Rare Earth Borohydrides—Crystal Structures and Thermal Properties

Abstract

Rare earth (RE) borohydrides have received considerable attention during the past ten years as possible hydrogen storage materials due to their relatively high gravimetric hydrogen density. This review illustrates the rich chemistry, structural diversity and thermal properties of borohydrides containing RE elements. In addition, it highlights the decomposition and rehydrogenation properties of composites containing RE-borohydrides, light-weight metal borohydrides such as LiBH₄ and additives such as LiH.

1. Introduction

Metal borohydrides are being intensively studied as potential hydrogen and thermal energy storage materials [1,2,3,4,5,6,7] as well as solid state electrolytes [8,9,10,11]. Alkali and alkaline earth borohydrides such as LiBH₄ and Mg(BH₄)₂ are of great interest due to their high gravimetric hydrogen storage capacity. Unfortunately, their decomposition temperatures are too high and rehydrogenation only occurs under severe conditions as a result of the strong ionic interaction between the metal cation and the BH₄⁻ anion. In contrast, complexes containing transition metal cations are too unstable with decomposition temperatures below or slightly above room temperature. In recent years, a correlation between the thermodynamic stability (decomposition temperature) and the Pauling electronegativity of the metal cation has been established [12]. According to reference [12] the hydrogen desorption temperature decreases with increasing electronegativity of the metal. This result suggests that the charge transfer from the metal cation to the BH₄⁻ anion is responsible for the stability of the metal borohydride. It has become apparent that borohydrides with Pauling electronegativities larger than 1.4 are too unstable for use in hydrogen storage applications. Tuning the thermodynamic properties of borohydrides has been attempted by the combination of different metal cations to form double-cation metal borohydrides Mⁿ⁺M^m+(BH₄)_(n+m) (Mⁿ⁺ = Li⁺, Na⁺, K⁺; M^m+ = transition metal cation). The search for new possible candidates for hydrogen storage applications has stimulated interest in materials containing rare earth (RE) metal cations. The gravimetric hydrogen storage density of monometallic RE-borohydrides ranges from 9.1 wt % H₂ for Y(BH₄)₃ to 5.6 wt % H₂ for Yb(BH₄)₃, respectively, and their thermodynamic stability is expected to lie between those of the alkali and transition metals. Compared to alkali or alkaline-earth borohydrides such as LiBH₄ (18.5 wt % H₂) or Mg(BH₄)₂ (14.9 wt % H₂), RE-borohydrides possess lower gravimetric hydrogen density. However, they are able to release a major amount of hydrogen between 200 and 300 °C, which is significantly lower than for LiBH₄ or Mg(BH₄)₂. As a result, a large number of novel RE-borohydrides have been synthesized and more than 50 papers have been published since 2008. The majority of publications are focused on the synthesis and structural characterization of the compounds and several extensive reviews already exist [1,4,7]. In contrast, less information is available about the thermal properties and rehydrogenation behavior of pure RE-borohydrides or as part of composites.

The current review article presents the state of the art on the crystal structure and thermodynamic properties of pure RE-borohydrides. In addition, we elucidate the thermal decomposition and rehydrogenation behavior of RE-borohydrides in composite mixtures containing LiBH₄ and additives such as LiH. The layout is as follows: Section 2 details commonly applied synthesis procedures for RE-borohydrides; Section 3 summarizes the crystal structures of mono-, di- and trimetallic borohydrides containing rare earth elements; Section 4 presents the thermal properties of RE-borohydrides starting from monometallic and halide substituted compounds, followed by composite mixtures containing RE-borohydrides, LiBH₄ and LiH. This is the first time that the state of the art for the decomposition and rehydrogenation behavior of composites containing RE-borohydrides has been summarized in literature.

2. Synthesis of Metal Borohydrides

The synthesis of borohydrides is mostly carried out by either solvent-based methods or by mechanochemistry. Both approaches offer distinct advantages, but also suffer from certain drawbacks. The first solvent-based synthesis of a metal borohydride, Mg(BH₄)₂, was reported by Wiberg and Bauer in 1950 and was based on the reaction of diethylmagnesium, Mg(CH₂CH₃)₂, and diborane, B₂H₆, in ether [13]. However, the authors were not aware of a stable solvent adduct that was formed under the experimental conditions described in their paper. A direct synthesis route for Mg(BH₄)₂ from the elements was described in a German patent from 1958 [14]. However, attempts to reproduce the procedure were not successful in obtaining sufficiently high yields and it was impossible to separate the product from the rest of the reaction mixture by solvent extraction [15]. In order to circumvent the high reactivity of diborane towards air and moisture and its inherent toxicity, Koester [16] developed a synthesis method in 1957 based on the reaction of the respective metal hydride with N-alkylborazanes instead of diborane. This method has since then been successfully adapted to substitute diborane by safer and easier to handle borane complexes such as triethylamine borane, (CH₃CH₂)₃NBH₃, or dimethylsulfide borane, (CH₃)₂SBH₃. The underlying principal of each of the synthesis methods named above, is that a metal hydride or metal alkyl reacts with a borane donor anion. This leads to the formation of a solvated metal borohydride complex, which in turn requires further thermal treatment in order to obtain the solvent-free product [15,17]. A major risk of this synthesis method lies in the decomposition of the borohydride during thermal treatment and solvent removal. The same risk also exists for the metathesis reactions between an alkali borohydride (mostly LiBH₄) and a metal chloride carried out in organic solvents such as ether or toluene. However, this strategy has been successful in obtaining the desired borohydride, e.g. in the case of Al(BH₄)₃ and other compounds according to Equation (1):

3LiBH_{4 (s)} + AlCl_{3 (s)} → Al(BH₄)_{3 (l)} + 3LiCl _(s)

(1)

The use of a weakly coordinating solvent that selectively dissolves the borohydride then allows for the removal of the metal halide by solvent extraction. The separation of the borohydride from the metal halide by-product is the major advantage of the solvent-based approach and allows for structural and thermal studies on the pure and halide-free materials.

Mechanochemistry is the most commonly applied approach for the preparation of borohydrides. High-energy ball milling can incite a chemical metathesis reaction between an alkali borohydride (Li, Na, K) and the respective metal chloride in the solid state. LiBH₄ has so far proven to be the most efficient precursor for these reactions, while attempts with other metal borohydrides, e.g., NaBH₄, generally offers poor yields at most [9,18,19,20,21,22,23,24,25,26,27,28,29,30,31].

The high energy impaction between the milling balls and the powder causes a continuous fracturing and welding process that can induce grain refinement, produce fresh and reactive surfaces and provide nucleation centers for the chemical growth of reaction products. The major advantage of the ball milling method lies in the possibility to obtain metastable compounds at ambient conditions as the process proceeds far away from thermal equilibrium. In addition, ball milling provides access to more complicated bi-/trimetallic borohydrides in contrast to the solvent-based approach, which typically only yields the monometallic compounds. However, special care has to be taken in order to avoid decomposition of the product during ball milling or the formation of halide-substituted compounds. Such unwanted reactions can be suppressed by adding sufficient pauses to the milling protocol and by avoiding prolonged milling times. The metathesis reaction involving stoichiometric ratios of a trivalent RE metal chloride and lithium borohydride results in the formation of a mono- or bimetallic RE-borohydride, or a chloride substituted RE-borohydride, LiRE(BH₄)₃Cl (

Table 1. Structure types for binary and ternary lithium rare earth borohydrides formed by metathesis reactions between LiBH₄ and RECl₃.

RE Ion	Ion Radius [Å]	LiRE(BH4)4 (CuAuCl4)	α-RE(BH4)3 (ReO3)	β-RE(BH4)3 (ReO3)	LiRE(BH4)3Cl (Spinel)
Sc3+	0.745	X
Y3+	0.90		X	X
La3+	1.172				X
Ce3+	1.15			X	X
Pr3+	1.13		X	X	X
Nd3+	1.123				X
Sm3+	1.098		X	X	X
Gd3+	1.078		X		X
Tb3+	1.063		X
Dy3+	1.052		X
Ho3+	1.041		X	X
Er3+	1.03		X	X
Yb3+	1.008	X	X	X
Lu3+	1.001	X

), according to Equations (2)–(4):

RECl₃ + 3LiBH₄ → RE(BH₄)₃ + 3LiCl

(2)

RECl₃ + 4LiBH₄ → LiRE(BH₄)₄ + 3LiCl

(3)

RECl₃ + 3LiBH₄ → LiRE(BH₄)₃Cl + 2LiCl

(4)

Unfortunately, the mechanochemical metathesis approach suffers from three major drawbacks. Firstly, the reaction products after ball milling are often not very crystalline and therefore require additional careful heat treatment in order to heal out structure defects and improve their crystallinity. Secondly, the product mixture always consists of the borohydride and a metal halide, the latter simply being “dead weight”, thereby effectively lowering the gravimetric hydrogen capacity. Finally, the metal halide may influence the chemical and physical properties of the borohydride and may affect its decomposition temperature or modify the decomposition pathway.

In order to avoid the drawbacks of the mechanochemical metathesis approach and to allow for halide-free products to be obtained, wet-chemical methods have been developed. They are based on the selective dissolution of the RE-borohydride in an organic solvent such as diethylether, toluene or dimethyl sulfide, followed by removal of the unwanted by-products (e.g., LiCl) through filtering, and careful drying of the solvated product in vacuo. The interested reader is referred to reference [1] for a more detailed description of the various synthesis procedures for metal borohydrides and their derivatives.

3. Crystal Structures

3.1. Monometallic RE-Borohydrides

All of the known monometallic RE-borohydrides are summarized in

Table 2. Monometallic rare earth borohydrides. RT = room temperature; HT = high temperature polymorph; meta = metastable.

Cation	Polymorph	Hydrogen Capacity (wt %)	Stability	Crystal System	Space Group	Structure Type	Ref.
Y3+	α-Y(BH4)3 β-Y(BH4)3	9.1	RT RT, HT	Cubic Cubic	Pa$\overline{3}$ Fm$\overline{3}$c	Distorted ReO3 ReO3	[18] [21]
La3+	La(BH4)3	6.6	RT	Trigonal	R$\overline{3}$c	Distorted ReO3	[34]
Ce3+	Ce(BH4)3	6.6	RT	Cubic Trigonal	Fm$\overline{3}$c R$\overline{3}$c	ReO3 Distored ReO3	[34] [34]
Pr3+	Pr(BH4)3	6.5	RT	Cubic	Pa$\overline{3}$ Pm$\overline{3}$m	Distorted ReO3 ReO3	[30]
Sm2+	o-Sm(BH4)2	4.5	~500 K	Orthorhombic	Pbcn	ort-Sr(BH4)2	[19,27]
Sm3+	α-Sm(BH4)3 β-Sm(BH4)3	6.2	RT RT	Cubic Cubic	Pa$\overline{3}$ Pm$\overline{3}$m	Distorted ReO3 ReO3	[19] [19]
Eu2+	o-Eu(BH4)2 t-Eu(BH4)2 c-Eu(BH4)2	4.4	~430 K >668 K >668 K	Orthorhombic Tetragonal Cubic	Pbcn P41212 Pm$\overline{3}$m	orthorombic Sr(BH4)2 tetragonal Sr(BH4)2 cubic Sr(BH4)2	[27] [35] [35]
Gd3+	Gd(BH4)3	6.0	RT	Cubic	Pa$\overline{3}$	Distorted ReO3	[18,19]
Tb3+	Tb(BH4)3	5.9	RT	Cubic	Pa$\overline{3}$	Distorted ReO3	[19]
Dy3+	Dy(BH4)3	5.8	RT	Cubic	Pa$\overline{3}$	Distorted ReO3	[18]
Ho3+	α-Ho(BH4)3 β-Ho(BH4)3	5.8	RT RT	Cubic Cubic	Pa$\overline{3}$ Fm$\overline{3}$c	Distorted ReO3 ReO3	[32] [32]
Er3+	α-Er(BH4)3 β-Er(BH4)3	5.7	RT RT	Cubic Cubic	Pa$\overline{3}$ Pm$\overline{3}$m	Distorted ReO3 ReO3	[19] [19]
Tm3+	α-Tm(BH4)3	5.7		Cubic	Pa$\overline{3}$	Distorted ReO3	[33]
Yb2+	α-Yb(BH4)2 β-Yb(BH4)2 γ-Yb(BH4)2	4.0	RT >523 K 473–573 K	Orthorhombic Tetragonal Orthorhombic	F2dd P$\overline{4}$ Pbca	α-Ca(BH4)2 β-Ca(BH4)2 γ-Ca(BH4)2	[1] [29] [29]
Yb3+	α-Yb(BH4)3 β-Yb(BH4)3	5.6	RT RT, meta	Cubic Cubic	Pa$\overline{3}$ Pm$\overline{3}$m	Distorted ReO3 ReO3	[29] [29]

and their crystal structures are discussed below.

3.1.1. Distorted ReO₃-Type (α-RE(BH₄)₃)

The distorted ReO₃-type structure (space group Pa$\overline{3}$) was suggested for Y(BH₄)₃, Gd(BH₄)₃ and Dy(BH₄)₃ from laboratory powder X-ray diffraction (PXD) data by Sato et al. [18]. The RE³⁺ ions form a pseudo-cube with BH₄⁻ approximately at the center of the pseudo-cube edges (Figure 1). Y(BH₄)₃ was further investigated as a hydride and deuteride with high-resolution synchrotron radiation powder X-ray diffraction (SR-PXD) and powder neutron diffraction (PND), respectively. However, the deuterated PND sample contained highly neutron-absorbing natural boron and they were thus unable to refine the atomic positions of boron or deuterium. Frommen et al. [21] performed combined SR-PXD and PND investigation on a double-isotope substituted sample, Y(¹¹BD₄)₃. Structure refinements revealed a significant distortion compared to the undistorted ReO₃-type structure (see Section 3.1.2). The B‒Y‒B angle is for instance 160.9° compared to 180° for an undistorted ReO₃-structure and the D‒B‒D angles take values between 103 and 115° (ideal value 109.5°).

The same structure type has been reported from SR-PXD data for Sm(BH₄)₃, Gd(BH₄)₃, Tb(BH₄)₃, Dy(BH₄)₃, Er(BH₄)₃, Ho(BH₄)₃, Tm(BH₄)₃, Yb(BH₄)₃ [19,29,30,32,33]. Due to the challenges with extreme neutron absorption in Sm, Dy and Gd, in addition to the general problem with absorption in natural boron and incoherent scattering from H, these phases have not been investigated by PND and thus accurate positions for B and H have not been obtained. The unit cell parameters obtained from PXD show a near linear dependence on the radius of the RE³⁺ cation (Figure 2).

3.1.2. ReO₃-Type (β-RE(BH₄)₃)

Differential scanning calorimetry (DSC) measurements of Y(BH₄)₃ indicated a structural phase transition between 199 and 488 °C [21]. The high-temperature phase was found to be metastable at ambient temperature in samples that were quenched after annealing above the phase transition temperature [21,22]. PXD data of the quenched material indicated a primitive cubic unit cell with a = 5.4547 Å and Rietveld refinements of the data showed excellent agreement with a regular ReO₃-type structure (space group Pm$\overline{3}$m). Y is positioned in the unit cell corners, octahedrally coordinated by BH₄⁻ at the center of each unit cell edge (Figure 3). In the reported space group symmetry, the BH₄⁻ units are disordered with 8 half-occupied D sites around each boron [22]. Further investigation of a Y(¹¹BD₄)₂ sample by PND measurement indicated a face-centered cubic unit cell with doubled unit cell axes (space group Fm$\overline{3}$c, a = 11.0086 Å). Rietveld refinement with this data yielded a model with identical Y and B distribution compared to the structure refined from PXD data alone, although the BH₄⁻ ions are now deemed to have an ordered orientation and are rotated by 90° relative to their closest neighbors in both the a, b and c-directions [21] (Figure 4). The strongest superstructure peak from the BH₄⁻ ordering in the PXD data has about 0.3% of the intensity of the strongest peak and is therefore virtually impossible to detect (Figure 5). In contrast, the superstructure peak with Miller index 531 is the strongest Bragg peak in the PND data (Figure 6).

SR-PXD measurements have revealed that β-RE(BH₄)₃ modifications also exist for Ce(BH₄)₃, Pr(BH₄)₃, Sm(BH₄)₃, Ho(BH₄)₃, Er(BH₄)₃ and Yb(BH₄)₃ [19,29,30,32,34]. Due to the chemical similarity of these RE ions and Y³⁺, it seems likely that these phases take structures with ordered BH₄⁻ orientations. However, due to lack of experimental verification, most of these structures were reported in the simplest model consistent with the PXD data, i.e., primitive cubic ReO₃-type unit cells (space group Pm$\overline{3}$m) with disordered BH₄⁻ units (Table 2). The correlation between the unit cell parameters and the cation radii are shown in Figure 2.

3.1.3. Alkaline Earth Borohydride-Types

Sm, Eu, Tm and Yb can exist in the 2+ oxidation state and divalent borohydrides have been reported for all of them except Tm. Sm²⁺ and Eu²⁺ have ionic radii of 1.2 Å and 1.17 Å, respectively, which are similar to that of Sr²⁺ (1.18 Å). Sm(BH₄)₂ [19,27] and Eu(BH₄)₂ [27,35] have been reported to crystallize in the same structure type as the room temperature (RT) modification of Sr(BH₄)₂ [35] (space group Pbcn, α-PbO₂ type structure). The RE²⁺ is octahedrally coordinated by 6BH₄⁻, with the RE(BH₄)₆ octahedra sharing all six vertices and two edges with other RE(BH₄)₆ octahedra, creating zigzag chains along the c-axis. Each chain is again linked by vertices to four other chains. The borohydride tetrahedron is surrounded by three M²⁺ ions and forms nearly perfect trigonal planar coordination.

Eu(BH₄)₂ forms two additional polymorphs at higher temperatures; one tetragonal (space group P4₁2₁2, superstructure of high temperature (HT)-ZrO₂-type) and one cubic (space group Fm$\overline{3}$c, CaF₂-related). Both are isostructural to HT polymorphs of Sr(BH₄)₂ [35]. Both HT polymorphs have 8-coordinated Eu²⁺. The tetragonal Eu(BH₄)₂ polymorph can be regarded as a distorted version of the CaF₂-type polymorph. Thus, both consist of distorted (tetragonal polymorph) or regular (cubic polymorph) Eu(BH₄)₈ cubes that share all edges with other Eu(BH₄)₈ cubes.

Yb²⁺ has a similar ionic radius to Ca²⁺ (1.16 Å and 1.18 Å, respectively) and the three reported Yb(BH₄)₂ modifications are isostructural to known Ca(BH₄)₂ phases. An orthorhombic α-Yb(BH₄)₂ (space group F2dd), which is isostructural to α-Ca(BH₄)₂ is reported to be the stable polymorph at ambient temperature [1]. The two high-temperature polymorphs are isostructural to β-Ca(BH₄)₂ (space group P$\overline{4}$) and γ-Ca(BH₄)₂ (space group Pbca) [29]. Both of these HT phases have been shown to have partial substitution of BH₄⁻ by Cl^‒ (37% and 15% substitution, respectively). However, it is not clear if Cl⁻ substitution is a prerequisite for their stabilities. All of the Yb(BH₄)₂ polymorphs (with and without chlorine substitution) have Yb²⁺ octahedrally coordinated by six BH₄⁻ units.

3.2. Halide-Containing RE-Borohydrides

The large RE elements La, Ce, Pr, Nd, Sm and Gd form cubic LiRE(BH₄)₃Cl compounds (space group I$\overline{4}$3m), usually by direct reaction between LiBH₄ and RECl₃. The first reported member of this series, LiCe(BH₄)₃Cl, was structurally characterized by combined SR-PXD and PND on a ¹¹B and D substituted sample [36]. Ce³⁺ is octahedrally coordinated by three Cl⁻ and three BH₄⁻ anions. The Ce³⁺ and Cl⁻ form distorted, isolated Ce₄Cl₄ cubes, and with the three BH₄⁻ anions coordinating each Ce³⁺, these units form [Ce₄(BH₄)₁₂Cl₄]⁴⁻ complex anions that are charge compensated by four Li⁺ cations (Figure 7). The Li⁺ was reported to fully occupy two crystallographic sites; one two-fold position (2a) in the center of the Ce₄Cl₄ cube and one six-fold position (6b) which is tetrahedrally coordinated four BH₄⁻ ions from two different [Ce₄(BH₄)₁₂Cl₄]⁴⁻ units. Occupation of these two sites is the only way to place the required eight charge compensating Li⁺ in the unit cell in an ordered way. Disordered models were not considered during structure determination. Neither PXD nor PND are particularly sensitive to Li, but the proposed structure was found to be stable from density functional theory (DFT) calculations [36]. An alternative study of LiCe(¹¹BD₄)₃Cl also employed SR-PXD, PND and DFT, but considered a variety of disordered distributions of Li⁺ within the structure. It was found that a 2/3 occupation of a 12-fold site (12d) both gave lower energy (DFT) and slightly better agreement with the diffraction data than the previously reported ordered Li-distribution. The fact that LiCe(BH₄)₃Cl exhibits a high Li-ion conductivity (0.1 mS/cm at RT), further supports the model with disordered Li⁺.

LiRE(BH₄)₃Cl with RE = La, Pr, Nd, Sm and Gd as well as LiLa(BH₄)₃X with X = Br and I have been investigated by lab- or SR-PXD confirming the same structure type as LiCe(BH₄)₃Cl for each of the heavy elements [9,19,37]. The Li-ion conductivities have been measured for RE = La, Gd and both compounds show higher conductivities (0.2 and 0.4 mS/cm, respectively) than LiCe(BH₄)₃Cl [9]. This indicates that the Li-ions also partially occupy the 12d site in these compounds.

Non-stoichiometric chloride-containing borohydrides, NaY(BH₄)_2−xCl_2+x, LiYb(BH₄)_4−xCl_x and Yb(BH₄)_2−xCl_x, where chloride anions partly substitute BH₄^- have also been reported [29,31,38].

The halide containing compounds are summarized in

Table 3. Halide containing rare earth borohydrides.

Cation		Composition	Crystal System	Space Group	Ref.
Li+	La3+	LiLa(BH4)3Cl	Cubic	I$\overline{4}$3m	[9,19,37]
Li+	La3+	LiLa(BH4)3Br	Cubic	I$\overline{4}$3m	[37]
Li+	La3+	LiLa(BH4)3I	Cubic	I$\overline{4}$3m	[37]
Li+	Ce3+	LiCe(BH4)3Cl	Cubic	I$\overline{4}$3m	[8,19,36]
Li+	Pr3+	LiPr(BH4)3Cl	Cubic	I$\overline{4}$3m	[19]
Li+	Nd3+	LiNd(BH4)3Cl	Cubic	I$\overline{4}$3m	[19]
Li+	Sm3+	LiSm(BH4)3Cl	Cubic	I$\overline{4}$3m	[19]
Li+	Gd3+	LiGd(BH4)3Cl	Cubic	I$\overline{4}$3m	[9]
Na+	Y3+	NaY(BH4)2−xCl2+x	Monoclinic	P2/c	[31,38]
Li+	Yb3+	LiYb(BH4)4−xClx	Tetragonal (x ~ 1.0)	P$\overline{4}$2c	[29]
	Yb2+	Yb(BH4)2−xClx	Orthorhombic (x = 0.3)	Pbca	[29]
			Tetragonal (x = 0.76)	P$\overline{4}$	[29]

.

3.3. Bi- and Trimetallic RE-Borohydrides

LiSc(BH₄)₄ (space group P$\overline{4}$2c) was the first reported bimetallic borohydride [39]. It consists of distorted tetrahedral [Sc(BH₄)₄]⁻ complex anions with charge compensating Li⁺ (Figure 8). The Li⁺ atoms have a disorder distributed over half-occupied sites (4k) along the tetragonal c-axis. The same structure type has also been found for LiRE(BH₄)₄ where RE = Y [10,40], Yb [29] (with 25% of BH₄⁻ substituted by Cl⁻) and Lu [19]. The only difference from the LiSc(BH₄)₄ structure is that Li⁺ were reported in a fully occupied 2-fold site (2a), tetrahedrally coordinated by 4BH₄⁻ for RE = Y and Yb. This site is situated in the middle of the partly occupied sites in LiSc(BH₄)₄. No structure refinement has been performed for LiLu(BH₄)₄. Li-ion conductivity has not been measured for LiSc(BH₄)₃, but LiY(BH₄)₄ shows low conductivity (10⁻⁶ S/cm at RT).

NaSc(BH₄)₄ exists in an orthorhombic crystal structure (space group Cmcm) which contains tetrahedral [Sc(BH₄)₄]⁻ complex anions that are less distorted than in LiSc(BH₄)₄ [41]. Na⁺ has a higher coordination number than Li⁺ due to its larger radius, and is octahedrally coordinated by six BH₄⁻ moieties. The Na(BH₄)₆ octahedra form edge-sharing chains along the c-axis, with the chains also being interconnected by egde-sharing Sc(BH₄)₄ tetrahedra. NaY(BH₄)₄ [10,40], NaEr(BH₄)₄ [42], KHo(BH₄)₄ [32], o-KY(BH₄)₄ [43], KEr(BH₄)₄ [44], and KYb(BH₄)₄ [45] are isostructural to NaSc(BH₄)₄. KSc(BH₄)₄ (space group Pnma) contains similar tetrahedral [Sc(BH₄)₄]⁻ complexes while the coordination number of alkali metal increases further to eight [46].

The BH₄⁻ anion is isoelectronic with O²⁻ and as such many borohydrides are structurally similar to oxides [33]. The various Ca(BH₄)₂ polymorphs are, for instance, similar to TiO₂ polymorphs (Ti⁴⁺ being isoelectronic to Ca²⁺) [47]. A series of 20 RE-borohydrides with perovskite-like structures have been reported [38,48,49,50,51,52,53]. Seven of the compounds exhibit face-centered cubic double-perovskite-type structures (space group Fm$\overline{3}$). Despite the high symmetry, the compounds form ordered structures with large unit cell volumes ranging from 1424 ų (Rb₂LiY(BH₄)₆) to 1710 ų (Cs₃Gd(BH₄)₆). LT-KYb(BH₄)₃ (space group P$\overline{4}$3m) exhibits a primitive cubic structure with a much smaller unit cell volume of 176 ų and β-CsSm(BH₄)₃ takes the ideal perovskite structure (space group Pm$\overline{3}$m). The rest of the compounds are tetragonal, orthorhombic or monoclinic (

Table 4. Bi- and tri-metallic rare earth borohydrides.

Cations			Polymorph	wt % H	Crystal System	Space Group	Structure Type	Ref.
Li+	Sc3+		LiSc(BH4)4	14.5	Tetragonal	P$\overline{4}$2c	New	[39]
Li+	Y3+		LiY(BH4)4	10.4	Tetragonal	P$\overline{4}$2c	LiSc(BH4)4	[10,40]
Li+	Yb3+		LiYb(BH4)4	6.7	Tetragonal	P$\overline{4}$2c	LiSc(BH4)4	[29]
Li+	Lu3+		LiLu(BH4)4	6.7	Tetragonal	P$\overline{4}$2c	LiSc(BH4)4	[19]
Na+	Sc3+		NaSc(BH4)4	12.7	Orthorhombic	Cmcm	HT-CrVO4	[41]
Na+	Y3+		NaY(BH4)4	9.4	Orthorhombic	Cmcm	NaSc(BH4)4	[10,40]
Na+	La3+		NaLa(BH4)4	7.3	Orthorhombic	Pbcn	New	[49]
Na+	Ce3+		NaCe(BH4)4	7.3	Orthorhombic	Pbcn	NaLa(BH4)4	[42]
Na+	Pr3+		NaPr(BH4)4	7.2	Orthorhombic	Pbcn	NaLa(BH4)4	[42]
Na+	Er3+		NaEr(BH4)4	6.5	Orthorhombic	Cmcm	NaSc(BH4)4	[42]
Na+	Yb3+		NaYb(BH4)4	6.3	Orthorhombic	Cmcm	NaSc(BH4)4	[45]
K+	Sc3+		KSc(BH4)4	11.3	Orthorhombic	Pnma	BaSO4	[46]
K+	Y3+		o-KY(BH4)4 m-KY(BH4)4	8.6	Orthorhombic Monoclinic	Cmcm C2/c	NaSc(BH4)4 rt-LaNbO4	[38,43] [38]
K+	La3+		K3La(BH4)6	7.0	Monoclinic	P21/n	Double-perovskite	[49]
K+	Ce3+		K3Ce(BH4)6	7.0	Monoclinic	P21/c	Double-perovskite	[48]
K+	Sm2+		KSm(BH4)3	5.2	Monoclinic	P21cn	Perovskite-related	[51]
K+	Gd3+		KGd(BH4)4 K2Gd(BH4)5 K3Gd(BH4)6	6.3 6.5 6.7	Monoclinic Monoclinic Monoclinic	P21/c P21/m P21/c	LiMnF4 New Double-perovskite	[52] [52] [52]
K+	Ho3+		KHo(BH4)4	6.1	Orthorhombic	Cmcm	NaSc(BH4)4	[32]
K+	Er3+		KEr(BH4)4	6.1	Orthorhombic	Cmcm	NaSc(BH4)4	[44]
K+	Yb2+		LT-KYb(BH4)3 HT-KYb(BH4)3	4.7	Cubic Orthorhombic	P$\overline{4}$3m Pmc21	Perovskite-related Perovskite-related	[53] [53]
K+	Yb3+		KYb(BH4)4	5.9	Orthorhombic	Cmcm	NaSc(BH4)4	[45]
Rb+	Y3+		o-RbY(BH4)4 m-RbY(BH4)4 Rb3Y(BH4)6	6.9 6.9 5.6	Orthorhombic Monoclinic Cubic	Pnma P21/c Fm$\overline{3}$	KSc(BH4)4 Distorted KSc(BH4)4 Double-perovskite	[38] [50] [38,53]
Rb+	Ce3+		Rb3Ce(BH4)6	5.0	Monoclinic	P21/c	Double-perovskite	[53]
Rb+	Sm2+		RbSm(BH4)3	4.3	Orthorhombic	Pbn21	Perovskite-related	[51]
Rb+	Eu2+		RbEu(BH4)3	4.3	Orthorhombic	Pna21	Perovskite-related	[53]
Cs+	Y3+		CsY(BH4)4 Cs2Y(BH4)5 Cs3Y(BH4)6	5.7 4.2	Tetragonal Hexagonal Cubic	I41/a P63cm Fm$\overline{3}$	CaWO4 new Double-perovskite	[50] [38] [38,53]
Cs+	Sm2+		α -CsSm(BH4)3 α’-CsSm(BH4)3 β-CsSm(BH4)3	3.7	Orthorhomic Tetragonal Cubic	P21212 P$\overline{4}$m2 Pm$\overline{3}$m	Perovskite-related Perovskite-related Ideal perovskite	[51] [51] [51]
Cs+	Eu2+		CsEu(BH4)3	3.7	Tetragonal	P4/mbm	Perovskite-related	[53]
Cs+	Gd3+		Cs3Gd(BH4)6	3.8	Cubic	Fm$\overline{3}$	Double-perovskite	[53]
Li+	K+	La3+	Li3K3La2(BH4)12	8.1	Cubic	Ia$\overline{3}$d	Garnet	[48]
Li+	K+	Ce3+	Li3K3Ce2(BH4)12	8.1	Cubic	Ia$\overline{3}$d	Garnet	[48]
Li+	Rb+	Y3+	Rb2LiY(BH4)6	6.8	Cubic	Fm$\overline{3}$	Double-perovskite	[38,50]
Li+	Cs+	Y3+	Cs2LiY(BH4)6	5.4	Cubic	Fm$\overline{3}$	Double-perovskite	[38,50]
Li+	Cs+	Ce3+	Cs2LiCe(BH4)6	4.8	Cubic	Fm$\overline{3}$	Double-perovskite	[53]
Li+	Cs+	Gd3+	Cs2LiGd(BH4)6	4.7	Cubic	Fm$\overline{3}$	Double-perovskite	[53]

). Using the same analogy to oxide structures, Li₃K₃La₂(BH₄)₁₂ and Li₃K₃Ce₂(BH₄)₁₂ (space group Ia$\overline{3}$d) are reported to have a garnet-type structure [48].

All other reported bi- and tri-metallic borohydrides with structure types not discussed above, are included in Table 4.

4. Thermal Properties of Rare Earth Borohydrides

4.1. Monometallic and Halide Substituted Rare Earth Borohydrides

The metathesis reaction between stoichiometric quantities of lithium borohydride LiBH₄ and a trivalent rare earth chloride RECl₃ results in principle in the formation of either a monometallic trivalent RE-borohydride, RE(BH₄)₃, LiRE(BH₄)₄ or a chloride substituted compound, LiRE(BH₄)₃Cl, according to Equations (2)–(4) (see above). During the past seven years extensive efforts have been made to understand the thermal decomposition and rehydrogenation behavior of these compounds. The following sections summarize the thermal behavior of two compounds: Y(BH₄)₃ and LiCe(BH₄)₃Cl, which are representative of the monometallic and halide substituted rare earth borohydrides adopting the same structure type.

4.1.1. Yttrium borohydride

Y(BH₄)₃ has a high gravimetric hydrogen capacity of 9.1 wt %, and has therefore received considerable interest since 2008 as a potential hydrogen storage material [18]. The stability of crystalline Y(BH₄)₃ has been assessed by density functional theory (DFT), and calculations predict the decomposition into boron, yttrium hydride and hydrogen at elevated temperatures. Dai et al. [54] have recently employed first principles calculations to predict the influence of additional alloying elements such as Li, Na, K, Ti, Mn and Ni on the stability and dehydrogenation properties of Y(BH₄)₃. The authors concluded that all alloyed systems show smaller enthalpy of decomposition than pure Y(BH₄)₃ and that the alkali metals (especially K) may be suitable for improving the dehydrogenation properties of Y(BH₄)₃.

Experimental studies regarding the thermal properties of Y(BH₄)₃ prepared by liquid phase and mechanochemical synthesis have been described in references [21,22,23]. According to these reports, α-Y(BH₄)₃, which is the stable modification at RT, undergoes a structural transition from the distorted ReO₃-type structure (cubic, Pa$\overline{3}$) into the ordered ReO₃-type structure (cubic, Fm$\overline{3}$c) upon heating to 157–182 °C. In situ SR-PXD measurements on a LiBH₄-Y(BH₄)₃ (3:1) sample heated under dynamic vacuum with a rate of 2 °C/min up to 450 °C shows the presence of three distinct temperature regions (Figure 9a,b) [21].

LiCl and α-Y(BH₄)₃ are the only crystalline phases present in region (I) which spans from RT to 200 °C. The peaks originating from Y(BH₄)₃ begin to decrease in intensity at 182 °C and completely disappear at 200 °C. The formation of β-Y(BH₄)₃ was not observed under the current experimental conditions, which suggests that the heating rate of 2 °C/min is probably too fast compared to the kinetics of the phase transformation. In region (III), above 262 °C, the pattern shows that LiCl and YH₃ were the only crystalline materials present. In region (II), between 200 and 247 °C, the major crystalline phases are LiCl and YH₃. Additionally, weaker peaks of a new phase emerge at around 200 °C gain in intensity up to 232 °C and diminish again until they finally disappear above 247 °C. Peak indexing and unit cell determination for this intermediate phase have indicated that it might be described in a primitive orthorhombic unit cell with lattice parameters a = 12.170(14) Å, b = 7.670(5) Å, and c = 7.478(6) Å. Similarly, Ravnsbæk et al. [22] have observed the formation of two unidentified compounds after the decomposition of Y(BH₄)₃ in the temperature range ~190 to 235 °C. The authors suggested that one of the phases has a composition similar to Y(BH₄)Cl₂ or Y(BH₄)₂Cl, while the other phase could be a decomposition product containing Y‒B‒Cl. Jaroń and Grochala [20] attempted to follow the thermal decomposition of Y(BH₄)₃ by heating samples in alumina crucibles inside a TG-DSC (Thermogravimetric-differential calorimeter) up to selected temperatures of 194, 210 and 229 °C followed by rapid cooling. The PXD patterns above 229 °C only show the presence of LiCl, with other decomposition products being amorphous. The partially decomposed samples possess a layered morphology and progressive color changes from white to light yellow, brown and finally black at high temperatures. This color change is attributed to the presence of amorphous decomposition products such as polymeric (B_xH_y)_n compounds as well as H-poor Y‒B‒H phases. The same authors also studied the hydrogen desorption kinetics of α-Y(BH₄)₃ as well as that of the HT modification of β-Y(BH₄)₃. They observed significant differences below 260 °C for both phases, the decomposition being faster and more facile for the HT phase which they attributed to a three times lower activation energy for thermal decomposition as compared to the low-temperature polymorph [24]. Remhof et al. studied the thermal decomposition of pure Y(BH₄)₃ prepared by direct, solvent-free synthesis from YH₃/YH₂, B₂H₆ and H₂ via reactive milling [55]. They observed the onset for hydrogen desorption at 187 °C and a maximum desorption rate of 0.12 wt % H₂/min at 250 °C when the materials were heated with 1 °C/min under 1 bar H₂. Time-resolved in situ PXD data with a low heating rate of 0.2 °C/min collected between RT and 450 °C under 1 bar of hydrogen showed only reflections from Y(BH₄)₃ and unreacted YH₃ up to 207 °C, while at 247 °C the Bragg peaks originating from Y(BH₄)₃ disappeared and only YH₃ remained visible. No structural transition from the α- to the β-phase was observed. Above 347 °C the decomposition was completed and no further change in the peak intensities was observed until the transformation of YH₃ into YH₂ proceeded at 447 °C. The authors also reported the appearance of at least one amorphous phase in addition to YH₃. However, they did not provide any insight as to the nature of these additional phases. Finally, Park et al. have undertaken a comparative study between Y(BH₄)₃ synthesized from LiBH₄-YCl₃ (3:1 and 4:1 ratios) by mechanochemistry and via direct synthesis from YH₃, B₂H₆ and H₂ [56]. The authors arrived at the conclusion that both methods result in products that possess distinctly different polymorphic phase transformation, melting behavior and even altered decomposition pathways. In fact, it turned out that the side product LiCl in the metathesis reaction cannot be regarded as inert as it shifts the melting temperature of Y(BH₄)₃ and apparently promotes the formation of YB₄ during decomposition which is otherwise absent in the case of Y(BH₄)₃ obtained from gas-solid reaction between YH₃, B₂H₆ and H₂.

4.1.2. Lithium Cerium Borohydride Chloride

LiCe(BH₄)₃Cl can be obtained via mechanochemical synthesis between three molar equivalents of LiBH₄ and one molar equivalent of CeCl₃ [8,36]. The thermal decomposition of LiCe(BH₄)₃Cl has been studied in reference [36] by a combination of TG-DSC and residual gas analysis (RGA) with a low heating rate of 2 °C/min under argon flow. According to this report, decomposition proceeds in a two-step process with minor gas release at about 215 °C followed by the major decomposition step at a peak temperature of 246 °C. RGA shows a more than 100 times higher partial pressure for hydrogen than for B₂H₆ and its fragments B₂H₅, B₂H₄ and BH₃. This indicates that the decomposition proceeds almost exclusively by release of hydrogen and that boron is thus kept in the solid state, which is one of the prerequisites for reversible hydrogen storage in borohydrides. Ley et al. [8] have performed Sieverts measurements on a sample containing only LiCe(BH₄)₃Cl and LiCl which show a gas release of 3.75 wt % H₂ during the first desorption and 1.25 wt % H₂ during the second and third desorption. Samples containing one equivalent excess of LiBH₄ show a different thermal behavior and exhibit a gas release of 0.5 wt % H₂ up to 220 °C, while 3.25 wt % H₂ are released between 220 and 270 °C. An additional amount of 0.8 wt % H₂ is released between 270 and 500 °C, while during a subsequent isotherm (3 h) about 0.15 wt % H₂ are released. Thus a total amount of 4.7 wt % H₂ are desorbed during heating to 500 °C, while after rehydrogenation at 400 °C and 100 bar for 24 h, about 1.8 wt % of H₂ are desorbed during the second dehydrogenation [8].

In situ SR-PXD experiments have been conducted on the hydride and deuteride form of LiCe(BH₄)₃Cl in order to determine the decomposition products and formation of possible intermediates [8,36]. Figure 10a shows the evolution of powder patterns between 80 and 600 °C with a heating rate of 2 °C/min for a (3LiBD₄-CeCl₃) mixture obtained after ball milling. Figure 10b displays powder patterns at selected temperatures in more detail.

The only crystalline phases present after ball milling are LiCe(BD₄)₃Cl and LiCl. In the low-temperature region between 80 and 150 °C, the crystallinity of the material increases upon heating. As a result the X-ray profiles become narrower and the overall peak intensity increases. Peak intensities for LiCe(BD₄)₃Cl decrease rapidly above 230 °C up to its decomposition peak temperature of 240 °C. Between 270 and 450 °C the SR-PXD pattern exhibits peaks belonging to LiCl only, and a modulated background with a broad and featureless region around 2θ = 10–15°, which is characteristic of some amorphous boron-containing compounds such as Ce‒B‒D or B‒D species. Above 450 °C, new peaks emerge and gain intensity up to the maximum temperature of 600 °C. They can be assigned to CeD₂, and have been highlighted in Figure 10b. The formation of cerium borides such as CeB₄ or CeB₆ could not be observed under the applied experimental conditions. Ley et al. performed in situ SR-PXD studies with a high heating rate of 8 °C/min up to 500 °C on several nLiBH₄-CeCl₃ mixtures (n = 3, 4) [8]. The authors observe the decomposition of LiCe(BH₄)₃Cl at 240 °C similar to [36], and the appearance of an unknown phase above 240 °C—possibly a ternary salt composed of lithium cerium and chloride. Major decomposition products at 500 °C are CeH₂ and CeB₆. Crystalline CeB₆ has also been obtained by Liu et al. [26] while heating up a 3LiBH₄-CeCl₃ mixture to 600 °C for an extended period of time. None of the studies found any indication that favored the formation of boron over CeB₆. Based on those findings the thermal decomposition of LiCe(BH₄)₃Cl is therefore believed to proceed via Equation (5) rather than via Equation (6):

LiCe(BH₄)₃Cl → ½CeH₂ + ½CeB₆ + 11/2H₂ + LiCl

(5)

LiCe(BH₄)₃Cl → CeH₂ + 3B + 5H₂ + LiCl

(6)

The first reaction pathway is furthermore supported by TG-DSC experiments which show a weight loss of 3.5 wt % H₂ for a 3LiBH₄-CeCl₃ hydride mixture after ball milling. This value is in excellent agreement with the expected weight loss of 3.6 wt % H₂, assuming the formation of CeH₂ and CeB₆ in Equation (5).

Trends in the thermal decomposition of the monometallic and halide substituted RE-borohydrides are difficult to establish despite an increasing number of publications in the field. This complication is mainly due to the use of different preparation methods e.g., mechanochemistry vs. wet-chemistry. Ball milling of LiBH₄-RECl₃ mixtures always yields LiCl in addition to the RE-borohydride which cannot be simply regarded as “dead weight”. In fact, its presence can affect the melting point, phase transformation temperatures, induce halide substitution and even alter the decomposition pathway. Wet-chemical methods have therefore been developed which can separate the alkali metal chloride from the metal borohydride, e.g., through dissolution and/or filtering. However, they lead to the formation of solvated RE-borohydrides which have to be carefully dried at elevated temperatures and in vacuo. Here arises another risk in which the solvent cannot be fully removed or that the metal borohydride partially decomposes during the drying process. A second factor is based on the use of different characterization methods such as calorimetry and in situ powder diffraction in which vastly different experimental conditions are employed. In the case of calorimetric studies the purge gas (Ar/He/H₂), heating rate, sample mass and choice of sample container (Al/Al₂O₃) have a profound effect on the number, intensity, shape and position of the registered heat flow signals. In the case of in situ diffraction the use of H₂-backpresure vs. vacuum is known to alter the decomposition pathway. Furthermore, reactions between the borohydrides and the sample holder (mainly sapphire) at high temperature in connection with micro-leaks have led to the formation of oxide species on several occasions [57].

A correlation between the decomposition temperature of the metal borohydrides and their Pauling electronegativity was already suggested in 1955 [58,59] and has been thoroughly investigated by several research groups both experimentally and theoretically [12]. The more stable alkali metal borohydrides tend to show a linear correlation between the decomposition temperature and the electronegativity of the metal cation.

However, compounds formed by metals with higher electronegativity tend to deviate from this linear behavior. An exponential correlation between the decomposition temperature and the electronegativity of the metal has therefore been suggested instead.

Table 5. Thermal properties of selected monometallic RE-borohydrides with trivalent cation.

Cation	Compound	Tdec. (°C)	Decompositon Products	Comments	Ref.
La3+	La(BH4)3	258	Not specified	Material obtained from thermolysis of La(BH4)3 nS(CH3)2; 8% mass loss due to simultaneous release of B2H6 and H2	[34]
Ce3+	Ce(BH4)3	251	Not specified	Material obtained from thermolysis of Ce(BH4)3 nS(CH3)2; 8.5% mass loss due to simultaneous release of B2H6 and H2	[34]
Pr3+	Pr(BH4)3	275	No crystalline products	Sample synthesized in toluene and DMS. In situ SR-PXD showed no crystalline decomposition product until 500 °C.	[30]
Gd3+	Gd(BH4)3	260	Not specified	Sample obtained from thermal decomposition of Gd(BH4)3*nS(CH3)2; DSC shows endothermic event at 260 °C related to hydrogen desorption. MS shows minor contributions of B2H6 between 225 and 275 °C.	[60]
Tb3+	Tb(BH4)3	213 and 253	Not specified	3LiBH4-TbCl3 mixture after ball milling (2 h). DSC trace shows two endothermic events with peak temperatures of 213 and 253 °C	[19]
Dy3+	Dy(BH4)3	No data	No data	Only structural information available; decomposition is believed to occur similar to Y(BH4)3 and Gd(BH4)3	[18]
Ho3+	Ho(BH4)3	170 225 255	No crystalline products	Three distinct endothermic events visible in the DSC up to 400 °C; onset of hydrogen release observed at 170 °C with the fastest rate at 255 °C	[32]
Er3+	Er(BH4)3	230	ErH2 and ErB4	Hydrogen desorption starts at 230 °C; multiple endothermic events visible in DSC; 3.2% mass loss observed up to 400 °C	[30]

and

Table 6. Thermal properties of lithium rare earth borohydride chlorides.

Cation1	Cation 2	Compound	Tdec. (°C)	Decompositon Products	Comments	Ref
La3+	Li+	LiLa(BH4)3Cl	226 and 266	LaH2 and LaB6	TG/DSC shows two endothermic events between 200 and 300 °C and total mass loss of 2.87%	[9]
Ce3+	Li+	LiCe(BH4)3Cl	a 215, a 246, b 240, b 266	CeH2 and CeB6	a 3LiBD4-CeCl3 mixture after ball milling; two endothermic events at 215 and 246 °C observed by DSC b 3LiBH4-CeCl3 mixture after ball milling: major DSC events at 213 (melting), 240, and 266 °C (both decomposition)	[8,36]
Pr3+	Li+	LiPr(BH4)3Cl	a 176 b 212 and 247	PrH3 and PrB6	a Synthesised in (C2H5)2O. Sample is a mixture between Pr(BH4)3 and LiPr(BH4)3Cl; Tdec determined by TG/DSC b 3LiBH4-PrCl3 mixture after ball milling.TG/DSC shows two endothermic events at 212 °C and 247 °C with 0.7% and 2.6% mass loss, respectively	[30]
Nd3+	Li+	LiNd(BH4)3Cl	223 and 241	Not specified	3LiBH4-NdCl3 mixture after ball milling. TG/DSC shows two endothermic events at 223 °C and 241 °C with 0.4% and 2.4% mass loss respectively	[19]
Gd3+	Li+	LiGd(BH4)3Cl	261	GdB4	Value determined from thermogravimetric experiment	[9]

summarize the thermal properties of selected monometallic rare earth borohydrides RE(BH₄)_n (n = 2, 3) and lithium rare earth borohydride chlorides LiRE(BH₄)₃Cl.

For the monometallic trivalent compounds the decomposition temperature decreases with increasing electronegativity of the metal. It reaches its highest value of 258 °C for La(BH₄)₃, 275 °C for Pr(BH₄)₃, and 260 °C for Gd(BH₄)₃ [30,34,60], and drops to 230 °C in the case of Er(BH₄)₃ [30]. Ultimately the decomposition reaction results in the formation of the corresponding rare earth hydrides and borides which is favored over the formation of elemental boron for technological applications. Crystalline decomposition products are not always formed, however, even when heating the materials to 400–500 °C, as is the case for Pr(BH₄)₃ and Ho(BH₄)₃ [30,32]. The thermal decomposition of the LiRE(BH₄)₃Cl-type compounds appears to be more complex and proceeds via a multi-step process involving at least two different steps over a temperature range of up to 50 °C. The highest decomposition temperature of 266 °C is observed for LiLa(BH₄)₃Cl while LiNd(BH₄)₃Cl possesses the lowest one with 241 °C. The variation in decomposition temperature is far less pronounced in the rare earth borohydrides as compared to alkali or alkaline earth borohydrides though. This is not unexpected since the range of electronegativity observed for the lanthanides is rather small and increases only moderately from 1.10 for Lanthanum to 1.23 for Erbium. In case of the LiRE(BH₄)₃Cl-type compounds the less electronegative alkali metal (Li) shows dominantly ionic interactions and the overall stability of compound is controlled by the more electronegative cation (RE) which is in fact part of a complex anion [RE₄Cl₄(BH₄)₁₂]⁴⁻. The difference of 25 °C in decomposition temperatures between LiLa(BH₄)₃Cl and LiNd(BH₄)₃Cl indicates that the decomposition temperatures are specific for these complex anions and directly reflects differences in their stability [1].

4.2. Composites Containing RE-Borohydrides and LiBH₄

LiBH₄ has the highest theoretical hydrogen storage capacity (18.5 wt % H₂) of all borohydrides of which 13.9 wt % H₂ are accessible through Equation (7):

LiBH₄ → LiH + B + 3/2H₂

(7)

However, harsh pressure and temperature conditions are necessary to reversibly release (>400 °C) and reabsorb (150 bar H₂; 650 °C) hydrogen from LiBH₄ due to its high thermal stability (ΔH = 74 kJ/mol H₂) [61]. The RE-borohydrides on the other hand decompose between 200 and 300 °C, which is significantly lower than the decomposition temperature of LiBH₄. There has been considerable interest in utilizing RE-borohydrides as part of a composite mixture with other light weight metal borohydrides such as LiBH₄. The general idea behind the use of composite mixtures over pure compounds is to reduce the enthalpy and consequently the hydrogen desorption temperature of the more stable LiBH₄. Indeed, additives such as MgH₂ [62,63], CaH₂ [64,65], CeH₂/LaH₂ [66] and Al [67,68] have been used successfully to lower the stability of LiBH₄ by the formation of MgB₂, CaB₆ and AlB₂, respectively. For a composite containing LiBH₄-MgH₂ (2:1), a two-step decomposition is observed at high temperatures (>400 °C) [62,63,69] while for LiBH₄-CaH₂ (6:1) decomposition occurs in a single step reaction, which is advantageous for reversible hydrogen storage applications [64]. In addition to hydrides, halides such as TiCl₃, TiF₃ and ZnF₂ have been recently employed to reduce the decomposition temperature of LiBH₄ through cation exchange and formation of an unstable transition metal borohydrides [70]. LiBH₄ has also been destabilized by the addition of FeCl₂, CoCl₂ and NiCl₂, however, in the case of CoCl₂, diborane is released during dehydrogenation which results in capacity loss [71]. Gennari et al. showed that composites of 6LiBH₄ + RECl₃ (RE = Ce, Gd) form RE-hydrides during thermal decomposition which in turn destabilize the remaining LiBH₄ [28]. A composite mixture of LiBH₄-CeCl₃ (6:1) showed a reversible capacity of 40% after rehydrogenation at 60 bar and 400 °C. In addition, a change in the decomposition pathway and increased formation of CeB₆ and GdB₄ has been reported for composites desorbed against 5 bar of hydrogen pressure as compared to vacuum [28]. In order to promote the reversibility of CeB₆ formed during the decomposition of Ce(BH₄)₃, the same authors added LiH to as-milled LiBH₄-CeCl₃ (6:1) composites and then further milled for 1 h. PXD data showed the presence of LiCl and CeH_2+x after the final milling step. Such a LiBH₄-CeCl₃-LiH (6:1:3) composite released 5.3 wt % H₂ during the first dehydrogenation and 4.4 wt % H₂ during the second cycle after rehydrogenation under relatively mild conditions (400 °C, 60 bar, 2 h) which corresponds to 80% reversible capacity.

Olsen et al. have intensively studied the structure, thermal behavior and hydrogen storage properties of a series of LiBH₄-RECl₃ (RE = La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Er, Yb and Lu) composites obtained by ball milling LiBH₄ and the respective RE-chlorides in a molar ratio of 6:1 [19]. The systematic study has shown that ball milling results in the formation of either LiRE(BH₄)₃Cl (RE = La, Ce, Pr, Nd, Sm, Gd), α,β-RE(BH₄)₃ (RE = Sm, Gd, Tb, Dy, Er Tm and Yb) or LiRE(BH₄)₄ (RE = Yb, Lu) together with excess LiBH₄ [19]. TG-DSC combined with in/ex situ PXD show that the composites which contain LiRE(BH₄)₃Cl decompose at around 200 °C, and exhibit multiple decomposition steps with the formation of several intermediate phases. The remaining LiBH₄ in these composites decomposes and releases hydrogen below 350 °C, which is much lower than the decomposition temperature of pure LiBH₄ [61]. In contrast, the thermal decomposition of the composites which contain α,β-RE(BH₄)₃ and excess LiBH_4, proceeds with only two major decomposition steps in the temperature range 250–350 °C. Common to all composites, is the formation of REH₂ as the main decomposition product and REB_n (n = 4, 6) species at temperatures in excess of 400 °C. It was concluded that the presence of REH₂ has a destabilizing effect on the remaining LiBH₄ in all the composites, as was already exemplary reported by Gennari et al. for RE = Ce and Gd [28]. Rehydrogenation of the La-system under 100 bar and 300 °C showed only very limited rehydrogenation capacity of 18%, whereas rehydrogenation of the Er-system (100 bar, 400 °C) showed a slightly higher capacity of 25% [19].

More recently, a selection of composites containing either La, Er or Pr have been synthesized to determine whether the composites obtained by mechanochemistry differ from those made by wet-chemical methods [30,72,73]; whether the composites which contain alkali chlorides as by-products show a different thermal behavior from those which are halide-free; and determine the role of additives such as LiH for the rehydrogenation behavior of the composites. Composite mixtures of RE-borohydrides of La, Er and LiBH₄ and LiH were obtained by mechanochemical methods [72]. Ex situ PXD and TG-DSC measurements showed that heating to 400 °C yielded LaB₆, ErB₄ and REH_2+δ (0 < δ < 1) as major decomposition products and that LiBH₄ was destabilized by REH_2+δ formed through decomposition of the parent borohydrides LiLa(BH₄)₃Cl and Er(BH₄)₃, respectively. Therefore, its decomposition temperature was reduced by as much as 100 °C compared to that of pure ball milled LiBH₄. The La-containing composite showed only a limited reversible capacity of 20% (340 °C, 100 bar), probably caused by hydrogen uptake of some amorphous boron-containing phases; similar to the limited rehydrogenation capacity found in the 6LiBH₄-LaCl₃ composite after ball milling that did not contain additional LiH [19]. In contrast, the erbium-containing 6LiBH₄-ErCl₃-3LiH composite showed more promising hydrogen storage properties and could be rehydrogenated at 340 °C and 100 bar hydrogen to a large extent. Such a composite mixture showed a reversible capacity of 66% when desorbed in vacuo, which increased to 80% when desorbed against 5 bar of hydrogen backpressure.

The structure and thermal properties of halide-free RE-borohydrides (RE = Pr, Er) have also been obtained by solvent-based synthesis instead of mechanochemistry [30,73]. Both compounds showed a simple decomposition pathway into the corresponding hydrides and borides according to Equations (8) and (9):

Er(BH₄)₃ → 0.75ErB₄ + 0.25ErH₂ + 5.75H₂

(8)

Pr(BH₄)₃ → 0.5PrB₆ + 0.5PrH₂ + 5.5H₂

(9)

Rehydrogenation of both materials was monitored by ex situ and in situ SR-PXD measurements at 340 °C and hydrogen pressures of up to 100 bar. However, no new crystalline phases were observed. In addition, TG-DSC experiments of the samples exposed to 100 bar of hydrogen for 21 h showed the absence of any mass loss and heat flow signals, indicating that rehydrogenation was unsuccessful. These findings are in disagreement with Gennari et al. who reported that Er(BH₄)₃ synthesized by ball milling had a reversible hydrogen capacity of 20% (rehydrogenation at 400 °C, 60 bar H₂) [74]. This “partial reversibility” for several RE-borohydride containing composites of the mechanomilled materials is most likely due to the rehydrogenation of LiBH₄ and not due to the formation of the RE-borohydrides from the corresponding RE-hydrides or borides.

The decomposition and rehydrogenation behavior of composite mixtures containing the halide-free Er(BH₄)₃ ball milled for one hour in the presence of LiBH₄ and/or LiH has been further investigated in reference [73]. Two composites were studied in detail: Er(BH₄)₃-LiH (1:6) and LiBH₄-Er(BH₄)₃-LiH (3:1:3). Hydrogen sorption measurements in a Sieverts-type apparatus for the 3:1:3 composite showed a hydrogen release of 4.2, 3.7 and 3.5 wt % H₂ during three desorption-absorption cycles (des: 400 °C, 5–10 bar H₂; abs: 340 °C, 100 bar H₂) and thus good rehydrogenation behavior with reversible hydrogenation capacity in the range of 80–85%. This halide-free composite showed similar behavior to a 6LiBH₄-ErCl₃-3LiH composite mixture [72] obtained by ball milling. In both systems, the major decomposition products were ErH_2+δ (0 < δ < 1) and ErB₄ while rehydrogenation leads to the formation of ErH₃ and the rehydrogenation of LiBH₄. Between the two studies, minor differences were observed in the number and variety of ErH₃ polymorphs formed upon rehydrogenation. While cubic-ErH₃ was the only polymorph reported in [72], a mixture of tetragonal-ErH₃ and cubic-ErH₃ polymorphs were reported after rehydrogenation (340 °C, 93 bar H₂) in [73].

The influence of additives such as LiH on the rehydrogenation behavior of Er(BH₄)₃ was investigated in reference [30]. For this purpose, halide-free Er(BH₄)₃ was decomposed at 400 °C and rehydrogenated at 340 °C and 100 bar of hydrogen. PXD experiments on the desorbed and rehydrogenated samples did not show any crystalline phases while TG-DSC data measured for the rehydrogenated sample did not show any mass loss and no heat flow signal up to 400 °C (Figure 11). In two control experiments, either 100 wt % of LiCl or 50 wt % of LiH were added to the halide-free Er(BH₄)₃ samples, and decomposed and rehydrogenated as before.

The composite mixture containing additional LiCl did not show any crystalline phases after rehydrogenation. Neither heat flow signals nor weight loss could be detected after rehydrogenation which clearly demonstrated that the presence of LiCl had no beneficial effect on the rehydrogenation behavior. The composite mixture containing 50 wt % LiH was studied by in situ SR-PXD (Figure 12). While the ball milled sample mainly showed the presences of LiH and a modulation of the background indicative of some amorphous boron-containing phases such as ErB₄ when heated from RT to 400 °C, the rehydrogenated material (100 bar H₂) showed the evolution of peaks belonging to ErH₃ at 400 °C and upon cooling to RT the appearance of peaks originating from h-LiBH₄.

The unsuccessful rehydrogenation of the composites containing LiCl vs. LiH clearly demonstrates that the influence of LiH as an additive goes beyond simply being a source of Li-ions. The successful formation of LiBH₄ for the LiH-containing composite is an analogous to that of similar reactive hydride composites with LiH, such as LiH + MgB₂. In the latter case, however, much harsher conditions or the use of additional catalysts is necessary [62,75].

5. Conclusions

The past 10 years have led to the discovery of a multitude of new borohydrides containing RE elements which show rich chemistry, structural diversity and a wide range of compositions. Their application ranges from hydrogen and thermal heat storage to solid state electrolytes. Lately, exciting new fluorescent and magnetocaloric properties have been discovered in several RE-borohydrides such as CsEu(BH₄)₃ and K₂Gd(BH₄)₅ [52,53]. In the future, it will hopefully be possible to design new advanced functional materials based on RE-borohydrides with desired compositions and new exciting properties that make their entrance into many more applications and research areas. A promising new field could be the use of RE-borohydrides as advanced fast neutron shields [76] for compact and portable neutron sources [77]. Such mobile units are in high demand for materials testing and neutron capture therapy, which is a promising approach to cancer therapy where conventional radiation therapies fail.