Magnetic and multiferroic properties of dilute Fe-doped BaTiO3 crystals

Magnetic and multiferroic properties of dilute Fe-doped BaTiO3 crystals M. Staruch, H. ElBidweihy, M. G. Cain, P. Thompson, C.A. Lucas, and P. Finkel U.S. Naval Research Laboratory, Washington, DC, U.S.A. U.S. Naval Academy, Annapolis, MD, U.S.A. Electrosciences Ltd., Farnham, Surrey, U.K. XMaS Beamline, European Synchrotron Radiation Facility, Grenoble, F-38043, France Department of Physics, University of Liverpool, Oliver Lodge Laboratory, Liverpool. L69 7ZE. U.K.


Introduction
BaTiO3 is well known as one of prototypical examples of materials demonstrating ferroelectricity that has a high dielectric constant, low loss tangent and high piezoelectric coefficient (d33 ~ 420 pC/N) at room temperature. Interest in doped BaTiO3 (BTO) single crystals has increased in the past decade after seminal results showing that (001) cut and poled crystals doped at the Ti 4+ site with either Fe 3+ or Mn 3+ were shown to have large and recoverable electrostrain of up to 0.8%, significantly higher than the approximately 0.05% which can be obtained reversibly from undoped BTO crystals. 1,2 The improvement in strain is thought to be due to the alignment of defects (i.e. O 2vacancies) with the crystallographic symmetry in the ferroelectric state when the material is aged through application of thermal treatments. This results in the ferroelectric domains favoring alignment with the defect dipoles which provides a restoring force, where recovery of the original ferroelectric domain pattern after non-180 o switching results allows for the high strains generated by this reversible domain motion. [3][4][5] There is also the possibility that the incorporation of a magnetic ion could give rise to magnetization in these samples and possibly be a new route to creating novel single phase multiferroic materials. Multiferroics demonstrate simultaneous ferroelectric and ferromagnetic properties with coupling between the magnetic and polar order parameters, paving the way for novel memory, sensor, and spintronics devices. 6,7 The possibility of coupling the large strain in these dilutely doped crystals to any magnetic properties remains as yet unexplored. Previous studies of iron doped BTO were focused on very large concentrations, above which the range where the defect dipole behavior is typically observed (x < 2%). 8 field of up to 3T was applied in the direction of polarization and counter to that direction and the resulting PE loop was measured to discern any differences in ferroelectric behavior as a function of applied external magnetic field.

Results and Discussion
After aging, the ferroelectric and piezoelectric properties of the sample were measured. A pinched hysteresis loop and recoverable large strain were observed, consistent with previous reports. However, it is important to note that after several electric field cycles, the pinching disappeared and the strain significantly decreased. Overall, with repeated application of the electric field, the ferroelectric properties moved towards those of an undoped BTO crystal. This suggests that the electric field allows for migration of the oxygen vacancies in the sample and therefore the defect dipole alignment to the ferroelectric dipole moments is not maintained. It was also found that the original behavior shown in Figure 1 can always be recovered through the completion of another aging process, so the material may be re-set. Further investigation into this This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002863
phenomenon is currently ongoing, but it is important to note when discussing the magnetic and multiferroic properties that the sample is expected to be very sensitive to the prehistory.
The current samples were confirmed to be fully tetragonal (P4mm) by room temperature powder x-ray diffraction as shown in Supplemental Information Figure S1 This is as expected based on previous reports and is consistent with the charge deficiency that would allow for the formation of oxygen vacancies and defect dipoles. While the pre-edge feature is not as sharp as the references, it is strongly dependent on site symmetry and may actually indicate a non-centrosymmetric position which could suggest a local polar moment extends to the Fe 3+ sites as well. 22 Zero field cooled (ZFC) and field cooled (FC) magnetization versus temperature data are plotted in Figure 3 Fe2O3 at high temperatures can lead to the mixture of these phases and has also been shown to slightly decrease the Verwey transition temperature as observed here. 24 The presence of these phases is also most likely what causes the split of the ZFC and FC curves for the 0.5 T data as well. Although iron impurities in nominally undoped samples have previously been reported, 1 it should be noted that the present undoped crystal does not show any defect dipole behavior, which would also be consistent with the iron forming as a small amount of secondary phase.
Beyond this, though, the diamagnetic response of BTO itself is the predominant contribution to the signal.
In contrast to the curves for undoped BTO, the BTFO sample does not show any magnetic anomalies related to iron oxide impurity phases as shown in Figure 3 7 matrix in this dilutely doped sample. This is consistent with a previous report of BaTiO3 doped with Co, Fe, and Cr (between 1.5% and 3.5%) showing paramagnetism at all temperatures measured that was also thought to originate from the transition metal ions. 25 No anomalies are present in the derivative of the M vs. T data that would otherwise suggest magnetic ordering within the range of temperatures measured.
Magnetization versus magnetic field data were also recorded at multiple temperatures and selected curves are plotted in Figure 4(a). Here it is instructive to note that even though there is no evidence of ferromagnetism in the M vs. T data that all of the curves have an S-shape with some small opening of the hysteresis loop. This is the case for both the BTO and the BTFO sample, and could be consistent with other reports of surface oxygen or oxygen vacancies and other defects in the lattice. 26,27 Yet once the paramagnetic or diamagnetic background is subtracted from these curves, we can see that there is no temperature dependence evident in the data for undoped BTO whereas there is a temperature evolution of the saturation magnetization (MS) that indicates a weak ferromagnetic ordering in the BTFO sample as shown in Figure 4(b).
Extrapolating the change in MS with temperature to reach the temperature-independent value of undoped BTO, this would suggest a Curie temperature (TC) of 550 -600 K assuming a power law dependence [ ∝ ( − ) ] with the critical exponent β~0.5. 28,29 This could explain why no evidence of this transition is observed in the magnetization versus temperature data, as we did not go up above TC during the measurements which would show the best evidence of this weak FM moment.
The values of MS are extremely low (~0.01 B/Fe), suggesting any long-range order is very weak. In that respect the magnetism here is very different from the very high moments observed in the dilute magnetic semiconductors, 30,31

so any FM interactions in the present sample
This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002863
are most likely not due to the formation of impurity bands or magnetic polarons as have been proposed for those systems as well as a 0.5% Mn doped BTO that also showed orders of magnitude higher magnetization that was explained through bound polarons 15  Lastly, as there is some evidence of a weak FM ordering, it is important to evaluate the multiferroic coupling. Capacitance versus temperature measurements with and without applied magnetic fields were made directly after the sample was re-aged. It should be noted that the capacitance magnitude includes stray capacitance and is not the absolute value of the samples. In addition, ferroelectric hysteresis loops were measured with an applied magnetic field, and both these results are shown in Figure 5. In the magnetocapacitance measurements, even at fields of up to 3 T there is no difference in capacitance with temperature. However, there is a statistically significant change in the ferroelectric response with applied magnetic field, notably in the material's electric coercive field EC which seems to contradict the capacitance measurements. It's important to note that there is no pinching in the ferroelectric hysteresis loop though, suggesting This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
9 that this sample has been de-aged leaving no net defect dipole alignment and therefore no longer any enhanced piezoelectricity. However, for the magnetocapacitance measurement, the sample was aged immediately prior to measurement and hence left in a stable polarization state. The two measurements differ in another important way: the ferroelectric measurement is a large field measurement whilst the capacitance measurement is that for a minor loop (1 V applied over 1 mm), and thus magnetocapacitance measurements without a bias electric field near EC is not likely to capture any evidence of multiferroicity. Near EC, the magnetoelectric (ME) coupling coefficient α = dP/dH (the derivative of polarization P with respect to applied magnetic field H) is within the range of other single phase multiferroic materials with a maximum value of ME = 2.3 x 10 -9 s/m. However, while non-zero, ME is considerably smaller at zero electric field. We must also consider the possibility that the presence of defect dipoles is incompatible with multiferroic behavior in this sample. To the best of our knowledge, there are no reports showing the coexistence of these phenomena and the only previous reports on multiferroicity do not measure or they explicitly state that there is zero magnetoelectric coupling. 15,16 Further study of these crystals with defect dipole alignment are necessary to evaluate if we would be able to tune the extraordinarily high 0.8% strains with a remote magnetic field.

Conclusion
In summary, undoped BaTiO3 and 0.5% Fe doped BaTiO3 single crystals were aged and the ferroelectric, magnetic, and multiferroic properties measured. After aging, pinched electric hysteresis loops were observed as well as large strains up to 0.8%. The introduction of Fe 3+ as confirmed by x-ray absorption spectroscopy had a significant impact on the magnetic properties, introducing paramagnetism from lone iron spins as well as what appears to be a weak ferromagnetism. Although evidence of multiferroicity and magnetoelectric coupling was This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.

PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002863
observed in the PE loops with applied magnetic field, there is not yet direct evidence that the ME nature can coexist with the defect dipole-aligned aged state of the crystal.

Supplementary Material
See supplementary material for the powder x-ray diffraction and Rietveld refinement of a crushed iron doped BaTiO3 sample.     This is the author's peer reviewed, accepted manuscript. However, the online version of record will be different from this version once it has been copyedited and typeset.
PLEASE CITE THIS ARTICLE AS DOI:10.1063/5.0002863