Dielectric and Spin-Glass Magnetic Properties of the A-Site Columnar-Ordered Quadruple Perovskite Sm2CuMn(MnTi3)O12

Perovskite-type ABO3 oxides show a number of cation-ordered structures, which have significant effects on their properties. The rock-salt-type order is dominant for B cations, and the layered order for A cations. In this work, we prepared a new perovskite-type oxide, Sm2CuMn(MnTi3)O12, with a rare columnar A-site order using a high-pressure, high-temperature method at about 6 GPa and about 1700 K. Its crystal structure was studied with synchrotron powder X-ray diffraction. The compound crystallizes in space group P42/nmc (No. 137) at room temperature with a = 7.53477 Å and c = 7.69788 Å. The magnetic properties of the compound were studied with dc and ac magnetic susceptibility measurements and specific heat. Spin-glass (SG) magnetic properties were found with TSG = 7 K, while specific heat, in the form of Cp/T, showed a strong, very broad anomaly developing below 20 K and peaking at 4 K. The dielectric constant of Sm2CuMn(MnTi3)O12 was nearly frequency and temperature independent between 8 K and 200 K, with a value of about 50. Cu2+ doping drastically modified the magnetic and dielectric properties of Sm2CuMn(MnTi3)O12 in comparison with the parent compound Sm2MnMn(MnTi3)O12, which showed a long-range ferrimagnetic order at 34–40 K. The antisite disorder of Cu2+ and Mn2+ cations between square-planar and octahedral sites was responsible for the SG magnetic properties of Sm2CuMn(MnTi3)O12.


Introduction
The properties of the perovskite-structure oxide material, ABO 3 , are controlled by their chemical compositions and degrees of cation orderings [1,2]. There are perovskites with B-site cation orderings, A-site cation orderings, and both types of orderings. In the case of B-site ordering, the rock-salt-type order is dominant [3]. In the case of A-site ordering in ABO 3 , the layered-type order is dominant [1,2,4], but there are other types of ordering [5,6]. There are also two special families of perovskites with A-site orderings: A-site-ordered quadruple perovskites, AA 3 B 4 O 12 [7][8][9], and A-site columnar-ordered quadruple perovskites, A 2 A A B 4 O 12 [10]. Quadruple perovskites can have ordered arrangements of 3d transition metals at the A (in general) perovskite sites in addition to the B sites. The resulting B-B, A-B, and A-A exchange interactions can produce complex interaction patterns and frustration networks and result in competing magnetic ground states, a large number of magnetic transitions and unexpected magnetism [11].
With A = R = rare earth elements and Bi and A = A = B = Mn, interesting classes of perovskite manganites are formed, namely RMn 7 O 12 [9] and RMn 3 O 6 (in a short formula) [12]. They show several magnetic transitions with spin reorientations [9], and some

Experimental
Sm 2 CuMn(MnTi 3 )O 12 was prepared using a high-pressure, high-temperature method using a belt-type high-pressure machine at 6 GPa and about 1700 K for 2 h in a Pt capsule. After annealing at 1700 K, the samples were quenched to room temperature (RT) by turning off the heating current, and the pressure was slowly released. Stoichiometric amounts of Sm 2 O 3 (99.9%), CuO (99.9%), MnO (99.99%), and TiO 2 (99.9%) were used as an initial oxide mixture with the 1:1:2:3 ratio, respectively. Commercial Sm 2 O 3 , CuO, and TiO 2 chemicals were used. A single-phase MnO oxide was prepared from a commercial MnO 2 chemical by annealing at 1273 K for 4 h in a 20% H 2 + 80% Ar gas flow.
X-ray powder diffraction (XRPD) data were collected at RT with a RIGAKU Mini-Flex600 diffractometer (CuKα radiation; a 2θ range of 8-100 • ; a step of 0.02 • , and scan speed of 1 • /min). The synchrotron XRPD data were collected at RT on the BL15XU beamline (the former NIMS beamline) of SPring-8 [26] between 2.04 • and 60.23 • at 0.003 • intervals in 2θ with a wavelength of λ = 0.65298 Å. The data between 6 • and 60.23 • were used in the refinements as no reflections were observed and expected below 6 • . The sample was inserted into a Lindemann glass capillary tube (inner diameter: 0.1 mm), which was rotated during the measurements. The Rietveld analysis of all XRPD data was performed using the RIETAN-2000 program [27].
Scanning electron microscopy (SEM) images and energy-dispersive X-ray (EDX) spectra were obtained on a Hitachi Miniscope TM3000 (operating at 15 kV). SQUID magnetometers (Quantum Design, MPMS-XL-7T and MPMS3) were used for the magnetic measurements. Temperature dependence was measured between 2 and 400 K in applied fields of 100 Oe and 10 kOe under both zero-field-cooled (ZFC) and field-cooled on cooling (FCC) conditions on an MPMS-XL-7T. Magnetic-field dependence was measured at T = 2 K and 5 K between −70 and 70 kOe on MPMS3. Frequency dependent alternating current (ac) susceptibility measurements were performed on cooling with a Quantum Design MPMS3 instrument at different frequencies (f ), different applied oscillating magnetic fields (H ac ), and different static dc field (H dc ). Relaxation curves were measured on MPMS3 using the following procedure: the sample was cooled down from 50 K to a measurement temperature at zero magnetic field, then a magnetic field of 100 Oe was applied, and magnetization was measured (as one scan within 2 s) as a function of time every 5 s.
Specific heat, C p , was measured by cooling from 270 K to 2 K at zero magnetic field and from 150 K to 2 K at magnetic field of 90 kOe by a pulse relaxation method using a commercial calorimeter (Quantum Design PPMS).
The dielectric constant and dielectric loss were measured on a NOVOCONTROL Alpha-A High Performance Frequency Analyzer in a frequency range from 100 Hz to 665 kHz in a temperature range from 8 K to 330 K (on heating) at zero magnetic field.

Results and Discussion
The as-synthesized Sm 2 CuMn(MnTi 3 )O 12 contained a small amount of CuO impurity. In addition, the synchrotron XRPD pattern showed the presence of Pt impurity. However, Pt appeared from Pt capsules used in the synthesis and can be considered as an extrinsic impurity. The presence of a CuO impurity suggests that the main phase should be slightly Cu-deficient in comparison with the target composition. The morphology of the sample is shown in Figure 1 in the refinements as no reflections were observed and expected below 6. The sample was inserted into a Lindemann glass capillary tube (inner diameter: 0.1 mm), which was rotated during the measurements. The Rietveld analysis of all XRPD data was performed using the RIETAN-2000 program [27].
Scanning electron microscopy (SEM) images and energy-dispersive X-ray (EDX) spectra were obtained on a Hitachi Miniscope TM3000 (operating at 15 kV).
SQUID magnetometers (Quantum Design, MPMS-XL-7T and MPMS3) were used for the magnetic measurements. Temperature dependence was measured between 2 and 400 K in applied fields of 100 Oe and 10 kOe under both zero-field-cooled (ZFC) and fieldcooled on cooling (FCC) conditions on an MPMS-XL-7T. Magnetic-field dependence was measured at T = 2 K and 5 K between −70 and 70 kOe on MPMS3. Frequency dependent alternating current (ac) susceptibility measurements were performed on cooling with a Quantum Design MPMS3 instrument at different frequencies (f), different applied oscillating magnetic fields (Hac), and different static dc field (Hdc). Relaxation curves were measured on MPMS3 using the following procedure: the sample was cooled down from 50 K to a measurement temperature at zero magnetic field, then a magnetic field of 100 Oe was applied, and magnetization was measured (as one scan within 2 s) as a function of time every 5 s.
Specific heat, Cp, was measured by cooling from 270 K to 2 K at zero magnetic field and from 150 K to 2 K at magnetic field of 90 kOe by a pulse relaxation method using a commercial calorimeter (Quantum Design PPMS).
The dielectric constant and dielectric loss were measured on a NOVOCONTROL Alpha-A High Performance Frequency Analyzer in a frequency range from 100 Hz to 665 kHz in a temperature range from 8 K to 330 K (on heating) at zero magnetic field.

Results and Discussion
The as-synthesized Sm2CuMn(MnTi3)O12 contained a small amount of CuO impurity. In addition, the synchrotron XRPD pattern showed the presence of Pt impurity. However, Pt appeared from Pt capsules used in the synthesis and can be considered as an extrinsic impurity. The presence of a CuO impurity suggests that the main phase should be slightly Cu-deficient in comparison with the target composition. The morphology of the sample is shown in Figure 1  All of the reflections on the laboratory and synchrotron XRPD patterns (except CuO and Pt) could be indexed in a tetragonal system in space group P42/nmc (No. 137) ( Figure  2). Sm2CuMnMnTi3O12 was found to crystallize in the parent structure of the A-site All of the reflections on the laboratory and synchrotron XRPD patterns (except CuO and Pt) could be indexed in a tetragonal system in space group P4 2 /nmc (No. 137) ( Figure 2). Sm 2 CuMnMnTi 3 O 12 was found to crystallize in the parent structure of the A-site columnarordered quadruple perovskites, A 2 A A B 4 O 12 [10]. Therefore, the structural data for the parent compound Sm 2 MnMn(MnTi 3 )O 12 [22,23] were taken as an initial starting model. columnar-ordered quadruple perovskites, A2A′A″B4O12 [10]. Therefore, the structural data for the parent compound Sm2MnMn(MnTi3)O12 [22,23] were taken as an initial starting model. In the structural analysis, we first assumed ideal cation distributions (that is, Sm at the A site, Cu at the square-planar A′ site, Mn at the tetrahedral A″ site, and 0.75Ti + 0.25Mn at the octahedral B site) and refined the occupation factors (g) together with all of the other structural and nonstructural parameters (except g(B): one cation occupation factor should always be fixed to avoid significant correlations among the refined g parameters). In addition, in the structural analysis, we always assumed that Ti 4+ cations were located at the B site, as Ti 4+ cations strongly prefer octahedral sites [28].
The refined g values were as follows: g(Sm-A) = 0.9167 (16), g(Cu-A′) = 0.914 (7), and g(Mn-A″) = 1.047 (7). These values suggest that the ideal cation distribution was not realized, and there were some antisite disorders. The g(Cu-A′) value suggested that this site should contain lighter elements that could only be Mn (with the above assumption on Ti). When only Mn was placed at the square-planar A′ site, the occupation factor was g(Mn-A′) = 1.112 (8), meaning that heavier elements should also be at this site. Because it was difficult to precisely refine the distribution of Mn and Cu with X-ray diffraction, we introduced a virtual atom: MC = 0.5Mn + 0.5Cu. The precise distribution of Mn and Cu could only be determined with neutron diffraction. The disordering of cations at the Cu site was also observed as in many cases of such perovskites [21,22,25,29].
The refined structural parameters and primary bond lengths and angles in Sm2CuMn(MnTi3)O12 are listed in Tables 1 and 2. The experimental, calculated, and difference synchrotron patterns are shown in Figure 3. The crystal structure of Sm2CuMn(MnTi3)O12 is illustrated in the inset of  In the structural analysis, we first assumed ideal cation distributions (that is, Sm at the A site, Cu at the square-planar A site, Mn at the tetrahedral A site, and 0.75Ti + 0.25Mn at the octahedral B site) and refined the occupation factors (g) together with all of the other structural and nonstructural parameters (except g(B): one cation occupation factor should always be fixed to avoid significant correlations among the refined g parameters).
In addition, in the structural analysis, we always assumed that Ti 4+ cations were located at the B site, as Ti 4+ cations strongly prefer octahedral sites [28].
The refined g values were as follows: g(Sm-A) = 0.9167 (16), g(Cu-A ) = 0.914 (7), and g(Mn-A ) = 1.047 (7). These values suggest that the ideal cation distribution was not realized, and there were some antisite disorders. The g(Cu-A ) value suggested that this site should contain lighter elements that could only be Mn (with the above assumption on Ti). When only Mn was placed at the square-planar A site, the occupation factor was g(Mn-A ) = 1.112 (8), meaning that heavier elements should also be at this site. Because it was difficult to precisely refine the distribution of Mn and Cu with X-ray diffraction, we introduced a virtual atom: MC = 0.5Mn + 0.5Cu. The precise distribution of Mn and Cu could only be determined with neutron diffraction. The disordering of cations at the Cu site was also observed as in many cases of such perovskites [21,22,25,29].
The refined structural parameters and primary bond lengths and angles in Sm 2 CuMn(MnTi 3 )O 12 are listed in Tables 1 and 2. The experimental, calculated, and difference synchrotron patterns are shown in Figure 3. The crystal structure of Sm 2 CuMn(MnTi 3 )O 12 is illustrated in the inset of Figure 3.
144.17 (9) Materials 2022, 15, x FOR PEER REVIEW 5 of 15 1.25 (7) Figure 4. There was a divergence between the 100 Oe ZFC and FCC curves at 7 K and a relatively sharp maximum on the 100 Oe ZFC curve at 7 K. A divergence between the ZFC and FCC curves almost disappeared under 10 kOe. These features are typical for spin-glass transitions [30][31][32]. Isothermal magnetization, M versus H, curves demonstrated an extended S-type shape with very weak and narrow hysteresis ( Figure 5). Almost no hysteresis was observed at 5 K because 5 K was close to its T SG = 7 K; on the other hand, the hysteresis was noticeably wider at a lower temperature of 2 K. Such M versus H curves are also typical for spin glasses [30][31][32].      The inverse magnetic susceptibilities (χ −1 versus T) followed the Curie-Weiss law at high temperatures ( Figure 4). To obtain the effective magnetic moment and the Curie-Weiss temperature, we performed fits between 250 and 345 K using the 10 kOe FCC curves (the fit and fitting parameters are summarized on Figure 4). The experimental effective magnetic moment was close to the expected one (8.803 µ B ; in the calculations we used 1.5 µ B for Sm 3+ [33]). The negative Curie-Weiss temperature shows that the main magnetic interactions were antiferromagnetic in nature. The ratio between the Curie-Weiss temperature (−81.5 K) and T SG (the so-called frustration ratio) was about 11, indicating a strong degree of magnetic frustration. We note that CuO was in an antiferromagnet with transition temperatures of 213 K and 230 K. Therefore, CuO impurity should not affect the reported magnetic properties at low temperatures.
To confirm the spin-glass nature of the sample, we measured ac magnetic susceptibility curves (Figures 6 and 7). We note that no dependence of the χ and χ values on the applied H ac field was observed (inset of Figure 6). We indeed observed typical features of spinglasses: peak positions were frequency-dependent and shifted to higher temperatures with increasing frequency; in addition, peak intensity was suppressed on the χ versus T curves and enhanced on the χ versus T curves with increasing frequency. All of these features are typical for spin glasses [30][31][32]. In addition, the shape of the χ versus T and the χ versus T curves was also typical for spin glasses. The criterion, which quantifies the relative change of the spin-glass temperature per frequency decade and is defined as ∆T SG /[T SG ∆log(f )], was about 0.023 for Sm 2 CuMn(MnTi 3 )O 12 (with T SG = 7.2 K at f = 2 Hz and T SG = 7.6 K at f = 500 Hz). This value is often observed in different spin-glass materials [30][31][32].
Sm 2 CuMn(MnTi 3 )O 12 shows time-dependent magnetic properties below T SG , namely magnetization relaxation (Figure 8). Above T SG , no noticeable relaxation of magnetization was detected. Time-dependent magnetic properties, such as relaxation, are typical features of spin-glass systems. Relaxation below T SG was fitted by the stretched exponential function, [30], and the resultant parameters are listed on Table 3. The most important parameter is the mean relaxation time, t r , and it decreases monotonically with increasing temperature.  (4)  6 12.34 (14) 13.01(9) 554(6) 0.4654 (4) The fitting equation is The specific heat data showed a noticeable magnetic contribution to the total specific heat below about 20 K, where it could be clearly seen as a rise in C p /T values below 20 K (Figure 9). No λ-type anomaly was detected in the C p versus T curve (inset of Figure 9, a green curve). Instead, a broad anomaly was seen in the C p versus T curve, which gave a broad peak centered at 4 K in the C p /T versus T curve. Therefore, specific heat measurements confirmed the absence of long-range magnetic ordering. A magnetic field of 90 kOe slightly suppressed the peak near 4 K and moved the magnetic entropy into the 14-40 K range. Sm2CuMn(MnTi3)O12 shows time-dependent magnetic properties below TSG, namely magnetization relaxation (Figure 8). Above TSG, no noticeable relaxation of magnetization was detected. Time-dependent magnetic properties, such as relaxation, are typical features of spin-glass systems. Relaxation below TSG was fitted by the stretched exponential function, [30], and the resultant parameters are listed on Table  3. The most important parameter is the mean relaxation time, tr, and it decreases monotonically with increasing temperature. The temperature dependence of the dielectric constant and dielectric loss is shown in Figure 10. The dielectric constant was nearly temperature and frequency-independent between 8 and 200 K. Above about 200 K, a sharp rise in the dielectric constant was observed, where the magnitude of the rise depended on frequency. This behavior typically originates from the Maxwell-Wagner contribution due to increased conductivity. No broad anomalies were observed in Sm 2 CuMn(MnTi 3 )O 12 in comparison with the parent compound Sm 2 MnMn(MnTi 3 )O 12 . This fact shows that Cu 2+ doping drastically modified the dielectric properties as well, in addition to the magnetic properties. We note that Pt impurity was only observed in a powder sample, which could contain parts from the surface. The surfaces of a pellet used for dielectric measurements were polished. Therefore, Pt impurity should not present in a pellet and affect dielectric measurements.  Table 3. 12.34 (14) 13.01(9) 554(6) 0.4654 (4) The The specific heat data showed a noticeable magnetic contribution to the total specific heat below about 20 K, where it could be clearly seen as a rise in Cp/T values below 20 K (Figure 9). No λ-type anomaly was detected in the Cp versus T curve (inset of Figure 9, a green curve). Instead, a broad anomaly was seen in the Cp versus T curve, which gave a broad peak centered at 4 K in the Cp/T versus T curve. Therefore, specific heat measurements confirmed the absence of long-range magnetic ordering. A magnetic field of 90 kOe slightly suppressed the peak near 4 K and moved the magnetic entropy into the 14-40 K range.  Table 3. The temperature dependence of the dielectric constant and dielectric loss is shown in Figure 10. The dielectric constant was nearly temperature and frequency-independent between 8 and 200 K. Above about 200 K, a sharp rise in the dielectric constant was observed, where the magnitude of the rise depended on frequency. This behavior typically origi- Spin-glass magnetic properties were also observed in Sm2MnZn(MnTi3)O12 at TSG = 6.5 K, with a significant antisite disorder [34]. This fact shows that antisite structural disorder should play a major role in the modification of magnetic properties of the parent Sm2MnMn(MnTi3)O12 compound, not the nature of dopant cations (magnetic as Cu 2+ or non-magnetic as Zn 2+ ). Both Sm2CuMn(MnTi3)O12 and Sm2MnZn(MnTi3)O12 demonstrated similar low-temperature specific heat features (Figure 9b).
The beneficial effects of Cu 2+ doping in RMn7O12 [13,14], RMn3O6 [11,12,15], and Y2MnGaMn4O12 [20,21] originate from the fact that Cu 2+ doping is aliovalent doping, which produces Mn 4+ cations. A mixture of Mn 3+ and Mn 4+ at the B sites of perovskites significantly enhanced the exchange interactions and magnetic transition temperatures. On the other hand, Cu 2+ doping in the parent Sm2MnMn(MnTi3)O12 compound was isovalent doping. Such doping did not change the oxidation state of Mn, while the antisite disordering "degraded" the magnetic properties.

Conclusions
A new member of the A-site columnar-ordered quadruple perovskite family, Sm2CuMn(MnTi3)O12, was prepared using a high-pressure, high-temperature method. Cu 2+ doping significantly modified the properties of the parent Sm2MnMn(MnTi3)O12 compound, as spin-glass magnetic properties at TSG = 7 K were observed in Sm2CuMn(MnTi3)O12 in comparison with the long-range ferrimagnetic order at TC = 34-40 K in Sm2MnMn(MnTi3)O12. In addition, relaxor-like dielectric properties of Sm2MnMn(MnTi3)O12 disappeared in Sm2CuMn(MnTi3)O12, which showed a nearly temperature and frequency-independent dielectric constant between 8 and 200 K with a value of about 50.   Spin-glass magnetic properties were also observed in Sm 2 MnZn(MnTi 3 )O 12 at T SG = 6.5 K, with a significant antisite disorder [34]. This fact shows that antisite structural disorder should play a major role in the modification of magnetic properties of the parent Sm 2 MnMn(MnTi 3 )O 12 compound, not the nature of dopant cations (magnetic as Cu 2+ or non-magnetic as Zn 2+ ). Both Sm 2 CuMn(MnTi 3 )O 12 and Sm 2 MnZn(MnTi 3 )O 12 demonstrated similar low-temperature specific heat features (Figure 9b).
The beneficial effects of Cu 2+ doping in RMn 7 O 12 [13,14], RMn 3 O 6 [11,12,15], and Y 2 MnGaMn 4 O 12 [20,21] originate from the fact that Cu 2+ doping is aliovalent doping, which produces Mn 4+ cations. A mixture of Mn 3+ and Mn 4+ at the B sites of perovskites significantly enhanced the exchange interactions and magnetic transition temperatures. On the other hand, Cu 2+ doping in the parent Sm 2 MnMn(MnTi 3 )O 12 compound was isovalent doping. Such doping did not change the oxidation state of Mn, while the antisite disordering "degraded" the magnetic properties.

Conclusions
A new member of the A-site columnar-ordered quadruple perovskite family, Sm 2 CuMn(MnTi 3 )O 12 , was prepared using a high-pressure, high-temperature method. Cu 2+ doping significantly modified the properties of the parent Sm 2 MnMn(MnTi 3 )O 12 compound, as spin-glass magnetic properties at T SG = 7 K were observed in Sm 2 CuMn(MnTi 3