Ba(Zn_1−2xMn_xCu_x)₂As₂: A Bulk Form Diluted Ferromagnetic Semiconductor with Mn and Cu Codoping at Zn Sites

Abstract

We report the synthesis and characterization of a bulk form diluted magnetic semiconductor Ba(Zn_1−2xMn_xCu_x)₂As₂ with the crystal structure identical to that of “122” family iron based superconductors and the antiferromagnet BaMn₂As₂. No ferromagnetic order occurs with (Zn, Mn) or (Zn, Cu) substitution in the parent compound BaZn₂As₂. Only when Zn is substituted by both Mn and Cu simultaneously, can the system undergo a ferromagnetic transition below T_C ~ 70 K, followed by a magnetic glassy transition at T_f  ~ 35 K. AC susceptibility measurements for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂ reveal that T_f strongly depends on the applied frequency with and a DC magnetic field dependence of , demonstrating that a spin glass transition takes place at T_f. As large as −53% negative magnetoresistance has been observed in Ba(Zn_1−2xMn_xCu_x)₂As₂, enabling its possible application in memory devices.

Introduction

The successful fabrication of III-V diluted magnetic semiconductors (In, Mn)As and (Ga, Mn)As through low temperature molecular beam epitaxy (LT-MBE) has opened up a new window for the study of magnetic semiconductors^1,2,3,4,5. It is proposed that the Curie temperature would reach room temperature with high enough spin and carrier densities⁶. Nevertheless, the low solid solubility of Mn²⁺ for Ga³⁺ makes it difficult to enhance the concentration of Mn. As of today, the highest Curie temperature, T_C, of (Ga, Mn)As films has been reported as 200 K⁷. On the other hand, Mn²⁺ substituting for Ga³⁺ introduces not only carriers but also local moments and some Mn²⁺ enter interstitial sites or even As sites, which makes it difficult to separate the charges and spins and investigate their individual influences on the ferromagnetism. Seeking for new DMS materials that have higher chemical solubility of magnetic atoms and whose carrier density and spin density can be controlled separately may be helpful to improve T_C and understand the mechanism of the ferromagnetic ordering⁸.

Recently, many novel DMSs that are derivatives of Fe-based superconductors have been reported^{8,9,10,11,12,13,14,15,16,17,18,19}. It has been shown from NMR²⁰ and μSR measurements that the ferromagnetism in (BaK)(ZnMn)₂As₂⁹, Li(Zn, Mn)As¹⁰, (La, Ba)(Zn, Mn)AsO¹² and Li(Zn, Mn)P²¹ are homogeneous, i.e., the long range ferromagnetic ordering is arising from the Mn atoms doped at Zn sites, instead of Mn related magnetic impurities. Furthermore, μSR results demonstrated that these bulk form DMSs share the same mechanism for the ferromagnetic ordering as that of (Ga, Mn)As²². These bulk form DMSs have the advantages of decoupled spin and carrier doping and the carrier densities can be controlled and tuned, which overcomes the low carrier densities encountered in II-VI DMS²³. These systems are isostructural to its variants, i.e., antiferromagnets and superconductors with lattice matching within 5%, which provides the the possibility to make junctions with these materials thorough the As layer¹⁹. In addition, the bulk form specimens would enable the magnetic techniques to provide complementary information at a microscopic level, such as nuclear magnetic resonance (NMR) and neutron scattering¹². Among them, the T_C of (Ba, K)(Zn, Mn)₂As₂ single crystal has been reported to reach 230 K²⁴. (Ba, K)(Zn, Mn)₂As₂ was synthesized by doping Mn and K into the parent compound β-BaZn₂As₂ which is a direct gap (0.2 eV) semiconductor²⁵, where the substitution of Mn for Zn and K for Ba introduces spins and hole carriers, respectively.

In this paper, we report the successful fabrication of a new DMS material with a rather new synthesize route, which is different to the previously reported ~10 DMSs^{8,9,10,11,12,13,14,15,16,17,18,19}. Instead of doping at different sites, we co-doped both Mn and Cu into the same Zn sites of BaZn₂As₂ to introduce local moments and carriers, respectively. A new series of DMS compounds Ba(Zn_1−2xMn_xCu_x)₂As₂ (0.025 ≤ x ≤ 0.2) have been successfully fabricated. While the system remains semiconducting, 20% Mn and Cu doping results in a ferromagnetic transition below T_C ~ 70 K, followed by a magnetic glassy transition below T_f  ~ 35 K. AC susceptibility measurements on an x = 0.125 sample indicate that T_f strongly depends on the applied frequencies and magnetic fields, which confirms the spin glass nature at T_f. In addition, as large as ~−53% negative magnetoresistance (MR) at a magnetic field H = 50 KOe has been achieved in Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂, which is attributed to the suppression of spin fluctuations by magnetic field. Future work is needed to gain deeper understanding of the magnetic behavior of this system and achieve higher T_C values.

Results and Discussion

Synthesis and structural characterization

The polycrystalline specimens of Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.200) were synthesized by the solid state reaction method. Details of the synthesis and facilities used for characterization are described in the Methods section. In Fig. 1, we show the X-ray diffraction patterns for polycrystalline Ba(Zn_1−2xMn_xCu_x)₂As₂ (0.025 ≤ x ≤ 0.200). The Rietveld refinement for Ba(Zn_0.85Mn_0.075Cu_0.075)₂As₂ with parameters R_WP = 10.52 %, R_P = 7.58 %, χ² = 1.348 shows that the Bragg peaks can be well indexed into the tetragonal structure with space group I4/mmm. The lattice parameter a increases and c decreases monotonically with the doping concentration x, indicating the successful doping of Mn and Cu into the lattice. We show the crystal structure in Fig. 1(c), which is isostructural to the parent compound of 122-type Fe-based superconductor Ba(Fe_1−xCo_x)₂As₂²⁶ with T_c = 22 K and antiferromagnet BaMn₂As₂ with Néel temperature T_N = 625 K²⁷. This feature provides the possibility to make junctions with these systems though As layer. No peaks of impurities are detected for the doping levels of x = 0.025 and x = 0.075. α-BaZn₂As₂ with space group of Pnma appears for x = 0.125 and becomes markable for x = 0.20, as marked by * in Fig. 1(a). Small traces of non-magnetic Ba₃As₁₄ impurity are marked as #. Both α-BaZn₂As₂ and Ba₃As₁₄ are Pauli paramagnetic, which will not affect the magnetic behavior of Ba(Zn_1−2xMn_xCu_x)₂As₂ discussed in the following.

Figure 1

Structural properties of Ba(Zn_1−2xMn_xCu_x)₂As₂.

(a) The X-ray diffraction patterns of Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20). Traces of BaZn₂As₂ with space group Pnma (*) and impurity Ba₃As₁₄ (#) are marked; (b) The systematic change of the lattice constants a (red filled circles) and c (blue filled squares) with x. (c) The layered crystal structure. (d) The Rietveld refinement of the powder X-ray diffraction for the x = 0.075 sample.

Resistivity

In Fig. 2, we show the temperature dependent resistivity of the parent compound BaZn₂As₂, Ba(Zn_0.9Cu_0.1)₂As₂ and Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20). The resistivity of the parent semiconductor BaZn₂As₂ displays a typical semiconducting behavior. With 10% Cu doping, the resistivity of Ba(Zn_0.9Cu_0.1)₂As₂ is heavily suppressed by an order of 4, indicating that carriers are doped. The semiconducting behavior for Mn and Cu codoped case has been conserved for x up to 20%, i.e., resistivity continuously increases with temperature decreasing from room temperature down to 4 K. The absolute value of resistivity at 4 K, however, drops from 10³ Ω cm for x = 0.025 to 10 Ω cm for x = 0.20. We roughly fit the resistivity of Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20) near room temperature in terms of a thermal activation function¹³. Similar approach has also been employed to (La, Sr)(Zn, Mn)AsO¹³. The fitting result for Ba(Zn_0.6Mn_0.2Cu_0.2)₂As₂ is shown in Fig. 2(b) as an example. The values of energy gap E_g are between 0.031 and 0.048 eV, which are about an order of magnitude smaller than that of the parent compound BaZn₂As₂²⁵. We have conducted Hall effect measurement on Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂, but the large resistivity prevents us to accurately determine the carrier density. A preliminary result shows that the carriers are p-type with the concentration in the order of p ~ 10¹⁹ cm⁻³. And the corresponding mobility is estimated to be in the order of 10⁻¹ cm²V⁻¹s⁻¹. This value of carrier density is not unusual in bulk form DMSs, which is comparable to that of (Ba_0.9K_0.1)(Cd_2−xMn_x)As₂¹⁶ and two orders of magnitude larger than that of Li(Zn, Mn)P¹¹, but an order of magnitude smaller than that of (Ba, K)(Zn, Mn)₂As₂⁹ and Li(Zn, Mn)As¹⁰.

Figure 2

Electrical properties of Ba(Zn_1−2xMn_xCu_x)₂As₂.

(a) The temperature dependence of electrical resistivity for BaZn₂As₂, Ba(Zn_0.9Cu_0.1)₂As₂ and Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20) in logarithmic scale; (b) The fit of resistivity for Ba(Zn_0.6Mn_0.2Cu_0.2)₂As₂ according to ρ ∝ exp(E_g/2k_BT).

Magnetization and hysteresis

In Fig. 3(a,b), we show the temperature dependence of magnetization for Ba(Zn_0.9Mn_0.1)₂As₂ and Ba(Zn_0.9Cu_0.1)₂As₂, respectively. No anomaly or transition is observed in the measured temperature range and the moment at 2 K is only ~0.001 μ_B/(Mn or Cu atom) for both Ba(Zn_0.9Mn_0.1)₂As₂ and Ba(Zn_0.9Cu_0.1)₂As₂. We fit the magnetization data to a Curie-Weiss law M = M₀ + C/(T − θ) and obtained C = 0.00456 μ_BK/Mn, θ = −2.74 K for Ba(Zn_0.9Mn_0.1)₂As₂ and C = 0.00028 μ_BK/Cu, θ = −1.45 K for Ba(Zn_0.9Cu_0.1)₂As₂, indicating the paramagnetic ground state. These results indicate that doping either Mn or Cu alone into BaZn₂As₂ can not form any type of magnetic ordering. This feature has also been observed in LaZnAsO, where doping Mn or Fe only does not result in ferromagnetic ordering^8,12. The magnetic character of Cu in 122-type arsenides has been investigated by density functional calculations²⁸ and intensive transport properties measurements²⁹. Cu 3d bands are ~3 eV below Fermi energy (E_F) and contribute little to the density of states at E_F²⁸. The 3d shell of Cu is completely filled with 3d¹⁰ electronic configurations^28,29. Therefore, the valence of Cu in 122-type arsenides is +1 with nonmagnetic state S = 0^28,29. The paramagnetic state of Ba(Zn_0.9Cu_0.1)₂As₂ are consistent with the previous reports^28,29.

Figure 3

Magnetic properties of Ba(Zn_1−2xMn_xCu_x)₂As₂.

Temperature dependent magnetization M for (a) Ba(Zn_0.9Mn_0.1)₂As₂ in an external field of H = 10 Oe and (b) Ba(Zn_0.9Cu_0.1)₂As₂ with H = 100 Oe. The solid lines represent the Curie-Weiss law M = M₀ + C/(T − θ). The star marks the signal from adsorbed oxygen. (c) T-dependent magnetization M for Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20) in the zero field cooling (ZFC) and field cooling (FC) modes with an external field of H = 100 Oe. (d) The isothermal magnetization measured at 2 K.

In Fig. 3(c), we show the magnetization of Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.20) with the same amount of Mn and Cu atoms doped into Zn sites of BaZn₂As₂. No magnetic transition has been observed for x = 0.025. A fit to Curie-Weiss law M = M₀ + C/(T − θ) shows that θ = −0.6 K, indicating the paramagnetic ground state. For the doping level of x = 0.075, a strong increase of magnetization at Curie temperature T_C = 33 K and a bifurcation of zero field cooling (ZFC) and field cooling (FC) curves at spin freezing temperature T_f = 12 K are observed. T_C and T_f are enhanced with increasing doping levels. With 20% doping, T_C increases to 70 K and T_f increases to 35 K. In Fig. 3(d), we present the results of isothermal magnetization measurements. For x ≥ 0.075, clear hysteresis loops have been observed at 2 K. The coercive field becomes lager for higher x and reaches 1600 Oe for x = 0.20. We should note that this value is much smaller than ~10⁴ Oe of (Ba, K)(Zn, Mn)₂As₂⁹. The contrasting ground states shown in Fig. 3(a,b,c) unequivocally demonstrate that only when Zn is substituted by both Mn and Cu simultaneously, can the ferromagnetic ordering develop, which also indicates that the ferromagnetic signals result from the doping of Mn and Cu rather than impurities.

We fit the T-dependent magnetization above T_C to the Curie-Weiss formula χ = χ₀ + C/(T − θ) in order to obtain the Weiss temperature (θ) and effective paramagnetic moment of Mn (μ_eff). The best fittings show that the effective moment μ_eff is 4.8 ~ 5.7 μ_B/Mn for 0.025 ≤ x ≤ 0.20, indicating the high spin state of Mn with the valence of +2 in the system of Ba(Zn_1−2xMn_xCu_x)₂As₂. We tabulate the Curie temperature T_C, the spin freezing temperature T_f (the temperature where ZFC and FC curves split), the base temperature moment μ_BT (the values at 2 K measured from FC curves with H = 100 Oe), the coercive field H_c and the energy gap E_g (fitted from the resistivity data) in Table 1. T_C, T_f, θ and H_c show a trend of increasing with higher doping level x, indicating the strengthening of ferromagnetic exchange interaction with higher concentration of Mn and Cu. Meanwhile, the systematic changes of these magnetic parameters suggest that the magnetic signals in this system are not caused by impurities. On the other hand, we notice that μ_BT first increases from 0.027 μ_B/Mn for x = 0.075 to 0.110 μ_B/Mn for x = 0.125, but decreases to 0.079 μ_B/Mn for x = 0.20. This may be due to the competition of ferromagnetic and antiferromagnetic exchange interactions between Mn atoms.

Table 1 Curie temperature (T_C), spin freezing temperature (T_f), Weiss temperature (θ), base temperature moment (μ_BT, the values at 2 K from FC curves with H = 100 Oe), the effective moment (μ_eff), coercive field (H_c) and energy gap (E_g, fitted from resistivity) for Ba(Zn_1−2xMn_xCu_x)₂As₂ (0.025 ≤ x ≤ 0.20).

To further investigate the valence of Cu and Mn, we conducted the X-ray photoelectron spectroscopy (XPS) measurements for Ba(ZnMn_0.2Cu_0.2)₂As₂. Ba and Zn have been observed from the peaks of binding energy. But unfortunately, after very careful comparison, we haven’t detected effective peaks of Cu or Mn from the binding energy. No conclusion about the valence of Cu or Mn has been achieved from the XPS measurements. We can’t obtain evidence from XPS that whether Cu contribute magnetic moments or not. Magnetic resonance techniques may be used in the future to further investigate the magnetic mechanism of this system.

Magnetoresistance

We measured the magnetoresistance for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂ under the applied fields of 0, 10, 30, 50 KOe and show the results in Fig. 4. The resistivity with different fields deviates from each other at ~22 K and the values of ρ at 7 K monotonically drop from 3363 Ω m at 0 Oe to 1587 Ω m at 50 KOe. The magnetoresistance (defined as [ρ(H) − ρ(0)]/ρ(0) at 7 K) reaches −53 % at 50 KOe. The large negative magnetoresistance has also been observed in other bulk form DMSs, such as (Ba_0.9K_0.1)(Cd_2−xMn_x)₂As₂¹⁶ and (Sr_0.9K_0.1)(Zn_1.8Mn_0.2)As₂¹⁷. We tentatively attribute the negative magnetoresistance to the suppression of spin fluctuations by applied field.

Figure 4

T-dependent magnetoresistivity for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂ under H = 0, 10, 30, 50 KOe.

The inset shows the field dependence of magnetoresistance at 7 K from −50 KOe to 50 KOe.

AC susceptibility

We measured the AC susceptibility, , for the x = 0.125 sample at various frequencies ν under zero external field and show the results in Fig. 5. We found that the maxima of the real part, , drop obviously and T_f shifts slightly to higher temperature with the increasing AC frequencies. This feature is typically taken as signs for spin glass systems^{30,31,32,33,34,35,36,37}. This kind of behavior has also been observed in CaNi_1−xMn_xGe³⁰, CeCu₄Mn³¹, La(Fe_1−xMn_x)_1.4Si_1.6³² and II-VI family DMS²³. The Vogel-Fulcher law^38,39,40,41 is usually used to describe the dependence between T_f and ν,

Figure 5

AC susceptibility at various frequencies for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂.

(a) The real component of AC magnetic susceptibility as a function of temperature T measured from 10 Hz to 10000 Hz. (b) Spin freezing temperature T_f as a function of 1/ln(ν₀/ν) fitted with the Vogel-Fulcher law. (c) A linear fit of ln(ν₀/ν) versus ln(T_f/T₀ − 1) with Eq. (2).

where E_a is the activation energy, T₀ is the Vogel-Fulcher temperature and ν₀ is the fitted frequency. We tried different values of ν₀ from 10¹⁰ Hz to 10¹³ Hz, which showed that the best linear fitting can be obtained when  Hz, in good agreement with expectation for a spin-glass Hz) rather than a cluster-glass  Hz)⁴². So ν₀ is considered as a constant value of 10¹³ Hz for this system in the following discussion. In Fig. 5(b), T_f is plotted as a function of 1/ln(ν₀/ν). The well fitted linear relation enables us to estimate the value of T₀ and E_a. T₀ is ~16 K and E_a is ~110.48 K, corresponding to eV because of E = k_BT with eV/K. So E_g = 2E_a = 0.02 eV, which is in the same order of magnitude as the value estimated from the fit of resistivity. The ratio is usually used to distinguish the frequency sensitivity of T_f in a spin glass^40,43. K is the order of 0.01 for spin glass systems, while K  > 0.1 for superparamagnets³⁶. For the x = 0.125 sample, K is estimated to be ~0.008 ± 0.002, in good agreement with the typical values reported for spin glasses^30,31,37,40.

The dynamical slowing down of spin fluctuations can also be expressed by the standard power dependence,

where τ = 1/ν is the relaxation time, τ₀ = 1/ν₀ is set as 10⁻¹³ s, T_G is the spin freezing temperature, η is the dynamic exponent. When T_f approaches T_G which is the zero frequency limit, the order of τ gets much larger than τ₀, indicating that spin fluctuations significantly slow down. A linear fit of ln(ν₀/ν) versus ln(T_f/T_G − 1) according to Eq. (2) is shown in Fig. 5(c), yielding T_G ~ 18.96 K and . The value of η falls into the range of 4–12 for spin glasses^{31,32,33,36,40,44,45}, which is not cluster-glass like character⁴⁶. is close to 7.9, the calculated value for the three-dimensional Ising spin-glass^47,48.

In Fig. 6(a), we show the measurements of T-dependent AC susceptibility at a fixed frequency of 500 Hz with various DC fields for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂. The AC susceptibility is strongly affected by the external DC fields, i.e., the cusps smear out, the peak value of decreases remarkably and T_f shifts to lower temperature with increasing DC fields. These are all characteristic features of spin glasses^30,32,49. The DC field dependence of the spin freezing temperature T_f can be described by the equation,

Figure 6

AC susceptibility with various DC fields for Ba(Zn_0.75Mn_0.125Cu_0.125)₂As₂.

(a) The temperature dependence of the AC susceptibility measured at a frequency of 500 Hz under different applied DC fields. (b) T_f as a function of H^δ with δ = 0.55. The straight line is a guide for eyes. (c) T-dependent imaginary part at 500 Hz with various DC fields from 0 to 3000 Oe.

A best fit of T_f versus H to Eq. (3) yields . We show the plot of T_f versus H^0.55 in Fig. 6(b). δ is ~2/3 for Ising spin glass systems and δ = 2 for Heisenberg systems^50,51. In the current case, δ is close to 2/3, indicating that the glassy state for Ba(Zn_1−2xMn_xCu_x)₂As₂ may be explained by mean-field theory with Ising model. In Fig. 6(c), we show the imaginary component of AC susceptibility at 500 Hz with DC fields up to 3000 Oe. Similar to the case of , T_f decreases noticeably with increasing fields. The T-dependent imaginary part of the AC susceptibility under 2000 Oe and 3000 Oe becomes almost undependent of T.

Conclusion

A bulk form diluted magnetic semiconductor Ba(Zn_1−2xMn_xCu_x)₂As₂ (0.025 ≤ x ≤ 0.2) with maximum T_C ~ 70 K has been successfully synthesized. It is the first time that ferromagnetic ordering is observed when Mn and Cu are codoped into the Zn sites, where Mn substitution for Zn introduces spin and Cu substitution for Zn introduces carriers, respectively. The new system displays large negative magnetoresistance while conserving the semiconducting behavior with the doping level up to 20%. The AC susceptibility measurements show that the spin freezing temperature T_f is dependent on frequency and external field, confirming the glassy nature below 35 K. Finally, the new DMS system has a tetragonal crystal structure identical to that of “122” family of Fe-based superconductors and the antiferromagnetic system BaMn₂As₂, which makes it possible to make various junctions of these systems through As layer. More theoretical and experimental work are expected to further understand the properties and physics of this system.

Methods

The polycrystalline specimens of Ba(Zn_1−2xMn_xCu_x)₂As₂ (x = 0.025, 0.075, 0.125, 0.200) were synthesized by the solid state reaction method. High purity elements of Zn (99.9%), Mn (99.99%), Cu (99.9%) and As (99%) were mixed, ground and pressed into pellets. The pellets were sealed in evacuated silica tubes and sintered at 800 °C for 10 hours to make the precursors (Zn_1−2xMn_xCu_x)As. The mixture of Ba (99.2%) and (Zn_1−2xMn_xCu_x)As were then slowly heated to 900 °C and held for 10 hours, then 1150 °C for 15 hours before cooling down to room temperature with the furnace turned off. The handling of materials were performed in a high-purity argon filled glove box (the percentage of O₂ and H₂O ≤ 0.1 ppm) to protect it from exposure to air. Powder x-ray diffraction was performed at room temperature using a PANalytical x-ray diffractometer (Model EMPYREAN) with monochromatic CuK_α1 radiation. The electrical resistance was measured on sintered pellets with the typical four-probe method. The DC magnetization measurements were conducted on a Quantum Design Magnetic Property Measurement System (MPMS-5). The AC susceptibility and magnetoresistance were measured on a Quantum Design Physical Property Measurement System (PPMS-9).