Hydrogen-Storage Salts Accessed via Metathesis ... - ACS Publications

Article pubs.acs.org/IC

Complete Series of Alkali-Metal M(BH3NH2BH2NH2BH3) HydrogenStorage Salts Accessed via Metathesis in Organic Solvents Rafał Owarzany,§ Karol J. Fijalkowski,*,‡ Tomasz Jaroń,‡ Piotr J. Leszczyński,‡ Łukasz Dobrzycki,§ Michał K. Cyrański,§ and Wojciech Grochala‡ ‡

Centre of New Technologies, University of Warsaw, ul. Zwirki i Wigury 93, 02-089 Warsaw, Poland Faculty of Chemistry, University of Warsaw, ul. Pasteura 1, 02-093 Warsaw, Poland

§

S Supporting Information *

ABSTRACT: We report a new efficient way of synthesizing high-purity hydrogenrich M(BH3NH2BH2NH2BH3) salts (M = Li, Na, K, Rb, Cs). The solventmediated metathetic synthesis applied here uses precursors containing bulky organic cations and weakly coordinating anions. The applicability of this method permits the entire series of alkali-metal M(BH3NH2BH2NH2BH3) salts (M = Li, Na, K, Rb, Cs) to be obtained, thus enabling their comparative analysis in terms of crystal structures and hydrogen-storage properties. A novel polymorphic form of Verkade’s base (C18H39N4PH)(BH3NH2BH2NH2BH3) precursor was also characterized structurally. For all compounds, we present a comprehensive structural, spectroscopic, and thermogravimetric characterization (PXRD, NMR, FTIR, Raman, and TGA/DSC/MS).

■

INTRODUCTION Protonic−hydridic compounds (XHxZHy) are among the most promising solid-state hydrogen-storage materials.1 The low temperature of hydrogen desorption results from the presence of both protonic (H+) and hydridic (H−) H atoms. Lightweight protonic−hydridic materials are based on N and B atoms as H+ and H− binding sites, respectively.2,3 Examples encompass NH4BH4,4 ammonia borane (NH3BH3),5,6 and metal amidoboranes (MNH2BH3).7−18 Most of these materials suffer from unfavorable thermodynamics (i.e., their rehydrogenation is difficult), an insufficient amount of hydrogen desorbed below 90 °C, and the presence of impurities in H2 gas.19,20 Recently, we reported two M(BH3NH2BH2NH2BH3) salts [abbreviated as M(B3N2)],21 where M = Li and Na, as novel highly efficient boron−nitrogen-based hydrogen-storage materials.21 M(BH3NH2BH2NH2BH3) salts sometimes appear as a contamination of amidoborane samples.19 We found X-ray reflections from Li(BH 3 NH 2 BH 2 NH 2 BH 3 ) and Na(BH3NH2BH2NH2BH3) in powder patterns contained in recently published papers.7,14,22,23 M(BH3NH2BH2NH2BH3) salts are of interest not only because of their high nominal hydrogen content but also because they constitute intermediates of thermal decomposition of metal amidoborane salts, thus testifying to various degrees of polymerization of the (BHxNHy) subunits. M(BH3NH2BH2NH2BH3) salts (M = Li, Na) were successfully synthesized using a wet-chemistry approach [in a tetrahydrofuran (THF) solvent]. We then attempted to extend these methods for the synthesis of heavier K, Rb, and Cs analogues, but the obtained products were of unsatisfactory purity. Therefore, we devised a novel method of synthesis of M(BH3NH2BH2NH2BH3) salts and applied it to the entire series of alkali-metal compounds. Synthesis proceeds © XXXX American Chemical Society

via a metathetic solvent-mediated reaction and utilizes precursors that are soluble in weakly coordinating organic solvents. As we will show, the method delivers desired products in a quite pure form and thus free from “dead mass” (as is typical of mechanochemical synthesis). The method is similar to that developed recently for complex metal borohydrides.24−26 As a source of BH3NH2BH2NH2BH3− anions, we used here Verkade’s base (abbreviated as VB) 27 ,2 8 derivative (C18H39N4PH)(BH3NH2BH2NH2BH3) [abbreviated as VBH(B3N2)] (Figure 1). As the second reactant, we used salts containing bulky weakly coordinating fluorinated organic anions and respective metal cations: M[Al{OC(CF3)3}4]29 (Figure 1). The reaction is performed in an anhydrous CH2Cl2 solvent. The new synthetic protocol is quite universal, and it carries the potential to be applied for the synthesis of compounds comprising cations other than the alkali-metal ones.

■

RESULTS AND DISCUSSION New Method of Metathetic Synthesis. As mentioned above, we recently prepared Li(BH3NH2BH2NH2BH3) and Na(BH3NH2BH2NH2BH3) salts via the reaction of alkali-metal hydride with 3 equiv of ammonia borane in THF.21 In this process, hydridic H atoms detaches protonic H atoms from ammonia borane molecules to form BH3NH2− groups (eq 1), which further react with excess ammonia borane to form longchain BH3NH2BH2NH2BH3− anions (eq 2). Received: July 26, 2015

A

DOI: 10.1021/acs.inorgchem.5b01688 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

(CF3)3}4] salts in molar ratio 1:1 using anhydrous CH2Cl2 as the solvent (eq 4). (C18H39N4PH)(BH3NH 2BH 2NH 2BH3) + M[Al{OC(CF3)3 }4 ] → M(BH3NH 2BH 2NH 2BH3) ↓ + (C18H39N4PH) [Al{OC(CF3)3 }4 ]

M(BH3NH2BH2NH2BH3) salts are white low-density solids that are sensitive to moisture; all are stable at room temperature. Among them, only Li(BH3NH2BH2NH2BH3) is well soluble in THF; the solubility for other alkali-metal salts is poor (scarce for Cs salt) but sufficient for performing NMR measurements. Spectroscopic Characterization. The 11B NMR spectra of all alkali-metal M(BH 3 NH 2 BH 2 NH 2 BH 3 ) salts and (C18H39N4PH)(BH3NH2BH2NH2BH3) are rather similar, as expected (Figure 2). All of them consist of a characteristic

Figure 1. (Top) General scheme of the new metathetic method of the synthesis of M(BH3NH2BH2NH2BH3) hydrogen stores. (Bottom) Chemical formulas of protonated VB (C18H39N4PH) as well as perfluorinated tetraalkoxyaluminate.

H− + NH3BH3 → BH3NH 2− + H 2

(4)

(1)

BH3NH 2− + 2NH3BH3 → BH3NH 2BH 2NH 2BH3− + H 2 + NH3

(2)

Using this method, we also tried to synthesize heavier alkalimetal M(BH3NH2BH2NH2BH3) salts (M = K, Rb, Cs), but the products obtained showed poor quality (cf. the Supporting Information, SI). Thus, we were in need of a new synthetic route. Here, we found (C18H39N4PH)(BH3NH2BH2NH2BH3) to be a universal soluble precursor for the metathetic synthesis of diverse M(BH3 NH2 BH 2 NH 2BH 3 ) salts. (C 18 H39 N 4 PH)(BH3NH2BH2NH2BH3) is a white solid that is well soluble in various formally aprotic organic solvents, i.e., THF, CH2Cl2, and toluene. It crystallizes triclinic with two polymorphs in the P1̅ space group; we will denote the one published recently as α28 and the one prepared here as β (see the section below). Both forms are equally suitable for the synthesis of M(BH3NH2BH2NH2BH3) salts in solution. We obtained (C18H39N4PH)(BH3NH2BH2NH2BH3) in a direct reaction of Verkade’s base with 3 equiv of ammonia borane (eq 3).30 Verkade’s base is a sufficiently strong base27,28 to deprotonate ammonia borane and form BH3NH2BH2NH2BH3− anions.

Figure 2. Comparison of 11B NMR spectra in THF-d 8 of (C18H39N4PH)(BH3NH2BH2NH2BH3) and all alkali-metal M(BH3NH2BH2NH2BH3) salts: M = Li−Cs. The chemical shift regions corresponding to the BH2 and BH3 groups are marked in gray. #, + , *, and ○ mark various impurities.

quartet at ca. −21 ppm representing [BH3] groups and a twice weaker triplet at ca. −8 ppm representing [BH2] groups. Such spectra confirm the presence of BH3NH2BH2NH2BH3− anions in the compounds studied. There are additional weak signals in the spectra [especially for Li(BH3NH2BH2NH2BH3)] attributed to unidentified impurities, which are largely amorphous and soluble in THF. All alkali-metal M(BH3NH2BH2NH2BH3) salts are characterized by similar and characteristic Fourier transform infrared (FTIR; Figure 3) and Raman spectra (see the SI). The most pronounced features in the IR spectra are sharp doublets coming from NH stretching vibrations (at 3256−3296 and 3302−3313 cm−1), similar narrow and intense doublets originating from BH stretching (at 2248−2310 and 2332− 2364 cm−1), and bands coming from the HNH bending mode (at 1556−1571 cm−1). The HNH band is diagnostic because it

C18H39N4P + 3NH3BH3 → (C18H39N4PH)(BH3NH 2BH 2NH 2BH3) + H 2 + NH3

(3)

M[Al{OC(CF3)3}4] salts comprising an alkali-metal cation and bulky weakly coordinating anions were used as the second precursor for the synthesis of M(BH3NH2BH2NH2BH3) salts. These salts are white puffy crystalline solids, with sufficient solubility in CH2Cl2 to serve as the sources of M+ cations for metathetic reactions.24−26 We performed the synthesis of alkali-metal M(BH3NH2BH2NH2BH3) salts at room temperature by reacting (C 18 H 39 N 4 PH)(BH 3 NH 2 BH 2 NH 2 BH 3 ) with M[Al{OCB


Article

Inorganic Chemistry

= Li) synthesized using metathetic protocol clearly demonstrated superior quality (cf. the SI). It may be expected that other M(BH3NH2BH2NH2BH3) salts, e.g., M(BH3NH2BH2NH2BH3)2, where M = Mg−Ba, could also be obtained using the novel method. Crystal Structures of M(BH3NH2BH2NH2BH3) Salts. All crystalline samples of alkali-metal M(BH3NH2BH2NH2BH3) salts were studied with powder X-ray diffraction (PXD). Previously, we described the crystal structures of Li(BH3NH2BH2NH2BH3) and Na(BH3NH2BH2NH2BH3), both containing substantially disordered n-pentane-like 2 9 BH3NH2BH2NH2BH3− anions.21 Now, having prepared highpurity samples of M = K, Rb, and Cs derivatives, we can determine their crystal structures (Figure 4) as well as draw a comparison for the entire alkali-metal series (Figure 5).

Figure 3. Comparison of FTIR spectra of (C 18 H 39 N 4 PH)(BH3NH2BH2NH2BH3) and all alkali-metal M(BH3NH2BH2NH2BH3) salts: M = Li−Cs. The NH and BH stretching regions are marked in gray. Compare the SI for expanded views of particular regions.

differs in wavenumber and intensity from that measured for respective amidoboranes. Likewise, Raman spectra of M(BH3NH2BH2NH2BH3) salts are different from those for the corresponding amidoborane salts (see the SI), and they constitute their unmistakable fingerprint. It is worth mentioning that for Rb(BH3NH2BH2NH2BH3) and Cs(BH3NH2BH2NH2BH3) each of the bands constituting a strong doublet in the NH stretching region is further split in two (by ca. ±14 cm−1 for Cs salt and ±4 cm−1 for Rb salt). This could come from group-factor (Davidov) splitting, but it is even more likely that this feature originates, in part, from a substantial deformation of BH3NH2BH2NH2BH3− anions, which contain four nonequivalent N−H bonds at the lowsymmetry site. We performed density functional theory (DFT) calculations to determine the frequencies of normal vibrations of selected M(BH3NH2BH2NH2BH3) salts. We focused on M = K and Rb a n a l o g u e s b e c a u s e K s a l t c o n t a i n s q u a s i - li n e a r BH3NH2BH2NH2BH3− anions, resembling the molecules of n-pentane in the solid state,31 while Rb salt exhibits the presence of bent anions with two inequivalent NH2 and two inequivalent BH3 groups (see the next section). DFT calculations reasonably reproduced the wavenumbers of all major NH stretching and BH stretching modes, albeit they yielded somewhat overestimated frequency values (cf. the SI). What is important is that the calculations revealed substantial splitting of the bands corresponding to NH stretching for the rubidium compound (up to ±5 cm−1) but much smaller ones for potassium (up to ±1 cm−1, i.e., below experimental resolution and originating from group-factor splitting), thus confirming our initial surmise. The FTIR and 11B NMR spectra may be used to assess the purity of M(BH3NH2BH2NH2BH3) salts obtained via two alternative synthetic methods: the published one using MH21 and the new method using the Verkade’s base derivative as described here. M(BH3NH2BH2NH2BH3) salts (except for M

Figure 4. Details of the crystal structures of M(BH3NH2BH2NH2BH3) salts: M = Li−Cs. Geometry of the BH3NH2BH2NH2BH3− anions (left) and the coordination polyhedra of the alkali-metal cations (right).

K(BH3NH2BH2NH2BH3) crystallizes in the base-centered orthorhombic cell. The BH3NH2BH2NH2BH3− anions have geometry close to that of crystalline n-pentane. The N−B distances are 1.609(14) Å (to the terminal B atom) and 1.551(14) Å (to the central B atom). K+ cations are coordinated by six hydride anions from the BH3 groups, at 2.395(16) Å (×4) and 2.54(2) Å (×2). The coordination number (CN) of potassium is rather low and typical of less ionic bonding because CN = 8 or even 12 would be expected in an ionic environment. The central BH2 units are not used for coordination to an alkali-metal cation, but they are involved in bifurcated dihydrogen bonding32 to the adjacent NH2 groups [at 2.29(2) Å]. In the crystal structures of three lightweight alkali-metal M(BH3NH2BH2NH2BH3) salts (M = Li, Na, K), one may distinguish layers of BH3NH2BH2NH2BH3− anions separated by flat layers of metal cations (Figure 5). The anions are oriented perpendicular to the layers of metal cations (somewhat reminiscent of self-organization of long aliphatic hydrocarbon chains); this results in large distances between the cationic layers, ranging from 8.5 to 9.4 Å. The length of the C


Article

Inorganic Chemistry

Crystal Structure of β-(C18H39N4PH)(BH3NH2BH2NH2BH3) Precursor. In addition to structural characterization of M(BH3NH2BH2NH2BH3) salts from powder data, we performed a single-crystal X-ray diffraction study of (C18H39N4PH)(BH3NH2BH2NH2BH3) crystals grown from toluene. It turned out that we obtained a new polymorphic form (β) that is triclinic, similar to the previously characterized α polytype,28 but it contains four and not two formula units inside the unit cell (Figure 6).28 (Further details

Figure 5. Comparison of the crystal structures of all alkali-metal M(BH3NH2BH2NH2BH3) salts: M = Li−Cs. Crystallographic unit cell parameters are shown. Figure 6. Projection of the asymmetric part of the unit cell of the β polymorph of (C18H39N4PH)(BH3NH2BH2NH2BH3), including the geometry of the BH3NH2BH2NH2BH3− anion. The displacement ellipsoids are shown at the 50% probability level.

BNBNB skeleton is similar in these compounds and equals ca. 5.4 Å. Rb(BH3NH2BH2NH2BH3) crystallizes monoclinic in the centrosymmetric P21/c space group. The bonding pattern is now different from that seen in its lighter homologues. The first marked difference is that the anions are not straight, but they adopt kinked geometry, which permits formation of intramolecular dihydrogen bonding (at ca. 2.345 Å; Figure 4). Interanion dihydrogen bonds are also present, at 2.2185, 2.244, and 2.2525 Å. The N−B bond lengths range between 1.62(2) and 1.65(2) Å (to the terminal B atom) and 1.60(2) Å (to the middle B atom). Such a bent geometry of the anion is preferred in the gas phase with respect to the quasi-linear conformer, as suggested by DFT calculations.21 As a consequence, the length of the BNBNB core (measured between the terminal B atoms) is only ∼4.8 Å, and the distance between the cationic layers is 7.6 Å (thus shorter than that for Li−K analogues). Moreover, the [NH2] groups are no longer equivalent, a feature that was suggested from FTIR spectra (above) and confirmed with DFT calculations. Last but not the least, the middle [BH2] group is now involved in bonding to an alkali-metal cation, which results in CN = 12 for a Rb+ cation, twice as large as that for its K+ analogue. Cs(BH3NH2BH2NH2BH3) crystallizes orthorhombic in the Pncn space group. The lattice parameters are similar to those of its Rb analogue (aCs ≈ cRb; bCs ≈ 2bRb; cCs ≈ aRb), and the heavy atom sublattice is close to Cmcm [as for Rb(BH3NH2BH2NH2BH3)]. Also, the geometry of an anion and the coordination sphere of a metal are similar to those found for Rb(BH3NH2BH2NH2BH3). The restrained N−B distances are within 1.56(7)−1.57(7) Å. Some of the hydridic H atoms that coordinate Cs atoms are bridging between two Cs cations. The shortest dihydrogen bond is at 2.0721 Å, but this value (as well as other distances involving hydrogen and reported above) must be taken with a grain of salt because of problems in determining the positions of H atoms from PXD, especially if heavy atoms are present in their vicinity.

of the crystal structure of β-(C18H39N4PH)(BH3NH2BH2NH2BH3) may be obtained from Cambridge Crystallographic Data Centre (12 Union Road, Cambridge CB2 1EZ, U.K.; e-mail [email protected]) on quoting the CCD number 1410414, while those of M(BH3NH2BH2NH2BH3) (M = K, Rb, Cs) may be obtained from Fachinformationszentrum Karlsruhe [76344 EggensteinLeopoldshafen, Germany; fax (+49)7247-808-666; e-mail crysdata@fizkarlsruhe.de] on quoting their CSD numbers 429844, 429845, and 429846, respectively.) Both cations creating the asymmetric part of the unit cell have different conformations of aliphatic isopropyl substituents. However, these molecules are not related by a pseudo inversion center; the structure of the main PN cage is preserved. What is interesting is that the conformation of the cation in the α polymorph is almost identical with that for molecule B in the β form. An overlay of cations demonstrating their structural differences is presented in Figure 7. Similar to that in the α form, the aminoborane anions take gauche conformations. This has been attributed to the stabilizing effect of intramolecular NH···HB dihydrogenbonding interactions in the solid state and supported by ab initio calculations.21 Another similarity between the polymorphs is that the average B−N bond lengths to the terminal B atoms (here 1.595 Å) are significantly longer than the average lengths of B−N bonds to their internal B atoms (here 1.575 Å). Notably, the central bonds in the β form are somewhat more altered compared with those in the α form, which reveals rather high structural flexibility of this molecular fragment. The respective distances are 1.565 and 1.586 Å (mean values) and 1.563−1.576 Å28 for the β and α forms, respectively. The anions interact with each other by sets of N−H···B bonds (2.712−3.141 Å) and B−H···H−N dihydrogen bonds (2.060− 2.251 Å). The latter were pointed out by Crabree et al.34 as D


Article

Inorganic Chemistry

Figure 7. Comparison of the geometry of cations in both polymorphs of (C18H39N4PH)(BH3NH2BH2NH2BH3). Overlay performed for PN cage atoms.33

very important for the stability of the systems constituting of BH−NH arrangements. Consequently, infinite chains composed of aminoborane anions are formed. They are propagated along the [001] crystallographic direction (cf. the SI). This kind of arrangement is very similar for both polymorphs; however, the shortest distance between the chains is significantly bigger in the β form [10.299(1) Å] compared with the α form [9.052(2) Å].28 Finally, the crystal structure is stabilized by weak C−H···B interactions (2.902−3.190 Å) and other types of weak contacts that are observed between the cations and anions. Thermogravimetric Analysis (TGA) Profiles and Evolved Gas Analysis (EGA). We investigated the thermal stability of heavier alkali-metal M(BH3NH2BH2NH2BH3) salts upon heating to 200 °C under an inert gas atmosphere (Figure 8) and compared their behavior to that of their lighter analogues. K(BH3NH2BH2NH2BH3), Rb(BH3NH2BH2NH2BH3), and Cs(BH3NH2BH2NH2BH3) exhibit similar TGA profiles and EGA results. They all decompose in the temperature range of 100−180 °C while evolving a small amount of H2 gas together with products of the degradation of BH3NH2BH2NH2BH3− anions such as diborane, ammonia, and other volatile B−N−H products, as seen in the mass spectrometry (MS) analysis (see the SI). Thermal decomposition of K(BH3NH2BH2NH2BH3), Rb(BH3NH2BH2NH2BH3), and Cs(BH3NH2BH2NH2BH3) results in substantial mass loss (19−36 wt %), much beyond what is expected for elimination of pure H2. Moreover, intensities from dihydrogen cations are rather small in the MS spectra. Given that only light elements (N, B, and H) constitute the gaseous impurities, one may notice that ca. 50−55% of the mass of the BH3NH2BH2NH2BH3− anions is volatilized during thermal decomposition. As far as the solid residue is concerned, borohydrides of respective metal cations are the only detected crystalline products of thermal decomposition of M(BH3NH2BH2NH2BH3) salts (see the SI). Because the borohydride anion stands for ca. 20% of the initial mass of the BH3NH2BH2NH2BH3− anion, one can calculate that approximately 30−35% of the initial mass of BH3NH2BH2NH2BH3− anions remains in the solid residue as an unknown amorphous B−N−(H) species. Indeed, the IR spectra of the solid residue (see the SI) reveal the presence of bands typical of metal borohydrides, as well as a broad band at

Figure 8. Comparison of TGA, DSC, and EGA profiles for Li(BH3NH2BH2NH2BH3) (three top graphs) and heavier alkalimetal M(BH3NH2BH2NH2BH3) salts: M = Na−Cs (two bottom graphs; the DSC profile for M = Cs is shown). Measurements performed at a scanning rate of 1 °C/min. The MS profiles of H2 (m/z 2), NH3 (m/z 16), and B2H6 (m/z 28) are shown in the logarithmic vertical scale to enhance the visibility of impurities.

ca. 1500 cm−1 characteristic of B−N.35 The NH stretching and HNH bending bands could not be detected, thus testifying to the absence of H atoms bound to N atoms in the solid residue. To account for all experimental observations, the following reaction equation might be proposed (eq 5): M(BH3NH 2BH 2NH 2BH3) → MBH4 + H 2 ↑ + NH3 ↑ 1 + B2H6 ↑ + BN where M = K, Rb, Cs (5) 2 The theoretical mass losses accompanying eq 5 are 30% (K), 21% (Rb), and 16% (Cs), compared to the measured values of 36% (K), 23% (Rb), and 19% (Cs). The discrepancies may be explained by partial thermal decomposition of MBH4 in the formally protic environment. Equation 5 does not cover the complexity of the thermal decomposition process. The heat effects associated with thermal decomposition of K(BH3NH2BH2NH2BH3), Rb(BH3NH2BH2NH2BH3), and Cs(BH3NH2BH2NH2BH3) are complex and indicative of multistep decomposition (see the SI). This is confirmed by the presence of several maxima in the MS profiles of the gases evolved (see the SI). Li(BH3NH2BH2NH2BH3) is quite different from its heavier analogues. It evolves 5 wt % (up to 140 °C) or 10 wt % (up to 200 °C) of H2 gas free from any contamination.21 Solid products of Li(BH3NH2BH2NH2BH3) thermal decomposition are LiBH4 and amorphous BN, similar to heavier M(BH3NH2BH2NH2BH3) salts (see the SI). The two steps of thermal decomposition might be described by the following reaction equations (eqs 6 and 7): E


Article

Inorganic Chemistry Li(BH3NH 2BH 2NH 2BH3) 2 → LiBH4 + (NHBH)n + 2H 2 n LiBH4 +

borane (with the very short intermolecular separation facilitating the head-to-tail N-to-B interaction) has been shown to decompose via a complex pathway in which the first key step consists of an intermolecular reaction.36,37

(6)

■

2 (NHBH)n + 2H 2 → LiBH4 + 2(BN) + 4H 2 n

CONCLUSIONS We have described a wet-chemistry metathetic approach toward hydrogen-rich M(BH3NH2BH2NH2BH3) compounds. The applicability of this method is exemplified by the entire series of alkali-metal salts, M = Li−Cs, with the derivatives of K, Rb, and Cs described here for the first time. The new method facilitates the preparation of compounds that could not be reached using previously described synthetic methods. M(BH3NH2BH2NH2BH3) salts exhibit some of the highest gravimetric hydrogen contents among inorganic compounds, which range from 5.8 wt % for Cs salt to 15.1 wt % for the Li one. The latter is particularly interesting because, as the only one in this family of compounds, it evolves very pure H2 gas, while decomposing to a mixture of crystalline LiBH4 and amorphous BN and releasing 10 wt % H2 in a low temperature range, 60−150 °C. On the other hand, thermal decomposition of its heavier analogues leads to evolution of H2 gas, which is highly contaminated with volatile low-molecular-mass NHx, BHx, and NBHx species. Supposedly, it is the sublattice composed of small Li+ cations that serves as a template for intermolecular coupling of BH3NH2BH2NH2BH3− anions within the anionic layers present in the crystal structure of this compound; this leads to the formation of high-molecular-mass fragments in the solid residue (oligomerization) and prevents the evolution of impurites.

(7)

each corresponding to a theoretical mass loss of 5 wt %. The mass loss accompanying thermal decomposition of Na(BH3NH2BH2NH2BH3) is smaller than that for the K analogue but larger than the one for Li salt. This suggests that the pathway of thermal decomposition is complex and likely a combination of those proposed for K salt (eq 5) and for the Li one (eqs 6 and 7). Li(BH3NH2BH2NH2BH3) versus M(BH3NH2BH2NH2BH3) (M = Na, K, Rb, Cs). Why Are They Different? It is puzzling at first sight that the presence of Li cations in the M(BH3NH2BH2NH2BH3) salt enables evolution of high-purity H and prevents elimination of volatile impurities (such as NH3, B 2 H 6 , etc.), which come from decomposition of the BH3NH2BH2NH2BH3− anions. There is clearly a transition from Li salt (pure H2 gas is evolved; the mass loss is small), via the Na one (impurities appear; the mass loss is larger), to the heaviest K, Rb, and Cs salts (lots of impurities in H2 gas; very large mass loss). This is associated by strikingly different thermal effects in the differential scanning calorimetry (DSC) profiles (Figure 8 and the SI): thermal decomposition of Li salt is exothermic, the ones of K as well as Rb and Cs salts are endothermic, while that of Na salt comprises both endo- and exothermic effects. One possible explanation of this behavior comes from analysis of the crystal structures of these salts. The average separation between BH3NH2BH2NH2BH3− anions within anionic layers in the crystal structures of Li, Na, and K salts is ∼4.0, ∼4.3, and ∼4.5−4.7 Å, respectively, as judged from relevant lattice constants of their unit cells (and obviously reflecting the size of the alkali-metal cations). Because the said distance in the crystal lattice of Li(BH3NH2BH2NH2BH3) is the shortest, the intermolecular recombination (N-to-B) connected with exothermic oligomerization of the B−N−(H) intermediate should be the most facile. On the other hand, when the said distance is larger, the recombination will be more difficult; concomitantly, the endothermic intramolecular degradation of the BH3NH2BH2NH2BH3− anion will be more favored kinetically, thus leading to low-molecular-mass volatile impurities of H2 gas. Another explanation of changes in the purity of hydrogen desorbed during heating of alkali-metal M(BH3NH2BH2NH2BH3) may lie in the size of the pores in the crystal structure of these compounds. A decrease of the lattice parameters of M(BH3NH2BH2NH2BH3) salts correlates with increasing purity of hydrogen evolved. The bigger the lattice pores, the more facile the evolution of BN intermediates. Thus, Li(BH3NH2BH2NH2BH3) with the smallest lattice pores evolves pure hydrogen. Understanding the role of inter- versus intramolecular coupling/oligomerization for the purity of H2 gas evolved during thermal decomposition of M(BH3NH2BH2NH2BH3) salts requires acquiring additional data for isotope-substituted samples and getting more theoretical (i.e., computational) insight into various aspects of this process, but it certainly merits a separate study. Here we notice that related ammonia

■

EXPERIMENTAL SECTION

Synthetic Procedures. Reactants. All procedures were performed under an argon atmosphere. We used commercially available reagents and solvent: NH3BH3 (98%, JSC Aviabor), Verkade’s base (VB; C18H39N4P, 97%, S.A.), LiH (95%, S.A.), LiNH2 (95%, S.A.), NaH (95%, S.A.), Rb, (99.6%, S.A.), Cs (99.5%, S.A.), LiAlH4 (97%, Alfa-Aesar), NaAlH4 (93%, S.A.), RbCl (99.8%, S.A.), CsCl (99.9%, S.A.) (CF3)3COH (97%, Fluorochem), toluene (99.8%, S.A.), DCM (CH2Cl2, anhydrous, 99.8%, S.A.), and THF (99.9%, S.A.), where S.A. stands for Sigma-Aldrich. Reactions were performed under an argon atmosphere. The products were analyzed without further purification. Samples were stored under an argon atmosphere in a Labmaster DP MBraun glovebox (O2 < 1.0 ppm; H2O < 1.0 ppm) at −35 °C. All further analyses were performed under an inert atmosphere or in a vacuum. Synthesis of (C18H39N4PH)(BH3NH2BH2NH2BH3) Precursor. We performed the synthesis according to the reaction (3). VB was added to the suspension of ammonia borane in toluene at room temperature. The reaction was completed within 3 days. The reaction mixture was left for crystallization overnight at room temperature. The crystals of the β form obtained in this way were washed three times with hexane. Synthesis of M[Al{OC(CF3)3}4] Precursors. We synthesized M[Al{OC(CF3)3}4] (M = Li, Na, K) by adding an excess of (CF3)3COH (ca. 150%) into the suspension of MAlH4 in hexane cooled to ca. 0 °C. The reaction mixture was afterward refluxed overnight and cooled, and the product was filtered, washed with hexane, and dried under a vacuum. Then Soxhlet extraction with perfluorohexane (for Li salt) or dichloromethane (DCM) was performed (for Na and K salts). The synthesis of M[Al{OC(CF3)3}4] (M = Rb, Cs) was performed in a mechanochemical way by milling Li[Al{OC(CF3)3}4] with an excess (200%) of MCl, followed by Soxhlet extraction with DCM.24 Synthesis of Alkali-Metal M(BH3NH2BH2NH2BH3) Salts (New Method). We synthesized alkali-metal M(BH3NH2BH2NH2BH3) salts F


Article

Inorganic Chemistry according to reaction (4). A DCM solution of (C18H39N4PH)(BH3NH2BH2NH2BH3) was added to a dispersion of M[Al{OC(CF3)3}4] in DCM in molar ratio of 1:1 and stirred for 1 h. The precipitate was filtered, washed off with an excess of fresh DCM, and dried under a vacuum. Synthesis of Alkali-Metal M(BH3NH2BH2NH2BH3) Salts (Published Previously). We synthesized M(BH3NH2BH2NH2BH3) salts also according to a previously applied method21 according to the following reaction equations (eqs 8 and 9):

0.71073 Å). A total of 1254 frames were collected with the Bruker APEX2 program.41 The frames were integrated with the Bruker SAINT software package42 using a narrow-frame algorithm. Integration of the data using a triclinic unit cell yielded a total of 106023 reflections to a maximum θ angle of 25.05° (0.84 Å resolution), of which 9341 were independent (average redundancy 11.350, completeness = 99.9%, Rint = 6.05%, and Rsig = 2.87%) and 7540 (80.72%) were greater than 2σ(F2). The final cell constants of a = 10.2994(7) Å, b = 16.7354(11) Å, c = 17.2881(11) Å, α = 109.271(2)°, β = 106.083(2)°, γ = 96.180(2)°, and V = 2636.9(3) Å3 are based on the refinement of the XYZ centroids of 9900 reflections above 20σ(I) with 4.896° < 2θ < 52.32°. Data were corrected for absorption effects using the multiscan method (SADABS).43 The ratio of minimum-to-maximum apparent transmission was 0.918. The structure was solved and refined using the Bruker SHELXTL software package44,45 using the space group P1̅, with Z = 4 for the formula unit C18H52B3N6P. The final anisotropic fullmatrix least-squares refinement on F2 with 527 variables converged at R1 = 3.81% for the observed data and wR2 = 9.58% for all data. The goodness-of-fit was 1.034. The largest peak in the final difference electron density synthesis was 0.335 e−/Å3, and the largest hole was −0.294 e−/Å3 with an root-mean-square deviation of 0.043 e−/Å3. On the basis of the final model, the calculated density was 1.048 g/cm3 and F(000) 928 e−. The non-H atoms were refined anisotropically. Most of the H atoms were placed in calculated positions and refined within the riding model. The temperature factors of H atoms were not refined and were set to be equal to either 1.2 or 1.5 times larger than Ueq of the corresponding heavy atom. The atomic scattering factors were taken from the International Tables for Crystallography.46 Molecular graphics was prepared using the program Mercury 3.3.47 Thermal decomposition was investigated using an STA 409 simultaneous thermal analyzer from Netzsch, in the temperature range −10 to +350 °C. STA 409 allows for simultaneous TGA, DSC, and EGA. The samples were loaded into alumina crucibles. High-purity 6 N argon was used as a carrier gas. The evolved gases were analyzed with a QMS 403C Aëolos mass spectrometer from Pfeiffer−Vacuum. The transfer line was preheated to 200 °C to avoid condensation of residues. DFT Calculations. DFT calculations were performed for K and Rb salts using the generalized gradient approximation with the PBEsol functional optimized for solids.48 We used a 500 eV cutoff energy and a density of the k grid of 0.3 Å−1. Full optimization of the unit cell was performed with the following convergence criteria: 10−7 eV/atom in energy and 10−3 eV/Å in forces. Then phonons at the center of the Brillouin zone were calculated. VASP49 and Phonon50 were used for calculations, as embedded in the Medea package.51

MH + 3NH3BH3 → M(BH3NH 2BH 2NH 2BH3) + 2H 2 + NH3 (8) M + 3NH3BH3 → M(BH3NH 2BH 2NH 2BH3) + 1.5H 2 + NH3 (9) Metal hydride (LiH, NaH, and KH) or metal (Rb and Cs) was covered with a solution of ammonia borane in THF at room temperature. The reaction was completed within a few hours. The solid products were washed several times with fresh portions of THF and left to dry. IR Absorption Spectroscopy. All substrates, products, and thermally decomposed samples were characterized with IR absorption spectroscopy in KBr pellets using a Vertex 80v vacuum FTIR spectrometer from Bruker. The samples sealed in a 0.6-mm-thick quartz capillary under argon were characterized by Raman spectroscopy using a Raman microscopy setup from Jobin Yvon T64000 with CCD detection. An excitation line of 514.5 nm was used. NMR. 1 1 B and 1 H NMR spectra of (C 1 8 H 3 9 N 4 PH)(BH3NH2BH2NH2BH3) and all M(BH3NH2BH2NH2BH3) salts were obtained using a NMR UnityPlus 200 MHz Varian spectrometer. BF3:C2H5OC2H5 (for 11B NMR) and tetramethylsilane (for 1H NMR) were used as external standards. THF-d8 (Aldrich, 99.5 atom % D), dried over metallic sodium, was used as a solvent. One should bear in mind that the NMR spectra probe not only the main phase but also its impurities (as a function of their relative solubilities in the organic solvent). PXRD. PXRD patterns of solids (sealed under argon inside 0.6-mmor 0.5-mm-thick quartz capillaries) were measured using two diffractometers: (a) a Panalytical X’Pert Pro diffractometer with a linear PIXcel Medipix2 detector (parallel beam; Co Kα1 and Co Kα2 radiation intensity ratio of ca. 2:1; λ ∼ 1.789 Å), denoted here as Co Kα; (b) a Bruker D8 Discover diffractometer with a 2D Vantec detector (parallel beam; Cu Kα1 and Cu Kα2 radiation intensity ratio of ca. 2:1; λ ∼ 1.5406 Å), denoted here as Cu Kα. All PXRD results are shown in copper wavelength scale. The PXRD patterns of M(BH3NH2BH2NH2BH3) (M = K, Rb, Cs) were indexed using the Accelrys X-cell program.38 The structures were solved in real space using FOX software, with M+ cations and flexible BH3NH2BH2NH2BH3− anions as building blocks (for M = Cs, the antibump restraints were necessary to achieve a realistic separation of the anions).39 The structures were then refined in Jana2006.40 For M = K, the B−H and N−H distances were restrained close to 1.100(5) and 0.900(5) Å, respectively, the H−X−H angles were restrained to tetrahedral geometry, and the z coordinate of a H atom bound to a N atom was kept the same as the z coordinate of the N atom; the positions of H atoms were then refined within these restraints. For M = Rb and Cs, the H atoms were constrained (no standard uncertainty has been calculated for their positions). In all the structures, the atomic displacement parameters of H atoms (ADP) were set to be 1.2 or 1.5 of the APD of the corresponding heavy atom. For M = Rb, the B−N distances were equalized (with ca. 0.01 Å tolerance), while for M = Cs, also the antibump intermolecular H···H restraints were necessary for refinement, as well as equalizing the intermolecular angles (with ca. 2° tolerance). In the latter structure, also the ADP parameters of all B and N atoms were equalized, and the B−N distances were kept close to 1.600(10) Å. The Rietveld plots are presented in the SI. Single-Crystal X-ray Diffraction. The X-ray measurement of (C18H39N4PH)(BH3NH2BH2NH2BH3) was performed at 100(2) K on a Bruker D8 Venture Photon100 diffractometer equipped with a TRIUMPH monochromator and a Mo Kα fine-focus sealed tube (λ =

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.5b01688. X-ray crystallographic data in CIF format (CIF) X-ray crystallographic data in CIF format (CIF) X-ray crystallographic data in CIF format (CIF) X-ray crystallographic data in CIF format (CIF) Spectroscopic data, Rietveld fits of PXRD profiles, selected TGA/DSC/MS data, and FTIR and Raman spectra (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: karol.fi[email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. G


Article

Inorganic Chemistry Notes

(21) Fijalkowski, K. J.; Jaroń, T.; Leszczyński, P. J.; Magos-Palasyuk, E.; Palasyuk, T.; Cyrański, M. K.; Grochala, W. Phys. Chem. Chem. Phys. 2014, 16, 23340. (22) Wu, C.; Wu, G.; Xiong, Z.; David, W. I. F.; Ryan, K. R.; Jones, M. O.; Edwards, P. P.; Chu, H.; Chen, P. Inorg. Chem. 2010, 49, 4319. (23) Ryan, K. R. Ph.D. Dissertation, University of Oxford, Oxford, U.K., 2011. (24) Jaroń, T.; Orłowski, P.; Wegner, W.; Fijalkowski, K. J.; Leszczyński, P.; Grochala, W. Angew. Chem., Int. Ed. 2015, 54, 1236. (25) Jaroń, T.; Orłowski, P.; Wegner, W.; Fijałkowski, K. J.; Leszczyński, P.; Grochala, W. Chem. - Eur. J. 2015, 21, 5689. (26) Starobrat, A.; Tyszkiewicz, M. J.; Wegner, W.; Pancerz, D.; Orłowski, P. A.; Leszczyński, P. J.; Fijałkowski, K. J.; Jaroń, T.; Grochala, W. Dalton Trans. 2015, 44, 19469. (27) Kisanga, P. B.; Verkade, J. G. Tetrahedron 2001, 57, 467. (28) Ewing, W. C.; Marchione, A.; Himmelberger, D. W.; Carroll, P. J.; Sneddon, L. G. J. Am. Chem. Soc. 2011, 133, 17093. (29) (a) Strauss, S. H.; Nolan, B. G.; Fauber, B. P. Colorado State University Research Foundation. U.S. Patent: WO 2000053611 A1 20000914, 2000; p 32. Krossing, I. Chem. - Eur. J. 2001, 7, 490. Krossing, I.; Reisinger, A. Coord. Chem. Rev. 2006, 250, 2721. (30) Because of facile reaction leading to the formation of M(BH3NH2BH2NH2BH3) salts, attempts to prepare analogous amidoborane precursors have failed, even for a reagent ratio of 1:1. (31) Boese, R.; Blaeser, D.; Weiss, H.-C. Angew. Chem., Int. Ed. 1999, 38, 988. (32) Custelcean, R.; Jackson, J. E. Chem. Rev. 2001, 101, 1963. (33) Verkade, J. G.; Kisanga, P. B. Tetrahedron 2003, 59, 7819. (34) Richardson, T. B.; de Gala, S. D.; Crabtree, R. H.; Siegbahn, P. E. M. J. Am. Chem. Soc. 1995, 117, 12875. Crabtree, R. H.; Siegbahn, P. E. M.; Eisenstein, O.; Rheingold, A. L.; Koetzle, T. F. Acc. Chem. Res. 1996, 29, 348. Klooster, W. T.; Koetzle, T. F.; Siegbahn, P. E. M.; Richardson, T. B.; Crabtree, R. H. J. Am. Chem. Soc. 1999, 121, 6337. (35) Frueh, S.; Kellett, R.; Mallery, C.; Molter, T.; Willis, W. S.; King’ondu, C.; Suib, S. L. Inorg. Chem. 2011, 50, 783. (36) Stowe, A. C.; Shaw, W. J.; Linehan, J. C.; Schmid, B.; Autrey, T. Phys. Chem. Chem. Phys. 2007, 9, 1831. (37) Patwari, G. N. J. Phys. Chem. A 2005, 109, 2035. (38) Neumann, M. J. Appl. Crystallogr. 2003, 36, 356. (39) Favre-Nicolin, V.; Č erný, R. J. Appl. Crystallogr. 2002, 35, 734. (40) Petricek, V.; Dusek, M.; Palatinus, L. Z. Kristallogr. - Cryst. Mater. 2014, 229, 345. (41) APEX2; Bruker AXS Inc.: Madison, WI, 2013. (42) SAINT; Bruker AXS Inc.: Madison, WI, 2013. (43) SADABS; Bruker AXS Inc.: Madison, WI, 2012. (44) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr. 1990, 46, 467. (45) Sheldrick, G. M. Acta Crystallogr., Sect. A: Found. Crystallogr. 2008, 64, 112. (46) Wilson, A. J. C., Ed. International Tables for Crystallography; Kluwer: Dordrecht, The Netherlands, 1992; Vol. C. (47) Macrae, C. F.; Edgington, P. R.; McCabe, P.; Pidcock, E.; Shields, G. P.; Taylor, R.; Towler, M.; van de Streek, J. J. Appl. Crystallogr. 2006, 39, 453. (48) Perdew, J. P.; Ruzsinszky, A.; Csonka, G. I.; Vydrov, O. A.; Scuseria, G. E.; Constantin, L. A.; Zhou, X.; Burke, K. Phys. Rev. Lett. 2008, 100, 136406. (49) Kresse, G.; Furthmüller, J. Comput. Mater. Sci. 1996, 6, 15. Kresse, G.; Furthmüller, J. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169. (50) Parlinski, K.; Li, Z.-Q.; Kawazoe, Y. Phys. Rev. Lett. 1997, 78, 4063. Parlinski, K. Phonon Software; Kraków, Poland, 2011. (51) MedeA: Materials Exploration and Design Analysis, version 2.14.5; Materials Design Inc.: Montrouge, France, 1998−2013.

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This research was funded from Grant 0122/IP3/2011/71 (“Iuventus Plus”) of the Polish Ministry of Science and Higher Education. DFT calculations were performed within the NCN project 2014/15/B/ST5/05012. The authors thank Dr. Dominik Kurzydłowski and Dr. Taras Palasyuk for fruitful discussions. Selected measurements were carried out using the CePT infrastructure financed by the EU European Regional Development Fund within the Operational Programme “Innovative economy” for 2007−2013 (POIG.02.02.00-14024/08-00). We thank ICM UW for time at supercomputers (G29-3).

■

REFERENCES

(1) Grochala, W.; Edwards, P. P. Chem. Rev. 2004, 104, 1283. Maj, Ł.; Grochala, W. Adv. Funct. Mater. 2006, 16, 2061. (2) Hamilton, C. W.; Baker, R. T.; Staubitz, A.; Manners, I. Chem. Soc. Rev. 2009, 38, 279. (3) Churchard, A. J.; Banach, E.; Borgschulte, A.; Caputo, R.; Chen, J.-C.; Clary, D.; Fijalkowski, K. J.; Geerlings, H.; Genova, R. V.; Grochala, W.; Jaroń, T.; Juanes-Marcos, J. C.; Kasemo, B.; Kroes, G. J.; Ljubic, I.; Naujoks, N.; Nørskov, J. K.; Olsen, R. A.; Pendolino, F.; Remhof, A.; Romanszki, L.; Tekin, A.; Vegge, T.; Zach, M.; Züttel, A. Phys. Chem. Chem. Phys. 2011, 13, 16955. (4) Parry, R. W.; Schultz, D. R.; Girardot, P. R. J. Am. Chem. Soc. 1958, 80, 1. (5) Shore, S. G.; Parry, R. W. J. Am. Chem. Soc. 1955, 77, 6084−6085. (6) Baumann, J.; Baitalow, E.; Wolf, G. Thermochim. Acta 2005, 430, 9. (7) Xiong, Z.; Yong, C. K.; Wu, G.; Chen, P.; Shaw, W.; Karkamkar, A.; Autrey, T.; Jones, M. O.; Johnson, S. R.; Edwards, P. P.; David, W. I. F. Nat. Mater. 2008, 7, 138. (8) Diyabalanage, H. V. K.; Nakagawa, T.; Shrestha, R. P.; Semelsberger, T. A.; Davis, B. L.; Scott, B. L.; Burrell, A. K.; David, W. I. F.; Ryan, K. R.; Jones, M. O.; Edwards, P. P. J. Am. Chem. Soc. 2010, 132, 11836. (9) Luo, J.; Kang, X.; Wang, P. Energy Environ. Sci. 2013, 6, 1018− 1025. (10) Diyabalanage, H. V. K.; Shrestha, R. P.; Semelsberger, T. A.; Scott, B. L.; Bowden, M. E.; Davis, B. L.; Burrell, A. K. Angew. Chem., Int. Ed. 2007, 46, 8995. (11) Spielmann, J.; Jansen, G.; Bandmann, H.; Harder, S. Angew. Chem., Int. Ed. 2008, 47, 6290. (12) Zhang, Q.; Tang, Ch.; Fang, Ch.; Fang, F.; Sun, D.; Ouyang, L.; Zhu, M. J. Phys. Chem. C 2010, 114, 1709. (13) Genova, R. V.; Fijalkowski, K. J.; Budzianowski, A.; Grochala, W. J. Alloys Compd. 2010, 499, 144. (14) Fijalkowski, K. J.; Genova, R. V.; Filinchuk, Y.; Budzianowski, A.; Derzsi, M.; Jaroń, T.; Leszczyński, P.; Grochala, W. Dalton Trans. 2011, 40, 4407. (15) Zhang, Y.; Shimoda, K.; Ichikawa, T.; Kojima, Y. J. Phys. Chem. C 2010, 114, 14662. (16) Li, W.; Miao, L.; Scheicher, R. H.; Xiong, Z.; Wu, G.; Araujo, C. M.; Blomqvist, A.; Ahuja, R.; Feng, Y.; Chen, P. Dalton Trans. 2012, 41, 4754. (17) Kang, X.; Luo, J.; Zhang, Q.; Wang, P. Dalton Trans. 2011, 40, 3799. (18) Wu, H.; Zhou, W.; Pinkerton, F. E.; Meyer, M. S.; Yao, Q.; Gadipelli, S.; Udovic, T. J.; Yildirim, T.; Rush, J. J. Chem. Commun. 2011, 47, 4102. (19) Fijalkowski, K. J.; Jurczakowski, R.; Kozminski, W.; Grochala, W. Phys. Chem. Chem. Phys. 2012, 14, 5778. (20) Fijalkowski, K. J.; Grochala, W. J. Mater. Chem. 2009, 19, 2043. H


Article

Inorganic Chemistry

■

NOTE ADDED AFTER ASAP PUBLICATION Due to a production error, this paper was posted on December 16, 2015, before the requested text corrections were implemented. The paper was reposted on December 17, 2015.

I


Hydrogen-Storage Salts Accessed via Metathesis ... - ACS Publications

Recommend Documents