Antiferromagnetic long-range spin ordering in Fe- and $\mathrm{NiF}{\mathrm{e}}_2$-doped $\mathrm{BaTi}{\mathrm{O}}_3$ multiferroic layers

We report on the Fe doping and on the comparative Ni-Fe codoping with composition close to $\mathrm{NiF}{\mathrm{e}}_2$ of fully oxidized $\mathrm{BaTi}{\mathrm{O}}_3$ layers (∼20 nm) elaborated by atomic oxygen plasma assisted molecular beam epitaxy; specifically any role of oxygen vacancies can be excluded in our films. Additionally to the classical in situ laboratory tools, the films were thoroughly characterized by synchrotron radiation x-ray diffraction and x-ray absorption spectroscopy. For purely Fe-doped layers, the native tetragonal perovskite structure evolves rapidly toward cubiclike up to 5% doping level above which the crystalline order disappears. On the contrary, low codoping levels $(\sim 5\%\mathrm{NiF}{\mathrm{e}}_2)$ fairly improve the thin film crystalline structure and surface smoothness; high levels (∼27%) lead to more crystallographically disordered films, although the tetragonal structure is preserved. Synchrotron radiation magnetic dichroic measurements reveal that metal clustering does not occur, that the Fe valence evolves from ${\mathrm{Fe}}^{2+}$ for low Fe doping levels to ${\mathrm{Fe}}^{3+}$ for high doping levels, and that the introduction of Ni favors the occurrence of the ${\mathrm{Fe}}^{2+}$ valence in the films. For the lower codoping levels it seems that ${\mathrm{Fe}}^{2+}$ substitutes ${\mathrm{Ba}}^{2+}$, whereas ${\mathrm{Ni}}^{2+}$ always substitutes ${\mathrm{Ti}}^{4+}$. Ferromagnetic long-range ordering can be excluded with great sensitivity in all samples as deduced from our x-ray magnetic absorption circular dichroic measurements. On the contrary, our linear dichroic x-ray absorption results support antiferromagnetic long-range ordering while piezoforce microscopy gives evidence of a robust ferroelectric long-range ordering showing that our films are excellent candidates for magnetic exchange coupled multiferroic applications.

Figure 1

RHEED patterns, showing the $\langle 110\rangle$ reciprocal space patterns, observed after 360 min of deposition on $\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$ of (a) a $\mathrm{BaTi}{\mathrm{O}}_3\left(001\right)$ layer doped with 4.7% $\mathrm{NiF}{\mathrm{e}}_2$, (b) a $\mathrm{BaTi}{\mathrm{O}}_3\left(001\right)$ layer doped with 27% $\mathrm{NiF}{\mathrm{e}}_2$, (c) a $\mathrm{BaTi}{\mathrm{O}}_3\left(001\right)$ layer doped with 2% Fe, and (d) a pure $\mathrm{BaTi}{\mathrm{O}}_3\left(001\right)$ layer.

Figure 2

Structural parameters of doped $\mathrm{BaTi}{\mathrm{O}}_3\text{/}\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$ layers. Reciprocal space scans for 4.7% and 27% $\mathrm{NiF}{\mathrm{e}}_2$ samples: (a) out-of-surface plane direction around the 001 Bragg peak and (b) low index in-surface-plane scan around the 220 Bragg peak. The reciprocal space is indexed using the $\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$ substrate reciprocal lattice units. Vertical lines are guides to the eye to evidence the lattice parameter changes. X-ray photon energy was 8 keV.

Figure 3

In- and out-of-surface plane lattice parameters observed for doped $\mathrm{BaTi}{\mathrm{O}}_3\left(001\right)$ layers. The abscissa axis to the right (left) reports the value of $\mathrm{NiF}{\mathrm{e}}_2$ percentage (Fe percentage) in doped films. For the sake of clarity the Fe doping were tenfold scaled on the abscissa, real doping levels being 1%, 2%, and 4%. A pure $\mathrm{BaTi}{\mathrm{O}}_3$ epitaxial layer is also included as reference. The bulk lattice parameters (right arrows) at room temperature for the $\mathrm{SrTi}{\mathrm{O}}_3$ substrate, tetragonal $\mathrm{BaTi}{\mathrm{O}}_3$, and half of the parameter of $\mathrm{NiF}{\mathrm{e}}_2{\mathrm{O}}_4$ are indicated.

Figure 4

XAS of Fe and Ni in doped $\mathrm{BaTi}{\mathrm{O}}_3$ layers deposited on $\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$. For the XAS measurements the samples were vertical, the beam polarization horizontal, and the glancing x-ray angle with respect to the surface was 30° (i.e., grazing incidence). $\mathrm{Fe}{L}_{2,3}$ spectra for samples with, from top to bottom, 5% Fe, 2% Fe, 1% Fe, 0.5% Fe, 27% $\mathrm{NiF}{\mathrm{e}}_2$, and 4.7% $\mathrm{NiF}{\mathrm{e}}_2$.

Figure 5

XMCD (circular right minus left light polarization) for Fe-doped $\mathrm{BaTi}{\mathrm{O}}_3$ films grown by AO-MBE. The spectra were normalized and vertically shifted for ease of shape comparison. The measurements were realized with the samples held at 4 K and under a magnetic field (oriented 30° from the surface) of 5 T. Magnetic saturation was not achieved. The peaks labeled A, B, and C indicate the position of, respectively, $\mathrm{F}{\mathrm{e}}^{2+}\left(O_h\right)$ (octahedral site), $\mathrm{F}{\mathrm{e}}^{3+}\left(T_d\right)$ (tetrahedral site), and $\mathrm{F}{\mathrm{e}}^{3+}\left(O_h\right)$ in the spinel structures. $\mathrm{XMCD}\mathrm{Fe}{L}_{2,3}$ spectra for samples with, from top to bottom, 10% Fe, 5% Fe, 4% Fe, 1% Fe, and 0.5% Fe doping of $\mathrm{BaTi}{\mathrm{O}}_3$. The curves were normalized for ease of comparison. The actual XMCD signals, with respect to the $\mathrm{Fe}\text{−}{L}_3$ line intensity, amounts 10%, 9%, 5%, 5%, and 3%, respectively.

Figure 6

XAS Ni in doped $\mathrm{BaTi}{\mathrm{O}}_3$ layers deposited on $\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$. For the XAS measurements the samples were vertical, the beam polarization horizontal, and the incidence angle (with respect to the surface normal) was 60°. $\mathrm{Ni}{L}_{2,3}$ spectra for samples with, from top to bottom, 27% $\mathrm{NiF}{\mathrm{e}}_2$, 4.7% $\mathrm{NiF}{\mathrm{e}}_2$, linear dichroic signal (linear horizontal minus vertical light polarization) signal for 4.7% $\mathrm{NiF}{\mathrm{e}}_2$, circular dichroic signal (circular right minus left light polarization). The XAS curves were normalized to their maximum values and shifted with respect to each other for the sake of clarity. The dichroic curves were scaled accordingly.

Figure 7

(a) Examples of magnetization loops realized using the $L_3$ maximum XMCD signal while looping the magnetic field. The Fe edge $L_3$ evolution is shown for purely Fe-doped samples and the $\mathrm{Ni}{L}_3$ edge for the highly doped $\mathrm{NiF}{\mathrm{e}}_2$ sample. The samples were held at 4 K and the glancing x-ray angle with respect to the surface was 30° (i.e., grazing incidence). (b) Remanence XMCD signal measured at 4 K. (Top) 5% Fe and (bottom) 4.7% $\mathrm{NiF}{\mathrm{e}}_2\mathrm{BaTi}{\mathrm{O}}_3$-doped layers. The $L_{2,3}$ XMCD spectra were recorded under a residual magnetic field of 0.001 T after applying a saturating magnetic field of 5 T.

Figure 8

$\mathrm{Fe}{L}_{2,3}$ x-ray linear dichroic (linear horizontal minus vertical light polarization) signal. The light impinges the sample at a glancing incidence angle of 30° with respect to the surface of the sample which was kept at 4 K.

Figure 9

PFM images. (a) Schematic drawing used by Nanoman to write to the sample. Black, white, and gray areas were polarized with a potential on the tip of −6 V, +6 V, and 0 V. (b) Ferroelectric hysteresis loop (PFM phase vs tip bias) for a 4% Fe-doped $\mathrm{BaTi}{\mathrm{O}}_3\text{/}\mathrm{Nb}(1\%)\text{−}\mathrm{SrTi}{\mathrm{O}}_3\left(001\right)$ sample. (c) Piezoresponse after writing PFM phase for a drive amplitude of 4000 mV to read the sample on a 0.5% Fe-doped $\mathrm{BaTi}{\mathrm{O}}_3$. (d) Piezoresponse after writing PFM phase measured for a drive amplitude of 4000 mV to read the sample on a 4% Fe-doped $\mathrm{BaTi}{\mathrm{O}}_3$.