Spin Frustration in Double Perovskite Oxides and Oxynitrides: Enhanced Frustration in La₂MnTaO₅N with a Large Octahedral Rotation

Abstract

The B-site sublattice in the double perovskite oxides A₂BB′O₆ (B: magnetic cation; B′: nonmagnetic cation) causes spin frustration, but the relationship between the structure and spin frustration remains unclear although a number of compounds have been studied. The present study systematically investigated A₂Mn^IIB′O₆ (S = 5/2) and found that the frustration factor, defined by $f=\left|{\theta }_W\right|/T_N$ (${\theta }_W$: Weiss temperature; $T_N$: Néel temperature), scales linearly with the tolerance factor t, i.e., octahedral rotation. Unexpectedly, La₂MnTaO₅N (space group: P2₁/n) synthesized under high pressure is more frustrated (f = 6) than oxides with similar t values, despite the large octahedral rotation due to the small t value of 0.914. Structural analysis suggests that the enhanced frustration can be attributed to the site preference of nitride anions at the equatorial positions, which reduces the variance of neighboring Mn–Mn distances. Our findings provide a new guide to control and improve spin frustration in double perovskites with multiple anions.

Graphical Abstract:

1. INTRODUCTION

Geometrical spin frustration usually occurs in compounds where concentrated magnetic ion arrangement involves equilateral triangular motifs.¹ Magnetic properties of such frustrated materials have been studied intensively in search of novel ground states including the spin liquid state² and noncollinear spin order.³ These nonclassical states are mainly derived from antiferromagnets with magnetic cations on two-dimensional (2D) triangular lattices,⁴ 2D kagomélattices (e.g., volborthite,⁵ herbertsmithite⁶), and 3D pyrochlore lattices.^7–13

In a cubic ABO₃ perovskite, the B-site cube lattice is not geometrically frustrated, but frustration can be introduced by allowing cation order, where B and B′ in A₂BB′O₆ (or A₂B′BO₆) are magnetic and nonmagnetic cations, respectively. The resulting B-site sublattice in the double perovskite forms a face-centered cubic (fcc) lattice having edge-shared tetrahedral (Figure 1). For example, Sr₂CaReO₆ and Sr₂MgReO₆ (B = Re^VI; S = 1/2) exhibit a divergence between field-cooled and zero-field-cooled susceptibilities, indicative of a spin glass state.^14,15 A collective spin-singlet state is observed for S = 1/2 Ba₂YMoO₆, La₂LiMoO₆ (B = Mo^V),¹⁶ and Ba₂NaOsO₆ (B = Os^VII)¹⁷ and S = 1 Ba₂YReO₆ and La₂LiReO₆ (B = Re^V),¹⁸ whereas S = 3/2 Ba₂YRuO₆ and La₂LiRuO₆ (B = Ru^V)¹⁹ and S = 5/2 A₂MnB′O₆ (B = Mn^II)^20,21 undergo long-range order (LRO) at temperatures lower than their Weiss temperatures. An oxynitride double perovskite Sr₂FeWO₅N (B = Fe^III; S = 5/2) with a Neél temperature (T_N) of 13 K, prepared by heating Sr₂FeWO₆ under NH₃ gas flow, has been recently added.²² Despite the large number of examples of double perovskites, the relationship between structure, e.g., structural distortion often discussed in simple perovskites,²³ and frustration is rarely investigated in the literature²⁴ and is still unclear.

Figure 1.

(a) Ideal cubit B-site ordered double perovskite A₂BB′O₆, in which magnetic B (purple) and nonmagnetic B′ (white) cations in octahedral coordination are ordered in a rock salt manner. The A-site atoms are omitted. (b) Edge-shared network of tetrahedra composed of magnetic B cations. The solid lines denote the unit cell.

In this paper, we focus on Mn-based double perovskites and oxynitrides (A₂Mn^IIB′O₆, A₂Mn^IIB′O₅N) to understand the structure–property relationship comprehensively. We found that in A₂Mn^IIB′O₆ there is an intimate correlation between the tolerance factor²⁵ (t) and frustration factor (f): f linearly decreases as t decreases.²⁶ However, in our newly synthesized oxynitride La₂MnTaO₅N, f turns out to be large in spite of the large octahedral rotation due to the small t. The structural analysis suggests that the frustration is enhanced by anion preferential occupation that suppresses the deviation of the Mn–Mn distances and hence reduces t.

2. EXPERIMENTAL SECTION

A polycrystalline sample of La₂MnTaO₅N was prepared by a high-pressure reaction using preheated La₂O₃ (99.999%, rare metallic), MnO (99.9%, rare metallic), and TaON. A phase-pure TaON was prepared by heating Ta₂O₅ (99.99%, rare metallic) at 850 °C for 15 h in flowing dry ammonia gas at a rate of 20 mL/min.²⁷ The starting reagents were intimately mixed in a nitrogen filled glovebox (O₂, H₂O < 0.1 ppm), pelletized, charged into a platinum capsule, and inserted into a pyrophyllite cell. A cubic anvil press was used to pressurize the sample to 1.5 GPa. The sample was then heated to a target temperature of 1400 °C in 14 min and stabilized for 2 h. After these procedures, it was quenched to room temperature within 7 min, followed by a slow release of pressure. As a reference for heat capacity, a nonmagnetic counterpart La₂ZnTaO₅N was prepared in a similar condition (7 GPa, 1200 °C) but using ZnO (99.999%, rare metallic) instead of MnO. We also prepared two Nd-substituted samples of LaNdMnTaO₅N and Nd₂MnTaO₅N at 7 GPa and 1400 °C using Nd₂O₃ (99.999%, rare metallic).

We checked the phase purity of all the samples obtained by powder X-ray diffraction (XRD) experiments at room temperature on a Bruker D8 Advance diffractometer equipped with a Cu Kα source. Energy dispersive X-ray spectroscopy (EDX) was conducted using an Oxford X-act detector mounted on a Hitachi S-3400N scanning electron microscope (SEM). For structural refinement, synchrotron XRD (SXRD) data were collected at SPring-8 (BL02B2) using a wavelength (λ) of 0.420731 Å. A large contrast in neutron scattering lengths between N (9.36 fm) and O (5.81 fm) allowed us to acquire N/O distributions at the anionic sites: neutron diffraction (ND) data was collected at room temperature using the high-resolution powder diffractometer (BT1) at the National Institute of Standards and Technology using a wavelength of 1.5406 Å. Structural refinements for the synchrotron and neutron data were carried out using JANA2006²⁸ by giving the composition as determined by the combustion analysis method. The nitride content of the specimen was determined by combustion analysis (Yanaco CHN Corder).

DC magnetic susceptibility of Ln₂MnTaO₅N (Ln₂ = La₂, LaNd, and Nd₂) was measured using a superconducting quantum interference device (SQUID; MPMS-XL, Quantum Design) in a temperature range of 2–350 K at a constant field of 0.1 T. To extract information on the magnetic order of La₂MnTaO₅N, we performed a time-of-flight (TOF) neutron diffraction study at 1.5 and 10 K using the GEM diffractometer at the ISIS neutron facility. Specific heat experiments were conducted using a commercial calorimeter (PPMS, Quantum Design) in the temperature range between 2 and 100 K.

3. RESULTS AND DISCUSSION

The XRD pattern (Figure S1) of the sample prepared in a stoichiometric condition (La₂O₃/MnO/TaON = 1:1:1) exhibited a monoclinic phase, along with an impurity phase of La₃TaO₇ (8 wt %). It is noticed that the color of the pellet surface (white gray) is different from the inner one (black gray). The presence of MnO was identified through powder XRD and SEM-EDX on the surface region. When excess MnO was used, the amount of La₃TaO₇ impurity decreased. A 20 mol % excess yielded the highest quality sample without impurity (Figure S1). The cation composition of the inner part of the sample examined with EDX gave an average cation ratio of La/Mn/Ta = 2.06:1:1.02, close to the target composition of La₂MnTaO₅N (Table S1).

La₂MnTaO₅N was found to have a $\surd 2a_{\text{p}}\times \surd 2b_{\text{p}}\times 2c_{\text{p}}$ cell, where $a_{\text{p}}\times b_{\text{p}}\times c_{\text{p}}$ represents the pseudocubic primitive cell of perovskite. The determined cell parameters are a = 5.6939(3) Å, b = 5.7924(2) Å, c = 8.0956(4) Å, and β = 89.976(12)°, similar to those of a double perovskite oxynitride La₂MgTaO₅N (a = 5.6678 Å, b = 5.6618 Å, c = 8.0639 Å, and β = 89.476°) with a P2₁/n space group (No. 14).²⁹ Accordingly, the La₂MgTaO₅N structure was employed as a structural model of La₂MnTa’O’₆ for Rietveld refinement of SXRD data (Figure 2a), where nitrogen was not taken into consideration owing to little difference in X-ray atomic scattering factors between O and N. The structural refinement was converged with adequate agreement parameters of R_wp = 7.37% and GOF = 3.78 (Figure 2a and Table 1). A variation of the occupation factor of each atom or taking into account the antisite disorder between Mn and Ta did not improve the results.

Figure 2.

Rietveld refinement of (a) SXRD data and (b) ND data for La₂MnTaO₅N at room temperature. Red crosses, green and blue lines, and blue ticks represent observed, calculated, and difference intensities and the Bragg peak positions of La₂MnTaO₅N. For the neutron data, La₃TaO₇ was included as a secondary phase since the sample was prepared in a stoichiometric ratio (see the text for details).

Table 1.

Structural Parameters of La₂MnTaO₅N from the Rietveld Refinement of SXRD (Upper) and ND (Lower) at 295 K^a

atom	site	g	x	y	z	Uiso (100 Å2)
La	4e	1	0.9932(3)	0.04354(9)	0.24851(9)	0.709(16)
			0.9902(7)	0.0431(4)	0.2460(6)	0.90(7)
Mn	2c	1	0.5	0	0.5	0.29(5)
						3.4(3)
Ta	2d	1	0.5	0	0	0.778(16)
						1.69(13)
O1	4e	1	0.0898(15)	0.4696 (12)	0.2530(11)	1.4(2)
		0.93(2)	0.0896(6)	0.4741(5)	0.2587(8)	0.26(12)
N1	4e	0.07(2)	0.0896(6)	0.4741(5)	0.2587(8)	0.26(12)
O2	4e	1	0.6950(14)	0.2902(14)	0.0506(14)	0.5(3)
		0.77(2)	0.7002(9)	0.2884(8)	0.0468(8)	1.00(17)
N2	4e	0.23(2)	0.7002(9)	0.2884(8)	0.0468(8)	1.00(17)
O3	4e	1	0.2104(15)	0.1974(15)	0.9621(17)	1.3(3)
		0.77(2)	0.2039(10)	0.1967(9)	0.9561(9)	2.0(2)
N3	4e	0.23(2)	0.2039(10)	0.1967(9)	0.9561(9)	2.0(2)

^aError bars indicate one standard deviation in the final digit.

The combustion experiment (elemental analysis) showed that the nitrogen content is close to 1. In the presence of three crystallographically independent anionic (4e) sites, the neutron analysis is of particular importance to see how oxide and nitride anions are distributed over these sites. The neutron refinement of La₂MnTaO₅N was performed using the P2₁/n structure. La₃TaO₇ was included as a secondary phase because of the early stage of the research (Figure 2b). During the refinement, the anion occupancy was constrained to unity ($g_{\text{N}i}+g_{\text{O}i}=1$, where i = 1, 2, 3) and the isotropic thermal parameters of O and N were set to be equal. When nitrogen atoms were occupied only at the “apical” O1 site (see Figure 3), the occupancy factor of g_N1 was unreasonably small (0.10(2)). On the contrary, the substitution of nitrogen atoms at the “equatorial” O2 and O3 sites yielded g_N2 = 0.37(3) and g_N3 = 0.43(4). These analyses strongly suggest preferential occupation of nitrogen atoms at the equatorial sites. The final refinement was conducted with a constraint of g_N2 = g_N3, yielding g_N1 = 0.07(2) and g_N2 = g_N3 = 0.23(2) with agreement parameters of R_wp = 4.80% and GOF = 1.82. It is notable that these values gave a nearly stoichiometric anion composition of La₂MnTaO_4.92(14)N_1.08. The refined parameters of synchrotron and neutron data are listed in Table 1. A similar preferential occupation of nitride anions at the equatorial site (23% and 2% at the equatorial and apical site, respectively) is also found in Sr₂FeMoO₅N with the I4/m space group (a ∼ 5.6 Å, c ∼ 7.9 Å).³⁰

Figure 3.

(a) La₂MnTaO₅N with the B-site ordered double perovskite structure (P2₁/n). Green, purple, brown, red, and gray spheres represent La, Mn, Ta, O, and N, respectively. The solid lines represent the unit cell. (b) Octahedral environment around Mn and Ta. The occupancies at the anion sites are g_O1/g_N1 = 0.93/0.07 and g_O2/g_N2 = g_O3/g_N3 = 0.77/0.23. Note that N/O atoms are disordered.

As a nonmagnetic reference for specific heat measurements, La₂ZnTaO₅N was synthesized under similar conditions. As shown in Figure S2 and Table S2, SXRD peaks were indexed in the monoclinic cell of √2a_p × √2b_p × 2c_p, as in the case of La₂MnTaO₅N. The Rietveld refinement was performed using the structural model of La₂MnTaO₅N obtained from the neutron diffraction data, which converged to the agreement parameters of R_wp = 4.36% and GOF = 2.09. The smaller cell parameters (a = 5.6536(3) Å, b = 5.7190(2) Å, c = 8.0244(4) Å, β = 89.734(4)°) than those of La₂MnTaO₅N are reasonable given the smaller ionic radius of Zn^II(0.74 Å) compared with Mn^II (0.83 Å). LaNdMnTaO₅N and Nd₂MnTaO₅N also have a monoclinic $\surd 2a_{\text{p}}\times \surd 2b_{\text{p}}\times 2c_{\text{p}}$ cell with smaller lattice parameters due to the smaller ionic radius of Nd^III (vs La^III): a = 5.629(1) Å, b = 5.795(1) Å, c = 8.029(3) Å, and β = 89.883(1)° for LaNdMnTaO₅N and a = 5.567(2) Å, b = 5.862(1) Å, c = 7.846(3) Å, and β = 89.693(1)° for Nd₂MnTaO₅N (Figure S3).

The DC magnetic susceptibility of La₂MnTaO₅N measured at 0.1 T is shown in Figure 4a. The reciprocal susceptibility in the temperature range of 170–300 K was fitted with the Curie–Weiss law, $\chi =C/\left(T–{\theta }_W\right)+{\chi }_0$, where $C$, ${\theta }_{\text{W}}$, and ${\chi }_0$ denote the Curie constant, the Weiss temperature, and a constant term. The Curie constant of 4.37(9) emu K/mol agrees well with the theoretical value of 4.375 emu K/mol for a high spin Mn^II ion (S = 5/2). This result also justifies the La₂MnTaO₅N stoichiometry. The negative value of ${\theta }_{\text{W}}$, −29(3) K, indicates the presence of antiferromagnetic interactions between adjacent Mn moments (i.e., the next-nearest neighbor B sites). As the temperature decreases, deviations from the Curie–Weiss formula are appreciated, which may be related to the development of spin–spin correlations. Upon further cooling, a kink appears around 4.5 K. The absence of any deviation between field cooling (FC) and zero-field cooling (ZFC) runs suggest that this anomaly is a transition to the antiferromagnetic state, rather than a spin glass state. The Neél temperature at T_N = 4.5 K is considerably smaller than the Weiss temperature, suggesting the presence of sizable magnetic frustration in this material with a frustration factor of $\left|{\theta }_W\right|/T_N=6$.

Figure 4.

(a) Inverse susceptibility (red) and the Curie–Weiss fit (black). The fitting gave C = 4.37(9) emu K/mol, θ_W = −29(3) K, and χ₀ = 3.7(2) × 10⁻³ emu/mol. (b) Magnetic susceptibility of La₂MnTaO₅N for 2–20 K (see Figure S4 for the full plot). Red and white symbols, respectively, represent data under ZFC and FC.

In order to confirm the existence of LRO, we conducted a neutron diffraction measurement. Figure 5 compares neutron diffraction patterns between 10 and 1.5 K in the low Q region. While no extra reflection was observed in the 10 K profile, the 1.5 K data has a peak centered at Q = 0.66(3) Å⁻¹ corresponding to (0 1/2 1/2). This propagation vector is generally observed in double perovskite oxides (Ca₂MnWO₆ and Ca₂CrSbO₆, for example)^31,32 and more recently in Sr₂FeWO₅N,²² where alternating (011) ferromagnetic planes couple antiparallely to form an A type antiferromagnet (see Figure S5). However, the magnetic moment estimated from the area of the (0 1/2 1/2) peaks is only 40% (∼2 μ_B) of the expected value from Mn^II (d⁵, high spin), implying the presence of the fluctuations is due to spin frustration. The small moment value may also be related to the measured temperature not being so far from the transition temperature. This magnetic peak is broad with a correlation length (ξ) of 46(1) Å estimated from the full width at half-maximum (fwhm).

Figure 5.

TOF neutron diffraction patterns at 1.5 K (red) and 10 K (blue) and residual curve (green). The black fitting curve is the Gauss function, and the center value Q = 0.66(3) Å⁻¹ corresponds to (0 1/2 1/2). The full-length plots are shown in Figure S6.

The results of the heat capacity measurement are shown in Figure 6a. Although broad, the heat capacity has a peak around 4.5 K, which is almost the same as the anomaly seen in the magnetic susceptibility. The magnetic component, C_mag, of the specific heat was estimated by subtracting the lattice contribution, for which we used the specific heat of a nonmagnetic La₂ZnTaO₅N from the total specific heat C_p of La₂MnTaO₅N. The magnetic entropy, S_mag, of La₂MnTaO₅N was calculated by integrating C_mag/T. Figure 6b shows that S_mag continues to increase and reaches 11.5 J K⁻¹ mol⁻¹ at 100 K. This corresponds to about 80% of the theoretical value of Rln6 (= 14.90 JK⁻¹ mol⁻¹; R is the gas constant), which means that a spin–spin correlation develops far beyond T_N.

Figure 6.

(a) Temperature dependence of specific heat C_p between 2 and 20 K for La₂MnTaO₅N (red) and La₂ZnTaO₅N (black). See also Figure S7. (b) Magnetic entropy S_mag for La₂MnTaO₅N. The broken line represents the theoretical value of 14.90 J K⁻¹ mol⁻¹ for Mn^II (S = 5/2). (c) Temperature dependence of specific heat C_p, in logarithmic scale for clarity, between 2 and 20 K for La₂MnTaO₅N (red), LaNdMnTaO₅N (green), and Nd₂MnTaO₅N (blue). The triangular marks indicate the temperature at maximum C_p (corresponding to T_N).

Although there have been a number of reports on Mn-based double perovskite oxides (A₂Mn^IIB′O₆ (S = 5/2) where B′ is a nonmagnetic cation) as moderately frustrated compounds,^{24,31,33–38} the frustration factor is varied (f = 4–8), and the origin of this variation is still unknown. Note that the A-site order types will not be considered here. It is reasonable to discuss various physical properties in connection with octahedral rotation or the tolerance factor t, as has been done for many perovskites and double perovskites.³⁹ As shown in Figure 7a, t is linearly related to the rotation angle ${\phi }_{\text{av}}$ of the octahedron in A₂Mn^IIB′O₆ oxides. For simple perovskites ABO₃ (B = magnetic ions), it is known that frustration can be introduced by making the next-nearest-neighbor super-super-exchange interaction comparable to the nearest-neighbor superexchange interaction by reducing t (thus increasing octahedral rotation). The reduction in t, for example, leads to multiferroic phenomena in LnMnO₃^40–43 and helical magnetic structure in MnTaO₂N.⁴⁴ In marked contrast, in the double perovskite structure, the ideal frustrated fcc lattice consisting of only one type of nearest-neighbor interaction by magnetic Mn^II ions is formed when the cubic structure has no octahedral rotation. Thus, we can imagine that the case of t = 1 is most frustrated, and as t deviates from 1; i.e., the fcc sublattice is distorted, and magnetic interactions become inequivalent, which suppresses spin frustration. The relationship between the octahedral rotation and physical properties have been pointed out in double perovskite; for example, the band gap in Cs₂AgBiBr₆⁴⁵ and the superexchange interaction in A₂CrOsO₆ (A = Sr, Ca)⁴⁶ scale to t. However, it is rather surprising that the relation between octahedral rotation and magnetic frustration has not been discussed in double perovskites. Only a tendency for T_N to decrease as the average size of the A-site cation is increased has been reported.²⁴ In fact, as can be seen in Figure 7b, the tendency of the frustration factor to drop with decreasing t is remarkably observed in A₂Mn^IIB′O₆. To our surprise, such relation has not been discussed in double perovskite oxides despite intensive research in the past.

Figure 7.

Relationship between the tolerance factor t and other structural and physical parameters in B-site ordered double perovskite A₂Mn^IIB′O₆ (B′: nonmagnetic cation)^31,33–37 and La₂MnTaO₅N (red). (a) The t dependence of ${\phi }_{\text{av}}\left({\phi }_{\text{av}}=\left({\phi }_{\text{a}}+{\phi }_{\text{b}}+{\phi }_{\text{c}}\right)/3\right)$. ${\phi }_{\text{a}}$, ${\phi }_{\text{b}}$, and ${\phi }_{\text{c}}$ are octahedral tilting angle around a_p, b_p, and c_p axes. (b) The t dependence of f. See also Table S3. (c) The t dependence of ${\text{Δ}}_{\text{Mn-Mn}}$ (${\text{Δ}}_{\text{Mn-Mn}}$: the degree of differentiated neighboring Mn–Mn distances). The dotted curve in each panel is a guide to the eyes.

Turning to our oxynitride, La₂MnTaO₅N has a smaller t of 0.914 than relevant oxides like Sr₂MnWO₆ (t = 0.949) because of a smaller ionic radius of La^III than Sr^II (136 pm vs 144 pm for a 12-fold coordination).⁴⁷ As shown in Figure 7a, La₂MnTaO₅N can be plotted on the same line and, therefore, is expected to be less frustrated. However, as already shown, La₂MnTaO₅N has a larger frustration factor of f = 6 than oxides with similar t values, such as Ca₂MnWO₆ (t = 0.916 and f = 4) and LaCaMnNbO₆ (t = 0.911 and f = 5). This suggests that there is an additional parameter in oxynitride other than the f–t correlation found in the oxide double perovskite.

In order to find the additional parameter that enhances frustration in our oxynitride, we examined the distances between adjacent Mn ions. Here, let us define the degree of Mn–Mn length differences, ${\text{Δ}}_{\text{Mn-Mn}}$, as a factor for quantifying the nonuniform Mn–Mn distance, which is given by the following equation

$${\text{Δ}}_{\text{Mn-Mn}}=\frac{1}{12}{\sum _i\left(\frac{d_i-d_{\text{av}}}{d_{\text{av}}}\right)}^2$$

where $d_i(i=1,2,\mathrm{...},12)$ is the individual bond length and $d_{\text{av}}$ is the average. Figure 7c displays ${\text{Δ}}_{\text{Mn-Mn}}$ as a function of t. ${\text{Δ}}_{\text{Mn-Mn}}$ is zero for the ideal double perovskite (Pm3m). As ${\text{Δ}}_{\text{Mn-Mn}}$ increases, the difference becomes uneven but is almost the same up to t = 0.95. With a further decrease in t, ${\text{Δ}}_{\text{Mn-Mn}}$ for Ca₂MnWO₆ and LaCaMnNbO₆ rapidly increases, giving 1.04 × 10⁻⁴ and 0.665 × 10⁻⁴, respectively. This causes the suppression of spin frustration (see Figure 7b). However, ${\text{Δ}}_{\text{Mn-Mn}}$ of La₂MnTaO₅N remains small (0.253 × 10⁻⁴), which explains the increased frustration in this material with the higher f despite the small value of t. Note that the existence of two kinds of superexchange interactions, i.e., Mn–O–Mn and Mn–N–Mn, may also cause the reduced T_N.

We have shown that the frustration in the oxynitride double perovskite La₂MnTaO₆N could be scaled differently from that of the double perovskite oxides. Accordingly, the magnetic properties of Nd-substituted LaNdMnTaO₅N and Nd₂MnTaO₅N (t = 0.899 and 0.884, respectively) were investigated by specific heat. Here, the ionic radius of Nd^III is 127 pm (136 pm for La), so the spin frustration is expected to be smaller. As shown in Figure 6c, these compounds also exhibit a peak corresponding to magnetic order. The peak temperature increases, c.f. 4.5 K for La₂MnTaO₅N, 8.5 K for LaNdMnTaO₅N, and 9.5 K for Nd₂MnTaO₅N. Note that the Mn-derived magnetism cannot be extracted from the magnetic susceptibility of these materials due to the presence of Nd^III moments (J = 9/2), so the heat capacity was measured in Nd substitutes.

4. CONCLUSION

The present study addresses the correlation between geometrical spin frustration and structural distortion in manganese double perovskite oxides and oxynitrides. In A₂Mn^IIBO₆, we found that the frustration factor shows a linear decrease with a decreasing tolerance factor t (i.e., increasing octahedral rotation angle). In contrast, the spin frustration of the newly synthesized monoclinic phase of La₂MnTaO₅N (P2₁/n) under high pressure increases, despite the small tolerance factor of this phase. This unexpected result may be rationalized by the preferential occupation of nitride anions in the equatorial sites, which suppresses the variance of the neighboring Mn–Mn lengths in the Mn sublattice. This means that substitution by different anions gives a new dimension, in addition to the tolerance factor, to control the structure and frustration in double perovskite systems.