Thesis Ezio Leone
Thesis Ezio Leone
Thesis Ezio Leone
Ezio Leone
Chemistry
October 2018
ii
Contents
Resumo ....................................................................................................................................... 9
Abstract .................................................................................................................................... 10
I. THEORY .......................................................................................................................... 11
Chemometrics....................................................................................................................... 11
Introduction ...................................................................................................................... 11
Design of Experiment....................................................................................................... 13
Mechanism of Adhesion............................................................................................... 25
Adjustability ................................................................................................................. 42
iii
Ingredients ........................................................................................................................ 43
Binder ........................................................................................................................... 43
Aggregates .................................................................................................................... 52
Redispersible Polymer.................................................................................................. 52
Fibers ............................................................................................................................ 54
Rheology .............................................................................................................................. 56
Rheological Models.......................................................................................................... 58
Viscosity ....................................................................................................................... 58
Viscoelasticity .............................................................................................................. 59
Materials ............................................................................................................................... 67
iv
Smooth Body Tiles ....................................................................................................... 67
Software ............................................................................................................................... 69
Instruments ........................................................................................................................... 70
Procedures ............................................................................................................................ 82
Evaluation of Deformability............................................................................................. 86
Sample Loading............................................................................................................ 87
v
Design of Experiment Results ............................................................................................ 103
24 h ................................................................................................................................. 120
At 5’............................................................................................................................ 132
vi
At 10’.......................................................................................................................... 133
At 15’.......................................................................................................................... 134
vii
Stress Sweep: G’ ½ ........................................................................................................ 166
viii
Resumo
Realizei meu projeto de tese no laboratório de P&D de adesivos de cimento da Mapei SpA,
empresa líder no campo de adesivos, selantes e produtos químicos para construção. O objetivo
do trabalho foi aplicar ferramentas quimiométricas para o estudo do comportamento mecânico
e reológico de adesivos à base de cimento em relação à sua formulação.
Em particular, dois Designs Experimentais foram definidos : um fatorial fracionário com de-
sign composto central e outro idêntico, mas com a adição de quatro pontos de verificação.
Foram considerando cinco ingredientes diferentes normalmente usados em formulações adesi-
vas para azulejos. Todas as formulações derivadas do modelo foram testadas através do pull-
off de tração de ladrilhos, de acordo com a ISO 13007, a fim de estudar seus desempenhos
mecânicos.
Propriedades frescas das mesmas formulações também foram estudadas através de medição
reológica utilizando reómetro rotacional. Além disso, usei a Análise de Componentes Princi-
pais para identificar possíveis correlações entre propriedades mecânicas e reológicas. Um ter-
ceiro design experimental foi projetado a fim de estudar a geração de poeira das misturas
secas.
Este trabalho tem como confirma que a quimiometria tem a possibilidade de melhorar signifi-
cativamente a informação gerada por experiências através de uma abordagem de análise mul-
tivariada. Esta disciplina, aplicada à ciência do produto à base de cimento, demonstra também
sua extrema flexibilidade e adaptabilidade a qualquer tipo de experiências e sua perfeita
aplicabilidade na ciência da formulação se for configurada corretamente, superando clara-
mente a abordagem clássica univariada.
9
Abstract
I carried out my thesis project at the R&D laboratory of cementitious adhesives of Mapei S.p.A,
leading company in the field of adhesives, sealants and chemical products for building.
The aim of the work was to apply chemometric tools to the study of mechanical and rheological
behavior of cement-based adhesives in relation to their formulation.
In particular, two Design of Experiment were defined: a fractional factorial with central com-
posite design and another identical but with the addition of four verification points. DoE were
defined considering five different ingredients normally used in tile adhesive formulations.
All the formulations derived from the model were tested through tile tensile pull-off, according
to ISO 13007, in order to study their mechanical performances.
Fresh property of the same formulations were also studied through rheological measurement
using rotational rheometer. Furthermore, I used Principal Component Analysis in order to iden-
tify possible correlations among mechanical and rheological properties.
Furthermore, a third DoE was also designed in order to study the dustiness generation of dry
blends.
This work has the goal to confirm that chemometrics has the possibility to strongly enhance
and improve the information generated by experimentations through the approach of multivar-
iate analysis. This discipline, applied to cement based product science, also wants to demon-
strate its extreme flexibility and adaptability to any kind of experimentation and its perfect ap-
plicability in formulation science if it is setup correctly, clearly surpassing the classical univari-
ate approach.
10
I. THEORY
Chemometrics
Introduction
Chemometrics has been defined as “A chemical discipline that uses statistical and mathematical
methods, to design or select optimum procedures and experiments, and to provide maximum
chemical information by analyzing chemical data.” In shorter words, it is focused as “Chemo-
metrics concerns the extraction of relevant information from chemical data by mathematical
and statistical tools.”[1] (Figure 1).
Knowledge
intelligence creativity
Information Hypotesis
data design
Experiments
11
Many in the field of analytical chemistry have found it difficult to apply chemometrics to their
work. The mathematics can be intimidating, and many of the techniques use abstract vector
spaces, which can seem counterintuitive. This has created a “barrier to entry” which has hin-
dered a more rapid and general adoption of chemometric techniques[2].
Despite the broad definition of chemometrics, the most important part of it is still the applica-
tion of multivariate data analysis to chemistry-relevant data. Chemical-physical systems of
practical interest are often complicated and relationships between available (measurement) data
and desired data (properties, origin) cannot be described by theory. Therefore, a typical chemo-
metric approach is not based on "first principles" but is "data driven" and has the goal to create
empirical models. A thorough evaluation of the performance of such models is essential for
new cases. Multivariate statistical data analysis has been proven as a powerful tool for analyzing
and structuring such data sets from chemistry and biochemistry[3].
A Brief History
The start of chemometrics dates back to the 1960s, when multivariate data analysis methods -
like for instance the "learning machine" - have been tried for solving rather complicated prob-
lems in chemistry, such as the automatic interpretation of molecular spectra. The name chemo-
metrics was first used by Svante Wold in 1972 (in Swedish, “kemometria”) and it was estab-
lished in 1974 by Bruce Kowalski.
The first years of chemometrics were characterized by rather uncritical use of machine learning
methods for complex - often too complex - tasks in chemistry and consequently sometimes
accompanied by ignorance and refusal of many chemists. However, in this time also falls the
presentation of the Partial Least Square regression method by chemometricians, which is now
the most used method for evaluation of multivariate data, not only in chemistry. During the next
decades chemometricians learned to use multivariate data analysis in a proper and safe way for
problems with a realistic chance for success, and also found back to the underlying statistical
concepts. Chemometrics contributed with valuable method developments and provided many
stimulants in the area. Furthermore, commercial software became available and nowadays sev-
eral basic chemometric methods, like principal component analysis, multivariate classification,
and multiple regression (by PLS and other approaches) are routinely used in chemical research
and industry. Admittedly, sometimes without the necessary elementary knowledge about the
used methods[3].
12
Multivariate Analysis
The use of chemometrics also implies the use of multivariate data analysis, in which several
related samples are analyzed simultaneously. A multivariate approach when handling and ex-
ploring complex chemical data and designing experiments is certainly part of the foundation of
chemometrics. Multivariate data analysis as opposed to using only one or a few variables in the
data analysis is based on the fact that complex problems - by nature - need multiple variables
to be described. Thus, by using and combining more variables, more information about the
chemical system can be retrieved. In standard multivariate data analysis, data are arranged in
a two-way structure, a table or a matrix. An example is a matrix in which each row corresponds
to a sample and each column to a variable describing the complex system. This is the typical
input for multivariate techniques: when these matrices are analyzed by means of chemometrics,
all the variables are considered at the same time and consequently the extracted information
represent a global overview of the system. Since chemometrics proved to be able to handle
large amounts of data and to extract useful information, it has been successfully applied in
different fields. During the last years, it has so increased in uses and applications that now
modern analytical techniques are usually combined with chemometric methods.
Design of Experiment
To explain satisfactorily this technique is better to introduce first what an experiment is. The
term experiment is defined as the systematic procedure carried out under controlled conditions
in order to discover an unknown effect, to test or establish a hypothesis, or to illustrate a known
effect. (Figure 2)[4].
Experimental Design allows to optimize the amount of available resources while carrying out
an experimental plan, having the possibility to predict the accuracy of the final model results
before starting any battery of tests, with a determination of the leverage level (model accuracy)
in every single point of the investigated space[5].
13
Process
Controllable
input factors Responses
Uncontrollable
input factors
Designed Experiments are also powerful tools to achieve manufacturing cost savings by mini-
mizing process variation and reducing rework, scrap, and the need for inspection (Figure 3).
Components of Design of Experiment (DoE):
Factors (inputs): Include controllable and uncontrollable variables. The former re-
fers to factors that we can control (e.g., a certain ingredient in dry blend formula-
tions). The latter refers to factors we cannot control. Human beings are generally
considered a noise factor, which is an uncontrollable factor that causes variability
under normal operating conditions; yet we can control these factors during the ex-
periment using blocking and randomization.
Levels (settings of each factor): One example would be the particular level of dosage
of an ingredient.
Response (output): Consider testing a new tile adhesive. The output could be its me-
chanical resistance against tensile test or its resistance against vertical slip, its vis-
cosity, etc… Experiments should avoid optimizing the process for one response at
the expense of another, and important outcomes are measured and analyzed to deter-
mine the factors and their settings in a way that will provide the best overall outcome.
14
Conduct experiment,
Define problems
collect data
Brainstorming Brainstorming
Design of Experi-
Verify predicted results
ment
Factorial Design
Experimenters are often concerned with how changes of certain factors affect a process. If the
experimenter has some control over the settings of these factors, then experimentation through
changing these settings can aid in understanding the process. By using settings of the factors at
a fixed number of values, referred to as levels, a factorial design is typically used as a plan to
conduct such an experiment. Instead of focusing on one factor at a time, a factorial experiment
varies the levels of these factors simultaneously. Depending on the different combinations of
factors used in the experiment, not only can the experimenter study how the factors impact the
response individually, but also their interaction.
15
Full Factorial Design
A full factorial design is a basic design that carry out all possible experiments with two-level
factors:
1 level =maximum level
-1 level=minimum level
If there are k factors, each at two levels, a full factorial design has 2k experiments. This method
allows to generate a linear model with interactions, for each variable.
For a three factors full-factorial design:
= + ∙ 1 + ∙ 2 + ∙ 3 + ∙ 1 2 + ∙ 1 3 + ∙ 2 3
(1.1)
All experiments at the boundaries of the design space are planned, as illustrated for three factors
in Figure 4. The corresponding experimental matrix with its encoding system is shown in
Table 1.
= , , (1.2)
I = 123 is called alias generator or design generator, a generating relation for this 23-1 design
(the dark-shaded corners of Figure 4b). Since there is only one design generator for this design,
it is also the defining relation for the design. Equally, I = -123 is the design generator (and
defining relation) for the light-shaded corners of Figure 4b. We call I = 123 the defining relation
for the 23-1 design because with it we can generate (by “multiplication”) the complete confound-
ing pattern for the design. That is, given I=123, we can generate the set of {1=23, 2=13, 3=12,
16
I=123}, which is the complete set of aliases, as they are called, for this 23-1 fractional factorial
design. With I = 123, we can easily generate all the columns of the half-fraction design 23-1.
With three factors, only one alias generator is allowed, dividing the number of experiments by
two for similar model accuracy.
+1 +1
X3 X3
-1 -1
+1 +1
X2 X2
-1 -1
-1 X1 +1 -1 X1 +1
a b
Figure 4. (a) Full Factorial Design; (b) Fractional Factorial Design
17
Where x'i = ( , ,…, )′, β = (β1 , β2 , … , βq )′, and β is a × matrix with β = βkk and
β = β = βkl /2 for < . Note that the model only includes the cross product for < ;
the matrix form with β includes both and , so the coefficients are halved to take this into
account.
Second-order models describes quadratic surfaces, and quadratic surfaces can take several
shapes. Figure 5 shows four of the shapes that a quadratic surface can take. First, we have a
simple minimum and maximum. Then we have a ridge; the surface is curved (here a maximum)
in one direction, but is fairy constant in another direction. Finally, we see a saddle point; the
surface curves up in one direction and curves down in another[7].
(b)
(a)
Y1 Y2
X2 X2
X1 X1
(d)
(c)
Y3 Y4
X2 X2
X1 X1
Figure 5. Sample second-order surfaces: (a) minimum, (b) maximum, (c) ridge, and (d) saddle.
18
There are several choices for second-order designs. One of the most popular is the central com-
posite design (CCD). A CDD is composed of factorial points, axial points and center points.
Summarizing, for a CDD there are three levels:
1 level=Maximum level
0 level= Average level
-1 level=Minimum level
Response = a + b ∙ V1 + c ∙ V2 + d ∙ V3 + e ∙ V1 V2 + f ∙ V1 V3 + g ∙ V2 V3 + h ∙ V1 + i ∙ V2 + j ∙ V3
(1.4)
Constant and linear terms: pla- Interaction terms: distorted re- Quadratic terms:
nar response-surface curve sponse-surface curve curve response
Table 1. Example of a 3 factors Full Factorial design with central composite addition.
Runs X1 X2 X3 Notes
1 -1 -1 -1 Full Factorial
2 -1 -1 1 Full Factorial
3 -1 1 -1 Full Factorial
4 -1 1 1 Full Factorial
5 1 -1 -1 Full Factorial
6 1 -1 1 Full Factorial
7 1 1 -1 Full Factorial
8 1 1 1 Full Factorial
9 0 0 0 Full Factorial
10 0 0 0 Central point
11 0 0 0 Central point
12 -1 0 0 Central point
13 1 0 0 Central Composite
14 0 -1 0 Central Composite
15 0 1 0 Central Composite
16 0 0 -1 Central Composite
17 0 0 1 Central Composite
19
Principal Component Analysis
Principal component analysis (PCA) is probably the oldest and best known of the techniques of
multivariate analysis. It was first introduced by Pearson (1901), and developed independently
by Hotelling (1933). Like many multivariate methods, it was not widely used until the advent
of electronic computers, but it is now well entrenched in virtually every statistical computer
package[8]. The central idea of PCA is to reduce the dimensionality of a data set consisting of a
large number of interrelated variables, while retaining as much as possible of the variation pre-
sent in the data set. This is achieved by using an orthogonal transformation to convert a set of
observations of possibly correlated variables into a set of values of linearly uncorrelated varia-
bles called principal components (PCs). The resulting vectors are an uncorrelated orthogonal
basis set.
Derivation of PCs
PCA can be thought of as fitting an n-dimensional ellipsoid to the data, where each axis of the
ellipsoid represents a principal component. If some axis of the ellipsoid is small, then the vari-
ance along that axis is also small, and by omitting that axis and its corresponding principal
component from our representation of the dataset, we lose only a commensurately small amount
of information.
To find the axes of the ellipsoid, we must first subtract the mean of each variable from the
dataset to center the data around the origin. Then, we compute the covariance matrix of the
data, and calculate the eigenvalues and corresponding eigenvectors of this covariance matrix.
Then we must normalize each of the orthogonal eigenvectors to become unit vectors. Once this
is done, each of the mutually orthogonal, unit eigenvectors can be interpreted as an axis of the
ellipsoid fitted to the data. This choice of basis will transform our covariance matrix into a
diagonalised form with the diagonal elements representing the variance of each axis. The pro-
portion of the variance that each eigenvector represents can be calculated by dividing the ei-
genvalue corresponding to that eigenvector by the sum of all eigenvalues.
Geometrical Representation
In order to understand PCA geometrically, let us consider a two dimensional data set I × J,
where I is the number of samples and J is the number of variables. In the present case, for
convenience, we have set the number of variables to two: J1 and J2. As shown in Figure 6, these
20
samples can be presented in a two dimensional space spanned by J1 and J2. The two axes J1 and
J2 are orthogonal to each other. The data set acquired for the samples have considerable amount
of variation along J1 and J2 axes. In other words, both the dimensions are significantly important
to have the complete information about the sample set.
A counterclockwise rotation of the J1 and J2 axes by an angle θ (45° in the present case) gener-
ates another pair of orthogonal axes T1 and T2. Mathematically, it could be shown using (1.5):
T cosθ sinθ J
=
T −sinθ cosθ J
The new variables (or dimensions) T1 and T2 are the linear combinations of J1 and J2 variables
with sine and cosine as coefficients
T = J cosθ + J sinθ (1.5)
T = −J sinθ + J cosθ (1.6)
Projection of the data set in space, spanned by the new variables T1 and T2 is shown in Figure
7. The data set has most the variations along T1 axis and is literally invariant along T2 axis.
Figure 6. Representation of a data set in the space spanned by J1 and J2. Data has significant variation
along the axes.
21
Figure 7. (a) Rotation of axes J1 and J2 by 45° to generate another pair of orthogonal axes T1 and T2.
(b) Representation of data set in the new space spanned by T1 and T2. The data set has variations along
T1 and T2. (c) Reducion of dimesnions. T2 is unimportant and hence could be removed, and T1 can be
taken as the approximation of data spanned in the two dimensional space spanned by J1 and J2.
In principle, variation along T1 axis can be taken as a good approximation of the two-dimen-
sional data set, and one can easily ignore the T1 and T2 axis. Thus, by projecting the data set in
a suitable space, it is possible to reduce the dimensions of the data set while retaining all the
information.
22
are not correlated to each other. Loadings which have little contribution to both components are
arranged at the origin.
On the other hand, score plot indicates the position of each single experiment in the same ref-
erence graph of loadings, indicating that tests located close to certain variables position have
higher values of these same variables and, with the same logic, experiments opposed to certain
variables will have lower values of these variables. Points close the average appear at the origin
of the score plot.
Loading Plot (68.4% of total variance) Score Plot (68.4% of total variance)
6
V7
V6 V8
0.4
4
9
13
Component 2 (22.9% of variance)
V14
V13 Component 2 (22.9% of variance) 18
11
10
V9 2
0.2
21
2216
29 33 32 6
V10 5
+
V1 7 30 14
31 8 24
+
20 23
0
12 17
0.0
V5 V11
V12 19 2
V2 V4 26 28 27
V3 25 1
-2
15
-0.2
4
-4
3
-0.4
-6
Side-by-side, these two plots took as example, very helpfully characterize all the observations
in the data set. In the loading plot we can assume that the variables arranged at the top (V6, V7,
V8) are correlated because they are close to each other but they are not correlated with the group
variables on the left (V2, V3, V4, V5) and on the right (V1, V9, V10, V11, V12) because the
groups are perpendicular to each other. On the other hand, these last two group of variables are
inversely correlated.
In the score plot, we can easily figure out a cluster of scores whose values do not differ greatly
from each other. Sample 11 as sample 3 are very distant from the cluster and this indicates that
the value of the score regarding the variable to which it refers is much higher than the other
scores.
23
It is no coincidence that we can mentally superimpose these two plots and come to exactly the
same conclusions, using only the plots (Figure 8). This result comes from the fact that the
scores (right) are just a linear combination of the raw data, with weighting given by the loadings
(left).
With this powerful mathematical instrument, it is possible to identify clearly any kind of corre-
lation among different tests. Furthermore, combining these information with experiments ob-
tained from a rational Experimental Design, it will be possible to correlate formulation variables
(input values in Design) to test results (output variables, elaborated with Principal Components
Analysis).
24
Cementitious Adhesives
Introduction to Adhesives
Adhesive, also known as glue, cement, mucilage, or paste, is any substance applied to one sur-
face, or both surfaces, of two separate items that binds them together and resists their separa-
tion[9]. Adhesives may be found naturally or produced synthetically. The earliest human use of
adhesive-like substances was approximately 200,000 years ago[10]. Adhesives are typically or-
ganized by the method of adhesion. These are then organized into reactive and non-reactive
adhesives, which refers to whether the adhesive chemically reacts in order to harden. Alterna-
tively they can be organized by whether the raw stock is of natural or synthetic origin, or by
their starting physical phase. Based on their chemical composition and mechanical properties,
the first important classification of gluing materials for construction, is as follows:
cementitious adhesives;
dispersion adhesives or ready to use;
organic adhesives;
this work will be focused on the cementitious adhesives category.
Mechanism of Adhesion
The mechanism of adhesion has been studied for years. In order to provide an explanation for
adhesion phenomena, several theories have been proposed. Because there is no unifying theory
that describes all adhesive bonds in a comprehensive manner, the categorization of adhesion
mechanisms often overlaps. The bonding of an adhesive to a substrate includes numerous me-
chanical, physical, and chemical forces that influence each other. As it is impossible to separate
these forces from each other, it can be divided into five different adhesion mechanisms, includ-
ing mechanical, electrostatic, adsorption, chemisorptions and diffusion theory[11].
Physical Absorption
The adhesion results from the molecular contact between two materials and these two materi-
als are held together by the van der Waals forces (Figure 9). These are weakest forces that
contribute to the adhesive bonding, but are quite sufficient to make strong joints[12].
25
Figure 9. Surface forces in physical absorption.
Chemical Bonding
The chemical bonding adhesion is attributed to the formation of either covalent, ionic or hydro-
gen bonds across the interface. Two materials form a compound at the joint by swapping elec-
tron (ionic bonding), sharing electron (covalent bonding) or the hydrogen atoms are attracted
to an atom of nitrogen, oxygen or fluorine (hydrogen bonding). Chemical bonds are strong and
have significantly contribution to the interior adhesion.
Diffusion Adhesion
Adhesion of polymeric materials is attributed to interpenetration of chains at the interface. This
theory requires both the adhesive and the substrate are polymers, which are both mobile and
can be soluble in each other. Figure 11 illustrates the interface between an adhesive and the
substrate before and after merged by diffusion. When a polymer adhesive and the substrate are
pressed together and heated, atoms diffuse from one particle to the neighbors. This creates the
26
adhesion. The diffusion adhesion is affected by the contact time, the temperature, molecular
weights of polymers and their physical form (liquid, solid).
Electrostatic Theory
Electrostatic adhesion theory invokes the forming of a difference in electrical charge at the
interface between two materials, in which electrons transfer from one to another. That gives a
force of attraction between these materials, which contribute to the resistance to the separation
of the adhesive and the substrate. Figure 12 illustrates an electrical double layer appeared when
an adhesive is brought into contact with a substrate. This theory cannot be applied if either one
or both materials are insulators.
Figure 12. Positive and negative electrical charge at the material joints.
Mechanical Adhesion
The mechanical interlocking theory of adhesion states that good adhesion occurs only when
an adhesive penetrates into the pores, holes and crevices and other irregularities of the adhered
surface of a substrate, and locks mechanically to the substrate. As illustrated in Figure 13, one
surface is never completely smooth. It always consists of a numerous of peaks and valleys.
27
According to this theory, when an adhesive is brought in contact with the substrate, it must
penetrate the cavities on the surface, displace the trapped air at the interface, and establish a
mechanical interlocking with the interface. It means that the adhesive must not only wet the
surfaces, but also have the right rheological properties to fill in the cavities and to be opened in
a reasonable time.
Figure 13. Mechanical interlocking between the adhesive and the substrate.
The surface roughness helps to increase the total contact area that the adhesion force can de-
velop. That will increase the total energy of surface interaction, which leads to a higher re-
sistance to separation of the joint. However, the adhesive must wet the substrate well in order
to have an efficiently joint.
28
Figure 14. Microscopic detail of cement hydration (Mapei).
The following pictures sequence shows how these filaments develop, from mortar applications
to mortar hardening. This process is known as hydration and filament interlocking occurs in the
final mortar-hardening phase, when most hydration has ended. This process provides the hard-
ened mortar with its mechanical properties and stiffness.
a b c d
Figure 15. (a) Tricalcium silicate hydration has not occurred. (a) After three days, the particles are
covered with hydration ‘filaments’. (a) Close-up of particle fibers or filaments. (a). Fully developed
interlocking between fibers and plate formation of Ca(OH)2 after 28 days.
Note that, though the lime mortar microstructure does not attain such complete interlocking as
that found in cement mortars, lime mortars achieve good mechanical adhesion and, in particu-
lar, greater deformability.
29
Mortars in History
Historically, the use of mortars dates back to thousands of years ago, when the architecture and
construction of the buildings began to be associated with the use of mineral mortars: the gyp-
sum-based mortars were already used by the Babylonians around 6,000 years ago and those
based on pozzolanic cement (volcanic ash powder) have existed for probably over 3,000 years
and were used to a large extent by the ancient Phoenicians and Greeks. In the Roman building
technique, cementitious adhesive was invented and used for the first time, i.e. a mixture of
mortar (lime mixed with sand or pozzolana) and stones or stone chips.
Mortars were applied up to the 1950s exclusively as job-site-mixed mortars. For these mortars,
the mineral binders (mostly cement) and aggregates (mostly quartz sand) are transported sepa-
rately to the job-site and are then mixed together by hand in the appropriate ratio. After mixing
with water, the wet mortar is ready for application. However, the on-site mixing leads to enor-
mous disadvantages, starting from the transport of raw materials and mixing them in the right
proportion directly to the workplace, up to the main disadvantage, which is the impossibility of
automating the entire process. During the 1950s and 1960s in Western Europe and in the USA,
but especially in Germany, there was a fast-growing demand in the construction industry for
new building materials and technologies[13]. Job-site mortar technology was and is notable to
adequately meet all these requirements. As a practical consequence, the development of the
modern construction and building chemical industry in the countries of the Western world from
the 1960s onwards was influenced mainly by three important trends, which can be seen nowa-
days worldwide:
Replacement of the job-site-mixed mortars by premixed and prepacked dry-mix mor-
tars.
Mechanization of mortar application, including bulk transportation systems (e.g., silos),
mechanical systems for automatic mixing of dry-mix mortar with water, and machine
application (spraying) of wet mortar.
Modification of mortars with polymer binders (redispersible powders) and special addi-
tives (e.g., cellulose ethers) and admixes to improve product quality and to meet the
requirements of the modern building industry.
The introduction of dry-mix mortar technology and the use of silo transport and machine appli-
cation of mortars made it possible that from 1960 to 1995 the volume of render and plaster
mortar application in Germany increased by 600 %, while the number of employees in this
30
sector decreased by 25 %, that means, productivity increased by 800 % 14. In Western Europe,
the consequences of this development were phenomenal. Since the 1960s, a huge number of
modern dry mortar plants have been established with millions of tons of capacity. In Germany,
for example, nowadays approximately 100 dry mortar plants exist, producing 10×106 tons per
year of dry-mix mortars. There was a tremendous boom in dry-mix mortar technology after
reunification of Germany after 1990, which now continues in the countries of Eastern Europe.
The average growth rate for dry-mix mortar applications in Europe is approximately 12% per
annum, based on a production of about 35–40×106 tons per year in 2000[15].
1) Mineral binders
Portland cement (OPC, ordinary Portland cement);
cement with high alumina content (HAC, high-alumina cement);
special cements;
hydrated lime;
sulfates.
31
2) Fillers
silica sand;
calcareous sand;
dolomite sand;
marble sand;
light fillers;
special and functional fillers.
3) Additives
cellulose ether;
pigments;
anti-foaming agents;
aerators;
retardants;
thickeners;
water repellents;
plasticizing;
superplasticisers;
mineral oils for dustiness abatement.
Uses of Mortars
The main applications of premixed mortars and their volume of use are shown as follows
1) Common products (about 70% of the total volume of the dry mixed mortar produced):
masonry mortars;
mortars for bricks laying;
adhesives for brick laying;
cement-based guides;
gypsum-based guides;
dry concrete;
32
concrete to be applied by spry;
mineral plasters.
2) Specialized products (about 30% of the total volume of dry mixed mortar produced):
ceramic tile adhesives;
building adhesives;
liquid mortars for tiles;
liquid mortars;
decorative mineral plasters;
powder paints;
outdoor thermal insulation composite system;
compounds to be applied with a trowel;
flooring compounds;
repair mortars.
These highly specialized mortars have characteristics that meet the requirements of the modern
building industry and have replaced the other building materials, such as, for example, ready-
to-use paste compounds and liquid mixtures used in combination with mineral mortars.
Among the commercially available specialist mortars on the market, there are cement-based
adhesives, which for this reason represent one of the most important Mapei sectors, context of
development of the present thesis work.
33
Figure 16. Example of tile adhesive application.
The research related to bonding materials has led to their commercialization as premix mortars,
characterized by properties dictated by the presence of special dehydrated polymers (redispersi-
ble powders) and applied with the thin-bed technique: this is the only way to satisfy the tech-
nical requirements regarding the functionality, efficiency and durability of the adhesive, regard-
less of the types of tiles used and the substrate on which they are glued. Traditionally tile adhe-
sives are a simple composition of cement and sand. Nowadays cement-based adhesives are
composed of cement, aggregates and organic additives (see the next chapters). They need to be
mixed with water or aqueous solutions just before use. This combination of two mineral mate-
rials can be applied using the thin bed spreading technique, especially if they are modified with
cellulose ethers, which significantly improve the workability and retention of water.
In general, the most important characteristics to be sought in an adhesive are the flexibility, the
deformability, the ease of application, the reliability and the good adhesion of all types of tiles
on all types of substrates. The tile adhesives are then classified according to their mechanical
properties and technical performances by the standard EN 12004, which defines and establishes
the criteria for the characterization of the different classes of products.
34
ISO 13007-2:2010 describes methods for determining characteristics for adhesives used in the
installation of ceramic tiles. The test methods described are determination of open time, deter-
mination of slip, determination of shear adhesion strength, determination of tensile adhesion
strength, determination of transverse deformation and determination of chemical resistance.
According to ISO 13007-1:2010 (E), cementitious adhesives shall comply with all of the C1
fundamental characteristics reported in Table 2. The additional characteristics for C2 (improved
performance) products are also contained in Table 2.
Further subclasses are defined as a function of the setting time (F), the extended open time (E),
the slippage resistance (T) and exterior glue plywood adhesion (P). These classification tests,
in accordance with the EN 1348 standard, are carried out by applying the adhesive on concrete
standardized slabs and gluing tiles with a low water absorption (less than 0.2%) and standard
dimensions (50x50 mm), on which pull-off tests are conducted in four different storage condi-
tions described below.
Open Time
The open time test is designed to study the behavior of the adhesive when, once applied to the
substrate, it remains exposed to the air for a long time: this happens when the operator tiles a
large surface and thus spends more time to apply all the tiles. In this test, as we will see later,
the fundamental ingredient is cellulose ether, which thanks to its property of incorporating wa-
ter into its molecular structure, prevents drying of the adhesive paste due to the hygroscopicity
of cement and filler and the formation of the “skin” due to evaporation which hinders a good
wettability.
Storage conditions: 7 days in standard conditions (T = 23 ± 2°C and 50% ± 5% of humidity).
Heat Aging
In hot countries, temperatures of façades exposed to the sunlight can easily reach 70 °C in
summer, leading to fast drying, accelerated setting of the cementitious tile adhesive and cement
dilation. These application conditions have a significant effect on the long-term secure bonding
of tiling. The aim of this test is to determine the critical factors for tiling under such harsh
conditions, and explore ways to increase the durability of the tile work.
Storage conditions: 14 days in standard conditions plus 14 days at 70°C plus one day in standard
conditions. Tiles are pulled off when the temperature of the slabs drops to 23 °C.
35
Water Immersion
Adhesives used to fix tiles externally and in wet duty installations such as swimming pools and
showers obviously have to have a high level of water resistance. In this context water resistance
defines the ability of the adhesive to withstand contact with water without deterioration and
should not be confused with the term “waterproof” which requires that the product should be
impervious to water in addition to it being water-resistant.
Storage conditions: 7 days in standard conditions plus 21 days of immersion in water. The test
is carried out immediately after the slab is removed from the water, when it is still wet, in order
to simulate the real conditions at best.
24 hours
Sometimes it is required that the tiled surface is walkable within a day. This test aims to simulate
the behavior of the adhesive after one day from its application. As already introduced, the ce-
ment starts its curing phase within few hours, but the hydration reactions are not completed yet
and therefore the adhesive does not yet have the characteristics of a completely hardened adhe-
sive which is obtained after many days.
Storage conditions: 24 hours in standard conditions.
Initial Adhesion
The test aims to study the performance of an adhesive, which, during the course of its working
life, does not undergo particular stress i.e. indoor applications.
Storage conditions: 28 days in standard conditions.
Freeze/Thaw Cycles
When water penetrates a tile or any part of the installation and freezes, it solidifies. As it solid-
ifies, it increases in volume. This volumetric change creates mechanical stresses that increase
and become more damaging as the freezing and thawing cycles continue. The stress can damage
the body of the tile, its surface (especially if it is glazed), the adhesive (in some cases creating
a loss of bond) and the grout. It is not constant freezing which is the condition that is most
searched but the action of freezing itself.
36
The ability to withstand these cycles without serious deterioration in adhesive strength are con-
sidered to be important requirements, particularly for Northern European countries.
For this test only, the adhesive is also applied to the back of the tiles, as well as to the slab
(“back buttering technique”).
Storage conditions: 7 days in standard conditions plus 21 days of immersion in water plus 25
cycles of freezing-thawing cycles: 2h at -15 ° C / 2 hours at 15 ° C. The test is carried out on
dry slabs.
Based on this classification, four main types of cementitious adhesives can be distinguished:
basic or C1 (1), improved (2), flexible (3) and highly flexible (4). The distinction between these
categories of adhesive depends on the amount of redispersible polymer powder, on the type and
amount of cellulose ether and cement.
1) The low-quality adhesives contain, in addition to the sand used as inert, about 30 - 35%
of Portland cement and about 0.3% of cellulose ether as a water retention agent: they do
not meet the requirements of the new European standard EN 12004. They are based on
a purely mechanical fixing mechanism, therefore suitable only for porous tiles on ma-
sonry substrates so that they do not give shrinkage or movement phenomena. Further-
more, there is not a high risk of failure due to exposure to very high temperatures or to
frost. Since the most commonly used tiles are not very porous and must be fixed on
different types of substrates, this adhesive must contain a higher polymer concentration
to reach the minimum standards required by the law. As a result, basic adhesives have
been overcome to significantly better quality products for practical applications that
meet particular needs of the market. These adhesives generally meet the requirements
of C1 class.
2) The improved adhesives show a much better bond strength than the previous one cate-
gory because of the use of a quantity equal to 1-3% of redispersible polymer in the
premixed powder. These adhesives generally meet the requirements of C2 class.
37
3) Flexible adhesives have been successful in Europe in the last decades. They are modi-
fied with a greater amount of polymer powder (3-8%) and respond promptly to the re-
quirements for class C2, according to EN 12004, and ensure excellent adherence to all
types of substrates and show a greater deformability. In general, they offer long-term
durability and reliability almost independently of the nature of the substrate.
4) The highly flexible cementitious adhesives contain up to 15% of polymer powder and
are produced for special applications16.
38
Table 3 gives the special characteristics that can be reported for cementitious adhesives.
39
Table 3. Specifications for cementitious adhesives - special characteristics.
40
the tensile adhesion strength measurement of cementitious adhesives, however, is used the new-
ton per square millimeter (N/mm2) or its SI counterpart, mega pascal (MPa). The equivalence
between these units being as follows:
According to standard EN 12004 (2017), adhesive strength is measured through tensile adhe-
sion strength or shear adhesion strength, expressed in terms of N/mm2 or MPa.
Adhesive failure: failure occurs at the interface between the adhesive and the substrate
(AF-S) or between the tile and the adhesive (AF-T). The test value is equal to the bond
strength. Failure sometimes occurs in the adhesive layer between the tile and the pull-
off head (pull stub or dolly). In this case, adhesive bond strength exceeds the test value.
Adhesive cohesive failure: failure occurs in the adhesive layer (CF‑A).
Cohesive failure in the substrate or in the tile: failure occurs in the substrate (CF-S)
or in the tile (CF-T). In this case, the bond strength is greater than the test value.
41
determines the deformability of a mortar which is the capacity of the hardened adhesive to be
deformed by stresses between the tile and the fixing surface without damaging the surface. The
deformation is recorded as deflection at the centre when a layer of the hardened adhesive is
subjected to a three-point loading.
According to the European standard EN 12002, an adhesive strip, 3 mm thick and 45 mm wide,
is deformed between supports, 200 mm apart, by an application of a progressive force in the
center of the strip. The deformation is measured according to the internal Mapei method (MAC
04): the deformation is measured at the maximum load as breaking criterion.
Slip Resistance
One performance requirement of tile adhesives is often that they resist “slip” i.e. once applied
they should retain their position and shape under their own weight. Slip behaviour can be eval-
uated according to EN 12004 where a 100 mm x 100 mm glassed-tile is applied on the adhesive
after two minutes from its layering, then its distance is measured from a reference bar, then it
is kept vertical for 20 minutes and finally its flow from the initial position is measured.
Adjustability
Adjustability-time is the time that the operator has available to adjust tiles once they are placed
into position on the adhesive bed without loss of adhesion strength. It is also dependent on water
42
retention. If the adhesive loses its water by evaporation, or by absorption into the substrate too
quickly, then the mobility of the adhesive is destroyed and attempts to adjust the position of the
tile will either not be possible or, if forced, will reduce the cohesive strength attained.
Ingredients
As we have seen in the previous chapter, cementitious adhesives can be quite different from
each other depending on the type and the proportions of the components, the mixing, the im-
plementation and the cure. We focus herein only on the ingredients that will be used in this
project.
Binder
Cement is the hydraulic part of a cementitious system. It is an hydraulic binder, a substance
used for construction that sets, hardens and adheres to other materials, binding them together.
Cement is seldom used on its own, but rather to bind sand and gravel together. Cement is used
with fine aggregate to produce mortar for masonry, or with sand and gravel aggregates to pro-
duce concrete. There are different types of cement, different for the composition, for the prop-
erties of strength and durability and therefore for the intended use. The most used cement is the
Portland cement, it is mostly used as a binder in the preparation of concrete and is considered
as the fundamental cement on which almost all modern hydraulic binders are based.
Due to the importance of Portland cements, and its complex chemical nature, has been intro-
duced the Cement Chemist Notation (CCN), in which the most common compound formulas
in cement chemistry are indicated with single letters to simplify the formulas. The most com-
mon abbreviations are listed in Table 4.
43
Table 4. Abbreviations of the most common compounds in cement chemistry.
Water H2 O H
Composition
ASTM C150[17] defines Portland cement as “hydraulic cement (cement that not only hardens
by reacting with water but also forms a water-resistant product) produced by pulverizing clink-
ers which consist essentially of hydraulic calcium silicates, usually containing one or more of
the forms of calcium sulfate as an inter ground addition”. The composition of the clinker is
listed in Table 5[18].
44
Table 5. Final composition of the clinker.
Gypsum CaSO4·2H2O C—
S H2 5
Note that the gypsum is added in order to regulate the setting time that, otherwise, it will be too
fast.
Mechanism of Hydration Reaction
The hydration of cement is a complex system of chemical reactions between cement and water,
thanks to which the cement is transformed from an initially plastic, and therefore easily mold-
able mass into a rigid and mechanically resistant material. The progression of these reactions
manifests itself through two distinct physical-mechanical variations.
Setting: a first gradual loss of workability of the cement until the moment when the
mixture is no longer workable.
Curing: the successive and progressive increase of mechanical resistance.
Between these two processes there is no discontinuity: the consistency of the material increases
progressively, going from the typical one of a mud to that of a compact rock. Distinguishing
the two processes responds above all to a practical need of the production process; with the
beginning of the setting phase, the time available to make use of the dough ends.
When water is added, the reactions which occur are mostly exothermic. We can get an indica-
tion of the rate at which the minerals are reacting by monitoring the rate at which heat is evolved
45
using a technique called conduction calorimetry. The heat generated is shown below as a func-
tion of time.
Figure 19. Rate of heat evolution during the hydration of Portland cement.
The stage I hydrolysis of the cement compounds occurs rapidly with a temperature increase of
several degrees. Stage II is known as the dormancy period. The evolution of heat slows dramat-
ically in this stage. The dormancy period can last from one to three hours. During this period,
the concrete is in a plastic state which allows the concrete to be transported and placed without
any major difficulty. This is particularly important for the construction trade who must transport
concrete to the job site. It is at the end of this stage that initial setting begins. In stages III and
IV, the concrete starts to harden and the heat evolution increases due primarily to the hydration
of tricalcium silicate. Stage V is reached after 36 hours. The slow formation of hydrate products
occurs and continues as long as water and unhydrated silicates are present. A more detailed
explanation of hydration reactions is given below.
When water is added to cement, the following series of reactions occur:
The tricalcium aluminate reacts with the gypsum in the presence of water to produce
ettringite and heat:
C3A + 3C—
S H2 + 26H C6AS3H32, H = 207 cal/g
Ettringite consists of long crystals that are only stable in a solution with gypsum (Figure 20).
The compound does not contribute to the strength of the cement glue.
46
Figure 20. Crystal structure of ettringite. (A) Structure of ettringite column, one-half unit cell. Struc-
ture is parallel to the c crystallographic axis. View of A – B plane. Circles represent ettringite col-
umns; regions between columns are channels containing water and sulfate molecules[19].
The tricalcium silicate (alite) is hydrated to produce calcium silicate hydrates, lime and
heat:
2C3S + 6H C3—
S 2H3 + 3CH, H = 120 cal/g
The CSH has a short-networked fiber structure which contributes greatly to the initial strength
of the cement glue.
Once all the gypsum is used up as in the first reaction, the ettringite becomes unstable
and reacts with any remaining tricalcium aluminate to form monosulfate aluminate hy-
drate crystals:
2C3A + 3C6A—
S 3H32 + 22H 3C4ASH18,
The monosulfate crystals are only stable in a sulfate deficient solution. In the presence of sul-
fates, the crystals resort back into ettringite, whose crystals are two-and-a-half times the size of
47
the monosulfate. It is this increase in size that causes cracking when cement is subjected to
sulfate attack.
The belite (dicalcium silicate) also hydrates to form calcium silicate hydrates and heat:
C2S + 4H C3—
S 2H3 + CH, H = 62 cal/g
Like in the second reaction, the calcium silicate hydrates contribute to the strength of the cement
paste. This reaction generates less heat and proceeds at a slower rate, meaning that the contri-
bution of C2S to the strength of the cement paste will be slow initially. This compound is how-
ever responsible for the long-term strength of Portland cement concrete.
C4AF + 3C—
S H2 + 3H C6(A,F) —
S 3H32 + (A,F)H3 + CH
- the ferrite further reacts with the ettringite formed above to produce garnets, i.e.
C4AF + C6(A,F) —
S 3H32 + 2CH +23H 3C4(A,F) —
S H18 + (A,F)H3
The garnets only take up space and do not contribute in any way to the strength of the cement
paste.
Finally, the hardened cement paste consists of the following[18][20][21]:
Ettringite- 15 to 20%
Calcium silicate hydrates, CSH- 50 to 60%
Calcium hydroxide (lime)- 20 to 25%
Voids- 5 to 6% (in the form of capillary voids and entrapped and entrained air)
48
Figure 21. SEM image of the complex structure of the hydration reaction products that coat the grains
of cement.
Figure 22. Cellulose, a linear polymer of D-glucose units (two are shown) linked by β(1→4)-glyco-
sidic bonds.
The cellulose fibers consist of bundles of parallel, non-branched chains joined by hydrogen
bonds between hydroxyl groups on adjacent chains, forming fibrils (Figure 23). This arrange-
ment of parallel chains in bundles due to the hydrogen bonds gives the cellulose fibers a high
mechanical strength. Cellulose can assume both an amorphous and crystalline conformation
(Figure 24).
49
Figure 24. Schematic illustration of amorphous and crystalline domains.
The crystalline part is hydrophobic. In order to obtain a hydrophilic product (such as the com-
mon cotton) it is necessary to subject the cellulose to a set of treatments called mercerization.
Cellulose ethers are characterized by the degree of substitution (DS), the average number of
substituents (MS) and the degree of polymerization (DP).
Several treatments in the literature show how cellulose ethers (CE) have a significant effect on
the physical properties of cement. In this treatment we restrict the investigation of this interac-
tion exclusively on a Portland system. Already in 1985 Hayakawa and Soshiroda[25][26] demon-
strated the ability of cellulose ethers to interact with cement, being able to bind the cement
matrix and the aggregates and thus prevent their gravimetric separation in the fluid paste.
Tanaka et al.[27], have patented an additive which contains CEs capable of increasing the fluidity
and workability and which, above all, confer a higher resistance to compression.
50
Yamamuro et al.[28] have shown that the polysaccharide derivatives contain ionic and hydro-
phobic functional groups that increase the viscosity of the cementitious suspension.
CE also improves the adhesion of cement to the substrate, which increases the cement's ability
to bond with many surfaces. Ghio et al.[29][30] have shown that in cements formulated with CEs
there is an increase in thixotropy that has effects on rheological properties and, therefore, de-
termines an improvement in workability. Some celluloses, in particular HEMC and HPMC[31],
are also able to stabilize air bubbles created during the mixing phase. The stabilization of the
air bubbles is important for a good workability: if, in fact, the air bubbles collapse during the
modeling processes, it would be impossible to use the mixture properly. On the other hand,
however, the air bubbles deteriorate the cohesive strength of the hardened mortar: the strength
of cohesion and the resistance to flexibility and compression are in fact exponential functions
of density. It is therefore necessary to optimize the system, looking for the best compromise
between good workability and compressive strength.
From a chemical point of view, cellulose ethers and water molecules have high affinity: each
EC molecule is surrounded by a hydration sphere. This immobilizes the fluid phase and gives
the cement the ability to "retain" water. Indeed cellulose ethers are used in tile adhesives in
order to confer sufficient water retentivity on them. This is necessary to ensure that the mixing
water is available to the tile adhesive long enough to hydrate the cement. High water retentivity
also ensures prolonged wetting capability of the adhesive once it has been combed onto the
substrate (tilers still refer to this period as the "open time"). Especially at high temperatures or
in windy conditions, the reliability of the bond is increased if the adhesive has prolonged wet-
ting capability, because once the adhesive starts to form a skin (because of evaporation), the
surface area available to the bonding will be less[28].
Another feature often required in cementitious adhesives is slip resistance. Indeed, a good slip
resistance is necessary when installing tiles on non-horizontal areas and, of course, on vertical
surfaces. This characteristic is achieved from certain rheological modifier present in the so-
called modified cellulose ethers that enhance the viscosity and the thickness of the fresh mortar
paste in order to give a greater consistency to the mixture. The modified cellulose ether consists
in cellulose ethers with specific molecular weight plus the addition of thickening substances as
starch and polyacrylic amide (PAA). Also the cellulose itself has a thickening effect on the
fresh mortar paste: the higher degree of polymerization and thus molecular weight, the higher
the solution viscosity and thickness[28].
51
Aggregates
Coarse Aggregates
Aggregates serve as reinforcement and structural component to add strength to the tile mortar.
They provide packing density by filling voids, flexural strength and durability. Aggregate may
be based on quartz, limestone and silica. Generally, silica sand (quartz) used for thin-bed adhe-
sives has a grain size distribution in the range of 0.05 to 0.5 mm.
Low-density aggregates like expanded glass granulates (granular density < 0.5 kg/l) are used in
tile adhesives instead of high-density silica sand (≈ 2.6 kg/l) for special applications[32].
Fine Aggregates
Another filler often used in cement-based tile adhesive is the calcium carbonate. The granu-
lometry can vary from 0.01 to 0.1 according to the use to which they are put. A fine calcium
carbonate gives a creamy consistency and an exceptionally smooth workability to the fresh
mixture. On the other hand, limestone powder tends to absorb water due to its hygroscopic
nature and, used in large quantities, gives brittleness to the hardened tile adhesive.
Redispersible Polymer
Introduction
The second generation of thin-bed tile adhesives, so-called flexible, polymer-modified tile ad-
hesives were polymer-modified tile adhesives were placed on the market in the beginning of
1980s. The main benefits of redispersible polymer powder additives are improved workability,
higher flexibility, and better adhesion. They made possible the safe laying of tiles on floor heat-
ing systems, terraces and balconies. Even fully vitrified and glass tiles could be laid without
problems due to excellent physical bonding of elasticized mortar to the reverse side of tiles.
Redispersible, elasticizing polymer powder invented by Wacker Chemie in 1953 was the key
for this quantum leap in thin-bed technology[32].
Production
Redispersible polymer powders are organic polymer materials produced from latex dispersion
by spry-drying. Latex dispersion are manufactured by emulsion polymerization. The starting
emulsion consists of water-insoluble monomers (i.e. vinyl acetate, vinyl versatate, ethylene,
styrene, methacrylic acid esters etc.) and surfactant (emulgator) in a continuous phase of water.
52
Water-soluble, so-called protective colloids (i.e. polyvinyl alcohol or cellulose ethers) are
added to stabilize the emulsion micelles. The polymerization is started by addition of persul-
fates or peroxides (initiators) to the emulsion. At the end of the polymerization process a milky
polymer dispersion containing 40 to 70 % of latex particles of diameter 0.1 to 20 μm is obtained.
a b
Lattices with particles above 1 μm are of milky-withe appearance. Latex particles in the range
of 0.1 to 1 μm give blue or brownish emulsions. Afterwards the emulsion is spray dried to
obtain polymer powder. For that purpose, the latex emulsion is atomized into fine drops by a
nozzle into the spray tower. An anti-caking agent is added during the spray-drying process in
order to avoid the agglutination of the still sticky redispersible powder. The dried polymer pow-
der is pneumatically carried out of the spray-tower and separated via a cyclone from the humid
air. Redispersibility means that the dried polymer particles of 100 to 500 μm size disperse into
the primary emulsion latex particles (0.1 to 20 μm) when stirred with water[32].
Uses
The polymer domains which work as an organic binder in the tile adhesive matrix develop after
evaporation of water of the mortar by coalescence of the individual latex particles. The polymer
domains provide better adhesion at the mortar/substrate and mortar/tile interface. The main
benefits of redispersible polymer powders in thin-bed tile adhesive are:
better interlocking of adhesive and substrate;
higher flexibility reduces shear stress in the composite substrate, tile adhesive and tile
and allows the use of large tiles;
53
improved rheological properties result in easier mixing, smother workability and good
wetting of the tiles.
Typical polymer emulsion used for tile adhesive are copolymers of ethylene and vinyl acetate
(EVA) or styrene and butyl acrylate.
Dosage of redispersible polymer powder in cementitious thin-bed tile adhesives range from 0
to 5 %[32].
Fibers
Cementitious materials are brittle in nature. Due to this behavior, short, randomly and distrib-
uted fibers are mostly being used to reinforce cementitious materials in the hardened state and
to avoid creeping in the fresh state. Added fibers enhance tensile strength and flexural toughness
and reduce crack creation and propagation in cement matrix. The major effect of fibers is to act
as bridging at crack tips to resist crack propagation. Fiber bonding to cement paste is an im-
portant factor that affects performance of the fiber reinforced cementitious composite (FRCC).
Bonding energy (adhesion) between these materials is composed of interfacial interactions
(chemical bonding) and mechanical interactions (interlocking)[33].
Figure 26. From the left: polypropylene mesh fiber; polypropylene fiber; cellulose fiber loose.
Source: asiafiberhk.com)
54
Fibers can be classified into two groups depend on their average length: long fibers with the
higher aspect ratio among 200 to 500, are mainly used for reinforcement of mortars; short fibers,
which have a general aspect ratio among 20 to 60, are used in influence wet-mortar properties
and water demand. Long fibers, typically over 40 mm’s length are also called macro fibers. A
typical dosage of macro fiber is 3-8 kg/m³. Whereas micro fibers are normally 6-12 mm’s length
with a typical dosage is 0.6-1.0 kg/m³. Macro fibers are primarily used to enhance the toughness
of a render or screed. Figure 26 shows some types of fibers, which can be used for reinforce-
ment of mortars, including polypropylene fiber, cellulose fiber, etc.[34]
55
Rheology
Importance of Rheology
In this section, we wish to recall the some basic definitions of the rheological science. For this
thesis it would be inappropriate to omit a chapter focused on rheology and therefore, on the
rheology that gravitates around the cementitious adhesives. This is because cementitious adhe-
sives must have suitable properties as:
workability;
flow and frictional resistance against surfaces;
adhesion;
resistance to sagging under self weight on a wall or inclined surface.
Introduction to Rheology
The term rheology was coined by Eugene C. Bingham, a professor at Lafayette College, in
1920, from a suggestion by a colleague, Markus Reiner[36]. The term was inspired by the apho-
rism of Simplicius (often attributed to Heraclitus), πάντα ῥεῖ, "everything flows"[37].
Rheology is defined as “the science of the deformation and flow of matter”[38], It is a branch of
physics and physical chemistry since the most important variables come from the field of me-
chanics: forces, deflections and velocities. This definition is very wide and formally would in-
clude studies such as those of hydrodynamics and aerodynamics, which in fact are not normally
regarded as coming within its scope.
In practice, rheology is principally concerned with extending continuum mechanics to charac-
terize flow of materials, that exhibits a combination of elastic, viscous and plastic behavior by
56
properly combining elasticity and (Newtonian) fluid mechanics. It is also concerned with es-
tablishing predictions for mechanical behavior (on the continuum mechanical scale) based on
the micro- or nanostructure of the material, e.g. the molecular size and architecture of polymers
in solution or the particle size distribution in a solid suspension. Materials with the characteris-
tics of a fluid will flow when subjected to a stress which is defined as the force per area. There
are different sorts of stress (e.g. shear, torsional, etc.) and materials can respond differently
under different stresses. Much of theoretical rheology is concerned with associating external
forces and torques with internal stresses and internal strain gradients and flow velocities[39][40].
Continuum mechanics
The study of the physics of continuous materials
Plasticity
Non-Newtonian fluids
Describes materials that
do not undergo strain
Elasticity permanently deform after
rates proportional to the Newtonian fluids
Describes materials that a sufficient applied
applied shear stress. undergo strain rates
return to their rest shape stress.
proportional to the
after applied stresses are
applied shear stress.
removed. Rheology
The study of materials with both solid and
viscoelastic behaviour.
Rheology unites the seemingly unrelated fields of viscosity and non-Newtonian fluid dynamics
by recognizing that materials undergoing these types of deformation are unable to support a
stress (particularly a shear stress, since it is easier to analyze shear deformation) in static equi-
librium. In this sense, a solid undergoing viscous deformation is a fluid, although no viscosity
coefficient is associated with this flow[41][42].
57
One of the major tasks of rheology is to empirically establish the relationships between defor-
mations (or rates of deformation) and stresses, by adequate measurements, although a number
of theoretical developments (such as assuring frame invariants) are also required before using
the empirical data. These experimental techniques are known as rheometry and are concerned
with the determination with well-defined rheological material functions. Such relationships are
then amenable to mathematical treatment by the established methods of continuum mechanics.
The characterization of flow or deformation originating from a simple shear stress field is called
shear rheometry (or shear rheology). The study of extensional flows is called extensional rhe-
ology. Shear flows are much easier to study and thus much more experimental data are available
for shear flows than for extensional flows.
Rheological Models
Viscosity
Viscosity is a fundamental characteristic property of all liquids. When a liquid flows, it has an
internal resistance to flow. Viscosity is a measure of this resistance to flow. Viscosity can also
be determined as a drag force and is a measure of the frictional properties of the fluid. Viscosity
is a function of temperature and pressure. Although the viscosities of both liquids and gases
change with temperature and pressure, they affect the viscosity in a different manner.
The fluid’s resistance to flow is caused by intermolecular friction exerted when layers of fluids
attempt to slide by one another[43].
When the viscosity of a liquid remains constant and is independent of the applied shear stress,
such a liquid is termed a Newtonian liquid. In the case of the non-Newtonian liquids, viscosity
depends on the applied shear force and time.
The shear viscosity of a fluid expresses its resistance to shearing flows, where adjacent layers
move parallel to each other with different speeds. If the speed of the top plate is low enough,
the fluid particles will move parallel to it, and their speed will vary linearly from zero at the
bottom to v at the top. Each layer of fluid will move faster than the one just below it, and friction
between them will give rise to a force resisting their relative motion. In particular, the fluid will
apply on the top plate a force F in the direction opposite to its motion, and an equal but opposite
one to the bottom plate. An external force is therefore required in order to keep the top plate
moving at constant speed. It can be expressed as following:
58
= (1.7)
Viscoelasticity
Viscoelasticity is the property of materials that exhibit both viscous and elastic characteristics
when undergoing deformation. A viscous material exhibits time-dependent behavior when a
stress is applied while under constant stress and deforms at a constant rate, and when the load
is removed, the material has ‘forgotten’ its original configuration, remaining in the deformed
state. On the other hand, an elastic material deforms instantaneously when stretched and ‘re-
members’ its original configuration, returning instantaneously to its original state once the
stress is removed. Viscoelastic materials have elements of both of these properties and, as such,
exhibit time-dependent strain showing a ‘fading memory’. Such a behavior may be linear (stress
and strain are proportional) or nonlinear.
The physical reasons for elastic behavior can be quite different for different materials. In metals,
the atomic lattice distances and its shape change when forces are applied (energy is added to
the system). When forces are removed, the lattice goes back to the original lower energy state.
For rubbers and other polymers, elasticity is caused by the stretching of polymer chains when
forces are applied[44].
In purely elastic materials, the stress and strain occur in phase, so that the response of
one occurs simultaneously with the other.
In purely viscous materials, there is a phase difference between stress and strain, where
strain lags stress by a 90 degree (π/2 radian) phase lag (Figure 27).
Viscoelastic materials exhibit behavior somewhere in between that of purely viscous
and purely elastic materials, exhibiting some phase lag in strain[44].
59
Purely Elastic Response Purely Viscous Response
= 0° = 90°
Stress
Stress
Strain
Strain
Stress and strain in a viscoelastic material can be represented using the following expressions.
Strain: = ( )
Stress: = ( + )
where
ω = 2 π f, where f is frequency of strain oscillation;
t is time;
δ is phase lag between stress and strain.
The elastic and viscous modulus in viscoelastic materials measure the stored energy, represent-
ing the elastic portion, and the energy dissipated as heat, representing the viscous portion[44].
The tensile storage and loss moduli are defined as follows.
Storage: =
Loss: =
Continuous Tests
In this section, we wish to recall some basic definitions of rheology, without particular reference
to mortar, including shear stress, shear rate, and yield stress and then we will focus on continu-
ous rotational tests.
Shear Stress
A shear stress, denoted τ, is defined as a stress, which is applied parallel or tangential to a face
of a material, as opposed to a normal stress, which is applied perpendicularly.
60
Figure 28. Simple schematic of shear stress.
In particular as shown in Figure 28Figure 29, it will result in a strain, or deformation, changing
the square into parallelogram. The formula to calculate average shear stress is:
= (1.8)
where τ is the shear stress, F is the force applied and A is the cross sectional area.
Shear Rate
Consider a volume material as a set of parallel molecular layers kept between two parallel
planes with the distance h between them as described in the Figure 29. The lower plane is fixed,
and the upper plane is displaced by a distance dh at a constant speed v.
s=0
Under the effect of tangential forces, the upper molecular layers move at the same speed of the
adjacent moving plane. The lower layers will move in the same direction but with smaller and
smaller velocities. They create a gradient of velocity between the two planes. The displacement
between two planes is defined as the deformation of the volume material, or the strain, denoted
γ, follows the relation:
61
= (1.9)
ℎ
The standard constant velocity gradient across the sample is defined as the shear rate . Also
called strain rate, it is the strain rate between two adjacent layers of the sheared fluid. It is often
presented as the derivative versus time of the deformation:
= = = = (1.10)
ℎ ℎ ℎ
Yield Stress
The yield stress is defined as the minimum applied shear stress that we observed a fluid flow in
the materials. When the applied shear stress is lower than this value, the material shows the
solid-like behavior (no flow, no permanent deformation). Pass through this threshold, there will
be a transition from solid-like to liquid-like behavior. The material will be sheared.
The yield is the intercept between the ordinate axis (shear stress) and the shear rate.
An approach is to start with the sample in its at-rest state (no permanent deformation) and in-
crementally increase the shear stress until we identify the value at which it starts to flow. It
means that the fluid sample goes from solid-like behavior to liquid-like behavior.
62
= + (1.11)
where τ is the shear stress applied to the material, τ0 is the Bingham yield stress, describing the
stress needed to initiate flow, η0 is the Bingham plastic viscosity, which is the resistance of the
material to flow, and is the shear strain rate.
Bingham model is used to characterize the fluids which have a constant viscosity value.
Shear thinning materials show a decrease in viscosity under increasing shear loads, but they do
not show a time dependent recovery of viscosity unless they are also thixotropic. Hence a thix-
otropic material is shear thinning, but a shear thinning material is not necessarily thixotropic.
In adhesive use, thixotropy can be a benefit. The lowering of viscosity as the adhesive is strained
will aid spreading but the subsequent recovery in structure and thickness will help retain adhe-
sive where it is desired. Ordinary cement-based adhesive is one example of a shear-thinning
fluid, while oobleck provides one realization of a shear-thickening fluid. In the equation 4.8,
when n = 1 and τ0 ≠ 0, the fluid behavior is Bingham. When n = 1 and τ0 = 0 the fluid is
Newtonian. By variation of n and the yield stress τ0, we can express the shear thinning as well
as the shear-thickening fluids.
Typical flow curves of shear stress versus shear rate for different rheological behavior models
are shown in the following figure.
Figure 30. Classification of fluids with shear stress as a function of shear rate.
Flow Test
In order to determine the general flow behaviour of a sample, the viscosity is measured as a
function of the shear rate in a rotational rheometer. For the presentation of the data, either the
63
viscosity or the shear stress is plotted against the shear rate. The thus obtained graph is called
flow curve. In many cases the flow curve is the first and usually also the most important rheo-
logical measurement. It shows the flow behaviour for low shear rates (slow motions) as well as
high shear rates (fast motion). Low shear rate appear e.g. at the levelling or sagging of paints,
sedimentation or slow flowing. High shear rates (> 1000 1/s) occur at the majority of technical
processing like coating, spraying or flowing though pipes.
By means of the flow curve it is possible to read the viscosity for the shear rates of interest. If
someone is interested in the flow behaviour during the use of cementitious adhesives the vis-
cosity at shear rates below 1 1/s are important.
Creep Tests
The creep test forms a simple and quick method used to find the viscoelastic properties of the
material under investigation. The mobile part of the measuring arrangement is loaded with a
constant shear stress (τx) for a certain period of time. The sample reacts on this force with a
deformation, i.e. the material starts to creep.
For this work, this test has been used to evaluate more accurately the slip resistance of cementi-
tious adhesives. In fact, one performance requirement of cement-based adhesives is often that
they resist ‘slump’ or ‘slip’ (i.e. once the tile is applied on a wall, it should retain its position
under its own weight). Slip behaviour can be studied through creep measurements of the long-
term flow of the paste under small but constant stress.
64
Slump resistant adhesives often take advantage of thixotropy: “a decrease of viscosity under
constant shear stress or shear rate, followed by a time dependent recovery when the shear is
removed”. Thixotropy is often confused with shear thinning or pseudo-viscosity.
Rheological Measurements
Cementitious adhesives are widely used for many rendering functions, both outdoor and indoor.
The rheological requirements associated with the various uses and processing combinations are,
therefore, expected to be diverse as well[45].
In cement adhesive laboratories, for each product under investigation, some properties of the
fresh adhesive paste are evaluated.
1. Viscosity: it is measured with the single point test using the Brookfield viscometer. This
is a very fast method because it provides an almost instantaneous response which must
fall within a certain range of viscosity in order to find the correct w/c rate and obtain a
good workability of the fresh paste.
2. Slip test according to ISO 13007 / EN 1308: it is a practical method used to measure the
vertical slip of the tile.
It is obvious that these tests were chosen to have rapid responses without the use of complex
and sensitive tools.
However, these measurements are rough, has a low reproducibility and are affected by the op-
erator’s error and thus, to fully understand the rheological properties of cement adhesives, this
method is not suitable.
Therefore, proper rheological evaluation is needed to provide a useful information for the opti-
mization of fresh cementitious adhesives performance. Rotational rheometry is used to measure
the viscosity of cement based adhesives by varying shear stress.
In this work, two geometries were used: the cross-hatched parallel-plate geometry and ball-
measuring system geometry.
Cross-hatched parallel-plate geometry (Figure 31a) is normally used either in rotational flow
or in oscillatory modes for the rheological evaluation of cementitious pastes. The cross-hatched
plate is needed because it guarantees a necessary grip for the reproducibility of the measure and
avoid slip. The gap between the two plate must be much larger than the particle size (at least
65
ten times) (sand grains in our case), otherwise the continuous medium hypothesis always
adopted in the rheometer theories is violated.
The ball-measuring system (BMS) (Figure 31b) is normally used for large-particulated fluids.
It consists of a sphere that is dragged through a sample volume of approximately 0.5 l. Imple-
mented in a rotational rheometer, this ball performs a rotational motion through the sample on
a circular path. Torques exerted on the sphere and the corresponding rotational speeds are rec-
orded within a wide measuring range.
(a) (b)
Cross-hatched plate
Because of its suitable features, the parallel-plate and the ball-measuring system geometries
have been extensively used in this work for the evaluation of cement adhesive pastes in creep,
stress sweep (parallel-plate only), and flow tests.
66
II. EXPERIMENTAL PART
Materials
Tile Adhesives
In this work we analyzed several components of typical tile adhesive formulation, we indicate
the range in which each component dosage has been varied in the tests.
Portland cement “Colacem 52.5R GREY”; batch: July 2017 (20% - 40%).
Cellulose ether (0.25% - 0.45%), modified or non-modified.
Redispersible polymeric powder (0% - 5%).
Fine calcium carbonate (<100 μm) (0% - 20%).
Coarse calcium carbonate (<600 μm) (0% - 20%).
Cellulose fibers (0% – 2%).
Silica sand (<450 μm).
Preconditioned water at 23 °C.
All the materials were conditioned at least for 8 hours under standard conditions (23 ± 1°C,
relative humidity of 50 ± 5%).
Concrete Slabs
All the concrete slabs used for tile adhesive testing were in compliance with ISO 13007 / EN
1323 in terms of water absorption, residual humidity and superficial resistance. They were all
coming from the same producer, with random production batches used during the experimental
campaign.
Ceramic Tiles
Smooth Body Tiles
All the tiles used for water immersion, heat aging, f/t cycles were, according to ISO
13006:2012, ceramic tiles, group BIII, belonging to non-porous tile, complying with ISO
67
13006:2012 of water absorption, cut to facial dimensions of (50 ± 1) mm × (50 ± 1) mm, with
a thickness in the range of 5 mm to 6 mm.
68
Software
An R-language based free Chemometrics software distributed from Italian Chemometrics
Group[46] was used to make Principal Component Analyses and to obtain data analysis, in order
to use the proper Design of Experiment and to obtain all the equations coefficients necessary to
describe the behavior of adhesives in the whole investigated domain.
An internal software based on Microsoft Excel and Microsoft Visual Basic has been devel-
oped[47] in order to obtain an intuitive graphical evaluation of the obtained results, making them
intuitively described by:
A fully rotatable tri-dimensional graph, representing one of the chosen response in func-
tion of two variables.
A bi-dimensional graph, with a plain representation of the response curves obtained
through the mathematical equation.
The possibility of graphs animation, in order to visually investigate the behavior of the
desired response in function of a third variable
Single point calculation of all the investigated responses in function of up to five input
composition variables.
69
Instruments
70
Deformability tests were performed
using “Uniframe-Mini” dynamome-
ter of Controls Group. Class 1 accu-
racy (load and deformation).
71
Rheological properties were evaluated using “AR-G2” of TA Instrument and “MCR
302” of Anton Paar. For the TA rheometer the parallel-plate, cross-hatched geometry
was used. For the Anton Paar rheometer the ball-measuring system was used.
72
Experimental Designs
First Model
In this work a Full-Factorial design, without Central Composite points, would have required
25=32 experiments which are considered excessive both for the high quantity of raw material
needed and also for the large amount of time required to conduct all the experiments. For this
reason a Fractional-Factorial with Central Composite Design model, was ideated in order to
understand the influence of five different ingredients in the performance of cementitious tile
adhesives. The model includes the following five formulation variables:
Ingredient -1 0 1
Portland cement 20% 27.50% 35%
Cellulose ether 0.25% 0.35% 0.45%
CE Modification Non modified 50/50 mix with CE Highly modified
Polymer 0% 2.50% 5%
Calcium Carbonate 0% 10% 20%
In this work, the quantity of sand (coarse filler) was added until that the sum of the percentage
values reach 100%.
All the samples contains 0.6% of calcium formate, a common accelerant used in the formula-
tions of cementitious adhesives.
For this Fractional Factorial Design the fifth column is generated by the alias generator intro-
duced in the theory part of this thesis. In this case the fifth column is generated from the multi-
plication of the first four column.
73
All the sample were prepared and analyzed following a random numeration, in order to mini-
mize the systematic error caused by the operator.
The experimental plan is showed in the following table:
1 -1 -1 -1 -1 1
2 -1 -1 -1 1 -1
3 -1 -1 1 -1 -1
4 -1 -1 1 1 1
5 -1 1 -1 -1 -1
6 -1 1 -1 1 1
7 -1 1 1 -1 1
8 -1 1 1 1 -1
9 1 -1 -1 -1 -1
10 1 -1 -1 1 1
11 1 -1 1 -1 1
12 1 -1 1 1 -1
13 1 1 -1 -1 1
14 1 1 -1 1 -1
15 1 1 1 -1 -1
16 1 1 1 1 1
17 0 0 0 0 0
18 0 0 0 0 0
19 0 0 0 0 0
20 0 0 0 0 -1
21 0 0 0 0 1
22 0 0 0 -1 0
23 0 0 0 1 0
24 0 0 -1 0 0
25 0 0 1 0 0
26 0 -1 0 0 0
27 0 1 0 0 0
28 -1 0 0 0 0
29 1 0 0 0 0
74
Second Model
The second model treated in this thesis is the Fractional-Factorial plus the Central Composite
Design seen above, with the addition of four verification points, for a total of 33 runs. It was
designed in order to check and verify the actual "operation" of the first model.
Verification point include the levels -0.5 and 0.5 which represent the central points between -1
and 0 (-0.5) or between 0 and 1 (0.5) (tab).
75
The final model including the verification points is:
1 -1 -1 -1 -1 1
2 -1 -1 -1 1 -1
3 -1 -1 1 -1 -1
4 -1 -1 1 1 1
5 -1 1 -1 -1 -1
6 -1 1 -1 1 1
7 -1 1 1 -1 1
8 -1 1 1 1 -1
9 1 -1 -1 -1 -1
10 1 -1 -1 1 1
11 1 -1 1 -1 1
12 1 -1 1 1 -1
13 1 1 -1 -1 1
14 1 1 -1 1 -1
15 1 1 1 -1 -1
16 1 1 1 1 1
17 0 0 0 0 0
18 0 0 0 0 0
19 0 0 0 0 0
20 0 0 0 0 -1
21 0 0 0 0 1
22 0 0 0 -1 0
23 0 0 0 1 0
24 0 0 -1 0 0
25 0 0 1 0 0
26 0 -1 0 0 0
27 0 1 0 0 0
28 -1 0 0 0 0
29 1 0 0 0 0
30 0 0 0 0 0.5
31 0 0 0 0.5 0
32 0 0 0.5 0 0
33 -0.5 0 0 0 0
Due to the large amount of sample needed to perform all the tests required, we decided to pre-
pare dry sample of 9 kg each.
76
The formulation of the samples is showed in the next table:
77
Table 11. Tile adhesive experimental plan raw results.
Water Sp. Mapei F/T Cy- Initial
EN Slip Adjust. 24h O.T. 5' O.T. 20' O.T. 30' H.A. 5' H.A. 10' H.A. 15' W.I. 5' W.I. 10' W.I. 15' Deform.
Test ratio Grav. Slip cles adhes. Load (N)
(mm) (min) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (N/mm2) (mm)
(%) (g/cm3) (mm) (N/mm2) (N/mm2)
1 21.5 1.6 0.3 0.4 10 0.65 1.18 0.00 0.00 0.74 0.00 0.00 0.57 0.44 0.31 1.32 0.39 1.51 7.60
2 19.0 1.32 2.1 10.0 40 0.89 2.57 1.30 0.56 2.11 1.35 1.31 0.79 0.57 0.54 1.15 2.79 3.50 8.44
3 23.5 1.5 0.1 0.2 5 0.30 0.89 0.00 0.00 0.31 0.00 0.00 0.50 0.47 0.32 1.09 0.51 1.27 3.73
4 24.0 1.6 0.3 0.5 5 0.74 2.81 0.55 0.14 2.51 0.20 0.14 0.66 0.52 0.39 0.60 2.78 3.03 9.86
5 25.0 1.3 5.0 10.0 50 0.60 1.35 0.90 0.59 0.54 0.49 0.41 0.67 0.59 0.39 0.93 0.97 1.72 4.51
6 25.5 1.37 5.0 10.0 50 0.89 2.34 0.85 0.60 2.20 1.89 0.90 0.68 0.51 0.46 1.24 2.61 3.24 8.35
7 27.5 1.54 0.2 0.4 10 0.28 0.64 0.48 0.18 0.24 0.12 0.00 0.58 0.27 0.00 1.03 0.75 1.38 4.43
8 24.5 1.35 0.5 0.9 15 0.40 2.08 1.71 1.06 1.78 1.54 1.35 0.61 0.50 0.45 1.02 1.65 3.65 6.62
9 21.5 1.59 0.8 1.2 15 1.10 1.10 0.27 0.00 1.26 0.55 0.17 1.44 0.81 0.60 2.13 1.86 1.54 12.87
10 22.0 1.65 3.0 10.0 25 0.99 1.55 0.18 0.00 2.46 0.18 0.00 1.12 0.47 0.00 2.27 2.71 2.53 16.47
11 25.0 1.84 0.1 0.2 2 0.40 0.00 0.00 0.00 0.00 0.00 0.00 0.80 0.37 0.00 1.87 0.45 1.57 16.22
12 23.5 1.69 0.3 0.6 5 1.08 2.27 0.18 0.00 3.68 1.36 1.08 1.50 0.95 0.61 1.98 3.48 2.62 14.41
13 27.0 1.56 5.0 10.0 25 0.83 1.80 0.29 0.00 1.62 1.08 0.23 1.29 0.80 0.52 1.66 2.17 1.92 10.47
14 26.0 1.43 5.0 10.0 50 1.36 3.48 1.33 0.81 3.34 2.94 2.36 1.32 1.17 0.74 1.82 3.45 2.84 12.08
15 29.0 1.52 0.4 0.7 15 0.41 1.00 0.26 0.18 0.68 0.58 0.27 1.16 0.97 0.71 1.80 1.04 1.61 10.41
16 29.0 1.61 0.4 0.8 18 1.14 2.54 0.52 0.22 3.61 1.89 0.00 0.93 0.72 0.44 2.39 2.32 2.86 10.31
17 25.0 1.48 0.8 1.3 18 1.11 1.69 0.46 0.17 2.45 1.23 0.87 1.08 0.71 0.52 1.75 2.68 2.27 9.23
18 24.5 1.47 0.7 1.2 20 1.06 1.87 0.72 0.32 1.87 1.43 1.06 1.19 0.66 0.51 1.85 1.90 2.24 9.28
19 24.5 1.53 0.6 1.4 22 0.88 1.87 0.81 0.22 2.50 1.47 1.15 1.00 0.77 0.52 1.54 2.17 2.39 9.02
20 23.5 1.45 1.0 2.5 20 1.00 2.06 0.88 0.56 1.59 1.45 0.73 1.02 0.73 0.50 1.33 2.10 2.29 9.82
21 25.0 1.56 0.6 1.0 15 1.11 1.66 0.53 0.04 2.21 0.90 0.00 0.68 0.53 0.60 1.86 2.38 2.23 12.09
22 25.0 1.56 0.6 1.0 12 0.57 0.85 0.27 0.00 0.54 0.21 0.00 0.60 0.44 0.41 1.81 1.09 1.81 7.40
23 24.0 1.49 1.6 3.2 25 1.29 3.07 0.81 0.47 3.43 2.12 1.70 1.19 0.68 0.52 1.44 3.55 2.99 11.43
24 23.5 1.48 5.0 10.0 30 1.14 2.27 0.80 0.41 2.34 1.80 1.11 1.11 0.56 0.50 1.50 2.91 2.18 10.48
25 26.0 1.56 0.2 0.5 12 1.10 2.41 0.94 0.51 2.42 1.18 0.92 1.09 0.60 0.47 1.87 2.56 2.19 12.51
26 22.5 1.59 0.3 0.7 10 1.17 1.55 0.20 0.00 2.15 0.68 0.21 0.96 0.68 0.44 2.02 2.23 2.17 11.47
27 26.5 1.49 2.0 7.5 40 0.89 2.07 1.06 0.64 2.11 1.79 0.98 1.01 0.58 0.45 1.26 2.10 2.50 8.48
28 24.0 1.48 0.8 1.4 20 0.89 1.86 0.89 0.46 1.77 1.39 1.11 0.79 0.63 0.29 1.22 1.95 2.28 7.06
29 25.0 1.61 0.6 1.1 15 1.33 1.49 0.57 0.17 2.04 1.26 0.00 1.24 0.74 0.43 2.08 2.54 2.11 11.47
30 25.0 1.51 1.0 2.6 25 1.14 2.22 0.65 0.37 2.13 1.59 0.77 1.08 0.69 0.45 1.49 2.33 2.30 8.66
31 23.5 1.46 1.8 3.4 30 1.22 2.36 0.75 0.39 2.58 1.33 1.36 0.97 0.70 0.52 1.53 2.49 2.62 7.79
32 23.0 1.48 2.2 6.0 35 1.11 2.15 0.78 0.48 2.10 1.34 0.86 1.08 0.69 0.49 1.38 2.22 2.29 8.84
33 23.0 1.49 0.9 1.5 20 0.88 2.05 0.84 0.29 1.75 1. 17 0.86 0.83 0.57 0.74 1.16 2.14 2.22 7.46
Table 12. Tile adhesive experimental plan raw results.
Ingredient -1 0 1
Portland cement 20% 30% 40%
Fine calcium
0% 10% 20%
carbonate
Coarse calcium
0% 10% 20%
carbonate
Polymer 0% 5% 10%
Cellulose fibers 0% 1% 2%
80
In the following table is listed the experimental plan and its raw results:
Exp. Cement Fine C. Coarse C. Polymer Fiber D’ (at 0s) D” (at 30s) Dust Index
This model does not include verification points that will be seen in future projects.
81
Procedures
All of the procedures were carried out by my with the exception of the pull-off tests which were
carried out by an operator.
Mixing Procedure
For each sample we found the volumes of water in such a way that the fresh paste has a
Brookfield viscosity between 450 and 550 mPa·s, calculated using the Brookfield viscometer.
Then we proceeded following the internal Mapei mixing procedure “MAC 01-E”.
The water is poured into the five-liter capacity planetary mixer (Automix); the mixer is started
at “slow” speed and 3 kg of the dry blend is added in a constant rate followed by mixing until
30 seconds; the bowl is scraped down with the paddle within 1 minute; the paddle is replaced
and the mixing is started again for 1 minute; the mixing is stopped to let the adhesive mature
for 10’; the mixing is started again for 15 seconds at speed 1.
Evaluation of Adhesion
All the investigated formulations were checked following the same preparation procedure and
pulled off with identical dynamometers (LBG M053, properly calibrated according to ISO
7500-1). The laboratory was maintained by a proper system fully conditioned, keeping temper-
ature, humidity and air speed within the range dictated by the norm: 23±1°C, relative humidity
of 50±5% and air speed lower than 0.2 m/s.
After the standardized mixing procedure, the adhesive is spreaded on a concrete slab in a single
pass, with a 6 mm-toothed square notched trowel. The trowel shall be held at an angle of ap-
proximately 60° to the substrate at a right angle to one edge of the slab and drawn across the
slab parallel to that edge (in a straight line).
After 5’, 10’ and 15’ or 5’, 20’, 30’ (for open time testing only), 50 mm x 50 mm tiles are
applied in the adhesive bed. Each tile is loaded with a 20 N weight for 30 sec (Figure 33).
The slabs prepared with each adhesive were submitted to all the curing procedures (Figure 34)
according to ISO13007 and listed in Table 15.
82
Figure 33. Laying of adhesive and tiles.
Figure 34. Slabs aging in different conditions and subsequent pull-off test.
83
Storage at room temperature (initial adhesion).
Porous tiles were used. The slabs are stored under standard conditions for 27 days. Then
the pull-head plates are bonded to the tiles and the slabs are stored for another day before
the strength measurement.
Heat aging.
Smooth tiles were used. The slabs are stored for 14 days under standard conditions and
then at 70°C in air-circulating oven for a further 14 days. After removal of the tests units
from the oven the pull-head plates are bonded to the tiles and the test units are stored at
standard condition for further 24 hours.
Freeze-thaw cycles.
Smooth tiles were used. The tiles are covered with a 1 mm thick layer of adhesive before
placed on the adhesive bed (buttering-floating technique). After the slabs have been
stored for 7 days under standard conditions and immersed in water for 21 days, 25
freeze-thaw cycles are carried out. One freeze-thaw cycle consists of storing the slab
test unit for 5 hours in a cold chamber at -15°C and afterwards for 2 hours in a water
quench. After the last cycle the test unit is removed from the water quench, wiped with
a cloth and the pull-head plates bonded to the tiles. The test units are stored another 24
hours at standard conditions before testing.
84
Measurement of Tensile Adhesion Strength
The tensile testing dynamometer is applied to the pull-head plates and the pull-off test is per-
formed with a constant load rate of 250 N/s. The pull-off dynamometers must satisfy the mini-
mum requirements for Class 1 according to ISO 7500-1, with a minimum sampling frequency
of 50Hz. These tests allow to dictate the adhesive mortar classes according to ISO 13007.
Figure 35. Evaluation of fresh properties (from left to right): Brookfield viscosity, EN slip, Mapei
slip.
85
Evaluation of Deformability
All the investigated samples were tested for their deformability, according to ISO 13007 / EN
12002 norms. After a standardized mixing procedure, six samples were prepared for each for-
mulation: the fresh adhesive is poured into the templates on a flat polyethylene film, and a flow
table is used to form a test strip of 45 mm width, 3 mm thickness and 280 mm length. The tests
specimens are placed in an air-tight plastic container for two days under their molds and then
conditioned at standard conditions for another 14 days. After that, the thickness of the tests
specimens is determined using a caliper with 0.01 mm precision. Samples which deviate more
than 0.1 mm from 3 mm are discarded.
Testing phase (3 points flexural test) has been performed using a Uniframe-Mini dynamometer
from Controls, with a sensitivity of 0.001N, a preload value of 0.1N and using deformability at
maximum load as breaking criterion.
Evaluation of Dustiness
The Dustiness has been evaluated with an AnaTec DustMon L machine. A fixed amount of
cementitious product drops in an 80 cm tube and then generates a cloud of dust inside a testing
chamber: the measurement is taken with a laser light detector, that transforms the amount of
non-passing light in a dustiness index. These values (%) are recorded immediately after the dust
cloud generation and after 30 second from the impact.
The so-called “Dust Index” is the sum of these two numbers and can be considered as an indi-
cator of air quality during the use of cementitious products.
86
Figure 37. Dustiness evaluation.
Mixing Procedure
All the paste batches were prepared using the same water ratio used in the evaluation of tile
adhesion strength. Different mixing batch were used for each test and repetition.
The parallel plate geometry requires a little amount of sample: 200 g of the dry blend were
poured in a 250 ml polypropylene container and mixed with the appropriate quantity of water
for 30” using a digital overhead stirrer with a dough hook bar. The paste was then maturated
for 10 minutes, mixed again for 30” and was ready for the measurement.
For the ball-measuring system measurements, 500 g of the dry blend were mixed with the ap-
propriate quantity of water using the planetary mixer (Automix - 65-L0006/AM) using the in-
ternal procedure “MAC 01-E”.
Sample Loading
For the parallel plate geometry, a silicon ring with 40 mm diameter and 2.8 mm height was used
for moulding the adhesive paste on the lower plate; the paste was placed with a spoon and the
excess material removed with a spatula; the ring was then removed.
For the ball-measuring system, the fresh sample was poured into a stainless steel bowl, with 10
cm diameter and 5 cm deep, until filling. The bowl was replaced in the appropriate compartment
of the rheometer and the spherical geometry was replaced and immersed in the dough.
87
Test Procedure
In this work, for each sample, four different rheological tests were carried out using two geom-
etry assembled on two rotational rheometer: TA rheometer used a cross-hatched parallel plate
geometry; Anton Paar rheometer used a ball-measuring system (BMS).
All tests were performed twice to confirm the repeatability of the measurement.
After each run, the geometries were cleaned with water and then dried with compressed air.
For tests requiring parallel-plate geometry, the upper cross-hatched plate was lowered until it
gently touched the flat surface of the sample at a gap distance of 1500 μm. This preparation
procedure ensured that initial geometry and volume of the samples were constant for all tests.
The Peltier plate maintained the temperature of the system at 23 °C for all the tests. In addition,
it is recommended to attempt disturbing the sample as little as possible when loading, in order
to maintain its original structure. This mainly consists of descending the upper plate slowly to
keep low shear rate and to avoid excessive increase of the normal force and possible modifica-
tion/destruction of the sample structure during the test.
For each test, the instrument performed an auto calibration and gap auto zero. This procedure
is identical for each test.
Flow curves were performed using both the geometries. The shear was set in the range
from 10-3 s-1 to 100 s-1 in 3 minutes of measurements. The instrument recorded 10 points
in series for every order of magnitude.
Oscillatory stress sweep tests were performed using the cross-hatched parallel plate ge-
ometry. The shear stress range was set from 1 to 1000 Pa in 3 minutes of measurement.
The instrument recorded the values of the elastic (G’) and viscous modulus (G”) 10
times in series for every order of magnitude.
Peak-hold tests were performed using the BMS. Three peak hold tests were performed
at 3 different shear: 1, 5 and 10 s-1. The rheometer performs more revolutions on the
same sample under a constant shear stress.
Creep tests were performed with the cross-hatched parallel plate geometry. A constant
shear (200 or 300 Pa) was set for 3 minutes after 3 minutes of waiting.
88
Evaluation of Dynamic Mechanical Properties
The samples already used for the deformability tests were collected and cut in order to obtain
samples measuring 60x10x4 mm.
DMA measurements were performed using DMA 242 C using the temperature scan method at
2° C/min. It started from 0°C and ended at 50°C using scan frequency of 1 Hz.
89
III. RESULTS AND DISCUSSIONS
Load
E'5
E'23
E'50
F/T Cycles Water 5'
Cement
0.3
Water 10' 28 gg
Delta E' H.A. 5'
0.2
Component 2 (27.9% of variance)
0.1
Water 15'
H.A.
O.T.10'
5'
+ Mapei
EN Deform.
Slip
Slip
H.A. 15'
0.0
Water ratio
Adjust.
-0.1
O.T. 20'
O.T. 30'
-0.2
Cellulose ether
-0.3
90
Score Plot (67% of total variance)
6 9
12
10
4
Component 2 (27.9% of variance)
29
11
25 14
26 13
2
16 24 23
21 17
18
15
+ 1920
30 3231
0
33 27
1 22
2
-2
28
6
8
-4
3
7
5
-6
-5 0 5
91
that are present in that area (i.e. 9, 11 and 12), they all have a high content of cement
and little amount of cellulose ether.
2) Open time at 20’, at 30’, adjustability and slip resistance are strictly correlated and they
are all related to the cellulose ether content. A high amount of cement and a high specific
gravity penalize these results.
3) Water immersion and heat aging at 5’ are correlated to the cement amount and at 10’
and especially at 15’ they move into the cellulose group.
4) Polymer generally improves all the performances, with a lower effect on water immer-
sion and freeze-thaw cycles and a very important contribution over heat, initial adhesion
and deformability results.
5) The modified cellulose ether improves only the slip test (that, naturally, has better re-
sults when the values are low) and strongly penalize all other results.
These considerations will be examined and studied later using Response Surface Curves (RSC).
92
Rheological Tests Correlations
General Overview
Before exposing the results using the chemometrics approach, it is better to focus on the classic
approach, in order to present, in a generic way, the individual graphs of the rheological tests
carried out on the fresh and hardened samples.
In order not to dwell too much on classic approach, only three samples with very different
formulations will be exposed in this chapter. A more detailed exposition and explanation of the
results of this work will be dealt using the chemometric approach, thanks to which it is possible
to analyze and understand the rheological behavior of cement adhesives, taking into consider-
ation all 33 samples simultaneously for each test.
For each characterization, the results of three different samples have been chosen which differ
considerably from the formulation and rheological point of view (see Table 16). All the de-
picted graphs, except the creep ones, are set up using a logarithmic scale for the ordinate and
the abscissae axis in order to have graphs that are more readable.
Flow Curves
The flow curves depicted in the next two figures, show the trend of the viscous evolution (vis-
cosity) (Pa∙s) with increasing the shear rate [1/s]. In this way, it is possible to visualize how the
viscosity of the adhesive changes in a range of shear rates unlike the Brookfield measurement
which calculates the viscosity only at a determined shear rate
The experiments were carried out using two different geometries and instruments: the graph in
Figure 40 is obtained from the TA® rheometer with the plate-plate cross-hatched system. The
93
graph in Figure 41 is obtained from the Anton Paar rheometer using the ball-measuring system.
Also in these test, three different sample were chosen: 6, 11 and 18.
Flow Curve TA
100000
10000
Viscosity (Pa·s)
1000
100
10
0,01 0,1 1 10 100
Shear rate (1/s)
Flow Curve AP
1,0E+05
1,0E+04
Viscosity (Pa·s)
1,0E+03
1,0E+02
1,0E+01
0,001 0,01 0,1 1 10 100
Shear rate (1/s)
94
As Bingham pseudoplastic fluid, the viscosity of the cementitious adhesives drops with the
increasing of the stress. However, it is clear the difference between the three samples, both in
the first and in the second graph: it is evident that sample 11 has a higher viscosity than the
sample 18 and 6. In this experiment the parameters that influence most are: the quantity of
cement, the water mix ratio and the quantity of cellulose. Indeed sample 11 has 35% of cement
instead the 27.5% of the sample 18 and the 20% of the sample 6 (see table xx of the formulas).
The flow curves obtained from the plate-plate cross-hatched geometry cross each other at 10
[1/s]. This happens because the sand could generate friction with the cross-hatched geometry.
The two tests both express the same property of an adhesive but the curves obtained are not the
same because different geometries were used.
95
Stress Sweep
1,00E+06 1,00E+06
1,00E+05 1,00E+05
G'' (Pa) ● ■ ▲
G' (Pa) ● ■ ▲
1,00E+04 1,00E+04
1,00E+03 1,00E+03
1,00E+02 1,00E+02
1 10 100 1000
Osc. Stress (Pa)
In all the samples G’ > G” until a certain value of stress: the fresh adhesives show an elastic
behavior when the oscillatory stress is low and it is almost linear until 100 Pascal. After a certain
value (that differs between the three samples) the viscous behavior becomes predominant
(G”>G’) because the structure of the adhesives begins to breakdown. The G’ and G” stress
sweep curve of the sample 11 his shifted in two orders of magnitude relative to the sample 18
and in three orders relative to the sample 6 and this result enlightens the big difference, both in
viscous and elastic modulus, between the three samples. This test shows that the fresh paste of
the sample 11, unlike the sample 6, is very structured and thicker due to the high quantity of
cement and modified cellulose ether, which are determinant in this test. The sample 18 shows
an intermediate behavior.
A more detailed analysis of the results will be discussed using the RSC in which we will be
able to understand the influence of all the other ingredients.
96
Creep 200 – 300 Pa
These tests are performed using a parallel plate geometry that rotate under a constant stress (200
and 300 Pa). This test is specifically used to investigate the slip resistance of adhesives and, in
our case, to simulate the slippage test performed with the normed tile (200 Pa) and the Mapei
tile (300 Pa) seen before.
Below is depicted a multiple line graph obtained from the creep tests carried out on samples 6,
11 and 18 at 200 Pa (Figure 43) and 300 Pa (Figure 44) respectively. % Strain is displayed on
the Y-axis (logarithmic) versus time on the X-axis.
Creep 200 Pa
1000
100
10
% Strain
0,1
0,01
0 0,5 1 1,5 2 2,5 3
Time (min)
97
Creep 300 Pa
1000
100
10
% Strain
0,1
0,01
0 0,5 1 1,5 2 2,5 3
Time (min)
As we can see, the curves follow the same trend but widely differ in strain percentage. The
ordinate axis is logarithmic so the values after the plateau differ in three orders of magnitude.
Both graphs show that the sample 6 has the highest value of strain, and thus means that it has
the lowest slip resistance. Sample 11 on the contrary, has. The sample 6 has a large quantity of
cellulose ether but without modifier and sample 11 has only an average quantity of modified
cellulose. These two ingredients contribute most for the slip resistance test and give opposite
results: the cellulose itself makes cement paste less thick and more fluent; the modified cellu-
lose gives solidity to the fresh adhesive.
These results are perfectly correlated with those obtained by slips tests: in fact, sample 6 pre-
sents a EN slip of 5 mm and Mapei slip of 10 mm; sample 11, 0.1 mm and 0.2 respectively;
sample 18, 0.7 mm and 1.2 mm respectively.
A more detailed analysis of the results will be discussed using the RSC in which we will be
able to understand the influence of all the other ingredients.
98
Dynamic Mechanical Analysis
This test is specific to observe the glass transition temperature of the polymer present in the
cement adhesive. It is also a test that can be compared to the deformability test seen above, but
with much smaller forces involved and without causing the material to break.
Among all the 33 samples analyzed, the results of 2, 18 and 11 were selected which have re-
spectively a high, medium and no polymer content.
E' /MPa
22000
20000 [3.2]
18000
[1.2] M 848 PR17ITAC0210-02.dm2
E' (1.0 Hz)
[2.2] M 843 PR17ITAC0210-18.dm2
16000 E' (1.0 Hz)
[3.2] M 840 PR17ITAC0210-11.dm2
E' (1.0 Hz)
14000
12000
[2.2]
10000
8000
6000
[1.2]
0 10 20 30 40 50
Temperatura /°C
The results demonstrate that the storage modulus G’ gradually decreases with increasing tem-
perature but it remains in the same order of magnitude. It is also clear the difference in elastic
modulus G’ between the three samples, mostly caused by the amount of cement that generates
a more rigid structure. Cellulose ether also contributes, even if in a minor way, to the lowering
of the elastic modulus as it acts as a retarder in the cement curing phase as discussed in the
theory part of this thesis[48]. One other important consideration concerns the glass transition
temperature: it is well evident in the sample 2 around 30°C and less evident but still present in
the sample 18, while it does not exist in the sample 11.
This test will be evaluated with the deformability results using a PCA, in the next chapter.
99
Principal Component Analysis
In situations like these, that is when you have a large amount of data, sometimes redundant, it
is very useful to use a PCA in order to get a better view of the information and it is easier to
interpret the results. For each test, the significant points of the graphs are chosen, necessary for
the realization of the PCA: in the creep test are chosen the last point of the graph for each run,
representing the total strain performed by the slip of the adhesive.
In the oscillatory test, are chosen G’ and G” at their plateau (maximum value), G’ and G” at
200 and 300 Pa in in such a way as to have a correlation with the results obtained by the creep
tests and G’1/2 which is the oscillation rate in which we have a halving of the elastic modulus.
This value indicates the stress necessary to flake apart the adhesive.
In the flow test are chosen the first point of the curve which represent the yield and the viscosity
at shear rate of 1s-1 to compare it with that of Brookfield.
In the peak-hold tests, all the runs shown constant values of viscosity therefore the mean values
were taken.
Using the Principal Component Analysis with the data obtained from the Experimental Design,
it is possible to evaluate the correlations among the whole rheological performance profile of
cementitious tile adhesives that exist between the various tests performed on the samples with
the addition of the pull-off results of the open-time test, obtaining the following result:
100
Loading Plot (77.9% of total variance)
ph 1s-1ph 5s-1
A ph 10s-1
0.4
1s-1 AP
1s-1TA
0.2
EN Slip
Mapei Slip
200 Pa
300
G'1/2Pa
(Pa) Adjust.
G"200
G'200 Pa +
B
Yield AP
G'Plateau
0.0
G'300 Pa
Yield TA
G"300 Pa
G"Plateau
O.T. 20'
O.T. 30'
C
-0.2
-0.4
11 9 13
10
18
22 21
16 29
7 33 325
+ 31
30 6
12 24 14
0
17 208 23
26 19
25 1
28 27 2
15
3 4
-5
-5 0 5
101
Variance explained by 5 components: 92.19%
% Variance explained by each component: 61.59; 16.34; 6.57; 4.17; 3.53.
Combining the loading and the score plots, we can deduce that the group A is totally uncorre-
lated to the other groups because these tests only represent the viscosity of the adhesive under
a specific stress. The viscosity values obtained from these tests are very similar to each other
because we have chosen to obtain dough with the same viscosity of Brookfield (by varying
water mix ratio) that can be compared to that obtained at 1s-1.
The group B has well correlated results because they all depend on the viscous and elastic
modulus which, in tile adhesives, follow the same trend. A high viscous or elastic modulus
means that also the value of G’1/2 will be high, this is the case of very structured adhesives: a
low amount of cellulose ether but with a high content of its modified formula. As we can see,
this group is inversely correlated to the group C where are depicted the loadings of open time,
slip, creep and registrability tests which are strictly correlated to each other: the adhesive with
a high amount of cellulose ether subjected to the extended open time test, will have a high pull-
off resistance (thanks to water retention and thus, good wettability), but it will slip more. The
slip resistant adhesive, contrariwise, will give high values to the loading group C and low values
to the group B, which is bad for open time test (low wettability) and good for slip test.
The group C strictly depends to the amount of rheological modifiers. Indeed the score plot
highlights that the samples with a high quantity of non-modified cellulose ether are on the left
of the plot and those with a high quantity of modified cellulose, on the right.
102
Design of Experiment Results
The experimental design based on a fractional factorial design explained in the previous chap-
ter, helped us to understand the variations of the performance by varying the concentration of
the five variables, which are the ingredient of our formulations.
The three-dimensional graphs that we are going to visualize are obtained from the elaboration
of the coefficients generated by the R-based chemometric program. Unlike how PCAs may
appear, they are easy to read for everyone.
We are going to see the response surface curves that express very interesting results and that
inspect real problems and on which we can make correlations. Also the dustiness DoE will be
analyzed.
Before the single discussion, coefficients (from the first and second model) generated from the
R-based program and their significance and predictions of verification points are illustrated.
*** 16
15
2.0
28
7
13
1.5
27
***
14
26
25
6
11
29
5
22
21
1.0
Fitted Value
***
30
8
19 17
18
31 4
***
33 23
24
32
3 28
20
12
24
*
0.5
*
26
10
22
1
9
0.0
** *
20
-0.5
1 2 3 4 5 6 7 8 9 11 13 15 17 19 20 22 24 26 28
Experimental Value
Table 17. Coefficients of water mix ratio. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0,03125 -0,03125 -0,09375 0,03125 0,34375 0,009894 0,009894 0,259894 0,009894 -0,24011
103
Significance of coefficients:
red: > 99,9%;
blue: between 99 and 99.9%;
green: between 95 and 99%.
Table of variables:
Variable Factor
V1 Cement
V2 Cellulose ether
V3 Modified cellulose ether
V4 Polymer
V5 Calcium carbonate
Table 18. Predicted vs. observed values of verification points (first model).
Test n° Predicted (%) Observed (%) Difference (%) Confidence interval (%)
30 24.8 24.5 0.3
31 24.3 23.5 0.8
± 7.2
32 24.0 23.0 1.0
33 24.1 23.0 1.1
Table 19. Coefficients of water mix ratio. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0,03125 -0,03125 -0,09375 0,03125 0,34375 0,00576 0,119778 0,273202 0,045627 -0,13876
104
Table 20. Predicted vs. observed values of verification points (second model).
105
Figure 47. Water mix ratio: filler amount vs. cement (top left); polymer amount vs. cellulose ether
(top right); CE modification vs. cement (bottom).
Specific Gravity
Significance of the coefficient and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
*** 11
***
1.8
***
0.05
1.7
12
10
29
Fitted Value
26116
94
1.6
13
25
21
0.00
30 7 22
15
1817 3 19
31
1.5
3233
23
***
2427
28
14 20
1.4
6
-0.05
***
2
5
1.3
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0,006875 -0,00687 0,008125 -0,00062 -0,01188 0,02095 0,01595 -0,00405 0,00095 -0,01905
106
Table 22. Predicted vs. observed values of verification points (first model).
Test n° Predicted (N) Observed (N) Difference (N) Confidence interval (N)
30 1.54 1.51 0.03
31 1.50 1,46 0.04
± 0.05
32 1.49 1.48 0.01
33 1.48 1.49 -0.01
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.006875 -0.00688 0.008125 -0.00062 -0.01188 0.025151 0.018479 -0.00206 -0.00096 -0.01997
107
Figure 48. Specific gravity RSC: cement amount vs. CE modification (left) and cellulose amount vs.
filler amount (right).
108
EN Slip
Significance of the coefficient and experimental vs. fitted plot (from R):
*** 6
5
14
13
***
0.2
24
0.5
10
2
23 32
27
0.0
*
31
Fitted Value
0.0
**
19 18 17 20
29 9 3330
21
22 28
8
-0.2
26
16
15
25
-0.5
1
4
12
7
-0.4
***
-1.0
3
11
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.12344 -0.00794 -0.00419 -0.00794 0.049063 -0.08467 -0.03617 0.074832 0.065832 -0.03617
Table 26. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.12346 -0.00793 -0.00418 -0.00793 0.049062 -0.09975 -0.05731 0.068481 0.069451 -0.04651
109
Table 28. Predicted vs. observed values (second model).
Figure 49 . Cellulose ether modification vs. cellulose ether. amount (left) and vs. polymer amount
(right) in slip testing.
110
Mapei Slip
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
0.4
5
14
6
*** ***
13
2
10
1.0
24
0.2
27 32
23
0.5
31
0.0
1817
19 20
Fitted Value
33 30
**
2129
26
22 9 28
8
0.0
-0.2
16
25 15
12
4
7
1
-0.4
-0.5
*** 3
11
-0.6
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.174 0.016375 -0.06463 0.0065 0.036875 -0.16137 0.105132 0.094632 -0.00237 -0.05587
Table 30. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.17398 0.016382 -0.0646 0.006484 0.036864 -0.19839 0.073166 0.114202 -0.01761 -0.06575
111
Table 32. Predicted vs. observed values (second model).
Figure 50. Cellulose ether modification vs. cellulose ether. amount (left) and vs. polymer amount
(right) in Mapei slip testing.
112
Creep Tests
At 200 Pa
Coefficient plot and experimental vs. fitted plot (from R):
2.5
***
6
14
0.4
*** ***
24
5
2.0
2
0.2
*
13
1.5
1 32
** *
2710
*
0.0
**
Fitted Value
9
1.0
31
2833
23
*
1918 1720
30
21 8
29
-0.2
26
0.5
25
22
12
4
15
-0.4
16
7
0.0
-0.6
-0.5
***
3
11
-0.8
-1.0
1 2 3 4 5 6 7 8 9 11 13 15 17 19 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
Experimental Value
Table 33. Coefficients of creep at 200 Pa. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.0778 -0.0538 0.075958 -0.02624 -0.04985 -0.0542 -0.01377 0.355235 -0.16428 -0.11033
Table 34. Predicted vs. observed values of verification points (first model).
Test n° Predicted (%) Observed (%) Difference (%) Confidence interval (%)
30 5.6 7.2 -1.6
31 8.0 12.2 -4.2
± 3.9
32 20.0 41.5 -21.5
33 7.4 8.8 -1.4
Table 35. Coefficients of creep at 200 Pa. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0,0778 -0,0538 0,075958 -0,02624 -0,04985 -0,06195 -0,02737 0,376314 -0,19224 -0,1138
113
Table 36. Predicted vs. observed values (second model).
At 300 Pa
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
**
24
0.5
2.5
14 5
2
*** ***
2.0
13
*
1 32
1.5
910
27
*
Fitted Value
*
0.0
31 23
2817 3320
**
19 18
30
21 29
1.0
25
22 826
12 4
15
16
0.5
7
-0.5
0.0
***
11
-0.5
Experimental Value
Table 37. Coefficients of creep at 300 Pa. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.09077 -0.01482 0.080802 -0.00224 -0.01161 -0.0759 -0.09141 0.424269 -0.15433 -0.09507
Table 38. Predicted vs. observed values of verification points (first model).
Test n° Predicted (%) Observed (%) Difference (%) Confidence interval (%)
30 13 17.6 -4.6
31 18.5 28.8 -10.3
± 8,7
32 49.6 107 -57.4
33 17.2 22.5 -5.3
114
Table 39. Coefficients of creep at 300 Pa. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.09077 -0.01482 0.080802 -0.00224 -0.01161 -0.08176 -0.10748 0.444427 -0.18583 -0.09926
Figure 51. CE modification vs polymer (left) and vs. cellulose (right) in creep test at 200 Pa.
115
Figure 52. Cement vs. cellulose (left); filler vs. polymer (right) in creep test at 300 Pa.
116
Adjustability
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
6
14
5
50
***
10
*
2
40
24
5
27
Fitted Value
32
30
23
0
13
10 31
20 19
17 18
33
28 30
16
20
26
29
21
-5
22 8
15
25 9
7
1
10
-10
***
4
12
3
11
0
-15
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0 10 20 30 40 50
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.625 -0.25 -3.375 2.5 1.625 -2.09367 5.406332 1.406332 -1.09367 -2.09367
Table 42. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.625 -0.25 -3.375 2.5 1.625 -3.72868 4.093238 2.679953 -1.74197 -2.83167
117
Table 44. Predicted vs. observed values (second model).
Figure 53. CE mod. vs. cellulose amount (left); filler vs. cement amount (right).
118
Figure 54. Polymer vs. cellulose amount.
Polymer always increases setting time but less than cellulose ether and especially when
there is a high amount of modified CE and low amount of cellulose ether (Figure 54
with high amount of modified CE, 0.25% of cellulose and 30% of cement).
119
24 h
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
1.4
14
***
0.2
24 29
16
1.2
1232 31
23
0.1
19 18 9
1730
20
10 21 26
33
1.0
27
25
28
Fitted Value
2
6
0.0
4 13
0.8
22
-0.1
***
1
0.6
5
-0.2
11
8
15
0.4
3
-0.3
0.2
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.02625 0.06 0.06375 0.06 0.0175 -0.00525 -0.08525 0.004749 -0.18525 -0.06025
Table 46. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.02625 0.06 0.06375 0.06 0.0175 -0.01886 -0.08429 -0.00026 -0.17488 -0.05072
120
Table 48. Predicted vs. observed values (second model).
Figure 55. Cement vs. CE modification (left) and vs. filler (right) in 24h testing.
121
CE modification penalize the result because it lowers the wettability of the adhesive.
Filler penalizes this pull-off test when it is present a high amount of cement: it with-
draws the water from the paste matrix penalizing the hydration of the cement and the
wettability. It also enhances the pull-off values when there is a low amount of cement
and a high amount of cellulose ether.
Figure 56. Cellulose vs. cement (left) and filler vs. polymer (right) in 24h testing.
122
Open Time at 5’
Coefficients Experimental vs. Fitted Values
***
0.8
14
0.6
4
3
23
16
2
*
24
***
0.4
6
1231
832 25
** *
20
Fitted Value
17 18
19 33
27
30
2
0.2
21 13
28
29
26
10
5
0.0
*
22
*
1
15 9
3
1
-0.2
7
-0.4
11
0
-0.6
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0 1 2 3
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.02375 0.04375 0.16625 0.08625 -0.0275 -0.29289 -0.15789 0.372111 -0.00789 -0.10789
Table 50. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.02375 0.04375 0.16625 0.08625 -0.0275 -0.29825 -0.18019 0.350492 -0.02924 -0.08924
123
Table 52. Predicted vs. observed values (second model).
Figure 57. O.T. 5': cement vs. cellulose amount (left); modified cellulose vs. polymer amount (right).
124
Open Time at 20’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
*** ***
0.3
1.5
0.2
**
14
2
0.1
Fitted Value
1.0
24 5
28
620 27
0.0
*
31
32 33 25
23
**
17 18 19
**
30
-0.1
1629
7 214
***
0.5
*** *
26
1513
-0.2
22
9
12
10
11
3
-0.3
0.0
1
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.03375 -0.065 0.00125 0.1175 -0.11 0.003879 -0.09612 0.143879 -0.18612 -0.02112
Table 54. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.03375 -0.065 0.00125 0.1175 -0.11 0.007753 -0.0973 0.142371 -0.19358 -0.0176
125
Table 56. Predicted vs. observed values (second model).
Figure 58. O.T. 20’: cement vs. cellulose amount (left); filler vs. polymer amount (right).
126
Open Time at 30’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
1.2
8
***
1.0
***
0.2
14
0.8
*
0.1
Fitted Value
56
0.6
2
24 27
0.0
20
28 25
*
23
0.4
33 31 32
17 19 18
***
16 30
-0.1
***
15
297
0.2
21
4
26
22
12
-0.2
3
13
0.0
9
10
11
1
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.065 -0.07625 -0.02 0.04125 -0.055 -0.01735 -0.01235 0.127652 -0.09735 -0.03235
Table 58. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.065 -0.07625 -0.02 0.04125 -0.055 -0.03515 -0.01968 0.133326 -0.10312 -0.02422
127
Table 60. Predicted vs. observed values (second model).
Figure 59. O.T. 30’: cement vs. cellulose amount (left); filler vs. polymer amount (right).
128
Initial Adhesion
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
1.0
12
3.5
14
23
3.0
4 24
31 102
6
0.5
32 29
2.5
***
18 19 16 25 17
20 30
Fitted Value
*
13
33
27 26 21
2.0
9 28
8
0.0
1.5
***
22
** 15
1.0
5
7
113
0.5
-0.5
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.21563 0.190625 0.081875 0.050625 -0.02063 -0.19873 -0.27873 0.291266 -0.12373 -0.20373
Table 62. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.21563 0.190625 0.081875 0.050625 -0.02063 -0.17543 -0.25542 0.276464 -0.133 -0.17194
129
Table 64. Predicted vs. observed values (second model).
Figure 60. Cement vs. polymer amount (left) and vs. cellulose amount (right).
130
Cellulose ether improve pull-off results only when there is a low amount of modified
CE, otherwise, it lowers the performance when it is present in more than 0.35% (Figure
49 on the right with low amount of modified CE, 2.5% of polymer and 20% of filler).
It also enhances the values when there is a high amount of filler, increasing the wetta-
bility of the adhesive.
131
Water Immersion
At 5’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
1.6
***
12
0.3
1.4
14
0.2
29
13
24
1.2
0.1
15
10
Fitted Value
32
25
3119
26 27 17 18
20
1.0
23
16
0.0
30
*
33
*
22
21 2
0.8
-0.1
11
28
46
8 5
0.6
-0.2
7
1
3
-0.3
0.4
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Experimental Value
Table 65. Coefficients of water immersion at 5’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.0575 0.05 0.045 -0.015 -0.01875 0.024551 -0.00545 0.109551 -0.09545 -0.14045
Table 66. Predicted vs. observed values of verification points (first model).
Table 67. Coefficients of water immersion at 5’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.0575 0.05 0.045 -0.015 -0.01875 0.016412 -0.0066 0.108472 -0.10365 -0.12455
132
Table 68. Predicted vs. observed values (second model).
At 10’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
1.2
***
0.15
14
0.10
**
1.0
12
15
* *
0.05
29
20 9
0.8
Fitted Value
13
16
0.00
27
182331
3217 19
24
26
30
33
2 2528
0.6
-0.05
**
4
21 5
22
86
3
10
1
***
-0.10
0.4
11
-0.15
7
0.2
Experimental Value
Table 69. Coefficients of water immersion at 10’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.00937 0.004375 0.033125 -0.00562 -0.00062 0.064881 0.009881 -0.04012 -0.06012 0.009881
Table 70. Predicted vs. observed values of verification points (first model).
133
Table 71. Coefficients of water immersion at 10’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.00938 0.004375 0.033125 -0.00562 -0.00062 0.055586 0.0045 -0.04205 -0.06152 0.015354
At 15’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
0.8
0.2
20 14
15
**
12
0.6
0.1
9
2423
31
27 3218 13 2
17
1 9
2930625 33
16 21
22
4 268
Fitted Value
0.0
0.4
28 5
1
*
3
***
-0.1
0.2
-0.2
11
7
10
0.0
Experimental Value
Table 73. Coefficients of water immersion at 15’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.01 0.03125 0.05875 -0.0175 0.00875 -0.11797 -0.03297 0.007032 -0.01297 0.072032
134
Table 74. Predicted vs. observed values of verification points (first model).
Table 75. Coefficients of water immersion at 15’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.01 0.03125 0.05875 -0.0175 0.00875 -0.09367 -0.04295 -0.00752 -0.02406 0.060088
135
RSC and Discussion
Also in this test, cement greatly increases pull-off resistance, especially at 5’. Increasing open
time, the adhesive needs more cellulose in order to obtain good performances, especially at 15’
and when the quantity of filler and cement are high (Figure 62, Figure 63, Figure 64 on the
left: average quantity of CE mod., 3% of polymer and 15% of filler). Filler penalize all the
results at 5’, 10’ and 15’. Cellulose ether modification does not seem to have an important role
in this experiment.
Figure 62. Water immersion at 5’: cellulose vs. cement amount (left); and CE mod. vs polymer
amount (right).
Figure 63. Water immersion at 10’: cellulose vs. cement amount (left); and CE mod. vs polymer
amount (right).
136
An interesting fact is that in this test the polymer seems to improve the performance of the
adhesive only when there is a high amount of modified CE at 5’, 10’, and especially at 15’. On
the other hand, it lowers the results when the quantity of modified CE is lower (Figure 62,
Figure 63, Figure 64, on the right with 30% of cement, 0.30% of cellulose and 15% of filler).
Figure 64. Water immersion at 15’: cellulose vs. cement amount (left); and CE mod. vs polymer
amount (right).
137
Heat Aging at 5’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
4
12
16
1.0
14
23
3
431
2410
0.5
6 25
***
29
32
18 27
226 17
19
Fitted Value
30
**
20 33 21
2
* * *
8
13 28
9
0.0
22
1
1
15
5
73
11
-0.5
0
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0 1 2 3 4
Experimental Value
Table 77. Coefficients of heat aging at 5’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.0375 0.18625 0.275 0.00875 0.00375 -0.25553 -0.03053 0.219472 -0.17553 -0.26053
Table 78. Predicted vs. observed values of verification points (first model).
Table 79. Coefficients of heat aging at 5’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.0375 0.18625 0.275 0.00875 0.00375 -0.25917 -0.01433 0.217344 -0.16761 -0.24113
138
Table 80. Predicted vs. observed values (second model).
Figure 65. Heat aging at 5’: cellulose vs. cement amount (left); filler vs. polymer amount (right).
139
Heat Aging at 10’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
14
3.0
0.6
***
2.5
0.4
**
6
2.0
*
16
*
0.2
23
Fitted Value
27
24
31 8
2932 20
1.5
218
17 12 19
25
33
0.0
30
28
21 13
** ** *
1.0
26
-0.2
22 5 15
9
0.5
4
10
-0.4
7
1
3
0.0
11
Experimental Value
Table 81. Coefficients of heat aging at 10’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.215625 0.144375 0.003125 0.056875 -0.16313 -0.05113 -0.14113 0.113865 -0.21113 -0.20113
Table 82. Predicted vs. observed values of verification points (first model).
Table 83. Coefficients of heat aging at 10’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.215625 0.144375 0.003125 0.056875 -0.16313 -0.05732 -0.13508 0.102246 -0.23759 -0.14882
140
Table 84. Predicted vs. observed values (second model).
Figure 66. Heat aging at 10’: cellulose vs. cement amount (left); filler vs. polymer amount (right).
141
Heat Aging at 15’
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
2.5
14
0.6
***
2.0
0.4
*
0.2
1.5
2 23
Fitted Value
24
0.0
31
6 12
3225
1.0
17 2718 19
**
20 33
-0.2
***
28
29 30
22
26
0.5
-0.4
5
4
21 15
13
-0.6
7
16 9
3
0.0
1
10
11
-0.8
Experimental Value
Table 85. Coefficients of heat aging at 15’. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.08375 -0.0525 -0.09125 0.035 -0.2775 -0.18365 -0.14365 0.276346 0.111346 -0.37365
Table 86. Predicted vs. observed values of verification points (first model).
Table 87. Coefficients of heat aging at 15’. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.08375 -0.0525 -0.09125 0.035 -0.2775 -0.19638 -0.15961 0.244488 0.127511 -0.36648
142
Table 88. Predicted vs. observed values (second model).
Figure 67. Heat aging at 15’: cellulose vs. cement amount (left); filler vs. polymer amount (right).
143
Also in this test, polymer improves the overall performance.
Cellulose ether modification is always negative in any formulation (Figure 68).
144
Freeze/Thaw Cycles
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
2.5
***
16
10
9
0.4
29
11 12
2.0
15
14 26
0.2
24 13 17 18
Fitted Value
32 2330
31
19 25
2221
27
20
1.5
33 1
0.0
3 6
2 28
7
1.0
8
5
-0.2
0.5
-0.4
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.09125 0.065 -0.015 -0.02875 0.0375 -0.02222 -0.03222 0.012784 -0.04722 -0.07722
Table 90. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.09125 0.065 -0.015 -0.02875 0.0375 -0.01723 -0.00053 0.013897 -0.02604 -0.06232
145
Table 92. Predicted vs. observed values (second model).
Figure 69. CE mod. Vs. cellulose amount (left); cement vs. polymer amount (right).
146
Deformability
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
*** 8
3.5
0.6
6
23
4
3.0
16
0.4
14
31
Fitted Value
12
10
2.5
27
3320 19
28
32
0.2
18
17
*
2421
26 30
***
25
**
29
2.0
13
* *
5
0.0
**
15 22
***
11
19
1.5
7
***
3
-0.2
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.010625 -0.00813 0.056875 0.005625 -0.07438 -0.07612 0.063879 -0.08612 0.128879 -0.01112
Table 94. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.010625 -0.00813 0.056875 0.005625 -0.07438 -0.0838 0.066869 -0.08121 0.126535 -0.00204
147
Table 96. Predicted vs. observed values (second model).
Figure 70. Cement vs. polymer amount (left) and filler vs. cellulose amount (right) in deformability
test.
148
Deformability Load
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
1110
16
3
12
14
2
**
29 14
12
*
Fitted Value
24 26 2125
1
16
4 15
13
31 30
32
10
1918 20
17 23
62
0
27
*
33
8
1
22
8
***
-1
28
6
7
5
3
-2
4
1 2 3 4 5 6 7 8 9 11 13 15 17 19 4 6 8 10 12 14 16
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.07625 -0.6725 -0.2175 0.04125 -0.235 -0.97964 -0.26964 1.250356 -0.82964 0.710356
Table 98. Predicted vs. observed values of verification points (first model).
Test n° Predicted (N) Observed (N) Difference (N) Confidence interval (N)
30 10.5 8.66 1.84
31 10.3 7.79 2.51
± 0.34
32 10.3 8.84 1.46
33 8.21 7.46 0.75
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.07625 -0.6725 -0.2175 0.04125 -0.235 -0.79843 -0.05004 1.340977 -0.86075 0.761907
149
Table 100. Predicted vs. observed values (second model).
Figure 71. Cement vs. polymer amount (left) and filler vs. cellulose amount (right) in deformability
load test.
150
DMA ΔE’ (5-50 °C)
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
4000
*** 12
1000
10
3000
82
14
500
23
31
Fitted Value
25
*
24
26 20
21 166
2000
30 32 19
29 1817
33
27
0
28
9
***
11
1000
13
15
3 1
-500
57 22
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-130.625 -66.875 75.875 -54.625 -49.75 -268.15 -78.6504 320.3496 -434.65 233.8496
Table 102. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-130.625 -66.875 75.875 -54.625 -49.75 -233.557 -17.8989 335.709 -407.547 237.6525
151
Table 104. Predicted vs. observed values (second model).
152
Figure 72. Difference of the module between 5 and 50 °C.
These results are related to those of deformability. An adhesive that contains a lot of polymer
will have a module that varies greatly due to the glass transition and will have a high deforma-
bility value (Figure 70).
153
Stress Sweep: G’ and G” at 200 and 300 Pa
G’ at 200 Pa
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
0.6
3
11
3.0
0.4
2.5
***
0.2
* * * **
2.0
Fitted Value
7
16
12
2622
15
25 214
0.0
1.5
29
*
30 20 10
828
9 18
331 1719
13
27 31
23
1.0
-0.2
32
5
*** *** ** 2
0.5
24
14
6
-0.4
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.105375 0.036587 -0.09113 -0.03127 0.035843 0.121995 0.144874 -0.29781 0.026969 0.186669
Table 106. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.105375 0.036587 -0.09113 -0.03127 0.035843 0.117337 0.146645 -0.30056 0.057345 0.166053
154
Table 108. Predicted vs. observed values (second model).
G’ at 300 Pa
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
0.6
3.0
0.4
2.5 11
** **
*
0.2
2.0
Fitted Value
7
16
0.0
1.5
*
26 12
22
29 15
4
21
259
28
30 20810
33 19
18
17
1.0
*** ***
1
-0.2
27 31
23 13
32
**
5
0.5
-0.4
14
2
246
0.0
-0.6
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.138689 0.04847 -0.08054 -0.03546 0.072024 0.180241 0.14495 -0.38951 0.039524 0.177841
Table 110. Predicted vs. observed values of verification points (first model).
155
Table 111. Coefficients of G’ at 300 Pa. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.138689 0.04847 -0.08054 -0.03546 0.072024 0.171538 0.151523 -0.3897 0.078978 0.156031
G” at 200 Pa
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
*** 11
2.5
3
0.4
*
2.0
**
0.2
**
7
*
26 12
1.5
Fitted Value
1516
29 2122
49
0.0
30 2025
10
1 28
33 19
18
17
278
1.0
31
13
2332
-0.2
*** ***
5
**
0.5
2
-0.4
24
6
14
0.0
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.050686 0.03331 -0.01697 -0.02138 0.030181 0.10652 0.200063 -0.36241 -0.05801 0.191763
156
Table 114. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.050686 0.03331 -0.01697 -0.02138 0.030181 0.114833 0.206406 -0.36988 -0.02867 0.170031
G” at 300 Pa
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
0.6
2.5
3
11
0.4
2.0
*
0.2
7
1.5
26 12
16
15
Fitted Value
0.0
4
29 22
20 928
21
8
2
***
1.0
3033 25
10 171819
271
-0.2
31
***
23
32
13
0.5
**
-0.4
14
0.0
-0.6
246
Experimental Value
157
Table 117. Coefficients of G” at 300 Pa. Model without verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.03355 0.095981 -0.09677 0.041477 0.002138 0.21997 0.193347 -0.45191 -0.01572 0.215366
Table 118. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.03355 0.095981 -0.09677 0.041477 0.002138 0.217003 0.20361 -0.46182 0.025833 0.200344
158
RSC Discussion
The results expressed in the next figures, represent how the elastic and viscous modulus at 200
and 300 Pa are influenced by the five ingredients.
It can immediately be deduced that the ingredients influence both the modules in the same way
at both 200 and 300 Pa.
What is different is that G' is always greater than G" both at 200 and at 300 Pa because the
oscillatory stress is not yet sufficient to breakdown the structure of the fresh adhesive. At these
values therefore, the elastic behavior prevails over that viscous one.
At 300 Pa both the modulus are always lower than those at 200 Pa (in equal formulations).
In particular, we can deduce that:
CE modification greatly increase, in all formulation, G’ and G” at 200 and 300 Pa (Fig-
ure 73 with: 30% of cement, 3% of polymer and 10% of filler. Figure 74 with: 30% of
cement, 0.35% of cellulose ether and 10% of filler).
Cellulose ether and polymer, on the contrary, are the two ingredients that decrease both
modules in both tests (Figure 73 and Figure 74).
k k
Figure 73. Elastic (on the left) and viscous (on the right) modulus at 200 Pa.
159
k k
Figure 74. Elastic (on the left) and viscous (on the right) modulus at 300 Pa.
Cement and filler, on the other hand, contribute less than the ingredients mentioned
above: both increase the modules a lot, but only when there is a low amount of polymer
and cellulose and a high amount of modification. This because, as previously men-
tioned, modification, cellulose and polymer have more influence than cement and car-
bonate and therefore tend to cover the effects of the latter (Figure 75 with: high quantity
of modified cellulose, 1% of polymer and 15% of filler) (Figure 76 with 35% of cement,
0.30% of cellulose and high quantity of modified cellulose).
k k
Figure 75. Elastic (on the left) and viscous (on the right) modulus at 200 Pa.
160
k k
Figure 76. Elastic (on the left) and viscous (on the right) modulus at 300 Pa.
It is interesting to note that these trends can be compared to those of slips and creeps, but the
values they present are opposite. In fact, the ingredients that decrease the value of slips and
creeps, increase those of the elastic and viscous modules; while on the contrary, the ingredients
that increase the values of the slips and creeps, decrease in turn G' and G".
161
Stress Sweep: G’ and G” at Plateau
G’ at Plateau
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
3.5
311
***
0.5
3.0
* * 2622 7
2.5
* **
21 124
15
16
30 20
2.0
Fitted Value
27 171819
0.0
25 23 9
*
31
1 33
5
13
32
1.5
2
*** ***
29 10
28
6
**
1.0
24
14
**
-0.5
0.5
8
0.0
1 2 3 4 5 6 7 8 9 11 13 15 17 19 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.08885 0.164374 -0.12305 0.153542 0.142202 -0.59759 0.365232 -0.48335 0.319856 0.405515
Table 122. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.08885 0.164374 -0.12305 0.153542 0.142202 -0.58256 0.353252 -0.49215 0.326945 0.376261
162
Table 124. Predicted vs. observed values (second model).
G” at Plateau
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
3 11
2.5
0.5
7
4
2622
2.0
12
21 15
20 16
8
Fitted Value
0.0
30
27
* **
23 17 19
18
25 33
1 9 31
1.5
28
29 5
2 1332
*
10
-0.5
1.0
14 24
0.5
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.00468 0.025391 0.026122 0.02946 0.010267 -0.20684 0.248355 -0.51823 0.211462 0.301675
Table 126. Predicted vs. observed values of verification points (first model).
163
Table 127. Coefficients of G” at plateau. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.00468 0.025391 0.026122 0.02946 0.010267 -0.20801 0.25065 -0.50879 0.218229 0.289684
164
k k
Figure 77. Elastic (on the left) and viscous (on the right) modulus at plateau.
This result confirms the correlation previously seen using the PCA: the loadings of the modules
are inversely correlated to the slip and creep ones. In fact, it is noteworthy that the graph has
the same shape as that of the slips and creeps (fig.). This means that an adhesive with high
modules will give low creep and slip values but also low performances in extended open time
tests.
In addition, cement would seem to increase both modules only up to a certain amount (30%)
above which it decreases them. Calcium carbonate, on the other hand, does not seem to make a
big contribution (Figure 78).
k k
Figure 78. Elastic (on the left) and viscous (on the right) modulus at plateau.
165
Stress Sweep: G’ ½
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
3
0.3
10
2.5
11
***
0.2
**
7
*
15
12
16
2.0
0.1
Fitted Value
29
26 25
4 22
21
13
30 8 28
17 18
19
33
0.0
23 31
20
27
32
1.5
1
9
** ****
5
2
6
-0.1
***
24
-0.2
1.0
14
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.06207 0.037412 -0.10542 -0.12522 0.087588 0.112873 0.024828 -0.07723 0.024828 0.024828
Table 130. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.06207 0.037412 -0.10542 -0.12522 0.087588 0.133532 0.040826 -0.10142 0.049851 0.024631
166
Table 132. Predicted vs. observed values (second model).
k
k
Figure 79. Modified C.E. vs. polymer (left). Cellulose ether vs. cement (right) in stress sweep testing
at G’ ½ .
167
Yield
Yield TA
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
0.4
3
0.3
5.5
10
11
0.2
***
7
***
15
Fitted Value
**
4
5.0
25
1612
*
0.1
8
22
26 21
29
0.0
2320 30 931
1719
18
33
4.5
28
*
**
32
5
27
-0.1
***
24 2
1
13
***
6
14
4.0
-0.2
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.04559 -0.02899 -0.1454 -0.05805 0.153364 0.016246 -0.06563 0.090162 0.090099 0.0899
Table 134. Predicted vs. observed values of verification points (first model).
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
-0.04559 -0.02899 -0.1454 -0.05805 0.153364 0.013868 -0.06723 0.087867 0.100659 0.08188
168
Table 136. Predicted vs. observed values (second model).
Yield AP
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
4.8
11
0.3
7
4.6
3
0.2
4.4
15
16
12
425
Fitted Value
4.2
*
9
***
8 22
0.1
21
*
26
20
*
4.0
30
29
18
17
19
33
23 28 31
27
10
0.0
*
3.8
32 52
1
*** *** 13
24
3.6
***
6
-0.1
14
3.4
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.004851 0.031323 -0.03591 0.087568 0.015089 -0.01883 -0.01101 0.001631 0.055871 0.083542
Table 138. Predicted vs. observed values of verification points (first model).
169
Table 139. Coefficients of yield AP. Model with verification points.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.004851 0.031323 -0.03591 0.087568 0.015089 -0.02225 -0.00648 -0.00701 0.06338 0.094118
Figure 80. CE mod. vs. cement amount in yield test using parallel-plate (left) and ball-measuring sys-
tem geometry (right).
The cement and the modified cellulose increase the yield value although the latter does it more
(especially using the parallel-plate geometry) (Figure 80). On the contrary, the polymer and
170
the cellulose lower it for the same reasons explained in the PCA considerations. In particular,
the cement gives a major contribution only when there is a low amount of cellulose (0.25% in
this case) and a high amount of its modified formula and the cellulose and polymer lower the
results more when there is a low amount of modified cellulose (Figure 81).
The same can be said for both instruments with the difference that the TA instrument measures
a higher yield than the Anton Paar does. This is because two different geometries were used.
Figure 81. . Cellulose ether vs. polymer amount in yield test using parallel-plate (left) and ball-meas-
uring system geometry (right).
Dustiness
E Min
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
4
6
25
***
3
4
12
2
20
1
16
Fitted Value
10
**
23
21
1
7
14
8 28
15
2519
2724 17 18
26
0
29 2
*
** *
11
20
13
-1
22
10
***
15
3
59
-2
1 2 3 4 5 6 7 8 9 11 13 15 17 19 5 10 15 20 25
Experimental Value
171
Table 141. Coefficients of E Min. Dustiness Model.
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.40625 0.26875 0.04375 -0.86875 -0.61875 0.354617 -0.09538 0.204617 0.054617 0.504617
E Max
Coefficient plot and experimental vs. fitted plot (from R):
Coefficients Experimental vs. Fitted Values
***
10
12
60
6
14
**
16
50
5
* *
21
29
23
8
10
47
Fitted Value
28
1817 13
19 20
40
27 25
0
24
1
*
1511 26
**
2
30
22
-5
9
20
35
-10
1 2 3 4 5 6 7 8 9 11 13 15 17 19 20 30 40 50 60
Experimental Value
V2V4 V2V5 V3V4 V3V5 V4V5 V1² V2² V3² V4² V5²
0.33125 1.43125 0.51875 -2.48125 -3.14375 4.440237 -4.55976 -2.30976 -2.25976 4.190237
172
Figure 82. Coarse carbonate vs. Cement amount (left) and fibers vs. polymer amount (right) in initial
dust formation.
173
The second mechanism mostly involves light components of the formulation, like fibers and
polymer that, after 30 second from powder pouring, are still fluctuating, generating the perma-
nent cloud of dust, as shown in the following figure.
Figure 83. Filler vs. Cement amount (left) and fibers vs. polymer amount (right) in permanent dust
development.
174
IV. CONCLUSIONS
This work confirms that chemometrics demonstrates to have the possibility to strongly improve
the knowledge generated by experimentations, as the definitions of rational Experimental De-
signs and the multivariate analysis approach can give a wide ranging among information that is
usually lost when evaluating tests in single points. In addition a univariate approach do not take
in consideration the interaction between the high amount of ingredients and variables of a com-
plex system.
This discipline, applied to cement based product science, also demonstrated its extreme flexi-
bility and adaptability to any kind of experimentation and its perfect applicability in formulation
science if it is setup correctly, clearly surpassing the classical univariate approach.
The tile adhesive model was able to predict the behavior of the adhesive from the rheological
and dynamic-mechanical point of view, in its fresh and hardened state.
The dustiness model, for its part, made us understand the way in which the dustiness of dry
blends depends on the fine and coarse components.
Many further studies can be carried in R&D laboratories of Mapei using these techniques; an
example could be the realization of a model able to foresee the coloring of the tile joints and
therefore able to predict the right dosage of the inorganic pigments.
175
References
1
Varmuza K.; Filzmoser P. (2009) Introduction to Multivariate Statistical Analysis in Chemometrics. Boca Raton,
FL. CRC Press.
2
Kramer R. (1998) Chemometric Techniques for Quantitative Analysis. New York, NY. Marcel Dekker, Inc.
3
Varmuza K. (2012) Chemometrics in Practical Applications. Rijeka. InTech.
4
Patricia M.; Sthor-Hunt (1996) An Analysis of Frequency of Hands-on Experience and Science Achievement. J.
of Research in Science Teaching. 33.
5
Monaco M.; Carrà S. (2018) Chemometrics Application in Cementitious Tile Adhesives Formulation.
6
Box G.E.; Hunter J.S.; Hunter W.G. (2005). Statistics for Experimenters: Design, Innovation, and Discovery,
2nd edition. Wiley.
7
Oehlert G.W. (2010). A First Course in Design and Analysis of Experiment. Minneapolis, MN. University of
Minnesota.
8
Joliffe I.T. (2010) Principal Component Analysis. 2nd edition. Springer.
9
Kinloch A.J. (1987). Adhesion and Adhesives: Science and Technology (Reprinted ed.). London. Chapman and
Hall. p. 1.
10
Mazza, P et al. (2006). A new Paleolithic discovery: tar-hafted stone tools in a European Mid-Pleistocene bone-
bearing bed. J. of Archaeological Science. 33 (9): 1310.
11
Van-Tien P. (2012) Relationship between the adhesive properties and the rheological behavior of fresh mortars.
Cachan. ENS.
12
Comyn J. (1997) Adhesion Science. Cambridge. RSC Paperbacks.
13
Bayer R.; Lutz H. (2005) Dry Mortars. Weinheim. Wiley-VCH Verlag GmbH & Co.
14
Konietzko A. (1985) The Application of Modern Dry, Factory Mixed, Mortar Products. ZKG International. 48,
no. 12, 625–659.
15
Data from EMO (European Mortar Organization) and Wacker Polymer Systems GmbH & Co. KG.
16
Fritze P. (2002) Deformability and Water Resistance of C1 and C2 Adhesives According to EN 12004 and EN
12002. Burghausen. Waker Polymer System.
17
ASTM C185-15a (2015) Standard Test Method for Air Content of Hydraulic Cement Mortar. www.ASTM.org.
West Conshohocken, PA. ASTM International.doi:10.1520/C0185-15A.
18
Mindess S.; Young F. J. (1981): Concrete. Englewood Cliffs, NJ. Prentice-Hall, Inc. pp. 671.
19
Cody A.M.; Lee H.; Cody R.D.; Spry P.G. (2003) The Effects of Chemical Environment on the Nucleation,
Growth, and Stability of Ettringite. Cement and Concrete Research. 34. pp. 869–881.
20
Kosmatka S.; Panarese W. (1988) Design and Control of Concrete Mixes, Portland Cement Association, Skokie,
Ill. pp. 205.
21
Mamlouk M.; Zaniewski J. (1999): Materials for Civil and Construction Engineers. Addison Wesley Longman,
Inc.
22
Crawford R.L. (1981). Lignin biodegradation and transformation. New York, NY. John Wiley and Sons.
23
Updegraff D.M. (1969). Semimicro Determination of Cellulose in Biological Materials. Analytical Biochemis-
try. 32 (3): 420–424.
24
Müller I. (2007) Influence of Cellulose Ethers on the Kinetics of Early Portland Cement Hydration. Karlsruher
Mineralogische und Geochemische Hefte: Schriftenreihe des Instituts für Mineralogie und Geochemie; 32.
25
Hayakawa K.; Soshiroda T. (1985) An Experimental Study on the Concrete Containing Cellulose Ether. Trans.
Japan Concrete Institute, 7, pp. 17-24.
26
Hayakawa K.; Soshiroda T. (1986) Effects of cellulose ether on bond between matrix and aggregate in concrete.
Adhesion Between Polymers and Concrete Proceedings. London. Chapman & Hall, pp. 22-31.
27
Tanaka Y.; Uryu T.; Yaguchi M. (1996) US Patent No. 5,573,589. Cement compositions containing a sulfated
polysaccharide and method.
176
28
Yamamuro H. (1999) Property of new polisaccaride derivative as a viscosity agent for self compacting concrete.
First International Rilem Symposium on Self Compacting Concrete. Stockholm, Sweden. RILEM Publications,
Bagneux, France. pp 449-459.
29
Ghio V.A.; Monteiro P. J. M. (1998) The Effects of Polysaccharide Gum Additives on the Shotcrete Process.
ACI Mater J. 95(2), pp. 152-157.
30
Ghio V.A.; Monteiro P. J. M.; Demsetz L. A. (1994) The Rheology of Fresh Cement Paste Containing Polyssa-
charide Gums. Cement and Concrete Research. 24(2), pp. 243-249.
31
Pourchez A. et al. (2006) HPMC and HEMC Influence on Cement Hydration. Cement and Concrete Research.
36(2), pp. 288-294.
32
Felixberger J.K. (2008) Polymer-Modified Thin-Bed Tile Adhesives. Augsburg. BASF SE – PCI.
33
Pakravan H.R.; Jamshidi M.; Latifi M. (2009) A Study on Bonding Strength of Polymeric Fibers to Cementitious
Matrix. PAN.
34
Van-Tien P. (2012) Relationship Between the Adhesive Properties and the Rheological Behavior of Fresh Mor-
tars. Cachan. ENS Cachan.
35
Eley R.R. (2005) Applied Rheology in the Protective and Decorative Coating Industry. Rheology Reviews. 173-
140.
36
Freeman Steffe J. (1996). Rheological Methods in Food Process Engineering. Freeman Press.
37
Beris A.N.; Giacomin A.J. (2014). πάντα ῥεῖ : Everything Flows. Applied Rheology. 24: 52918.
38
BS 5168: (1975) Glossary of Rheological Terms, London, British Standards Institution.
39
Schowalter W.R. (1978) Mechanics of Non-Newtonian Fluids. Pergamon Press.
40
Bird R.B.; Stewart W.E.; Lightfoot E.N. (1960) Transport Phenomena. John Wiley & Sons.
41
Bird R.B.; Curtiss C.F.; Armstrong R.C. (1989) Dynamics of Polymeric Liquids. Vol 1 & 2. Wiley Interscience.
42
Morrison F.A. (2001) Understanding Rheology. Oxford University Press.
43
Viswanath D.S. et al. (2007) Viscosity of Liquid. Dordrecht. Springer.
44
Meyers M.A.; Chawla K.K. (2008) Mechanical Behavior of Materials. Cambridge, Cambridge University Press.
pp. 98-103
45
Pileggi R.G. et al. Characterisation of Rendering Mortars by Squeeze-Flow and Rotational Rheometry. Depart-
ment of Construction Engineering, Escola Politécnica, University of São Paulo
46
Gruppo Italiano di Chemiometria, www.gruppochemiometria.it.
47
Monaco M. (2018) C2 Tile Adhesive General Model v1.17. Mapei S.p.A.
48
Monti M. (2008) Meccanismo di interazione di un etere di cellulosa con le fasi costituenti il clinker di un ce-
mento Portland 52,5 durante il processo di idratazione. Mapei S.p.A.; Unimib.
177