A Self‐Catalyzing Hydrogen‐Storage Material

Greater than the sum of its parts: Hydrogen storage in complex hydrides is accelerated by using the ternary composite 2 LiNH2/LiBH4/MgH2, which exhibits a “self-catalyzing” reaction pathway that results in faster H2 desorption, lower desorption temperatures, and suppression of NH3 release in comparison to the constituent binary composites (see diagram). The enhanced properties arise from the incorporation of an ionic liquid phase (Li4BH3H10) and from ancillary-reaction seeding of a reversible H2 storage reaction. Conventional (e.g. MgH2) and complex hydrides (e.g. alanates, borohydrides, and amides) are the two primary classes of solid-state hydrogen-storage materials.1–3 Many of these “high-density” hydrides have the potential to store large amounts of hydrogen by weight (up to 18.5 wt % for LiBH4) and/or volume (up to 112 g L−1 for MgH2), values that are comparable to the hydrogen content of gasoline (15.8 wt %, 112 g L−1). However, all known hydrides are inadequate for mobile storage applications due to one or more of the following limitations: a) unfavorable thermodynamics (they require high temperatures to release hydrogen4), b) poor kinetics (low rates of hydrogen release and uptake), c) decomposition pathways involving the release of undesirable by-products (e.g. ammonia), and/or d) an inability to reabsorb hydrogen at modest temperatures and pressures (i.e. “irreversibility”). In spite of these drawbacks, renewed interest in complex hydrides has been stimulated recently by substantial improvements in their kinetics and reversibility5, 6 provided by catalytic doping (e.g. TiCl3-doped NaAlH4),7, 8 and by thermodynamic enhancements achieved through reactive binary mixtures9 such as LiNH2/MgH2,10, 11 LiBH4/MgH2,12 and LiNH2/LiBH4.13, 14 These compositions, previously termed “reactive hydride composites”,15 represent the state-of-the-art in hydrogen-storage materials; compared to their constituent compounds, they exhibit improved thermodynamic properties, higher hydrogen purity, and, in some cases, reversibility. The desorption behavior of these previously studied composites is illustrated in Figure 1 a. It is evident from the hydrogen desorption profile (top panel) that the composites generally desorb hydrogen at significantly lower temperatures than their individual components. For example, the lowest temperature reaction, which involves a 2:1 mixture of LiNH2 and MgH2, has a desorption temperature more than 100 °C lower than that of either MgH2 (approx. 350 °C) or LiNH2 (which releases only NH3) alone. Although these binary reactions present significant benefits, they all have well-known disadvantages: a) their lowered desorption temperatures are still too high, b) the reaction involving LiNH2 and LiBH4 is irreversible, c) the nitrogen-containing binaries LiNH2/MgH2 and LiNH2/LiBH4 emit a significant amount of ammonia (a proton-exchange membrane fuel cell (PEM-FC) poison) together with the hydrogen (Figure 1 a, bottom panel), and, most significantly, d) hydrogen desorption/uptake in all of these binary composites is too slow. Therefore, further improvement in these areas is highly desirable. a) Hydrogen (top) and ammonia (bottom) kinetic desorption data as a function of temperature (5 °C min−1 to 550 °C) for the ternary composition (blue trace) and its unary and binary constituents. Hydrogen desorption is measured in weight percent (wt %) to 1 bar whereas relative ammonia release is measured as partial pressure (torr) in a flow-through set-up (100 sccm Ar). b) Ternary phase space defined by unary compounds (nodes), LiBH4 (pink), MgH2 (purple), and LiNH2 (orange) and the binary mixtures (edges), LiBH4/MgH2 (gray), MgH2/LiNH2 (green), and LiNH2/LiBH4 (red). The present ternary composition, which is a 2:1:1 mixture of LiNH2, LiBH4, and MgH2, and previously investigated binaries, are identified. Herein we present a strategy for enhancing the properties of binary composites through the creation of a multi-component composite of three hydride compounds (2 LiNH2/LiBH4/MgH2). The improved properties of this system arise almost entirely from a “self-catalyzing” reaction pathway that results in faster kinetics and lower desorption temperatures than for the binary composites and almost complete suppression of ammonia release. The key elements that contribute to the enhanced properties are the incorporation of a low melting temperature ionic liquid (Li4BH3H10) and a mechanism for seeding the products of a reversible hydrogen desorption reaction.16 The choice of the 2 LiNH2/LiBH4/MgH2 stoichiometry is based on several factors: a) the constituent hydrides all possess high gravimetric/volumetric capacities, b) binary mixtures of these hydrides are among the best known hydrogen-storage materials (see Figure 1 a, top panel), c) mixtures containing MgH2 are known to suppress ammonia release from nitrogen-containing hydrides such as LiNH2 (Figure 1 a, bottom panel), and d) a stable, lightweight compound, namely lithium magnesium boron nitride (LiMgBN2), which contains N, B, and Mg in a 2:1:1 ratio (the same as our stoichiometric composite), is known which could serve as a potential dehydrogenated product phase. The compositional phase space of the ternary composite is shown in the Gibbs triangle in Figure 1 b and information regarding the sample synthesis and preparation conditions is given in the Experimental Section. We begin our discussion of this system by summarizing its principal hydrogen-storage attributes in relation to those of the unary and binary components. Employing a wide range of experimental phase analyses and first-principles evaluation of reaction thermodynamics, we subsequently identify a complex, “cascading” sequence of reactions that explain the observed properties. We conclude with a more detailed discussion of the proposed “self-catalyzing” mechanism. Lowered desorption temperatures: The measured kinetic desorption behavior (5 °C min−1, 1 bar) of the ternary composite is compared with the constituent unary and binary components in the upper panel of Figure 1 a. The ternary system rapidly releases hydrogen in a process that begins at 150 °C (top panel), which is about 50–200 °C lower than the binary composites, thereby indicating significantly improved kinetics and/or thermodynamics. The total capacity of the ternary composite is 8.2 wt %. Improved hydrogen purity: The composition of the gas released from the ternary composite while heating at 5 °C min−1 in a flow of 100 sccm argon is plotted in comparison with the binary composites in the lower panel of Figure 1 a. The ammonia released by the ternary composite is less than the 100 ppm detection limit of our instrument; the ammonia released from the nitrogen-containing binaries was found to be more than an order of magnitude larger. No other volatile boron- and/or nitrogen-containing by-products were detected throughout the desorption process. Reversibility: The reversible storage capacity and response to cycling were determined from a series of charge/discharge experiments performed with a Sievert-type PCT apparatus at 160 °C and charging (discharging) at 100 (1) bar. The results for five charge/discharge cycles (Figure 2 a) show that the as-prepared material rapidly releases approximately 3.0 wt % of hydrogen within 20 minutes. After recharging, the second through fifth desorption cycles consistently liberate around 2.8 wt % of hydrogen, a reversible capacity at moderate temperature that is among the best for solid-state hydrogen storage.5–8 a) Hydrogen desorption kinetics at 160 °C and 1 bar over five cycles after subsequent charging (160 °C and 100 bar hydrogen). b) Reversible isothermal kinetic hydrogen desorption data (to 1 bar) for the ternary composition based on the second desorption cycle at 140 (•), 150 (▾), 160 (□), and 180 °C (⧫) versus time. Kinetics: The reversible isothermal kinetic desorption profiles for the second desorption cycle (to 1 bar) were collected at 140, 150, 160, and 180 °C (Figure 2 b). The ternary composite is capable of desorbing more than 2.5 wt % hydrogen in times ranging from 10 min (180 °C) to 2.5 h (140 °C) in this temperature range. The remaining hydrogen is liberated in a second step at higher temperatures for a total hydrogen capacity of 8.2 wt % (Figure 3). The initial release of hydrogen at both 260 and 320 °C is dramatically accelerated, with 3.2 wt % released within minutes, while the subsequent desorption steps are more influenced by temperature, reaching full desorption after 1.5 and 14 h at 320 and 260 °C, respectively. Isothermal kinetic hydrogen desorption to 1 bar for the ternary composition at 260 and 320 °C versus time. The unique desorption behavior described above strongly suggests that the reaction mechanism(s) of the ternary composite is not a simple superposition of the known binary reactions. To understand its hydrogen-release characteristics, we therefore collected temperature-programmed-desorption mass spectrometry (TPD-MS) data at a constant heating rate and carrier gas flow (5 °C min−1, 100 sccm argon flow; Figure 4 a). Four distinct hydrogen-release events occur (maxima at 180, 190 (shoulder), 310, and 560 °C, respectively), with an initial desorption onset at 110 °C.17 TPD-MS data were also collected for the cycled/recharged material (See Supporting Information). These data clearly show that the first steep desorption step (at 180 °C) in the as-prepared sample is no longer observed in the recharged sample. Instead, the peak temperature for the recharged sample is now shifted to the shoulder region for the fresh material (approx. 190 °C), thus indicating that the reaction corresponding to the shoulder is reversible, which is consistent with the powder X-ray diffraction (PXRD) and IR spectroscopic analyses (discussed below). a) TPD-MS data for the ternary composition depicting hydrogen (m/z 2, blue) and ammonia (m/z 17, pink) as a function of temperature (heating at 5 °C min−1 to 575 °C). b) Phase composition as a function of desorbed hydrogen amount (wt %) and temperature (°C) as determined from the static PXRD and IR data. Phase identification: Phase-composition studies were carried out for identically prepared samples, which were desorbed to varying degrees at 1 bar hydrogen by heating at 5 °C min−1 in a water displacement apparatus (see Supporting Information) to identify the species involved in the various desorption reactions. Following desorption, each sample was quenched and analyzed by PXRD and IR spectroscopy. The results are summarized in Figure 4 b, and raw data and phase assignments are provided as Supporting Information. The as-prepared sample (ball milling 2 g of LiNH2, LiBH4, and MgH2 in a 2:1:1 ratio for 5 h) contains two new species (Mg(NH2)2 and Li4BN3H10) and no residual LiNH2, which is indicative of milling-induced transformations. Residual MgH2 and LiBH4 starting materials are also present. Growth of Mg(NH2)2 and (weakly crystalline) LiH is detected upon initial heating to 140 °C before any appreciable amount of hydrogen is released. At the same time, the diffraction peaks for Li4BN3H10 disappear. As the characteristic symmetric and asymmetric amide NH IR frequencies (observed: 3301 and 3242 cm−1; literature:18 3303 and 3243 cm−1) persist, we conclude that Li4BN3H10 has melted. Further heating to 180 °C results in the release of 2.0 wt % hydrogen (first low temperature event in Figure 4 a) and the formation of Li2Mg(NH)2, based on its three characteristic peaks at 30.7°, 51.3°, and 60.9° in the PXRD pattern19 and the signature NH stretch in the IR spectrum (observed: 3178 cm−1; literature:20 3187 cm−1). This phase continues to grow in intensity until 255 °C, at which point 4.0 wt % H2 has desorbed. At this stage, MgH2 and Mg(NH2)2 have been completely consumed while Li4BN3H10 is significantly depleted. The second major hydrogen releasing event occurs between 255 and 375 °C and corresponds to a total of 8.2 wt % desorbed hydrogen. Li2Mg(NH)2 and LiBH4 are consumed during this stage and Mg3N2 and Li3BN2 are formed. Trace amounts of LiH and an unknown phase (denoted as “Phase X”) are also detected by PXRD.21 Further heating to 500 °C does not produce additional hydrogen but rather an observed phase transformation consistent with the consumption of Li3BN2, Mg3N2, and LiBH4 and the production of LiH and LiMgBN2.22 The final hydrogen releasing step (>500 °C) is attributed to decomposition of LiH (third major event in Figure 4 a). Variable-temperature in situ PXRD was used to validate the above phase assignments and to provide phase transformation information (see the Supporting Information for instrument set-up and data collection protocol). Figure 5 a shows the raw PXRD data as a function of temperature (25–450 °C) and Figure 5 b shows the two-dimensional contour plot. The phase assemblage as a function of temperature is shown in Figure 5 c. The data reveal that the sequence and relative phase contributions are identical to those observed by static PXRD, thereby confirming the proposed reaction sequence. Furthermore, the in situ data reveal that the Li4BN3H10 and MgH2 phases disappear rapidly (by 100 and 150 °C, respectively) during initial heating of the as-prepared material and prior to any hydrogen release. The observed melting of Li4BN3H10 at 100 °C occurs at a significantly lower temperature than previously reported (150 °C).23 This low temperature melt may serve as an effective mass transfer medium or homogenizing agent and aid in the distribution of Li2Mg(NH)2 (produced in the first desorption step reaction between Li4BN3H10 and MgH2), which would in turn serve as Li2Mg(NH)2 nucleation seeds for a second step reaction between Mg(NH2)2 and LiH. The presence of the ionic liquid may therefore positively influence the desorption kinetics of the initial hydrogen release reactions. a) Raw PXRD data for the ternary composition as a function of temperature (25–450 °C). b) The corresponding two-dimensional contour plot derived from the raw patterns in (a). c) Plot of the relative amounts of individual phases as a function of temperature. Reaction pathway: A set of proposed reactions are summarized in Figure 6. Our assignment of reactions takes into account the observed and theoretical hydrogen capacity for each step, the reversible amount of stored hydrogen, and the phase compositions (obtained from both quenched/static and in situ PXRD and IR spectroscopy). A reaction scheme flowchart is included in the Supporting Information. The TPD-MS curve from Figure 4 a is incorporated to indicate the temperature region under which each reaction occurs. The reaction enthalpies (ΔHcalcd) and free energies (ΔGcalcd) at 300 K obtained by density functional theory calculations are also included in this table. The fact that all the calculated free energies are negative suggests that the proposed reactions are thermodynamically reasonable. (The activation energies for reactions (2) and (4) are given as Supporting Information.) Proposed reaction pathway for the ternary composite, including the observed/theoretical hydrogen capacity, reaction enthalpy (ΔH), free energy (ΔG) (both in kJ mol−1 for products at 300 K), and the corresponding temperature range (coupled to the TPD-MS curve). The reaction numbers in this figure correspond to those used in the text. We refer to the ternary composite as “self-catalyzed” in the sense that one reaction [reaction (2)] pre-forms the product nuclei (Li2Mg(NH)2) for the subsequent reaction [reaction (3)], which results in an enhancement of the overall kinetic properties. A separate study has confirmed the beneficial effects of product seeding in improving the desorption kinetics of the Mg(NH2)2/LiH system.24 It should be emphasized that the thermodynamics of the binary reaction between Mg(NH2)2 and LiH [reaction (3)] indicate that it should proceed at a lower temperature than observed. Our results suggest a new rational route by which the kinetic properties of existing hydrogen-desorption reactions can be enhanced, namely by coupled self-catalyzing reactions. In conclusion, our study of the ternary LiBH4/2 LiNH2/MgH2 composite has led to the discovery of a new “self-catalyzing” strategy for enhancing the kinetics of hydrogen storage in complex hydride composites. We have demonstrated through a wide-ranging experimental and first-principle computational analysis that this self-catalyzing mechanism arises from a set of coupled, ancillary reactions that yield both a homogenizing ionic liquid phase and product nuclei for a subsequent reversible hydrogen-storage reaction. These effects combine to yield enhanced low-temperature desorption kinetics and a significant reduction in ammonia liberation relative to the state-of-the-art binary constituent composites. The strategy of utilizing built-in, ancillary reactions to catalyze a primary hydrogen-storage reaction suggests prospective routes for advancing existing and future storage materials. Sample Preparation: LiNH2 (95 % purity, Sigma-Aldrich), MgH2 (95 % purity, Gelest), and LiBH4 (95 % purity, Sigma-Aldrich) were used as received. All sample handling was performed in an MBraun Labmaster 130 glove box maintained under argon with less than 0.1 ppm O2 and H2O vapor. The binary composites 2 LiNH2/LiBH4, 2 LiNH2/MgH2, and 2 LiBH4/MgH2 were prepared according to literature protocols.10, 12, 13 For the ternary composite, two grams of LiNH2, LiBH4, and MgH2 in a 2:1:1 molar ratio was loaded into a milling vial containing three stainless steel balls weighing 8.4 g each. Mechanical milling was carried out using a Spex 8000 high-energy mixer/mill for 1–20 h. Characterization and Property Evaluation: All methods relating to sample characterization and property evaluation, including powder X-ray diffraction (PXRD), IR spectroscopy, kinetic hydrogen desorption/absorption studies (PCT, TPD-MS, and WDD), density functional theory (DFT) calculations, and activation energy calculations are described in detail in the Supporting Information. Supporting information for this article is available on the WWW under http://www.wiley-vch.de/contents/jc_2002/2008/z703756_s.pdf or from the author. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.