An Old Dog with New Tricks – Additions to the Cesium Lithium Chloride System: Cs₃Li₂Cl₅ and the Hydrated Cs₃LiCl₄ · 4H₂O

Abstract

The CsCl/LiCl system has been studied for over a century now. Numerous phases have been predicted – only three have hitherto been found. We present the synthesis and single-crystal structure of the cesium lithium pentachloride Cs₃Li₂Cl₅, predicted earlier but with a different structure. The anhydrous new phase readily reacts to Cs₃LiCl₄ · 4H₂O in air. The tetrahydrate can also be obtained through the simplest, most inexpensive and green synthesis possible: an immediate, room-temperature mechanosynthesis from only CsCl and LiCl (3 : 1) in air. Differential scanning calorimetry (DSC) and thermogravimetric analyses (TGA), as well as in situ temperature-dependent powder X-ray diffraction studies on this second ever reported ternary alkali chloride hydrate allowed for a revision of the CsCl/LiCl phase diagram. Density of states and total energy calculations further elucidate the new alkali chlorides and update the relative stability of previous structure predictions.

Introduction

The CsCl/LiCl system has been studied for over a century now, starting with the observation of CsLiCl₂ and “Cs₂LiCl₃” in 1915.^[5] The first single-crystal structure was reported in 1983: the high-temperature modification β-CsLiCl₂ in the tetragonal space group P4/nmm.^[6] Jansen et al.^[7] proposed a multitude of structures based on theoretical predictions in 2007: Cs₃LiCl₄ in Amm2, Cs₂LiCl₃ in Cmcm, and Cs₃Li₂Cl₅ in Cm, they computationally confirmed β-CsLiCl₂ in P4/nmm, and further predicted Cs₂Li₃Cl₅ in Imm2, CsLi₂Cl₃ in Immm, and CsLi₃Cl₄ in the space group Cmcm. Among these predicted structures, only Cs₂LiCl₃, the already experimentally found β-CsLiCl₂, and CsLi₂Cl₃ are thermodynamically stable with respect to decomposition into the binary compounds CsCl and LiCl.^[7] In 2012, the same group experimentally found the monoclinic low-temperature modification of β-CsLiCl₂, namely α-CsLiCl₂ in C2/c, and the predicted CsLi₂Cl₃ in Pbcn, raising the number of experimentally observed single-crystal structures of CsCl/LiCl compounds to three.^[8] The discovery of α-CsLiCl₂ allowed to adjust the empirical potential for related structure predictions. As a result, the predicted structures for Cs₂LiCl₃ and CsLi₂Cl₃ were re-evaluated and pushed to a lower local energy minimum.^[8] The Cs₃Li₂Cl₅, predicted over ten years ago,^[7] was now experimentally realized, albeit, with a completely different crystal structure. This cesium lithium chloride we herein report is the cesium-richest representative obtained to date. Cs₃Li₂Cl₅ adds to the plentiful system, where the number of predicted compounds still far exceeds the number of experimentally observed ones.

Surprisingly, in the wide field of hydrated ternary purely alkali chlorides, only two crystal structure reports have been given to this date. Romankiw et al.^[9] restudied^[10] the CsCl/NaCl/H₂O system in 1984, and described CsNa₂Cl₃ · 2H₂O, found to crystallize in the monoclinic space group I2/c. The second known structure pertains to the CsCl/LiCl/H₂O system. In 1953, Cs₂LiCl₃ · 4H₂O was observed for the first time.^[11] In 2016, an initial structure description of “LiCl · 2CsCl · 4H₂O” was reported by Khripun et al.^[12] Recently, the same group published a study of the enthalpies of dissolution of Cs₃LiCl₄ · 4H₂O and dry mixtures of lithium and cesium chlorides in the molar ratio of 1 : 3.^[13] In the following, we firstly present a complete single-crystal structure of the hydrated cesium lithium chloride. Cs₃LiCl₄ · 4H₂O is only the second representative of a hydrated, pure alkali-metal chloride. It is the first representative of the Cs-Li-Cl-H₂O substance class.

In this publication, we report on the synthesis, single-crystal structure, and properties of the new cesium lithium chloride Cs₃Li₂Cl₅ and on the first cesium lithium chloride hydrate Cs₃LiCl₄ · 4H₂O. Furthermore, we have studied the thermal decomposition of the hydrate through differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA), as well as in situ temperature-dependent powder X-ray diffraction. The results led to a revision of the CsCl/LiCl phase diagram. In addition, using density functional theory (DFT), we re-evaluated the stabilities of the predicted structures by Jansen et al.^[7–8] with respect to the new Cs₃Li₂Cl₅.

Results

Crystal structure of Cs₃Li₂Cl₅

The water-free cesium lithium chloride Cs₃Li₂Cl₅ crystallizes in the monoclinic, centrosymmetric space group P2₁/m (no. 11) with a = 677.78(8), b = 707.54(8), c = 1329.5(2) pm and β = 103.65(1)°. A different structure in a subgroup of P2₁/m, Cm, was predicted for Cs₃Li₂Cl₅ by Jansen et al. in 2007.^[7] The single-crystal data collection and evaluation of Cs₃Li₂Cl₅ is specified in Table 1. The according Wyckoff positions, atomic coordinates, and isotropic displacement parameters of Cs₃Li₂Cl₅ are listed in Table S1 of the Supporting Information (SI). Table S2 of the SI gives the equivalent anisotropic displacement parameters.

Table 1.

Crystallographic and structure refinement data of Cs₃Li₂Cl₅.

Empirical formula	Cs3Li2Cl5
Molar mass /g mol-1	589.86
Crystal system	monoclinic
Space group	P21/m (no. 11)
Formula units per cell	Z = 2
Temperature /K	100(2)
Single crystal diffractometer	Bruker APEX II CCD
Radiation	Mo-Kα (λ = 71.075(1) pm)
	(graphite monochromator)
a /pm	677.78(8)
b /pm	707.54(8)
c /pm	1329.5(2)
β /°	103.65(1)
V /10-6 pm3	619.6(2)
Calculated density /g cm-3	3.162
Crystal size /mm3	0.28 × 0.14 × 0.12
Absorption coefficient /mm-1	9.792
F(000) /e	512
θ range /°	1.6 – 33.3
Index rage	-10 ≤ h ≤ 10,
	0 ≤ k ≤ 10,
	0 ≤ l ≤ 20
Reflections total / independent	2533 / 2528
Rint with I ≥ 3σ(I) (all domains)	0.0382
Reflections with I ≥ 2σ(I)	2458
Data / ref. parameters	2528 / 53
Absorption correction	multi-scan (twinabs[14])
Final R1 / wR2 [I ≥ 2σ(I)]	0.018 / 0.037
Final R1 / wR2 (all data)	0.019 / 0.038
Goodness-of-fit on Fi2	1.40
Largest diff. peak and hole /e 10-6 pm-3	0.857 / -1.259

The new Cs₃Li₂Cl₅ structure is built up from [Li₂Cl₆]^4- edge-sharing double-tetrahedra (Figure 1, left) that are further interconnected through a common chloride position Cl4 (Figure 2). Distances in the tetrahedra range from 230.7(6) pm for Li2–Cl1 to 235.1(6) pm for Li1–Cl4, which is in good agreement with commonly reported values for multinary compounds containing Li–Cl bonds.^[15] In α-CsLiCl₂, a related structure in C2/c, these edge-sharing lithium chloride tetrahedra are also present, with Li–Cl bond lengths of 228(2) - 237(2) pm.^[8] In pure LiCl, reported distances are larger, ranging from 254.0^[16] to 266.0 pm,^[17] comparable to the Li–Cl bond lengths of 259.39 - 266.55 pm in the theoretical structure of Cs₃Li₂Cl₅.^[7] In the predicted Cs₃Li₂Cl₅, the lithium position is similarly coordinated by four chlorine atoms in a tetrahedral manner as depicted for comparison in Figure 1 on the right. The tetrahedra, however, are connected through one corner atom shared by three tetrahedra, unlike in the experimentally obtained structures, where the tetrahedra share corners as well as edges (cf. Figure 2). Relevant distances and angles of the experimentally observed lithium chloride tetrahedra in Cs₃Li₂Cl₅ are given in Table S3 of the SI.

Figure 1.

Edge- and corner-sharing [LiCl₄]^3- tetrahedra in Cs₃Li₂Cl₅ (left) and trimeric building blocks in the predicted structure (right).

Figure 2.

Cs₃Li₂Cl₅ viewed along the b-axis; Lithium chloride double-tetrahedra (grey polyhedra) form chains along the a-axis.

Through the corner-sharing [LiCl₄]^3- tetrahedra, the computed structure forms corrugated lithium chloride layers in the ac-plane. The synthesized equivalent, with its corner- and edge-sharing tetrahedra, forms 1D chains along the a-axis instead, as shown in Figure 2. The closest interatomic distance for Li1–Li2 in the experimentally obtained Cs₃Li₂Cl₅ is 296.8(9) pm, crossing the shared tetrahedra edge, and 707.54(8) pm between the symmetry equivalent Li–Li positions in separate lithium chloride chains. The minimum Li–Li distance of 404.90 pm, predicted for Cs₃Li₂Cl₅ in Cm, is much larger.^[8]

The lithium chloride chains in Cs₃Li₂Cl₅ are separated by three crystallographically independent rows of cesium cations, aligned also along [100], as depicted in Figure 3. Figure 4 shows the face- and edge-sharing coordination polyhedra of the cesium cations. Cs1 is nine-fold coordinated, Cs2 is coordinated in a distorted hexagonal bi-pyramid by 8 chlorine positions. Cs3 is surrounded by 7 chlorine atoms, roughly forming a pentagonal bi-pyramid. Interatomic distances in the cesium chloride polyhedra are listed in Table S3. The shortest Cs–Cl bond of 329.45(8) pm is much smaller than in the model structure (355.75 pm), but is well in the range of precedented bond-lengths.^[18] The smallest Cs–Cs distance of 433.74(4) pm is slightly larger than 413.60 pm for the theoretical structure. All chlorine anions are shared between lithium and cesium cations. All faces of the [Li₂Cl₆]^4- double tetrahedra are shared with the cesium chloride polyhedra.

Figure 3.

Cs₃Li₂Cl₅ viewed along the a-axis; Lithium chloride double-tetrahedra strings (grey polyhedra) and cesium cations are arranged alternatingly in layers within the ac-plane.

Figure 4.

Cesium coordination polyhedra in Cs₃Li₂Cl₅.

The bond-valence sums in Cs₃Li₂Cl₅ were calculated from the single-crystal data using the bond-length/bond-strength concept (Ʃ).^[19] The following values were obtained: Cs1 +1.003(1), Cs2 +0.962(1), Cs3 +1.009(1), Li1 +1.245(9), Li2 +1.26(1), Cl1 -1.181(6), Cl2 -1.028(5), Cl3 -1.108(5), Cl4 -1,140(1). The obtained sums are within the expected range for the formal ionic charges of the atoms.

Crystal structure of Cs₃LiCl₄ · 4H₂O

Figure 5 shows the experimental powder pattern of Cs₃LiCl₄ · 4H₂O compared to the theoretical pattern derived from single-crystal data. In the X-ray experiment, no side phases or impurities are visible. T-C-H Pseudo-Voigt functions described the experimentally found peak shapes best. A preferred orientation in [00l] was refined to account for the plate-like crystal shape. Details of the powder data collection and Rietveld refinement are given in Table 2.

Figure 5.

Experimental (red) and theoretical powder pattern of Cs₃LiCl₄ · 4H₂O (black curve, peak positions as green lines) calculated from the single-crystal data; the difference curve is given in blue.

Table 2.

Crystallographic and structure refinement data of Cs₃LiCl₄ · 4H₂O.

Empirical formula	Cs3LiCl4 · 4H2O
Molar mass /g mol-1	619.53
Crystal system	tetragonal
Space group	P4/nmm (no. 129c)
Formula units per cell	Z = 2
Temperature /K	293(2)
Powder data
Powder diffractometer	RIGAKU MiniFlex 600
Radiation	Cu-Kα1 (λ = 154.059(1) pm)
θ range /°	2 – 85
Step width /°	0.02
a /pm	865.91(5)
c /pm	1000.03(4)
V /10-6 pm3	748.14(1)
RBragg	13.4
Rf	13.3
Single crystal data
Single crystal diffractometer	RIGAKU XtaLABmini
	Mo-Kα (λ = 71.075(1) pm)
Radiation	(graphite monochromator)
a /pm	858.29(3)
c /pm	992.00(6)
V /10-6 pm3	730.77(7)
Calculated density /g cm-3	2.816
Crystal size /mm3	0.10 × 0.08 × 0.06
Absorption coefficient /mm-1	8.15
F(000) /e	552
θ range /°	3.1 – 33.9
	-12 ≤ h ≤ 12,
Index rage	-9 ≤ k ≤ 13,
	-12 ≤ l ≤15
Reflections total / independent	6065 / 884
Rint	0.0352
Reflections with I ≥ 2σ(I)	762
Data / ref. parameters	884 / 28
Absorption correction	analytical
Final R1 / wR2 [I ≥ 2σ(I)]	0.022 / 0.054
Final R1 / wR2 (all data)	0.028 / 0.057
Goodness-of-fit on Fi2	1.11
Largest diff. peak and hole /e 10-6 pm-3	1.48 / -0.58

Specifications on the single-crystal data collection and evaluation of Cs₃LiCl₄ · 4H₂O are listed in Table 2. In Table S4 of the SI, Wyckoff positions, atomic coordinates, and isotropic displacement parameters of Cs₃LiCl₄ · 4H₂O are given. Table S5 (SI) lists the respective anisotropic displacement parameters. Belonging to the ditetragonal dipyramidal point group 4/mmm or D_4h, Cs₃LiCl₄ · 4H₂O displays a centrosymmetric crystal structure. The anhydrous Cs₃LiCl₄ was predicted by Jansen et al.^[7] in Amm2, a rhombic pyramidal space group of C_2v symmetry pertaining to the Laue group mm2, which is polar. Already this quick comparison makes clear, how defining the water molecules are for the new Cs₃LiCl₄ · 4H₂O structure type.

“LiCl · 2CsCl · 4H₂O”, as reported by Khripun et al. in 2016,^[12] shows strikingly similar lattice parameters to the herein reported Cs₃LiCl₄ · 4H₂O: a = 857.16(3), c = 995.59(5) pm. “LiCl · 2CsCl · 4H₂O”, however, reportedly crystallizes with 16 formula units per unit cell in an equally tetragonal space group of lower symmetry: P4/n. The brief structure portrayal of “LiCl · 2CsCl · 4H₂O”, lacking essential crystallographic information like atomic parameters, suggests the group intended to describe the here reported Cs₃LiCl₄ · 4H₂O. Additionally, the same authors refer to “LiCl · 2CsCl · 4H₂O” as “LiCl · 3CsCl · 4H₂O” in their later study of its dissolution enthalpy.^[13] We have hence reason to assume that the previously described “LiCl · 2CsCl · 4H₂O” is in fact the same compound we herein give a full structure report on.

Cs₃LiCl₄ · 4H₂O is built up from two different layers in the ab-plane: an alternatingly arranged lithium hydrate and cesium chloride hydrate as one layer-type around z = 0, and cesium chloride as the second layer centered at z = 0.5. Figure 6 shows these layers in the structure along [110].

Figure 6.

Crystal structure of Cs₃LiCl₄ · 4H₂O viewed along the face diagonal [110]. Hydrogen bonds are given in dashed orange lines, Li–O coordination bonds are given in grey dashed lines.

Within the cesium chloride layer, the Cs2 and Cs3 atoms are each coordinated by Cl1 as a slightly distorted cube, as shown in Figure 7. Atomic distances in the cation coordination polyhedra are given in Table S6 of the SI. Even though Cs3 is surrounded by Cl1 at one single distance, the edges of the formed coordination polyhedron vary in length (452.59(7) and 405.70(7) pm). In pure CsCl,^[20] the Cs cations are coordinated in a cubic fashion by eight chlorine atoms (Cs–Cl 356.37 and Cl–Cl 411.50 pm).

Figure 7.

Face-sharing, slightly distorted cubic chloride coordination pertaining to the cesium cation positions Cs2 and Cs3 in Cs₃LiCl₄ · 4H₂O.

The Cs1 cation resides in the mixed layer of cesium chloride hydrate and lithium hydrate. It is coordinated square-antiprismatically by four chloride ions forming one of the square faces, and four oxygen atoms in the opposite face, as shown in Figure 8 on the left. All Cs–Cl interatomic distances (cf. Table S6) are shorter than those reported for the predicted, anhydrous Cs₃LiCl₄, where the shortest distance is found for Cs1–Cl3 with 363.96 pm.^[7] Pure CsCl shows a Cs–Cl bond length of 356.37 pm, very similar to the herein reported values.^[20] In β-CsLiCl₂, which is reported in the same space group as Cs₃LiCl₄ · 4H₂O, P4/nmm, the Cs–Cl distances are also comparable with 345.7(5) - 349.12(5) pm.^[6] In contrast to the hydrated Cs₃LiCl₄ · 4H₂O, in Cs₃LiCl₄, the two crystallographically independent cesium positions are coordinated seven- instead of eight-fold.^[7]

Figure 8.

Square-antiprismatic hetero-anionic coordination environment of Cs1 (left) and tetrahedrally coordinating water positions around Li1 (right) in Cs₃LiCl₄ · 4H₂O.

The lithium cations are coordinated tetrahedrally by four water positions (Figure 8), which are shared with the [CsCl₄(H₂O)₄]^3- polyhedra. This highly symmetric lithium coordination environment has been reported before e.g. for Li(H₂O)₄B(OH)₄ · 2H₂O^[21] and was also mentioned for “LiCl · 2CsCl · 4H₂O”.^[12] The Li–O distances and O–Li–O angles found for Cs₃LiCl₄ · 4H₂O are 4 × 192.4(2) pm with 4 × 105.20(7)° and 2 × 118.4(2)°, respectively. This tetrahedral coordination is slightly more dense and distorted than those reported for Li(H₂O)₄^[21] and Li₂(H₂O)₇^[22] building blocks in lithium hydroborate hydrates with ranges of 192.9 - 201.0 pm and 106.4-116.6°. In the anhydrous, theoretical Cs₃LiCl₄,^[7] the lithium position is surrounded by four chloride anions in a disphenoidal or highly distorted, tetrahedral polyhedron. The Li1 positions in Cs₃LiCl₄ · 4H₂O are aligned along the face diagonal [110] at a comparably far minimum distance of 606.90(2) pm from each other. This results from the [Li(H₂O)₄]⁺ tetrahedra not sharing any faces unlike the purely cesium chloride coordination polyhedra (cf. Figure 7). The cesium positions are situated closer to each other at a minimum 424.02(6) pm (Cs1–Cs2). In the computed phase, Li1 has a much shorter minimum Li–Li distance of 453.00 pm.^[7]

The layers in Cs₃LiCl₄ · 4H₂O are interconnected through hydrogen-bonds formed between water and chloride anions as shown in Figure 6. OH-bonds are 83(3) pm long, the donor-acceptor and proton-acceptor distances of 310.3(2) and 229(3) pm, respectively, are within the expected range for moderately strong hydrogen-bonds with a Cl^- acceptor.^[22–23] The hydrogen-bond angle O1–H1–Cl1 of 171(3)° is slightly wider than those reported for e.g. hydrogen bonding found in chloride water clusters (168.0 - 168.9 pm).^[23a]

The bond-valence sums in Cs₃LiCl₄ · 4H₂O were calculated from the single-crystal data using the bond-length/bond-strength concept (Ʃ) to support the structure determination.^[19] The following values were derived: Cs1 +0.892(2), Cs2 +0.931(1), Cs3 +0.9559(4), Li1 +1.160(3), O1 -2.3(2), H1 +1.0(4), Cl1 -0.750(1). Values with less significant digits result from higher errors in distances to hydrogen atoms. The obtained sums correspond well with the expected values for the formal ionic charges of the atoms.

Vibrational spectroscopy of Cs₃LiCl₄ · 4H₂O

An FTIR ATR spectrum of Cs₃LiCl₄ · 4H₂O in the range of 400 to 4000 cm^-1 is given in the SI Figure S1. Overlapping absorptions, assignable to Li–O vibrations^[24] and water modes,^[25] were found from 445 cm^-1 to 695 cm^-1. Alkaline metal chloride vibrations, typically found around 714 - 753 cm^-1,^[26] possibly overlapping with water torsion modes,^[25] were detected. Water absorbed on the surface of the specimen is most likely causing absorptions detected at 2158 cm^-1.^[27] A dominant, broad band around 3318 cm^-1 and its second harmonic at ~1630 cm^-1 can be assigned both to Cs–Cl^[27] and to water vibrations,^[28] supporting the water molecules postulated in the single-crystal structure solution.

The Raman spectrum of a single-crystal of Cs₃LiCl₄ · 4H₂O in the range of 100 - 4000 cm^-1 was collected (Figure S2 of the SI). Shifts from 105 up to a broader band at 442 cm^-1 can be attributed to a plethora of Cs–Cl vibrations.^[29] The most pronounced, broad band at 3339 cm^-1, with its much weaker second harmonic at 1670 cm^-1, stems from vibrations of crystal water.^{[28a, 28b]}

Thermal properties of Cs₃LiCl₄ · 4H₂O

To study the thermal decomposition of Cs₃LiCl₄ · 4H₂O thermogravimetric analyses (TGA), differential scanning calorimetry (DSC), and in situ temperature dependent powder X-ray diffraction (TD-PXRD) were performed. The TGA shows a constant weight loss due to adhesive and hydrate water loss under a constant nitrogen flow (cf. Figure S3 of the SI). Figure 9 shows the endothermic DSC signals embedded in a waterfall plot of the TD-PXRD data up to 670 K including multiple reactions. Figure S4 of the SI shows the entire DSC plot separately including the cooling curve. Peaking around 333 K, a sharp signal indicates a structural change of Cs₃LiCl₄ · 4H₂O. The broad signal around 375 K shows the evaporation of surface water. From approximately 425 to 458 K, loss of the remaining water molecules and the formation of various anhydrous Cs-Li chlorides is observed. In the corresponding cooling curve, none of these dehydration peaks are observed. After the complex dehydration and decomposition processes, a transition from monoclinic α-CsLiCl₂ to the tetragonal, high-temperature β-CsLiCl₂ at 544 K, previously observed around 583 K,^{[6, 8]} can be assumed. With its maximum at approximately 588 K, a DSC signal indicates the peritectoid decomposition of CsLi₂Cl₃ to β-CsLiCl₂ and LiCl, as reported by Jansen et al. at 593 K.^[8] Near 606 K, the eutectic mixture of β-CsLiCl₂ and LiCl melts, as reported by E. Korreng around 605 K.^[5a] The signal at ~ 630 K, with a shoulder near 632 K, correlates well with a peritectic reaction: remaining β-CsLiCl₂ transforming to “α-Cs₂LiCl₃” and a melt, and the enantiotropic phase transition from “α-Cs₂LiCl₃” to “β-Cs₂LiCl₃” (cf. E. Korreng at 624 and 633 K, respectively).^[5a] Around 647 K, we see another peritectic reaction of “β-Cs₂LiCl₃” decomposing to CsCl and a melt, as reported around 653 K before.^{[5a, 8]}

Figure 9.

DSC heating curve with endothermic signals in blue to the right, and TD-PXRD waterfall diagram with square root intensity representation of Cs₃LiCl₄ · 4H₂O up to 670 K.

The collected temperature dependent PXRD data mirrors the decompositions and phase transitions of the starting material Cs₃LiCl₄ · 4H₂O detected by DSC. A waterfall plot summarizing the results is given in Figure 9. Upon heating up to 670 K, we found several changes in the diffraction patterns with strongly varying crystallinity of the specimen. Reflections indicative of the new Cs₃Li₂Cl₅ appear in diffractograms between 430 and 600 K. Selected diffractograms collected at 430, 550, and 600 K showing reflections from the decomposition products are given in the SI, Figures S5, S6, and S7.

From the observed reactions, we conclude that Cs₃LiCl₄ · 4H₂O decomposes into CsCl, LiCl, α-CsLiCl₂, β-CsLiCl₂, CsLi₂Cl₃, and “α-Cs₂LiCl₃”. The TD-PXRD data further suggest additional, still unknown decomposition products.

Theoretical results

Total energy (E_TOT) calculations were performed with the Vienna ab initio simulation package (VASP).^[30] Thereby, we relate the overall stability of the newly synthesized Cs₃Li₂Cl₅ to comparable, predicted cesium-rich ternary chlorides by Jansen et al., as well as to α-CsLiCl₂ and the binary salts CsCl and LiCl.^{[7, 31]} The results, listed in Table S7 of the SI, confirm that the experimentally observed structure of Cs₃Li₂Cl₅ in P2₁/m is more stable than the predicted equivalent in Cm by 0.54 eV/formula unit (f.u.) (52.0 kJ/mol).

Subsequently, the formation probabilities of the here reported Cs₃Li₂Cl₅ and three cesium-rich structures predicted in the CsCl/LiCl system were analyzed by comparing their enthalpies of formation Δ_fH (Table 3).^{[7, 31]} Sums for Δ_fH at 0 K were derived from calculated E_TOT values with respect to the binary compounds CsCl and LiCl (cf. Table S7 of the SI for full list). As reported by Jansen et al.,^[7] our results show that the predicted Cs₃Li₂Cl₅ and Cs₃LiCl₄ tend to decompose to CsCl and LiCl, while the theoretical structure of Cs₂LiCl₃ should be stable with respect to decomposition into the binaries. Remarkably, the synthesized Cs₃Li₂Cl₅ shows a particularly negative enthalpy of formation of ∆_fH = -35.1 kJ/mol, rendering it much more stable i.e. likely to form than its predicted structure with ∆_fH = +16.9 kJ/mol.

Table 3.

Calculated enthalpies of formation Δ_fH in eV/f.u. (values in kJ/mol in parentheses) at 0 K for experimentally observed and predicted cesium-rich CsCl/LiCl ternary compounds.

3 CsCl	+	2 LiCl	→	Cs3Li2Cl5 (this work)	ΔfH
-19.593		-14.774		-34.731	-0.364 (-35.1)
3 CsCl	+	2 LiCl	→	Cs3Li2Cl5 (predicted)
-19.593		-14.774		-34.192	+0.175 (+16.9)
3 CsCl	+	LiCl	→	Cs3LiCl4 (predicted)
-19.593		-7.387		-26.646	+0.334 (+33.2)
2 CsCl	+	LiCl	→	Cs2LiCl3 (predicted)
-13.062		-7.387		-20.482	-0.033 (-3.18)

To further evaluate the stabilities of the structures, the decomposition energies Δ_rH of the experimentally found Cs₃Li₂Cl₅ and the computed Cs₃Li₂Cl₅, Cs₃LiCl₄, and Cs₂LiCl₃ were calculated at 0 K as listed in Table 4. Again, the calculated E_TOT values were used for this assessment (cf. Table S7). The results show that both Cs₃LiCl₄ and Cs₂LiCl₃ would decompose to Cs₃Li₂Cl₅ in P2₁/m and CsCl. Importantly, although the predicted Cs₂LiCl₃ is stable with respect to its decomposition into CsCl and LiCl, the decomposition into CsCl and the newly synthesized Cs₃Li₂Cl₅ is very favorable with ∆_rH = -14.5 kJ/mol. Therefore, the predicted Cs₂LiCl₃ is unlikely to form. Opposingly, the synthesized Cs₃Li₂Cl₅ has a positive decomposition enthalpy of 3.86 kJ/mol, meaning it would not decompose into the two neighboring compounds, CsCl and α-CsLiCl₂, in the phase diagram.

Table 4.

Calculated decomposition enthalpies Δ_rH in eV/f.u. (values in kJ/mol in parentheses) at 0 K of the experimentally observed Cs₃Li₂Cl₅, as well as the predicted Cs₃Li₂Cl₅, Cs₃LiCl₄ and Cs₂LiCl₃.

Cs3Li2Cl5 (this work)	→	2 α-CsLiCl2	+	CsCl	ΔrH
-34.731		-28.160		-6.531	+0.040 (+3.86)
Cs3Li2Cl5 (predicted)	→	2 α-CsLiCl2	+	CsCl
-34.192		-28.160		-6.531	-0.499 (-48.2)
Cs3LiCl4 (predicted)	→	1/2 Cs3Li2Cl5 (this work)	+	3/2 CsCl
-26.646		-17.366		-9.796	-0.516 (-49.8)
Cs2LiCl3 (predicted)	→	1/2 Cs3Li2Cl5 (this work)	+	1/2 CsCl
-20.482		-17.366		-3.266	-0.150 (-14.5)

In addition, the electronic structures of the new compounds Cs₃Li₂Cl₅ and Cs₃LiCl₄ · 4H₂O were analyzed through density of states (DOS) and band structure calculations. The DOS of Cs₃Li₂Cl₅, shown in Figure 10 on the left, reflects its ionic character with localized states of all elements. All Cs 5p electrons (36 e^-/unit cell (u.c.), 6 e^-/atom) occupy states between -7 and -6 eV, while Cs 6s states are unoccupied with high energies (from 5.0 to 7.5 eV) indicative of Cs⁺. States below the Fermi level, down to -2.5 eV, mostly belong to fully occupied Cl 3p orbitals with 60 e^-/u.c. (6 e^-/atom), resulting in Cl⁻. The band structure of Cs₃Li₂Cl₅, given in Figure S8 of the SI, indicates its insulator behavior with a direct band gap of ~4.9 eV.

Figure 10.

Density of states (DOS) for one unit cell (u.c.) of Cs₃Li₂Cl₅ (left) and Cs₃LiCl₄ · 4H₂O (right). The Fermi levels are indicated as a dashed line.

The DOS of Cs₃LiCl₄ · 4H₂O in Figure 10 on the right opens a large gap at the Fermi level, again indicating insulating behavior. The corresponding band structure (cf. Figure S9 of the SI) shows a direct band gap of ~5.0 eV. The states of all elements are localized, demonstrating the ionic feature of the compound. In the energy windows of -9 to -8 eV, -5.2 to -4 eV and -3 to -2 eV, the states mainly pertain to H₂O. The states of water are distanced from the Fermi level, thus having barely any impact on the direct band gap. States around -6 eV show fully occupied Cs 5p orbitals with 36 e^-/u.c. (6 e^-/atom). Cs 6s states, situated between 6 and 8.5 eV, are unoccupied, indicating Cs⁺. The states between -1 eV and the Fermi level pertain to Cl 3p, occupied with 48 e^-/u.c. (6 e^-/atom). Therefore, all 3p orbitals of the chlorine atoms are occupied, resulting in Cl^-.

Concluding, both Cs₃Li₂Cl₅ and Cs₃LiCl₄ · 4H₂O are ionic insulators.

Discussion

Revised CsCl/LiCl phase diagram

Based on the observation of the new phase Cs₃Li₂Cl₅, thermal decomposition studies of the tetrahydrate Cs₃LiCl₄ · 4H₂O, as well as total energy calculations, we propose a revised phase diagram for the CsCl/LiCl system (Figure 11), adapting the most recently published diagram by Jansen et al.^[8] For comparison, the SI includes the previously published phase diagrams we combined (Figures S10 and S11).^{[5b, 8]} We propose that the double-salt “LiCl · 2CsCl” (or “α-Cs₂LiCl₃”), as reported in 1915, corresponds to the herein reported Cs₃Li₂Cl₅.^[5a] The early reported phase “LiCl · 2CsCl” contains 66.7 wt.-% CsCl, while the herein reported Cs₃Li₂Cl₅ contains 60 wt.-% CsCl – a then minor difference in composition. Likewise, no crystal structure was ever experimentally found for “LiCl · 2CsCl” or “Cs₂LiCl₃”. Unlike for both modifications of “Cs₂LiCl₃”, all other experimentally found phases in the CsCl/LiCl system have been characterized by single-crystal diffraction, which confirmed their exact compositions. Furthermore, according to our calculations, the predicted crystal structure of Cs₂LiCl₃^[7] is energetically unstable towards decomposition into the experimentally observed Cs₃Li₂Cl₅ and CsCl. The observed liquidus temperature of CsCl/LiCl at x = 0.4 is approx. 860 K, while the solidus temperature is at about 650 K. Consequently, the melting temperature of Cs₃Li₂Cl₅ must be within that temperature range. However, no additional, unassigned signal was observed in our DSC studies. We hence conclude that one of the already assigned signals corresponds to the melting of Cs₃Li₂Cl₅. Consequently, we propose that previously reports on “α-Cs₂LiCl₃” and Cs₃Li₂Cl₅ reported here correspond in fact to the same substance with the sum formula Cs₃Li₂Cl₅.

Figure 11.

Revised phase diagram for the CsCl/LiCl system.

One could argue that the observed DSC signals at 624 K and at 633 K (cf. Figure 9) occur due to the formation of two different substances with different compositions (“Cs₂LiCl₃” and Cs₃Li₂Cl₅ or vice versa). E. Korreng, however, reported a change in the birefringence of the studied “Cs₂LiCl₃” crystals, whereby the specimen did not decompose whatsoever.^[5a] This observation could not be made for a substances undergoing a change in chemical composition. Instead, we assume that Cs₃Li₂Cl₅ is equally dimorphic-enantiotropic and was in fact studied by E. Korreng. Unfortunately, isolating crystals of the high-temperature polymorph remains a major challenge due to its small temperature region. To complete the revised phase diagram, we added all known phase transitions in the pseudo-binary CsCl/LiCl system.

We feel obligated to mention that the single-crystal structure of Cs₃Li₂Cl₅ was determined at 100 K under a nitrogen flow to protect the extremely hygroscopic crystals from moisture. Consequently, we acknowledge the slim chance for the reported Cs₃Li₂Cl₅ to decompose between 100 K and room temperature into another cesium-rich phase and α-CsLiCl₂.^[7]

Experimental section

Synthesis

Both Cs₃Li₂Cl₅ and Cs₃LiCl₄ · 4H₂O were synthesized from the binary compounds cesium chloride (99.99%, Acros Organics, New Jersey, USA) and lithium chloride (99+%, Acros Organics, New Jersey, USA) at a molar ratio of 3 : 1 without any solvent. To obtain Cs₃Li₂Cl₅ single-crystals, the educt mixture was ground together and pelletized in a glovebox (N₂ atmosphere, H₂O and O₂ <0.5 ppm). After sealing the pellet in a dried quartz tube, flushed with argon repeatedly and evacuated to 50 mbar, the sample was heated with 5 K/min to 973 K and kept at that temperature for 4 days. It was then cooled with five degrees per hour to 473 K. Transparent, colorless, Cs₃Li₂Cl₅ single-crystals could be isolated under inert conditions from the product mixture also containing CsCl^[20] and α-CsLiCl₂.^[8] Cs₃Li₂Cl₅ is extremely sensitive to moisture and instantly decomposes when exposed to air. The product mixture containing Cs₃Li₂Cl₅, CsCl, and α-CsLiCl₂, when ground up in air, reacts to form the stable, hydrated Cs₃LiCl₄ · 4H₂O.

For a direct synthesis of Cs₃LiCl₄ · 4H₂O, CsCl and LiCl 3 : 1 were thoroughly ground together in air at room temperature in an agate mortar. Cs₃LiCl₄ · 4H₂O was observed, with CsCl as an impurity, after letting the ground educts sit at room temperature for 5 min. In order to obtain phase pure and high-quality single-crystals of Cs₃LiCl₄ · 4H₂O, the same educt mixture was heated to 973 K and cooled to 473 K under inert conditions, as explained in detail above. After opening the ampoule in air, transparent, regularly shaped, colorless platelets of Cs₃LiCl₄ · 4H₂O were attained, coating the top end of the quartz tube. The compound shows strong hygroscopicity, but is, kept in a desiccator, stable in air for months. Cs₃LiCl₄ · 4H₂O is practically insoluble in water and readily dissolved by hydrochloric acid.

Crystal structure analyses

For the single-crystal structure determination of Cs₃Li₂Cl₅, a regularly shaped block (284 × 143 × 123 µm) was selected under a polarized-light microscope. The X-ray experiments were conducted on a Bruker APEX II CCD with graphite monochromatized Mo-K_α (λ = 71.075(1) pm) radiation, whereby the crystals were kept under constant N₂ flow at 100 K to prevent them from decomposing. All crystals were subject to heavy twinning; The obtained data was treated as two twin components. The structure solution and refinement (full matrix least squares on F²) were performed with the Shelx-2008 software suite.^[32] The centrosymmetric space group P2₁/m (no. 11) was found to be correct during the refinement. This was confirmed through the ADDSYM and STRUCTURE TIDY routines of the program PLATON^[33] implemented in WinGX 2014.1.^[34] A multi-scan absorption correction for twins was applied to the intensity data.^[14] Cesium and chlorine atoms were refined with anisotropic displacement parameters, lithium positions were refined isotropically.

The powder X-ray diffraction pattern of Cs₃LiCl₄ · 4H₂O was collected with a RIGAKU MiniFlex 600 benchtop diffractometer used in reflection geometry, equipped with graphite monochromatized Cu-K_α1 (λ = 154.059(1) pm) radiation and a scintillation counter (SC-70) detector. The data were analyzed using the FullProf suite 2017^[35] and plotted with WinPLOTR.^[36]

For the single-crystal structure determination of Cs₃LiCl₄ · 4H₂O, a regularly shaped platelet (103 × 80 × 56 µm) was selected under a polarized-light microscope. The single-crystal intensity data were collected on a Rigaku XtaLABmini benchtop diffractometer with graphite monochromatized Mo-K_α (λ = 71.075(1) pm) radiation with SHINE optics. The structure solution and refinement were performed with the same software as mentioned above. The centrosymmetric space group P4/nmm (no. 129c) with the origin at 2/m and 4mm symmetry element located at ¼, ¼, z was found to be correct during the refinement. ADDSYM and STRUCTURE TIDY routines did not show any obvious structural changes needed.^[33] An analytical numeric absorption correction was applied to the intensity data.^[37] All non-hydrogen atoms were refined with anisotropic displacement parameters, the hydrogen positions were refined isotropically without any restrictions.

All relevant details of the crystal data collection and evaluation of Cs₃Li₂Cl₅ are specified in Table 1. The according Wyckoff positions, atomic coordinates, and isotropic displacement parameters of Cs₃Li₂Cl₅ are listed in Table S1 of the SI. In Table S2 (SI), the equivalent anisotropic displacement parameters of Cs₃Li₂Cl₅ are given. The crystal data collection and evaluation details of Cs₃LiCl₄ · 4H₂O are listed in Table 2. Table S4 of the SI gives according Wyckoff positions, atomic coordinates, and isotropic displacement parameters of Cs₃LiCl₄ · 4H₂O. In Table S5 of the SI, the equivalent anisotropic displacement parameters are given. Additional information on the crystal structure investigations is to be obtained from the Fachinformationszentrum Karlsruhe, D-76344 Eggenstein-Leopoldshafen, Germany (fax: +49-7247-808-666; e-mail: crysdata@fiz-karlsruhe.de, https://icsd.fiz-karlsruhe.de/search/dbinfo.xhtml) on quoting the ICSD collection codes CSD-1895758 and CSD-1895757 for Cs₃Li₂Cl₅ and Cs₃LiCl₄ · 4H₂O, respectively.

Vibrational spectroscopy

An infrared characterization of Cs₃LiCl₄ · 4H₂O was performed on the bulk sample. A Bruker Alpha ATR FTIR spectrometer (spectral resolution higher than 2 cm^-1) equipped with a high sensitivity L-alanine doped triglycine sulfate (DLaTGS) detector, a KBr beam splitter and a RockSolid™ interferometer was employed. The sample was placed directly on the ZnSe analyzer crystal, 4 scans were acquired for the sample as well as the background. The primary data was recorded using the spectrometer software Opus^[38]; an atmospheric correction was applied.

A Raman spectrum of a single-crystal of Cs₃LiCl₄ · 4H₂O, sealed in a capillary to prevent accumulation of surface water, was recorded with a Horiba LabRam HR confocal microscope from 100 - 4000 cm^-1. The sample was excited using the green emission line (532 nm) of a frequency-doubled 100 mW Nd-YAG-laser. The scattered light was dispersed by an optical grating with 1800 lines mm^-1 and collected at a 1024 × 256 Andor CCD Peltier cooled detector. The spectral resolution, determined by measuring the Rayleigh line, was in the order of 2 cm^-1. The accuracy of the Raman shifts was about 0.5 cm^-1. A two-point calibration, measuring the grating zero-order line and the 520.7 cm^-1 silicon line of a silicon standard, was used to determine this accuracy. The spectra were recorded unpolarized at ambient conditions, while the crystals were sealed in a capillary to protect them from moisture.

Differential scanning calorimetry and thermogravimetric analysis

The thermal stability of Cs₃LiCl₄ · 4H₂O was studied by differential scanning calorimetry with a DSC 214 Polyma (Netzsch GmbH, Selb, Germany) and the provided Proteus 7.0 software package. The specimen (about 4.5 mg) was placed in a concave aluminum pan with a pierced lid, which was cold-sealed. Under a constant nitrogen flow of 40 ml/min, the sample was heated to 673 K with 10 K/min, then cooled to RT at the same rate. The data was baseline corrected.

To investigate the water loss in Cs₃LiCl₄ · 4H₂O, thermogravimetric analyses were performed on a Netzsch TG 209 F1 Libra. The specimen (10.5366 mg) was placed in an open alumina crucible, data was collected under constant nitrogen flow of 50 ml/min. The sample was heated to 298 K with 10 K/min, then cooled to RT again at the same rate. A baseline correction measurement mode was chosen.

In situ high-temperature powder X-ray diffraction

Cs₃LiCl₄ · 4H₂O was studied in an in situ heating powder XRD experiment in the range between 300 and 670 K. A PANalytical X’Pert Pro diffractometer equipped with an Anton Paar HTK-1200N furnace was used in Bragg-Brentano geometry in θ/2θ mode with Cu-K_α1,2 radiation, and an X’Celerator multi-strip detector. Diffraction patterns were collected every 10 K with a scan step width of 0.0167 °2θ and 90.17 s of total counting time per step. The used ramp rate between set points was 10 K/min with 10 min dwell time at the set points before starting the measurement. The total data collection per scan was approx. 0.5 h.

Theoretical Methods

Electronic band structure and total energy calculations were performed using the projector augmented wave method of Blöchl^[39] coded in the Vienna ab initio simulation package (VASP).^[30] All VASP calculations employed the generalized gradient approximation (GGA) with exchange and correlation treated by the Perdew-Burke-Enzerhoff functional.^[40] The 5s, 5p, and 6s states for Cs, the 1s and 2s states for Li, the 1s state for H, the 2s and 2p states for O, as well as 3s and 3p states for Cl were treated as valence states. The cutoff energy for the plane wave calculations was set to 500 eV and the Brillouin zone integrations were carried out using a 7 × 7 × 6, 8 × 8 × 4, 8 × 4 × 6, 13 × 13 × 13, 5 × 11 × 4, 11 × 2 × 6, and 7 × 2 × 9 k-point mesh for Cs₃LiCl₄ · 4H₂O, Cs₃Li₂Cl₅, α-CsLiCl₂, CsCl, and the predicted Cs₃LiCl₄, Cs₂LiCl₃,and Cs₃Li₂Cl₅ by Jansen et al., respectively.^[7]

Conclusions

Pertaining to the long-studied, simple CsCl/LiCl system, Cs₃Li₂Cl₅ was synthesized from a mixture of its binary salts without any solvent. Close comparison to the compound’s computed model showed a completely different, unprecedented crystal structure. Through our addition to this still poorly explored system, we help to improve future parameter fits for empirical potentials aiming to study this very promising field of alkali-metal halides. Emblematically, our energy calculations revealed that past predictions in the system can now be revised. Our work also looks at a deeper understanding of an otherwise less explored family of compounds: A full structure report was given on the second hitherto reported ternary alkali-metal chloride containing water. Surprisingly, Cs₃LiCl₄ · 4H₂O could also be obtained through the most inexpensive, fast and green chemical synthesis possible: simply grinding a stoichiometric mixture of CsCl and LiCl at room temperature in air. This discovery makes the search for related compounds for applications that require large-scale synthesis immensely appealing. Thermal analyses of Cs₃LiCl₄ · 4H₂O, which is stable in air, resulted in an amendment of the CsCl/LiCl phase diagram. This goes to show that the simplest systems, with the most straight forward synthesis techniques, still hold interesting compounds that are yet to be discovered. Further studies on the anhydrous CsCl/LiCl system and related ternary halides containing cesium are underway.