Antiferromagnetic structure and electronic properties of BaCr2As2 and BaCrFeAs2

TL;DR

This study investigates the antiferromagnetic and electronic properties of BaCr2As2 and BaCrFeAs2, revealing their magnetic structures, metallic nature, and the effects of atomic disorder and magneto-structural coupling.

Contribution

It provides detailed experimental and theoretical insights into the magnetic ordering and electronic behavior of these chromium-arsenide compounds, highlighting their potential as parent materials for superconductivity.

Findings

BaCr2As2 exhibits G-type antiferromagnetic order at 580 K.

BaCrFeAs2 shows G-type order below 265 K with small Fe moments.

Both compounds are metallic with enhanced specific heat coefficients.

Abstract

The chromium arsenides BaCr2As2 and BaCrFeAs2 with ThCr2Si2 type structure (space group I4/mmm; also adopted by '122' iron arsenide superconductors) have been suggested as mother compounds for possible new superconductors. DFT-based calculations of the electronic structure evidence metallic antiferromagnetic ground states for both compounds. By powder neutron diffraction we confirm for BaCr2As2 a robust ordering in the antiferromagnetic G-type structure at T_N = 580 K with mu_Cr = 1.9 mu_B at T = 2K. Anomalies in the lattice parameters point to magneto-structural coupling effects. In BaCrFeAs2 the Cr and Fe atoms randomly occupy the transition-metal site and G-type order is found below 265 K with mu_Cr/Fe = 1.1 mu_B. 57Fe Moessbauer spectroscopy demonstrates that only a small ordered moment is associated with the Fe atoms, in agreement with electronic structure calculations with mu_Fe ~ 0. The temperature dependence of the hyperfine field does not follow that of the total moments. Both compounds are metallic but show large enhancements of the linear specific heat coefficient gamma with respect to the band structure values. The metallic state and the electrical transport in BaCrFeAs2 is dominated by the atomic disorder of Cr and Fe and partial magnetic disorder of Fe. Our results indicate that Neel-type order is unfavorable for the Fe moments and thus it is destabilized with increasing iron content.

Tables

Table 1. Table 1: Results of the crystal structure refinements for BaCr 2 As 2 and BaCrFeAs 2 . Powder neutron diffraction data were collected at T = 2 𝑇 2 T=2  K on the instrument E9. The refinements of the data sets were carried out in the tetragonal space group I ​ 4 / m ​ m ​ m 𝐼 4 𝑚 𝑚 𝑚 I4/mmm . The isotropic temperature factors of the different atoms were constrained to be equal.

BaCr2As2 at 2 K, $R_N$ = 0.0480
$a=3.9503\left(2\right)$ Å, $c=13.6047\left(10\right)$ Å, $Z=2$, $V=212.30\left(3\right)$ Å3
atom	site	$x$	$y$	$z$	$B$ (Å2)
Ba	$2a$	0	0	0	0.32(3)
Cr	$4d$	0	$\frac{1}{2}$	$\frac{1}{4}$	0.32(3)
As	$4e$	0	0	0.36092(14)	0.32(3)
BaCrFeAs2 at 2 K, $R_N$ = 0.0402
$a=3.9793\left(1\right)$ Å, $c=13.2532\left(4\right)$ Å, $Z=2$, $V=209.86\left(1\right)$ Å3
atom	site	$x$	$y$	$z$	$B$ (Å2)
Ba	$2a$	0	0	0	0.19(2)
Cr	$4d$	0	$\frac{1}{2}$	$\frac{1}{4}$	0.19(2)
Fe	$4d$	0	$\frac{1}{2}$	$\frac{1}{4}$	0.19(2)
As	$4e$	0	0	0.35779(10)	0.19(2)

Table 2. Table 2: Spin magnetic moment of the Cr atoms, the DOS at the Fermi energy n ​ ( ϵ F ) 𝑛 subscript italic-ϵ F n(\epsilon_{\mathrm{F}}) , and the coefficient of the electron specific heat γ 0 subscript 𝛾 0 \gamma_{0} for different calculation schemes for G 𝐺 G -type afm ordered BaCr 2 As 2 . All calculations were performed including spin-orbit interaction (SO).

	$U$	$m_{\mathrm{Cr}}$	$n\left({ϵ}_F\right)$	${\gamma }_0$
method	eV	${\mu }_B$	eV-1	mJ mol-1 K-2
GGA with SO	0	2.4	3.37	7.95
LDA$+U$	1	2.8	3.67	8.65
	2	3.1	2.45	5.77
	3	3.3	2.91	6.85
	4	3.4	2.85	6.71
	5	3.5	1.84	4.35

Table 3. Table 3: Magnetic properties of BaCrFeAs 2 . Tabulated are the DOS just above ϵ F subscript italic-ϵ F \epsilon_{\mathrm{F}} ( n max subscript 𝑛 max n_{\mathrm{max}} ), the absolute values of the magnetic moments at Cr and Fe species ( m Cr / Fe subscript 𝑚 Cr Fe m_{\mathrm{Cr/Fe}} ), and the calculated Néel temperatures ( T N , calc subscript 𝑇 N calc T_{\mathrm{N,calc}} ).

	ordered	disordered
$n_{\max }$	6.92	4.99	(eV-1)
$m_{\mathrm{Cr}}$	2.74	2.76	(${\mu }_B$)
$m_{\mathrm{Fe}}$	2.53	0.00	(${\mu }_B$)
$T_{N,\mathrm{calc}}$	965	260	(K)

Full text

I Introduction

The discovery of high-$T_{\mathrm{c}}$ superconductivity in iron-based pnictides and chalcogenides has initiated intense efforts to investigate and theoretically model the magnetic states adopted by the non-superconducting relatives of these compounds [Bascones et al., 2016; Dai, 2015; Si et al., 2016]. Similar as for the oxocuprate superconductors, there is a subtle interplay between magnetism and superconductivity, and spin-fluctuations are considered as the mechanism enabling the formation of Cooper pairs [Fernandes and Millis, 2013]. The possible coexistence of specific types (spin-density wave, SDW) of magnetic order and superconductivity on the microscale in the Fe-based systems is an ongoing topic of research.

The antiferromagnetic (afm) parent compounds of cuprate superconductors are Mott (charge transfer) insulators. The situation is less clear for their Fe-based counterparts. For instance, non-superconducting pnictides like LaOFeAs or BaFe2As2 are metallic, itinerant antiferromagnets with low ordered moments at the iron sites ($<1\mu_{\mathrm{B}}$), whereas chalcogenides like K0.8Fe1.6Se2 are insulators with large Fe moments of $\approx 3\mu_{\mathrm{B}}$. Strictly, however, half-filled systems with $d^{5}$ configuration have to be regarded as the Mott-type parent compounds, which is confirmed by the semiconducting properties of e.g. BaMn2As2 [An et al., 2009; Singh et al., 2009a, b; Johnston et al., 2011; Zhang et al., 2016] as well as of LaOMnAs and LiMnAs [Beleanu et al., 2013; Dong et al., 2014].

There is an ongoing debate about the proper description of the electronic structure of Fe-based pnictides and chalcogenides, in particular about the importance of electron correlation and the degree of electron itinerancy [Dai et al., 2012; Mannella, 2014]. Since the basic Fe-As layers with formally Fe2+ ions in the parent compounds constitute a multi-orbital system, orbital ordering is also believed to be of importance. Two opposing views have been adopted: On the one hand, the iron pnictides have been considered as weakly correlated metals where a SDW magnetic state is formed due to Fermi surface nesting. On the other hand, in particular the more strongly correlated selenides have been discussed as Mott-type insulators and the magnetism was modeled in terms of Heisenberg-type exchange interactions between localized moments. However, none of the limiting views can describe the many facets of magnetism which have emerged from experimental studies and actually it is believed that the Fe-based pnictides and chalcogenides are in between these extremes. For instance, even itinerant electronic systems may show pronounced correlation effects arising from Hund’s rule coupling (Hund’s metals [Si et al., 2016; Backes et al., 2015]), which rationalizes the existence of large magnetic moments on short timescales of the order of femtoseconds, as derived from photoemission experiments [Mannella, 2014].

An important prototype system is BaFe2As2 which adopts the tetragonal ThCr2Si2-type crystal structure at room temperature and features a concomitant structural and magnetic transition at $T_{\mathrm{N}}=132$ K [Rotter et al., 2008]. Below $T_{\mathrm{N}}$ a stripe-like afm order occurs with saturated Fe moments of about $0.9\mu_{\mathrm{B}}$ being aligned along the $a$ axis of the low-temperature orthorhombic crystal structure [Huang et al., 2008]. BaFe2As2 can be converted into a high-$T_{\mathrm{c}}$ superconductor by hole or electron doping [Rotter et al., 2008; Chen et al., 2008; Sasmal et al., 2008] or by high pressure [Alireza et al., 2008]. While substitution of Fe by some transition metals like Co, Ni or even Ru [Sefat et al., 2008; Leithe-Jasper et al., 2008; Schnelle et al., 2009] induces superconductivity, substitution by others like Mn does not lead to superconducting states [Kasinathan et al., 2009]. In contrast to BaFe2As2, the half-filled Mn-analogue BaMn2As2 remains tetragonal down to low temperatures and adopts a $G$-type afm spin structure with a high ordered moment ($3.9\mu_{\mathrm{B}}$) and a high ordering temperature of 625 K [Singh et al., 2009b; Johnston et al., 2011]. Similarly high Néel temperatures (692–758 K) and $G$-type afm have been observed for the Cr species in the isostructural silicides $R$Cr2Si2 ($R$ = Tb, Ho, Er) [Moze et al., 2003].

Superconductivity in manganese and chromium compounds is very rare and only recently representatives have been found. The first, binary CrAs, has been long known for its structural and helimagnetic ordering transition at 240–190 K with sizable Cr moments of $1.67\mu_{\mathrm{B}}$ [Boller and Kallel, 1971]. The ordering temperature decreases dramatically to zero for pressures of 0.7–0.8 GPa and superconductivity with critical temperature up to 2.2 K appears [Wu et al., 2014; Kotegawa et al., 2014]. At ambient pressure superconductivity has been investigated in the new $A_{2}$Cr3As2 ($A$ = K, Rb, Cs) compounds featuring quasi onedimensional Cr3As3 tubes [Bao et al., 2015; Cao et al., 2016].

The Mott scenario for the transition-metal arsenides [Ishida and Liebsch, 2010] considers a mirror-symmetry in the many-body physics arising when the half-filled $d^{5}$ configuration is either doped by electrons or by holes. The Cr compounds formally have a $d^{4}$ configuration and in this sense are the hole-doped analogues to the electron-doped $d^{6}$ system BaFe2As2. Recently, some theoretical studies explored the possibilities to find superconductivity for Ba$M_{2}$As2 and LaO$M$As with transition-metals $M$ (or mixtures) with less than 5 electrons [Edelmann et al., 2016; Pizarro et al., 2016]. Emphasis in these works is put on the strength of correlations in dependence of the band filling. It is hoped that correlations similar to those in the actual iron-based high $T_{\mathrm{c}}$ materials lead to superconductivity.

Here, we focus on the system Ba(Fe1-xCrx)2As2, where no superconducting compositions have been found so far [Kasinathan et al., 2009; Sefat et al., 2009; Marty et al., 2011]. Theoretical calculations predicted an afm checkerboard ($G$-type) ordering in the end member BaCr2As2 [Singh et al., 2009c] and a ferromagnetic (fm) ground state in atomically ordered BaCrFeAs2 [Singh et al., 2009b]. Another electronic structure calculation predicted ordered BaCrFeAs2 as a fully compensated antiferromagnet with an iron moment of 2.6 $\mu_{\mathrm{B}}$ [Hu and Hu, 2010]. Such a material could be relevant for spintronics applications. Neutron diffraction studies on iron-rich BaFe2-xCrxAs2 single crystals with $0<x<0.94$ indicated that near $x=0.6$ the SDW ground state is replaced by a $G$-type afm state [Marty et al., 2011], but spin structure and ordering temperature of the end member BaCr2As2 have not been studied experimentally yet. Also electronic structure calculations of EuCr2As2 suggested a stable $G$-type afm order of the Cr sublattice (in addition, the Eu2+ ions show fm ordering below 21 K) [Paramanik et al., 2014]. A recent experimental study confirmed these predictions [Nandi et al., 2016].

We investigated the detailed crystal and magnetic structures of BaCr2As2 and BaCrFeAs2 by temperature dependent powder neutron diffraction. Magnetic susceptibility, electrical transport, specific heat, and 57Fe Mössbauer spectroscopy measurements complement the study. Our theoretical studies predict a random occupation of Cr and Fe on the transition-metal site for BaCrFeAs2 and metallic afm $G$-type ordered ground states for both compounds. Experimentally, it is shown that both compounds in fact feature $G$-type afm order, where the Néel temperature $T_{\mathrm{N}}=580$ K of BaCr2As2 is nearly as high as that of BaMn2As2, although the magnetic moment is only half as large. Both compounds are metallic conductors. BaCrFeAs2 turns out to be atomically disordered, and thus only average magnetic moments are obtained from neutron diffraction. Most interestingly, the Fe Mössbauer spectra indicate that the Fe magnetic moments are much smaller than those of Cr and of similar size as in BaFe2As2. $T_{\mathrm{N}}$ decreases with increasing iron content, thus the $G$-type order is unfavorable for the Fe moments. The itinerant character of the magnetism persists in the whole stability range of the $G$-type order, whereas BaMn2As2 with the same nominal $d$ electron count as BaCrFeAs2 is a semiconductor.

II Experimental & Calculation Details

BaCr2As2 and BaCrFeAs2 were synthesized by reacting CrAs and FeAs with Ba in an alumina crucible. The binaries were synthesized according to Singh et al. [Singh et al., 2009c] using Cr (chemPUR, 99.99 %), Fe (chemPUR, 99.9 %), and As (Alfa Aesar, 99.999 %). To control the harsh reaction between the binaries and Ba (Alfa Aesar, 99.999 %), only 1/3 of the barium was added initially. The mixture was heated to 1423 K and held for 12 h. Then the product was ground and additional Ba was added. These steps were repeated until the powder x-ray diffraction pattern (PXRD) did not show any CrAs or FeAs impurities. In both cases an excess of $\approx 5$ % Ba was added to react all of the strongly magnetic CrAs or FeAs. Single crystals were grown using the Bridgman technique with alumina crucibles under argon atmosphere. The crucible was heated to 1823 K and held for 24 h, then it was moved with 1 mm h*-1* out of the hot zone of the furnace. Single crystals (Figure 1b) up to 10$\times$5$\times$4 mm could be grown. For the neutron diffraction experiments $\approx 4$ g of polycrystalline sample was prepared.

The crystal structure (ThCr2Si2 type, $tI10$, space group (SG) $I4/mmm$, No. 139, Ba on $2a$, Cr on $4d$, As on $4e$) was confirmed by PXRD using Cu-Kα radiation ($\lambda=1.54056$ Å), with a Huber G670 camera (Guinier technique). The refined lattice parameters are $a=3.9667$ and $c=13.6214$ Å for BaCr2As2 and $a=3.986$ and $c=13.2939$ Å for BaCrFeAs2. Additional reflexes suggest the presence of traces of BaAl2O4 in some samples. High-resolution transmission electron microscopy (HR-TEM) microstructure studies were performed using a FEI Titan 80-300. In the high-angle annular dark field (HAADF) images (shown in Figures 1c and d for BaCr2As2; not shown for BaCrFeAs2) no signs of disorder on the Ba $2a$(0,0,0), Cr/(CrFe) $4d$ ($\tfrac{1}{2},\tfrac{1}{2},\tfrac{1}{2}$) or As $4e$ (0,0,$z$) sites could be found. In agreement with the PXRD pattern, segregates of BaAl2O4 were seen on the surface for some BaCrFeAs2 crystals. The average composition of the crystals was found to be Ba1.03(4)Cr2.01(5)As2.03(5) and Ba1.01(6)Cr0.96(8)Fe1.08(2)As2.10(1) by chemical analysis with inductively coupled plasma optical emission spectroscopy (ICP-OES).

Powder neutron diffraction experiments on BaCr2As2 and BaCrFeAs2 have been carried out on the instruments E2, E6, and E9 at the BER II reactor of the Helmholtz-Zentrum Berlin. The instrument E9 uses a Ge-monochromator selecting the neutron wavelength $\lambda=1.7973$ Å, while the instruments E2 and E6 use a pyrolytic graphite monochromator selecting the neutron wavelength $\lambda=2.38$ Å and 2.42 Å, respectively. On these instruments powder patterns were recorded in the following ranges of diffraction angles: $7.7^{\circ}\leq 2\theta\leq 83.4^{\circ}$ (E2), $7.5^{\circ}\leq 2\theta\leq 136.5^{\circ}$ (E6), and $8^{\circ}\leq 2\theta\leq 141^{\circ}$ (E9). In case of BaCr2As2 powder patterns were collected between 2 and 750 K, for BaCrFeAs2 between 2 and 275 K. The refinements of the crystal and magnetic structure were carried out with FullProf [Rodríguez-Carvajal, 1993] using the nuclear scattering lengths $b$(O) = 5.805 fm, $b$(Cr) = 3.635 fm, $b$(Fe) = 9.54 fm, $b$(Ba) = 5.25 fm [Sears, 1995]. The magnetic form factors of the Cr3+ and Fe3+ ions were taken from Ref. Brown, 1995.

Magnetization was measured in two magnetometer systems (MPMS-XL7 and MPMS3, Quantum Design). Only small differences were seen between data taken in warming after zero-field cooling (zfc) and during cooling in field (fc). For some crystals such differences seem to be due to tiny CrAs precipitations (see below). For one BaCr2As2 crystal the oven option of the MPMS3 was employed. For these measurements the crystal was first measured at low temperatures and then cemented to a heater stick for the high-temperature run. The electrical resistivity, transverse magnetoresistance, Hall effect, and heat capacity were determined with a measurement system (ACT and HC options, respectively, PMMS9, Quantum Design) between 1.8 and 320 K.

Temperature-dependent Mössbauer spectra of BaCrFeAs2 were measured with a standard WissEl spectrometer and a Janis closed cycle refrigerator (SHI-850-5) using a 57Co/Rh source and a drive system operating in constant acceleration mode. The Mössbauer absorber consisted of an acrylic glass sample container with a sample density of $\approx 17$ mg per cm2. In order to ensure homogeneous distribution the sample was mixed with boron nitride. Spectra were collected between 5 and 292 K and evaluated with MossWinn [Mos, ]. Hyperfine field distributions were extracted using the Hesse-Rübartsch method. Isomer shifts are given relative to $\alpha$-iron.

The electronic structures were calculated self-consistently in the local spin density approximation (LSDA) using the full potential linearized augmented plane wave (FLAPW) method including spin-orbit (SO) interaction as implemented in Wien2k [Blaha et al., 1999; Schwarz and Blaha, 2003; Blaha et al., 2014]. For atomically disordered BaCrFeAs2 the calculations were carried out using the full potential, spin polarized relativistic Korringa–Kohn–Rostocker method (SPRKKR) [Ebert, 1999; Ebert et al., 2011]. All calculations were performed using the PBE exchange-correlation functional [Perdew et al., 1996] and the generalized gradient approximation (GGA). To estimate the influence of a static correlation, the LDA$+U$ method was applied.

For BaCr2As2 the calculations were started with the values $a=3.96$, $c=13.6$ Å, and $z=0.361$ [Pfisterer and Nagorsen, 1980]. A $G$-type afm order was assumed and the lattice was set up in SG $I\bar{4}m2$ (119) and the Cr atoms were located on the split $2c$ and $2d$ sites to allow for the anti-parallel orientations of their magnetization. This space group concerns the atomic positions, the final magnetic symmetry depends on the direction of the quantization axis and is described by a Shubnikov color group (here $P\>4^{\prime}/m^{\prime}m^{\prime}m$ (123.344)). A full structural optimization showed that the above given parameters are nearly relaxed, in agreement with previous observations [Singh et al., 2009c].

In the calculations of the electronic structure for BaCrFeAs2 the same basic atomic and magnetic structures were used. Two cases were assumed, an ordered and an alloyed structure. In the ordered structure, the initial $4d$ position of the Cr atom was split and half of the new positions was occupied by Fe. This results in SG $I\bar{4}2m$ (119) with Ba $2a$, Cr $2c$, Fe $2d$, and As $4e$. Starting from the experimental lattice parameters, an optimization of the ordered structure resulted in $a_{\mathrm{opt}}=4.012$ Å, $c_{\mathrm{opt}}=13.3819$ Å, and $z_{\mathrm{opt}}=0.357$. The strong deviations of these optimized parameters from the experimental ones suggest that the ordered case is not realized, i. e. in the actual material Cr and Fe are disordered. For the alloyed structure it was assumed that Fe and Cr occupy both the $2c$ and $2d$ position in a 50:50 ratio.

III Results

III.1 Crystal structures

The crystal structures of BaCr2As2 and BaCrFeAs2 were investigated by powder neutron diffraction. Note that below $T_{\mathrm{N}}=580$ K, respectively 265 K, the patterns also included magnetic Bragg intensity (see Section III.5). Both compounds crystallize in the tetragonal SG $I4/mmm$ (No. 139), where the Ba, Cr(Fe), and As atoms occupy the Wyckoff positions $2a$(0,0,0), $4d$(0,$\tfrac{1}{2}$,$\tfrac{1}{4}$), $4e$(0,0,z), respectively. The same SG was reported earlier for BaMn2As2 [Singh et al., 2009a]. From the data sets collected on the fine-resolution powder diffractometer E9 we could not find any additional peak splitting indicating a lower crystal structure symmetry. The refinements of the crystal structure of BaCr2As2 from data sets recorded in the temperature range from 2 up to 750 K resulted in residuals between $R_{\mathrm{N}}=0.048$ and 0.088 [defined as $R_{\mathrm{N}}=(\Sigma||F_{\mathrm{obs}}|-|F_{\mathrm{calc}}||)/\Sigma|F_{\mathrm{obs}}|$]. These values are somewhat larger than expected. This can be ascribed to the fact, that the investigated sample contained an impurity, which could not be characterized so far (see Figure 2). Further, additional impurity peaks of the sample container were observed in the diffraction patterns using the high temperature furnace on E9. Nevertheless, the lattice parameters, as well as the positional parameters of BaCr2As2 could be refined with good accuracy. In the case of BaCrFeAs2 the crystal structure was investigated in the temperature range between 2 and 275 K. Due to the higher purity of this sample the refinements resulted in smaller residuals between $R_{\mathrm{N}}=0.035$ and 0.043. The results of the Rietveld refinements of the data sets collected at 2 K are given in Table 1. It is emphasized that there are no indications of superstructure reflections which would point to atomic order of Cr and Fe atoms.

Figure 3a shows the temperature dependence of the cell parameters as well as the cell volume. For BaCrFeAs2 the lattice parameters $a$ and $c$ show a continuous increase from 2 up to 275 K, which is above the magnetic ordering temperature $T_{\mathrm{N}}=265(5)$ K. A similar trend was found for BaMn2As2 [Singh et al., 2009a]. A continuous increase of $a$ and $b$ was also observed for BaCr2As2 but only in the range from 2 K up to about 300 K. Further heating up to 700 K leads to a decrease of the $c$ parameter, while the increase of the $a$-parameter becomes somewhat more pronounced (see Figure 3). As a consequence, the ratio $c/a$ decreases from 3.43 at 300 K to 3.34 at 750 K. The former value is in agreement with literature data [Pfisterer and Nagorsen, 1980]. These results indicate that the magnetic transition in BaCr2As2 is accompanied by subtle structural modifications, in contrast to BaMn2As2 [Singh et al., 2009a]. The anomaly in the $c$ parameter is reflected in the change of the distance between the arsenic atoms lying along the $c$ direction. Namely, one observes a decrease of the distance $d$(As-As) between about 300 K and the Néel temperature $T_{\mathrm{N}}=580(10)$ K as shown in Fig. 4. Above $T_{\mathrm{N}}$ finally $d$(As-As) seems to show a slight increase up to 750 K. The bond length $d$(Cr-As) only shows a slight increase from 2 K up to 750 K, but a weak anomaly is observed near $T_{\mathrm{N}}$. On the other hand in BaCrFeAs2 both, $d$(As-As) and $d$(Cr-As), only show a slight increase in the temperature range from 2 K and 275 K without anomalies, indicating that magnetic ordering does not influence the structural properties significantly. In Fig. 4 we have also plotted the changes of the two different bond angles $\angle_{1}$(As-Cr-As) and $\angle_{2}$(As-Cr-As) in the CrAs4 (and Cr/FeAs4) tetrahedra, which form a twodimensional layer in the $ab$ plane. The structural anomalies accompanying the magnetic transition in BaCr2As2 are also reflected in the bond angles. Thus, the bond angles $\angle_{1}$(As-Cr-As) and $\angle_{2}$(As-Cr-As) show a strong increase and decrease in the magnetically ordered state down to 2 K. Above 650 K both bond angles seem to reach saturation values. In contrast, the two bond angles for BaCrFeAs2 are practically unchanged between 2 and 275 K. However, both pnictides do not reach the ideal tetrahedron angle of $109.47^{\circ}$. The CrAs4 tetrahedra in BaCr2As2 are stronger distorted than the Cr/FeAs4 tetrahedra in BaCrFeAs2.

III.2 Electronic structure of BaCr2As2

In Figure 5 the Fermi surface of BaCr2As2 is shown. Three bands are crossing the Fermi energy $\epsilon_{\mathrm{F}}$, resulting in three iso-surfaces. The innermost one is closed and has a cushion shape. It is related to a hole pocket around the $\Gamma$ point. The outer two are open at top and bottom and have corrugated cylinder shapes with fourfold arranged bulges. The effective band mass $m^{\star}/m_{e}$ of the innermost hole pocket was determined for the $\Delta$ and $\Sigma$ directions revealing values of $m^{\star}_{\Delta}=-0.63$ and $m^{\star}_{\Sigma}=-0.39$ at the $\Gamma$ point. This hints that this band is in the GGA-LSDA not responsible for the observed enhanced value of the electronic specific heat coefficient $\gamma$ (cf. Ref. Singh et al., 2009c and our specific heat results in section III.4).

The evolution of the DOS with increasing Coulomb parameter $U$ in the LDA$+U$ calculation is given in Figure 6. Note that the unoccupied minority states of one of the Cr atoms correspond to the unoccupied majority states of the other, and vice versa for the occupied states due to the anti-symmetric spin densities of an antiferromagnet. The increased $U$ results in an increase of the magnetic moment at the Cr atoms (cf. Table 2). Their values are in the range 2.4-3.5 $\mu_{\mathrm{B}}$.

The “bare” coefficient of the electron specific heat $\gamma_{0}=\pi^{2}k_{B}^{2}n(\epsilon_{\mathrm{F}})/3$ for BaCr2As2 is $\approx 8$ mJ mol*-1K-2*. For the various calculation schemes the results are also given in Table 2. The inclusion of SO interaction has only a weak effect on $\gamma_{0}$. For LDA$+U$, it is found that $n(\epsilon_{\mathrm{F}})$ – and thus $\gamma_{0}$ – increases slightly and has a maximum at $U\approx 1$ eV, but then decreases with further increase of $U$. Using SPRKKR and the mean field approximation a Néel temperature $T_{\mathrm{N,calc}}\approx 880$ K can be calculated.

III.3 Electronic structure of BaCrFeAs2

The electronic structure of the ordered variant of BaCrFeAs2 should be a completely compensated half-metallic ferromagnet as reported in Reference [Hu and Hu, 2010]. The peculiarity of this type of magnetism is that the moment of two different types of atoms – here Cr and Fe – compensate each other similar to a antiferromagnet. The different symmetry, however, allows that the DOS in the two spin channels are different. Moreover, a band gap at $\epsilon_{\mathrm{F}}$ appears in one of the spin densities similar to a half-metallic ferromagnet. The DOS for the $\uparrow$ spin channel has a pronounced maximum just above $\epsilon_{\mathrm{F}}$ with a value of $n(\epsilon_{\mathrm{F}}+\delta)=6.92$ eV*-1* ($\delta\approx 50$ meV). This value is twice as high as the one calculated for BaCr2As2. It arises from flat bands around $\epsilon_{\mathrm{F}}$. A detailed analysis of the DOS reveals that those states are localized at the Fe atoms. In many cases, compounds with such a peaked DOS at $\epsilon_{\mathrm{F}}$ are not stable.

The electronic structure and the DOS of the alloy variant of BaCrFeAs2 are shown in Fig. 7. The result of the chemical disorder scattering are strongly broadened electronic states, in particular close to $\epsilon_{\mathrm{F}}$. The broadening causes a reduction of the maximum of the DOS just above $\epsilon_{\mathrm{F}}$ to about 5 eV*-1*, which is still higher compared to BaCr2As2.

The magnetic properties of ordered and alloyed BaCrFeAs2 are compared in Table 3. The moment for the atomically ordered phase compares well with 2.6 $\mu_{\mathrm{B}}$ reported in Ref. Hu and Hu, 2010. The moments in the ordered case are not completely the same, the vanishing total moment is guaranteed by the polarization of the interstitial and the atoms in the vicinity of the magnetic Cr and Fe atoms. The orbital magnetic moments are very small, in the order of 0.002 $\mu_{\mathrm{B}}$ or 0.06 $\mu_{\mathrm{B}}$.

Interestingly, for the random distribution of Fe and Cr, a stable solution is found with vanishing moments at the Fe atoms at the same site. The calculation was started with a starting moment at the Cr atoms ($4\mu_{\mathrm{B}}$) but zero moment at Fe. No magnetic moment was induced during the self consistent cycles. The result is an average moment of $1.38\mu_{\mathrm{B}}$ for the $2c$/$2d$ positions.

Also for BaCrFeAs2 the Néel temperature can be calculated in the mean field approximation using SPRKKR. The value for the atomically ordered structure is very high, however, a much lower Néel temperature of about 260 K is obtained for the disordered version.

III.4 Magnetic and thermal properties, electrical transport

The temperature dependence of the magnetic susceptibility, $\chi=M/H$, at $\mu_{0}H=1$ T of a BaCr2As2 crystal platelet ($ab$ plane) was measured with the magnetic field parallel and perpendicular to the $c$ axis (Fig. 8). The small difference between the directions above 600 K suggests a rather weak anisotropy in the paramagnetic regime. However, no Curie-Weiss behavior could be observed up to $T=800$ K. For $H\parallel c$ a clear kink is visible, indicating the afm ordering of the Cr sub-lattice at $T_{\mathrm{N}}=575(10)$ K, as determined from the peak in the derivative d($\chi T$)/d$T$. The decrease of $\chi(T)$ for $H\parallel c$ towards low temperatures is much stronger than for $H\perp c$, eventually suggesting an afm order with magnetic moments aligned along the $c$ axis. The behavior of $\chi(T)$ and the ordering temperature is very similar to that of BaMn2As2 [Johnston et al., 2011]. The nearly isotropic but non-Curie-Weiss behavior above $T_{\mathrm{N}}$ is observed for several afm compounds with ThCr2Si2-type structure. It is typical for the somewhat twodimensional character of the magnetic interactions and has been treated in detail theoretically for BaMn2As2 [Johnston et al., 2011]. For BaCr2As2 the smooth maximum of $\chi(T)$ may be anticipated for temperatures around 900 K.

A small upturn of $\chi(T)$ in both directions for temperatures below $\approx 50$ K suggests the presence of paramagnetic impurities, e.g. from point defects. Magnetization loops taken at $T=1.8$ K (Fig. 8 inset) show a very small fm-like component of $<5\times 10^{-4}\mu_{\mathrm{B}}$ for both field along or perpendicular to the $c$ axis. Also, in measurements of $M(T)$ at $\mu_{0}H=0.01$ T, a small sharp transition from a fm impurity phase becomes visible at $T_{imp}\approx 76$ K. We assign this transition and the fm-like signals to the structural and helimagnetic ordering transition of CrAs [Boller and Kallel, 1971]. The ordering temperature of CrAs is known to decrease dramatically to zero for pressures of 0.7-0.8 GPa [Wu et al., 2014; Kotegawa et al., 2014]. We speculate that CrAs on the crystal surface is strained and its ordering temperature is therefore reduced to $T_{\mathrm{imp}}$.

The magnetic susceptibility of sintered BaCrFeAs2 (Fig. 9) displays a broadened cusp at $T_{\mathrm{N}}\approx 269(2)$ K (midpoint of step in d($\chi T$)/d$T$), but there is only a slight decrease of the $\chi(T)$ curve below this temperature. Instead, with decreasing temperature the susceptibility increases again, indicating a strong paramagnetic contribution following a Curie law. Above the Néel temperature (range 300-400 K) the data are well fitted by a Curie-Weiss law with effective moment $\mu_{\mathrm{eff}}=3.70\,\mu_{\mathrm{B}}$ and Weiss temperature $\theta_{\mathrm{P}}=-273$ K. The BaCrFeAs2 crystal shows a similar cusp, however at $\approx 225$ K. The lower $T_{\mathrm{N}}$ of the single crystal is probably due to a slightly lower Fe content compared to the sinter sample. The decrease of $\chi(T)$ is more pronounced for $H\parallel c$, suggesting an afm ordered structure with the magnetic moments lying in the crystallographic $c$ direction. The paramagnetic contribution in the magnetically ordered state of BaCrFeAs2 is discussed in connection with the Mössbauer results (Sec. III.6).

The electrical resistivity, $\rho(T)$, of the BaCr2As2 crystal measured for current in the $ab$ plane (Fig. 10) increases almost linearly with temperature and indicates metallic behavior of the compound ($\rho_{\mathrm{300\,K}}=136$ $\mu\Omega$ cm). The residual resistance ratio RRR $\approx 3.8$ indicates a fair quality of the specimen and the residual resistance is already reached at $\approx 13$ K. Interestingly, below this temperature $\rho(T)$ increase very slightly (by $\approx 0.1$ % of $\rho_{0}$). The magnetoresistance (MR, Fig. 10 inset) is very small and positive for high temperatures. Surprisingly, for $T=2.0$ K it is negative and quite strong (-1.7 % at $\mu_{0}H=9$ T). This might indicate the damping of an additional scattering mechanism at very low temperatures compared to the $T_{\mathrm{N}}$. The origin of both the upturn of $\rho(T)$ below 13 K and of the negative MR might be due to a Kondo-hole effect. Non-magnetic atoms replacing magnetic Cr in the afm ordered lattice can give rise to a Kondo-like upturn of resistivity at low temperatures. This scattering is weakened by the application of a magnetic field resulting in a negative MR proportional to the field. The Hall resistivity curves $\rho_{ab}(H)$ are linear, the Hall constants are positive and do almost not vary with temperature. This observation is consistent with the hole pocket around the $\Gamma$ point (cf. Fig. 5). Within a one-band model the Hall constant corresponds to a hole density $n_{\mathrm{h}}\approx 1.5\times 10^{22}$ cm*-3* and low mobility ($b_{\mathrm{h}}=11$ cm2 V*-1* s*-1* at $T=2$ K).

For polycrystalline BaCrFeAs2 (Fig. 10) as well as for a single crystal (not shown) the electrical resistivity is about five time higher. Interestingly, it displays only a weak temperature dependence and increases continuously with decreasing temperature. Similar to BaCr2As2 there is an upturn below a certain temperature (here, $\approx 40$ K). The absence of a typical metallic resistivity behavior is probably due to scattering of charge carriers on the disordered Cr/Fe species of the $4d$ Wyckoff site. The order of magnitude suggests however that BaCrFeAs2 is still a metal. Nominally isoelectronic BaMn2As2 crystals grown in MnAs flux show an about 100 times higher resistivity [Singh et al., 2009a]. The magnetoresistance of BaCrFeAs2 is small ($<0.15$ %) for all temperatures, indicating that magnetic scattering due to the atomic Cr/Fe disorder is unimportant. Also for BaCrFeAs2 the Hall isotherms $\rho_{xy}(H)$ are linear, however Hall constants are negative and vary with temperature. They correspond to electron densities $n_{e}=8.3\times 10^{21}$ cm*-3* at $T=300$ K and only $1.5\times 10^{21}$ cm*-3* at 2 K with $b_{e}=6$ cm2 V*-1* s*-1*.

The specific heat, $c_{p}(T)$, of the two compounds is presented in Fig. 11. For BaCr2As2 no anomalies from transitions are visible up to 320 K. However, $c_{p}(T)$ for BaCrFeAs2 is larger than that of BaCr2As2 in the covered temperature range which might also be due to the Cr/Fe disorder. A very small steplike anomaly can be seen at $\approx 260$ K (barely standing out of the noise). The transition temperature is in agreement with our magnetization and neutron diffraction data (see Section III.5). The small entropy change connected with this magnetic ordering is due to the low ordered moments of Cr/Fe and the predominantly itinerant character of the magnetic system. The Dulong-Petit limit $c_{p}=3nR$ ($R$ = molar gas constant, $n$ = number of atoms) is reached by both compounds at around room temperature.

At low temperatures the specific heats (see Fig. 11 inset) are well described by $c_{p}(T)=\gamma T+\beta T^{3}+\delta T^{5}$ where $\gamma$ is the coefficient of the linear term $\gamma T$ (usually assigned to conduction electrons) and $\beta T^{3}$ and $\delta T^{5}$ are the first two terms due to the harmonic theory of lattice specific heat. For BaCr2As2 (BaCrFeAs2) least-squares fits in the range 1.9–7 K result in $\gamma=18.8$ (64.9) mJ mol*-1* K*-1*, $\beta=0.51$ (0.56) mJ mol*-1* K*-4* corresponding to an initial Debye temperature of 268 (259) K, and $\delta=0.6$ (0.0) $\mu$J mol*-1* K*-6*. The $\gamma$ value obtained for BaCr2As2 is close to the value deduced by Singh et al. [Singh et al., 2009c], however the linear term of BaCrFeAs2 is very large. The $\gamma$ value for BaCrFeAs2 is confirmed through measurements on the single crystal for which we show the susceptibility in Fig. 9. The lattice properties (Debye temperatures) of the two compounds are similar, as expected from the small atomic mass difference of Cr and Fe. Also, afm spin waves do not play a role due to the high Néel temperatures.

The linear specific heat contribution is insensitive to magnetic fields (cf. Fig. 11 inset). For BaCr2As2 in a field $\mu_{0}H=9$ T the linear coefficient $\gamma$ is not changed at all. For BaCrFeAs2 the specific heat increases slightly in $\mu_{0}H=9$ T for $T<5$ K (by maximally 1.8 % at 3.3 K), which might be due to a Schottky-type anomaly due to minor impurities. Thus, the enhanced $\gamma$ values should either not be due to spin fluctuations at all or the spin fluctuations are of too high in energy. Especially BaCr2As2 has a very high Néel temperature and the fluctuations might not be quenchable by such a small field.

III.5 Magnetic structures

In order to investigate the magnetic structure of BaCr2As2 we have collected a powder neutron pattern at 2 K. In comparison to the data collected at 750 K, well above the Néel temperature $T_{\mathrm{N}}=580$ K, it was shown that magnetic intensities appear at the positions of nuclear Bragg reflections indicating a magnetic structure, which can be described with the propagation vector $\bm{k}=0$. This shows that the translation $\bm{t}=(\tfrac{1}{2},\tfrac{1}{2},\tfrac{1}{2})$ associated with the $I$ cell is not lost. In Figure 2 it can be seen that the strongest magnetic intensity is observed at $2\theta=27.4^{\circ}$, which is the position of the Bragg reflection 101. Strong intensity could be generated at this position using a magnetic structure model, where the chromium atoms in the Wyckoff positions $4d$, located at $(0,\tfrac{1}{2},\tfrac{1}{4})$ and $(\tfrac{1}{2}$,0,$\tfrac{1}{4})$, are coupled antiparallel. Due to the $I$ centering the operation $\bm{t}=(\tfrac{1}{2},\tfrac{1}{2},\tfrac{1}{2})$ does not change the direction of the spin. Therefore one finds the spin sequence $+-+-$ for the Cr atoms in the positions $(0,\tfrac{1}{2},\tfrac{1}{4})$, $(\tfrac{1}{2},0,\tfrac{1}{4})$, $(\tfrac{1}{2},0,\tfrac{3}{4})$, and $(0,\tfrac{1}{2},\tfrac{3}{4})$.

Using this structure model the magnetic structure could be successfully refined, when the magnetic moments are aligned parallel to the $c$ axis. The same type of magnetic ordering is found for BaMn2As2 and also for the Cr sublattice in EuCr2As2 [Nandi et al., 2016]. Assuming an additional magnetic component within the $ab$ plane, magnetic intensity should appear at the position of the reflection 002. Magnetic intensity could not be easily determined from the difference patterns collected at 2 and 750 K, in the fully ordered and the paramagnetic state, because the strong structural changes lead to a change of the nuclear intensity of the reflection 002. Therefore we have refined simultaneously the crystal and the magnetic structure at 2 K. The calculated nuclear intensity of the 002 reflection is even slightly larger than the observed one. This clearly indicates the absence of an additional magnetic component within the $ab$ plane, i. e. the absence of any spin canting to the $c$ axis. From the data set collected at 2 K on instrument E9 we finally found for the Cr atoms a magnetic moment of $\mu_{\mathrm{exp}}=1.88(2)\mu_{\mathrm{B}}$ resulting in a residual $R_{\mathrm{M}}=0.0268$ [defined as $R_{\mathrm{M}}=(\Sigma||I_{\mathrm{obs}}|-|I_{\mathrm{calc}}||)/\Sigma|I_{\mathrm{obs}}|$]. From the data set collected on instrument E2 the magnetic moment $\mu_{\mathrm{exp}}=1.80(2)\mu_{\mathrm{B}}$ was found to be slightly smaller (residual $R_{\mathrm{M}}=0.118$). It has to be mentioned that the moment value determined from the E9 data is more reliable, since the overall scale factor could be determined with higher accuracy (a much larger number of nuclear Bragg reflections was available). From the data collected on E2, E6, and E9 we are able to determine the temperature dependence of the magnetic moment up to the Néel temperature $T_{\mathrm{N}}=580(10)$ K (see Figure 12). The results from the neutron diffraction experiments compare well with those from magnetization data of the BaCr2As2 single crystal (Fig. 8). The experimental magnetic Cr moments at 2 K compare reasonably well with the values obtained from our (cf. Table 2 and previous [Singh et al., 2009c] electronic structure calculations. The experimental Cr moment is even slightly smaller than the calculated moments. This shows that electronic correlation effects reflected in the parameter $U$ of the LDA$+U$ calculations do not play a role as increasing $U$ lead to larger Cr magnetic moments.

Accordingly we have determined the magnetic structure of BaCrFeAs2. As it can be seen in Figure 2 the strongest magnetic intensity could also be observed at the position of the reflection 101. The magnetic intensities of BaCrFeAs2 are found to be much weaker than those in BaCr2As2 indicating that the averaged magnetic moments of the Cr and Fe atoms are strongly reduced. A moment $\mu_{\mathrm{exp}}=1.09(3)\mu_{\mathrm{B}}$ was refined resulting in a residual $R_{\mathrm{M}}=0.068$. The Néel temperature $T_{\mathrm{N}}=265(5)$ K compares well with our magnetic susceptibility data (Fig. 9).

III.6 Mössbauer spectroscopy on BaCrFeAs2

Representative 57Fe Mössbauer spectra of BaCrFeAs2 are shown in Figure 13. The room temperature spectrum appears as a broadened line which can be fitted by a quadrupole doublet with an isomer shift IS of 0.44 mm s*-1* and a quadrupole splitting QS of 0.13 mm s*-1*. By contrast, the spectrum at 5 K features a complex broad pattern which evidences the presence of magnetic hyperfine splitting. The spectrum was fitted by assuming a distribution of hyperfine fields, yielding IS = 0.59 mm s*-1* and a peak hyperfine field $B_{\mathrm{hf}}=5.4$ T. The isomer shifts are slightly larger than those of BaFe2As2 [Błachowski et al., 2011] which adopts the SDW phase below the magnetic ordering temperature. Nevertheless the local electronic structure is quite similar in the two compounds. Most remarkably, the peak $B_{\mathrm{hf}}$ of about 5 T in BaCrFeAs2 is nearly the same as the $B_{\mathrm{hf}}$ in the magnetically ordered phases of BaFe2As2 [Błachowski et al., 2011] and LaOFeAs [McGuire et al., 2008]. From the neutron diffraction study we obtained an average magnetic moment of 1.09 $\mu_{\mathrm{B}}$ per magnetic ion. Assuming that the Cr moment is the same as in BaCr2As2 (1.8 $\mu_{\mathrm{B}}$) we estimate the iron moment as 0.4 $\mu_{\mathrm{B}}$.

It has been pointed out that a direct estimation of Fe moments from $B_{\mathrm{hf}}$ is questionable due to spin-orbit induced contributions to $B_{\mathrm{hf}}$ in the iron pnictides [Derondeau et al., 2016]. Nevertheless, a small iron moment is in qualitative agreement with the small $B_{\mathrm{hf}}$ for BaCrFeAs2. A similarly small iron moment of 0.4 $\mu_{\mathrm{B}}$ was reported for LaOFeAs [de la Cruz et al., 2008], whereas a larger moment of 0.9 $\mu_{\mathrm{B}}$ was derived for BaFe2As2 [Huang et al., 2008]. The spectra of BaCrFeAs2 are, however, much less resolved than those of BaFe2As2 and LaOFeAs which reflects the Fe-Cr disorder in the present material giving rise to a broad $B_{\mathrm{hf}}$ distribution. Both, neutron diffraction and Mössbauer data confirm that Fe and Cr atoms are disordered, while some previous electronic structure calculations assumed an ordered arrangement [Singh et al., 2009a; Hu and Hu, 2010]. A large ordered Fe moment of 2.5–2.6 $\mu_{\mathrm{B}}$ predicted from the electronic structure calculations for atomically ordered BaCrFeAs2 (Ref. Hu and Hu, 2010 and Table 3) is incompatible with the neutron diffraction and the Mössbauer data.

Indications for an unusual behavior of the Fe moments is obtained from the temperature dependence of $B_{\mathrm{hf}}$. With increasing temperature a pronounced decrease in hyperfine splitting is apparent even in the temperature region 5-100 K (Figure 13), which is unexpected as the Néel temperature is much higher (265 K). At 100 K the $B_{\mathrm{hf}}$ distribution even extends down to 0 T (Figure 14) indicating that a fraction of the iron atoms is already non-magnetic. These observations are in contrast to the behavior of BaFe2As2 where in this temperature range only a minor reduction of $B_{\mathrm{hf}}$ was observed although $T_{\mathrm{N}}$ is considerably smaller ($\approx 140$ K) [Błachowski et al., 2011]. The strong decrease of the peak $B_{\mathrm{hf}}$ from 5.4 T at 5 K to 2.4 T at 100 K is also in sharp contrast to the temperature dependence of the total magnetic moments derived from neutron diffraction (Figure 12).

A possible scenario is that the Fe moments do not participate in the $G$-type afm structure and rather freeze in the low temperature range which is consistent with the fact that no anomaly is apparent in the temperature dependence of the moments between 2 and 100 K. The formation of the $G$-type afm order then is still driven by the large Cr moments. Further on, the upturn in the susceptibility $\chi(T)$ below $T_{\mathrm{N}}=265$ K (Fig. 9), which is in contrast to the behavior of BaCr2As2, is a further indication that the iron moments are at least partially disordered. As demonstrated above electronic structure calculations of atomically disordered BaCrFeAs2 starting with zero Fe moment result in a stable solution with $\mu_{\mathrm{Fe}}\sim 0$ and $\mu_{\mathrm{Cr}}=2.76$ $\mu_{\mathrm{B}}$. The resulting average moment of 1.38 $\mu_{\mathrm{B}}$ per formula unit of BaCrFeAs2 is not too far from the experimental moment of 1.09 $\mu_{\mathrm{B}}$. The mismatch of the Fe moments with $G$-type antiferromagnetism explains why $T_{\mathrm{N}}$ in Fe-substituted BaCr2As2 decreases drastically from 580 K in the parent compound to $\approx 50$ K for an iron content of about 70 % where finally the $G$-type order becomes unstable and is replaced by the SDW structure [Marty et al., 2011]. At higher temperature the hyperfine splitting further decreases and for $T\geq 200$ K essentially a broad unstructured line is found. In this temperature range the Fe atoms are possibly polarized by the Cr moments. Above 260 K only minor changes in line broadening occur which compares well with $T_{\mathrm{N}}=265$ K obtained from neutron diffraction and susceptibility data.

IV Conclusions

We have studied in detail the structural, electronic, and magnetic properties of BaCr2As2 and BaCrFeAs2. BaCr2As2 as well as LaOCrAs [Edelmann et al., 2016; Pizarro et al., 2016] have been discussed recently in connection with the Mott scenario (see Section I) of transition-metal arsenide superconductivity. Our powder neutron diffraction studies verify that BaCr2As2 adopts the theoretically predicted $G$-type afm order at $T_{\mathrm{N}}=580$ K and an ordered moment $\mu_{\mathrm{Cr}}=1.9$ $\mu_{\mathrm{B}}$ at 2 K. Evidence for magneto-structural coupling effects was found and it remains to be clarified whether this reflects some type of electronic instability. The experimental magnetic Cr moment agrees well with LDA electronic structure calculations whereas incorporation of electron correlation within the LDA$+U$ scheme leads to Cr moments which are too high. $T_{\mathrm{N}}$ is 1/3 lower than the ordering temperature calculated within mean-field approximation, underlining the somewhat twodimensional character of the magnetic system.

BaCrFeAs2 still adopts this $G$-type afm structure but the ordering temperature is less than half of that of BaCr2As2. The small average moment $\mu_{\mathrm{Cr/Fe}}\approx 1.1$ $\mu_{\mathrm{B}}$ in connection with the small hyperfine field from Mössbauer spectra is in agreement with the results of our calculations for atomically disordered BaCrFeAs2 and an Fe moment $\mu_{\mathrm{Fe}}\sim 0$. The Néel temperature calculated within mean-field approximation is in good agreement with the experimental value (260 K and 265 K, respectively). Also the average Cr/Fe moment is in fair agreement with the calculation (2.76 $\mu_{\mathrm{B}}$, distributed on two atoms). These findings, together with an anomalous temperature dependence of the hyperfine field, indicate that the small Fe moments are not incorporated into the $G$-type afm order. Thus it may be conjectured that Fe favors the stripe-type and Cr the checkerboard-type spin fluctuations, the latter being considered to be detrimental to superconductivity [Fernandes and Millis, 2013]. In case of electron pairing mediated by spin-fluctuations this finding would be in agreement with the fact that so far no superconductivity was observed for BaCr2As2 or Cr-substituted BaFe2As2. Similarly, no superconductivity was found in Mn-substituted BaFe2As2, where the pure Mn compound BaMn2As2 adopts a stable $G$-type afm order as well. Accordingly not only the doping level but also the type of spin fluctuations has to be considered for predictions of new superconducting compositions.

Both BaCr2As2 and BaCrFeAs2 are metals. The linear specific heat (Sommerfeld) coefficients $\gamma$ are much larger than expected from band theory, by a factor of 2.4 for BaCr2As2 and a huge factor of 5.5 for BaCrFeAs2. The discrepancies for both compounds cannot be resolved by reasonable strengths of correlations. However it seems that for BaCrFeAs2 the disorder of the magnetic Cr and Fe atoms to some extent additionally boosts the $\gamma$ value.

Our findings may be compared with results for the better investigated Mn 122 system. In contrast to BaCr2As2 BaMn2As2 is semiconducting [An et al., 2009; Singh et al., 2009a, b; Johnston et al., 2011; Zhang et al., 2016]. The nominal $3d^{5}$ compound has a Néel temperature which is slightly higher than that of BaCr2As2. It is possible to induce a metallic state in BaMn2As2 by application of pressure [Satya et al., 2011] or by hole-doping with as little as 1.6 % of K [Pandey et al., 2012]. In this metallic state Ba1-xKxMn2As2 is still $G$-type ordered [Lamsal et al., 2013] and is thus similar to BaCr2As2. But there is one important difference, the Sommerfeld $\gamma$ in the metallic Mn system is much smaller (8.4 mJ mol*-1* K*-1* for $x=0.05$ [Pandey et al., 2012]) than in BaCr2As2. This suggests that electronic correlations are weak in the BaMn2As2 materials. A recent photoemission (ARPES) study corroborates that there is almost no band renormalization with respect to the DFT-based band structure in BaMn2As2 [Zhang et al., 2016]. An ARPES study on BaCr2As2 would therefore be highly welcome in order to shed more light on the development of electronic correlations in the 122 family of transition-metal arsenides.

For Ba1-xKxMn2As2 the Néel temperature and the ordered Mn moment remain almost constant up to high substitution levels ($x\leq 0.4$) [Lamsal et al., 2013]. In contrast, in the substitution series BaFe2-xMnxAs2 for a low Mn substitution level $x=0.15$ a clear competition between stripe-type and checkerboard-type spin fluctuations could be observed by inelastic neutron scattering [Tucker et al., 2012], quite similar to the BaCr2-xFexAs2 series [Marty et al., 2011]. For BaFe2-xMnxAs2 a miscibility gap prevents the synthesis of single-phase material for Mn contents $x>0.24$ [Pandey et al., 2011].

Although both modified BaMn2As2-based ($3d^{5}$) materials [Satya et al., 2011; Pandey et al., 2012; Lamsal et al., 2013] as well as BaCr2As2 are afm-ordered metals, until now there are no ways known to obtain superconductors based on these materials. The partial substitution of Ba by K in BaCr2As2 (hole doping) showed to be unsuccessful, but surprisingly resulted in the discovery of superconducting K2Cr3As3 [Bao et al., 2015; Cao et al., 2016]. Another way to possibly generate superconductivity in Mn or Cr-based 122 or 1111-type arsenides is the application of high pressure. With high pressure, in transition-metal arsenides structural instabilities come into play. We have now undertaken a structural and electrical transport study on one of our BaCr2As2 crystals under high pressure, the results of which will be reported elsewhere [Medvedev et al., ].

Acknowledgements.

We thank Ralf Koban for assistence and Igor Veremchuk for spark plasma sintering of a sample. MG acknowledges support of NSF-DMR 1507252 grant.