You are on page 1of 19

Chemical Geology 167 2000.

141159
www.elsevier.comrlocaterchemgeo

Dehydrationrrecrystallization mechanisms, energetics, and


kinetics of hydrated calcium silicate minerals: an in situ
TGArDSC and synchrotron radiation SAXSrWAXS study
S. Shaw a,b,) , C.M.B. Henderson a,b, B.U. Komanschek b
a

Department of Earth Sciences, Uniersity of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, UK
b
Daresbury Laboratory, CCLRC, Warrington WA4 4AD, UK
Received 11 January 1999

Abstract

The dehydration and recrystallization of the natural hydrated calcium silicates xonotlite wCa 6 Si 6 O17 OH. 2 x, 11A
anomalous tobermorite wCa 5 Si 6 O 16OH. 2 P 4H 2 Ox, and hillebrandite wCa 2 SiO 3OH. 2 x were studied in situ by dynamic
heating 58C miny1 . differential scanning calorimetryrthermogravimetric analysis TGArDSC. and synchrotron small- and
wide-angle X-ray scattering SAXSrWAXS.. All have structures based on silicate chains with a repeat unit of three
tetrahedra dreierketten.. Room-T infrared data for the tobermorite sample suggest that it is essentially OH.-free and that it
is likely to contain double chains in addition to the normal single chain units. The in situ techniques allow the mechanisms
and kinetics of dehydration and recrystallization processes to be studied directly or deduced. Xonotlite and tobermorite
recrystallise to low-temperature T.-wollastonite CaSiO 3 . with two-phase regions existing over the temperature ranges
7808358C and 8358558C, respectively. Hillebrandite breaks down via a two-phase region between 5208C and 5908C to
form larnite b-Ca 2 SiO4 . which appears to retain some OH. in its structure. This hydroxylated larnite in turn recrystallises
to aXL-Ca 2 SiO4 between 7758C and 9008C. The presence of OH. in the hydroxylated larnite appears to extend its stability
limit to higher Ts than those shown for anhydrous larnite. In each case, the widths of the two-phase regions are dependent on
the extents of structural changes involved. DSC data for xonotlite, hillebrandite and hydroxylated larnite give a mean energy
for dehydroxylationrrecrystallization of 4000 " 1200 Jrg OH as H 2 O. lost. Radius of gyration values for hillebrandite
this is believed to represent the lengthscale of domains between zones
provide estimates of particle sizes up to about 640 A;
of defects formed by displacements of structural units along the dreierketten chains. q 2000 Elsevier Science B.V. All rights
reserved.
Keywords: Xonotlite; Tobermorite; Hillebrandite; CSH; SAXS; Reaction mechanisms

1. Introduction
)
Corresponding author. Department of Earth Sciences, University of Manchester, Williamson Building, Oxford Road, Manchester M13 9PL, UK. Tel.: q44-958-917783.
E-mail addresses: shaw@fs1.ge.man.ac.uk S. Shaw .,
chenderson@fs1.ge.man.ac.uk C.M.B. Henderson..

Naturally occurring calcium silicate hydrates


CSH minerals. include over 30 chemically and
structurally distinct phases; some contain molecular
water, some hydroxyl, and others contain both hy-

0009-2541r00r$ - see front matter q 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 9 - 2 5 4 1 9 9 . 0 0 2 0 6 - 5

142

S. Shaw et al.r Chemical Geology 167 (2000) 141159

drous species. Despite the wide range of minerals


belonging to the CSH-group, they are of limited
occurrence. The least hydrated forms e.g., xonotlite,
Ca 6 Si 6 O 17 OH. 2 . are generally found in metamorphosed carbonate rocks, often associated with anhydrous calcium silicates such as wollastonite CaSiO 3 .
and larnite Ca 2 SiO4 . in the highest grade rocks
Tilley, 1951.. By contrast, the more hydrated forms
e.g., tobermorite, Ca 5 Si 6 O 16 OH. 2 P 4H 2 O., are
formed in relatively low temperature - 2008C., hyper-alkaline, hydrothermal environments, for example, where hydrous fluids react with basic igneous
rocks e.g., Okayama, Japan; Henmi and Kusachi,
1992.. Anhydrous CS-minerals are also known to
occur in calcined cements, slags and bricks, while
the more hydrated phases and their gel precursors
form in Portland cements Taylor, 1990., in the
hyper-alkaline environments surrounding cementitious nuclear and toxic waste sites Atkinson et al.,
1995., and in fly-ash waste deposits following reaction with ground water Johnson and Schweizer,
1998.. Thus, as well as the stabilities and reactionr
dehydration mechanisms of CSH-minerals being of
intrinsic mineralogical interest, an increased understanding of the processes that form these cement
minerals is particularly important for determining
the chemical evolution of the cementrock interface
of toxic waste repositories.
There have been many studies of the structural
relations of natural CSH minerals, many of which
have been stimulated by an assumption that such
minerals represent reasonable analogues for the
structures of poorly ordered, CSH-gels in Portland
cement. Natural samples usually occur as fibrous or
small platey crystals and single crystal X-ray and
electron diffraction patterns show effects of stacking
disorder e.g., streaking, defects: McConnell, 1954;
Hamid, 1981.. In addition, there have been many ex
situ studies of phase stabilities in which samples are
characterised at room temperature and pressure after
quenching from the hydrothermal synthesis or heattreatment conditions e.g., Mitsuda and Taylor, 1975;
El-Hemaly et al., 1977; Jauberthie et al., 1996..
Despite this earlier work, much remains to be established regarding the structures, reaction mechanisms,
thermochemical properties, reaction kinetics and
phase transitions of these structurally complex minerals and we have attempted to provide a new ap-

proach by carrying out in situ, elevated temperature,


dynamic studies at the Daresbury synchrotron.
An in situ experimental study of the stabilities and
reaction rates for the hydrothermal formation of tobermorite and xonotlite from synthetic, amorphous
starting materials, is the subject of a companion
paper in this volume Shaw et al., 1999.. In this
paper, we describe an in situ dynamic study of the
dehydration, structural collapse and recrystallization
of natural samples of tobermorite, xonotlite, and
hillebrandite Ca 2 SiO 3 OH. 2 . over the temperature
range from room temperature up to about 10008C,
with the ultimate formation of high-temperature T.,
anhydrous CS phases. Experiments were carried out
in the laboratory using combined differential scanning calorimetryrthermogravimetric analysis DSCr
TGA., and at the Daresbury synchrotron using combined small-angle X-ray scatteringrwide-angle X-ray
scattering SAXSrWAXS. Bras et al., 1993..
The WAXS technique provides information on the
internal atomic structure, while SAXS monitors the
external properties of particles within heterogeneous

samples on length scales from about 1600 to 40 A,


providing information on the densities, sizes and
shapes of particles suspended in a matrix of contrasting electron density. The ability to obtain continuous SAXSrWAXS records of the phases present,
as a function of time at controlled heating rates, with
a time resolution of about 1 min, allows the mechanisms and kinetics for continuous and stepwise reactions to be assessed.
Although we have employed the SAXSrWAXS
technique in this study to characterise the breakdown
of hydrated minerals, the technique is equally applicable to the study of hydrothermal systems; such
work would require a different environmental heating cell.

2. Structures, stabilities and phase transitions of


CSH- and CS-minerals

pseudo-orthorhombic
The basic structure of 11-A
tobermorite Hamid, 1981. and clinotobermorite
Hoffman and Armbruster, 1997. consists of infinite
hydroxylated silicate chains with a three tetrahedra,
dreierketten repeat wSi 3 O 9 x, similar to that of wollastonite; the three tetrahedra repeat can be consid-

S. Shaw et al.r Chemical Geology 167 (2000) 141159

ered as consisting of a pair of tetrahedra Si 2 O 7 .


pointing in the same direction, linked by a bridging
tetrahedron pointing in the opposite direction. Note
that the hydroxyls occupy the apex of a bridging
tetrahedron in the dreierketten chains. The silicate
chains are aligned parallel to the b axis and are
linked by Ca`O,OH. chains to form sandwich
layers in the ab plane. These layers are stacked
along c with additional Ca and H 2 O molecules in
the interlayers wsee Fig. 2a in the companion paper in
this volume Shaw et al., 1999.x. During heating up
to 3008C, the basal spacing of normal tobermorite
resulting from water
decreases from 11.3 to 9.3 A

loss from the inter-layers to form riversideite 9.3-A


tobermorite. Mitsuda and Taylor, 1978.. wNote that
a variety of tobermorite plombierite. has an even
higher content of interlayer water 8H 2 O. and a
basal spacingx. Another variant of 11.3-A to14-A
bermorite, the anomalous form, loses interlayer
water under the same conditions as normal tobermorite but shows no contraction of the basal spacing
Mitsuda and Taylor, 1978. up to 7008008C. Mitsuda and Taylor 1978. suggested that the absence of
a collapse in the basal spacing during dehydration
could be due to some cross linking of single chains
via the bridging tetrahedra, rather than the presence
of discrete double dreierketten chains as proposed by
Wieker 1968.. Hoffmann and Armbruster 1997.
suggested that clinotobermorite could contain double
chain units, the end member for which would have
the stoichiometry Ca 5 Si 6 O 17 5H 2 O with no OH possible where all the bridging oxygens are shared. Note
that Wieker et al. 1982. and Komarneni et al.
tobermorites using
1985. studied synthetic 11-A
29
Si MAS NMR and reported the presence of peaks
for both Q2 and Q3 silicons, with peak intensities
suggesting double-chain structures, so confirming
such structural complexities.
Hillebrandite also contains silicate dreierketten
chains that occupy channels in a linked Ca`O,OH.
three-dimensional polyhedral sheets; as in tobermorite these sheets are stacked along c Dai and
Post, 1995.. Two-thirds of the hydroxyls are bonded
to Ca atoms in the Ca`O sheet and one-third are part
of the dreierketten chain, adjacent to vacant Si sites.
Note that the fundamental differences between the
Si`O chainrCa`O polyhedral sheet structures in
hillebrandite and tobermorite are related to the SirCa

143

ratios hillebrandite 1:2; tobermorite 1:0.83.. Xonotlite is structurally similar to tobermorite except that
double dreierketten chains build up sheets in the ab
plane Mamedov and Belov, 1955. which, in turn,
link sheets of Ca`O polyhedra parallel to 001. see
Fig. 2b; Shaw et al., 1999.. X-ray structural studies
place two hydroxyls in the Ca`O sheets Kudoh and
Takeuchi, 1979. although H-NMR studies suggest
that about 18% of the OH. might be associated with
the silicate chains Noma et al., 1998..
Anhydrous calcium silicate minerals include lowT-wollastonite also known as a- or para-wollastonite. and high-T-wollastonite also known as b-,
pseudo-, and cyclo-wollastonite. CaSiO 3 ; CS., larnite b-Ca 2 SiO4 ; C 2 S. and bredigite a-C 2 S.. wNote
that, for brevity, the cement chemist notation is used
for describing the stoichiometry of phases with CaO
denoted as C, SiO 2 as S, and H 2 O as Hx. Synthetic
polymorphs of C 2 S also include a low-T form g and
high-T forms aX , aXL and aXH . Low-T-wollastonite
consists of infinite dreierketten chains parallel to b
which are linked by Ca cations in octahedral coordination Mamedov and Belov, 1956. while high-Twollastonite pseudo-hexagonal. contains rings of
three tetrahedra alternating with layers of Ca`O 8
polyhedra Yamanaka and Mori, 1981.. Low-T-wollastonite transforms to high-T-wollastonite at about
11258C by a reconstructive phase transition. g-C 2 S
is based on an olivine structure with isolated Si`O4
tetrahedra linked by Ca cations in octahedral coordination. The a-, aX- and b-forms of C 2 S also consist
of isolated Si`O4 tetrahedra linked by Ca atoms and
are topologically similar to each other but are structurally distinct from g-C 2 S. g-C 2 S is the stable
low-T form and the pure end-member C 2 S a-, aXand b-polymorphs only occur at high temperature
although the presence of impurities allows their
metastable persistence at room temperature.
Because of their technological importance, many
studies have been carried out on the structures and
phase transitions of the C 2 S system e.g., Barnes et
al., 1980; Kim et al., 1992; Mumme et al., 1995,
1996; Remy et al., 1997.. The following transformations have been identified at successively higher
temperatures: g-C 2 S aXL at ; 7808608C; irreversible, first-order reconstructive. aXH at ;
11608C; reversible, displaceablerferroelastic. a
at ; 14258C, ferroelastic, first-order semi-recon-

144

S. Shaw et al.r Chemical Geology 167 (2000) 141159

structive, Eysel and Hahn, 1970.. Also, b-C 2 S transforms to aXL at ; 7008C via a first-order ferroelastic
phase transition, and on cooling b-C 2 S inverts reconstructively and irreversibly to the g-polymorph at
- 5008C by a first order phase transition.
Dent and Taylor 1956. found that xonotlite dehydrates to low-T-wollastonite at about 7758008C,
while Meyer and Jaunarajs 1961. report a temperature of about 8008C for the loss of hydroxyls and the
recrystallization to low-T-wollastonite. Meyer and
Jaunarajs 1961. also report that tobermorite shows a
partial dehydration at about 5008C and recrystallises
to low-T-wollastonite at about 8108C. Gard and Taylor 1957. studied the dehydration of tobermorite by
heating at constant temperature for different times
and then used X-ray techniques to characterise the
sample at room temperature. The main water loss
occurred up to 2003008C followed by smaller losses
with the sample becoming fully dehydrated at about
8908C. This sample was still tobermorite after 24 h
at 7508C but had crystallised to low-T-wollastonite
after 96 h at 9008C. Mitsuda and Taylor 1978. show
data for the dehydration in five tobermorite samples;
all lost about 80% of their H 2 O component by
2503008C, with the remaining lost, during dehydroxylation, up to 7308C and 8308C depending on
the sample. Mitsuda and Taylor 1978. also report
that DTA exotherms at 8008C to 8508C, which vary
in intensity and sharpness, are associated with breakdown to low-T-wollastonite.
Ishida et al. 1992. reported that synthetic hillebrandite began to dehydrate at about 4908C with
90% of the initial water component lost by 5788C;
dehydration was complete by 6658C. Their data show
that decomposition to larnite occurred between 5308C
and 6108C, and inferred that further weight loss up to
6658C was from a protonated larnite cf. Okada et
al., 1998.. Ishida et al. 1992. also showed that
larnite began to break down to aXL-C 2 S at about
6708C, going to completion at about 8408C. By
contrast, Kim and Kriven 1995. found that, in ex
situ experiments, hillebrandite decomposition to larnite was not observed for samples heated at 5008C,
was observed in some experiments at 6008C, and
was substantially complete at 7008C. Kim and Kriven
1995. also report TEM in situ heating experiments
in which a phase forming over 6008008C was
tentatively identified as low-T-wollastonite; this sur-

prising result was interpreted as reflecting the breakdown of the Ca`O,OH. polyhedral network to CaO
with only a proportion of this CaO reacting with the
silicate chains to form the CS phase, rather than
larnite C 2 S..
The dynamic nature of our in situ synchrotron
experiments with the ability to obtain full diffraction
patterns in short times, probing different lengthscales, provides much new information on these
systems, in particular allowing knowledge of the
reaction mechanisms to be refined and providing
new data on reaction kinetics.

3. SAXS
Small-angle scattering of X-rays is observed when
electron density inhomogeneities on a colloidal scale
exist in the sample. Samples can be studied as
amorphous and crystalline solids, solutions and slurries. An analogy with SAXS is the formation of a
lunar corona by scattering of moonlight from water
droplets in the atmosphere on misty, cloudless nights
with the formation of a halo with intensity decreasing away from the light source. A synchrotron X-ray
source is ideal for small-angle scattering studies
because of its very low divergence allowing measurements to be made at very low 2 u angles at least
down to 0.058.. A brief summary of SAXS is given
below but further information can be obtained from
books and publications e.g., Glatter, 1992; Colyer et
al., 1996; de Moor et al., 1997..
The scattering intensity as a function of the scattering angle u , described in terms of the scattering
vector q 4prl s 2prd ., can be written as:
I q. s f P P q. P S q.

1.

where f is the number density of the particles or


individual scatterers in the sample. The form factor
P q . describes the relation between the geometry of
the individual particles and their scattering. The
structure factor S q . contains information on the
correlation of these particles. Form factors can be
defined mathematically for particles of uniform shape
and size allowing the SAXS pattern to be calculated
and analysed. Thus, in monodisperse systems, where
all particles in the scattering volume are assumed to

S. Shaw et al.r Chemical Geology 167 (2000) 141159

have the same size, shape, and internal structure,


analysis of the scattering pattern provides information on these properties. For polydisperse systems,
scattering information is highly convoluted, but data
can be obtained on size distribution if a certain shape
for the particles is assumed. In samples where the
particles might vary in size and shape, only idealized, averaged size information can be obtained for a
hypothetical particle shape i.e., the radius of gyration, R G ..
Although the scattering function I q . varies with
particle shape its integral can be assumed to remain
invariant allowing definition of the SAXS invariant
Q as follows:
Qs I q. Pq2 Pd q

2.

The dependence of the invariant intensity Q on


the electron density contrast D r . between particles
and matrix, and the proportions j . of these in a
binary mixture is defined as:
Q s j 1 y j . Dr .

3.

which has a maximum for a 50:50 mixture Colyer et


al., 1996.. Thus, variation of Q with reaction stage
provides some information on the proportions of the
phases of interest.
The intensity scattered as a function of scattering
vector q, for small scattering angles, is given by the
approximate formula:
ln I q . s ln I 0 . y 4p 2rl2 P R G2 2 u .

145

electron microprobe fitted with a Link Systems ED


detector. A 15-keV electron beam was defocused to
give a spot of about 20-mm diameter; in each case
the average of eight analyses is reported. Unit cell
parameters were determined using a Philips vertical
goniometer and a computer controlled data acquisition and analysis system; either silicon or quartz
were used as internal standards. Cell parameters
were calculated using the non-linear, least-squares
program Unit cell Holland and Redfern, 1997..
Infrared spectra were obtained on a Bio Rad
FTS-60A Fourier Transform spectrometer run in
transmission mode. Samples were finely ground and
mixed with spectroscopy-grade KBr in the proportion 1:100 and pressed into 13-mm discs; these were
dried at 1108C before running.
4.2. TGA r DSC studies
Combined TGArDSC data were obtained at the
University of Manchester using a Rheometric Scientific STA 1500H TGArDSC. The temperature and
heat energy scales were regularly calibrated with
silver nitrate melting 2098C; 67.8 mJrmg., ab
quartz phase transition 5738C; 10.5 mJrmg., and
NaCl melting 8018C; 481.8 mJrmg.. Samples about
1520-mg aliquots. were run in Pt crucibles and
heated at 58Crmin from 25 to 11008C with the
weight and heat flow monitored every 2 s. The
sample atmosphere was Ar gas flowing at 30 mlrmin.

4.

where I 0. is the intensity at zero scattering angle. A


Guinier plot of ln I q . vs. 2 u . 2 for small values of
q is linear with slope 4p 2rl2 P R G2 . from which the
idealized radius of gyration can be estimated.

4. Methods and materials


4.1. Samples
The samples studied were natural specimens of
tobermorite from Portree, Isle of Skye, UK BM
1959,552., xonotlite from Hindubagh, W. Pakistan
M.2776, Harwood Collection, University of Manchester. and hillebrandite from Terara, Velardera,
Durango, Mexico BM 1933,458.. Minerals were
analysed using a Cambridge Instruments Geoscan

4.3. SAXS r WAXS experiments


4.3.1. Station 8.2
The optical system of Station 8.2 at the Daresbury
Synchrotron Radiation Source consists of an asymmetrically cut a s 10.58. monochromator as the
first, and a mirror as the second optical element. The
triangular bent Ge 111. monochromator crystal is
used to focus the X-rays in the horizontal plane and
monochromate the white X-ray beam to a wave The subsequent vertical focussing
length of 1.54 A.
is obtained by using an uncoated fused quartz mirror
under grazing incidence. The total flux at the sample
is 5 = 10 10 photonsrs in a spot size of 4 = 0.4 mm
H = V . with the SRS operating at 200 mA and 2
GeV. The horizontal and vertical divergence of the
beamline are 5 and 0.25 mrad, respectively.

146

S. Shaw et al.r Chemical Geology 167 (2000) 141159

The schematic set-up of the SAXSrWAXS camera is shown in Fig. 1a. The length of the vacuum
beampipe to the SAXS detector is usually 3.5 m
., but a
providing data up to about 28 2 u d s 44.1 A
shorter beampipe of 1m extends the range to about
.. Samples were prepared by pressing
108 d s 8.8 A
about 30-mg aliquots into 13-mm diameter discs
using an IR sample preparation die. These samples
are mounted in a high-T furnace, built in cooperation
with the Royal Institution, London, which reaches
temperatures up to 12008C Dent et al., 1995.. Platinumr13% rhodium wire is used for heating elements and the aluminium body of the furnace is
water-cooled. Muscovite mica was used as a window
and the temperature is controlled by a separate combined power supplyrcontroller unit. The samples
were usually heated at 58Crmin in an atmosphere of
flowing nitrogen with the SAXSrWAXS data monitored every 1 min.
Both the SAXS and WAXS data are collected in
transmission geometry. The position-sensitive, quadrant detector used for the SAXS data collection is a
multi-wire, linear detector with an opening angle of
708 and an active length of 20 cm. This detector
measures the X-ray intensity in a radial direction and
is only suitable for isotropic scatterers. It has an
advantage over the usual linear detectors in that the
active area increases radially, therefore compensating

for the weaker scattering towards higher angles. The


quadrant detector has a spatial resolution of 250 mm,
a dynamic range of 1:10 6 and an overall count rate
limit of 300,000 countsrs at 1 MHz. A beam-stop
prevents the direct beam from striking the quadrant
detector and cuts out the central part of the SAXS
peak. Fig. 1b shows a SAXS spectrum for tobermorite obtained using a short 1 m. camera length.
The INEL position sensitive detector used for the
WAXS data collection is a curved, pressurised gas
detector with a working pressure of 5.2 bar. The
global count rate limit over the complete 1208 opening angle is 40,000 countsrs at 20 kHz, which is
reduced to 20,000 countsrs because only a 608
opening angle is physically used with this experimental set-up. The dynamic range is 1:10 4 and the
angular resolution is about 0.18. In order to obtain
the best compromise with respect to the spatial resolution of both SAXS and WAXS data sets, the X-ray
beam was focussed at the halfway point between the
two detectors. Standard Daresbury VME data acquisition systems were used for the data acquisition and
time framing of the two data sets in order to ensure
exact time synchronisation.
WAXS and SAXS 2 u values were calibrated
using a quartzSi mixture and collagen from a rats
tail or silver behanate, respectively. Backgrounds
were determined for both detectors without a sample

Fig. 1. a. Schematic diagram of the SAXSrWAXS set up on Station 8.2 of the SRS, Daresbury. b. SAXS diffraction pattern for
tobermorite obtained with a 1m camera length. The sharp decrease at the low 2 u end marks the position of the beam stop. The logarithmic
decrease in intensity with increasing 2 u defines the SAXS scattering invariant. Note that the 002. Bragg peak is superimposed on the
SAXS invariant.

S. Shaw et al.r Chemical Geology 167 (2000) 141159

147

in place. The response of the INEL detector was


calibrated by scanning an 55 Fe source through its
effective angular range, with the Fe-isotope source
stationary for calibrating the area detector.
In the furnace cage, the control thermocouple was
buried in the pyrophyllite ceramic so its temperature
lagged behind that of the sample in the dynamic
heating experiments; sample temperatures were calibrated by replacing a sample disc with an extra
thermocouple giving a temperature correction which
increased up to q508 at 5008C, decreasing to q208
at 9008C.

based program XFIT Cheary and Coelho, 1992..


Cell parameters at 258C or 508C intervals were calculated using the program Unit Cell in order to
observe the changes in the unit cell dimensions of
the particular mineral phase with increasing temperature. Note that the cell parameter errors involved are
moderately high about 1 part in 1000. because of
the short data collection times and non-optimum
focussing geometry.

4.4. SAXS r WAXS data analysis

5.1. Room temperature characterisation of starting


samples

Raw SAXS data were background subtracted and


corrected for detector alinearities, using the one-dimensional data manipulation package XOTOKO see
www.dl.ac.ukrSRSrNCDrmanual.otoko.html. .
Guinier plot analysis for data sets was also carried
out using XOTOKO. The WAXS data were correct
for background scattering and detector alinearities
using the XOTOKO package in the same way as the
SAXS data. Selected spectra from these corrected
data were indexed and then peak-fitted using the PC

5. Results and interpretation

Chemical analyses and unit cell parameters of


samples used for this study are reported in Table 1.
Both xonotlite and hillebrandite have stoichiometric
end-member compositions within one sigma error.
Tobermorite departs from the stoichiometric end
member formula reported in the introduction by containing Al substituted for Si in the silicate chains,
however, the tetrahedral cations total 6.027, within
error of the stoichiometric six atoms per formula

Table 1
Analyses and cell parameters for starting materials

Analysis wt.%.
SiO 2
Al 2 O 3
CaO
H 2 Ob
Total
Cell formula
Si
Al
Ca
H

Xonotlite

Tobermorite

Hillebrandite

Av. eight spots 1 s .


50.3 0.25.
n.d.a
46.9 0.16.
2.5
99.7
19 O.
6.002

Av. eight spots 1 s .


47.0 0.6.
3.71 0.14.
35.2 0.3.
12.2
98.11
21 O.
5.514
0.513
4.424
9.558

Av. eight spots 1 s .


31.6 0.1.
n.d.
58.7 0.2.
9.5
99.8
4 O.
1.002

11.1833.
7.3753.

16.3185.
3.6432.

22.6887.

11.8394.

1871.07.

703.84.

6.001
1.992

1.993
2.010

Cell parameters

aA

bA

cA
b8
3
VA
a
b

17.0408.
7.3616.
7.0023.
90.334.
878.28.

n.d.s not detected Ti, Fe, Mn, Mg, Na, and K were also not detected..
Total H as H 2 O wt.%. determined by TGA.

148

S. Shaw et al.r Chemical Geology 167 (2000) 141159

unit. The Al content wAlrAl q Si. s 8.5%x for our


tobermorite is at the upper end of the compositions
reported by Mitsuda and Taylor 1978. and is about
half of the maximum degree of substitution wAlrAl
q Si. s 1520%x reported by Tsuji et al. 1991..
Both Ca and total water are about 10% lower than
the stoichiometric values; the Ca deficiency at least
is real and presumably reflects variable Ca occupancies in alternating Ca`O polyhedral sheets Faucon
et al., 1998.. Perhaps coincidentally, the analysed Ca
content of our sample is the same as that determined
by Hamid 1981. from X-ray site-occupancies. Cell
parameters for xonotlite and hillebrandite show close
agreement with JCPDS entries 29-0379 and 42-0538,
respectively. Cell parameters for our tobermorite have
lower a and higher c values than those for JCPDS
entry 45-148 which are likely to be accounted for by
different degrees of hydration.
Xonotlite shows a single, sharp peak in the IR
spectrum at about 3610 cmy1 which can be assigned
to structural OH., and a further sharp peak at 1200
cmy1 which is due to the stretching vibration of the
bridging Si`O`Si bonds in the silicate double chains
Henning, 1974.. Tobermorite IR spectra show a
broad band from 3700 to 3100 cmy1 due to molecular water and a further water peak at about 1620
cmy1 ; no sharp peaks near 3620 cmy1 are observed
on top of the water band. Interestingly, a clear
shoulder is seen at 1200 cmy1 , which we correlate
with the presence of double chains. Hillebrandite has
four sharp peaks over the range 3500 to 3600 cmy1
indicating the presence of OH in distinct structural
sites; a peak at 1600 cmy1 could be due to the
presence of some molecular water in hillebrandite or
to the presence of a hydrated impurity.
5.2. Dynamic high-T experiments
5.2.1. Xonotlite
Combined TGArDSC heating experiments on
xonotlite Fig. 2. show that between about 7708C
and 8808C a weight loss of 2.5% occurs 99% of the
stoichiometric H 2 O content.; this weight represents
the loss of structural hydroxyls as water. The bulk of
this loss occurs between about 7808C and 8208C.
The accompanying endothermic peak is well defined
over about 308C peak centred at 8158C. and gives a
D H of about 88 Jrg 63 kJrmol..

Fig. 2. DSCrTGA scan for xonotlite obtained at a heating rate of


58C miny1 . The main weight loss and endotherm near 8158C are
related to loss of OH. from xonotlite and simultaneous recrystallization to the low-T-wollastonite polymorph.

The WAXS data Fig. 3a. show little change in


peak intensities up to 7608C but these decrease over
the next 158C. At about 7808C, the peaks for low-Twollastonite appear. As temperature increases, the
peak intensities for wollastonite increase progressively and those for xonotlite decrease until xonotlite
peaks are no longer detectable at about 8358C, i.e.,
the reaction goes to completion in about 11 min. The
wollastonite peak intensities continue to increase after xonotlite peaks can no longer be detected, reaching maximum intensities at about 8558C. Note that
the CarSi ratio in xonotlite and wollastonite is the
same 1:1. so a single phase anhydrous product is
expected. The cell parameters for xonotlite increase
linearly with temperature up to 7708C see Table 2..
The increase in the a, b and c parameters over the
observed temperature is 0.65%, 0.24% and 1.15%,
respectively. This anisotropic thermal expansion reflects the nature of the xonotlite structure. The relatively small increases in the a and b directions
reflect the fact that the ab plane consists of a rigid
silicate sheet so that thermal expansion is relatively
small especially parallel to b. The large increase in c
over the observed temperature range is due to the
relatively weak bonding between the sheets formed
by the adjacent calcium polyhedra and silicate chains.
Although the starting sample is a single phase,
well-crystalline, compressed powder disc of
xonotlite, the initial intensity of the SAXS invariant
Fig. 3b. is quite high reflecting a significant electron density contrast in a relatively porous hydroxylated sample i.e., the sample is heterogeneous showing significant electron density contrast.. The SAXS

S. Shaw et al.r Chemical Geology 167 (2000) 141159

149

Fig. 3. Stacked WAXS a.rSAXS b. plots vs. temperature for xonotlite; the vertical scales are intensities in arbitrary units. SAXS data
were collected with a 3.5-m camera length. Every third spectrum is giving temperature steps of 158C. The recrystallization of xonotlite to
low-T-wollastonite is seen at about 8008C in the WAXS plot.

S. Shaw et al.r Chemical Geology 167 (2000) 141159

150

Table 2
Linear regression of temperature dependence of cell parameters
y s mx q c .
c

R2

Xonotlite (20 7708C)


A
1.434=10y4
B
2.680=10y5
C
1.130=10y4
b
2.322=10y4
V
2.489=10y2

17.0
7.364
7.004
90.19
876.9

0.904
0.912
0.994
0.747
0.992

Hillebrandite (20 5308C)


A
2.516=10y4
B
3.516=10y5
C
1.724=10y4
V
2.809=10y2

16.333
3.644
11.814
703.1

0.883
0.835
0.953
0.987

invariant intensity is fairly constant up to about


7958C and then increases rapidly reaching a maximum value at 8308C, as the electron density contrast
increases in an increasingly heterogeneous product.
The SAXS intensity remains the same up to 8908C
where the beam was lost. Guinier plots not figured.
define R G values showing little variation around 250
"0.2 A ..
A
The xonotlite to low-T-wollastonite phase transition at about 8108208 involves shearing the double
dreierketten chains in xonotlite along the b axis to
form the single chains of wollastonite, together with
associated tetrahedral tilting and Ca atomic displacements cf. Taylor 1990... The reconstructive nature
of this transition is reflected in the presence of a

two-phase field in our experiment. Despite these


significant structural changes, the transition takes
place rapidly and is complete in about 11 min. The
fact that only a single heat effect can be identified
suggests that dehydroxylation and recrystallization
occur simultaneously.
Although the sample consists of single phase
low-T-wollastonite above about 8358C, the SAXS
invariant does not pass through a maximum as expected when the sample consists of equal proportions
of xonotlite and wollastonite at about 8158C. It
seems that the wollastonite product has not densified
so that the SAXS intensity reflects scattering from a
low-density sample disc, with the electron density
contrast being between crystallites and pore space;
no significant change in particle size was detected.
5.2.2. Tobermorite
TGArDSC experiments for tobermorite Fig. 4.
show that the main weight loss 9 wt.%; 73% of the
stoichiometric amount. occurs between 508C and
2508C with a further 3.2% weight loss up to 7008C
27% of the stoichiometric amount.. The weight
decrease and broad endothermic hump between 508C
and 2508C presumably result from the loss of loosely
bound molecular waters. Between 2508C and about
7508C, any remaining molecular water is lost; dehydroxylation presumably would occur at higher temperatures still. In the DSC spectrum, a very broad
endothermic hump occurs between about 508C and
2508C, followed by a small endothermic peak 45

Fig. 4. DSCrTGA scan for tobermorite obtained at a heating rate of 58C miny1 . The main weight loss below 3008C and associated
endotherm are due to loss of molecular water. The exotherm near 8608C is due to recrystallization to low-T-wollastonite.

S. Shaw et al.r Chemical Geology 167 (2000) 141159

151

Fig. 5. Stacked WAXS a.rSAXS b. plots vs. temperature for tobermorite; the vertical scales are intensities in arbitrary units. SAXS data
were collected with a 1-m camera length. Every third spectrum is plotted giving temperature steps of 158C. The recrystallization of
tobermorite to low-T-wollastonite is seen at about 8508C in the WAXS plot. The sharp drop in the SAXS invariant intensity at 8108C marks
the final collapse of the tobermorite structure. See text for further discussion.

152

S. Shaw et al.r Chemical Geology 167 (2000) 141159

Jrg; 33 kJrmol. at 8268C which occurs just before


an exothermic peak 189 Jrg; 138 kJrmol. at 8578C.
The WAXS peak heights increase and then decrease over the temperature range from 2508C to
4508C; these changes are presumably related to the
combined effects of sample densification and water
loss. Little further change in peak heights occur up to
about 8058C Fig. 5a.. Above this temperature, the
tobermorite peaks start to decrease in intensity again.
New peaks characteristic of low-T-wollastonite are
detectable at about 8408C and at 8558C only wollastonite peaks can be detected. Thus, a very narrow
- 208C. field of coexisting tobermorite and wollastonite occurs with broadened diffraction peaks
indicating that both phases are disordered as the
structural transformation proceeds. These dynamic
heating experiments show that the transformation is
complete within - 4 min and can be placed at
845 " 58C. The wollastonite peaks increase in intensity, becoming sharper, and reach a maximum intensity at about 8608C. Curiously, up to about 9208C, a
very small peak remains in the position of the tobermorite basal 002. reflection, suggesting that a small
residuum of the tobermorite layer structure persists
metastably for about 30 min, well into the wollastonite stability field. Fig. 6a,b,c,d shows the changes
in the tobermorite cell parameters with temperature.
There is a sharp decrease in a up to 2008C, which is
related to the loss of molecular water from the
structure. Above 2008C, this decrease slows down
and a eventually levels off at approximately 6008C.
The temperature evolution of b and c are very
similar in that they remain almost constant up to
3708C, then decrease before levelling off at approximately 6008C The cell volume shows a steady linear
decrease up to 6008C after which it remains constant.
These cell parameter changes can mainly be attributed to the lost of water from the tobermorite
structure. The reason for the main phase of shrinkage
varying between the different cell parameters is
probably due to different structural water species
being lost at different temperatures. The lack of a
major reduction in the c parameter during dehydration on heating to 3008C shows that our sample is an
tobermorite.
anomalous 11-A
The SAXS invariant shows little change in intensity up to about 8008C Fig. 5b. before starting to
decrease drastically at 8108C reaching a minimum at

Fig. 6. Tobermorite cell parameters determined from the WAXS


data as a function of temperature. Error bars are 1 s ; curves are
fitted using a third order polynomial. The trends are mainly
controlled by loss of molecular water with the main effect being
shown for the a parameter. Note that c shows relatively little
change up to 3008C indicating that the sample has the properties
typical of anomalous tobermorite.

8608C. Radius of gyration values are fairly constant

at a rather small value of 85 A.


The endothermic hump in the DSC trace and the
accompanying weight loss for tobermorite up to
2508C clearly suggest the loss of four loosely held
molecular waters per formula unit. The remaining
water loss is completed well before the endothermic
peak at 8268C. This endotherm is not accompanied
by any detectable weight loss suggesting that this
endothermic heat effect is not due to dehydroxylation and is most likely due to breakdown of anhy-

S. Shaw et al.r Chemical Geology 167 (2000) 141159

drous tobermorite to an intermediate amorphous


phase. The main exothermic peak at 8578C 189
Jrg. presumably represents the recrystallization of
the sample to low-T-wollastonite after complete dehydration. Tobermorite seems to transform essentially continuously into low-T-wollastonite, presumably reflecting the fact that the silicate chains in both
samples have the same general topologies, although
the possible presence of double chains in tobermorite
see below. requires a more complex structural
change for these units at least. Nucleation of the
high-T phase requires only small changes in the
tilting of tetrahedra and limited diffusion of Ca
cations and this accounts for the very rapid transformation - 4 min.. The fact that the CarSi ratio in
tobermorite 5:6. is slightly smaller than that in
xonotlite 1:1. indicates that a silica phase should be
present in the products; none was identified suggesting that it was either amorphous or present in too
small an amount to detect -; 3%.. The collapse of
the SAXS invariant intensity to about 1r3 of the
maximum value Fig. 5b. reflects the formation of a
more homogeneous final product.
5.2.3. Hillebrandite
The TGArDSC plots for hillebrandite are shown
in Fig. 7. The weight losses up to about 4808C are
believed to be due to the dehydration of minor
impurity phases mainly foshagite. andror the loss
of molecular water from hillebrandite. The main

153

weight change due to loss of structural hydroxyl


species occurs over the temperature ranges 480
5708C 6.8%. and 5706708C 1.9%.. A further
0.8% weight loss occurs between 6708C and 8308C
bringing the total to 9.5%, the stoichiometric water
content of hillebrandite. The minor heat effects between 3508C and 4508C in the DSC plot are due to
the dehydration of impurity phases. In addition to
shoulders at about 4858C and 5258C, clear endothermic features occur with peaks at 5508C and 6508C
with D H values of 363 Jrg 69 kJrmol. and 59
Jrg 11.3 kJrmol., respectively. These features most
likely represent the losses of different OH. groups
whose presence is shown by the infrared spectra see
before..
WAXS data Fig. 8a. show that hillebrandite peaks
begin to decrease in intensity at about 5308C and that
larnite peaks appear at 5408C, with hillebrandite
peaks no longer detectable at 5908C; the reaction
goes to completion in about 10 min. The stacked
plots show that the main larnite peak at about 328 2 u
due to unresolved y301, y121, 002, 121. is broad
becoming sharper as temperature increases. At about
348 2 u , another less intense larnite peak 301. grows
to a maximum size at about 5908C and then decreases in intensity disappearing at about 9008C; this
peak is the best indicator of the growth and breakdown of larnite. Up to 7758C larnite is the only
phase present with peaks for aXL-C 2 S appearing at
this temperature. At this stage the main peak at 328

Fig. 7. DSCrTGA scan for hillebrandite obtained at a heating rate of 58C miny1 . Features below 4508C are believed to be due to impurity
phases. The main weight loss and endotherm around 5205508C are related to loss of OH. molecules from different structural sites in
hillebrandite and simultaneous recystallization to larnite. The endotherm at about 6508C is believed to represent loss of OH from
hydroxylated larnite.

154

S. Shaw et al.r Chemical Geology 167 (2000) 141159

Fig. 8. Stacked WAXS a.rSAXS b. plots vs. temperature for hillebrandite; the vertical scales are intensities in arbitrary units. Every third
spectrum is plotted giving temperature steps of 158C. SAXS data were collected with a 3.5-m camera length. WAXS data show the
X
recrystallization of hillebrandite to larnite and of larnite to a L -Ca 2 SiO4 at around 5508C and 8008C, respectively. SAXS data are considered
further in Fig. 9. See text for further discussion.

S. Shaw et al.r Chemical Geology 167 (2000) 141159

sharpens and splits with a clear shoulder appearing


on the high 2 u side. Note that the most intense peaks
for larnite cover the same 2 u range as those for
aXL-C 2 S complicating peak assignments. Over the
temperature range 7758908C, the peak intensities
for larnite decrease and those for aXL-C 2 S increase,
at 8958C, aXL-C 2 S is the only phase present and
persists up to the highest temperature studied
11008C.. All the cell parameters of hillebrandite
show a linear increase with temperature see Table
2. due to thermal expansion.
The stacked SAXS plot Fig. 8b. shows a complex variation in intensity of the invariant with clear
features occurring at temperatures of 5708C, 6908C
and 8808C. This series of SAXS diffraction patterns
has been analysed using Guinier plots; the variation
of the intensity of the invariant at q s 0 I0 . and the
variation of the radius of gyration R G . with increasing temperature are shown in Fig. 9. The three
features mentioned above are now seen more clearly
as peaks in I 0 with maxima at 5708C, 7008C and
8908C and as sharp decreases or peaks in the radius
of gyration plot at 5508C, 7008C and 9408C. Mean
and
radius of gyration values vary around 250 A
show clear trends related to the dehydration and
recrystallization stages; further implications of the
length-scale established will be addressed in the
discussion.
Combining the data for the TGArDSC and
SAXSrWAXS, the stepwise reactions can be characterised. The main weight change due to loss of OH
as H 2 O and the reaction of hillebrandite to larnite

155

occur at 5205908C; this reaction is endothermic


with an energy of 363 Jrg. Above 5908C, larnite is
the only phase present up to 7758C. The peak at
5705808C in the invariant intensity can be correlated with the stage at which the reaction mixture
consists of 50% of each of hillebrandite and larnite.
Larnite is nominally anhydrous yet a substantial
further amount of water is lost up to about 6708C
with the possibility of a small amount being lost at
even higher temperatures. It seems that the larnite
formed initially is likely to contain a significant
amount of residual hydroxyl and that this phase does
not become fully dehydroxylated until it starts to
transform to aXL-C 2 S. The endothermic peak 59
Jrg. centred at 6508C can be correlated with the
dehydration of the hydroxylated larnite. The peak in
the invariant intensity at 6907008C is likely to
represent the temperature at which the hydroxylated
and anhydrous larnite phases form equal proportions
of the two-phase reaction mixture with the larnite
becoming completely dehydrated at about 6707008C
The larnite to aXL-C 2 S transformation takes place
progressively over the temperature range 7759008C
25 min. accompanied by a continuous increase in
the intensity of the SAXS invariant up to 8908C. The
two-phase region and extended time over which the
phase transition occurred reflects the different
topologies of the silicate monomers and associated
Ca`O polyhedra in the two C 2 S polymorphs involved. The peak in the invariant at 8908C occurs in
a single phase region and presumably represents
electron density contrast between crystalline sample

Fig. 9. Plots of calculated paramaters based on Guinier plots for hillebrandite as a function of temperature. I0 values show maxima at
5708C, 7008C and 8908C which reflect differences in electron density within the sample as it recrystallizes. Radius of gyration values R G .
show sharp decreases or maxima at 5508C, 7008C and 9408C which are related to the variation of particle size during recrystallization
reactions. See text for further discussion.

156

S. Shaw et al.r Chemical Geology 167 (2000) 141159

and pore space; the decrease towards a minimum


value is likely to be due to a progressive sintering of
the sample at temperatures up to 11008C, with production of a denser, more homogeneous sample.
6. Discussion
Most of our TGArDSC and SAXSrWAXS experiments were carried out with the temperature
being ramped at 58Crmin, although some SAXSr
WAXS experiments were also run with a slower
heating rate of 28Crmin. With such dynamic experiments, it would seem likely that only instantaneous
structural changes would be observed to occur at the
appropriate equilibrium temperature. Less rapid
changes would be expected to be delayed and
smeared out over a range of temperature and time.
Thus, reaction temperatures and the ranges of temperature and times. over which these reactions occur are likely to be high estimates.
6.1. Dehydroxylation of CSH phases and double
chain tobermorite
Based on their stoichiometries, both xonotlite and
hillebrandite contain hydroxyl as the only H-component with equivalent H 2 O values of 2.46 and 9.46
wt.%, respectively. Our TGArDSC experiments for
these minerals demonstrate the conditions of dehydroxylation and show that well-defined endothermic
peaks characterise the total heat effects resulting
from dehydroxylation and simultaneous recrystallization. The D H values observed for xonotlite and
hillebrandite are 88 and 363 Jrg of solid sample,
respectively. Taking account of the different OH.
contents per formula unit, the equivalent values are
3550 and 5340 kJrg water loss observed; the similarity of these values might be coincidental as the
measured heats are combined dehydroxylationrrecrystallization transition heats. Stoichiometric
tobermorite is supposed to contain both molec11-A
ular water 9.85 wt.%. and hydroxyl 2.46 wt.% as
H 2 O. and it might be expected that dehydroxylation
of tobermorite would be accompanied by a well-defined endothermic peak of a similar size to that
shown by xonotlite also 2.46% OH as H 2 O. accompanied by the appropriate weight loss. Our
TGArDSC experiments on tobermorite show an endothermic heat effect of about 45 Jrmg sample at

8268C which, by comparison with xonotlite, would


be equivalent to about 1.2% H 2 O, yet no significant
weight loss is detected near 8268C. We conclude that
this endothermic peak is not related to loss of OH.
and indeed it seems that there is no evidence for
OH. in our tobermorite sample. This is backed up
by a lack of sharp OH. peak around 3500 cmy1 in
the infrared spectrum. These points, together with
the absence of a collapse in the c axis when the
molecular water is lost on heating, might suggest that
our natural sample contains a significant proportion
of double dreierketten chains, as suggested from the
presence of the 1200 cmy1 peak in the infrared
spectrum, and in this respect is similar to many
tobermorites Wieker et al., 1982..
synthetic 11-A
The presence of double chains linked via the
bridging tetrahedra would reduce the content of
structural OH., perhaps accounting for the low total
water content of our sample. Lower OH. has been
correlated with lower Ca Hamid, 1981. in line with
the relatively low Ca and water contents of our
sample.
6.2. Larnite ( b-C2 S) to a LX -C2 S transition
Experiments studying the phase transition of anhydrous b-C 2 S to aXL-C 2 S identify a 508C range
over which both phases coexist we.g., 7207708C
heating rate 0.28 miny1 .; Barnes et al., 1980x. The
transition from protonated-b-C 2 S to aXL-C 2 S appears to occur over a much wider temperature range
w6708408C heating rate 108miny1 .; Ishida et al.,
1992x. We find that in our hydroxylated b-C 2 S, the
start of the breakdown to aXL-C 2 S is delayed to
about 7758C and is only complete at about 9008C.
Although it seems likely that the transition might be
extended in dynamic heating experiments compared
to isothermal ones, it is clear that the presence of
residual hydroxyl inherited from precursor hillebrandite increases the thermal stability of the hydroxylated larnite relative to its anhydrous equivalent.
6.3. Variation of the SAXS inariant during dehydration r recrystallization of CSH minerals
We have pointed out that the SAXS invariant
intensity is a direct monitor of the degree of homogeneity of the sample and of the electron density
contrast in it on a length scale of a few tens to
hundreds of angstroms. The final anhydrous product

S. Shaw et al.r Chemical Geology 167 (2000) 141159

of this type of heating experiments would be expected to be a dense, annealed sample with a relatively low invariant intensity. This is the case for the
wollastonite and larnite products for the tobermorite
and hillebrandite experiments, respectively. However, the wollastonite formed from xonotlite shows
no such collapse in the invariant suggesting that the
final product at about 8908C is a single-phase sample
with substantial internal electron density contrast.
Dent and Taylor 1956. showed that a single crystal
of xonotlite was transformed to a single crystal of
low-T-wollastonite preserving the orientation parallel
to b and 100., essentially maintaining the original
shape and size of the initial xonotlite crystallite
despite the higher density of the wollastonite 2.71
and 2.75 grml, respectively.. They suggested that
the size of the crystallite was maintained during the
dehydroxylation and shrinkage of the a axis by the
opening up of minute spaces between the particle
fibres parallel to b. It seems likely that this type of
porosity is also present in our sample and that the
high electron density contrast in the SAXS reflects
scattering from the wollastonite fibres and pore-space
combined.
6.4. Radius of gyration (R G ) ariation
The R G values for xonotlite and tobermorite and
their wollastonite breakdown products are fairly constant with varying temperature at 250 " 3 and 86 " 2
respectively. As shown in Fig. 9, the hillebranA,
dite, and its dehydration products larnite and aXL-C 2 S,
show a systematic variation over the range 230290
with aXL-C 2 S showing the longest lengthscale.
A
These values are clearly too small to be considered
as discrete grain sizes but fall at the lower end of the
. recently detersilicalite particle sizes 151750 A
mined by SAXS and TEM de Moor et al., 1997.,
and are similar to those determined for pore-space in
. by SAXS Hall et al., 1986..
shales 90150 A
Assuming that the samples can be considered to
be monodisperse to a first approximation, the R G
values can be corrected to actual mean geometric
shapes as follows:
Sphere radius R . R G s 5r3 . R 2
and
Cylinder semiaxis r , height h . R G s r 2r2 q h 2r12,
after Glatter, 1992..

157

for hillebranTaking a mean R G value of 250 A


dite as an example, the mean radius for spherical
640 A diameter..
particles would be about 320 A
CSH minerals generally have fibrous or platey morphologies. If we assume a cylindrical fibre with
semi-axis r and height h s 10r, r s 84 and h s 840
equivalent to a fibre of dimensions 168 A wide =
A,
long. Assuming a platey particle with semi840 A
and h s 35
axis r and height h s 0.1r, r s 350 A
equivalent to a plate 700 A wide = 35 A thick.
A,
TEM studies of CSH minerals e.g., Hamid, 1981;
Kim et al., 1993. and wollastonite e.g., Wenk et al.,
1976. show many features attributed to stacking
faults and defects. Such features in hillebrandite have
been explained by displacement of blocks consisting of one or several units along the fibre axis b .
Kim et al., 1993.. It is possible that the electron
densities being probed by the SAXS data reported
here result from such stacking domain heterogeneities. We hope to test this hypothesis using high
resolution TEM on the CSH minerals and dehydration products.
7. Conclusions
1. Our dynamic, in situ heating experiments provide temperatures marking the beginning of dehydrationrrecrystallization in generally good agreement
with earlier studies. However, in all cases two-phase
fields are present, the temperature width of which
can be correlated with the degree of structural rearrangement necessary to convert the CSH mineral to
an anhydrous or less-hydrated higher temperature
phase.
2. Irrespective of the detailed structural rearrangements, all of the CSH mineral breakdown reactions take place on a time scale of minute: xonotlite
to wollastonite double chain to single chain. occurs
between 7808C and 8358C and is complete in about
11 min; tobermorite to wollastonite single and double chains to single chains. occurs between 8408C
and 8558C 3 min.; and hillebrandite to larnite single
chains to monomers. takes place over the range
5208C to 5908C 14 min..

3. Our natural tobermorite is an anomalous 11-A


form with no evidence of hydroxyl groups in the
structure suggesting that a proportion of the principal

158

S. Shaw et al.r Chemical Geology 167 (2000) 141159

silicate units may be double rather than single dreierketten chains as is confirmed by the presence of an
infrared peak at 1200 cmy1 , a mode characteristic of
the double chains in xonotlite.
4. We find that hillebrandite breaks down to
larnite b-C 2 S. rather than to wollastonite plus portlandite confirming the findings of Ishida et al. 1992.
rather than those of Kim et al. 1992.. The larnite
formed initially appears to be a hydroxylated form as
has been reported by other workers e.g., Ishida et
al., 1992.. However, we find that the presence of this
hydroxyl extends the upper stability limit of larnite
to higher temperatures than previously reported.
5. The dehydroxylationrrecrystallization of
xonotlite and hillebrandite both show endothermic
peaks with similar energies per g OH as H 2 O. lost:
3550 and 5340 Jrg H 2 O, respectively. Dehydration
of hydroxylated larnite at about 6508C is accompanied by an endothermic peak of energy 59 Jrg
sample, equivalent to 3105 Jrg H 2 O lost, a value
fairly similar to those given above. The recrystallization of dehydrated tobermorite to wollastonite is
characterised by an exothermic heat of 189 Jrg
sample.
6. Guinier analysis of the SAXS data suggest
diameter. This
particle sizes up to about 640 A
length scale is unlikely to represent the size of
discrete grains and possibly gives information on the
sizes of domains between zones of defects formed by
displacements of structural units along the dreierketten chains, which are the basic structural feature in
this group of minerals.

Acknowledgements
We thank Dr Paul Schofield, Natural History
Museum, London, for providing the samples of tobermorite and hillebrandite, Dr G. Sankar for arranging use of the Royal Institution furnace, Dr G. Mant
for advising on use of the OTOKO computer package, and Dr W. Bras for advice on SAXS interpretation. We also thank Cathy Davies for carrying out
the TGArDSC experiments, Drs Simon Clark and
Simon Redfern for advice on X-ray diffraction theory, and Prof. John Parise and Simon Clark for their
constructive reviews on the paper. NERC financed a
research studentship for Sam Shaw GT4r96r202r

E. and provided beamtime support on the block


grant to Daresbury Laboratory.
References
Atkinson, A., Harris, A.W., Hearne, J.A., 1995. Hydrothermal
alteration and ageing of synthetic calcium silicate hydrate gels.
UK Nirex Report NSSrR374.
Barnes, P., Fentiman, C.H., Jeffery, J.W., 1980. Structurally
related dicalcium silicate phases. Acta Cryst. A36, 353356.
Bras, W., Derbyshire, G.E., Ryan, A.J., Mant, G.R., Felton, A.,
Lewis, R.A., Hall, C.J., Greaves, G.N., 1993. Nuclear Instruments and Methods in Physics Research A 326, 587591.
Cheary, R.W., Coelho, A.A., 1992. A fundamental parameters
approach of X-ray line-profile fitting. J. Appl. Cryst. 25,
109121.
Colyer, L.M., Greaves, G.N., Komanschek, B.U., Roberts, M.A.,
Dent, A.J., Fox, K.K., Carr, S.W., 1996. Dynamic observation
of crystallographic and micro structural changes associated
with the collapse and recrystallization of Cd-exchanged zeolite-A. Mater. Sci. Forum 228231, 363368.
Dai, Y., Post, J.E., 1995. Crystal structure of hillebrandite: a
natural analogue of calcium silicate hydrate CSH. phases in
Portland cement. Am. Mineral. 80, 841844.
de Moor, P.-P.E.A., Beelen, T.P.M., Komanschek, B.U., Diat, O.,
van Santen, R.A., 1997. In situ investigation of Si-TPA-MFI
crystallization using ultra-. small- and wide-angle X-ray scattering. J. Phys. Chem. B 101, 1107711086.
Dent, L.S., Taylor, H.F.W., 1956. The dehydration of xonotlite.
Acta Cryst. 9, 10021004.
Dent, A.J., Greaves, G.N., Roberts, M.A., Sankar, G., Wright,
P.A., Jones, R.H., Sheehy, M., Madill, D., Catlow, C.R.A.,
Thomas, J.M., Rayment, T., 1995. A new furnace design for
use in combined X-ray-absorption and diffraction of catalysis
and ceramics studies formation from carbonate precursors
of Cu, Co, Mn spinels for the oxidation of CO and the
formation of PLZT, a piezoelectric ceramic. Nuclear Instruments and Methods in Physics Research B 97, 2022.
El-Hemaly, S.A.S., Mitsuda, T., Taylor, H.F.W., 1977. Synthesis
of normal and anomalous tobermorite. Cem. Concr. Res. 7,
429432.
Eysel, W., Hahn, T., 1970. Polymorphism and solid solution of
Ca 2 GeO4 and Ca 2 SiO4 . Zeits. Krist. 131, 322341.
Faucon, P., Adenot, F., Jacquinot, J.F., Pettit, J.C., Cabrillac, R.,
Jorda, M., 1998. Long term behaviour of cement pastes used
for nuclear waste disposal: review of physico-chemical mechanisms of water degradation. Cem. Concr. Res. 28, 847857.
Gard, J.A., Taylor, H.F.W., 1957. A further investigation of
tobermorite from Loch Eynort. Scotland Mineral. Mag. 31,
361370.
Glatter, O., 1992. Small-angle scattering. In: Wilson, A.J.C. Ed..,
International Tables for Crystallography. Kluwer Academic
Publishers, Dordrecht, pp. 89112.
Hall, P.L., Mildner, D.F.R., Borst, R.L., 1986. Small-angle scattering studies of the pore spaces of shaly rocks. J. Geophys.
Res. 91, 21832192.

S. Shaw et al.r Chemical Geology 167 (2000) 141159

natural
Hamid, S.A., 1981. The crystal structure of the 11 A
tobermorite Ca 2.25 wSi 3 O 7.5 ` OH.1.5 xP1H 2 O. Zeits. Krist. 154,
189198.
Henmi, C., Kusachi, I., 1992. Clinotobermorite, Ca 5 Si 6 O,OH.18
P5H 2 O, a new mineral from Fuka, Okayama Prefecture, Japan.
Mineral. Mag. 56, 353358.
Henning, O., 1974. Cements; the hydrated silicates and aluminates. In: Farmer, V.C. Ed.., Infrared Spectra of Minerals.
Mineral Society, London, pp. 445463.
Hoffman, C., Armbruster, T., 1997. Clinotobermorite,
Ca 5 wSi 3 O 8 OH.x2 P4H 2 O ` Ca 5 wSi 6 O17 xP5H 2 O, a natural C-SHI. type cement mineral: determination of the substructure.
Zeits. Krist. 212, 864873.
Holland, T.J.B., Redfern, S.A.T., 1997. Unit cell refinement from
powder diffraction data: the use of regression diagnostics.
Mineral. Mag. 61, 6577.
Ishida, H., Mabuchi, K., Sasaki, K., Mitsuda, T., 1992. Low-temperature synthesis of b-Ca 2 SiO4 from hillebrandite. J. Am.
Ceram. Soc. 75, 24272432.
Jauberthie, R., Temimi, M., Laquerbe, M., 1996. Hydrothermal
tobermorite. Cem.
transformation of tobermorite gel to 10 A
Concr. Res. 26, 13351339.
Johnson, C.A., Schweizer, C., 1998. Weathering processes in
municipal solid waste incinerator bottom ash deposits. Min.
Mag. 62, 723724.
Kim, Y.J., Kriven, W.M., 1995. A transmission electron microscopy study on the decomposition of synthetic hillebrandite
Ca 2 SiO4PH 2 O.. J. Mater. Res. 10, 30843095.
Kim, Y.J., Nettleship, I., Kriven, W.M., 1992. Phase transformations in dicalcium silicate: II. TEM studies of crystallography,
microstructure, and mechanisms. J. Ceram. Soc. 75, 2407
2419.
Kim, Y.J., Kriven, W.M., Mitsuda, T., 1993. TEM study of
synthetic hillebrandite Ca 2 SiO4 H 2 O.. J. Mater. Res. 8,
29482953.
Komarneni, S., Roy, R., Roy, D.M., Fyfe, C.A., Kennedy, G.J.,
Bothner-By, A.A., Dadok, J., Chesnick, A.S., 1985. 27Al and
29
Si magic angle spinning nuclear magnetic resonance spectroscopy of Al-substituted tobermorites. J. Mater. Sci. 20,
42094214.
Kudoh, Y., Takeuchi, Y., 1979. Polytypism of xonotlite: 1.
Structures of an A-1 polytype. Min. J. 9, 349373.
Mamedov, Kh.S., Belov, N.V., 1955. Structure of xonotlite. Dokl.
Akad. Nauk SSSR 104, 615618.
Mamedov, Kh.S., Belov, N.V., 1956. The crystal structure of
wollastonite. Dokl. Akad. Nauk SSSR 107, 463466.
McConnell, J.D.C., 1954. The hydrated calcium silicates riversideite, tobermorite and plombierite. Mineral. Mag. 30, 293
305.
Meyer, J.W., Jaunarajs, K.L., 1961. Synthesis and crystal chemistry of gyrolite and reyerite. Am. Mineral. 46, 913933.

159

Mitsuda, T., Taylor, H.F.W., 1975. Influence of aluminium on the


toberconversion of calcium silicate hydrate gel into 11 A
morite at 908C and 1208C. Cem. Concr. Res. 5, 203210.
Mitsuda, T., Taylor, H.F.W., 1978. Normal and anomalous tobermorites. Mineral. Mag. 42, 229235.
Mumme, W., Hill, R.J., Bushnell-Wye, G., Segnit, E.R., 1995.
Rietveld crystal structure refinements, crystal chemistry and
calculated powder diffraction data for the polymorphs of
dicalcium silicate and related phases. Neues Jahrb. Mineral.,
Abh. 169, 3568.
Mumme, W., Cranswick, L., Chakoumakos, B., 1996. Rietveld
crystal structure refinements from high temperature neutron
powder diffraction data for the polymorphs of dicalcium silicate. Neues Jahrb. Mineral., Abh. 170, 171188.
Noma, H., Adachi, Y., Yamada, H., Nishino, T., Matsuda, Y.,
Yokoyama, T., 1998. 29 Si MAS NMR spectroscopy of poorly
crystalline calcium silicate hydrates C-S-H.. In: Colombet, P.,
Grimmer, A.-R., Zanni, H., Sozzani, P. Eds.., Nuclear Magnetic Resonance Spectroscopy of Cement-based Materials.
Springer-Verlag, Berlin, pp. 159168.
Okada, Y., Masuda, T., Takada, M., Xu, L., Mitsuda, T., 1998.
Relationship between NMR 29 Si chemical shifts and FTIR
wave numbers in calcium silicates. In: Colombet, P., Grimmer, A.-R., Zanni, H., Sozzani, P. Eds.., Nuclear Magnetic
Resonance Spectroscopy of Cement-based Materials.
Springer-Verlag, Berlin, pp. 6978.
Remy, C., Reynard, B., Madon, M., 1997. Raman spectroscopic
investigations of dicalcium silicate: polymorphs and high-temperature phase transitions. J. Am. Ceram. Soc. 80, 413423.
Shaw, S., Clark, S.M., Henderson, C.M.B., 1999. Hydrothermal
formation of the calcium silicate hydrates tobermorite and
xonotlite: an in situ synchrotron study. This volume.
Taylor, H.F.W., 1990. Cement Chemistry. Academic Press, San
Diego, CA.
Tilley, C.E., 1951. A note on the progressive metamorphism of
siliceous limestones and dolomites. Geol. Mag. 88, 175178.
Tsuji, M., Komarneni, S., Malla, P., 1991. Substituted tobermorites: 27Al and 29 Si MASNMR, cation exchange, water
sorption studies. J. Am. Ceram. Soc. 74, 274279.
Wenk, H.-R., Muller, W.F., Liddell, N.A., Phakey, P.P., 1976.
Electron Microscopy in Mineralogy. Springer, Berlin, 324 pp.
Tobermorits
Wieker, W., 1968. Silikatanionenstruktur des 14 A
von Crestmore und seiner Entwasserungsprodukte.
Z. Anorg.

Allg. Chem. 360, 307316.


Wieker, W., Grimmer, A.R., Winkler, A., Magi, M., Tarmak, M.,
Lipmaa, E., 1982. Solid-state high-resolution 29 Si NMR spec 11 A and 9 A tobermorites. Cem.
troscopy of synthetic 14 A,
Concr. Res. 12, 333339.
Yamanaka, T., Mori, H., 1981. The structure and polytypism of
alpha-CaSiO3 pseudowollastonite.. Acta Cryst. B37, 1010
1017.

You might also like