Content uploaded by Eric Doehne
Author content
All content in this area was uploaded by Eric Doehne on Dec 14, 2017
Content may be subject to copyright.
SALT WEATHERING: INFLUENCE OF EVAPORATION RATE,
SUPERSATURATION AND CRYSTALLIZATION PATTERN
CARLOS RODRIGUEZ-NAVARRO*{AND ERIC DOEHNE
The Getty Conservation Institute, 1200 Getty Center Drive, Suite 700, Los Angeles, CA 90049, USA
Received 11 September 1997; Revised 15 May 1998; Accepted 19 July 1998
ABSTRACT
Micro- and macroscale experiments which document the dynamics of salt damage to porous stone have yielded data
which expose weaknesses in earlier interpretations. Previously unexplained differences are found in crystal
morphology, crystallization patterns, kinetics and substrate damage when comparing the growth of mirabilite
(Na
2
SO
4
.10H
2
O) and thenardite (Na
2
SO
4
) versus halite (NaCl). The crystallization pattern of sodium sulphate was
strongly affected by relative humidity (RH), while a lesser RH effect was observed for sodium chloride. Macroscale
experiments confirmed that mirabilite (crystallizing at RH> 50 per cent) and thenardite (crystallizing at RH< 50 per
cent) tend to form subflorescence in highly localized areas under conditions of constant RH and temperature. This
crystallization pattern was more damaging than that of halite, since halite tended to grow as efflorescence or by filling
the smallest pores of the stone in a homogeneous fashion, a result which contradicts Wellman and Wilson's theoretical
model of salt damage. Low RH promoted rapid evaporation of saline solutions and higher supersaturation levels,
resulting in the greatest damage to the stone in the case of both sodium sulphate and sodium chloride crystallization.
At any particular crystallization condition, sodium chloride tended to reach lower supersaturation levels (resulting in
the crystallization of isometric crystals) and created negligible damage, while sodium sulphate reached higher
supersaturation ratios (resulting in non-equilibrium crystal shapes), resulting in significant damage. ESEM showed no
damage from sodium sulphate due to hydration. Instead, after water condensation on thenardite crystals, rapid
dissolution followed by precipitation of mirabilite took place, resulting in stone damage by means of crystallization
pressure generation.
It is concluded that salt damage due to crystallization pressure appears to be largely a function of solution
supersaturation ratio and location of crystallization. These key factors are related to solution properties and
evaporation rates, which are constrained by solution composition, environmental conditions, substrate properties,
and salt crystallization growth patterns.
When combined with a critical review of salt damage literature, these experiments allow the development of a
model which explains variations in damage related to combinations of different salts, substrates and environmental
conditions. Copyright #1999 John Wiley & Sons, Ltd.
KEY WORDS salt weathering; environmental conditions; crystallization pressure; stone decay; sodium sulphate; sodium chloride
INTRODUCTION
Over the last few decades, many field and laboratory studies have shown that salt weathering is an
important mechanism of rock decay in a range of environments (Evans, 1970; Pye and Sperling, 1983;
Goudie, 1985; Trenhaile, 1987; Yatsu, 1988; Matsukura and Kanai, 1988; Cooke et al., 1993; Goudie
and Viles, 1997). Salt weathering is thought to play a role in the development of many geomorphologic
features such as alveoles or honeycombs (Mustoe, 1982), tafoni or cavernous weathering (Bradley et al.,
1978) and pedestal rocks (Chapman, 1980). It is also considered to be an important contributor to the
production of rock debris, especially in arid regions (Beaumont, 1968; Goudie, 1977; Goudie and Day,
1980), coastal areas (Mottershead, 1989) and in Antarctica (Prebble, 1967). Indeed, salt weathering may
be a common planetary phenomenon, since Malin (1974) has suggested it could be responsible for some
aspects of Martian landscapes (see also Clark and Van Hart, 1981; Clark, 1993). In addition, salt
Earth Surface Processes and Landforms
Earth Surf. Process. Landforms 24, 191±209 (1999)
CCC 0197-9337/99/030191±19 $17.50
Copyright #1999 John Wiley & Sons, Ltd.
* Correspondence to: C. Rodriguez-Navarro, The Getty Conservation Instutute, 1200 Getty Center Drive, Suite 700, Los Angeles, CA 90049,
USA.
{Present address: Dpto. Mineralogia y Petrologia, Universidad de Granada, Fuente Nueva s/n, 18010 Granada, Spain.
Contract/grant sponsor: Getty Conservation Institute
weathering of building stones is recognized as one of the primary agents in the loss of historic
architecture (Evans, 1970; Winkler, 1973; Amoroso and Fassina, 1983; Smith et al., 1988; Winkler, 1994;
Cooke, 1994; Price, 1996; Cooke and Gibbs, 1995); and many structures in archaeological sites have also
been damaged by this process (Lucas, 1925; Winkler, 1965; Goudie, 1977). Salts have had a significant
impact on civil engineering, since they have been found to be responsible for damage to roads, highways,
runways, dams and building foundations (Netterberg and Loudon, 1980; Sayward, 1984; Obika et al.,
1989; Doornkamp and Ibrahim, 1990; Goudie and Viles, 1997). In hot, arid areas, such as the Middle
East, the full significance of salt weathering as an engineering hazard became evident after the
construction boom of the 1970s and 1980s (Doornkamp and Ibrahim, 1990). Additionally, salt
crystallization tests have been used since 1828 to evaluate the durability of construction materials
(Evans, 1970; Price, 1978). However, recent debate as to the suitability of the standard sodium sulphate
test for aggregate durability (ASTM, BRE) has also brought into focus our imperfect knowledge of this
process (McMahon et al., 1992). From the above it is obvious that salt weathering is not only of
academic interest but, in some circumstances, it is a hazard with significant cultural and economic
implications.
Nevertheless, while many researchers have studied this subject, the processes and pathways of salt
damage are still poorly understood (Zehnder and Arnold, 1989; Doehne, 1994; Rodriguez-Navarro et
al., 1996). In fact, there appear to be many unverified assumptions as to how and why salt weathering
occurs. This has led to some general, well accepted salt damage models that do not explain many
observed phenomena. For instance, it is widely accepted that salts cause more damage to rocks
containing a large proportion of micropores connected to macropores, because of thermodynamic
assumptions that salts growing in the macropores have a lower free energy of formation and tend to be
supplied with solution from the micropores. A theoretical model for this assumption was promulgated
by Wellman and Wilson (1965). This notion appears to be well accepted in the literature (Fitzner and
Snethlage, 1982; Zehnder and Arnold, 1989; Rossi-Manaresi and Tucci, 1991; Fitzner et al., 1996),
despite a paucity of detailed experimental observations as to where and how salts grow in porous
materials, as well as how the resulting damage takes place. Furthermore, a clear explanation has been
lacking for the contrasting damage potential of different salts. Many empirical studies have ranked the
ability of various salts to cause damage to porous stones (Goudie and Cooke, 1970; Kwaad, 1970;
Goudie, 1974, 1986, 1993; McGreevy and Smith, 1982). However, no convincing and experimentally
verified answer yet exists to the following question: Why are some salts more damaging than others? In
addition, there is little agreement on the importance of the various parameters contributing to salt
weathering. For instance, some consider as more determinant the stone properties (McGreevy, 1996),
others, the environmental conditions (Sperling and Cooke, 1985; Davison, 1986), while still others focus
on the salt composition (Kwaad, 1970; Goudie, 1993). While all these variables have some significance,
and studying each in isolation yields useful information, such studies may generate results that cannot
readily be compared and may further complicate our lack of knowledge concerning the key parameters
controlling salt weathering and the dynamics of this decay process.
To date, few studies have been performed on the dynamics (and the effects) of crystal growth within a
capillary or a pore considering the particular crystallographic behaviour of each salt involved in the
process of salt weathering. Zehnder and Arnold (1989) and Obika et al. (1989) showed that the
crystallographic characteristics of a particular salt may control the pattern and degree of damage created
when crystallization and growth take place within a pore. Thus, the morphology and growth rate of the
newly formed salt crystals appear to be important keys to understanding this particular decay
phenomenon. Sunagawa (1981) demonstrated that crystal shape and morphology are dependent on the
degree of supersaturation reached when crystallization takes place, and this parameter is mainly
controlled by the initial solution concentration and environmental/physical parameters such as
temperature, relative humidity, surface area and roughness of the substrate. Thus, the physical
properties of a salt solution (i.e. viscosity, surface tension and vapour pressure) will have a critical effect
on the dynamics of solution flow and evaporation within the stone, and therefore on the degree of
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
192
supersaturation, the dynamics of precipitation and salt growth, and the resulting damage to porous host
materials (Puehringer and Engstrom, 1985).
Another issue to be dealt with when addressing salt decay problems is the difference between the
process of salt crystallization on the surface of a porous material (efflorescence) and the crystallization
within the porous system (subflorescence). The former, while appearing impressive, typically results in
little damage to the stone, whereas, by contrast, the latter typically creates extensive damage (Cooling,
1930). Certain salts are prone to effloresce, while others crystallize more readily inside a pore
(Rodriguez-Navarro et al., 1996), and it is important to understand why. Environmental changes
(especially relative humidity) can promote changes in the location and/or in the way salt crystallization
and growth take place, either promoting the formation of efflorescence or generating subflorescence.
Additionally, little attention has been paid to the fact that a liquid (salt solution) in a porous rock can
reduce the mechanical strength of the rock. Dunning and Huf (1983) demonstrated experimentally that
the presence of a liquid in the pores of a rock submitted to stress, greatly increased the rate of crack
propagation. They explained this phenomenon as a reduction in the interfacial tension between rock
minerals, due to wetting by the liquid. Walder and Hallet (1985) and Selby (1993) pointed out that stress
corrosion cracking phenomena that occur in the presence of a liquid are important decay processes,
which have received little attention in geomorphological literature. Thus, when considering the damage
produced by salts, the crack-enhancing effects of the solution itself must be taken into account when salt
crystallization or hydration pressures are generated. This means that a rock can experience considerable
damage even if little pressure is exerted by the salts.
Considering the importance of this decay mechanism, not only from a geomorphologic point of view,
but also from a cultural and economic standpoint, it is essential to reveal the fundamental mechanisms
leading to salt damage, identify the key parameters controlling this process, and determine how a better
understanding of salt weathering can lead to new methods of mitigating this problem in the fields of
engineering and heritage conservation.
In this paper, we undertake a series of dynamic experiments to begin to resolve these problems by
comparing the crystallization behaviour of two typical salts under conditions of low and high relative
humidity. The existing theories and models suggested for this process are also critiqued by combining
information from studies in geomorphology, conservation science and engineering. An analysis of the
results reveals some key parameters which control salt damage of porous stone, how and where
crystallization occurs and in what kind of pores, elucidates why sodium chloride is less damaging than
sodium sulphate and, finally, permits a critical evaluation of existing salt weathering models and
prompts suggestions for future research. This work finds that the supersaturation ratio, the kinetics of
nucleation and salt growth, the influence of environemtnal conditions (such as relative humidity), and
how and where supersaturation is reached in a porous support are important parameters in the salt
damage process.
SALT WEATHERING: THEORIES AND MODELS
Several salt damage mechanisms and corresponding theoretical models have been postulated to explain
the widespread damage observed in field and laboratory studies of the problem. Primarily, salt
weathering has been reported to be due to the build-up of crystallization pressure against pore walls
when salt nucleation and growth take place in a confined space (Evans, 1970). Correns (1949) postulated
that a salt crystal can grow against a confining pressure (as demonstrated by the experiments carried out
by Taber (1916) if a thin, supersaturated solution film is maintained at the salt±pore wall interface. The
existence of the film is determined by the interfacial tension differences at the salt±solution, the salt±rock
and the solution±rock interfaces. If the salt±rock interfacial tension is bigger than the sum of the other
two terms, the solution can penetrate between the salt crystal and the pore wall, permitting the continued
growth of the salt. The pressure is transmitted across the solution film, which may be semi-rigid under
these conditions (Robin, 1978; Walder and Hallet, 1985). Correns (1949) proposed the following
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
193
equation which allows calculation of the pressure P(atm) exerted by a crystal growing from a saturated
solution:
P=(RT/Vs) ln (c/c
s
)(1)
where Vs (cm
3
mol
±1
) is the molar volume of the crystal substance, (c/c
s
) is the ratio of actual to
saturation concentration (supersaturation ratio), Ris the gas constant (cm
3
atm K
±1
g-mol
±1
) and Tis
the temperature (K). On the basis of this equation, Winkler and Singer (1972), calculated the pressures
potentially exerted by common salts and concluded that, under certain supersaturation states, the
pressures generated by salts undergoing crystallization inside a pore could be high enough to disintegrate
most commonly used porous building materials. This model has also successfully explained the
`displacive growth' of minerals such as pyrite in confined sediments (Carstens, 1986).
Weyl (1959) proposed a general model to understand the `force of crystallization' of minerals as the
opposite of the pressure solution process. This model considers Correns' model essentially as a special
case. Weyl postulated that growth at a stressed cystal face requires the presence of a supersaturated
solution film between the crystal face and its constraint. The thickness of this film, which is necessary for
the diffusion of solutes to the precipitation site, was assumed by Weyl (1959) and Tada et al. (1987) to be
of the order of several nanometres. The force of crystallization is the result of the deposition of matter on
the growing crystal surface, at the crystal±pore wall interface. If the solution is supersaturated,
preciptiation in the area of contact between the crystal and the pore wall will take place if the ratio of the
supersaturation to the stress coefficient of solubility is greater than the average normal stress between the
crystal and the pore wall. If the stress is increased above this limit, pressure solution will take place. This
model, in fact, explains the experimental results from Taber (1916) who observed that alum crystals grew
faster under low pressure than under high pressures. Weyl (1959) provides different expressions that
permit the forces of crystallizations to be calculated. This model, according to Evans (1970), is the only
acceptable explanation for the observed force of crystallization (salt damage). Nevertheless, Yatsu (1988)
criticized Weyl's model for its assumption that the concentration of solute at the crystal±rock film is a
linear function of the effective normal stress across the film, since the validity of this assumption has not
been verified. However, data from Tada et al. (1987), Maliva and Siever (1988) and Dewers and
Ortoleva (1990) corroborate the validity of Weyl's model, and its applicability to explain various natural
phenomena. A comparable model has been applied recently to explain frost weathering (Walder and
Hallet, 1985, 1986).
Wellman and Wilson (1965) developed a thermodynamic model for calculating the crystallization
pressure of a salt, considering that the free energy of a solid increases with its surface area. Thus, to keep
the free energy of a system at a minimum, the area of its interfaces should tend to be minimized as well.
The work required to be done during crystal growth on one face of the crystal is equal to (P
l
±P
s
)dV,
where P
l
is the pressure in the liquid, P
s
is the pressure in the solid (in atmospheres) and dVis the
increase of crystal volume (cm
3
). This work should be equal to the work required to extend the surface,
which is equal to sdA, where sis the interfacial tension between the solution and the growing crystal
(dyne cm
±2
) and dAis the increase of crystal surface area (cm
2
). sis independent of V, therefore:
P
l
± P
s
=sdA/dV(2)
Wellman and Wilson (1965) concluded that salt crystallization will therefore take place initially in the
larger pores (forming large crystals, growing at the expense of the smaller) from solution being supplied
from the smaller capillaries. When the larger capillaries are filled, the crystals will not grow into the
smaller ones because this will require a significant increase of surface area relative to the small volumetric
increase, resulting in a large increase in the chemical potential of the crystal. Thus, the crystals in the
large pores will continue growing, generating pressure against the pore walls and eventually resulting in
damage to the host material. Wellman and Wilson (1968) proposed that the excess pressure, DP(in
atmospheres) built up in a large pore when crystallization takes place is equal to:
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
194
DP= 2s(1/R ± 1/r) (3)
where Ris the radius of the large pore connected by a capillary of radius r(cm). Thus, when DPexceeds
the tensile strength of the rock, fracture of the material will take place. This suggests that rocks with
large capillaries separated by areas of micropores are most susceptible to disruption by crystallization of
salts. The Wellman and Wilson model is based on the theoretical freezing model developed by Everett
(1961) who established that freezing results in the generation of `capillary pressure' by a solid growing at
the expense of a liquid in a capillary. However, the thermodynamic background for this theory,
according to Evans (1980), is valid for gas±liquid and liquid±liquid interactions, but its use for solid±
liquid interactions has not yet been validated. Moreover, Tiller (1991) stated that the `capillary pressure',
which is valid for liquids, is a fictitious quantity when applied to solids.
Several other salt damage mechanisms have been proposed and are still somewhat controversial:
hydration pressure (Kwaad, 1970; Sperling and Cooke, 1985; Fahey, 1986; Charola and Weber, 1992;
Doehne, 1994), thermal expansion (Cooke and Smalley, 1968; Chapman, 1980; Johannessen et al., 1982),
chemical weathering (McGreevy, 1985; Young, 1987; Mottershead and Pye, 1994), osmotic pressure
(Winkler, 1970, 1994), and hydraulic pressure (McMahon, 1992). These mechanisms are assumed to have
little relevance to the current experiments, for which crystallization due to evaporation appears to be the
most important process. A general review of the problem of salt damage and its complexities can be
found in the monograph by Goudie and Viles (1997).
MATERIALS AND METHODS
Stone type and properties
The damaging effects of salt crystallization within a naturally occurring porous medium were studied
using an oolitic limestone (Monks Park stone, UK), which has been used in building construction for
centuries (Leary, 1983). Previous experiments have shown that this stone is sensitive to salt decay
(Schaffer, 1932, 1955; Honeyborne and Harris, 1958; Leary, 1983; Rodriguez-Navarro et al., 1996). It is
a fine-grained, homogeneous, buff-coloured stone. The oolites, which comprise c. 60±70 per cent of the
stone, range in size from 300 to 600mm, and are cemented by sparry calcite. Optical microscopy and
scanning electron microscopy show that the stone has a high intergranular porosity concentrated mainly
in the micritic oolites, and in the contact between the sparitic calcite crystals of the cement. The median
pore radius is 0´25+0´05mm, and the average porosity is 20+2 per cent as determined by mercury
intrusion porosimetry (MIP). This limestone exhibits a small capillary suction coefficient (0´09g cm
±2
),
slow capillary rise (0´32cm min
±1/2
) and slow drying (determined according to RILEM, 1980). These
characteristics may help determine the sensitivity of the stone to salt damage.
Microscale experiments
The aim of the microscale experiments was to study and determine the microscopic characteristics of salt
crystallization and growth in both non-porous and porous media. Two salt solutions were chosen for
this study: saturated sodium chloride (NaCl) and saturated sodium sulphate (Na
2
SO
4
) solutions. The
crystal phases expected to develop were halite (NaCl) and mirabilite (Na
2
SO
4
.10H
2
O). Eventually
thenardite (Na
2
SO
4
) will form by mirabilite dehydration. These salts are well known for their presence in
a wide range of locations and environments (Doorkamp and Ibrahim, 1990). Additionally, the two
solutions allow comparison (in a narrow temperature range) of salt behaviour without hydrates (halite)
and with different hydrated phases (thenardite±sodium sulphate heptahydrate±mirabilite).
Three sets of experiments were carried out. (i) Crystallization from a droplet of solution on a glass
plate at different relative humidity (RH) conditions (20, 35, 50 and 65 per cent RH); (ii) crystallization
within glass capillary tubes of different dimensions (50mm up to 400mm diameter): the course of these
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
195
two experiments was observed and documented at regular intervals via polarizing light microscopy and a
computer-based, time-lapse video system. Phase analysis was performed by X-ray diffractometry (XRD,
Siemens, model D5000). (iii) Crystallization of salts in a stone block observed by environmental scanning
electron microscope (ESEM, Electroscan model E-3). An in situ, high magnification, dynamic
monitoring of the crystallization process was performed in the ESEM chamber, with evaporation and/or
condensation cycling, following the methodology developed by Danilatos (1993) and Doehne (1994).
For studying the dynamics of crystal growth at high magnification, ESEM is a useful new tool because it
allows the direct observation of the process of dissolution and precipitation of salts in porous supports
(Doehne and Stulik, 1990; Doehne, 1994). Additional analyses of the salt crystals formed in stone blocks
under varying RH conditions (macroscale experiments described below) were performed using the
ESEM coupled with energy dispersive microanalysis (EDAX).
Macroscale experiments
The macroscopic characteristics of sodium chloride and sodium sulphate crystallization and growth
within the oolitic limestone, and the resulting damage, were studied and evaluated. Stone blocks were
partially immersed in a salt solution and, following capillary rise, crystallization occurred in the upper
exposed stone areas (the `wick effect', according to Goudie (1986)). The surface of the saline solution was
covered with melted paraffin wax to promote the migration of the solution through the pore system of
the stone, to avoid `creeping' of the salt, and to minimize solution evaporation. Crystallization of the salt
and the resulting damage took place under low RH (35 per cent) and medium-high RH (60 per cent) in a
Micro Climate Technology (MCT) controlled environmental chamber. The temperature was kept
constant at 208C during all the experiments. No RH cycling or temperature variation was promoted
during the experiment, thus, only continuous evaporation, supersaturation and salt crystallization were
expected. Neither deliquescence of the existing salts followed by reprecipitation, nor any hydration
process, would take place in such conditions. Figure 1 shows the experimental set-up which was intended
to reproduce a common natural decay situation (Goudie, 1985, 1986), and at the same time to simulate a
single damage process due solely to crystallization pressure.
The changes in appearance and salt morphologies were documented by means of a time-lapse video
(TLV) system that recorded at a rate of one frame per minute. Details of the TLV system can be found
elsewhere (Rodriguez-Navarro and Doehne, 1998). The extent of damage was evaluated by weighing the
Figure 1. Macroscale salt experiment set-up for crystallization under contrasting RH conditions
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
196
material that fell off of the stone surface (after salt removal by immersion in distilled water). Total
weight loss was expressed as a percentage of the original weight of the stone block. The changes in
porosity and pore size distribution, as well as the distribution of the salts in the decayed stones, were
evaluated by means of mercury intrusion porosimetry. Samples of the unweathered stone, the stone
submitted to salt weathering, and samples taken after salt extraction by immersion in distilled water were
analysed using a mercury intrusion porosimeter (Micromeritic model 3300). Up to four replicate
analyses were performed for each case. The evaporation rate of the two solutions under the two RH
conditions mentioned above was determined by measuring the weight change of the beaker plus the
stone block at constant intervals.
RESULTS
Microscale experiments
Crystallization from a droplet at different RH. Initially, NaCl crystallization always resulted in tiny
euthedral, cubic crystals, which grew at the border and the centre of the droplet. In a second stage,
mostly larger hopper crystals were formed at the edge of the droplet. Finally, massive skeletal and
dendritic crystals grew at the outside of the edge of the droplet, and produced efflorescence-like
precipitates (Figure 2a). This crystallization pattern was not affected by RH conditions. However, the
number of well developed crystals was less numerous at the low RH.
At RH above 50 per cent sodium sulphate solution showed two stages of crystallization. First,
prismatic euhedral mirabilite crystals grew within the limits of the droplet (Figure 2b). Second, acicular
and dendritic aggregates of mirabilite were formed in the centre of the drop. In most cases, no
appreciable dendritic growth was detected at the outside of the droplet at RH values higher than 50 per
Figure 2. Optical microscopy micrographs of salt crystallization in a saturated solution droplet: (a) sodium chloride; (b)
mirabilite growing at 60 per cent RH; (c) dendrite of mirabilite growing at the drop edge after dehydration of the crystals (high
RH); (d) thenardite crystals growing inside the droplet at low RH
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
197
cent. However, dendritic, branch-like crystals grew from the edges of the droplet in a small number of
cases (Figure 2c). These crystals grew at the final stage of evaporation of the droplet, once the
dehydration of mirabilite had begun. It should be noted that when dehydration of mirabilite occurred,
the release of water of crystallization produced small, episodic pulses of water at the edge of the drop,
which generated the secondary growth of efflorescence-like crystals outside the droplet. This is a well
known phenomenon that was described by Washburn (1927) and Hazlehurst et al. (1936) and is
considered to be responsible for the so-called `creeping' of salt solutions.
In most of the experiments carried out at RH below 50 per cent, prismatic, acicular or bulky
thenardite crystals (identified by XRD) grew inside the drop (Figure 2d). Thenardite crystals grew
directly from the solution and they were not formed from dehydration of mirabilite. Crystallization at 20
per cent RH resulted in the formation of either euhedral, rhombohedral thenardite crystals (identified by
Figure 3. Optical microscopy micrographs of salt crystals growing in glass capillaries: (a) halite; (b) mirabilite
Figure 4. ESEM micrographs of waxy halite crystals filling micropores (a) and halite cubic efflorescence aggregates developed on
the surface of the stone (b); prismatic, hollow-faced and hopper halite aggregates (c) and anhedral halite crystals growing in
fractures (d), developed at low RH conditions
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
198
XRD) or of large prismatic or hexagonal crystals. The latter crystals underwent a very slow dehydration
(when compared with mirabilite) and were, presumably, Na
2
SO
4
.7H
2
O, a metastable phase reported by
Sperling and Cooke (1985). However their identification by XRD was not successful.
Crystallization within capillary tubes. Halite crystals usually grew at the air±solution interface. Typically,
a single large crystal was found on the surface of the solution plugging the capillary (Figure 3a). In
contrast, a large number of crystals of mirabilite, with size equal to the capillary diameter, grew inside
the solution (Figure 3b). No mirabilite crystals were found growing at the air±solution interface.
ESEM study. The dynamic ESEM study of the sodium chloride crystallization process showed that
halite tended to grow within the pores of the stone as micrometre-sized, smooth-faced or waxy crystals
filling the smaller capillaries and cementing the micritic calcite grains (Figure 4a). Halite was also seen as
large euhedral cubic efflorescence aggregates on the surface of the stone (Figure 4b). The presence of
micrometre-sized, halite crystals within the smallest pores of the stone (within the micritic oolites) was
confirmed by EDAX microanalysis, and seems to be quite common (Eswaran et al., 1980; Obika et al.,
1989). The rate of halite precipitation and growth within the pores of the stone was very slow (minutes).
According to data from Sunagawa (1981), the cubic euhedral halite crystal morphology results from
precipitation at low supersaturation ratios.
Samples collected from the stone submitted to sodium chloride crystallization in the environmental
chamber (macroscale experiments) showed the same type of crystals developed in the ESEM
experiments. However, after sodium chloride crystallization at 35 per cent RH, hollow-faced, hopper
and prismatic halite crystals were found (Figure 4c). The latter morphologies may indicate slightly higher
supersaturation ratios (Attolini et al., 1986). Aditionally, at low RH, some halite crystals were found
growing inside fractures, specially at the oolite±cement interfaces (Figure 4d). At high RH conditions,
halite tended to grow as perfect, smooth-faced cubic crystals, mainly as efflorescence or as waxy,
micrometre-sized aggregates filling the stone pores.
Figure 5. ESEM micrographs of sodium sulphate growing in a microcrack before (a) and (b) dehydration. Micrometre-sized
sodium sulphate crystals (thenardite?) developed in fractures and on the faces of spalled sparitic calcite grains (c). Whisker-like
sodium sulphate crystals (mirabilite) during dehydration (d)
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
199
Mirabilite growth from solution within the stone was extremely rapid (almost instantaneous),
resulting in three different morphologies. (i) Inside the pores, tiny anhedral or euhedral, plate-like and
prismatic mirabilite crystals developed (Figure 5a), evolving into flake-like thenardite crystals once
dehydration took place (Figure 5b). The growth of these crystals resulted in the formation of cracks
within the sparitic calcite crystals cementing the oolites. (ii) In some cases, the development of anhedral,
micrometre-sized sodium sulphate crystals was observed at the oolite±cement interfaces, or at the
contact between sparitic calcite grains (Figure 5c). The growth of these crystals, which are probably
thenardite crystals since no dehydration features were observed, resulted in the detachment of the oolites
and the development of fractures within the calcite cement. The same type of crystals were observed in
the delaminated stone scales collected from the macroscale experiments run at low RH conditions. After
a second condensation and evaporation cycle in the ESEM, dissolution of the thenardite crystals took
place, followed by precipitation of mirabilite crystals. No hydration of thenardite to form mirabilite was
observed. (iii) Both on the internal faces of c. 2mm up to c. 3mm thick delaminated stone scales
(collected from the macroscale experiments) and on the external surface of the stone, large `whisker-like'
crystals were found (Figure 5d). These `whiskers', which were abundant on the surface of the stone
submitted to salt decay at high RH, grew very slowly (c. 3mm per day, as observed by the TLV
recording), and generally after the scales were formed. This contributed to the mechanical separation of
scales from the undamaged stone and eventually resulted in their falling off the stone. According to
Sunagawa (1981), anhedral or non-equilibrium crystal habits indicate high supersaturation ratios. This is
the case of morphologies (i) and (ii). Zehnder and Arnold (1989) reported the growth of euhedral and
whisker-like salt crystals (corresponding to morphology (iii) in our experiment) at low supersaturation
ratios.
Figure 6. General view of the limestone blocks: (a) after sodium chloride crystallization at low RH; (b) after sodium sulphate
crystallization at 35 per cent RH; (c) after sodium sulphate crystallization at 60 per cent RH
Table I. Weight loss (DW) and porosity of limestones after salt crystallization text
Salt RH conditions
(%)
DW*(%) Porosity (%)
with salt
Porosity (%)
without salt
NaCl 60 0´68 8´8 19´3
35 2´67 10´3 21´9
Na
2
SO
4
60
35
7´06
21´43
20´2
21´4
20´5
20.8
*DW=100(W
0
±W
1
)/W
0
, where W
0
is the initial weight and W
1
is the weight of the salt-free decayed sample
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
200
Macroscale experiments
Table I shows the average weight loss and porosity values for the macroscale experiments (data
collected at the end of the experiment, which lasted 30 days). Regardless of RH conditions, an impressive
volume of efflorescence covered the faces of the stone in which NaCl crystallization occurred (Figure 6a).
Little damage, mainly due to powdering of the stone surface, was observed in these blocks. However, the
stone submitted to sodium chloride crystallization at low RH showed granular disintegration and thin
(less than 0´5mm thick) scale formation when salt efflorescence was removed by immersion in distilled
water. Massive loss of scales (more than 2mm in thickness) and stone surface without development of
noticeable efflorescence was observed in samples where sodium sulphate crystallization occured at low
RH. One sample was almost totally destroyed after 30 days from the start of the experiment (Figure 6b).
Curiously, the shape of the top of the damaged block was rounded (probably resembling the shape of the
capillary fringe). In comparison, large amounts of efflorescence and less weight loss were observed when
crystallization took place at a relative humidity of 60 per cent (Figure 6c). Mercury intrusion
porosimetry (MIP) of both surface and core samples of the stone submitted to NaCl crystallization
shows an important porosity reduction, due to the filling of the smallest pores (Figure 7a). MIP analysis
of both superficial and core samples collected after sodium sulphate crystallization shows little change
when compared with the blank stone sample (Figure 7b).
As a general trend, and under similar conditions (temperature, RH, experimental set-up and stone
support), it was observed that at the beginning of the experiment the evaporation rate of sodium
sulphate solution was slower than the evaporation rate of sodium chloride solution (Figure 8). However,
after some days the evaporation rate of NaCl solution was steadily reduced, indicating that the pores
were filled with precipitated halite crystals, somehow preventing further damage (cementing action).
Both NaCl and Na
2
SO
4
solution evaporated faster at low RH than at high RH (Figure 8). Therefore,
under constant capillary rise, the highest supersaturation ratios should occur at low RH.
Thenardite crystals may form at low RH, as shown by the experiments on crystallization from a
solution droplet. Kwaad (1970) and Sperling and Cooke (1985) reported greater crystallization pressure,
and damage, for thenardite crystallization than for mirabilite crystallization. However, it was not
possible to identify which sodium sulphate phase was responsible for the initial damage to the stone.
After collection of the salts found in lost scales, only thenardite was present (XRD data), but if mirabilite
was the original salt, dehydration may have occurred since the experiments were run at RH below the
transition point between thenardite to mirabilite, which is given as 71 per cent at 208C by Arnold (1976).
Only in the case of whisker growth was mirabilite identified. However, a few minutes after the collection
of these latter crystals, dehydration and transformation to thenardite were observed.
Figure 7. MIP data of limestone before and after crystallization of sodium chloride (a) and sodium sulphate (b) at different RH
conditions (data collected before salt removal)
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
201
DISCUSSION
Important and significant differences have been found in the crystallization and growth patterns of
sodium chloride and sodium sulphate that may contribute to the extreme difference in damage potential
commonly observed for these salts. Growth in solution droplets and in capillary tubes shows that halite
tends to grow as perfect cubic crystals (initially forming inside the droplet), while the larger volume of
material crystallizes as hopper crystals or as efflorescence-like, dendritic aggregates generally at the air±
solution interface. In contrast, mirabilite (high RH conditions) and thenardite (low RH conditions) tend
to grow within the solution as single, fast-growing prismatic, well developed crystals. However, at high
RH condition, mirabilite sometimes develops dendritic aggregates outside the solution droplet. This may
indicate that halite is prone to form efflorescence (air±solution interface) while mirabilite and thenardite
are prone to form subflorescence (within the solution).
Under similar environmental conditions and using the same stone type, it was observed that negligible
(high RH conditions) or little (low RH conditions) damage took place from halite growth. However,
impressive damage occurs due to mirabilite and thenardite growth. Halite forms a large volume of
efflorescence, and within the stone this salt is found filling the micropores in a uniform fashion. In the
case of massive damage due to mirabilite/thenardite growth, little efflorescence growth was detected
(under low RH conditions). Sodium sulphate was not homogeneously distributed through the stone pore
system. No appreciable quantities of sodium sulphate were found filling stone pores (MIP results), but a
high concentration of salt was found in a very narrow layer at the subsurface of the stone, where scales
were formed and had been detached. At high RH, mirabilite growth results either in the development of
efflorescence or the formation of scales, but with relatively less damage. However, the damage was still
much greater than in the case of halite.
Sodium chloride versus sodium sulphate
Halite has been found to be responsible for salt damage in many locations and in many rock types
(Evans, 1970; Chapman, 1980; Winkler, 1994). Nevertheless, in a large number of laboratory tests in
which the effects of different salts were compared, NaCl crystallization produced little damage when
compared with other salts, such as alkali metal sulphates, carbonates or nitrates (Kwaad, 1970; Goudie
and Cooke, 1970; Smith and McGreevy, 1983, 1988; Goudie, 1986, 1993).
In the case of sodium sulphate, it should be noted that the high damage potential of this salt is a well
documented observation that has been discussed by many authors (e.g. Schaffer, 1932; Evans, 1970;
Figure 8. Evaporation rate from a limestone slab for NaCl and Na
2
SO
4
saturated aqueous solutions at different RH conditions
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
202
Marschner, 1978; Price, 1978; Cooke, 1979; Sperling and Cooke, 1985; Goudie and Viles, 1997). This
damage potential has led to the widespread use of this salt for testing the durability of building materials
(Price, 1978). However, according to many authors (Kwaad, 1970; Charola and Weber, 1992; McMahon
et al., 1992; Doehne, 1994) the exact mechanism of sodium sulphate-induced decay is still not well
understood. Cooke (1981) concluded that sodium sulphate is so damaging because: (a) sodium sulphate
undergoes a high degree of volume change when hydrated; (b) sodium sulphate suffers a decrease of
solubility at temperatures above (slow decrease) and below (rapid decrease) 32´48C; (c) sodium sulphate
is very soluble, thus a substantial quantity of solid salt is available for crystals to grow when evaporation
takes place; and (d) the needle-shaped nature of sodium sulphate crystals might tend to increase their
destructive force due to the crystallization pressure being concentrated over a smaller surface area.
Like Cooke (1981), Winkler (1973), Goudie (1977) and Marschner (1978) assumed that most of the
damage attributed to this salt was caused by hydration pressure generated by the volume expansion as
thenardite is hydrated to form mirabilite. However, no hydration took place in our experiments since the
temperature and RH were kept constant. Only crystallization and growth (followed by eventual
dehydration of mirabilite to form thenardite) took place, and still the damage was impressive.
Additionally, no cooling or heating took place, and the solubility of Na
2
SO
4
at 208C (experimental
conditions) is smaller than that of NaCl (16´1 weight per cent vs. 26´4 weight per cent respectively). Thus,
most of Cook's hypotheses for the high damage potential of sodium sulphate do not explain our
experimental results.
It seems likely that explanations for the extreme differences in damage potential between sodium
sulphate and sodium chloride are associated with the differential crystallization pattern and growth
kinetics of the two salts, as well as the solution physical properties (surface tension, vapour pressure and
evaporation rate).
Influence of crystallization pattern on salt weathering
Mirabilite and thenardite tend to precipitate inside the solution creating subflorescence within the
pores of the stone, while halite tends to precipitate at the air±solution interface, preferentially creating
harmless efflorescence growing on the stone surface. However, MIP and ESEM data show that halite
can also fill the micropores of the stone, while causing little evident damage. These results appear to
conflict with the Wellman and Wilson (1965) theory for salt crystallization, since according to this
theory, salts should grow in the larger pores, while in this experiment halite grew in the smallest. There
does not seem to be a clear explanation for the negligible damage generated by the growth of halite
inside the smallest pores, but it seems that low supersaturation ratios reached before NaCl nucleation,
and a slow rate of halite precipitation and growth (observed in the ESEM), may hinder the build-up of
enough pressure to damage the stone. As observed in the experiment carried out by precipitation on a
glass plate, halite was observed to grow both at the edge of the droplet (a process which may induce
efflorescence growth on a porous support) as well as in the centre of the drop (which may result in halite
precipitation filling the smallest capillaries in a stone). It was generally observed that cubic halite crystals
grew both inside the solution droplet as well as in the pores of the stone (macroscale experiment and
ESEM microscale experiments). This crystal shape, according to data from Sunagawa (1981),
corresponds to halite precipitation at low supersaturation. This implies that the crystallization pressure
generated by this salt should be small (Correns, 1949; Winkler and Singer, 1972).
Eswaran et al. (1980) observed halite as cubic crystal aggregates in most saline crust in soils, reporting
that hair-like or acicular halite crystals appeared only where the evaporation rates were extremely high.
Obika et al. (1989) reported that significant damage to roads by sodium chloride is only observed when
prismatic or acicular halite crystals occur, with these crystals growing at very high supersaturation ratios.
Therefore, it can be hypothesized that according to the model proposed by Weyl (1959), under the low
supersaturation ratios of our experiments halite crystals cannot generate enough pressure to produce the
breakdown of the stone, and crystal growth should stop or else pressure-induced dissolution would take
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
203
place. It seems that a high supersaturation cannot be sustained in NaCl solution in the presence of a
large surface area (pore walls) for heterogeneous nucleation to take place. Since growth occurs
preferentially at the air±solution interface or at the interface with a solid, nucleation should be easy at
low supersaturation. However, in some instances, it was observed that prismatic and hopper halite
crystals were formed when crystallization took place at low RH. Under this condition, damage due to
halite growth, while small if compared with sodium sulphate crystallization, was indeed more evident
than in the case of halite growth at high RH. These observations confirm that low RH conditions
promote higher supersaturation ratios with the result that halite can generate damage when it grows in
non-equilibrium forms.
From dynamic ESEM crystallization experiments it is observed that mirabilite can grow very rapidly
even in the smallest capillaries. Salt crystal morphologies indicate that sodium sulphate crystallization
generally takes place at very high supersaturation ratios. This may be due to the presence of a wide
metastable zone between saturation and the maximum supersaturation that sodium sulphate solutions
can reach before crystallization begins (Winkler and Singer, 1972). In the case of sodium chloride it
seems that this metastable zone is reduced. This factor seems to be an important contributor to the
differential damage of both salts.
To explain the ability of these salts to create damage, it should also be considered that the presence of
saline solution while crystals precipitate enhances crack propagation (as discussed by Dunning and Huf
(1983) and Selby (1993). Additionally, in the case of sodium sulphate, after the crack is generated and
dehydration of mirabilite to form thenardite is completed, a second condensation±evaporation cycle may
result in the dissolution of the already formed sodium sulphate crystals, and the precipitation of new
mirabilite crystals that generate additional damage. This experimental result suggests that hydration
pressure is not an active mechanism in this decay process. Crystallization pressure appears to be the only
mechanism generating stone weathering in our experiments.
Effect of relative humidity and physical properties of the saline solutions
Important differences in damage level were found after crystallization of sodium sulphate under
varying RH conditions. Efflorescence growth, occurring essentially at high RH, produces less damage
than subflorescence growth, which takes place mainly at low RH. This may be explained partly by (i)
differences in evaporation rate (and supersaturation reached) and (ii) differences in the phase formed.
(i) Low RH conditions will enhance the evaporation rate of the saline solution. Because the capillary
rise is constant, since it depends on the solution surface tension, contact angle, viscosity and the pore
radius ± values that are constants for a given salt solution and stone type ± an increase of the
evaporation rate means that the solution will not reach the stone surface, but will evaporate below
the stone surface. In fact, it was observed that the contour of the stone's damaged surface
corresponds to a convex surface that may be representative of the limits of the capillary fringe. Cooke
(1994) reported that under natural conditions salt weathering was concentrated at the limits of the
capillary fringe, where high supersaturation is reached rapidly. Hence subflorescence, and damage
resulting from high crystallization pressures (due to high supersaturation), will be promoted.
(ii) On the other hand, at low RH direct precipitation of thenardite is observed (in the droplets), and the
crystallization pressure of thenardite is greater than that of mirabilite for equal supersaturation ratios
(Winkler and Singer, 1972). Thus, thenardite may promote greater damage than mirabilite, as
observed also by Sperling and Cooke (1985) and Fahey (1985). However, it is not clear which factor
is determinant, i.e. the high supersaturation ratio reached by rapid evaporation, or the crystallization
of thenardite instead of mirabilite. Perhaps both act together at low relative humidity.
The slow capillary flow rate of saturated sodium sulphate solution allows evaporation to occur faster
than the rate of replenishment of the solution by capillary migration from the inside of the stone. This
will result in rapid supersaturation and subsequent damage due to salt crystallization under the stone's
surface (Lewin, 1981). In contrast, sodium chloride saturated solution, having faster capillary rise (at
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
204
least at the beginning of the experiment, before plugging of the micropores takes place), will be prone to
evaporate at the surface of the stone, precipitating mainly as efflorescence. The differential flow
behaviour of the two saline solutions is due mainly to the higher surface tension of the saturated sodium
chloride solution, compared with the saturated sodium sulphate solution (Gummerson et al., 1980;
Garrecht et al., 1991). Once the solutions reach the evaporation front (located at or below the surface of
the stone), the evaporation rate will depend on the solution vapour pressure. Because the vapour
pressure is higher in the case of sodium sulphate solution, faster evaporation will occur and the
supersaturation ratio will be higher than in the case of sodium chloride solution, resulting in higher
crystallization pressures. Pertinent physical properties of the two solutions are given in Table II.
Crystallization of sodium sulphate solutions at medium-high RH (60 per cent) results in the
development of massive efflorescence and relatively little damage. This may be due partly to the ability
of mirabilite to develop dendritic efflorescence-like crystals at the edges of the solution droplet at high
RH, and because the solution replenishment towards the stone faces may be faster than the evaporation
rate. Thus, due to the slow evaporation, low supersaturation will be reached and precipitation of the salt
will take place mostly at the surface of the stone.
Sodium chloride can also generate more damage at low RH conditions. As in the case of sodium
sulphate, rapid evaporation can induce the growth of subflorescence. Under these conditions, high
supersaturation ratios can be reached before halite crystallization begins. Thus, enhancing the
evaporation rate appears to increase the damage potential of this salt. According to Sperling and Cooke
(1985), crystallization of salts and the resulting high crystallization pressures are favoured by rapid
evaporation, thereby increasing the supersaturation of the solution. Rapid evaporation is enhanced by
low RH, high temperature and wind flow. However, even at low RH conditions, the amount of damage
due to halite growth is much less than that for sodium sulphate. In fact, halite efflorescence growth is
dominant even at 35 per cent RH. These results show that differences in solution physical properties and
crystallization patterns between sodium chloride and sodium sulphate can be responsible for their
contrasting weathering properties.
Influence of the stone pore system
The following observations should be considered in order to evaluate the influence of the pore system
characteristics on stone decay due to salt weathering: Schaffer (1932) demonstrated that stones with a
high proportion of micropores were more susceptible to salt decay than those with high proportions of
large pores. This fact has been confirmed by several studies (Honeyborne and Harris, 1958; Fitzner and
Snethlage, 1982). Punuru et al. (1990) developed a model for durability factor calculations of natural
stone based on the pore size distribution. They found that stones having a large number of pores with
radii less than 0´5mm were less durable than others. Wellman and Wilson (1965) offered a
thermodynamic model aiming to explain these observations. However, our experimental results imply
that a reduction of the solution flow rate by the presence of micropores is partly responsible for the
increase of damage in this type of stone, because slow solution flow will result in the crystallization of the
Table II. Comparison between key physical properties of sodium chloride and sodium sulphate saturated solutions. Data source:
Lide (1996). Surface tensions measured by the authors (DuNouy Tensiometer, model CSC No. 70535)
Saturated
solution*
Concentration
(weight %)
Density
(g cm
±3
)
Surface tension
(Pa)
Viscosity
(mPa s)
Vapour pressure{
(kPa)
RHeq (%)
NaCl 26´41 1´1978 8´35 1´986 1´7634 75´4
Na
2
SO
4
16´13 1´1503 7´66 1´812 2´1750 93{
* All values at 208C
{Vapour pressure of distilled water at 208C is 2.3388 kPa
{Data from Price and Brimblecombe
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
205
salt deep under the stone surface. In fact, the hydric properties of a particular stone can be used to
predict its resistance to salt decay phenomena (Honeyborne and Harris, 1958; Everett, 1961), since
porosity and pore size distribution are the principal factors controlling the uptake and transport of
liquid within a stone (Vos, 1976). The Wellman and Wilson model for salt decay does not take into
account that, according to the Laplace and Washburn equation, the capillary (suction) pressure P(atm)
is inversely proportional to the pore radius:
P=2scosy/r (4)
where sis the interfacial tension and yis the contact angle of the meniscus at the pore wall of radius r
(cm). This means that as evaporation progresses, the saturated solution will be suctioned from the larger
pores towards the smaller ones, where it will concentrate. Thus, once a critical supersaturation is
reached, crystallization will take place in the smaller pores, not in the larger ones. The larger pores will
act as reservoirs supplying solution to the smaller ones where high supersaturation ratios can be reached.
Tiller (1991) reported that heterogeneous nucleation of a crystal will occur in the smallest pores, where
the solution will be withdrawn and eventually reach very high supersaturation ratios (thus generating
high crystallization pressures, according to Correns, 1949).
The two factors discussed above may explain why stones with a high proportion of micropores
connected to large pores are very susceptible to salt weathing. Micropores are very common in Monks
Park limestone. Our experimental results confirm that these micropores result in a larger surface area for
evaporation and slower solution transport, thus increasing the chances that high supersaturation ratios
will be reached below the stone surface. In stones with larger pores, capillary rise is limited, surface area
is lower, and solutions will reach the surface more readily, without achieving a high supersaturation, and
resulting in efflorescence growth.
CONCLUSIONS
The extent of salt damage in porous stone due to crystallization pressure appears to be largely a function
of solution supersaturation ratio and the location of crystallization. These key factors are in turn related
to important properties and parameters, such as salt type and evaporation rates. Based on the
experiments presented here, differential damage behaviour between sodium chloride and sodium
sulphate appears largely to be due to differences in crystallization patterns, dynamics, and the location of
the precipitates. Solution physical properties (e.g. surface tension and vapour pressure), environmental
factors and pore-system characteristics of the stone support play key roles in determining the flow rate
and evaporation rate of the saline solution and thus the resulting supersaturation ratio at the moment of
crystallization. Combination of these factors determines whether crystallization will result in damaging
subflorescence or almost harmless efflorescence. Furthermore, the dynamics and kinetics, as well as the
crystallization pattern of each salt, will determine the location of the salt crystals, the crystallization
pressure, and therefore the resulting damage.
The Wellman and Wilson (1965) model for salt damage does not seem to explain our experimental
observations. However, our experimental results are in agreement with the original model for generation
of crystallization pressure proposed by Correns (1949), and completed by Weyl (1959).
Nevertheless, for a better understanding of salt weathering processes, additional work needs to be
done to verify our observations using other salt types (and mixtures), stone types and environmental
conditions (RH, temperature ranges/cycles, and air flow). In particular, the crystallization pattern of
salts and the kinetics of the salt crystallization process should be considered in detail in future studies.
These studies should also include tests to evaluate the significance of cycling (i.e. crystallization/
dissolution and/or hydration/dehydration cycles) and fatigue or threshold effects on salt weathering.
Development of a model that can be used to explain the structure of some natural weathering forms and
their weathering rates, as well as the degradation modes and rates of archaeological and architectural
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
206
stone affected by salt weathering, will be of importance in the design of conservation procedures for the
protection of our cultural heritage.
ACKNOWLEDGEMENT
We are indebted to Dr William S. Ginell, Dr Eric Hansen and Dr Blythe McArthy for criticisms of an
earlier version of the manuscript. The manuscript was greatly improved thanks to the comments and
thorough critical review performed by Dr Ian S. Evans and an anonymous referee. The time-lapse video
system was designed and set up by Adrian Heritage (now at English Heritage). This paper was financed
by the Getty Conservation Institute under the research project entitled `Preservation of Porous
Calcareous Stone'.
REFERENCES
Amoroso, G. G. and Fassina, V. 1983. Stone Decay and Conservation: Atmospheric Pollution, Cleaning, Consolidation and
Protection, Elsevier Science Publishers, Amsterdam, 454 pp.
Arnold, A. 1976. `Behavior of some soluble salts in stone deterioration', in Skoulikidis, T. H. (Ed.), 2nd International Symposium
Deterioration of Building Stones, Technical University of Athens, Athens, 27±36.
Attolini, G., Paorici, C. and Ramasamy, P. 1986. `Skeletal and hollow crystals of cadmium sulfide grown under time-increasing
supersaturation', Journal of Crystal Growth, 78, 181±184.
Beaumont, P. 1968. `Salt weathering on the margin of the Great Kavir, Iran', Geological Society of America Bulletin, 79, 1683±
1684.
Bradley, W. C., Hutton, J. T. and Twidale, C. R. 1978. `Role of salts in development of granitic tafoni, South Australia', Journal
of Geology, 86, 647±654.
Carstens, H. 1986. `Displacive growth of authigenic pyrite', Journal of Sedimentary Geology, 56, 252±257.
Chapman, R. W. 1980. `Salt weathering by sodium chloride in the Saudi Arabian desert', American Journal of Science, 280, 116±
129.
Charola, A. E. and Weber, J. 1992. `The hydration±dehydration mechanism of sodium sulphate', in Delgado-Rodrigues, J.,
Henriquez, F. and Jeremias, J. T. (Eds), Proceedings of VII International Congress on Deterioration and Conservation of Stone,
Lisbon, LNEC, Lisbon, 2, 581±590.
Clark, B. C. 1993. `Geochemical components in Martian soil', Geochimica et Cosmochimica Acta, 57, 4575±4581.
Clark, B. C. and Van Hart, D. C. 1981. `The salts of Mars', Icarus. 45, 370±378.
Cooke, R. U. 1979. `Laboratory simulation of salt weathering processes in arid environments', Earth Surface Processes, 4, 347±
359.
Cooke, R. U. 1981. `Salt weathering in deserts', Proceedings of the Geologist Association, London, 92, 1±16.
Cooke, R. U. 1994. `Salt weathering and the urban water table in deserts', in Robinson, D. A. and Williams, R. B. G. (Eds), Rock
Weathering and Landform Evolution, John Wiley and Sons, Chichester, 193±205.
Cooke, R. U. and Gibbs, G. B. 1995. Crumbling Heritage? Studies of Stone Weathering in Polluted Atmospheres, National Power
plc and PowerGen plc, Swindon, 68 pp.
Cooke, R. U. and Smalley, I. J. 1968. `Salt weathering in deserts', Nature, 220, 1226±1227.
Cooke, R., Warren, A. and Goudie, A. 1993. Desert Geomorphology, UCL Press, London 526 pp.
Cooling, L. F. 1930. `Contribution to the study of florescence. II. The evaporation of water from bricks', Transaction of the
British Ceramic Society, 29, 39±54.
Correns, C. W. 1949. `Growth and dissolution of crystals under linear pressure', Discussions of the Faraday Society, 5, 267±271.
Danilatos, G. D. 1993. `Introduction to the ESEM instrument', Microscopy Research Technique, 25, 354±361.
Dewers, T. and Ortoleva, P. 1990. `Force of crystallization during the growth of siliceous concretions', Geology, 18, 204±207.
Doehne, E. 1994. `In situ dynamics of sodium sulfate hydration and dehydration in stone pores: observation at high
magnification using Environmental Scanning Microscopy', in Fassina, V., Ott, H. and Zezza, F. (Eds), Proceedings of IIIrd
International Congress on the Conservation of Monuments in the Mediterranean Basin, CMU, Venice, 143±150.
Doehne, E. and Stulik, D. 1990. `Applications of the environmental scanning electron microscope to conservation science',
Scanning Microscopy, 4, 275±286.
Doornkamp, J. C. and Ibrahim, H. A. M. 1990. `Salt weathering', Progress in Physical Geography, 14, 335±348.
Dunnig, J. D. and Huf, W. L. 1983. `The effects of aqueous chemical environments on crack and hydraulic fracture propagation
and morphologies', Journal of Geophysical Research, 88, 6491±6499.
Eswaran, H., Stoops, G. and Abtahi, A. 1980. `SEM morphologies of halite (NaCl) in soils', Journal of Microscopy, 120, 343±
352.
Evans, I. S. 1970. `Salt crystallization and rock weathering: A review', Revue de Ge
Âomorphologie Dynamique, 19, 155±177.
Everett, D. M. 1961. `The thermodynamics of frost damage to porous solids', Transactions of the Faraday Society, 57, 2205±
2211.
Fahey, B. D. 1985. `Salt weathering as a mechanism of rock breakup in cold climates: an experimental approach', Zeitscrift fu
Èr
Geomorphologie N.F., 29, 99±111.
Fahey, B. D. 1986. `A comparative laboratory study of salt crystallization and salt hydration as potential weathering agents in
deserts', Geografiska Annaler, 68A, 107±111.
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
207
Fitzner, B. and Snethlage, R. 1982. `Uber Zusammaen Hange Zwischen Saltzkristallisationsdruck und porenradienverteilung',
GP Newsletter, 3, 13±24.
Fitzner, B., Heinrichs, K. and Volker, M. 1996. `Model for salt weathering at Maltese globigerina limestones', in Zezza, F. (Ed.),
Origin, Mechanisms and Defects of Salts on Degradation of Monuments in Marine and Continental Environments, Protection
and Conservation of the European Cultural Heritage Research Report 4, Tecnomack, Bari, 333±344.
Garrecht, H., Hilsdorf, H. K. and Kropp, J. 1991. `Hygroscopic salts: influence on the mosture behavior of structural elements',
in Proceedings of 5th International Conference on Durability of Building Materials and Components, Brighton, Chapman and
Hall, 313±324.
Goudie, A. S. 1974. `Further experimental investigation of rock weathering by salt and other mechanical processes', Zeitschrift
fu
Èr Geomorphologie, Supplementband, 21, 1±12.
Goudie, A. S. 1977. `Sodium sulfate weathering and the disintegration of Mohenjo-Daro, Pakistan', Earth Surface Processes, 2,
75±86.
Goudie, A. S. 1985. Salt Weathering, Research Paper, School of Geography, University of Oxford, 33, 31 pp.
Goudie, A. S. 1986. `Laboratory simulation of ``the wick effect'' in salt weathering of rock', Earth Surface Processes and
Landforms,,11, 275±285.
Goudie, A. S. 1993. `Salt weathering simulation using a single-immersion technique', Earth Surface Processes and Landforms,,18,
369±373.
Goudie, A. S. and Day, M. J. 1980. `Disintegration of fan sediments in Death Valley, California, by salt weathering', Physical
Geography, 1, 126±137.
Goudie, A. S. and Viles, H. 1997. Salt Weathering Hazards, John Wiley and Sons, Chichester, 241 pp.
Goudie, A. S., Cooke, R. and Evans, I. S. 1970. `Experimental investigation of rock weathering by salts', Area, 4, 42±48.
Gummerson, R. J., Hall, C. and Hoff, W. D. 1980. `Water movement in porous building materials: II. Hydraulic suction and
sorptivity of brick and other masonry materials', Building and Environment, 17, 101±108.
Hazlehurst, T. H., Martin, H. C. and Brewer, L. 1936. `The creeping of saturated salt solutions', Journal of Physical Chemistry,
40, 349±452.
Honeyborne, D. B. and Harris, P. B. 1958. `The structure of porous building stone and its relation to weathering behavior', in
Everett, D. H. and Stone, F. S. (Eds), Proceedings 10th Symposium of the Colston Research Society, Butterworths Scientific
Publications, London, 343±365.
Johannessen, C. L., Freiereisen, J. J. and Wells, A. N. 1982. `Weathering of ocean cliffs by salt expansion in a mid-latitude
coastal environment', Shore and Beach, 50, 26±34.
Kwaad, F. J. P. M. 1970. `Experiments on the granular disintegration of granite by salt action', Fysisch Geografisch en
Bodemkundig Laboratorium, 16, 67±80.
Leary, E. 1983. The Building Limestones of the British Isles, Building Research Establishment Report, London, 78 pp.
Lewin, S. Z. 1981. `The mechanism of masonry decay through crystallization', in Barkin, S. M. (Ed.), Conservation of Historic
Stone Buildings and Monuments, National Academy of Sciences, Washington DC, 120±144.
Lide, D. R. (Ed). 1996. Handbook of Chemistry and Physics, 77th edn, CRC Press, Boca Raton.
Lucas, A. 1925. Antiques, their Restoration and Preservation, Arnold E. and Co., London, 136 pp.
Malin, M. C. 1974. `Salt weathering on Mars', Journal of Geophysical Research, 79, 3888±3894.
Maliva, R. G. and Siever, R. 1988. `Diagenetic replacement controlled by force of crystallization', Geology, 16, 688±691.
Marschner, H. 1978. `Application of salt crystallization test to impregnated stones', in UNESCO/RILEM International
Symposium on Deterioration and Protection of Stone Monuments, Reliure, Paris, 3.4, 16 pp.
Matsukura, Y. and Kanai, H. 1988. `Salt fretting in the valley cliff of the Asama volcano region, Japan', Earth Surface Processes
and Landforms, 13, 85±90.
McGreevy, J. P. 1985. `A preliminary scanning electron microscope study of honeycomb weathering of sandstone in a coastal
environment', Earth Surface Processes and Landforms, 10, 509±518.
McGreevy, J. P. 1996. `Pore properties of limestone as controls on salt weathering susceptibility: a case study', in Smith, B. J. and
Warke, P. A. (Eds), Processes of Urban Stone Decay, Donhead, London, 150±167.
McGreevy, J. P. and Smith, B. J. 1982. `Salt weathering in hot desert: observation on the design of simulation experiments',
Geografiska Annaler, 64A, 161±170.
McMahon, D. J., Sandberg, P., Folliard, K. and Mehta, P. K. 1992. `Deterioration mechanisms of sodium sulfate', in Delgado-
Rodrigues, J., Henriques, F. and Jeremias, F. (Eds), Proceedings of VII International Congress on Deterioration and
Conservation of Stone, Lisbon, LNEC, Lisbon, 2, 705±714.
Mottershead, D. N. 1989. `Rates and patterns of bedrock denudation by coastal salt spray weathering: a seven-year record',
Earth Surface Processes and Landforms, 14, 383±398.
Mottershead, D. N. and Pye, K. 1994. `Tafoni on coastal slopes, South Devon, U.K.', Earth Surface Processes and Landforms,
19, 543±563.
Mustoe, G. E. 1982. `The origin of honeycomb weathering', Geological Society of America Bulletin, 93, 108±115.
Netterberg, F. and Loudon, P. A. 1980. `Simulation of salt damage to roads with laboratory experiments', in Proceedings of 7th
Regional Conference for Africa on Soil Mechanics and Foundation Engineering, Accra, 355±361.
Obika, B., Freer-Hewish, R. J. and Fookes, P. G. 1989. `Soluble salt damage to thin bituminous road and runway surfaces',
Quarterly Journal of Engineering Geology, 22, 59±73.
Prebble, M. M. 1967. `Cavernous weathering in the Taylor Dry Valley, Victoria Land, Antarctica', Nature, 216, 1194±1195.
Price, C. A. 1978. `The use of sodium sulphate crystallization test for determining the weathering resistance of untreated stone',
UNESCO/Rilem International Symposium Deterioration and Protection of Stone Monuments, Reilure, Paris, 3.6, 23 pp.
Price, C. A. 1996. Stone Conservation: an Overview of Current Research, Research in Conservation, The Getty Conservation
Institute, Los Angeles, 73 pp.
Price, C. A. and Brimblecombe, P. 1994. `Preventing salt damage in porous materials', in Roy, A. and Smith, P. (Eds), Preprints
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
C. RODRIGUEZ-NAVARRO AND E. DOEHNE
208
Contribution Ottawa Congress on Preventive Conservation Practice Theory and Research, International Institute of
Conservation, Ottawa, 90±93.
Puehringer, J. and Engstrom, L. 1985. `Unconventional methods for the prevention of salt damage', in Furlan, V. (Ed.),
Proceedings of Vth International Congress on Deterioration and Conservation of Stone, LCP, Laussane, 1, 241±250.
Punuru, A. R., Chowdhury, A. N., Kulshreshtha, N. P. and Gauri, K. L. 1990. `Control of porosity on durability of limestone at
the Great Sphinx, Egypt, Environmental Geology and Water Science, 15, 225±232.
Pye, K. and Sperling, H. B. 1983. `Experimental investigation of silt formation by static breakage processes: the effect of
temperature, moisture and salt on quartz dune sand and granitic regolith', Sedimentology, 30, 49±62.
RILEM 1980. `Recommended tests to measure the deterioration of stones and assess the effectiveness of treatment methods',
Commission 25-PEM: Protection et Erosion des Monuments, 175±253.
Robin, P. Y. F. 1978. `Pressure solution at grain-to-grain contacts', Geochimica et Cosmochimica Acta, 42, 1383±1389.
Rodriguez-Navarro, C. and Doehne, E. 1998. `Time-Lapse video and ESEM microscopy: integrated tools for understanding
processes in-situ', American Laboratory (in press).
Rodiguez-Navarro, C., Doehne, E., Ginell, W. S. and Sebastian, E. 1996. `Salt growth in capillary and porous media', in
Sebastian. E., Valverde, I. and Zezza, U. (Eds), Proceedings of 3rd International Congress on Restoration of Architectural
Heritage and Building, Universidad de Granada, Granada, 509±514.
Rossi-Manaresi, R. and Tucci, A. 1991. `Pore structure and the disruptive or cementing effect of salt crystallization in various
types of stone', Studies in Conservation, 36, 53±58.
Sayward, J. M. 1984. Salt Action on Concrete, US Army Corps of Engineers, Special Report 84±25, Cold Region Research &
Engineering Laboratory, Hanover (NH), 69 pp.
Schaffer, R. J. 1932. The Weathering of Natural Building Stones, DSIR, Building Research Special Report 18, Stationery Office,
London, 34 pp.
Schaffer, R. J. 1955. `Stone in Architecture: I. Stone as a building material', Journal of the Royal Society of Arts, 837±867.
Selby, M. J. 1993. Hillslope Materials and Processes, 2nd edn, Oxford University Press, Oxford, 451 pp.
Smith, B. J. and McGreevy, J. P. 1983. `A simulation study of salt weathering in hot deserts', Geografiska Annaler, 65A, 127±133.
Smith, B. J. and McGreevy, J. P. 1988. `Contour scaling of a sandstone by salt weathering under simulated hot desert conditions',
Earth Surface Processes and Landforms, 13, 697±705.
Smith, B., Whalley, B. and Fassina, V. 1988. `Elusive solution to monumental decay', New Scientist, 118, 49±53.
Sperling, C. B. and Cooke, R. V. 1985. `Laboratory simulation of rock weathering by salt crystallization and hydration processes
in hot, arid environment', Earth Surface Processes and Landforms, 10, 541±555.
Sunagawa, I. 1981. `Characteristics of crystal growth in nature as seen from the morphology of mineral crystals', Bulletin
Mineralogie, 104, 81±87.
Taber, S. 1916. `The growth of crystals under external pressure', American Journal of Science, 41, 532±556.
Tada, R., Maliva, R. and Siever, R. 1987. `A new mechanism, for pressure solution in porous quartzose sandstone', Geochimica
et Cosmochimica Acta, 51, 2295±2301.
Tiller, W. A. 1991. The Science of Crystallization: Microscopic Interfacial Phenomena, Cambridge University press, Cambridge,
391 pp.
Trenhaile, A. A. 1987. The Geomorphology of Rocky Coasts, Clarendon Press, Oxford, 384 pp.
Vos, B. H. 1976. `Water absorption and drying of materials', in Rossi-Manaresi, R. (Ed.), The Conservation of Stone I, Centro
per la Conservazione delle Sculture All'Apperto, Bologna, 679±694.
Walder, J. and Hallet, B. 1985. `A theoretical model of the fracture of rock during freezing', Geological Society of America
Bulletin, 96, 336±346.
Walder, J. and Hallet, B. 1986. `The physical basis of frost weathering: toward a more fundamental and unified perspective',
Arctic and Alpine Research, 18, 27±32.
Washburn, E. R. 1927. `The creeping of solutions', Journal of Physical Chemistry, 31, 1246±1248.
Wellman, H. W. and Wilson, A. T. 1965. `Salt weathering, a neglected geological erosive agent in coastal and arid environments',
Nature, 205, 1097±1098.
Wellman, H. W. and Wilson, A. T. 1968. `Salt weathering or fretting', in Fairbridge, R. W. (Ed.), The Encyclopedia of
Geomorphology, Stroudsburg, Pennsylvania, 968±970.
Weyl, P. K. 1959. `Pressure solution and the force of crystallization ± A phenomenological theory', Journal of Geophysical
Research, 64, 2001±2025.
Winkler, E. M. 1965. `Weathering rates as exemplified by Cleopatra's Needle in New York City', Journal of Geological
Education, 13, 50±52.
Winkler, E. M. 1970. `Salt action on stone in urban buildings', in Application of Science in the Examination of Work of Art,
Museum of Fine Arts, Boston, 139±146.
Winkler, E. M. 1973. Stone: Properties, Durability in Man's Environment, Springer-Verlag, Berlin, 250 pp.
Winkler, E. M. 1994. Stone in Architecture, Springer-Verlag, Berlin, 313 pp.
Winkler, E. M. and Singer, P. C. 1972. `Crystallization pressure of salt in stone and concrete', Geological Society of America
Bulletin, 83, 3509±3514.
Yatsu, E. 1988. The Nature of Weathering: an Introduction, Sozosha, Tokyo, 624 pp.
Young, A. R. M. 1987. `Salt as an agent in the development of cavernous weathering', Geology, 15. 962±966.
Zehnder, K. and Arnold, A. 1989. `Crystal growth in salt efflorescence', Journal of Crystal Growth, 97, 513±521.
SALT WEATHERING
Copyright #1999 John Wiley & Sons, Ltd. Earth Surf. Process. Landforms 24, 191±209 (1999)
209
