Journal of Cultural Heritage 11 (2010) 10–18

Original article

Desalination of masonry structures: Fine tuning of pore size

distribution of poultices to substrate properties
Barbara Lubelli a,b,∗ , Rob P.J. van Hees a,b
a Faculty of Architecture, Delft University of Technology, Julianalaan, 134, 2628 BL, Delft, The Netherlands
b TNO Built Environment and Geosciences, van Mourik Broekmanweg, 6, 2628 XE, Delft, The Netherlands
Received 17 November 2008; accepted 5 March 2009
Available online 13 November 2009

Abstract
Desalination is a relatively new intervention in the field of conservation of architectural heritage. Especially the desalination of immovable
objects, such as masonry structures, is still a trial-error practice. In the field, different desalination materials and methods are used, sometimes
with unsatisfactory results. Better understanding of the desalination process is needed in order to support the conservator with clear guidelines for
choosing a suitable desalination material and method. The research presented in this paper constitutes the first step towards the development of a
modular system of poultices, which can be adapted, i.e. fine-tuned to different types of substrates. Starting point is to make use of advection, i.e. the
transport of salt ions with water flow. This transport mechanism is faster than diffusion and the application on immovable objects is relatively easy.
In order to optimize salt extraction, a poultice working by advection should have smaller pores than the substrate. Starting from this principle the
pore sizes of different desalination materials (sand, cellulose, kaolin and bentonite) mixed in different proportions, have been measured. Interesting
results were obtained, showing that the desalination materials commonly used in the field are often not the most suitable ones. On the basis of the
results, recipes for poultices, adapted to a specific substrate, can be formulated.
© 2009 Elsevier Masson SAS. All rights reserved.

Keywords: Salt damage; Desalination; Poultices; Pore size distribution; Advection

1. Introduction dated practice with generally satisfactory results, desalination

Salt crystallization is one of the most recurrent causes of dam-
age in ancient buildings. Historic monuments are, up to a higher
extent than new constructions, threatened by salt decay because
of the large amount of salts accumulated in the course of time,
the frequent lack of maintenance, as well as the occurrence of
calamities in the past. Prevention of salt decay is therefore an
important issue in the preservation of architectural heritage. For
salt damage to occur, both salt and moisture need to be present.
This implies that removing one of these two factors will be
enough to stop the development of salt decay. In some cases,
i.e. in situations where no salt supply is present anymore and
the moisture source can not be easily removed, (partial) desali-
nation of the masonry can slow down the development of salt
damage and preserve the original materials. While desalination
of movable objects following the bath method is a consoli-
∗ Corresponding author.
E-mail address: barbara.lubelli@tno.nl (B. Lubelli).
of masonry structures is still a trial-error practice. Different
desalination materials and methods are used, sometimes with
questionable results, as shown by recent reviews of this subject
[1,2]. Better understanding of the desalination process is there-
fore needed in order to give a reliable, scientific basis for the
choice of the most suitable desalination material and method.
Desalination, i.e. the process of removing salt from a material,
can occur by two transport mechanisms: diffusion or advection.
Diffusion consists in the transport of a substance (salt ions, in
this case) in a liquid from high to low concentration; advection
consists in the transport of a substance with a moving liquid
(water, in this case). Diffusion is the main mechanism working
in desalination by the bath method or by wet poultices; advec-
tion is the main mechanism working in desalination by drying
poultices. Advection is a much faster mechanism than diffusion
and its application on immovable objects is relatively easy. Nev-
ertheless, desalination of porous material by advection needs to
be optimized in order to achieve good results in practice.
It is known that during drying of a material, coarse pores
will be empty first; small pores will dry out later. In lay-

1296-2074/$ – see front matter © 2009 Elsevier Masson SAS. All rights reserved.
doi:10.1016/j.culher.2009.03.005
 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18 11

ered systems, as for example the substrate/desalination poultice
combination, water will move from the coarse to the small
pores and salt will accumulate in the fine porous material
[3,4]. Therefore, in order to improve salt extraction, a poul-
tice working by advection should have smaller pores than the
substrate. Nevertheless, it is convenient that the pores of the
poultice are not too small, to avoid slowing down of the process
[5,6].
From these principles the idea of the development of a
modular system of poultices came forward. The foreseen
modular desalination system implies the use of poultices
whose pore system is adapted to the pore system of the
salt laden substrate, in order to optimize the desalination
efficiency.
This research, developed in the framework of the EU project
Desalination (EC contract no. 022714), constitutes a first step
towards the development of such a modular system. This scope
has been pursued step-by-step, starting from the study of simple
(single component) poultices and moving then to mixes of two
and three components. Poultices with different pore sizes have
been obtained by varying the components (sand, cellulose,
kaolin and bentonite), the ratio between them and the grain
size distribution of the sand (sections 2 and 3). The obtained
results are discussed with respect to the practice of desalination
(section 4).
2. Experimental section
2.1. Selection of the components of the poultices
The research started with the selection of the base materials
to be used for the desalination poultices. The following materials
were used:
• siliceous sand: Standard sand CEN-NORMSAND DIN-EN
196-1 which has been sieved to obtain;
• fine sand (0.08–0.5 mm);
• coarse sand (0.5–1 mm);
• kaolin: Polywhite (by Imerys);
• bentonite: Ceratosil® R (by Sud Chemie);
• cellulose: Arbocell BW 40 (fibres length 200 ␮m) and BC
1000 (fibres length 700 ␮m).
Most of the selected materials (sand, cellulose and bentonite)
have been chosen because they are often used in the practice of
conservation. Kaolin is not generally used in the field, since the
white deposit it leaves after removal is usually considered not
acceptable. Despite its limited use in practice, kaolin has been
selected for this research because of its limited shrinkage and
good desalination efficiency mentioned in literature [7].
2.2. Design of desalination poultices
Using the components mentioned in 2.1, different poultices
were prepared. At first cellulose-, kaolin- and bentonite-only
poultices were prepared and their pore size distribution was mea-
sured. Then kaolin was mixed in different proportions with sand
of different grain size. The same was on a more limited selec-
tion of bentonite/sand mixtures. At this stage, three components
poultices were prepared, including cellulose/kaolin/sand and
cellulose/bentonite/sand. The complete set of poultices studied
in this research is reported in Table 1.
2.3. Specific weight
The specific weight of the single materials used for making
the poultices was measured (Table 2). A container of a known

Table 1
Composition and water content of the selected poultices.
Acronym Composition Ratio (by weight) Sand grain size (mm) Water content (weight water/weight dry poultice)

SS Sand 0.08–0.5 –
SL Sand 0.5–1 –
K Kaolin 0.5
B Bentonite 0.6
Ca Cellulose BW40 5
Cb Cellulose BC1000 5
KS13 S Kaolin/sand 1:3 0.08–0.5 0.22
KS13 L Kaolin/sand 1:3 0.5–1 0.22
KS15 S Kaolin/sand 1:5 0.08–0.5 0.2
KS15 L Kaolin/sand 1:5 0.5–1 0.2
KS17 S Kaolin/sand 1:7 0.08–0.5 0.21
KS17 L Kaolin/sand 1:7 0.5–1 0.16
BS13 S Bentonite/sand 1:3 0.08–0.5 0.3
BS15 S Bentonite/sand 1:5 0.08–0.5 0.26
BS17 S Bentonite/sand 1:7 0.08–0.5 0.23
CKS211 L CellBW40/kaolin/sand 2:1:1 0.5–1 1.6
CKS219 L CellBW40/kaolin/sand 2:1:9 0.5–1 0.7
CKS121 L CellBW40/kaolin/sand 1:2:1 0.5–1 0.8
CKS129 L CellBW40/kaolin/sand 1:2:9 0.5–1 0.4
CKS129 S CellBW40/kaolin/sand 1:2:9 0.08–0.5 0.45
CBS121 L CellBW40/bentonite/sand 1:2:1 0.5–1 0.96
CBS129 L CellBW40/bentonite/sand 1:2:9 0.5–1 0.4
12 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18

Table 2 were applied on a fired-clay brick substrate instead of casted in

Specific weight of the single materials. a mould. The thickness of the poultices was about 2 cm. The
Material Volume [dm3 ] Weight [kg] Density [kg/m3 ] brick/poultice combinations were dried at 20 ◦ C/50% RH. After
Sand 0–0.5mm 0.0005 0.8113 1622.6
2 weeks, the poultice slabs were detached from the brick and
Sand 0.5–1mm 0.0005 0.8019 1603.8 dried in an oven at 70 ◦ C until they reached a constant weight.
Kaolin 0.0005 0.2795 559 At this point, samples for MIP were collected from the part of
Bentonite 0.0005 0.5133 1026.6 the poultice slab near the brick/poultices interface, in order to
Cellulose BC1000 0.0005 0.471 94.2 minimize differences and to obtain reproducible results. In fact,
Cellulose BW40 0.0005 0.1013 202.6
the pore size distribution of the poultice at the brick interface
may differ from the pore size of the poultice at its evaporation
surface, due to the suction exerted by the brick immediately after
size was filled up with the material and weighed: no pressure was applying of the poultice [9].
applied but the material was compacted by the use of a shaking
table.
2.5. Mercury intrusion porosimetry
Then, the volume of each component as % of the total volume
was calculated for each of the studied poultices (Table 3). Since MIP measurements were performed on the single compo-
the components have a very different specific weight, know- nents (sand, kaolin, bentonite, cellulose BW40 and BC 1000) as
ing the volume ratios helps interpreting the mercury intrusion well as on all the poultices reported in Table 1.
porosimetry (MIP) results. A mercury intrusion porosimeter, Autopore IV9500 from
the Micromeritics, has been used. All measurements were per-
2.4. Preparation of the poultices formed in twofold. Measurements of pore size distribution of
sands have been performed by a special penetrometer for pow-
The water content to be used to provide the poultice with der samples; all the other measurements were performed with a
a good workability depends on the type and particle size of penetrometer for solid samples.
the components and the ratio between them. The water content MIP showed to be a possible method for measuring the pore
necessary for each poultice was determined by measurements size distribution of poultices: in spite of the weakness of some
performed by the “Laboratoire de Recherche des Monuments of the poultices, no abrupt change in the incremental volume
Historiques” (France), in the framework of the Desalination curve was measured, indicating that cracking of the poultices
project [8]. due to the pressure has never occurred. A further proof is the
All the poultices were prepared according to a standard pro- very good reproducibility of the results observed in all measure-
cedure which was set up by the “Laboratoire de Recherche ments performed in twofold. It might be questioned whether the
des Monuments Historiques” (France) in the framework of the pore size distribution of dry poultices is relevant and how much
Desalination project [8] and consisted of: mixing of the dry this differs from the pore size of the poultice at the wet state.
components for 30 seconds for each added component, addi- This seems to depend on the type of components. In most of the
tion of water at ambient temperature, mixing of the poultices cases (cellulose, kaolin and sand based poultices) no significant
for 1 minute, storing of the poultices in a closed container for differences are expected between the pore size of the dry poul-
24 hour, mixing for 1 minute before application. tice and the pore size of the poultice when it starts working by
The pore size of a fresh mortar/poultice being affected by the advection. In fact, the shrinkage is limited and happens in the
suction of the material on which it is applied [9], all poultices first period following the application, when the poultice is still
Table 3
Composition of the poultices expressed as volume %.
Poultice Mass ratio Vol % cell Vol % kaolin Vol % bentonite Vol % sand

KS13 S Kaolin/sand 1:3 49 51

KS13 L Kaolin/sand 1:3 49 51
KS15 S Kaolin/sand 1:5 37 63
KS15 L Kaolin/sand 1:5 36 64
KS17 S Kaolin/sand 1:7 29 71
KS17 L Kaolin/sand 1:7 29 71
BS13 S Bentonite/sand 1:3 35 65
BS15 S Bentonite/sand 1:5 24 76
BS17 S Bentonite/sand 1:7 18 82
CKS211 L Cellulose/kaolin/sand 2:1:1 80 15 5
CKS219 L Cellulose/kaolin/sand 2:1:9 57 10 33
CKS121 L Cellulose/kaolin/sand 1:2:1 54 39 7
CKS129 L Cellulose/kaolin/sand 1:2:9 35 25 40
CKS129 S Cellulose/kaolin/sand 1:2:9 35 25 40
CBS121 L Cellulose/bentonite/sand 1:2:1 66 26 8
CBS129 L Cellulose/bentonite/sand 1:2:9 40 16 44
 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18
delivering water to the substrate. At the moment advection from
the substrate to the poultice starts, no more significant shrinkage
occurs and the pore size of the poultice does not significantly
change. Measurements reported in literature [10] of pore size
distribution of kaolin-based poultices at different stages of dry-
ing confirm the limited change of pore size between wet and dry
stage. In the case of bentonite-based poultices, due to the high
shrinkage, it cannot be excluded that the pore size distribution
measured on the dry poultices may differ from that at the initial
wet state.
3. Results of the mercury intrusion porosimetry
measurements
The results of the MIP measurements are presented in this
section. At first, results obtained on single components are
reported (section 3.1), then the measurements performed on two
and three components poultices (sections 3.2 and 3.3).
3.1. Single components
3.1.1. Sand
The incremental intrusion curve (Fig. 1) of the sand fraction
0.08–0.5 mm shows a clear peak in the pore size distribution at
80 ␮m, with most of the pores between 30 and 180 ␮m; the sand
fraction between 0.5 and 1 mm shows a peak at 180 ␮m with
most of the pores between 100 and 400 ␮m. Despite the similar
density of the two sand fractions (Table 2) a much lower total
porosity is measured for the coarse sand: this suggests that part
of the porosity in the coarse sand is larger than 400 ␮m and is
not measured by MIP.
It can be concluded that the range of pore sizes which can
be obtained using these two different classes of sand is always
larger than 30 ␮m. Considering that many types of porous build-
ing materials have a pore size smaller than 30 ␮m, the pore size
given by the sand only will not be effective for desalination by
advection. In order to get a range of pores effective for desalina-
tion by advection, the porosity between the sand grains should
be (partially) filled in by other components (e.g. clay).
Fig. 1. Total porosity and pore size distribution of the sand fractions
0.08–0.5 mm and 0.5–1 mm.
13
Fig. 2. Total porosity and pore size distribution of the clay poultices.
3.1.2. Clay
Kaolin- and bentonite-only poultices have been mea-
sured. The pore size distributions of the kaolin-only and
bentonite-only poultice are reported in Fig. 2. The pore size
distribution of the kaolin poultice shows a peak at 0.25 ␮m,
with pores ranging mainly between 0.05 and 0.3 ␮m. The
interstitial porosity in kaolin particles will be thus effec-
tive on advection (for materials with pores larger than about
0.3 ␮m). The bentonite poultice shows mainly pores smaller
than 0.1 ␮m. The interstitial porosity in bentonite particles
will be thus not really effective for advection, as this pore
class only plays a minor role in capillary transport. Bentonite
will have mainly the role of increasing the water retention
of the poultice (useful for desalination by diffusion) and to
bind ions.
3.1.3. Cellulose
The pore size distribution of two types of cellulose fibres
has been studied: ARBOCELL BW40 (fibre length 200 ␮m)
and ARBOCELL BC1000 (fibre length 700 ␮m). Both types
are very often used in the field as desalination materials. The
short fibre type helps to shape the poultice in the irregularities
of the substrate, the long fibre type to give cohesion to the mix.
The two types of cellulose are often mixed together in a 1:1
proportion.
The pore size distribution of both cellulose poultices is uni-
modal; the BW40 cellulose has a sharp peak at 15–20 ␮m; the
BC1000 cellulose shows pores distributed in a slightly wider
range, between 10 and 30 ␮m (Fig. 3).
3.2. Clay/sand mixtures
Kaolin and bentonite based poultices have been studied
(Table 1).
3.2.1. Kaolin/sand poultices
The pore size distribution of all kaolin/sand poultices is
reported in Fig. 4.
14 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18
Fig. 3. Total porosity and pore size distribution of the cellulose poultices.
3.2.1.1. KS 15 L. The KS 15L poultice has a bimodal pore size
distribution, with two peaks at 0.35 ␮m and at 180 ␮m. These
two peaks are determined by the interstitial porosity in the kaolin
and the sand respectively, as shown by the fact that the two
peaks are exactly the same as those measured for the single
components (Figs. 1 and 2). The pores between 1 and 100 ␮m
are due to partial filling by the kaolin of the interstitial porosity
in the sand. The volume fraction of these pores is small, due
to the little amount of kaolin present in this poultice (36% by
volume).
3.2.1.2. KS 15S. The use of 0.08–0.5 mm sand shifts the pores
partially filled with kaolin to smaller sizes (2–4 ␮m). The peak
due to kaolin is still present (0.35 ␮m), as well as some pores
due to interstitial porosity of the sand (> 30 ␮m).
3.2.1.3. KS 13L. An increase of the kaolin content leads to
a larger volume of pores in the 0.2–0.3 ␮m range (pores due
to interstitial porosity in the kaolin). The amount of pores in
Fig. 4. Pore size distribution of kaolin/sand poultices.
the sand which are partially filled with kaolin (10–100 ␮m)
increases, while the amount of interstitial pores which remains
empty (> 100 ␮m) decreases (compare with KS 15L).
3.2.1.4. KS 13S. The use of 0.08–05 mm sand shifts the peak
of the pores partially filled with kaolin to 3–5 ␮m, while the
interstitial pores in the sand disappear. This means that, thanks
to the large amount of kaolin used (40% by volume) and to the
fine sand, there are no pores between the grains which are not
(or at least partially) filled with kaolin.
3.2.1.5. KS 17L. The pore size distribution of the KS 17L
poultice is very similar to the one of the KS 15L poultice.
There are some interstitial pores in the sand which are par-
tially filled by the kaolin (2–50 ␮m range), but most of them are
empty (> 50 ␮m). The volume of interstitial pores in the kaolin
decreases as consequence of the decreased kaolin content.
3.2.1.6. KS 17S. The pore size distribution of the KS 17S poul-
tice shows three peaks. The peak at 0.3 ␮m is due to interstitial
porosity in the kaolin; the other two are due to partial filling of
the interstitial porosity in the sand with kaolin. The reason for
these two peaks is not clear. A hypothesis is that the larger pores
(30 ␮m) are the result of interstitial pores which are just coated
at the surface with kaolin. The pores in the range 2–3 ␮m might
be due to actual partial filling.
On the basis of the obtained results it can be concluded
that the total porosity does not vary much between the poul-
tices. Values between 27 and 33% by volume are generally
obtained. As expected, a higher initial water content in the
poultice corresponds to a higher total porosity in the dried
out material. Regarding the pore size distribution it can be
concluded that all poultices show a bimodal pore size distri-
bution, except for KS 17S having three peaks. The smaller
pores always have a size of about 0.3 ␮m and are due to the
 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18 15

Fig. 5. Total porosity and pore size distribution of bentonite/sand poultices.

interstitial porosity between the kaolin particles. This implies
that the amount of these fine pores increases with the clay
amount. The second peak in the pore size distribution is deter-
mined by the grain size of the sand and the kaolin/sand ratio.
This peak varies between 2 and 4 ␮m in the case 0.08–0.5 mm
sand is used and between 20 and 250 ␮m when 0.5–1 mm sand
is used. The use of finer sand, keeping the clay/sand ratio
constant, moves the peak in the pore size distribution towards
smaller pores. We can conclude that, if 0.08–0.5 mm sand is
used, a kaolin/sand ratio of 1:3 by weight (= 1:1 by volume)
is enough to fill, at least partially, all the interstitial pores
in the sand, so that no pores coarser than 10 ␮m are present
anymore; if the sand is coarser, larger amounts of kaolin are
needed.
3.2.2. Bentonite based poultices
Three bentonite/sand poultices have been studied (Table 1)
(Fig. 5).
3.2.2.1. BS 13L. The pore size distribution of the BS 13L poul-
tice shows a peak at about 30–40 ␮m due to partial filling of
the interstitial porosity between the sand grains (the sand only
has a peak at 60–70 ␮m). The interstitial pores in the bentonite
(< 0.1 ␮m) are not found anymore.
3.2.2.2. BS 15S. The pore size distribution of the BS 15S poul-
tice is similar to the one of BS 13S. The total porosity is higher
than for BS 13S poultice, most probably because of the higher
water content used.

Fig. 6. Porosity and pore size distribution of cellulose BW40/kaolin/sand 0.5–1 mm poultices: CKS 211 L (a), CKS 219 L (b), CKS 121 L (c), CKS 129 L (d).
16 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18

3.2.2.3. BS 17S. The pore size distribution of the BS 17S poul-

tice is similar to the one of BS 13S and BS 15S. The total
porosity is proportional to the water content: higher than in
BS 13S and lower than in BS 15S.
On the basis of these results, it can be concluded that the
total porosity increases with the water content. The pore size
distribution does not significantly change when varying the ben-
tonite/sand ratio.

3.3. Cellulose/clay sand poultices

As the last step of this research, three-components poul-

tices, based on cellulose, clay (kaolin or bentonite) and sand,
have been studied. For these experiments, only short fibre cel-
lulose (BW40) has been used. Because of the very similar pore
size distribution of these two cellulose types (Fig. 3), no large
differences were expected when replacing BW40 by BC1000.
3.3.1. Cellulose/kaolin/sand poultices
Different cellulose/kaolin/sand ratios have been used (Fig. 6).
The effect of the grain size distribution of the sand on the pore
size distribution has been tested only for the mix having the high-
est sand amount: cellulose BW40/kaolin/sand with ratio 1:2:9.
3.3.1.1. CKS 211 L. The pore size distribution of the
CKS 211 L poultice shows a clear peak at 15 ␮m, due to the
cellulose (Fig. 3). Only a very limited amount of porosity is
due to interstitial porosity in the kaolin. It can be supposed that
the small amount of kaolin present does not agglomerate but
tends to cover the cellulose fibres.
3.3.1.2. CKS 219 L. The CKS 219 L poultice shows a peak at
15 ␮m due to the cellulose. The interstitial pores of the kaolin
are not present, whereas only few empty interstitial pores in the
sand are measured. These constitute a very small part of the total
porosity.
3.3.1.3. CKS 121 L. In the CKS 121 L poultice, in which the
kaolin content is high (39% by volume), the interstitial porosity
of the kaolin appears. The peak of the cellulose is not present
anymore: we can suppose that the kaolin fills (almost) com-
pletely the spaces between the cellulose fibers, originating the
porosity range between 0.3 and 2 ␮m.
Fig. 7. Porosity and pore size distribution of KS13 L, CKS 129 L, CKS 219 L
poultices.
poultices can be further explained when comparing CKS129 L,
CKS219 L and KS13 L, all having a CK:S or K:S ratio equal
to 1:3 (Fig. 7). The addition of cellulose leads to an increase in
the total porosity, due to the higher water content necessary for
obtaining a good workability.
The effect of kaolin depends on its amount. Kaolin will pref-
erentially arrange to coat the sand grains and the cellulose fibres.
Whenever sufficient kaolin is present, part of it will agglomerate
and the interstitial porosity between kaolin particles will appear,
as clearly visible in KS13 L. By the addition of cellulose to the
poultice, the surface area will drastically increase and the pos-
sibility for kaolin to agglomerate will be limited. This is clear
when comparing CKS219 L with CKS129 L: in the first case
the amount of kaolin is only 10% (volume kaolin/total volume
poultice) and no kaolin will be left to agglomerate, after coating
of cellulose and sand; in the CKS129 L the amount of kaolin
is 25% and there will be kaolin left to agglomerate, as shown
by the presence of interstitial porosity between kaolin particles
(0.3 ␮m).
The effect of the grain size of the sand can be evaluated by
comparing CKS 129 L and CKS 129 S (Fig. 8). The use of
finer sand moves the pores towards smaller size, while the total
porosity remains about the same. This is in agreement with what
observed for kaolin/sand poultices.

3.3.1.4. CKS 129 L. In the pore size distribution of the

CKS 129 L poultice different pore classes can be recognized:

• pores between 0.1 and 0.3 ␮m, due to interstitial porosity in

the kaolin;
• pores between 0.3 and 10 ␮m, due to interstitial porosity
between kaolin, cellulose and sand (partially filled pores);
• pores between 80 and 400 ␮m, due to the interstitial porosity
in the sand. These last, however, constitute only a very small
part of the total porosity.

The role of the cellulose and the kaolin with respect to Fig. 8. Total porosity and pore size distribution of the CKS 129S and CKS 129L
the pore size distribution and porosity of cellulose/kaolin/sand poultices.
 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18
Fig. 9. Porosity and pore size distribution of cellulose BW40/bentonite/sand 0.5–1 mm poultices: CBS 121 L (left), CBS 129 L (right).
3.3.2. Cellulose/bentonite/sand
For some of the ratios already studied for cellulose/kaolin/
sand, cellulose/bentonite/sand poultices have been prepared
and their pore size distribution measured (Fig. 9). Only coarse
sand (0.5–1 mm) has been used.
3.3.2.1. CBS 121L. The pore size distribution of the CBS 121L
poultice shows a large amount of pores in the range 0.5–5 ␮m.
The interstitial porosity between the cellulose is partially filled
up by the bentonite, so that the peak due to cellulose (15 ␮m) is
shifted to smaller pores. The interstitial porosity in the bentonite
is still present (see the increase in total porosity for pores smaller
than 0.005 ␮m).
3.3.2.2. CBS 129L. The pore size distribution of the CBS 129L
poultice shows a large amount of coarse pores (5–100 ␮m). The
interstitial pores in the bentonite are not found anymore and the
partial filling of the interstitial porosity in the cellulose is much
less than in the CBS 121L poultice. It can be supposed that the
low amount of bentonite present (16 vol %) tends to create a
layer on cellulose and sand.
When compared to CKS poultices, CBS poultices have a
lower total porosity. The interstitial porosity in bentonite is not
found in CBS poultices (similarly to what was observed in BS
poultices), while the interstitial porosity of kaolin in CKS poul-
tices is clearly visible. Both these findings may be explained by
the higher shrink/swell capacity of bentonite in comparison with
kaolin.
4. Discussion and conclusions
The research reported in this paper constitutes an important
step forward in the development of a modular system of poultices
which can be adapted, i.e. fine-tuned to different types of sub-
strates. This research has shown which components can be used
and in which proportion in order to obtain a certain pore size of
the poultice. It has become clear that just varying the grain size of
the sand (0.08–1 mm) is not sufficient to obtain a pore size effec-
tive for desalination by advection on most of the porous building
materials. A filler, as for example clay, is needed to shift the pore
size of the poultice towards smaller pores. Also, the porosity
17
present in cellulose poultices is not small enough to guarantee
capillary suction from substrates with pores smaller than 15 ␮m,
i.e. for example no advection will occur when applying a cel-
lulose poultice on a fired-clay brick substrate. This points out
how the practice of desalination, mainly relaying on cellulose
poultices, is far from being optimized.
Bentonite/sand poultices showed no significant differences
in pore size when changing the ratio between components: the
pores in the dry poultices were always too large to be effective
for desalination by advection. In the case cellulose is added to
bentonite and sand, pore size ranges are obtained which may
be suitable for desalination of substrates with pores larger than
1–2 ␮m.
Kaolin is a good candidate for the design of desalination
poultices having different pore sizes. First of all the interstitial
porosity in kaolin (0.3 ␮m) has the right size to be effective in
capillary transport and thus advection [11]. Besides, by vary-
ing the ratio between kaolin/sand and cellulose/kaolin/sand a
large range of pore sizes can be obtained, among which it is
possible to select the most suitable one for each substrate. The
problem of the whitish deposit left by kaolin on the surface of
the desalinated material is known and needs still to be solved.
In the coming period, different materials alternative to kaolin,
i.e. having similar grain size, will be tested. The possibility of
safely removing the deposit will be also investigated.
References
[1] V. Vergès-Belmin, H. Siedel, Poultice desalination of masonries and
monumental sculptures: a review, Restor. Build. Monum. 11 (6) (2004)
1–18.
[2] A. Heritage, A. Sawdy, F. Funke, V. Vergès-Belmin, A. Bourgès.
How do conservators tackle desalination? An international survey of
current poulticing methods, in Postprints from the 8th European Con-
ference on Research for Protection, Conservation and Enhancement of
Cultural Heritage, 10–12 November 2008 Ljubljana, Slovenia, (2008)
forthcoming.
[3] J. Petkovic. Moisture and ion transport in layered porous building materials:
a nuclear magnetic resonance study. PhD thesis, Eindhoven University of
Technology (2005).
[4] J. Petković, H.P. Huinink, L. Pel, K. Kopinga, R.P.J. Van Hees, Salt transport
in plaster/substrate layer, Mater Stru 40 (5) (2007) 475–490.
18 B. Lubelli, R.P.J. van Hees / Journal of Cultural Heritage 11 (2010) 10–18
[5] L. Pel, A. Sawdy, V. Voronina. Physical principles and efficiency of salt
extraction by poulticing, forthcoming J. Cultural Heritage.
[6] A. Sawdy, B. Lubelli, V. Voronina, F. Funke, L. Pel. Optimising the extrac-
tion of soluble salts from porous materials by poultices, forthcoming,
Studies in Conservation.
[7] W. Domaslowski, M.K. Lewandowska, J.W. Lukaszewicz, Badanie nad
Konserwacja murow ceglanych (Research on the conservation of brick
masonries), Torun University Publishers, 1998.
[8] A. Bourgès, V. Vergès-Belmin. Personal communication.
[9] M. de Rooij, C.W.P.J. Groot, A closer look at salt loaded microstructure,
Heron 51 (1) (2006) 49–62.
[10] M. Auras, Poultices and mortars for salt contaminated masonry and stone
objects, in: L.M. Ottosen, I. Rori-Dalgaard, P.K. Larsen, I. Brajer, P.
Bollingstoft, M. Marciniak, M. Svane (Eds.), Salt Weathering on build-
ings and stone sculptures, Proceedings of the Int. Conf, Copenhagen,
October 22–24, Technical University of Denmark, Copenhagen, 2008,
pp. 197–217.
[11] B. Meng, Moisture-transport-relevant characterization of pore structure,
in: J.D. Delgado Rodrigues, F.M.A. Henriques, F. Telmo Jeremias (Eds.),
Proceedings of the 7th International Congress on Deterioration and Conser-
vation of Stone, Lisbon, June 15–18, Laboratorio Nacional de Engenharia
Civil, Lisbon, 1992, pp. 387–396.