Determination of reaction kinetics and mechanisms of 1.13 NM

DETERMINATION OF REACTION KINETICS AND
MECHANISMS OF 1.13NM TOBERMORITE BY
IN-SITU NEUTRON DIFFRACTION
Saskia Bernstein
Dissertation
Fakultät für Geowissenschaften
der Ludwig-Maximilians-Universität
München
vorgelegt von
Saskia Bernstein
aus Schönebeck
München den 12.Januar 2011
Erstgutachter:
Prof. Dr. Karl Thomas Fehr
Zweitgutachter:
Prof. Dr. Herbert Pöllmann
Tag der Disputation:
27. Mai 2011
the cupola of the pantheon in Rome, build by the Romans using the oldest
concrete-like material of the world - Opus Caementitium
3
Table of Content
i)
List of figures...........................................................................................6
ii)
Danksagung............................................................................................7
iii)
Summary..................................................................................................9
iv)
Zusammenfassung.................................................................................11
Chapter 1: Introduction
1.1.
Calcium silicate hydrates..............................................................................14
1.2.
Aerated autoclaved concrete........................................................................15
1.2.1 History...................................................................................................15
1.2.2 Industrial production............................................................................16
1.2.3 Properties..............................................................................................18
1.3.
Tobermorite and AAC....................................................................................19
Chapter 2: Crystal Chemistry of Xonotlite Ca6Si6O17(OH)2. Part I: Determination
of Polytypes using X-Ray Powder Diffraction
2.1. Introduction........................................................................................................22
2.2. Structure and polytypism of Xonotlite............................................................23
2.3. Assignment of polytypes..................................................................................25
2.4. Correlation of chemical composition and assigned polytypes....................27
2.5. Conclusion.........................................................................................................28
4
Chapter 3: HAND: An Hydrothermal Autoclave for Neutron Diffraction
3.1. Introduction........................................................................................................29
3.2. Design of HAND.................................................................................................30
3.3 The applicability of HAND for neutron diffraction experiments.....................31
3.4. Conclusion.........................................................................................................32
Chapter 4: Reaction and growth kinetics of 1.13nm tobermorite crystallizing in
AAC
4.1. Introduction........................................................................................................34
4.2. Influence of quartz grain size...........................................................................36
4.3. Influence of temperature...................................................................................38
4.4. Quenching experiments....................................................................................39
4.5. Calculated rate constants and activation energies........................................40
4.6. Conclusion.........................................................................................................41
Chapter 5:outlook
5.1. Outlook...............................................................................................................42
5.2. References.........................................................................................................43
5.3. Appendix............................................................................................................45
A
Internship-report, Xella Thermopierre, (unpublished)...................46
B
Crystal Chemistry of Xonotlite Ca6Si6O17(OH)2. Part I:
Determination of Polytypes using X-Ray Powder Diffraction.......61
5
C
A Hydrothermal Autoclave for Neutron Diffraction (HAND) –
Design, Technique and Applicability................................................86
D
The Formation of 1.13nm Tobermorite under Hydrothermal
Conditions: 1. The influence of quartz grain size within
the system CaO-SiO2D2O.....................................................................................................103
E
The hydrothermal formation of 1.13nm tobermorite within the
system CaO-SiO2-D2O: a kinetic study by in situ neutron
diffraction..........................................................................................113
5.4. Curriculum Vitae....................................................................................................
6
List of figures
Chapter 1.
Figure 1.1 ternary system CaO-SiO2-H2O......................................................14
Figure 1.2 development of YTONG-products from 1930 to 1975...................15
Figure 1.3 pore formation in AAC after Hofmann (2008) ...............................16
Figure 1.4 hydration of CaO and clinker phases.............................................17
Figure 1.5 formation of 1.13nm tobermorite....................................................17
Figure 1.6 REM- picture of tobermorite crystals..............................................19
Figure 1.7 p-T diagram of tobermortie stability...............................................19
Figure 1.8 structure of tobermorite and xonotlite............................................20
Chapter 2.
Figure 2.1 polytypes of xonotlite.....................................................................24
Chapter 3.
Figure 3.1 schematic drawing of HAND..........................................................30
Figure 3.2 time resoluted diffraction patterns..................................................31
Figure 3.3 onset of hk0 and 00l reflections of 1.13nm tobermorite.................32
Chapter 4.
Figure 4.1 REM picture of reacted quartz and newly evolved
phases...........................................................................................37
Figure 4.2 transition times and portlandite expense and tobermorite
occurrence versus reaction temperature
38
Figure 4.3 rate constant versus specific
surface..........................................................................................40
7
Danksagung
Jetzt, da diese Arbeit anscheinend doch noch ein gutes Ende findet, ist es an der
Zeit, denjenigen zu danken, die mich auf diesem doch recht langen und oft auch
steinigen Weg begleitet haben. Es sind in 6 Jahren so viele gewesen, dass Ihr mir
hoffentlich nicht böse seid, wenn ein paar ungenannt bleiben.
Zunächst einmal gilt mein innigster Dank meinen Eltern, die mich immer auf jede
erdenkliche Weise unterstütz haben und die mir wenn ich es brauchte zur Seite
standen. Ohne euch wäre ich nicht da wo ich jetzt bin.
Meinem Betreuer möchte ich danken für die Möglichkeit zu dieser Doktorarbeit. Ich
danke ihm für all die Möglichkeiten zur Diskussion, für die vielen ermutigenden,
manchmal auch strengen Worte, für den Freiraum den er mir gelassen hat und nicht
zuletzt für den Kaffee, der immer bereit stand. Danke Tommi !
Ein ganz besonderes Dankeschön geht an Andi Laumann, dem besten Zuhörer den
ich kenne. Dafür das er wusste, mit meinen Launen umzugehen aber vor allem dafür,
dass er die vergangenen 10 Jahre immer ein guter Freund war und es hoffentlich
auch noch in den folgenden sein wird.
Meiner Nuria danke ich für ihr Verständnis, für die stundenlangen Gespräche und für
den Spaß den wir zusammen hatten... un besito
Dir Pablo danke ich einfach nur fürs zur Seite stehen, manchmal sicher die
schwerste Aufgabe.
Danke an Yaping, Andi, Christoph, Michi, Felix und Linda, dass sie mein kreatives
Chaos ertragen haben, welches manchmal doch recht raumfüllend war und einfach
dafür dass sie die besten Büro-Kollegen sind die man sich vorstellen kann.
8
Dem Freistaat Bayern möchte ich danken, dass er mir diese Arbeit überhaupt
ermöglicht hat durch die finanzielle Unterstützung mittels einem Stipendium nach
dem „Bayerischen Eliteförderungsgesetz“
Dr. Rupert Hochleitner, Kurator der Mineralogischen Staatssammlung danke ich für
die Bereitstellung zahlreicher Mineralien, die ich für meine Untersuchungen
benötigte.
Ich danke auch allen, die an der Entwicklung und des Baus der für die Experimente
benötigten Autoklaven beteiligt waren.
Herrn Tom Hansen danke ich ganz besonders für die Unterstützung zu jeder Tages. und Nachtzeit während der Messzeiten am ILL. Ohne ihn gäbe es die Daten nicht,
die den Kern dieser Arbeit bilden.
9
Summary
1.13nm tobermorite belongs to the mineral group of calcium-silicate-hydrates or short
CSH-phases. Natural CSH-phases are due to the specific areas of formation rare in
nature but synthetic members have applications in many different sections of the
industry. 1.13nm tobermorite for example is the main binding phase in aerated
autoclaved concrete (AAC), a building material which became more and more
important over the last decades. It is easy to process due to its low density but
nevertheless offers an excellent strength resistance caused by the interlocking
lathlike tobermorite crystals. The embodied air filled pores effect very good insulation
behaviour, which is of great importance for an ecologically sensitive method of
construction. Despite the high significance of AAC for the building industry the
standard of knowledge referring to the ongoing processes during production is still
not satisfying. The applied conditions during the production are in part based on
empirical values and just the macroscopic properties of the product are controlled.
The macroscopic properties are primary determined by the microscopic ones, like
type and amount of formed mineral phases and their structure and texture. Several
scientific studies dealing with this topic are already published but the hence resulting
set of data on mineral forming processes and their kinetics is still deficient. Previous
studies have shown that the formation of 1.13nm tobermorite is just metastable under
the conditions present during productions. In equilibrium 1.13nm tobermorite
decomposes to xonotlite and quartz, which has a fatal influence on the strength
resistance of the building material. Therefore the present work addresses in detail on
the determination of the reaction kinetics of the formation of 1.13nm tobermorite
within the system CaO-SiO2-H2O by in situ neutron diffraction. To assure a
successful
interpretation
of
the
experimental
data,
crystallographic
well
10
characterised sample material is needed. Therefore the crystal chemistry of natural
xonotlites was previously studied and the results are integrated in the presented
work. The neutron diffraction experiments were conducted at three different
temperatures and to different grain sizes of quartz to determine the influence of these
parameters. The experiments were carried out at the D20 powder diffractometer of
the the Institute Laue Langevin research reactor in Grenoble (France). For this
purpose an autoclave (HAND) was designed which enables an investigation of the
previously synthesized greenbodies under saturated steam pressure. The high
neutron flux at the D20 provides a time resolution of 1 exposure per minute, hence
sufficient data for the early state of the reaction could be collected as well. The
obtained diffraction pattern were evaluated with respect to the reaction progress and
subsequently interpreted in terms of the present reaction process by using a kinetic
model. It could be shown that the formation of 1.13nm tobermorite is a non-isokinetic
process with changes in the reaction mechanism from solution control to diffusion
control and in most of the cases back to diffusion control. The determined transition
times and points of portlandite expense and tobermorite occurrence were compared
with respect to the influence of reaction temperature and grain size of quartz. Based
on the data for the reaction progress, the rate constant for the different sections of
the reaction were calculated. Using the rate constants determined at different
temperatures the activation energies of the tobermorite formation were calculated.
11
Zusammenfassung
1.13nm Tobermorite gehört zur Mineralgruppe der Calciumsilikathydrate oder kurz
CSH-Phasen. Natürliche Vertreter dieser Gruppe sind aufgrund der speziellen
Bildungsräume in der Natur eher selten, synthetische CSH-Phasen finden jedoch
Anwendung in den verschiedensten Bereichen der Industrie. 1.13nm Tobermorite
zum Beispiel ist die primäre Phase in Porenbeton, einem Baustoff der in den letzen
Jahrzehnten immer mehr an Bedeutung gewann. Er ist aufgrund seiner geringen
Dichte
leicht
zu
Festigkeitseigenschaften
verarbeiten,
durch
bietet
die
aber
trotzdem
Verzahnung
der
hervorragende
lattenförmigen
Tobermoritkristalle. Die enthaltenen luftgefüllten Poren bewirken außerdem sehr gute
Dämmeigenschaften, welche für eine ökologische umweltbewusste Bauweise von
hoher Bedeutung sind. Trotz des hohen Stellenwertes von Porenbetonprodukten in
der Bauindustrie ist der Wissensstand über die während der Herstellung ablaufenden
Prozesse noch unbefriedigend. Die während der Herstellung herrschenden
verwendeten Produktionsbedingungen beruhen meist auf empirischen Werten und
überprüft werden nur die makroskopischen Eigenschaften des fertigen Baustoffs. Die
makroskopischen Eigenschaften werden aber primär durch die mikroskopischen
Eigenschaften, sprich Menge und Art der entstehende Mineralphasen und deren
Struktur und Textur, bestimmt werden. Mehrere wissenschaftliche Arbeiten zu
diesem Thema wurden bereits veröffentlicht, der daraus resultierende Datensatz
bezüglich der ablaufenden Mineralbildungsprozesse und deren Kinetik ist aber immer
noch unzureichend. In vorangegangenen Arbeiten wurde gezeigt, dass sich 1.13nm
Tobermorit im Porenbeton unter den gegebenen Bedingungen nur metastabil bildet.
Unter Gleichgewichtbedingungen zerfällt dieser zu Xonotlite und Quarz, was sich in
verheerendem Maße negativ auf die Festigkeitseigenschaften des Baustoffs
12
auswirken würde. Die vorliegende Arbeit befasst sich daher eingehend mit der
Bestimmung der Reaktionskinetik von 1.13nm Tobermorit im System CaO-SiO2-H2O
mittels in situ Neutronenbeugung..Um die in den Versuchen gewonnenen Daten
erfolgreich auswerten zu können, erfordert es kristallographisch gut charakterisiertes
Probenmaterial. Aus diesem Grund wurde die Kristallchemie von natürlichen
Xonotliten im Vorfeld eingehend studiert und die gewonnenen ergebnisse sollen mit
in die vorliegende Arbeit einfließen. Die Neutronenbeugungs-Experimente wurden
bei drei verschiedenen Temperaturen und mit zwei unterschiedlichen Korngrößen der
Quartzfraktion durchgeführt, um den Einfluss dieser Parameter beurteilen zu können
Die
in
die
Auswertung
einbezogenen
Experimente
wurden
am
D20
Pulverdiffraktometer des Forschungsneutronenreaktors des Instituts Laue-Langevin
in Grenoble (Frankreich) durchgeführt. Hierfür wurde eigens ein Autoklav (HAND)
entwickelt, der es ermöglicht die zuvor hergestellten Grünkörper unter gesättigter
Dampfatmosphäre zu untersuchen. Der hohe Neutronenfluss am D20 ermöglicht
eine zeitliche Auflösung von 1 Diffraktogramm pro Minute, so konnten genügend
Daten für das Frühstadium der Reaktion gesammelt werden. Die so erhaltenen
Aufnahmen wurden bezüglich des Reaktionsumsatzes ausgewertet und mithilfe von
kinetischen Modellen hinsichtlich des Reaktionsmechanismus interpretiert. Es zeigte
sich das die Bildung von 1.13nm Tobermorit nicht isokinetisch ist, sondern zunächst
ein
lösungskontrollierter
Abschnitt
vorliegt
welcher
übergeht
in
einen
diffusionskontrollierten Teil und in den meisten Fällen erneut wechselt zu einem
lösungskontrollierten Abschnitt. Die so erhaltenen Üebergangszeiten und Zeitpunkte
des vollständigen Verbrauchs von Portlandit und des ersten Auftretens von 1.13nm
Tobermorit wurden hinsichtlich des Einflusses von Reaktionstemperatur und
Quarzkorngröße miteinander verglichen. Aus den Daten wurden anschließend die
Geschwindigkeitskonstanten
für
die
verschiedenen
Abschnitte
der
Reaktion
13
berechnet. Aus den so erhaltenen Geschwindigkeitskonstanten bei verschiedenen
Temperaturen konnten die Aktivierungsenergien bestimmt werden.
14
Chapter 1: Introduction
1.1 Calcium Silicate Hydrates (CSH)
Calcium-silicate-hydrates are mineral phases mostly formed in weathered and
hydrothermally altered basic rocks. One of the most famous deposits is known from
Maroldsweisach,
Bavaria,
where
calcium-
natural
silicate-hydrates are found
in xenolites inside basalt.
Due to the spatial and in
terms
of
geology
short
occurrence of the needed
formation
calcium
conditions,
silicate
hydrates
are rather rare. The huge
Fig.1.1.: ternary system CaO-SiO2-H2O with some of the most
important natural CSH-phases and the composition of
recipes of common building materials
variety and the complicate
crystal chemistry despite a
simple composition of calcium silicium and water (Fig:1.1) made them subject of
several studies. But their main importance lies in the construction industry. Calciumsilicate-hydrates are the main binding phases in many building materials. The basic
principal is the bonding of grained materials by inorganic or hydraulic binders like
lime and portland cement, respectively. As most common and widespread material
concrete needs to be mentioned. Concrete can be described as artificial chemical
sedimentary rock where aggregates like sand or gravel are bonded by the
crystallizing phases in the cement after adding water. The acicular crystals are
interlock with each other causing the desired resistance and constructive strength of
15
the material. Already 2000 years ago the Romans had this knowledge and used the
first concrete like building material known as OPUS CAEMENTITIUM for their
monumental examples of architecture like the pantheon (cover picture) or the
colosseum in Rome. But beside concrete ,steam cured building materials like limesand bricks and aerated autoclaved concrete gained more and more importance in
our era.
1.2. Aerated autoclaved concrete
1.2.1 History
Fig.1.2: increase of YTONG products in 10³m³ from 1930 to 1975 and improvement of thermal
conductivity from 1975 to 2000 for the strength categories 2, 4 and 6 madified after Dubral
(1992)
16
Due to the scarcity of energy and recourses after the First World War the
governments of many European countries tightened the requirements of building
materials in terms of insulation behaviour and cost of production and supported a lot
of scientific research. In 1924 the Swedish scientist A. Eriksson developed the basic
method of producing aerated autoclaved concrete based on the preliminary work of
Zernikov and Michaelis on lime sand mortar combined with the pore-forming method
of Aylsworth and Dyer by adding metal powders to the mixture. The material is
outstanding for its good heat insulation behaviour caused by the pores (Fig.1.2.) and
a high compressive strength despite a low density.
1.2.2 Industrial production
The first industrial production started in 1929 in Yxhult (Sweden) intrducing the trade
name YTONG (YXHULTS ÅNGEHÄRDADE GASBETONG). The number of plants
increased over the thirties due to the huge demand. The Second World War
temporarily haltered the quick spread of AAC but nevertheless the development of
Fig.1.3.:formation of pores by the alkaline reaction of Al and Ca(OH) after Homann (2008)
AAC product worldwide increased immensely between 1929 and 1975 as depicted in
Fig.1.2. Today the annual production In western Europe add up to 8.65 mio m³ over
50% of which are produced in Germany (Dubral, 1992).The global success was first
17
Fig.1.4: hydration of CaO and cement clinker phases to form ca(OH) and Ca(OH) and tricalcium
silicate hydrate,respectively, modified after Homan (2008)
of all established by the companies YTONG, Siporex and Durox.
Over the years some improvements were made on the industrial production of AAC
like cutting with wires instead of saw but the main concept stayed the same. After
grinding the raw materials lime, sand and cement are mixed followed by an
exhaustive dispersing with water. To safe cost the waste from cutting and recycled
Fig.1.5.: formation of 1.13nm tobermorite by the reaction of ground quartz and Ca(OH) or calcium
silicate hydrate, respectively, modified after Homan (2008)
AAC is added to the raw materials as well as anhydrite or gypsum to advance the
formation of the desired mineral phase. Right before pouring the paste through the
moulds aluminium is added to the mixture as pore-builder. The paste rests for about
18
two hours at elevated temperatures. During that time pores are formed by the
reaction of aluminium with calcium hydroxide and water (see Fig 1.3) causing an
expansion of the cake. Meanwhile the quicklime and the cement clinker phases
dicalciumsilicate (C2S) and tricalciumsilicate (C3S) react with water to form calcium
hydroxide and calciumhydroxide and tricalciumsilicatehydrate, respectively (see
fig.1.4) and a basic resistance is reached. The greenbody is removed from the
mould, cut and stored in the autoclaves for hydrothermal curing. The curing normally
takes place at temperatures around 190°C and saturated steam pressure (12.5bar)
over a time span of 6 up to 12h. During autoclaving the crystalline phase 1.13nm
tobermorite is formed by the reactions shown in (Fig 1.5)
1.2.3 Properties
The material AAC is outstanding for its high compressive strength despite a very low
weight and density due to the pores. The pores causing as well a extremely good
heat and acoustic insulation behaviour. Important values to determine quality of the
building material are the shrinkage, the e-modul and the heat conductivity which are
monitored regularly during production. Those values are optimized in the industrial
production process by changing the ratio of the different raw materials and therefore
the raw density of the “cake” as explained in the unpublished report resulting from an
internship in an AAC-plant in Bourgoin Jallieu, France (appendix A).
This empiric method in fact leads to the desired results but the scientific research of
the last decades has shown that the macroscopic properties of AAC and other
construction materials are controlled by the microscopic ones, speaking of the type
and structure of the evolving CSH-phases. A targeted control of the ongoing mineral
forming reactions and therefore the macroscopic properties of AAC is just possible
19
with
an
knowledge
extensive
of
the
quantitative
structural
characteristics, the thermodynamic
parameters and the reaction kinetic.
Despite the intense research of the
last decades this is still insufficient
and should be the main aim of the
Fig.1.6: network of lathlike tobermorite crystals in
AAC
presented work. The influence of
temperature and grain size of quartz on the formation of 1.13nm tobermorite was
studied by in-situ neutron diffraction. Based on the experiments detailed information
on the reaction mechanisms were obtained and beyond that rate constants and
activation energies were calculated as explained in the chapter 4.
1.3 Tobermorite in AAC
As mentioned before (Fig.
1.4), 1.13nm tobermorite is
the main phase evolving
during
the
hardening
hydrothermal
of
AAC.
The
lathlike crystals interlock to
each other to form a network
causing
the
good
Fig.1.7: p-T diagram of tobermorite stability
compressive strength of the
product (Fig. 1.6). The tobermorite-forming reactions does not reach the chemical
equilibrium within the technical time scales, thus under the chosen curing conditions
20
the formed 1.13nm tobermorite is stabilized metastable (Gabrosek et al., 1993; Fehr
and Zuern, 1997; Zuern and Fehr,2000.). Under equilibrium conditions 1.13nm
tobermorite decomposes to xonolite and quartz (fig.1.7). Xonotlite has a fibrous
structure which would remarkably decrease the compressive strength of the brick.
Both 1.13nm tobermorite
and
xonotlite
show
characteristic
the
structural
features of infinite silicate
double chains of a type
called Dreier-Doppelketten
built
up
of
condensed
Dreierketten
common
almost
CSH-Phases
all
to
(fig.1.8). The chains are
intercalated
by
a
Ca-O
layer (portlandite layer) so
the structure consists of a
central layer of calcium
Fig.1.8: structure of tobermorite (a) and xonotlite (b) modified
after Bonaccorsi et al (2005) and Hejny & Armbruster (2001)
octahedra
which
has
silicate sheets on each
side. The calcium octahedra share oxygens with the silicate tetrahedra, the distance
between two edges in the calcium octahedral layer is about the same length as a
silicate Dreierketten unit. This feature enables differing linkage possibilities of Ca and
Si layers and therefore polytism occurs. For both xonotlite and tobermorite several
polytypes are known. Often the crystals show an intergrowth of more then one
polytype which complicates an assignment by diffraction techniques. An intergrowth
21
in nanoscale causes order/disorder phenomena which also hinder a clear description
of the structure. These rare minerals are often too small in size for single crystal
diffraction which is needed to obtain a detailed solution of the structure. For this
reason we wanted to find a way to use the simple and fast method of X-ray powder
diffraction to describe the structural characteristics of crystalline material obtained by
synthesis without limitations by crystal size or time consuming Rietveld analysis of
the patterns. The method was proofed based on natural samples of xonotlite and will
be explained in chapter 2.
Coming back to tobermorite, here the composite layers of one calcium and two
silicate layers are bound together by an interlayer containing calcium ions and water
molecules. The grade of hydration affects the basal spacing of the structure in [001].
Based on that three members are known from the tobermorite group 0.9nm
tobermorite or riversideite, the 1.13nm tobermorite or tobermorite senso stricto (part
of this study) and the 1.4nm tobermorite also named as plombierite. Tobermorite
1.4nm transforms into the 1.13nm one by heating to 100°C, further heating up to
300°C leads to the 0.9nm tobermorite. by proceeding dehydration (Merlino et al.
2001). it is known from some 1.13 nm tobermorites to not shrink on dehydration and
are therefore called “anomalous (Mitsuda & Taylor 1978). The average structure was
described by Hamid but the real structure was solved by Merlino et al. (2001, 1999)
which is based on two polytypic modification of orthorhombic and monoclinic
symmetry leading to a disordered structure (O/D character). In AAC, 1.13nm
tobermorite is close to the composition Ca5Si6O16 (OH)2 *4H2O and occurs in
association with semi-crystalline CSH-phases CSH (I ) and CSH (II) as minor
components. In contrast to tobermorite these phases are highly disordered and
display a wide range of compositions.
22
Chapter 2: Crystal Chemistry of Xonotlite Ca6Si6O17(OH)2. Part I: Determination
of Polytypes using X-Ray Powder Diffraction
2.1. Introduction:
As already mentioned in chapter 1 1.13nm tobermorite is formed metastable under
the normally conditions during production of AAC. The thermodynamically stable
phase is xonotlite (fig 1.7). Xonotlite differs from 1.13nm tobermorite by its structure
and texture. While tobermorite forms lathlike crystals which influence the pressure
resistance of the product in a positive way due to the interlocking “house of card”
structure, xonotlite crystallizes more needle-like what extensively decreases the
pressure resistance. Despite this knowledge just a few data exist on the reaction
kinetics and mechanisms present during the formation of 1.13nm tobermorite in AAC.
There is a strong need of collecting kinetic data to understand and control the
processes during the hydrothermal curing of AAC, which are the major aim of this
work and will be discussed in detail in chapter 4.
The determination of reaction and growth kinetics of CSH-phases and the calculation
of thermodynamic equilibrium demands material well characterized by its crystal
chemistry. Single crystal diffraction would be the most exact method to determine the
structure of CSH-phases but is often not applicable due to the lack of single crystals
of sufficient size. Beyond that one has to deal with short range order effects due to
the order disorder behaviour of many CSH-phases and the presence of one ore more
polytypes intergrown in one crystal. Applying the Rietveld method on data obtained
by synchrotron or neutron diffraction is often time consuming and reaches the limits if
more then two polytypes occur in the same crystal. Therefore another aim of the
present work was to find a fast and easy method to distinguish between different
23
polytypes by x-ray powder diffraction, which would be suitable for fine grained
crystals as well. In the presented paper this method was tested on different natural
polytypes. Based on the work of Hejny and Armbruster deriving structural data for the
different polytypes present in xonotlite by single crystal x-ray diffraction, the
theoretical powder diffraction patterns were calculated. The polytypes of natural
xonotlites were determined by matching the measured patterns with the model
patterns Characteristic peaks of lower intensities were chosen in the range of 10 and
35° 2Θ to distinguish between the different polytypes. Ten natural xonotlite samples
from seven different localities and different lithologies were investigated. After
determining the present polytypes a correlation to the conditions of formation was
done.
2.2. Structure and polytypism of xonotlite
Mamedov & Belov (1955, 1956) were the first to propose a structure model for
xonotlite which was later confirmed by Eberhard et al. (1981). The structure of
xonotlite consists of Ca-O-polyhedral layers in both sevenfold and octahedral
coordination and [Si6O17]-Dreier-Doppelketten. The CaO polyhedras are edgesharing to form infinite chains in b-direction and joined together resulting in layers
parallel to (001). Between these layers the [Si6O17]-Dreier-Doppelketten are located.
Each of these double chains consists of two wollastonite-like Dreier-Einfachketten
with two paired tetrahedra and one bridging tetrahedron (Fig.: 1.8). The structural
units were confirmed by extended X-ray absorption fine structure (EXAFS)
investigations of Ca (Lequeux et al., 1999) and by
Noma et al., 1998) on synthetic material, respectively.
29
Si NMR (Cong et al., 1996;
24
Due to the same length of [Si6O17]-Dreier-Doppelketten and two Ca-polyhedra there
exist two different ways of attachment of the double chains to the polyhedral layers
and hence various polytypes are possible (Gard, 1966; Kudoh & Takeuchi, 1979).
Based on the structure model of Mamedov & Belov (1955, 1956a) and the
confirmation of Eberhart et al (1981) six different polytypes (four ordered and two
one-dimensional disordered) were suggested for xonotlite. These Polytypes can be
seen as different stacking in [100]- and [001]-direction of a protoxonotlite-cell
introduced by Kudoh & Takeushi (1979). In [100]-direction a continuous shift of +b/4
or –b/4 or an alternating shift of +b/4 and –b/4 is possible. In [001]-direction the
protoxonotlite-cells are either in juxtaposed positions or shifted by b/2. The
combination of these different stacking modes leads to four ordered polytypes
Fig 2.1: polytypes of xonotlite
M2a2bc, M2a2b2c, Ma2bc and Ma2b2c as shown in Figure 2.1. The letter M
indicates the monoclinic symmetry of the protoxonotlite-cell and the three lower case
letters, with numerical values in front if necessary, indicate the periodicity of the three
directions in space according to the modified Gard-notation (Guinier et al., 1984). The
different cell parameters were determined by Hejny & Armbruster (2001)
For M2a2bc and M2a2b2c twinning is possible if one species displays intergrowth of
domains with continuous shift of +b/4 in a-direction and continuous shift of –b/4.
Streaks parallel to a* observed in single crystal patterns by Gard (1966) were
assigned to the two known disordered polytypes P∞21 and A∞22 ( Corresponding to
Mad2bc and Mad2b2c in modified Gard-notation). Hejny & Armbruster (2001)
25
extended the group of possible polytypes by Ma2bcd and M2a2bcd, which have onedimensional disorder in c-direction as indicated by streaks observed parallel to c*
(Chisholm, 1980; Eberhard et al.,1981). Short spikes recorded perpendicular to the c*
streaks have been interpreted in terms of two-dimensional disorder (DornbergerSchiff, 1964), and they are termed with the corresponding symbol Mad2bcd (Hejny &
Armbruster, 2001).
2.3. Assigned polytypes
X-ray powder diffraction (XRPD) data were determined with a STOE STADI Pdiffractometer using a SIEMENS KRISTALLOFLEX 710/710H generator operating at
the following conditions: 40 kV, beam current 30mA, curved Germanium
monochromator, step scan in the 2Θ region 10-60° with 0,01 2Θ steps and Cu-Karadiation (l=1.5406 Å). Most of the xonotlites show intergrowth of two or three
different polytypes. As the characteristic peaks are of very low intensity sometimes
not all are detectable in the investigated patterns. Due to occurring texture effects
some peaks have highly increased and others decreased intensities. Some of the
characteristic peaks can coincide with those of other polytypes (if present in the same
sample). For this reason different peaks had to be used for polytype assignment in
the different samples. The reproducibility of the obtained results was verified by
means of random sampling.
In all samples the M2a2b2c polytype was found but in Xon2, Xon8 and Xon9 with
only minor amounts. This predominance is an effect of specific growth conditions
availing the development of one polytype. Hejny and Armbruster (2001) explain this
preferred development by the more balanced and therefore favourable distribution of
OH-groups at the free apices of each Ca-octahedron in the structure of this polytype.
26
The Ma2b2c-polytype could be detected in two samples found in Russian rodingites
of Bazhenovskoe (Xon1) and Chukotka (Xon3) and in two samples found in Mn-ore
deposits from the Wessels Mine located in the Kalahari Manganese Field of South
Africa (Xon8) and from Franklin, New Jersey (Xon9), respectively.
Ma2bc is developed in the xonotlite from Mäntijärvi (Finland) with an exceptionally
high amount and in the well crystalline sample from N’Chwaning Mine (Xon4)
investigated by Hejny & Armbruster (2001). In addition this polytype does exist with
minor amounts in xonotlites of Bazhenovskoe (Xon1) and Chukotka (Xon3).
The occurrence of M2a2b2c and Ma2bc in Xon4 is in good agreement with the
investigations of Hejny & Armbruster (2001) on xonotlites from the same locality.
They also reported the presence of Ma2b2c polytype which could not be confirmed in
this study. This can be explained by different intergrown polytypes even in samples
from the same locality due to small scale fluctuations in the physico-chemical
conditions during formation.
The M2a2bc-polytype was first detected in natural xonotlite from Chukotka (Russia)
by Garbev (2004) using Rietveld-modelling of diffraction data obtained by syncrotron
radiation. In this study the M2a2bc polytype could clearly be detected in xonotlites
from Mäntijärvi (Xon2) and Chukotka (Xon3). Esteban et al. (2003) made their
assignment in Xonotlites from Carratraca (Spain) by use of X-ray powder diffraction,
too. Due to the missing description of the occurrence of M2a2bc polytype in natural
xonotlites in literature, this polytype was not taken into account by Esteban et
al.(2003). This is in contrast to the results of this study where the development of
M2a2bc polytype in xonotlite from Carratraca (Xon5) could be confirmed by the
presence of three characteristic reflections.
The above mentioned lower intensity of (h0l)-reflections in all patterns can be
explained by a preferred orientation of the acicular crystals along their elongated b-
27
axes during preparation. The observed phenomenon of inverse intensity-ratio in (0kl)
and (hkl)-reflections in M2a2bc-polytype may be caused by preferred orientation of
more disk-shaped crystals.
2.4. Correlation of chemical composition and assigned polytypes
Quantitative chemical data for xonotlites were obtained by electron microprobe
analysis (EMPA) using a CAMECA SX100 operated at 15 keV acceleration voltage
and 20 nA beam current. Synthetic wollastonite (Ca,Si), periclase (Mg), corundum
(Al), hematite (Fe), escolaite (Cr), natural ilmenite (Mn,Ti), albite (Na) and osumilite
(K) were used as standards and matrix correction was performed by the PAP
procedure (Pouchou & Pichoir, 1984). The reproducibility of standard analyses was
<1% for each element routinely analysed. For a detailed description of the chemical
composition of all investigated xonotlite samples the reader is referred to the
appended article (appendix B).
Due to the different lithologies of the localities the investigated samples were divided
into three different groups. Xonotlites formed in kimberlites (1), xonotlites in rodingites
(2) and xonotlites formed by metasomatic processes close to Mn- and Mn-Zn- oredeposits (3), respectively. Group 1 is only represented by the xonotlite from Mäntijärvi
(Finland). This sample exhibits a low SiO2- and CaO-content and a slight
enhancement of Na2O up to 0.09 wt%.
Group 2-xonotlites were formed in rodingites (Xon1, Xon3, Xon5, Xon6) and are
characterized by the highest CaO-amount of 47.07 up to 47.59 wt%. Analyses of
Carratraca-xonotlites (Xon5) are in good agreement to those of acicular crystals
replacing hydrogrossular of the same locality published by Esteban et al. (2003)
28
Samples belonging to group 3 are formed by hydrothermal alteration (250-400ºC) of
the primary sedimentary and low-grade metamorphic Mn-ores (Kalahari Manganese
Field ; Xon4, Xon8) or high-grade metamorphic Mn-Zn-ores (Franklin, New Jersey;
Xon9) and showing slightly higher Mn-content. This could be explained by a preferred
integration of Mn on the Ca-positions in Xonotlite-structure, likewise indicated by a
lower Ca-content. A substitution of Al for Si on tetrahedral-sites, indicated by the
higher amount of Al2O3 could be detected noticeably only in the sample from
Carratraca and one from Wessels mine. De Bruiyn et al. (1999) described a higher
SiO2- and CaO-content in xonotlites of N’Chwaning Mine in comparison to those of
Wessels Mine, which could be confirmed in this investigation. In addition De Bruiyn et
al. (1999) detected slightly higher FeO-contents which could not be verified in the
samples of this study. In all investigated xonotlites Al, Na and Mn are the only
elements which are enriched in remarkable amounts. A enhancement of Mg known
from synthetic xonotlites (Quian et al., 1997) could not be detected in the investigated
natural xonotlites
2.5. Conclusion:
A clear coherence of different lithology of the habitat and the developed polytype
could not be confirmed. Xon2 from kimberlites (Mäntijärvi, Finland) is in a special
position referring to this question. It displays a very high amount of Ma2bc polytype
compared to all other investigated samples. This can be linked to the special growth
conditions in kimberlites.
The results of this study clearly demonstrate that X-ray powder diffraction is a useful
and fast method to distinguish the different polytypes developed in xonotlite. It is the
preferable option if crystallite size is too small for X-ray single crystal diffraction as it
is typical for natural and synthetic xonotlites
29
Chapter 3: HAND - An Hydrothermal Autoclave for Neutron Diffraction
3.1. Introduction:
As described in chapter 1 1.13nm tobermorite is formed during the hydrothermal
curing of AAC. To optimize the reaction conditions and control the formation of
phases the exact determination of thermodynamic data and the kinetics of the
reaction are needed. Most commonly hydrothermal reactions and their evolving
phases at a certain temperature and pressure are studied by performing experiments
using so called Parr-bombs. Initial materials and water are filled in the bomb and kept
under the desired temperature for a certain amount of time. Subsequently the bomb
is quenched with water and the products are investigated by the usual methods. This
method reaches the limit when it comes to the determination of reaction kinetics. The
scientist has to struggle with several problems. First of all, quenching effects can
influence the final phase relations in particular and there is no guarantee to freeze
the process exactly at one stage and to quench it without change. Furthermore a
huge number of experiments at different compositions and temperatures are needed
to obtain a sufficient amount of data-points which means an enormous expenditure of
time. Reliable kinetic data can only be obtained by performing in-situ experiments by
x-ray or neutron diffraction. For those experiments special reaction cells are needed
tailored to the particular requirements of the scientific problem and of course the
instrument, respectively. Therefore an autoclave was designed to perform neutron
diffraction experiments on the formation of 1.13nm tobermorite.
30
3.2. Design of HAND
The hydrothermal autoclave for
neutron diffraction (HAND) was
designed to be a simple and cheap
reaction cell fitting to the wellestablished ILL D20 (Walton and
O’Hare, 2000; Hansen et al., 2008)
station with its vanadium furnace.
For a detailed description of the
instrument the reader is referred to
the appended article (appendix C)
Changes
of
samples
and
apparatus must be possible fast
and easy. Therefore the apparatus
is mainly an upright steel tube
Fig.3.1: schematic drawing of HAND
closed at both ends. The steam necessary for the hydrothermal reaction is generated
inside this tube during heating, so no separate steam supply is needed. The material
chosen for the autoclave is cobalt-free stainless steal ( 4301, Linster, Aschau). The
thickness of the walls is a compromise between the demands of a stability at an
internal pressure of up to 40 bars and the aim to obtain a maximum penetration of the
neutron beam. The schematic diagram of the reaction cell is given in Figure 3.1.
HAND consist of three parts: bottom, sample support and cover (Fig.3.1.). The
bottom is fixed inside the vanadium furnace device below the neutron beam. It serves
as reservoir for D2O and contains the bushings for the internal thermocouples. Inside
the bottom the sample support is placed above the water reservoir. The cover is a
31
tube of 14 cm in length and 2.5cm in diameter which is closed at the upper end. It is
simply screwed upon the bottom and can easily be replaced. Its walls have a
narrowing down to 1 mm at the level of the neutron beam to maximize the intensity of
the neutron flow through the sample.
3.3. Applicability for neutron diffraction experiments
To proof the applicability of HAND the pure system CaO-SiO2-D2O was chosen to be
studied first. The fast and easy sample preparation allows it to easily add different
additives to the system. The bulk composition was set to a Ca/Si ratio of 0.5
projecting on the join tobermorite-quartz and a chosen D2O/solids ratio of 0.8
resembles the recipes of industrially
manufactured steam cured building
materials [Fehr & Zuern, 2000].
Until now HAND was used for
several beam times at ILL to perform
experiments
to
influence
temperature,
of
investigate
the
quartz
grain size and Al-content on the
formation of 1.13nm tobermorite.
The results for these experiments
are summarized and interpreted in
chapter
4
and
related
articles
(appendix D and E). The monitoring
of the inner and outer thermocouple
reveals
an
accuracy
in
sample
Fig.3.2: time resolved diffraction patterns of an
experiment at 190°C and 16µm quartz grain
size (modified after Zuern, Fehr & Hansen,
2002)
temperature of ± 0.5°C and the desired reaction temperature of 190° was reached
32
after 60 min of heating up. The expense of the initial solid phases quartz and
portlandite with reaction time and the formation of 1.13nm tobermorite can be
observed by the decrease of their Bragg-peaks in the time-resolved neutron
diffraction pattern as demonstrated in Figure 3.2. within the range of 40° to 55° 2-Θ.
Sequential
individual
fitting
of
Bragg-peaks
multiple,
of
every
powder pattern were performed by a
procedure programmed to perform
this task from inside the ‘Large Array
Manipulation
Program’
(LAMP,
http://wwwold.ill.fr/data_treat/lamp/la
Fig.3.3: different onset of hk0 and 00l reflections of
1.13nm tobermorit during crystallisation
mp.html), the data-visualization and
treatment system used at ILL. The
main diffraction peak of iron (mantle of HAND) did not interfere with any peaks of the
phases of the sample and was used to calibrate the intensities of the phases of
interest. After 200 min. portlandite was dissolved completely, but crystallization of
1.13nm tobermorite did not start until 331 min. at 190°C (Fig.3.2.). The amount of
quartz did not remain constant after the consumption of complete portlandite,
indicating a reaction of quartz and initially formed semi-crystalline Ca-rich C-S-H(I).
The first detectable reflections of tobermorite were those of (hk0) planes, (00l)
reflections follow with a time lag of about 60 minutes (see fig. 3.3).
3.4. Conclusion
The high flux instrument D20 enables a time resolution of one minute for recording
one diffraction pattern with a good peak/background ratio. Each single diffractogram
allows an exact determination of the amount of phases and the decrease or increase
33
of phases as a function of time. The low scattering of the data on the amount of
phases involved indicates, that a detailed kinetic modeling (e.g. using the model of
Chan et al. (1978) or an Avrami-equation (Shaw et al., 2000) is possible on data
obtained by HAND experiments. The steel used for HAND has the advantage to
behave chemically inert and derived kinetic data correspond to the pure system SiO2CaO-D2O. Furthermore, steel is a cheap material and easy to handle in contradiction
to gold-coated Ti-Zr alloys used by Walton et al (1999).
The detailed compilation of information obtained by HAND-experiments leads to a
better insight in the reaction kinetics and mechanisms of CSH-formation. Its
applicability has been proofed in a variety of experiments studying the influence of
different parameters on the reaction kinetics of 1.13nm tobermorite as will be shown
in the next chapter. Beyond that, this autoclave offers a multitude of other possible
applications in geo.- and material sciences. The mature design of HAND allows an
easy adaptation on powder diffraction devices of other neutron sources assumed that
they can provide a sufficient neutron flux.
34
Chapter 4: reaction and growth kinetics of 1.13nm tobermorite crystallizing in
AAC
4.1. Introduction:
1.13nm tobermorite (Ca5Si6O16 (OH)2 *4H2O) is known to be formed during the
hydrothermal hardening of aerated autoclaved concrete (AAC), a widely-used
building material for light weight constructions. In consequence of the rapid increase
in applications of such materials during the last 10 years a strong need of more
detailed scientific research arose simultaneously. Fundamental knowledge on the
nature of CSH-phases had been given by Taylor (1964) with his studies on portland
cement phases but there is still a demand of further investigations. The existence of
various poorly ordered and metastable phases in the CSH-system hinder
experimental work thus the thermodynamics, kinetics and structural features of
1.13nm tobermorite and its neighbours are still poorly understood. The knowledge of
these properties is of essential importance as the mechanical properties of the
mentioned building materials are strongly dependent from the type, amount and
texture of the evolving CSH-phases. In AAC, 1.13nm tobermorite is close to the
composition Ca5Si6O16 (OH)2 *4H2O and occurs in association with semi-crystalline
CSH-phases CSH (I ) and CSH (II) as minor components. In contrast to tobermorite
these phases are highly disordered and display a wide range of compositions. They
are classified by their Ca:Si ratio: CSH (I) with a Ca:Si ratio <1.5 and CSH (II) with a
Ca:Si ratio > 1.5 according to Taylor (1950,1968). There has been a lot of work in this
field aimed at understanding the formation mechanisms and growth kinetics of CSHphases (Chan & Mitsuda., 1978; Klimesch & Ray,2002). But little quantitative data
exist on the kinetics of 1.13 nm tobermorite formation. In addition there is no
accordance on the nature of the reaction mechanism because some studies
35
proposed being solution controlled and others being diffusion controlled as pointed
out in detail by Klimesch et al. (1996). The reaction mechanism and kinetics of the
formation of 1.13 nm tobermorite in the pure cement-free system CaO-SiO2-H2O from
lime, silica and water under hydrothermal conditions were determined by quenching
experiments at 180°-190°C/Psat (Taylor,1968; Chan et al. ,1978; Zürn & Fehr, 2000)
and by an in-situ Neutron diffraction experiment (Fehr et al., 2002) as well. As
mentioned in chapter 3, quenching experiments reveal the disadvantage of missing
data for the early evolution of phases in time and have prevented a quantitative
kinetic description so far.
The major aim of this study was to determine the influence of reaction temperature
and quartz grain size on the formation of 1.13nm tobermorite in terms of reaction
mechanism and reaction rate. Therefore the reaction mechanism was determined
from in situ- neutron diffraction experiments and reaction constants were calculated
for the pure system CaO-SiO2-D2O at different temperatures (170,190,210°C, Psat)
and the employment of two different grain sizes of the quartz component with 16 and
8 µm, respectively.
The time-resolved neutron diffraction pattern were taken within the range of 8° to
153.6° 2-θ at λ = 2.4 Å to allow the analysis of d-spacing up to 11.3 Å, where the
basal (002) reflection of the evolving 1.13nm tobermorite is expected. The
mechanisms of the 1.13nm tobermorite forming reaction can be evaluated on the
basis of the reaction conversion of quartz according to Chan et al (1978) assuming
that there are no seeds in the reactants and the growth rate is low:
1 − (1 − α ) = kt 1 n
1
3
(1)
where α gives the fractional reaction conversion of quartz, k the reaction constant
and t the reaction time. According to equation (1) the factor n reveals information on
36
the reaction mechanism. If n=1 the reaction is solution controlled (phase boundary
model), if n=2 the reaction is diffusion controlled (Jander equation),(Hancock &
Sharp, 1972). Values for α were calculated from the decreasing integral intensity of
the (101)-Bragg reflection of quartz. Rate constants for the overall reaction progress
were calculated using equation (1) assuming slopes of 1 (n = 1) and 0.5 (n = 2) for a
solution and diffusion controlled reaction mechanism, respectively. Based on the
calculated rate constants for the three temperatures a first attempt to determine
activation energies was done. Therefore activation energies EA can be determined
according to the Arrhenius equation (2) when data were plotted according to equation
as follows
k = A⋅e
EA
− RT
(2)
where k is the rate constant, EA the activation energy, T the temperature in Kelvin, R
the gas constant and A the pre-exponential factor.
4.2. Results
The results for the neutron diffraction experiments are described and discussed in
detail in the related articles (see appendix D and E). At this point, only the main
findings will be mentioned. After determining the reaction conversion of quartz for all
experiments and plotting the results terms of equation (1) as explained in the
introduction, a change in slope can be seen. This could now lead to the conclusion
that the chosen kinetic model of Chan et. al. (1978) is not valid for the investigated
reaction but if one survey the single segments of each curve they are either
described with a slope of 1 or 0.5 referring to an exponent of n= 1 or n= 2 in equation
(1), respectively. Interpreting this in terms of the reaction mechanism it implies
changing reaction mechanisms with the reaction progress. All experiments show a
37
change from a solution controlled mechanism to a diffusion controlled mechanism. In
some of them a third segment could be assigned with a further solution controlled
reaction mechanism. 1.13nm tobermorite is not formed directly, it is just found after
the
complete
expense
of
portlandite.
By
determining
the
Ca/Si ratio of the
evolving
phases
over the reaction
progress one can
see
that
the
phases formed first
Fig.4.1:electron optical picture of a quartzgrain surrounded by a rim
of freshly formed 1.13nm tobermorite and semicrystaline
CSH-phases (Zürn,1997)
are richer in Ca
then expected for
1.13nm tobermorite. The Ca/Si ratio first reaches a maximum of 1.4 and then
converges to the theoretical value of 1.13nm tobermorite of 0.83
This was first
detected by our workgroup from neutron diffraction experiments in 2002 (Fehr et al.,
2002). The initial step of the reaction is controlled by the solution of quartz and its
reaction with portlandite, leading to the formation of a layer of semi crystalline CSHphases surrounding the quartz grains as shown in Figure 4.1 for AAC steam cured at
190°C/Psat(Zürn,1997). The second part of the reaction is controlled by the diffusion
of SiO2 through this layer of CSH-phases, portlandite is expensed completely and the
Ca/Si ratio decreases. 1.13nm tobermorite is then formed by the reaction of quartz
with the previously formed CSH-phases.
38
Comparing the transition times, the point of portlandite expense and the first
occurrence of 1.13nm tobermorite-reflections the strong influence of temperature and
grain size of quartz becomes apparent (Fig.4.2).
4.3. Influence of temperature and grain size of quartz
On the first glance, the use of finer quartz generally accelerates the reaction,
portlandite is expensed earlier and also 1.13 tobermorite crystallizes faster. This is
expected as a change in grain size from 16 to 8µm increases the specific surface
Fig. 4.2: changing transition time with reaction temperature and points of expense of
portlandite and occurrence of 1.13nm tobermorite determined from experiments with
16µm quartz (a and b) and 8µm quartz (c)
about 44% resulting in a higher reactive area. But an increase in specific surface do
39
not only accelerate the reaction, it also has a strong influence on the present reaction
mechanisms as well, demonstrated by the missing second diffusion controlled
segment in the experiments at 170 and 190°C/Psat applying the 8µm quartz.
Applying 16µm quartz all experiments show a second change back to a solution
controlled segment. With increasing reaction temperature the length of the diffusion
controlled segment increases but the time for portlandite out and tobermorite in
decreases. Initial semicrystalline CSH starts to react earlier with quartz to form 1.13
tobermorite indicated by the longer persistence of a diffusion controlled mechanism.
Experiments with 8µm quartz show a differing behaviour. For 170 and 190°C the
second change to a solution controlled mechanism is missing or not detectable. Only
at 210°C this change is present. But again the transition time from solution to
diffusion control decreases with increasing reaction temperature. The point of
portlandite expense slightly decreases with rising temperature. For 170 and 190° this
is also valid for the point of tobermorite occurrence but at 210°C the time increases
again, due to the present second change in reaction mechanism. This behaviour was
assigned to the clear metastable formation of 1.13nm tobermorite under these
conditions (Zürn & Fehr, 2000).
4.4 Quenching experiments
Based on this study kinetic data obtained from quenching experiments were
recalculated in terms of equation (1) and can be now interpreted by applying two
different slopes (n=1 and n=2) due to two distinct mechanisms during the reaction
progress. Literature values (Klimesh & Ray, 2002; Zürn & Fehr 1997) were used to
determine the transition temperatures ( see appendix D, Fig.3) and interpreted
concerning to the grain size dependency ( see appendix D, Fig.4) The results are in
40
good agreement with findings of this study, showing an increase of reaction time with
increasing grain size of quartz
4.5. Calculated rate constants and activation energies
The influence of both grain size and reaction temperature can as well be seen in the
calculated rate constants and activation energies
The change of the rate constants with increasing specific surface for the three
investigated reaction
temperatures clearly
shows an increase
at
a
given
temperature.
strongest
The
increase
occurs at 210°C, the
rate
constant
changes
Fig.4.3: changing rate constant with increasing specific surface
(decreasing grain size of quartz)
2.795*10-4
from
s-1
to
3.214*10-4 s-1 for the
first solution controlled part and from 2.784*10-4 s-1 to 2.794*10-4 s-1 for the diffusion
controlled part with a decreasing grain size of quartz. Likewise the rate constant
increases with increasing temperature what can be extracted from Figure 4.3. Based
on the calculated rate constants for the three temperatures a first attempt to
determine activation energies was done. The determined activation energies for the
experiments containing the 16µm quartz are with 0.2 kJ/mol for the solution
controlled segment and 1.8 kJ/mol for the diffusion controlled segment. They reveal
values which were remarkably lower than those determined from the 8µm quartz
41
experiments of 6.2 kJ/mol and 7.4 kJ/mol, respectively. These values are
considerably below 26 kJ/mol, 37 kJ/mol and 33kJ/mol determined for the system
CaO-Al2O3-SiO2-H2O assuming an isokinetic behaviour (Shaw et al., 2000).
4.6. Conclusion
It could be shown that in-situ neutron diffraction is a very suitable method to
investigate the kinetics of the 1.13nm tobermorite formation. The non isokinetic
behaviour of the reaction could be evidenced by combining the high intensity of the
D20 powder diffractometer at ILL together with an improved hydrothermal autoclave
(manuscript 2) allowing constant reaction conditions and a fast and easy sample
exchange.
Furthermore exact times for the transition and the consumption of
portlandite and the occurrence of 1.13nm tobermorite could be determined. Based on
the data obtained by applying the kinetic model of Chan et al. (1978) on the values
for the overall reaction progress rate constants could be determined for the first time.
Likewise Shaw et al. calculated rate constants for the Tobermorite forming reaction
but did not interpret their date in terms of the present reaction mechanism. By
conducting experiments at three different temperatures, the temperature dependence
and hence activation energies could be determined. The results of this study yield
detailed kinetic data on the 1.13nm tobermorite formation, which were just insufficient
investigated in the past. These data give a better understanding of the processes
present during the production of AAC and could help to optimize production
conditions and recipes resulting in shorter production times and an optimal exploit of
the available resources.
42
5.1.Outlook
The presented work successfully studied the reaction kinetics of the 1.13nm
Tobermorite formation in the system SiO2-CaO-H2O. But answering one question
leads to several new ones, that is how science works. Future work should therefore
focus on the influence of additives like Al, SO4 and K to the reaction kinetics. They
can enter the production process by the natural sand minerals like feldspar and mica.
High temperature long term experiments could give insights in the decomposition of
1.13nm tobermorite to xonotlite and quartz.
The possibility to use porous samples for neutron diffraction experiments would
improve the comparability to the production of AAC but is difficult to implement so far.
The pattern matching method to characterise the material in terms of polytism like
done for natural xonotlite in the presented work should be assigned to neutron
diffraction experiments as well. At the same time more analytical methods should be
used for the characterisation. First attempts with FTIR and Raman yield promising
result but need further work.
43
5.2.References:
ALEXANDERSON, A (1979): Relations between Structure and Mechanical Properties of
Autoclaved Aerated Concrete. Cem. Concr. Res. 9(4),. 507-514
AYLSWORTH, J.W. AND DYER, F. A. (1914): US Patent 1.087.098
E. BONACCORSI,, S. MERLINO, A. R. KAMPF (2005): The crystal structure of Tobermorite 14Ǻ, a CSHPhase. – J. of Am. Ceram. Soc., 88 (3),505-512
DE BRUIYN, H.; SCHOCH, A.E.; VAN DER WESTHUIZEN, W.A & BEUKES, G.J.
(1999): The chemical
composition of xonotlite and associated inesite from the Nchwaning and Wessels mines, Kalahari
manganese field, South Africa. N.Jb.Miner.Mh., 5: 212-222
CHAN, F., MITSUDA, T. (1978): Formation of 11 Å tobermorite from mixtures of lime and colloidal silica
with quartz, Cem.Concr.Res. 8 135-138.
CHAN, C.F.,SAKIYAMA, M.; MITSUDA, T. (1978), Kinetics of CaO-Quartz-H2O reaction at 120°C in
suspension. - Cem. Con. Res. 8, 1-6.
CHISHOLM, J.E. (1980): Polytism in xonotlite, Ca6Si6O17(OH)2. - In Electron Microscopy and Analysis,.
Inst. Phys. Conf. Ser., 52, 109-112
CONG, X. & KIRKPATRICK, R.J. (1996): 29Si and 17O NMR Investigation of the Structure of Some
Crystalline Calcium Silicate Hydrates. - Advn. Cem. Bas. Mat., 3, 133-143
DORNBERGER-SCHIFF K. (1964):Grundzüge einer Theorie von OD-Strukturen aus Schichten. - Abhandl.
d. Deutsch. Akad. d. Wiss. zu Berlin. Klasse f. Chemie, Geol. u. Biol., 3, 1–107
DUBRAL, W. (1992): On production and application of AAC worldwide. – Advances in Autoclaved
Aerated Concrete, Wittmann (ed.), 3-9
EBERHART; E.; HAMID, S. & RÖTTGER, B. (1981): Strukturverfeinerung und Polytypie von Xonotlite, - Z.
Krist., 154, 271-272
ERIKSSON, J. A. (1924): DRP 447 194 of 12.6.1924
ESTEBAN, J.J.; CUEVAS, J. & TUBIA, J.M. (2003): Xonotlite in rodingite assemblages from the Ronda
peridotite Betic Cordilleras, Southern Spain. –Can. Mineral., 41: 161-170
FEHR, K-T.; HUBER, M.; ZUERN, S.G.; HANSEN, T. Proc. 5th ICSTR (2002) 37-40
FEHR, K.T. AND ZUERN, S.G. (1997): Phase relations of 1.13 nm tobermorite, xonotlite, truscottite and
gyrolite under hydrothermal conditions, - in Proc. 5th Int. Symp. Hydrotherm. Reactions, 225-227.
GABROVSEK, R., KURBUS B., MUELLER D. AND W. WIEKER (1993) “Tobermorite formation in the system
CaO, C3S-SiO2-Al2O3-NaOH-H2O under hydrothermal conditions,” Cem. Con. Res. 23, 321.
GARBEV, K. (2004): Struktur, Eigenschaften und quantitative Rietveldanalyse von hydrothermal
kristallisierten Calciumsilikathydraten (CSH-phasen),Thesis, University of Karlsruhe
GUINIER , A.; BOKIJ, G.B.; BOLL-DORNBERGER, K.; COWLEY, J.M.; DUROVIC, S.; JAGODSINSKI, H.; KRISHNA,
P.; DEWOLFF, P.M.; ZVYAGIN, B.B.; COX, D.E.; GOODMAN, P.; HAHN, TH.; KUCHITSU, K. & ABRAHAMS,
S.C. (1984): Nomenclature of Polytype Structures - Report of the International Union of
Crystallography Ad.Hoc Committee on the Nomenclature of Disordered, Modulated and Polytype
Structures. - Acta Cryst, A40: 399-404
HAMID S.A. (1981): The crystal structure of the 11Å natural tobermorite Ca2.25[Si3O7.5(OH)1.5]. 1H2O. Z. Kristallogr. 154 189–198.
44
HANCOCK AND J.H. SHARP (1972): Method of Comparing Solid-State Kinetic Data and Its Application to
the Decomposition of Kaolinite, Brucite, and BaCO3, J. Amer.Ceram.Soc. 55, 74-77.
HANSEN, T.C.; HENRY, P.F.,FISCHER, H.E. (2008): The D20 instrument at the ILL: a versatile highintensity two-axis neutron diffractometer. - Measurement Science & Technology.
HEJNY,C. & ARMBRUSTER, T. (2001): Polytypism in Xonotlite Ca6Si6O17(OH)2. - Z.Kristallogr., 216: 396408
HOMAN, M. (2008): Porenbeton Handbuch- planen und bauen mit System,Bundesverband Porenbeton,
6. Auflage, Bauerverlag, Gütersloh
KALOUSEK , G.L.; NELSON, E.B (1978) : Hydrothermal reactions of dicalcium silicate and silica. - Cem.
Concr. Res. 8(3), 283-289
KLIMESCH D. S.; RAY A.; SLOANE B. (1996): Autoclaved cement-quartz pastes: The effects on chemical
and physical properties when using ground quartz with different surface areas .1. Quartz of wide
particle size distribution. - Cem.Concr.Res. 26, 1399-1408.
KLIMESCH D. S., RAY A. (2002): Effects of silica reactivity on the nature and formation of Al-1.1nm
tobermorite. - J.Therm.Anal.Calorim., 995-1003
KUDOH, Y. & TAKEUSHI, Y. (1979): Polytypism of xonotlite: (I) Structure of an A1 polytype. - Min.Journal,
9, 349-373
LEQUEUX, N., MORAU, A.; PHILIPPOT, S. & BOCH, P. (1999): Extended X-ray Absorbtion Fine Structure
Investigation of Calcim Silicate Hydrates. – J. Am. Ceram. Soc., 82 (5), 1299-1306
MAMEDOV,KH. & BELOV, N.V. (1955): Crystal structure of xonotlite. – Dokl.Akad.Nauk SSSR, 104: 615618
MAMEDOV, KH. & BELOV, N.V. (1956): Crystal structure of the mineral group wollastonite. 1 structure of
xonotlite. - Zapiskii Vsesoyuz. Miner. Obshchest, 85: 13-38
MERLINO S., BONACCORSI, E. AND. ARMBRUSTER. T. (1999): Tobermorites: Their real structure and
order-disorder (OD) character, Amer.Miner. 84, 1613-1621
MERLINO S., BONACCORSI E. AND ARMBRUSTER TH. (2001): The real structure of tobermorite 11
angstrom: normal and anomalous forms, OD character and polytypic modifications. - Eur. J.Min.,
13, 577-590
MICHAELIS, W. (1881): DRP 14195 vom 2.6.
NOMA, H. ADACHI, Y.; MITSUDA, Y. & YOKOYAMA, T. (1998):
Xonotlite. - Chemistry Letters, 219-220
29
Si and 1H NMR of Natural and Synthetic
POUCHOU, J.L. & PICHOIR, F. (1984): A new model fro quantitative X-ray microanalysis Part I:
application to the analysis of homogenious samples. -La Recherche Aérospatiale, 3: 13-38
QUIAN, G.; XU, G.; LI, H. & LI, A. (1997): Mg-Xonolite and its coexisting phases. –Cem. Concr. Res., 27
(3), 315-320
SHAW, S., CLARK S. M. (2000): Hydrothermal formation of the calcium silicate hydrates, tobermorite
(Ca5Si6O16(OH)2* 4H2O) and xonotlite (Ca6Si6O17(OH)2): an in situ synchrotron study. - Chemical
Geology 167(1-2) 129-140.
SHAW, S. HENDERSON, C.M.B. & KOMANSCHEK, B.U. (2000): Dehydration/recrystallization mechanism ,
energetics, and kinetics of hydrated calcium silicate minerals: an in situ TGA/DSC and synchrotron
radiation SAXS/WAXS study. - Chemical Geology, 167: 141-159
TAYLOR, H.F.W (1950): Hydrated calcium silicates. Part I. Compound formation at ordinary
temperatures. - J. Chem. Soc. 30, 3682.
45
TAYLOR, H.F.W Proc. Vth Int. Symp. Chem. Cem. Vol II (1968) 1-26
TAYLOR, H.F.W. (1964): The calcium silicate hydrates. in ‘The Chemistry of Cements' H.F.W. Taylor,
Ed., Academic Press, London, 167–232.
WALTON, R.I. AND O’HARE, D. (2000), Watching solids crystallise using in situ powder diffraction. Chem. Commun. 23, 2283 – 2291.
WALTON, R.I., FRANCIS, R.J., HALASYAMANI, P.S., O’HARE, D., SMITH, R.I., DONE, R. AND HUMPHREYS, R.J.
(1999): Novel apparatus for the in situ study of hydrothermal crystallizations using time-resolved
neutron diffraction, - Rev. Sci. Instr. 70, 3391.
ZERNIKOW (1877): DRP 502 of 2.7.1877, Odenberg
ZUERN, S.G. AND FEHR, K.T. (2000): Phase relations and thermodynamic properties of 1.13 nm
tobermorite and xonotlite. in Proc. 6th Int. Symp. Hydrotherm. Reactions, 286-289.
ZUERN, S.G., FEHR, K.T. AND HANSEN, T. (2002): Determination of the reaction kinetics and reaction
mechanisms of 1.13nm tobermorite by means of in-situ neutron diffraction. in Proc Vth ICSTR,
East Brunswich, 37-40
ZUERN, S.G. (1997): Einfluss der Sandminerale auf die Bildung von Calciumsilikathydraten (CSHPhasen), das Gefüge und die mechanischen Eigenschaften von Porenbeton, Inauguraldissertation,
Logos-Verlag Berlin
46
5.3 Appendix
Appendix A
Report of the industrial internship at Xella Thermopierre, Bourgoin- Jallieu,
France (unpublished)
Praktikumsbericht
Indusrtiepraktikum 13.09.-06.10.2006
YTONG/ Xella Thermopierre
St Savin ( Bourgoin-Jallieu)
Herstellung von Porenbeton
Zur Produktion von Porenbeton wird eine
Mischung aus Branntkalk, Quarzsand und Zement
verwendet. Die Sandfraktion wird fein gemahlen
um die Kristallisation zu begünstigen. Im Werk von
St Savin sollen in Zukunft 2 Sande verwendet mit
unterschiedlichen Quarzgehalten. Der eine wird in
der Nähe des Werks aus einem See gefördert und
ist Quarzärmer der andere stammt aus Bedoin mit
SiO2 – Gehalten um die 93 %. Ein höherer
Quarzgehalt
wirkt
sich
positiv
auf
die
Druckfestigkeit aus.
Das beim Zuschneiden der Kuchen anfallende
Material wird als sogenanntes Rückgut dem
Produktionsprozess wieder zugeführt. Für ein Abb.1: Verwendung von Porenbeton
besseres Ansteifen der Masse wird zusätzlich
noch Anhydrit (CaSO4) beigemengt. Die Rohstoffe werden in den gewünschten
Anteilen vermischt und mit Wasser zu einer Mörtelmischung angemacht in der der
Kristallisationsprozess beginnt. Erst kurz vor dem Gießen wird fein gemahlenes
Aluminiumpulver (2 Korngrößen) zugegeben das mit der alkalischen Mörtelmischung
unter Bildung von Wasserstoff reagiert. Dadurch kommt es zur Porenbildung und
zum Aufschäumen der ansteifenden Masse.
Die Formen (4x1x0.8m) werden bis zu ca. 2/3 mit der fertigen Masse befüllt und
ruhen anschließend bei ca 37°C für 115 min in Kammern. Nach ca. 50 Minuten ist
der „Kuchen“ bis zum Rand der Formen aufgegangen. Nach dem Ansteifen werden
die Formen um 90° gedreht und der grünfeste Kuchen wird auf einen Wagen
umgelagert. Anschließend erfolgt das Zuschneiden der Blöcke auf das gewünschte
Maß mit Hilfe von Drähten. Am Ende des Produktionsprozesses steht die
47
Autoklavierung. Hier werden die Kuchen bei ca. 190°C und einem
Sättigungsdampfdruck von 10-12 bar für 8-12 Stunden hydrothermal gehärtet.
Das fertige Produkt, YTONG® (SIPOREX®), zeichnet sich bei relativ geringer Dichte
durch hohe Wärmedämmung (geringe Wärmeleitfähigkeit) und ausreichender
Drückfestigkeit aus und findet im Hausbau auch aufgrund der einfachen Verarbeitung
(geringes Eigengewicht und leicht zu schneiden) Anwendung.
Die Wärmeleitfähigkeit und die Druckfestigkeit des Materials sind von der Rohdicht,
dem Quarzgehalt, der Feinheit des Sandes und der Menge an Bindemittel abhängig
und werden im Werk täglich kontrolliert.
Kontrolle und Optimierung von Wärmeleitfähigkeit, Druckfestigkeit und
Korngröße der Sandfraktion
Zur Optimierung dieser beiden Größen
werden
Versuchsgießungen
mit
bestimmten Gehalten an Bindemittel,
Mischungsverhältnissen der beiden Sande
und definierter Rohdichte hergestellt. Pro
Gießung werden zwei Blöcke entnommen
und aus ihnen drei Prüfwürfel mit 10 cm
Kantenlänge herausgesägt. Die Richtung
des entweichenden Wasserstoffs, die
Nummer und der Tag der Gießung werden
auf den Blöcken vermerkt (Abb.2).
Anschließend werden die Würfel von 4
Seiten geschliffen um eine glatte
Oberfläche zu erhalten die für die
Messungen nötig ist. Die Würfel werden
dann im Trockenschrank bei ca. 80°C
getrocknet und vor den Messungen eine
Nacht lang im Exikator gelagert um die
Restfeuchte zu beseitigen.
Abb.2: Prüfwürfel mit Beschriftung, Flächen zur
Messung von λ (1) und A (2) und Angaben
über Abmessungen
.
Messung der Wärmeleitfähigkeit:
Die Wärmeleitfähigkeit λ gibt an welche Wärmemenge Q in der Zeit t und bei einem
Temperaturunterschied dT durch die Fläche A strömt. Die Prüfwürfel werden in der
Reihenfolge B (bas) M (millieu) H (haut) in einem Exikator gestapelt. Der Wert für λ
wird einmal zwischen den Würfeln B und M und einmal zwischen M und H auf der in
Abb.2 mit 1 gekennzeichneten Fläche (vert × horiz 1) mittels einer Sonde gemessen.
Bei der letzten Messung werden die Steine zusätzlich mit einem Gewicht beschwert.
Aus dem Unterschied zwischen der vorgegebenen Temperatur und der in den
48
Steinen gemessenen Temperatur berechnet das Gerät den wert für λ in W/min*K.
Die am in St Savin vorhandenen Messgerät ermittelten Werte werden anschließend
korrigiert anhand einer Eichkurve die aus Messungen an einem genormten Gerät in
Paris ermittelt wurde.
2.2
Prüfung der Druckfestigkeit:
Die Druckfestigkeit des Materials wird an der Fläche Nr 2 in Abb.2 gemessen. Es
wird dabei die auf den Prüfquerschnitt (horiz1 × horiz2 siehe Abb2) wirkende Kraft
ermittelt wen es zum Bruch kommt.
Die Druckfestigkeit des Materials steigt im Allgemeinen mit zunehmender Rohdichte
(abnehmende Porosität). Um Gießungen unterschiedlicher Rohdichte miteinander
vergleichen zu können muss noch auf diese normiert werden. Zur Ermittlung der
Rohdichte werden zusätzlich Gewicht und Kantenlänge des Prüfkörpers gemessen.
Der auf die Rohdichte normierte Wert wird als A-Zahl angegeben.
A-Zahl =
Rc
const. • ρ²
Rc : Druckfestigkeit in MPa/dm²
ρ
: Rohdichte, trocken in kg/dm³
const: 62500000
Korngrößenbestimmung mittels Laser – Granulometrie
Eine weitere Größe die Einfluss auf die Eigenschaften des Porenbetons hat ist die
Mahlfeinheit (Korngröße) der Sandfraktion. Hier gilt je feiner der Sand gemahlen ist
desto besser verläuft die Kristallisation und desto höher ist die Druckfestigkeit des
Produkts.
Es wurden zwei verschiedene Feinheiten des Sandes aus St Savin verwendet. Zum
einen eine Mischung mit fein gemahlenem Sand aus St Savin wie er bisher auch in
der normalen Produktion verwendet wurde. Zum anderen eine Mischung mit gröber
gemahlenem Sand aus St Savin.
Die Korngröße des Sandes wird mit der Methode der Laser-Granulometrie bestimmt.
Die Korngrößenbestimmung mittels Laser-Granulometrie ist eine Methode, die die
Projektionsfläche der Teilchen anhand der Beugung des Laserstrahls an den
Teilchenrändern ermittelt. Über die Fläche wird auf das Volumen der Teilchen
geschlossen. Bei der Berechnung der Volumenanteile einer bestimmten Korngröße
wird davon ausgegangen das die Teilchen annähernd kugelförmig sind.
Aufgaben und Ziele des Praktikums
Während meines Praktikums wurden mehrere Versuchsgießungen produziert die ich
dann hinsichtlich Wärmeleitfähigkeit und Druckfestigkeit untersuchen sollte. Dazu
habe ich, wie unter 2. beschrieben, Prüfwürfel aus den Blöcken gesägt, diese
anschließend geschliffen und für 5 Tage im Trockenschrank bei ca. 80°C gelagert.
Nach der Trocknung habe ich die Werte für die Wärmeleitfähigkeit und die
Druckfestigkeit an den dafür vorgesehenen Messgeräten (siehe Punkt 2) ermittelt.
49
Ziel dieser Untersuchungen war es neue Rezepturen mit Verwendung des SiO2
reichen Sandes zu testen und somit die Werte für Wärmeleitfähigkeit, Druckfestigkeit
und auch der Korngröße der Sandfraktion zu optimieren. Hierfür wurden die beiden
Sande in verschiedenen Mischungsverhältnissen verwendet. Bisher konnte nur eine
Mischung in einem extra Silo hergestellt werden, die dann in allen Gießungen des
Tages verwendet wurde. Die Installation eines neuen Programms ermöglicht es jetzt
in jeder einzelnen Gießung die Anteile der beiden Sande zu variieren. Dies ist
vielleicht für die normale Produktion weniger von Bedeutung, bietet aber für die
Versuchsgießungen einen großen Vorteil.
Am Ende des Praktikums soll anhand der in dieser Zeit ermittelten Daten und anhand
von Daten aus vorangegangenen Versuchen Optimierungskurven erstellt werden. In
diesen werden die ermittelten A-Zahlen und Werte für λ gegen das CaO/SiO2 –
Verhältnis und zusätzlich den prozentualen Anteil an Bindemittel aufgetragen. Die
Wärmeleitfähigkeit λ sollten idealer weise einen Wert 0.098 W/k*m nicht
überschreiten. Denn je geringer die Wärmeleitfähigkeit umso besser ist die
Wärmedämmung des Baumaterials. Für die A-Zahl gilt generell, je größer desto
besser, sie sollte am Optimum aber über 1450 liegen. Einfluss auf diese Größen hat
die Rohdichte des Materials, der Anteil an Calciumbinder, die Feinheit und der
Quarzanteil in der Sandfraktion.
Versuche
Es wurden generell 2 Versuchsansätze verwendet. Zum einen wurden Gießungen
produziert in denen, bei konstanten Mischverhältnis der beiden Sande der
prozentuale Anteil an Bindemittel (taux de liant) variiert. Im zweiten Ansatz wurden
die Mischverhältnisse variiert und der Anteil an Calciumbinder konstant gehalten. Die
verwendeten Sandmischungen lagen bei 20, 30, 40, 45, 50,60 und 70% des Sandes
aus Bedoin (quarzreich). Die Menge an Calciumbinder variierte zwischen 25 und
35% wobei ein Hauptaugenmerk auf den Bereich zwischen 28 und 30% gelegt
wurde. Das CaO/SiO2 Verhältnis wurde im Nachhinein aus den reellen
Mengenangaben der verschiedenen Bestandteile berechnet und variiert in beiden
Versuchsansätzen.
Die Mischverhältnisse der Sandfraktion sind durch die verschiedenen Farben der
Zeilen zu erkennen (siehe Legende)
Um das CaO/SiO2 Verhältnis zu berechnen wurde die chemischen Analysen der
Firma Brück vom 25.06. 2006 verwendet
Der Wert für λ und ρ ist der Durchschnitt aus den Werten der bei jeder Gießung
hergestellten 3 Prüfwürfeln (H M B).
Ergebnisse und Auswertung
Ohne Rücksicht auf eine Optimierung der Kenngrößen lässt sich zunächst einmal
sagen das der Wert für λ mit abnehmender Rohdichte sinkt: Bei sinkender Rohdichte
werden aber ebenfalls die werte für die Druckfestigkeit kleiner, was bis zu einem
gewissen Grad tolerierbar ist.
Betrachtet man für die einzelnen Sandmischungen die besten erreichten Werte für λ
und A so zeigt sich das unabhängig vom Mischungsverhältnis dieses Maximum sich
50
immer bei ca. 28 % Bindemittel befindet. Dies erscheint zunächst ein wenig seltsam
da bei Zunahme des Quarzanteils auch der Anteil an Bindemittel steigen müsste.
Erklären lässt sich dies meiner Ansicht nach durch das Verhältnis von CaO zu SiO2.
Dieses sinkt mit zunehmendem Anteil von Quarz bei konstanter Bindemittelmenge
(siehe Abb 3).
29% taux de liant
0.56
0.55
0.54
CaO/SiO2
0.53
0.52
0.51
0.50
0.49
0.48
0.47
20
25
30
35
40
45
50
55
% Sand Bedoin
Abb.3: Änderung des CaO/SiO2 Verhältnis mit steigendem
Anteil an Sand Bedoin bei konst taux de liant
Für die Maximalwerte heißt das, dass mit zunehmendem Anteil an Quarz im Sand
die besten Werte für λ und A mit einem kleineren CaO/SiO2 Verhältnis zu finden sind.
Generell liegen das Maximum für die Quarzreichen Mischungen bei wesentlich
höheren Werten als bei den Quarzarmen (Abb.4).
1550
0.55
1500
0.5
1450
0.45
1400
0.4
1350
0.35
1300
Ca/Si
A-Zahl
Änderung A-Zahl
A max
CA/Si
0.3
20
30
40
45
50
60
70
melange
Abb.4: Zunahme der max A-Zahl mit steigendem Quarzanteil
51
a)
b)
Melange %
Sable bedoin
Gießung
Bindemittel %
20
30
40
45
50
60
70
6
26
13
7
35
6
10
26,28,29
23-33
26-30
28,29
23-35
26,27,28
27-31
Melange
Melange %
Sable bedoin
20
30
40
45
50
60
70
A
Rc
l
r
1350
2.64
0.093
350
1442
2.9
0.094
354
1445
3.29
0.1
377
1456
3.09
0.098
364
1559
3.47
0.103
373
1481
2.98
0.097
355
1498
3.04
0.096
356
Fit
liant
20
30
40
c)
Maximal
45
50
60
70
A
liant
Ca/Si
A
liant
Ca/Si
A
liant
Ca/Si
A
liant
Ca/Si
A
liant
Ca/Si
A
liant
Ca/Si
A
liant
Ca/Si
1350
1333
26,4
26.2
0,52
0.52
1442
1381
27
28.1
0.51
0.53
1445
1405
27.9
27.9
0.51
0.51
1456
1437
28.9
28.6
0.51
0.5
1559
1416
26.9
27.7
0.46
0.47
1481
1471
27.1
27.6
0.45
0.45
1498
1498
27.1
27.1
0.42
0.42
Tabelle 1
a)
Anzahl der Versuchsgießungen mit den verschiedenen Sandmischungen und den abgedeckten
Bereich an Bindemittelanteil
b) Maximal ermittelte Werte für A (mit liant und Ca/Si) in den einzelnen Sandmischungen und werte
der Maxima aus den Näherungskurven
c) Druckfestigkeit (Rc), Wärmeleitfähigkeit (λ) und Rohdichte (r) der Versuche mit maximaler AZahl in den verschiedenen Sandmischungen
Für die Auswertung der Ergebnisse wurden die ermittelten A-Zahlen gegen das
CaO/SiO2 Verhältnis aufgetragen. Um die Optimierungskurve zu erhalten wurden
diese Daten mit einem Polynom 2. Ordnung angefittet.
Eine Auswertung hinsichtlich der Änderung von λ mit steigendem CaO/SiO2
Verhältnis ist aufgrund der stark schwankenden Rohdichte schwierig.
Mit eine hohen Rohdichte nimmt zwar die Druckfestigkeit zu es wird aber auch der
Wert für λ zu groß und umgekehrt.
Die Schwankungen in der Rohdichte sind auf unterschiedliches Aufgehen des
Kuchens zurückzuführen. Geht der Kuchen stark auf wird in der Produktion viel
Material weggeschnitten und der Wert für die Rohdichte wird geringer als der
gewünschte.
52
Mit zunehmendem Quarzanteil im Sand sinkt der Grenzwert für die Rohdichte bei
dem noch gute werte für λ erzielt werden.
Das bedeutet bei gleicher Rohdichte ist der Wert für λ besser bei niedrigeren
Quarzgehalten der Sandmischung.
Betrachtet man das Maximum der ermittelten Optimierungskurven so wird deutlich das die AZahl mit steigendem Quarzgehalt im Sand zunimmt.
Das CaO/SiO2 Verhältnis nimmt jedoch ab
Für 20, 45 und 60 % S-Bedoin wurden keine vollständigen Optimierungskurven
ermittelt da nur Werte für einen Ast der Kurve vorliegen und die A-zahlen erst
beginnen wieder kleiner zu werden.
Deswegen sind in der Auswertung die Maximal ermittelten Werte berücksichtigt
worden. Die maximal und aus den Optimierungskurven ermittelten Werte sind mit
den dazugehörigen CaO/SiO2 und Bindemittelanteil und zusätzlich ρ, λ und Rc für
die Maxima aus den Tabellen 1a) und b) zu entnehmen.
20% Bedoin-Sand
Mit einer Mischung mit 20 % S-Bedoin wurde die größte A-Zahl (1350) bei einem
Bindemittelanteil von 26.4% und einem CaO/SiO2 Verhältnis von 0.52 ermittelt. A
liegt somit noch deutlich unterhalb der geforderten 1450. Die Rohdichte dieser
Gießung beträgt 350 kg/dm³ und λ ist gleich 0.093, beide Werte liegen im optimalen
Bereich.
melange 20 %
1360
1340
1320
1300
28
1260
liant in %
A-Zahl
1280
1240
1220
1200
1180
A-Zahl
taux de liant
1160
0.520
0.525
0.530
0.535
26
0.540
0.545
CaO/SiO2
Abb.3: Optimierungskurve für A-Zahl bei 20% Bedoin-Sand
30% Bedoin-Sand
53
Bei 30%-iger Mischung liegt die aus der Optimierungskurve entnommene A-Zahl bei
1385 mit einem Bindemittelanteil von 27% und einem CaO/SiO2 von 0.52. Auch der
maximal ermittelte Wert liegt mit 1442 noch knapp unter dem Zielwert. Sowie r als
auch λ sind mit 354 und 0.094 als gut einzustufen.
melange 30 %
1500
1480
34
1460
1440
32
1420
1400
30
1360
1340
28
1320
1300
26
1280
1260
A-Zahl
taux de liant
1240
1220
1200
0.40
0.45
0.50
0.55
liant in %
A-Zahl
1380
24
22
0.65
0.60
CaO/SiO2
Abb.4: Optimierungskurve für A-Zahl bei 30% Bedoin-Sand
40% Bedoin-Sand
Mit 40% Bedoin-Sand beträgt der Wert aus der Kurve bei 1405 bei einem CaO/SiO2
von 0.507 und einem Bindemittelanteil von. Der maximal ermittelte Wert liegt bei
1445 mit aufgrund der hohen ρ von 377einem zu großen λ von 0.100, dafür aber mit
3.29 bei einem sehr guten Wert für Rc. Die Rohdichte und somit auch λ sind besser
bei geringeren Bindemittelanteilen.
1460
melange 40%
1440
30
1420
1380
1360
28
1340
1320
1300
1280
A-Zahl
1260
taux de
liant
0.46
0.48
0.50
0.52
0.54
CaO/SiO2
Abb.5: Optimierungskurve für A-Zahl bei 40% S-Bedoin
45 % Bedoin-Sand
0.56
26
liant in %
A-Zahl
1400
54
Bei 45% liegen sowohl λ als auch Rc im akzeptablen Bereich und auch die A-Zahl
übersteigt knapp das Limit von 1450 des zu erzielenden Werts.
Ein anfittten der Werte ist nicht möglich da die Versuche ganz knapp das Maximum
erreicht haben und noch nicht genügend Werte für den „absteigenden Ast“ zur
Verfügung stehen
50% Bedoin-Sand
Bei 50% Bedoin-Sand beträgt der Höchstwert von A (1416) laut Optimierungskurve
bei einem CAO/SiO2 von 0.47 und einem Bindemittelanteil von 27.7%. Die maximal
ermittelte Wert 1559 liegt bei einem CaO/SiO2 von 0.45 und einem Bindemittelanteil
von 26.9 %. Dieser sehr hohe Wert ist auf die ebenfalls hohe ρ zurückzuführen was
auch der schlechte Wert für λ von 0.103 bestätigt.
1600
melange 50 %
36
1550
34
1500
32
30
1400
28
1350
26
liant in %
A-Zahl
1450
1300
24
1250
1200
0.35
A-Zahl
taux de liant
0.40
0.45
0.50
0.55
22
0.60
CaO/SiO2
Abb.6: Optimierungskurve für A-Zahl bei 50% S-Bedoin
Sandfeinheit (Lasergranulometrie)
Der Einfluss der verschiedenen Sandfeinheiten wird in Abb. 5 deutlich. Es wurden zwei
verschiedene Mahlfeinheiten des Sandes aus St Savin verwendet. In den Mischungen mit dem
feinen Sand sind die Korngrößenverteilungen der beiden Sande ungefähr gleich in den
anderen ist der Sand aus Bedoin relativ zum Sand aus St Sv feiner. Diese Gießungen mit dem
liegen mit ihren A-Zahlen im Durchschnitt um 30 höher als die gleichen Gießungen mit
feinerem Sand aus St Savin (siehe Abb.5). Auch hier zeigt sich wieder eine Zunahme der
Differenz der A-Zahlen mit steigendem Quarzanteil im Sand Die große Abweichung des
Wertes bei 50%-iger Mischung in der Reihe 2 ist auf eine zu hohe Rohdichte zurückzuführen.
Eine größere Feinheit des quarzreichen Sandes begünstigt somit die Kristallisation
des Tobermorits und führt zu besseren Werten für die Druckfestigkeit.
55
Abb.7: Ergebnisse der Lasergranulometrie fur den Bedoin-Sand
56
Abb.8: Ergebnisse der Lasergranulometrie für den feinen Savin-Sand (Standard)
57
Abb.9: Ergebnisse der Lasergranulometrie für den groben Savin-Sand
58
Abb.10: Granulometriekurven der verwendeten Sande
Tabelle 2
Ergebnisse der Lasergranulometrie
Sand
St Savin fein
St Savin grob
Bedoin
D < 10
D < 50
D < 90
4.72
5.67
6.08
43.88
58.10
45.34
125.97
181.94
125.18
A-Zahl
Azahl gegen Sandfeinheit
1510
1490
1470
1450
1430
1410
1390
1370
1350
fein Savin (normal)
grob Savin
30 30 40 40 50 50 60 60 70 70
% sable Bedoin
Abb.5: A-Zahl aufgetragen gegen die verschiedenen Sandmischungen mit feinem und
gröber gemahlenem Sand aus St Savin
59
Zusammenfassung
Zusammenfassend ist zu sagen, dass eine sehr gute Optimierung der A-Zahlen fur die
Mischung mit 30% Bedoin-Sand leider nicht möglich war. Es konnten nur Maximalwerte von
1380 erreicht werden.
Dieses Maximum erreicht man mit einem Bindemittelanteil von ca. 28% und einem
entsprechenden CaO/SiO2 Verhältnis von 0.53.
Mit quarzreicheren Mischungen können wesentlich bessere Ergebnisse erzielt werden.
Hierbei ist aber auf den Grenzwert der Rohdichte zu achten. Es fällt auf das mit höherem
Anteil an BR, Zement und Kalk die Rohdichten oft zu hoch liegen.
Sehr deutlich in den Versuchen mit 50% Bedoin-Sand zu erkennen.
Es sollte daher beim Erstellen der Rezepte darauf geachtet werden diese werte möglichst
gering zu halten und Bindemitttelanteil und CaO/SiO2 über die Gesamtmasse an festem
Material zu steuern.
Es reicht anscheinend nicht aus für eine kleinere rohdichte nur die Gesamtmasse an festen
material zu verringern.
Weitere Versuche die durchgeführt werden sollten sind meiner Meinung nach auch
Gießungen mit 100% Quarzsand. Hierzu können die Rezepte aus Montreau mit leichten
Modifikationen übernommen werden. Dort verwendet man für die Rohdichte von 350
ausschließlich 100% Quarzsand da dieser in der werkseigenen Grube zur Verfügung steht.
Rezepte
Tabelle 3
liant
Ca/Si
Ms
Ss
BR
Cim
Chaux
Anh
Al1
Al2
20
26.3
30
28.0
40
28.1
45
28.5
50
27.6
60
27.6
70
27.1
0.52
0.53
0.51
0.50
0.47
0.45
0.42
1940
1070
280
360
150
80
1.75
0.75
1930
1030
280
370
170
80
1.75
0.75
1940
1035
280
380
165
80
1.75
0.75
1930
1020
280
380
170
80
1.75
0.75
1940
1045
280
375
160
80
1.75
0.75
1940
1045
280
380
155
80
1.75
0.75
1940
1055
280
375
150
80
1.75
0.75
In Tabelle 3 sind die theoretischen Rezepte aufgeführt, die zum erreichen der jeweiligen
Maximumswerte verwendet werden sollten.
In Tabelle 4 sind die Rezepte aufgelistet die tatsächlich verwendet wurden für die Gießungen
mit den maximalen A-Zahlen.
Die dazugehörigen Nummern und das Datum der Gießung sind ebenfalls aus der Tabelle 4 zu
entnehmen.
Diese Rezepte können immer nur Richtwerte sein da sie den Schwankungen der
Mischmaschine unterliegen.
60
Tabelle 4
melange
taux de liant
CaO/SiO2
Ms
Ss
Br
Cim
Chaux
Anh
Al1
Al2
Gießung
Datum
20
30
40
45
50
60
70
26.4
0.52
1900
1037.7
280
358.5
143.8
27.9
0.51
1940
999.2
320
380.8
160
80
1.75
0.75
75
28.9
0.51
1980
1026.8
300
413.2
160
80
1.7
0.75
78
26.9
0.45
1980
1087.2
280
391
141.8
80
1.7
0.75
76
27.2
0.45
1950
1059
280
371
160
80
1.75
0.75
81
30.2
0.49
1940
80
1.75
0.75
64
27.3
0.51
1900
1021
280
367
152
80
1.75
0.75
101
19.09.2006
14.09.2006
21.09.2006
24.08.2006
07.09.2006
27.09.2006
29.09.2006
975
300
399
186
80
1.75
0.75
59
61
Appendix B
Crystal Chemistry of Xonotlite Ca6Si6O17(OH)2. Part I: Determination of
Polytypes using X-Ray Powder Diffraction
published in: Neues Jahrbuch für Mienralogie, Vol. 186/2, p 153-162, August 2009
Crystal Chemistry of Xonotlite Ca6Si6O17(OH)2. Part I:
Determination of Polytypes Using X-Ray Powder Diffraction
(XRPD)
Saskia Bernstein, Karl Thomas Fehr, Rupert Hochleitner
Abstract: The crystal chemistry of xonotlite is mainly controlled by its four different polytypes. Ten natural
xonotlites from three different lithologies were studied in order to determine their polytypes by powder methods
as X-ray powder diffraction (XRPD). The chemical compositions were obtained by electron microprobe analysis
(EMPA) and show constant compositions with minor substitutions of Mn and Al. To determine the 4 ordered
polytypes known for xonotlite (M2a2b2c, M2a2bc, Ma2b2c, Ma2bc), the theoretical diffraction patterns were
calculated based on atom coordinates of Hejny & Armbruster (2001). For each polytype characteristic reflections
can be chosen. The assignments were conducted by means of pattern matching. The results reveal that xonotlite
mainly occurs in nature as intergrowths of two up to four polytypes. It could be demonstrated that X-ray powder
diffraction is a useful and fast method to determine the different polytypes of xonotlite.
Key words: xonotlite, crystal chemistry, polytypism, XRPD, pattern matching
Introduction
The physico-mechanical properties of steam-cured building materials are determined by the
type and the structure of the Calcium-Silicate-Hydrates or CSH phases using the notation of
cement chemistry for CaO, SiO2 and H2O, respectively. These binders are formed during the
hydrothermal curing at elevated temperatures under saturated steam pressure. They could
account for 10 up to 80 weight percent of the solid phases of the product. Depending on the
62
type of material and hardening temperature 1.13 nm tobermorite or xonotlite are the
predominant phases with semi crystalline CSH-phases as minor components. In addition to
steam cured building materials like light weight RPC there is a wide range of technical
applications of xonotlite ranging from storage of hazardous wastes to insulating material and
flame retardants. Xonotlite (Ca6Si6O17(OH)2), was described first by Rammelsberg (1866) in
contact-metamorphic limestones of Tetela de Xonotla, Mexico and can be found in nature as a
vein-forming mineral in many different localities. Xonotlite is formed mainly as a product of
Ca-metasomatism in the contact-zone of Ca-bearing rocks with igneous (often ultramafic)
rocks (Brown, 1978, Henry, 1999; Marincea et al., 2001; Esteban et al., 2003). Xonotlite
crystallizes in monoclinic symmetry and is usually forming acicular to fibrous crystals up to
centimetre size.
Mamedov & Belov (1955, 1956a) were the first to propose a structure model for xonotlite
which was later confirmed by Eberhard et al. (1981). The structure of xonotlite consists of CaO-polyhedral layers and [Si6O17]-Dreier-Doppelketten. Two of the Ca-atoms are in sevenfold
coordination surrounded by 6 oxygens in form of a trigonal prism and one additional oxygen
on one prism plane, the third Ca-atom is in octahedral coordination (Fig. 1). CaO6-octahedra
and CaO7-polyhedra in sevenfold coordination are both edge-sharing to form infinite chains in
the b-direction. These structural elements were confirmed by extended X-ray absorption fine
structure (EXAFS) investigations of Ca by Lequeux et al. (1999). The different chains are
joined together by sharing edges and build up layers parallel to (001). Between these layers
the [Si6O17]-Dreier-Doppelketten are located. Each of these double chains consists of two
wollastonite-like Dreier-Einfachketten with two paired tetrahedra and one bridging
tetrahedron as shown in Figure 1. These structural units were also confirmed by
29
Si NMR
(Cong et al., 1996; Noma et al., 1998) on synthetic material. In comparison to Cong et al.
(1996), Noma et al. (1998) observed a splitting of Q2 sites which was attributed to different
Si-O bond lengths and Si-O-Si angles of the paired tetrahedra and additional Q1 sites which
63
were interpreted in terms of disorder and may be due to the synthesis path. In xonotlite the
bridging tetrahedra are connected to the Ca-polyhedral layers. Due to the same length of
[Si6O17]-Dreier-Doppelketten and two Ca-polyhedra there exist two different ways of
attachment of the double chains to the polyhedral layers and hence various polytypes are
possible (Gard, 1966; Kudoh & Takeuchi, 1979). The OH-group is located at the free apex of
a CaO6-octahedron where no bridging tetrahedra are attached (Kudoh & Takeuchi., 1979).
Either every CaO6-octahedron carries one OH-group, or CaO6-octahedra containing two OHgroups are alternating with non-hydroxylated ones depending on the polytype developed.
Noma et al. (1998) measured a 1H NMR signal of 1.86 ppm in synthetic xonotlite. This proton
was attributed as a component of a silanol-group. According to the xonotlite structure silanolgroups should not exist and their occurrence was not confirmed by other NMR studies
(Grimmer & Wieker, 1971; Cong et al., 1996). In addition Noma et al. (1998) detected a
broad shoulder at 5.26 ppm assigned to interlayer water, which was not verified by NMR(Cong et al., 1996), IR- (Kalousek & Roy, 1957) and TGA/DSC-studies (Shaw et al., 2000).
Polytypism of xonotlite
Based on the structure model of Mamedov & Belov (1955, 1956a) and the confirmation of
Eberhart et al (1981) six different polytypes (four ordered and two one-dimensional
disordered) were suggested for xonotlite. These Polytypes can be seen as different stacking in
[100]- and [001]-direction of a protoxonotlite-cell introduced by Kudoh & Takeushi (1979).
In [100]-direction a continuous shift of +b/4 or –b/4 or an alternating shift of +b/4 and –b/4
is possible. In [001]-direction the protoxonotlite-cells are either in juxtaposed positions or
shifted by b/2. The combination of these different stacking modes leads to four ordered
polytypes M2a2bc, M2a2b2c, Ma2bc and Ma2b2c as shown in Figure 2. The letter M
indicates the monoclinic symmetry of the protoxonotlite-cell and the three lower case letters,
with numerical values in front if necessary, indicate the periodicity of the three directions in
64
space according to the modified Gard-notation (Guinier et al., 1984). The different cell
parameters were determined by Hejny & Armbruster (2001) and are listed in Table 1. For
M2a2bc and M2a2b2c twinning is possible if one species displays intergrowth of domains
with continuous shift of +b/4 in a-direction and continuous shift of –b/4. Streaks parallel to a*
observed in single crystal patterns by Gard (1966) were assigned to the two known disordered
polytypes P∞21 and A∞22 ( Corresponding to Mad2bc and Mad2b2c in modified Gardnotation). Hejny & Armbruster (2001) extended the group of possible polytypes by Ma2bcd
and M2a2bcd, which have one-dimensional disorder in c-direction as indicated by streaks
observed parallel to c*
(Chisholm, 1980; Eberhard et al.,1981). Short spikes recorded
perpendicular to the c* streaks have been interpreted in terms of two-dimensional disorder
(Dornberger-Schiff, 1964), and they are termed with the corresponding symbol Mad2bcd
(Hejny & Armbruster, 2001).
Scope of this study
The determination of reaction and growth kinetics of CSH-phases and the calculation of
thermodynamic equilibria demands material well characterized by its crystal chemistry.
Single crystal diffraction is the most exact method to determine the different polytypes but
these examinations are very time consuming and require crystals of a certain size. ufficiently
big crystals of xonotlite are rare in nature and usually do not occur in synthesis experiments or
technical applications. Therefore a less time-consuming determination method is required
suitable for fine grained samples as well. Esteban et al (2003) could confirm the occurrence of
M2a2b2c and Ma2b2c in the investigated sample from Carratraca (Spain) by comparing
recorded and calculated diffraction patterns. Another attempt to determine the different
polytypes was carried out by Garbev (2004) by using Rietveld-analysis and a structural model
containing all polytypes. In the first part of this study the main focus is put on X-ray powder
diffraction and pattern matching techniques of natural xonotlites. In addition a systematic
65
study on the composition of xonotlite from different localities is missing and only one report
on the chemical variation in xonotlite from the Kalahari Manganese Field is given by de
Bruiyn et al. (1999). Therefore a further aim of this study is to check for a dependency of the
polytypes evolved and the chemical composition.
Experimental
Ten natural xonotlite samples from seven different localities and different lithologies were
investigated (Tab. 2). All samples form polycrystalline aggregates of white to whitish crystals.
Most of them consist of fine fibres sometimes displaying radial growth. In the sample from
Mäntijärvi, Finland (Xon2) xonotlite is filling small cavities in a kimberlite and is
accompanied by calcite. Xonotlites from Bazhenovskoe and Chukotka, Russia (Xon1, Xon3,
Xon6, Xon7) and Carratraca, Spain (Xon5) occur in rodingite veins crosscutting serpentinites.
Xon4, Xon8 and Xon10 from the Kalahari Manganese Field, RSA and Xon9 from Franklin,
New Jersey, are products of metasomatism caused by low temperature hydrothermal fluids in
manganese- and zinc-manganese-deposits respectively. Xon4 is a well crystallized xonotlite
forming acicular crystals up to 30 mm long and 1mm in diameter, which were previously used
by Hejny & Armbruster (2001) for single crystal diffraction.
Quantitative chemical data for xonotlites were obtained by electron microprobe analysis
(EMPA) using a CAMECA SX100 operated at 15 keV acceleration voltage and 20 nA beam
current. Synthetic wollastonite (Ca,Si), periclase (Mg), corundum (Al), hematite (Fe),
escolaite (Cr), natural ilmenite (Mn,Ti), albite (Na) and osumilite (K) were used as standards
and matrix correction was performed by the PAP procedure (Pouchou & Pichoir, 1984). The
reproducibility of standard analyses was <1% for each element routinely analysed.
66
X-ray powder diffraction (XRPD) data were determined with a STOE STADI Pdiffractometer using a SIEMENS KRISTALLOFLEX 710/710H generator operating at the
following conditions: 40 kV, beam current 30mA, curved Germanium monochromator, step
scan in the 2Θ region 10-60° with 0,01 2Θ steps and Cu-Kα-radiation (λ=1.5406 Å). For
determination of the present polytype or polytypic intergrowth the theoretic powder
diffraction patterns of the four ordered polytypes were simulated for Cu-Kα-radiation. For
calculations the tool “Visualizer” of ICSD was applied using the atomic coordinates for
M2a2bc, Ma2bc and Ma2b2c of Hejny & Armbruster (2001). The diffraction pattern for the
polytype M2a2b2c described by Kudoh & Takeushi (1979) was calculated with the software
package “Fullprof” for Rietveld-refinement (Rodriguez-Carvajal; 1993). The obtained
patterns were used to define every polytype by a set of diagnostic peaks in the range of 10-35°
2Θ. The polytypes of natural xonotlites were determined by matching the measured patterns
with the model patterns. To revise the quality of the results Rietveld-analysis was performed
for a selected sample (Figure 5) by using the FULLPROF-software (Rodriguez-Carvajal;
1993)
Results
Chemical composition
The chemical compositions of xonotlites under investigation are shown in Table 4. The data
are the mean out of 10 analyses except for Xon2 where only three analyses were taken into
account. All xonotlites are predominately Calciumsilikates, other elements only occur as
minor components. The amount of SiO2 ranges from 49.37 wt% up to 51.10 wt%. The CaO
contents vary between 45.76 wt% and 47.48 wt%. Samples from the Kalahari Manganese
Field (N’Chwaning and Wessels Mine) show the highest amounts of SiO2 (49.37-51.10 wt%)
67
accompanied by a lower amount of CaO (45.82-47.48 wt%). Xonotlites from rodingites
(Bazhenovskoe, Chukotka and Carratraca) contain the highest amounts of CaO up to 47.37
wt%. The content of Al2O3 is low and varies between 0.02 wt% and 0.04 wt% except for a
sample from the Wessels Mine (Xon10 in Tab.4) and from Carratraca (Xon5 in Tab.4)
revealing 0.12 and 0.06 wt%, respectively. The contents of FeO are negligible except for
samples Xon2 showing a slightly higher value of 0.10wt%. MnO is considerably enriched in
samples from the Kalahari Manganese Field (Xon4,Xon8 in Tab. 4) and the Franklin (Xon9),
revealing contents of MnO from 0.18 up to 0.36 wt% as shown in Table 3. Only xonotlite
from a kimberlite (Xon2 in Tab.4) and from rodingites in Carraraca (Spain, Xon5 in Tab.4)
contains slightly higher alkali contents of 0.09 and 0.04 wt% Na2O, respectively. The contents
of MgO, K2O, TiO2 and Cr2O3 are negligible in all samples.
Simulated diffraction patterns
The calculated diffractograms of all four polytypes are shown in Figure 3. The patterns are
displayed in two different scales (Figure 3) ranging from 10 to 25° 2Θ and 25 to 35° 2Θ,
respectively. At the first sight the four diffraction patterns are quite similar. Seven peaks of
high intensity and five of lower intensity are common for all four polytypes. Characteristic
differences can only be found by a detailed investigation of peaks with lower intensities in the
region between 10 and 35° 2Θ as depicted in Table 5. Peaks which allow distinction between
the four polytypes are called "characteristic peaks". The indication corresponding to the
values of ° 2Θ used in the text can be taken likewise from Table 5.
The calculated pattern of the M2a2b2c polytype shows ten characteristic peaks. Most
demonstrative is the (013) peak at 22.7°2Θ. Weaker characteristic peaks are those at 13.8,
15.7, 18.8, 23.9, 26.1, 27.2, 29.2, 32.0 and 34.3° 2Θ.
68
The M2a2bc polytype shows nine characteristic peaks. The strongest are the (011), (-111) and
(012) peaks at 17.6, 19.1 and 28.4° 2Θ, respectively. Further weaker peaks are those at 12.3,
14.3, 21.8, 29.3, 31.1 and 33.6° 2Θ.
The pattern of the Ma2b2c polytype has two stronger characteristic peaks at 23.1° 2Θ (113)
and 29.6° 2Θ (511). Weaker peaks are to be found at 22.5, 30.8, 24.9, 34.5 and 14.6° 2Θ.
The pattern of the Ma2bc polytype shows the strongest peak at 18.2° 2Θ (111) and also
distinct characteristic peaks at 17.4, 18.2, 20.4 and 28.6° 2Θ. Weaker peaks are at 12.0, 13.1,
15.9, 19.7, 23.6, 28.2, 30.2 and 32.3° 2Θ.
Polytypes in natural xonotlites
The characteristic peaks of the different polytypes in the investigated samples are summarized
in Table 5. Most of the xonotlites show intergrowth of two or three different polytypes. As the
characteristic peaks are of very low intensity sometimes not all are detectable in the
investigated patterns. The assignement is demonstrated in Figure 4 for Xon2 (Mäntijärvi,
Finland). Due to occurring texture effects, which are discussed later, some peaks have highly
increased and others decreased intensities. Some of the characteristic peaks can coincide with
those of other polytypes (if occurring in the same sample). For this reason different peaks had
to be used for polytype assignment in the different samples. The reproducibility of the
obtained results was verified by means of random sampling.
M2a2b2c
69
Every investigated xonotlite shows the peaks at 13.8, 22.7 and 34.3° 2Θ, characteristic for the
M2a2b2c-polytype. These three peaks often have broad, asymmetrical shapes. They show a
weaker intensity in the patterns of Xon2 (Mäntijärvi, Finland, Fig.4) and Xon8 (Wessels
Mine, South Africa). This is due to a minor amount of the M2a2b2c polytype in these
samples. In addition there is a distinct reflection at 26.1° 2Θ in Xon2 (see Fig.4). This
reflection is also visible in the patterns of Xon3 (Chukotka, Russia) and Xon10 (Wessels
Mine, South Africa) but here with a medium intensity. The characteristic peaks at 18.8 and
23.9° 2Θ are detectable only in the pattern of Xon5 (Carratraca, Spain). The (-111) peak at
15.7° has been found only subordinate in the patterns of Xon1 and Xon3, both from Russian
rodingites. In sample 1, 3 and 10 the (2-13) peak is weakly developed. The (3-11) peak is
detectable in patterns of Xon1, Xon4, Xon6 and Xon8 (Tab. 6). Sample 6, 7 and 10 show only
characteristica of the M2a2b2c polytype. Whereas the peak at 13.8° 2Θ is weak in the
calculated pattern, nearly all natural xonotlites show a much higher intensity. In contrast the
2-13 peak is of medium intensity in the calculated pattern, whereas in Xon5 it shows only a
weak intensity.
M2a2bc
The M2a2bc polytype could be found in three of the investigated xonotlite samples. All three
patterns show the peaks at 28.4, 29.3 and 33.6° 2Θ which have been used for the assignment
of this polytype. Xon3 (Chukotka, Russia) shows every peak characteristic for the M2a2bc
polytype (see Table 5) with exception of the (011) peak at 17.6° 2Θ, whereas the latter peak is
strongly developed in the pattern of Xon2 (Mäntjärvi, Finland, Fig. 4).
The normally weak (-112) peak is highly increased in sample 5 from Carratraca (Spain),
additional characteristic peaks of the M2a2bc polytype have been found only between 25 and
35° 2Θ.
70
Ma2b2c
This polytype could be found in four xonotlite samples. Two of them come from Russian
rodingites (Xon1, Xon3) and the two remaining from manganese ore deposits at Wessels
Mine in South Africa (Xon8) and at Franklin (New Jersey) in the USA (Xon9). All four
patterns show the (213) and (511) reflections at 24.9 and 29.6° 2Θ. The (-111) peak at 14.6
2Θ, which has a very low intensity in the calculated pattern, is missing only in Xon9
(Franklin, New Jersey). The detection of the (013) and (611) characteristic peaks at 22.5 and
34.5° 2Θ is difficult due to the superposition with characteristic peaks of the M2a2b2c
polytype at 22.7 and 34.3° 2Θ. The peaks at 23.1 2Θ and 30.8 2Θ were detected in the
patterns of Xon9 and Xon3, respectively.
Ma2bc
This polytype has been found in four of the investigated xonotlites (Xon1, Xon2,Xon3,
Xon4). The highest amount is detected in Xon2 due to very strong peaks at 30.2 and 32.3° 2Θ
and a medium peak at 20.4° 2Θ which are to be seen with much lower intensities in the other
patterns.
The peaks at 13.1, 28.2 and 28.6° 2Θ are missing in Xon2 but developed in the remaining
three xonotlites. The peak at 12.0° 2Θ (weak in the calculated pattern) is detectable in Xon1
and Xon2, the peak at 18.2° 2Θ in Xon2 and Xon3 patterns. Both xonotlites from Russian
rodingites show the peak at 15.9° 2Θ. The very weak peaks at 19.7 and 23.6° 2Θ of the
calculated pattern are only developed in Xon3 and Xon1, respectively. The peak at 17.4 2Θ
could not be detected.
71
Discussion
In this study ten different natural xonotlites were investigated concerning chemical
composition and polytypism. Due to the different lithologies of the localities the samples
could be assigned to three different groups. Xonotlites formed in kimberlites (1), xonotlites in
rodingites (2) and xonotlites formed by metasomatic processes close to Mn- and Mn-Zn- oredeposits (3), respectively.
Chemical composition
Group 1 is only represented by the xonotlite from Mäntijärvi (Finland). This sample exhibits a
low SiO2- and CaO-content and a slight enhancement of Na2O up to 0.09 wt%.
Group 2-xonotlites were formed in rodingites (Xon1, Xon3, Xon5, Xon6) and are
characterized by the highest CaO-amount of 47.07 up to 47.59 wt%. Analyses of Carratracaxonotlites (Xon5) are in good agreement to those of acicular crystals replacing hydrogrossular
of the same locality published by Esteban et al. (2003) as shown in Table 3 and 4.
Samples belonging to group 3 are formed by hydrothermal alteration (250-400ºC) of the
primary sedimentary and low-grade metamorphic Mn-ores (Kalahari Manganese Field ;
Xon4, Xon8) or high-grade metamorphic Mn-Zn-ores (Franklin, New Jersey; Xon9) and
showing slightly higher Mn-content. This could be explained by a preferred integration of Mn
on the Ca-positions in Xonotlite-structure, likewise indicated by a lower Ca-content.
A substitution of Al for Si on tetrahedral-sites, indicated by the higher amount of Al2O3 could
be detected noticeably only in the sample from Carratraca and one from Wessels mine.
De Bruiyn et al. (1999) described a higher SiO2- and CaO-content in xonotlites of
N’Chwaning Mine in comparison to those of Wessels Mine, which could be confirmed in this
investigation, as depicted in Table 3 and 4. In addition De Bruiyn et al. (1999) detected
slightly higher FeO-contents which could not be verified in the samples of this study.
72
In all investigated xonotlites Al, Na and Mn are the only elements which are enriched in
remarkable amounts. A enhancement of Mg known from synthetic xonotlites (Quian et al.,
1997) could not be detected in the investigated natural xonotlites.
Polytypes
Due to the minor crystal size of xonotlites X-ray single crystal diffraction is often not
applicable to distinguish between the different polytypes. Therefore the determination in this
study is made by pattern matching powder diffractograms of investigated xonotlites and
calculated patterns of the four ordered polytypes. Most of the investigated xonotlites show
intergrowth of two or more polytypes.
In all samples the M2a2b2c polytype was found but in Xon2, Xon8 and Xon9 with only
minor amounts. This predominance is an effect of specific growth conditions availing the
development of one polytype. Hejny and Armbruster (2001) explain this preferred
development by the more balanced and therefore favourable distribution of OH-groups at the
free apices of each Ca-octahedron in the structure of this polytype.
The Ma2b2c-polytype could be detected in two samples found in Russian rodingites of
Bazhenovskoe (Xon1) and Chukotka (Xon3) and in two samples found in Mn-ore deposits
from the Wessels Mine located in the Kalahari Manganese Field of South Africa (Xon8) and
from Franklin, New Jersey (Xon9), respectively.
Ma2bc is developed in the xonotlite from Mäntijärvi (Finland) with an exceptionally high
amount and in the well crystalline sample from N’Chwaning Mine (Xon4) investigated by
Hejny & Armbruster (2001). In addition this polytype does exist with minor amounts in
xonotlites of Bazhenovskoe (Xon1) and Chukotka (Xon3).
The occurrence of M2a2b2c and Ma2bc in Xon4 is in good agreement with the investigations
of Hejny & Armbruster (2001) on xonotlites from the same locality. They also reported the
presence of Ma2b2c polytype which could not be confirmed in this study. This can be
73
explained by different intergrown polytypes even in samples from the same locality due to
small scale fluctuations in the physico-chemical conditions during formation.
The M2a2bc-polytype was first detected in natural xonotlite from Chukotka (Russia) by
Garbev (2004) using Rietveld-modelling of diffraction data obtained by syncrotron radiation.
In this study the M2a2bc polytype could clearly be detected in xonotlites from Mäntijärvi
(Xon2) and Chukotka (Xon3). Esteban et al. (2003) made their assignment in Xonotlites from
Carratraca (Spain) by use of X-ray powder diffraction, too. Based on the by then missing
description of the occurrence of M2a2bc polytype in natural xonotlites in literature, this
polytype was not taken into account by Esteban et al.(2003). This is in contrast to the results
of this study where the development of M2a2bc polytype in xonotlite from Carratraca (Xon5)
could be confirmed by the presence of three characteristic reflections.
The above mentioned lower intensity of (h0l)-reflections in all patterns can be explained by a
preferred orientation of the acicular crystals along their elongated b-axes during preparation.
The observed phenomenon of inverse intensity-ratio in (0kl) and (hkl)-reflections in M2a2bcpolytype may be caused by preferred orientation of more disk-shaped crystals.
The powder diffraction data of sample 8 from Wessels mine ( South Africa) were used
additionally to perform a Rietveld refinement (Figure 5). The two polytypes determined by
pattern matching were used as phases for the refinement. We obtained the best results by
taking into account a preferred orientation in [010] which corresponds to the elongated bdirection of the crystals. But there is also a recognizable influence of the above mentioned
[h0l] direction which could explain the observed difference in the region of the (102)
reflection. The results show the exigence of further refinement with main focus on the
different preferred orientations. For this reason we abandoned a quantitative analysis for now.
74
Conclusions
A clear coherence of different lithology of the habitat and the developed polytype could not
be confirmed. Xon2 from kimberlites (Mäntijärvi, Finland) is in a special position referring to
this question. It displays a very high amount of Ma2bc polytype compared to all other
investigated samples. This can be linked to the special growth conditions in kimberlites.
The results of this study clearly demonstrate that X-ray powder diffraction is a useful and fast
method to distinguish the different polytypes developed in xonotlite. It is the preferable option
if crystallite size is too small for X-ray single crystal diffraction as it is typical for natural and
synthetic xonotlites. The results show also the need of further investigations by other powder
methods like FTIR and Raman spectroscopy. FTIR-data of xonotlite were published by
Hochleitner & Fehr (2002) and an extensive investigation will be published in a forthcoming
paper.
Acknowledgements
We would like to thank Prof. Dr. Thomas Armbruster and Dr. Clivia Hejny for kindly
providing a natural xonotlite sample from N’Chwaning (South Africa). Detailed comments by
an anonymous reviewer and Prof. Dr. Wolfgang Schmahl were also greatly appreciated.
References
BROWN, P.A. (1978): Xonotlite: a new occurrence at Rose Blanche, Newfoundland.- Can.Mineral.,16: 671-672
CHISHOLM, J.E. (1980): Polytism in xonotlite, Ca6Si6O17(OH)2. In Electron Microscopy and Analysis, 1979. Inst.
Phys. Conf. Ser., 52: 109-112
CONG, X. & KIRKPATRICK, R.J. (1996): 29Si and 17O NMR Investigation of the Structure of Some Crystalline
Calcium Silicate Hydrates. - Advn. Cem. Bas. Mat., 3: 133-143
DANA, E.S. & FORD, W.E. (1892): The system of mineralogy if James Dwight Dana, 6th edition: 569; publisher
unknown, New York, United States (USA)
75
DE BRUIYN, H.; SCHOCH, A.E.; VAN DER WESTHUIZEN, W.A & BEUKES, G.J.
(1999): The chemical composition
of xonotlite and associated inesite from the Nchwaning and Wessels mines, Kalahari manganese field, South
Africa. - N.Jb.Miner.Mh., 5: 212-222
DORNBERGER-SCHIFF, K. (1964): Grundzüge einer Theorie der OD-Strukturen aus Schichten. -Abhandlungen der
Deutschen Akademie der Wissenschaften zu Berlin. Klasse für Chemie, Geologie und Biologie, 3:1-107
EBERHART; E.; HAMID, S. & RÖTTGER, B. (1981): Strukturverfeinerung und Polytypie von Xonotlite:- Z. Krist.,
154: 271-272
ESTEBAN, J.J.; CUEVAS, J. & TUBIA, J.M. (2003): Xonotlite in rodingite assemblages from the Ronda peridotite
Betic Cordilleras, Southern Spain. –Can. Mineral., 41: 161-170
GARBEV, K. (2004): Struktur, Eigenschaften und quantitative Rietveldanalyse von
hydrothermal kristallisierten Calciumsilikathydraten (CSH-phasen),Thesis,
University of Karlsruhe
GARD, J.A.(1966): A System of Nomenclature for the Fibrous Calcium Silicates, and a Study of Xonotlite
Polytypes. –Nature, 211: 1078-1079
GRIMMER, A.-R. & WIEKER, W. (1971): Bestimmung der Art der Wasserbindung im Xonotlit 6CaO 6SiO2 H2O.
–Z.anorg.allg.Chem., 384: 34-42)
GUINIER , A.; BOKIJ, G.B.; BOLL-DORNBERGER, K.; COWLEY, J.M.; DUROVIC, S.; JAGODSINSKI, H.; KRISHNA, P.;
DEWOLFF, P.M.; ZVYAGIN, B.B.; COX, D.E.; GOODMAN, P.; HAHN, TH.; KUCHITSU, K. & ABRAHAMS, S.C.
(1984): Nomenclature of Polytype Structures. Report of the International Union of Crystallography Ad.Hoc
Committee on the Nomenclature of Disordered, Modulated and Polytype Structures. -Acta Cryst, A40: 399404
HEJNY,C. & ARMBRUSTER, T. (2001): Polytypism in Xonotlite Ca6Si6O17(OH)2.- Z.Kristallogr., 216: 396-408
HENRY, D.A. (1999): Cuspidine-bearing skarn from Chesney Vale, Victoria. –Australian J. of Earth Sciences,
46: 251-260
HOCHLEITNER, R. & FEHR, K.T. (2002): Infrared spectroscopy of natural xonotlites. –Beih .z. Eur. J. Mineral.,
14: 70
KALOUSEK, G.L. & ROY, R. (1957): Crystal Chemistry of Hydrous Calcium Silicates: II, Characterization of
interlayer Water. - J. Am. Ceram. Soc., 40 [7]: 236-239
KUDOH, Y. & TAKEUSHI, Y. (1979): Polytypism of xonotlite: (I) Structure of an A1 polytype - Min.Journal, 9:
349-373
LEQUEUX, N., MORAU, A.; PHILIPPOT, S. & BOCH, P. (1999): Extended X-ray Absorbtion Fine Structure
Investigation of Calcim Silicate Hydrates. –J. Am. Ceram. Soc., 82 [5]: 1299-1306
MAMEDOV, KH. & BELOV, N.V. (1956a): Crystal structure of the mineral group wollastonite. 1 structure of
xonotlite. - Zapiskii Vsesoyuz. Miner. Obshchest, 85: 13-38
MAMEDOV,KH. & BELOV, N.V. (1955): Crystal structure of xonotlite. –Dokl.Akad.Nauk SSSR, 104: 615-618
MARINCEA, S.; BILAL, E.; VERKAEREN, J.; PASCAL, M.-L; FONTEILLES, M. (2001): Superimposed parageneses in
thee spurrite-, tilleyite- and gehlenite-bearing skarns from Cornet Hill Apuseni Mountains, Romania. –Can.
Mineral., 39: 1435-1453
NOMA, H. ADACHI, Y.; MATSUDA, Y. & YOKOYAMA, T. (1998): 29Si and 1H NMR of Natural and Synthetic
Xonotlite. - Chemistry Letters, 219-220
POUCHOU, J.L. & PICHOIR, F. (1984): A new model fro quantitative X-ray microanalysis Part I: application to the
analysis of homogenious samples. -La Recherche Aérospatiale, 3: 13-38
76
PRODAN, A. MARINCOVIC, V.; VENE, N.; KURBUS, B. & BOSWELL, F.W. (1983): On structural relations between
the CaO-SiO2-H2O phases. –Z. Kristallogr., 164: 189-209
RAMMELSBERG, C. (1866): Über den Xonaltit, ein neues wasserhaltiges Kalksilikat, und den Bustamit aus
Mexico. - Z. deutsch. geol. Ges. 18: 33-34
SHAW, S. HENDERSON, C.M.B. & KOMANSCHEK, B.U. (2000): Dehydration/recrystallization mechanism ,
energetics, and kinetics of hydrated calcium silicate minerals: an in situ TGA/DSC and synchrotron radiation
SAXS/WAXS study. - Chemical Geology, 167: 141-159
QUIAN, G.; XU, G.; LI, H. & LI, A. (1997): Mg-Xonolite and its coexisting phases. –CCR, 27 [3]: 315-320
RODRIGUEZ-CARVAJAL, J. (1993): Recent advances in magnetic structure determination by neutron powder
diffraction.- Physica B., 192: 55 - 69
Authors’ address:
SASKIA BERNSTEIN, KARL THOMAS FEHR: Department für Geo- und Umweltwissenschaften, Ludwig
Maximilians Universität, Theresienstraße 41, 80333 München
RUPERT HOCHLEITNER: Mineralogische Staatssammlung, Theresienstraße 41, 80333 München
Corresponding author’s e-mail: bernstein@min.uni-muenchen.de
Table headings:
TABLE 1: Lattice parameter and symmetry of the ordered polytypes
after Hejny & Armbruster (2001)
TABLE 2: Localities of investigated xonotlites
TABLE 3: Literature values for chemical composition of natural xonotlites
TABLE 4: Average composition of investigated xonotlites obtained by EMPA,
n= number of analyses taken into account, numbers in parentheses
denote for standard deviation
TABLE 5: Characteristic peaks found in the 10 diffraction patterns and the resulting assigned
polytypes (w=weak, m= medium, s=strong, b=broad)
the dominating polytype is marked by “+” those of minor amounts by “*”
Figure captions
Fig 1: structure of xonotlite
Ca-polyhedra: dark grey in octahedral coordination and light grey in sevenfold
coordination
Si6O17 tetrahedral-double chains in middle grey
77
protoxonotlite-cell with white dashed outlines
notice the two different possibilities (light and middle grey) of connecting a SiO4tetrahedral chain to the Ca-polyhedra
Fig 2: Four ordered polytypes shown as different stacking of the protoxonotlite-cell (dark
outlines) after Hejny & Armbruster (2001).
Cells with displacement in c-direction are drawn with dashed outlines
Fig 3: Calculated powder diffraction patterns for all ordered polytypes (Cu-Kα) in the range
of 10-25 and 25-35° 2Θ
Fig 4: Diffraction pattern of Xon2 (Mäntijärvi, Finland), peaks taken for assignment are
marked by arrows and the number of corresponding polytype
1= M2a2b2c, 2=M2a2bc, 3=Ma2b2c, 4=Ma2bc
Fig 5: Rietveld-refinement for Xon8 (Wessels Mine, Souh Africa)
TABLE 1
protoxonotlite cell
M2a2bc
M2a2b2c
Ma2bc
Ma2b2c
a [A]
8.516
8.712
8.712
17.032
17.032
b [A]
7.363
7.363
7.363
7.363
7.363
c [A]
7.012
7.012
14.023
7.012
14.023
α [°]
90.00
89.99
89.99
90.00
90.00
β [°]
90.37
90.36
90.36
90.36
90.36
TABLE 2
sample
Xon 1
Xon 2
Xon 3
Xon 4
locality
Bashenovskoje, Russia
Mäntijärvi, Finland
Tschukotka, Russia
N’Chwaning, South Africa
reference
a)
MSM 30538
a)
MSM 30536
a)
MSM 27382
Fe175b)
Xon 5
Xon 6
Xon 7
Xon 8
Xon 9
Xon 10
Carratraca, Malaga,Spain
Bashenovskoje, Russia
Bashenovskoje, Russia
Wessels Mine, South Africa
Franklin, New Jersey USA
Wessels Mine, South Africa
MSM 30537
MSM
a)
MSM 28676
Fe50
a)
MSM 1218
Fe
a)
a) Mineralogische Staatssammlung München
b) Hejny & Armbruster
γ[°]
90.00
102.18
102.18
90.00
90.00
space group
P1
A1
P2/a
A2/a
78
TABLE 3
Carratraca
(1)
SiO2
CaO
Al2O3
FeO
MnO
MgO
Na2O
K2O
TiO2
Cr2O3
H2O
total
50.07
47.39
0.04
0.07
0.03
0.00
0.02
0.01
0.00
0.02
N´chwaning
South Africa
(2)
50.02
46.15
n.m
0.28
0.18
0.05
n.m
n.m.
n.m.
n.m.
2.50
99.18
Wessels mine South Africa Heguri Japan
(2)
(3)
49.96
45.85
n.m.
0.62
0.21
0.05
n.m.
n.m.
n.m.
n.m.
2.50
99.19
49.99
46.19
n.m.
0.36
0.16
n.m.
0.17
0.02
n.m.
n.m.
3.05
99.94
Ohmi-Machi
Japan (4)
50.80
44.70
0.38.
0.04
0.01
n.m.
0.78
0.02
n.m.
n.m.
3.18.
99.92
(1) Esteban et al., 2003; (2) de Bruiyn et al, 1999; (3) Kudoh et al, 1979;(4) Noma et al., 1998:
(5) Garbev, 2004; (6) Dana,E.S., 1892; (7) Marincea et al., 2001
Tchukotka
Russia
(5)
49.52
46.26
0.29
n.m.
n.m.
n.m.
0.01
0.01
n.m.
n.m.
96.11(incl. 0.02SO2)
Tetela de Xonotla Mexico Cornett Hill, Romania
(6)
(7)
49.58
43.56
n.m.
1.31
1.79
n.m.
n.m.
n.m.
n.m.
n.m.
3.70
99.94
49.79
47.28
0.20
0.03
0.01
0.08
0.00
n.m.
n.m.
n.m.
2.51
99.90
79
TABLE 4
K2O
Na2O
MgO
Al2O3
SiO2
CaO
TiO2
MnO
FeO
Cr2O3
Total
Xon1
b.d.
b.d.
b.d.
0.04(02)
49.80(36)
47.37(22)
b.d.
b.d.
b.d.
b.d.
97.31(19)
Xon2
b.d.
0.09(01)
b.d.
b.d.
479.37(05)
45.76(37)
b.d.
b.d.
0.10(05)
b.d.
95.37
Xon3
b.d.
b.d.
b.d.
0.03(01)
49.71(16)
47.37(29)
b.d.
b.d.
b.d.
b.d.
97.21(30)
Xon4
b.d.
b.d.
b.d.
b.d.
51.10(18)
47.01(28)
b.d.
0.36(03)
b.d.
b.d.
98.53(20)
Xon5
0.00
0.03
0.03
0.08
49.93
47.08
0.00
0.01
0.02
0.01
97.19
Xon6
b.d.
b.d.
b.d.
0.02(01)
50.11(18)
47.10(16)
b.d.
b.d.
b.d.
b.d.
97.31(23)
Xon7
b.d.
b.d.
b.d.
0.02(02)
49.37(23)
47.48(18)
b.d.
b.d.
b.d.
b.d.
96.99(39)
Xon8
b.d.
0.05(03)
b.d.
b.d.
50.21(62)
46.73(41)
b.d.
0.18(02)
b.d.
b.d.
97.20(45)
Xon9
b.d.
b.d.
b.d.
0.02(01)
49.41(23)
46.55(25)
b.d.
0.36(16)
b.d.
b.d.
96.42(48)
Xon10
b.d.
b.d.
b.d.
0.12(06)
50.08(81)
45.82(77)
b.d.
b.d.
b.d.
b.d.
96.09(12)
80
TABLE 5
hkl
° 2Θ
13.8
15.7
18.8
22.7
23.9
26.1
27.2
29.2
32
34.3
011
-111
111
013
1-13
113
211
2-13
3-11
015
12.3
14.3
17.6
19.1
21.8
28.4
29.3
31.1
33.6
010
1-10
011
-111
111
012
-112
-1-12
-212
14.6
22.5
23.1
24.9
29.6
30.8
34.5
-111
013
113
213
511
-413
611
12
13.1
15.9
17.4
18.2
19.7
20.4
23.6
28.2
28.6
30.2
32.3
010
110
210
011
111
310
211
311
012
-112
212
312
M2a2b2c
Xon1 Xon2 Xon3 Xon4 Xon5 Xon6 Xon7 Xon8 Xon9 Xon10
M2a2b2c
s,b
w
s,b
s
m
s,b
s,b
w
m
s,b
w
s
m
s,b
w
s.b
s
s
s,b
s,b m,b
m
s,b
w
s
m
m
w
s
s
m
m
w
ww
w
w
ww
w
s,b
w
s
s,b
s
m
s
ww
w
m
M2a2bc
w
m
m
m
s
m
m*
s*
m
m
w,b
m
s
m
ss
m-s
m
m
m
w
ww
Ma2b2c
w
s
w
m
m,b
s,b
s
m
w
m
w
w
s
s
w,b
w
s
m
w
Ma2bc
m
w,b
m
w
m
w
m
m
w
ww
w
m
m
s
w
m
m
w
m
w
m
w
m
w
w,b
w
m,b
ss
m
w
w
s
w
m
m*
-
+
M2a2bc
Ma2b2c
*
Ma2bc
*
*
+
*
*
+
*
+
+
+
*
*
+
*
*
+
+
*
+
81
Figure 1
82
Figure 2
83
Figure 3
84
Figure 4
85
Figure 5
86
Appendix C
A Hydrothermal Autoclave for Neutron Diffraction (HAND) –
Design, Technique and Applicability
submitted to: Journal of Powder Diffraction, under rewiev
A Hydrothermal Autoclave for Neutron Diffraction (HAND) –
Design, Technique and Applicability
K.T. Fehr(1), S. Bernstein(1), M. Huber(1), E. Peters(2), B. Walk-Lauffer(3), S.G. Zuern(4)
(1)
Department of Earth & Environmental Sciences, Ludwig-Maximilians Universität, Munich,
Germany
(2)
YTONG Holding AG, R&D Center, Schrobenhausen, Germany; currently at Muenchener
Stadentwaesserung, Munich, Germany
(3)
Institute of Building- and Materials Chemistry, University of Siegen, Germany; currently at
Rhein-Chemotechnik GmbH, Breitscheid, Germany
(4)
YTONG Holding AG, R&D Center, Schrobenhausen, Germany; currently at PCI,
Development Department, Augsburg, Germany
Abstract
An autoclave cell has been designed for performing time-resolved neutron diffraction
analyses of dynamic processes occurring during hydrothermal reactions in the presence of a
hydrous fluid. The hydrothermal autoclave for neutron diffraction (HAND) is described and
its successful use on the crystallization of 1.13 nm tobermorite from lime and silica at 190°C
and saturation pressure is demonstrated. The reaction time was set to 8 hours and the reaction
product consisted of tobermorite, semi-crystalline calcium-silicate-hydrate C-S-H(I) and
quartz. Tobermorite is formed on the expense of portlandite and quartz and by the reaction of
semi-crystalline calcium-silicate-hydrate C-S-H(I) with quartz.
87
Keywords: Hydrothermal, autoclave, neutron diffraction, kinetic, tobermorite
1. Introduction
Hydrothermal syntheses under saturated steam pressure are usually performed by using so
called Parr-bombs. Once the experiment is completed the sample is quenched and the reaction
products are analysed. By this way of synthesizing one always has to deal with several
problems. Quenching effects can influence the final phase relations in particular and there is
no guarantee to freeze the process exactly at one stage and to quench it without change.
Furthermore a huge number of experiments at different compositions and temperatures are
needed to obtain a sufficient amount of data points which means an enormous expenditure of
time. This is hardly possible by using Parr bombs without an enormous amount of both time
and work. Reactions with fast kinetics are impossible to record by this method and reliable
data especially from the early state of the reaction are missing. The best way to obtain kinetic
data of a reaction is to observe it in-situ by x-ray or neutron diffraction. For those experiments
special reaction cells are needed tailored to the particular requirements of the scientific
problem and of course the applied instrument, respectively.
The reaction we are focussed on is the formation of 1.13nm tobermorite taking place during
the hydrothermal hardening of aerated autoclaved concrete, a building material used
worldwide due to its excellent mechanical properties. Corresponding to the production
process of AAC or lime silicate bricks the hydrothermal hardening of lime silica based
samples in the vapour phase and not in suspension should be examined. Former
thermodynamic and experimental studies [Gabrosek et al., 1993; Fehr and Zuern, 1997; Zuern
and Fehr, 2000a] have shown that tobermorite is metastable in the presence of quartz at
temperatures of production and decomposes to the equilibrium phases xonotlite (Ca6Si6(OH)6)
88
or gyrolite according to the bulk composition. This fact shows the strong need of kinetic
models in order to describe the metastable formation of 1.13 nm tobermorite.
Requirement for successful in-situ experiments according to reaction kinetics is a high-flux
neutron source [Polak et al., 1990;Walton et al., 2000; Walton and O’Hare, 2000] that enables
an adequate time resolution in spite of the absorption of X-rays by the reaction cell and a
diffractometer with a sufficient 2Θ range to detect the basal reflections of tobermorite. All
this is fulfilled at the D20 powder diffractometer of the neutron reactor at ILL. Based on that
background a functional and cheap reaction cell was designed. The hydrothermal autoclave
for neutron diffraction (HAND) allows a fast sample exchange and can easily be fitted to the
D20 or by minor modification to powder diffraction devices at other neutron sources.
.
2. The Neutron Powder Diffractometer D20 at ILL
D20 [Walton and O`Hare, 2000, Hansen et al., 2008] is a medium to high-resolution 2-axis
diffractometer at the high flux reactor source at the ILL in Grenoble, providing a neutron flux
of up to 108 ns-1cm-2 at the sample position. The schematic set up is given in figure 1. A
stationary curved linear position sensitive detector (PSD),consisting of 48 precisely cut
microstrip gas chamber detector (MSGC) plates provides a usable aperture of 153.6° (2Θ).
The polygonal arrangement of the juxtaposed plates enables the continuous and homogeneous
coverage of the whole 2Θ range, each plate covering 3.2°. The gas filling of 3 bar ³He and 1
bar CF4 and the detection gap of 5 cm results in a neutron detection efficiency from 60% (λ=
0.8Å) to 90% (λ=2.4Å).
A vertically focusing monochromator of pyrolytic graphite HOPG (002) in reflection position
offers λ = 2.4 or 2.5 Å at a take-off angle of 42° or 44°. It is equipped with graphite filters to
suppress harmonics. A copper monochromator Cu (200) in transmission gives wavelengths of
λ ≈ 0.82, 0.88, 0.94 or 1.3Å at take-off angles of 26°, 28°, 30° or 42°. At λ = 1.3 Å the
89
monochromatic beam has its highest flux of about 9.8⋅107 n⋅cm-2⋅s-1. Soller collimators allow
to reduce the divergence of the incident polychromatic beam (27’) to α1 = 10’ or 20’.
Beyond that, D20 is equipped with a furnace consisting of a heating element,45 mm in
diameter, made of a vanadium sheet. As vanadium has a very low coherent scattering length
for neutrons this material only adds constant incoherent scattering to the powder diffraction
pattern. This heating device is placed in a large vacuum vessel, to avoid scattering by air and
oxidizing of the vanadium sheet and to improve thermal insulation. It is equipped with
neutron-absorbing B4C screens and a direct beam-stop to avoid neutrons diffracted by the
aluminium walls of vessel to propagate to the detector.
All the above mentioned characteristics enable a large choice in Q-space, resolution,
wavelengths and flux and accomplish high precision in intensity measurements. This makes
D20 adaptable to various levels of crystallographic complexity and rapidity of the observed
phenomenon and therefore an ideal tool for in-situ diffraction studies with time constants
even below a second.
3. The Hydrothermal Autoclave for Neutron Diffraction (HAND)
The hydrothermal autoclave for neutron diffraction was designed to be a simple and cheap
reaction cell fitting to the well-established ILL D20 [ Walton and O’Hare, 2000; Hansen et
al., 2008] station with its vanadium furnace. Changes of samples and apparatus must be
possible fast and easy. Therefore the apparatus is mainly an upright steel tube closed at both
ends. The steam necessary for the hydrothermal reaction is generated inside this tube during
heating, so no separate steam supply is needed. The material chosen for the autoclave is
cobalt-free stainless steal ( 4301, Linster, Aschau). The thickness of the walls is a
compromise between the demands of a stability at an internal pressure of up to 40 bars and
the aim to obtain a maximum penetration of the neutron beam. The schematic diagram of the
90
reaction cell is given in Figure 2. HAND consists of three parts: bottom, sample support and
cover. The bottom is fixed inside the vanadium furnace device below the neutron beam. It
serves as reservoir for D2O and contains the bushings for the internal thermocouples. Inside
the bottom the sample support is placed above the water reservoir. The cover is a tube of 14
cm in length and 2.5cm in diameter which is closed at the upper end. It is simply screwed
upon the bottom and can easily be replaced. Its walls have a narrowing down to 1 mm at the
level of the neutron beam to maximize the intensity of the neutron flow through the sample.
During the first experiments HAND was equipped in addition with a valve to evacuate the
interior but following experiments have shown that this is not necessary. In Contrary the
valve on top causes problems with leak tightness. Therefore we decided to remove it in the
new version (see Fig 2)
HAND is mounted in the vertical axis of the diffractometer inside the furnace device available
at the instrument D20. Due to the geometric arrangement of the there existing neutron option
(fig. 1) the lower limit of the 2Θ range is 8° using HAND with a sample diameter of 20 mm.
Below this angle the obligatory beamstop cuts of the diffracted intensity.
4. Experimental
To proof the applicability of HAND we have chosen the pure system CaO-SiO2-D2O to be
studied first. D2O instead of H2O was chosen due to the lower interaction between deuterium
and neutrons. The fast and easy sample preparation allows us to add different additives to the
system easily. The bulk composition was set to a molar Ca/Si ratio of 0.5 projecting on the
join tobermorite-quartz and a chosen D2O/solids ratio of 0.8 resembles the recipes of
industrially manufactured steam cured building materials [Fehr and Zuern, 2000].
The starting materials were mixed with mortar and pestle, the compound was poured into a
mold (diameter 2cm, height 8cm) and then dried for 1 hour at 60°C to obtain the mandatory
91
solidity to place it on the sample support. For detailed description of used materials and
sample preparation it shall be referred to the studies of Fehr and Zuern (2000) and Bernstein
and Fehr (2010).
The temperature of the sample is controlled and monitored with thermocouples using a WEST
controller. To determine the temperature gradient through the autoclave-cross-section an
additional thermocouple was fixed on the exterior of HAND. Due to the sensitivity of
hydrothermal reactions to temperature this is of great importance. Therefore the adjustment
control parameter were optimized so that the desired temperature was reached after 1 hour
and kept constant for 8 h of reaction time. The rather high background from heavy water and
the presence of semi-crystalline phases demands good counting statistics Therefore the typical
acquisition time for one powder pattern was set to one minute.
To observe a d-spacing up to 11.3 Å, where the (002) reflection of the evolving tobermorite is
expected, the wavelength was set on 2.4 Å instead of 1.3 Å (highest flux). The option of 10’
Soller collimators was abandoned, as this would decrease the intensity by a factor of four
without an enormous gain in angular resolution. This is mainly limited by the large sample
size [Caglioti et al., 1958].
5. Results
Until now HAND was used for several beam times at ILL to perform experiments under
different conditions. The detailed description and discussion of the experiments was not the
main scope of this paper and for further information we would like to draw the attention on
other publications of our workgroup [Fehr & Zuern, 2000; Bernstein & Fehr, 2010]. The
maximum reaction temperature of 190 °C was reached after 60 minutes of up heating (see fig.
3). The monitoring of the inner and outer thermocouple reveals an accuracy in sample
temperature of ± 0.5°C depicted at the bottom part of figure 3. The homogeneity of the
92
sample cross-section and in the field of the incoming beam was verified by electron
microprobe analyses. The reproducibility of standard analyses was <1% for each element
routinely analysed.
The expense of the initial solid phases quartz and portlandite with the reaction time and the
formation of 1.13 nm tobermorite can be observed by the decrease of their Bragg-peaks in the
time-resolved neutron diffraction pattern as demonstrated in Fig. 4 within the range of 40° to
55° 2-Θ.
Sequential fitting of multiple, individual Bragg-peaks of every powder pattern were
performed by a procedure programmed to perform this task from inside the ‘Large Array
Manipulation Program’ (LAMP, http://wwwold.ill.fr/data_treat/lamp/lamp.html), the datavisualization and treatment system used at ILL. To take correctly into account the asymmetry
due to the ‘umbrella’-effect of detecting sections of the Debye-Scherrer cones with a linear
PSD of a certain height, the correction of Finger et al. [1994] was applied to the data. The
instrumental peak-shape of a Bragg-peak was described using a pseudo-Voigt function. The
main diffraction peak of iron (mantle of HAND) did not interfere with any peaks of the
phases of the sample and was used to calibrate the intensities of the phases of interest. The
decrease of the calibrated intensities of quartz and portlandite is depicted in Fig. 4. After 200
min. portlandite was dissolved completely, but crystallization of 1.13nm tobermorite did not
start until 331 min. at 190°C. The amount of quartz did not remain constant after the
consumption of complete portlandite, indicating a reaction of quartz and initially formed
semi-crystalline Ca-rich C-S-H. The first detectable reflections of tobermorite were those of
(hk0) planes, (00l) reflections follow with a time lag of about 60 minutes (see fig. 5).
6. Discussion
The high flux instrument D20 provides a time resolution of one minute for recording one
diffraction pattern with a good peak/background ratio. Each single diffractogram allows an
93
exact determination of the amount of phases and the decrease or increase of phases. The low
scattering of the data on the amount of phases involved indicates, that a detailed kinetic
modeling (e.g. using the model of Chan et al. [1978] or an Avrami-equation [Shaw et al.,
2000] is possible on data obtained by HAND experiments.
The steel used for HAND remains chemically inert during an experiment. An autoclave cell,
designed for D1B diffractometer at ILL by Polak et al. [1990], consisted of aluminum.
Synthesis of CSH-phases in a hydrous atmosphere takes place at high pH-values (>12). At
such conditions metallic aluminum does not behave chemically inert and will be dissolved.
The presence of Al-ions in the system will change the reaction mechanism of 1.13 nm
tobermorite formation [Huber et al., 1998; Mitsuda and Taylor, 1975; Klimesch and Ray,
1999, Bernstein and Fehr. 2010]. In addition, Al3+ is substituted for Si4+ in the tobermorite
structure and the intra-crystalline order state of 1.13nm tobermorite will be changed resulting
in an increase of the reaction rate [Fehr et al., 2003]. The steel used for HAND has the
advantage to behave chemically inert and derived kinetic data correspond to the pure system
SiO2-CaO-D2O. Furthermore, steel is a material cheap and easy to handle in contradiction to
gold-coated Ti-Zr alloys used by Walton et al.[1999].
. The detailed compilation of information obtained by HAND-experiments leads to a better
insight in the reaction kinetics and mechanisms of CSH-formation. Until now a large number
of experiments has successfully be performed studying the pure system and the influence of
different ions like Al3+.or a varrying grain size of quartz [Fehr and Zuern, 2000; Bernstein and
Fehr, 2010] Beyond that, this autoclave offers a multitude of other possible applications in
Geo.- and Material sciences. The mature design of HAND allows an easy adaptation on
powder diffraction devices of other neutron sources assumed that they can provide a sufficient
neutron flux.
94
References
1. Journal article
BERNSTEIN, S.; FEHR, K.T. (2010), “The Formation of 1.13nm Tobermorite under
Hydrothermal Conditions:1. The influence of quartz grain size within the
system CaO-SiO2-D2O” Progress in Crystall Growth and Characterization of
Materials , accepted
CAGLIOTI, G., PAOLETTI, A. and RICCI, F.P. (1958), “Choise of collimators for a crystal
spectrometer for neutron diffraction” Nucl. Instr. and Meth. 35 223-228.
CHAN, C.F., SAKIYAMA, M. and MITSUDA, T. (1978), “Kinetics of CaO-Quartz-H2O
reaction at 120°C in suspension” Cem. Con. Res. 8, 1-6.
FINGER, L.W., COX, D.E. and JEPHCOAT, A.P. (1994) “A correction for powder
diffraction peak asymmetry due to axial divergence,” J. Appl. Cryst. 27, 892-900.
GABROVSEK, R., KURBUS B., MUELLER D. and W. WIEKER (1993) “Tobermorite
formation in the system CaO, C3S-SiO2-Al2O3-NaOH-H2O under hydrothermal
conditions,” Cem. Con. Res. 23, 321.
HANSEN, T.C.; HENRY, P.F., FISCHER, H.E. (2008), “The D20 instrument at the ILL: a
versatile high-intensity two-axis neutron diffractometer,” Measurement Science &
Technology.
KLIMESCH, D.S and RAY, A.S. (1999), “Effect of quartz content in the nature of alsubstituted 11A tobermorite in hydrothermaly treated CaO-Al2O3-SiO2-H2O
system,” Advanc. Cem. Res. 11, 179.
MITSUDA T. and TAYLOR, H.F.W. (1975) “Influence of aluminum on the conversion of
calcium silicate hydrate gels into 11Å tobermorite at 90° and 120°,”Cem. Con. Res. 5,
203-210.
POLAK, E., MUNN, J., BARNES, P., TARLING, S.E. and RITTER, C. (1990), Timeresolved neutron diffraction analyses of hydrothermal Synthesis using a novel
autoclave cell,” J. Appl. Cryst. 23, 258-262.
SHAW, S, CLARK, S.M. and HENDERSON, C.M.B. (2000) “Hydrothermal formation of the
calcium silicate hydrates, tobermorite (Cs5Si6O16(OH)2) 4H2O and xonotlite
(Ca6Si6O17(OH)2): an in situ synchrotron study,”Chem. Geol. 167, 129-140.
WALTON, R.I., SMITH, R.I., MILLANGE, F.,. CLARK, I.J,. SINCLAIR D.C and O'HARE,
D. (2000), An in-situ time resolved neutron diffraction study of the hydrothermal
crystallisation of barium titanate,” Chem. Commun. 14, 1267 – 1268.
WALTON, R.I. and O'HARE, D. (2000), Watching solids crystallise using in situ powder
diffraction, Chem. Commun. 23, 2283 – 2291.
WALTON, R.I., FRANCIS, R.J., HALASYAMANI, P.S., O'HARE, D., SMITH, R.I., DONE, R. and
HUMPHREYS, R.J. (1999) Novel apparatus for the in situ study of hydrothermal
crystallizations using time-resolved neutron diffraction, Rev. Sci. Instr. 70, 3391.
95
2. Selections from an anthology
FEHR, K.T. and ZUERN, S.G. (1997), “Phase relations of 1.13 nm tobermorite, xonotlite,
truscottite and gyrolite under hydrothermal conditions,” in Proc. 5th Int. Symp.
Hydrotherm. Reactions, 225-227.
FEHR, K.T. and ZUERN, S.G. (2000), Mechanisms of calcium-silicate-hydrates under
hydrothermal conditions,” in Proc. 6th Int. Symp. Hydrotherm. Reactions, 278-281.
FEHR, K.T; HUBER, M.; ZÜRN, S.G. and PETERS E. (2003): Determination of the reaction
kinetics and reaction mechanisms of Al-tobermorite under hydrothermal conditions by
in-situ neutron diffraction. In FENG, S.H.; CHEN, J.S. & SHI, Z. (eds.): Hydrothermal
Reactions and Techniques. World Scientific, New Jersey, 19-26
HUBER, M., FEHR, K.T. and ZUERN, S.G. (1998), Kinetische Studien zur Bildung von 1.13
nm Tobermorit unter hydrothermalen Bedingungen,” in: Bauchemie von der
Forschung bis zur Praxis“, Monogr. 11 d. GDCh , 29-32, edited by W. Hiller.
ZUERN, S.G. and FEHR, K.T. (2000a), “Phase relations and thermodynamic properties of
1.13 nm tobermorite and xonotlite,” in Proc. 6th Int. Symp. Hydrotherm. Reactions,
286-289.
3. Computer programs
LAMP,
The
Large
Array
http://wwwold.ill.fr/data_treat/lamp/lamp.html
Manipulation
Program.
96
Figure captions
Figure 1 Schematic diagram of the setup on the ILL D20 station
Figure 2 Schematic diagram of the HAND reaction cell (vertical section).
Figure 3 Cumulative distribution (Q3) and density distribution (q3) of quartz grainsize
Figure 4 Time resolved neutron diffraction pattern in the 2-Theta range 40 – 55 ° at T =
190°C.Time resolution is one minute.
Figure 5 Variation of the integral intensity of portlandite (●)and quartz (○) and the maximum
intensity of tobermorite (002) (∗) with time, temperature profile for the experiment
is shown in the lower part of the diagram.
Figure 6 occurrence of (hk0) and (00l) reflections in the course of the experiment
97
Figure 1
98
Figure 2
99
Figure 3
100
Figure 4
101
Figure 5
102
Figure 6
103
Appendix D
The Formation of 1.13nm Tobermorite under Hydrothermal Conditions:
1. The influence of quartz grain size within the system CaO-SiO2-D2O
Progress in Crystall Growth and Characterisation of Materials, in press
The Formation of 1.13nm Tobermorite under Hydrothermal Conditions:
1. The influence of quartz grain size within the system CaO-SiO2-D2O
S. Bernsteina,*, K.T. Fehra
a
Ludwig Maximilians University, Department of Earth and Environmental Sciences,
Theresienstraße 41/III, 80333 Munich, Germany
1. Abstract
The influence of grain size of quartz on the formation of 1.13nm tobermorite in aerated
autoclaved concrete was investigated by applying in-situ neutron diffraction. Experiments
were performed at 210°C/Psat employing quartz of 8µm and 16µm, respectively. The results
reveal changes in the reaction mechanism from solution control to diffusion control. The grain
size of the quartz fraction clearly influences the occurrence of those changes. Based on those
results an interpretation of former not clearly interpretable quenching experiments was
performed. An interpretation using different reaction mechanisms for those experiments leads
to a coherent picture of the reaction.
Keywords: 1.13nm tobermorite, kinetic, aerated autoclaved concrete, in situ neutron
diffraction
1. Introduction
Calcium Silicate Hydrates (CSH-phases) are very rare in nature but one deploys their
properties in several technical applications. Up to now the main scope lies in the production
of steam cured building materials. For the fabrication of aerated autoclaved concrete
(AAC),one of the most popular building materials in Europe for lightweight mode of
construction, 1.13 nm tobermorite is the predominant phase. The evolved crystal texture
mainly controls the mechanical and thermal properties of the product, like high pressure
resistance and low thermal conductivity.
1.13 nm tobermorite crystallizes in a layered structure, stacked along [001] with a basal
spacing of 1.13 nm. The average structure was described by Hamid [1] but the real structure
was solved by Merlino et al. [2,3] which is based on two polytypic modification of
orthorhombic and monoclinic symmetry leading to a disordered structure (O/D character).
The common structural feature is characterized by infinite silicate double chains of a type
called Dreierdoppelketten built up of condensed dreierketten (kinked to repeat at intervals of
104
three tetrahedra) along [010]. The chains are intercalated by a Ca-O layer (portlandite layer)
so the structure consists of a central layer of calcium octahedra which has silicate sheets on
each side. The calcium octahedra share oxygens with the silicate tetrahedra, the distance
between two edges in the calcium octahedral layer is about the same length as a silicate
dreierketten unit. This type of structural unit is characteristic for most of all CSH-phases. In
1.13nm tobermorite the composite layers of one calcium and two silicate layers are bound
together by an interlayer containing calcium ions and water molecules. The interlayer contains
variable amounts of calcium so that charge balance is achieved by variation of hydrogen
atoms bonded to the silicate chains. Therefore, the variable occupancy of calcium in these
layers allows the Ca:Si ratio to vary from Ca5Si6O16 (OH)2 *4H2O (C5S6H5) to Ca4Si6O(OH)2
*2H2O (C4S6H3) [2,3]. In AAC, 1.13nm tobermorite is close to the composition Ca5Si6O16
(OH)2 *4H2O and occurs in association with semi-crystalline CSH-phases CSH (I ) and CSH
(II) as minor components. In contrast to tobermorite these phases are highly disordered and
display a wide range of compositions. They are classified by their Ca:Si ratio: CSH (I) with a
Ca:Si ratio <1.5 and CSH (II) with a Ca:Si ratio > 1.5 according to Taylor [4,5].
There has been a lot of work in this field aimed at understanding the formation mechanisms
and growth kinetics of CSH-phases [e.g.6,7]. But little quantitative data exist on the kinetics
of 1.13 nm tobermorite formation. In addition there is no accordance on the nature of the
reaction mechanism because some studies proposed being solution controlled and others
being diffusion controlled as pointed out in detail by Klimesch et al. [8].
The reaction mechanism and kinetics of the formation of 1.13 nm tobermorite in the pure
cement-free system CaO-SiO2-H2O from lime, silica and water (CaO + SiO2 + H2O) under
hydrothermal conditions were determined by quenching experiments at 180°-190°C/Psat
[5,6,9] and by an in-situ Neutron diffraction experiment [10] as well. Quenching experiments
reveal the disadvantage of missing data for the early evolution of phases in time and have
prevented a quantitative kinetic description so far.
The formation of tobermorite is, in addition to reaction temperature and the amount of Al in
the initial mixture, mainly affected by the grain size of quartz [8,11] .Therefore, the major aim
of this investigation was to determine reaction mechanism and kinetics of the formation of
1.13 nm tobermorite under hydrothermal conditions as a function of the grain size of quartz.
Experiments were conducted at 210°C employing quartz with a grain size of 8µm and 16 µm,
respectively. In order to avoid quenching effects in short-time runs the experiments were
conducted at in-situ conditions and data will be collected by means of neutron-diffraction
with the HAND apparatus [12].
3. Experimental
In-situ experiments under hydrothermal conditions were conducted at 210°C under saturation
pressure and within a time-range of up to 10 hours by applying a hydrothermal autoclave cell
for neutron diffraction HAND [12] at the D20 powder diffractometer of the high-flux
neutron-source (4.2* 107 n cm-2 s-1) at Institute Laue-Langevin (ILL), Grenoble. In order to
obtain neutron diffraction spectra with a low background noise, hydrogen-free substances
have to be used, and contact of the materials with humidity must be minimized. Thus the
experiments were carried out with heavy water (D2O) instead of H2O as hydrous reactant. To
control the influence of D2O on the kinetics of the tobermorite-forming reaction two
preliminary quenching experiments were conducted in cold seal pressures vessel at 190°C/Psat
for 8 hours, using D2O and H2O as hydrothermal fluid, respectively. Analyses of the final
products by X-rax diffraction demonstrate no differences between both runs as indicated by
similar phase assemblages and phase amounts.
Pure SiO2, CaO and D2O were used as starting materials in order to determine the reaction
kinetics in the simple system CaO-SiO2-D2O according to the reaction:
105
5CaO + 6SiO2 + 5H 2 O ⇒ Ca 5 Si6 O16 (OH )2 ∗ 4 H 2 O (1)
CaO used was produced by heating calcium carbonate (Merck p.a.), having a particle size of
<0.090mm at 1250° C for 14 h. The source of SiO2 was ground Miocene quartz sand (SiO2 >
99.5 wt.-%) supplied by the Quarzwerke Company in Frechen, Germany, revealing a medium
grain size of 8µm (SH500) and 16µm (W12), respectively. Pure deuterium oxide (Merck,
Uvasol, D2O > 99.8 %) was used as liquid reactant. The mixture was prepared at D2O/solids
ratio of 0.8 with a Ca:Si ratio of 0.5.
The experiments were designed to examine in-situ the hydrothermal hardening of lime-silica
based samples in the vapour phase and not in suspension like Shaw et al. [13] corresponding
to the production process of AAC or lime-silicate bricks. Therefore the initial specimen has to
reveal a mechanical stability that is great enough to be placed on the sample support. For this
purposes the raw materials were mixed using mortar and pestle to obtain a homogeneous
paste. The mass was poured into a previously prepared paper wrapper (diameter 1.5 cm,
height 8 cm) and stored at 60° C for 120 minutes in a sealed container. The raised temperature
resulted in an acceleration of the binding process. After this thermal treatment, the solid and
hardened sample was placed on the sample holder in the autoclave. The desired temperature
was reached by 1 hour of controlled heating up and then held for the defined length of time (8
up to 10h). The inner and outer temperature of the sample was monitored and regulated with
thermocouples inside and on the bottom of the sample. The temperature gradient across as
well as along the sample determined in some preliminary tests is 2°C.
The time-resolved neutron diffraction pattern were taken within the range of 8° to 153.6° 2-θ
at λ = 2.4 Å to allow the analysis of d-spacing up to 11.3 Å, where the basal (002) reflection
of the evolving tobermorite is expected. Due to a rather high background from heavy water
and the presence of semi-crystalline phases in the initial steps of the reaction, good counting
statistics are necessary. Therefore the typical acquisition time for one powder pattern was set
to one minute.
The mechanisms of the tobermorite forming reaction can be evaluated on the basis of the
reaction conversion of quartz according to Chan et al (1978)[6] assuming that there are no
seeds in the reactants and the growth rate is low:
1 − (1 − α ) 3 = kt 1 n
1
(2)
where α gives the reaction conversion of quartz, k the reaction constant and t the reaction
time. According to equation 2 the factor n reveals information on the reaction mechanism. If
n=1 the reaction is solution controlled (phase boundary modell), if n=2 the reaction is
diffusion controlled (Jander equation) [14]. Values for α were calculated from the decreasing
integral intensity of the (101)-Bragg reflection of quartz. The integral intensity was obtained
by using the peakfitting-routine “STR_fit” implemented in the Large Array Manipulation
Program (LAMP) provided by the ILL. The diffraction patterns were fitted in the range from
41 to 44 ° 2theta, the peak shape was set to Pseudo-Voigt and a linear background was
chosen. By applying equation 2 to these data and plotting them in logarithmic scale, one
obtains information on the reaction mechanism from the slope of the data points. The
complete expense of portlandite was determined by the disappearance of the (101) reflection
at 54.3° 2Θ. The first occurrence of tobermorite was determined by the appearance of (hk0)
reflections at 46° 2Θ.
4. Results
106
The results for the experiments are depicted in Figure 1a) and b). Both experiments don’t
show a curve linearity but different slopes of 1 and 0.5 within the reaction progress.
According to equation (2) the slope of the data points yields information on the reaction
mechanism. Therefore the reaction can be split into 3 sections with a diffusion or solution
controlled mechanism. The times where the changes in slope occur are labelled here with
t1_transit and t2_transit respectively. Determined transition temperatures, the moment of
portlandite expense and 1.13nm tobermorite occurrence are summarized in Table 1.
For the first experiment (Fig. 1a) a quartz with a mean grainsize of 16µm was used in the
initial mixture. The desired reaction temperature of 210°C was reached after 60 min of
heating up. Up to 30 min (t1_transit) after starting the experiment the reaction is solution
controlled, followed by a diffusion controlled period. After 228 min a further change back to
a solution controlled reaction mechanism can be observed. Portlandite is consumed after 142
minutes of the running experiment within the diffusion controlled section of the reaction. Just
after 250 min, within the second solution controlled period, the first reflections of 1.13nm
tobermorite could be observed.
On the first glance the second experiment using quartz with a mean grain size of quartz at
8µm shows a similar curve progression. The differences becomes just apart if one compares
the transition times and the expense and occurrence of protlandite and tobermorite,
respectively. The reaction starts again with a solution controlled segment but the first change
to a diffusion controlled mechanism already occurs after 18 minutes (t1_transit). The second
change back to solution control is detectable after 180 min, compared to the 16µm experiment
almost 50 min earlier. Also the consumption of portlandite after 60 min and the first
appearance of tobermorite is accelerated. The transition times, the moment of the expense of
portlandit and the occurrence of tobermorite show a clear increasing trend with increasing
grain size of quartz ( Fig.2)
5. Discussion
In situ neutron diffraction experiments conducted in this study at 210°C with two different
grain sizes of quartz revealed a non isokinetic behaviour of the tobermorit forming reaction in
AAC. The reaction mechanism changes from solution control to diffusion control and back to
a solution controlled segment. The duration of these segments is strongly influenced by the
grain size of quartz employed in the initial mixture. If the grain size is decreased to 8µm, the
first solution controlled part is reduced from 30 to 18 minutes. Likewise the following
diffusion controlled stage is just present up to 180 min (t2_transit) compared to 228 min
(t2_transit) in the experiment with the coarser quartz. With the finer quartz portlandite is
expensed earlier and 1.13 tobermorite can be detected already after 140 min in the diffusion
controlled segment of the reaction. Lasaga & Luettge [15] used a modified Gibbs Thomson
equation to describe the dissolution of crystals (equation 3).
∆G = ∆G 0 + RT ln al a s + σ∆A + u (r )∆V
(3)
By neglecting the term for the strain field of dislocation defects, this equation can also be
used to see the influence of different factors to the formation of 1.13nm tobermorite.
This equation shows the dependence of solubility to the activity (a) of the liquid and the solid
reactants and the surface free energy( σ). the surface free energy depends on the grain size of
the solids and the activity (a) on the composition of the reactants (in case of tobermorite if Al
is added to the system. Equation (3) shows a linear correlation between the solubility and the
influencing factors. To generalize this for 1.13nm tobermorite formation, the reaction is
constrained by the grain size of the reactants (this study) and the composition of the initial
mixture (forthcoming paper)
107
With the results from this study one can make interpretations in terms of the reaction path.
The changing reaction mechanism denote that tobermorite is not formed initially but by an
intermediate step. The initial reaction is controlled by the solution of quartz to form
semicrystalline CSH-phases. Such phases are evident by a bumb in the background of the
diffraction patterns. The second step is controlled by the diffusion of quartz through a layer of
semicrystalline CSH-phases and the last step, after the expense of portlandite,is controlled by
the reaction of still existing quartz with the CSH-phases to form tobermorite. This is in good
agreement with results of Fehr et al. [10]
Quenching experiments conducted in the past [6,7,9] could not be interpreted in terms of a
reaction displaying a isokinetic behaviour with just a single slope. Interpretations are often not
unambiguous due to a low density of data points. Furthermore information for the initial part
of the reaction are missing due to lacking short time experiments.
Based on the results of this study some quenching experiments [6,7,9] were recalculated in
terms of reaction conversion of quartz according to equation (2) and reinterpreted. Nonisokinetic reaction mechanisms of the quenching experiments becoming evident by applying
the findings from in-situ experiments as demonstrated in Figure 3. the determined transition
temperatures are summarized in Table 2. The quenching experiments at 180°C/Psat [6,7] are
carried out in the system CaO-SiO2-H2O beyond the stability field of 1.13 nm tobermorite [9].
The experiments of Chan et al. [6] were carried out with different Ca:Si ratios (0.8 and 1.0)
but there are no differences in reaction mechanism as displayed in Figure 2. Their
experiments applying quartz < 10 µm only exhibit a diffusion controlled mechanism and
higher values for the reaction conversion of quartz (α) compared to the results on 8 µm-quartz
used by Klimesch & Ray [7] as shown in Figure 3. The reaction mechanism for 8 µm-quartz
[7] is solution controlled up to 5.0 hours (t3_transit time) followed by diffusion control and
indicating an estimated t2_transit time below 1 hour.
The reaction mechanism for 16 µm-quartz [6] is solution controlled up to 5.2 hours (t3_transit
time) followed by diffusion control and indicating an estimated t2_transit time below 1.5
hours. Finally, the reaction mechanism for 35 µm-quartz [7] is diffusion controlled up to 2.1
hours (t2_transit time) followed by solution control up to 7.2 hours (t3_transit time) reaching
diffusion control at the final stage.
The quenching experiments at 190°C/Psat [9] are carried out in the system CaO-SiO2-H2O
beyond the stability field of 1.13 nm tobermorite [9] and this phase crystallizes metastable.
The quartz used in these experiments displays a grain size of 45 µm [9]. The reaction
mechanism for 45 µm-quartz is diffusion controlled up to 6 hours (t2_transit time) followed
by solution control up to 9.8 hours (t3_transit time) ending in diffusion control at the final
stage. Summarizing the quenching experiments at 180°C/Psat [6,7] no t1_transit time dividing
solution from diffusion control can be observed due to the lack of short-time experiments.
Two additional transition times occur dividing diffusion from solution control (t2_transit time)
and finally solution from diffusion control (t3_transit time).
Comparing the transition times from experiments with different grin sizes of quartz the same
increasing trend with increasing grain size becomes apparent as shown in Figure 4.
The reaction temperature seems to have just a minor influence on the reaction mechanism
though data from experiments with different temperatures are compareable and are plotted in
one diagram ( Fig.3).
Experiments with quartz < 10µm at 180 °C [6] just show a slope indicating a diffusion
controlled mechanism. for the early stage of the reaction, where solution controlled
mechanism is expected, no data are available thus one can just assume the same progress as
found for experiments performed by Klimesch & Ray [7]. In addition data from Chan et al [6]
show higher values for the reaction progress then the one from Klimesch & Ray [7]. This
could be ascribed to the different upheating and the different water /solid ratios in the
experimental setup. Data from the experiment with 35mm quartz at 180 °C are ambiguous but
108
transition times were constructed with regard to the results from in situ experiments. As the
initial step of the reaction is very fast and the density of data points obtained by quenching
experiments is low the t1_transit time can not be determined from those but an additional
change from solution to diffusion control (t3_transit) can be assigned. Due to lacking data for
short time experiments at 180°C with a quartz grain size of 8µm [7] and 10-20µm [6], t2transition times where estimated to be under 0.8 and 1.4h respectively. Comparing the results
for different grain sizes of quartz one can observe the same trend of decreasing transition
times with increasing grain size of quartz (Fig.4) as found by in situ experiments of this study.
The t2_transit times obtained by quenching experiments are much lower than the one from insitu experiments. This can be assigned to the different calcination temperatures for the CaO
used. There is a strong influence of the calcination temperature to the reactivity of CaO as
shown by .Moropoulou et al. [16]. As the reaction is solution driven, the solution of the
reactants plays an important role for the reaction progress. The solution (reactivity) is for both
reactants quartz and CaO is primarily influenced by their specific surface area but for CaO the
reactivity is additionally increased by lower calcination temperatures.
6. Conclusion
The formation of 1.13 tobermorite in AAC can be described as a non isokinetic reaction. One
can observe changes in the reaction mechanism between solution and diffusion controlled
when plotting the reaction conversion of quartz against reaction time in terms of equation (2).
Based on experiments conducted at 210°C with two different grain sizes of quartz it could be
shown that a finer quartz fraction is decreasing the several transition times and therefore
accelerating the reaction due to an increase in reactivity. The moment of portlandite expense
and 1.13 tobermorite occurrence and the related change in reaction mechanism could be
assigned to the reaction path. 1-13nm tobermorite is not formed initially but by an
intermediate step of semicrystalline CSH-phases.
The effect of decreasing transition temperatures when using a smaller grain size of quartz was
also detectable in quenching experiments (6,7) after recalculating them by applying equation
(2). The 1.13 nm tobermorite forming reaction is mainly controlled by the solubility of the
involved phases and their speciation. This solubility is influenced by the specific surface area
of the reactant and in case of CaO on the calcination temperature.
* corresponding author: Tel.: +49 89 2180 4276; fax: +49 89 2180 4176
email address: bernstein@min.uni-muenchen.de
References
[1]
S.A. Hamid, Z. Krist. 154 (1981) 189.
[2]
[3]
[4]
[5]
[6]
S. Merlino, E. Bonaccorsi and Th. Armbruster, Amer.Miner. 84 (1999) 1613.
S. Merlino, E. Bonaccorsi and Th. Armbruster, Eur. J.Min. 13 (2001) 577.
H.F.W. Taylor, J. Chem. Soc. 30 (1950) 82.
H.F.W. Taylor, Proc. Vth Int. Symp. Chem. Cem. Vol II (1968) 1.
C. F.Chan, M. Sakiyama and T. Mitsuda , Cem.Concr.Res. 8 (1978) 1.
[7]
D. S. Klimesch and A.Ray , J.Therm.Anal.Calorim. (2002) 995.
[8]
D. S. Klimesch and A.Ray and B.Sloane , Cem.Concr.Res. 26 (1996) 1399.
[9]
[10]
S.G. Zuern and K.T. Fehr., Proc. Joint ISHR & ICSTR, Kochi (2000) 286.
K.T. Fehr, M: Huber, S.G. Zuern and T. Hansen, Proc. 5th ICSTR (2002) 37.
109
[11]
N.Isu, H. Ishida and T. Mitsuda , Cem.Concr.Res. 25 (1995) 243-248
[12]
S.G. Zuern, K.T. Fehr and T. Hansen, Proc. 5th ICSTR (2002) 33-36
[13]
[14]
S. Shaw, S.M. Clark and C.M.B. Henderson, Chem. Geol. 167 (2000) 129.
J.D. Hancock and J.H. Sharp, J. Amer.Ceram.Soc. 55 (1972) 74.
[15]
[16]
C. Lasaga and A.Luettge, Eur.J.Mineral. 15 (2003) 603.
A. Moropoulou, A. Bakolas, E. Aggelak, Cem.Concr.Res.31 (2001) 633.
Tables
Table 1: net weight, determined transition times and points of portlandite out and tobermorite
in
experiment.
W12
W12_210
g
16
-
SH500_210
-
16
SH500 CaO
g
g
t1
min
t2
min
portlandite
out
tobermorite
in
8
30
228
142
250
8
18
180
60
140
Table 2: employed grain size of quartz and reaction temperatures and determined transition
times of recalculated quenching experiments from literature
experiment
grain size
temperature
(µm)
(°C)
1
K&R02
8
180
35
180
K&R02 1
16
180
C&M78 2
1
2
Klimesch &Ray (2002) [7]; Chan et al. (1978)[6]
t2
(h)
~0.8
2.1
~1.4
t3
(h)
5.0
7.2
5.2
110
Figures:
Figure 1:
Reaction conversion of quartz in the system CaO-SiO2-D2O at 210°C/Psat and employing
quartz of 16µm (a) and 8µm (b) according to equation (2)
Figure 2:
111
Grain size dependency of t1_transit (squares) and t2_transit (circles) determined from in situ
experiments
Figure 3:
Recalculated data from quenching experiments [6,7,9] in terms of equation (2) assigned
transition times are displayed in boxes
112
Figure 4:
Grain size dependency of t2_transit (circles) and t3_transit (squares) determined from
recalculated quenching experiments [6,7]
113
Appendix E
The hydrothermal formation of 1.13nm tobermorite within the system CaOSiO2-D2O: a kinetic study by in situ neutron diffraction
submitted to: Cement and Concrete research (under review)
The hydrothermal formation of 1.13nm tobermorite within the system CaO-SiO2-D2O: a
kinetic study by in situ neutron diffraction
Saskia Bernsteina,*, Karl Thomas Fehra
a
Ludwig Maximilians Universität München, Department of Earth and environmental
Sciences, Theresienstrasse 42/III, 80333 München, Germany
*corresponding
author:
Fax:00498921804176
saskia.bernstein@gmx.de;
Tel.:
00498921804276,
Abstract
1.13 nm tobermorite was synthesized hydrothermal within the system CaO-SiO2-D2O
simulating the conditions during the hydrothermal hardening of aerated autoclaved concrete.
The reaction was monitored in situ by neutron diffraction at the D20 powder diffractometer of
the ILL (Grenoble) and the influence of reaction temperature ( 170-210 °C) and grain size of
quartz (8 and 16µm) was investigated. The experiments revealed the non isokinetic nature of
the mineral forming reaction with changes in the dominant reaction mechanisms between
solution and diffusion control. Based on these insights rate constant (9.924·10-5 – 5.5·10-3 s-1)
and for the first time activation energies (16.5-33.8 kJ/mol*K) were calculated.
keywords: 1.13nm tobermorite, neutron diffraction, reaction mechanism, kinetics
Introduction
The group of Calcium Silicate Hydrates (CSH-phases) is a comparatively rare mineral family
with 40 members known from natural sources, nevertheless their synthetic equivalents are
114
used for a multitude of applications. One of the phases of most interest is the 1.13nm
tobermorite (Ca5Si6O16 (OH)2 *4H2O) which is known to be formed during the hydrothermal
hardening of aerated autoclaved concrete (AAC), a widely-used building material for light
weight constructions. In consequence of the rapid increase in applications of such materials
during the last 10 years a strong need of more detailed scientific research arose
simultaneously. Fundamental knowledge on the nature of CSH-phases had been given by
Taylor [1] with his studies on portland cement phases but there is still a demand for further
investigations. The existence of various poorly ordered and metastable phases in the CSHsystem hinders experimental work, thus the thermodynamics, kinetics and structural features
of 1.13nm tobermorite and its neighbours are still poorly understood. The knowledge of these
properties is of essential importance as the mechanical properties of the aforementioned
building materials are strongly dependent from the type, amount and texture of the evolving
CSH-phases.
For the tobermorite-family three members are known, the 0.9 nm tobermorite or riversideite,
the 1.13 nm tobermorite or tobermorite sensu stricto and the 1.4 nm tobermorite also called
plombierite. The numerical value in the name indicates the different d[002]-spacing due to
different water contents in the structure. Tobermorite 1.4nm transforms into the 1.13 nm one
by heating to 100 °C, further heating up to 300 °C leads to the 0.9nm tobermorite due to
proceeding dehydration [2]. Some tobermorites are known not to shrink on dehydration and
are therefore called “anomalous [3]. Recently a new member is described, crystallizing in
monoclinic symmetry and hence called clinotobermorite. This polytype was first found in
Fuka (Japan) [4] and also at Wessels mine (South Africa) [5]. The d-spacing in [001] is
similar to the one of 1.13 nm tobermorite which is in focus of this study. An accurate
description of the tobermorite structure in general and the one of 1.13 nm tobermorite in
particular is complicated by presence of structural disorder evidenced by diffuse streaks or
spots in X-ray or electron diffraction patterns [6]. A first structural model was given by
115
Megaw & Kelsey [7] in which they just outlined the main modules of the average structure
without presenting any quantitative structural data. First cell dimensions are given by Hamid
[6], determined by single crystal diffraction on a specimen from Zeilberg (Germany).
According to the most recent structural refinement the structure of 1.13 nm tobermorite
consists of layers of Ca2O7-polyhedras flanked on both sides by wollastonite-like SiDreiereinfachketten running along [010] [8] as the basic structural unit.
Two alternative occupied positions of the silicate chains shifted by b/2 and two opposite
orientations of the bridging tetrahedron in the chain [6] lead to long range stacking disorder in
both natural and synthetic 1.13nm tobermorite. Merlino et al. [2;8] successfully described the
order-disorder character of 1.13nm tobermorite (normal and anomalous) by means of ODtheory [9]. Moreover they have shown that the difference in the thermal behaviour is not
related to a different arrangement of the tetrahedral chains [10] but rather by the presence of
interlayer Si-O-Si linkage in the anomalous tobermorites and their absence in the normal
ones.
In 1.13nm tobermorite the composite layers of one calcium and two silicate layers are bound
together by an interlayer containing calcium ions and water molecules. The interlayer contains
variable amounts of calcium so that charge balance is achieved by variation of hydrogen
atoms bonded to the silicate chains. Therefore, the variable occupancy of calcium in these
layers allows the Ca:Si ratio to vary from Ca5Si6O16 (OH)2 *4H2O (C5S6H5) to Ca4Si6O(OH)2
*2H2O (C4S6H3) [8;11]. In AAC, 1.13 nm tobermorite is close to the composition Ca5Si6O16
(OH)2 *4H2O and occurs in association with semi-crystalline CSH-phases CSH (I ) and CSH
(II) as minor components. In contrast to tobermorite, these phases are highly disordered and
display a wide range of compositions. They are classified by their Ca/Si ratio: CSH (I) with a
Ca:Si ratio <1.5 and CSH (II) with a Ca/Si ratio > 1.5 according to Taylor [11;12].
There has been a lot of work in this field aimed at understanding the formation mechanisms
and growth kinetics of CSH-phases [13; 14]. But little quantitative data exist on the kinetics
116
of 1.13 nm tobermorite formation. In addition there is no agreement on the nature of the
reaction mechanism. Some studies proposed it being solution controlled and other authors
determined a diffusion controlled mechanism. as pointed out in detail by Klimesch et al.[15].
The reaction mechanism and kinetics of the formation of 1.13 nm tobermorite in the pure
cement-free system CaO-SiO2-H2O from lime, silica and water under hydrothermal
conditions were determined by quenching experiments at 180°-190 °C/Psat [12;13;16] and by
an in-situ Neutron diffraction experiment [17] as well. Quenching experiments reveal the
disadvantage of missing data for the early evolution of phases in time and have prevented a
quantitative kinetic description so far.
The major aim of this study was to determine the influence of reaction temperature and quartz
grain size on the formation of 1.13 nm tobermorite in terms of reaction mechanism and
reaction rate. Therefore, the reaction mechanism was determined by in situ neutron diffraction
experiments at different temperatures (170,190,210 °C) and a varying grain size of the quartz
component (16 and 8 µm). Based on this experiments, reaction constants and activation
energies were calculated.
Methods and experimental setup:
Pure SiO2, CaO and D2O were used as starting materials to prepare the samples for neutron
diffraction. In order to obtain a low background noise in the diffraction patterns hydrogen free
substances are necessary. Therefore heavy water (Merck, Uvasol, >99.8%) was used as liquid
reactant. CaO used was produced by heating calcium carbonate (Merck p.a.), having a particle
size of <0.090 mm at 1250 °C for 14 h. The source of SiO2 was ground Miocene quartz sand
(SiO2 > 99.5 wt.-%) supplied by the Quarzwerke Company in Frechen, Germany, revealing a
medium grain size of 8 µm (SH500) and 16 µm (W12) and a specific surface of 0.9 m²/g and
117
1.6 m²/g, respectively. The grain size and the specific surface was determined by Gasabsorption (BET) at Quarzwerke Frechen.
The solid initial materials were mixed well using mortar and pistil then the required amount
of D2O for a water/solid ratio of 0.8 was added. The CaO/SiO2 ratio was set to 0.5, the exact
initial weight can be taken from Table 1. The compound was poured into a mould and stored
in a drying furnace for 60 minutes to obtain a cylindrical greenbody of 8 cm in length and 1.5
cm in diameter of sufficient solidness.
For neutron diffraction experiments a hydrothermal autoclave was designed consisting of a
steal tube and a bottom part with the sample support and the D2O reservoir as described by
Zürn et al. [18] with modifications concerning the housing of the thermocouples and the
evacuating valve (see Fig. 1). The temperature during the experiments was controlled and
regulated by two thermocouples on the bottom and inside the sample. The whole cell was
heated using a vanadium furnace. The desired temperatures of 170, 190 or 210 °C were
reached after a heating period of 60 minutes and subsequently the real experiment started.
After the experiment was completed, the furnace was turned off for cooling down. The
wavelength was set to 2.4 Å by applying a pyrolitic graphite monochromator (HOPG (002))
at a take-off angle of 42°. This set up leads to an effective neutron flux of 4.2x107 n s-1cm-2
which provides the possibility to measure a wide 2Θ range (8-156 °2Θ) accompanied by a
sufficient time resolution of one diffraction pattern per minute and good counting statistics. A
wide 2Θ range is important especially for the low values of 2Θ to record also the basal
reflections of 1.13 nm tobermorite which are expected at d-values of 11.3 Å. Once the
experiment was succesfully finished, the now hydrothermal hardened sample and the
remaining eluate was kept for possible further investigation.
The obtained enormous amount of neutron diffraction data of up to 600 patterns for
experiments conducted at 170 °C/Psat were handled with the LAMP software provided by
ILL [19].
118
The time-resolved neutron diffraction patterns were taken within the range of 8° to 153.6° 2-θ
at λ = 2.4 Å to allow the analysis of d-spacing up to 11.3 Å, where the basal (002) reflection
of the evolving 1.13 nm tobermorite is expected. The mechanisms of the 1.13 nm tobermorite
forming reaction can be evaluated on the basis of the reaction conversion of quartz according
to Chan et al [13] assuming that there are no seeds in the reactants and the growth rate is low:
1 − (1 − α ) 3 = kt 1 n
1
(1)
where α gives the reaction conversion of quartz, k the reaction constant and t the reaction
time. According to equation 1 the factor n reveals information on the reaction mechanism. If
n=1 the reaction is solution controlled (phase boundary model), if n=2 the reaction is
diffusion controlled (Jander equation) [20]. Values for α were calculated from the decreasing
integral intensity of the (101)-Bragg reflection of quartz. The integral intensity was obtained
by using the peak fitting-routine “STR_fit” implemented in the Large Array Manipulation
Program (LAMP) provided by the ILL [19]. The diffraction patterns were fitted in the range
from 41 to 44 °2Θ, the peak shape was set to Pseudo-Voigt and a linear background was
chosen. By applying equation (1) to these data and plotting them in logarithmic scale, one
obtains information on the reaction mechanism from the slope of the data points. The
complete consumption of portlandite was determined by the disappearance of the (101)
reflection at 54.3° 2Θ. The first occurrence of 1.13 nm tobermorite was determined by the
appearance of (hk0) reflections at 46° 2Θ.
Results:
The results for the reaction conversion of quartz in terms of equation (1) for each experiment
are depicted in Figure 2.
119
What immediately becomes apparent is that the experiments cannot be described applying one
slope over the whole reaction progress. The experiments can be divided into different
segments and compared to each other by the changing transition times from one segment to
the other. The different transition times for all experiments are summarized in Table 2.
(Table 2 here)
The changes in slope could now lead to the assumption that the chosen kinetic model of Chan
[13] is not valid for the investigated reaction. However, if one surveys the single segments of
each curve they are either describable with a slope of 1 or 0.5 referring to an exponent of n= 1
or n= 2 in equation (1), respectively. Interpreting this in terms of the reaction mechanism it
implies changing reaction mechanisms with the reaction progress. This is also confirmed by
the changing Ca/Si ratio during the proceeding reaction depicted for the run at 190 °C/ Psat
and16mm quartz (Fig. 3). First the Ca/Si ratio of the evolving products increases to a
maximum of 1.4 after 230 min and then decreases again and converge to the theoretical Ca/Si
ratio of 1.13 nm tobermorite after 500 min of reaction [17]. The same effect of first increasing
and then decreasing Ca/Si Ratio was described by Klimesch & Ray [14]
For all experiments one can observe changes from solution controlled mechanisms to
diffusion controlled ones. An additional change back to a solution controlled part is present
for experiment 1, 2, 3 and 4 (Table 1). Comparing the transition times, the consumption of
portlandite and the occurrence of 1.13 nm tobermorite, one can notice a strong influence of
the employed grain size of quartz and the reaction temperature.
For the experiment conducted at 210 °C/Psat (Exp 3 in Table 1, Fig. 2e), applying a grain size
of quartz of 16 µm, the first transition from a solution to a diffusion controlled mechanism
occurs after 30min and the second one back to solution control after 228min. Portlandite is
expensed after 142 min within the diffusion controlled part and first reflexions of 1.13 nm
tobermorite could be assigned after 250 min. Using the fine grained quartz (8µm) at the same
conditions, (exp.6 in Table 1 and 2, Fig. 2f) a considerable acceleration of the reaction can be
120
detected. The first transition occurs at 12min and the second at 150 min of reaction time, 18
and 78 min earlier as in experiment 3. Likewise, portlandite is already expensed after 60 min
and 1.13 nm tobermorite occurs after 140min, at the end of the diffusion controlled part.
Plotting the transition times against the reaction temperature a trend of increasing transitions
times with increasing temperature becomes apparent for the experiments employing quartz 0f
16 µm grain size (Fig 4a). T1_transit ascends from 10 min at 170 °C to 30 min at 210 °C
(Table 2). The influence of increasing temperature becomes more distinct at t2_transit, it
increases from 30 min at 170 °C over 190 min at 190 °C up to 230 min at 210 °C (Fig 4a,
Table 2).
Whereas the times of portlandite consumption and 1.13 nm tobermorite occurrence shows an
opposite trend. The time decreases from 254 min and 374 min at 170 °C/Psat to142 min and
250 min at 210 °C/Psat with decreasing temperature, respectively (see Fig. 4b, Table 2). One
could also detect a temperature dependence of the length of the diffusion controlled segment.
An increase in temperature leads to an extension of the period of diffusion control (Fig. 4a).
Experiments employing the 8 µm quartz ( Exp. 4,5,6 in Table 1 and 2, Fig 2 b,d,f) show a
slightly different behaviour.
What stands out is the missing second change back to a solution controlled mechanism for the
experiments at 190 and 170 °C/Psat (exp. 4,5 in Tab 1&2, Fig. 2d,f). Hence, just t1_transit
could be determined for all three experiments revealing a strong decrease from 93 min over
65 min down to 12 min with rising temperature (see exp 6,5,4 in Table 2, Fig.5). Likewise,
the consumption of portlandite is accelerated from 70 to 60 min with increasing reaction
temperature (see exp 4,5,6 in Table 2; Fig 5). The same trend can be determined comparing
the first occurrence of 1.13 nm tobermorite at 87min at 190 °C/Psat and 105 min at 170 °C/Psat
(exp 4 and 3 in Table 2, Fig. 5). For the experiment at 210 °C/Psat, a change back to a solution
controlled reaction can be detected and the occurrence of 1.13 nm tobermorite seems to be
121
retarded (140 min) compared to the experiments at lower reaction temperatures (see exp. 6 in
Table 2, Fig 5).
Rate constants for the overall reaction progress were calculated using equation (1) assuming
slopes of 1 (n = 1) and 0.5 (n = 2) for a solution and diffusion controlled reaction mechanism,
respectively (9.924·10-5 – 5.5·10-3 s-1,see Table 2).
The change of the rate constants with increasing specific surface for the three investigated
reaction temperatures clearly shows an increase as depicted in Figure 6. Based on the
calculated rate constants at three different temperatures, activation energies (EA) can be
determined according to the Arrhenius equation (2) by plotting the data according to equation
as follows
k = A⋅e
EA
− RT
(2)
where k is the rate constant, EA the activation energy, T the temperature in Kelvin, R the gas
constant and A the pre-exponential factor. The pre-exponential factor A is equivalent to the
amount of collisions between reactants with energies higher than the activation energy. Its
temperature dependence can be neglected due to the fact that the exponential part changes
much stronger with temperature. From experiments with coarse quartz EA could be
determined for the three different segments of the reaction. Whereas for the experiments with
fine quartz, EA could just be determined for the first solution controlled segment, due to the
fact that just one experiment shows two changes in the reaction mechanism. Experiments
using fine quartz reveal remarkably lower values for EA of 16.5 kJ/mol*K compared to 30.8
kJ/mol*K for the ones using coarse quartz (Fig. 7).
122
Discussion
Neutron diffraction experiments were conducted to determine the reaction kinetics of 1.13nm
tobermorite formation in the system CaO-SiO2-D2O. The varying slopes for the reaction
conversion of quartz after applying the kinetic model of Chan et al. [13] indicate a change in
the present reaction mechanism between solution and diffusion control during the proceeding
reaction. The initial step of the reaction is controlled by the solution of quartz and its reaction
with portlandite, leading to the formation of a layer of semicrystalline CSH-phases
surrounding the quartz grains as shown in Figure 8 for AAC steam cured at 190 °C/Psat,
previously published by Zürn et al [18]. The second part of the reaction is controlled by the
diffusion through this layer of CSH-phases and portlandite is expensed completely, however,
the two experiments applying the 8µm quartz conducted at 170 and 190 °C/ Psat do not show
further changes in the reaction mechanism. In contrast, all other experiments of this study
show a second change back to diffusion control (see Fig. 2). If this second change is present
the first reflections of 1.13 nm tobermorite occur within the last diffusion controlled segment
of the reaction. Implying that 1.13 nm tobermorite is formed by the reaction of quartz with the
primarily formed semicrystalline CSH-phases.
This acceleration of the reaction can be explained by the increase in specific surface of quartz
with decreasing grain size. A 50% reduction of the mean grain size results in a 44% increase
of the reactive surface. An increase in specific surface does not only accelerate the reaction,
but it also has a strong influence on the present reaction mechanisms as well, demonstrated by
the missing second diffusion controlled segment in the experiments at 170 and 190 °C/Psat
applying the 8µm quartz. At 210 °C/ Psat with a grain size of quartz of 8 µm again two
changes are present.
Elevating the reaction temperature using 16µm quartz causes an increase in the transition
times between the different reaction mechanisms. The period of diffusion control seems to
123
extend (see Fig 4a), whereas the consumption of portlandite and the formation of 1.13 nm
tobermorite is accelerated by rising the reaction temperature (Fig. 4b).
In experiments with the fine quartz fraction, t1_transit clearly shows a decrease by increasing
the reaction temperature (Fig. 5). The time needed to consume portlandite is as well decreased
by elevating the reaction temperature but this trend could not be detected for the 1.13 nm
tobermorite occurrence. At first the formation of 1.13 nm tobermorite was slightly accelerated
by increasing the temperature but at 210 °C it could not be detected before 140 minutes of
reaction. This means a retarding of tobermorite formation at high temperatures using fine
quartz powder. This effect can be attributed to the metastable formation of 1.13 nm
tobermorite beyond its stability field under these conditions [16]. Comparing the increasing
intensities of 1.13 nm tobermorite hk0 and 00l reflections, it stands out that the (hk0)
reflections occur before the (00l) reflections (see Fig: 3). This can be interpreted in terms of
how the crystallization of 1.13 nm tobermorite takes place. It starts with the of nanoscale abplanes which then, at a later stage of the reaction, start to stack along [001] to form the
characteristic lathlike tobermorite crystals.
With decreasing grain size of quartz the calculated rate constants are increasing at a given
temperature(Fig. 6). Based on these data, activation energies for the pure system CaO-SiO2D2O were calculated for the first time. The calculated values (Table 2) for coarse quartz
experiments, 30.8 kJ/mol*K (solution controlled segment) and 33.8 kJ/mol*K (diffusion
controlled segment), are within the range of activation energies for the Al-bearing system
determined by Shaw et al. [21]. Activation energies for experiments using fine quartz (16.5
kJ/mol*K, Tab. 2) are considerably below those values. This may be due to the fact that Shaw
was assuming an isokinetic reaction and applying a slope of 1 to the Arrhenius equation.
There exist several studies trying to determine the reaction kinetics of 1.13 nm tobermorite
and describe the influence of the reaction temperature and grain size of quartz by performing
quenching experiments. Those experiments were not interpretable at all with the hitherto level
124
of knowledge of an isokinetic reaction. Now, taking into account that the reaction mechanism
changes between solution and diffusion control, this is possible. Based on the results and new
insights in the1.13 nm tobermorite forming reaction kinetic data obtained from quenching
experiments [14;22,23] were recalculated in terms of equation (1,) applying two different
slopes (n=1 and n=2). The transition temperatures were determined (Fig.9) and interpreted
concerning to the grain size and temperature dependency (Fig.10). The results are in good
agreement with findings of this study, showing an increase of reaction time with increasing
grain size of quartz.
Conclusion:
It could be shown that in-situ neutron diffraction is a very suitable method to investigate the
kinetics of the 1.13 nm tobermorite formation. For the first time the non isokinetic behaviour
of the reaction could be evidenced by combining the high intensity of the D20 powder
diffractometer at ILL together with an improved hydrothermal autoclave (HAND, Fig. 1)
allowing constant reaction conditions and a fast and easy sample exchange. Furthermore exact
times for the transition and the consumption of portlandite and the occurrence of 1.13 nm
tobermorite could be determined. Based on the data obtained by applying the kinetic model of
Chan et al. [13] on the values for the overall reaction progress rate constants could be
determined for the first time. Likewise Shaw et al. [21] calculated rate constants for the
tobermorite forming reaction but did not interpret their date in terms of the present reaction
mechanism. By conducting experiments at three different temperatures, the temperature
dependence and hence activation energies could be determined. The results of this study yield
detailed kinetic data on the 1.13 nm tobermorite formation, which were just insufficient
investigated in the past. Beyond that the conducted experiments yield information on the
crystallisation path of 1.13 nm tobermorite as well. These data give a better understanding of
the processes present during the production of AAC and could help to optimize production
125
conditions and recipes resulting in shorter production times and an optimal exploit of the
available resources.
References:
[1] Taylor, H.F.W. ‘The Chemistry of Cements, Ed., Academic Press, London, 1992.
[2] Merlino S., Bonaccorsi E. and Armbruster Th., The real structure of tobermorite 11
angstrom: normal and anomalous forms, OD character and polytypic modifications,
Eur. J.Min. 13 (2001) 577-590
[3] Mitsuda T. & Taylor H.W.F., Normal and anomalous tobermorites, Min. Mag.42 (1978)
229-235,
[4] Henmi C. & Kusachi I., Monoclinic tobermortie from Fuka, Journal of the Japanese
Association of Mineralogists, Petrologists and Economic Geologists 84 (1989) 374-379.
[5] Hofmann C. & Armbruster T., Clinotobermorite, Ca5[Si3O8(OH)]2.4H2O Ca5[Si6O17].5H2O, a natural C-S-H(I) type cement mineral: determination of the substructure,
Z. Kristallogr. 212 (1997) 864–873.
[6] Hamid S.A., The crystal structure of the 11Å natural tobermorite Ca2.25[Si3O7.5(OH)1.5].
1H2O, Z. Kristallogr. 154 (1981) 189–198.
[7] Megaw H.D., Kelsey C.H., Crystal structure of tobemorite, Nature 177 (1956) 390–391.
[8] Merlino S., Bonaccorsi E. and. Armbruster Th, Tobermorites: Their real structure and
order-disorder (OD) character, Amer.Miner. 84 (1999) 1613-1621.
[9] Dornberger-Schiff K.:Grundzüge einer Theorie von OD-Strukturen aus Schichten,
Abhandl. d. Deutsch. Akad. d. Wiss. zu Berlin. Klasse f. Chemie, Geol. u. Biol., 3 (1964) 1–
107.
10] Wieker, W., Grimmer, A.-R., Winkler, A., Mägi, M., Tarmak, M., Lippmaa, E., Solid
state high-resolution 29Si NMR spectroscopy of synthetic 14Å, 11 Å and 9Å tobermorites.
Cem. Concr. Res., 12 (1982) 333–339.
[11] H.F.W. Taylor, Hydrated calcium silicates. Part I. Compound formation at ordinary
temperatures, J. Chem. Soc. 30 (1950) 3682.
[12] H.F.W. Taylor, Proc. Vth Int. Symp. Chem. Cem. Vol II (1968) 1-26
[13] Chan F., Sakiyama M. and Mitsuda T., Kinetics of the CaO-quartz-H2O reaction at
120°C to 180°C in suspension, Cem.Concr.Res., 8 (1978) 135-138.
[14] Klimesch D. S., Ray A., Effects of silica reactivity on the nature and formation of Al1.1nm tobermorite, J.Therm.Anal.Calorim., (2002) 995-1003.
126
[15] Klimesch D. S.; Ray A.; Sloane B., Autoclaved cement-quartz pastes: The effects on
chemical and physical properties when using ground quartz with different surface areas .1.
Quartz of wide particle size distribution, Cem.Concr.Res. 26 (1996) 1399-1408.
[16] Zuern S.G. & K.T. Fehr., Phase Relations and Thermodynamic Properties of
1.13 nm Tobermorite and Xonotlite, Proc. Joint ISHR & ICSTR Kochi, (2000) 286-289
[17] Fehr K.T., Huber M, Zuern S.G.; Hansen T., Determination of the reaction kinetics and
reaction mechanisms of 1.13 nm tobermorite by means of in-situ neutron diffraction, Proc. 5th
ICSTR, (2002) 37-40
[18] Zürn S., Fehr K.T., Hansen T., Design, Technique and Applicability of a Hydrothermal
Autoclave for Neutron Diffraction (HAND) for Analyzing the Reaction Process of Steam
Cured Building Materials Proc. Vth ICSTR East Brunswick, (2002) 33-36.
[19] LAMP, the Large Array Manipulation Program.
http://www.ill.fr/data_treat/lamp/lamp.html
[20] D. Hancock and J.H. Sharp, Method of Comparing Solid-State Kinetic Data and Its
Application to the Decomposition of Kaolinite, Brucite, and BaCO3, J. Amer.Ceram.Soc. 55
(1972) 74-77.
[21] Shaw, S., Clark S. M., Hydrothermal formation of the calcium silicate hydrates,
tobermorite (Ca5Si6O16(OH)2* 4H2O) and xonotlite (Ca6Si6O17(OH)2): an in situ synchrotron
study, Chemical Geology 167(1-2) (2000) 129-140.
[22] Zürn, S. Fehr, K.T., Kinetic study on the crystallisation of calcium-silicatehydrates
in steam cured building materials, Intern. Symp. Hydrotherm. Reactions,
Gatlinbrurg, (1997).
[23] Chan, F., Mitsuda, T., Formation of 11 Å tobermorite from mixtures of lime and
colloidal silica with quartz, Cem.Concr.Res., 8 (1978) 135-138.
127
Table 1
Reaction conditions, composition, employed grain size of quartz and corresponding specific
surface of the conducted experiments.
run
temperature
1
2
3
4
5
6
(°C)
170
190
210
170
190
210
quartz grain size
(µm)
16
16
16
8
8
8
specific surface
(m²/g)
0.9
0.9
0.9
1.6
1.6
1.6
SiO2
(g)
20
30
30
16
16
16
CaO
(g)
10
15
15
8
8
8
W/M
0.8
0.8
0.8
0.8
0.8
0.8
Table 2
Transition times, portlandite expense, tobermorite occurrence and calculated rate constants
and activation energies for the conducted experiments
ru
n
1
2
3
4
5
6
t1_transit t2_transit port_ou
(min)
(min)
t (min)
15
42
30
93
65
12
36
150
228
150
254
120
142
70
64
60
tob_in
(min)
374
360
250
105
87
140
rate
rate
EA
constants P1 constants P2 solution
(s-1, x 10-5)
controlled
(s-1, x 10-3)
part
(kJ/mol*K)
5.529(39)
0.885(03)
4.754(14)
2.398(01)
30.8(16.04)
7.667(10)
3.343(02)
6.882(08)
4.863(03)
9.924(17)
5.599(02)
16.5(11.15)
8.935(21)
2.651(01)
EA
diffusion
controlled part
(kJ/mol*K)
33.8(10.68)
-
128
Figure headings
Figure 1:
mechanical drawing of the hydrothermal autoclave for neutron diffraction designed for this
study
Figure 2
Consumption of quartz versus reaction time for different reaction temperatures and grain sizes
of quartz in terms of equation (1) (Chan et al., 1978) for the conducted experiments. Time of
portlandite expense and 1.13nm tobermorite occurrence are marked by arrows, slopes with n=
1 or n=2 indicating a solution or diffusion controlled reaction mechanism, respectively.
Figure 3
Changing Ca/Si ration with increasing reaction time for the run at 190°/ Psat and 16µm grain
size of quartz after [17]
Figure 4
Temperature dependence of t1_transit and t2_transit (a) and portlandite expense and
tobermorite occurrence (b) for experiments conducted with 16µm grain size of quartz
Figure 5
Temperature dependence of t1_transit (small graphic) ,portlandite expense and tobermorite
occurrence (b) for experiments conducted with 8µm grain size of quartz
Figure 6
Rate constants versus specific surface for the first solution controlled part of the reaction
Figure 7
calculated activation energies (EA) versus specific surface for the solution controlled segment
of the reaction.
Figure 8:
Backscattered electrons image of AAC showing a large quartz grain (qz) next to
semicrystalline CSH-phases (CSH) and 1.13nm tobermorite (Tob) crystallizing into the gap
between qtz and CSH.
Figure 9
Recalculated quenching experiments [14;22,23] according to equation (1) ,determined
transition times are given in boxes
Figure 10
129
transition time versus grain size of quartz from recalculated quenching experiments
[14,22,23]
Figure 1
130
Figure 2
131
Figure 3
Figure 4
132
Figure 5
Figure 6
133
Figure 7
Figure 8
134
Figure 9
Figure 10
Download PDF
Similar pages