Kinetic instability of nanocrystalline aluminum prepared by chemical synthesis; facile room-temperature grain growth

Article abstract of DOI:10.1021/ja981972y

Nanocrystalline aluminum (nano-Al) is synthesized by two chemical methods. Method A consists of the following: reaction of LiAlH₄ and AlCl₃ at 164°C in 1,3,5-trimethylbenzene produces nano-Al with an average mean coherence length (crystallite size) of 160 ± 50 nm. The byproduct LiCl is removed by washing with MeOH at -25 or 0°C. Method B consists of the following: nano-Al is produced by decomposition of H₃Al(NMe₂Et) under reflux in 1,3,5-trimethylbenzene (ca. 100-164°C), with or without added decomposition catalyst Ti(O-i-Pr)₄. Here the mean particle size (40-180 nm) and degree of aggregation of the nano-Al depend on the mole percent of decomposition catalyst used (0 - 1%). The nano-Al produced by method A contains 3-4 wt % each of C, O, and Cl; whereas that produced by method B contains only ≤0.25 wt % each of C, O, and Cl and is >99 wt % Al. nano-Al produced by both methods has been consolidated by uniaxial pressing at 350 MPa for 1 h at 25, 100, or 300°C. Rapid grain growth is observed at each pressing temperature in method-B powders. The mean grain size doubles at 25°C within 1 h and continues to increase beyond the nanometer-size regime over longer periods. The low-temperature grain-size instability of the chemically synthesized (method-B) nano-Al contrasts markedly with the high- temperature stability (≥300°C) of nano-Al prepared by gas condensation or mechanical attrition. Facile grain growth in the chemically synthesized (method-B) nano-Al is attributed to higher grain-boundary purity and to aggregate structures that minimize adventitious oxidation. The results establish that barriers to grain growth in pure nano-Al are intrinsically low.

Full text of DOI:10.1021/ja981972y

J. Am. Chem. Soc. 1998, 120, 10847-10855
10847
Kinetic Instability of Nanocrystalline Aluminum Prepared by
Chemical Synthesis; Facile Room-Temperature Grain Growth
Joel A. Haber and William E. Buhro*
Contribution from the Department of Chemistry, Washington UniVersity, St. Louis, Missouri 63130-4899
ReceiVed June 5, 1998
Abstract: Nanocrystalline aluminum (nano-Al) is synthesized by two chemical methods. Method A consists
of the following: reaction of LiAlH4 and AlCl3 at 164 °C in 1,3,5-trimethylbenzene produces nano-Al with
an average mean coherence length (crystallite size) of 160 ( 50 nm. The byproduct LiCl is removed by
washing with MeOH at -25 or 0 °C. Method B consists of the following: nano-Al is produced by
decomposition of H3Al(NMe2Et) under reflux in 1,3,5-trimethylbenzene (ca. 100-164 °C), with or without
added decomposition catalyst Ti(O-i-Pr)4. Here the mean particle size (40-180 nm) and degree of aggregation
of the nano-Al depend on the mole percent of decomposition catalyst used (0-1%). The nano-Al produced
by method A contains 3-4 wt % each of C, O, and Cl; whereas that produced by method B contains only
e0.25 wt % each of C, O, and Cl and is g99 wt % Al. nano-Al produced by both methods has been
consolidated by uniaxial pressing at 350 MPa for 1 h at 25, 100, or 300 °C. Rapid grain growth is observed
at each pressing temperature in method-B powders. The mean grain size doubles at 25 °C within 1 h and
continues to increase beyond the nanometer-size regime over longer periods. The low-temperature grain-size
instability of the chemically synthesized (method-B) nano-Al contrasts markedly with the high-temperature
stability (g300 °C) of nano-Al prepared by gas condensation or mechanical attrition. Facile grain growth in
the chemically synthesized (method-B) nano-Al is attributed to higher grain-boundary purity and to aggregate
structures that minimize adventitious oxidation. The results establish that barriers to grain growth in pure
nano-Al are intrinsically low.
Introduction
Nanocrystalline metals and ceramics are expected to exhibit
improved mechanical properties over conventional coarse-
grained materials.1,2,9-13 Reduction of grain dimensions into
the nanometer regime may provide a means to simultaneously
strengthen1,9,10,12,13 and toughen1,9,12,13 structural materials. In
conventional coarse-grained materials having grain dimensions
of tens to hundreds of micrometers, increases in strength are
generally achieved at the expense of fracture toughness and vice
versa.14 However, predicted changes in deformation and
fracture mechanisms should effect increased strength with
retention of ductility and toughness (or increased ductility and
toughness with retention of strength) in fully dense nanocrys-
talline materials having grain dimensions of e100 nm.1,9,12,13
These and related expectations have prompted studies of a large
number of nanocrystalline ceramics, intermetallics, and metals,
with emphasis on how their properties change as a function of
grain size.1,2,9-13
We show that the intrinsic barrier to grain (crystallite) growth
in nanocrystalline aluminum (nano-Al) is low. Our results add
to the accumulating evidence that pure nanocrystalline metals
are not inherently kinetically stable.1-8 The kinetic stability
often observed likely results from extrinsic origins such as grain-
boundary impurities1,2 rather than from new physical phenomena
emerging in ultrafine-grained materials.9-13 The nano-Al we
prepared by chemical synthesis has sufficiently low impurity
content that intrinsic grain-growth behavior is enabled.
(1) Morris, D. G.; Morris, M. A. Materials Science Forum 1997, 235-
228, 861-872.
(2) Malow, T. R.; Koch, C. C. Mater. Sci. Forum 1996, 225-227, 595-
604.
(3) Gu¨nther, B.; Kumpmann, A.; Kunze, H.-D. Scr. Metall. Mater. 1992,
27, 833-838.
(4) Kumpmann, A.; Gu¨nther, B.; Kunze, H.-D. Mater. Sci. Eng., A 1993,
A168, 165-169.
The predicted property enhancements in nanocrystalline
metals have been only partially realized. Nanocrystalline metals
have hardnesses that are 2-7 times greater9 and compressive
strengths that are similarly greater1,17,18 than those of the same
coarse-grained metals. However, to date nanocrystalline metals
have typically been weak and brittle under tensile loading.1,9
(5) Gertsman, V. Y.; Birringer, R. Scr. Metall. Mater. 1994, 30, 577-
581.
(6) Lian, J.; Valiev, R. Z.; Baudelet, B. Acta Metall. Mater. 1995, 43,
4165-4170.
(7) Valiev, R. Z.; Kozlov, E. V.; Ivanov, Yu. F.; Lian, J.; Nazarov, A.
A.; Baudelet, B. Acta Metall. Mater. 1994, 42, 2467-2475.
(8) Akhmadeev, N. A.; Kobelev, N. P.; Mulyukov, R. R.; Soifer, Ya.
M.; Valiev, R. Z. Acta Metall. Mater. 1993, 41, 1041-1046.
(9) Siegel, R. W. Mater. Sci. Forum 1997, 235-228, 851-860.
(10) Gryaznov, V. G.; Trusov, L. I. Prog. Mater. Sci. 1993, 37, 289-
401.
(14) Strength and fracture toughness in conventional materials depend
largely on dislocation mobility. Dislocations are immobile in strong, brittle
materials such as ceramics. Dislocations are very mobile in ductile, soft
but tough metals. Chemical or microstructural changes that limit dislocation
mobility strengthen but embrittle metals.12,13,15,16
(15) Gordon, J. E. The New Science of Strong Materials, 2nd ed.;
Princeton University Press: Princeton, 1988; pp 91-98, 117, 209-230.
(16) Geselbracht, M. J.; Ellis, A. B.; Penn, R. L.; Lisensky, G. C.; Stone,
D. S. J. Chem. Educ. 1994, 71, 254-261.
(11) Gleiter, H. Prog. Mater. Sci. 1989, 33, 223-315.
(12) Siegel, R. W.; Fougere, G. E. Nanostruct. Mater. 1995, 6, 205-
216.
(13) Siegel, R. W.; Fougere, G. E. In Nanophase Materials, Synthesis,
Properties, Applications; Hadjipanayis, G. C., Siegel, R. W., Eds.; NATO
ASI Series, Vol. 260; Kluwer Academic: Dordrecht, 1994; pp 233-261.
10.1021/ja981972y CCC: $15.00 © 1998 American Chemical Society
Published on Web 10/13/1998

10848 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Haber and Buhro
Neither hardness nor compressive strength is very sensitive to
the porosity or other internal flaws present in nanocrystalline
consolidates. Most nanocrystalline consolidates are obtained
by compaction of fine powders consisting of nanometer-sized
crystallites in micrometer-sized aggregates, which resist con-
solidation to full density. Densities of only 75-95% of
theoretical values are typically achieved; considerable nano- and
microporosity remain.1 Tensile strength, ductility, and fracture
toughness are sensitive to porosity and related flaws. Therefore,
improvements in consolidation or other means of reducing
residual porosity are required to allow the ductility and fracture-
toughness expectations to be properly tested.1 Elucidation of
the mechanisms governing hardening, plastic deformation,
fracture, and microstructural stability of nanostructured materials
continues to be an active research area of fundamental impor-
tance.1,2,9,10,11,19
and thermal processing.27-31 In some cases the ultrafine-grained
microstructure is retained after melting.30,31
However, as described herein the nano-Al prepared by
chemical synthesis undergoes rapid coarsening upon room-
temperature consolidation. We attribute the facile grain growth
to the purity of the chemically produced powders and to
favorable aggregate structures that protect nanocrystallite sur-
faces from adventitious oxidation. To our knowledge, this work
constitutes the first isolation of nano-Al from chemical syn-
theses.36,37
Results
Preparative Method A. Reduction of AlCl3 with LiAlH4
afforded nano-Al according to eq 1. The nano-Al was obtained
1,3,5-Me3C6H3
3LiAlH4 + AlCl3
8
Retention of properties due to nanoscale grains requires grain-
size stability. Nanocrystalline materials are thermodynamically
unstable because large fractions of the constituent atoms reside
in (high-energy) grain boundaries.10,11,20-22 Grain growth, which
decreases grain-boundary volume, decreases total free energy
substantially. However, most nanocrystalline metals and ceram-
ics behave as though they are deeply metastable and resist grain
growth to high temperatures.10,11 At issue is whether high
barriers to grain growth are intrinsic12,13,20,21 or extrinsic1,2,22 in
origin. Solute or impurity segregation at grain boundaries, grain-
boundary pinning by additive or impurity (second-phase)
particles (Zener drag23), and pores are effective impediments
to grain growth.1,2 Recent observations of low-temperature grain
growth in nanocrystalline copper,3-8 silver,3,4 and palladium3,4
suggest that intrinsic barriers to grain growth in nanocrystalline
metals are actually low.
164 °C
4nano-Al + 3LiCl + 6H2 (1)
as a mixture with the LiCl byproduct, which was removed by
extraction with cold (0 °C) MeOH. We succeeded in extracting
up to 2.5 g of the mixture in a single operation, provided that
the temperature was maintained between -25 and 0 °C.
However, attempts to extract larger quantities resulted in
runaway exothermic oxidation of nano-Al by MeOH. Conse-
quently, method A (eq 1) was inconvenient for production of
large quantities of nano-Al. Additionally, the nano-Al contained
significant amounts of C, O, and Cl as determined by analysis
of a pellet consolidated from method-A powder (see below).
We surmised that eq 1 proceeded by the intermediate
formation and decomposition of alane, AlH3. Alane may be
isolated as an Et2O adduct from reactions of LiAlH4 and AlCl3
(as in eq 1) in Et2O near room temperature.38,39 Stable alane-
amine adducts have been prepared similarly.40 We expected
that LiCl-free nano-Al would be produced by decomposition
of an appropriate, purified alane adduct, H3Al(L). The adduct
H3Al(NMe2Et) was selected for study because it has proven
convenient for the chemical vapor deposition of Al films at
g100 °C.41
Nanocrystalline aluminum has presented a dramatic coun-
terexample to the facile coarsening found in nanocrystalline
copper, silver, and palladium.24 Samples produced by inert-
gas condensation26-31 and mechanical attrition (high-energy ball
milling)31-35 undergo minimal grain growth during consolidation
(17) Suryanarayanan, R.; Frey, C. A.; Sastry, S. M. L.; Waller, B. E.;
Buhro, W. E. In Processing and Properties of Nanocrystalline Materials;
Suryanarayana, C., Singh, J., Froes, F. H., Eds.; Minerals, Metals and
Materials Society: Warrendale, PA, 1996; pp 407-413.
(18) Suryanarayanan, R.; Frey, C. A.; Sastry, S. M. L.; Waller, B. E.;
Bates, S. E.; Buhro, W. E. J. Mater. Res. 1996, 11, 439-448.
(19) Schiøtz, J.; Di Tolla, F. D.; Jacobsen, K. W. Nature 1998, 391,
561-563.
Preparative Method B. H3Al(NMe2Et) was prepared by the
method of Frigo and co-workers.40 We found that H3Al(NMe2-
Et) decomposed under reflux in 1,3,5-trimethylbenzene solution
with or without added decomposition catalyst39 Ti(O-i-Pr)4
according to eq 2. In the absence of Ti(O-i-Pr)4 decomposition
(20) Rivier, N. In Physics and Chemistry of Finite Systems: From
Clusters to Crystals; Jena, P., Khanna, S. N., Rao, B. K., Eds.; Kluwer
Academic: Dordrecht, 1992; pp 189-198.
1,3,5-Me3C6H3
H3Al(NMe2Et) w/ or w/o cat. 8
(21) Fecht, H. J. Phys. ReV. Lett. 1990, 65, 610-613.
(22) Ga¨rtner, F.; Bormann, R.; Birringer, R.; Tscho¨pe, A. Scr. Metall.
Mater. 1996, 35, 805-810.
Ti(O-i-Pr)4
e164 °C
nano-Al + 3/2H2 + NMe2Et (2)
(23) Morris, D. G.; Morris, M. A. Mater. Sci. Eng., A 1991, A134, 1418-
1421.
(24) Birringer indicated in a review25 that nanocrystalline aluminum
underwent grain growth at room temperature, but to our knowledge, the
experiments upon which that conclusion was based have not been described.
Our current results agree with Birringer’s indication.
(25) Birringer, R. Mater. Sci. Eng., A 1989, A117, 33-43.
(26) Hahn, H.; Averback, R. S. J. Appl. Phys. 1990, 67(2), 1113-1115.
(27) Gutmanas, E. Y.; Trusov, L. I.; Gotman, I. Nanostruct. Mater. 1994,
4, 893-901.
(33) Al Sadi, A.; Bonetti, E.; Mattioli, P.; Valdre`, G. J. Alloys Compd.
1994, 211/212, 489-493.
(34) Bonetti, E.; Scipione, G.; Valdre`, G.; Enzo, S.; Frattini, R.; Macr`ı,
P. P. J. Mater. Sci. 1995, 30, 2220-2226.
(35) Oleszak, D.; Shingu, P. H. J. Appl. Phys. 1996, 79, 2975-2980.
(36) Rieke, R. D.; Chao, L. Synth. React. Inorg. Met.-Org. Chem. 1974,
4, 101.
(28) Qin, X. Y.; Wu, X. J.; Cheng, L. F. Nanostruct. Mater. 1993, 2,
99-108.
(37) Tsai, K.-L.; Dye, J. L. J. Am. Chem. Soc. 1991, 113, 1650-1652.
(38) Finholt, A. E.; Bond, A. C.; Schlesinger, H. I. J. Am. Chem. Soc.
1947, 69, 1199-1203.
(29) Wu¨rschum, R.; Greiner, W.; Schaefer, H.-E. Nanostruct. Mater.
1993, 2, 55-62.
(39) Brower, F. M.; Matzek, N. E.; Reigler, P. F.; Rinn, H. W.; Roberts,
C. B.; Schmidt, D. L.; Snover, J. A.; Terada, K. J. Am. Chem. Soc. 1976,
98, 2450-2453.
(30) Sa´nchez-Lo´pez, J. C.; Ferna´ndez, A.; Conde, C. F.; Morant, C.;
Sanz, J. M. Nanostruct. Mater. 1996, 7, 813-822.
(31) Eckert, J.; Holzer, J. C.; Ahn, C. C.; Fu, Z.; Johnson, W. L.
Nanostruct. Mater. 1993, 2, 407-413.
(40) Frigo, D. M.; van Eijden, G. J. M.; Reuvers, P. J.; Smit, C. J. Chem.
Mater. 1994, 6, 190-195.
(32) Eckert, J.; Holzer, J. C.; Krill, J. C., III.; Johnson, W. L. J. Mater.
Res. 1992, 7, 1751-1761.
(41) Simmonds, M. G.; Phillips, E. C.; Hwang, J.-I.; Gladfelter, W. L.
Chemtronics 1991, 5, 155-158.

Facile Room-Temperature Grain Growth
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10849
Figure 1. XRD patterns of method-A powders. (a) As precipitated:
(1) nano-Al, (2) LiCl, (3) LiCl‚H2O. (b) After MeOH wash; nano-Al
reflections are labeled by their Miller indices, hkl.
Figure 2. Aggregates of the two types found in method-A nano-Al
powders: (a) larger faceted crystallites from a sample having an XRD
coherence length of ∼200 nm; (b) smaller rounded crystallites from a
sample having an XRD coherence length of ∼100 nm.
of H3Al(NMe2Et) occurred at ca. 164 °C. The resulting nano-
Al was obtained as a spongy mass coating the bottom of the
reaction vessel, which could be ground in a mortar and pestle
into coarse flakes. In the presence of catalytic amounts of Ti-
(O-i-Pr)4, decomposition of H3Al(NMe2Et) occurred well below
164 °C, and the nano-Al was precipitated as a powder. The
spongy and powder forms of the product suggested that the
decomposition occurred upon the heated vessel surface without
added catalyst and in solution dispersion with added catalyst.
The purity of method-B powder was considerably greater than
the purity of method-A powder (see below).
Characterization of nano-Al Produced by Method A. An
X-ray powder diffraction (XRD) pattern of the eq 1 product
prior to extraction contained reflections for LiCl and nano-Al
(Figure 1a). After MeOH extraction of the powder, nano-Al
was the only crystalline phase detected (Figure 1b), indicating
that the C-,O-, and Cl-containing impurities present were poorly
crystalline or in amounts below the (few %) XRD detection
limit. The mean coherence length determined by Scherrer
analysis of XRD line widths (see the Experimental Section)
varied from trial to trial over the range of 90 to >200 nm. We
were unable to identify the reaction parameter responsible for
this large variation. The average in the mean coherence length
was 160 ( 50 nm over sixteen trials, establishing Al crystallite
sizes of that approximate magnitude and size distribution.
Microscopic investigation of the method-A powder revealed
aggregates of small Al crystallites and larger nonaggregated
polyhedral Al crystallites. Scanning electron microscopy (SEM)
images established a powder microstructure consisting mainly
of particle aggregates having dimensions of 3-150 µm. These
large aggregates were partially dismantled and suspended in
pyridine in an ultrasonic cleaning bath. Transmission electron
microscopy (TEM) images of the suspended material showed
smaller residual aggregates of faceted (Figure 2a) and rounded
(Figure 2b) particles. The dimensions of the particles were
consistent with the mean XRD coherence length (160 ( 50 nm,
Figure 2b). The TEM images also consistently revealed
minority fractions of nonaggregated polyhedral crystallites
including tetrahedra, octahedra, and cubooctahedra (Figure 3).
Although the method-A powders contained broad distributions
of particle sizes as evidenced by Figures 2 and 3, individual
aggregates tended to contain similarly sized particles.
Figure 3. Nonaggregated polyhedral crystallites found as a minor
fraction in method-A and method-B nano-Al powders: (a) and (h)
cubooctahedra, (b) tetrahedron, (c) tetrahedron and cuboctahedron with
rounded edges, (d) and (e) octahedra, (f) and (g) other polyhedra.
2 depended upon the amount of Ti(O-i-Pr)4 decomposition
catalyst used (see Table 1 and Figure 4). When no catalyst
was added, the mean coherence lengths were 44-82 nm.
Addition of 0.01 mol % of Ti(O-i-Pr)4 catalyst caused the mean
coherence length to increase to 197 nm. The mean coherence
length then decreased with increasing amount of catalyst to 50-
75 nm at 1 mol % catalyst (Table 1 and Figure 4). The initial
jump in the nano-Al coherence length with added catalyst was
ascribed to the change in the mechanism of H3Al(NMe2Et)
decomposition noted previously. The decrease in coherence
length with increasing amount of catalyst was ascribed to an
increasing crystallite nucleation rate.
In general, with decreasing size the particles adopted non-
faceted morphologies and, as expected,31 formed larger, more
densely packed aggregates. The large flakes produced by
grinding the spongy eq 2 product obtained without added
catalyst consisted of dense aggregates 1-2 µm in diameter with
nanosized features and loose aggregates of rounded nanometer-
sized particles (Figure 5a). As above, the aggregates were
partially dismantled by sonication in pyridine. TEM images
Characterization of nano-Al Produced by Method B. The
crystallite sizes, estimated by mean XRD coherence lengths,
and the degree of aggregation of the nano-Al produced by eq

10850 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Haber and Buhro
Table 1. Mean Coherence Lengths of Nano-Al Prepared by
Method B
concentration of mol % Ti(O-
Scherrer (XRD)
powder H3Al(NMe2Et)a
i-Pr)4
coherence lengthc (nm)
b
B1
B2
B3
B4
B5
B6
B7
B8
2
2.4
2.4
2.5
2.0-2.4
1
3.2-4.0
2.4
2.6
2.4
2.6
1.6-2.1
2.5-3.0
2.5-3.0
2.4
2.4
2.5
1.6-2.1
0
0
0
0
45 ( 3
67 ( 4
82 ( 20
48 ( 2
0.0035-0.014
0.07-0.14
0.16
0.17
0.18
0.20
0.29
0.41
0.58
0.51-0.67
0.64
0.71
0.97
197 ( 20
128 ( 11
116 ( 10
97 ( 7
B9
128 ( 1
95 ( 8
B10
B11
B12
B13
B14
B15
B16
B17
B18
108 ( 5
73 ( 7
48 ( 8
62 ( 11
74 ( 4
50 ( 10
62 ( 6
1.15
99 ( 16
a Moles of H3Al(NMe2Et) per L of 1,3,5-trimethylbenzene. b On the
basis of H3Al(NMe2Et). cCoherence lengths and errors were extracted
from the XRD data as described in the Experimental Section. Size and
strain contributions to the XRD line widths were deconvoluted using
the method applied to the data from consolidated pellets (see the
Experimental Section), but the resulting sizes were within the error
limits of the simple Scherrer values recorded here, indicating that strain
contributions were negligible.
Figure 5. SEM (left) and TEM (right) images of method-B nano-Al
powders. From samples prepared by eq 2 using: (a) and (b) 0 mol %,
(c) and (d) 0.007-0.29 mol %, (e) and (f) 0.64 mol %, and (g) and (h)
0.71 mol % Ti(O-i-Pr)4 catalyst.
larger particles (1.0-2.0 µm in diameter) making up these
aggregates were polycrystallites that in some cases exhibited
partially formed facets. The internal structure (polycrystallinity)
of the particles was established by TEM images of sonicated,
suspended samples (Figure 5f, h); tightly aggregated primary
crystallites sharing angular interfaces were evident (Figure 5h).
The primary crystallite sizes in Figure 5h are 50-200 nm.
The facets observed upon some of the aggregates (such as in
Figure 5e) likely indicate the onset of recrystallization and grain
growth within them. The faceted constituents of the aggregates
resemble the previously discussed polyhedral single crystallites
(Figure 3) obtained as minor product fractions in both method
A and B syntheses, suggesting that clearly observable facets
emerge only in larger crystallites having dimensions of ap-
proximately g200 nm.
Figure 4. Dependence of the XRD coherence length (a measure of
mean crystallite size) in method-B nano-Al powders upon the quantity
of Ti(O-i-Pr)4 catalyst used in eq 2.
(Figure 5b) of the suspended fraction revealed that the tight
aggregates were composed of 20-50 nm crystallites, consistent
with the XRD coherence length.
In contrast, the eq 2 products with the largest mean coherence
lengths, those produced using up to 0.29 mol % of catalyst,
were the finest powders and frequently passed through medium
porosity glass frits. SEM images (Figure 5c) indicated that these
fine powders consisted largely of faceted crystallites 150-500
nm in diameter. A minority fraction of 0.5-1.5 µm aggregates
having 20-50 nanometer-sized rounded features was also
present. TEM images (Figure 5d) of sonicated, suspended
material showed that the faceted crystallites were readily
dispersed and that the tight aggregates again consisted of 20-
50 nm diameter crystallites.
Characterization of Consolidated nano-Al. Pellets were
consolidated by uniaxial pressing for 1 h at 350 MPa and 25,
100, or 300 °C. The pressing was conducted in a hot press
assembled in an argon-filled glovebox to minimize adventitious
oxidation associated with powder transfers and die loading. The
pellets produced from both method-A and method-B powders
When greater than 0.41 mol % of catalyst was used in eq 2,
dense aggregates were again formed (Figure 5e, g), similar to
those from the catalyst-free eq 2 synthesis (Figure 5a). The

Facile Room-Temperature Grain Growth
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10851
Table 2. Properties of Consolidated nano-Al
pelleta
pellet mass (g)
Trb
% densityc
mean grain sized (nm)
rms straine (10-4
)
microhardness (MPa)
A-100
A-300
B13-100
B15-RT
B15-100
B15-300
1.27
0.77
3.53
0.75
0.79
0.75
0.40
0.61
0.40
0.32
0.40
0.61
84
89
90
75
88
95
270 ( 160
190 ( 50
>335
153 ( 14
168 ( 24
200 ( 16
3.6 (( 120%)
2.1 ((110%)
410 ( 20
550 ( 60
470 ( 14
560 ( 50
570 ( 40
760 ( 50
3.8 ((30%)
3.0 ((50%)
2.9 ((25%)
a The pellet label combines the powder synthesis method (and run), and the pressing temperature (in °C). b Reduced pressing temperature, Tr, is
the absolute pressing temperature divided by the absolute melting temperature of bulk Al; the reduced recrystallization temperature of bulk Al is
0.54.32 c On the basis of the density of bulk Al; we estimate a 3% error in these geometrically determined values. d Grain sizes and errors were
determined from XRD line widths according to eq 3 and as described in the Experimental Section. In one case, grain size was estimated as greater
than the coherence length for the least strain-broadened reflection, the 111. e Strains were determined from XRD line widths as described in the
Experimental Section.
were polished, without disintegration, to a metallic luster,
indicating at least moderate interparticle bonding. The density,
mean grain size, rms strain, and microhardness of several pellets
are given in Table 2. (Pellets were labeled according to the
source of the powder and the pressing temperature. Thus, pellet
B15-25 was pressed from method-B run-15 powder, see Table
1, at 25 °C.) As described below, density, porosity, and
microhardness appeared to correlate with the purity of the
starting nano-Al powder and the pressing temperature. The
most significant result was the observation of grain growth upon
consolidation of the pellets, even at room temperature (see
below).
Elemental analysis performed by glow-discharge mass spec-
trometry on portions of the pellets A-300 and B13-100
established that the nano-Al produced by method B was purer
by an order of magnitude than that produced by method A. Pellet
A-300 was only 87.0 wt % Al and contained 3.0, 4.0, and 3.3
Figure 6. SEM images at two magnifications of unpolished, air-
unexposed surfaces of pellets pressed from powder B15 (see Tables 1
and 2). Pressing temperature: (a) and (b) 25 °C, (c) and (d) 100 °C,
and (e) and (f) 300 °C. Note that the highest pressing temperature
removed micrometer-sized pores, but not nanometer-sized pores.
wt % of C, O, and Cl, respectively. Pellet B13-100 was 99 wt
% Al and contained only 0.23, 0.25, 0.14, and 0.32 wt % of C,
O, Cl, and Ti, respectively. Pellet A-300 contained 0.12 wt %
of Li, in significant molar deficiency relative to the residual
Cl. Thus, the Cl present was not due to substantial amounts of
residual LiCl, but more likely to unreacted AlCl3. Energy-
dispersive X-ray spectroscopy (EDS) conducted in the SEM on
pellets A-100 and A-300 gave compositional data consistent
with those from glow-discharge mass spectrometry. Further-
more, EDS on these samples established that Cl was distributed
heterogeneously, and in polished specimens that the Cl and O
impurities were concentrated near cracks and pores. Small
amounts (<0.3 atom %) of heterogeneously distributed Si were
also detected, which presumably derived from glass surfaces,
fritted disks, or silicone grease applied to stopcocks. EDS of
B15-25, B15-100, and B15-300 gave compositional data
consistent with those from glow-discharge mass spectrometry
on B13-100, except that Ti and O were present in slightly greater
amounts, possibly in part because of the slightly larger mol %
of Ti(O-i-Pr)4 used in the synthesis of the B15 powder. Only
30 to 50% of the Ti added in the catalyst was incorporated into
the nano-Al product; the rest of the Ti presumably remained
soluble and was removed in the washing step. The greater purity
of the method-B powders, revealed by analyses of the consoli-
dated pellets, was presumably due to the absence of Cl in the
method-B, H3Al(NMe2Et) precursor and the omission of the
MeOH wash used in the method-A synthesis.
lower densities of method-A pellets may have been due to
inhibition of densification by impurity phases.
SEM images of unpolished pellet surfaces showed that
porosity scaled with density. Pore sizes ranged from 5 µm to
< 500 nm in diameter. Pellet A-100 contained extensive µm-
scale porosity, whereas the denser pellet A-300 contained only
nanometer-scale pores. The pellets pressed from powder B15
showed the same trend (see Figure 6). Pellet B15-300 retained
some nanometer-scale porosity; thus, even the higher-purity
pellets pressed at relatively high temperature were not fully
dense. SEM images confirmed the trend in the geometrically
determined densities.
SEM images of polished pellet surfaces gave misleading
estimations of porosity. Polishing greatly increased the apparent
porosity by increasing the number of submicrometer-scale pores
observed. The polished pellets appeared to contain 30-50%
porosity by optical microscopy (metallography), which were
values inconsistent with the geometrically determined densities.
We believe that this apparent porosity increase resulted from
removal of particles and aggregates from the surface during
polishing and from reaction of the incompletely densified nano-
Al with the water used in polishing. Thus, the polishing media
used in conventional polishing techniques are not ideal for
incompletely densified, reactive nanostructured materials.
The Vickers microhardnesses of the pellets scaled roughly
with density (Table 2) and were greater than the microhardness
of pure, coarse-grained Al (290 MPa).32 Our microhardnesses
(410-760 MPa) were less than that of sintered aluminum
Pellet densities were measured at 75-95% of the theoretical
Al density (Table 2), and scaled roughly with powder purity
and pressing temperature. Pellets pressed from method-A
powders were less dense than those pressed from method-B
powders under the same conditions. The observed density range
was typical of consolidated nanocrystalline materials.1,2,9-11 The

10852 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Haber and Buhro
powder (1450 MPa),27 which contains a large volume fraction
of oxides, and less than the previously reported microhardness
of nano-Al (120-nm grain size) prepared by gas-phase conden-
sation (1700 MPa).27 Because of the ease of measurement,
microhardness is often used as an approximate indicator of
strength in nanocrystalline materials. However, microhardness
and tensile strength respond differently to flaw content and the
degree of interparticle bonding. Many factors including grain
size, grain-boundary structure, porosity, and impurity-derived
dispersoid populations influence microhardness.1 Therefore,
differences in microhardness among conventional aluminum,
sintered aluminum powder, and fine-grained aluminum are not
readily interpretable. We may not conclude that the somewhat
enhanced microhardnesses of our samples were due to small
grain sizes.
Grain Growth in Consolidated nano-Al. During consoli-
dation the nano-Al grain sizes grew beyond the nanocrystalline
size regime (to 150 to >330 nm; see Table 2). Substantial grain
growth was most evident in the pellets pressed from method-B
powders because of their smaller initial particle sizes and likely
also because of their greater purities, as compared to the
method-A powders. The mean grain size more than doubled
in pellet B15-25, pressed at room temperature, and increased
by factors of 3-4 in some method-B pellets pressed at higher
temperatures. The large uncertainties in the final grain size for
the larger-grained consolidates (Table 2) were due to increas-
ingly narrowed XRD line widths, which failed to provide precise
measures of grain size for grains larger than ca. 200 nm, and to
crystallite-shape anisotropies, which were not accounted for by
our analytical model (see the Experimental Section). The rms
microstrains of 2-4 × 10-4, also extracted from XRD line
widths (see the Experimental Section), were much less than the
maximal rms microstrains of 2 × 10-3 obtained by mechanical
attrition of aluminum,31 indicating that the nano-Al was only
moderately strained as a result of consolidation.
Figure 7. TEM images from a sliced and thinned specimen from pellet
B15-300 (see Table 2) after the pellet aged for 3 months at room
temperature. Note the presence of many grains with ca. 1-µm
dimensions, and the near absence of grains having e100-nm dimen-
sions.
Discussion
Grain sizes continued to increase when the pellets were stored
at room temperature as evidenced by further narrowing of XRD
line widths. After three months, XRD analyses of pellets B15-
25, B15-100, and B15-300 indicated that grain sizes had
surpassed 200 nm, and were thus outside the regime for accurate
grain-size measurement by XRD. Pellet B15-300 was selected
for TEM analysis because it was the densest and best able to
survive slicing, polishing, and ion milling. TEM images (Figure
7) of the thinned specimen revealed enlarged grains. Irregular
crystallite shapes and contrast prevented a detailed statistical
analysis of the grain-size distribution, but the mean grain size
appeared to be g300 nm with several single-crystalline regions
of ca. 1 µm in diameter also evident. Thus, the mean grain
size was found to be larger than that determined by XRD
immediately after consolidation, confirming continued grain
growth.
Grain growth occurs significantly faster and at lower tem-
peratures in the chemically synthesized (method-B) nano-Al
described here than in nano-Al prepared by gas condensation
or mechanical attrition.23 The kinetic stability of nanocrystalline
metals has been previously attributed to small grain sizes and,
more precisely, to narrow grain-size distributions.12,13,20 How-
ever, we argue below that such stability is likely due to extrinsic
impurities. Enhanced grain growth in the chemically synthe-
sized (method-B) nano-Al is likely due to greater grain-boundary
purity and the arrangement of nanocrystallites into favorable
aggregate structures.
Whereas we report grain-size doubling at 25 °C for chemi-
cally synthesized (method-B) nano-Al, previous studies reported
no or minimal grain growth up to temperatures of 300,27 330,28
or 800 °C31 for nano-Al prepared by the other methods.
Annealing at 610 °C was required to double grain sizes in nano-
Al produced by gas condensation.28 Apparently, one or more
of the factors that stabilize nanometer-scale grain structures is
lacking in the chemically synthesized nano-Al. Grain-size
distribution may be eliminated as a likely factor.
Because of the grain growth that occurred during and after
pressing, the consolidated pellets were not nanocrystalline. The
incompletely densified pellets were embrittled by the submi-
crometer porosity retained and were therefore unsuitable for
measurement of tensile properties. Although the pellets could
have been pressed at higher temperatures or pressures to attempt
the removal of residual porosity, grain growth beyond the
nanometer size regime had already occurred. Thus, our chemi-
cal synthesis of nano-Al followed by compaction apparently
cannot produce fully dense, pure nanocrystalline Al. If facile
room-temperature grain growth is an intrinsic characteristic of
pure nanocrystalline Al, as our results suggest, then fully dense,
pure nanocrystalline Al is not practicably achievable.
Rivier proposed that nanocrystalline materials are kinetically
stabilized by narrow grain-size distributions.20 According to
this model, the local driving force for grain coarsening is the
difference in chemical potential between neighboring grains.
Chemical potential is grain-size dependent, such that a smaller
grain would be readily consumed to enlarge an already larger
neighboring grain. However, neighboring grains of the same
size are trapped in a metastable equilibrium without sufficient

Facile Room-Temperature Grain Growth
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10853
driving force to enlarge one at the expense of the other. If this
is the origin of increased grain-size stability in nano-Al prepared
by gas condensation or mechanical attrition, then we should
observe significantly broader grain-size distributions in our
samples than were observed in the former.
However, the grain-size distributions achieved in the gas-
condensation and mechanical-attrition syntheses are fairly broad
and not markedly narrower than those achieved in the chemical
syntheses. A statistical analysis from TEM images of nano-Al
prepared by gas condensation gave a grain-size distribution of
10-60 nm (30 nm ( 100%).30 Microscopic images of samples
prepared in other gas-condensation27 and mechanical-attrition
syntheses reveal similar and broader size distributions.32,33 As
established above, the method-B powders subjected to consoli-
dation possessed grain-size distributions of 20-200 nm, and
mean crystallite sizes of 48 and 74 nm (for samples B13 and
B15, respectively; see Table 1). These distributions are shifted
to slightly larger mean sizes than those achieved by gas
condensation and mechanical attrition and are asymmetric,
possessing a minority fraction of larger crystallites in the size
range of 100-200 nm. However, the relative magnitudes of
the size distributions from the chemical syntheses, relative to
the midpoint of the observed size ranges, are comparable to
those from gas condensation and mechanical attrition. Conse-
quently, the Rivier model20 does not account for the differing
grain-size stabilities of nano-Al prepared by the various
methods. The validity of the model is neither supported nor
refuted, but the model does not distinguish the present com-
parison. Presumably, none of the nano-Al samples have
sufficiently narrow grain-size distributions to be stabilized by
Rivier’s proposed effect.
The high grain-size stability of nano-Al prepared by gas
condensation or mechanical attrition, in some cases to temper-
atures above the melting point of bulk Al (660 °C),31 exceeds
reasonable expectation for pure nano-Al. Stabilities for nano-
crystalline phases should scale roughly with melting points and
recrystallization temperatures for the bulk phases. Gleiter11 and
Birringer25 stated that 10-nm grains in metals having bulk
melting points of about 600 °C or less double within 24 h at
room temperature (e.g., Sn, Pb, Al,24 or Mg). Nanocrystalline
metals with higher bulk melting points have higher grain-size
stability. Nanocrystalline Cu (bulk Tm ) 1083 °C) is stable to
e100 °C, nanocrystalline Pd (bulk Tm ) 1552 °C) to e150 °C,
and nanocrystalline Fe (bulk Tm ) 1535 °C) to e200 °C.11,25
The recrystallization temperature of coarse-grained (bulk) Al
is only 80 °C.32 Because Al is one of the lowest-melting fcc
metals, nano-Al should exhibit less resistance to grain growth
than do other nanocrystalline metals but, apart from the present
findings, has instead exhibited more.
Figure 8. Scheme depicting proposed oxidation and consolidation
behavior of nano-Al powders. Heavy black borders represent oxide
coatings. (a) nano-Al prepared by gas condensation. (b) nano-Al
prepared by chemical synthesis (present work).
impurities, usually adventitious and unrecognized, are implicated
as the extrinsic origin of stability in nano-Al prepared by gas
condensation or mechanical attrition.
Aluminum is extremely oxophilic. Surface oxidation occurs
at O2 partial pressures of g10-41 Torr,45 which is exceeded
under typical gas-condensation and mechanical-attrition condi-
tions. The loose nanocrystallite aggregates formed by gas
condensation30,31,44 are susceptible to surface oxidation because
a large surface area is exposed (Figure 8a). During mechanical
attrition, low levels of adventitious O2 and N2 in the milling
atmospheres often result in finely dispersed oxide and nitride
impurity phases in Al and Al-containing intermetallics,31,46,47
presumably because surface impurities are kneaded into the
nanostructured aggregate particles. These impurities stabilize
the nanocrystallites against grain growth during compaction and
thermal treatments (Figure 8a).
The chemically prepared nano-Al is also adventitiously
oxidized, but the extent and distribution of oxide impurities
varies from the above. We propose the following rationale.
The reducing reaction conditions prevent oxidation of nano-
crystallite surfaces during chemical synthesis. Other likely
oxidation-prevention factors are the small reactor (reaction-flask)
volume and short preparation time in comparison to the other
methods, which further limit exposure to adventitious oxygen.
Tight aggregates precipitate from the solvent dispersion (Figure
5e, f, and h), in which the internal nanocrystallite interfaces
are protected (Figure 8b). The dense aggregate structures
contrast with the open, porous structures formed by gas
condensation (Figure 8a).30,31,44 Subsequent manipulation of
the nano-Al after chemical synthesis allows adventitious oxida-
tion of the external aggregate surfaces. Upon compaction, grains
grow to sizes approaching the sizes of the initial aggregates
(Figure 8b). Because most of the nanocrystallite surfaces in
the chemically synthesized nano-Al remain unexposed and thus
free of stabilizing oxide layers, the expected intrinsically rapid
(room-temperature) grain growth is observed.
This unexpected resistance to grain growth suggests extrinsic
stability origins. Indeed, nano-Al prepared by gas condensation
and oxidatively passivated with 4-nm aluminum-oxide coating
layers retained 30-nm grain sizes after heating to 725 °C.30,44
In a separate study, Al nanocrystallites prepared by gas
condensation were found to be adventitiously coated with 6-nm
aluminum-oxide layers, to which was attributed the grain-size
stability of the sample at 800 °C.31 Second phases (impurities)
at grain boundaries are known to inhibit coarsening in nano-
crystalline materials.1,2,23 Consequently, grain-boundary oxide
Conclusions
Chemical synthesis produces nano-Al having sufficient purity
to exhibit intrinsically facile grain growth. Unlike chemical
synthesis, gas-condensation and mechanical-attrition syntheses
do not employ hydrocarbon solvents or heteroatom-containing
reagents and, consequently, are not susceptible to impurities
derived from them. However, these nano-Al syntheses employ
processing methods and produce powder microstructures that
(42) So¨hnel, O.; Garside, J. Precipitation: Basic Principles and Industrial
Applications; Butterworth and Heinemann: Oxford, 1992; pp 163-180.
(43) Callister, W. D. Mater. Sci. Engineering, 3rd ed.; Wiley: New York,
1994; p 171.
(44) Sa´nchez-Lo´pez, J. C.; Gonza´lez-Elipe, A. R.; Ferna´ndez, A. J. Mater.
Res. 1998, 13, 703-710.
(45) Fujii, H.; Nakae, H.; Okada, K. Acta Metall. Mater. 1993, 41, 2963-
2971.
(46) Suryanarayana, C.; Froes, F. H. AdV. Mater. 1993, 5, 96-106.
(47) Itsukaichi, T.; Masuyama, K.; Umemoto, M.; Okane, I.; Caban˜as-
Moreno, J. G. J. Mater. Res. 1993, 8, 1817-1828.

10854 J. Am. Chem. Soc., Vol. 120, No. 42, 1998
Haber and Buhro
Energy dispersive spectroscopy (EDS) with light-element detection
was performed in conjunction with TEM and SEM using Noran
Instruments Voyager II X-ray quantitative microanalysis system with
digital imaging.
Vickers microhardnesses were measured on polished and/or unpol-
ished pellet surfaces using a LECO M-400-H1 instrument with 50-
and/or 100-g loads. The polished surfaces were prepared by successive
polishing with 320, 600, 1000, and 2000 grit paper with flowing water,
followed by lapping with 1.0 µm and 0.3 µm Al2O3 paste. Each
hardness reported is the mean of 10 hardness measurements distributed
over the entire surface of each pellet. Asymmetrical indentations were
ignored.
Starting materials were obtained as follows. LiAlH4 (95% purity,
Aldrich) was used without further purification. NMe2Et (99% purity,
Aldrich) was distilled under nitrogen. AlCl3 (Matheson, Coleman &
Bell) was purified by a single sublimation from Al and NaCl. 1,3,5-
trimethylbenzene (97%, Aldrich) was washed in a separatory funnel
with concentrated H2SO4 until colorless, and then washed successively
with deionized H2O, 5% NaOH, and deionized H2O. The 1,3,5-
trimethylbenzene was then predried over CaH2, distilled from sodium/
potassium benzophenone ketyl and stored over type 4A sieves. Pentane
was refluxed over sodium benzophenone ketyl with a few drops of
tetraethylene glycol dimethyl ether (99+%, Aldrich), distilled, and
stored over type 4A sieves. MeOH was refluxed over magnesium
turnings activated with iodine, distilled, and stored over type 3A sieves.
Pyridine (99.99%, Fisher) was refluxed over KOH for 4 days, distilled,
and stored over type 4A sieves. Titanium(IV)isopropoxide (98% purity,
Strem) was used as received.
N,N-dimethylethylamine alane, H3Al(NMe2Et), was prepared by the
method of Frigo et al.40 Warning: a powerful explosion occurred
during the synthesis of H3Al(NMe2Et). The AlCl3/LiAlH4 slurry must
be stirred thoroughly and rapidly to ensure initiation of the reaction
upon slow addition of NMe2Et. Failure to initiate leads to a mixture
of exothermic-reaction partners in large, concentrated quantities, which
is a dangerous situation prone to violent explosion. Complete details
are given in the Supporting Information.
H3Al(NMe2Et) is a colorless liquid that partially decomposed to a
gray precipitate when stored at room temperature. When stored in
liquid N2 in an evacuated Schlenk flask, a cloudy white impurity
developed, which settled upon standing at room temperature. In both
cases, the decomposition was accompanied by a build-up of overpres-
sure, which in one case fractured the top of the storage flask. H3Al-
(NMe2Et) was usually repurified after several days of storage by
recrystallization from pentane (-20 °C). The highest yields of nano-
Al were obtained using freshly recrystallized H3Al(NMe2Et).
Preparation of nano-Al. Method A. A stirred slurry of AlCl3 (3.11
g, 23.3 mmol) and LiAlH4 (2.66 g, 70.1 mmol) in 100 mL of 1,3,5-
trimethylbenzene was refluxed for 16 h. The crude product (5.18 g,
94.5% yield) was collected on a glass frit, washed with two 10-mL
portions of hexane, and dried in vacuo. The crude product was divided
into three fractions (0.57, 2.24, and 2.43 g), which were washed with
several portions of 0 °C MeOH (4-10 10-mL portions, respectively),
dried in vacuo, and combined (overall yield of nano-Al: 2.24 g, 83.0
mmol, 89% with respect to AlCl3). Elemental analysis of method-A
nano-Al was conducted on a consolidated pellet (see below).
Preparation of nano-Al. Method B. The decomposition of H3-
Al(NMe2Et) was performed using a range of molar ratios of decom-
position catalyst, Ti(O-i-Pr)4, on a variety of reaction scales using
several different batches of H3Al(NMe2Et) (see Table 1).
In general, 25 to 100 mL of H3Al(NMe2Et) and 75 to 300 mL of
1,3,5-trimethylbenzene (approximately 2.5 mol of H3Al(NMe2Et)/L of
1,3,5-trimethylbenzene) were placed in a Schlenk flask. Then Ti(O-
i-Pr)4 (0-1.2 × 10-2 moles per mole of H3Al(NMe2Et)) was added to
the room-temperature solution. The clear, colorless solution turned
tan to dark brown, depending upon the amount of added Ti(O-i-Pr)4,
and fumed. In the runs using the Ti(O-i-Pr)4 catalyst, rapid gas
evolution ensued when the solution was warmed to reflux (ca. e100
°C). Gas evolution was complete within 15 min. and heating was
continued for an additional g20 min. In the runs without the Ti(O-
i-Pr)4 catalyst, complete decomposition required several h at higher
reflux temperatures (≈164 °C). Powders or spongy deposits of nano-
are inherently more prone to adventitious oxidation. The
reducing reaction conditions and tightly packed aggregate
microstructures achieved in the method-B chemical synthesis
produce a lower oxide content and coarser oxide distribution
than has been found in the nano-Al produced by the other
methods.31
nano-Al prepared by gas condensation or mechanical attrition
is extrinsically stabilized against grain growth by grain-boundary
oxide impurities. Sa´nchez-Lo´pez et al.30 and Eckert et al.31
independently established that oxide passivation of nano-Al
prepared by these methods provides grain-size stability to
temperatures far beyond expectation for pure nano-Al. We have
now shown the corollary; reduction of oxide-impurity levels
activates room-temperature grain growth, as predicted.11,25 That
narrower grain-size distributions may enhance stability20 remains
to be tested because such distributions are not practicably
achievable by current methods. Barriers to grain growth in pure
nano-Al having typical grain-size distributions are intrinsically
low.
Metals having widely varying oxophilicities and other proper-
ties, that is, aluminum, copper,3-8 silver,3,4 palladium,3,4 etc.,11,25
are now known to exhibit low kinetic stabilities in nanocrys-
talline form. Consequently, low thermal stability should perhaps
be a general expectation for pure nanocrystalline metals.
Production of nanostructured materials with useful thermal
properties will likely require nanocomposite strategies in which
very finely dispersed second phases are purposefully incorpo-
rated to stabilize nanoscale grains.
Experimental Section
General Procedures and Starting Materials. All ambient-pressure
synthetic manipulations were conducted under dry N2 using standard
inert-atmosphere techniques. Ground-glass joints were sealed with
Teflon sleeves to minimize Si contamination of the products. Glow-
discharge mass-spectrometric analyses were conducted by Shiva
Technologies (Cicero, NY). X-ray diffraction procedures and data
analysis are described in a separate section below.
Transmission electron microscopy (TEM) was performed on a JEOL
2000 FX instrument operating at 200 keV. Powder samples were
prepared by combining 10-25 mg of powder with 10-30 mL of dry
pyridine (under N2). This mixture was then sonicated for 15-250 min
in an ultrasonic cleaning bath until the powder was well suspended.
Powders that were loosely agglomerated suspended immediately without
sonication, whereas heavily aggregated powders required sonication
for extended periods. While the suspension continued to be sonicated,
a few drops were removed by syringe and placed on a holey-carbon-
coated TEM grid in air. The grid was immediately placed into a tightly
capped vial with 2-5 drops of the pyridine suspension and immediately
loaded into the TEM antechamber while still wet with pyridine. This
method proved quite effective at limiting oxidation, with many particles
containing less than 10% oxygen.
The sample for TEM analysis from the consolidated pellet was
prepared by cutting a thin slice from pellet B15-300 (see below) using
a low-speed diamond wheel saw (South Bay Technology model 650)
and an aqueous cooling solution (South Bay Technology P/N 0202460).
The slice was then thinned by polishing on 1500-grit carborundum paper
wet with MeOH and Ar-ion milled (Gatan dual ion mill model 600) at
a 12° angle in a liquid-nitrogen-cooled specimen holder.
Scanning electron microscopy (SEM) was performed on a Hitachi
S-4500 field-emission instrument. SEM specimens were prepared by
dusting powders on carbon tape on Al stubs in the N2 glovebox, sealing
the stubs in tightly capped vials and performing SEM immediately after
removing the vials from the glovebox. The stubs were removed from
the vials and placed in the sample-exchange chamber immediately,
keeping air exposure to less than one minute. Consolidated specimens
were similarly examined before extended air exposure and/or after
polishing as described below for microhardness testing.

Facile Room-Temperature Grain Growth
J. Am. Chem. Soc., Vol. 120, No. 42, 1998 10855
Al were then precipitated with or without catalyst, respectively. The
coarseness of the powder depended upon the amount of catalyst used
(see Results); in some cases, the product was so fine it passed in part
or entirely through a medium porosity glass frit. The spongy product
produced without catalyst or the collected powders were washed with
hexane or pentane and dried in vacuo. The isolated nano-Al yields
(with respect to starting AlCl3) ranged from 28 to 83% for uncatalyzed
runs and to 60-99% for Ti(O-i-Pr)4-catalyzed runs. Elemental analysis
of method-B nano-Al was conducted on a consolidated pellet (see
below).
Consolidation of nano-Al. Pellets 13 mm in diameter and 2 to 12
mm in thickness were pressed from the nano-Al produced by methods
A and B using a uniaxial hot press in an Ar-filled glovebox. The press
employed TZM (a molybdenum-based alloy) plungers and die with a
disposable graphite sleeve for lubrication and ease of pellet removal.
All pellets (0.75-3.53 g) were pressed at 350 MPa for 1 h at room
temperature, 100 °C, or 300 °C. The pellets were labeled A-100-
B15-300 to specify the source of the powder and the pressing
temperature (see Tables 1 and 2 and the Results).
The mean coherence lengths (crystallite sizes) in the as-precipitated
powders were calculated using the Scherrer formula L ) (1.0λ)/(â cos
θ),48 where λ is the X-ray wavelength in nm, â is the intrinsic peak
width in radians (2θ), θ is the Bragg angle, and 1.0 is the Scherrer
constant. This calculation assumed that the nano-Al powders were
strain-free, which was supported by subjecting the data to the size-
strain deconvolution process described below. The sizes determined
from the deconvolution calculations were within error limits of the
values determined from the Scherrer formula, and the strain values were
small (10-4) and both positive and negative. Thus, strain was negligible.
The errors in the mean coherence lengths (sizes) were estimated from
the variation in coherence length calculated from the 111, 200, 220,
311, and, when collected, the 222 diffraction intensities and do not
include the systematic error introduced by assuming pure Cauchy peak
shapes. Note that, as the intrinsic broadening increases, the assumptions
about line shape become less important. These results are recorded in
Table 1.
Both size and strain broadening contributed to the intrinsic peak
widths of the consolidated pellets. The broadening due to crystallite
size, (δs)size, (where s ) 2 sin θ/λ) is independent of s, whereas the
strain broadening, (δs)strain, is proportional to s.32 Thus, the dependence
of the corrected peak width (δs)0 on s enables the separation of size
and strain broadening. The broadening due to particle size is best
approximated by a Cauchy function, whereas the broadening due to
strain is best approximated by a Gaussian function.32,48 The contribu-
tions of size and strain broadening were separated following Eckert et
al.32 who substituted the functional dependence of (δs)size and (δs)strain
on s into an approximation for a Cauchy/Gaussian convolution to obtain
eq 3
Elemental analysis by glow-discharge mass spectrometry gave the
following results. The major impurities in sample A-300 were (wt
%): C, 3.00; N, 0.19; O, 4.00; Si, 0.24; Cl, 3.30; K, 2.10; Li, 0.12; Al,
87.0 (by difference). The minor impurities were (ppm by wt): B, 190;
Na, 280; Mg, 28; S, 14; Ca, 2.3; Ti, 11; V, 0.19; Cr, 30; Mn, 2.5; Fe,
90; Ni, 27; Cu, 6; Zn, 8.6; In, 73; Ta, <28; W, 0.12.
The major impurities in B13-100 were (wt %): C, 0.23; O, 0.25; N,
0.055; Cl, 0.14; Ti, 0.32; and Al, 99.0 (by difference). The minor
impurities were (ppm by wt): Li, 1.3; B, 0.15; F, <0.05, Na, 2.9; Mg,
0.37; Si, 40; S, 1.8; K, 4.8; Ca, 0.61; Cr, 0.5; Mn, 0.01; Fe, 1.5; Ni,
0.35; Cu, 0.4; Zn, 0.27; Ga, 0.12; Mo, 0.9; In, 1.9; Ta, <1; W, 0.05.
X-ray Diffraction (XRD) Procedures and Data Analysis. XRD
patterns were obtained using a Rigaku vertical powder diffractometer
with Cu KR (λ ) 1.541 845 Å) radiation and Materials Data Inc. (MDI)
automation and software. All XRD patterns were collected in a high-
resolution configuration with a 0.040°-2θ step and a 1-second count.
The XRD samples (1.5 × 2 cm smear mounts) were prepared in a
N2-filled glovebox and coated with a thin film of polyvinyltoluene
(PVT) to minimize air exposure. The PVT film was applied by dripping
a toluene solution of PVT onto the prepared XRD slide and evaporating
it to dryness. XRD line widths were determined using the MDI
software to electronically subtract the background and KR2 peaks. A
background level was drawn across each peak, which the software
integrated to determine the integral breadth (which the software terms
fwhm) by dividing the area by the peak height. The instrumental
broadening was determined by collecting a pattern of the NIST Standard
Reference Material (SRM) 660, (LaB6), using a step of 0.020° 2θ, with
a 1-second count. In subsequent calculations, see below, the fwhm of
the SRM 660 peak closest to the experimental peak was used as the
instrumental broadening.
In the high-resolution configuration we employed, the instrumental
broadening and size-distribution broadening of the XRD reflection
profiles are better approximated by Cauchy (or Lorentzian) peak shapes
than by Gaussian peak shapes.48 Therefore, Cauchy peak shapes were
assumed, and the intrinsic breadth was extracted using B ) â + b,
where B is the experimental (observed) breadth, â is the intrinsic breadth
in the sample, and b is the instrumental breadth.48 This method resulted
in narrower intrinsic peak widths and hence larger apparent coherence
lengths than were obtained when Gaussian peak shapes were assumed.
Because the experimental peak shapes had a small Gaussian component,
the mean coherence lengths obtained from the pure Cauchy analysis
were overestimates; however, these coherence lengths were more
consistent with TEM and SEM particle sizes than were those calculated
assuming pure Gaussian peak shapes (B2 ) â2 + b2).48
2
1
(δs)0
s
≈ d - 6.25 e2
d
(3)
[
]
(δs)0
where d is the volume-averaged grain size, and e2
is the rms
1/2
microstrain. The parameters d and e2
were determined from the
1/2
linear least-squares fit of 1/(δs)0 vs [s/(δs)0]2 from all of the diffraction
peaks collected from the sample, whereas the error in the fit estimated
the error in these parameters. The results are recorded in Table 2.
The values obtained for the grain size and rms strain have large
uncertainties for two reasons. (1) For large crystallite sizes, the intrinsic
peak broadening approaches the error in the measurement. (2) The
powders and pellets contained a complex distribution of particle shapes,
which varied from sample to sample. The breadth of XRD reflections
is also influenced by particle shape. Shape anisotropy variously
broadens lines corresponding to different crystallographic directions
within the crystallites. We were unable to account for this shape effect
in our analysis.
Acknowledgment. This work was funded by NSF grants
CHE-9158369 and CHE-9709104, with additional support from
the Exxon Education Foundation, Emerson Electric, Monsanto,
and Eastman Kodak. J.A.H. was partially supported by a
Department of Education GAANN grant (P200A40147). We
thank Nilesh V. Gunda for assistance with pressing and
microhardness testing, and W. L. Gladfelter for suggesting the
use of the Ti(O-i-Pr)4 catalyst. We also thank S. Hariprasad for
advice on polishing pellets and making microhardness measure-
ments, and P. C. Gibbons, K. F. Kelton, and J. A. Hollingsworth
for helpful discussions.
Supporting Information Available: Precautions for the
preparation of H3Al(NMe2Et) (2 pages, print/PDF). See any
current masthead page for ordering information and Web access
instructions.
(48) Klug, H. P.; Alexander, L. E. X-ray Diffraction Procedures for
Polycrystalline and Amorphous Materials, 2nd ed.; Wiley-Interscience: New
York, 1974, Ch. 9.
JA981972Y