1.2. Structural and Magnetic Properties of RMn2 Compounds
The
RMn
2 compounds have been widely studied because of their very interesting structural and magnetic properties. Changes of the unit cells for
RMn
2 as a function of the temperature are presented in
Figure 1a
[16]. For the (Tb, Gd, Sm, Nd, Pr, Y)Mn
2 compounds, the first-order phase transition is observed, and temperature dilatometer and X-ray diffraction measurements revealed significant increases in the volume of unit cells of these alloys at temperatures below 105 K. The magnetic ordering was also observed below the temperatures of the structural changes
[17][18][19]. The largest, an almost 5% change in volume, was recorded for YMn
2. This significant magneto-volume effect is related to the appearance of a well-localized magnetic moment in the Mn sublattice (~2.7μ
B) below the magnetic ordering temperature
[20]. The (Er, Ho, Dy)Mn
2 compounds do not show changes in the volume cell; they undergo a second-order phase transition in the temperature range of 15–35 K
[14][21].
Figure 1. (
a) Changes of the unit cells for
RMn
2 as a function of the temperature for
RMn
2 [16]; (
b) Relation between the
55Mn hyperfine field
Hself and the lattice parameter of
RMn
2 at 4.2 K. The transferred hyperfine fields from rare-earth moments are subtracted from the observed ones. The effective lattice parameter of C14 compounds (closed circles) is given by
aeff = (3
a2c)
1/3. The scale for the estimated Mn moment from
Hself is given on the right-hand side
[22].
The type of the phase transition can be related to the occurrence of a magnetic moment of Mn atoms or its absence. The measurements of the hyperfine field on the
55Mn nuclei in
RMn
2 compounds carried out by Yoshimura (
Figure 1b) indicate that below a certain distance of ~2.66 Å between the nearest Mn atoms (
dMn-Mn), the localized magnetic moment of the Mn atoms is not observed. This distance is called the critical distance and denoted as
dc [22]. Based on the
dMn-Mn distance,
RMn
2 compounds can be divided into three groups
[23]:
-
dMn-Mn <
dc, (LuMn
2, ErMn
2, TmMn
2), where the Mn subnet is non-magnetic, and the Er and Tm sublattices are ferromagnetic
[24];
-
dMn-Mn >
dc, (NdMn
2, PrMn
2), where the dominant Mn magnetic sublattice is stable and enforces antiferromagnetic ordering in the sublattice of
R-atoms with the
R-Mn type interaction
[25];
-
dMn-Mn ~
dc, (GdMn
2, TbMn
2, DyMn
2, HoMn
2, YMn
2), where the Mn sublattice is on the verge of stability. Generally, non-collinear magnetic structures are observed in this group of compounds. In some cases (TbMn
2, DyMn
2, HoMn
2), only a part of the Mn atoms carries a magnetic moment
[20][26][27][28][29][30][31][32][33][34].
The magnetism of the RMn2 compounds is related to the electronic structure of Mn and R atoms. Mn has an unfilled outer 3d electron shell affecting the crystalline field from other ions. However, in the R atoms, the unfilled inner 4f electron shell is shielded by the 5p outer shell, which facilitates the behavior similar to that noted for the free ion.
In the RMn2 compounds, three types of exchange interactions can be distinguished as follows:
-
Between 3d–3d magnetic moments (Mn–Mn),
-
Between 3d–4f moments (Mn–R),
-
Between localized 4f–4f moments (R–R).
The strongest among the above is the 3d–3d exchange interaction, as evidenced by the high temperature of ordering of the magnetic relationship with non-magnetic rare earths (YMn2, TN ≈ 105 K). The band model, which is applicable here, assumes that the difference in concentration between electrons with spins ↓ and ↑ is responsible for the formation of the magnetic moment.
The weakest is the 4f–4f interaction, as indicated by low temperatures of magnetic ordering of the ErMn2 and TmMn2 compounds with the non-magnetic Mn sublattice (Figure 1a,b). The R atoms having an unfilled 4f electron shell are characterized by well-localized magnetic moments; the wave functions that describe them have a short range in comparison to the distance between atoms (magnetic moments). The magnitude of the interaction of 4f–4f depends on the density of the polarized conduction electrons; their oscillating nature as a function of distance is described by the RKKY theory.
The magnetic interaction 3
d–4
f has an intermediate value between 3
d–3
d and 4
f–4
f. It occurs mainly through the polarization of electrons of 5d rare earth atoms
[35].
1.3. Hydrogen in Metals
Rare earth (
R) metals react with hydrogen forming stable di- and trihydrides. The
RH
2 dihydrides are very stable, with reported formation enthalpies of about −200 kJ/molH
2; thus, they decompose in a vacuum, releasing hydrogen and forming individual RE metals at very high temperatures only
[36]. On the other hand, manganese, which is also the main building block of the discussed
RMn
2H
x structures, behaves differently under the influence of hydrogen in comparison to RH
x. At ambient pressure conditions and room temperature, the subhydride Mn
2H is the only stable compound. With increasing pressure, MnH is stabilized, and then hydrogen-rich hydrides appear at higher pressure
[37].
A simplified scheme of hydrogen penetration into the crystal structure of the metal (intermetallic compound) can be divided into three stages (
Figure 2): (a) Physisorption: gaseous hydrogen molecules interact with electrons on the metal surface using weak attractive van der Waals forces; (b) Chemical adsorption: hydrogen molecules dissociate, overcoming the surface energy barrier, and form a metallic hydrogen bond; (c) Absorption: in this step, the chemisorbed hydrogen atom can jump to the subsurface layer to finally diffuse through the crystalline structure of the metallic host
[38].
Figure 2. Simplified mechanism of hydrogen absorption in metal. (a) Physical adsorption: hydrogen molecules adhere to the metal surface, forming several layers; (b) Chemical adsorption: dissociated hydrogen molecules of the layer in direct contact with the metal form chemical bonds with it; (c) Absorption: hydrogen atoms diffuse into the parent metal lattice.
After absorption, the hydrogen atoms/ions remain in the sites of the crystal structure until they are given a surplus of energy necessary to overcome the diffusion barrier as a result of thermal activation. Under these conditions, the diffusion coefficient satisfies the classical Arrhenius relationship:
D =
D0 exp (−
Ea/
kT), where
D0 is the diffusion constant. Assuming the simple activated diffusion, the typical diffusion constant for H in intermetallic compounds equals 1–10
3 mm
2/s and
Ea = 45 kJ/mol at room temperature
[39][40]. Researchers also point to the dependence of the
D0 and the
T on the structure on the structure of the metal. For example, for Pd (
fcc structure),
D0 = 2.9 × 10
−3 cm
2/s and
Ea = 0.230 eV, while
Do = 4.2 × 10
−4 cm
2/s and
Ea = 0.40 eV for Fe (bcc structure)
[41].
One of the first experiments indicating the dissociation of H
2 molecules in the process of its absorption by metals was the study of the
p-c-T relationship (hydrogen pressure-concentration-temperature)
[39]. In numerous metal hydrides, hydrogen atoms are characterized by high mobility already at room temperature, especially when the number of available positions is greater than the number of hydrogen atoms. Hydrogen atoms can jump from one position to another without creating a crystallographic ordered structure—the positioning is accidental. The frequency of hydrogen jumps is huge (e.g., in vanadium hydrides at room temperature, it is ~10
10–10
12 Hz)
[39]. With the lowering of the hydride temperature, the frequency decreases. Therefore, it may also happen that hydrogen atoms occupy selected positions in an orderly manner, and they create a superstructure. Skripov states that in many hydrides of the Laves phase, even two types of hydrogen jumping motions with different characteristic jump rates coexist. While the slower jump process is responsible for long-distance diffusion, the faster process corresponds to the localized movement of hydrogen over small groups of interstitial sites. In some of the Laves-phase compounds, the localized hydrogen motion is not ‘
frozen out’ on the frequency scale of 10
7 to 10
8 Hz down to 20 K
[42]. At low temperatures, there is also the possibility of non-activation, tunnel transitions of the hydrogen atom from one position to another
[38][43][44].
Figure 3a shows the typical hydrogen absorption isotherms in metal. The isotherms form a phase diagram of the equilibrium pressure as a function of the hydrogen concentration Peq (x), in which three essential areas can be distinguished: (i) the α region is the “solid” solution of hydrogen in the parent metal, where the hydrogen pressure p increases with the hydrogen content x according to Sievert’s law p = (Ks c)2 (K—Sievert’s constant, c—hydrogen concentration), (ii) the two-phase region (α + β) and (iii) the β region—metal hydride. As the temperature increases, the latter narrows and disappears at the critical temperature Tcrit, which is characteristic for a given element. The isotherms in the middle region show a hydride characteristic plateau, corresponding to the equilibrium pressure as a function of the hydrogen concentration.
Figure 3. (
a) Schematic plot of hydrogen absorption in metal (
p-c-T isotherms).
Peq—gas pressure,
c—hydrogen concentration in the metal; (
b) Construction of the Van’t Hoff relation. Based on
[45][46].
The equilibrium pressure zone is ideally suited for the isothermal absorption and desorption of hydrogen from metal. The state of equilibrium can be described by a phase diagram analogous to the Van der Waals diagram of the non-ideal gas.
Peq equilibrium pressure is associated with changes in enthalpy (Δ
H) and entropy (Δ
S) in the two-phase region via the empirical temperature of the Van’t Hoff relationship (
Figure 3b)
[45][46]. For a typical equilibrium pressure, e.g., for PdH
x—hydride at 423 K, it is 0.1 MPa, and it is 1.0 MPa at 518 K. The critical point of this hydride is
Tcrit = 565 K at
x ≈ 0.25 H/f.u.
[39].
2. RMn2Hx Hydrides
2.1. Localization of Hydrogen
The physical properties of rare earth hydrides with manganese
RMn
2H
x largely coincide with the properties of metal hydrides (MH
x). Also, alloys of
RMn
2 compounds absorb hydrogen very easily. At close to atmospheric hydrogen pressure and temperatures above 100 °C, it is possible to achieve a hydrogen concentration of ~4.5 H/f.u.
[47]. As noted in metal hydrides, hydrogen occupies the interatomic sites of the elements that comprise the
RMn
2 unit cell. There are three types of tetrahedral sites in
RMn
2 compounds: A2B2, AB3 and B4, where A and B represent atoms of
R and Mn, respectively, surrounding the site (
Figure 4). The largest and at the same time most willingly filled by hydrogen site (below
x ≈ 3.5 H/f.u.) is the A2B2 type. For
x ≳ 3.5 H/f.u., there is also a tendency to fill smaller sites, such as AB3 sites
[48][49][50][51]. However, no B4 site filling was observed. The same types of sites occur in both C14 and C15 types of Laves phases. When filling the sites, hydrogen occupies crystallographic positions (
Table 1).
Figure 4. Three crystallographic positions accessible for hydrogen in C15 Laves phase:
96g (inside of A2B2 tetrahedra),
32e (inside of AB3 tetrahedra) and
8b (inside of B4 tetrahedra). The
Fd3𝑚3¯� space group. Some atoms were removed for clarity. Big, dark balls: R; small, bright balls: Mn. Based on
[50].
Table 1. Hydrogen positions in C14 and C15 phases.
Although the number of sites is significant, every site cannot be occupied by hydrogen. Two criteria are useful: (i) the Switendick criterion
[52] states that the distance
dH-H must be larger than ~2.1 Å and (ii) the Westlake criterion indicates that the radius of the sphere available for hydrogen cannot be less than ~0.37 Å.
[53]. As an effect, criteria preclude occupancy of A2B2 tetrahedrons that have a common wall, and the B4 site is not available for hydrogen due to size restrictions. The Switendick criterion, confirmed in neutron measurements
[54][55], largely determines the maximum content of absorbed hydrogen. The theoretically determined maximum hydrogen concentration in
RT2H
x (
R: rare earth,
T: transition metal = Y, Mn) compounds is therefore 6.0 H/f.u.
[56]. Only the use of ultra-high hydrogen pressures of the order of 10
3 MPa led to the formation of
RMn
2H
6 hydrides; the first being YMn
2H
6 [57][58]. Highly hydrogenated (
x ≳ 4.5 H/f.u.) compounds show different physical properties compared to low-hydrogen compounds (
x ≲ 4.5 H/f.u.)
[59][60] and will not be discussed in this study.
2.2. Sample Preparation and Dependence of ΔV/V versus Hydrogen Content
In the discussed hydrides, the sample preparation was similar and usually involved the following scenario.
The host materials (
RMn
2) were prepared from high purity elements using the standard induction melting technique under argon atmosphere. Next, materials were usually annealed to obtain a single-phase compound. The
RMn
2 samples were saturated with hydrogen using a standard volumetric method (Sievert) to obtain
RMn
2H
x hydrides in the typical range of
x: 0 <
x < 4.3–4.5
[7][8][9][54][61][62][63][64][65][66]. The first test after hydrogen saturation was XRD measurements at 300 K. As a representative example, the HoMn
2H
x_C15 diffractogram is presented in
Figure 5a. The most visible effect of hydrogen absorption by the sample is the shift of the diffraction lines towards smaller angles (in 2 theta), which means an increase in the distance between atoms in the cell, and consequently an increase in the unit volume of the cell with
x.
Figure 5. HoMn
2H
x. (
a) The XRD patterns for different
x; here, α represents the cubic phase (
𝐹𝑑3), and
β represents the rhombohedral phase (
𝑅3); (
b) Relative change in the unit cell volume as a function of
x [65].
Even at room temperature, the
RMn
2H
x hydrides do not persist in the crystallographic structure of the host compound. For the HoMn
2H
x_C15 hydride, a combination of the cubic (
α) and rhombohedral distortion (
β) is observed (
Figure 5a). The obtained results are in agreement with
[67]. To compare lattice parameters of hydrides with different crystallographic structures, their parameters were converted into parameters of the pseudo-cubic cell considering abundance phases, if the system is not a single-phase system. For example, the lattice parameters of the rhombohedral phase were converted into lattice parameters
𝑎*𝛿 of the pseudo-cubic phase according to the relationship reported below:
where
aδ and
cδ are lattice parameters of the rhombohedral phase (space group:
𝑅3, nr166,
setting: 1)
[65].
The relative volume change in unit cell versus hydrogen content in HoMn
2H
x_C15 is depicted in
Figure 5b. The first ‘model’ that attempted to describe the relation between Δ
V/
V0 and
x was just a linear relationship. The next one, the Hirata model, only correctly reproduces the lowest range of
x (0 <
x < 3.0) (
Figure 5b, dashed line)
[68], and its development was proposed by Figiel et al.
[50]. In the last one (farther: Hirata–Figiel model), the authors assume a two-stage approach to occupying interstitial sites (
Figure 5b red line). Initially, with a characteristic parameter
Xc < 3.0–3.5 H/f.u. (see below), the A2B2 type sites are filled (this part is identical with Hirata model). Next, above
Xc > 3.0–3.5 H/f.u., the AB3 type sites are also occupied. It is represented by a characteristic break in the red curve (
x~3.2 H/f.u.,
Figure 5). The Hirata-Figiel model is described according to the relationship described below
[50].
where Δ
V/
V0 is the relative increase in unit cell volume;
B0 is a parameter of the lattice contraction, a quantity analogous to the bulk modulus at starting concentration related to the intrinsic pressure;
b is the first derivative with respect to concentration
x: (
Bx =
B0 + bx); and
P is the probability of filling the AB3 sites by hydrogen atoms for concentrations greater than
Xc (
P = 0 for
x ≤ Xc)
[34][50]. Fitted parameters for the
RMn
2H
x (
R: Y, Dy, Gd, Tb, Ho) hydrides are presented in
Table 2. For all compounds, the characteristic
Xc belongs to the region of ~3.0–3.5 and probabilities of entering the AB3 interstitial sites
P~0.6–0.9. This type of filling sites was also postulated in
[49][50].
Table 2. Fitted parameters for the Hirata-Figiel model.
The relative volume increase (ΔV/V0) of unit cells of hydrides determined at room temperature and depending on the hydrogen content in the sample is non-linear and can be described by the same function, indicating that this effect is determined only by the presence of hydrogen. The fits of the Hirata–Figiel model to the change in (ΔV/V0) depending on the hydrogen concentration for the various studied hydrides RMn2Hx are shown in Figure 6.
Figure 6. Comparison of relative volume change in unit cells for
RMn
2H
x versus hydrogen content at RT = 300 K. Based on
[7][50][64][65][66].
For x < 3.0 H/f.u., the curves, except for those corresponding to (Gd and Nd)Mn2Hx, are almost identical. At values greater than x ≈ 3.0 H/f.u., distinct behavior becomes visible with increasing x, especially for concentrations close to the maximum, where the relative increase in volume increases with the atomic number of the rare earth element.
2.3. Examples of Structural and Magnetic Transformations in RMn2Hx
With a temperature change, the
RMn
2H
x hydrides undergo complex phase transformations in many cases, which, combined with the often occurring multi-phase states, makes the interpretation of their X-ray diffraction magnetic transformation results extremely difficult. Some authors have even argued that it will never be possible
[39]. One of the first hydride families to be studied over a wide range of
x and temperature changes was YMn
2H
x, showing a very wide range of structural transformations as a function of temperature. In contrast, TbMn
2H
x hydride families show a very ‘modest’ spectrum of structural transformations. Very interesting are the structural and magnetic properties of HoMn
2H
x hydrides, in which the base compound can crystallize in a C14 or C15 Laves phase structure, depending on the sample preparation
[10].