betaFeedback

WorksPhysics and Astronomy

Tuning charge-transport properties and magnetic order in metallic EuTiO3−δ_{3-δ}

Abstract

The stoichiometric antiferromagnetic insulator EuTiO3_3 is proximate to a ferroelectric phase. Whereas cation substitution has been used as a tuning parameter to introduce charge carriers and manipulate the magnetism, the effects of oxygen-vacancy doping have been less explored. Here we report a detailed study of the charge transport and magnetic properties of metallic, oxygen-vacancy-doped EuTiO3−δ_{3-δ}. Using CaH2_2 as an oxygen getter to achieve a higher carrier concentration than previously reported, we find that the phase diagram of the oxygen-vacancy-doped system is distinct from that obtained via cation doping. In particular, we uncover a change from antiferromagnetic to ferromagnetic order in the metallic state, with a maximum Curie temperature of TC ≈\approx 11 K at the highest carrier concentration of n ≈\approx 1021^{21} cm−3^{-3}. These findings are supported by density functional theory calculations, which indicate a significant change in the nearest-neighbor magnetic exchange constant with increasing electron doping. We also present x-ray diffuse scattering and complementary first-principles results that reveal that, similar to the prominent incipient ferroelectric perovskite SrTiO3_3, the data for EuTiO3_3 are consistent with thermal diffuse scattering and with the absence of quasi-elastic contributions. Finally, we report specific heat measurements that confirm the magnetic transition temperatures deduced from magnetization measurements and corroborate the lattice dynamics picture inferred from the diffuse scattering data.

Introduction

Ferroelectric (FE) materials that approach a paraelectric (PE) phase at zero temperature have garnered interest due to their proximity to a quantum phase transition [1, 2]. Two prominent examples are the perovskite oxides SrTiO3\mathrm{SrTiO}_{3} (STO) and KTaO3\mathrm{KTaO}_{3} (KTO), which exhibit a zone-center soft mode whose frequency remains nonzero at low temperatures, accompanied by a saturation of the permittivity, a phenomenon known as incipient ferroelectricity [3, 4, 5, 6, 7]. Notably, the emergence of bulk superconductivity in STO at low carrier concentrations [8, 9] and of interfacial superconductivity in KTO, introduced via gating and at heterostructure interfaces [10, 11], has been proposed to be closely related to FE soft-mode and Rashba spin-orbit coupling effects [8, 12, 13, 14, 15, 16]. In the case of STO, alloying with Ca or Ba [17, 18], 18O^{18}\mathrm{O} isotope substitution [19, 20], as well as strain [21, 22] have been shown to introduce ferroelectricity and to enhance the superconducting transition temperature. Similarly, spin fluctuations and their connection to superconductivity have been extensively studied in magnetic materials [23, 24]. An interesting possibility that involves both FE and magnetic fluctuations is multiferroic quantum criticality, which could potentially provide a novel platform for the exploration of the interplay of both properties and of functional behavior under applied fields [5].

Stoichiometric, insulating EuTiO3\mathrm{EuTiO}_{3} (ETO) features a band gap in the 0.2-0.9 eV range [25, 26, 27, 28]. Optical measurements typically report a higher gap of ~0.8-0.9 eV [25, 28], whereas electrical measurements and density functional theory (DFT) give smaller values of ~0.2-0.3 eV [26, 27]. Like STO and KTO [6, 33], ETO is an incipient FE with an incomplete zone-center transverse optic (TO) mode softening upon cooling [29, 30, 31]. Yet ETO also shows antiferromagnetic (AFM) order associated with Eu2+\mathrm{Eu}^{2+} moments, with a Néel temperature of TN∼5.5T_N \sim5.5 K [32, 33]. The proximity to both FE and magnetic transitions at zero temperature makes ETO a promising multiferroic quantum criticality candidate, with alloying and strain as possible tuning parameters [5]. In fact, it was recently reported that tuning through a quantum critical point was achieved via alloying, for the solid solution Eu0.3Ba0.1Sr0.6TiO3\mathrm{Eu}_{0.3}\mathrm{Ba}_{0.1}\mathrm{Sr}_{0.6}\mathrm{TiO}_{3} [34]. Moreover, due to spin-phonon coupling, ETO exhibits magnetoelectric effects: the dielectric constant decreases at the Néel temperature and is enhanced in an applied magnetic field [35]. Furthermore, under biaxial strain, the ground state of ETO can transform from AFM-PE to ferromagnetic-ferroelectric (FM-FE), as shown both experimentally and via first-principles calculations [36, 37, 38], with a colossal magnetoelectric response at the AFM-PE / FM-FE boundary [39].

Metallic, electron-doped ETO exhibits itinerant electrons in Ti-3d3d conduction bands, which affects the magnetic properties, as Eu-4f4f orbitals hybridize with Ti-3d3d orbitals [40]. An increase in the carrier concentration can shift the ground-state spin configuration from AFM to FM, and thus turn the system into a FM metal, as indicated theoretically [41] and demonstrated experimentally, with electron doping achieved via substitution on both the perovskite (ABO3)(\mathrm{ABO}_{3}) A- (La(\mathrm{La} [42, 43, 44], Gd [45]) and B-site (Nb [46], Al [47]), and via hydrogen substitution (EuTiO3−xHx\mathrm{EuTiO}_{3-x}\mathrm{H}_{x}) [48]). However, whereas FM order has been observed in oxygen-deficient thin films [51, 52], this has not yet been explored in bulk crystals [53]. Another intriguing aspect of doped ETO is the evidence of incommensurate superstructure correlations due to antiferrodistortive and competing instabilities, which were proposed to be highly sensitive to defects such as Eu3+^{3+} and oxygen vacancies [54, 55], although the exact nature of these correlations has remained unclear. A comprehensive investigation of the transport, magnetic, and structural properties of bulk oxygen-vacancy doped (OVD) ETO is therefore highly desirable.

Here we investigate the charge transport and thermodynamic properties of OVD EuTiO3−δ_{3-\delta} single crystals with increasing oxygen deficiency. In addition, we present synchrotron x-ray diffuse scattering data for stoichiometric, insulating ETO, and we complement our experimental work with first-principles (DFT + UU) calculations. The diffuse scattering data confirm the high structural quality of our samples and, together with the DFT + UU results, indicate that the structural response originates from phonon contributions, with a lattice dynamics that is similar to STO, as both perovskites exhibit a cubic-tetragonal phase transition associated with a zone-boundary soft-mode [40, 56]. We find that oxygen-vacancy doping with calcium hydride (CaH2_2) as a getter material can result in electron-carrier concentrations that exceed previously reported values [53] and that the magnetic ground state evolves from AFM to FM. This change in the metallic state is supported by our DFT + UU calculation, which indicates a significant modification of the Eu-Eu nearest-neighbor exchange constant with increasing doping. The theoretical analysis furthermore indicates that, although the three-fold degeneracy of the conduction band at the Γ\Gamma-point is lifted by spin-orbit coupling (SOC), nearly all doped samples studied in the present work fall into the regime in which all three bands are active. Finally, specific heat measurements corroborate the lattice dynamics picture inferred from diffuse scattering and confirm the magnetic transition temperatures observed via magnetometry. The present work lays the foundation for further studies. For example, analogous to Ba/Ca and Nb-doped STO [17, 18], Ba/Ca-substituted EuTiO3−δ_{3-\delta} might host a polar metallic state. In this state, spin-phonon coupling might allow the manipulation of polar properties with an applied magnetic field [35], while itinerant electron-spin coupling could enable magnetic-field-tuning of charge-transport properties [43, 45, 46]. Moreover, our diffuse x-ray scattering data provide a benchmark for future studies of antiferrodistortive and competing instabilities and their relationship with Eu3+^{3+} defects and oxygen vacancies.

This paper is organized as follows: In Section II, we briefly describe the techniques used for sample preparation and measurement, as well as DFT + UU calculations, and we provide additional details in the Supplemental Material [57]. Section III describes our results in five sub-sections: structural characterization and lattice dynamics, charge transport and insulator-metal transition, magnetic phase diagram, specific heat, and carrier concentration effects on spin interaction and ground state. We then discuss and summarize our results in Sections III and IV.

Methods

Single crystals of ETO were grown with the floating-zone method. The crystals were cut into small samples, and their orientation was determined via Laue x-ray diffraction. Other techniques, including in-house powder and single-crystal x-ray diffraction as well as synchrotron x-ray diffuse scattering were used to characterize the samples, as described below. We used CaH2\mathrm{CaH}_{2} as a getter material to create oxygen vacancies in ETO. Nominally stoichiometric, as-grown ETO was sealed with CaH2\mathrm{CaH}_{2} powder (without direct contact) in quartz tubes under vacuum (pressure below 100 mTorr), with a 1:12 mass ratio of ETO to CaH2\mathrm{CaH}_{2}. The sealed quartz tubes were heated for 12 h at temperatures between 700°C and 1050°C, depending on the desired carrier concentration, with higher temperatures resulting in higher oxygen vacancy densities. Resistivity and Hall measurements were conducted with the standard four-point van der Pauw method using a Quantum Design, Inc. Physical Property Measurement System (PPMS) in AC drive mode. A wire bonder was used to make contacts with aluminum wires. For metallic samples, the contact resistance was about 1 Ω\Omega. DC magnetic-susceptibility measurements were performed with a Quantum Design, Inc. Magnetic Property Measurement System (MPMS). A 20 Oe DC field was applied along either <100><100> or <110><110> directions. Synchrotron x-ray diffuse scattering measurements were performed at beamline QM2 ID4B of the Cornell High Energy Synchrotron Source (CHESS). DFT + UU calculations were performed at the Agate cluster of the Minnesota Supercomputing Institute. Electronic and magnetic properties were calculated using VASP [58, 59] in the cubic phase of ETO. The phonon properties were calculated with the finite displacement method, on the tetragonal phase with a 2×2×22 \times2 \times2 supercell, using Quantum Espresso [60, 61] and the Phonopy [62] package. Specific heat measurements were made in a PPMS using relaxation calorimetry. Apiezon N grease was used to affix crystals to PPMS platforms for measurements between 1.85 K and 300 K, and 2% temperature pulses were used. At each temperature, three measurements were performed in order to achieve improved accuracy. Other experimental and computational details are described in the Supplemental Material [57].

Results

Structural characterization and lattice dynamics

The perovskite ETO undergoes a phase transition from simple cubic (Pm3‾mPm\overline{3}m structure) to tetragonal (I4/mcmI4/mcm structure) somewhat below room temperature [31, 55, 56, 63, 64] (Figure 1(a)). The conventional tetragonal cell parameters are related to the cubic lattice constant a0a_0, with a∼2a0a \sim\sqrt{2}a_0, b∼2a0b \sim\sqrt{2}a_0, and c∼2a0c \sim2a_0. This phase transition involves octahedral rotation, as in the case of STO [65, 66], and it is associated with the condensation of a zone-boundary RR-point mode [31]. Furthermore, undoped ETO undergoes a PM-AFM transition with a Néel temperature of TN∼5.5T_N \sim5.5 K. The observed GG-type AFM order is illustrated in Figure 1(b), with spins aligned along <100>cubic<100>_{\mathrm{cubic}} (equivalently, <110>tetragonal<110>_{\mathrm{tetragonal}}), as determined from neutron powder diffraction and resonant x-ray scattering [32]. In this GG-type AFM state, nearest-neighbor and second-nearest-neighbor Eu moments are antiparallel and parallel to each other, respectively. The three nearest exchange interactions J1J_1, J2J_2, and J3J_3 are indicated in Figure 1(b).

Crystal structure and x-ray characterization of as-grown EuTiO3

Figure 1. Crystal structure and x-ray characterization of as-grown EuTiO3_3. (a) Schematic representation of the Pm3ˉmPm\bar{3}m cubic and I4/mcmI4/mcm tetragonal unit cells. Eu: purple; Ti: blue; O: red. (b) Illustration of GG-type AFM order in a 2×2×22 \times2 \times2 supercell, with a single cubic cell indicated on the bottom left. The nearest, second-nearest, and third-nearest interactions J1J_{1}, J2J_{2}, and J3J_{3} are indicated by black lines. Ti and O atoms are omitted for clarity. Neighboring spins are antiparallel along ⟨100⟩\langle100\rangle. (c) Laue x-ray diffraction pattern obtained with the photon beam along [1], showing the diffraction pattern in the (HK0)(HK0) plane. (d) Room-temperature powder x-ray diffraction result (obtained with Cu K-α\alpha photons) compared with calculated Bragg-peak scattering angles and intensities. Inset: two as-grown EuTiO3_3 crystals, with surfaces polished perpendicular to [100] crystalline direction.

Our as-grown single crystals were characterized with several x-ray techniques. The Laue backscattering x-ray pattern of an ETO crystal is shown in Figure 1(c). The resultant diffraction pattern exhibits sharp Bragg peaks. Additionally, Cu K-α\alpha powder x-ray diffraction (XRD) was performed on ground single-crystal samples, and the pattern shown in Figure 1(d) matches well with the calculated 2θ2\theta positions. The high crystalline quality was further confirmed via single-crystal XRD characterization (see Figure S1 [57]) and synchrotron x-ray measurements (see below).

Single-crystal x-ray diffraction plots: (a) rocking curve around the (200) Bragg peak and (b) theta–2theta scan showing (200), (300), and (400) peaks

Figure S1. Single-crystal x-ray diffraction result for as-grown EuTiO3\mathrm{EuTiO}_{3}. (a) Rocking curve (ω\omega-scan) across the (200) Bragg peak. (b) θ/2θ\theta/2\theta scan along [100].

In order to characterize the atomic dynamics and probe structural correlations, we obtained synchrotron x-ray diffuse scattering data for an as-grown ETO crystal (additional data for an OVD ETO crystal are shown in Figure S2 [57]). While x-ray diffuse scattering was measured previously [54], this prior work did not cover a sufficiently large momentum space volume to enable a detailed characterization of the lattice dynamics. Figure 2(a) compares the diffuse scattering pattern in the (HK0HK0) plane, obtained at 175 K and 300 K, with a DFT +U+ U calculation of the thermal diffuse scattering (TDS) due to phonons. Since the cubic phase of ETO exhibits an unstable phonon mode [67], it is necessary to calculate phonons in the tetragonal phase for TDS simulations. The resultant phonon dispersions and phonon density of states (PDOS) are shown in Figure 2(b). We note that, in Figure 2(b), we use high-symmetry kk-point labels for the body-centered tetragonal cell (c/a>1c/a > 1), whereas in the remainder of the manuscript the wavevectors are defined with respect to the five-atom perovskite cell. Phonon eigenvalues and eigenvectors were subsequently used to calculate the dynamic structure factor, S(Q,ω)S(\mathbf{Q},\omega), where Q\mathbf{Q} and ℏω\hbar\omega are the momentum and energy transfers, respectively, as detailed in [57]. The TDS simulations were then generated by integrating the calculated S(Q,ω)S(\mathbf{Q},\omega) over the ℏω=0\hbar\omega= 0–30 meV energy range. In the (HK0HK0) plane, streaks between Bragg peaks along ⟨110⟩cubic\langle110\rangle_{\mathrm{cubic}} are clearly seen in both the diffuse scattering data and the results of the theoretical model. The observed diffuse scattering pattern is similar to that for STO [68, 69], owing to the same perovskite structure and structural instabilities associated with octahedral rotations, indicative of closely similar atomic dynamics. By investigating the phonon eigenvector components along high-symmetry directions, as shown in Figure 2(b) (see also Figure S8 [57]), it can be deduced that the diffuse scattering along ⟨110⟩cubic\langle110\rangle_{\mathrm{cubic}} (Γ\Gamma–MM) originates from nearly dispersionless modes around 11 meV that are dominated by Eu motion. These low-energy modes correspond to the first peak in the atom-projected phonon density of states (DOS) in Figure 2(b), in agreement with a Eu-specific PDOS measurement [64], and consistent with our DOS calculation and thermal diffuse scattering result. As shown below, our specific heat data provide additional thermodynamic validation of this low-energy phonon contribution.

Comparison of synchrotron x-ray diffuse scattering data and first-principles modeling for undoped ETO

Figure 2. Comparison of synchrotron x-ray diffuse scattering data and first-principles modeling for undoped ETO. (a) X-ray data for the (HK0)(HK0) plane at 300 K and 175 K, above and below the cubic-tetragonal structural transition temperature Ts∼285T_s \sim285 K, compared with calculated TDS. High-symmetry points (in cubic (P3ˉm3)(\mathrm{P}\bar{3}\mathrm{m}\mathrm{3}) notation): Γ\Gamma (0 0 0)(0\ 0\ 0), X (0.5 0 0)(0.5\ 0\ 0), M (0.5 0.5 0)(0.5\ 0.5\ 0), and (not shown) R (0.5 0.5 0.5)(0.5\ 0.5\ 0.5). (b) Calculated phonon band structure and density of states, using body-centered tetragonal (I4/mcm, c/a>1)(I4/mcm,\ c/a > 1) notation (we use cubic notation throughout the rest of the paper), with high-symmetry points Γ\Gamma (0 0 0)(0\ 0\ 0), X (0 0 0.5)(0\ 0\ 0.5), P (0.25 0.25 0.25)(0.25\ 0.25\ 0.25), N (0 0.5 0)(0\ 0.5\ 0), M (0.5 0.5 −0.5)(0.5\ 0.5\ -0.5), and S (0.374 0.626 −0.374)(0.374\ 0.626\ -0.374). The phonon dispersion and eigenvectors were used to calculate the dynamical structure factor S(Q,ω)S(\mathbf{Q},\omega) and the TDS. The different colors indicate the predominant character of the dynamical matrix eigenvectors and not the force constants matrix eigenvectors. Since dynamical matrix eigenvectors include a factor of the square root of masses, the weights of heavier atoms are overestimated. As a result, the acoustic branch near the Γ\Gamma point, which should have equal weight contributions from all atoms, appears Eu-heavy. (c,d) X-ray data for (HK2.5)(HK2.5) and (HK0.5)(HK0.5) planes compared with DFT + UU. The left panels show the data at 300 K and 175 K, whereas the right panels show the calculated TDS with temperature occupation factor at 300 K. (e) One-dimensional cuts along [H −1.5 0.5][H\ -1.5\ 0.5]. At 300 K, the R-points show diffuse scattering associated with the soft-mode due to out-of-phase octahedral rotation. At 175 K, R-point superlattice peaks appear as the system has transitioned into the tetragonal phase. (f) Illustration of phonon eigenvectors and structural distortion associated with the R-point mode.

Synchrotron x-ray scattering intensity maps in the (a) HK0, (b) HK0.5, and (c) HK2.5 planes

Figure S2. Synchrotron x-ray scattering data for the OVD ETO sample with nH=2.2×1021 cm−3n_H = 2.2 \times10^{21}\ \mathrm{cm}^{-3} in the (a) HK0HK0, (b) HK0.5HK0.5, and (c) HK2.5HK2.5 planes.

Band structures of undoped FM ETO without SOC

Figure S8. Band structure of undoped FM ETO without SOC from DFT + UU. (a) U=5.7U = 5.7 eV. (b) U=6.2U = 6.2 eV. The energy zero is shifted to the Fermi energy EfE_f.

Figure 2(c,Figure 2d) compares the experimental diffuse scattering patterns for the half-integer planes (HK2.5HK2.5) and (HK0.5HK0.5) at 175 K and 300 K with our DFT +U+ U result. We note that the latter does not capture the experimental background intensity modulation at low Q\mathbf{Q} that results from strong air scattering. These data again show similarities with STO [68, 69] and reveal several characteristic features. First, the diffuse peaks at the half-integer positions (h/2 k/2 l/2h/2\ k/2\ l/2) (with integer hh, kk, and ll) are directly related to the cubic-tetragonal transition associated with the soft acoustic phonon at RR-points. At 300 K, this low-energy phonon has not yet condensed, leading to thermal diffuse scattering. In contrast, at 175 K, below TsT_{\mathrm{s}}, the phonon has condensed, and we observe superlattice Bragg peaks at the RR points (see also Figure 2(e)). The eigenvector of the RR-point phonon corresponds to out-of-phase rotation of the octahedra in the aa-bb plane, as depicted in Figure 2(f). Neighboring octahedra rotate in opposite directions about the cc-axis, in a GG-type pattern. This distortion drives the transition from the cubic Pm3ˉmPm\bar{3}m to tetragonal I4/mcmI4/mcm structure. Second, we observe weak cross-shaped diffuse scattering near the RR-points along ⟨100⟩cubic\langle100\rangle_{\mathrm{cubic}} (Figure 2(c)), e.g., around Q=(−1.5±1.5 2.5)\mathbf{Q}=(-1.5 \pm1.5\ 2.5). The cross-shaped features can be better seen in Figure 3(a-Figure 3b), and we attribute them to phonons with wavevectors between the MM and RR points. These phonons also correspond to octahedral rotations, but with a different spatial periodicity than simple in-phase or out-of-phase rotations (i.e., with long-wavelength modulations along the axis of rotation). These features are also present in the calculated oxygen-only diffuse scattering map in Figure 3(c). Third, streaks through (h k l/2)(h\ k\ l/2) positions (with integer hh, kk, and ll) and the curvilinear features of the diffuse pattern in half-integer planes (Figure 2(c) and (d)), analogous to STO [68], are well reproduced by theoretical modeling and attributed to Eu and Ti distortions (Figure 3(d) and Figure S9 [57]). Fourth, we did not observe incommensurate peaks in the as-grown ETO sample (at 175 K and 300 K), likely as a result of differences in defect types or densities compared to prior work [54, 55]. Lastly, we note that we also measured a metallic OVD ETO sample and observed closely similar diffuse scattering patterns (Figure S2 [57]).

Zoom of synchrotron x-ray diffuse scattering data and element-decomposed thermal diffuse scattering

Figure 3. Zoom of synchrotron x-ray diffuse scattering data (same data as in Figure 2) and element-decomposed thermal diffuse scattering in the (HK2.5)(HK2.5) plane. The data exhibit cross-shape patterns around R-points at both (a) 300 K and (b) 175 K. (c) The DFT + UU result of oxygen-only thermal diffuse scattering shows cross-shape patterns at half-integer HH and KK. (d) The DFT + UU result of (Eu + Ti)-decomposed thermal diffuse scattering shows curvilinear features, reflecting valleys and ridges in reciprocal space associated with low-energy phonon modes. The overall agreement between experiment and theory demonstrates that the observed diffuse scattering arises from thermal diffuse scattering rather than static disorder.

Spin-projected band structure of undoped FM ETO with SOC from DFT + $U$

Figure S9. Spin-projected band structure of undoped FM ETO with SOC from DFT + UU. (a) U=5.7U = 5.7 eV. (b) U=6.2U = 6.2 eV. The energy zero is shifted to the Fermi energy EfE_f.

Charge transport and insulator-metal transition

We conducted resistivity and Hall-effect measurements of both as-grown and OVD ETO samples with varying doping levels. The respective results are shown in Figure 4(a) and (Figure 4b). The resistivity of the as-grown sample is ρ≈100 Ωcm\rho\approx100\ \Omega\mathrm{cm} at 300 K and becomes immeasurably large at low temperatures. With increasing electron density, oxygen-deficient ETO is seen to evolve to a metallic state. At the highest doping level (Hall carrier concentration nH≈2.2×1021 cm−3n_H \approx2.2 \times10^{21}\ \mathrm{cm}^{-3}), ρ≈3.33×10−4 Ωcm\rho\approx3.33 \times10^{-4}\ \Omega\mathrm{cm} at 300 K, and the residual resistance ratio (RRR) is 3.3. The resistivities of the six samples with the highest doping levels exhibit a T2T^2 relationship at low temperatures and cusps between 5.5 and 12 K, indicative of a magnetic phase transition (Figure 4(a)). The sample with the seventh-highest doping level (nH≈7.3×1018 cm−3n_H \approx7.3 \times10^{18}\ \mathrm{cm}^{-3}) shows an upturn in resistivity upon cooling, deviating from typical metallic behavior. We therefore identify the metal-insulator transition (MIT) carrier concentration is between the seventh- and sixth-highest doping level (7.3×1018−5.5×1019 cm−37.3 \times10^{18} - 5.5 \times10^{19}\ \mathrm{cm}^{-3}). For the undoped sample (nH<1.7×1016 cm−3n_H < 1.7 \times10^{16}\ \mathrm{cm}^{-3}), the resistivity increases monotonically upon cooling, consistent with insulating behavior. As seen from Figure 4(b), the as-grown sample shows a decrease in nHn_H upon cooling, reflecting its insulating behavior, whereas the OVD samples exhibit nearly temperature-independent carrier concentrations, consistent with prior work [53].

Resistivity and Hall carrier concentration measurements

Figure 4. Resistivity and Hall carrier concentration measurements. (a) Resistivity (ρ\rho) as a function of temperature (TT) for varying Hall carrier concentration (nHn_H), with nHn_H values specified in (b). With increasing doping, the system gradually evolves to a metallic state. Inset: Zoom of low-temperature resistivities of the four samples with the highest carrier concentrations. The cusps indicate a magnetic transition. (b) Carrier concentration as a function of temperature. The as-grown sample exhibits insulating behavior, with a decreasing carrier concentration upon cooling, whereas nHn_H is nearly temperature-independent for the OVD samples. The specified carrier concentration of the as-grown sample is the room-temperature value, whereas in all other cases it is the temperature-average. (c) ρ\rho vs. T2T^2 and results of fits (dashed lines) to ρ=ρ0+AT2\rho= \rho_0 + AT^2 for the six highest doping levels of the present study (nHn_H values specified in (b)). Inset: Extended result for nH=5.5×1019 cm−3n_H = 5.5 \times10^{19}\ \mathrm{cm}^{-3}. (d) Log-log plot of A(nH)A(n_H). Also shown are prior experimental results [53], the simple heuristic n−1n^{-1} dependence, and the result of our three-band model calculation.

The resistivity of the six metallic samples exhibits quadratic temperature dependence over an extended temperature range, and the results of fits to the form ρ=ρ0+AT2\rho=\rho_0 + AT^2 for the six highest doping levels are shown in Figure 4(c) and (Figure 4d). With increasing carrier concentration, the quadratic relationship extends to higher temperatures (∼150\sim150 K, 220 K, 240 K, 280 K, 300 K, and 300 K; note that our data do not extend above 300 K; for additional information regarding these fits, see Figure S17 and Figure S18 [57]). As seen from Figure 4(d), the extracted A(n)A(n) values are in overall good agreement with prior results [53] and extend this earlier work to both lower and higher doping levels.

Resistivity fit results

Figure S17. Resistivity fit results. The colors and the markers are the same as the ones in the main text Figure 3, Figure 4, and Figure 5, and the carrier concentrations are shown in (d). (a) ρ\rho vs. T2T^{2} and results of fits (dashed lines) to ρ=ρ0+AT2\rho= \rho_{0} + AT^{2} for the five highest doping levels of the present study. Inset: ρ\rho vs. T2T^{2} and fit for the sixth highest doping level. (b) Fits were performed within a temperature range from 2 K to Tmax⁡T_{\max}. The fit parameter AA changes as the fit range (i.e., Tmax⁡T_{\max}) changes. (c) Goodness of fit. With increasing doping, the R2R^{2} value is closest to 1 for Tmax⁡,opt=150T_{\max,\mathrm{opt}} = 150, 220, 240, 280, 300, and 300 K. These “optimal” values were used in (a) and are shown in (d). Note that the maximum temperature of our measurements was 300 K, and thus this analysis is less reliable at the highest doping levels.

Fit results for the residual resistivity at the six highest doping levels

Figure S18. Fit results for the residual resistivity at the six highest doping levels. The colors and the markers are the same as the ones in main text Figure 3, Figure 4, and Figure 5.

The band structure of cubic ETO with AFM order, from which parameters were obtained for the three-band model, is shown in Figure 7(c-Figure 7d). Results from DFT + UU calculations performed with and without SOC are given to illustrate the dependence of the Ti-dd band splitting on relativistic effects in the system. Interestingly, with SOC included, the Ti-dd bands at Γ\Gamma split into a single band that is 31 meV higher than the other two lower energy bands, while the splitting of the two lowest energy bands is negligible. This behavior of cubic ETO with AFM order, including SOC is similar to that of cubic STO [70]. In STO, the tetragonality splits the two lowest energy bands further by about 5 meV [71], and we adopt this for our ETO three-band model instead of calculating tetragonal AFM ETO, for simplicity. In contrast, the band structure results obtained for ETO with FM order are different from AFM, with large splitting between all three bands (as shown in Figure S8 and Figure S9). See the Supplemental Material [57] for additional information and a discussion of the effect of varying the on-site Coulomb interaction on the hybridization between the Eu-4ff and Ti-3dd orbitals.

DFT + $U$ results for magnetic exchange constants, energy difference between AFM and FM states, and electronic band structures

Figure 7. DFT + UU results for magnetic exchange constants, difference in ground state energy between the FM and GG-type AFM states, and band structure. (a) First-, second-, and third-nearest-neighbor exchange interactions vs. electron carrier concentration. Bottom scale: number of electrons per Eu. Top scale: carrier density, denoted here as nn (in contrast to the Hall concentration nHn_H obtained experimentally). (b) Energy difference between AFM and FM states vs. electron carrier concentration. The experimental AFM-FM transition region is indicated as well (see also Figure 5(e, Figure 5f)). (c,d) Electronic band structure of undoped cubic ETO in the AFM state with and without spin-orbit coupling (SOC), respectively. SOC lifts the degeneracy at the Γ\Gamma-point of the three lowest conduction bands. In the AFM state, spin-up and spin-down bands are degenerate. The energy is expressed with respect to the Fermi energy EFE_F.

We adopted the simple parabolic three-band model of Ref. [53], but rather than using band minima parameters and effective mass values for STO [53], we adjusted these to match ETO based on our DFT calculation. The carrier concentrations of each band is:

ni(EF)=13π(2miℏ2)3/2∫EiEFE−Ei dE(1)n_i(E_F) = \frac{1}{3\pi}\left(\frac{2m_i}{\hbar^2}\right)^{3/2}\int_{E_i}^{E_F}\sqrt{E-E_i}\,\mathrm{d}E \tag*{(1)}

with Fermi energy EFE_F, band minima at E0E_0, E1E_1, and E2E_2, effective masses m0m_0, m1m_1, and m2m_2, and total electron density n(EF)=n0+n1+n2n(E_F)=n_0+n_1+n_2. We used m0=1.24mem_0=1.24m_e, m1=3.47mem_1=3.47m_e, m2=1.62mem_2=1.62m_e, E0=0E_0=0 meV, E1=5E_1=5 meV (approximate), and E2=31E_2=31 meV from our DFT + UU calculation (see also below and Figure S13 and Figure S14 [57]). Upon inverting EF(n)E_F(n), we obtain A(n)∝[EF(n)]−2A(n)\propto[E_F(n)]^{-2} and, consequently, A∝nαA\propto n^\alpha, where α\alpha changes as nn increases. In the three-band regime (EF>E2E_F>E_2), α\alpha initially reaches a peak value of -1 and then gradually decreases to -4/3 as nn increases (see also Figure S14 [57]). Thus, we first fit the data to A(n)=F1⋅n−1A(n)=F_1\cdot n^{-1} at the six highest doping levels, and then adjust the pre-factor F2F_2 of the three-band result A(n)=F2⋅nαA(n)=F_2\cdot n^\alpha to coincide with F1⋅n−1F_1\cdot n^{-1} at nH=5.5×1019 cm−3n_H=5.5\times10^{19}\ \mathrm{cm}^{-3}, as shown in Figure 4(d). We note that the simple heuristic behavior A(n)∝n−1A(n)\propto n^{-1} is roughly in agreement with our data, but that the three-band model provides a better description as AA decreases faster at higher nHn_H as α\alpha becomes smaller than -1 (Figure 4(d)).

Fits to the lowest three conduction bands along the Γ–X direction

Figure S13. Fits to the lowest three conduction bands along the Γ−X\Gamma- X direction using the band structures of undoped AFM ETO using U=6.2U = 6.2 eV with SOC. The blue solid lines are the DFT + UU band structure. The gray dashed lines are fitted parabolic curves. The fit range is from 0 to 0.06 A˚−1\mathrm{\AA}^{-1}. The energy is shifted to the conduction band minimum (CBM).

Plots of $A(n)$ and exponent $\alpha$ for the three-band model

Figure S14. (a) A(n)A(n) for the three-band model. (b) The corresponding exponent α\alpha in A∝nαA \propto n^{\alpha}. The three-band model is used for the analysis in Figure 4. The Lifshitz transition carrier concentrations are marked by vertical dashed lines.

Magnetic phase diagram

We conducted detailed magnetization measurements to investigate the effect of doping on the magnetic properties of OVD ETO. The field-cooled (FC) magnetic susceptibility (χ\chi), inverse susceptibility (1/χ1/\chi), and temperature derivative (dχ/dT\mathrm{d}\chi/\mathrm{d}T) are presented in Figure 5(a-Figure 5c), respectively. Below nH=3.5×1020 cm−3n_H=3.5\times10^{20}\ \mathrm{cm}^{-3}, χ(T)\chi(T) exhibits a maximum indicative of AFM order [26, 72, 73]. The Néel temperature extracted in this manner is nearly independent of carrier concentration and in the TN=5.2T_N=5.2 - 5.5 K range, consistent with prior studies of La-doped [42, 43, 50], Al-doped [47], Nb-doped [74], and OVD [53] ETO that found mild or no doping dependences. For nH>3.5×1020 cm−3n_H>3.5\times10^{20}\ \mathrm{cm}^{-3}, the data are consistent with FM rather than AFM order, with a nearly temperatureindependent susceptibility below TCT_C (Figure 5(a)). Similar to work on FM EuTi1−x_{1-x}Alx_xO3_3 [74] and EuTiO3−x_{3-x}Hx_x [48], we extract the Curie temperature from the minimum of dχ/dTd\chi/dT (Figure 5(c)).

Magnetization measurements and phase diagram of OVD ETO

Figure 5. Magnetization measurements and phase diagram of OVD ETO. (a) Temperature dependence of magnetic susceptibility (χ\chi) for samples with varying carrier concentration, obtained with a 20 Oe field under field-cooled conditions along either <100> or <110> crystalline direction. Inset: Zoom of the low-temperature region. The nHn_{\mathrm{H}} values are indicated in (c). (b) Inverse of magnetic susceptibility as a function of temperature. The inset shows Curie-Weiss fits. (c) Temperature derivative of magnetic susceptibility. (d) χ\chi and dχ/dT\mathrm{d}\chi/\mathrm{d}T for the sample with nH≈3.5×1020 cm−3n_{\mathrm{H}} \approx3.5 \times10^{20}\ \mathrm{cm}^{-3}; dχ/dT\mathrm{d}\chi/\mathrm{d}T shows a minimum at ∼\sim5.5 K. (e) Temperature-doping phase diagram showing the relationship between carrier concentration and magnetic order. The AFM transition temperatures TNT_{\mathrm{N}} are determined from the cusps in χ(T)\chi(T) in (a). The FM transition temperatures TCT_{\mathrm{C}} are determined from the minima of dχ(T)/dT\mathrm{d}\chi(T)/\mathrm{d}T in (c). The sample with nH≈3.5×1020 cm−3n_{\mathrm{H}} \approx3.5 \times10^{20}\ \mathrm{cm}^{-3} exhibits the characteristics of a Néel transition, with a kink at TN∼5.25T_{\mathrm{N}} \sim5.25 K, as well as a paramagnetic (PM) feature down to the lowest measured temperature of ∼3.8\sim3.8 K, possibly an indication of macroscopic two-phase co-existence. This sample also shows a local minimum in dχ(T)/dT\mathrm{d}\chi(T)/\mathrm{d}T at TC∼5.5T_{\mathrm{C}} \sim5.5 K. The large open symbols indicate the results from specific heat measurements. The white region represents the AFM-FM-PM transition regime, defined by the nearest AFM and FM carrier concentrations around nH≈3.5×1020 cm−3n_{\mathrm{H}} \approx3.5 \times10^{20}\ \mathrm{cm}^{-3}. Inset: TCT_{\mathrm{C}} of the four FM samples with a linear carrier-concentration scale. (f) Curie-Weiss temperatures obtained from the fits to the data in (b). In (e) and (f), the Lifshitz transition carrier concentrations nc1n_{\mathrm{c1}} and nc2n_{\mathrm{c2}}, where the Fermi energy crosses the second and third bands, as determined from our three-band model, are indicated as vertical dashed lines. The thick gray band in (f) marks the AFM-FM-PM transition region, as defined same as the white region in (e).

The susceptibility and its temperature derivative for the sample with the intermediate carrier concentration nH=3.5×1020n_H = 3.5 \times10^{20} cm−3^{-3} are shown separately in Figure 5(d). χ(T)\chi(T) still exhibits a kink at ∼5.25\sim5.25 K, suggesting the presence of residual AFM order. In addition, dχ/dTd\chi/dT also shows a local minimum at ∼5.5\sim5.5 K, indicative of FM order. Moreover, upon further cooling, χ(T)\chi(T) increases monotonically, and dχ/dTd\chi/dT remains negative and does not cross zero, indicative of paramagnetic behavior down to the lowest measured temperature of 3.8 K. We considered whether this sample resides in a miscibility gap, i.e., if its response indicates a first-order phase transition with macroscopic AFM and FM regions. To this end, we modeled this sample’s susceptibility as a linear superposition of the susceptibilities of the two samples with adjacent doping levels, nH=5.5×1019n_H = 5.5 \times10^{19} cm−3^{-3} and 9.0×10209.0 \times10^{20} cm−3^{-3}, with respective weights of about 65% and 35% (Figure S6 [57]). Although this superposition captures the response above ∼5.5\sim5.5 K, it does not describe the observed low-temperature behavior that is indicative of a third, paramagnetic contribution. Adding an extra paramagnetic component captures this low-TT trend better (Figure S6 [57]). Assuming that we can rule out macroscopic chemical inhomogeneity, the evidence for AFM, FM, and paramagnetic contributions might be understood within a three-dimensional percolation picture, whereby two distinct species (with AFM and FM order) can simultaneously percolate at intermediate concentrations and co-exist with non-percolating zero-dimensional clusters (paramagnetic regions) [75, 76]. As demonstrated in Figure S3 [57], we consistently observed this complex magnetic behavior in multiple samples with approximately the same doping level.

Resistivity data for samples S1 and S2 reduced at 900 °C for 12 h

(a)

Magnetic susceptibility data for samples S1 and S2 reduced at 900 °C for 12 h

(b)

Figure S3. Example of data reproducibility for two samples reduced at 900 °C for 12 h. (a) Resistivity. (b) Corresponding magnetic susceptibility data for the same two samples. The results for sample S2 are the same as those shown in the main text (nH=3.5×1020 cm−3n_H = 3.5 \times10^{20}\ \mathrm{cm}^{-3}).

Magnetic susceptibility of three samples near the Hall concentration where the AFM-FM transition occurs

Figure S6. Magnetic susceptibility of three samples near the Hall concentration where the AFM-FM transition occurs. In order to test if the response of the sample with nH=3.5×1020 cm−3n_{\mathrm{H}} = 3.5 \times10^{20}\ \mathrm{cm}^{-3} can be interpreted as a mixed state of states with the two measured adjacent doping levels, its susceptibility is modeled as a linear superposition of the susceptibilities of samples with 5.5×1019 cm−35.5 \times10^{19}\ \mathrm{cm}^{-3} and 9.0×1020 cm−39.0 \times10^{20}\ \mathrm{cm}^{-3}, depicted as a dashed blue line (see main text). The superposition of AFM and FM contributions alone does not capture the monotonic increase of χ\chi at very low temperatures. Adding a paramagnetic contribution χ=C/T\chi= C/T (orange curve), leads to a better description in the low-temperature region (dashed line).

The susceptibility was fit to a Curie-Weiss form (Figure 5(b) inset), and the doping dependence of the extracted Curie temperature, θCW\theta_{\mathrm{CW}}, is summarized in Figure 5(f). Similar to our analysis of the resistivity data, we investigated how the fit temperature range affects θCW\theta_{\mathrm{CW}}. Specifically, by varying Tmin⁡T_{\min} from 10 K to 20 K, and fitting the data between Tmin⁡T_{\min} and 50 K, we obtained reliable error estimates (see also Figure S5 [57]). For as-grown ETO, we find θCW=2.97(4)\theta_{\mathrm{CW}} = 2.97(4) K, which is overall consistent with the prior results θCW=3.6\theta_{\mathrm{CW}} = 3.6 K [28], 3.8 K [33], 3.5 K [73], and 3 K [47]. Notably, although ETO exhibits AFM order, θCW\theta_{\mathrm{CW}} has a small positive value, likely due to magnetic frustration [47, 73]. FM samples exhibit a concomitant increase of TCT_C and θCW\theta_{\mathrm{CW}}, with TC=11.4(9)T_C = 11.4(9) K and θCW=8.64(3)\theta_{\mathrm{CW}} = 8.64(3) K at the highest electron concentration of the present study. TCT_C exhibits an approximately linear doping dependence (see Figure 5(e), inset). This behavior is distinct from the effects of doping via cation substitution (A-site: La [42, 44, 49], Gd [45]; and B-site: Nb [46], Al [47]) and hydrogen substitution (EuTiO3−x_{3-x}Hx_x [48]), in which cases TCT_C was found to remain constant or decrease as nn increases. Figure 5(e) shows the magnetic phase diagram of OVD ETO established in the present work.

Curie-Weiss fit results

Figure S5. Curie-Weiss fit results. (a) Data for 1/χ1/\chi and fits (dashed lines, fit range 12.5 - 50 K). The colors and the markers are the same as the ones in Figure 3, Figure 4, and Figure 5. (b) Fit results for θCW\theta_{\mathrm{CW}} for different fit interval cut-offs Tmin⁡T_{\min} in the 10 K to 20 K range. (c) R2R^2 values corresponding to (b).

We note here that we also obtained MM vs. HH data at 4.2 K for the as-grown (AFM) and most highly doped (FM) samples (see Figure S15 [57]). The observed steep increase in MM at low HH and the clear saturation at high HH exhibited by the nH=2.2×1021n_H = 2.2 \times10^{21} cm−3^{-3} sample indicates robust FM order. The as-grown sample exhibits a smaller slope at low fields and, moreover, a convex dependence on HH. The latter is consistent with the AFM ground state and associated with spin-flips, as observed previously [26, 48]. For both samples, the magnetization gradually saturates at high fields and approaches the ideal Eu2+^{2+} (S=7/2S = 7/2) moment of 7 μB\mu_{\mathrm{B}} at 5 T. We note that the AFM sample shows a slower, more gradual change compared with the FM sample, similar to H-doped ETO [48] and Nb-doped ETO [74]. Consistent with prior studies of undoped and doped EuTiO3_3 [26, 48, 77, 78], we observe hardly any hysteresis. While a lack of hysteresis is expected in the AFM case, the weak hysteresis in the FM case is unusual. However, this observation for OVD ETO is consistent with prior measurements of La-doped [42, 45], Nb-doped [74, 77], and H-doped [48] samples. Small hysteresis loops have been observed in other Eu-base ferromagnets (e.g., Eu8_8Ga16_{16}Ge30_{30} [79]), as they are soft ferromagnets with a small coercive field.

Magnetization versus magnetic field for as-grown and OVD samples

Figure S15. MM-HH data at 4.2 K for the as-grown sample and an OVD sample with the highest doping level. Inset: Zoom of low-field region. The gray dashed line is a guide to the eye, indicating its deviation from the linear behavior. For both samples, the magnetization gradually increases and approaches the ideal Eu2+\mathrm{Eu}^{2+} (S=7/2S = 7/2) spin moment of 7 μB\mu\mathrm{B} at 5 T.

Specific heat

We further employed specific heat measurements to probe the thermodynamic properties of undoped and OVD ETO. We measured three samples on which we had carried out the charge transport and magnetization measurements described above: the as-grown AFM crystal and the two FM samples with Hall carrier concentrations of nH=2.0×1021n_{\mathrm{H}} = 2.0 \times10^{21} cm−3^{-3} and nH=2.2×1021n_{\mathrm{H}} = 2.2 \times10^{21} cm−3^{-3}. Figure 6(a,Figure 6b) show CpC_{\mathrm{p}} and Cp/TC_{\mathrm{p}}/T, respectively. For all three samples, CPC_{\mathrm{P}} decreases upon cooling and exhibits distinct peaks indicative of their respective magnetic transition temperatures, as highlighted in the inset of Figure 6(a). Above the magnetic transition temperatures, CPC_{\mathrm{P}} and CP/TC_{\mathrm{P}}/T overlap for all three samples due to the small electronic and magnetic contributions at higher temperatures. In the 12–300 K range, CP(T)C_{\mathrm{P}}(T) for the undoped sample was fit to a Debye-Einstein model (Eq. (2) [57]), which captures both the dispersive acoustic and the relatively dispersionless optic phonon modes (Figure 6(c)). The resultant characteristic temperatures are ΘD=245(10)\Theta_{\mathrm{D}} = 245(10) K and ΘE=635(18)\Theta_{\mathrm{E}} = 635(18) K, with corresponding energies ℏωD=kBΘD∼21\hbar\omega_{\mathrm{D}} = k_{\mathrm{B}}\Theta_{\mathrm{D}} \sim21 meV and ℏωE=kBΘE∼55\hbar\omega_{\mathrm{E}} = k_{\mathrm{B}}\Theta_{\mathrm{E}} \sim55 meV. The Debye model effectively captures the low-energy Eu and Ti modes, whereas the Einstein model captures the high-energy oxygen modes (see also Figure 2(b)). Also shown in Figure 6(c) is CpC_{\mathrm{p}} obtained from our DFT + UU phonon calculation by summing over all modes across the Brillouin zone (Eq. (3) [57]). Both the Debye-Einstein model and the DFT + UU calculation show good agreement with our specific heat data.

Specific heat measurements

Figure 6. Specific heat measurements. (a) Cp(T)C_p(T) for one as-grown (nH=1.7×1016 cm−3n_H = 1.7 \times10^{16}\ \mathrm{cm}^{-3}) and two FM metallic (nH=2.0×1021 cm−3n_H = 2.0 \times10^{21}\ \mathrm{cm}^{-3} and 2.2×1021 cm−32.2 \times10^{21}\ \mathrm{cm}^{-3}) samples compared with a prior result for undoped ETO [26]. The three samples are the same as those for which charge transport and magnetization data are shown in Figure 3 and Figure 4. Inset: Low-temperature result with logarithmic temperature scale. The downward arrows indicate the peak positions of Cp(T)C_p(T): 5.0±0.95.0 \pm0.9 K, 6.3±0.76.3 \pm0.7 K, and 8.6±0.98.6 \pm0.9 K; these characteristic temperatures are slightly lower than TNT_N and TCT_C extracted from the magnetization data (see Figure 5(e)). (b) Cp/TC_p/T vs. TT. Inset: Low-temperature results with logarithmic temperature scale. (c) Cp(T)C_p(T) for the undoped sample compared with DFT +U+ U modeling and a Debye-Einstein fit. The modeling includes only the phonon contribution. The Debye-Einstein model gives Debye and Einstein temperatures of ΘD∼245\Theta_D \sim245 K and ΘE∼635\Theta_E \sim635 K, respectively. Inset: Cp/TC_p/T vs. TT compared with DFT +U+ U and Debye-Einstein model. (d) Cp/T3C_p/T^3 vs. TT at low temperatures. All three samples show a hump at about 25 K, consistent with a prior study of undoped ETO [26], which corresponds to the lowest-energy Eu mode in the phonon DOS (see Figure 2(b)). At low temperatures, Cp/T3C_p/T^3 is compared with the DFT +U+ U phonon calculation, a Debye-only fit (resulting in ΘD∼436\Theta_D \sim436 K), and a Debye fit plus an Einstein mode at 11 meV.

Our data for as-grown ETO are consistent with previous reports [26, 64]. For FM OVD ETO, they furthermore exhibit characteristic magnetic transition temperatures consistent with our transport and magnetization results. In an effort to further explore the thermodynamics of as-grown and doped ETO and the relationship with the lattice dynamics, we show Cp/T3C_{\mathrm{p}}/T^3 vs. TT in Figure 6(d). All three samples exhibit a distinct increase at low temperatures, with respective magnetic contributions, but share a common hump at about 25 K. This hump corresponds to the dispersionless 11 meV Eu phonon mode, as will be discussed below.

Carrier concentration effects on spin interaction and ground state

In order to better understand the changes in the magnetic ground state with increasing carrier concentration, we performed DFT + UU calculations of ETO in the cubic phase, with nn varied from zero to 0.1 electrons per Eu, and calculated the first-, second-, and third-nearest-neighbor Eu-Eu exchange interactions, J1J_1, J2J_2, and J3J_3 (see Figure 1(b)). As in previous work on cubic ETO, the on-site Coulomb interaction, UEuU_{\mathrm{Eu}}, and on-site exchange (Hund’s exchange), JEuJ_{\mathrm{Eu}}, were chosen to be 5.7 eV and 1.0 eV, respectively, to obtain the experimental ratio of Néel to Curie temperatures [36, 39, 67]. We note that UEuU_{\mathrm{Eu}} values in the 5.57 to 8.0 eV range were used in prior first-principles studies to better capture experimental findings for the lower-symmetry tetragonal phase [22, 37, 41, 43, 80]. We found the consideration of the cubic phase to be sufficient for capturing the carrier-concentration dependence of the exchange interactions. For JEuJ_{\mathrm{Eu}}, a value of 1.0 eV is a common choice [39, 43, 67, 80]. The dependence of the exchange constants on doping is summarized in Figure 7(a). Notably, most prior studies of undoped ETO only included exchange interactions up to the second-nearest neighbor, and, consequently, J1J_1 and J2J_2 were assigned to be negative (FM) and positive (AFM), respectively [41, 81]. We emphasize that the third-nearest-neighbor exchange interaction is crucial to stabilize the AFM ground state in the undoped case [40, 82]. As a result, our calculation shows that J1J_1 is positive (AFM), J2J_2 is negative (FM), and J3J_3 is positive (AFM) in undoped ETO, with J1J_1 close to zero, while the former calculation without including J3J_3 results in a negative J1J_1 [41]. As seen in Figure 7(a), as nn increases, J1J_1 decreases significantly and changes sign, whereas J2J_2 and J3J_3 increase slowly. Figure 7(b) shows the energy difference between the AFM and FM states as the carrier concentration increases. In the undoped case, this difference is negative, and the AFM ground state is favored. As nn increases, the energy difference between the two states decreases in magnitude, changes sign at n∼5.5×1020 cm−3n \sim5.5 \times10^{20}\ \mathrm{cm}^{-3}, and the FM state becomes energetically preferred. This trend agrees well with our experimental findings and with a prior computational study [41]. The calculated “critical” carrier concentration is very close to our experimental result of n∼3.5×1020 cm−3n \sim3.5 \times10^{20}\ \mathrm{cm}^{-3}. It is interesting that J1J_1 is the most sensitive to nn, whereas J3J_3 is somewhat unaffected. This is likely a result of the geometry of the perovskite structure, which dictates that J3J_3 is dominated by Ti-mediated exchange processes (Figure 1(a,Figure 1b)), whereas J1J_1 has contributions from both O- and Ti-mediated hopping processes. The sensitivity of J1J_1 to electron-doping is equivalent to increased occupation of Ti-dd orbitals, as discussed below.

Discussion

Seeking to relate our exchange interaction results and charge transport measurements to the electronic structure, we calculated the electronic band structure for cubic ETO with AFM order, both with and without spin-orbit coupling (SOC) effects (Figure 7(c,Figure 7d)). We performed the band structure calculations for the cubic phase, consistent with our DFT + UU calculations. Consideration of the higher-symmetry cubic phase reduces the number of exchange interactions that need to be accounted for, yet still provides insight into which exchange constants and mechanisms are affected by changes in the carrier concentration. In the AFM state, the spin-up and spin-down bands are degenerate, as expected. Without SOC, the lowest conduction bands are triply degenerate at the Γ\Gamma point (Figure 8(c)), and SOC lifts this degeneracy (Figure 8(d)). The effective masses m0=1.24mem_0 = 1.24m_e, m1=3.47mem_1 = 3.47m_e, m2=1.62mem_2 = 1.62m_e, and band splitting energies E0=0E_0 = 0 meV, E1=5E_1 = 5 meV (approximate as described earlier), and E2=31E_2 = 31 meV were used for our simple parabolic threeband model to calculate the transport coefficient A(n)A(n) in Figure 4(d). The full AA vs. nn relationship of the three-band model is illustrated in Figure S14 [57]. The Lifshitz transition carrier concentrations nc1n_{c1} and nc2n_{c2}, where the Fermi energy crosses the second and the third bands, are indicated in Figure 5(e,Figure 5f). As seen in Figure 4(d), for metallic ETO, the observed AA vs. nn relationship follows expectations for a three-band model, similar to metallic STO at high carrier concentrations (nH≥5.5×1019 cm−3n_H \ge5.5 \times10^{19}\ \mathrm{cm}^{-3}) [53]. Regardless of the overall similarities (perovskite structure, incipient ferroelectricity, structural phase transition, and band structure), ETO and STO show somewhat distinct charge transport behavior. Notably, the carrier concentration in nn-type STO can be easily tuned to the highly dilute one-band regime [9, 83, 84]. The sixth highest doped ETO sample lies at border between the two- and three-band regimes, whereas the five highest doped samples lie in the three-band regime. We have not succeeded in preparing metallic samples with lower carrier concentrations. We note that in our DFT + UU band structure calculation, the electrons are directly added to the system without oxygen vacancy defects. As the carrier concentration increases, the Fermi energy gradually crosses the first, the second, and the third conduction bands. However, OVD and cation doping introduce point disorder, and the carriers are localized at dilute doping levels, which cannot be captured solely by band physics. When the carrier concentration nn reaches the Mott criterion nc1/3aB∗≈0.25n_c^{1/3}a_B^* \approx0.25, where aB∗a_B^* is the effective Bohr radius, the doped electrons delocalize and become conductive [85]. ETO exhibits a considerably smaller low-temperature permittivity than STO (ε∼400\varepsilon\sim400 vs. ∼20,000\sim20{,}000 [35]) and, consequently, a smaller effective Bohr radius aB∗=a0 ε me/mia_B^* = a_0\ \varepsilon\ m_e/m_i, where a0a_0 is the standard Bohr radius. As a result, the insulator-metal transition in OVD ETO occurs at a higher carrier concentration (between the sixth- and seventh-highest doping levels: 7.3×1018 cm−3∼5.5×1019 cm−37.3 \times10^{18}\ \mathrm{cm}^{-3} \sim5.5 \times10^{19}\ \mathrm{cm}^{-3}) than in nn-type STO [43, 53], which has been reported to exhibit metallic behavior down to nH=8.5×1015 cm−3n_H = 8.5 \times10^{15}\ \mathrm{cm}^{-3} [86]. In perovskite oxides, where there is no sharp MIT, the carrier concentration required for metallicity is much higher than the Mott criterion [86]. In practice, nc1/3aB∗≈10n_c^{1/3}a_B^* \approx10 has been able to give a good prediction for STO, KTO, and ETO [53]. When using ε∼400\varepsilon\sim400 and m1=1.24 mem_1 = 1.24\,m_e, we obtain a MIT carrier concentration 2.0×1020 cm−32.0 \times10^{20}\ \mathrm{cm}^{-3}, which is close to our experimental findings (7.3×1018 cm−3∼5.5×1019 cm−37.3 \times10^{18}\ \mathrm{cm}^{-3} \sim5.5 \times10^{19}\ \mathrm{cm}^{-3}). The behavior that MIT carrier concentration is much higher than the Mott criterion in STO semiconductor systems has been explained by strong compensation effects coming from the background charged defects [87].

Band structure plots of undoped AFM ETO with SOC projected to Ti-d and Eu-f orbitals

Figure 8. Band structure of undoped AFM ETO with SOC from DFT + UU projected to (a) Ti-dd and (b) Eu-ff orbitals. The energy zero is set to the Fermi energy EFE_{\mathrm{F}}.

As demonstrated by our charge transport and magnetization measurements, the magnetic order involving Eu moments exhibits a significant doping dependence. In order to further establish how the doped electrons couple with the magnetic properties, we calculated the atomic-orbital-projected band structure. As seen from Figure 8, the low-energy conduction bands have Ti-3dd character, and the Eu-4ff band lies just below, as previously noted based on first-principles calculations [41, 43]. As a result, when nn and the Fermi energy increase, itinerant electrons fill the Ti-3dd bands and interact with Eu spins, analogous to the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction, as proposed for rare-earth or Nb-doped bulk ETO and Nb-doped oxygen-deficient ETO films [45, 51, 88]. The nearest-neighbor interaction constant J1J_1 has contributions from both Oand Ti-3d hopping processes, and, consequently, the itinerant electrons in the Ti-3d bands cause the magnitude and sign of J1J_1 to change. We note that charge conduction, originating from Ti-3d electrons, can be tuned by an external magnetic field due to the strong coupling to local spins, similar to La-doped and Nb-doped ETO [45, 88]. As shown in Figure S16 [57], we measured the longitudinal resistance RxxR_{xx} of the nH=2.0×1021 cm−3n_{\mathrm{H}} = 2.0 \times10^{21}\ \mathrm{cm}^{-3} sample in different applied magnetic fields. In zero field, RxxR_{xx} shows a slight increase upon cooling from 30 K to TC=7.5T_{\mathrm{C}} = 7.5 K, and a sharp cusp at TCT_{\mathrm{C}}, followed by a decrease below TCT_{\mathrm{C}}. With increasing magnetic field, the resistance close to the transition temperature is suppressed, and the cusp gradually disappears, indicating a large negative magnetoresistance. This behavior agrees well with prior studies of La-doped [42, 44, 45] and H-doped [48] ETO, and it reflects the FM nature of OVD ETO at higher electron concentrations as well as the coupling between itinerant electrons in Ti-3d orbitals and Eu moments.

Longitudinal resistance of the $n_H = 2.0 \times 10^{21}\ \mathrm{cm}^{-3}$ sample under different applied magnetic fields

Figure S16. Longitudinal resistance of the nH=2.0×1021 cm−3n_H = 2.0 \times10^{21}\ \mathrm{cm}^{-3} sample under different applied magnetic fields.

Lastly, we analyzed the low-temperature properties of the specific heat and the corresponding lattice dynamics. In Figure 6(d), CP/T3C_P/T^3 vs. TT of one undoped and two metallic OVD ETO are compared with a DFT + UU phonon calculation of undoped ETO. As noted, CP/T3C_P/T^3 for the three measured samples shows somewhat distinct increases below ∼12\sim12 K due to magnetic contributions, but the data share a common hump near 25 K. To elucidate the origin of this hump, we first fit CPC_P in the 12 - 100 K range to a Debye-only model. This model, which captures the contribution from acoustic modes, approaches a constant in the low-temperature limit and fails to reproduce the hump at ∼25\sim25 K. To account for this feature, we include an additional Einstein term with ΘE∼128\Theta_E \sim128 K (following a similar approach used for undoped/doped STO [89]). This Einstein temperature corresponds to a flat phonon mode with energy ℏωE=kBΘE=11\hbar\omega_E = k_B\Theta_E = 11 meV, and successfully reproduces the observed behavior of CP/T3C_P/T^3. This low-energy Einstein mode is distinct from the higher Einstein temperature ΘE=635(18)\Theta_E = 635(18) K obtained from full-temperature fits, which accounts for the high-energy oxygen vibrations. Instead, the 11 meV mode corresponds to the Eu-dominated peak in the phonon density of states as seen in Figure 2(b) and is independent of doping, consistent with its nonpolar character and its distinction from the ferroelectric soft mode. We further compare this behavior with our DFT + UU phonon calculation. While the calculations contain a magnetic specific heat contribution, and therefore do not capture the low-temperature upturn associated with magnetic order, they reproduce both the Debye-like low-temperature flattening and the hump near 25 K that arises from the flat phonon branch at ∼11\sim11 meV. This agreement supports our assignment of the 25 K feature to the low-energy Eu-dominated phonon mode.

The hump observed in CP/T3C_P/T^3 of ETO is consistent with prior measurements on undoped ETO [26]. Similarly, undoped and doped STO [89] also exhibit a hump, which is independent of doping and corresponds to the Sr-dominated phonon peak at ∼12.9\sim12.9 meV in the phonon density of states [90]. In ETO, this feature occurs at slightly lower energy, reflecting the larger atomic mass of Eu relative to Sr, and likewise exhibits no doping dependence. Our analysis of the low-temperature specific heat, therefore, highlights the close similarity of the lattice dynamics in ETO and STO. Moreover, the quantitative agreement between DFT + UU calculation, including the reproduction of the Einstein-like feature near 25 K, provides further validation of the computed phonon density of states and the interpretation of the synchrotron x-ray diffuse scattering results.

Conclusion

In conclusion, we successfully employed CaH2CaH_2 as an oxygen getter to tune the carrier concentration of the perovskite EuTiO3−δEuTiO_{3-\delta} up to an unprecedented electron density of n≈1021 cm−3n \approx10^{21}\ \mathrm{cm}^{-3}. This allowed us to uncover a magnetic ground-state transition of the metallic system from AFM to FM order at nH∼3.5×1020 cm−3n_H \sim3.5 \times10^{20}\ \mathrm{cm}^{-3}, with a maximum Curie temperature of TC∼11T_C \sim11 K at the highest electron density, comparable to prior results obtained via cation substitution. A magnetic ground-state change at about the same doping level is also predicted by our complementary DFT + UU calculation. Our theoretical analysis of the charge transport properties, considering spin-orbit coupling, indicates three non-degenerate active bands in the doping range of the present study and yields an AA-coefficient that is consistent with the data. We find good agreement between our x-ray diffuse scattering data and thermal diffuse scattering calculation, which indicates that the response is dominated by phonon contributions. The diffuse scattering patterns share similarities with SrTiO3SrTiO_3, which reflects the closely similar lattice dynamics of these two perovskites, with a common cubic-tetragonal phase transition associated with a zone-boundary soft-phonon mode. Moreover, our diffuse scattering and DFT + UU phonon calculation reveals a dispersionless Eu mode at ∼11\sim11 meV, consistent with our specific heat data for as-grown and OVD samples. The present study reveals that oxygen reduction provides a knob to simultaneously tune the electrical and magnetic properties in metallic EuTiO3EuTiO_3. It furthermore lays the foundation for a range of queries, such as searches for a multiferroic metal and superconductivity, the further fine-tuning of the carrier concentration and magnetic order, and the systematic study of the interplay between the incommensurate structural instability, antiferrodistortive modes, and oxygen vacancies and other defects.

Acknowledgments

We thank D. Pelc, S. Sharma, Y. Tao, and C. Leighton for discussions, and Y. Tao and C. Leighton for their help with the specific heat measurements. This work was supported by the Department of Energy through the University of Minnesota Center for Quantum Materials, under grant number DE-SC-0016371. Research conducted at the Center for High-Energy x-ray Science (CHEXS) is supported by the National Science Foundation (BIO, ENG, and MPS Directorates) under award DMR-2342336.

Supplemental Material for Tuning charge-transport properties and magnetic order in metallic EuTiO3−δ_{3-\delta}

Xing He; Chiou Yang Tan; Issam Khayr; Zach Van Fossan; Richard J. Spieker; Dayu Zhai; Sarah Anderson; Dinesh Shukla; Suchismita Sarker; Javier Garcia-Barriocanal; Turan Birol; Martin Greven

Supplementary text: A. Sample preparation. B. In-house x-ray characterization. C. Resistivity and magnetization. D. Synchrotron diffuse x-ray scattering. E. Specific heat. F. Data reproducibility. G. Curie-Weiss fits. H. Spin exchange constants, ground-state energy, and electronic band structure. I. Phonon dispersion, phonon density of states, thermal diffuse scattering. J. Three-band model. K. Additional magnetization and transport data. L. Fits to transport data.

Supplementary references S1-21

Supplementary figures S1-18

Sample preparation

EuTiO3\mathrm{EuTiO}_{3} (ETO) crystals were grown with the floating-zone method, as in prior studies [S1,S2]. Eu2O3\mathrm{Eu}_{2}\mathrm{O}_{3} (Thermo Scientific Chemicals, 99.99%) and TiO2\mathrm{TiO}_{2} (Sigma-Aldrich, rutile, 99.98%) powders with stoichiometry Eu2Ti2O7\mathrm{Eu}_{2}\mathrm{Ti}_{2}\mathrm{O}_{7} were mixed, ground, and then compressed at 60 MPa to form a 6 mm-diameter feed rod of ∼\sim8-10 cm length. The feed rod was sintered in a tube furnace at 1050 °C for 10 h, with flowing 5%H2\mathrm{H}_{2}/95%Ar gas to avoid the formation of Eu2Ti2O7\mathrm{Eu}_{2}\mathrm{Ti}_{2}\mathrm{O}_{7} with tri-valent Eu3+\mathrm{Eu}^{3+}. A small part of each sintered feed rod was used as a seed. Single crystals with a diameter of ∼\sim5 mm were grown in a 5%H2\mathrm{H}_{2}/95%Ar atmosphere with a growth speed of 10 mm/h and a relative rotation of 60 rpm between the feed rod and seed. The atmospheric pressure was maintained at 4-5 bar. A picture of an as-grown crystal is shown in Figure 1(d) (inset). The crystals were cut into small pieces and characterized with a Laue x-ray diffractometer. Each piece was aligned along ⟨110⟩\langle110\rangle or ⟨100⟩\langle100\rangle for further measurements. Our reduction technique utilizes CaH2\mathrm{CaH}_{2} as a getter material [S3,S4]. As-grown ETO crystals were sealed with CaH2\mathrm{CaH}_{2} in a quartz tube under vacuum (pressure below 100 mTorr), with an ETO to CaH2\mathrm{CaH}_{2} mass ratio of 1:12. The ETO crystals were situated in a molybdenum boat above the CaH2\mathrm{CaH}_{2} to avoid direct contact. The sealed quartz tubes were heated for 12 h at temperatures between 700 °C and 1050 °C, depending on the desired carrier concentration, with higher temperatures resulting in higher carrier concentrations. There exist several possible reduction mechanisms [S5]: CaH2\mathrm{CaH}_{2} decomposes and generates H2\mathrm{H}_{2} gas, and the H2\mathrm{H}_{2} gas reduces the oxygen from ETO and forms water; CaH2\mathrm{CaH}_{2} takes up oxygen and forms Ca(OH)2\mathrm{Ca(OH)}_{2}.

In-house x-ray characterization

Portions of the as-grown crystals were ground for powder x-ray diffraction with a Rigaku Smartlab SE diffractometer, which used Cu KK-α\alpha photons (λ=1.5406\lambda= 1.5406 Å). Single crystal Laue diffraction patterns such as those in Figure 1 were taken with a Photonic Science back-reflection system. A typical single-crystal with a polished ⟨100⟩\langle100\rangle surface was characterized via single-crystal diffraction with a Rigaku Smartlab XE diffractometer (Cu KK-α\alpha photons), as shown in Figure S1.

Resistivity and magnetization

Resistivity and Hall data were obtained with the standard four-point van der Pauw method using a Quantum Design Inc. Physical Property Measurement System (PPMS) in AC drive mode. The four contacts were made with a wire bonder and aluminum wires were used. For metallic samples, the typical contact resistance was 1 Ω\Omega. The resistivity and Hall data are shown in Figure 3 and Figure 4.

Magnetization measurements were performed with a Quantum Design Inc. Magnetic Property Measurement System (MPMS) in a 20 Oe DC field. The field was applied at 50 K, and data were collected upon cooling to about 3.8 K. The magnetization data are shown in Figure 5.

Synchrotron diffuse x-ray scattering

Synchrotron diffuse x-ray scattering measurements was performed at beamline QM2^2 ID4B at the Cornell High Energy Synchrotron Source (CHESS). The incident photon energy was 50 keV. Three-axis rotation was used to cover a large Q\mathbf{Q}-space volume. An Oxford Cryosystems N2_2 Cryostream was used to cool the sample. Data for an as-grown sample, measured at 300 K and 175 K, are shown in Figure 2 and Figure 3. We also measured the highest doped sample (nH=2.2×1021n_H = 2.2 \times10^{21} cm−3^{-3}) at 300 K, and the diffuse scattering pattern is shown in Figure S3. The doped sample exhibits features that are similar to those for undoped ETO, including sharp Bragg peaks in the HK0HK0 plane (Figure S3) indicative of high crystalline quality, and thermal diffuse scattering with streaks and curvilinear features in the half-integer planes HK0.5HK0.5 (Figure S3) and HK2.5HK2.5 (Figure S3). However, we note that the doped sample does not show diffuse features at RR points, likely because of the lower antiferrodistortive (AFD) transition temperature [53] and the resultant stiffer room-temperature RR-point soft mode, similar to OVD SrTiO3_3 [S7]. We did not determine the doping dependence of the AFD transition temperature in the present study.

Specific heat

We used relaxation calorimetry in a PPMS to measure the specific heat (CpC_p) of an as-grown sample that exhibits AFM order (nH=1.7×1016n_H = 1.7 \times10^{16} cm−3^{-3} at 300 K) and of two OVD samples that exhibit FM order (nH=2.0×1021n_H = 2.0 \times10^{21} cm−3^{-3} and 2.2×10212.2 \times10^{21} cm−3^{-3}). The data are shown in Figure 6. Apiezon N grease was used to affix crystals to PPMS platforms and 2% temperature pulses were used. The measurements were conducted between 300 K and 2 K and performed three times at each temperature for improved accuracy. The data for the undoped sample were fit between 12 and 300 K to a standard Debye-Einstein model [S8]:

Cp(T)=9(1−X)NNAkB(TΘD)3∫0ΘD/Texx4(ex−1)2 dx+3XNNAkBeΘE/T(eΘE/T−1)2(ΘET)2(2)C_p(T) = 9(1-X)NN_Ak_B\left(\frac{T}{\Theta_D}\right)^3\int_0^{\Theta_D/T}\frac{e^x x^4}{(e^x-1)^2}\,dx + 3XNN_Ak_B\frac{e^{\Theta_E/T}}{(e^{\Theta_E/T}-1)^2}\left(\frac{\Theta_E}{T}\right)^2 \tag*{(2)}

where XX and 1−X1-X are the respective weights of the Einstein and Debye contributions, NN is the number of atoms in the unit cell, NAN_A is Avogadro’s number, kBk_B is Boltzmann’s constant, and x=ℏω/kBTx = \hbar\omega/k_BT, where ω\omega is the phonon frequency. The fit results for the Debye and Einstein temperatures are ΘD=245(10)\Theta_D = 245(10) K and ΘE=635(18)\Theta_E = 635(18) K, respectively. We also estimated CpC_p from our DFT + UU phonon calculation by using the equation:

Cp=∑qvkB(ℏω(qv)kBT)2exp⁡(ℏω(qv)/kBT)[exp⁡(ℏω(qv)/kBT)−1]2(3)C_p = \sum_{qv} k_B\left(\frac{\hbar\omega(qv)}{k_BT}\right)^2\frac{\exp(\hbar\omega(qv)/k_BT)}{[\exp(\hbar\omega(qv)/k_BT)-1]^2} \tag*{(3)}

where qq is the wavevector, vv is the mode, and ω\omega is the phonon frequency.

Data reproducibility

Samples reduced under the same nominal conditions generally exhibited closely similar resistivities and magnetic susceptibilities, as demonstrated in Figure S3 and Figure S4.

Resistivity data for samples S3 and S4 reduced at 1050 °C for 12 h

(a)

Magnetic susceptibility data for samples S3 and S4 reduced at 1050 °C for 12 h

(b)

Figure S4. Example of data reproducibility for two samples reduced at 1050 °C for 12 h. (a) Resistivity. (b) Corresponding magnetic susceptibility data. The results for sample S3 are the same as those shown in the main text (nH=2.2×1021 cm−3n_H = 2.2 \times10^{21}\ \mathrm{cm}^{-3}).

Curie-Weiss fits

The magnetic susceptibility was fit to the Curie-Weiss form C=χ/(T−θCW)C = \chi/(T - \theta_{\mathrm{CW}}), and the fit results and extracted Curie temperatures θCW\theta_{\mathrm{CW}} are summarized in Figs. 4 and S5. As EuTiO3−δ_{3-\delta} transitions from the AFM to the FM state, θCW\theta_{\mathrm{CW}} is seen to increase. Similar to our analysis of the resistivity data (see below), we investigated how the lower limit Tmin⁡T_{\min} of the fit temperature interval [Tmin⁡T_{\min}, 50 K] affects the extracted θCW\theta_{\mathrm{CW}} and the R2R^2 value. By varying Tmin⁡T_{\min} from 10 K to 20 K, we obtained the error estimates in Figs. 4(f) and S5(b). The corresponding R2R^2 values are shown in Fig. S5(c).

We also considered if the sample with the Hall concentration nH=3.5×1020 cm−3n_{\mathrm{H}} = 3.5 \times10^{20}\ \mathrm{cm}^{-3} resides in a miscibility gap, i.e., if its response reflects a mixed state associated with adjacent doping levels. To this end, we modeled this sample’s susceptibility as a linear superposition of the susceptibilities of the two measured samples with adjacent doping levels, nH=5.5×1019 cm−3n_{\mathrm{H}} = 5.5 \times10^{19}\ \mathrm{cm}^{-3} and 9.0×1020 cm−39.0 \times10^{20}\ \mathrm{cm}^{-3}, χ(3.5×1020 cm−3)=Coef1χ(5.5×1019 cm−3)+Coef2χ(9.0×1020 cm−3)\chi(3.5 \times10^{20}\ \mathrm{cm}^{-3}) = \mathrm{Coef}_1\chi(5.5 \times10^{19}\ \mathrm{cm}^{-3}) + \mathrm{Coef}_2\chi(9.0 \times10^{20}\ \mathrm{cm}^{-3}), as shown in Fig. S6 (blue dashed curve). The fit values of coefficients are Coef1≈0.65\mathrm{Coef}_1 \approx0.65 and Coef2≈0.35\mathrm{Coef}_2 \approx0.35. This superposition captures the response above the magnetic transition temperature, but it does not describe the paramagnetic trend at low temperatures. Adding an extra paramagnetic contribution, χ=C/T\chi= C/T (orange curve in Fig. S6), approximately captures the low-temperature trend (green dashed curve in Fig. S6). As described in the main text, this can be understood within a three-dimensional percolation picture.

Spin exchange constants, ground state energy, and electronic band structure

Density functional theory (DFT) simulations were performed with VASP [58, 59] using the Projector-Augmented Wave (PAW) method and Perdew–Burke–Ernzerhof (PBE) Generalized Gradient Approximation (GGA) [S11,S12] exchange correlation functional. In order to account for the strong interactions between Eu ff-electrons, the DFT + UU formalism was used with the on-site Coulomb interaction set to UEu=5.7U_{\mathrm{Eu}} = 5.7 eV for calculations related to the magnetic exchange constants, while values of 5.7 eV and 6.2 eV were used for electronic band structure calculations. The experimentally observed room-temperature lattice constant of about 3.90 Å [56] was used for all calculations, along with a plane-wave cutoff energy of 600 eV, on-site exchange of JEu=1.0J_{\mathrm{Eu}} = 1.0 eV, and a 6×6×66 \times6 \times6 kk-point mesh.

For the determination of the magnetic exchange constants, electronic relaxation was performed for multiple collinear magnetic orders in a 2×2×22 \times2 \times2 40-atom supercell. The Heisenberg model was then used to perform a linear regression in which the coefficients associated with first-, second-, and third-nearest-neighbor exchange interactions J1J_1, J2J_2, and J3J_3 were determined. The electronic band structure was obtained using a two-step procedure in which a self-consistent electronic relaxation was performed for a cubic 2×2×22 \times2 \times2 supercell with spin-orbit coupling, followed by a non-self-consistent electronic relaxation along desired high symmetry directions in the Brillouin zone using the converged charge density. The bands of the primitive 5-atom unit cell were identified from the band structure of the 2×2×22 \times2 \times2 supercell using the unfolding method provided by the PyProcar software package [S14]. While a value of UEu=5.7U_{\mathrm{Eu}} = 5.7 eV offers experimental agreement with the ratio of the Curie and Neel temperatures of cubic ETO, this was found to cause unwanted hybridization between the Eu-4ff and Ti-3dd bands, which makes the system metallic. A value of UEu=6.2U_{\mathrm{Eu}} = 6.2 eV was therefore used to remove the hybridization between the Eu-4ff and Ti-3dd bands and create a gap which would allow the use of the three-band model considered in this study. Given the close similarity between the curvature of the Ti-3dd bands for the two different values of UEuU_{\mathrm{Eu}}, we expect the three-band model fitted using UEu=6.2U_{\mathrm{Eu}} = 6.2 eV to retain interpretability. The calculated unfolded AFM band structures using UEu=5.7U_{\mathrm{Eu}} = 5.7 eV and 6.2 eV without and with SOC are shown in Figure S7. The calculated FM band structures using UEu=5.7U_{\mathrm{Eu}} = 5.7 eV and 6.2 eV without and with SOC are shown in Figure S8 and Figure S9.

Unfolded band structures of undoped AFM ETO

Figure S7. Unfolded band structure of undoped AFM ETO from DFT + UU. (a) U=5.7U = 5.7 eV without SOC. (b) U=6.2U = 6.2 eV without SOC. (c) U=5.7U = 5.7 eV with SOC. (d) U=6.2U = 6.2 eV with SOC. The thin solid lines are folded band structures from the 2x2x2 supercell. The thick solid lines are unfolded band structures for a 5-atom unit cell. The energy zero is shifted to the Fermi energy EfE_f.

Phonon dispersion, phonon density of states, thermal diffuse scattering

DFT simulations were performed using Quantum Espresso (QE) [60, 61] for phonon property simulations. A 6 x 6 x 5 Monkhorst-Pack electronic kk-point mesh and plane-wave cutoff energy of 480 Ry were used for a tetragonal unit cell (20 atoms). The generalized gradient approximation (GGA) in the Perdew–Burke–Ernzerhof revised for solid (PBE-sol) was utilized. We used the DFT + UU formalism with an on-site Coulomb parameter U=8U = 8 eV and Hund’s exchange JH=1.0J_H = 1.0 eV. The lattice constant of the tetragonal cell was fully relaxed with energy criteria of 10−510^{-5} Ry and force criteria of 10−410^{-4} Ry/Bohr (relaxed lattice constants: a=b=5.487a = b = 5.487 Å, c=7.823c = 7.823 Å are close to the experimental lattice constants at 175 K: a=b=5.515a = b = 5.515 Å, c=7.805c = 7.805 Å) [56]. Harmonic phonon calculations were performed using the nonzero-displacement approach as implemented in the Phonopy package [62]. A 2 x 2 x 1 supercell, a 3 x 3 x 5 Monkhorst-Pack electronic kk-point mesh, and a plane-wave cutoff energy of 480 Ry were used in QE. The phonon calculation was also used to compare with specific heat measurements. We also calculated the phonon band structure with U=6.8U = 6.8 eV, and the obtained dispersions are very similar to those obtained with U=8.0U = 8.0 eV, as shown in Figure S10. This indicates that the phonon properties are not very sensitive to the choice of UU value. Phonon energy, eigenvectors, and x-ray form factors were used to calculate the dynamic structure factor S(Q,ω)S(\mathbf{Q},\omega) [S18], which is defined as:

Phonon band structures of EuTiO3

Figure S10. Phonon band structures of EuTiO3, calculated from DFT with U=8U = 8 and 6.86.8 eV. The weak instability near the Gamma point is an error that emerge as a result of the manner in which the acoustic sum rule is treated and does not affect any of the results discussed in the present work.

S(Q,ω)∝∣F(Q)∣2n+1ω(4)S(\mathbf{Q},\omega) \propto\lvert F(\mathbf{Q}) \rvert^2 \frac{n+1}{\omega} \tag*{(4)}
F(Q)=∑j(Q⋅ej)bjMjeiQ⋅Rje−Wj(5)F(\mathbf{Q}) = \sum_j (\mathbf{Q} \cdot\mathbf{e}_j) \frac{b_j}{\sqrt{M_j}} e^{i\mathbf{Q}\cdot\mathbf{R}_j} e^{-W_j} \tag*{(5)}

where Q\mathbf{Q} is the momentum transfer, ℏω\hbar\omega the phonon energy, ℏ\hbar the reduced Planck constant, nn the Bose-Einstein occupation factor n=[exp⁡(ℏωkBT)−1]−1n = \left[\exp\left(\frac{\hbar\omega}{k_B T}\right)-1\right]^{-1}, ej\mathbf{e}_j the phonon eigenvector, bjb_j the x-ray form factor, jj the atomic index, RjR_j the atomic position in the unit cell, and WjW_j the Debye-Waller factor. The dynamic structure factor was calculated with the Phonopy package [62], and the thermal diffuse scattering intensity was obtained by integrating ℏω\hbar\omega from zero to 30 meV. Eu, Ti, Eu+Ti, and O only diffuse scattering patterns were obtained by setting the x-ray form factors of other elements to be zero, as shown in Figure S11 and Figure S12.

Thermal diffuse scattering simulation of decomposed elements in the $HK0$ plane, showing Eu only, Ti only, Eu and Ti only, and O only

Figure S11. Thermal diffuse scattering simulation of decomposed elements in the HK0HK0 plane. (a) Eu only; (b) Ti only; (c) Eu and Ti only; (d) O only. The decomposed thermal diffuse scattering patterns were calculated by assuming that the form factors of other elements are zero. For clarity, the Ti and O contributions in (b) and (d) are multiplied by a factor of 10. As shown, Ti and O contribute only at certain integer reciprocal lattice points based on the selection rules and do not contribute to thermal diffuse scattering streaks in the HK0HK0 plane.

Thermal diffuse scattering simulation of decomposed elements in the HK2.5 plane

Figure S12. Thermal diffuse scattering simulation of decomposed elements in the HK2.5\mathit{HK}2.5 plane. (a) Eu only; (b) Ti only; (c) Eu and Ti only; (d) O only. The decomposed thermal diffuse scattering patterns were calculated by assuming that the form factors of other elements are zero. For clarity, the Ti and O contributions in (b) and (d) are multiplied by a factor of 10. As shown, the combination of Eu and Ti results in the curvilinear pattern, and O contributes to the diffuse scattering at half-integer points in the HK2.5\mathit{HK}2.5 plane.

Three-band model

As discussed in the main text, we adopted the simple parabolic three-band model from Ref. [53], which studied ETO, but used parameters for SrTiO3_3, and modified this to the undoped cubic AFM ETO case. Considering the simple case of three parabolic bands with band minima at E0E_0, E1E_1, and E2E_2, and effective masses m0m_0, m1m_1, and m2m_2, the carrier concentration of each band is

ni(EF)=13π2(2miℏ2)3/2∫EiEFE−Ei dE,(6)n_i(E_F) = \frac{1}{3\pi^2}\left(\frac{2m_i}{\hbar^2}\right)^{3/2}\int_{E_i}^{E_F}\sqrt{E-E_i}\,\mathrm{d}E, \tag*{(6)}

and the total electron density is n(EF)=n0+n1+n2n(E_F) = n_0 + n_1 + n_2. We used m0=1.24 mem_0 = 1.24\,m_e, m1=3.47 mem_1 = 3.47\,m_e, m2=1.62 mem_2 = 1.62\,m_e, E0=0E_0 = 0 meV, E1=5E_1 = 5 meV, and E2=31E_2 = 31 meV from our DFT +U+ U calculation, where mem_e is the electron mass. The effective masses were obtained by fitting the lowest three conduction bands along the Γ−X\Gamma-X direction to a parabolic function,

E(k)=Ec+ℏ2k22mi,(7)E(k) = E_c + \frac{\hbar^2 k^2}{2m_i}, \tag*{(7)}

where E(k)E(k) is the band energy, kk the wavevector in units of Å−1^{-1}, EcE_c the minimum energy, and mim_i the effective mass (see Figure S13). The effective mass mim_i is expressed in units of electron mass mem_e, with ℏ2/(2me)=3.81\hbar^2/(2m_e) = 3.81 eV·Å. We note that for undoped cubic AFM ETO, the band structures, including SOC, show no visible splitting in the two lowest conduction bands, similar to STO [70]. Including tetragonality in STO results in a small splitting about 5 meV [71], and thus we approximate the splitting of ETO by E1−E0=5E_1 - E_0 = 5 meV.

From the inverse function EF(n)E_F(n), we calculated A(n)∝EF−2(n)A(n) \propto E_F^{-2}(n). The exponent α\alpha in A∝nαA \propto n^\alpha was obtained from the gradient of a log-log plot of A(n)A(n). Figure S14 shows both A(n)A(n) and α(n)\alpha(n). Similar to the three-band model in Ref. [53], when the carrier concentration is low, and only one band is filled, α\alpha is proportional to n−4/3n^{-4/3}. As the second band starts to be filled, α\alpha suddenly increases to n−2/3n^{-2/3} and gradually reaches n−1n^{-1}. In the three-band regime, n−1n^{-1} slowly changes back to n−4/3n^{-4/3}. The Lifshitz transitions occur at the carrier concentrations nc1=1.48×1018 cm−3n_{c1} = 1.48 \times10^{18}\,\mathrm{cm}^{-3} and nc2=9.91×1019 cm−3n_{c2} = 9.91 \times10^{19}\,\mathrm{cm}^{-3} (see Figure S14).

Additional magnetization and transport data

We performed additional MM vs. HH measurements at 4.2 K of the as-grown and most highly doped samples, as shown in Figure S15. For both samples, the magnetization gradually saturates at high fields and approaches the ideal Eu2+^{2+} (S=7/2S = 7/2) spin moment of 7 μ\muB at 5 T.

We performed longitudinal resistance measurements of the nH=2.0×1021 cm−3n_H = 2.0 \times10^{21}\,\mathrm{cm}^{-3} sample under different applied magnetic fields, as shown in Figure S16. As the magnetic field increases, the cusp in the resistance gradually disappears, and the longitudinal resistance decreases, reflecting the large magnetoresistance and the interaction between Eu spins and itinerant Ti 3d3d electrons.

Fits to transport data

In an effort to evaluate the “optimal” temperature range for fits of the resistivity to the Fermi-liquid relationship ρ=ρ0+AT2\rho= \rho_0 + A T^2, fits were performed within varying temperature ranges from 2 K to Tmax⁡T_{\max} (100 K < Tmax⁡T_{\max} < 300 K) for the data obtained at the six highest doping levels. The fit parameter AA changes as the fitting range changes, as shown in Figure S17. With increasing doping level, R2R^2 values closest to one were obtained for Tmax⁡,optT_{\max,\mathrm{opt}} = 150 K, 220 K, 240 K, 280 K, 300 K, and 300 K, as summarized in Figure S17,Figure S17). Note that the maximum temperature of our measurements was 300 K, and thus, this analysis is less reliable at the highest doping levels. These Tmax⁡,optT_{\max,\mathrm{opt}} values were used for the fits in Figure S17 and to obtain A(n)A(n). The resultant residual resistivity estimates are shown in Figure S18.

Supplemental references

[S1] D. S. Ellis, H. Uchiyama, S. Tsutsui, K. Sugimoto, K. Kato, D. Ishikawa, and A. Q. R. Baron, Phonon softening and dispersion in EuTiO3\mathrm{EuTiO}_{3}, Phys. Rev. B 86, 220301 (2012).

[72] A. Midya, P. Mandal, Km. Rubi, R. Chen, J.-S. Wang, R. Mahendiran, G. Lorusso, and M. Evangelisti, Large adiabatic temperature and magnetic entropy changes in EuTiO3\mathrm{EuTiO}_{3}, Phys. Rev. B 93, 094422 (2016).

[S3] D. Li, K. Lee, B. Y. Wang, M. Osada, S. Crossley, H. R. Lee, Y. Cui, Y. Hikita, and H. Y. Hwang, Superconductivity in an infinite-layer nickelate, Nature 572, 624 (2019).

[S4] J. Zhao, Y.-L. Gan, G. Yang, Y.-G. Zhong, C.-Y. Tang, F.-Z. Yang, G. N. Phan, Q.-T. Sui, Z. Liu, G. Li, X.-G. Qiu, Q.-H. Zhang, J. Shen, T. Qian, L. Lu, L. Yan, G.-D. Gu, and H. Ding, Continuously Doping BiM2Sr2CaCu2O8+δ\mathrm{BiM}_{2}\mathrm{Sr}_{2}\mathrm{CaCu}_{2}\mathrm{O}_{8+\delta} into Electron-Doped Superconductor by CaH2\mathrm{CaH}_{2} Annealing Method, Chinese Phys. Lett. 39, 077403 (2022).

[S5] M. A. Hayward, M. A. Green, M. J. Rosseinsky, and J. Sloan, Sodium hydride as a powerful reducing agent for topotactic oxide deintercalation: synthesis and characterization of the nickel(I) oxide LaNiO2\mathrm{LaNiO}_{2}, J. Am. Chem. Soc. 121, 8843 (1999).

[53] J. Engelmayer, X. Lin, C. P. Grams, R. German, T. Fröhlich, J. Hemberger, K. Behnia, and T. Lorenz, Charge transport in oxygen-deficient EuTiO3\mathrm{EuTiO}_{3}: The emerging picture of dilute metallicity in quantum-paraelectric perovskite oxides, Phys. Rev. Mater. 3, 051401 (2019).

[S7] Q. Tao, B. Loret, B. Xu, X. Yang, C. W. Rischau, X. Lin, B. Fauqué, M. J. Verstraete, and K. Behnia, Nonmonotonic anisotropy in charge conduction induced by antiferrodistortive transition in metallic SrTiO3\mathrm{SrTiO}_{3}, Phys. Rev. B 94, 035111 (2016).

[S8] Y. Zhang, A. Saha, F. Tutt, V. Chaturvedi, B. Voigt, W. Moore, J. Garcia-Barriocanal, T. Birol, and C. Leighton, Thermal properties of the metallic delafossite PdCoO2\mathrm{PdCoO}_{2}: A combined experimental and first-principles study, Phys. Rev. Mater. 6, 115004 (2022).

[58] G. Kresse and J. Furthmüller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996).

[59] G. Kresse and J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54, 11169 (1996).

[S11] J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou, and K. Burke, Restoring the density-gradient expansion for exchange in solids and surfaces, Phys. Rev. Lett. 100, 136406 (2008).

[S12] P. E. Blöchl, Projector augmented-wave method, Phys. Rev. B 50, 17953 (1994).

[56] M. Allieta, M. Scavini, L. J. Spalek, V. Scagnoli, H. C. Walker, C. Panagopoulos, S. S. Saxena, T. Katsufuji, and C. Mazzoli, Role of intrinsic disorder in the structural phase transition of magnetoelectric EuTiO3\mathrm{EuTiO}_{3}, Phys. Rev. B 85, 184107 (2012).

[S14] U. Herath, P. Tavadze, X. He, E. Bousquet, S. Singh, F. Muñoz, and A. H. Romero, PyProcar: A Python library for electronic structure pre/post-processing, Computer Physics Communications 251, 107080 (2020).

[60] P. Giannozzi et al., QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials, J. Phys.: Condens. Matter 21, 395502 (2009).

[61] P. Giannozzi et al., Advanced capabilities for materials modelling with Quantum ESPRESSO, J. Phys.: Condens. Matter 29, 465901 (2017).

Supplemental figures

References

  1. [1]M. J. Coak, C. R. S. Haines, C. Liu, S. E. Rowley, G. G. Lonzarich, and S. S. Saxena, Quantum critical phenomena in a compressible displacive ferroelectric, Proceedings of the National Academy of Sciences 117, 12707 (2020).
  2. [2]S. E. Rowley, L. J. Spalek, R. P. Smith, M. P. M. Dean, M. Itoh, J. F. Scott, G. G. Lonzarich, and S. S. Saxena, Ferroelectric quantum criticality, Nature Phys 10, 367 (2014).DOI
  3. [3]K. A. Müller and H. Burkard, SrTiO₃: An intrinsic quantum paraelectric below 4 K, Phys. Rev. B 19, 3593 (1979).DOI
  4. [4]X. Li, B. Fauqué, Z. Zhu, and K. Behnia, Phonon thermal hall effect in strontium titanate, Phys. Rev. Lett. 124, 105901 (2020).DOI
  5. [5]A. Narayan, A. Cano, A. V. Balatsky, and N. A. Spaldin, Multiferroic quantum criticality, Nature Mater 18, 3 (2019).
  6. [6]Y. Yamada and G. Shirane, Neutron scattering and nature of the soft optical phonon in SrTiO₃, J. Phys. Soc. Jpn. 26, 396 (1969).DOI
  7. [7]T. Esswein and N. A. Spaldin, Ferroelectric, quantum paraelectric, or paraelectric? Calculating the evolution from BaTiO₃ to SrTiO₃ to KTaO₃ using a single-particle quantum mechanical description of the ions, Phys. Rev. Res. 4, 033020 (2022).
  8. [8]M. N. Gastiasoro, J. Ruhman, and R. M. Fernandes, Superconductivity in dilute SrTiO₃: A review, Annals of Physics 417, 168107 (2020).DOI
  9. [9]X. Lin, G. Bridoux, A. Gourgout, G. Seyfarth, S. Krämer, M. Nardone, B. Fauqué, and K. Behnia, Critical doping for the onset of a two-Band superconducting ground state in SrTiO₃-δ, Phys. Rev. Lett. 112, 207002 (2014).
  10. [10]C. Liu, X. Zhou, D. Hong, B. Fisher, H. Zheng, J. Pearson, J. S. Jiang, D. Jin, M. R. Norman, and A. Bhattacharya, Tunable superconductivity and its origin at KTaO₃ interfaces, Nat Commun 14, 951 (2023).
  11. [11]C. Liu, X. Yan, D. Jin, Y. Ma, H.-W. Hsiao, Y. Lin, T. M. Bretz-Sullivan, X. Zhou, J. Pearson, B. Fisher, J. S. Jiang, W. Han, J.-M. Zuo, J. Wen, D. D. Fong, J. Sun, H. Zhou, and A. Bhattacharya, Two-dimensional superconductivity and anisotropic transport at KTaO₃ (111) interfaces, Science 371, 716 (2021).
  12. [12]M. N. Gastiasoro, M. E. Temperini, P. Barone, and J. Lorenzana, Generalized Rashba electron-phonon coupling and superconductivity in strontium titanate, Phys. Rev. Res. 5, 023177 (2023).
  13. [13]M. N. Gastiasoro, T. V. Trevisan, and R. M. Fernandes, Anisotropic superconductivity mediated by ferroelectric fluctuations in cubic systems with spin-orbit coupling, Phys. Rev. B 101, 174501 (2020).
  14. [14]G. Venditti, M. E. Temperini, P. Barone, J. Lorenzana, and M. N. Gastiasoro, Anisotropic Rashba coupling to polar modes in KTaO₃, J. Phys. Mater. 6, 014007 (2023).
  15. [15]K. Dunnett, A. Narayan, N. A. Spaldin, and A. V. Balatsky, Strain and ferroelectric soft-mode induced superconductivity in strontium titanate, Phys. Rev. B 97, 144506 (2018).
  16. [16]J. M. Edge, Y. Kedem, U. Aschauer, N. A. Spaldin, and A. V. Balatsky, Quantum critical origin of the superconducting dome in SrTiO₃, Phys. Rev. Lett. 115, 247002 (2015).
  17. [17]C. W. Rischau, X. Lin, C. P. Grams, D. Finck, S. Harms, J. Engelmayer, T. Lorenz, Y. Gallais, B. Fauqué, J. Hemberger, and K. Behnia, A ferroelectric quantum phase transition inside the superconducting dome of Sr₁−xCaxTiO₃−δ, Nature Phys 13, 643 (2017).
  18. [18]Y. Tomioka, N. Shirakawa, and I. H. Inoue, Superconductivity enhancement in polar metal regions of Sr₀.₉₅Ba₀.₀₅TiO₃ and Sr₀.₉₈₅Ca₀.₀₁₅TiO₃ revealed by systematic Nb doping, Npj Quantum Mater. 7, 1 (2022).DOI
  19. [19]A. Stucky, G. W. Scheerer, Z. Ren, D. Jaccard, J.-M. Poumirol, C. Barreteau, E. Giannini, and D. van der Marel, Isotope effect in superconducting n-doped SrTiO₃, Sci Rep 6, 37582 (2016).DOI
  20. [20]Y. Tomioka, N. Shirakawa, K. Shibuya, and I. H. Inoue, Enhanced superconductivity close to a non-magnetic quantum critical point in electron-doped strontium titanate, Nat Commun 10, 738 (2019).DOI
  21. [21]C. Herrera, J. Cerbin, A. Jayakody, K. Dunnett, A. V. Balatsky, and I. Sochnikov, Strain-engineered interaction of quantum polar and superconducting phases, Phys. Rev. Mater. 3, 124801 (2019).DOI
  22. [22]K. Ahadi, L. Galletti, Y. Li, S. Salmani-Rezaei, W. Wu, and S. Stemmer, Enhancing superconductivity in SrTiO₃ films with strain, Sci. Adv. 5, eaaw0120 (2019).DOI
  23. [23]D. J. Scalapino, A common thread: The pairing interaction for unconventional superconductors, Rev. Mod. Phys. 84, 1383 (2012).DOI
  24. [24]P. Gegenwart, Q. Si, and F. Steglich, Quantum criticality in heavy-fermion metals, Nature Phys 4, 186 (2008).DOI
  25. [25]J. H. Lee, X. Ke, N. J. Podraza, L. F. Kourkoutis, T. Heeg, M. Roeckerath, J. W. Freeland, C. J. Fennie, J. Schubert, D. A. Muller, P. Schiffer, and D. G. Schlom, Optical band gap and magnetic properties of unstrained EuTiO₃ films, Applied Physics Letters 94, 212509 (2009).DOI
  26. [26]A. Jaoui, S. Jiang, X. Li, Y. Tomioka, I. H. Inoue, J. Engelmayer, R. Sharma, L. Pätzold, T. Lorenz, B. Fauqué, and K. Behnia, Glasslike thermal conductivity and narrow insulating gap of EuTiO₃, Phys. Rev. Mater. 7, 094604 (2023).DOI
  27. [27]O. I. Malyi, X.-G. Zhao, A. Bussmann-Holder, and A. Zunger, Local positional and spin symmetry breaking as a source of magnetism and insulation in paramagnetic EuTiO₃, Phys. Rev. Mater. 6, 034604 (2022).arxiv.org/abs/2107.00142
  28. [28]H. Akamatsu, K. Fujita, H. Hayashi, T. Kawamoto, Y. Kumagai, Y. Zong, K. Iwata, F. Oba, I. Tanaka, and K. Tanaka, Crystal and electronic structure and magnetic properties of divalent europium perovskite oxides EuMO₃ (M = Ti, Zr, and Hf): experimental and first-principles approaches, Inorg. Chem. 51, 4560 (2012).
  29. [29]V. Goian, S. Kamba, J. Hlinka, P. Vaněk, A. A. Belik, T. Kolodiazhnyi, and J. Petzelt, Polar phonon mixing in magnetoelectric EuTiO₃, Eur. Phys. J. B 71, 429 (2009).DOI
  30. [30]S. Kamba, D. Nuzhnyy, P. Vaněk, M. Savinov, K. Knı́žek, Z. Shen, E. Šantavá, K. Maca, M. Sadowski, and J. Petzelt, Magnetodielectric effect and optic soft mode behaviour in quantum paraelectric EuTiO₃ ceramics, EPL 80, 27002 (2007).DOI
  31. [31]D. S. Ellis, H. Uchiyama, S. Tsutsui, K. Sugimoto, K. Kato, D. Ishikawa, and A. O. R. Baron, Phonon softening and dispersion in EuTiO₃, Phys. Rev. B 86, 220301 (2012).DOI
  32. [32]V. Scagnoli, M. Allieta, H. Walker, M. Scavini, T. Katsufuji, L. Sagarna, O. Zaharko, and C. Mazzoli, EuTiO₃ magnetic structure studied by neutron powder diffraction and resonant x-ray scattering, Phys. Rev. B 86, 094432 (2012).
  33. [33]T. R. McGuire, M. W. Shafer, R. J. Joenk, H. A. Alperin, and S. J. Pickart, Magnetic structure of EuTiO₃, Journal of Applied Physics 37, 981 (1966).DOI
  34. [34]D. Repček, P. Proschek, M. Savinov, M. Kachlík, J. Pospíšil, J. Drahokoupil, P. Doležal, J. Prokleška, and S. Kamba, Multiferroic quantum criticality in a (Eu,Ba,Sr)TiO₃ solid solution, Phys. Rev. B 109, 214109 (2024).DOI
  35. [35]T. Katsufuji and H. Takagi, Coupling between magnetism and dielectric properties in quantum paraelectric EuTiO₃, Phys. Rev. B 64, 054415 (2001).DOI
  36. [36]J. H. Lee et al., A strong ferroelectric ferromagnet created by means of spin–lattice coupling, Nature 466, 954 (2010).
  37. [37]C. J. Fennie and K. M. Rabe, Magnetic and electric phase control in epitaxial EuTiO₃ from first principles, Phys. Rev. Lett. 97, 267602 (2006).pubmed.ncbi.nlm.nih.gov/17280465
  38. [38]X. Ke, T. Birol, R. Misra, J.-H. Lee, B. J. Kirby, D. G. Schlom, C. J. Fennie, and J. W. Freeland, Structural control of magnetic anisotropy in a strain-driven multiferroic EuTiO₃ thin film, Phys. Rev. B 88, 094434 (2013).DOI
  39. [39]T. Birol, N. A. Benedek, H. Das, A. L. Wysocki, A. T. Mulder, B. M. Abbett, E. H. Smith, S. Ghosh, and C. J. Fennie, The magnetoelectric effect in transition metal oxides: Insights and the rational design of new materials from first principles, Current Opinion in Solid State and Materials Science 16, 227 (2012).DOI
  40. [40]T. Birol and C. J. Fennie, Origin of giant spin-lattice coupling and the suppression of ferroelectricity in EuTiO₃ from first principles, Phys. Rev. B 88, 094103 (2013).DOI
  41. [41]Z. Gui and A. Janotti, Carrier-Density-Induced Ferromagnetism in EuTiO₃ Bulk and Heterostructures, Phys. Rev. Lett. 123, 127201 (2019).DOI
  42. [42]Y. Tomioka, T. Ito, and A. Sawa, Magnetotransport Properties of Eu₁−ₓLaₓTiO₃ (0 ≤ x ≤ 0.07) Single Crystals, J. Phys. Soc. Jpn. 87, 094716 (2018).DOI
  43. [43]K. Maruhashi, K. S. Takahashi, M. S. Bahramy, S. Shimizu, R. Kurihara, A. Miyake, M. Tokunaga, Y. Tokura, and M. Kawasaki, Anisotropic quantum transport through a single spin channel in the magnetic semiconductor EuTiO₃, Advanced Materials 32, 1908315 (2020).
  44. [44]K. S. Takahashi, M. Onoda, M. Kawasaki, N. Nagaosa, and Y. Tokura, Control of the anomalous Hall effect by doping in Eu₁−ₓLaₓTiO₃ thin films, Phys. Rev. Lett. 103, 057204 (2009).pubmed.ncbi.nlm.nih.gov/19792531
  45. [45]T. Katsufuji and Y. Tokura, Transport and magnetic properties of a ferromagnetic metal: Eu₁−ₓRₓTiO₃, Phys. Rev. B 60, R15021 (1999).DOI
  46. [46]L. Li, J. R. Morris, M. R. Koehler, Z. Dun, H. Zhou, J. Yan, D. Mandrus, and V. Keppens, Structural and magnetic phase transitions in EuTi₁−ₓNbₓO₃, Phys. Rev. B 92, 024109 (2015).
  47. [47]D. Akahoshi, S. Koshikawa, T. Nagase, E. Wada, K. Nishina, R. Kajihara, H. Kuwahara, and T. Saito, Magnetic phase diagram for the mixed-valence Eu oxide EuTi₁−ₓAlₓO₃, Phys. Rev. B 96, 184419 (2017).
  48. [48]T. Yamamoto, R. Yoshii, G. Bouilly, Y. Kobayashi, K. Fujita, Y. Kususe, Y. Matsushita, K. Tanaka, and H. Kageyama, An antiferro-to-ferromagnetic transition in EuTiO₃−ₓHₓ induced by hydride substitution, Inorg. Chem. 54, 1501 (2015).
  49. [49]K. Maruhashi, K. S. Takahashi, M. S. Bahramy, S. Shimizu, R. Kurihara, A. Miyake, M. Tokunaga, Y. Tokura, and M. Kawasaki, Anisotropic Quantum Transport through a Single Spin Channel in the Magnetic Semiconductor EuTiO₃, Advanced Materials 32, 1908315 (2020).
  50. [50]H. Shin, C. Liu, F. Li, R. Sutarto, B. A. Davidson, and K. Zou, Controlling the electrical and magnetic ground states by doping in the complete phase diagram of titanate Eu₁−ₓLaₓTiO₃ thin films, Phys. Rev. B 101, 214105 (2020).DOI
  51. [51]K. Kugimiya, K. Fujita, K. Tanaka, and K. Hirao, Preparation and magnetic properties of oxygen deficient EuTiO₃−δ thin films, Journal of Magnetism and Magnetic Materials 310, 2268 (2007).DOI
  52. [52]D. Shin, I. Kim, S. Song, Y.-S. Seo, J. Hwang, S. Park, M. Choi, and W. S. Choi, Defect engineering of magnetic ground state in EuTiO₃ epitaxial thin films, Journal of the American Ceramic Society 104, 4606 (2021).DOI
  53. [53]J. Engelmayer, X. Lin, C. P. Grams, R. German, T. Fröhlich, J. Hemberger, K. Behnia, and T. Lorenz, Charge transport in oxygen-deficient EuTiO₃: The emerging picture of dilute metallicity in quantum-paraelectric perovskite oxides, Phys. Rev. Mater. 3, 051401 (2019).DOI
  54. [54]J.-W. Kim, P. Thompson, S. Brown, P. S. Normile, J. A. Schlueter, A. Shkabko, A. Weidenkaff, and P. J. Ryan, Emergent superstructural dynamic order due to competing antiferroelectric and antiferrodistortive instabilities in bulk EuTiO₃, Phys. Rev. Lett. 110, 027201 (2013).pubmed.ncbi.nlm.nih.gov/23383935
  55. [55]V. Goian, S. Kamba, O. Pacherová, J. Drahokoupil, L. Palatinus, M. Dušek, J. Rohlíček, M. Savinov, F. Laufek, W. Schranz, A. Fruith, M. Kachlík, K. Maca, A. Shkabko, L. Sagarna, A. Weidenkaff, and A. A. Belik, Antiferrodistortive phase transition in EuTiO₃, Phys. Rev. B 86, 054112 (2012).DOI
  56. [56]M. Allieta, M. Scavini, L. J. Spalek, V. Scagnoli, H. C. Walker, C. Panagopoulos, S. S. Saxena, T. Katsufuji, and C. Mazzoli, Role of intrinsic disorder in the structural phase transition of magnetoelectric EuTiO₃, Phys. Rev. B 85, 184107 (2012).DOI
  57. [57]See Supplemental Materials for details regarding sample preparation, measurements, data analysis, and DFT calculations.
  58. [58]G. Kresse and J. Furthmüller, Efficiency of ab-initio total energy calculations for metals and semiconductors using a plane-wave basis set, Computational Materials Science 6, 15 (1996).DOI
  59. [59]G. Kresse and J. Furthmüller, Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set, Phys. Rev. B 54, 11169 (1996).pubmed.ncbi.nlm.nih.gov/9984901
  60. [60]P. Giannozzi et al., QUANTUM ESPRESSO: a modular and open-source software project for quantum simulations of materials, J. Phys.: Condens. Matter 21, 395502 (2009).DOI
  61. [61]P. Giannozzi et al., Advanced capabilities for materials modelling with Quantum ESPRESSO, J. Phys.: Condens. Matter 29, 465901 (2017).
  62. [62]A. Togo and I. Tanaka, First principles phonon calculations in materials science, Scripta Materialia 108, 1 (2015).arxiv.org/abs/1506.08498
  63. [63]B. J. Kennedy, G. Murphy, E. Reynolds, M. Avdeev, H. E. R. Brand, and T. Kolodiazhnyi, Studies of the antiferrodistortive transition in EuTiO₃, J. Phys.: Condens. Matter 26, 495901 (2014).DOI
  64. [64]D. Bessas, K. Z. Ruschchanskii, M. Kachlik, S. Disch, O. Gourdon, J. Bednarcik, K. Maca, I. Sergueev, S. Kamba, M. Ležaić, and R. P. Hermann, Lattice instabilities in bulk EuTiO₃, Phys. Rev. B 88, 144308 (2013).DOI
  65. [65]G. Shirane and Y. Yamada, Lattice-Dynamical Study of the 110°K Phase Transition in SrTiO₃, Phys. Rev. 177, 858 (1969).DOI
  66. [66]K. Hirota, J. P. Hill, S. M. Shapiro, G. Shirane, and Y. Fujii, Neutron- and x-ray-scattering study of the two length scales in the critical fluctuations of SrTiO₃, Phys. Rev. B 52, 13195 (1995).pubmed.ncbi.nlm.nih.gov/9980508
  67. [67]K. Z. Rushchanskii, N. A. Spaldin, and M. Ležaić, First-principles prediction of oxygen octahedral rotations in perovskite-structure EuTiO₃, Phys. Rev. B 85, 104109 (2012).DOI
  68. [68]M. Kopecký, J. Fábry, and J. Kub, X-ray diffuse scattering in SrTiO₃ and model of atomic displacements, J Appl Cryst 45, 393 (2012).DOI
  69. [69]M. Zacharias, G. Volonakis, F. Giustino, and J. Even, Anharmonic electron-phonon coupling in ultrasoft and locally disordered perovskites, Npj Comput Mater 9, 1 (2023).DOI
  70. [70]Z. Zhong, A. Tóth, and K. Held, Theory of spin-orbit coupling at LaAlO₃/SrTiO₃ interfaces and SrTiO₃ surfaces, Phys. Rev. B 87, 161102 (2013).DOI
  71. [71]D. van der Marel, J. L. M. van Mechelen, and I. I. Mazin, Common Fermi-liquid origin of T² resistivity and superconductivity in n-type SrTiO₃, Phys. Rev. B 84, 205111 (2011).
  72. [72]A. Midya, P. Mandal, Km. Rubi, R. Chen, J.-S. Wang, R. Mahendiran, G. Lorusso, and M. Evangelisti, Large adiabatic temperature and magnetic entropy changes in EuTiO₃, Phys. Rev. B 93, 094422 (2016).DOI
  73. [73]A. M. Iqbal and G. H. Jaffari, Effect of stoichiometry on electrical response and polydispersivity related to hopping polarization in EuTiO₃, Journal of Applied Physics 125, 114102 (2019).
  74. [74]L. Li, H. Zhou, J. Yan, D. Mandrus, and V. Keppens, Research Update: Magnetic phase diagram of EuTi₁−ₓBxO₃ (B = Zr, Nb), APL Materials 2, 110701 (2014).DOI
  75. [75]R. I. Dass and J. B. Goodenough, Itinerant to localized electronic transition in Sr₂FeMo₁−ₓWₓO₆, Phys. Rev. B 63, 064417 (2001).DOI
  76. [76]K.-I. Kobayashi, T. Okuda, Y. Tomioka, T. Kimura, and Y. Tokura, Possible percolation and magnetoresistance in ordered double perovskite alloys Sr₂Fe(W₁−ₓMoₓ)O₆, Journal of Magnetism and Magnetic Materials 218, 17 (2000).DOI
  77. [77]Y. Kususe, H. Murakami, K. Fujita, I. Kakeya, M. Suzuki, S. Murai, and K. Tanaka, Magnetic and transport properties of EuTiO₃ thin films doped with Nb, Jpn. J. Appl. Phys. 53, 05FJ07 (2014).DOI
  78. [78]A. M. Iqbal, G. H. Jaffari, M. Saleemi, and A. Ceylan, Relaxation dynamics and polydispersivity associated with defects and ferroelectric correlations in Ba-doped EuTiO₃, J. Phys.: Condens. Matter 29, 465402 (2017).DOI
  79. [79]B. C. Sales, B. C. Chakoumakos, R. Jin, J. R. Thompson, and D. Mandrus, Structural, magnetic, thermal, and transport properties of X₈Ga₁₆Ge₃₀ (X = Eu, Sr, Ba) single crystals, Phys. Rev. B 63, 245113 (2001).DOI
  80. [80]K. Ahadi, Z. Gui, Z. Porter, J. W. Lynn, Z. Xu, S. D. Wilson, A. Janotti, and S. Stemmer, Carrier density control of magnetism and Berry phases in doped EuTiO₃, APL Materials 6, 056105 (2018).DOI
  81. [81]H. Akamatsu, Y. Kumagai, F. Oba, K. Fujita, H. Murakami, K. Tanaka, and I. Tanaka, Antiferromagnetic superexchange via 3d states of titanium in EuTiO₃ as seen from hybrid Hartree-Fock density functional calculations, Phys. Rev. B 83, 214421 (2011).
  82. [82]P. J. Ryan, J.-W. Kim, T. Birol, P. Thompson, J.-H. Lee, X. Ke, P. S. Normile, E. Karapetrov, P. Schiffer, S. D. Brown, C. J. Fennie, and D. G. Schlom, Reversible control of magnetic interactions by electric field in a single-phase material, Nat Commun 4, 1334 (2013).DOI
  83. [83]A. Najev, N. Somun, M. Spaić, I. Khayr, M. Greven, A. Klein, M. N. Gastiasoro, and D. Pelc, Electronic spin susceptibility in metallic strontium titanate, Npj Quantum Mater. 10, 4 (2025).DOI
  84. [84]B. Fauqué, C. Collignon, H. Yoon, Ravi, X. Lin, I. I. Mazin, H. Y. Hwang, and K. Behnia, Electronic band sculpted by oxygen vacancies and indispensable for dilute superconductivity, Phys. Rev. Res. 5, 033080 (2023).
  85. [85]P. P. Edwards and M. J. Sienko, Universality aspects of the metal-nonmetal transition in condensed media, Phys. Rev. B 17, 2575 (1978).
  86. [86]A. Spinelli, M. A. Torija, C. Liu, C. Jan, and C. Leighton, Electronic transport in doped SrTiO₃: Conduction mechanisms and potential applications, Phys. Rev. B 81, 155110 (2010).DOI
  87. [87]Y. Huang, Y. Ayino, and B. I. Shklovskii, Metal-insulator transition in n-type bulk crystals and films of strongly compensated SrTiO₃, Phys. Rev. Mater. 5, 044606 (2021).DOI
  88. [88]S. Roy, N. Khan, and P. Mandal, Unconventional transport properties of the itinerant ferromagnet EuTi₁₋ₓNb₃(x=0.10−0.20), Phys. Rev. B 98, 134428 (2018).DOI
  89. [89]E. McCalla, M. N. Gastiasoro, G. Cassuto, R. M. Fernandes, and C. Leighton, Low-temperature specific heat of doped SrTiO₃: Doping dependence of the effective mass and Kadowaki-Woods scaling violation, Phys. Rev. Mater. 3, 022001 (2019).DOI
  90. [90]N. Choudhury, E. J. Walter, A. I. Kolesnikov, and C.-K. Loong, Large phonon band gap in SrTiO₃ and the vibrational signatures of ferroelectricity in ATiO₃ perovskites: First-principles lattice dynamics and inelastic neutron scattering, Phys. Rev. B 77, 134111 (2008).DOI

Paper details

vlnc:W_nv4jtb8x
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 9

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents