Strain transfer in porous multiferroic composites of CoFe₂O₄ and PbZr_xTi_1−xO₃

Multiferroics are materials that simultaneously exhibit more than one ferroic order parameter, such as ferromagnetism or ferroelectricity; they are of interest because of their potential applications in a wide range of nanoscale devices.^1–3 In particular, magnetoelectric multiferroic materials couple a magnetic and an electric polarization, but single-phase materials that show this property at room temperature are rare.^1,2,4–7 Composite materials, however, offer another route to magnetoelectric behavior. Such materials generally use strain-coupling and consist of layers of piezoelectric and magnetostrictive materials. As an electric field is applied to the composite, the piezoelectric is strained, and this strain is transferred to the magnetostrictive material, which in turn affects the magnetization. This coupling allows the magnetization to be controlled by applied electric fields and vice versa.^8–16 Because this technique requires intimate coupling between the two phases, numerous synthetic methods and architectures of strain-coupled multiferroics have been investigated in the literature, including sequentially deposited two-dimensional stacks, spontaneously phase-separated nanopillar arrays, and other three-dimensional arrangements.^10–27 

We have recently shown that porosity is an important control parameter in the synthesis of multiferroic composites.¹⁶ In our previous work, a mesoporous, magnetostrictive cobalt ferrite (CoFe₂O₄ or CFO) film was filled using atomic layer deposition (ALD) with piezoelectric lead zirconate titanate (PbZr_xTi_1−xO₃ or PZT). The result was an interconnected, three-dimensional network containing both CFO and PZT. The final porosity in such a composite can be controlled by the initial pore size, which is determined by the polymer template used in the synthesis of the initial mesoporous CFO, and the thickness of the deposited PZT layer. Our results showed that the final composite porosity was correlated with a change in magnetic saturation that could be achieved upon electrical poling. It was hypothesized that this correlation was due to a link between porosity and mechanical flexibility of the composite, but no direct evidence for that hypothesis was obtained. Here, we examine the mechanism of magnetoelectricity in these thin film composites by depositing a range of thicknesses of PZT in the mesoporous CFO framework and measuring the resultant voltage-dependent strain in the CFO framework.

The mesoporous CFO framework was synthesized using block copolymer–templating of sol–gel films, a technique that has been used to produce a wide range of metal oxide materials of varying nanoarchitectures.^28–37 The CFO sol was templated with an amphiphilic diblock copolymer that forms micelles in solution. As the solution is deposited, the micelles self-assemble into periodic structures within the film. Upon pyrolysis, the polymer is removed, leaving a stable porous network of CFO.

This porous film is then conformally coated with PZT using ALD, which grants uniformity over the entire structure because this technique obtains layer-by-layer growth through a self-limiting surface reaction. Alternating pulses of gaseous precursors completely saturate all available surface sites, allowing conformal deposition over the entire porous network. ALD also allows for fine control over the thickness deposited and, thus, over the final porosity of the composite material. This method, thus, provides additional functionality compared to composites in the literature, which, thus far, have been dense structures that lack porosity. Here, we aim to use high-resolution x-ray diffraction on films as a function of an ex situ poling field to explore the mechanisms of strain coupling in this porous composite.

Synthetic details for both CFO and PZT have been discussed previously.^16,28,38 Briefly, poly(ethylene-co-propylene)-block-poly-(ethylene oxide) with a mass ratio of PEP(3900)-b-PEO(4000) was used to template a sol based on nitrate salts of Co and Fe. Films were dip-coated onto silicon wafers in a humidity-controlled chamber set to 10%–20% relative humidity. The withdrawal rate was usually 2 mm/s but can be varied, depending on desired thickness. To form rigid inorganic/organic structures, the films were calcined in air at 80 °C for 6 h, at 130 °C for 8 h, and at 180 °C for 6 h for a total heating time of 24 h including temperature ramps. Once calcined, films were annealed at 550 °C with a 10 °C/min ramp for 5 min.

PZT was deposited via ALD using Pb(TMHD)₂, Ti(O-i-Pr)₂(TMHD)₂, and Zr(TMHD)₄ as precursors. PZT was deposited at no more than 180 °C in an amorphous form and then crystalized into tetragonal PZT by rapid thermal annealing at 700 °C under oxygen for one minute. Here, the PZT layer thicknesses range from 3 to 10 nm.

The morphology and thickness of the nanocomposites were confirmed using a JEOL JSM-6700F field-emission scanning electron microscopy (FE-SEM). Ellipsometric porosimetry (EP) was performed on a Semilab PS-1100 in the spectral range of 1.24–4.5 eV. A UV-vis CCD detector adapted to a grating spectrograph analyzes the signals reflected by the sample from a 75 W Hamamatsu Xe lamp. Toluene was used as the adsorbent, and the EP analysis was performed using the associated SEA software. Angular-dependent x-ray diffraction (XRD) was collected at the Stanford Synchrotron Radiation Laboratory (SSRL) using beamline 7–2 at wavelengths λ = 0.9919 and 1.0332 Å. Magnetic measurements were carried out on a Quantum Design MPMS 5T SQUID magnetometer with RSO detection.

Because our previous work indicated the importance of residual porosity in multiferroic composites, we first characterized the porosity of the composites using ellipsometric porosimetry. EP adsorption/desorption curves for samples with various PZT layers [Fig. 1(a)] show that as thicker PZT layers are deposited, less porosity is observed. The samples with 0 and 3 nm of PZT show a distinct type IV isotherm, which signifies an interconnected porous network. The calculated porosity values are 26.0% for the 0 nm PZT sample, 15.3% for 3 nm, 6.6% for 6 nm, and 0.03% for 10 nm. The PZT in these as deposited films is amorphous, but previous work¹⁶ has shown that the PZT can be crystallized to the ferroelectric tetragonal phase. We find that redistribution of the PZT in the pores can block the small necks in the structure, impeding toluene access to the pores. As a result, SEM was used to characterize the samples after crystallization. SEM images of the samples with crystallized PZT layers [Fig. 1(b)], from top to bottom, show reduced porosity as thicker PZT layers are deposited. The unfilled CFO framework exhibits ordered porosity, which is distorted by grain growth upon annealing of the PZT layer. For this reason, the 10 nm sample still appears to be somewhat porous by SEM, even though access into the porous interior is stopped by pore necks that had been completely stoppered by PZT, as seen from the EP adsorption/desorption curves.

To determine the magnetoelectric coupling of these thin film composites, they were electrically poled ex situ with the electric field applied perpendicular to the sample surface (henceforth referred to as out-of-plane). The films were covered with a 13 μm polyvinylidene chloride spacer and physically sandwiched between two Al electrodes 1.28 cm in diameter. The nanocomposite was electrically poled for 10 min with applied electric fields ranging from 0 to 1.42 MV m⁻¹. As such, the strains and polarizations explored in this paper are remanent ones. While it is true that much of the strain and polarization will be lost upon removal of the applied field, the remanent polarization stabilizes within milliseconds and can be assumed to be constant throughout the measurement.^39,40

Magnetization measurements show a decrease in out-of-plane saturation magnetization upon electrical poling, which is correlated with porosity of the composite (Fig. 2). The sample with the thinnest PZT layer shows the largest change in saturation magnetization, and the sample with the thickest PZT shows hardly any change. Because polarization in ultrathin PZT is known to decrease with thickness,^41,42 this trend is likely due to the mechanical properties of the porous composite, rather than any favorable change in the PZT itself. The films with the thinnest PZT are also the ones with the highest porosity and, therefore, the greatest mechanical flexibility, as pore flexion accommodates significant strain changes in the material.^28,29,43

The role of porosity in magnetoelectric coupling is corroborated by strain analysis of the CFO layer. Synchrotron high-resolution XRD was used to probe the differences in both out-of-plane and in-plane (parallel to the substrate) lattice spacings. The CFO{311} and PZT{200} peaks were relatively well resolved and were treated as representative of overall strain changes in both materials. Because these films consist of polycrystalline CFO and PZT with no preferred orientation with respect to the substrate, any lattice plane can be used to report on the overall strain state of the material. As shown in Fig. 3 and expected based on the magnetization data, the CFO{311} out-of-plane lattice spacing increased upon ex situ electrical poling, and the magnitude of the change was directly correlated with the porosity of the composite. As the porosity decreased, the strain transferred upon electrical poling also decreased (Fig. 3). CFO exhibits negative magnetostriction, and so out-of-plane tension directly corresponds to the reduced magnitude of the change in out-of-plane magnetization saturation.

Even though CFO is not a piezoelectric, it is strain-coupled to one, and so we can calculate the strains when 1 MV m⁻¹ has been applied and then removed from the sample. While this strain is not a real piezoelectric coefficient, it relates a remanent strain to an ex situ electric field, and so we give it the symbol $d33′$⁠. Values of $d33′$ are ranging from $d33′$ = 590 × 10⁻¹² mV⁻¹ for the composite with the highest porosity (3 nm PZT) to $d33′$ = 130 × 10⁻¹² mV⁻¹ for the composite with the lowest porosity (10 nm PZT), which is comparable to true piezoelectric coefficients of PZT.^44–46 These values demonstrate more than a fourfold reduction in strain transferred when porosity is removed from the sample. Again, we emphasize that these calculated values are not true piezoelectric coefficients because they relate the remanent strain to an ex situ applied field instead of the instantaneous strain to an in situ field; the instantaneous piezoelectric coefficient should be higher indeed.

No significant change upon electrical poling was found in the in-plane saturation magnetization nor in the CFO in-plane strain. The CFO framework is covalently bound to the substrate and is unable to move in plane because of substrate clamping. Because its strain is unchanged, the CFO in-plane magnetization is also unchanged. However, the PZT layer is deposited onto the CFO framework itself, and as such is not constrained by the substrate. As the PZT deforms due to the out-of-plane electric field, strain can be expressed as out-of-plane tension or in-plane compression. This strain is transferred to the clamped CFO framework and can be expressed only as the aforementioned out-of-plane tension. Interestingly, analysis of PZT strain reveals contribution from both in-plane compression and out-of-plane tension. Similar $d′$ coefficients calculated for PZT show comparable strains to the CFO, but with more noise, because the PZT layer is mere nanometers thin and, thus, diffracted intensity is weaker. The greatest PZT strains are in the most porous sample (3 nm PZT) and are shown in Fig. 4. The data show changes in both in-plane and out-of-plane lattice constants and demonstrate that the PZT is not at all substrate clamped. The strains are calculated to be $d31′$ = −670 × 10⁻¹² m V⁻¹ in-plane and $d33′$ = 130 × 10⁻¹² m V⁻¹ out-of-plane. These values are comparable to that of the CFO, suggesting that much of the strain had indeed been transferred. Thus, from strain analysis of this free PZT layer, we see that three-dimensional porosity has an advantage over traditional two-dimensional structures where multiple layers are clamped together and to the substrate. In a three-dimensional structure like this one, the pre-filling material can remain unclamped if sufficient residual porosity is retained.

Overall, these experiments have allowed us to explore the mechanism of strain-coupling in porous magnetoelectric CFO/PZT composites. These thin films are composed of a templated mesoporous CFO framework, which is subsequently filled by ALD PZT of varying thicknesses. As the samples are electrically poled out-of-plane, x-ray diffraction shows that the piezoelectric PZT layer may exhibit both out-of-plane tension and in-plane compression. This strain is transferred to the magnetostrictive CFO layer, which results in decreased out-of-plane saturation magnetization as measured by SQUID magnetometry. The strain transfer is greatest in samples with the greatest porosity, as pore flexion accommodates greater strains in the material. This porous architecture, thus, not only offers greater mechanical flexibility than traditional composite architectures, but also mitigates the effects of substrate clamping for the ALD layer. Perhaps more importantly, the observation of in-plane compression in what could have been a clamped PZT layer provides insight into the use of porosity in the design of future porous multiferroic composites.

This work was supported by the NSF Nanosystems Engineering Research Center for Translational Applications of Nanoscale Multiferroic Systems under Cooperative Agreement Award No. EEC-1160504. Additionally, author S.K.P. acknowledges support from a National Science Foundation Graduate Research Fellowship under Grant Nos. DGE-1650604 and DGE-2034835. This work made use of the UCLA Molecular Instrumentation Center. This manuscript contains data collected at the Stanford Synchrotron Radiation Lightsource, experimental station 7-2. Use of the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-76SF00515.