17234 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 This journal is c the Owner Societies 2011

Cite this: Phys. Chem. Chem. Phys., 2011, 13, 17234–17241

Formation of Al2H7� anions — indirect evidence of volatile AlH3 on

sodium alanate using solid-state NMR spectroscopywzMichael Felderhoff* and Bodo Zibrowius*

Received 9th June 2011, Accepted 10th August 2011

DOI: 10.1039/c1cp21877h

After more than a decade of intense research on NaAlH4 doped with transition metals as

hydrogen storage material, the actual mechanism of the decomposition and rehydrogenation

reaction is still unclear. Early on, monomeric AlH3 was named as a possible transport shuttle for

aluminium, but never observed experimentally. Here we report for the first time the trapping of

volatile AlH3 produced during the decomposition of undoped NaAlH4 by an adduct of sodium

alanate and crown ether. The resulting Al2H7� anion was identified by solid-state 27Al NMR

spectroscopy. Based on this indirect evidence of volatile alane, we present a simple description of

the processes occurring during the reversible dehydrogenation of NaAlH4.

Introduction

Complex aluminium hydrides have been regarded as potential

hydrogen storage materials for many years,1 but it was an experi-

mental finding by Bogdanovic and Schwickardi2 in 1997 that

marked a milestone on the way to a technical application of this

class of materials. They observed that doping NaAlH4 with

catalytic amounts of titanium compounds renders dehydrogenation

and rehydrogenation sufficiently fast at reasonable temperatures. It

is generally accepted that the reversible decomposition of NaAlH4

after doping with transition metals proceeds via the same two steps

that are found for the decomposition of the pristine material:3,4

3NaAlH4 ! Na3AlH6 + 2Al + 3H2 (1)

Na3AlH6 + 2Al ! 3NaH + 3Al + 32H2 (2)

Although we are not aware of any convincing evidence of an

incorporation of Ti or other transition metals into the NaAlH4

framework, explanations of their catalytic effect based on such

incorporation have become rather popular, in particular among

theoreticians.5–7 A very recent TEM and high-resolution X-ray

diffraction study confirms that even after ball-milling and repeated

dehydrogenation/rehydrogenation cycles Ti is located in

Al1�xTix nanoparticles on the surface of NaAlH4 grains.8

Despite more than a decade of intense research both in

academia and industry, the actual mechanism of the (reversible)

dehydrogenation of NaAlH4 is also not sufficiently well

understood.9 In particular, the long-range transport process

that converts the perfect mixture of aluminium and sodium in

NaAlH4 into well separated NaH and aluminium particles10–12

during dehydrogenation and vice versa during rehydrogenation

is a matter of controversy. The elemental segregation during

dehydrogenation is indeed remarkable, since significant

amounts of titanium are only found in the aluminium rich

particles.10–12 Early on, monomeric aluminium hydride or

alane (AlH3) was named as a possible transport species.13

Later, Walters and Scogin14 strongly advocated in favour of

AlH3 as the transport species and proposed a complete

mechanism for the reversible dehydrogenation of NaAlH4

based on this hypothesis. However, all attempts to detect this

volatile species in experiments have failed so far. Strong

support for volatile alane as the transport species comes from

a combined experimental and theoretical study on the role of

titanium during rehydrogenation.12 Summarizing their

experimental results, the authors state that the effect of

titanium can be limited to activating aluminium for hydrogen

chemisorption and catalyzing the formation of aluminium

hydride. Molecular-dynamic calculations show that the AlH3

so formed is stable and mobile on an aluminium surface.

On the other hand, there seems to be some experimental

support for vacancy-mediated transport processes which are

discussed by a growing number of authors.15–18 The results of

hydrogen–deuterium exchange experiments using mass

spectrometry and Raman spectroscopy corroborate the view that

NaAlH4 decomposes at NaAlH4/Al interfaces via the formation of

AlH3 vacancies which diffuse to NaAlH4/Na3AlH6 interfaces.9

This interpretation would be in line with first-principle calculations

that suggest that the dehydrogenation rate of Ti-doped NaAlH4

is limited by the diffusion rate of AlH3 vacancies.17

Solid-state NMR spectroscopy cannot compete with

Raman spectroscopy in sensitivity and time resolution, but it

Max-Planck-Institut fur Kohlenforschung, Kaiser-Wilhelm-Platz 1,45470 Mulheim an der Ruhr, Germany.E-mail: felderhoff@mpi-muelheim.mpg.de,zibrowius@mpi-muelheim.mpg.dew In memoriam Borislav Bogdanovic and Manfred Schwickardi.z Electronic supplementary information (ESI) available. See DOI:10.1039/c1cp21877h

PCCP Dynamic Article Links

www.rsc.org/pccp PAPER

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online / Journal Homepage / Table of Contents for this issue

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 17235

has proven to be a rather useful tool for studying complex

aluminium hydrides.19,20 In particular, the application of magic-

angle spinning (MAS) at moderate magnetic fields allows subtle

structural details21 and the fate of various nuclei during

dehydrogenation and rehydrogenation to be studied.22–24

Recently, Conradi and co-workers25 reported the discovery

of a mobile Al-bearing species formed in NaAlH4 at elevated

temperature and pressure that can be retained under ambient

conditions. Because of the position of a relatively narrow 27Al

NMR line at 105 ppm, the new species was named S105 and

‘‘identified as highly defective NaAlH4 likely having a large

AlH3 vacancy concentration’’. A solid-state 27Al NMR line

with a similar position and width was reported by Simagina

et al.26 for NaAl2H7 stabilized with the crown ether 18-crown-

6 (18C6) almost two decades ago. This coincidence prompted

us to investigate the possibility of trapping volatile AlH3 by

means of crown-ether complexes.

In the present paper we demonstrate that Al2H7� anions are

indeed generated during the thermal decomposition of

NaAlH4 in the presence of 18C6. We include some results of

the preliminary characterisation of Na(18C6)Al2H7 by solid-state27Al NMR spectroscopy. Based on our experimental results, we

outline a simple mechanism that can explain some of the

ill-understood peculiarities of the (reversible) dehydrogenation

of NaAlH4 in the presence of a catalyst. In this mechanism,

which is in fact not entirely new, but is related to the ones

proposed by Walters and Scogin14 and Chaudhuri et al.,12

volatile alane plays the central role as the transport species.

Furthermore, we attribute the unexpectedly high thermal

stability of pristine sodium alanate to the trapping of released

AlH3 molecules by neighbouring AlH4� units leading to a

disproportionation equilibrium on the surface.

Experimental

Sample preparation and handling

NaAlH4 (Chemetall, 82–85%) was purified by dissolving it in

THF and filtrating off the insoluble portions. The pure NaAlH4

was precipitated from the solution by addition of pentane and

carefully dried under vacuum. 18-Crown-6 (18C6, 99%) was

purchased from Aldrich and used without further purification.

Na(18C6)Al2H7 was prepared from stoichiometric amounts of

NaAlH4 and 18C6 by adding a twofold amount of freshly prepared

AlH3 in ether solution according to a literature procedure.27

For the synthesis of Na(18C6)AlH4 0.27 g (5 mmol)

NaAlH4 were suspended in toluene and a solution of 2.64 g

(10 mmol) 18C6 in toluene was added at 0 1C. The mixture was

stirred for 24 h at 0 1C, filtered and dried in vacuo.

For ball milling a Fritsch Pulverisette P7 was used. In a

typical experiment 0.27 g (5 mmol) NaAlH4 and 0.13 g

(0.5 mmol) 18C6 were ball milled for 5 min at 200 rpm in a

12 mL stainless steel vessel with five balls of 4 g each.

Cryomilling was performed with a Retsch Cryomill at 77 K

using a stainless steel vial and two balls of 14 g each.

All syntheses and operations were performed under argon

using dried and oxygen-free solvents. The MAS rotors were

filled and capped in a glove box and transferred to the

spectrometer in argon-filled vials.

NMR spectroscopy

The solid-state 27Al NMR spectra were recorded on a Bruker

Avance 500WB spectrometer at a resonance frequency of

130.3 MHz using a double-bearing MAS probe (DVT BL4).

The chemical shift was referenced relative to an external 1.0 M

aqueous solution of aluminium nitrate. The same solution was

also used for determining the flip-angle.

For the MAS NMR spectra, single p/12 pulses (tp = 0.6 ms,nrf E 70 kHz) under high-power proton decoupling (cw, nrf,H =

50–70 kHz) were usually applied at a repetition time of 1 s

(800 scans) and a spinning frequency (nMAS) of 13 kHz.

Deviations from these standard conditions are indicated in the

captions to the figures.

For the measurements of the ball-milled samples, special

zirconia rotors with a reduced volume (ca. 50 ml, 2 mm) and

zirconia rotor caps with a small centric bore (0.3 mm) were

used.28 Variable-temperature measurements were carried out

in the range between 295 K and 360 K. A heating rate of

5 K min�1 was applied. The measurements were started at the

earliest 10 min after a constant reading (�0.5 K) of the

thermocouple had been reached.

The temperature was calibrated using the temperature

dependence of the 207Pb resonance line of Pb(NO3)2.29 This

calibration had to be performed for all spinning speeds used

because different spinning speeds require different nitrogen

flows and the influence of frictional heating has to be taken

into account.29,30 Without additional cooling, i.e. when all gas

flows are applied at room temperature (295 K), the lowest

accessible temperature at a spinning rate of 13 kHz is 311 K.

Even though the above-mentioned rotor with a reduced

volume was used, a significant temperature gradient over the

length of the sample (12 mm) was observed. From the width of

the low-frequency tailing of the 207Pb NMR line of Pb(NO3)2we estimate that at a spinning speed of 13 kHz and a nominal

temperature of 413 K some parts of the sample are as much as

22 K cooler. The temperatures given always relate to the

hottest parts of the sample (accuracy: better than 2 K and

5 K at temperatures close to room temperature and close to

400 K, respectively).

Experimental results and discussion

Fig. 1 displays 27Al MAS NMR spectra of NaAlH4 mixed

with crown ether 18C6 (molar ratio: 10 : 1) after ball-milling

for just 5 min at low speed (200 rpm). The spectrum of the

as-prepared material measured at 311 K exhibits three

resonance lines. The broad line centred at about 94 ppm stems

from NaAlH4.19,21,23 Its asymmetric shape and width (full

width at half height (FWHH): 750 Hz) can be easily explained

by second-order quadrupolar and dipole–dipole interaction

(vide infra). We assign the more intense of the two narrow lines

in this spectrum, the line at 102.3 ppm, to Na(18C6)AlH4. This

line is extremely narrow (FWHH: 85 Hz) for a solid-state NMR

line of a quadrupolar nucleus, indicating that the aluminium

atoms in the tetrahedral AlH4 units are experiencing much

smaller quadrupolar and dipolar couplings than in sodium

alanate. The spectrum measured for Na(18C6)AlH4 prepared

from a suspension of NaAlH4 in a toluenic solution of 18C6

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

17236 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 This journal is c the Owner Societies 2011

(Fig. 1g) reveals that the conversion of NaAlH4 is incomplete.

However, a comparison with the spectrum in Fig. 1a shows a

perfect match of the lines discussed so far. It should be noted

that the exact position and width of the resonance line of

Na(18C6)AlH4 depend on the temperature. When spinning the

sample at only 3 kHz (T = 296 K), a line at 102.6 ppm

(FWHH: 40 Hz) is observed.

The third line in the top spectrum shown in Fig. 1, the line at

101.1 ppm, is slightly wider (FWHH: 120 Hz) than the one

assigned to Na(18C6)AlH4. Its intensity varied from being

hardly detectable to being larger than that of the line obtained

for Na(18C6)AlH4 for as-prepared materials from different

runs of this experiment. We assign this line to Na(18C6)Al2H7

formed according to eqn (3):

Na(18C6)AlH4 + AlH3 - Na(18C6)Al2H7 (3)

We propose that the decomposition of part of the NaAlH4 into

NaH and AlH3 is the source of the monomeric alane. It seems

unlikely that the crown ether initiates the decomposition of

NaAlH4. Rather, we suppose that a certain amount of volatile

alane is always present, at least on the surface of any NaAlH4

particle. The function of the crown ether is that the sodium

complex with the crown ether stabilizes the Al2H7� anion. The

mechanistic issues will be discussed in more detail later.

As Fig. 1 shows, a moderate increase in temperature leads to

a conversion of Na(18C6)AlH4 into Na(18C6)Al2H7. The

conversion appears to be complete after 100 min at 338 K

(Fig. 1c). Interestingly, in an earlier run of this experiment we

observed the lines of both crown-ether complexes after cooling

down the sample at this point of the experiment. This finding is

due to the slightly different influence of temperature on the line

positions of the two crown-ether adducts. Although both lines

show a low-frequency shift with increasing temperature, the

shift is more pronounced for Na(18C6)AlH4 than for

Na(18C6)Al2H7. At about 340 K the lines almost merge,

making the detection of small quantities of residual

Na(18C6)AlH4 very difficult. We therefore heated the sample

to 343 K and kept it at this temperature for further 20 min

(Fig. 1d) before leaving it to cool down to room temperature

overnight.

The position and width of the resonance line at 101.1 ppm

(Fig. 1e) fit extremely well with those of the line obtained for

Na(18C6)Al2H7 (cf. Fig. 1f) prepared in a solution process

according to the literature. For a 0.5 M solution of

Na(18C6)Al2H7 in tetrahydrofuran a chemical shift of

98 ppm was observed.27 The only previous paper26 on

solid-state NMR studies of this compound that we are aware

of reports a 27Al MAS NMR line at 95 ppm (FWHH: about

300 Hz, nMAS = 2 kHz, B0 = 7.05 T). Because of the

narrowness of the line the discrepancy in its position cannot

be explained by the different magnetic fields used, i.e. by the

different quadrupole-induced shifts.31 Inspection of the

corresponding figure in ref. 26 suggests that the position of this

line is at about 100 ppm rather than 95 ppm. Unfortunately,

without additional information we are not able to explain this

difference satisfactorily.

Fig. 1 27Al MAS NMR spectra of a 10 : 1 mixture of NaAlH4 and

18C6 after ball-milling recorded (a) at 311 K, (b) after 10 min at 338 K,

(c) after 100 min at 338 K, (d) after additional 10 min at 343 K, and

(e) after overnight cooling to 311 K. For comparison, 27Al MAS NMR

spectra of (f) Na(18C6)Al2H7 prepared according to the literature and

(g) Na(18C6)AlH4 are included.

Fig. 2 Spinning sideband manifolds of the satellite transitions in the27Al MAS NMR spectra of (a) a 10 : 1 mixture of NaAlH4 and 18C6

after ball-milling and subsequent heating to 343 K (the same spectrum

as in Fig. 1e) and (b) Na(18C6)Al2H7 prepared according to the

literature (T = 311 K). The asterisks indicate the positions of the

maxima of the inner satellite transition of NaAlH4. The centre bands

of the central transitions are cut off at about 10% of their maximum

intensities.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 17237

However, the assignment of the resonance line at 101.1 ppm

that is observed for the thermally treated NaAlH4/18C6

mixture (Fig. 1e) to Na(18C6)Al2H7 is not solely based on

its position. Fig. 2 demonstrates that the pattern of the

spinning sidebands caused by the satellite transitions31

matches that observed for Na(18C6)Al2H7 prepared according

to the literature.27 It should be noted that the outer parts of the

spectrum shown in Fig. 2a are dominated by the characteristic

pattern of the inner satellite transition of NaAlH4. The

separation of the maxima of this transition (465 kHz,

determined at nMAS = 3 kHz and marked in Fig. 2) yields a

good estimate of the quadrupole coupling constant of

Cq = 3.1 MHz, a value close to the one given by Zhang

et al.21 (3.15 MHz). From the envelope of the sideband pattern

of Na(18C6)Al2H7 it follows that the quadrupole coupling is

much weaker than in NaAlH4 and far from being axially

symmetric. The nonaxiality of the electric field gradient at

the Al sites in Na(18C6)Al2H7 is not surprising.

A single 27Al resonance line is observed for Na(18C6)Al2H7.

Since there are two aluminium nuclei in the Al2H7� anion,

they are probably equivalent. Simagina et al.26 have proposed

an H3Al–H–AlH3 complex, i.e. two AlH3 units bridged by an

additional hydrogen atom that undergoes fast intramolecular

rearrangements, resulting in equivalence of all Al–H bonds in

the AlH4 tetrahedra. The equivalence of the two aluminium

nuclei would imply that the corresponding Na+ cation lies

side-on to this complex in the mirror plane between the two

aluminium nuclei. The small quadrupole coupling points to a

rather symmetric environment and/or some motional averaging.

Ab initio studies of the structure and stability of the Al2H7�

anion32 support this view. In particular, they indicate that the

Li+ cation in LiAl2H7 is located above the Al–H–Al bridge.

Our own investigations concerning the structure of

Na(18C6)Al2H7 are underway.

Apart from the resonance lines shown in Fig. 1, all spectra

measured for thermally treated NaAlH4/18C6 mixtures exhibit an

additional resonance line at �42.5 ppm, which is characteristic of

the hexahydride Na3AlH6.21,23 The increase in intensity of the

Na3AlH6 line during the thermal treatment parallels that of

the line assigned to Na(18C6)Al2H7 (cf. Fig. S1, ESIz). On the

other hand, the spectrum in Fig. 2a shows that even after the

thermal treatment there is no metallic aluminium detectable.

The expected position of the metal (ca. 1639 ppm) is right in

the middle between the 15th and the 16th high-frequency

spinning sideband.

It should be noted that the spectra shown in Fig. 1 are not

quantitative. To avoid saturation effects for pure NaAlH4,

relaxation delays of more than 30 s are necessary, even for

flip-angles as small as p/12. Furthermore, while the proton

decoupling power applied for most of the measurements

(nrf,H E 50 kHz) is sufficient for the two crown-ether

complexes, it is insufficient for NaAlH4. Fig. 3 demonstrates

both effects. With stronger decoupling the characteristic

second-order splitting of the central transition31 is observed

for NaAlH4. By using the value given above for the quadrupole

coupling constant Cq and the axial symmetry of the field

gradient tensor (Z = 0), fitting of the lineshape33 allows the

isotropic 27Al chemical shift of NaAlH4 to be determined with

rather high accuracy. In accordance with earlier studies21 we

obtain diso(NaAlH4) = (97.4 � 0.2) ppm at T = 311 K. Since

the lower decoupling power and the shorter relaxation delays

did not interfere with the resonance lines of the crown-ether

complexes, we used these experimental conditions to avoid

additional heating and to reduce the measuring time.

To gather more evidence for the presence of volatile AlH3,

we added freshly prepared polymeric AlH3 to a mixture of

NaAlH4 and 18C6. Because of the low thermal stability of

polymeric AlH3 the final mixture (composition: 10 : 1 : 1) was

milled at 77 K for 30 min. Fig. 4 shows the 27Al MAS NMR

spectra recorded immediately after the cryomilling. Similar to

the experimental result obtained for NaAlH4 ball-milled with

18C6 at room temperature, Na(18C6)AlH4 (line at 102.4 ppm

in Fig. 4a) is formed and a large percentage of it is converted

to Na(18C6)Al2H7 (line at 101.2 ppm), most likely during the

time in which the sample was filled into the rotor and

transferred into the probe. Frictional heating caused by

spinning the sample at frequencies of up to 13 kHz is sufficient

to convert all remaining Na(18C6)AlH4 into Na(18C6)Al2H7

within about 1 h. The interpretation that the polymeric AlH3 is

the source of the volatile alane is corroborated by the fact that

no Na3AlH6 is detected in this experiment. Although the

observed decrease in the intensity of the broad line of

polymeric AlH3 at 6 ppm (cf. Fig. S2, ESIz) supports this

view, this is only a weak argument, since the line of the

metallic aluminium at ca. 1639 ppm (cf. Fig. S2, ESIz)increases in the course of the measurement. Furthermore, a

quantitative evaluation of the rather broad resonance line of

AlH3 is hampered by the superposition with the first-order

spinning sidebands of the lines shown in Fig. 4.

Attempts to produce detectable amounts of NaAl2H7 without

18C6 by ball-milling NaAlH4 and AlH3 at 77 K failed. This

result is not surprising, but rather in line with earlier observations

that stable Al2H7� anions are only formed from AlH4

� anions

under conditions in which the interaction between the charge-

compensating metal cation and the anion is weakened.27,34

Fig. 3 Influence of repetition time (D1) and decoupling power (nrf,H)on the relative intensities and lineshapes of the 27Al MAS NMR

spectra of a 10 : 1 mixture of NaAlH4 and 18C6 after ball-milling:

(a) D1 = 16 s, nrf,H E 100 kHz; (b) D1 = 16 s, nrf,H E 70 kHz; and

(c) D1 = 1 s, nrf,H E 70 kHz. All spectra (400 scans) were recorded at

nMAS = 13 kHz and T = 311 K.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

17238 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 This journal is c the Owner Societies 2011

Proposed mechanism

Pristine NaAlH4

We regard the formation of Al2H7� anions from AlH4

� anions

stabilized by a sodium crown-ether complex (eqn (3)) as strong

evidence for the presence of volatile AlH3 on solid sodium

alanate. We therefore propose that, at least on the surface of

the alanate particles, the following chemical equilibrium exists:

NaAlH4 ! NaH + AlH3 (4)

A direct decomposition of NaAlH4 into NaH instead of a

transformation according to eqn (1) has already been suggested

on the basis of a correlation between the temperature of thermal

decomposition of alkali alanates and the corresponding standard

enthalpy of decomposition.35 At room temperature the

decomposition level is very low, i.e. the equilibrium in eqn (4)

lies far to the left. The AlH3 formed is much too reactive to

escape from the surface. It is trapped by a neighbouring AlH4�

anion, forming an Al2H7� anion, in a similar way to what we

observed in the presence of the crown ether 18C6:

NaAlH4 + AlH3 ! NaAl2H7 (5)

In contrast to the situation occurring with the rather bulky

sodium crown-ether complex, the Al2H7� anion is not stable

when the charge is compensated by the small Na+ cation.34

Hence, eqn (5) describes a dynamic equilibrium, i.e. a very

effective diffusion mechanism for AlH3, at least on the surface

of the sodium alanate particles. By combining eqn (4) and (5)

one obtains a disproportionation equilibrium:

2NaAlH4 ! NaH + NaAl2H7 (6)

Again, the equilibrium lies to the left. Since only a surface

layer may be involved the overall concentration of both NaH

and NaAl2H7 is far too low to be detected by conventional

MAS NMR spectroscopy. The situation might be different in

the case of more elaborate NMR techniques specifically

dedicated to the detection of surface species. Eqn (6) describes

the steady state on or close to the surface of pristine sodium

alanate particles. As long as the concentration of impurities

that are able to ease either the decomposition of AlH3, or its

transition into the gas phase, is sufficiently low, there is no

driving force for further decomposition of NaAlH4. Purified

sodium alanate can be stored for months (at room temperature

and ambient pressure) without any detectable decomposition.

This stability is indeed remarkable since the equilibrium

pressure for NaAlH4 at 303 K is 0.1 MPa.36 Moreover,

pristine NaAlH4 melts above 450 K (457 K,4 453.8 K37)

without immediate decomposition and release of hydrogen.

This is the reason why sodium alanate was regarded as

unsuitable for reversible hydrogen storage. The proposed

trapping reaction (eqn (5)) might be the key to understanding

the unexpectedly high thermal stability of sodium alanate,

since without the loss of AlH3 from the surface, there is no

driving force for a further decomposition of sodium alanate

according to eqn (4). The situation changes dramatically if an

effective dehydrogenation catalyst is present on the surface of

the sodium alanate particles, as for example after the addition

of titanium compounds and ball-milling.

Dehydrogenation of NaAlH4 in the presence of a catalyst

In the presence of a catalyst, the volatile aluminium hydride

generated according to eqn (4) on the surface of the alanate

particles is effectively decomposed into aluminium and

hydrogen at the catalytically active sites:

AlH3 ! Al + 32H2 (7)

While the hydrogen is immediately released into the gas phase,

the aluminium deposits on the surface of the alanate particles,

presumably close to the active site. The catalytic decomposition

of sodium alanate can be observed at room temperature,38 but

it is much faster at elevated temperatures because of the shift

of the equilibrium and the faster transport of AlH3. The first

step of the decomposition is usually carried out at temperatures

around 393 K.39–41 According to eqn (4), the more the AlH3 is

consumed, the more the NaAlH4 decomposes. The sodium

hydride so produced is consumed by subsequent reaction with

neighbouring NaAlH4:35

2NaH + NaAlH4 ! Na3AlH6 (8)

The reaction most likely proceeds via intermediates such as

[Na2AlH5]14 or [Na5Al3H14],

42 although neither of these has

been detected experimentally. However, as far as our present

discussion is concerned, the actual elementary steps are not

relevant. In a thorough thermodynamic assessment of this

system Lee et al.43 derived equations for the temperature

dependence of the Gibbs energy of formation of the three

compounds involved. From these equations it follows that the

Gibbs energy of the reaction according to eqn (8) is negative in

the whole temperature range used for the reversible

Fig. 4 27Al MAS NMR spectra of a 10 : 1 : 1 mixture of NaAlH4,

18C6 and polymeric AlH3 after ball-milling at T = 77 K recorded

(a) at 304 K, (b) after 10 min at 311 K, (c) after 22 min at 311 K, and

(d) after 50 min at 311 K. For comparison, the 27Al MAS NMR

spectrum of (e) Na(18C6)Al2H7 prepared according to the literature is

included. The spectrum in (a) was recorded at a spinning frequency

nMAS = 10 kHz, all others at nMAS = 13 kHz.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 17239

dehydrogenation of sodium alanate. The lower the temperature

the further the equilibrium in eqn (8) lies to the right. This

provides a simple explanation as to why, despite the direct

decomposition of NaAlH4 into NaH, substantial amounts of

the alkali hydride are only detected after most of the starting

material has been converted to Na3AlH6.13,23,44 Since both

AlH3 and NaH are consumed in different reactions it is not

clear which of these reactions is the actual driving force for

further decomposition of NaAlH4 according to eqn (4).

As far as the decomposition of Na3AlH6 is concerned, i.e.

the second step in the decomposition of sodium alanate

(eqn (2)), we propose that it proceeds via the reactions

discussed already. Although the equilibrium described in

eqn (8) lies to the right, any NaAlH4 formed according to this

equilibrium would most probably immediately decompose into

NaH and AlH3 (eqn (4)) at the higher temperature used for the

second step of the dehydrogenation (typically 453 K39–41). This

drain-off effect would lead to a complete decomposition of

Na3AlH6. Hence, the decompositions of Na3AlH6 and NaAlH4

are able to proceed via the same mechanism. According to this

mechanism, the difference in the activation energies of both

steps of the catalysed decomposition of sodium alanate45

could be caused by the activation energy of the reaction given

in eqn (8).

Clearly, all equilibria discussed here are understood as local

equilibria. At the beginning of the dehydrogenation only the

outer rims of the alanate particles are involved. At a certain

level of decomposition, i.e. at a certain level of depletion of

aluminium, patches with a Na3AlH6 crystal structure will

develop. Until the first step of dehydrogenation is completed,

a rather inhomogeneous distribution of the elements involved

is very likely. Since the crystal structures of NaAlH4 and

Na3AlH6 do not match, mechanical stress will eventually lead

to the formation of cracks. Starting from the new surfaces so

generated, the same processes can take place until the particles

are completely converted into Na3AlH6. Similar arguments

hold true for the second step of the decomposition, namely,

the coexistence of Na3AlH6 and NaH. The transport of AlH3

in the former alanate particles thus proceeds via the reactions

given in eqn (4) and (8). Alane chemisorbed on the surface of

aluminium particles has been shown to be mobile.46

Although we propose the same mechanism for both steps of

the catalytic dehydrogenation of sodium alanate, the different

compositions of the solid during these steps can easily be

explained. In accordance with the phenomenological eqn (1)

and (2), we should expect diminishing amounts of NaAlH4 and

growing amounts of both metallic aluminium and Na3AlH6 in

the first step (eqn (1)) and diminishing amounts of Na3AlH6

and growing amounts of both metallic aluminium and NaH in

the second step (eqn (2)). The whole process is summarized in

Scheme 1. Omitting the actual equilibrium character of most of

the reactions involved and some of the intermediates discussed

above, the scheme depicts the stepwise release of aluminium and

hydrogen from the solid sodium aluminium hydride.

Rehydrogenation in the presence of a catalyst

The equations describing the dehydrogenation process

(eqn (4), (7) and (8)) in the presence of a catalyst have

deliberately been written as equilibria. The direction in which

a particular reaction proceeds depends on the concentrations

of the compounds involved and the temperature. Hence,

reversing all arrows in Scheme 1 yields an outline of the

rehydrogenation process.

Under typical conditions for the dehydrogenation of

NaAlH4 (hydrogen pressure of around 0.1 MPa and a

temperature of 393 K or higher39–41), the equilibrium of the

decomposition reaction of AlH3 (eqn (7)) lies far to the right.

The evolving hydrogen escapes into the gas phase and the

metallic aluminium is deposited close to the catalytic site. The

situation changes considerably at a hydrogen pressure that is

two orders of magnitude higher and a somewhat lower

temperature, i.e. under conditions typically used for rehydro-

genation of the depleted material.39–41 Under these conditions,

the backward reaction of eqn (7) produces significant amounts

of AlH3. The alane is adsorbed on the surrounding aluminium

surface. Chemisorbed alane molecules can be assumed to be

highly mobile.12,46 Without NaH particles in close contact, the

formation of alane would stop as soon as the aluminium surface

is covered. However, at the interface between aluminium and

NaH particles the alane is consumed by the backward reaction

of eqn (4). In the surrounding NaH matrix, the NaAlH4 so

formed is immediately transformed into Na3AlH6, according

to the forward reaction of eqn (8). The consumption of AlH3 is

the driving force for further hydrogenation of the metallic

aluminium. The process continues until the NaH particles of

the depleted material have been transformed into Na3AlH6

particles, i.e. until the second step of dehydrogenation has

been reversed.

The rehydrogenation can proceed further, since according

to eqn (8) there is always a small amount of NaH present that

can take up further AlH3 owing to the backward reaction of

eqn (4). The ongoing supply of alane generated at the catalytic

sites on the shrinking aluminium particles finally leads to the

conversion of almost all Na3AlH6 into NaAlH4, i.e. to a

reversal of the first step of dehydrogenation. Regarding the

transport of alane, one has to bear in mind that the equilibrium

according to eqn (8) leads to an averaging of the concentrations

of the sodium-containing compounds throughout the particles.

This equilibrium can be regarded as a very effective transport

mechanism for aluminium. Hence, the uptake of alane at the

surface of the particles in the backward reaction of eqn (4) is

Scheme 1 Catalysed decomposition of sodium alanate. For clarity,

some intermediate species discussed in the text and the double arrows

to indicate the equilibria involved are omitted.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

17240 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 This journal is c the Owner Societies 2011

sufficient to account for the complete rehydrogenation. For a

more detailed description of the process, elucidation of the

elementary steps of the reaction in eqn (8) is necessary. Finally,

it is worth noting that the higher the concentration of the

NaAlH4 is, the more effective should be the transport of the

alane via the proposed equilibrium in eqn (5).

Conclusions and outlook

The crown-ether adduct Na(18C6)AlH4 can be produced by

ball-milling a mixture of NaAlH4 and the crown-ether. This

approach also works for other alkali aluminium hydrides and

crown ethers.

The Na(18C6)AlH4 so formed is completely transformed into

Na(18C6)Al2H7 by trapping molecular AlH3 at temperatures

not exceeding 343 K. This trapping reaction can be regarded as

strong evidence for the presence of volatile alane on pristine

NaAlH4.

Both crown-ether adducts can easily be identified by

solid-state 27Al NMR spectroscopy. The structural elucidation

and a more comprehensive characterisation are underway.

We propose that a disproportionation equilibrium

contributes to the unexpectedly high thermal stability of

pristine sodium alanate.

Based on the experimental evidence for the presence of

volatile alane, we propose a simple description of the catalysed

dehydrogenation and rehydrogenation of sodium alanate. In

this mechanism volatile alane plays the central role as the

transport species for aluminium. The role of the catalyst is

limited to facilitating the decomposition and formation of

alane during dehydrogenation and rehydrogenation,

respectively.

As distinct from earlier proposals we suggest that both steps

of the dehydrogenation and rehydrogenation of sodium

alanate proceed via the same mechanism.

Although the mechanism outlined above needs to be refined

and the thermodynamics and kinetics of the coupled equilibria

involved have yet to be studied in detail, we feel that pursuing

this concept should contribute considerably to a greater

understanding of the reversible dehydrogenation of complex

aluminium hydrides.

Acknowledgements

We thank Dr Richard Goddard (Mulheim) and Dr Marcus

Zibrowius (Cambridge) for their valuable comments on the

manuscript. Furthermore, we wish to thank one of the referees

for encouraging us to add a schematic description of the

proposed reaction mechanism to the paper.

Notes and references

1 P. J. Herley, O. Christofferson and R. Irwin, J. Phys. Chem., 1981,85, 1874 and references therein.

2 B. Bogdanovic and M. Schwickardi, J. Alloys Compd., 1997,253, 1.

3 J. A. Dilts and E. C. Ashby, Inorg. Chem., 1972, 11, 1230.4 T. N. Dymova, Yu. M. Dergachev, V. A. Sokolov andN. A. Grechanaya, Dokl. Akad. Nauk SSSR, 1975, 224, 591.

5 C. M. Araujo, R. Ahuja, J. M. O. Guillen and P. Jena, Appl. Phys.Lett., 2005, 86, 251913.

6 A. Peles and C. G. Van de Walle, Phys. Rev. B: Condens. MatterMater. Phys., 2007, 76, 214101.

7 Z. Lodziana, A. Zuttel and P. Zielinski, J. Phys.: Condens. Matter,2008, 20, 465210.

8 P. E. Vullum, M. P. Pitt, J. C. Walmsley, B. C. Hauback andR. Holmestad, J. Alloys Compd., 2011, 509, 281.

9 A. Borgschulte, A. Zuttel, P. Hug, G. Barkhordarian, N. Eigen,M. Dornheim, R. Bormann and A. J. Ramirez-Cuesta, Phys.Chem. Chem. Phys., 2008, 10, 4045.

10 G. J. Thomas, K. J. Gross, N. Y. C. Yang and C. Jensen, J. AlloysCompd., 2002, 330–332, 702.

11 M. Felderhoff, K. Klementiev, W. Grunert, B. Spliethoff,B. Tesche, J. M. Bellosta von Colbe, B. Bogdanovic, M. Hartel,A. Pommerin, F. Schuth and C. Weidenthaler, Phys. Chem. Chem.Phys., 2004, 6, 4369.

12 S. Chaudhuri, J. Graetz, A. Ignatov, J. J. Reilly andJ. T. Muckerman, J. Am. Chem. Soc., 2006, 128, 11404.

13 K. J. Gross, S. Guthrie, S. Takara and G. Thomas, J. AlloysCompd., 2000, 297, 270.

14 R. T. Walters and J. H. Scogin, J. Alloys Compd., 2004, 379, 135.15 C. M. Araujo, S. Li, R. Ahuja and P. Jena, Phys. Rev. B: Condens.

Matter Mater. Phys., 2005, 72, 165101.16 A. J. Du, S. C. Smith and G. Q. Lu, Appl. Phys. Lett., 2007,

90, 143119.17 H. Gunaydin, K. N. Houk and V. Ozolins, Proc. Natl. Acad. Sci.

U. S. A., 2008, 105, 3673.18 G. B. Wilson-Short, A. Janotti, K. Hoang, A. Peles and C. G. Van

de Walle, Phys. Rev. B: Condens. Matter Mater. Phys., 2009,80, 224102.

19 V. P. Tarasov and G. A. Kirakosyan, Russ. J. Inorg. Chem.(Transl. of Zh. Neorg. Khim.), 2008, 53, 2048.

20 M. H. W. Verkuijlen, P. J. M. van Bentum, E. R. H. van Eck,W. Lohstroh, M. Fichtner and A. P. M. Kentgens, J. Phys. Chem.C, 2009, 113, 15467.

21 J. Zhang, M.-A. Pilette, F. Cuevas, T. Charpentier, F. Mauri andM. Latroche, J. Phys. Chem. C, 2009, 113, 21242.

22 V. P. Balema, J. W. Wiench, K. W. Dennis, M. Pruski andV. K. Pecharsky, J. Alloys Compd., 2001, 329, 108.

23 B. Bogdanovic, M. Felderhoff, M. Germann, M. Hartel,A. Pommerin, F. Schuth, C. Weidenthaler and B. Zibrowius,J. Alloys Compd., 2003, 350, 246.

24 J. W. Wiench, V. P. Balema, V. K. Pecharsky and M. Pruski,J. Solid State Chem., 2004, 177, 648.

25 T. M. Ivancic, S.-J. Hwang, R. C. Bowman, Jr., D. S. Birkmire,C. M. Jensen, T. J. Udovic and M. S. Conradi, J. Phys. Chem.Lett., 2010, 1, 2412.

26 V. I. Simagina, V. M. Mastikhin, I. V. Stoyanova andV. A. Likholobov, Koord. Khim., 1992, 18, 120.

27 V. V. Gavrilenko, L. A. Chekulaeva, V. A. Antonovich andL. I. Zakharkin, Izv. Akad. Nauk. SSSR, Ser. Khim., 1987, 969.

28 AVT GmbH, Bleichstr. 12, 45468 Mulheim, Germany,Schulz@avtgmbh.de.

29 G. Neue and C. Dybowski, Solid State Nucl. Magn. Reson., 1997,7, 333.

30 J. Brus, Solid State Nucl. Magn. Reson., 2000, 16, 151.31 D. Freude, Quadrupolar Nuclei in Solid-state Nuclear Magnetic

Resonance, in Encyclopedia of Analytical Chemistry, ed.R. A. Meyer, John Wiley and Sons Ltd, Chichester, 2000,pp. 12188–12224.

32 O. P. Charkin, M. L. McKee and P. von R. Schleyer, Russ. J.Inorg. Chem. (Transl. of Zh. Neorg. Khim.), 1996, 41, 1973.

33 K. Eichele and R. E. Wasylishen, WSolids1, version 1.17.30,Dalhousie University, Halifax, 2001.

34 V. V. Gavrilenko, L. A. Chekulaeva, V. A. Antonovich andL. I. Zakharkin, Izv. Akad. Nauk. SSSR, Ser. Khim., 1977, 2409.

35 W. Grochala and P. P. Edwards, Chem. Rev., 2004, 104, 1283.36 B. Bogdanovic, R. A. Brand, A. Marjanovic, M. Schwickardi and

J. Tolle, J. Alloys Compd., 2000, 302, 36.37 P. Claudy, B. Bonnetot, G. Chahine and J. M. Letoffe, Thermochim.

Acta, 1980, 38, 75.38 G. Sandrock, K. Gross, G. Thomas, C. Jensen, D. Meeker and

S. Takara, J. Alloys Compd., 2002, 330–332, 696.39 B. Bogdanovic, M. Felderhoff, S. Kaskel, A. Pommerin,

K. Schlichte and F. Schuth, Adv. Mater., 2003, 15, 1012.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 17234–17241 17241

40 B. Bogdanovic, M. Felderhoff, A. Pommerin, F. Schuth andN. Spielkamp, Adv. Mater., 2006, 18, 1198.

41 B. Bogdanovic, M. Felderhoff, A. Pommerin, F. Schuth,N. Spielkamp and A. Stark, J. Alloys Compd., 2009, 471, 383.

42 J. G. O. Ojwang, R. van Santen, G. J. Kramer and X. Ke, J. SolidState Chem., 2008, 181, 3037.

43 B. M. Lee, J. W. Jang, J. H. Shim, Y. W. Cho and B. J. Lee,J. Alloys Compd., 2006, 424, 370.

44 K. J. Gross, G. Sandrock and G. J. Thomas, J. Alloys Compd.,

2002, 330–332, 691.45 G. Sandrock, K. Gross and G. Thomas, J. Alloys Compd., 2002,

339, 299.46 J. G. O. Ojwang, S. Chaudhuri, A. C. T. van Duin, Y. J. Chabal,

J.-F. Veyan, R. van Santen, G. J. Kramer and W. A. Goddard III,

J. Chem. Phys., 2010, 132, 084509.

Dow

nloa

ded

by N

ew Y

ork

Uni

vers

ity o

n 20

/05/

2013

12:

53:4

0.

Publ

ishe

d on

30

Aug

ust 2

011

on h

ttp://

pubs

.rsc

.org

| do

i:10.

1039

/C1C

P218

77H

View Article Online