It is an everlasting dream to make electronic devices miniaturization, multi-functionalization, intellectualization, and low-power consumption in this current information society^[1]. But many bottleneck problems occur when the fabrication technology encounter nanometer scale^[2]. As one branch of spintronics, diluted magnetic semiconductors (DMSs) offers intriguing opportunities to combine semiconductor properties and ferromagnetism in a single material that would enable to fabricate new devices to process and store information simul- taneously^{[3, 4]}.

DMSs researches date back to the concentrated magnetic semiconductors compounds, like EuX^[5] (X = O, S, Se and Te) & CdCr ₂X₄^[6] (X = S and Se). However, the complicated growth and the relatively low Curie temperature (T_C) for these compounds limited these materials to fundamental studies. In 1980s, people payed much attention to Mn doped II–VI group DMSs, such as (Zn,Mn)Te, (Zn,Mn)Se, (Cd,Mn)Te, etc^[7] and PbSnMnTe^[8]. Yet T_C is still only several Kelvins due to lack of effective approach to dope carriers^[9]. Benefiting from the low temperature molecular beam epitaxy (LT-MBE) technique, Mn doped III–V group DMSs were explored in 1990s, like (Ga,Mn)As^[10–12] and (In,Mn)As^[13]. The highest T_C could reach 200 K in heavy Mn doped (Ga,Mn)As film with proper annealing procedure^[14–16].

It is of great fundamental significance and practical value to further improve T_C of (Ga,Mn)As^{[17, 18]}. However, in (Ga,Mn)As, the Ga³⁺/Mn²⁺ substitution offers hole carriers and spins simultaneously, which make it difficult to individually control charge and spin concentrations and more importantly to improve T_C. Since 2011, a new type of DMSs with independent spin and charge doping, such as the “111” type Li(Zn,Mn)As^[19] and “122” type (Ba,K)(Zn,Mn)₂As₂ (BZA)^[20] named by its stoichiometry ratio, have been discovered. In these DMSs^[21], such as BZA, spins are introduced by isovalent Zn²⁺/Mn²⁺ doping, and charge carriers are introduced by heterovalent Ba²⁺/K¹⁺ substitution. Taking the advantage of freedom to individually control of charge and spin, BZA exhibit ferromagnetic order with T_C up to 230 K^[22] which is record of controlled reliable T_C among carrier mediated ferromagnetic DMSs^[16]. In this review, we demonstrate the discovery of three basic groups of new DMSs, namely the “111”, “122” and “1111” families (Fig. 1). Then we focus on physical properties of BZA and prototype device based on BZA single crystal. In the last section, we discuss the prospective of new type of DMSs with independent spin and charge doping.

LiZnAs^{[23, 24]} is a direct-gap semiconductor that has a cubic crystal structure with space group F-43m as shown in Figs. 1(a) & 1(b) and a band gap (1.61 eV) similar to that of GaAs^[25] (1.52 eV). Note that LiZnAs could be viewed as a zinc blende structure, in which Ga sites in (Ga,As) are replaced with Li and Zn. Benefitting on the LiFeAs^[26] iron superconductor fabrication experiences and facilities. Jin’s group firstly reported the successful synthesis of bulk form polycrystals Li(Zn,Mn)As^[19]. In principle, the Zn²⁺/Mn²⁺ replacement offers spins, and the off-stoichiometrical Li cation adjust the carrier’s type and concentration. But the long range ferromagnetic ordering was only observed with the excess Li doping and it is a p-type rather than n-type as the theory prediction^[27]. The reason is that the excess Li¹⁺ prefers to occupy the Zn²⁺ sites based on a DFT calculation^[28]. The temperature dependence of M at 2 kOe (no difference between field cooling and zero field cooling procedures) was illustrated in Fig. 2(a), and external field H from 0 to 20 kOe dependence of M at 2 K is shown in Fig. 2(b). Clear signatures of ferromagnetic order were seen in these figures with the highest T_C of 50 K in Li_1.1(Zn_0.85Mn_0.15) As sample. Semiconducting behavior of resistivity of LiZnAs (green line in Fig. 2(c)) changes to T-independent behavior for Li deficient systems, wheras much smaller resistivity and definite metallic behavior were found for Li excess samples. The resistivity increases monotonically with increasing Mn concentration in Fig. 2(d) which suggests that Mn acts as a scattering center. The onset of magnetic order reduces this scattering rate, as can be seen in the negative magnetoresistance of Li_1.1(Zn_0.9Mn_0.1)As sample in Fig. 2(e) below T_C. Fig. 2(f) shows representative anomalous Hall effect of Li_1.1(Zn_0.95Mn_0.05)As at 2 K, which exhibits p-type carriers with concentrations of n ~10²⁰ cm^–3.

In order to examine volume fraction and the ordered moment size, μSR measurements were performed on sintered polycrystalline specimens of Li_1.1(Zn_0.95Mn_0.05)As^[29]. The time spectra in zero field (ZF), shown in Fig. 3(a), clearly show an increase of the relaxation rate below ~25 K on Fig. 3(b), and measurements in longitudinal fields confirmed that the increase is from static magnetic order. This is in correspond with the earlier result in (Ga,Mn)As^[30]. The volume fraction of regions with static magnetism on Fig. 3(c) indicates a sharp transition at T_C ~ 30 K, and full volume magnetic order achieved when T < 10 K. Further studies on fabrication methods and heat treatments might help improve spatial homogeneity of the transition. The weak transverse field (WTF) of 30 Oe, shown in Fig. 3(d), also provide direct signal of the magnetic volume fraction^[31]. When the internal magnetic field is much larger than the applied external field, the scale of oscillation can reflect the paramagnetic volume. The consistenct of the ordered fraction from the measurements in ZF and WTF, shown in Fig. 3(c), supports the validity of our analysis of ZF-μSR spectra.

The discovery of Li(Zn,Mn)As sparked extensive researches in this 111 type DMSs^{[28, 32–36]}. Fig. 4 shows only three of these new DMSs with independent spin and charge doping, e.g. Li(Zn,Mn)P, Li(Cd,Mn)P and Li(Zn,Co,Mn)As. Isostructural to Li(Zn,Mn)As^[19], Li(Zn,Mn)P^[28] is also a p-type DMSs with excess lithium providing charge doping. The highest T_C could reaches 34 K. The saturation moment per Mn (M_sat) is about 1μ_B–2μ_B, which is comparable to that in (Ga,Mn)As^[11], Li(Zn,Mn)As^[19], etc. ρ(T) of Li(Zn,Mn)P decreases with increasing temperature, which shows semiconductor behaviors^[28]. Li(Zn,Mn)P is a soft magnetic material with about less than 100 Oe, shown in Fig. 4(a), based on the magnetic hysteresis loop and magnetoresistance curves. Magnetoresistivity ρ_H(T) of Li_1.04(Zn_0.9Mn_0.1)P at different external fields are shown in Fig. 4(b). ρ_H(T) increases monotonically with decreasing temperature, showing a rapid rise below T_C. As shown in the inset of Fig. 4(b), the negative magnetoresistance is far from saturation in rather high magnetic field, in which spin orientation is fully aligned. In this condition, the negative magnetoresistance presumably results from the weak localization effects^[37]. Systematic μSR measurements^[32] also confirmed that the magnetic volume fraction on Li_1.15(Zn_0.9Mn_0.1)P could reaches nearly 100% at 2 K. Compared to Li(Zn,Mn)P, DMSs Li(Cd,Mn)P^[36] with optimum doping exhibits a higher T_C of 45 K as shown in Fig. 4(c). But more than 80% negative magnetoresistance, shown in Fig. 4(d), is a record in this 111 type DMSs. For this new type DMSs, spins and carries are indispensable.

Usually, the carriers are induced in Li site while spins in Zn site in all these above systems. Different from that idea, a new DMSs Li(Zn,Co,Mn)As^[33] was reported in which carriers and spins are both induced in Zn site simultaneously. Fig. 4(e) displays the tempertature dependence of M in H = 100 Oe for Li(Zn_1–x–0.15Co_xMn_0.15)P. There is no difference in ZFC and FC procedures for small coercive fields with the highest T_C of 40 K in LiZn_0.80Co_0.05Mn_0.15P. The resistivity of Li(Zn_1–x–0.15Mn_0.15)P decreases with increase of Co doping. This means much more carries are induced successfully. A field induced insulator-to-metal like transition around T_C can be observed with the external field of 1 T, which is due to the suppression of magnetic fluctuations below T_C. This feature is also observed in Li(Zn,Mn)As system^[19]. Fig. 4(d) demonstrates the Hall resistivity of Li(Zn_0.8Co_0.1Mn_0.1)As at 2 K, which exhibits p-type carriers with concentrations of n ~ 7.74 × 10¹⁹ cm^–3 together with the anomalous Hall effect due to spontaneous magnetization at H = 0.

BaZn₂As₂ is a semiconductor synthetized at high temperature (>900 °C) with the tetragonal ThCr₂Si₂ crystal structure. A new type “122” DMSs (Ba,K)(Zn,Mn)₂As₂ has been synthesized with the Ba²⁺/K¹⁺ substitution (hole carries) and Zn²⁺/Mn²⁺ (spins) doping. Fig. 5(a) shows the temperature dependence of M in H = 500 Oe for (Ba_0.7K_0.3)(Zn_0.85Mn_0.15)₂As₂ at ZFC and FC procedures with T_C 230 K. The hysteresis curves M(H) are shown in the inset of Fig. 5(a). Fig. 5(b) exhibits the spontaneous magnetization curve under 5 Oe of (Ba_0.7K_0.3)- (Zn_0.85Mn_0.15)₂As₂, showing T^3/2 dependence in low temperature, as expected for a homogeneous ferromagnet^[38]. Volume fraction of regions with static magnetic order, estimated by μSR measurements in ZF and WTF of 50 Oe are shown in Fig. 5(c). The μSR results indicate that static magnetic order develops in the entire volume with a sharp onset around T_C. Resistivity of (Ba_1–xK_x)(Zn_1–yMn_y)₂As₂ for with several different charge doping levels are shown in Fig. 5(d). For BaZn₂As₂ semiconductor, doping K atoms into Ba sites introduces hole carriers, leading to metallic behavior in (Ba,K)Zn₂As₂. The resistivity curves of (Ba_1–xK_x)(Zn_1–yMn_y)₂As₂ for selected values of x up to 0.3, exhibit a small increase at low temperatures due to spin scattering effect caused by Mn dopants. This variation of resistivity is often observed in heavily doped semiconductors^[39]. Strictly metallic behavior (with monotonic decrease of resistivity with decreasing temperatures) is not a precondition of having a ferromagnetic coupling between Mn moments mediated by RKKY interaction^{[12, 30]}. Fig. 5(e) shows the magnetoresistance curve measured in the external field up to 7 T at different temperatures, showing obvious negative magnetoresistance below T_C. Fig. 5(f) shows the Hall effect results of (Ba_0.85K_0.15)(Zn_0.90Mn_0.10)₂As₂ at several temperatures with hole concentration about 4.3 × 10²⁰ cm^–3.

Different from (Ba,K)(Zn,Mn)₂As₂, another “122” type DMSs with hexagonal CaAl₂Si₂ was reported subsequently, such as (Ca,Na)(Zn,Mn)₂As₂^[40], (Sr,Na)(Zn,Mn)₂As₂^[41] and (Sr,Na)(Cd,Mn)₂As₂^[42], etc. Fig. 6(a) shows the temperature dependence of M in H = 500 Oe for (Ca_0.9Na_0.1)(Zn,Mn)₂As₂ with several different charge doping levels at ZFC and FC procedures. The highest T_C is 33 K. In Fig. 6(b), the temperature dependence of the volume fraction of regions with static magnetic order, derived from μSR measurements in ZF, is consistent with that of spontaneous magnetization under 5 Oe. The latter shows T^3/2 dependence in low temperature expected for a homogeneous ferromagnet^{[22, 30, 38]}. Due to the competition between nearest-neighbor antiferromagnetic interactions and ferromagnetic interactions from remote Mn moments, M_sat per Mn decreases with increasing Mn concentration. μSR results, shown in Fig. 6(d) are also consistent with the spontaneous magnetization under 10 Oe. This means (Sr_0.8Na_0.2)(Zn_0.85Mn_0.15)₂As₂ is also a homogeneous ferromagnet with almost 100% ordered volume fraction at low temperatures. Fig. 6(e) shows the temperature dependence of M in H = 500 Oe for (Sr_1–xNa_x)(Cd_1–xMn_x)₂As₂ with several different charge doping levels x at ZFC and FC procedures with the highest T_C 13 K. Magnetotrasport measurements performed on (Sr_0.8Na_0.2)(Zn_0.8Mn_0.2)₂As₂ at 2 K under the field of up to 7 T are shown in Fig. 6(f). The negative magnetoresistance reached –23% at 2 K and 7 T Taking the orbital effect into consideration, the negative magnetoresistance data are fit by kB^1/2 rule, indicating weak localization magnetoresistance at low temperature. The similar phenomenon is found in (Ga,Mn)As system^{[37, 43]}.

Isostructural to the well-studied iron-based superconductor LaFeAs(O_1–xF_x)^[44], a new kind of “1111” DMSs were reported after the “111” and “122” DMSs, e.g. (La,Ca)(Zn,Mn)SbO^[45] and (Ba,K)F(Zn,Mn)As^{[46, 47]}, etc. Fig. 7(a) shows the M(H) for (La_0.95Ca_0.05)(Zn_0.9Mn_0.1)SbO measured at 25 and 100 K, respectively. An abrupt increase of magnetization at 25 K suggests that a ferromagnetic state and paramagnetic state transition (T_C) occurs. The highest T_C reaches 40 K with proper doping in (La,Ca)(Zn,Mn)SbO material. The resistivity increases monotonically with increasing Mn concentration, suggesting that scattering center of Mn. The resistivity of (La_0.95Ca_0.05)(Zn_0.925Mn_0.075)SbO is shown in Fig. 7(b), with magnetic field up to 5 T. A negative magnetic resistance is clearly observed in a wide temperature region. This behavior can be well explained by the field suppression of the spin fluctuation^[48]. Fig. 7(c) shows the M(T) in ZFC and FC procedures under 500 Oe for the (Ba_0.8K_0.2)F(Zn_1–yMn_y)As samples with y = 0.025, 0.05, 0.075, 0.1 and 0.15, respectively. The highest T_C reaches 30 K for optimal Mn doping (y = 0.1). Above T_C, the susceptibility is fit to Curie-Weiss law as shown in the inset of Fig. 7(c), which indicates a ferromagnetic interaction between Mn²⁺. The temperature dependence of the fast relaxation rate Λ is plotted in Fig. 7(d), exhibiting a monotonic increase with decreasing temperature and reaching a maximum value at the lowest measurement temperature (2 K). The relationships between hole concentration and T_C of “111”, “122” and “1111” DMSs and other diluted ferromagnetic semiconductor systems are plotted in Fig. 7(e)^{[19, 20, 22, 28, 36, 45, 49, 50]}. From Li(Zn,Mn)P to (Ba, K)(Zn,Mn)₂As₂, T_C is considerably improved. As the Zener model predicted^{[51, 52]}, the ferromagnetic ordering in DMSs is mediated by hole carriers, and the Curie temperature is positive correlated with hole concentration.

The origin of magnetic ordering on DMSs is still full of debates^[53–56] meanwhile the general understanding for most common (III,Mn)V–based DMSs invokes As-derived valence-band states^{[57, 58]} as mediators of ferromagnetic interactions. As for Mn-doped II–II–V semiconductors (Ba_1–xK_x)(Zn_1–yMn_y)₂As₂, the details of the electronic structure and the role of As mediating (hole) states remain unresolved both in experiment and theory. A theory^[59] predicts that the competition between the short-range antiferromagnetic (superexchange) interaction and thelonger-range ferromagnetic interaction mediated by the itinerant holes determines the final ground state of BZA. Thus it is of great significance to verify the electronic states in (Ba,K)(Zn,Mn)₂As₂.

Fig. 8(a) exhibits the X-ray absorption spectroscopy (XAS) measurements^[60] of the Mn L_2,3 edge on (Ba_0.7K_0.3)(Zn_0.85Mn_0.15)₂As₂. The line shapes of (Ba_0.7K_0.3)(Zn_0.85Mn_0.15)₂As₂ are similar to (Ga_0.922Mn_0.078)As^[61] and (Ga_0.958Mn_0.042N)^[62], which indicates that the valence of Mn atoms is 2+ and that Mn 3d orbitals strongly hybridize with the As 4p orbitals as in (Ga,Mn)As and (Ga,Mn)N system. The strength of hybridization is weaker than (Ga,Mn)As but stronger than (Ga,Mn)N according to the shoulder structures around hν = 640 and 643 eV. The line shape shows a more localized nature than the Mn metal^[63] and metallic Mn doped into BaFe₂As₂^[64]. Different from LaMnO₃ and MnO^[65], (Ba_0.7K_0.3)(Zn_0.85Mn_0.15)₂As₂ does not show the clear multiple structures, consistent with the semi-metallic conductivity in Mn-BaZn₂As₂^[20]. Figs. 8(b)–(d) display the electronic structure of (Ba_0.904K_0.096)(Zn_0.805Mn_0.195)₂As₂ by soft X rays angle-resolved photoemission spectroscopy (ARPES)^[66]. The results clarify the host valence-band electronic structure is primarily from the As 4p states. Two hole pockets around the Γ point explain the metallic behavior. The impurity band is well below the valence-band maximum (VBM), unlike that in (Ga,Mn)As, which is around the VBM. We conclude that the strong hybridization between the Mn 3d and the As 4p orbitals in (Ba,K)(Zn,Mn)₂As₂ plays a key role in creating the impurity band and inducing high temperature ferromagnetism.

Compared with polycrystals, single crystals are ideal research platforms due to fewer defects. (Ba,K)(Zn,Mn)₂As₂ single crystal^{[67, 68]} is grew with flux method as shown in the inset of Fig. 9(a). Fig. 9(b) demonstrates the magnetization curves in different directions with T_C of about 50 K. Large anisotropic behavior are shown in Fig. 9(c) with easy axis along c. The semiconductor behavior on Fig. 9(d) arises from the localization effect^[69]. Based on the Magnetoresistance R_xx and Hall effect R_xy measurements, the hole carrier density increase from 2.82 × 10²⁰ cm⁻³ at 2 K to 4.80 × 10²⁰ cm⁻³ at 130 K. The arise is from the enhanced thermal excitation from the impurity band to the conduction band.

Andreev refection spectroscopy (AR spectroscopy) is commonly used to achieve spin polarization (P) in various materials, e.g., (Ga,Mn)As^[70], (Ga,Mn)Sb^[71], (La,Sr)MnO₃^[72], CrO₂^[73], EuS^[74], and HgCr₂Se₄^[75]. A schematic view of the BZA/Pb heterojunction is shown in Fig. 10(a). A “clean” interface, represented by the parameter Z, is a crucially required in spectral analysis. For this the experiment, the small Z value (Z = ) implies the manifestation of a clean and transparent interface between BZA single crystal and Pb film. One key parameter for analysis is the differential conductance, G(V) = dI(V)/dV. It is measured as a function of dc-bias voltage (V) crossing the AR junction. Fig. 10(b) shows normalized differential conductance G/G₀ spectra (red dot) and their best fits to the modified Blonder–Tinkham–Klapwijk (BTK)^[76] theory (blue line) at 1.7 K. Spin polarization of 66% is obtained. Besides, about 40%–60% spin polarization is also achieved in a new DMSs (Ba,Na)(Zn,Mn)₂As₂^[77]. The success on AR junction paves a solid route to fabricate multilayer heterojunctions based on BZA.

Compared to classical DMSs, new type of DMSs has one great advantage, existence of numerous isostructual function materials. As shown in Fig. 11, (Ba,K)(Zn,Mn)₂As₂ shares the same tetragonal ThCr₂Si₂ type structure with semiconductor BaZn₂As₂, antiferromagnetic BaMn₂As₂^[78] and superconductor (Ba,K)Fe₂As₂^[79]. The negligible mismatch of lattice constants in the a-b plane (less than 5%)^[20] makes the above materials promising to fabricate multilayer functional heterojunctions. These heterojunctions should have near perfect interface that enable deep insight of many new physical pheromones and physical rules. Recently, a series of single-phased, single-oriented thin (Ba,K)(Zn,Mn)₂As₂ films are successfully fabricated by pulsed laser deposition (PLD) on different substrates, e.g. LSAT, SrTiO₃, LaAlO₃, Si and MgAl₂O₄. A n-type polycrystalline BZA DMSs Ba(Zn,Co)₂As₂^[80] was reported recently. All these results offers much research room for many kinds of heterojunctions in the future^[4].

There are three main group of new diluted magnetic semiconductors with independent charge and spin doping, i.e. the “111”, “122” and “1111” type, as listed in Table 1. BZA is unique in that it has T_C above 200 K. Note that there are other four DMSs which are isostructural to BZA but have low T_C. What makes BZA so special is one open question. The accurate answer to the question could be the clue to seek mechanism of DMSs and the guiding light to search for room temperature ferromagnetic DMSs.

Taking the advantage of lattice matched superconductors, semiconductors and antiferromagnetic compounds, these new types of DMSs have potential to fabricate multilayer heterojunctions^{[3, 4]}. In the roadmap^[81] for spintrionic materials, BZA is proposed as one of the most promising DMSs materials.Two major research fields are recommended for future studies of BZA: (I) to search for Curie temperature higher than room temperature based on the enhancing ferromagnetism interactions; (II) to fabricated isostructural DMSs junctions with various lattice matched materials.

We would like to thank the collaborators Prof. Y. J. Uemura, Prof. Yong Qing Li, Prof. A. Fujimori, Prof. S. Maekawa, Prof. Bo Gu, Prof. Lixin Cao, et al. We are grateful to Prof. L. Yu, F. C. Zhang, J. H. Zhao and F. L. Ning for the helpful discussions. This work is financially supported by Ministry of Science and Technology of China (Nos. 2018YFA03057001, and 2017YFB0405703), National Natural Science Foundation of China through the research projects (No. 11534016).