equilibrium conditions of marine originated salt mixtures: an ecos application at the archaeological...

SWBSS, Copenhagen 2008

1

Equilibrium conditions of marine

originated salt mixtures: An ECOS

application at the archaeological site of

Delos, Greece

Petros Prokos

Hellenic Ministry of Culture, 26th Ephorate, Piraeus, Greece.

Abstract

Sea spray comprises a multi salt solution. Despite the quantitative dominance of

halite, monuments in the coastal zone contain various marine originated mixtures due

to fractionated infiltration. The presence of other species with varied solubility

influences considerably the phase transition equilibrium of halite, which should not be

considered descriptive of the weathering environment. The thermodynamic

assessment of salt mixtures has been suggested as a diagnostic tool against salt

weathering. In order to understand the mechanism that triggers salt weathering at the

archaeological site of Delos this new methodology was applied, based on quantitative

analysis and phase transitions assessment with the use of ECOS software. The results

suggest that the presence of minor species in sea brines plays an important role in the

generation of damage by shifting the equilibria to a wider range on environmental

conditions.

Keywords

Salt weathering, salt mixtures, Delos, ECOS, marine aerosols, equilibrium conditions

1. Introduction

The presence of marine aerosols constitutes a major risk for the preservation of

architectural heritage along coasts. Salt solutions of marine origin, in the form of spray

produced on the surface of the sea, are transferred by the wind for long distances


2

inland and deposited by inertial impaction on vertical surfaces. This process

continuously supplies salts to the masonry, resulting in severe weathering at a fast

rate. Halite is the dominant precipitant of sea spray and the most abundant

efflorescence on coastal monuments. In addition, sea salt contains other species as

well in smaller concentrations. Although the thermodynamic behaviour of halite does

not significantly alter in the presence of these minor species, weathering simulations

have shown that sea spray is more destructive than halite alone [Rivas et al, 2000].

However, the fractionated infiltration of sea spray results in varied concentration

gradients, and that might be responsible for damage. Recent developments in salt

weathering research have proposed the investigation of equilibrium conditions of salt

mixtures as means of defining the optimal conditions for preservation. The present

work investigates the equilibrium conditions of marine originated salt mixtures at the

archaeological site of Delos and relevant for the preservation of the Hellenistic wall

paintings.

2. Methodology

Phase transitions occur at a specific relative humidity for each salt, at a given

temperature, called equilibrium relative humidity (RHeq). Single salts, however, are

rarely found in nature. In practice, buildings are contaminated by salt mixtures, which

present a totally different behaviour. As a matter of fact, salt mixtures do not present

individual RHeq but rather a range of relative humidity within which progressive and

multiple crystals growth will occur. In the context of salt weathering, the interaction of

salts has been studied by Price and Brimblecombe [1994] and Steiger and Zeunert

[1996], using the approach of Pitzer [1973], which triggered the development of the

computer software ECOS (Environmental Control of Salts) that predicts crystal

volumes in a given system under given environmental conditions [Price, 2000].

The present research aims to diagnose the weathering environment, which triggers the

action of salts resulting in the deterioration of wall paintings at Delos. Taking

advantage of the above-mentioned recent developments concerning the interaction of

salts, the equilibrium conditions that influence a generation of damage were assessed

in situ. The investigation followed a comparative approach guided by spatial variables.

Two monuments have been selected for the purpose of this investigation, the House of

Hermes (areas HD, HG) and the House of Masks (areas MD, MA). The sampling sites

were selected according to their orientation and the degree of exposure in relation to

sheltered and exposed areas. The samples were extracted from the wall paintings in

the form of micro-drills in depth sequence. The mortar samples were weighed and

diluted in 20ml of distilled water. The aqueous extraction was then analysed by Ion

Chromatography(IC) for the anions and Inductively Coupled Plasma-Atomic Emission

Spectroscopy (ICP-AES) for the cations at the laboratories of the Geology

Department, Royal Holloway College, University of London. The raw data was

calculated in molar/1000 (mM), which was used as input for ECOS programme.

Sampling was repeated four times in order to follow the annual climatic cycle. The

thermodynamic assessment was supported by aerosols sampling and analysis,

environmental monitoring, crystallographic analysis and microscopy of salt crusts and

ex situ experiments [Prokos, 2005].


3

3. Results and discussion

3.1 Overall results

Both the sheltered and the exposed areas presented similarities, at least in terms of

enrichment trends and concentrations. Sea spray appeared to be dominant in the

exposed areas but area HD facing the main wind direction and receiving the highest

amount of aerosols presented higher concentrations during the dry period. The

sheltered areas and the exposed area facing the opposite direction received almost

1/10 of the exposed areas’ aerosols deposition. The height distribution was

homogenous in the exposed areas, quantitatively and qualitatively. The sheltered areas

presented higher concentrations, and halite was dominant only in the upper zone. The

height distribution in the sheltered areas revealed contamination by ground moisture

capillary rise. Qualitatively, the salt distribution followed the solubility model [Arnold

and Zehnder, 1991], while quantitatively, the concentrations decreased with height

during the rainy season and increased during the dry season. The areas HD (exposed)

and MD (sheltered) will be presented in detail. Due to technical difficulties gypsum

was not included in the diagrams.

3.2 Area MD

In the sampling campaign of November, the lower zone in MD (fig. 1a) was

quantitatively dominated by gypsum. By extracting 1.1 mol of gypsum the significant

surplus of sulphate results in hydrated salts of Na, Mg and K. According to ECOS,

mirabilite is the first salt that precipitates from the mixture at 87.8%RH, significantly

lower than its RHeq as a single salt (95%). The next step at 71.7%RH indicates the

major equilibrium point of the system where four species precipitate simultaneously

and quickly reach their highest volumes. Mirabilite dehydrates to thernadite while

simultaneously bonding with magnesium and potassium sulphates to form bloedite

and aphthitalite respectively. Chloride is entirely consumed in halite, which is also

slightly delayed in relation to its RHeq as single salt. Nitrate bonds as NaNO3 with

NaSO4 to form darapskite at 68.5%RH and as nitre at 40.9%. The latter represents the

last step of the sequence.

The mid zone presents a slightly different mixture. The initial calcium extract and

gypsum are both decreased to less than half the quantity of the lower zone. The

equilibrium of mirabilite is thus shifted slightly lower to 86.5%RH and the next step

occurs exactly as presented above at 71%RH. The most significant change is the shift

of nitre towards higher values (63%) closer to its RHeq. Gypsum is sparingly present in

the upper zone. The dominance of chloride primarily as halite and secondly as sylvite

does not permit the precipitation of mirabilite or thernadite (see Figure 1b).

Alternatively, sulphate is altered to aphthitalite, which precipitates along with halite at

72%RH, bloedite, which is slightly delayed, and picromerite. Sylvite equilibrium is

significantly decreased (46.4%RH). The consumption of sodium as halite also permits

the presence of Mg-SO4 – independently from bloedite – as hexahydrite (62%),

starkeyite (58%) and kieserite (22.4%).


4

The height distribution and the potential of the salt mixtures present a very interesting

fractionation, which probably relates to the presence of ground moisture. As the

moisture content assumingly decreases with height the composition of the salt

mixtures leads to more soluble species. The presence of moisture at the lower and mid

zones permits the precipitation of mirabilite, which is absent in the higher zone.

Gypsum also decreases with height, eventually permitting the thermodynamic

dominance of halite at the upper zone.

Figure 1. ECOS diagrams. November, area MD, lower (a) and upper (b) zones

Autumn is characterised in Delos by a long period of transition from the dry and hot

summer to the mild winter. The sampling campaign was carried out near the mean

annual values of atmospheric temperature and relative humidity and at the starting

point of the rainy season. The fluctuation of RH over the period of sampling indicated

diurnal phase transitions of mirabilite while the other salts probably stayed in solution.

The following period of very high relative humidity, in which condensation was

reached during December and January, might have caused deliquescence of gypsum as

well. After a short period of frost, the temperature returned to the values of November

and relative humidity was slightly decreased. From the end of January until the next

sampling period of March, the RH rarely surpassed the RHeq of mirabilite but dropped

frequently below the RHeq of halite at diurnal rate. On the other hand, several rain

events occurred in Delos at the time of sampling. We can assume that mirabilite is

more likely to crystallise at the mid zone than at the lower one. The presence of

ground moisture probably delays the precipitation of mirabilite. Gypsum should

precipitate as well at the mid zone. It is not clear whether the rapid drops of RH

permit the crystallisation of halite. It is very interesting to notice that the less soluble

species of the higher zone are kept in solution under conditions where they should

crystallise as single salts. We can assume that this is favourable for preservation. As

the average RH decreases towards the summer the diurnal fluctuations reach more

frequently the RHeq of the salts found in the higher zone.

After a long period of warm and dry weather, which started spontaneously around

May, the sampling campaign of August revealed a totally different thermodynamic

potential (see Figure 2a). Gypsum significantly decreased in the lower zone (0.1 mol)

while thenardite was totally absent. Chlorides corresponded primarily to halite,

sulphates to starkeyite and nitrates to nitre. However, the presence of the very soluble

carnallite and nitromagnesite shifted the equilibrium towards much lower values. Thus

halite precipitated at 59.2%RH while – due to the higher temperature and the absence


5

of ground moisture – starkeyite precipitated immediately at the same value without the

prior presence of epsomite. Nitre precipitated slightly later at 55%RH. The next

significant step at 30.4%RH starkeyite was dehydrated to kieserite while part of nitre

diluted and recrystallised as carnallite and nitromagnesite. Similarly, the mid zone

presented a different potential dominated by more soluble salts. Contrary to the lower

zone, gypsum maintained the high concentration of November. Halite crystallised at

61.5%RH due to the presence of the very soluble carnallite, bischofite, nitromagnesite

and anhydrous nitrocalcite. In the upper zone, the dominant halite maintained its RHeq

close to the normal values (71.4%) despite the presence of the same very soluble salts

(see Figure 2b).

Figure 2. ECOS diagrams. August, area MD, lower (a) and upper (b) zones

Obviously the withdrawal of ground moisture governed the alterations observed in

MD from the previous campaign. Mirabilite was probably withdrawn from the

mixture as efflorescence. Gypsum precipitated in the mid zone while it was dissolved

by ground moisture and probably diffused back to the foundations in the lower zone.

This can be supported by the constant presence of a gypsum crust that was observed in

the mid zone and the periodic efflorescence in the lower zone. As expected, the

ground moisture carried sulphates that crystallised slowly in the mid part while

simultaneously caused phase transitions of less soluble salts in the lower zones. The

presence of moisture in the deeper parts was probably responsible for the sand

disintegration in the lower zone. The external efflorescence of gypsum in the mid zone

during the warm months probably prevented damage from occurring. On the other

hand, deliquescence played the primary role in the upper zone where halite was drawn

slowly towards the surface and crystallised as efflorescence. Furthermore, the diurnal

fluctuations of RH during the warm months probably caused transitions of halite,

which might have been responsible for the exfoliation of the external layer. The low

supply of moisture was presumably the cause of the whisker like appearance of the

halite efflorescence (see Figire 5a). The rest of the species in the lower and the mid

zone more likely stayed in solution since their extremely low RHeq was not reached in

the sheltered areas. The evident marine origin of the species in the upper zone

indicated that sea spray dominated as contaminant in the absence of ground moisture.

The next sampling campaign of January confirmed the ground moisture origin of

sulphates. In the lower zone (see Figure 3a), where damage was probably deactivated

during summer, the rising ground moisture offered the appropriate conditions for the

generation of hydrated salts. It is quite clear that the thermodynamic potential of


6

August, which permitted the crystallisation of starkeyite, had altered into a more

aggressive form, favouring the production of epsomite and hexahydrite as well. Thus,

as predicted by ECOS, the first salt that precipitated was epsomite at 89.4%RH,

followed by hexahydrite at 84.5%RH. At lower values, the dehydration of hexahydrite

should potentially occur at 52% (starkeyite) and 20% (kieserite). Halite found in very

small concentration was kept at very low RHeq (58%) as are the rest of the very

soluble salts such as carnallite and bischofite. Gypsum was again present at almost

double the proportion than August, as in November. The most interesting observation

was that the species found in August in the lower parts were transported to the mid

zone, altering again the thermodynamic potential. The slower supply of moisture,

howeve,r does not favour the generation of epsomite, as in the lower zone. Thus, the

mixture crystallised initially as hexahydrite (71%) followed by halite (67%). The

lower concentration of magnesium and sulphates not only restricted the generation of

epsomite, but also significantly lowers the RHeq of hexahydrite. The less soluble salts

again dominated the potential of the mid zone. Alternatively, the higher zone, which

as presumed so far is mainly influenced by atmospheric RH deliquescence, did not

present any particular alteration.

Figure 3. ECOS diagrams. January, area MD(a) and August, area HD(b), lower zones

The fluctuations of RH during this period favoured phase transitions only at the lower

part, and specifically between epsomite and hexahydrite. The mid zone again

represented the height of gypsum accumulation, with the rest of the species probably

in solution. At the upper part, halite was progressively dissolved and withdrawn to the

deeper parts. It is very interesting to observe that the exfoliation of the upper zone was

observed particularly during winter. We can assume that the withdrawal of halite

releases the white flakes that were generated by diurnal fluctuations during summer.

3.3 Area HG

The thermodynamic potential in room HD was in most cases dominated by halite.

This finding is in agreement with the aspect and exposure variables of this site. The

deposition investigation indicated that HD is more subjected to sea spray than the

other locations. ECOS results for the lower zone in November indicated the dominant

presence of halite along with much smaller quantities of gypsum and other sulphates

and nitrates. The precipitation of mirabilite was not favoured and thus thernadite

precipitated simultaneously with halite and bloedite at 71.4%RH. Similarly, in the


7

upper zone, halite precipitated at 71.7%RH, followed by thernadite at 69.4%RH. The

dominance of halite and the similarity of the potential in both zones indicated that

direct deposition of sea spray prevails as source and pathway of contamination.

During the sampling campaign of March the composition of the minor species

changed significantly, although halite was still dominant. Sulphates were withdrawn

from the system, replaced mainly by chlorides. As a result the RHeq of halite was

slightly reduced to 69.7%RH in both zones. The most sensible explanation for the

withdrawal of the sulphates is that they precipitated as efflorescence due to

fractionated crystallization and were washed off by rain. This explanation is also

justified by the depth distribution according to which the less soluble sulphates stay at

the surface while the more soluble chlorides advance inwards.

During the next campaign of August sulphates were again present, but only in the

lower zone (see Figure 3b). Halite was still dominant but precipitated at a lower RHeq

(67.8%) due to the presence of the very soluble bischofite and carnallite. In the mid

zone halite precipitated at 68.1% deviating as well from its RHeq as a single salt (see

Figure 4a). The most significant deviation though can be observed at the upper zone

where the RHeq of halite dropped to 55% (see Figure 4b). The same phenomenon was

observed at the upper zone of the other exposed room MA. Probably during the warm

period halite crystallised as efflorescence at the upper part as a result of the intense

solar radiation and the absence of ground moisture. According to the climatic data the

salt mixtures of the lower and the mid zone must have stayed in solution as well.

Figure 4. ECOS diagrams. August, area HD, mid (a) and lower (b) zones

During January, halite’s RHeq drops even lower than that predicted for August

(61.2%). It is obvious that the proportion between halite and the more soluble species

had changed, while the sulphates withdrew from the system. Although the

environmental conditions did not predict precipitation for any salt in the lower zone in

August, we can assume that due to the intense solar radiation the less soluble

sulphates along with some halite might have crystallised on the surface. On the other

hand, the continuous supply of sea spray might have altered the potential that was

predicted for August, permitting the precipitation of the less soluble species. The

withdrawal of the salts that presumably precipitated could have been achieved by rain-

off since the rainy period had already started during the sampling campaign of

January.


8

A slightly similar case is predicted by ECOS for the mid zone. Halite is significantly

reduced while bischofite, present in the form of tachydrite in the previous campaign,

has retained its volume. As a result halite deviates to 71.4%RH. A similar assumption

of halite precipitation and rain-off is plausible in this case although it contradicts the

environmental conditions, according to which halite should stay in solution. Another

fact, which indicates the influence of rain, is the reduction of gypsum during January

in the upper and the mid zone. The upper zone presents a considerably altered

resultant mainly characterised by sulphate and nitrate enrichment. Thus halite’s RHeq

is still kept very low while niter and hexahydrite precipitate at 69.7% and 64.7%

respectively.

In general, ECOS predictions for HD presented some contradictions between the

potential of the mixtures and the climatic conditions. According to these predictions

all species should stay in solution throughout the sampling period. From one of view,

this presumption is in accordance with the good preservation state of the wall

paintings. On the other hand, in order to explain the fluctuations of the species

concentrations it was assumed that – at least during the warm period – the less soluble

salts should crystallise.

Figure 5. SEM images. Area MD upper zone, whisker-like halite efflorescence (a/200µm)

and area HD, halite and gypsum efflorescence (b/50µm)

The depth variable revealed a constant presence of halite and gypsum at the

interface between the external lime wash layer and the mortar in areas receiving direct

sea spray, which was also testified by microscopic observations (see Figure 5b).

Although the results were confirmed by laboratory simulations (Prokos 2005), the

conditions of damage could not be defined precisely. Another experiment conducted

by Environmental Scanning Electron Microscopy revealed a very unstable behaviour

of the mixture at very low RH conditions during the hydration of the anhydrite. It was

thus presumed that there was an influence of a kinetic factor, which according to

hypothetical ECOS calculations should supply higher temperatures. A logical

explanation is the direct solar radiation, which, as recorded by pyranometer during the

warm period in Delos, can supply energy of 900w/m2. The sharp rise of temperature

on the rendering surface generates fast evaporation and high supersaturations. The


9

crystallisation site lies at the interface of the wall painting’s layers due to hydraulic

discontinuity. It is presumed that these conditions are met in Delos.

Since the ECOS results in relation to the climatic data did not clearly indicate when

phase transitions were triggered, it is very difficult to define the rate of this

mechanism. The only indication of crystallisation was the reduction of salts during the

warm period. The cold season represented a period of slow salt accumulation and

transport to the interface, disturbed by rain-wash. After the rainy season, sea spray

deposits on the surface were undisturbed, allowing the slow formation of a crust.

During this period, the superficial deposits supplied the deeper parts with solution.

According to the data, RH was still fluctuating higher than the RHeq of the mixtures,

building significant concentrations on the interface due to hydraulic discontinuity. The

transition from the winter to summer is very fast in the Aegean. Presumably, the fast

evaporation enhanced by the solar radiation leaves solution islands at the interface that

cannot follow the withdrawal of moisture to the surface. As described by Scherer

[2004] for a random case of variable pore size distribution, the liquid pockets trapped

between small radius pores, will eventually dry up when the local RH becomes low

enough for the menisci to pass through the pore entries. In the case of layered

structures we suggest that this mechanism is localised on the interface. The isolated

liquid pockets will eventually generate localised damage on the interface, separating

the layers. Progressively, this mechanism will produce exfoliation of the external

layer, which carries the primary information.

4. Conclusions

According to the previous discussion we can draw a number of conclusions for the

preservation of the investigated monuments in Delos. Primarily, it is evident that

preservation is impossible in outdoor conditions. Sea spray can cause severe damage

to paint layers, especially in combination with direct solar radiation. Area HD, which

receives the larger amount of sea spray presented a critical relative humidity near the

RHeq of halite. Damage, probably in diurnal rate, more likely occurs during the

warmer months under the influence of direct solar radiation. Sheltering in this case

will definitely provide both protection against sea spray deposition and solar radiation.

Considering the state of the adjacent sheltered room special care should be given to

avoid the undesirable effects. The absence of wind and direct solar radiation enhance

ground moisture capillary rise. Damage in the sheltered areas is a result of the

particular microenvironment that has been created under the shelters in response to the

particular mixtures that evolved due to fractionated infiltration. The variability of the

salt content suggests that the environmental control of salt damage in real conditions

of various contamination pathways and sources cannot be restricted to a single optimal

range. The present investigation underlined the role of minor species in the generation

of damage. The thermodynamic assessment is an essential diagnostic tool especially in

buildings contaminated by marine aerosols, which – despite the quantitative

dominance of halite – contain complex salt mixtures. The produced damage model

can be used for various conservation treatments from environmental control to

desalination and drainage.


10

Acknowledgments

This project has been carried out at the Institute of Archaeology, University College

London under the supervision of Professor Clifford Price (UCL) and funded by the

Greek State Scholarships Foundation, the A.G. Leventis Foundation and The Natural

Environment Research Council (UK). We would like to thank the 21st Ephorate of

Antiquities (Cyclades) and the personnel of Delos site, Dr Panagiotis Hatzidakis, head

of Delos excavations (Ministry of Culture) and Mr Nikos Minos, head of the

Conservation Directorate (Ministry of Culture).

References

Arnold, A., Zehnder, K. 1991. Monitoring wall paintings affected by soluble salts. The

Conservation of Wall Paintings, The Getty Conservation Institute, Los Angeles, 103-

136.

Pitzer, K.S. 1973. Thermodynamics of electrolytes: 1.Theoritical basis and general

equations. Physical Chemistry, 77, 268-277.

Price, C.A. 2000. Salt damage in porous materials. An expert model for determining

the environmental conditions needed to prevent salt damage in porous materials,

Archetype publications ltd, London, 3-11.

Price, C.A., Brimblecombe P. 1994. Preventing salt damage in porous materials.

Preventive Conservation: Practice, Theory and Research, International Institute for

Conservation, London, 80-93.

Prokos, P. 2005. Salt weathering in the coastal environment: The deterioration of wall

paintings at Delos, Greece. Ph.D. dissertation, Institute of Archaeology, University

College London, United Kingdom.

Rivas, T., Prieto, B., Silva, B., Birginie, J.M. 2000. Comparison between traditional

and chamber accelerated ageing tests on granitic rocks. Proc. 9th Symposium for the

Conservation of Stone, Venice, 171-180.

Scherer, G.W. 2004. Stress from crystallization of salt. Cement and Concrete

Research, 34[9], 1613-1624.

Steiger, M., Zeunert, A. 1996. Crystallisation properties of salt mixtures: Comparison

of experimental results and model calculations. Proc. 8th International Congress on

Deterioration and Conservation of Stone, Berlin, 535-544.

equilibrium conditions of marine originated salt mixtures: an ecos application at the archaeological...

Documents