Zurbriggen R.,Wetzel A.,Harzer S., Herwegh M.,Kaufmann J.,Winnefeld F., Pass K. & Waser H. «Failure mechanisms of outdoor applied large sized fully-vitrified tiles»

Abstract

There is an increasing number of damages related
to the exterior application of large formatted porcelain
tiles. The ongoing trend towards such large-sized fullyvitrified
ceramic tiles produces two technical problems.
The less porous and the smoother the porcelain tile, the
more difficult it is for the mortar to provide adhesion.
Secondly, the larger the tile format, the longer the
distance from center to grout along which the strain
increments sum up to become largest at the tile edges.
This paper describes first results from a research program
on the mechanisms of shrinkage and adhesion of
large-sized porcelain tiles. Simple see-through experiments
allow for macroscopic observation of micronsized
shrinkage cracks and their propagation along the
tile-mortar interface. Due to an interference colour effect
these micro-cracks can be seen by the naked eye
in their early stage of formation when they are 0,5 to 5
microns in width. Continuous field observations of outdoor
applied glass tiles (ordinary float glass as used for
windows) indicate that the percolation rate for meteoric
water drastically increases along these micro-cracks
and that this is an important factor in the continued
failure history of the future damage.

1. Introduction
Technical recommendations

There are clear recommendations how to minimize
the risk for damages for the outdoor application of tiles (Reinmann, 2001). The first recommendation
is to provide proper wetting of the tile
by the mortar. Especially, for wall tile applications
where the mortar is more viscous and has a higher
yield for the purpose of anti-sagging properties, the
mortar ripples are not completely pressed, resulting
in a reduced wetting percentage (Pic. 2). It is
therefore recommended to apply the tiles by the
floating-buttering technique (DIN 18 157). Where
the mortar is applied on both sides, on the wall and
on the back side of the tile. This technique has two
advantages: (a) The fresh mortar is pressed onto
the back-side of the tile providing an ideal wetting
of the tile-mortar interface, (b), the tile is successively
layed after the fresh mortar has been applied
on its back-side, resulting in an improved bond to
the mortar which was trowelled onto the substrate
before. However, the floating-buttering technique
requires more material and consumes more time
compared to the simple floating technique. This is
the reason why the floating-buttering technique
is often not practiced, where the recommendation
clearly requires for it.

In case of floor tiling proper wetting can also be
achieved by the use of lower viscous mortars, so-called
flow-bed mortars (German: «Fliessbettmurtel»).

The second recommendation is to apply a flexible
membrane to absorb the differential shrinkage
between the substrate (e.g., screed) and the rigid tiles.
Again, this recommendation is often not followed under
the pressure of time and money.

A third important recommendation is to use
adhesive mortars with an extended open time. Especially,
when temperatures are high and/or wind
causes the surface of the freshly applied mortars to
dry much quicker than under normal laboratory conditions,
skin formation becomes the limiting factor for
adhesion properties (Oberste-Padtberg & Sieksmeier,
2007).

Market trend and material damage risk

The ongoing trend for large porcelain tiles might
explain the increase in complains, because the risk for
damages is increased (i) if the tile is large and (ii) the
material is fully-vitrified porcelain.

The larger the tile, the longer the distance (from
centre to rim) along which the differential strain increments
between tile and mortar are summing up,
which results in highest strain at the tile edges. Flexible
mortars can deform and reduce the stresses at
the mortar-tile interface. Rigid mortars do not so and
become stressed. Thus, highest stresses occur at the
tile edges (Felixberger, 2003). This explains why tiles
generally first fail at the edges.

Fully-vitrified ceramic tiles or fully-vitrified
stoneware tiles (German: «Feinsteinzeugfliesen»)
are dust-pressed ceramic tiles with water absorption levels of less than 0.5 wt.%, also known as impervious
stoneware tiles or porcelain stoneware. They are also
referred to as porcelain tiles, China tiles or with the
Italian term «Gres Porcellanato» because their raw
materials are similar to those used for manufacturing
China Porcelain. The term porcelain refers to a
wide range of ceramic products that have been fired
at temperatures above 1200°C to achieve vitreous or
glassy qualities (due to a high degree of sintering)
such as translucence and low porosity. Because of
the reduced water uptake capability fully-vitrified
ceramic tiles are frost proof and can be used in exterior
areas. On the other side, the smooth and glasslike
surface makes it very difficult for the mortar to
adhere, which is a reason for the increased risk for
adhesion failure.

The «failure history»

The failure of a tile is the final event of a longer
«failure history». Without knowing this history, at
least the critical parts of it, it is not possible to know
the reasons for the damage. In many cases the application
conditions are not reported and therefore the history
of failure lacks important information from the
very beginning. Due to the lack of such information
the discussion about the reasons for failure gets very
difficult. Especially, because relatively expensive repair
cost ask to be paid by someone. Experts are then
called for advice. But, as they first try to reconstruct
the failure history they will struggle with the lack of
information, as well.

Nobody wants damages to occur, but if they occur,
they are not only bad, but also interesting for
technicians and scientists, who want to learn more
about these materials and why they failed in this
particular case. The reconstruction of «failure histories
» and the resulting better understanding of damages
is also key for material improvements and new
developments. Therefore, Elotex has initiated a research
program with two research partners, namely
the University of Bern and the Swiss Federal Laboratories
for Material Testing and Research (Empa in
Dubendorf) with the goal to produce damages under
known conditions in order to be able to reconstruct
typical «failure histories».

2. The research program

Pic. 1 is an illustration of the different approaches
within this research program. Tile adhesives are
tested according to industrial standards (point 1.1
in Pic. 1), between single-layed large tiles and concrete
slabs (1.2) and in a large-scale field study (1.3),
where wall and floor area of several square meters
are covered with tiles. All these samples are studied
by different analytical techniques with the goal
to reconstruct the evolution of material properties
and structures on all scales (macro, micro and nanoscale;
points 2-5 in Pic. 1). Because drying is directional
from rim to centre and from substrate to tile,
the system develops different compositional and structural gradients with respect to humidity and
degree of hydration (point 6 in Pic. 1). All these material
parameters build the numeric base for computer
modelling of stress distributions (point 7 in Pic.
1), which deliver an important part of the «failure
history». Finally, everything should point to the figure
in the centre (point 8), the evolution of the distribution
of adhesion strength («adhesion mapping»),
another critical aspect of the «failure history».

Pic. 1 An illustration of the multi-approach research collaboration between Empa, University of Bern and Elotex with the goal to investigate the long-term mechanisms of shrinkage and adhesion of outdoor applied fully-vitrified ceramic tiles

Three young people, one PhD student and two
master students will perform their theses within this
project.

This paper describes results gained from test series
of single layed large (30×30 cm) porcelain tiles and
first results from the outdoor field study where the
same tiles are applied in areas of several square meters.

3. Failure studies on plain glass tiles

When tile adhesive mortars are tested according
to the industrial standard EN 12004, then, given by
the small tile size of 5×5 cm, the mortar looses most
of its unbound water by drying within 48 hours. Due
to this fast drying rate cement hydration stops after
two days and the resulting degree of hydration
is very low in the order of 30% and less (Jenni et al.,
2005). But usually, compared to real situations on the
construction site, tiles are much larger (e.g., 30×30
cm), the resulting drying rates are much slower and
the cement has weeks to hydrate and to reach a significantly
higher degree of hydration. However, the
standard test using 5×5 cm small tiles mimics the situation
at the edge of a large tile. There, in the outermost
2 cm from the grout the mortar dries relatively
early and the resulting degree of hydration is quite
low. On the other side, early drying accelerates film
formation of the polymers which can partly compensate
for the local lack of hydration. Even though
these peripheral parts remain critical because of
drying shrinkage and thermal stresses which are
highest in these zones.

The reason for localised shrinkage stresses is basically
given by the different material properties of the
porcelain tile compared to the substrate materials, including
the tile adhesive mortar. The rigidity (relatively
high E modulus) of the porcelain tile material, its lack
of capillarity, which makes it resistant against drying
shrinkage and swelling under wet conditions and its
significantly different thermal expansion coefficients,
causes strong differential movements at the tile mortar
interface. If adhesion resists these shear movements,
then shear stresses are generated at the tile-mortar interface.
Felixberger (2003) calculated these shear forces
and could demonstrate that they are zero in the centre
of the tile and increase constantly as the increments
(small parts with a given differential stress value) sum
up along the distance from centre to the grout to reach
maximum values at the edges of the tile.

Table 1 Non-accelerated simple mortar formulations used in this study

Pic. 2 Glass tile in outdoor field study. Application was done (against the recommendation!) by floating technique. The result is an incomplete wetting of the glass tile by formation of hollows in between the partly pressed mortar ripples. It could be observed that rain water intruded the system by the grout and caused a wetting of the outermost 4 cm. This caused the mortar to swell near the grout. As a result, the glass tile detaches from the mortar and micro-cracks are formed. The crack width can be estimated from interference colours to be in the range of one micron (within the zone of second order red).

In fact, the formation of shrinkage or expansion
cracks at the edges can be observed in a simple
see-through experiment with transparent glass tiles,
made from ordinary window glass (float glass; Pic. 2).

The formation of micro-cracks can be macroscopically
observed by an optical effect called interference.
When the crack is dry (filled with air)
and has a width in the order of 0,5-3 microns, white
sun light (composed of rainbow colours) is reflected
at the upper and lower side of the crack. The light
from the lower side travels a longer path through
the material and is swinging out of phase with the
light reflected at the upper side of the crack. If the travel distance is 0,5, 1,5, 2,5, 3,5…of red, then this
particular colour is extinct and we see the complementary
colour of red. Extinction of red occurs
where ever the travel distance is an exact multitude
of the half wavelength of red. But if the crack is
more than 3 microns wide, the interference colours
become more pastel-like until the effect is not seen
anymore.

For our study we used ordinary float glass for
such experiments. But it must be stated at this point
that the adhesion to float glass is much more difficult
for the mortar as in case for porcelain. However, the
interference patterns observed through glass tiles
tell us something about where the system undergoes
highest stresses and where failure occurs first. Furthermore,
the findings of such glass tile experiments
confirm observations that failure generally first occurs
at the edges and propagates from there inwards.
Following figure shows a mapping of the failures with
time.

Pic. 3 Glass tile (30x30 cm) applied by flow-bed mortar on a concrete plate (40x40 cm). After 5 weeks of normal storage conditions (23°C/50% r. h.) first isolated detachments form at the edges. With time these zones of detachment coalesce. After 15 weeks most of the glass tile is detached, and only an inner area 8x9 cm holds the tile in place

The micro-cracks shown in figures 3 form under
normal climate conditions (23°C/50% r. h.) where
the mortar under a 30×30 cm large tile takes several
weeks to dry. Therefore, we interpret these detachments
to be related with drying shrinkage of the mortar
and concrete plate underneath. This is confirmed
by drying rate curves showing that drying takes at
least 10 weeks until a stable weight of the sample is
reached.

The fact that the outermost rim of the glass
tile remains adhered is interpreted to be an artefact
related to the dishing and buckling of the concrete
slab underneath. Shrinkage measurements showed
that the concrete slab is buckling as it is wetted by
the freshly applied mortar. The concrete slab remains
buckled until the mortar sets. In the following
weeks the concrete-mortar-tile system dries and the
concrete slab dishes back. But the rigid tile resists
compression. In this particular geometric setting the
back-dishing causes a compression perpendicular to
the mortar bed near the tile edges and an increasing
tensile force perpendicular to the mortar bed in
more central parts.

This buckling-dishing scenario is unique to the
set-up of single layed tiles. We therefore follow the
approach of the field study, where the same tiles are
applied in larger areas and the substrate is shrinking
and expanding in the direction parallel to the mortar
bed, which is closer to reality. However, test series of
single layed large tiles bear other advantages as they
can for example be handled for tests under different
defined climatic conditions.

4. Adhesion mapping

In both approaches, applied as single large tile on
a separate concrete slab or as part of a larger area in
the outdoor field test, we test adhesion strength as a
function of position and time. This allows the reconstruction of adhesion maps and the evolution of characteristic
adhesion patterns.

Pic. 4 Mosaic pattern for adhesion mapping (a) as a function of position and time (b). Sample was stored at 23°C/50% r. h.. From Wetzel et al. 2008a and 2008b

Pic. 4 presents a typical adhesion strength
distribution with three characteristic zones, each
of which has its own characteristic evolution (Pic.
4b). It must be stated that the heterogeneity in the
distribution of adhesion strength is quite large and
that one cannot predict where strength is highest in
a particular case. Even though, the picture can be
generalised as follows. There are peripheral parts
(hatched areas in Pic. 4) with lowest adhesion values
and central parts where adhesion is significantly
higher.

From gradient studies we know that degree
of hydration is lowest in these peripheral parts and
increases continuously towards inner parts, where
hydration of cement can last longer as free water is
available as such.

Interestingly strength decreases after 7 weeks of
dry storage probably due to increasing shrinkageriemes
and the formation of micro-cracks at the mortar-
tile interface, an information gained from seethrough
experiments with glass tiles.

5. Conclusions

A pull-out test is telling about the additional load
a system can carry before it fails. Thus, it is only the
apparent strength we measure, and we do not know
if the system is pre-stressed, e.g., due to shrinkage
stresses and can therefore only be loaded with additional,
e.g., 1,5 N/mm2 until failure, or if the system is
not pre-stressed and the measured load is therefore
describing the bulk strength of the system.

From damage cases we definitely know, that
pre-stressing can be in the order of the bulk adhesion
strength, and from micro-structural investigations,
we know that micro-cracking at the tile-mortar interface
is most critical for the system and can hardly be
avoided. The question is more, how a micro-cracked
system can be stabilised to avoid the final stage of a
through-going fracture. It seems that in some cases
local cracking causes a decrease in stresses and crackpropagation
is stopped. In other cases, the microcracks
propagate along the entire interface.

Modelling of stress distributions and how they
evolve in cases of micro-cracking should bring more
light in the understanding of the failure history of
these multiple layer systems.

Especially, the lowering of the E modulus by
redispersible polymer powders is an interesting approach
to avoid the build up of high differential
shrinkage forces at the tile-mortar interface.

Acknowledgements

Financial support by Swiss Commission for Technology
and Innovation is gratefully acknowledged
(CTI project № 8605.1 EPRP-IW). We would like to
thank Jurg Lang for technical advice.

Bibliography

1. Felixberger J. (2003): Kleine Teilchen mit grosser
Wirkung. Fliesen und Platten, 5, 48-51.

2. Jenni A., Holzer L., Zurbriggen R. and Herwegh
M. (2005): Influence of polymers on microstructure and
adhesive strength of cementitious tile adhesive mortars.
Cement and Concrete Research, 35, 35-50.
3. Oberste-Padtberg R. and Sieksmeier J. (2007):
Zugig einlegen. Fliesen und Platten, 4, 22-24.

4. Reinmann Ch. (2001): Vermeidung von Frostschuden.
Vuterchen Frost aussperren. Fliesen und
Platten, Vol. 10, 2001, 18-21.

5. Wetzel A., Zurbriggen R., Herwegh M. (2008a):
Setup for investigations of tile-mortar interface of
large-sized tiles. Proceedings of the GDCh Baufachtagung
in Koblenz, Germany, 22.-24.9.08.

6. Wetzel A., Zurbriggen R., Herwegh M. (2008b):
Spatial and temporal evolution of physico-chemical
properties of polymer-modified mortar at the interfaces
of large-sized tiles. Proceedings of the GDCh Baufachtagung
in Koblenz, Germany, 22.-24.9.08.


Media partners