Diluted magnetic semiconductors (DMS) have capability to control spins and charges. Thus, these materials have great potential applications in spintronic devices^[1−5]. To date one of the major issues for DMS is to obtain high Cuire temperatures (T_C)^[6]. The ferromagnetic ordering in a DMS is mediated by its carriers, thus techniques to increase the carrier density have been widely used to enhance T_C of a DMS^[7−9]. For the parent phases of these DMS, their high-pressure structures and pressure-induced phase transitions have been extensively studied^[10]. For example, GaAs and other Ⅲ−Ⅴ semiconductors undergo a transform from zincblende structure (ambient phase, space group F-43m) to denser phase upon compression^[11]. GaAs undergoes a direct-to-indirect bandgap transition at ~5 GPa^[12]. At higher pressure of about 16 GPa, it collapses into a Cmcm structure. Below the transition pressure, the lattice constant a decrease monotonically with increasing pressure. Because of the fix atom sites of Ga (4a: 0, 0, 0) and As (4c: 1/4, 1/4, 1/4), the bond length of Ga−As keeps decreasing without any anomaly^[13]. External pressure can extend electronic band and then modify the band configuration around Fermi energy without any impurity doping. Thus, pressure effects on DMS have been extensively studied in classic Ⅲ−Ⅴ based DMS, e.g. (Ga,Mn)As, (Ga,Mn)Sb, and new type DMS e.g., Ⅱ−Ⅱ−Ⅴ based (Ba,K)(Zn,Mn)₂As₂^[14−22]. However, both positive and negative pressure dependent T_C were reported.

Crystal structure is one the fundamental factors to determine the materials’ properties^[23]. In particularly, the samples are isolated from external conditions during compression. Then structure is the only variable factor to induce changes of band structure and then other properties. Thus, it is essential to study structural parameters and the possible phase transition under external pressures. Aforementioned (Ba,K)(Zn,Mn)₂As₂ belongs to a new type of DMS with independent carrier and spin doping^{[24, 25]}, which is a key feature distinguished from (Ga,Mn)As^[26]. Its parent material BaZn₂As₂ has two phases. The orthorhombic phase is stable at ambient temperature and pressure, while the tetragonal one is stable at high temperature or high pressure^[18]. The latter can also be stabilized by doping at ambient conditions. Heavier doping levels produce more stable tetragonal lattice.

Our previous work focused on heavy doped samples, e.g., (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂, which is stable up to 25 GPa^[18]. On the other hand, light doped samples are also worth investigation because they are close to the boundary of the two phases and could generate rich transformations under compression. In this work, we study the high-pressure effects on the crystal structure of Ba(Zn_0.95Mn_0.05)₂As₂. An isostructural phase transition occurs at around 19 GPa. Below the transition pressure, Zn/Mn−As bonds are lengthened by anisotropic compression on Zn/MnAs layers. After the transition, strong interlayer As−As dimers form, and lead to shortened Zn/Mn−As bonds by compensation the anisotropic compression on the Zn/MnAs layers.

The high-pressure synchrotron X-ray diffractions (SXRD) were performed at room temperature at beamline 16 BM-D of the APS (at Argonne National Laboratory) using Mao-Bell symmetric diamond anvil cells (DAC) with 300 μm culets anvils. Tiny ruby grains were loaded as pressure markers. The wavelength of X-ray was 0.4246 Å, and SXRD rings were collected with a mar3450 image plate detector. The patterns were transferred with FIT2D software. Rietveld refinements were performed using the GSAS software package^[27].

In-situ high-pressure SXRD measurements were conducted with pressure up to 46.9 GPa. At ambient condition, the crystal structure of Ba(Zn_0.95Mn_0.05)₂As₂ belongs to the ThCr₂Si₂-type structure as shown in Fig. 1(a). It is worth noting that As (0, 0, z) is the only variable atom site in Ba(Zn_0.95Mn_0.05)₂As₂ lattice. Ba and Zn/Mn are fixed to the sites of (0, 0, 0) and (0, 1/2, 1/4) respectively. Fig. 1(b) shows the SXRD patterns of title material under pressures from 0 to 43.4 GPa. Within the pressure range, all the peaks can be indexed into space group of I4/mmm (no. 139) within the measuring pressure range, indicating the preservation of the ambient structure. With increasing pressures, all the diffraction peaks shift to higher angles, indicating lattice shrinkage as expected. Nevertheless, one could roughly notice that low pressures move the peaks faster than high pressures do. Such changes can be clearly observed in Fig. 1(b), where blue and red dash lines are eye-guides of peaks shifts under low and high pressures respectively. Since pressures are increased in a nearly constant rate (about 2 GPa), the apparent slope differences between blue and red dash lines indicate an isostructural phase transition occurs around 19 GPa.

To obtain precise structure parameters, the Rietveld refinements were performed on the entire SXRD patterns. The refinement results at the pressures of 1.0 and 43.4 GPa are present in Figs. 2(a) and 2(b) as two typical examples. The lattice parameters under all the pressures are listed in Table 1. R_wp and R_p are two key parameters to judge the quality of a refinement. R_wp and R_p of all the refinements are less than 4% and 2%, respectively, indicating reliable fittings. The lattice constants a, c and their ratios (c/a) under all the pressures are plotted in Fig. 2(c). It is clear that decrease of a upon compression becomes slower over around 19 GPa, while c shows steady shrink rate. Correspondingly, c/a versus pressure displays a distinct downturn at 19 GPa. These features indicate that the high-pressure phase is a collapsed ThCr₂Si₂-type structure^[28]. This isostructural phase transition is further evidenced by the change of bulk modulus. Fig. 2(d) presents the pressure dependent unit cell volume (V(P)). Without any phase transition, the V(P) curve can be well fitted by the third-order Birch-Murnaghan equation of states,

1P=32B0[(V0V)7/3−(V0V)5/3]×{1−(3−34×B0')×[(V0V)2/3−1]}.

The low-pressure fitting becomes divergent from high-pressure data over about 19 GPa as shown by the red curve in Fig. 2(d). Thus, the high-pressure data are fitted independently. The two fittings yield B0low−P = 47.3 GPa, B0′low−P = 5.4, B0high−P = 60.5 GPa and B0′high−P = 4.1. As expect, the bulk modulus of high-pressure phase is larger than that of low-pressure one.

For materials with ThCr₂Si₂-type structure, the bond angles and bond length in the CrSi layer are essential to their properties. The pressure dependent Zn/Mn−As bond length (L_Zn/Mn−As) is plotted in Fig. 3(a). Chemical intuition tells us that bond in compressed materials should be shorter than those under ambient pressure. Surprisingly, Fig. 3(a) shows that Zn/Mn−As bond is extended upon compression within low-pressure phase. Such counterintuitive bond lengthening was rarely reported before our work. It was neither found in spin and charge doped (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂ that is isostructural to Ba(Zn_0.95Mn_0.05)₂As₂. Above 19 GPa, Zn/Mn−As bond length is reduced with increasing pressures. Eventually, L_Zn/Mn−As bond at 46.9 GPa is shorter than ambient one.

The As−Zn/Mn−As bond angles (α, β as marked in Fig. 1(a)) are highly correlated with Zn/Mn−As bond length within Zn/MnAs₄ tetrahedra. In an idea tetrahedron, α = β ≈ 109.5°. The pressure dependent α and β are plotted in Fig. 3(b). Both decrease of α and increase of β indicate that the Zn/MnAs layer is anisotropic compressed. Namely, compression along ab-plane is stronger than that along c-axis. A similar but slighter anisotropic compression was observed in aforementioned (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂, where α decrease from 108° (0 GPa) to 105° (25 GPa) with a rate of Δα/ΔP ≈ −0.13°/GPa. On the other hand, low-pressure phase Ba(Zn_0.95Mn_0.05)₂As₂ has significantly larger Δα/ΔP ≈ −1.04°/GPa. Above 20 GPa, α starts enlarging with increasing pressures, indicating compression along c-axis becomes dominating. At 46.9 GPa, α = 96.7° and β = 116.2° as shown in Fig. 4.

Interlayer As−As distance (d_As−As) has received much attention in the previous works of (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂, because the band gap is strongly influenced by interlayer As−As hybridization. Theoretical calculations showed this hybridization contribute As 4p_z states near the valence band maximum even under ambient pressure with d ≈ 3.7 Å. It was proposed that decease of d is related with the semiconductor-to-metal transition in compressed (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂. In Ba(Zn_0.95Mn_0.05)₂As₂, the change of d_As−As with pressure is even more substantial with Δd/ΔP ≈ −0.081 Å/GPa below 20 GPa, which is about two times larger than (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂ (−0.028 Å/GPa).

Additional in-situ resistivity measurements are required to clarify this scenario. Here we propose a hypothesis to explain the differences between Ba(Zn_0.95Mn_0.05)₂As₂ and (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂. They crystalize into a layered structure, so interlayer interaction could be an important factor for local structural changes. It is clear that Ba-layers are 2+ and Zn/MnAs layers are 2−. Furthermore, the Zn/MnAs-layer itself is a As^3-−Zn/Mn²⁺−As^3- sandwich-like stacking along c-axis (Fig. 2(d)). Ref. [29] stated that Ba 5d and As 4p states are nonbonding states due to the lattice symmetry. Thus, the interaction between As-layers and Ba-layers is Coulomb attraction in principle. The Coulomb attraction always try to pull As-layers toward Ba layers, namely away from Zn/Mn-layers, and make interlayer As−As closer (as marked by the red arrows in Fig. 2(d)). With 20% K doping into Ba-site, the charge amount of Ba/K layer decreases. As evidenced by X-ray absorption measurements, Mn keeps 2+ in K doped (Ba,K)(Zn,Mn)₂As₂, the charge amount of As also decreases^[30]. Compare to (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂, Ba(Zn_0.95Mn_0.05)₂As₂ has stronger attraction to pull As toward Ba layer. The fact that interlayer As−As distance of (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂ and (Ba_0.8K_0.2)(Zn_0.95Mn_0.05)₂As₂ are 3.7 and 3.9 Å respectively supports this scenario^[18].

Once the lattice is perturbated by external pressures, the Coulomb attraction between As-layers and Ba-layers may become more pronounced to further push As away from Mn and then produced lengthening Zn/Mn−As bond. This trend is slowed down by forming of As−As interlayer bond. Above the transition pressure (19 GPa), the decease of d becomes steady. The possible reason is the forming of strong interlayer As−As bonds or As−As dimers, which compensates the interlayer compression along c-axis as shown in Fig. 4. Phenomenally interlayer As−As bonds provide inner-stresses to resist compression of Zn/MnAs₄ tetrahedra along ab-plane and consequently lead to a near isotropic compression of the tetrahedral.

In this work, we present the crystal structural evolution of the magnetic doped semiconductor Ba(Zn_0.95Mn_0.05)₂As₂ with external pressure up to 46.9 GPa. The slight doped materials crystalize into a ThCr₂Si₂-type structure at ambient pressure and undergo an isostructural phase transition at around 19 GPa. The high-pressure phase can be considered as a collapsed tetragonal lattice as shown in Fig. 4. Before the phase transition, the most remarkable feature is the lengthening of Zn/Mn−As bond within Zn/MnAs layers, since chemical bonds are generally shortened with applying pressures. Anisotropic compression of the Zn/MnAs layers is confirmed by distorted As−Zn−As bond angle. Accompanied with the bond stretch, interlayer As become closer and the As−As dimers eventually formed when the phase transition occurs. With increasing pressures, the strong interlayer As−As bonds compensate the anisotropic compression of the Zn/MnAs layers, and result in the recovery of isotropic compression. In this process, the Zn/Mn−As bond become shortened as expected.