You are on page 1of 16

Valencia 1

Dry Mechanochemical Synthesis of Alane from LiH and AlCl3

Research by:
Hlova, Ihor; Gupta, Shalabh; Goldston, Jennifer; Kobayashi, Takeshi; Pruski, Marek; Pecharsky,
Vitalij. The Royal Society of Chemistry. Faraday Discuss., 2014, 170, 137-153.

Commentary by:
Joseph E. Valencia
Chem 228HApril 18, 2015
Dr. David Bergbreiter

Valencia 2
Summary
This study presents the discovery of a dry mechanochemical synthesis of the potential
Hydrogen fuel source, Aluminum trihydride (alane or AlH3), from LiH and AlCl3 that can be
carried out at room temperature. This synthesis, performed by ball-milling in a hydrogen
pressurized (at 300 bar), stainless steel container, is accomplished in 3 Stages. In the first stage, 9
molar equivalents of LiH are added to 1 molar equivalent of AlCl3 (a 9:1 ratio), and the resulting
products are LiCl, Li3AlH6, and (unreacted) 3 LiH. In the second stage, 1.25 molar equivalents of
AlCl3 are added to the products of Stage 1 so that the ratio is reduced to 4:1, and the products of
the Stage 2 reaction are LiCl and LiAlH4. In the third stage, 0.75 molar equivalents of AlCl3 are
added to the products of Stage 2 (creating a 3:1 final ratio) in order to yield LiCl and AlH3 in
several polymorphic forms. Any attempt to achieve AlH3 with any lesser initial, stoichiometric
ratio results in the significant formation of Aluminum metal, thereby reducing the overall
hydrogen-carrying capacity of the product.
Background
In recent times, there has been a large push to move away from fossil fuels and towards
Hydrogen as the main source of energy for automobiles. Currently, solid state hydrogen storage
is safer for transportation and storage than pressurized or cryogenic materials. In addition, an
advantage of solid state hydrogen storage is that it can store much more hydrogen per unit
volume than can pressurized or cryogenic approaches. The U.S. Department of Energy (DOE)
has set gravimetric and volumetric hydrogen targets for solid state hydrogen fuel cells of 7.5 wt
% and 70 g/L for the year 2017 (2). Among the solid state approaches, alane appears to be
advantageous because it surpasses these targets with flying colors, containing hydrogen in
quantities of 10.1 wt% and 148 g/L. Alane also has two other advantages that suggest it should

Valencia 3
be a frontrunner in the race for the development of a hydrogen fuel cell: first, it releases
hydrogen at around 100 C, which is an optimal temperature for a reaction under the hood of an
automobile; and second, decomposition yields only elemental hydrogen and Aluminum, which
makes the long-term sustainability of an alane fuel source both environmentally friendly and
feasible.
Historically, alane has only been synthesized in a few ways. It was first prepared in 1942 as
an amine (3). The next approach, undertaken by Finholdt et al, accomplished the synthesis of an
ethereal solution via an exchange reaction between LiH and AlCl3 (4). In 1955, Chisinky et al.,
(5) and Brower et al., (6) synthesized alane organometallically at Dow Chemical. Another
popular approach to synthesizing alane required a diethyl ether solvent, and was accomplished
via the two chemical reactions shown below (4).
1) 3 LiH + AlCl3 + nEt2O AlH3 + 3 LiCl + nEt2O
2) 3 LiAlH4 + AlCl3 + nEt2O 4 AlH3 + 3 LiCl + nEt2O
Currently, there are 7 known forms of AlH3, but only the , , , and forms have been
studied (6). All of these forms decompose into elemental Al and H2, however, the form is the
only one which does so in a single step mechanism. There are a number of other ways to make
alane. Unfortunately, direct hydrogenation is not one of these ways because it is not
thermodynamically favorable; and in order accomplish this, the reaction would require incredibly
high pressures and would not be feasible on an industrial scale. In addition to the wet, diethyl
ether approach, amines have also been used as an organic solvent (10). An issue arises with this
chemistry, however, because this reaction forms the AlH3 in an adduct form; requiring further
purification; and in the process of this purification, much of the alane is destroyed due to heating.
Other current/recent research regarding the synthesis of alane includes electrochemical

Valencia 4
regeneration, the hydrogenation of Aluminum metal in supercritical fluid (11 & 12), direct
thermal conversion (17), and solvent-free mechanochemical synthesis.
Current mechanochemical synthesis generates AlH3, and LiCl; however, in the process,
Aluminum metal is also formed, which results in an un-optimized yield and an overall loss in
Hydrogen carrying capacity for the reaction. Though it is not empirically confirmed, it is
currently hypothesized that the heat from decomposing AlH3 amidst the formation reaction
causes the formation of Aluminum metal. Procedures to counteract this decomposition by
lowering the temperature could involve cryogenics and would not be favorable if used in mass
production due to their complexity and large costs. Thus, the researchers chose a different
approachto attempt to prevent the formation of Aluminum metal by stoichiometric control of
the LiH and AlCl3. Thus, the main goal of this study was to determine the ratio of LiH to AlCl3
that prevented the formation of Aluminum metal in the one pot mechanochemical synthesis of
alane (starting with only LiH in the pot). Additional goals of this study included finding a
procedure that optimized the yield of AlH3, finding feasible (and inexpensive) ambient
conditions in which this synthesis could occur, and to gain an understanding of the reaction
mechanism of the mechanochemical synthesis of alane.
Researchers Approach & Results
The researchers used LiH (Aldrich, 98%) and AlCl3 (Aldrich, 99.99%). These reactants,
as well as the AlH3 product, are both sensitive to air and moisture. Thus, all handling of these
substances occurred in a purified, argon-filled glove box, where the air and water concentrations
were controlled to less than 5 ppm. Approximately 1 gram of the starting material (a
predetermined combination of LiH and AlCl3) was placed in a stainless steel container with 20
chrome steel balls (each having a diameter of 11.9 mm and weighing 7 grams each). The

Valencia 5
container was then sealed, and the argon was replaced with H2 gas (Linweld, 99.999%) until the
desired pressure was reached. (Pressures ranged from between 1 to 350 bar.) The stainless steel
container was then rotated at 150 to 300 rpms in a two-station, horizontal planetary mill in 2
minute intervals. The reaction was considered complete after 48 min (15).
The products of the mechanochemical reaction were characterized in 3 ways: X-ray
powder diffraction (XRD) analysis, Solid-state Nuclear Magnetic Resonance (SSNMR)
Spectroscopy, and temperature programmed desorption. The XRD analysis was taken with a Cu
K1 radiation with a 0.02degree 2 step, in a range of Bragg angles 2 from 10 degrees to 80
degrees. The SSNMR used a 104.2 MHz frequency for the 27Al and 400.0 MHz frequency for
the 1H. A 1.0 M aqueous solution of Al(NO3)3 was used as a standard for all 27Al spectra. The
27

Al spectra are referred to as DPMAS and the 1H spectra are referred to as CPMAS. For analysis

by temperature programmed desorption, the thermal decomposition of the material was


monitored while being heated from room temperature to 200 at a rate of 4/min. Once the
temperature reached 200, it remained there until the remainder of the material decomposed,
and then the gasses were analyzed.
The above procedure (excluding characterization) was followed for ratios of LiH to AlCl3
of 3:1, 4:1, 6:1, 9:1, and 12:1 (15). The 3:1 and 4:1 ratios yielded a very large amount of
Aluminum metal, which initially suggested two (possible) limitations of the reaction: first, that
for quantitative reactions, the only way to stabilize the hydride intermediate was via the
formation of an alane-ether adduct (in an ether solvent); and/or second, that a mechanochemical
reaction must follow thermodynamic pathways, which would not allow for the direct
hydrogenation of the AlCl3 (15). The 12:1 ratio gave some hope, however, as Li3AlH6 and
(unconsumed) LiH were the only major constituents of the product. The reaction of the 6:1 ratio

Valencia 6
also produced Aluminum metal, however, the reaction of the 9:1 ratio did not produce any
Aluminum metal. (One key condition that affected the reactions of all ratios was the pressure of
the H2 gasall ratios required at least 300 bar of pressure; otherwise, they formed Aluminum
metal, LiCl, and LiH.) Since this ratio would cost less than a higher ratio in production (because
those making the alane would not have to spend as much on the starting LiH), it was considered
the optimal ratio for the remainder of the study (15).
After the optimal ratio was determined, the researchers decided that it was necessary to
add more AlCl3 to the container in order to quantitatively form AlH3. In the end, it was
determined that the AlCl3 had to be added in 3 stages in order to form the desired AlH3 product,
where stage 1 was the (previously completed) reaction of the 9:1 mixture, stage 2 was the
addition of AlCl3 so that the ratio was reduced to 4:1, and stage 3 was the addition of AlCl3 so
that the ending ratio was 3:1 (which was the ratio necessary for the quantitative formation of
AlH3). These stages can be summarized by the following set of chemical reactions, where A
Products are 3 LiCl + Li3AlH6 + 3 LiH (the products from the reaction at stage 1), and B
Products are 6 LiCl + 9/4 LiAlH4 (the products from the reaction at stage 2).
Stage 1)

9 LiH + AlCl3 3 LiCl + Li3AlH6 + 3 LiH

Stage 2)

A Products + 5/4 AlCl3 6 LiCl + 9/4 LiAlH4

Stage 3)

B Products + 3/4 AlCl3 9 LiCl + 3 AlH3

XRD and SSNMR characterizations were performed on the contents of the Stage 1 reaction
as a function of milling time. Samples were analyzed at times of 16 min, 24 min, and 48 min.
The graphical results of these characterization tests are shown in Figure 1 (next page).

Valencia 7
When the reaction had progressed for 16 min, there were 3 types of Bragg peaks shown
LiCl, LiAlCl4, and LiH; however, the DPMAS indicates that there were 5 species of Aluminum
present in the mixture in the forms of LiAlCl4, LiAlH4, AlVI, AlCl3, and Li3AlH6. The CPMAS
can only detect compounds that contain Hydrogen (due to the frequency), and thus, AlCl3 is not
detected by the CPMAS spectroscopy (15). Note that the LiAlCl4 and the LiAlH4 have given
spectroscopy values that are both very close to 95 ppm, and thus, their spectral overlap makes
Figure 1 (15)

quantifying the relative abundance of these two


compounds challenging. It can be determined
that both compounds are present, however,
because the LiAlH4 shows up in the CPMAS
spectrum (whereas the LiAlCl4 does not show
up in this spectrum) and the DPMAS spectrum
indicates that other compounds besides LiAlH4
are present around 95 ppm. The minute presence
of the six-coordinated AlVI suggests that AlH3 or
derivatives containing at least 3 Chlorine atoms
have been formed. In order for these derivatives
to have been formed, AlH3 would have to have
been formed previously in the reaction,
suggesting the following stoichiometric
equation:
3) LiH + AlCl3 LiAlCl4 + AlH3

Valencia 8
According to Figure 1, when the reaction has progressed for 24 min., several things can be
noted: first, LiAlH4 now is present enough to be apparent in the Bragg peaks (and that LiAlCl4 is
overshadowed by the presence of LiAlH4); second, the presence of LiCl has increased, which is
one of the final products; third, the DPMAS spectrum shows a greater presence of LiAlH4 than
what was present at 16 min.; and fourth, that the presence of AlVI (any form of AlH3) has
disappeared. The evidence for the increase in the amount of LiAlH4 present and the
disappearance of the AlVI means that the products of reaction 3 were merely intermediates in the
formation of LiAlH4. This suggests the following intermediate reaction (15):
4) AlH3 + LiH LiAlH4
Equations 3 and 4, together, account for the initial formation of AlH3, its disappearance,
along with the appearance of LiAlCl4 (in the Bragg peaks at 16 min. and the SSNMRs) and
LiAlH4 (in the Bragg peaks at 24 min. and an increased presence at the SSNMRs at 24 min.). It
is important to note that the resonance on the graph labeled ssb is a spinning sideband, which
is generated by the LiAlH6 structure. The reaction reached completion after 48 min., and the
corresponding XRD for this time shows only Li3AlH6, LiCl, and LiH present at the completion of
the stage 1 reaction. These conclusions are supported in several ways by the corresponding
SSNMR data. First, both SSNMR graphs show a large amount of resonance for Li3AlH6 (with an
SDP of approximately 90%). Second, there is a small amount of resonance at 95 ppm on the
DPMAS graph (with an SDP of approximately 10%), which corresponds only to LiAlCl4 because
no LiAlH4 shows up on the CPMAS graph (15). Third, the presence of ssb near 115 ppm ensures
that LisAlH6 was present in the product of the Stage 1 reaction. To ensure that the end of stage
one had occurred, the researchers extended the milling time up to 3 hours, and no significant
changes occurred.

Valencia 9
For the initial attempt of the Stage 2 reaction, AlCl3 was added until the ratio of 3 LiH:1
AlCl3 (the quantitative ratio) was attained. However, black powder aluminum metal formed after
15-20 min. of the reaction. The amount of AlCl3 added was reduced so that the final ratio of the
Figure 2 (15)

stage 2 reaction was 4 LiH:1 AlCl3, and the result was


that no aluminum metal was formed. Again, the
XRD and SSNMR characterizations were
performed on the contents of the Stage 2
reaction. The only difference between these
characterizations in Stage 1 and Stage 2 is that
they were performed after the reaction had
progressed for 8 min. and 48 min. in stage two.
The graphical results of these test are shown in
Figure 2 (15).

Valencia 10
After the reaction had progressed for 8 min., the XRD shows that only LiAlCl4 and LiCl
were present. The DPMAS showed that LiAlH4 and LiAlCl4 were present in the spectrum in a
relative abundance of 66%, that AlVI and AlCl3were present in the spectrum in a relative
abundance of 8% each, and that the Li3AlH6 was present in the spectrum in a relative abundance
of 17%. Again, the exact abundances of LiAlCl4 and LiAlH4 were difficult to determine, the
shape of the signal near 95 ppm is very similar to that of the Stage 1 95 ppm signal at 16 min.,
suggesting that the Stage 1 and Stage 2 reactions commence via a similar initiation mechanism.
This is further supported by the small signals that arise in both of these spectra at 13 ppm and
correspond to AlVI (which formed early on in Stage 1 and later diminished). In addition, this is
also supported by the fact that unused AlCl3 is present at -2 ppm in both of the aforementioned
spectra for Stage 1 and Stage 2 (and is later used up in the Stage 1 reaction).
After the reaction had gone to completion, the combination of LiAlCl4 and LiAlH4 around 95
ppm comprise 97% of both SSNMR spectra. Both the shape of the DPMAS at 95 ppm and the
CPMAS indicate that this combination is more so LiAlH4 at this point in time. Both SSNMR
spectra also indicate that there is only a small amount of Li3AlH6 present. The XRD data shows
that only LiAlH4 and LiCl were present in significant amount at the end of the 48 min. of the
Stage 2 reaction. The data from the SSNMR and XRD indicate that LiAlH4 was the only major
product present at the end of the Stage 2 reaction. This is further supported by the equation used
in the historical synthesis of LiAlH4:
5) 4 LiH + AlCl3 LiAlH4 + 3 LiCl
For the Stage 3 reaction, 0.75 molar equivalent of AlCl3 was added to give the final molar
ratio of 3 LiH:1 AlCl3. XRD and SSNMR characterizations were collected as the reaction after

Valencia 11
the reaction had progressed for 10 min. and 48 min.. Figure 3 (on the following page) contains
the graphical results of these characterization tests.
After the reaction had progressed for 10 min., the XRD showed only Bragg peaks for LiCl
and LiAlCl4. Though there were no Bragg peaks for AlH3, the absence of metallic Al and AlCl3
indicated that AlH3 had formed in a non-expected complex. Both SSNMR spectra reveal that
there are very large increases in the presence of AlVI, which indicates that AlH3 formed in an
amorphous or nanocrystalline form. The signal at 95 ppm corresponds to the LiAlCl4 and
LiAlH4, which took up the other 45% of the DPMAS spectrum. Due to the similarities in Stage 1
and Stage 2 spectra, it is likely that there was a greater amount of LiAlH4 present.
After the reaction had progressed for 48 min., the Bragg peaks only registered that LiCl
was present in the product, which suggests that the reaction had been driven to completion. The
Figure 3 (15)

SSNMR only have 1 major signal at this point, which

corresponds to AlVI. It is important to note that the shape of the AlVI is different in the DPMAS
and CPMAS spectra, and because of the rightward shift in the CPMAS form, the product formed
is likely a composed of several polymorphic forms of AlH3 (15). Some of these could contain
trace amounts of Chlorine.

Valencia 12
Figure 4 (next page) shows the thermal
decomposition of the products of each of the 3 stages
of the mechanochemical reaction. Each of these
products were analyzed after 48 min. of the stage.
The 9:1 product has a single desorption step that
commences at 180 C, and has a net evolution 1.4%
net evolution of Hydrogen by wt% (15). The 4:1
product decomposes in 2 stages, with initiating
temperatures of 130 C and 180 C, which consist
with the known LiAlH4 path (16):
6) LiAlH4 1/3 Li3AlH6 + 2/3 Al + H2
7) 1/3 Li3AlH6 LiH + 1/3 Al + H2
Figure 4 (15)

The net weight of hydrogen evolution


for the curve 4:1 product was around
1.8 wt% (accepted value 1.87 wt%)
(15). The stage 3 reaction had a onestep breakdown that was initiated at 90
C, and had net hydrogen evolution of
1.9 wt%. The 90 degree Celsius
initiation is consistent with the literature value (18).
The reaction mechanisms were difficult to determine. The products of the (initially failed)
1 AlCl3:3 LiH reaction were characterized by XRD at 6, 12, and 18 min., and it was determined

Valencia 13
that the LiAlCl4 was initially formed. This then became AlCl2H and/or AlClH2, which are
unstable and yielded Aluminum metal upon percussive impact. Thus, the researchers
hypothesized that an excess of LiH was needed to prevent the formation of large, local AlCl3:LiH
ratios, in accordance with the conclusions of Mikheeva et al (19). In another strange turn of
events, another 3:1 LiH:AlCl3 reaction was performed at 150 rpms for 20 hours (pressure was not
changed). Contrary to expectation, this did not result in the formation of Aluminum metal,
however, the reaction occurred very slowly, so that it was not completed even after 20 hours of
milling. Ergo, the researchers were unable to precisely determine the reaction mechanism of their
mechanochemical synthesis of alane, however, they were able to hypothesize that the H2 gas
acted either as a stabilizer for solid state diffusion of ions (to stabilize hydride ions) or a heat
transport medium that created new thermodynamic pathways (15).
Commentary
This chemistry could grow increasingly important in the near future as Hydrogen fuel
cells become more popular and necessary. Currently, the mechanochemical synthesis of AlH3 is
successful in the sense that it avoids the expenses, dangers, and wastes associated with the use of
solvents, can be carried out at room temperature, and that it can be synthesized without forming
any Aluminum metal.
There are a number of improvements that need to be made to this chemistry. The
researchers also admit this for several facets of this chemistry, stating first that the product is
80% LiCl by weight, and that the AlH3 needs to be separated . Currently, no organic solvent has
been able to completely separate these compounds without altering their chemical makeup (15).
In order for this chemistry to be truly dry, a solvent-free separation method for the products of
this synthesis must be achieved. Since the melting point of LiCl is much higher than the

Valencia 14
temperature at which AlH3 decomposes (13), melting LiCl in order to separate it from AlH3
would be ineffective (and wet). However, Lithium metal is slightly paramagnetic (14). Perhaps
some gaseous catalyst could be introduced in the first stage of the reaction which would convert
the LiCl into Lithium metal and Chlorine gas. If so, the Lithium metal could be separated at each
phase by use of its magnetic properties, and the Cl2 gas could be removed after each stage of the
synthesis. This could also prove to be advantageous because it could drive the reaction toward
completion by preventing (or hindering) Lithium from forming the LiAlH4, LiAlCl2, and Li3AlH6
complexes with aluminum. If combustion of H2 with Cl2 becomes problematic, another gas, such
as Cl2 or Helium, could be used in place of H2 to provide the necessary 300 bar of pressure. It
would also be advantageous to find a mechanochemical catalyst that would form only the form
of AlH3. The wt% of Hydrogen in the product must also be perfected in order to meet the goals
set by the D.O.E.
The researchers also note that the gas pressure of 300 bar is an imperative condition for
the synthesis to proceed without the formation of Aluminum metal, but this chemistry is not yet
understood. The fact that it could play a role in heat transfer or provide a fluid medium for solid
state diffusion indicates that more research needs to be done. Gases could hold thermodynamic
or mechanistic secrets for reactions with solids that could revolutionize this chemistry.

References
(1) J. Graetz, J. Reilly, V. Yartys, J. Maehlen, B. Bulychev, V. Antonov, B. Tarasov and I. Gabis,
J. Alloys Compd., 2011, 509, S517.
(2) http://www1.eere.energy.gov/hydrogenandfuelcells/storage/pdfs/targets_onboard_
hydro_storage_explanation.pdf.
(3) O. Stecher and E. Wiberg, Ber. Dtsch. Chem. Ges. B, 1942, 75, 2003.

Valencia 15
(4) A. E. Finholt, A. C. Bond and H. I. Schlesinger, J. Am. Chem. Soc., 1947, 69, 1199.
(5) G. Chizinsky, J. Am. Chem. Soc., 1955, 77, 3164.
(6) F. Brower, N. Matzek, P. Reigler, H. Rinn, C. Roberts, D. Schmidt, J. Snover and K. Terada,
J. Am. Chem. Soc., 1976, 98, 2450.
(7) E. C. Ashby, J. R. Sanders, P. Claudy and R. Schwartz, J. Am. Chem. Soc., 1973, 95, 6485.
(8) A. E. Finholt, G. D. Barbaras, G. K. Barbaras, G. Urry, T. Wartik and H. I. Schlesinger, J.
Inorg. Nucl. Chem., 1955, 1, 317.
(9) B. M. Bulychev, A. V. Golubeva and P. A. Storozhenko, Russ. J. Inorg. Chem., 1998, 43,
1765.
(10) R. Zidan, B. L. Garcia-Diaz, C. S. Fewox, A. C. Stowe, J. R. Gray and A. G. Harter, Chem.
Commun., 2009, 25, 3717.
(11) G. S. McGrady, US Patent US 2008/0241056 A1, 2 October, 2008.
(12) L. V. Dinh, D. A. Knight, M. Paskevicius, C. E. Buckley and R. Zidan, Appl. Phys. A:
Mater. Sci. Process., 2012, 107, 173.
(13) Material Safety Data Sheet: Lithium Chloride, Reagent ACS (Crystals), 99% (Titr.). (2000,
August 2). Retrieved April 17, 2015, from http://avogadro.chem.iastate.edu/MSDS/LiCl.htm
(14) Magnetic Susceptibilities of Paramagnetic and Diamagnetic Materials at 20C. (n.d.).
Retrieved April 17, 2015, from http://hyperphysics.phy-astr.gsu.edu/hbase/tables/magprop.html

Valencia 16
(15) Hlova, I.; Gupta, S.; Goldston, J.; Kobayashi, T.; Pruski, M.; Pecharsky, V. Dry
mechanochemical synthesis of alane from LiH and AlCl3. The Royal Society of Chemistry.
Faraday Discuss., 2014, 170, 137-153.
(16) J. A. Dilts and E. C. Ashby, Inorg. Chem., 1972, 11, 1230.
(17) V. P. Balema, V. K. Pecharsky and K. W. Dennis, J. Alloys Compd., 2000, 313, 69.
(18) G. Sandrock, J. Reilly, J. Graetz, W.-M. Zhou, J. Johnson and J. Wegrzyn, Appl.
Phys. A: Mater. Sci. Process., 2005, 80, 687.
(19) V. I. Mikheeva, E. M. Fedneva and Z. L. Shnitkova, Zh. Neorg. Khim., 1956, 1,
2440.

You might also like