Rapid solid-state metathesis routes to aluminum nitride

Article abstract of DOI:10.1021/ic026143z

Metathesis (exchange) reactions offer the possibility of controlling temperature through a judicious choice of precursors. Here, a reaction between AlCl₃ and Ca₃N₂ is found to produce phase-pure aluminum nitride (AIN) in seconds. The CaCl₂ byproduct salt, whose formation drives this highly exothermic reaction, is simply washed away after reaction completion. SEM images demonstrate that the AIN product is a micron-sized powder, while TEM shows well-formed crystallites. Thermodynamic calculations indicate that a reaction temperature of 2208 K could be reached under adiabatic conditions. Using an in situ thermocouple and a stainless steel reactor vessel to hold the precursors, a reaction temperature of 1673 K is measured 0.8 s after initiation. Switching to a thermally insulating ceramic vessel produces a maximum reaction temperature of 2010 K because of the more nearly adiabatic conditions. The high reaction temperature appears to be critical to forming phase-pure AIN. Experiments with Li₃N, instead of Ca₃N₂, produce lower temperatures (1513 K), resulting in both Al and Al₂O₃ impurities.

Full text of DOI:10.1021/ic026143z

Inorg. Chem. 2003, 42, 2714−2719
Rapid Solid-State Metathesis Routes to Aluminum Nitride
Rebecca A. Janes, Madeleine A. Low, and Richard B. Kaner*
Department of Chemistry and Biochemistry, and Exotic Materials Institute,
UniVersity of California, Los Angeles, California 90095-1569
Received October 29, 2002
Metathesis (exchange) reactions offer the possibility of controlling temperature through a judicious choice of precursors.
Here, a reaction between AlCl3 and Ca3N is found to produce phase-pure aluminum nitride (AlN) in seconds. The
2
CaCl2 byproduct salt, whose formation drives this highly exothermic reaction, is simply washed away after reaction
completion. SEM images demonstrate that the AlN product is a micron-sized powder, while TEM shows well-
formed crystallites. Thermodynamic calculations indicate that a reaction temperature of 2208 K could be reached
under adiabatic conditions. Using an in situ thermocouple and a stainless steel reactor vessel to hold the precursors,
a reaction temperature of 1673 K is measured 0.8 s after initiation. Switching to a thermally insulating ceramic
vessel produces a maximum reaction temperature of 2010 K because of the more nearly adiabatic conditions. The
high reaction temperature appears to be critical to forming phase-pure AlN. Experiments with Li3N, instead of
Ca3N , produce lower temperatures (1513 K), resulting in both Al and Al2O impurities.
2
3
Introduction
consuming and/or expensive. Syntheses of AlN often involve
processes with complex equipment, such as ion beam
evaporation4 or DC arc plasma, which forces aluminum
ingots to react with ammonia or nitrogen gas under extreme
conditions.7 Another synthetic method, carbothermal reduc-
tion, involves the conversion of Al2O3 and C in a N2 or NH3
atmosphere into AlN at high temperatures (>1600 K) for
several hours.8 Self-propagating high-temperature synthesis
represents an additional route to AlN, in which Al metal
powder is reacted with elemental nitrogen or NaN3 at high
pressures (>1000 atm) and temperatures (up to 1900 K).9
One drawback to this synthesis is incomplete nitridation,
especially at lower pressures. A major challenge is to obtain
a high yield of pure AlN at moderate temperatures and low
pressures.
Miniaturization of high-power electronic devices greatly
increases the amount of thermal energy produced per unit
area during their operation. This creates a need for materials
that can dissipate heat rapidly to avoid electronic or me-
chanical failure.1 Aluminum nitride (AlN) is an important
thermal management material for silicon-based electronics
because of its low coefficient of thermal expansion (4.3 ×
10-6 K-1), closely matching that of silicon, and its very high
thermal conductivity (320 W m-1 K-1).2 AlN is an insulator
with a large band gap (Eg ) 6.4 eV), high resistivity (>1011
Ω m), and low dielectric constant (8.6).3 These physical
properties make AlN promising for use as an electronic
substrate and as a packaging material for circuits.4 Other
applications for AlN include UV photodetectors, pressure
sensors, thermal radiation sensors, and field-effect tran-
sistors.5,6
Solid-state metathesis (exchange) reactions offer the
advantages of a rapid, high-yield method that starts from
room-temperature solids and needs little equipment. The idea
behind metathesis reactions is to use the exothermicity of
There is a need for new synthetic approaches to AlN and
other nitrides, because most current methods are time-
* To whom correspondence should be addressed. E-mail: kaner@
chem.ucla.edu. Phone: 310-825-1346. Fax: 310-206-4038.
(1) Jagannadham, K.; Watkins, T. R.; Dinwiddie, R. B. J. Mater. Sci.
2002, 37, 1363-1376.
(5) Fuflyigin, V.; Salley, E.; Osinsky, A.; Norris, P. Appl. Phys. Lett. 2000,
77, 3075-3077.
(6) Djurisic, A. B.; Bundaleski, N. K.; Li, E. H. Semicond. Sci. Technol.
2001, 16, 91-97.
(2) Kitagawa, H.; Shibutani, Y.; Ogata, S. Model. Simul. Mater. Sci. Eng.
1995, 3, 521-531.
(7) Li, H. D.; Zou, G. T.; Wang, H.; Yang, H. B.; Li, D. M.; Li, M. H.;
Yu, S.; Wu, Y.; Meng, Z. F. J. Phys. Chem. B 1998, 102, 8692-
8695.
(8) Contursi, L.; Bezzi, G.; Beghelli, G. (TEMAV S.p.A.). Process for
preparing fine aluminum nitride powder from an inorganic flocculant.
Venice Porto Marghera EP0481563, Apr 22, 1992.
(9) Munir, Z. A.; Holt, J. B. J. Mater. Sci. 1987, 22, 710-714.
(3) Goldberg, Y. In Properties of AdVanced Semiconducting Materials:
GaN, AlN, InN, BN, SiC, SiGe; Levinshtein, M. E., Rumyantsev, S.
L., Shur, M. S., Eds.; John Wiley & Sons: New York, 2001; pp 31-
47.
(4) Zhu, Q.; Jiang, W. H.; Yatsui, K. J. Appl. Phys. 1999, 86, 5279-
5285.
2714 Inorganic Chemistry, Vol. 42, No. 8, 2003
10.1021/ic026143z CCC: $25.00 © 2003 American Chemical Society
Published on Web 03/25/2003

Metathesis Routes to Aluminum Nitride
byproduct salt, the acid prevents base-catalyzed hydrolysis of the
product.20,21 To reduce the amount of time the AlN is in the aqueous
solvent, the product is immediately vacuum filtered and then dried
in a furnace at 450 K. Because cellulose filter paper cannot
withstand the acid washing, glass microfiber filter paper (Millipore)
is used instead.
In situ reaction temperature measurements were made by
modifying the stainless steel reactor. A hole was drilled through
the bottom of both the steel canister and the reaction cup. A
thermocouple was threaded through the hole and placed directly
into the reaction mixture and secured with ceramic paste. A
computer was connected to the thermocouple and programmed to
record one data point every millisecond.
salt formation to rapidly create a desired product. A metal
halide is combined with an alkali or alkaline earth main-
group compound to produce the desired product plus a salt
that is then washed away with water or alcohol. Metathesis
reactions have proven to be successful in the synthesis of a
number of crystalline refractory materials including borides,10
chalcogenides,11,12 and nitrides.13-15
Once initiated, metathesis reactions reach high tempera-
tures (>1200 K) in a fraction of a second and cool very
quickly (often <5 s). Because of the rapid nature of these
reactions, nucleation and growth are quickly terminated,
generally resulting in small crystallites and occasionally
forming metastable phases.11,16 Previous attempts to make
AlN via metathesis reactions resulted in oxide impurities.17
Here, phase-pure AlN is synthesized in seconds from a
metathesis reaction between Ca3N2 and AlCl3. The use of
Ca3N2, which increases the temperature of the reaction, is
critical to avoiding impurities. Products are characterized
using powder X-ray diffraction, scanning electron micros-
copy, transmission electron microscopy, and in situ temper-
ature analysis.
Product Characterization
Powder X-ray diffraction was performed on the washed products
using a Crystal Logic θ-2θ diffractometer with a graphite mono-
chromator and Cu KR ) 1.5418 Å radiation. The scans were taken
between 10 and 100° 2θ at 0.1° intervals with a 3-s count time.
Least-squares refinement was carried out using MacDiff (http://
www.geol.uni-erlangen.de/html/software/Macdiff.html) to fit the
X-ray diffraction peaks and Unit Cell (http://www.esc.cam.ac.uk/
astaff/holland/UnitCell.html) to then calculate the lattice parameters.
In situ reaction temperatures were measured in a modified reactor
with 0.03-in.-diameter C-type (Omega, 26% rhenium/84% tungsten
versus 5% rhenium/85% tungsten) high-temperature thermocouples,
which were placed directly into the reaction mixture. Scanning
electron microscopy (SEM-Stereoscan 250) was used to characterize
surface structure and particle size of the product. Transmission
electron microscopy (TEM JEOL 100CX) provides a fuller picture
of the morphology of the AlN crystallites. Thermogravimetric
analysis (TGA) was carried out on a Perkin-Elmer Pyris Diamond
TG/DTA, from room temperature to 1770 K, at increments of 10
°C/min.
Experimental Section
The precursors AlCl3 (Strem, 99.99%), Al2S3 (Cerac, 99.9%),
Li3N (Cerac, 99.5%), and Ca3N2 (Cerac, 99%) were used as
received. AlI3 was formed by heating its constituent elements (Al,
Cerac, 99.5%; I2, Fisher) through a vapor-transport reaction in an
evacuated, sealed Pyrex tube using a temperature gradient from
463 to 623 K, as modified slightly from a literature preparation of
GaI3.18
The synthesis of aluminum nitride was carried out in a helium-
filled glovebox (Vacuum Atmospheres MO-40). The amounts of
reactants were adjusted to produce 0.20 g (4.8 mmol) of AlN
product. Stoichiometric amounts of the reactants were weighed and
ground together with an agate mortar and pestle. The reactants were
then transferred to a stainless steel (or ceramic) cup and placed
within a larger capped steel reaction vessel, modeled after a bomb
calorimeter.19 This allows for containment of any gases produced.
The reaction is initiated through the use of a resistively heated
Nichrome wire and is complete in less than a second. Warning:
Solid-state metathesis reactions are highly exothermic and can
initiate as the reagents are being ground together. Precautions should
be taken before performing this type of reaction, and extreme care
should be used when scaling up reactions. The reaction products
are then removed from the drybox and washed in 0.5 M HCl or
1.0 M H3PO4. While the aqueous washing solution removes the
Results and Discussion
Solid-state metathesis reactions between the aluminum
precursors (AlCl3 or Al2S3) and the nitriding agents (Li3N
or Ca3N2) proceed in a rapid, exothermic manner upon
initiation with a resistively heated Nichrome wire, as follows
AlCl3 + Li3N f AlN + 3LiCl
0.5Al2S3 + Li3N f AlN + 1.5Li2S
AlCl3 + 0.5Ca3N2 f AlN + 1.5CaCl2
(1)
(2)
(3)
The real driving force behind each reaction is the formation
of an ionic salt (LiCl, Li2S, or CaCl2), which is so favorable
that the reactions become self-propagating. The salt that is
produced can then be washed away.
(10) Rao, L.; Gillan, E. G.; Kaner, R. B. J. Mater. Res. 1995, 10, 353-
361.
When choosing reagents for metathesis reactions, several
important factors must be considered, including the stability
of products versus reactants, the temperatures at which the
precursors change phase, and the maximum reaction tem-
perature. Clearly, the products must be considerably more
stable than the reactants to create an exothermic reaction that
will self-propagate. A phase change of one of the precursors
is generally needed to initiate a solid-state metathesis
(11) Gillan, E. G.; Kaner, R. B. J. Mater. Chem. 2001, 11, 1951-1956.
(12) Hector, A.; Parkin, I. P. Polyhedron 1993, 12, 1855-1862.
(13) Rao, L.; Kaner, R. B. Inorg. Chem. 1994, 33, 3210-3211.
(14) Wallace, C. H.; Reynolds, T. K.; Kaner, R. B. Chem. Mater. 1999,
11, 2299-2301.
(15) Fitzmaurice, J. C.; Hector, A.; Parkin, I. P. Polyhedron 1993, 12,
1295-1300.
(16) Jarvis, R. F.; Jacubinas, R. M.; Kaner, R. B. Inorg. Chem. 2000, 39,
3243-3246.
(17) Ponthieu, E.; Rao, L.; Gengembre, L.; Grimblot, J.; Kaner, R. B. Solid
State Ionics 1993, 63-5, 116-121.
(18) Corbett, J. D.; McMullan, R. K. J. Am. Chem. Soc. 1955, 77, 4217-
4219.
(20) Bowen, P.; Highfield, J. G.; Mocellin, A.; Ring, T. A. J. Am. Ceram.
Soc. 1990, 73, 724-728.
(21) Krnel, K.; Kosmac, T. J. Eur. Ceram. Soc. 2001, 21, 2075-2079.
(19) Bonneau, P. R.; Wiley, J. B.; Kaner, R. B. Inorganic Synthesis 1995,
30, 33-37.
Inorganic Chemistry, Vol. 42, No. 8, 2003 2715

Janes et al.
precursor during reaction. Alumina, on the other hand, likely
forms as a result of the aqueous wash, either from unreacted
precursors or from hydrolysis of small, poorly formed AlN
particles that are more susceptible to oxidation. Note that
alumina is not observed in powder X-ray diffraction patterns
taken before washing, but Al metal is.
Changing the precursor from AlCl3 to AlI3 results in
products and impurities similar to those found with AlCl3.
Because the melting point of AlI3 is 461 Ksversus 453 K
for the sublimation temperature of AlCl3sthe reaction
initiates at nearly the same temperature. A comparable Tmax
is observed, along with aluminum and alumina impurities.
To achieve greater nitridation of the product, Al2S3 was
substituted for AlCl3. The idea is that, because Al2S3 has an
anion with a doubly negative charge, a more stable byproduct
salt, Li2S, is expected. The ∆Hf for Li2S is -446.9 kJ/mol,
compared to a ∆Hf of -408.4 kJ/mol for LiCl.23 Aluminum
sulfide was mixed and ground together with Li3N in an agate
mortar and pestle, the mixture was placed in the stainless
steel reactor, and the reaction was initiated with a heated
filament. The products were then removed from the He
atmosphere drybox, filtered, and dried in an oven at 450 K
(Figure 1, middle). The AlN produced from the reaction of
Al2S3 and Li3N is slightly more crystalline than that formed
from AlCl3 and Li3N (Figure 1, top), and the Al2O3 impurity
is eliminated. However, unwanted Al metal still presents a
problem.
Figure 1. Powder X-ray diffraction patterns of the washed products from
the following solid-state metathesis reactions that produce AlN: AlCl3 +
Li3N (top), 0.5 Al2S3 + Li3N (middle), and AlCl3 + 0.5 Ca3N2 (bottom).
The Miller indices (hkl) for hexagonal AlN are given in the highly crystalline
pattern produced by reacting AlCl3 with Ca3N2 (bottom). Aluminum (#)
and alumina (*) impurities are found in the other reactions.
The persistence of metallic Al in the product at first
suggested that more nitriding agent might be needed. Thus,
several nitrogen-containing salts including NH4Cl, NaN3, and
LiNH2 were added into the reactions. As the reactions release
heat after initiation, these salts decompose, and can supply
either nitrogen or ammonia to aid in nitride formation.14,24
However, instead of increasing the crystallinity of the nitride
product, all of these additives resulted in less-crystalline
products with more impurities. Adding an excess of nitride
starting material also yielded only negative results. The
commonality among these attempts to improve the reactions
is that excess reagents act as heat sinks and lower reaction
temperatures. Because this leads to less-crystalline products,
a method was sought that would instead raise the reaction
temperature. Aside from the possibility that AlN might
simply form more readily at elevated temperatures, higher
temperatures have other advantages, including more energy
available to surmount activation barriers and an increased
probability of reactants finding each other in the short time
of reaction. Additionally, once the product forms, crystal
growth will continue as long as the byproduct salt remains
molten due to Ostwald ripening.25,26 Because hotter reactions
take longer to cool, more time will be available for crystal
growth, generally leading to higher crystallinity along with
greater yields.
reaction.22 Once initiated, the heat generated by salt formation
needs to be sufficient to keep the salt produced in a molten
state. If this is the case, the reaction will be self-propagating.10
Theoretical reaction temperatures can be calculated using the
enthalpy of reaction (∆Hrxn), temperature-dependent heat
capacities (Cp), and enthalpies of phase changes. The
adiabatic maximum reaction temperatures, referred to as Tmax
,
assume that reactions proceed to completion with no heat
loss.
The synthesis of AlN was approached by considering
several possible precursors, including aluminum halides or
chalcogenides in combination with alkali or alkaline earth
nitrides. Figure 1 shows a progression of products from a
low-quality nitride contaminated with both aluminum metal
and aluminum oxides to a phase-pure, high-quality nitride.
The metal chloride (AlCl3) was chosen first because of
its relatively low cost and low sublimation temperature (453
K). AlCl3 was ground together with Li3N, the mixture was
placed in a stainless steel reactor, and the reaction was
initiated with a Nichrome wire (eq 1). This reaction produces
moderately crystalline AlN (Figure 1, top); however, there
is contamination from Al metal and alumina. The aluminum
impurity most likely results from reduction of the AlCl3
(23) Chase, M. W., Jr.; Daview, C. A.; Downey, J. R., Jr.; Frurip, D. J.;
McDonald, R. A.; Syverud, A. N. JANAF Thermochemical Tables;
National Bureau of Standards: Washington, DC, 1985; Vol. 14.
(24) O’Loughlin, J. L.; Wallace, C. H.; Knox, M. S.; Kaner, R. B. Inorg.
Chem. 2001, 40, 2240-2245.
(25) Ostwald, W. Z. Phys. Chem. 1901, 37, 385.
(26) Baldan, A. J. Mater. Sci. 2002, 37, 2171-2202.
(22) Treece, R. E.; Gillan, E. G.; Kaner, R. B. Comments Inorg. Chem.
1995, 16, 313-337.
2716 Inorganic Chemistry, Vol. 42, No. 8, 2003

Metathesis Routes to Aluminum Nitride
Table 1. Calculated and Measured Reaction Temperatures of
AlN-Producing Solid-State Metathesis Reactions
Tmax,calc
Tmax,exp
Tmax,salt
reaction
(K)
(K)
(K)
AlCl3 + 0.5Ca3N2 f AlN + 1.5CaCl2
2208
1673
1427
(2010)a
1513
AlCl3 + Li3N f AlN + 3LiCl
0.5Al2S3 + Li3N f AlN + 1.5Li2S
1656
2073
1503
1086
1373
a Reaction carried out in insulated ceramic reactor.
Using Al2S3 but changing the nitriding agent to Ca3N2 was
expected to achieve a higher reaction temperature with the
possibility of producing AlN with enhanced crystallinity
because the calcium sulfide byproduct salt is very stable (∆Hf
) -473.2 kJ/mol). Unfortunately, the reaction of Ca3N2 and
Al2S3 did not propagate. This is likely related to the very
high melting point of the byproduct salt CaS (mp ) 2524
K). Previous experiments have shown that, if the energy
generated by salt formation only is insufficient to keep the
salt produced in a molten state, propagation will cease. This
value, called Tmax,salt, was calculated for each reaction and
presented in Table 1. For the reaction between Al2S3 and
Ca3N2, Tmax,salt is 1202 K, well below the melting point of
CaS, and therefore, it is not surprising that this reaction does
not self-propagate.
The aluminum precursor was therefore changed back to
AlCl3 while maintaining Ca3N2 as the nitriding agent. The
maximum adiabatic temperature predicted for the reaction
of Ca3N2 and AlCl3 is 2208 Ksover 550 K above the value
calculated for the reaction between Li3N and AlCl3. This
predicted increase in reaction temperature is mainly due to
the higher thermodynamic stability of the byproduct salt
CaCl2 (∆Hf ) -795 kJ/mol) versus LiCl (∆Hf ) -408.4
kJ/mol).23 Because the value for Tmax,salt (1427 K) is now
above the melting point of the CaCl2 salt byproduct (1045
K), this reaction is expected to propagate. As hoped, this
reaction not only propagates rapidly but succeeds in produc-
ing single-phase, crystalline hexagonal AlN (Figure 1,
bottom). A least-squares refinement of the X-ray data gives
lattice parameters of a ) 3.1090 Å and c ) 4.9749 Å
((0.0012 Å), which closely matches the literature values of
a ) 3.1114 Å and c ) 4.9792 Å (JCPDS no. 25-1133). The
Al content of the AlN is 48 mol % (50% expected) based
on thermogravimetric analysis carried out in air to 1773 K,
which converts all of the AlN to Al2O3 (corundum). The
slightly lower than expected Al content might be due to some
surface phosphate left from the acid wash. The yield of the
product is typically about 80%, based on a reaction scale of
0.2 g. A larger reaction scale is likely to increase the yield
significantly.27 At the 0.2-g reaction scale, mechanical losses
of product due to transfers and material trapped in the filter
paper is significant. The relative amount of these losses
would be less important at larger scales. However, caution
is advised in scaling up these highly exothermic reactions.
To measure the actual temperatures achieved in the AlN-
forming reactions, type C thermocouples were inserted
directly into the stainless steel reaction vessels. Temperature
Figure 2. In situ temperature measurements for the following AlN
reactions: (a) AlCl3 + 0.5 Ca3N2 in an insulating cup, (b) AlCl3 + 0.5
Ca3N2, (c) AlCl3 + Li3N, and (d) 0.5 Al2S3 + Li3N.
profiles for each reaction are presented in Figure 2. As
expected, the least energetic reaction, Al2S3 + Li3N (Figure
2d), produced the lowest Tmax (1373 K). The reaction of AlCl3
with Li3N increased the measured Tmax to 1513 K (Figure
2c). Switching from Li3N to Ca3N2 resulted in a further
increase in Tmax to 1673 K, which was reached 0.8 s after
initiation of the reaction (Figure 2b).
Although the reaction with calcium nitride reached higher
temperatures than the reactions with Li3N, as expected from
the Tmax calculations (see Table 1), it is apparent that the
measured temperatures are considerably lower than the
calculated theoretical maximum temperatures. This is most
likely due to nonadiabatic conditions. The high thermal
conductivity of the stainless steel reaction vessel suggested
that this might be largely responsible. Thus, in the hope of
improving the experimental Tmax value, the reaction of AlCl3
+ Ca3N2 was repeated in a thermally insulating ceramic cup.
The result was an increase of over 300 K in Tmax to 2010 K.
This result is much closer to the calculated adiabatic value
of 2208 K. Additionally, an increase in the crystallinity of
the product was observed (Figure 1, bottom). Because slower
heat dissipation results in the product remaining in the molten
salt matrix for longer times, more Ostwald ripening, and
hence enhanced crystallinity, occurs. This, in turn, makes
the AlN product more resistant to hydrolysis in water.
From the in situ temperature measurements, it appears that
Tmax g 1673 K is needed to produce phase-pure AlN via
metathesis reactions. Either products formed in cooler
reactions never attain enough heat to produce AlN or the
low crystallinity of the AlN particles formed make them
vulnerable to attack during the aqueous wash. These results
explain why Ca3N2 is preferable to Li3N as a nitriding agent
for reactions with aluminum halides.
Thermal dissipation explains why the Tmax results are lower
than predicted for AlCl3 + Ca3N2 and AlCl3 + Li3N, but
does not entirely describe the case of Al2S3, where in situ
measurements are found to be only about one-half of the
calculated adiabatic values. Because the Tmax,salt of this
reaction is less than the melting point of Li2S (the byproduct
salt), this reaction was not expected to propagate. However,
some AlN is produced through this reaction. To understand
(27) Gillan, E. G.; Kaner, R. B. Inorg. Chem. 1994, 33, 5693-5700.
Inorganic Chemistry, Vol. 42, No. 8, 2003 2717

Janes et al.
Figure 3. Scanning electron microscopy (SEM) images of the AlN product formed in the reaction between AlCl3 and Ca3N2: (a) before acid washing, the
product appears as a melt of CaCl2, and (b) after acid washing, pure AlN appears as a fine micron-sized powder.
why this occurs, it is worth considering the temperatures at
which the reactants change phase. AlCl3 sublimes at ap-
proximately 453 K, whereas Al2S3 does not melt until 1645
K.23,28 Because the Nichrome wire detonator heats to near
1100 K when current is run through it, there is ample energy
to vaporize some of the molecular solid AlCl3, but not the
extended network structure of Al2S3, which remains in solid
form. Thus, the stability of Al2S3 leads to an incomplete
reaction with Li3N, while the high melting point of Li2S (the
byproduct salt) prevents self-propagation because a molten
salt does not form (the melting point for Li2S > Tmax,salt).
This explains why the observed Tmax value is much lower
than expected.
The morphologies of the products were characterized using
scanning electron microscopy (SEM) and transmission
electron microscopy (TEM). An SEM image of the AlCl3 +
Ca3N2 reaction product is shown before (Figure 3a) and after
(Figure 3b) washing in acid. Figure 3a demonstrates that,
before the byproduct CaCl2 is washed away, the product
appears as a molten salt (CaCl2). After being washed in 1
M phosphoric acid to remove the CaCl2 byproduct salt, the
product appears as expectedsa white, fluffy, micron-sized
powder. The AlN product is composed of well-formed,
micron-sized crystallites as imaged with TEM (Figure 4).
The edges are well-defined, thus demonstrating that hydroly-
sis has not destroyed the surface morphology of the sample.
Conclusions
Figure 4. Transmission electron microscopy (TEM) image of representa-
tive AlN product demonstrating well-formed crystallites and micron-sized
A solid-state metathesis reaction between AlCl3 and Ca3N2
is shown to be a fast, reliable method for producing high-
quality aluminum nitride. The driving force behind the
reaction is the formation of a thermodynamically stable
byproduct salt that is easily removed from the nitride product
domains.
through an acid wash/filtration process. This highly exo-
thermic reaction reaches a temperature above 2100 K within
a fraction of a second and then quickly cools to room
temperature. Theoretical maximum reaction temperature
(28) Lide, D. R. CRC Handbook of Chemistry and Physics, 3rd electronic
ed.; CRC Press: Boca Raton, FL, 2001.
2718 Inorganic Chemistry, Vol. 42, No. 8, 2003

Metathesis Routes to Aluminum Nitride
calculations based on thermodynamic data and in situ
temperature measurements were used to gain an understand-
ing of these metathesis reactions. Higher sustained reaction
temperatures are necessary for the formation of phase-pure
AlN. At lower temperatures, incomplete nitridation leads to
aluminum impurities as well as aluminum oxide, which forms
upon washing. Extension of this method to produce other
nitrides using Ca3N2 as the nitriding agent along with a
thermally insulating ceramic reaction vessel is now in
progress.
Acknowledgment. The authors thank Richard G. Blair
for his help with the in situ temperature measurements. This
work was supported by the National Science Foundation,
Grant DMR-0073581.
IC026143Z
Inorganic Chemistry, Vol. 42, No. 8, 2003 2719