Unusual Electrical Transport Driven by the Competition between Antiferromagnetism and Ferromagnetism in Antiperovskite Mn₃Zn_1−xCo_xN

Abstract

The magnetic, electrical transport and thermal expansion properties of Mn₃Zn_1−xCo_xN (x = 0.2, 0.4, 0.5, 0.7, 0.9) have been systematically investigated. Co-doping in Mn₃ZnN complicates the magnetic interactions, leading to a competition between antiferromagnetism and ferromagnetism. Abrupt resistivity jump phenomenon and negative thermal expansion behavior, both associated with the complex magnetic transition, are revealed in all studied cases. Furthermore, semiconductor-like transport behavior is found in sample x = 0.7, distinct from the metallic behavior in other samples. Below 50 K, resistivity minimum is observed in samples x = 0.4, 0.7, and 0.9, mainly caused by e-e scattering mechanism. We finally discussed the strong correlation among unusual electrical transport, negative thermal expansion and magnetic transition in Mn₃Zn_1−xCo_xN, which allows us to conclude that the observed unusual electrical transport properties are attributed to the shift of the Fermi energy surface entailed by the abrupt lattice contraction.

1. Introduction

As a strongly correlated electron system, antiperovskite compounds with a chemical formula Mn₃XN (X: transition metals or semiconducting elements) and with a noncollinear magnetic ground state induced by the geometric frustration in the Mn₆N octahedron have been shown to exhibit fascinating physical properties, such as abnormal thermal expansion including negative thermal expansion (NTE) and zero thermal expansion etc. [1,2,3], near-zero temperature coefficient of resistivity (TCR) [4,5,6], magnetostriction [7], spin-glass (SG) behavior [8,9,10] and magnetocaloric effect [11,12]. It has been found that these interesting physical properties are sensitive to the number of the valence electrons of metal X located at the corners of antiperovskite unit cell, which contributes itinerant electrons at the Fermi level [13]. Hence, any change in carrier concentration of Mn₃XN has a significant impact on its electronic structure, and may produce great diversity of its magnetic structures and related novel physical phenomena [14].

Among these antiperovskite compounds, Mn₃ZnN with the so-called noncollinear Γ^5g antiferromagnetic (AFM) structure, has attracted considerable attention [15,16]. Previous reports have demonstrated that Mn₃ZnN undergoes an AFM transition at around 180 K, responsible for a resistive-switching phenomenon [17], and two cubic phases appear in a given temperature range [18]. Chemical doping and defect based on Mn₃ZnN could lead to additional interesting magnetic and electronic properties [1,19,20,21,22]. For example, the zero thermal expansion (ZTE) of Mn₃[Zn-(Ag,Ge)]_xN compounds has been investigated, and is closely correlated with the magnetic structure and can be tuned by adjusting the chemical and vacancy concentrations [1] and the crystallite size [23]. Magnetoresistance reversal has been reported in Mn₃Cu_0.5Zn_0.5N, which is thought of as the reconstruction of the Fermi surface accompanied by an AFM- ferromagnetic (FM) transition [24]. Moreover, recent reports have shown that the introduction of Co could effectively tune the physical properties in antiperovskites, such as the near zero TCR in Mn_3−xCo_xCuN [25] and the AFM-FM transition in Mn₃Ag_1−xCo_xN [26]. Since Co bears a similar electronic structure to that of Zn, introducing magnetic Co in Mn₃ZnN may provide new insight into the understanding of the origin of these novel physical properties.

In this study, we present the magnetic, electrical transport and thermal expansion properties of Mn₃Zn_1−xCo_xN compounds. The doping of magnetic Co at the Zn site in Mn₃ZnN can effectively modify the magnetic interactions and trigger strong AFM and FM competition. The competing interaction may prompt the unusual transport properties in Mn₃Zn_1−xCo_xN.

2. Materials and Methods

2.1. Sample Preparation

Polycrystalline samples Mn₃Zn_1−xCo_xN (x = 0.2, 0.4, 0.5, 0.7, 0.9) were synthesized by solid-state reaction from stoichiometric mixtures of Mn₂N, Zn, and Co powders. These powders were carefully mixed and ground in a mortar, and then pressed into pellets. The pellets were wrapped in Ta foils and sealed in vacuum (p < 10⁻⁵ Pa) into quartz tubes. The tubes were then sintered at 1073.15 K for 80 h, and cooled down to room temperature.

2.2. Characterization

Variable-temperature XRD experiments in the temperature range 10–300 K were carried out on a Bruker D8 ADVANCE diffractometer (Bruker Corporation, Billerica, MA, USA ) with Kα1 radiation selected by a Ge (111) primary beam monochromator. The measurements of the temperature-dependent magnetization from 10 K to 350 K were performed on Magnetic Property Measurement System (MPMS) (Quantum Design, San Diego, CA, USA). Both zero-field-cooled (ZFC) and field-cooled (FC) magnetization curves were measured from 5 K to 350 K under external magnetic field of 600 Oe. Magnetic hysteresis loops between 0 and 5 T were recorded at 50, 100, 150, 200, and 300 K. The electrical resistivity was measured using the standard four-probe method with a commercial (Quantum Design, Inc.) physical property measurement system (PPMS). Differential Scanning Calorimeter (DSC200F3, NETZSCH, Gebrüder-Netzsch-Straße, Selb, Germany) was used to measure the specific heat of the samples in the temperature range 110–300 K.

3. Results and Discussion

3.1. Crystal Structure

The crystal structures of all samples were investigated using powder XRD at room temperature, as shown in Figure 1a. All samples crystallize in the cubic antiperovskite structure with the space group Pm-3m. The XRD pattern of all the Mn₃Zn_1−xCo_xN (0 < x ≤ 1) samples were analyzed using the Fullprof software [27]. An initial analysis of all patterns was carried out by assuming a space group Pm-3m with N, Zn/Co, and Mn atoms at the 1a site (0, 0, 0), 1b site (1/2, 1/2, 1/2), and 3d site (1/2, 0, 0), respectively. The good agreement between the observed and calculated patterns from the Rietveld refinement indicates that the Pm-3m model is suitable for Mn₃Zn_1−xCo_xN compounds. Figure 1c,d shows the refined results for samples x = 0.2 and 0.9, respectively. A small amount of impurity phase MnO was marked in Figure 1 and was not considered in the refinement. The lattice parameter as a function of the Co concentration is shown in Figure 1b. The lattice constant decreases monotonically with increasing Co content, as the atomic radius of Co is smaller than that of Zn. These results indicate that Co replaces Zn in Mn₃ZnN as designed.

3.2. Magnetic Properties

The temperature dependence of the magnetization curves M (T) of the polycrystalline Mn₃Zn_1−xCo_xN under ZFC and FC processes is shown in Figure 2. In sharp contrast to the host material Mn₃ZnN where an AFM transition occurs at ~183 K [15], the samples x = 0.2, 0.4, 0.5, 0.7 and 0.9 show a magnetic state featured by a canted AFM magnetic ground state, resulting from the competition between FM and AFM interactions. The irreversibility between ZFC and FC curves probably implies the presence of FM components related to canting phenomenon. It is worth noting that the magnetization decreases as the Co-doping proceeds, indicating the suppression of FM interactions and the enhancement of AFM interactions. In order to further study the magnetic properties, we performed specific heat measurement. All temperature-dependent specific heat C_P curves (Figure 2) exhibit an obvious peak around the magnetic transition T_N, in good accordance with the magnetization results. The small cusp in specific heat curves at ~115 K is a contribution of impurity phase MnO which undergoes a magnetic phase transition at 115 K.

To obtain information on the nature of the magnetic order, the spin-only expression: χ(T) = C/(T − Θ_W), where C is the Curie constant and Θ_W is the Weiss temperature, was applied to fit the paramagnetic linear region of the magnetic susceptibility curves, as shown in Figure 2f. The fitting results for all samples are presented in

Table 1. Parameters obtained by the fitting of Curie-Weiss of Mn₃Zn_1−xCu_xN compounds.

Co (x)	Weiss Temperature (K)	Effective Moment μeff (μB)
0.2	200	2.00
0.4	115	2.14
0.5	−22	2.76
0.7	−220	3.41
0.9	−380	3.68

. For samples x = 0.2 and 0.4, positive Θ_W of 200 and 115 K were obtained, respectively, indicating the dominant FM interactions. However, negative Θ_W were obtained for x ≥ 0.5, suggesting the governing role of the AFM interactions. Such a variation of the sign of Θ_W implies the enhancement of the AFM interactions with increasing Co concentration. Moreover, the effective magnetic moment μ_eff of all the compounds is much lower than the theoretical value of 4 μ_B/Mn [28], which is consistent with the magnetism from itinerant electrons, as expected for these compounds. As shown in Figure 3a–e, the presence of a small fraction of ferromagnetic components can be confirmed in the isothermal M-H curves. For x = 0.2, the magnetization does not reach saturation, but has a remnant magnetization value of 0.32 µ_B/f.u. as the magnetic field increases below T_N. The remnant magnetization decreases with increasing Co content. These features imply that the Mn₃Zn_1−xCo_xN compounds possess a canted AFM magnetic ground state with x > 0. Regarding the magnetic interaction pathways, previous studies have shown consistently that the triangular lattice composed of Mn atoms is uniquely responsible for the magnetic properties. It is hence reasonable to believe that the significant spin interactions arise from Mn-Mn atoms. However, based on our current experimental results, we cannot rule out the possibility of the existence of exchange interactions between Co-Co atoms, which may affect the remnant magnetization. Further verification about this point requires advanced experimental techniques such as neutron diffraction.

3.3. Electrical Transport Behavior and Negative Thermal Expansion

Figure 4 shows the temperature-dependent resistivity ρ (T) of Mn₃Zn_1−xCo_xN (x = 0.2, 0.4, 0.7, 0.9) in the temperature range of 5–350 K. No large magnetoresistance behavior was observed in any of the studied samples (see Figure S1 in Supplemental Materials). For x = 0.2, the resistivity first decreases upon cooling from 300 K, then increases abruptly at T_N = 175 K, reaching a maximum, and finally decreases with a further decrease in temperature. The resistivity exhibits a temperature-dependent fluctuation up to 20% (evaluated by the function (ρ_max − ρ_mim)/ρ_mim), which is much higher than that in Mn₃ZnN [17]. Sample with x = 0.4 also shows abrupt resistivity change at T_N, but with a very slight decrease after a resistivity maximum. Namely, the transport behavior at low temperature is metallic for Mn₃Zn_1−xCo_xN (x = 0.2 and 0.4) compounds. For x = 0.7, there is an obvious increase in resistivity around T_N = 260 K with decreasing temperature down to 175 K, as shown in Figure 4c. Then the resistivity increases slowly but monotonically to the lowest measured temperature. In contrast to x = 0.2 and 0.4, the temperature dependence of the resistivity of Mn₃Zn_0.3Co_0.7N shows a semiconductor-like transport behavior which is likely a result of the change of its energy band structure [29]. This gives rise to a metal-to-semiconductor-like change from sample x = 0.2 to x = 0.7.

It has been previously shown that Mn₃ZnN undergoes a resistive switching phenomenon driven by AFM phase separation below 190 K [17]. In our work, the behavior of electrical transport in all investigated samples (from x = 0.2 to 0.9) shows an abrupt resistivity jump around T_N. Among them, the sample with x = 0.9 exhibits an abrupt resistivity change at the much higher temperature T_N = 276 K. In all cases, the abrupt resistivity jump phenomenon is accompanied by magnetic ordering, indicating a strong correlation between the magnetism and the electrical transport.

To explain the unusual electrical transport behavior, we investigated the thermal expansion behavior of Mn₃Zn_1−xCo_xN series using variable temperature XRD. XRD data collected at different temperatures reveal that all samples crystallize into cubic cells with space group Pm-3m, and no structural transition can be observed over the whole measured temperature range. The refinement of the XRD data yields the temperature dependence of the lattice constant, as shown in Figure 5. It can be seen that all samples display a negative thermal expansion behavior in their specific temperature range. For samples x = 0.2, 0.4, 0.5, and 0.7, the temperature range could be estimated at about 125–180 K, 180–230 K and 150–240 K, 153–223 K, respectively, with respective linear coefficients of thermal expansion −5.53 × 10⁻⁵ K⁻¹, −3.1 × 10⁻⁵ K⁻¹, −1.67 × 10⁻⁵ K⁻¹ and −0.68 × 10⁻⁵ K⁻¹. Obviously, the introduction of Co effectively broadens the temperature range of NTE. This is because of the negative thermal expansion associated with the magnetic phase transition through the magneto-volume effect, as suggested in several antiperovskites [1,2,3]. On the basis of this lattice change, we may discuss slightly the underlying cause for unusual electrical transport properties in Mn₃Zn_1−xCo_xN. The abrupt change in the lattice parameter, caused by the magnetic transition, may lead to the shift of Fermi level. Therefore, an abrupt decrease in the DOS near the Fermi level could be generated through the shift of the Fermi energy surface, leading to a pronounced decrease of the effective number of conduction electrons, as explained in Ref. [29]. Therefore, the resistivity can be enhanced significantly at the magnetic transition temperature. In addition, note that the grain size and grain boundaries of the samples can also play a role in producing the abnormal change in resistivity. The reason why only sample x = 0.7 shows semiconductor-like transport behavior can also be addressed based on negative thermal expansion. The linear coefficient of NTE of sample x = 0.7 is smaller compared to other samples, as shown in Figure 5. Therefore, we believe that gradual change of lattice parameter as a function of temperature is a key ingredient for the occurrence of semiconductor-like feature.

Another marked feature is the appearance of electrical resistivity minima below 50 K in the Mn₃Zn_1−xCo_xN (x = 0.4, 0.7, 0.9) samples, which reflects the involvement of additional scattering factors. Generally, besides the well-known Kondo mechanism [30,31], there are other possible models that could account for the ρ minima, such as e-e interaction and electron-phonon interactions. In a strong electron-correlated system, the e-e interaction should play an important role in the electronic transport. To make a quantitative analysis, taking into account these mechanisms for minimum resistivity, the following equation was therefore taken to fit the low temperature resistivity data [30,31,32,33]

$${\mathsf{\rho }=\mathrm{A}+\mathrm{BT}}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}{-\mathrm{ClnT}+\mathrm{DT}}^5$$

(1)

where the coefficients A, B, C, and D represent the contributions from the residual resistivity, electron-electron (e-e) interactions, Kondo-like spin-dependent scattering, and electron-phonon (e-p) interactions, respectively. The curves are fitted well with Equation (1), as shown in the inset of Figure 4. The fitting parameters are listed in

Table 2. Table 2 Parameters used to fit the resistivity data (H = 0 kOe) of the Mn₃Zn_1-xCo_xN (x = 0.4, 0.7 and 0.9) compound.

x	A	B	C	D
0.2	-	-	-	-
0.4	1.69 × 10−4	6.69 × 10−8	9.06 × 10−7	9.95 × 10−16
0.7	1.02 × 10−4	5.41 × 10−8	1.96 × 10−7	1.49 × 10−15
0.9	7.14 × 10−5	2.31 × 10−8	2.43 × 10−7	1.14 × 10−15

. The coefficient D related to e-p interactions is much smaller than B and C and can be neglected, suggesting that the behavior of electrical resistivity minima is mostly determined by the Kondo-like scattering and the e-e interactions. The coefficient B decreases as the Co concentration increases, indicating the suppression of the e-e scattering. This suggests that the appearance of the FM state may constrain the local spin directions and suppress the e-e scattering [33]. Because the Kondo-like scattering plays only a minor role, the good agreement between the experimental and fitted result suggests that the e-e interactions should be mainly responsible for the electrical resistivity minima. Even though the phenomenological fitting provides a reasonable explanation for the occurrence of resistivity minimum, other possibilities of generating such phenomenon such as chemical disorder or defects cannot be ruled out.

4. Conclusions

In summary, Mn₃Zn_1−xCo_xN (x = 0.2, 0.4, 0.5, 0.7, 0.9) were synthesized by solid-state reaction in vacuum. The effect of Co doping on the magnetic, thermal expansion and resistivity properties of antiperovskite Mn₃Zn_1−xCo_xN compounds was investigated. As a consequence of the doping of Co at the Zn site, all samples show competition between AFM and FM interactions, which are associated with the observed unusual electrical transport and negative thermal expansion behavior. The Mn₃Zn_1−xCo_xN (x = 0.2, 0.4, 0.7, 0.9) compounds exhibit an abrupt resistivity jump phenomenon near the magnetic phase transition, which originates from the shift of the Fermi surface triggered by negative thermal expansion. Resistivity minima at low temperatures were observed in Mn₃Zn_1−xCo_xN (x = 0.4, 0.7, 0.9) and the e-e interaction is likely responsible.