Relationship between crystallite size and bond lengths in boehmite

Article abstract of DOI:10.1006/jssc.2001.9124

Boehmite with a crystallite size between 1 and 27 nm was prepared by annealing a boehmite precipitate under hydrothermal conditions at different temperatures. The crystalline structure, measured with X-ray powder diffraction, was refined with the Rietveld method. From the refinement bond lengths and bond angles between oxygen and aluminum atoms were calculated, which revealed that boehmite bond lengths and local symmetry depended on crystal dimensions. When boehmite's crystallite size decreased, Al-OH interaction increased, and the local octahedral symmetry tended to be tetrahedral. The hydrogen bond, sustaining boehmite's crystalline structure, and the bonds between oxygen atoms and hydroxyls inside the octahedra double layers, building the crystalline structure, became weaker as crystallite size decreased, which could explain why boehmite's transformation temperature into a transitional alumina also decreased with crystallite size. Alumina crystallite size also depended of the corresponding size of boehmite. It was shown that desorbed water from annealed samples came from the crystal surface and not from the bulk.

Full text of DOI:10.1006/jssc.2001.9124

Journal of Solid State Chemistry 159, 32}40 (2001)
doi:10.1006/jssc.2001.9124, available online at http://www.idealibrary.com on
Relationship between Crystallite Size and Bond Lengths in Boehmite
X. Bokhimi,*ꢀ-ꢀꢁ J. A. Toledo-Antonio,* M. L. Guzm aH n-Castillo,* and F. Hern aH ndez-Beltr aH n*
*
Instituto Mexicano del Petr o& leo, Eje Central L. Ca&rdenas 152, A.P. 14-805, 07730 M e& xico, DF, Mexico; and -Institute of Physics,
The National University of Mexico (UNAM), A.P. 20-364, 01000 M e& xico, DF, Mexico
Received November 16, 2000; in revised form January 17, 2001; accepted February 9, 2001; published online May 11, 2001
(9), the reported models only provide a "rst idea about real
Boehmite with a crystallite size between 1 and 27 nm was atomic ordering. The most recent report about the crystal-
prepared by annealing a boehmite precipitate under hydrother- line structure of these aluminas corresponds to the theta
mal conditions at di4erent temperatures. The crystalline struc-
ture, measured with X-ray powder di4raction, was re5ned with
the Rietveld method. From the re5nement bond lengths and bond
angles between oxygen and aluminum atoms were calculated,
which revealed that boehmite bond lengths and local symmetry
depended on crystal dimensions. When boehmite:s crystallite size
decreased, Al+OH interaction increased, and the local octahed-
ral symmetry tended to be tetrahedral. The hydrogen bond,
sustaining boehmite:s crystalline structure, and the bonds be-
phase (10), which, however, is only a new approximation but
not the correct solution.
Aluminas are commonly obtained from four precursors:
two hydroxides, gibbsite (11) and bayerite (12); and two
oxyhydroxides, diaspore (13) and boehmite. Although this
last phase has been widely used for preparing aluminas,
a detailed characterization of boehmite is lacking. In litera-
ture boehmite is classi"ed in two big groups: the one where
tween oxygen atoms and hydroxyls inside the octahedra double the samples are well crystallized and the named
layers, building the crystalline structure, became weaker as pseudoboehmite group (14). Recently we have suggested
crystallite size decreased, which could explain why boehmite:s that boehmite and pseudoboehmite represent the same
transformation temperature into a transitional alumina also
decreased with crystallite size. Alumina crystallite size also de-
pended of the corresponding size of boehmite. It was shown that
desorbed water from annealed samples came from the crystal
surface and not from the bulk. ( 2001 Academic Press
phase but with di!erent crystallite size (15). In order to
provide more evidence to justify this proposition, we have
prepared boehmite samples with well-de"ned and control-
led crystallite size and have analyzed in detail the corre-
sponding crystallography.
When boehmite is the precursor for alumina synthesis, its
characteristics determine crystal and particle properties of
the synthesized transitional and alpha aluminas; for
example, sintering and transformation temperatures and
their thermal stability. The transformation temperature of
boehmite into a transitional alumina (16) and of h- into
a-alumina (17) increases as boehmite crystallite size in-
creases; the origin of this temperature dependence, however,
has not been explained. It could be associated to changes in
atom bonds, because detailed studies in natural and syn-
thetic boehmite show di!erent bond lengths in both systems
Key Words: Boehmite:s crystallography; Rietveld re5nement;
bond lengths; crystal:s surface; transitional alumina.
INTRODUCTION
Aluminum oxide is widely used in industry, for example,
for re"ning petroleum (1}4), as a substrate in integrated
circuits (5), or for reducing pollution (6). Considering atom
ordering in its crystalline structure, this oxide can be classi-
"
ed into two groups: the one represented by alpha alumina
with its atoms in a hexagonal arrangement (7) and the group
of the transitional aluminas (s-, g-, c-, i-, and h-alumina),
where atoms are ordered in a spinel-based structure (8).
Alpha alumina, which is free of hydroxyls, is the most
stable phase and is obtained after annealing samples at
temperatures higher that 8003C. The crystalline structure of
transitional aluminas, which are characterized by contain-
ing hydroxyls in its structure, has been only approximated
(18, 19).
In order to understand boehmite properties in more detail,
we have prepared it with di!erent crystallite sizes by growing
a boehmite precipitate under hydrothermal condition at dif-
ferent temperatures and have re"ned the crystalline structure
with the Rietveld method, starting from X-ray powder dif-
fraction patterns. From the re"nement we calculated bond
lengths and bond angles between oxygen and aluminum
atoms. Since it was possible to prepare boehmite with crystal-
lite sizes between 1 and 27 nm, the present study suggests
a possible behavior of atoms near boehmite's surface.
ꢁ
To whom correspondence should be addressed. E-mail: bokhimi@
fenix.i"sicacu.unam.mx.
3
2
0
022-4596/01 $35.00
Copyright ( 2001 by Academic Press
All rights of reproduction in any form reserved.

CRYSTALLITE SIZE AND BOND LENGTHS IN BOEHMITE
33
EXPERIMENTAL
TABLE 1
Boehmite, Space Group Cmcm: Atom Fractional Coordinates
Synthesis. Boehmite crystal growth was performed un-
der hydrothermal conditions. Seeds were precipitated at
room temperature (233C) by mixing, dropwise on 50 mL of
distilled water at pH 8, an aqueous solution of aluminum
Atom
Site
x
y
z
Al
O1
4c
4c
4c
0.0
0.0
0.0
y
y
y
0.25
0.25
0.25
ꢂꢀ
-
-
ꢁ
complex ions and a solution rich in hydroxyls. The "rst O2
solution (300 mL, 0.3 M Alꢄ>) was prepared from
AlCl ) 6H O (J. T. Baker) and distilled water; the second
ꢃ
Note. Since hydrogen X-ray di!raction is negligible, it was not con-
sidered.
ꢄ
ꢃ
one (100 mL) from NH OH (J. T. Baker, 50 vol%). The
ꢅ
suspension (450 mL) with the boehmite precipitate was
placed in a 600-mL autoclave and heated and stirred at
a "xed temperature between 23 and 2403C for 18 h under resolution electron microscopy (26). For the re"nement, this
autogenous pressure. Thereafter, it was "ltered and washed form anisotropy was modeled by using FULLPROF code
thoroughly with distilled water until it was chlorine free (21), which showed that crystals grew with their plates
(
silver nitrate test), and dried overnight at 1103C in air.
parallel to (020) planes. The only adjustment coe$cients
used for the re"nement where the four de"ning the poly-
nomial to model background. For the samples with large
crystallites, 32 independent variables were used; this number
was reduced to 17 for the samples with small crystallites,
because the anisotropic temperature factors were "xed. Two
more constrains were used in the re"nement of all samples:
The < and = parameters associated to the gaussian contri-
bution of the peak pro"le were "xed to their values deter-
mined from the re"nement of standard samples with large
crystallites. Figure 4 displays a typical Rietveld re"nement
plot.
For all samples the crystal shortest dimension was the
one parallel to [020] direction, which for boehmite's crystal-
line structure is perpendicular to [020] planes. This prop-
erty can be explained by analyzing the crystalline structure
made of oxygen octahedra with aluminum atom near their
centers and oxygen atoms in their vertices (Fig. 5); two of the
oxygen atoms belong to hydroxyl ions. Octahedra form
double layers perpendicular to b axis (Fig. 1) that interact
between each other via an hydrogen bonding OH2O; this
kind of bond has typical bond energies about 20 kJ/mol
Characterization. X-ray di!raction patterns of the sam-
ples packed in a glass holder were recorded at room temper-
ature with CuKa radiation in a Bruker Advance D-8
di!ractometer having theta}theta con"guration and
a graphite secondary-beam monochromator. Di!raction
intensity was measured by step scanning in the 2h range
between 103 and 1273, with a 2h step of 0.023 for 8 s per
point. Crystalline structures were re"ned with the Rietveld
technique by using DBWS-9411 (20) and FULLPROF-
V3.5d (21) codes; peak pro"les modeled with a pseudo-Voigt
function (22) contained average crystallite size as one of its
characteristic parameters (23). Standard deviations, which
show the last "gure variation of a number, are given in
parentheses; when they correspond to re"ned parameters,
their values are not estimates of the probable error in the
analysis as a whole, but only of the minimum possible
probable errors based on their normal distribution (24).
RESULTS AND DISCUSSION
Crystallography and Crystal Morphology
(27). In the present case, this bonding is destroyed between
All fresh samples contained only boehmite. Its crystalline 300 and 5503C, which is the temperature range where boeh-
structure was re"ned by modeling it with an orthorhombic mite is transformed into a transitional alumina. Detailed
unit cell having atoms in the positions given in Tables 1 and
2
, and the symmetry described by space group Cmcm. Until
now, this is the best model reported for boehmite's crystal-
line structure (25). In Table 1, O1 corresponds to the posi-
tion of oxygen atoms shared by four octahedra, and O2
represent the position of oxygen atoms of hydroxyls, which
participate in the hydrogen bonding that sustain boehmite's
crystalline structure (Fig. 1).
For a given sample, the peak widths of its di!raction
pattern were so di!erent (Figs. 2 and 3) that they could not
be explained by only considering strain and crystallite size 100
TABLE 2
Atom Fractional Coordinate y of Atoms for the Di4erent
Heating Temperatures
¹
(3C)
y
y
y
ꢂꢀ
-ꢁ
-ꢃ
2
3
50
3
0
!0.3239(3)
!0.3238(3)
!0.3247(3)
!0.3227(3)
!0.3192(2)
!0.3182(2)
!0.3180(1)
0.3100(5)
0.3104(5)
0.3043(4)
0.2982(4)
0.2927(4)
0.2907(3)
0.2905(2)
0.0841(4)
0.0843(4)
0.0860(3)
0.0856(4)
0.0780(3)
0.0797(2)
0.0795(2)
1
1
2
40
80
40
e!ects; the widest peak corresponded to (020) re#ection.
This width di!erence was an e!ect of crystal form anisot-
ropy, because they are platelet-like, as observed by high-

3
4
BOKHIMI ET AL.
FIG. 1. Projection of boehmite crystalline structure perpendicular to
the a axis. The hydrogen bonding between the double layers is shown.
information (covalent O}H and electrostatic O}H2O
FIG. 3. X-ray di!raction patterns of samples heated under hydrother-
bond lengths) about this boehmite bonding is reported for mal conditions at temperatures between 23 and 503C and the one prepared
neutron di!raction experiments (25); it is similar to those via sol}gel technique. Tick marks correspond to boehmite; the Miller
indices, (020), of the "rst peak are shown.
published for its isomorphous phase, lepidocrocite (c-
FeOOH), which has been also characterized with neutron
di!raction (28, 29). The discussion in the next paragraphs
about bond lengths in the basic octahedra (Fig. 5) will make
FIG. 4. Rietveld re"nement plot of the sample heated at 2403C
FIG. 2. X-ray di!raction patterns of the samples heated under hy-
drothermal conditions at temperatures between 100 and 2403C. Tick marks
correspond to boehmite; the Miller indices, (020), of the "rst peak are also
shown.
(
R
"0.145). Circles correspond to the experimental data, and the con-
UN
tinuous line to the calculated one. Tick marks represent boehmite
(
R ꢁ!''"0.028).

CRYSTALLITE SIZE AND BOND LENGTHS IN BOEHMITE
35
TABLE 4
Boehmite Lattice Parameters as a Function of Crystallite Size
d
(nm)
a(nm)
b(nm)
c(nm)
(
ꢆꢃꢆ)
1
1
2
2
6
.13(1)
0.2851(1)
1.212(1)
0.3736(1)
0.3761(1)
0.3733(1)
0.3715(1)
0.36945(4)
0.36941(2)
0.36941(4)
.56(2)
.04(4)
.42(4)
.90(8)
0.28796(8)
0.28675(9)
0.28686(9)
0.28695(3)
0.28681(1)
0.28678(1)
1.2205(9)
1.2274(9)
1.2265(9)
1.2232(2)
1.22256(8)
1.22188(4)
1
2
4.2(2)
6.3(5)
3
.21(8) nm and for 2403C it was 49(1) nm. The smallest
dimension on these plates, 3.21(8) nm, however, was nearly
0 times the unit cell dimension along them (Table 4),
1
indicating that boehmite crystals grew with preference along
FIG. 5. Octahedron representative of boehmite crystalline structure.
plates. This fact was more spectacular in sol}gel samples
(30); in this case, di!raction patterns showed only the re#ec-
evident why boehmite crystals grow mainly along the a}c tions associated to these dimensions (Fig. 3). Because in the
plane, where atomic bonds are stronger than interlayer present work crystallite thickness is representative of crys-
hydrogen bonds.
tallite size, in the rest of the discussion this thickness will be
used to represent crystallite size when its e!ect on para-
meters will be discussed.
Crystallite Dimensions
Boehmite crystallite growing under hydrothermal condi-
Boehmite crystallite dimensions depended on the heating tions was similar to the one reported for other compounds
temperature for the hydrothermal treatment (Table 3): Crys- grown with the same method (31): crystallite size increases
tallite thickness varied from 26.3(5) nm for the samples when heating temperature is increased. This size can also be
treated at 2403C to the dimensions of a unit cell, 1.13(1) nm, increased by changing the solvent or the mineralizer in the
for those treated at 233C. It is worth mentioning that boeh- hydrothermal treating (32, 31).
mite prepared with the sol}gel technique can have a &&crys-
tallite'' thickness smaller than a unit cell (30), as it is
concluded from the absence of (020) re#ection peak in the
Atom Bond Lengths
X-ray di!raction pattern (Fig. 3). For the used hydrother-
The octahedra building boehmite double layers are
mal conditions, large heating times produced only small oriented parallel to unit cell axis. Along the c axis, octahedra
thickness increases; for example, in the boehmite heated at share vertices (Fig. 1), and along the a axis they share their
2
403C, crystallite thickness grew from 26.5(3) to 34.3(3) nm edges parallel to the b axis (Fig. 6); in this way, each oxygen
when heating time varied from 18 to 138 h. atom (not belonging to hydroxyls) is shared by four oc-
Crystallite dimension along plates also grew as heating tahedra. In contrast, the oxygen atom of hydroxyls is only
temperature was increased (Table 3): for 233C it was shared by two octahedra (Figs. 1 and 6). This reveals that
the interactions along the a}c plane, given by the interaction
between the oxygen atoms shared by four octahedra and
TABLE 3
Boehmite Crystallite Dimensions for the Di4erent Heating
Temperatures
aluminum atoms, should be stronger than the interaction
between two octahedra double layers, favoring crystal
growing along planes and not perpendicular to them.
Lattice parameters a and b were almost independent on
boehmite's crystallite size (Table 4); in contrast the lattice
parameter c increased as crystallite size diminished. This
result is useful in proposing a model for the atom behavior
near boehmite's crystal surface. As it will be evident from the
discussion below, when the crystallite size decreases the
expansion along c axis is related to stronger interactions
between aluminum and oxygen atoms participating in hy-
drogen bonding.
¹
(3C)
d
(nm)
d
(ꢃꢆꢆ)"d(ꢆꢆꢃ)(nm)
(ꢆꢃꢆ)
2
3
5
00
40
80
40
3
0
0
1.13(1)
1.56(2)
2.04(4)
2.42(4)
6.90(8)
3.21(8)
3.53(8)
4.0(1)
5.4(1)
16.6(4)
43(2)
1
1
1
2
14.2(2)
26.3(5)
49(1)

3
6
BOKHIMI ET AL.
TABLE 5
Bond Lengths as a Function of Boehmite Crystallite Size
d
(nm)
d (nm)
d (nm)
d (nm)
d (nm)
(
ꢆꢃꢆ)
ꢁ
ꢃ
ꢄ
ꢅ
1
1
2
2
6
.13(1)
0.1810(1)
0.1825(1)
0.1805(1)
0.1823(1)
0.1892(1)
0.1901(1)
0.1904(1)
0.2160(1)
0.2181(1)
0.2136(1)
0.2063(1)
0.1983(1)
0.1957(1)
0.1953(1)
0.18756(6)
0.18876(6)
0.18832(6)
0.18815(6)
0.18755(6)
0.18774(6)
0.18774(6)
0.2738(2)
0.2760(2)
0.2680(2)
0.2607(2)
0.2602(2)
0.2580(2)
0.2578(2)
.56(2)
.04(4)
.42(4)
.90(8)
1
2
4.2(2)
6.3(5)
of the HO}Al}OH plane correspond to d . Finally, the
ꢉ
hydrogen bond length between contiguous octahedra
double layers is named d
(Fig. 1). The angle between an
2
-
& -
aluminum atom and the two next neighboring hydroxyls
bonds in the plane HO}Al}OH (Fig. 7a) was represented
with Greek letter d. The angle in the same plane between the
aluminum atom and the hydroxyl and oxygen atom bonds
in the quasi-diagonal was represented with g (Fig. 7b). The
angle formed by the interaction between aluminum atom
and the oxygen atoms in vertices was represented with
ꢀ
(Fig. 7c).
The bond lengths between the aluminum and oxygen of
hydroxyls and the oxygen in the HO}Al}OH plane, d and
ꢁ
d , respectively (Figs. 5 and 7 and Table 5), tended to be
ꢃ
FIG. 6. Projection of boehmite crystalline structure along the [001]
direction.
equal for large crystal dimensions. Since in the present work
the thickest crystals were only 26.3(5) nm, it is expected that
for boehmite microcrystals both bond lengths to be equal.
To understand more about boehmite's behavior, we cal- This and the fact that the angle g between these bonds
culated some characteristic atom bond lengths and atom (Table 5) moved to 1803 as this thickness increased suggest
bond angles in the octahedron representative of its crystal- that the orthorhombic symmetry of boehmite unit cell
line structure (Fig. 5). If readers are interested in other would probably change to a tetragonal symmetry for micro-
distances and angles di!erent to those listed in the tables, crystals.
the information about boehmite crystallography we are
When boehmite crystallite dimensions diminished, the
providing in the present work is enough to do it. To avoid interaction between aluminum atom and hydroxyls was
misunderstandings, we will clearly de"ne the di!erent dis- stronger, producing shorter Al}OH bond lengths and larger
tances and angles reported in tables and "gures (Tables d angles between them (Table 6); this angle moved from
5
through 7 and Figs. 1 and 5 through 7). The bond length 97.73(6) for the largest crystal thickness, 26.3(5) nm, to the
between an aluminum atom and the oxygen of hydroxyl
Al}OH) is named d ; d is the bond length between alumi-
(
ꢁ
ꢃ
num and oxygen atoms (Al}O) in the octahedra's plane
de"ned by hydroxyls and aluminum atoms (HO}Al}OH).
The bond length between aluminum and oxygen atoms in
an octahedron vertex that is outside of the HO}Al}OH
plane (along text, this oxygen will be named as the oxygen in
octahedron vertex) corresponds to d ; d is the bond length
TABLE 6
Some Bond Angles (Fig. 7) in the Representative Octahedron
as a Function of Boehmite Crystallite Size
d
(nm)
ꢀ
g
d
(
ꢆꢃꢆ)
1.13(1)
.56(2)
ꢄ
ꢅ
169.69(3)
170.06(3)
164.72(3)
161.62(3)
160.10(3)
159.37(3)
159.38(3)
190.68(6)
190.74(6)
190.44(6)
187.85(6)
182.98(6)
181.84(6)
181.62(6)
103.93(6)
104.16(6)
105.20(6)
103.80(6)
98.65(6)
97.93(6)
97.73(6)
between oxygen and hydroxyls in HO}Al}OH plane, while
1
d is the bond length between hydroxyl and the oxygen in
2.04(4)
2.42(4)
ꢇ
the octahedron vertex. The bond length between oxygen
6
.90(8)
14.2(2)
6.3(5)
atoms in the plane HO}Al}OH is designated by d , which is
equal to the bond length between hydroxyls in the same
plane; the bond length between oxygen atoms in and outside
ꢈ
2

CRYSTALLITE SIZE AND BOND LENGTHS IN BOEHMITE
37
TABLE 7
methodology used here represents an alternative method for
studying the surface properties of solids, especially their
crystallography.
The oxygen atoms not associated to hydroxyls interact
with four aluminum atoms: two of them in the HO}Al}OH
planes of edge-sharing octahedra (Figs. 5 and 6), and the
other two belonging to vertex-sharing octahedra (Figs.
Bond Lengths as a Function of Boehmite Crystallite Size
d
(nm)
d (nm)
d (nm)
d (nm)
d
(nm)
2
}
(
ꢆꢃꢆ)
ꢇ
ꢈ
ꢉ
- & -
1
1
2
2
6
.13(1)
0.2677(1)
0.2695(1)
0.2712(1)
0.2746(1)
0.2810(1)
0.2825(1)
0.2827(1)
0.28510(4)
0.28796(4)
0.28675(4)
0.28686(4)
0.28695(4)
0.28681(4)
0.28677(4)
0.2763(1)
0.2790(1)
0.2705(2)
0.26277(9)
0.25616(8)
0.25414(8)
0.25391(8)
0.2765(1)
0.2788(1)
0.2818(1)
0.2803(1)
0.2691(1)
0.2685(1)
0.2681(1)
.56(2)
.04(4)
.42(4)
.90(8)
4.2(2)
6.3(5)
1 and 5). The bond length (d ) associated to the interaction
ꢃ
of aluminum atom with oxygen atom O1 in the HO}Al}OH
1
2
plane decreased when the boehmite dimensions increased
(
Table 5); the bond length, d , representing the interaction of
ꢄ
aluminum atom with the oxygen atom O1 outside of this
plane, however, was shorter and independent of boehmite
ideal angle (109.53) for the small crystallites, which corres- crystal dimensions (Table 5). That means that the interac-
ponds to a tetrahedral symmetry. It is worth commenting tion along the [001] direction between oxygen and alumi-
that the bond lengths of the small crystallites could be num atoms (Fig. 7c) is stronger, and therefore, it will
representative of their values on boehmite crystals surface of predominate during crystal growing. Consequently, crystals
any dimension. If this is true, the kind of study reported in "rst grow along this direction, explaining the observed
the present paper could be done in any system, and the preferential crystal growing along plates. The above results
FIG. 7. Projections of the representative octahedron in order to detail bond lengths and bond angles: (a) and (b) along the [001] direction, (c) along the
[
100] direction.

3
8
BOKHIMI ET AL.
also suggest that plate dimension along the c axis should be lated between boehmite octahedral double layers. Lippens
longer as observed by electron microscopy (26). This also (32) and Baker et al. (33) based their analysis on only the
explains why in the present study, the (002) peak of the (020) re#ection of the di!raction pattern (Baker correlated
di!raction pattern could not be "tted well (Fig. 4). The X-ray powder di!raction with NMR experiments). This
corresponding calculated peak was always broader than the kind of analysis with only one di!raction peak gives many
experimental one, because the software used in the re"ne- degrees of freedom when models for the crystalline structure
ment assumes that crystal dimensions along the [100] and are constructed. The Rietveld re"nement, however, is a full
[
001] directions are equal.
It is interesting to note that for the thickest crystal bond crystalline structure must be compatible with all observed
lengths d and d , between the oxygen in the HO}Al}OH di!raction peaks; therefore, any model derived from this
pattern analysis technique, and any proposed model for the
ꢅ
ꢉ
plane and the oxygen of hydroxyls and oxygen in the vertex kind of analysis will be more realistic than those derived
respectively were notoriously shorter than the correspond- from the information involved in only one di!raction peak.
ing length of the hydrogen bond between octahedra double In the present work, we observed a maximal variation of
layers (Figs. 4 and 5 and Tables 5 and 7); this di!erences b lattice parameter of 0.005 nm between the samples with
were larger than 0.01 nm. This strong interaction causes average crystallite size of 26.3(5) and those with crystallite
octahedra deformation in direction to the center of the size of 2.04 nm. This variation is 16 times smaller than the
octahedra double layer (Fig. 8a). The shortness of these one reported by Lippens (32), which was 0.08 nm. Consider-
lengths suggests the possible existence of hydrogen bonds ing this small b lattice variation, it is impossible to interca-
between these oxygen atoms; therefore, the hydrogen bond late water molecules between octahedra double layers. For
traditionally thought to occur in boehmite's crystalline that reason, the water retained in boehmite samples, as
structure needs to be revised. This analysis should be done proposed by Baker et al. for boehmite samples with a very
via neutron di!raction in samples with large crystals. Bond small crystallite size (33), has another origin, which is related
lengths d and d increased as crystallite size diminished to the hydroxyls on a crystal surface as is explained in the
ꢅ
ꢉ
(
Table 5), indicating a weaker interaction between the corre- following paragraph.
sponding oxygen atoms, which increased octahedra sym-
metry around the c axis (Fig. 8b).
Boehmite crystals are thin plates with its thickness along
(020) direction; therefore, most crystal surface is made of
planes perpendicular to it. Because the interaction between
octahedra double layers is weaker than the interaction in-
side the layers, crystal cleavage should occur between
Water Adsorption
The crystallography results of the present analysis ques- double layers and not inside them, producing crystal surfa-
tion the claim that water is intercalated between boehmite ces totally covered with hydroxyls bonded to aluminum
octahedra double layers (32) and extend the fact that water atoms (Fig. 1). The oxygen atoms of these surface hydroxyls
from samples with small crystallite size comes from crystal- have a free orbital that give rise to the hydrogen bonds when
lites surface (33). By correlating the maximum position of crystals continue growing perpendicular to the surface. If
(020) re#ection with the amount of water retained in the crystals do not grow any more, this oxygen can react easily
samples as a function of boehmite crystallite size, Lippens with atoms of an aqueous environment; when during syn-
concluded that, like in other silicates (34), water was interca- thesis the environment is acidic, the free orbital will react
with a proton, forming an aquo ligand with an aluminum
atom (35); if the environment is basic, which corresponds to
the experimental condition used for the synthesis, it will
form a hydrogen bonding with the hydroxyls of the environ-
ment. In this last case, a lot of weakly bonded water is
formed on the crystal surface, which will be eliminated at
relatively low temperatures. The above analysis explains
why boehmite with a low crystallite size lost a lot of water
when samples were annealed at temperatures below 2003C
(14, 36) and discards the claim that pseudoboehmite is boeh-
mite with intercalated water between the octahedra double
layers (8).
Transitional Alumina Derived from Boehmite
FIG. 8. Projection of the representative octahedra along the [001]
direction: (a) for the sample heated at 2403C (large crystallite size); (b) for
the one heated at 503C (small crystallite size).
From Al}O}OH systems, transitional aluminas are of big
interest, not only from the scienti"c point of view (16, 37) but

CRYSTALLITE SIZE AND BOND LENGTHS IN BOEHMITE
39
also from their applications (1}4). They are characterized by dimensions. Al}OH interaction was stronger for small crys-
having very small crystallite size and by adsorbing hy- tallite sizes and the angle between these Al}OH bonds
droxyls on their surface (38). Since the crystallography of tended to a tetrahedral symmetry. The hydrogen bonding
these aluminas is not well known (although many authors sustaining boehmite's crystalline structure and the bonding
have proposed models for it (39}41)), it is impossible to start between oxygen atoms and hydroxyls inside the double
from alumina crystallography to determine the way hy- layers were weaker as crystallite size decreased, which could
droxyls are on crystal surface. The description for boehmite explain why its transformation temperature into
surface given in the last paragraph, however, could be an a transitional alumina also decreases with the crystallite
alternative and realistic model for describing these hy- size. Since boehmite crystals were made of plates perpen-
droxyls on an alumina surface, because, like in boehmite, dicular to [020] direction, crystal surface is full of hydroxyls
this surface is made only of aluminum and oxygen atoms interacting with water molecules, which explains the large
and hydroxyls.
amounts of desorbed water when samples were annealed.
The transitional aluminas derived from boehmite had The crystallite size of alumina depended on boehmite crys-
a crystallite size that depended on the precursor crystal tallite size. The results of the present study suggests some
dimensions: Boehmite with large (small) crystals gave rise to tendencies about boehmite behavior, which could be veri"-
alumina with large (small) crystals (Table 8). The grain size ed by preparing new samples and by using additional char-
of the transitional alumina was similar to the crystallite size acterization techniques; for example, the possibility that
of the precursor boehmite, because the transformation of microcrystalline boehmite had a tetragonal symmetry, or
boehmite into a transitional alumina, occurring by dehyd- that the characteristics determined for the boehmite with
roxylation at temperatures below 8003C, is pseudomorphic, very small crystallite size could correspond to the behavior
and it involves atom displacements within only a single of boehmite or transitional alumina surface.
boehmite crystal. At higher temperatures, aluminum and
oxygen atoms move between transitional alumina crystals,
producing a-alumina with large crystallite dimensions. In
order to get a "rst approximation for the crystallite size of
transitional alumina, its crystalline structure was modeled
with the monoclinic unit cell reported for h-alumina (10):
Atom positions have the symmetry described by space
group C2/m. The residue obtained with this model by com-
paring experimental and the calculated data was 10%
smaller than what one got with a nondeformed cubic unit
cell having the spinel structure.
ACKNOWLEDGMENTS
We thank Mr. A. Morales and Mr. Manuel Aguilar for technical assist-
ance. This work was "nancially supported via Grants IMP-D.01024 and
IMP-D.01234.
REFERENCES
1
. Z. R. Ismagilov, R. A. Shkrabina, and N. A. Koryabkina, Catal. ¹oday
7, 51 (1999).
. J. Hietala, A. R. Root, and P. Knutila, J. Catal. 150, 46 (1994).
. F. Barath, M. Turki, V. Keller, and G. Maire, J. Catal. 185, 1 (1999).
4
2
3
CONCLUSIONS
4. M. F. L. Johnson, J. Catal. 123, 245 (1990).
5
6
7
. J. Singh, J. Mater. Eng. Perform. 3, 378 (1994).
. A. C. Pierre, E. Elaloui, and G. M. Pajonk, ¸angmuir 14, 66 (1998).
. A. S. Brown, M. A. Spackman, and R. J. Hill, Acta Crystallogr. 49, 513
When boehmite precipitated at room temperature was
heated between 23 and 2403C under hydrothermal condi-
tions, its crystallite size grew between 1 and 27 nm, respec-
tively. The Rietveld re"nement of the crystalline structure
revealed that boehmite bond lengths depended on crystallite
(1993).
8
. J. J. Fitgerald, G. Piedra, S. F. Dec, M. Seger, and G. E. Maciel, J. Am.
Chem. Soc. 119, 7832 (1997).
9
. R.-S. Zhou and R. L. Snyder, Acta Crystallogr. B 47, 617 (1991).
1
1
0. E. Husson and Y. Repelin, J. Solid State Inorg. Chem. 33, 1223 (1996).
1. I. Seyssiecq, S. Veesler, G. Pepe, and R. Boistelle, J. Cryst. Growth
96, 174 (1999).
`
1
TABLE 8
12. Y. Cesteros, P. Salagre, F. Medina, and J. E. Sueiras, Chem. Mater.
11, 123 (1999).
13. S. Keyser, G. K. Shter, Y. De Hazan, Y. Cohen, and G. S. Grader,
Chem. Mater. 9, 2464 (1997).
Transitional-Alumina Average Crystallite Size as a Function
of Boehmite Crystal Dimension
1
4. G. Chuah, S. Jaenicke, and T. H. Xu, Microporous Mesoporous Mater.
37, 345 (2000).
d
(nm)
d(nm)
(ꢆꢃꢆ)
1
5. M. L. Guzma
Blasquez, and F. Herna
2001).
6. T. Tsukada, H. Segawa, A. Yasumori, and K. Okada, J. Mater. Chem.
, 549 (1999).
7. H.-L. Wen and F.-S Yen, J. Cryst. Growth 208, 696 (2000).
8. L. Farkas, P. Gado, and P.-E. Werner, Mat. Res. Bull. 12, 1213 (1977).
9. W. O. Milligan and J. L. McAtee, J. Phys. Chem. 60, 273 (1956).
H
n-Castillo, X. Bokhimi, A. Toledo-Antonio, J. Salmones-
1
1
2
2
6
.13(1)
2.65(7)
2.69(7)
3.0(1)
3.3(1)
4.5(1)
6.2(2)
6.6(2)
H
H
ndez-Beltran, J. Phys. Chem. B 105, 2099
H
.56(2)
.04(4)
.42(4)
.90(8)
(
1
9
1
1
1
1
2
4.2(2)
6.3(5)
H

4
0
BOKHIMI ET AL.
2
2
0. R. A. Young, A. Sakthivel, T. S. Moss, and C. O. Paiva-Santos, J.
Appl. Crystallogr. 28, 366 (1995).
1. J. Rodr mH guez-Carbajal, Laboratoire Leon Brillouin (CEA-CNRS),
31. L. N. Dem'yanets and A. N. Lobachev, &&Crystallization Process under
Hydrothermal Conditions'' (A. N. Lobachev and G. D. Archard, Eds.),
p. 1. Consultants Bureu, New York, 1973.
France Tel: (33) 1 6908 3343, Fax: (33) 1 6908 8261, E-mail: juan@ 32. B. C. Lippens, &&Structure and Texture of Aluminas.'' Doctoral thesis,
llb.saclay.cea.fr.
Technical University of Delft, Amsterdam, 1961.
2
2. P. Thompson, D. E. Cox, and J. B. Hasting, J. Appl. Crystallogr. 20,
33. B. R. Baker and R. M. Pearson, J. Catal. 33, 265 (1974).
34. A. De Roy, C. Forano, K. El Malki, and J. P. Besse, in &&Procc.
Synthesis of Microporous Materials'' (M. L. Occelli and H. Robson,
Eds.), Vol. 2, p. 108. Van Nostrand Reinhold, New York, 1992.
35. I. P. Kuz'mina, B. N. Litvin, and V. S. Kurazhkovskaya, in &&Crystal-
lization Process under Hydrothermal Conditions'' (A. N. Lobachev
and G. D. Archard, Eds.) p. 185. Consultants Bureu, New York, 1973.
36. D. L. Coke, E. D. Johnson, and R. P. Merril, Catal. Rev. Sci. Eng.
26, 163 (1984).
7
9 (1987).
2
2
2
3. R. A. Young and P. Desai, Arch. Nauki Mat. 10, 71 (1989).
4. E. Prince, J. Appl. Crystallogr. 14, 157 (1981).
5. C. E. Corbato
Miner. 33, 71 (1985).
6. X. Bokhimi, D. Acosta, M. L. Guzma
and F. Hernandez-Beltran, in preparation.
7. D. Chamma, Doctoral thesis, &&Les resonances de Fermi en liaison
hydrogene X}H2Y: pro"ls theoriques du spectre infrarouge de
l'elongation X}H.'' Universite de Perpignan, France, 1999.
H , R. T. Tettenhorst, and G. C. Christoph, Clays Clay
2
2
H n-Castillo, A. Toledo-Antonio,
H
H
H
`
H
37. K. Sohlberg, S. J. Pennycook, and S. T. Pantelides, J. Am. Chem.
Soc. 121, 7493 (1999).
H
H
2
2
3
8. H. Christensen and N. Christensen, Acta Chem. Scand. A 32, 87 38. X. Liu and R. E. Truit, J. Am. Chem. Soc. 119, 9856 (1997).
(
1978).
9. A. Ole
1970).
0. A. Vazquez, T. Lo
1 (2001).
39. H. Knozinger and P. Ratnasamy, Catal. Rev. Sci. Eng. 17, 31 (1978).
s, A. Szytula, and A. Wanic, Phys. Status Solidi 41, 173 40. V. E. Henrich and P. A. Cox, &&The Surface Science of Metal Oxides,''
Section 2.3, Cambridge Univ. Press, New York, 1994.
H
(
H
H
pez, R. Go
H
mez, and X. Bokhimi, J. Mol. Catal. A 167, 41. A. A. Tsyganenko and P. P. Mardilovich, J. Chem. Soc. Faraday ¹rans.
9
92, 4843 (1996).