming single AO and BO2 terminated regions; however, an incomplete chemical separation, beyond the lateral resolution of KPFM, and persisting after thermal treatments, cannot be totally excluded. If it were the case, the SP measured on each terrace would be an effective average, thus resulting in a smaller ΔSP. Finally, because of the finite size of the nanostructures, the corrugation of the surface potential itself decays rapidly perpendicularly to the surface plane.25 Moving now to the bare LSAT substrates, the most striking observation when analyzing the data is that the ΔSP contrast (Figure 7a, left) is considerably larger than, but coincides in sign to, that of STO (Figure 7a, left), that is, SP(A′O) > SP(B′O2). As mentioned before, in clean LSAT, these terminations are polar, and therefore, the larger magnitude of ΔSP could arise from additional ionic and/or electronic compensation mechanisms. We turn next to the SP profiles of LSMO grown on STO and LSAT sketched in Figure 7b (central and left panels, respectively). In both cases, the SP of LSMO films replicates the SP contrast of the corresponding substrate, that is, SP(A′O) > SP(B′O2) for both LSMO//STO and LSMO//LSAT surfaces. The values of ΔSP measured at the LSMO surfaces are quite similar, about 30 and 35 mV, and smaller than ΔSP at the substrates, ΔSP(LSAT) ∼ 65 mV and ΔSP(STO) ∼ 45 mV. To understand the physical origin of the observed contrast, we invoke again two different extreme scenarios where the LSMO overlayer behaves as (i) a metal or (ii) an insulator. In case (i), the Fermi level of different LSMO regions are aligned, implying that right below the topmost LSMO surface the electrostatic potential is uniform. Then, the only source of surface potential modulation should originate from the termination-dependent local work function of LSMO, regardless of the underlying substrate. The similar ΔSP measured in LSMO//STO and LSMO//LSAT indeed suggests that this might be a reasonable hypothesis. Conversely, in the scenario (ii), we expect the electronic levels of the LSMO film to align separately to the underlying substrate. In this case, in addition to the termination dependence of the LSMO work function described above, we must also take into account the band offset at the buried LSMO//STO (and LSMO//LSAT) interface. To get some insight into Schottky barrier formation for the LSMO case, we refer to recent works addressing the importance of the STO interfacial termination on the band alignments and its influence on the Schottky barrier height (SBH) of La0.6Sr0.4MnO3−TiO2 and La0.6Sr0.4MnO3−SrO.31,32 Minohara et al.31 used photoemission spectroscopy to determine the potential at the interface of La0.6Sr0.4MnO3 films grown on TiO2 and SrO terminated STO substrates. They found a reduction on the SBH of 0.5 eV and −0.4 eV, at the /.../MnO2/ La1−xSrxO//TiO2 and /.../La1−xSrxO/MnO2//SrO interfaces, respectively, and suggested that this reduction may be due to build-in dipoles. These, microscopically, would originate from

■

SUMMARY AND CONCLUSIONS We have shown here that the distinct chemical terminations (AO and BO2) of (001)LSAT and (001)STO substrates can be induced to self-order forming distinctive regions some tens of nanometers wide and separated by steps of 1/2 uc height. Associated with the topographic nanostructuration, the chemical terminations of the substrates promote the formation of a patterned surface potential. We have found that the SP contrast at surfaces of STO and LSAT substrates originates from the different electrostatic potential at the (possibly reconstructed and adsorbate-covered) AO and BO2 terminations. To further elucidate the fundamental mechanisms that are at the origin of the measured contrast, we performed firstprinciples calculations of selected SrTiO3 surface models. While a substantial disagreement remains between the calculated ionization potential differences and the measured ΔSP, our analysis constitutes an important first step toward a better theoretical understanding of Kelvin-probe measurements performed on insulating surfaces. These nanostructured surfaces have been used as templates for the epitaxial growth of functional perovskite oxides (La2/3Sr1/3MnO3 and BaTiO3). By exploiting a layer-by-layer 4182



Article

to correlate local electric properties with the chemically different terminations. Experimental details of these measuring modes and setup are given elsewhere.10,39,40 Conducting B-doped diamond coated Si tips (by Nanosensors) mounted on cantilevers with k = 40 N/m were employed. KPFM is a dynamic mode that measures the so-called contact potential difference or surface potential (SP) through measurement of the electrostatic force between a metallic tip and the sample. The tip and sample are not in direct contact but are electrically connected via electronics. As a result, their Fermi levels align, creating a SP, and thus, a tip−sample electrostatic force develops. Though the method has been applied to a wide variety of materials, including insulators,20 KPFM was first introduced to investigate the work function of metals,41 and it is understood as follows. If the work function of the KPFM tip (vibrating electrode) is φtip and φsample is that of the sample, then the contact potential difference between tip and sample is SPsample = (φtip − φsample)/e = Δφs/e, where e is the electric charge. The tip, which is at some distance (∼15 nm) above the sample surface, is driven by an oscillating voltage Vtip = Vdc + Vac sin(ωt), and as a consequence, the electrostatic force between them is Fes = −1/2 ∂C/∂z [(Vdc − Δφs/e) + Vac sin(ωt)]2, such that the Vdc nullifying the first harmonic of this electrostatic force is just the SPsample. Often, as in the present work, in order to minimize cantilever contributions, the same minimization procedure is used but applied to the force gradient measurement:42

growth mode, the morphology of the substrate consistently propagates through the film, and allows obtaining patterns of well-defined chemical terminations (A′O and B′O2) at their surface. The surface potential profiles of LSMO grown on STO and LSAT mimic those of the substrate. Therefore, the substrate acts not only as template determining morphology and chemical termination but also allows for the lateral tuning of the electrical properties of the film surface. We have also shown that the SP contrast of the nanostructured LSMO films is slightly smaller than that of the underlying substrates; however, ΔSP shows only a weak dependence on the nature of the substrate, suggesting that the distinct (La,SrO) and (MnO2) termination might have the dominant role on the SP contrast. Interestingly, the SP contrast at the surface of BTO appears reversed with respect to that of the underlying substrate and electrode, suggesting that the ferroelectric nature of BTO, e.g. via a hypothetical selective formation of ferroelectric domains on distinct polar terminations, could play a role. The conductance maps of all the investigated surfaces indicate that higher conductance is ubiquitously associated to higher SP. This close correlation, that is interpreted by arguing that high(low) surface potential correspond to low(high) local electrostatic barrier height, should contribute to a better understanding of the KPFM on insulating materials. From a more general perspective, the results reported here well illustrate that nanoscale lateral and vertical patterning of functional properties can be achieved by exploiting selforganization of chemical terminations in perovskite oxides.

■

∂Fω ∂ 2C =− [(Vdc − Δφs /e) + Vac sin(ωt )]2 ∂z ∂z 2 When the sample surface has regions with different electronic properties, as the two chemical terminations (AO and BO2) of interest here, the contact potential of each region would be given by SPA = (φtip − φA)/e and SPB = (φtip − φB)/e, respectively. Consequently, the contact potential difference between them is ΔSP = SPA − SPB = − (φA − φB), independent of the material the tip is made of. Therefore, the contrast of the surface potential maps obtained by scanning the tip over the surface directly reflects the local variations of the surface work function. Note that, by definition, the higher the local SP the lower the local effective work function (ΔSP = −ΔφAB). The method lacks the desired lateral resolution for measuring nanostructures43 but avoids problems derived from tip details (end atoms and geometry) appearing when attempting to determine absolute SP values. Contrary to KPFM, for CSFM, the conducting tip is placed in direct contact with the sample. To ensure the acquiring of current maps in a noninvasive manner (no sample indentation), CSFM is performed under controlled constant load, that is, by using a force feedback while measuring the current between tip and sample. Simultaneous topographic images z(x,y) and current maps I(x,y) were recorded over a given region at a fixed voltage. In our setup, the sample was always grounded and the voltage was applied to the tip (Vtip). Provided the insulating character of the substrates used (STO and LSAT), the direct electric contact to ground is established through a metallic clamp (counter electrode) firmly attached to the surface film at the sample border (millimeters apart from the tip−surface contact). An external I−V converter (Stanford Research Systems) was used to access a wide range of compliance currents (1 pA to 1 mA). Whereas for topographic and lateral force images the color code is the commonly used, bright for high and dark for low, for the current maps, it depends on the voltage sign. Thus, higher currents appear darker in CSFM images taken at negative Vtip, while brighter currents appear for positive V tip . At selected positions of interest (AO and BO2 terminations), I−V characteristics were measured as a function of Vtip, starting from negative tip voltages. For reliable data comparison, the same tip has been used in all the CSFM experiments of at least one series of samples. Note that combining KPFM and CSFM to characterize the same surface locations guaranties local properties correlation but is challenging, since it requires passing from noncontact dynamic conditions at which the tip oscillates at some distance away from the surface to direct tip−surface contact at which the electric current

EXPERIMENTAL SECTION

Material Elaboration. (001)SrTiO3 (STO) and (001)La0.18Sr0.82)(Al0.59Ta0.41)O3 single-crystalline substrates from CrysTec GmbH33 were thermally treated at high temperature (up to 1300 °C) in air in a dedicated tubular furnace. La2/3Sr1/3MnO3 (LSMO) films and BaTiO3 (BTO) films were grown by pulsed laser deposition (PLD), with a KrF excimer laser (λ = 248 nm).34 A differentially pumped 30 keV RHEED system was used to monitor the growth in real-time. We denote by films//substrate, the distinct films-substrate combinations described here: LSMO//LSAT, LSMO//STO, and BTO/LSMO//STO. The lattice parameter of LSMO (pseudocubic cell), BTO (a-parameter), LSAT, and STO are 3.89 Å, 3.994 Å, 3.868 Å, and 3.905 Å, respectively. The resulting lattice mismatches f (calculated as 100 × (asubs − afilm)/afilm) for LSMO//LSAT, LSMO//STO, and BTO// STO are f = −0.57%, 0.39%, and −2.23%, respectively). High quality epitaxial growth in a layer-by-layer fashion (one-by-one uc stacking) is made possible by the moderately small lattice mismatches. Scanning Probe Microscopy Characterizations. Scanning probe microscopy (SPM) measurements were carried out at room temperature using different instruments (5100SPM and 5500LS from Agilent Technologies and Cervantes from Nanotec Electrónica). All data were analyzed by using the WSxM software.35 Amplitude modulation atomic force microscopy (AM-AFM, dynamic mode with fixed frequency) working in ambient conditions was used to investigate the surface morphology and the lateral distribution of the chemical terminations by phase-shift imaging.36,37 Friction force microscopy (FFM) was used (under low humidity conditions RH ≤ 5% in a N2 flux) to assess the chemical differentiation obtained from phase-shift data in dynamic mode.38 Conducting scanning force microscopy (CSFM) was used to obtain direct contact tip−surface responses, whereas phase-shift contrast (in AM-AFM) and Kelvin probe force microscopy (KPFM), the latter in the retrace mode, were used to accurately determine local variations in surface contact potential difference (ΔSP). KPFM data were crosschecked by point electrostatic force spectroscopy. At each specific surface, distinct nanoscale characterizations were made on exactly the same locations 4183



Article

between tip and surface is measured. Another advantage of combining contact and noncontact measurements, lateral force and phase-shift contrasts with morphological details (step heights, step ledges shape, and local roughness) is that it allows recognition and spatial localization of the different chemical terminations coexisting on the surfaces. Apart from high stability, this combination of experimental techniques implies a compromise when choosing the tips and cantilevers employed. To overcome resolution related drawbacks, the morphological details were cross-checked separately by high resolution topographic data obtained by using nonconducting ultrasharp Si tips with nominal radius of 10 nm.

■

(17) Ohnishi, T.; Takahashi, K.; Nakamura, M.; Kawasaki, M.; Yoshimoto, Y.; Koinuma, H. Appl. Phys. Lett. 1999, 74, 2531. (18) Fompeyrine, J.; Berger, R.; Lang, H. P.; Perret, J.; Mächler, E.; Gerber, Ch.; Loquet, J. P. Appl. Phys. Lett. 1998, 72, 1697. (19) Koster, G.; Rijnders, G.; Blank, D. H. A.; Rogalla, H. Phys. C 2000, 339, 215. (20) Barth, C.; Foster, A. S.; Henry, C. R.; Shluger, A. L. Adv. Mater. 2011, 23, 477. (21) König, T.; Simon, G. H.; Heinke, L.; Lichtenstein, L. Belstein J. Nanotechnol. 2011, 2, 1. (22) Rosenwaks, Y.; Shikler, R.; Glatzel, Th.; Sadewaser, S. Phys. Rev. B 2004, 70, 085320. (23) Barth, C.; Henry, C. R. Phys. Rev. Lett. 2007, 98, 136804. (24) Kalinin, S. V.; Bonnell, D. A. Nano Lett. 2004, 4, 555. (25) Kalinin, S. V.; Bonnell, D. A. Phys. Rev. B 2001, 63, 125411. (26) Stengel, M. Phys. Rev. B 2011, 84, 205432. (27) Stengel, M. in preparation (28) Eglitis, R. I.; Vanderbilt, D. Phys. Rev. B 2008, 77, 195408. (29) Mrovec, M.; Albina, J. M.; Meyer, B.; Elsasser, C. Phys. Rev. B 2009, 79, 245121. (30) Guhl, H.; Miller, W.; Reuter, K. Phys. Rev. B 2010, 81, 155455. (31) Minohara, M.; Yasuhara, R.; Kumigashira, H.; Oshima, M. Phys. Rev. B 2010, 81, 235322. (32) Hikita, Y.; Nishikawa, M.; Yajima, T.; Hwang, H. Y. Phys. Rev. B 2009, 79, 073101. (33) Crystec GmbH, http://www.crystec.de. (34) Bachelet, R.; Pesquera, D.; Herranz, G.; Sánchez, F.; Fontcuberta, J. Appl. Phys. Lett. 2010, 97, 121904. (35) Horcas, I.; Fernández, R.; Gómez-Rodríguez, J. M.; Colchero, J.; Gómez-Herrero, J.; Baro, A. M. Rev. Sci. Instrum. 2007, 78, 013705. (36) Cleveland, J. P.; Anczykowski, B.; Schmid, A. E.; Elings, V. B. Appl. Phys. Lett. 1998, 72, 2613. (37) García, R.; Pérez, R. Surf. Sci. Rep. 2002, 47, 197. (38) Carpick, R. W.; Salmeron, M. Chem. Rev. 1997, 97, 1163. (39) Moreno, C.; Munuera, C.; Valencia, S.; Kronast, F.; Obradors, X.; Ocal, C. Nano Lett. 2010, 10, 3828. (40) Munuera, C.; Barrena, E.; Ocal, C. Nanotechnology 2007, 18, 125505. (41) Nonnenmacher, M. O.; O’Boyle, M. P.; Wickramasinghe, H. K. Appl. Phys. Lett. 1991, 58, 2921. (42) Albrecht, T. R.; Grütter, P.; Horne, D.; Rugar, D. J. Appl. Phys. 1991, 69, 668. (43) Glatzel, Th.; Sadewasser, S.; Lux-Steiner, M. Ch. Appl. Surf. Sci. 2003, 210, 84.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; [email protected]; fontcuberta@ icmab.es. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Financial support by the Spanish Government [Projects MAT2010-20020, MAT2010-18113, MAT2011-29269-C03, and NANOSELECT CSD2007-00041], Generalitat de Catalunya (2009 SGR 00376), and European Commission (Project NMP3-SL-2009-228989 “OxIDes”) is acknowledged.

■

REFERENCES

(1) Lee, J. H.; Fang, L.; Vlahos, E.; Ke, X.; W. Jung, Y.; Kourkoutis, L. F.; Kim, J. W.; Ryan, P. J.; Heeg, T.; Roeckerath, M.; Goian, V.; Bernhagen, M.; Uecker, R.; Hammel, P. C.; Rabe, K. M.; Kamba, S.; Schubert, J.; Freeland, J. W.; Muller, D. A.; Fennie, C. J.; Schiffer, P.; Gopalan, V.; Johnston-Halperin, E.; Schlom, D. G. Nature 2010, 466, 954. (2) Garcia, V.; Bibes, M.; Bocher, L.; Valencia, S.; Kronast, F.; Crassous, A.; Moya, X.; Enouz-Vedrenne, S.; Gloter, A.; Imhoff, D.; Deranlot, C.; Mathur, N. D.; Fusil, S.; Bouzehouane, K.; Barthélémy, A. Science 2010, 327, 1106. (3) Ohtomo, A.; Hwang, H. Y. Nature 2004, 427, 423. (4) Bert, J. A.; Kalisky, B.; Bell, Ch.; Kim, M.; Hikita, Y.; Hwang, H. Y.; Moler, K. A. Nat. Phys. 2011, 7, 767. (5) Kawasaki, M.; Takahashi, K.; Maeda, T.; Tsuchiya, R.; Shinohara, M.; Ishiyama, O.; Yonezawa, T.; Yoshimoto, M.; Koinuma, H. Science 1994, 266, 1540. (6) Koster, G.; Kropman, B. L.; Rijnders, G. J.; Blank, H. M.; Rogalla, D. H. A. Appl. Phys. Lett. 1998, 73, 2920. (7) Leca, V.; Blank, D. H. A.; Rijnders, G. Termination Control of NdGaO3 Crystal Surfaces by Selective Chemical Etching. 2012, arXiv:1202.2256v2.arXiv.org e-Print archive http://arXiv.org/pdf/ 1202.2256v2. (8) Bachelet, R.; Sánchez, F.; Santiso, J.; Munuera, C.; Ocal, C.; Fontcuberta, J. Chem. Mater. 2009, 21, 2494. (9) Bachelet, R.; Sánchez, F.; Palomares, F. J.; Ocal, C.; Fontcuberta, J. Appl. Phys. Lett. 2009, 95, 141915. (10) Paradinas, M.; Garzón, L.; Sánchez, F.; Bachelet, R.; Amabilino, D. B.; Fontcuberta, J.; Ocal, C. Phys. Chem. Chem. Phys. 2010, 12, 4452. (11) Kleibeuker, J. E.; Koster, G.; Siemons, W.; Dubbink, D.; Kuiper, B.; Blok, J. L.; Yang, C. H.; Ravichandran, J.; Ramesh, R.; Ten Elshof, J. E.; Blank, D. H. A.; Rijnders, G. Adv. Funct. Mater. 2010, 20, 3490. (12) Bachelet, R.; Ocal, C.; Garzón, L.; Fontcuberta, J.; Sánchez, F. Appl. Phys. Lett. 2011, 99, 051914. (13) Noguera, C. In Physics and Chemistry at Oxide Surfaces; Cambridge University Press: New York, 1996. (14) Noguera, C. J. Phys.: Condens. Matter 2000, 12, R367. (15) Bonnell, D. A.; Garra, J. Rep. Prog. Phys. 2008, 71, 044501. (16) Deak, D. S. Mater. Sci. Technol. 2007, 23, 127. 4184


Nanoscale Laterally Modulated Properties of Oxide Ultrathin Films by

Recommend Documents