Electric and Mechanical Switching of Ferroelectric and ...€¦ · Indeed, the flexoelectric...

(1 of 10) 1701614© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.com

Electric and Mechanical Switching of Ferroelectric and Resistive States in Semiconducting BaTiO3–δ Films on Silicon

Andrés Gómez, José Manuel Vila-Fungueiriño, Rahma Moalla, Guillaume Saint-Girons, Jaume Gázquez, María Varela, Romain Bachelet, Martí Gich, Francisco Rivadulla, and Adrián Carretero-Genevrier*

DOI: 10.1002/smll.201701614

Materials that can couple electrical and mechanical properties constitute a key element of smart actuators, energy harvesters, or many sensing devices. Within this class, functional oxides display specific mesoscale responses which often result in great sensitivity to small external stimuli. Here, a novel combination of molecular beam epitaxy and a water-based chemical-solution method is used for the design of mechanically controlled multilevel device integrated on silicon. In particular, the possibility of adding extra functionalities to a ferroelectric oxide heterostructure by n-doping and nanostructuring a BaTiO3 thin film on Si(001) is explored. It is found that the ferroelectric polarization can be reversed, and resistive switching can be measured, upon a mechanical load in epitaxial BaTiO3−δ/La0.7Sr0.3MnO3/SrTiO3/Si columnar nanostructures. A flexoelectric effect is found, stemming from substantial strain gradients that can be created with moderate loads. Simultaneously, mechanical effects on the local conductivity can be used to modulate a nonvolatile resistive state of the BaTiO3−δ heterostructure. As a result, three different configurations of the system become accessible on top of the usual voltage reversal of polarization and resistive states.

Semiconducting Films

A. Gómez, Dr. J. Gázquez, Dr. M. GichInstitut de Ciència de Materials de Barcelona ICMABConsejo Superior de Investigaciones Científicas CSICCampus UAB 08193, Bellaterra, Catalonia, Spain

Dr. J. M. Vila-Fungueiriño, Dr. A. Carretero-GenevrierInstitut d’Électronique et des Systèmes (IES)UMR 5214CNRS-UM2.Batiment 5, 860 Rue Saint Priest 34095, Montpellier, FranceE-mail: [email protected]

Dr. R. Moalla, Dr. G. Saint-Girons, Dr. R. BacheletInstitut des Nanotechnologies de Lyon (INL)CNRS - École Centrale de Lyon36 avenue Guy de Collongue 69134, Ecully, France

Dr. M. VarelaDepartamento de Física de Materiales and Instituto PluridisciplinarUniversidad Complutense de MadridMadrid 28040, Spain

Prof. F. RivadullaCentro de Investigación en Química Biolóxica e Materiais Moleculares (CIQUS), and Departamento Química-FísicaUniversidade de Santiago de Compostela15782, Santiago de Compostela, Spain

1. Introduction

The development of new functional materials with multi-stable states that can be switched by external stimuli

constitutes a most promising avenue toward improved sensors and also for information storage technologies. Transition metal oxides are robust materials, highly appealing for these purposes. Their chemical compatibility

small 2017, 1701614

www.advancedsciencenews.com

http://doi.wiley.com/10.1002/smll.201701614

full papers

1701614 (2 of 10) www.small-journal.com © 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

with silicon makes them attractive for integration within complementary metal-oxide semiconductor device tech-nologies. On the other hand, the strong electronic corre-lation characteristic of 3d electrons, along with the large coupling between the charge, spin, and orbital degrees of freedom introduce cross-coupled ferroic orders[1] which can be tuned under externally applied fields. In fact, the sensitivity to external stimuli can be signifi-cantly enhanced in epitaxial oxide thin films and hetero-structures, where low dimensionality and strain may be responsible for the emergence of unexpected mesoscale responses, absent in the corresponding bulk phases.[2,3] Thin films of ferroelectric oxides epitomize this scenario: they show cross-coupled responses by confronting nano- and mesoscale phenomena with a high sensitivity to the material microstructure.[4] For example, the absence of an intrinsic ferroelectric size effect in thin-film ferroelectric perovskites[5] spurred research into the use of these mate-rials as tunnel barriers, leading to the discovery of tun-neling electroresistance.[6]

In fact, at small sizes, intrinsic changes of polariza-tion are coupled to other extrinsic effects such as defects or microstructure.[7] For instance, point defects such as oxygen vacancies, which promote n-type doping, may have a strong influence on the dielectric properties of ferroelectric films. They are at the origin of a memris-tive behavior which can be exploited for next generation nonvolatile resistive random access memory devices.[8] Regarding microstructure, a flexoelectric effect (cou-pling polarization and strain gradients) has recently been claimed to explain how a large local deformation per-formed by the tip of an atomic force microscope (AFM) can mechanically induce ferroelectric polarization[9] and resistive switching[10] in BaTiO3−δ (BTO) ferroelectric tunnel junctions. Indeed, the flexoelectric coefficients are expected to be larger for epitaxial strained insulator BTO,[11] and even much higher for n-type semiconducting BaTiO3−δ single crystals.

[12] Therefore, highly deformable structures such as 1D columnar nanostructured epitaxial BTO ferroelectric films may facilitate the switching at a lower force threshold, allowing the mechanical switching in thicker films. Moreover, application of a force on top of nonhomogeneous columnar-like structures with a varying thickness along their length will naturally produce a strain gradient.[13]

Based on these ideas, we designed a nanostructured multilayer to amplify the flexoelectric response of n-type BaTiO3−δ films. We combine chemical solution deposition and molecular beam epitaxy (MBE) to grow columnar hetero structures of epitaxial BTO/La0.7Sr0.3MnO3 (LSMO)/SrTiO3 (STO) on Si(001) wafers. We demonstrate that the columnar BTO/LSMO/STO heterostructures show a simultaneous and independent mechanically controlled switching of the ferroelectric polarization and the resis-tive state at room temperature. This mechanically assisted switching of polarization and electrical conductivity is compatible with the application of a voltage to reverse those bistable states, giving rise to three different configu-rations of the system.

2. Preparation of Epitaxial Heterostructures and Their Microstructure

To fully exploit the flexoelectricity and resistive switching properties of ferroelectrics into future devices, there are three main challenges to overcome: (i) to perform the com-plete polarization of the material at low force thresholds, (ii) to investigate the mechanical switching between different resistive states, and (iii) to further integrate these films on silicon platforms.

Regarding the last challenge, the monolithic integration of ferroelectric oxides on silicon is difficult due to the uneasy matching of dissimilar (structurally, thermally, and in gen-eral chemically reactive) oxides with silicon. Nevertheless, McKee et al.[14] demonstrated the possibility to grow crys-talline STO films on Si, setting the basis of oxide integration by MBE.[15] However, MBE presents several disadvantages for the epitaxial growth of ternary and quaternary metal oxide nanostructures.[16] Developing such epitaxial oxide nanostructures with controllable shapes and morphologies requires top-down approaches, consisting of expensive lithog-raphy or sophisticated and tedious electron and ion beam lithography.[17]

On the contrary, chemical solution deposition (CSD) appears as a straightforward bottom-up approach to design advanced functional materials with a high degree of control over the stoichiometry, microstructure, and texture[18] over large areas.[16,19] Unfortunately, direct CSD deposition of many ferroelectric oxides on Si/SiO2 is hampered by chem-ical reactivity of alkaline-earth ions.[18d,20] Therefore, the mul-tilayered nanostructures studied in this work were prepared by an innovative process which combines the deposition of oxide films by MBE and polymer assisted deposition (PAD) processes. We selected the PAD methodology because among the soft chemistry techniques that can be combined with physical deposition processes,[16] it is the most suitable to produce high quality epitaxial complex and multilayer struc-ture films.[16,21]

The complex epitaxial heterostructures were grown in three different stages, as illustrated in Figure 1 and Figure S1 (Supporting Information): (1) growth of epitaxial SrTiO3(001) layer (≈15 nm bottom layer) on Sr-passivated Si(001) substrates by MBE; (2) deposition of an epitaxial LSMO(001) interlayer (≈30 nm interlayer) by spin coating using PAD; and (3) growth of epitaxial BaTiO3−δ layer (≈45 nm top layer) by a second high-vacuum MBE process to complete the final BTO/LSMO/STO/Si(001) heterostructure.

The resulting multilayer system has a rather complex microstructure. Field emission scanning electron microscopy images show a columnar microstructure of the BTO/LSMO/STO/Si(001) heterostructures, (see Figure 2a). Studies of the structural features of a multilayer by X-ray diffraction (XRD) (see the Supporting Information) and scanning trans-mission electron microscopy (STEM) confirmed this unique microstructure which is reminiscent of a distorted truncated pyramid (see Figure 2b,c).

The atomic resolution Z-contrast image shown in Figure 1 reveals that a controlled excess of Sr during the first growth step produces stacking faults in the crystalline STO

small 2017, 1701614


(3 of 10) 1701614© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.comsmall 2017, 1701614


Figure 1. Schematic diagram illustrating a new preparation approach combining MBE and PAD. The deposition parameters were optimized to produce epitaxial growth of columnar porous BTO/LSMO/STO/Si(001) heterostructures. A cross-sectional atomic resolution Z-contrast image of the crystalline STO layer obtained after the step (1) is also shown.

Figure 2. a) 45°-tilted low magnification FEG-SEM image of a columnar like porous epitaxial BTO/LSMO/STO/Si(001) thin film grown by the combination of MBE and PAD methods. b) Low magnification Z-contrast image of a porous epitaxial BTO/LSMO/STO/Si(001) multilayer viewed along the [110] Si axis (left panel) along with a sketch of the heterostructure (right panel). c) High resolution Z-contrast image of a single BTO column viewed along the BTO [100]-crystallographic direction. It shows the epitaxial growth of the BTO onto the LSMO layer, grown by MBE and PAD, respectively. d) High resolution Z-contrast image of the BTO/LSMO interface (yellow dashed line) viewed along the Si [110]-crystallographic direction. e) Higher resolution Z-contrast image of the coherent interface between the BTO and LSMO layers. The inset shows the region were an EEL spectrum image was acquired. Color elemental mapping produced by overlaying the Ti L2,3 (yellow), Ba M4,5 (blue), Mn L2,3 (green), and La M4,5 (red) elemental maps, displaying a high quality and abrupt chemical interface between BTO and LSMO layers, the structure of which are sketched above and below the elemental map, respectively.

full papers


matrix (see Figures S1 and S2 in the Supporting Informa-tion for more details). The resulting microstructure (sketched also in Figure 1) consists of antiphase domains, separated by SrO stacking faults, with an average size of about 25 nm (see Figure S2b in the Supporting Information). A detailed study of the crystallization and growth of STO layer on Si sub-strates can be found elsewhere.[22] The result is an array of STO grains slightly misaligned on either side of the stacking fault yielding a high mosaicity,[22] also evidenced by means of XRD, see Figure S2c (Supporting Information). This struc-tural feature can be used to induce a columnar porous struc-ture to the LSMO film subsequently deposited by PAD in step (2) (see Figure 2b and Figures S3 and S4 in the Sup-porting Information for more details).

The final step (3) comprises a second MBE high-vacuum deposit of a BTO film. The conformal growth of the MBE technique enables the transfer of LSMO columnar structure to the BTO layer, as confirmed by the low-magnification Z-contrast image shown in Figure 2b. Despite the porous character of both the LSMO (≈30 nm inter layer) and the BTO (≈45 nm top layer) and the resulting columnar struc-ture of the latter, high quality and epitaxial BTO grains grow onto the LSMO layer, as evidenced in the high-resolution Z-contrast image of the LSMO/BTO interface and in the detailed structural study shown in Figure 2c and Figure S5 (Supporting Information). Electron energy-loss spectroscopy (EELS) measurements with atomic resolution (Figure 2e) show some cation intermixing restricted to the interface, in agreement with the contrast observed in the Z-contrast image (Figure 2d).

These results, along with the thermal stability shown by the STO layer (see Figure S6 in the Supporting Information), demonstrate that combining MBE and PAD allows circum-venting MBE limitations on growing good quality LSMO on STO/Si, which had not been achieved by MBE so far.

3. Characterization of Ferroelectricity and Transport Properties of Heterostructures

The electromechanical properties of the heterostructure were investigated using piezoresponse force microscopy (PFM)[23] at room temperature. Ferroelectricity was tested by recording a 4 × 4 µm squared area with a positive sample bias of +7 VDC. Next, a 2 × 2 µm square was recorded inside the large one using the opposite bias, −7 VDC. The recorded areas were then scanned using PFM with 2 VAC tip bias, as shown by the PFM amplitude and phase images in Figure 3a,b, respectively. The clear distinction of two zones with opposite poling with a full 180° phase shift upholds the ferroelectricity of the sample.[24] Additional spectroscopy sweeps were performed by biasing the tip with 2 VAC, while at the same time the DC bias was swept from 0 VDC to +5 VDC then to −5 VDC and, finally, back to +5 VDC, as shown in Figure 3d. Butterfly like loops corroborated the ferroelec-tricity of the columnar BTO film, which presented a coercive field of 422 kV cm−1 and a small imprint field (see Figure 3c).

Following an approach similar to that of Lu et al.,[9] we studied the flexoelectric properties of our heterostructures

(see Figure 3d). First, a 4 × 4 µm square was poled up by applying +7 VDC to the Si substrate and subsequently, two smaller inner rectangles were recorded in this poled area. The left rectangle was recorded by applying a force of 1000 nN, while the right rectangle was recorded with a voltage of −5 VDC. Notice that neither the mechanical nor electrical poling caused an irreversible damage to the sample, as shown by the topographic images (see Figure S7e,f in the Supporting Information). The same area was then rescanned by PFM, applying 200 nN and with 2 VAC. The phase image of this scan (Figure 3d) shows that in the inner rectangles, the up polarization has been reversed, by the applied stress in the left rectangle and by a negative DC bias in the right rec-tangle, revealing the mechanical switching of the polarization.

So far, this type of mechanical switching of the polariza-tion has been only achieved on ultrathin BTO films[9,25] or, very recently, in thicker BiFeO3 films but applying substan-tially larger forces.[26]

Also, very importantly, we were unable to induce the mechanical switching of the polarization in dense BTO/LSMO heterostructures fabricated by similar PAD–MBE method, but directly on STO(001) substrates (see Figure S8 in the Supporting Information). Therefore, we ascribe the flexoelectric response of the columnar BTO nanostructured films to their truncated pyramid geometry, which leads to

small 2017, 1701614


Figure 3. SPM measurements at room temperature showing the electrically and mechanically ferroelectric polarization in eptiaxial columnar BTO/LSMO/STO/Si(001) films. a) PFM amplitude and b) phase images of the 45 nm thick BaTiO3 film on silicon after the generation of the downward-oriented domains by electrical poling. c) Local PFM hysteresis loops measured in the same BaTiO3 film (top: phase signal, bottom: amplitude signal). d) Phase image of the 45 nm thick BaTiO3 film after the generation of the downward-oriented domains by mechanical loading force (left) and electrical poling (right). A saturated amplitude signal indicates fully polarized states. e) Schematic diagram illustrating the switching of the polarization with an electric field or a mechanical force across the columnar BaTiO3/La0.7Sr0.3MnO3/SrTiO3/Si(001) heterostructure at room temperature.


substantial strain gradients under moderate loads. Indeed, compressing materials of truncated square pyramid geom-etries has become a standard approach to generate strong strain gradients and study the flexoelectric effect,[11b,13] even though the quantitative analysis of the experiments is non-trivial due to the complexity of the stress fields developed in such geometries.[27] We can, though, estimate the strain gra-dient produced by applying a force of 1000 nN to a truncated pyramid with the dimensions of that displayed in Figure 2c to be ≈105 m−1 (see details in Figure S10 in the Supporting Information). As the strain gradient is inversely propor-tional to the elastic modulus, another way of understanding the development of a large flexoelectric polarization in the columnar BTO film is considering that such microstructure presents an elastic modulus which is significantly smaller than that of the dense BTO. To find evidence of this, we performed AFM indentation experiments on the dense and columnar BTO films, as well as on the much harder (softer) lithium nio-bate (low density polyethylene (LDPE) polymer). Figure 4a shows the indentation areas at the maximum tip penetra-tion, used to compute the stress also from the force and the tip penetration. As expected, the curves of stress versus the AFM piezo displacement in Z during the unloading show qualitatively that the elastic modulus of the nanostructured BTO film is significantly smaller than that of the dense BTO film (Figure 4b).

Here, we would like to remark that a larger degree of relaxataion of the porous BTO layer with respect to the dense one will also add to the reduction of the bulk mod-ulus induced by the porous nanostructure. Although from the analysis of the reciprocal space map (RSM) we did not find any significative difference between the degree of epitaxial relaxation between dense/porous BTO, a small contribution from this effect cannot be completely excluded.

Finally, we would like to note that since the tip size is comparable to the typical sizes of the columns, it is plausible that upon entering or leaving an isolated pyramidal column, the strain gradient exceeds that created by the vertical load

of the tip at the center of the structure. This demonstrates that our nanostructuring strategy constitutes an effec-tive approach to the flexoelectric control of polarization in thicker films at moderate loads.

Additional ferroelectric and flexoelectric switching exper-iments were performed at different temperatures. Whereas the polarization can be switched by an electric field at 223 K, the mechanical switching of ferroelectric polarization is not possible (see Figure S9d in the Supporting Information) at this temperature, since the phase transition of the BTO from tetragonal to orthorhombic is accompanied by a decrease of the flexoelectric coupling coefficient.[28] Moreover, at 438 K (above Curie temperature of BTO) the PFM measurements do not show the characteristic ferroelectric response.

3.1. Resistive Switching Behavior of Semiconducting Nanostructured BTO Films

Besides controlling the polarization state of BTO, electric fields and strain gradients can also switch the resistive state of the columnar heterostructure. The high vacuum synthesis in the MBE introduces a certain amount of oxygen vacancies and therefore a finite conductivity of the BTO film.

The charge transport and resistive switching behavior of nanostructured BTO films was investigated using con-ductive atomic force microscopy (CAFM) with the setup of Figure 5a. By biasing the LSMO layer with −6 V, the hetero-structure is set into a low resistance state (LRS) while a +6 V bias sets a high resistance state (HRS) (Figure 5b).

The force dependence of BTO conductivity (see Figure 5c) was tested by recording a LRS on a 4 × 4 µm squared area with a negative bias of −7 VDC applied to the LSMO layer and, subsequently, two smaller inner rectangles: the left rectangle in 0 VDC with an applied force of 1000 nN and the right rectangle with +7 VDC applied to LSMO. Next, the same area was scanned in current sensing mode with −2 VDC bias and with 200 nN load to obtain the image

small 2017, 1701614


Figure 4. a) Comparison of the mechanical properties on columnar and dense BTO films, hard lithium niobate, and LPDE soft polymer by indentation technique. b) Z piezo displacement versus pressure curves of the different materials that show the material strain given by the elastic deformation plus the bending of the tip cantilever. Notice that the measurements indicate that the elastic modulus for columnar BTO is significantly lower than that of dense BTO and thus at larger strain gradient can also be expected for it for given applied load as the strain gradient is inversely proportional to the elastic modulus.

full papers


shown in Figure 5c. It shows that the area mechanically recorded and the area poled with +7 VDC present similar contrast, i.e., comparable average currents which are char-acteristic of the HRS. In order to check the repeatability of this effect, we renewed to LRS a previously mechanically recorded HRS, by applying a negative bias (see Figure S7d in the Supporting Information). The effect was also studied with current versus force including time axis spectroscopy experiments. The spectroscopy data were obtained within an area prerecorded with −7 VDC, see Figure 5d, which shows that once the tip engages the sample (point A), a cur-rent starts flowing. However, at a force of 1500 nN (point B), the current decreases abruptly. Typically, the measured current in CAFM increases with the applied load due to an enhancement of the tip–sample contact area. In contrast, here we observe a sharp decrease of the current which can only be ascribed to the mechanical switching from LRS to HRS. Subsequent sweeps performed at the same point dem-onstrated that, contrary to what is observed in the so-called piezotronic effect,[29] the mechanical writing of HRS state is retained. This finding demonstrates that the force can induce a transition from a LRS to a HRS state, and that mechanical stress represents a new degree of freedom to control the resistive state in nanostructured and semicon-ducting BTO films.

The experiments described above show that nanostruc-tured columnar BTO films can be switched between two polarization states (up, down) and two resistance states (HRS, LRS) by applying a DC bias or a mechanical load. Figure 6 summarizes the configurations that the system can

adopt upon application of those externally controlled stimuli which drive the switching interconversion between them.

One might expect that the combinations of two bistable states would allow setting the system into four different

small 2017, 1701614


Figure 6. Upon application of external electrical biases to the LSMO layer and mechanical loads as external stimuli, the resistive and polarization states of columnar BTO nanostructures can be set into three different configurations summarized in this schema. In particular, the (DOWN, HRS) configuration, can be only set by means of a mechanical load, either from a (UP, HRS) configuration by the flexoelectric switching of the polarization, or from a (DOWN, LRS) configuration by switching the resistive state from LRS to HRS in a process of strain-driven migration of oxygen vacancies.

Figure 5. a) Schematic of the conductive AFM system used for nanoscale electrical measurement. b) The resistive switching behavior of epitaxial columnar BTO/LSMO/STO/Si(001) films. The curve started with −6 VDC, where a low resistance state (LRS) was recorded. Then, the voltage was sweeped, at a rate of 12 V s−1 to +6 VDC where a high resistance state (HRS) was recorded, and back to −6 VDC to recover the LRS state. c) Mapping of electrical current of the 45 nm thick BaTiO3 film showing the piezo-driven resistance switching between a low resistance state (LRS) and a high resistance state (HRS) domains by electrical poling (right) and mechanical loading force (left). d) Current versus force (nN) including time axis spectroscopy curves of the BTO/LSMO/STO/Si(001) films. Notice that the curve shows that the mechanically assisted switching from an ON to an OFF resistive state takes about 50 ms in a curve recorded with force rate of 5 µN s−1. The curves were performed on a −7 VDC poled area, with −2 VDC reading voltage.


configurations, i.e., (P-up, LRS), (P-down, LRS), (P-up, HRS), and (P-down, HRS). In practice, however, an upward polari-zation (P-up) and a LRS cannot be achieved simultaneously. Indeed, such (P-up, LRS) configuration should be estab-lished either by switching the polarization of a (P-down, LRS) configuration or by resistive switching from a (P-up, HRS) configuration. In order to switch the polarization from a (P-down, LRS) configuration, a positive bias would simul-taneously reverse polarization and resistive state, while the application of a force would merely switch the resistive state to HRS with no effect on the polarization, because the flexo-electric switching can only reverse an upward polarization state. For the second possibility, switching the resistive state of a (P-up, HRS) configuration, both the voltage and force-assisted switching change the polarization and resistance states at the same time, turning the system configuration into (P-down, LRS).

Before concluding, we briefly discuss the mechanisms that may be involved in the electric and mechanical switching of resistive states in our semiconducting BTO films. Regarding the voltage controlled resistive switching, it can be under-stood as a modulation of the Schottky barrier height at the metal–semiconductor interfaces due to the accumulation or depletion of carriers induced by the ferroelectric polari-zation, being the associated bending of the BTO band that determines the conductivity of the interface.[30]

Regarding the mechanically assisted RS effect, one has to consider the effect of a strain on the oxygen vacancies and their mobility. A doubly ionized oxygen vacancy intro-duces two extra electrons into nonbonding orbitals of the perovskite structure, resulting in an expansion of the unit cell, the so-called chemical-expansivity mechanism.[31] Accord-ingly, a compressive strain is known to force the migration of oxygen vacancies against the strain gradient.[32] That is, the tip decreases the local density of carriers by a strain-mediated removal of oxygen vacancies, which sets a HRS. A similar effect has been reported for NiO thin films, in which a strain generated by an AFM tip produced a metal (NiO1−x) to insulator (NiO) transition.[33] In our samples, the retention of HRS stems from the low diffusion coefficient of oxygen vacancies at room temperature in BTO,[34] which undergo an extremely low relaxation after removing the mechanical stress. Besides, the highly defective microstructure of our BTO films favors the pinning of oxygen vacancies. It is worth pointing out that in n-type transition-metal semiconducting oxides with a high concentration of oxygen vacancies, the resistive switching can be controlled by its migration under an applied voltage.[35] However, such a scenario is at odds with our experimental results. Indeed, in a resistive switching controlled by the migration of oxygen vacancies, a down polarization state would set a HRS, in contrast to our obser-vations. This means that the oxygen vacancy density of our heterostructures is much lower than that of highly n-doped BTO of ref. [8c] and supports our interpretation of a resis-tive switching mechanism based in the modulation of the Schottky barrier height by accumulation/depletion of carriers induced by the direction of ferroelectric polarization. It is worth pointing out that the mechanical switching from LRS to HRS, which we explain by to the migration of vacancies

driven away from the compressive strain induced by the AFM tip toward the BTO/LSMO, also involves an increase of electron density, driven by donor oxygen vacancies, at the BTO/LSMO. This accumulation of electrons at the inter-face also occurs in the case of the electrical switching from LRS to HRS by applying a +VDC at the LSMO electrode (induced by the ferroelectric up polarization). Therefore, the observed increase in the resistivity is due to the increase of the Schottky barrier height at the BTO/LSMO interface, in both mechanisms. From this discussion, one would expect the HRS obtained mechanically and electrically to be different. As a matter of fact, in Figure 5c the averaged intensities in the high resistivity areas recorded mechanically and electri-cally differ by a factor of two.

4. Conclusion

In summary, a mechanically assisted switching of bistable polarization and the conductivity states in columnar BTO/LSMO heterostructures on Si is reported. The mechanical switching mechanisms are independent of the ordinary voltage-controlled reversal of the four polarization and the conductivity states also present in the material, and thus these four states can combine into three different configurations of the system. Our study shows the enormous potential of com-bining physical and chemical deposition methods to obtain complex epitaxial nanostructured oxides with new func-tionalities. In particular, it provides an innovative pathway toward the design and integration on silicon of nonvolatile multilevel devices with mechanoelectric control, which could find applications in data storage and sensing.

5. Experimental Section

Synthesis of Nanostructured STO Thin Films (MBE): Nano-structured epitaxial SrTiO3 thin films were deposited by MBE on 500 µm thick (001) oriented silicon wafers, as reported previ-ously.[22] Knudsen effusion cells were used for Sr and Ti evapora-tion. Before introduction in the reaction chamber, the Si substrates were cleaned by exposition to UV-O3 ambient and in a buffered oxide etching solution to remove the native SiO2 surface layer. A controlled and clean SiO2 thin layer was then formed at the bare Si surface by exposing again the surface to UV-O3, and was then removed from the Si surface in the growth chamber by using the Sr-catalyzed desorption procedure consisting of annealing under ultrahigh vacuum (UHV) at ≈770 °C, leading to a self-limited cov-erage of 1/3 monolayers (ML) of Sr on the Si surface. A completion to ½ ML of Sr at 500 °C was then performed. After this treatment, the substrate temperature was ramped down to 360 °C, and the samples were exposed to an O2 partial pressure of 5 × 10−8 Torr for ≈1 min and 10 ML of Sr-rich SrTiO3 were then deposited (codepo-sition of Sr and Ti) at this temperature and O2 partial pressure, leading to the formation of partially amorphous SrTiO3 layers.

[22] The samples were then crystallized by annealing at 460 °C during ≈5 min under UHV, and subsequent stoichiometric SrTiO3 layer was epitaxially grown at 420 °C under an oxygen partial pressure up to 5 × 10−7 Torr at a growth rate of ≈1.5 ML min−1. Sr-rich SrTiO3 was

small 2017, 1701614


full papers


deposited at the early stage of the growth in order to crystallize the SrTiO3 layer and to form nano-/macrodomains about 25 nm sepa-rated by antiphase boundaries with tunable size depending on the Sr ratio.[22]

Synthesis of LSMO Nanostructured Thin Films: Thin films of LSMO were grown by spin coating deposition of aqueous solu-tions. The precursor solutions of individual cations (La3+, Sr2+, and Mn2+) were obtained by dissolving, respectively, nitrates in deion-ized water with ethylenediaminetetra acetic acid (EDTA; 1:1 molar ratio to La and Sr and 1:2 to Mn) and polyethylenimine (1:1 mass ratio to EDTA for Sr and Mn, and 2.5:1 for La). The pH of these solu-tions was 8.45 for La, 4.96 for Sr, and 4.68 for Mn. These solutions were filtered with membranes of 10 000 nominal molecular weight limit for ensuring that all cations in solution were coordinated to the polymer. After this purification process, the retained portions were analyzed by inductively coupled plasma (ICP) yielding a final concentration of [La3+] = 96.105 × 10−3 m, [Sr2+] = 51.016 × 10−3 m, and [Mn2+] = 175.064 × 10−3 m. The solution with the final stoichi-ometry (La:Sr:Mn 0.7:0.3:1) was obtained by mixing the purified precursors, obtaining a final concentration of 171 × 10−3 m, which is low enough to keep the porous structure of under layer STO, but high enough to guarantee a homogeneous coverage, 30 nm thick, over the whole wafer.

Solutions were spin coated at 4000 rpm during 15 s on STO/Si (001) substrates, obtaining aqueous polymeric layers that finally were annealed at 800 °C for 2 h in air. Heating and cooling ramps were set at 3 °C min−1. LSMO thicker films were prepared by repeating deposition and annealing steps.

Synthesis of Nanostructured BTO Thin Films: Epitaxial columnar BaTiO3 thin films were grown by MBE on LSMO/STO/Si (001) with Knudsen effusion cells, at T = 450 °C, P(O2) = 1 10−7 Torr, and at a growth rate of ≈1.5 ML min−1.

Characterization: Film structure was investigated using a field emission gun scanning electron microscope (FEG-SEM), Model FEI Quanta 200 environmental scanning electron microscopy (ESEM) FEG. Cross-sectional STEM analysis of BTO/LSMO/STO/Si het-erostructures was performed using a probe aberration corrected microscope, a JEOL ARM 200cF STEM with a cold field emission source operated at 200 kV and equipped with a Gatan Quantum energy filter spectrometer for EELS measurements, at the Univer-sidad Complutense de Madrid, Spain. STO layers grown on Si were studied using an aberration corrected NION UltraSTEM 100 oper-ated at 100 kV, at the Oak Ridge National Laboratory, USA. All STEM images shown in this work were acquired in high angular annular dark field imaging mode, in which the contrast results from the high angle scattering strength. The intensity in the images is approximately proportional to Z2, giving rise to so called Z-con-trast imaging.[36]

Specimens for TEM observation were prepared by conventional methods, by grinding, dimpling, and Ar ion milling. X-ray diffraction measurements were carried out using a PANalytical Empyrean dif-fractometer, with an Euler cradle and a wavelength of 1.540598 Å, equipped with a PIXcel 3D detector. Tapping AFM images where obtained using a Keysight 5100 scanning probe microscopy (SPM) unit in constant amplitude feedback mode, using AppNano FORT probes with a nominal spring constant of 2.8 N m−1. PFM measure-ments were performed on an Agilent 5500LS instrument using a solid platinum conducting tip with a spring constant of 18 N m−1 and tip length 80 µm tip, reference RMN-25PT300 with the aim of

preventing artifacts not related to piezoelectricity.[37] The tip was biased with 2 VAC, while the DC bias was applied to the sample. In order to optimize signal-to-noise ratio, contact resonance was used for all PFM measurements. Force calibration was performed using a force versus distance curve for each tip used to obtain the deflec-tion sensitivity and hence multiplying it by cantilever force constant in order to obtain the applied load. The indentation experiments were performed using diamond-coated Si tips with a tetrahedral shape and tip diameter of 5–10 nm. The stress upon unloading of the indentation test was calculated from the force over the cross-sectional area of the tip, obtained from the penetration depth and the maximum indentation areas (Figure 4a). The Z piezo displace-ment accounts for the material strain under the applied load but it is also affected by the bending of the tip cantilever. However, it is noted that while indenting the very soft LDPE, the tip bending is negligible and no force can be exerted on the material, resulting in a flat curve. On the other hand, in the case of the hard lithium niobate, the Z piezo displacement is dominated by tip bending with a negligible contribution of the material strain, which gives an esti-mate of the cantilever bending at a given force and allows sepa-rating this contribution from the material strain in the BTO curves.

Current AFM images were obtained with an additional acces-sory, “Resiscope” from CSI instruments. In such equipment, bias is applied to the LSMO layer by contacting it laterally using silver paint, while the tip is grounded. In order to avoid short circuit due to the fact that silver paint was also in contact with BTO film, the measurements were performed at the film side opposite to the contact. The same RMN-25PT300 tip was used to obtain the cur-rent maps and spectroscopy experiments. In order to perform tem-perature dependent experiments, a homemade sample heater and cooler was used. The cooler is a 4-stage Peltier stack with a liquid cooling system that can be used up to −60 °C, while the heater is a resistive based heater with a PT100 sensor to read temperature. For all PFM and CAFM measurements, low humidity, less than 5%, was maintained inside AFM box.[38] The tip was brought into feedback position in the contact mode with a constant force of 200 nN to read images which were maintained between the tip and the sample sur-face. An AC voltage VAC = 2 V was applied between the tip and the bottom electrode within the specific contact resonance frequency of the cantilever. Additional ferroelectric experiments were performed by increasing the temperature above BTO Curie temperature transi-tion, i.e., up to 438 K (see Figure S9a–c in the Supporting Informa-tion) in order to confirm the absence of ferroelectricity.[39]

Supporting Information

Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements

A.C.-G. acknowledges the financial support from the French Agence Nationale pour la Recherche (ANR), projet Q-NOSS ANR ANR-16-CE09-0006-01 and the Ecole Centrale de Lyon under the BQR 2016 project. The authors thank P. Regreny, C. Botella, J. B. Goure

small 2017, 1701614


(9 of 10) 1701614© 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.small-journal.comsmall 2017, 1701614


for technical assistance on the Nanolyon technological platform. ICMAB acknowledges MINECO in co-funding with European Social funds through the MAT2014-56063-C2-1-R, the Severo Ochoa Pro-gram for Centers of Excellence in R&D (SEV-2015-0496) and the Ramón y Cajal program (J.G., RyC-2012-11709) and the General-itat de Catalunya (2014SGR213). The STEM microscopy work was conducted in the ICTS-CNME at the UCM as well as at the Center for Nanophase Materials Sciences, which is a DOE Office of Sci-ence User Facility. Authors acknowledge the ICTS-CNME for offering access to their instruments and expertise. This research was sup-ported by the European Research Council (ERC StG-2DTHERMS), the Ministerio de Economía y Competitividad of Spain (Project No. MAT2016-80762-R), the Consellería de Cultura, Educación e Ordenación Universitaria (ED431F 2016/008, and the Centro singular de investigación de Galicia accreditation 2016–2019, ED431G/09), and the European Regional Development Fund (ERDF)), and Fundación BBVA.

Conflict of Interest

The authors declare no conflict of interest.

[1] a) M. Dawber, K. M. Rabe, J. F. Scott, Rev. Mod. Phys. 2005, 77, 1083; b) R. Ramesh, D. G. Schlom, MRS Bull. 2008, 33, 1006.

[2] a) A. Ohtomo, H. Y. Hwang, Nature 2004, 427, 423; b) D. G. Schlom, L.-Q. Chen, X. Pan, A. Schmehl, M. A. Zurbuchen, J. Am. Ceram. Soc. 2008, 91, 2429.

[3] J. L. MacManus-Driscoll, A. Suwardi, H. Wang, MRS Bull. 2015, 40, 933.

[4] L. Liang, X. Kang, Y. Sang, H. Liu, Adv. Sci. 2016, 3, 150069.[5] a) T. Tybell, C. H. Ahn, J.-M. Triscone, Appl. Phys. Lett. 1999,

75, 856; b) J. Junquera, P. Ghosez, Nature 2003, 422, 506; c) D. D. Fong, G. B. Stephenson, S. K. Streiffer, J. A. Eastman, O. Auciello, P. H. Fuoss, C. Thompson, Science 2004, 304, 1650.

[6] a) P. Maksymovych, S. Jesse, P. Yu, R. Ramesh, A. P. Baddorf, S. V. Kalinin, Science 2009, 324, 1421; b) V. Garcia, S. Fusil, K. Bouzehouane, S. Enouz-Vedrenne, N. D. Mathur, A. Barthelemy, M. Bibes, Nature 2009, 460, 81; c) A. Gruverman, D. Wu, H. Lu, Y. Wang, H. W. Jang, C. M. Folkman, M. Y. Zhuravlev, D. Felker, M. Rzchowski, C. B. Eom, E. Y. Tsymbal, Nano Lett. 2009, 9, 3539.

[7] A. Gruverman, A. Kholkin, Rep. Prog. Phys. 2006, 69, 2443.[8] a) X. Chen, C. H. Jia, Y. H. Chen, G. Yang, W. F. Zhang, J. Phys.

D: Appl. Phys. 2014, 47, 365102; b) M. P. Singh, L. Méchin, W. Prellier, M. Maglione, Appl. Phys. Lett. 2006, 89, 202906; c) M. Li, J. Zhou, X. Jing, M. Zeng, S. Wu, J. Gao, Z. Zhang, X. Gao, X. Lu, J. M. Liu, M. Alexe, Adv. Electron. Mater. 2015, 1, 150069.

[9] H. Lu, C.-W. Bark, D. Esque de los Ojos, J. Alcala, C. B. Eom, G. Catalan, A. Gruverman, Science 2012, 336, 59.

[10] H. Lu, D. J. Kim, C. W. Bark, S. Ryu, C. B. Eom, E. Y. Tsymbal, A. Gruverman, Nano Lett. 2012, 12, 6289.

[11] a) S. Choudhury, Y. L. Li, L. Q. Chen, Q. X. Jia, Appl. Phys. Lett. 2008, 92, 142907; b) L. E. Cross, J. Mater. Sci. 2006, 41, 53.

[12] J. Narvaez, F. Vasquez-Sancho, G. Catalan, Nature 2016, 538, 219.[13] W. Zhu, J. Y. Fu, N. Li, L. Cross, Appl. Phys. Lett. 2006, 89, 192904.[14] R. A. McKee, F. J. Walker, M. F. Chisholm, Phys. Rev. Lett. 1998, 81,

3014.[15] a) V. Vaithyanathan, J. Lettieri, W. Tian, A. Sharan, A. Vasudevarao,

Y. L. Li, A. Kochhar, H. Ma, J. Levy, P. Zschack, J. C. Woicik, L. Q. Chen, V. Gopalan, D. G. Schlom, J. Appl. Phys. 2006, 100, 024108, b) L. Largeau, G. Patriarche, G. Saint-Girons, G. Delhaye, G. Hollinger, Appl. Phys. Lett. 2008, 92, 031904; c) G. Niu, S. Yin,

G. Saint-Girons, B. Gautier, P. Lecoeur, V. Pillard, G. Hollinger, B. Vilquin, Microelectron. Eng. 2011, 88, 1232.

[16] J. M. Vila-Fungueiriño, R. Bachelet, G. Saint-Girons, M. Gendry, M. Gich, J. Gazquez, E. Ferain, F. Rivadulla, J. Rodriguez-Carvajal, N. Mestres, A. Carretero-Genevrier, Frontiers in Physics 2015, 3, 38.

[17] S. R. Burns, J. M. Gregg, V. Nagarajan, Adv. Funct. Mater. 2016, 26, 8367.

[18] a) A. Carretero-Genevrier, G. L. Drisko, D. Grosso, C. Boissiere, C. Sanchez, Nanoscale 2014, 6, 14025; b) Y. Xia, G. M. Whitesides, Annu. Rev. Mater. Sci. 1998, 28, 153; c) C. J. Brinker, Y. F. Lu, A. Sellinger, H. Y. Fan, Adv. Mater. 1999, 11, 579; d) A. Carretero-Genevrier, M. Gich, L. Picas, J. Gazquez, G. L. Drisko, C. Boissiere, D. Grosso, J. Rodriguez-Carvajal, C. Sanchez, Science 2013, 340, 827.

[19] a) F. F. Lange, Science 1996, 273, 903; b) R. W. Schwartz, T. Schneller, R. Waser, C. R. Chim. 2004, 7, 433; c) M. L. Calzada, I. Bretos, R. Jimenez, H. Guillon, L. Pardo, Adv. Mater. 2004, 16, 1620.

[20] A. Carretero-Genevrier, J. Oro-Sole, J. Gazquez, C. Magen, L. Miranda, T. Puig, X. Obradors, E. Ferain, C. Sanchez, J. Rodriguez-Carvajal, N. Mestres, Chem. Mater. 2014, 26, 1019.

[21] a) Q. X. Jia, T. M. McCleskey, A. K. Burrell, Y. Lin, G. E. Collis, H. Wang, A. D. Q. Li, S. R. Foltyn, Nat. Mater. 2004, 3, 529; b) I. Lucas, J. M. Vila-Fungueiriño, P. Jiménez-Cavero, B. Rivas-Murias, C. Magén, L. Morellón, F. Rivadulla, ACS Appl. Mater. Interfaces 2014, 6, 21279; c) J. M. Vila-Fungueiriño, B. Rivas-Murias, B. Rodríguez-González, O. Txoperena, D. Ciudad, L. E. Hueso, M. Lazzari, F. Rivadulla, ACS Appl. Mater. Interfaces 2015, 7, 5410.

[22] G. Saint-Girons, R. Bachelet, R. Moalla, B. Meunier, L. Louahadj, B. Canut, A. Carretero-Genevrier, J. Gazquez, P. Regreny, C. Botella, J. Penuelas, M. G. Silly, F. Sirotti, G. Grenet, Chem. Mater. 2016, 28, 5347.

[23] D. A. Bonnell, S. V. Kalinin, A. L. Kholkin, A. Gruverman, MRS Bull. 2009, 34, 648.

[24] C. Dubourdieu, J. Bruley, T. M. Arruda, A. Posadas, J. Jordan-Sweet, M. M. Frank, E. Cartier, D. J. Frank, S. V. Kalinin, A. A. Demkov, V. Narayanan, Nat. Nanotechnol. 2013, 8, 748.

[25] Z. Wen, X. Qiu, C. Li, C. Zheng, X. Ge, A. Li, D. Wu, Appl. Phys. Lett. 2014, 104, 042907.

[26] L. Chen, Z. Cheng, W. Xu, X. Meng, G. Yuan, J. Liu, Z. Liu, Sci. Rep. 2016, 6, 19092.

[27] A. Abdollahi, D. Millán, C. Peco, M. Arroyo, I. Arias, Phys. Rev. B 2015, 91, 104103.

[28] W. Ma, L. E. Cross, Appl. Phys. Lett. 2006, 88, 232902.[29] Z. L. Wang, Nano Today 2010, 5, 540.[30] P. W. M. Blom, R. M. Wolf, J. F. M. Cillessen, M. P. C. M. Krijn, Phys.

Rev. Lett. 1994, 73, 2107.[31] S. B. Adler, J. Am. Ceram. Soc. 2001, 84, 2117.[32] J. R. Petrie, C. Mitra, H. Jeen, W. S. Choi, T. L. Meyer,

F. A. Reboredo, J. W. Freeland, G. Eres, H. N. Lee, Adv. Funct. Mater. 2016, 26, 1563.

[33] Y. Kim, S. J. Kelly, A. Morozovska, E. K. Rahani, E. Strelcov, E. Eliseev, S. Jesse, M. D. Biegalski, N. Balke, N. Benedek, D. Strukov, J. Aarts, I. Hwang, S. Oh, J. S. Choi, T. Choi, B. H. Park, V. B. Shenoy, P. Maksymovych, S. V. Kalinin, Nano Lett. 2013, 13, 4068.

[34] M. Kessel, R. A. De Souza, M. Martin, Phys. Chem. Chem. Phys. 2015, 17, 12587.

[35] M. Bowen, P. Brinks, W. S. Choi, M. Dawber, R. Dittmann, V. Esposito, D. D. Fong, I. Garbayo, H. Guo, M. Huijben, A. Janssen, A. P. Kajdos, G. Koster, Y. Krockenberger, H. N. Lee, C. Leon, S. Linderoth, R. V. K. Mangalam, L. W. Martin, J. A. Moyer, M. Naito, N. Pryds, G. Rijnders, J. Santamaria, S. S. A. Seo, S. Stemmer, J. E. ten Elshof, A. Vailionis, H. Yamamoto, J. Zhang, H. Zhou, in Epitaxial Growth of Complex Metal Oxides

full papers

1701614 (10 of 10) www.small-journal.com © 2017 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim small 2017, 1701614


(Eds: G. Koster, M. Huijben, G. Rijnders), Woodhead Publishing, Campbridge, UK 2015, p. xi.

[36] S. J. Pennycook, D. E. Jesson, Ultramicroscopy 1991, 37, 14.[37] S. V. Kalinin, D. A. Bonnell, Phys. Rev. B 2002, 65, 125408.[38] N. Balke, P. Maksymovych, S. Jesse, A. Herklotz, A. Tselev, C.-B. Eom,

I. I. Kravchenko, P. Yu, S. V. Kalinin, ACS Nano 2015, 9, 6484.[39] a) M. J. Polking, M.-G. Han, A. Yourdkhani, V. Petkov,

C. F. Kisielowski, V. V. Volkov, Y. Zhu, G. Caruntu, A. Paul Alivisatos,

R. Ramesh, Nat. Mater. 2012, 11, 700; b) B. Maria Teresa, B. Vincenzo, V. Massimo, P. Jan, S. Maxim, M. Liliana, T. Andrea, N. Paolo, H. Catalin, Z. Zhe, N. Mats, Nanotechnology 2004, 15, 1113.

Received: May 16, 2017Revised: June 30, 2017Published online:

Electric and Mechanical Switching of Ferroelectric and ...€¦ · Indeed, the flexoelectric...

Documents

Transcript of Electric and Mechanical Switching of Ferroelectric and ...€¦ · Indeed, the flexoelectric...