Effect of chemical pressure on the electronic phase transition in Ca_1−xSr_xMn₇O₁₂ films

There is growing interest in quadruple perovskites, (AA′₃)B₄O₁₂,^1–3 due to the wide range of physical properties found in this material class. These properties include giant dielectric permittivity in CaCu₃Ti₄O₁₂,^4–6 giant magnetoresistance in CaCu₃Mn₄O₁₂,⁷ heavy fermions in CaCu₃Ru₄O₁₂,^1,8 inter-site charge transfer in ACu₃Fe₄O₁₂,^9,10 bifunctional catalytic activity in AMn₇O₁₂ perovskites,¹¹ and a charge-ordering metal-insulator transition and multiferroicity in CaMn₇O₁₂.^12–14 The variety of functionalities present in (AA′₃)B₄O₁₂ compounds is enabled by their structural framework that can accommodate transition metals on both the A′ and B positions allowing both A-site and B-site ordering.^3,15 Similar to ABO₃ perovskites, the quadruple perovskites accommodate substitution on the cation sites to form solid solutions, allowing for a large variety of material systems, with unique and customizable properties for various applications.

Of the AMn₇O₁₂ compounds, CaMn₇O₁₂ has attracted considerable attention surrounding the four distinct phase transitions it exhibits. At T* = 440 K, a simultaneous charge-ordering¹³ and structural phase transition occurs from the distorted cubic (⁠$Im$3́) to rhombohedral (⁠$R$3́) structure.^1,13 The charge-ordering occurs on the B-site Mn cations with a 3:1 ratio of Mn³⁺ and Mn⁴⁺, respectively, as evidenced by local changes in the Mn–O bond lengths determined from powder diffraction measurements.^13,16 An orbital-ordering transition is observed at ∼250 K (T_OO) through scattering and Raman spectroscopy measurements.^17,18 Below 90 K (T_N1), non-collinear helical magnetic and ferroelectric ordering is observed,^12,14,19 with a second helical magnetic transition at ∼45 K (T_N2).^{1,14,16,20,21} Similar transitions are also observed in bulk SrMn₇O₁₂, with little change in the magnetic transition temperatures but a decrease of ∼60 K reported in T*.^22,23 These intriguing properties have motivated many experimental and theoretical studies exploring the nature of multiferroicity in CaMn₇O₁₂, in which the ferroelectricity and helical magnetism are believed to be intimately coupled.^{12,14,19,23–29} Most previous studies have focused on the properties of ternary CaMn₇O₁₂ and SrMn₇O₁₂; in contrast, the effect of chemical substitutions on functional behavior in quaternary compounds is not yet fully understood.^7,30 A few examples have been reported, where substitution of La for Ca on the A-site was shown to enhance the electric polarization³¹ and doping Fe³⁺, Al³⁺, and Cr³⁺ on the Mn sites suppresses charge-ordering and leaves only the cubic structure.³⁰ Finally, insertion of Cu on the B-site yields significant magnetoresistance.^7,32–36 Of particular relevance to this report, polycrystalline Ca_1−xSr_xMn₇O₁₂ ceramics with 0.05 ≤ x ≤ 0.2 exhibit an enhanced low temperature magnetization compared to the undoped (x = 0) compound.³⁷ Most recently, a second study focused on Ca_1−xSr_xMn₇O₁₂ ceramics with x less than 0.05 that showed enhanced polarization.³⁸ 

However, the effect of Sr substitution on the electronic properties remains largely unexplored, specifically at larger Sr concentrations (x > 0.2) and in thin films. Given the growing interest in electronics that utilize the correlated electron behavior and/or abrupt phase transitions of complex oxides,^39–43 a detailed understanding of how A-site substitution can be used to tune the charge-ordering transition is needed to engineer quadruple perovskites for such applications. We have synthesized a set of Ca_1−xSr_xMn₇O₁₂ thin films with various concentrations of Sr up to x = 0.6 on (LaAlO₃)_0.3(SrAl_0.5Ta_0.5O₃)_0.7 (LSAT) substrates. Sr is isovalent to Ca but has a larger ionic radius (1.31 Å and 0.99 Å for Sr²⁺ and Ca²⁺, respectively) and thus allows for the effect of chemical pressure to be studied without changing the nominal Mn valence. In conventional AMnO₃ manganites, substitution of Sr for Ca on the A-site acts to increase the tolerance factor and reduce rotations of the corner-connected MnO₆ octahedra.⁴⁴ This is known to have significant effects on physical properties, including a greater propensity for charge-ordering as the magnitude of rotations (tolerance factor) is increased (decreased).^45–47 Herein, we demonstrate that similar trends are operative in quadruple manganites. We confirm a lattice expansion with increasing Sr concentration and show that T* decreases systematically with increasing x. Our work provides insight into how to tune electronic phase transitions in quadruple perovskite thin films, which is important to engineer materials with a specific electronic switching functionality for devices or sensors.

70 nm thick Ca_1−xSr_xMn₇O₁₂ films were synthesized using oxide molecular beam epitaxy (MBE). Single crystal (001)-oriented LSAT substrates were used. The calculated lattice mismatch between the LSAT (a = 3.868 Å) substrates and CaMn₇O₁₂ (a = 3.682 Å in a pseudocubic perovskite lattice) is 5.1%. The large mismatch leads to strain relaxation of the films at the thicknesses studied here.²⁵ Growth conditions and post-growth anneals were set similar to the CaMn₇O₁₂ film growth²⁵ with an O₂ chamber pressure of 2 × 10⁻⁶ Torr and a substrate temperature of ∼$650°$C. The Ca and Sr content was controlled by tuning the effusion cell temperatures and shutter times. The films were annealed ex-situ in a tube furnace with flowing oxygen at $850°$C for 3 h and then at $200°$C for 1 h in a flowing mixture of O₂:O₃ (∼95:5).

The films were characterized by the Rutherford backscattering spectrometry (RBS), X-ray diffraction (XRD), reciprocal space mapping (RSM), and high temperature resistivity measurements. For the RBS measurements, as-grown films deposited on the MgO (001) substrates directly adjacent to the LSAT substrates were used. The SIMNRA software package was used to simulate and analyze the RBS data.⁴⁸ XRD and X-ray reflectivity (XRR) measurements were carried out using a four-circle high-resolution X-ray diffractometer (X’Pert Pro, Panalytical) with Cu–Kα₁ radiation. Reciprocal space maps were used to determine the in-plane lattice parameters in the thin films. To investigate thermal expansion and the structural phase transition, we performed temperature dependent XRD equipped with a controlled temperature stage (DHS 1100, Anton Paar) in air. We used silver paste to adhere the sample to the heating plate. Data were collected at nine different temperatures above room temperature with a step size of $25°$C. Sample realignment was conducted at each temperature to maximize the XRD intensities. We note that the temperature of the hot stage has an estimated error of ∼$10°$C. We performed high temperature in-plane DC resistivity measurements with a custom-built setup using a Keithley 6220 current source, a Keithley 2148 nanovoltmeter, and a Lakeshore 211 temperature monitor by first heating the sample to 600 K and collecting data while the sample is cooled slowly at a rate of approximately 5 K/min to room temperature. For these measurements, the temperature sensor was placed adjacent to the sample and silver paint was used to make a van der Pauw geometry electrical contact. In addition, resistivity data were collected below room temperature using a Quantum Design Physical Property Measurement System (PPMS).

To confirm the Sr-induced lattice expansion, XRD was measured as a function of x. Figure 1(a) displays room temperature XRD data from Ca_1−xSr_xMn₇O₁₂ films grown on LSAT substrates. All samples exhibit a single diffraction peak indexed to the pseudocubic 002 reflection, without any additional peaks over the 2θ range of $46°$ – $52°$⁠. The black line and arrow show the peak shifting towards smaller 2θ values with increasing x. Figure 1(b) presents the calculated c-axis lattice parameter from the 002 pseudocubic reflection as a function of Sr concentration in the films on LSAT substrates. We observe a linear lattice expansion with increasing Sr concentration, given by c = 3.683 + 0.021x Å. These results are consistent with previous work on bulk Ca_1−xSr_xMn₇O₁₂³⁷ over the range of x = 0 – 0.2 and comparable to bulk SrMn₇O₁₂, which has a pseudocubic lattice parameter of 3.702 Å.²³ Figure 1(c) shows reciprocal space maps for x = 0 and x = 0.6 samples to determine the in-plane pseudocubic lattice parameter of the samples. The obtained c/a ratios are 0.999 (x = 0) and 1.001 (x = 0.6), consistent with the films being unstrained. Based on this, we believe that the out-of-plane parameter obtained from the 002 peak is a good measure of the general pseudocubic lattice parameter of the films. We note that there is a subtle splitting of the 103 peak along L for the CaMn₇O₁₂ film, the origin of which is unknown and merits future study.

Temperature dependent XRD data were also measured to identify the effect of Sr doping on the structural phase transition temperature. Figure 2 shows the temperature dependence of the c-axis lattice parameter of the Ca_1−xSr_xMn₇O₁₂ films grown on LSAT substrates as obtained from the (002) diffraction peak. We observe a discontinuity of the c-axis parameter in the films near 400–440 K; consistent with the presence of a phase transition, the c-axis increases abruptly above the transition. While our measurements do not provide direct evidence that the phase transition is due to charge-ordering, we denote this as a charge-ordering transition based on previous experimental evidence and density functional theory calculations that revealed the state below phase transition temperature (T*) to be charge-ordered.^13,25 Also consistent with bulk CaMn₇O₁₂,⁴⁹ we observe an increase in the lattice parameter on warming across the phase transition. The temperature of the discontinuity in lattice expansion decreases with increasing x, indicating that Sr-doping decreases the structural phase transition temperature. Linear fits above and below the discontinuity were performed to determine the rate of change of the c-axis parameter on both sides of the phase transition with a 3σ error. In the charge-ordered phase (below T*), the c-axis parameter expands at a rate of 1.5(1), 1.1(1), and 1.1(1) × 10⁻⁵ Å/K for the x = 0, 0.21, and 0.6 samples, respectively. The high temperature data (above T*) have significantly more variation; the x = 0 and 0.6 samples expand at 2.0(1) and 2.2(1) × 10⁻⁵ Å/K, while the x = 0.21 sample expands at 0.9(1) × 10⁻⁵ Å/K. We believe that the high value [∼2.0(1) × 10⁻⁵ Å/K] is more accurate, as this was found in more of our samples and in the x = 0.6 film, which has the most data points above T*. In previous work on bulk CaMn₇O₁₂, the non-charge-ordered phase was found to expand at a greater rate with increasing temperature than the charge-ordered phase.⁴⁹ In contrast, the LSAT substrates exhibit simple linear expansion [∼0.9(1) × 10⁻⁵ Å/K] over the measured temperature range, as shown in the bottom panel of Fig. 2.

Temperature dependent resistivity (ρ) measurements were performed to determine how the Sr content alters the charge-ordering transition and to determine the conduction mechanisms. As can be seen in Fig. 3, Sr doping does not appear to systematically alter the magnitude of the resistivity and all samples exhibit insulating (dρ/dT < 0) behavior. While differences in resistivity at 300 K and 500 K are observed across the samples, we speculate that these variations occur due to differences in oxygen vacancy concentration or to varying defect densities and microstructural quality within the films. We note that similar variability in the resistivity magnitude was observed in undoped CaMn₇O₁₂ films, but T* was relatively insensitive to cation and/or oxygen off-stoichiometry with measured T* values varying from 420 to 429 K. To determine T* as a function of Sr doping, we fit d(ln ρ)/dT as a function of temperature⁵⁰ to a bi-gaussian function as shown in Fig. 3(c); the center position is taken as T*. Figure 3(d) shows the charge-ordering transition temperature as a function of x. Increasing the Sr concentration leads to a continuous decrease in the charge-ordering transition temperature. The T* values determined from the fits are 426, 405, 401, and 385 K for the x = 0, 0.21, 0.41, and 0.6 samples, respectively. The calculated 3σ error for all the T* values obtained from the resistivity data is ±1 K. The results are in qualitative agreement with the high temperature XRD, but there is some discrepancy in the obtained T* values between the two measurements. We believe these relate to the experimental conditions and likely do not represent the decoupling of the electronic and structural transitions in our films. We first note that our method of extracting T* from the electrical measurements yields T* values that are lower than the onset of the resistivity changes by 5–10 K. Additionally, the resistivity measurements are taken while cooling the sample, thus if the cooling rate of the film is slower than that of the thermocouple, this will also act to depress the T* value obtained from electrical measurements. In contrast, the XRD measurements are taken while heating the sample and thus may lead to enhanced T* values if the film temperature is slightly less than the sample stage. Finally, we note the fundamentally different nature of the two characterization techniques, where resistivity requires a local percolation path of low resistance while diffraction measures the global average.

To determine the dominant transport mechanism above and below T*, the temperature dependent resistivity data were fit to numerous conduction models: (A) activated [ρ = ρ₀ exp(E_A/k_BT)], (B) 3D variable range hopping (VRH) [ρ = ρ₀ exp(T₀/T)^(1/4)], (C) 2D VRH [ρ = ρ₀ exp(T₀/T)^(1/3)], (D) adiabatic polaron [ρ = ρ₀T exp(E_A/k_BT)], (E) non-adiabatic polaron [ρ = ρ₀T^(3/2) exp(E_A/k_BT)], and (F) power law [ρ = ρ₀T^m]. We find that conduction above T* can be best represented by the polaron models, consistent with our previous work on CaMn₇O₁₂,²⁵ with comparable R² values obtained from the two polaron models, thus precluding differentiation between adiabatic and non-adiabatic (D) and (E) models. Representative fits obtained above T* from the non-adiabatic polaron model (E) are shown in Fig. 4(a). Activation energies obtained from the fits range from 233 to 260 meV, as displayed in Fig. 4(b). Below T*, the activated model (A) provides a good fit to the transport, again in agreement with CaMn₇O₁₂. Figure 4(b) also presents the activation energies from these fits, which are between 180 and 210 meV. The error from the fits is on order of 1-2 meV, while the experimental error is estimated to be on order of 10 meV, which was the standard deviation obtained from measuring two CaMn₇O₁₂ films. These results indicate that while Sr doping yields a systematic reduction in T*, it does not have a significant effect on either the dominant conduction mechanism above and below T* or the relevant activation energies.

We have carried out an experimental study on the effect of Sr doping in CaMn₇O₁₂ films. The structural phase transition temperature from the rhombohedral to the cubic phase is shown to decrease with Sr doping. This structural change is attributed to the increase in the tolerance factor by substitution of Ca²⁺ with Sr²⁺. Furthermore, we relate these changes in structure to the charge-ordering transition temperature as indicated by the temperature dependent resistivity measurements, which confirmed the presence of an electronic phase transition ranging from 385 K < T* < 426 K for the Ca_1−xSr_xMn₇O₁₂ films. An increase in Sr doping leads to a decrease in T* but does not significantly alter electronic conduction on either side of the transition. Our results suggest structural control by chemical doping as a route to tune T* in AMn₇O₁₂ compounds, an important parameter in future device structures based on control of phase transitions above room temperature.

We thank Boris Yakshinkiy for RBS measurements at the Rutgers University Laboratory for Surface Modification. A.H. was supported by the U.S. Department of Energy (DOE), Office of Science (OS), Office of Workforce Development for Teachers and Scientists, Office of Science Graduate Student Research (SCGSR) program. The SCGSR program is administered by the Oak Ridge Institute for Science and Education for the DOE under Contract No. DE-SC0014664. S.J.M. was supported by the Army Research Office (No. W911NF-15-1-0133). The work at ORNL was supported by U.S. DOE, OS, Basic Energy Sciences, Materials Sciences and Engineering Division (high temperature XRD characterization) and the Scientific User Facilities Division.