Cement and Concrete Research 40 (2010) 1248–1254

Contents lists available at ScienceDirect

Cement and Concrete Research

j o u r n a l h o m e p a g e : h t t p : / / e e s. e l s ev i e r. c o m / C E M C O N / d e f a u l t . a s p

Synthesis and characterization of the LDH hydrotalcite–pyroaurite

solid-solution series
K. Rozov a,c,⁎, U. Berner a, C. Taviot-Gueho b, F. Leroux b, G. Renaudin b, D. Kulik a, L.W. Diamond c
a
Waste Management Laboratory, Paul Scherrer Institute, 5210 Villigen, Switzerland
b
Laboratoire de Matériaux Inorganiques, CNRS UMR 6002, Université Blaise Pascal, F-63177 Aubière Cedex, France
c
Rock-Water Interaction Group, Institute of Geological Sciences, University of Bern, Baltzerstrasse 3, CH-3012, Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: A layered double hydroxide (LDH) hydrotalcite–pyroaurite solid-solution series Mg3(AlxFe1 − x)(CO3)0.5
Received 5 June 2009 (OH)8 with 1 − x = 0.0, 0.1……1.0 was prepared by co-precipitation at 23 ± 2 °C and pH = 11.40 ± 0.03. The
Accepted 25 August 2009 compositions of the solids and the reaction solutions were determined using ICP-OES (Mg, Al, Fe, and Na)
−
and TGA techniques (CO2− 3 , OH , and H2O). Powder X-ray diffraction was employed for phase identification
Keywords: and determination of the unit cell parameters ao and co from peak profile analysis. The parameter ao = bo
Hydrotalcite
was found to be a linear function of the composition. This dependency confirms Vegard's law and indicates
(B) Characterization
Thermodynamic Calculations
the presence of a continuous solid-solution series in the hydrotalcite–pyroaurite system. TGA data show that
(E) Waste Management the temperatures at which interlayer H2O molecules and CO2− 3 anions are lost, and at which dehydroxylation

Modeling of the layers occurs, all decrease with increasing mole fraction of iron within the hydroxide layers.
Features of the Raman spectra also depend on the iron content. The absence of Raman bands for Fe-rich
members (xFe > 0.5) is attributed to possible fluorescence phenomena.
Based on chemical analysis of both the solids and the reaction solutions after synthesis, preliminary Gibbs
free energies of formation have been estimated. Values of ΔG°f(hydrotalcite) = − 3773.3 ± 51.4 kJ/mol and
ΔG°f(pyroaurite) = − 3294.5 ± 95.8 kJ/mol were found at 296.15 K. The formal uncertainties of these
formations constants are very high. Derivation of more precise values would require carefully designed
solubility experiments and improved analytical techniques.
© 2009 Elsevier Ltd. All rights reserved.

1. Introduction presently hampered by: 1) scarce information on thermodynamic and

mixing properties of LDH solid solutions, and 2) lack of detailed
The naturally occurring hydrotalcite–pyroaurite minerals Mg3(Al, knowledge of the atomic-scale uptake mechanisms for cationic and
Fe)(OH)8(CO3)0.5 nH2O belong to the layered double hydroxide (LDH) anionic contaminants.
family, which is also known as the “anionic clays”. The hydroxide layers The aim of the present work was to carry out a careful synthesis
have a brucite-like crystal structure with a permanent positive charge and characterization of the binary hydrotalcite–pyroaurite series
due to the presence of trivalent Al3+ (Fe3+) ions. This charge is Mg3Al(OH)8(CO3)0.5 nH2O–Mg3Fe(OH)8(CO3)0.5 nH2O with varying
compensated by anions – carbonate anions in the present case – located mole fractions of Fe (0.0, 0.1……1.0). Using various experimental
in the interlayer space along with water molecules. LDH solid solutions techniques (X-ray power diffraction, Raman scattering spectroscopy,
are common secondary phases in the contact zone between clays and and thermogravimetric analysis), the primary objective was to
cementitious materials, a situation which may arise in the engineered ascertain the presence of a continuous solid-solution series in this
barrier system of nuclear waste disposal sites. There is growing evidence system. The second objective was to quantify the Gibbs free energy of
that LDH phases may play an important role in the retention of formation and to improve the chemical thermodynamic database [5]
hazardous cations and especially anions (I−, SeO2− 2− 2−
4 , SeO3 TcO4 , etc.) with the help of a thermodynamic model, based on the chemical
[1]. We believe that LDH phases can be treated as solid solutions, with analysis of solid and liquid phases at equilibrium.
substitution in divalent and trivalent cationic positions within the
hydroxide layer and, possibly, anion exchange in the interlayer space.
However, the understanding of retention properties of LDH phases is 2. Materials and methods

2.1. LDH synthesis

⁎ Corresponding author. Waste Management Laboratory, Paul Scherrer Institute,
5210 Villigen, Switzerland. Samples of hydrotalcite–pyroaurite series with varying mole
E-mail address: urs.berner@psi.ch (K. Rozov). fraction of iron (0.0; 0.1; 0.2; 0.3; 0.4; 0.5; 0.6; 0.7; 0.8 0.9; and 1.0)

0008-8846/$ – see front matter © 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2009.08.031
 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254 1249

have been prepared by a co-precipitation method according to [2] at

23 ± 2 °C and pH = 11.40 ± 0.03.
The syntheses were performed under N2 gas flow by slow addition
(at a rate of 0.04 mL/min ) of 60 mL 0.1 M metal-nitrate solution with
Mg/(Al + Fe) = 3.0 ± 0.1 into a reactor containing 300 mL of stirred
0.025 M Na2CO3 solution (Fig. 1). In order to produce increased
amounts of solid, a second set of syntheses was performed using a six-
fold amount of the solutions in a 2000 mL reactor. To avoid
precipitation of iron hydroxides, sufficient amounts of concentrated
HNO3 had previously been added to the Fe-containing nitrate stock
solutions. To maintain pH ≈ 11.40, a 2 M NaOH solution was added
simultaneously (at a maximum rate of 0.05 mL/min) into the reactor
using an automated titrator “Titroprocessor 670” (Metrohm AG). All
stock solutions were prepared using Merck® “pro analysis” chemicals
and MilliQ water generated by the “Millipore” water purification
system. The overall reaction can be described as:

3MgðNO3 Þ2 þ xAlðNO3 Þ3 þ ð1−xÞFeðNO3 Þ3 þ 8NaOH

þ 0:5Na2 CO3 →Mg3 ½Alx Fe1−x ðOHÞ8 ðCO3 Þ0:5 d nH2 O þ 9NaNO3 ð1Þ

After the addition step, the suspension was aged for 24h in the
stirred reactor under controlled pH (pH = 11.40 ± 0.03). It was then Fig. 1. Experimental setup for the co-precipitation synthesis of hydrotalcite–pyroaurite
solids.
separated from the mother solution by centrifuging at 95,000 g for 1 h.
About 40 mL of MilliQ water was used to wash about 500 mg of the
precipitate and the mixture was then shaken and centrifuged again at The cell parameters were also determined from the Bragg
95,000 g for 10 min. This washing procedure was repeated 3 times until equation, using the relationship between interlayer distances (dhkl),
the electrical conductivity of the supernatant was below 12 µS, which the h,k,l indexes, and the lattice parameters for hexagonal symmetry
means that the concentration of NaNO3 (provided that sodium nitrate is based only the first 4 reflections. The results of both methods are
the only solute) in these solutions is below 7.2 10− 5 mol/L. This in turn represented in Figs. 4 and 5.
means that NaNO3 impurities in the solid phase were below 0.05 wt.%.
The washed solids were finally dried in an oven at 60 °C for 24h. 2.2.5. Thermogravimetric analyses (TGA)
TGA was carried out in order to determine the contents of
interlayer water, and of hydroxide and carbonate anions in the solids.
2.2. Analyses and characterization
Measurements were performed on a Mettler Toledo TGA instrument.
Before the measurements the samples were dried at 60 °C for 15 min.
2.2.1. ICP-OES analyses
The weight loss of the solids in air was analyzed from 60 °C to 1000 °C
The contents of Mg, Al, Fe and Na in the solid and liquid phases
with a heating rate of 5 °C/min.
were determined by ICP-OES using a Perkin-Elmer (type Vista Pro)
instrument. The synthesis solutions were analyzed after acidifying
with some drops of concentrated nitric acid (dilution was recorded by 2.2.6. Raman spectroscopy
a balance). The solids were analyzed after dissolving 20 mg of solid in Raman scattering spectra were acquired using a Jobin Yvon Horiba
60 g of 0.5 M HNO3. Internal standards were prepared from ICP quality LabRam HR 800 spectrometer equipped with an Olympus BX41
(Merck®) multi-element standard solutions of Mg, Al, Fe and Na. petrographic microscope and a 532.12 nm (green) frequency-doubled
Nd:YAG laser. The spectra were calibrated using silicon and Ne standards.
Before measuring, the samples were dried at 60 °C for 6 h. The spectra
2.2.2. Ion chromatography were manipulated and recorded using the LabSpec™ v. 4.14 software. The
Nitrate anions in the product liquid phase were analyzed band component analysis was carried out using the “PeakFit” software
spectrophotometrically after separation by ion chromatography package, and the band fitting and smoothing were done using Gaussian
(Dionex DX-600 Ion Chromatograph with IonPac AS16/AG16 chro- functions. The spectra were recorded from 250 to 3800 cm− 1.
matographic columns). Solutions of KNO3 (Merck®) were used for the
preparation of the internal standards.
2.2.7. Determination of solid compositions
In order to calculate the compositional formulae of the solids we
2.2.3. Total inorganic carbon analysis (TIC) combined the ICP-OES analyses of Mg, Al, Fe and Na with the TGA
A Shimadzu TOC-V analyzer was used to determine the carbonate measurements: we assumed that the solid products after the TGA
content in the liquid phase. Internal standard solutions were prepared analysis (at 1000 °C) are composed of mixtures of MgO, Al2O3 and
using Na2CO3 (Merck®) and degassed MilliQ water. Fe2O3 and that the removal of interlayer water, hydroxyl and
carbonate groups can be described as follows:
2.2.4. Powder X-ray diffraction (PXRD)
The powder X-ray diffraction (PXRD) technique was applied to Mg3 ðAlx Fe1−x ÞðCO3 Þ0:5 ðOHÞ8 d nH2 O→Mg3 ðAlx Fe1−x ÞðCO3 Þ0:5 ðOHÞ8
þ nH2 O ð2Þ
characterize the precipitated solids using a Phillips XPert-Pro
diffractometer. No internal standards for possible specimen displace-
Mg3 ðAlx Fe1−x ÞðCO3 Þ0:5 ðOHÞ8 →Mg3 ðAlx Fe1−x ÞðCO3 Þ0:5 O4 þ 4H2 O ð3Þ
ment were used. The diagrams were recorded with CuKα radiation at
ambient temperature within a 2Θ-range from 5° to 70°, using a step
size of 0.0168° and a counting time of 20 min. Mg3 ðAlx Fe1−x ÞðCO3 Þ0:5 O4 →Mg3 ðAlx Fe1−x ÞO4:5 þ 0:5CO2 ð4Þ
The cell parameters were determined from peak profile analysis
using the program “FullProf” (full pattern matching — pseudo–Voigt The stoichiometric coefficients given in Table 1 can be derived
profile function) [3]. using the Mg/(Al + Fe) ratio determined by ICP-OES, the amount of
1250 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254

Table 1
Stoichiometric formulae, estimated standard Gibbs free energies and total solubility products of the synthesized solids.

Mole fraction of iron in solids. (xFe) Stoichiometric formulaea (including analytical errors) MgII/(AlIII + FeIII) G°f kJ/mol (without interlayer water) logKsp
in solid phase

0 Mg3 Al1:019 ðCO3 Þ0:472 ðOHÞ8:114 ⋅2:53H2 O 2.94 ± 0.29 − 3773.43 ± 124.15 − 68.92 ± 3.50
ð0:102Þ ð0:031Þ ð0:243Þ

0 Mg3 Al1:021 ðCO3 Þ0:666 ðOHÞ7:730 ⋅2:46H2 O 2.94 ± 0.29 − 3818.71 ± 118.97 − 69.52 ± 3.54
ð0:099Þ ð0:019Þ ð0:258Þ

0.098 ± 0.007 Mg3 Al0:896 Fe0:097 ðCO3 Þ0:336ðOHÞ8:305 ⋅2:51H2 O 3.02 ± 0.27 − 3671.87 ± 112.49 − 68.76 ± 3.55
ð0:083Þ ð0:019Þ ð0:017Þ ð0:271Þ

0.192 ± 0.012 Mg3 Al0:827 Fe0:192ðCO3 Þ0:536 ðOHÞ7:987⋅2:62H2 O 2.94 ± 0.24 − 3690.29 ± 112.49 − 67.79 ± 3.62
ð0:080Þ ð0:020Þ ð0:017Þ ð0:267Þ

0.203 ± 0.013 Mg3 Al0:802 Fe0:205 ðCO3 Þ0:537 ðOHÞ7:946 ⋅2:67H2 O 2.98 ± 0.24 − 3701.77 ± 111.44 − 71.61 ± 3.75
ð0:080Þ ð0:020Þ ð0:017Þ ð0:267Þ

0.304 ± 0.019 Mg3 Al 0:695 Fe 0:304 ðCO3 Þ 0:481 ðOHÞ8:034⋅2:52H2 O 3.00 ± 0.23 − 3623.89 ± 104.99 − 70.83 ± 3.61
ð0:267Þ
ð0:069Þ ð0:030Þ ð0:016Þ
2.94 ± 0.21 − 3604.44 ± 103.58 − 68.7 ± 3.65
0.391 ± 0.024 Mg3 Al 0:625 Fe0:394ðCO3Þ0:553ðOHÞ 7:951 ⋅2:50H2 O
ð0:062Þ ð0:039Þ ð0:019Þ ð0:267Þ
3.01 ± 0.21 − 3499.21 ± 95.95 − 69.25 ± 3.63
0.497 ± 0.030 Mg3 Al0:502Fe0:496ðCO3Þ0:361ðOHÞ 8:276 ⋅2:45H2 O
ð0:050Þ ð0:050Þ ð0:013Þ ð0:278Þ
0.603 ± 0.037 Mg3 Al 0:415 Fe 0:619 ðCO3 Þ0:531ðOHÞ 8:039 ⋅2:47H2 O 2.90 ± 0.21 − 3513.97 ± 95.17 − 69.51 ± 3.74
ð0:041Þ ð0:062Þ ð0:020Þ ð0:270Þ
2.99 ± 0.23 − 3387.14 ± 85.67 − 70.09 ± 3.68
0.701 ± 0.044 Mg3 Al0:299Fe0:703ðCO3 Þ0:248ðOHÞ8:509⋅2:55H2 O
ð ð0:030Þ ð0:070Þ ð0:009Þ ð0:281Þ
0.805 ± 0.052 Mg3 Al 0:207 Fe 0:839 ðCO3 Þ 0:491 ðOHÞ 8:157 ⋅2:55H2 O 2.87 ± 0.24 − 3418.14 ± 86.37 − 70.23 ± 3.81
ð 0:021Þ ð0:084Þ ð0:019Þ ð0:275Þ
0.897 ± 0.060 Mg3 Al 0:108 Fe 0:902 ðCO3 Þ 0:342 ðOHÞ 8:344 ⋅2:64H2 O 2.97 ± 0.27 − 3323.14 ± 71.95 − 70.38 ± 3.53
ð 0:005Þ ð0:090Þ ð0:013Þ ð0:260Þ
2.76 ± 0.28 − 3321.52 ± 78.70 − 72.36 ± 3.99
1 Mg3 Fe 1:086 ðCO3 Þ 0:343 ðOHÞ 8:570 ⋅2:15H2 O
ð0:108Þ ð0:014Þ ð0:296Þ

a
Calculated mole numbers of components are given to three decimal places to reach electrical neutrality. The real certainty of the analyses is lower, as indicated by numbers in
brackets, e.g. (0.102) denotes ±0.102.

CO2 and H2O lost between 60 °C and 1000 °C, and considering the a–e: stoichiometric coefficients; μi: chemical potentials, evaluated
electroneutrality of the solid. The uncertainties are based on the from the synthesis solution.
assumption that the analytical error of the applied ICP-OES technique The total solubility products of individual solids were also
is ±5% for Mg, Al and Fe and that the error of TGA analyses is well calculated based on the activities of the relevant components in the
below 1%. Note that the analysis of Na in all samples was less than “synthesis” solutions, using Eq. (6):
0.015 ± 0.005 wt.%, which is a very strong indication that Na-
containing phases (i.e., NaNO3) are absent after the washing and 2þ a 3þ b 3þ c −d 2− e
drying procedure (see above). log10 Ksp ¼ log10 ½Mg Al Fe OH CO3 ; ð6Þ

2.2.8. Estimation of Gibbs free energies

Standard Gibbs free energies of formation of the solids were 3. Results
estimated according to the following scheme: 1) the speciation of the
synthesis solutions was modelled in order to obtain the activities and All the solids display powder X-ray diffraction spectra typical
chemical potentials of relevant solutes (i.e., Mg2+, Al3+, Fe3+, OH−, of LDH materials (Fig. 2). However, the peaks are broad, pre-
and CO2− 3 ). To perform these calculations we used the Gibbs free sumably due to the simultaneous effects of small coherent domain
energy minimization software GEMS [4], including the thermody- size and structural disorder. This prevented us from performing
namic data given in [5]. From the chemical potential of the solutes and Rietveld structural refinements. The diffraction peaks were in-
from the stoichiometric coefficients in Table 1, the Gibbs free energies dexed on a hexagonal unit cell with the space group R-3m [6,7].
were obtained according to: The cell parameters gathered in Table 2 were obtained from pro-
file peak analysis and from a Bragg-type evaluation. An example of
−
Gf ðHtlc−PyrÞ ¼ aμðMg Þ þ bμðAl Þ þ cμðFe Þ þ dμðOH Þ
2þ 3þ 3þ
the unit cell refinement is given in Fig. 3. It is also important to
2−
þ eμðCO3 Þ; ð5Þ note that none of the spectra indicates the presence of a second
phase.
The variation of the cell parameter a as a function of the iron
content x (Fig. 4) is in excellent agreement with the Vegard's law
[8], thereby demonstrating the existence of a continuous solid
solution throughout the hydrotalcite–pyroaurite series. By linear
regression it was found that ax = (3.0600 ± 0.0004) + (0.0500 ±
0.0007) × x3+
Fe (Å).
This equation can be compared to that expected for an idealized
octahedral layer structure built with regular M(OH)6 octahedra. In
this case, p
the
ffiffiffi cell parameter a of the−rhombohedral cell must be
equal to 2dM–OH , dM–OH (=Rn+ M + ROH) being pthe
ffiffiffi mean M–OH
da
distance within the hydroxide layer and dR = 2. On the other
hand, one can calculate a mean ionic radius within the hydroxide
layers equal to:

2þ 3þ 3þ
Rm ¼ 0:75RMg þ 0:25xRAl þ 0:25ð1−xÞRFe ð7Þ

Fig. 2. X-ray diffractograms of synthetic hydrotalcite–pyroaurite solid-solution series in the 3:1 hydrotalcite–pyroaurite structure and dRm
dx
= 0:25⋅ðR3+
Al +
with xFe = 0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0 (bottom – up). R3+
Fe Þ.
 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254 1251

Table 2
Cell parameters determined from peak profile analysis and from Bragg evaluation and
refined in the space group R-3m.

Approximate ao = bo., Å co, Å

mole fraction
W-B Rv W-B Rv
of iron in solid

0.0 3.063 ± 0.004 3.0615 ± 0.0003 23.445 ± 0.074 23.380 ± 0.006

0.1 3.064 ± 0.001 — 23.346 ± 0.018 —
0.2 3.071 ± 0.001 3.0694 ± 0.0002 23.380 ± 0.024 23.352 ± 0.003
0.3 3.074 ± 0.001 — 23.354 ± 0.014 —
0.4 3.081 ± 0.001 3.0758 ± 0.0002 23.428 ± 0.014 23.392 ± 0.004
0.5 3.085 ± 0.001 — 23.409 ± 0.014 —
0.6 3.093 ± 0.001 3.0909 ± 0.0003 23.434 ± 0.020 23.382 ± 0.005
0.7 3.096 ± 0.001 — 23.439 ± 0.012 —
0.8 3.101 ± 0.001 3.0995 ± 0.0002 23.460 ± 0.012 23.392 ± 0.003
0.9 3.105 ± 0.001 — 23.481 ± 0.010 —
1.0 3.110 ± 0.001 3.1107 ± 0.0002 23.539 ± 0.016 23.537 ± 0.003

W-B — results obtained from the Bragg evaluation, Rv — results obtained from profile
peak analysis. Fig. 4. Unit cell parameters (a) as a function of the mole fraction of iron. The shaded area
represents the fitted slope of 0.050 and encompasses all experimental points including
their uncertainties. The dashed line represents the theoretical curve based on a regular
octahedral layer. Filled diamonds represent refined parameters determined from peak
Since da
dx
= dR ⋅ dx and using R3+
da dR
Al (in octahedral coordination) = profile analysis and filled circles represent parameters evaluated using the Bragg-type
3+
0.535 Å, RFe (in octahedral coordination) = 0.645 Å [9] and Mg2+/ equation. Open squares represent solids produced in a reactor vessel with increased
(Al3+ + Fe3+) = 2.91 ± 0.15, one can calculate the slope dao/dx: volumes (2000 mL).

da pffiffiffi According to the literature [11–13] the thermal decomposition of

= 2⋅ð0:26F0:01Þ⋅ðRFe3 + −RAl3 + Þ = 0:042F0:002 ð8Þ
dx LDHs includes three main stages, i.e., the loss of interlayer water, the
decomposition of structural hydroxyl groups and finally the decom-
We observe a significant difference between the theoretical slope position of interlayer carbonate anions.
of 0.042 based on regular M(OH)6 octahedra and the observed slope For the present compounds, these three decomposition steps were
of 0.050. Such a difference can be easily explained by the fact that evaluated by determining the second derivative of the TGA-curve (Fig. 7).
M(OH)6 octahedra within LDH hydroxide layers are distorted, It was observed that an increase of the mole fraction of iron causes a
being flattened in the direction normal to the layers and extended decrease of the temperature of dehydroxylation of brucite-like layers
in a plane parallel to the layers, even more than is found in brucite from 230 to 180 °C, in agreement with the fact that the short Al–OH bond
or gibbsite [10]. distance is stronger than the longer Fe–OH bond distance. Similarly, the
The lattice parameter c (Fig. 5 and Table 2, see also Section 2.2.4.) temperature of decarbonation decreases from 440 to 407 °C with
corresponds to the 3 layer rhombohedral polytype 3R (Fig.6). The increasing mole fraction of iron.
resulting interlayer distance d003 = c/3 depends on the layer charge Raman spectroscopy is specifically suited to identify the nature of
density, the nature of the interlayer anion (CO2−
3 ) and on the number the interlayer anion and to investigate its interactions with the rest of
of water molecules in the interlayer space. Since the Mg2+/(Al3+ + the structure, particularly the hydrogen-bond network.
Fe3+) ratio is kept constant, one can reasonably attribute the slight Raman bands (around 540 cm− 1, 1060 cm− 1, 1370–1390 cm− 1
increase of d003 as a function of x from 7.793 to 7.845 Å to the and 3400–3700 cm− 1) which are typical [14] for hydrotalcite structure
replacement of Al3+ (R3+ Al = 0.535 Å) by Fe
3+
, which displays a were observed only in solids with xFe ≤ 0.5 (see discussion below). The
3+
higher ionic radius (RFe = 0.645 Å) and thus longer M–OH dis- band around 540 cm− 1 (Fig. 8A) originates from the interlayer
tances. This interpretation is consistent with the chemical analysis carbonate-water unit, where the two hydrogen atoms of the H2O
showing a constant Mg2+/(Al3+ + Fe3+) ratio and no remarkable molecule are bridged to two oxygen atoms of the carbonate anion,
change of the composition of the interlayer space with respect to
both water molecules (approx. 2.51 ± 0.13 per formula unit) and
interlayer anions (Table 3).

Fig. 5. Dependence of the unit cell parameter co on the mole fraction of Fe. Filled
Fig. 3. Results of the profile analysis of XRD patterns for Mg3Al0.823Fe0.191(CO3)0.532 diamonds represent refined parameters determined from peak profile analysis and
(OH)7.978·2.615H2O (x = 0.191) in the space group R-3m: experimental X-ray filled circles represent parameters evaluated using the Bragg-type equation. Open
diffraction (cross), calculated (line), Bragg reflections (ticks) and difference profiles; squares represent solids produced in a reactor vessel with increased volumes
the refined cell parameters are a = 3,0694(2) Å and c = 23,352(3) Å. (2000 mL).
1252 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254

Fig. 6. Layer–interlayer configuration in R-3m space group: OH groups from adjacent

octahedral layers superimpose, forming a prismatic environment around interlayer
carbonate anions.

strongly indicating the presence of carbonate in the system [14]. The

band around 1060 cm− 1 (Fig. 8B) corresponds to interlayer carbonate
anions (υ1) associated with [M(II), M(III)](OH)6 structural units in the
brucite-type layers. From x = 0 to 0.3 the position of this peak remains Fig. 7. TGA curves of hydrotalcite–pyroaurite samples. The applied heating rate was
5 °C/min. Samples were dried for 15 min at 60 °C before analysis.
almost constant, indicating no change in symmetry of the CO2− 3 anions.
The very weak band around 1370–1390 cm− 1 (υ4) (Fig. 8C) also be-
longs to the carbonate anions, either free or bound to interlayer water for example, that Mg varies over the range 5.7 to 12.0 μmol/kg in two
molecules or hydroxide groups of the brucite-like sheet. The large band identical experiments (first two rows of Table 3). Hence, we
around 3400–3700 cm− 1 (Fig. 8D) represents OH-stretching vibrations concluded that the analytical error for Mg in the “synthesis” solution
from OH groups as well as interlayer water molecules. is substantially higher than 5%. From the average of Mg analyses
The formation of a unique phase of LDH type throughout the whole provided in Table 3, an analytical error of at least ±50% was
composition range has been confirmed by PXRD analysis. Hence, the estimated. For Al, the measured concentrations vary in the range 2.1
disappearance of the Raman signals above x = 0.5 cannot be ascribed to 8.5 μmol/kg but do not show a clear tendency as a function of the
to a structural change. It is likely due to masking by fluorescence of the mole fraction of Fe, as could be expected from general chemical
iron-rich samples. Attempts to acquire spectra using a less absorbing considerations. Hence, from the standard deviation it could be
633 nm He–Ne (red) laser failed to improve the situation. Finally, it is inferred that the analytical error of the Al determinations is also
worth mentioning the absence of a peak at 1080 cm− 1, attributed to about ±50%. However, in the case of pyroaurite the analysis shows
interlayer nitrate anions [14], indicating that the washing procedure traces of Al in the range 0.001 μmol/kg, although Al should not be
was successful and that the investigated solids are free of NaNO3, in present at all in this particular “synthesis” solution. Contrary to the
accordance with the low conductance (12 µS) measured in the above statement this low detection limit indicates that the analytical
washing solution and with the fact that no Na was found in the error of Al is most likely less than ±50%.
analysis of the solid phase. In the case of iron a similar situation was observed. A clear trend of
Chemical compositions of the “synthesis” solutions and solids are analytical results was hard to see in the sequence of Fe concentrations
provided in Tables 1 and 3. It should be noted that the synthesis was from hydrotalcite to pyroaurite, although the values varied from 0.09
carried out in solutions with “high” ionic strength between 0.1 and to 4.9 μmol/kg. A simple estimate based on the standard deviation and
0.2 mol/kg. As a consequence, one measures concentrations at the assuming no trend in the data reveals that the analytical error could
mmol/kg-level for Na, NO− 2−
3 and CO3 . The analytical error for these be as high as ±66%. From simultaneously performed analyses of
concentrations is assumed to be about ±5%. The metals Mg, Al and Fe internal standards, it was inferred that the analytical error for Fe is
were found at concentrations between 0.1 and 22 μmol/kg. According about 15% and for Mg the error is about 10%. However, using these
to [15] the analytical error of the ICP-OES technique is more than ±5% latter uncertainties it turned out that the analytical errors of synthesis
when Fe lies in the range from 10− 8 to 10− 6 mol/kg. A further solution compositions do not significantly contribute to the total error
complication is that these low concentrations of Mg, Al and Fe must be of calculated standard Gibbs free energies (see below).
determined in a matrix of high Na concentration. It was recognized, The determination of the stoichiometric composition of the solid
phases has been outlined in Section 2.2.7. Table 1 reveals that the
average Mg(II)/(Al(III) + Fe(III)) ratio is close to the desired value of
Table 3
3 (2.91 ± 0.15). The analyses of water, hydroxide and carbonate are
Composition of solutions after synthesis at 23–24 °C and pH = 11.40 ± 0.05.
remarkably constant in the range from hydrotalcite to pyroaurite. It was
Mg Al Fe Na NO−
3 CO2−
3 Approximate found that each formula unit contains 2.51± 0.13 H2O, 8.15 ±0.24 OH−
mole fraction and 0.45 ± 0.12 CO2−
μmol/kg mmol/kg 3 . Note that, when contemplating varying composi-
of iron in solid
tions, the uncertainties in the values of carbonate and hydroxide anions
12.0 2.67 Below dl 82.0 37.2 19.1 0.0 must correspond to each other because there are no other anions
5.7 2.05 Below dl 78.0 35.6 20.3 0.0
available (charge balance). It has already been demonstrated above that
8.6 4.83 0.28 120.0 74.5 18.6 0.1
21.8 8.48 1.67 132.0 76.0 18.3 0.2 in washed solids the impurities of NaNO3 are not significant.
1.2 2.08 Below dl 128.0 76.5 20.0 0.2 The solids as described in Table 1 precipitated from the solutions
2.3 7.31 0.09 119.0 74.6 18.9 0.3 presented in Table 3. In this first study we neglected possible effects of
13.6 7.18 2.73 136.0 79.2 18.5 0.4
oversaturation and we assumed equilibrium between the precipitated
10.1 6.97 1.68 123.0 73.6 19.1 0.5
16.1 5.18 3.77 129.0 76.1 24.3 0.6
solids and their corresponding “synthesis” solutions. Based on the
9.8 5.42 2.46 123.0 75.3 20.7 0.7 uncertainties of the stoichiometric coefficients many possible com-
16.2 5.10 4.92 129.0 77.0 20.2 0.8 positions exist which are compatible with the formulae presented in
9.8 2.42 2.67 122.0 74.1 19.4 0.9 Table 1. In order to estimate the range of possible free energies,
16.8 <0.001 4.68 129.0 76.8 20.0 1.0
calculations were performed for the most extreme compositions
 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254 1253

Fig. 8. Raman spectra of hydrotalcite–pyroaurite samples with low mole fraction of Fe (xFe ≤ 0.5; left column). A more detailed description of individual spectra is given in the text.

within the analyzed range. As an example, Gibbs free energies of the There is a fair agreement of ΔG°f of hydrotalcite with the value
solid with xFe = 0.192 (row 4 of Table 1) are shown: from [16]. These authors provided − 1043.1 ± 2.1 kJ/mol for a solid

having the stoichiometry Mg0.74Al0.26 (OH) 2 (CO 3 )0.13 0.39H2 O.
Mg3 Al0:910 Fe0:211 ðCO3 Þ0:553 ðOHÞ8:258 ΔGf ¼ −3802:8kJ=mol ð9Þ

Mg3 Al0:745 Fe0:173 ðCO3 Þ0:518 ðOHÞ7:717 ΔGf ¼ −3577:8kJ=mol ð10Þ

As a result, the average of −3690.3 ± 112.5 kJ/mol was selected for

the solid with xFe = 0.192.
Note that the interlayer water (“nH2O”, corresponding to n times
237.182 kJ/mol) is not included in the calculation of the Gibbs free
energy in order to obtain results that can be compared on a common
“water-free” basis. It is evident that the interlayer water is essential for
stabilizing the structure of LDH, but it is also essential that comparisons
within the solid-solution series are made with an identical amount of
interlayer water (i.e. we have arbitrarily selected n = 0).
The estimated Gibbs free energies were regressed against the mole
fraction of iron. The weighted linear regression produced:
 2
ΔGf ¼ ð−3773:3F51:4Þ þ ð478:8F80:8Þd xFe ½kJ=mol; R ¼ 0:981

(see Fig. 9), from which we obtained the standard Gibbs free energies
Fig. 9. Gibbs free energies of “water-free” solids calculated as a function of the mole
of the end-members: fraction of iron. Filled symbols are calculated including uncertainties of stoichiometric
coefficients (see text above). Open symbols are calculated including only uncertainties
ΔGf ðhydrotalciteÞ ¼ −3773:3F51:4kJ=mol; of concentrations of relevant dissolved components (Mg, Al ± 50%, Fe ± 66%) and using
idealized stoichiometric coefficients Mg3Al1 − xFex(OH)8(CO3)0.5. The diamond symbol
represents a result from Allada et. al [2005], converted to water-free conditions and
ΔGf ðpyroauriteÞ ¼ −3294:5F95:8kJ=mol: normalized to Mg3-stoichiometry. Dashed lines indicate the 95% confidence interval.
1254 K. Rozov et al. / Cement and Concrete Research 40 (2010) 1248–1254

Converting this value to the “water-free” basis and normalizing the There is, however, another point that should be taken into account.
stoichiometry to three Mg (i.e., Mg3Al1.054(OH)8.108(CO3)0.527) gives The present Gibbs free energy analysis started from the assumption
−3853.8 ± 8.4 kJ/mol. that the system can be described using a simple and ideal binary solid-
solution model with the two end-members hydrotalcite and pyroaur-
4. Conclusions and discussion ite. Most likely, this view is too simplistic. From ongoing investigations
on C–S–H systems there are strong indications that enhanced solid-
Hydrotalcite–pyroaurite samples containing different mole frac- solution models (e.g. [18]) may better describe the hydrotalcite–
tions of Fe were successfully synthesized by a co-precipitation method pyroaurite system. Work on this topic is ongoing.
at ambient conditions (23 ± 2 °C, 1 bar) and constant pH ≈ 11.40 ± To conclude, we clearly see the need to perform more precise
0.03. The Mg2+/(Al3++Fe3+) ratios of the precipitated solids were analytical work and to reverse the experiments. Solubility experi-
kept constant at 2.91 ± 0.15; only the Fe3+/Al3+ ratio was varied ments on hydrotalcite–pyroaurite solid solutions including iron-55
between 0 and 1. tracers have therefore been started. The results of combining these
All PXRD patterns indicate the formation of a single phase modified approaches will be the subject of a forthcoming publication.
belonging to the hydrotalcite–pyroaurite solid-solution series [6,7].
A clear linear relationship was observed between the lattice
parameter ao (equal to 3.060 ± 0.001 for hydrotalcite and 3.110 ± Acknowledgements
0.001 Å for pyroaurite) and the mole fraction of iron, in agreement
with Vegard's law, thereby confirming the existence of a solid We wish to thank Dr. U. Eggenberger, M. Eggimann, C. Wanner —
solution. The variation observed for the c0 parameter is quite small for their help with PXRD analyses; M. Fisch, M. Painsi — for their
but higher than the uncertainty, so the trend seems correct. From the support with Raman spectroscopy; Dr. Th. Arlt — support with TGA
available data it is difficult to distinguish variations caused by longer measurements; and for financial support the Nationale Genos-
M–OH distances from those caused by interlayer water content. senschaft für die Lagerung radioaktiver Abfälle (NAGRA).
The thermal behavior is that expected for LDH with 3 main
decomposition steps. These 3 events are delayed when the iron
content is decreased, reflecting some strengthening of binding forces References
at high mole fractions of Al. This is consistent with the fact that Fe–OH [1] F. Cavani, F. Trifiro, A. Vaccari, Hydrotalcite-type anionic clays: preparation,
distances are longer than Al–OH distances. TGA measurements were properties and applications, Catalysis Today 11 (1991) 173–301.
used here successfully for measuring the water- and carbonate [2] A. de Roy, C. Forano, M. EL Malki, J.-P. Besse, Anionic clays: trends in pillaring
chemistry, in: M.L. Occelli, H.E. Robson (Eds.), Synthesis of Microporous Materials,
content of the precipitated solids. For hydrotalcite-like phases this is Expanded Clays and Other Microporous Solids, Van Nostrand Reinhold, New York,
usually difficult to achieve because the three different decomposition 1992, p. 108.
steps are often not well separated. [3] J. Rodriguez-Carvajal, FULLPROF: A Program for Rietveld Refinement and Pattern
Matching Analysis, Abstracts of the Satellite Meeting on Powder Diffraction of the
The low stability of iron-rich samples under laser irradiation XV Congress of the IUCr, Toulouse, France, 1990, p. 127.
precludes a complete characterization of the solid-solution series by [4] Research package for thermodynamic modelling of aquatic (geo)chemical systems
laser-Raman spectroscopy. The typical peaks around 540 cm− 1 result by Gibbs Energy Minimization, http://gems.web.psi.ch.
[5] Wolfgang Hummel, U. Berner, Enzo Curti, F.J. Pearson, Tres Thoenen, Nagra/PSI
from the interaction of H2O with the carbonate in the interlayers. Chemical Thermodynamic Data Base 01/01, Universal Publisher/uOublish.com,
Unfortunately the Raman bands disappear in the whole wavenumber Parkland, Florida, 2002.
region from 250 to 3800 cm− 1 when the mole fraction of iron exceeds [6] H.C.B. Hansen, C.B. Koch, Synthesis and characterization of pyroaurite, Applied
Clay Science 10 (1–2) (August 1995) 5–19 (Synthesis and Application of Anionic
about 0.5. As mentioned above, TGA experiments reveal that iron-rich Clays).
compositions are less stable with respect to temperature. Similarly, de [7] G.J. Ross, H. Kodama, Properties of a synthetic magnesium–aluminum carbonate
Faria et al. [17] showed that the energy deposited by the laser beam hydroxide and its relationship to magnesium–aluminum double hydroxide,
manasseite and hydrotalcite, American Mineralogist 52 (1968) 1036–1047.
leads to disappearance of bands in the magnetite–hematite system.
[8] A.R. Danton, Ashcroft, Vegard's law, Physical Review 43 (6) (1991) 3161–3164.
Since we used beam powers of around 40 mW on the sample as well [9] R.D. Shannon, Revised effective ionic-radii and systematic studies of interatomic
as long acquisition times in our own measurements, we cannot rule distances in halides and chalcogenides, Acta Crystallographica Section A 32 (1976)
out similar effects of thermal decomposition during recording of the 751–767.
[10] M. Bellotto, B. Rebours, O. Clause, J. Lynch, D. Bazin, E. Elkaïm, A reexamination of
spectra. hydrotalcite crystal chemistry, Journal of Physical Chemistry 100 (1996)
As outlined in the introduction, an extended aim of the present 8527–8534.
study included quantifying the Gibbs free energies of members of the [11] S. Miyata, Physicochemical properties of synthetic hydrotalcites in relation to
composition, Clays and Clay Minerals 28 (1980) 50–56.
solid-solution series. This can be achieved by measuring solution- and [12] V. Vagvolgyi, S.J. Palmer, J. Kristof, R.L. Frost, E. Horvath, Mechanism for
solid components under the assumption that thermodynamic hydrotalcite decomposition: a controlled rate thermal analysis study, Journal of
equilibrium between precipitated solid and mother solution was Colloid and Interface Science 318 (2008) 302–308.
[13] V. Rives, Characterisation of layered double hydroxides and their decomposition
reached. All these analyses were performed and the results are products, Materials Chemistry and Physics 75 (2002) 19–25.
illustrated in Fig. 9. It turned out that the precision obtained is not yet [14] R.L. Frost, B.J. Reddy, Thermo-Raman spectroscopic study of the natural layered
satisfactory for applications of the data to problems of waste double hydroxide manasseite, Spectrochimica Acta. Part A: Molecular and
Biomolecular Spectroscopy 65 (2006) 553–559.
management. The uncertainties of the estimated Gibbs free energies, [15] National Water Quality Laboratory, Technical Memorandum 94-08, http://nwql.
for example for the two end-members hydrotalcite and pyroaurite, lie usgs.gov/Public/tech_memos/mwql.94-08.html.
between 50 and 100 kJ/mol. Converted into the usually used log10- [16] R.K. Allada, A. Navrotsky, J. Boerio-Goates, Thermochemistry of hydrotalcite-like
phases in the MgO–Al2O3–CO2–H2O system: a determination of enthalpy, entropy,
units this corresponds to uncertainties of 10 to 20 orders of magnitude
and free energy, American Mineralogist 90 (2005) 329–335.
in solubility. A quick analysis shows that these large uncertainties [17] D.L.A. Faria, S. Venăncio Silvia, M.T. de Oliveira, Raman microspectroscopy of some
arise from uncertainties in the analyses of the solid phases (see iron oxyhydroxides, Journal of Raman spectroscopy 28 (1997) 873–878.
Table 3). Uncertainties in composition of the solutions, although [18] J. Bruno, D. Bosbach, D. Kulik, A. Navrotsky, Chemical thermodynamics of solid
solutions of interest in radioactive waste management: a state-of-the art report,
substantial from the point of view of absolute values, are much less in: F.J. Mompean, M. Illemassene, J. Perrone (Eds.), Chemical Thermodynamics
significant with respect to the total uncertainties. Series, vol. 10, OECD, Paris, 2007, pp. 6–100.