Hydrogen generation from ammonia borane and water through

Combustion and Flame 162 (2015) 1498–1506

Contents lists available at ScienceDirect

Combustion and Flame j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m b u s t fl a m e

Hydrogen generation from ammonia borane and water through combustion reactions with mechanically alloyed AlMg powder Daniel A. Rodriguez a, Edward L. Dreizin b, Evgeny Shafirovich a,⇑ a b

Department of Mechanical Engineering, The University of Texas at El Paso, El Paso, TX 79968, USA Department of Chemical, Biological, and Pharmaceutical Engineering, New Jersey Institute of Technology, Newark, NJ 07102, USA

a r t i c l e

i n f o

Article history: Received 5 September 2014 Received in revised form 13 November 2014 Accepted 13 November 2014 Available online 2 December 2014 Keywords: Heterogeneous combustion Energetic materials Gas generators Metal combustion Laser ignition Hydrogen

a b s t r a c t It is known that ammonia borane (AB) forms combustible mixtures with gelled water and nanoscale aluminum powder. The reaction of nanoaluminum with water serves as a source of heat for ammonia borane thermolysis and hydrolysis, also releasing additional hydrogen from water. Nanoaluminum, however, has drawbacks such as high cost and reduced amount of free metallic aluminum. The present paper investigates a feasibility of using a mechanically alloyed AlMg powder instead of nanoaluminum in these mixtures. Initial experiments showed that mixtures of mechanically alloyed AlMg powder with gelled water are combustible. The velocities of combustion front propagation exceed those obtained for mixtures of nano-Al powder with gelled water. Then, combustion experiments were conducted with mixtures of AB, mechanically alloyed AlMg powder, and gelled heavy water (D2O). Heavy water was used to investigate the reaction mechanisms through mass-spectroscopy of released H2, HD, and D2 gases. The isotopic tests have shown that AB participates in two parallel processes – thermolysis and hydrolysis, thus increasing hydrogen yield. Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction

ð1Þ

Here the byproducts in the first and second steps are polymeric aminoborane and polymeric iminoborane, respectively, but some other byproducts such as monomeric aminoborane, borazine, ammonia, and traces of diborane were also detected [2–6]. The byproduct in the third step is hexagonal boron nitride [7]. Thermal analysis of AB decomposition has shown that the first step occurs near the melting point of AB (104 °C). At a low heating rate in the test, it may decompose while it is still solid, while at higher heating rates decomposition starts after melting [3]. The second, broader step occurs at 125–200 °C, where the upper limit corresponds to higher heating rates [2]. The third step starts at 1170 °C and it is not finished by 1500 °C, the maximum temperature in the test [7]. Another method for hydrogen release from AB is hydrolysis, which occurs when a suitable catalyst (typically, an acid or a noble metal) is added to the mixture of AB and water [8–11]. This reaction can be described by the equation:

ð2Þ

NH3 BH3 þ 2H2 O ! NHþ4 þ BO2 þ 3H2

ð3Þ

Recently, hydrothermolysis, i.e., combination of thermolysis and hydrolysis, of AB has been proposed [12]. The idea of that method was to increase the temperature of water to the values at which the first and second steps of AB thermolysis occur and possibly conduct both thermolysis and hydrolysis with no catalyst. The desired

There is demand for hydrogen storage systems with high hydrogen densities and rapid generation rates. For such systems, chemical hydrogen storage is a promising alternative to compressed gas and liquid storage. Chemical hydrogen storage systems typically include a solid compound with a high content of hydrogen. This compound should release hydrogen either upon heating or as a result of reactions with other components. One attractive compound is ammonia borane (AB, H3NBH3), which contains 19.6 wt% hydrogen and is a stable powder under standard conditions [1,2]. Hydrogen release from AB can be achieved by different methods. One of them is thermal decomposition (thermolysis). This process includes three steps, each releasing about one of the three equivalents of hydrogen available in AB [2]:

1 H3 N BH3 ! ðH2 N BH2 Þx þ H2 x 1 1 ðH2 N BH2 Þx ! ðHN BHÞx þ H2 x x 1 ðHN BHÞx ! BN þ H2 x ⇑ Corresponding author. Fax: +1 915 747 5019. E-mail address: eshafi[email protected] (E. Shafirovich).

http://dx.doi.org/10.1016/j.combustflame.2014.11.019 0010-2180/Ó 2014 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

ð4Þ

D.A. Rodriguez et al. / Combustion and Flame 162 (2015) 1498–1506

elevated water temperatures can be achieved at increased inert gas pressures in the reactor. Experiments showed that both thermolysis and hydrolysis occur under such conditions. Along with increasing the ambient gas pressure, another way for simultaneously conducting thermolysis and hydrolysis of AB was proposed in the same paper [12]. It was shown that the mixtures of AB, water, and nanoscale aluminum powder were combustible. They could be readily ignited and burned while releasing hydrogen. The combustion wave was driven by a highly-exothermic reaction between Al and water. The reported hydrogen yield exceeded the maximum theoretical hydrogen yield of Al/H2O reaction, clearly indicating that hydrogen was released from both AB and water. However, the mechanism of hydrogen release from AB in the nano-Al, AB, and water system remains unclear. Because of high temperatures in the combustion wave, it may be primarily thermolysis of AB. On the other hand, the presence of water may lead to its hydrolysis as well. Note that combustible mixtures of Al with liquid water and ice have been studied by several research teams for space propulsion and hydrogen generation applications [13–29]. In these mixtures, nanoscale aluminum powders were used. When micron-sized Al replaced over 20% of nanoscale Al, the burn rate of the mixture decreased significantly [22,23]. Nanoscale aluminum powders, however, are relatively expensive, difficult to handle, and have substantially reduced active aluminum content, compared to conventional micron-scale powders. Replacing nano-aluminum with a micron-sized metal powder is, therefore, desirable provided this does not deteriorate the combustion performance. In several studies, the reaction of aluminum with water was greatly improved by the addition of Li [30–32] or Ga [33–36]. These approaches, however, are not applicable to storable Al/water mixtures because the reaction occurs vigorously at room temperature. Recently, mechanically alloyed AlMg powders were developed as promising metal fuels for oxygen-generating compositions based on sodium chlorate [37]. Mechanically alloyed powders were shown to be more reactive than regular alloys or pure metals [38–41]. Their developed surface area accelerates ignition, which is also assisted by weakly exothermic, subsolidus intermetallic reactions. It is important that a mechanically alloyed AlMg powder consists of relatively large particles (the mean volume diameter is around 10 lm [40,41]) and the fabrication method is simple and readily scalable. Also, the rate of Mg reaction with distilled water is very low [42] so that it is expected that mixtures of Mg/Al powder with water will be stable. Thus, it is of interest to consider this powder as a substitute for nanoscale Al in the mixtures of AB with water. One objective of the present work was to investigate mixtures of ammonia borane, water, and mechanically alloyed AlMg powder, specifically, to explore their combustion and determine hydrogen yield. This investigation involved laser ignition of the mixtures in an argon environment. Preliminarily, mixtures of the AlMg powder with water were tested and then AB was added to the system. Another objective was to clarify the mechanisms of hydrogen release from AB during its combustion with water and mechanically alloyed AlMg powder. This was achieved through isotopic tests that used heavy water (D2O) instead of H2O. Mass-spectroscopic analysis of the gaseous combustion products was used to identify the source of hydrogen (or deuterium). Combustion of NH3BH3/D2O/ AlMg system may produce H2 from AB thermolysis, D2 from the reaction of AlMg with D2O, and HD from AB hydrolysis. Analysis of the released gases reveals relative contributions of these three processes.

1499

2. Experimental 2.1. Preparation of mixtures Mechanically alloyed AlMg (1:1 mass ratio) powder was prepared in a planetary ball mill (Retsch PM-400 MA) equipped with an air conditioner that cools the milling compartment. Starting materials included elemental powders of Al (Atlantic Equipment Engineers, 99.8% pure, 325 mesh) and Mg (Alfa-Aesar, 99.8% pure, 325 mesh). The AlMg powder was prepared following a twostage procedure [40,41]. Powders of Al and Mg and 9.5-mm diameter hardened steel balls were loaded in steel milling vials in argon. The powder charge was 30 g per vial and the ball-to-powder mass ratio was 10. Hexane (50 mL) was added to each milling vial as a process control agent (PCA). Milling time was 120 min. The rotation speed was set at 350 rpm and the rotation direction changed every 15 min. The first stage produced a coarse, mechanically alloyed powder. The second stage of milling, aimed to reduce the particle size, involved the addition of a new PCA, iodine (I2, chips, Sigma Aldrich, 99% pure), at 4 wt% of the initial powder load. The 9.5-mm balls were removed and replaced with the same mass of 3-mm hardened steel balls. The duration of the second milling stage varied between 65 and 125 min, depending on the effectiveness of the air-conditioner cooling which was affected by the air humidity. One important issue for applications is stability of the prepared materials in aqueous mixtures. Reaction of the mechanically alloyed AlMg powder with hot water was investigated with a setup that was previously used in the kinetic studies of the reaction between activated Al powder and water [43]. The setup includes a digital hotplate (Scilogex MS7-H550-Pro) and a system for measuring the volume of released gas based on water displacement in an inverted graduated cylinder. A sample (0.7 g) of the AlMg powder was submerged in 750 mL of deionized water at 80 °C. After 24 h of heating, no reaction was detected. Apparently, the formation of an oxide film on the particles prevented further oxidation. This implies that the mixtures of this powder with water may remain stable for a long time. The obtained mechanically alloyed AlMg powder was mixed with deionized water. To prevent sedimentation of particles, water was gelified by adding polyacrylamide (PAM, linear formula (C3H5NO)n, mass average molecular mass 5 106–6 106, Sigma). First, the gellant was added to water and mixed manually for several minutes. Then the AlMg powder was mixed (also manually) with the obtained gel. The mass fraction of water in the metal–water mixture was varied from 10% to 60%. A sample of the resulting mixture was then placed in a quartz tube (inner diameter 7.5 mm, thickness 1 mm, height 25 mm) for the combustion experiments. In the mixtures that contained both mechanically alloyed AlMg powder and ammonia borane, heavy water (D2O, Aldrich, 99.9 at% D) was used. First, D2O was mixed with PAM. Then the obtained gel was mixed with the AlMg powder and ammonia borane using an acoustic mixer (Resodyne LabRAM). The mixture compositions are described in Section 3.2. Mixing for 1 min at 50% maximum intensity produced a uniform mixture. A sample (1.1–1.2 g) of the obtained mixture was then placed in a quartz tube (same dimensions as shown above) with the addition of a booster pellet (0.2– 0.3 g) on top of the sample. The booster pellet was a stoichiometric mixture of D2O and mechanically alloyed AlMg powder. It was used for facilitating ignition of the main mixture. 2.2. Combustion experiments Prepared mixtures were ignited using a CO2 laser. The experimental setup shown schematically in Fig. 1 includes a stainless

1500


steel chamber (volume: 11.35 L), equipped with a door port and three windows for observation and video recording. The chamber is connected to a mass-spectrometer (Pfeiffer Omnistar GSD 320) for analysis of gases generated during combustion. Before each experiment, a tube with a sample was installed on a brass pedestal inside the chamber, which was three times evacuated and filled with ultra-high purity argon (99.999%) to establish an Ar pressure of 1 atm. During the evacuation, the chamber pressure was decreased to 15 kPa, while still remained well above the saturation pressure of water at room temperature (3.2 kPa). Thus, boiling was avoided and only a relatively slow evaporation occurred. An infrared beam (wavelength: 10.6 lm, diameter: 2.0 ± 0.3 mm) of a CO2 laser (Synrad Firestar ti-60) was introduced into the chamber vertically through a zinc selenide (ZnSe) window, located at the chamber lid, and directed to the top of the sample. For alignment of the optical system, a laser diode (Synrad Diode Pointer) was used. The power of the infrared beam after passing the beam delivery system and ZnSe window was measured with a powermeter (Synrad PW-250) and controlled using a laser controller (Synrad UC-2000), while the duration of the laser pulse was controlled using LabVIEW (National Instruments) software. In all reported experiments, the power and duration of the laser pulse were 39 ± 1 W and 2 s, respectively. After ignition, the combustion front propagated downward through the sample. A digital video camera (Sony XCD-SX90CR) recorded the experiment. The pressure increase due to the released gases was recorded with a pressure transducer (Omegadyne PX409-030AI). Mass-spectroscopic analysis of the released gases was performed after cooling to room temperature. To enable quantitative analysis, the mass-spectrometer was calibrated using hydrogen (H2, 99.999% pure, Airgas), deuterium hydride (HD, 96 mol% HD, 98 at% D, Aldrich), and deuterium (D2, 99.8 at% D, Aldrich). Condensed products were characterized using X-ray diffraction analysis (Bruker D8 Discover XRD).

energy input was about 80 J, while the released heat of combustion was several kJ. Figure 2 shows average velocities of the combustion front measured in mixtures with 1 and 3 wt% PAM. Here and throughout the paper, the percentage of polyacrylamide is given with respect to water. Each test was repeated three times and the error bars show standard deviation for the observed flame velocities. At 3 wt% PAM, a sharp peak in the flame velocity as a function of water concentration is observed. At less than 20 wt% H2O or more than 40 wt% H2O, the front velocity was relatively low (see also Fig. 3 and Supplementary Video 1). At 30–40 wt% H2O, the velocity was much higher (see also Supplementary Video 2). The peak velocity observed at 30 wt% of H2O with 3 wt% PAM was consistently observed in three repeated tests. The accelerating effect of polyacrylamide on the combustion may be associated with its decomposition and possible formation of chelate compounds that inhibit growth of a protective oxide film on the metal surface [21]. The measured combustion front velocities were compared with the data obtained for stoichiometric nano-Al–H2O compositions at a pressure of 1 atm (or 0.1 MPa) where water was gelled with 3 wt% PAM [14,18]. The comparison shows that the front velocity for the mechanically alloyed AlMg powder significantly exceeds those for 120-nm Al powder (2 mm/s [14]) and 38-nm Al powder (7 mm/s [18]). It is important that the 38-nm powder had an active aluminum content of only about 54.3 wt% [18], while the fraction of oxidized metal in the mechanically alloyed AlMg powder is not expected to exceed 2 wt%. Figure 4 shows the values of hydrogen yield, calculated from the pressure increase measured after combustion and cooling, with the assumption that the released gas is pure hydrogen. For

3. Results and discussion 3.1. Mixtures of mechanically alloyed AlMg powder and water The attempts to ignite stoichiometric mixtures of mechanically alloyed AlMg powder and water with no gellant were unsuccessful. Thickening water with 1 wt% polyacrylamide made the mixtures combustible over a range of water concentrations from 20 to 60 wt%. The combustion was self-sustained, i.e., the front propagated after turning off the laser, and the laser beam

Fig. 1. Schematic diagram of the experimental setup.

Fig. 2. Combustion front velocities in AlMg–water mixtures vs water concentration, at (a) 1 wt% and (b) 3 wt% polyacrylamide.

1501


0

1s

2s

3s

4s

5s

6s

7s

Fig. 3. Combustion of AlMg–H2O mixture at 20 wt% H2O. Time zero was selected arbitrarily.

comparison, the figure also shows theoretical values calculated with assumption of complete conversion of the following reactions:

2Al þ 3H2 O ! Al2 O3 þ 3H2

ð5Þ

Mg þ H2 O ! MgO þ H2

ð6Þ

It is seen that most experimental values exceed the theoretical ones, and the difference increases with increased concentration of polyacrylamide. These observations are explained by PAM decomposition, which generates gases and increases pressure in the chamber. Note that flash pyrolysis of PAM at 700 °C led to the formation of 24 organic products [21]. Such a variety of possible products makes it difficult to estimate the pressure increase based on the mass of decomposed PAM for comparison with the experimental data. Mass-spectroscopic analysis of the gas environment in the

chamber after combustion has detected additional peaks along with those of argon, hydrogen, and air traces, which confirms the presence of gases evolved from polyacrylamide. Further mass-spectroscopic analysis was focused on the measurements of hydrogen concentration. Using binary Ar/H2 gas mixtures with different hydrogen contents, the mass-spectrometer was calibrated to measure H2 concentration in the gas environment after combustion. Figure 5 shows the obtained values of hydrogen yield in comparison with theoretical values. It is seen that the efficiency of hydrogen release from the stoichiometric mixture is about 80%. Apparently, part of water escaped from the mixture because of vaporization during laser heating and as an unreacted water vapor during the combustion front propagation. Figure 6 presents an XRD pattern of the solid products collected after combustion of mechanically alloyed AlMg powder with water at 46.5 wt% water. This concentration corresponds to the stoichiometric ratios for reactions (Eqs. (5) and (6)). The identified compounds in the products are MgAl2O4 and MgO. The absence of Mg peaks indicates that the reaction of Mg with water was complete (within the limits of XRD accuracy). According to the XRD database, the peaks of Al overlap with those of MgAl2O4 so that no conclusion about the presence of Al in the products can be made from the XRD pattern.

3.2. Mixtures of ammonia borane, mechanically alloyed AlMg powder, and water Thermodynamic calculations of the adiabatic flame temperatures and combustion product compositions were conducted using THERMO (version 4.3) software, which is based on the Gibbs free energy minimization and contains a database of approximately

Fig. 4. Hydrogen yield determined from pressure measurements for AlMg–water mixtures vs water concentration at (a) 1 wt% and (b) 3 wt% polyacrylamide, in comparison with theoretical values.

Fig. 5. Hydrogen yield determined by mass-spectrometry for AlMg–water mixtures vs water concentration at 3 wt% polyacrylamide, in comparison with theoretical values.

1502


Fig. 6. XRD pattern of combustion products for the stoichiometric mixture of mechanically alloyed AlMg powder and water.

3000 compounds [44]. Since the available THERMO database did not contain thermochemical properties of ammonia borane, the formation enthalpy of this compound, 178 kJ/mol [45], was added to the database. The amount of H2O corresponded to the stoichiometry of reactions with the AlMg powder (Eqs. (5) and (6)). No extra water for reactions with AB was added. The amount of AB was varied from 0 to 100%. Pressure was equal to 1 atm in all calculations. Figure 7 shows the obtained adiabatic flame temperature and gravimetric hydrogen yield (the mass of released hydrogen per unit mass of the initial mixture). It is seen that the dependence of hydrogen yield on AB concentration is almost linear, while the temperature curve has three parts. The middle part, from 10 to 50 wt% AB, has a much smaller slope than two others (50 wt%). Since a self-sustained combustion is possible at sufficiently high adiabatic flame temperatures, it is reasonable to test mixtures in the range of AB from 10 to 50 wt%. The goal is to obtain the highest hydrogen yield while maintaining the mixture’s ability to sustain the reaction. It should be noted that although the calculated product compositions include many compounds (H2, BN, Al2MgO4, MgO, Al2O3, AlOH, Al, AlN, Mg, MgOH, MgH, Mg(OH)2, BHO, BHO2, B2O2, B2O3, BO, N2, H2O, OH, H, O), some byproducts of AB thermolysis, such as aminoborane, iminoborane, borazine, and ammonia borane itself, are absent in the database and thus cannot appear among the calculated products. Because of this, in the calculations, 100% ammonia borane decomposes to boron nitride and hydrogen with

Fig. 7. Adiabatic flame temperature and gravimetric hydrogen yield calculated for mixtures of ammonia borane, AlMg, and H2O vs the concentration of ammonia borane in the initial mixture.

a heat release, while in reality it is a stable compound at room temperature. Thus, although thermodynamic calculations can be used to initially select the most promising mixtures for experiments, caution should be exercised while using the predictions to interpret the experimental results. To enable isotopic tests, combustion experiments with the mixtures were conducted using D2O instead of H2O. Table 1 shows compositions of the tested mixtures, both in mass and mole fractions. Note that use of D2O instead of H2O changes the mass fractions, but does not change the mole ratio. Mixtures 1, 2, and 3 were designed based on the assumption that AB decomposes thermally, while the amount of water (in moles) corresponds to the stoichiometric reactions with Al and Mg (Eqs. (5) and (6)). Mixture 4 included more water to enable hydrolysis of AB while maintaining enough water for the reaction with AlMg. The attempts to ignite mixture 1 were unsuccessful, while three other mixtures were combustible. Figure 8 shows combustion of mixture 3. The first and second images correspond to combustion of the booster pellet. The next four images show propagation of the combustion front over the main mixture. They also show formation of a wide glowing zone at the sample top, i.e., behind the front. This zone continued to emit light for several seconds after the front reached the bottom (see also Supplementary Video 3). Although this zone was located at the top, it could not be related to the booster pellet because its solid combustion products (oxides of Al and Mg) were ejected by the released gas from the tube. The formation and continuous glowing of a wide luminous zone was observed in all experiments with mixtures that contained AB. On the other hand, this phenomenon was not observed in the experiments with AlMg–H2O (see Fig. 3 and Supplementary Videos 1 and 2), Al–H2O (see images in Refs. [17–19,22]), and Mg–H2O mixtures [46]. Apparently, the glowing is associated with ammonia borane or its products. The rates of processes that involve AB are probably lower than the burn rate of AlMg particles with H2O. Note that multi-zone reaction waves have been observed in many combustion systems, particularly, in self-propagating high-temperature synthesis (SHS) [47]. Figure 9 shows the combustion front velocities for mixtures 2, 3, and 4. Each point was obtained based on the results of three tests. It is seen that mixture 2, with the highest concentration of AB, has the lowest velocity. Also, comparison with the data for AlMg–water mixtures (Fig. 2) shows that the addition of AB significantly decreases the front velocity. This is not unexpected because the heat release values of AB thermolysis and hydrolysis are much smaller than the heat release of AlMg reaction with water. Further, mixture 4 exhibits a lower front velocity than mixture 3. This is apparently explained by a lower combustion temperature in mixture 4 due to the larger amount of water there. Figure 10 shows XRD patterns of solid products obtained after combustion of mixtures 2, 3, and 4. All three patterns indicate the presence of MgAl2O4 and MgO, produced by the reaction between AlMg particles and water. Also, in each pattern, a compound that contains a borate anion is detected: ammonium pentaborate tetrahydrate NH4B5O84H2O for mixture 2, magnesium orthoborate Mg3(BO3)2 for mixture 3, and magnesium metaborate Mg(BO2)2 for mixture 4. Note that, since heavy water was used in these experiments, the actual formula of detected ammonium pentaborate tetrahydrate probably contains D2O instead of H2O. The presence of a borate anion in each pattern is explained by AB hydrolysis (see, for example, Eq. (4) where formation of (BO2) ion is shown). A type of cation bound with this anion depends on the mixture composition: it is ammonium for mixture 2 and magnesium for mixtures 3 and 4. The presence of ammonium cation in the products of mixture 2 is apparently associated with the higher concentration of AB in this mixture. The absence of BN in any pattern indicates that the third step of AB decomposition (Eq. (3)) was not achieved. The condensed

1503

D.A. Rodriguez et al. / Combustion and Flame 162 (2015) 1498–1506 Table 1 Tested mixture compositions. No.

Composition in mass fractions

1 2 3 4

Composition in mole fractions

H3NBH3 (wt%)

D2O (wt%)

AlMg (wt%)

H3NBH3 (wt%)

D2O (wt%)

AlMg (wt%)

33 25 12.5 12.5

32.96 36.9 43.05 49.4

34.04 38.1 44.45 38.1

26.4 20 9.4 9.2

40.7 44 50 56.5

32.9 36 40.6 34.3

0

1s

2s

3s

4s

5s

6s

7s

8s

9 s

10 s

11 s

12 s

13 s

14 s

15 s

Fig. 8. Combustion of mixture 3. Time zero was selected arbitrarily.

Fig. 9. Combustion front velocity for each mixture.

products of the first and second stages of AB thermolysis (Eqs. (1) and (2)) are amorphous according to the literature [1–6] and hence they cannot be seen in XRD patterns. The concentrations of H2, HD, and D2 in the gaseous products of each mixture were determined by mass-spectroscopic analysis. Figure 11 shows the measured amounts of H2, HD, and D2 per unit mass of the sample. Note that each sample included a booster pellet (see Section 2.1), which slightly decreased the difference between the results obtained for different mixtures. Each value in Fig. 11 was obtained based on three tests. It is seen that each mixture generates comparable concentrations of the three gases. This clearly indicates that all three possible pathways for hydrogen release take place: thermolysis of H3NBH3 (produces H2), reaction between H3NBH3 and D2O (produces HD), and reaction between AlMg and D2O (produces D2). Mixture 2 produced twice more H2 than mixtures 3 and 4. This corresponds to the twofold content of AB in mixture 2. Despite the lower content of AB, however, mixture 3 produced as much HD as mixture 2. This indicates that hydrolysis increased the efficiency of using AB in this mixture. The largest

amount of D2 was obtained from mixture 4. This is understandable because in mixtures 2 and 3 there was not enough D2O for AlMg combustion because part of D2O was consumed by AB hydrolysis, while mixture 4 had a twofold amount of D2O. On the other hand, the amount of HD for mixture 4 was less than for mixture 3. This is apparently explained by the lower hydrolysis reaction rate due to the lower combustion temperature in mixture 4. For better clarity, based on the measured amounts of H2 and HD, a plot was generated (Fig. 12) that shows, for each mixture, the fraction of the total available H that was detected in gas phase. The total available amount of H in the calculations was determined for each sample based on the actual mass of H3NBH3 in the sample. In addition, the same figure shows what percentage of the produced H came from thermal decomposition of AB and how much came from AB hydrolysis as ‘‘H’’ part of the formed HD. It is seen that for the mixtures 2 and 4 only 65–66% of H, available in AB, was released to gas phase and detected as either H2 or part of HD, while for mixture 3, the released fraction of H increased to 88%. The rest of hydrogen apparently remained in the condensed products as part of polymeric boron-hydrogen compounds and ammonium cation (see Fig. 10 and discussion of XRD results). It is also clearly seen in Fig. 12 that the increase in H yield for mixture 3 was caused by the increased role of AB hydrolysis. Similarly, based on the measured amounts of HD and D2, a plot was generated (Fig. 13) that shows, for each mixture, the fraction of the total available deuterium that was detected in gas phase as either D2 or part of HD. The total available amount of D in the calculations was determined for each sample based on the actual mass of D2O in the sample. In addition, the same figure shows what percentage of the released deuterium came from the reaction of AlMg with water and how much came from AB hydrolysis. It is seen that the maximum fraction of deuterium released to gas phase was only 66% (mixture 2). This is apparently associated with

1504


Fig. 11. Measured amounts of (a) H2, (b) HD, and (c) D2 per unit mass of the sample.

Fig. 10. XRD pattern of the products obtained after combustion of mixtures (a) 2, (b) 3, and (c) 4.

the loss of D2O during ignition and combustion. Indeed, part of vaporized D2O could escape from the sample and then condense in the chamber. The efficiency of releasing D in mixtures 2 and 3 is about the same despite the larger amount of D2O in mixture 3. It is interesting that in mixture 4, the attempt to compensate the loss of heavy water and increase the amount of D2O available for AB hydrolysis resulted in a lower efficiency of D release because of the lower yield from AB hydrolysis, apparently, explained by the lower combustion temperature. To determine the maximum hydrogen yield that could be achieved for the tested compositions if H2O were used instead of D2O, the measured masses of released H2, HD, and D2 as well as the sample masses were recalculated for H as the only isotope of hydrogen. Figure 14 shows the resulting total hydrogen yield

Fig. 12. Released fraction of total available H via AB thermolysis and AB hydrolysis.

(hydrogen-to-mixture mass ratio) for each composition. Note that the recalculation procedure accounted for the fact that each sample included a booster pellet in addition to the main mixture. The maximum theoretical yield for each sample (based on the assumption that all hydrogen is released) is also shown in this figure. Here each sample also included both the main mixture and booster pellet. Standard deviations of the theoretical values were caused by the actual variation of the booster pellet mass fraction


1505

4. Conclusions

Fig. 13. Released fraction of total available D via reaction of AlMg with D2O and AB hydrolysis.

A novel approach to hydrogen release from ammonia borane has been tested that involves the reaction of mechanically alloyed AlMg powder with water as a source of heat for AB thermolysis and hydrolysis. This reaction also releases hydrogen from water, thus increasing the total hydrogen yield. Experiments have shown that mixtures of mechanically alloyed AlMg powder with gelled water are combustible. The velocities of combustion front propagation exceed the values obtained for mixtures of nanoscale Al powder with gelled water. At the same time, no reaction occurs between mechanically alloyed AlMg powder and hot (80 °C) water for 24 h, which indicates that the mixtures can remain stable for long time. Experiments have been conducted with mixtures of AB, mechanically alloyed AlMg powder, and heavy water (D2O), where the latter was used for investigating the reaction mechanisms through mass-spectroscopy of released H2, HD, and D2 gases (isotopic tests). The addition of ammonia borane to the AlMg–water mixture increased the total hydrogen yield. The isotopic tests have shown that AB participates in two parallel processes – thermolysis and hydrolysis. Because of this, as much as 88% of hydrogen contained in AB was released in one of the tested mixtures, which significantly exceeds the amount released in the first and second steps of AB thermolysis (35–70%). Tuning the composition and scaling up to a practical hydrogen-generating reactor may further increase hydrogen yield in these mixtures. Acknowledgment

Fig. 14. Theoretical gravimetric hydrogen yield of AB–H2O–AlMg mixtures and its experimental values obtained by recalculating the data obtained in experiments with AB–D2O–AlMg mixtures.

in experiments. It is seen that mixture 2 provides the highest hydrogen yield (both experimentally and theoretically), while mixture 3 is the most efficient from the standpoint of approaching the theoretical limit in the experiments. Comparison of hydrogen yield for mixtures 2 and 3 with that for the mixtures of mechanically alloyed AlMg powder with water (Fig. 5) shows that the addition of AB increases hydrogen production. Tuning the composition and scaling up to a practical hydrogen-generating reactor may further increase hydrogen yield. Concentration of AB could be further increased to provide more hydrogen from AB while keeping a sufficiently high exothermicity of the mixture. A larger diameter of the reactor would decrease the heat losses, leading to a higher combustion temperature and hence to a possibility of adding more AB. Also, an increase in the reactor height may decrease the loss of water during ignition and combustion of the mixture. Finally, for a larger reactor, the mass fraction of the booster pellet would be smaller, leading to a higher hydrogen yield. Note that compositions 2 and 3 were designed assuming that no hydrolysis of AB occurs and all water is consumed by the reaction with AlMg. In reality, AB hydrolysis occurs and consumes some water, shifting the water-metal ratio from the stoichiometry to having more water in the mixture. For the same amount of AB, the production of hydrogen may, therefore, increase. On the other hand, the shift from stoichiometry reduces the combustion temperature, leading to a lower rate of AB hydrolysis and a lower hydrogen yield (see the result for mixture 4). Thus, there should be an optimal concentration where there is enough water for the reactions with both AlMg and AB, while the combustion temperature remains sufficiently high. Because this optimization is affected by the reaction temperature, it is also affected by specific heat transfer conditions and thus by the sample mass and shape.

This research was supported by the U.S. Department of Defense (Grants Nos. W911NF-12-1-0056 and W911NF-14-1-0034; Grant Officer’s Representative: Dr. Ralph A. Anthenien of the Army Research Office; Co-GOR: Dr. Clifford D. Bedford of the Office of Naval Research). Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.combustflame. 2014.11.019. References [1] F.H. Stephens, V. Pons, R.T. Baker, Dalton Trans. (2007) 2613–2626. [2] A. Staubitz, A.P.M. Robertson, I. Manners, Chem. Rev. 110 (2010) 4079–4124. [3] G. Wolf, J. Baumann, F. Baitalow, F.P. Hoffmann, Thermochim. Acta 434 (2000) 19–25. [4] F. Baitalow, J. Baumann, G. Wolf, K. Jaenicke-Rößler, G. Leitner, Thermochim. Acta 391 (2002) 159–168. [5] J. Baumann, E. Baitalow, G. Wolf, Thermochim. Acta 430 (1–2) (2005) 9–14. [6] M.R. Weismiller, S.Q. Wang, A. Chowdhury, S.T. Thynell, R.A. Yetter, Thermochim. Acta 551 (2013) 110–117. [7] S. Frueh, R. Kellett, C. Mallery, T. Molter, W.S. Willis, C. King’ondu, S.L. Suib, Inorg. Chem. 50 (2011) 783–792. [8] M. Chandra, Q. Xu, J. Power Sources 156 (2006) 190–194. [9] M. Chandra, Q. Xu, J. Power Sources 159 (2006) 855–860. [10] M. Chandra, Q. Xu, J. Power Sources 168 (2007) 135–142. [11] F. Cheng, H. Ma, Y. Li, J. Chen, Inorg. Chem. 46 (2007) 788–794. [12] M. Diwan, V. Diakov, E. Shafirovich, A. Varma, Int. J. Hydrogen Energy 33 (2008) 1135–1141. [13] V.G. Ivanov, S.N. Leonov, G.L. Savinov, O.V. Gavrilyuk, O.V. Glazkov, Combust. Explos. Shock Waves 30 (1994) 569–570. [14] V.G. Ivanov, O.V. Gavrilyuk, O.V. Glazkov, M.N. Safronov, Combust. Explos. Shock Waves 36 (2000) 213–219. [15] E. Shafirovich, V. Diakov, A. Varma, Combust. Flame 144 (2006) 415–418. [16] E. Shafirovich, V. Diakov, A. Varma, Int. J. Hydrogen Energy 32 (2007) 207–211. [17] G.A. Risha, S.F. Son, R.A. Yetter, V. Yang, B.C. Tappan, Proc. Combust. Inst. 31 (2007) 2029–2036. [18] G.A. Risha, J.L. Sabourin, V. Yang, R.A. Yetter, S.F. Son, B.C. Tappan, Combust. Sci. Technol. 180 (2008) 2127–2142.

1506


[19] J.L. Sabourin, G.A. Risha, R.A. Yetter, S.F. Son, B.C. Tappan, Combust. Flame 154 (2008) 587–600. [20] T.L. Connell Jr., G.A. Risha, R.A. Yetter, G.A. Young, D.S. Sundaram, V. Yang, Proc. Combust. Inst. 33 (2011) 1957–1965. [21] Y. Sun, B. Zhu, H. Dang, H. Sun, J. Mater. Sci. 46 (2011) 4471–4478. [22] Y. Sun, B. Zhu, Ind. Eng. Chem. Res. 50 (2011) 14136–14141. [23] T.L. Connell Jr., G.A. Risha, R.A. Yetter, V. Yang, S.F. Son, Int. J. Energ. Mater. Chem. Propul. 11 (2012) 259–273. [24] J. Wan, S.Z. Cai, Y. Liu, C.S. Xie, X.P. Xia, D.W. Zeng, Mater. Chem. Phys. 136 (2012) 466–471. [25] D.S. Sundaram, V. Yang, T.L. Connell Jr., G.A. Risha, R.A. Yetter, Proc. Combust. Inst. 34 (2013) 2221–2228. [26] D.E. Kittell, L.J. Groven, T.R. Sippel, T.L. Pourpoint, S.F. Son, Combust. Sci. Technol. 185 (2013) 817–834. [27] T.R. Sippel, T.L. Pourpoint, S.F. Son, Propell. Explos. Pyrot. 38 (2013) 56–66. [28] T.X. Phuoc, R.H. Chen, Appl. Energy 101 (2013) 567–571. [29] D.S. Sundaram, V. Yang, Combust. Flame 161 (2014) 2215–2217. [30] V. Rosenband, A. Gany, Int. J. Hydrogen Energy 35 (2010) 10898–10904. [31] M.-C. Lin, J.-Y. Uan, T.-C. Tsai, Int. J. Hydrogen Energy 37 (2012) 13731–13736. [32] X. Chen, Z. Zhao, M. Hao, D. Wang, Int. J. Energy Res. 37 (2013) 1624–1634. [33] O.V. Kravchenko, K.N. Semenenko, B.M. Bulychev, K.B. Kalmykov, J. Alloy. Compd. 397 (2005) 58–62. [34] A.V. Parmuzina, O.V. Kravchenko, Int. J. Hydrogen Energy 33 (2008) 3073–3076.

[35] A.V. Ilyukhina, O.V. Kravchenko, B.M. Bulychev, E.I. Shkolnikov, Int. J. Hydrogen Energy 35 (2010) 1905–1910. [36] J.T. Ziebarth, J.M. Woodall, R.A. Kramer, G. Choi, Int. J. Hydrogen Energy 36 (2011) 5271–5279. [37] M.A. Machado, D.A. Rodriguez, Y. Aly, M. Schoenitz, E.L. Dreizin, E. Shafirovich, Combust. Flame 161 (2014) 2708–2716. [38] Y.L. Shoshin, R.S. Mudryy, E.L. Dreizin, Combust. Flame 128 (2002) 259–269. [39] Y.L. Shoshin, E.L. Dreizin, Combust. Flame 145 (2006) 714–722. [40] Y. Aly, M. Schoenitz, E.L. Dreizin, Combust. Flame 160 (2013) 835–842. [41] Y. Aly, V.K. Hoffman, M. Schoenitz, E.L. Dreizin, J. Propul. Power 30 (2014) 96–104. [42] O.V. Kravchenko, L.G. Sevastyanova, S.A. Urvanov, B.M. Bulychev, Int. J. Hydrogen Energy 39 (2014) 5522–5527. [43] A.K. Narayana Swamy, E. Shafirovich, Combust. Flame 161 (2014) 322–331. [44] A.A. Shiryaev, Int. J. SHS 4 (1995) 351–362. [45] A.T. Raissi, Technoeconomic analysis of area II hydrogen production – Part II: hydrogen from ammonia and ammonia–borane complex for fuel cell applications, in: Proceedings of the 2002 US DOE Hydrogen Program Review, NREL/CP-610-32405, . [46] M. Diwan, D. Hanna, E. Shafirovich, A. Varma, Chem. Eng. Sci. 65 (2010) 80–87. [47] A.G. Merzhanov, B.I. Khaikin, Prog. Energy Combust. Sci. 14 (1988) 1–98.