IV Transport Phenomena and Durability

Materials for Buildings and Structures. Edited by F. H. Wittmann

Copyright  2000 Wiley-VCH Verlag GmbH, Weinheim.
ISBN: 3-527-30125-9
The Ion Mobility of Deteriorating Salts in Masonry Materials of
Different Moisture Content

Anja Buchwald, Christian Kaps

Bauhaus-Universität Weimar, Professur Bauchemie

1 Introduction

The existence and movement of water and damaging salts are at the origin of numerous types
of decay observed in masonry. Salt migration can be subdivided at least into two processes
based totally on different mechanisms. Ions of dissolved salts can be transported with the mi-
grating water. The second transport mechanism is the ion diffusion driven by a concentration
gradient. In practice this mechanism comes into play to remove dissolved ions from salt con-
taining masonry by application of a compress. For the calculation of the duration of such a
desalination process the knowledge about salt diffusion coefficient is necessary. The diffusion
coefficient of the ion pair of Na+ and SO42-- in fully saturated brick material were recently
presented [1]. Nevertheless in reality the water saturation is much less than 100 %. Therefore
diffusion measurements were carried out to determine the influence of the degree of water
saturation on the observed diffusion coefficient of salt ions in brick and mortar material. Ad-
ditionally to this the impedance spectroscopy were used as an independent method to measure
the ion mobility and therefore the diffusion coefficient of these and other materials also.

2 Determined Investigations

2.1 Three-Layer Diffusion Experiment

In order to measure the diffusivity of salt ions at different moisture levels, sandwich speci-
mens have been prepared. These layered specimens consist of three plates, each of them of
about 1,7 cm (brick) or 3 cm (mortar) thickness. The central layer was impregnated by a salt
solution (Na2SO4). All three plates are equilibrated with the same moisture content by micro-
wave impulse drying. Paper filters have been inserted between two adjacent plates in order to
ensure a better contact at the interfacial zones. This sandwich construction is mechanically
compressed in order to enhance the contact between the three layers. Immediately all the faces
of the sandwich specimens are sealed with wax in order to prevent any moisture exchange
between specimens and the surrounding atmosphere. At given duration’s of diffusion, the
three plates were disconnected and drilled down step by step up to defined depths in order to
used the brick powder to analyze the water-soluble ion content of Na+ and SO42--. Assuming
that no advection phenomenon and no chemical reaction can take place during the run, the
transient transfer of the ions held in porewater of the porous system can be described by a
diffusion equation given by the second Fick’s law. The solution of this equation is done nu-
merically and already proposed in [2]. To find the best ion diffusion coefficient, the computer

Materials for Buildings and Structures. Edited by F. H. Wittmann

Copyright  2000 Wiley-VCH Verlag GmbH, Weinheim.
ISBN: 3-527-30125-9
158

program [3] fits the calculated concentration profile to the experimental profile by minimizing
the sum of squares.

2.2 Impedance Spectroscopy (IS)

Analogous prepared cylinder shaped samples with 18 mm in diameter and 5 to 35 mm thick-
ness were used for impedance spectroscopy in a SI 1260 impedance analyzer. For best electri-
cal contact between samples and electrode silver coated gum were used to avoid artifacts due
to surface resistance [4]. If plotted in the complex impedance plane (Nyquist-plot), the IS-data
form a depressed arc in the high-frequency zone which goes over in a straight line at lower
frequencies. A typical curve of a salt and moisture penetrated sample is shown in Figure 1.
The bulk resistance is then found as the interception of the high frequency-arc and the real
axis using the indicated equivalent circuit for fitting the experimental data. The arc-depressing
factor n of the CPE which symbolizes the distribution of relaxation time is related to pore size
distribution [5,6]
10

R
Impedance Zs’’ [10 Ω]

+ experimental data CPE

8
3

curve fitted using CPE

CNLS-method
6

material electrode

15 MHz
fit parameter for material:
f R= 8,7 kΩ
2

CPE: C= 0,1 pF; n = 0,74

0

0 2 4 6 8 10 12 14
Impedance Zs’ [103 Ω]
Figure 1. Nyquist plot and equivalent circuit of a typical sample (f-frequency; Circuit elements: R-resistance,
CPE-constant phase element with C-capacity and n- arc-depressing factor).

The conductivity σeff of the material can then be determined using R and the geometry
h (thickness of cylinder) and A (area of cylinder):
1 h
σ eff = (1)
RA
Using Nernst-Einstein-relation, the diffusion coefficient D can be calculated from σeff and
the salt concentration in the sample csalt [7]:
k ⋅T
D = σ eff 2
(2)
z ⋅ F ⋅ e ⋅ c salt

3 Materials

The investigated building materials include brick, mortar and sandstone. Beside these materi-
als porous glass as a model material were include, which has a similar pore structure at differ-
ent pore diameters. An overview about the porosity characteristics are given in Table 1. It can
easily be seen, that the porous glass material represent a broad spectrum of porosity properties
 159

where the building materials can be inserted. The pore structure of the sandstone and the po-
rous glass materials are similar. The porosity of the brick material is formed by small gaps
mainly orientated in extrusion direction. The mortar contains two different kind of porosity,
ones the capillaries and as second the big air holes (compare the medium pore diameter meas-
ured by MIP).

Table 1. Porosity characteristics of the used materials

bulk density true apparent coefficient of permeabil- medium
density [g/cm³] porosity porosity water adsorp- ity to gases pore
[g/cm³] [vol.- [vol.-%] tion [kg/m²√h] [10-14 m²] diameter
%] [µ m]

brick Z1 1.81 2.81 36 34 19 0.45 2

lime mortar M 1.80 2.65 32 31 6 6.8 2 / 200
sandstone 1 S1 2.15 2.63 18 17 1...4 1...3 0.5
sandstone 2 S2 2.07 2.69 23 22 9...20 8...14 50
porous glass 1 P1 1.35 2.39 43 39 - 11.6 100
porous glass 2 P2 1.27 2.33 45 43 - 11.2 80
porous glass 3 P3 1.70 2.33 27 23 - 5...8 22
porous glass 4 P4 1.69 2.31 27 24 - 6.5 10
porous glass 5 P5 1.68 2.28 26 24 - 0.25 0.9

4 Results

The absolute values of D for all materials at 100% water saturation measured by impedance
spectroscopy are given in
Table 2. The influence of pore morphology on the diffusion coefficient has been already dis-
cussed in [8].

Table 2. Diffusion coefficient of Na2SO4 in fully saturated material by impedance spectros-

copy
Z1 M S1 S2 P1 P2 P3 P4 P5
D [10-6 cm²/s] 2.6 4.2 3.9 2.9 3.1 3.3 3.2 3.4 3.4

Figure 2 shows the diffusion coefficient of Na2SO4 in brick material (left side) and in the
lime mortar (right side). The diffusion coefficient of the salt ions in the fully saturated materi-
als are between 2×10-6 and about 4.5 ×10-6 cm²/s. The diffusion coefficient in the brick mate-
rial measured by impedance spectroscopy seems to be higher than by diffusion experiment.
There might be additional ions solvated by pore solution out of bulk brick material increase
the measured conductivity and therefor affect the diffusion coefficient. The influence of the
water saturation on the diffusion coefficient seems to be almost linear down to about 30 %.
Afterwards the diffusion coefficient decreases less rapidly and comes to almost zero. This
behavior is similar to already described potential relation between R and SW in sandstone [9]
Other authors found exponential or linear relation between D and SW in soils [10,11]
160

5 5
Brick Lime Mortar
3-layer diffusion experiment
4 4
Impedance spectroscopy
D [10-6 cm²/s]

D [10-6 cm²/s]
3 3

2 2

1 1

0 0
100% 80% 60% 40% 20% 0% 100% 80% 60% 40% 20% 0%
water saturation SW water saturation SW

Figure 2. Comparison of diffusion coefficients of Na2SO4 determined in diffusion experiment as well as by

impedance spectroscopy in relation to water saturation

The values differ from each other significant especially in the diffusion experiment. That
might first be caused by taking always new samples for each experiment and each moisture
content and second by difficulties in contacting the three plates as well as unpleasant advec-
tive and convective water movement.
It can be said that the usage of both diffusion experiment and impedance spectroscopy re-
sult in the same diffusion coefficients in the relation to moisture content. The diffusion coeffi-
cients of the two involved methods are consistent. Each method contains advantages and
disadvantages (Table 3). A combination of both increases not only the numbers of measure-
ments and therefor the consistency of the assumption. Both methods complete each other in
the light of minimizing the maximal error. For instance the Three-Layer Diffusion Experi-
ment is difficult to manage at moisture saturation’s smaller then 30 %. The Impedance spec-
troscopy consist of errors at high saturation’s because of stray and cable inductance’s.
If comparing the diffusion coefficient of all materials in relation to the moisture content in
Figure 3, it can be seen that there are 2 typical groups of curves formed by the data. If
coarsely looked, the diffusion coefficient decrease almost 1:1 to water saturation. But in detail
the decreasing of D in brick (Z1) and mortar (M) is more rapidly then in sandstone S1 and all
porous glasses and shows potential relation. The other extreme symbolized the sandstone S2
because D decreases more moderate at higher water saturation. In-between 20 % - 80 % water
saturation the curves in Figure 3 (left side) follow the sequence S2>S1>M>Z1.
 161

Table 3. Comparison of the different methods to determine Diffusion coefficient in relation to

moisture content (advantages and disadvantages)
3-layer diffusion experiment impedance spectroscopy
pros • concentration curve is visible (mistakes • quick and easy measurement
influence the curve shape, but it can be • all moisture degrees are meas-
taken into account) urable on only one sample
cons • different moisture content of the three • incorrect electrode contact leads
plates leads to moisture transport and to measurement artifacts
change the salt concentration • if the electrical behavior of the
• the contact between the plates can limit the material comes in the same or-
salt transport der of magnitude then cable and
• low moisture content are difficult to man- stray immitance’s which lead to
age in experiment measurement errors in the high
• immense scope of sample preparing and frequency zone (high conduct-
analysis ing materials)
• every moisture content needs new samples
and an own diffusion experiment

1 1
y=x y=x

0,8 0.8
D/DSw=100%
D/DSw=100%

0,6 0.6

0.4
0,4

0.2 P1 P2 P3
0,2 S1 S2
P4 P5
Z1 Mö
0
0
100% 80% 60% 40% 20% 0%
100% 80% 60% 40% 20% 0%
water saturation SW
water saturation SW

Figure 3. Relative values of diffusion coefficients in relation to water saturation of different materials measured
by impedance spectroscopy

This corresponds with the arc-depression factor n and therefore with distribution of relaxa-
tion time. The expected curve shape should theoretically follow an straight line up to a certain
water saturation due to reducing of solution volume. Underneath the concentration of the pore
solution increases rapidly which causes in non-proportional decreasing of material conductiv-
ity. This curve shape show all the porous glasses and the sandstone. But the potential relation
between D and Sw of brick and mortar can’t be explained by this hypothesis. Obviously there
is an additional cause which decreases D. This might either due to pore morphology (connec-
tivity of pores, pore size distribution) or to ion adsorption/desorption at inner surface. The two
groups of materials with brick and mortar on one side and the sandstone’s and the porous
glasses on the other differ both in pore morphology and chemistry of inner surface. The ar-
gument that the pore morphology with the connectivity of the pores have a great influence on
162

the water distribution and so on the conductivity seems to be more plausible. Further investi-
gation will verify this hypothesis.
This work was supported by the Deutsche Bundesstiftung Umwelt and by the Deutsche
Forschungemeinschaft at the SFB 524 “Materials and constructions for the revitalization of
buildings”.

5 References:

1. Buchwald, Ch. Kaps, GDCh-Monographie 15, 1999, 46-51

2. H. Sadouki, A. Buchwald, F.H. Wittmann, Ch. Kaps, International Journal for Restoration
of Building and Monuments, 1999, 5, 3-14
3. R. Kriegel, ”Win_Diff – a diffusion simulation program”, Weimar 1999
4. J.-H. Hwang, K.S. Kirkpatrick, T.O. Mason, E.J. Garboczi, Solid state ionics. 1997, 98,
93-104
5. P. Gu, P. Xie, Y. Fu, J.J. Beaudoin, Cement and Concrete Research, 1994, 24, 86-88
6. B.J. Christensen, T.O. Mason, H.M. Jennings, J. Am. Ceram. Soc. 1992, 75, 4, 939-945
7. T. Kudo, K. Fueki, Solid state ionics, VCH, Tokyo, 1990, p.62
8. Buchwald, Ch. Kaps: Zum Ionentransport bauschädlicher Salze in feuchtebelasteten
Bauwerkstoffen. 5th International Colloquium -Materials Science and Restoration-
30.nov.-2.dec.1999, in progress
9. C.F. Rust, Petroleum Transactions, AIME, 1952, 195, 217-224
10. M.J.M. Romkens, R.R. Bruce, Soil Science, 1964, 98, 332-337
11. L.K. Porter, W.D. Kemper, R.D. Jackson, B.A. Stewart, Soil Sci. Soc. Proc. 1960, 24, 4,
460-463