Introduction

This research is concerned with the study of the ongoing decay and degradation of the Byzantine Wall, which surrounds what was once Thevest, modern-day Tebessa, which is situated in the northern Aures Mountains in northeastern Algeria (Fig. 1a). Many weathering forms are clearly visible in the studied structure, ranging from slight discolouration up to the loss of material and granular disaggregation. At present, no deep investigation has been carried out to study the weathering damage of the Byzantine Wall. In this work, our goal is to characterise the extent to which differences in the properties of limestone and carbonate tufa lead to differences in weathering characteristics, including the identification of the main deteriorating agents. To demonstrate the effects of complex features such as bioturbation, fissures, micro- and macroporosity on stone durability, physical, morphological properties and salt crystallisation tests have been carried out.

Fig. 1
figure 1

Location of Tebessa in northeastern Algeria (a) , schematic map of the Byzantine Wall including the location of the study area into the north facade (b), the north part of the Byzantine Wall (c) , and a statuette of Septimus Emperor Wife’s in the Caracalla Gate (d)

Full size image

Since ancient times, monuments have been widely constructed with carbonate rocks, which are the most used rocks for construction in the world due to their high availability (Goudie and Viles 1997; May 1997; Török 2006; McCabe et al. 2007; Cassar et al. 2011; Espinosa-Marzal and Scherer 2010; Smith et al. 2010; Garcia-Guinea et al. 2013; La Russa et al. 2013; Ruffolo et al. 2015a, b). During the Roman occupation of North Africa, many monuments, such as the Byzantine Wall, the Basilica, the Temple of Minerve and many more in Tebessa, were constructed with carbonate rocks. Like all rocks, the stones of monument walls have been subjected to weathering by the atmosphere and hydrosphere since their exhumation.

In the context of construction material weathering and decay in historical monuments, a great deal of research has been undertaken to identify and understand the agents and mechanisms of degradation to devise better practices for eventual restoration procedures. In many instances, when restoration works have been undertaken without a prior understanding of the degradation process, the restoration has aggravated the decay (Přikryl et al. 2010; Boumezbeur et al. 2015). It is thus worth conducting a deeper investigation into the nature of weathering and the agents causing it. Several weathering forms, such as flaking, efflorescence, exfoliation and granular disaggregation, demnonstrate the state of stone weathering and deterioration in built environments (Schaffer 1932; Price 1975, 2006, 2007; Winkler 1997; May 1997, 1998a; Doehne and Price 2010; Ruffolo et al. 2013, 2015a, 2015b; Emmanuel and Levenson 2014; Boumezbeur et al. 2015). In fact, the petrophysical properties of rocks are determined by their mineralogy and structure (Ruedrich et al. 2011); in turn, these complex properties are reflected in the stone weathering responses to the natural and built environments (May 1994, 1997, 1998b; Kramar et al. 2010).

Salt damage is considered a common hazard, playing a major role in the decay of stones under a wide range of environmental conditions (Goudie and Viles 1997; Doehne 2002; Gôtze and Siedel 2004), including the semi-arid region of NE Algeria where the Byzantine Wall is located. The durability of carbonate rocks is controlled mainly by soluble salt (Ruffolo et al. 2015b; Al Omari et al. 2016). Salt damage is controlled by multiple stone properties, and their actions occur through physicochemical reactions (Charola 2000). Thus, pore size distribution can strongly control the transport of liquids, which influences the stability of stones through the growth of their crystals and the pressure generated on the porous networks in the walls (Nicholson 2001; Benavente et al. 2007b; Yu and Oguchi 2010; Koniorczyk and Gawin 2012; Ruffolo et al. 2015b). According to Lόpez-Doncel et al. (2016), pore size distribution controls the effects of salt damage in stones, and stone durability is mainly related to the proportion of micropores. However, the petrophysical properties of rocks change with the presence of fissures, which give rocks a double porosity response (Benavente et al. 2007b). In particular, the action of salt crystallisation in bioturbated zones in stone causes the development of alveoli close to the stone’s surface (Cassar 2002). Limestone and carbonate tufa in the Byzantine Wall consist mainly of calcite, but they present contrasting weathering behaviours in response to the significant variations in their structure and texture (Kramar et al. 2010).

Materials and methods

Study site and stone materials

In 146 BCE, Tebessa became part of the Roman Empire with the name Thevest. Since that time, several buildings have been erected, such as the amphitheatre, the Temple of Minerve, the Arch of Caracalla, the Basilica and the Roman Wall. The latter was destroyed and rebuilt in the sixth century by the Byzantine General Solomon for security and political purposes (Castel 1912), and, as a result, the Wall is now called the Byzantine Wall. It is rectangular in shape; and spans 298 m in length, 260 m in width, and is 7 m high. The Byzantine Wall, which surrounds the city, contains four gates and fourteen square towers for surveillance (Fig. 1b, c). The main gate, the Caracalla Gate, was erected in 214 AD, faces north (Fig. 1d) and since 1982 has been classified as one of the cultural heritage treasures of the world. The Byzantine Wall was built entirely of carbonate stones and exhibits severe decay in its northern and western facades.

Carbonate sedimentary rocks are widespread in the Tebessa area, with the oldest outcrops being Albian, while the most recent outcrops are Quaternary (Durozoy 1953) (Fig. 2a). For rock monuments, the most used stratigraphical units are those of the middle Turonian for the limestone and the actual formation for the carbonate tufa (Coquand 1862). Limestone and tufa are the main rock types that were used as building materials for the Byzantine Wall. Hence, these materials are selected in this study due to their prevalence in the Wall and due to their wide range of weathering intensities. The limestone is pinkish in colour, middle Turonian in age and contains intensive bioturbations that are highly stained by iron oxides. The iron surfaces often show a few dolomite crystals. The petrographical analysis of the stones reveals that the main mineral component is calcite, which is accompanied by bioclasts and small percentages of quartz, dolomite and goethite (Fig. 2e). The observed bioclasts are the debris of lamellibranches and the spicules of echinoids, millioles, ostracods and gastropods. Recrystallised calcite grains range in size from 80 to 200 μm and sometimes reach 750 μm. Automorphic dolomite crystals have a diameter of 250–400 μm with a clear growth zone impregnated by iron oxides. These crystals are clearly seen as red patches in the bioturbation structures. This limestone presents vuggy and fissural porosity. According to Allag (2016), clay minerals such as montmorillonite and kaolinite can be observed in the Turonian limestone.

Fig. 2
figure 2

Geological map of Tebessa (Durozoy 1953) (a) and the view of the tufa quarry in Djebel Djoua (b) and the limestone quarry in Djebel Doukkane (c). Optical microphotographs of carbonate tufa (d) and limestone (e) textures

Full size image

Carbonate tufa is another stone that is used in the Wall and is extracted from nearby quarries close to the monuments. The tufa is mainly composed of calcite and is deposited by springs, which were common near the city of Thevest (Durozoy 1953). Tufa is a very porous, soft and spongy rock that often shows plant moulds (Fig. 2d). Tufa can be easily cut and reworked into cubic blocs. Generally, tufa is commonly formed by the precipitation of calcite from normal freshwater in lakes and shallow ponds. This carbonate tufa has an arborescent texture and a complex structure.

Methods

Fresh stones (quarries)

Fresh stones were sampled from the ancient Roman quarries assumed to be quarried for the construction of the Byzantine Wall and other monuments (Coquand 1862). Carbonate tufa was extracted from the quarry of Djebel Djoua (Fig. 2b), which is located approximately 1 km east of the Byzantine Wall, while fresh limestone was extracted from an ancient quarry in Djebel Doukkane (Fig. 2c), which is located approximately 6 km WSW.

Petrophysical and durability characterisations were carried out on representative fresh limestone and carbonate tufa samples. This characterisation is essential to understand the stone behaviours under different factors and their properties, which play a great role during the weathering process. Pore sizes were measured on unweathered samples to determine the petrophysical properties of the studied rocks. These properties were open porosity, capillary imbibition and P wave velocity. Porosity and ultrasonic wave velocity were measured a second time after durability tests to evaluate the changes in the rock properties.

Total porosity, ϕT, defined as the ratio of the total volume of pore space to the bulk material volume, is calculated using the relationship between the bulk and grain densities. Bulk density, ρbulk, was determined through direct measurements of the dried weights and dimensions of samples. The grain, or real density, ρ, of a material is defined as the ratio of its mass to its solid volume. Both stones are composed of calcite (more than >98%); therefore, the grain density was considered to be the density of calcite (2.71 g/cm3). Open porosity, ϕO, was obtained using the vacuum water saturation test (UNE-EN 1936 2007). The connected porosity, ϕc, and pore size distribution were calculated via the mercury intrusion porosimetry (MIP) technique. A PoreMaster 60 GT (Quantachrome Instruments) mercury porosimeter was used, and a surface tension and contact angle of mercury of 480 mN/m and 130°, respectively, were chosen. The pore radius interval ranges from 0.002 to 200 μm, which corresponds to the highest and lowest head pressures, and, consequently, MIP cannot characterise large vug pores. The samples used are prismatic in shape and have a size of 0.216 cm3.

Capillary absorption coefficients, C, were determined using the standard method for limestone in accordance with UNE-EN-1925 ( 1999). The continuous method was used to characterise tufa varieties because of their high absorption rates (C > 150 g/m2 s0.5) (Benavente et al. 2007a). To analyse the capillary water absorption, six cubic samples of 5 cm each were cut for every stone type. These specimens were soaked in approximately 0.2 cm of water. For data acquisition, weight measurements were taken every 30 s for the carbonate tufa until sample saturation was achieved (approximately 10 min). For the limestone, weight measurements were taken every 3 min for the first hour, then every 1 h for 8 h, and then 1 measurement was taken every day up to the 7th day. The capillary absorption coefficient was calculated from the weight increase per unit area versus the square root of time. The results were plotted as the mass of water absorbed per area of sample during imbibition versus the square root of time.

P wave velocities, vP, were measured using signal emitting–receiving equipment (NDT 5058PR; Panametrics,) and an oscilloscope (TDS 3012B; Tektronix). The uniaxial compressive strength, σC, was evaluated according to the standard test EN-1926 (2007) using a servo-controlled compression machine with a loading capacity of 200 kN.

The mineral compositions of both stones (limestone and tufa) in their fresh states were determined by X-ray diffraction (XRD) on a powder of a crushed sample (63 μm size) using a Philips PW 1840 diffractometer, and the patterns were run with Ni-filtered Cu Kα radiation (λ = 1.54056 Å) at 40kv and 10 mA. The scanning range was from 2θ = 1 to 80°. To characterise the microtextural properties of the stones, the limestone and tufa were examined with a scanning electron microscope (SEM), and SEM microprobe analyses were performed by an SEM JEOL JSM 6400 coupled with an energy dispersive X-ray spectrometer (EDX) system, detector model 6587, to identify the textural and mineralogical composition of rocks. The samples were coated with gold under a 10-kV voltage.

Durability tests

The salt crystallisation test was carried out in accordance with EN-12370 (1999) recommendations. Cubic samples of the 4-cm side were cut from the studied stone types and soaked in a saline solution of 14% w/w Na2SO4 at 20 °C for 4 h, followed by a drying period at 105 °C for 16 h and cooling at 20 °C for 4 h. This test was repeated for 15 cycles. The evolution of the open porosity, bulk density, P wave velocities and the dry weight loss, ΔW (%), were calculated at the 15th cycle after cleaning and oven-drying the samples. After each cycle, the weight of each sample was determined to an accuracy of ±0.01 g. The specimens were visually inspected to quantify the salt action.

Weathering characterisation

Weathering occurs because of the interactions of environmental factors with stone materials and consequently brings mineralogical, textural and morphological changes to the surface of the rock. The superficial alteration of the stone presents different weathering forms (Heinrichs 2008; Siedel et al. 2011). These weathering forms have been observed at the mesoscale (cm to metre), with different lengths and widths. To determine the weathering forms on the Byzantine Wall facades, the mapping techniques of Fitzner et al. (1995) and Siedel et al. (2011) were used, in which weathering damage is mapped and classified into different categories, where each shows the dominant weathering form, the intensity and the state of the monument preservation.

The study area is situated along the north part of the Byzantine Wall where the Wall face is clearly visible and accessible. This method was carried out on an area of approximately 20 m2 of the Wall.

The Schmidt hammer test was carried out on the surface of the weathered and unweathered stones to establish the hardness values of their surfaces according to ISRM (1978) and ASTM (2001) recommendations. A Schmidt hammer provides an inexpensive and fast measure of the surface strength of the rocks, and is in fact a measure of the surface hardness of building stones (Queisser 1986, Schneider et al. 2008, Nazir et al. 2013). These values are widely used to estimate the effect of weathering on the mechanical properties of rock material (Schneider et al. 2008; Aydin 2009).

Weathered samples were extracted from the north and west facades of the Byzantine Wall, where the main weathering forms are the most representative. These samples were characterised by XRD and electron microscopy.

Results and discussion

Mineralogical and textural characterisation

The results of the XRD analysis of limestone and tufa show that the main mineral constituent is calcite, whereas the secondary minerals in this limestone are quartz, dolomite and goethite.

On the limestone, a brownish colour and fissures are widely exhibited on bioturbation surfaces. Inside the bioturbation structures, the brownish colour is the result of iron oxide (goethite), which was detected by the XRD analysis and supported by SEM observations (Fig. 3a). Under SEM, the fissures are of different lengths and widths without preferential orientation and are generally less than 200 μm in width. In this case, these fissures are assumed to be original. The fissure surfaces show the presence of clay minerals between the grains, which was detected by SEM (Fig. 3a) and confirmed by EDX microanalysis.

Fig. 3
figure 3

Scanning electron microscope images (SEM) and pore size distribution of the investigated samples. The limestone structure shows: fissures, porosity in fissures, clays and iron oxides (a); the tufa structure shows porosities and their interconnection (b). Cumulative mercury intrusion curves and pore size distribution curves (MIP) for limestone (c) and carbonate tufa (d)

Full size image

Under SEM observations, the porosity is more important in bioturbation structures than in other parts of the stone, and the porosity is below 200 μm (Fig. 3a); however, surfaces without bioturbation or fissures have pore radii of less than 10 μm. Intercrystalline porosity can be seen in this limestone and is related to the fissure surfaces and bioturbation. In particular, clay with a low thickness can play a role in cementing grains together and reducing the porosity. Additionally, in very porous media, clay deposited in intercrystalline spaces stops the circulation of water (Siegesmund and Török 2011). The arrangement and the spatial distribution of the fissures and bioturbations within the limestone seem to play an important role in the control of petrophysical and mineralogical properties. The presence of bioturbation, fissures and clay minerals inside the fissures plays a great role in the control of water movement, causing detachment, fracturation and swelling, respectively (Nicholson 2001; Jiménez González and Scherer 2004; Benavente et al. 2007a; Siegesmund and Dürrast 2011).

In carbonate tufa, however, plant moulds create macropore-scale pores and voids. The investigations of the carbonate tufa (Fig. 3b) show the presence of large pores that can reach 500 μm, as shown by the SEM observations. The significant interconnection between the pores in the carbonate tufa can result in high values of pore connectivity as well as in high capillary imbibition coefficients (García-del-Cura et al. 2012).

Petrophysical characterisation of fresh samples

The main agent of weathering and degradation of stones is water, which plays an important role in the transfer and deposition of particles within the stone (Beck and Al-Mukhtar 2010). For this purpose, petrophysical characterisation is essential for understanding the behaviour of stones and evaluating their durability. In fact, bulk density, porosity and pore size distribution influence rock susceptibility to weathering because nearly all the weathering processes rely on the chemical reactions of salts or other contaminants with the stone material in the porous media (Nicholson 2001). The petrophysical characteristics of limestone and carbonate tufa are shown in Table 1. The porosity of carbonate tufa is more important than that of limestone because carbonate tufa is characterised by its macropore volume, which is reflected by the fact that the total porosity values are higher than the connected porosity values. The limestones, with a 2.58 ± 0.02 g/cm3 bulk density, represent heavy samples, while the carbonate tufas, with a 1.47 ± 0.12 g/cm3 bulk density, represent the lightest samples.

Table 1 Petrophysical values of the limestone and carbonate tufa. ρbulk Bulk density, ϕO open porosity, ϕT total porosity, ϕc connected porosity, vP compressional wave velocity, σC uniaxial compressive strength, C capillary absorption coefficient
Full size table

The presence of macropores in the stone is shown by significant mercury intrusions beyond the detection limit of MIP (García-del-Cura et al. 2012). MIP shows different porous networks between the limestone and the carbonate tufa. Figure 3c shows that the limestone exhibits a polymodal pore size distribution, where the thin pores (0.001-0.01 μm) represent 72% of the pore population. This kind of pore size distribution corresponds mainly to interparticle pores because the larger pores indicate the interparticle pores around bioturbations and fissures (Fig. 3a). In the carbonate tufa (Fig. 3d), the pore size distribution is mainly unimodal, in the range 1–100 μm, and its porosity can exceed 1 cm, which has been shown in macroscale investigations and is supported by SEM observations (Fig. 3d). The connected porosity, ϕC, coincides with the open porosity in the limestone rocks but differs greatly in the carbonate tufa facies. When the micro- and the macropores are joined, the open porosity gives more realistic results for interpreting the porosity of the carbonate tufa (García-del-Cura et al. 2012).

The capillary absorption coefficient, C, of the limestone is less than that of the porous carbonate tufa, which reflects the pore networks of these stones (Table 1). Capillary imbibition occurs mainly in pore sizes between 1 μm and 1 mm and provides high water absorption rates for a well-connected pore structure (Benavente et al. 2007b, 2015). Thus, the large pores in carbonate tufas play a major role in the overall capillary absorption, as seen in the C–ϕo graph where the high porosity values show a high capillary absorption coefficient (Fig. 4a). The action of capillary transport is negligible for pore radii of up to 1 mm, and the mechanism that predominantly controls fluid movement is gravity. Hence, the large pores in carbonate tufa are revealed by the difference between the total and connected porosities. In fact, the main mechanisms that control the fluid in the carbonate tufa are capillary action and gravity, while capillary action is the main mechanism in the limestone.

Fig. 4
figure 4

The correlations of petrophysical properties. Capillary coefficient absorption versus open porosity (a), P wave velocities versus total porosity (b), and compressive strength versus total porosity (c) of limestone and carbonate tufa

Full size image

The vP–ϕT graph shows that P wave velocity increases when porosity decreases according to a linear relationship (Fig. 4b) (García-del-Cura et al. 2012). The propagation of ultrasonic waves is hindered in stones with high porosities. According to Martínez-Martínez (2008) and Martínez-Martínez et al. (2011), the lower vP obtained for some samples of limestone can be explained by the presence of large calcite crystals inside the textures and the fissures. Limestone and tufa present medium to very low uniaxial compressive strengths according to Deere and Miller (1966), ranging from 56 to 70 MPa for limestone and from 1.5 to 3 MPa for carbonate tufa. The relationship between strength and porosity shows an exponential trend for both stones (Fig. 4c), where the porosity provides preliminary information on strength values (Török 2006; García-del-Cura et al. 2012).

Porosity values of less than 10% indicate interparticle pores (intercrystalline and intergranular), while values higher than 45% are associated with macropores; thus, stone strength is more sensitive to macropores than to interparticle porosity (García-del-Cura et al. 2012; Benavente et al. 2018). These types of porosities highly influence the mechanical properties of stones in the sense that they determine the weakness of the structure. For low porosity rocks, small changes in porosity usually involve considerable changes in strength (García-del-Cura et al. 2012).

Durability test

Salt damage is the main mechanism responsible for stone weathering in the Byzantine Wall. Fresh limestone and carbonate tufa were subjected to an accelerated weathering process using the salt crystallisation test, which is widely used for stone durability assessments. The weathering of these stones was monitored using five parameters: (1) weight loss in the sample, (2) visual deterioration, (3) ultrasonic response, (4) bulk density and (5) open porosity. These parameters were studied to compare the variations in petrophysical values and to assess the degree of decay. Figure 5 shows the variation in weight loss after each cycle of the salt weathering test performed on the untreated samples. In general, the carbonate tufa presents excellent durability against salt crystallisation compared to that of the limestone, such that the tufa samples are structurally sound at the mesoscale, while the limestone samples show different types of weathering forms.

Fig. 5
figure 5

Salt crystallisation test: curve representing the trend of weight loss of limestone (L) and carbonate tufa (T) during the 15 cycles of salt crystallisation, and their state before and after tests for limestone (a) and carbonate tufa (b)

Full size image

Due to the formation of salt crystals inside the pore spaces, some limestone samples initially display an increase in their weights until the sixth cycle. Other samples continued to increase in weight until the tenth cycle. From the seventh cycle onwards, the limestones show a decrease in weight and begin to degrade, exhibiting several weathering forms, such as efflorescence, fracture, flaking and loss of materials, mainly in the bioturbations (Fig. 5a). All the limestone samples display a superficial granular disaggregation, which contributes to the decrease in their weight after the durability test. In limestone, some superficial flaking is observed, which causes the loss of thick sheets on the sample surfaces. Fissures may coalesce and then turn into fractures and cause sample disaggregation and flaking. The formation of fissures and fractures involves both the initiation and propagation of pre-existing fissures or extensions of new fractures (Nicholson 2001), resulting in an increase in open porosity and a decrease in bulk density after the salt crystallisation test. Table 2 shows the variation of petrophysical properties after the salt crystallisation test. The variations are lower in both limestone and carbonate tufa. In limestone, this lower variation is controlled by fissures and loss of materials. An increase in fissure density also increases water absorption, which makes weathering more severe (Eppes and Keanini 2017). According to Thomachot and Jeannette (2002), wetting–drying cycles can increase the porous network density, which can lead to material damage.

Table 2 Petrophysical properties before and after the salt crystallisation test for the limestone and carbonate tufa. Bulk density, ρbulk; open porosity, ϕO; compressional wave velocity, vP
Full size table

Discolouration is another weathering form and can appear in the form of colour accentuation in bioturbation structures. Discolouration is caused by the reaction of iron oxides with water during the salt crystallisation test. The colours propagate around the bioturbation surface and result in the appearance of surface disaggregation. The presence of iron oxides associated with bioturbations increases local porosity and decreases rock strength because the chemical reactions between water and iron oxides through salt action lead to stone weakness and degradation. The influence of iron oxides on this type of weathering pattern has also been reported by Akin and Özsan (2011) and Benavente et al. (2018) in coloured travertine. The durability test shows the importance of this bioturbation, which is observed in the samples (L3 and L4) with high levels of bioturbation. As a consequence, these samples show more weathering, which is reflected in granular disaggregation and fracture formation. At the end of the test, the limestones show an average weight loss of approximately 1%.

In addition, the tufa samples show a sudden increase in weight after the first cycle because the amount of salt retained inside the stone is easily deposited without any detrimental effect (Fig. 5b). The stone undergoes no degradation, and a weight reduction is observed only in one sample, in which a small detachment on the surface is observed after the first cycle. The pressure generated by salt crystals is not sufficient to cause significant damage in this stone type because the pores are large enough to contain the pressures generated by salt growth, which can be seen from the fact that there is a 1% gain in weight. The decrease in the carbonate tufa porosity can be explained by the salt crystals that form inside the pore network, which is indicated by the decrease in bulk density (Table 2) and the increase in weight without weathering after the salt crystallisation test.

The mechanical strength evolution quantified by vP is insignificant after the durability test, and, in most cases, the values fall between the maximum and minimum resistances of the unweathered samples (Table 2). Therefore, the more limited variation of the ultrasonic velocity corroborates that the internal changes alone (fissures and pores) are not sufficient to cause any deterioration in material properties. During the drying step of the test, the thermal expansion behaviour is at least partially controlled by single-crystal properties (Malaga-Starzec et al. 2002) because calcite strongly influences weathering due to the anisotropy of the coefficient of the thermal expansions of different mineral axes (Eppes and Griffing 2010). Therefore, the expansion and contraction of minerals within the stone induce the appearances of fissures and fractures, which weaken and accelerate weathering, especially in limestone with low porosity (Gomez-Heras et al. 2009; Benavente et al. 2018). Consequently, limestone and carbonate tufa present different decay mechanisms. Although the mechanical resistance of carbonate tufa is lower than that of limestone, tufa is more resistant to salt crystallisation action (Benavente et al. 2018). The resistance of the material to crystallisation pressure creates tensile stress over the pore surface (Benavente et al. 2004).).

However, the carbonate tufa exhibits larger pores (Fig. 3a), and thus, the weathering effect of salt pressure decreases considerably. Crystals may pass through large pores without generating significant levels of crystallisation pressure on the pore walls (Benavente 2011; Benavente et al. 2011). However, limestone has thin pores, where crystal growth produces an effective crystallisation pressure on the pore wall. The pressure generated by the growth of salt crystallisation produces a connection between isolated pores and open pores, which can enlarge fissures and create fractures (La Russa et al. 2013). A few clay minerals are also observed between calcite crystals and on surface fissures. Thus, both iron oxides and clay minerals may also enhance the effectiveness of salt crystallisation and contribute to granular disaggregation and fissuration.

Weathering characterisation

The Byzantine Wall of Tebessa exhibits a wide range of weathering forms that affect the main stones used in its construction (Fig. 6). Limestone has several weathering forms and generally presents soiling and efflorescence in the lower parts of the Wall. Efflorescence is related to water and fluid infiltration by capillary action, whereas soiling is caused by rainwater, which makes dust stick toi the lower parts of the stones. In particular, rising damp is a major problem of damage of the Byzantine Wall, which contributes to salt dissolution and transport through the pore structure of stones. Fissures affect some ashlars and are independent of limestone structure. At a height of 60 cm above the ground, the limestone mainly presents flaking, white crust, back weathering, alveolisation, discolouration and scaling (Fig. 6). Flaking is related to alveolisation, which is a step in the process of the formation of alveoli (Cassar 2002). Discolouration is associated with bioturbation and is widely found on the limestone surfaces because their bioturbations are rich in iron oxides. Moreover, the chemical erosion of these bioturbation structures affects the calcite and iron fractions prior to discolouration. The bioturbation surfaces show granular disaggregation, and the stones often present a great tendency to become discoloured.

Fig. 6
figure 6

Weathering forms obtained from the north side of the Byzantine Wall, which show only the limestones. The main weathering forms are flaking, alveolization, white crust, efflorescence, fracture, and discoloration. The damage categories of the Byzantine Wall are very severe, severe, moderate, slight and no damage

Full size image

X-ray diffraction and SEM analysis revealed that the efflorescence, white crusts and flaking are composed of gypsum and halite. The gypsum crystals are oriented at different angles to the surface of the pores (Fig. 7). The halite crystals present dissolution (corrosion) habits related to dissolution–precipitation cycles in the surface stone. The flakes, which are thin slivers of material that detach as salt crystallises behind them, are dominantly composed of gypsum. The detachment of the flakes suggests rhythmic fluctuations in the solution supply and salt crystallisation (Kramar et al. 2010). The fissures are 100–200 μm thick and oriented in different directions from the surface. The surface fissures present iron oxides and clay minerals and reduce the strength of the stone. The flow of water and salt solutions weakens the planes, and the cyclic swelling of clay contributes to the cracking and delamination of the stone (Rodriguez-Navarro et al. 1997; Sebastián et al. 2008).

Fig. 7
figure 7

Scanning electron microscopy (SEM) images for the weathered limestone surface. Gypsum crystals present a lenticular habit whereas halite crystals show dissolution forms. Interparticle porosity is partially closed by the clays and the salts

Full size image

Carbonate tufas represent approximately 50 stones in the entire Wall. This carbonate tufa generally presents two weathering forms, namely, loss of material (powdering and granular disaggregation) and fracturing when the stone is used in the lower part of the Wall close to the bottom. The investigation by the X-ray diffractometer for the weathering products shows salt absence. Meteoric waters may also have washed them out because they present high values of porosity and large pore sizes. Beyond a height of 5 m, no weathering forms appear in the stones, which tend to resist degradation at this height, whereas the weathering in the lower parts may be related to low mechanical resistance.

Damage categories are determined by comparing individual damage rates according to Fitzner and Heinrichs (2002) and Fitzner et al. (2002). The following five categories of damage were defined: very severe, severe, moderate, slight and no damage (Fig. 8). The Byzantine Wall revealed damage categories that range from moderate to severe. However, the most important weathering, including flaking, white crust, alveolisation, discolouration, granular disaggregation and back weathering, is found in the north and west facades. The weathering forms, which are recorded during the durability test, are showed in the Byzantine Wall for both stones, and mainly for flaking, alveolisation, efflorescence, fracture and loss of material.

Fig. 8
figure 8

Schmidt hammer rebound values of different conservation states, such as weathered and unweathered surfaces for limestone and carbonates tufa of the byzantine Wall

Full size image

Figure 8 shows the values of the Schmidt hammer rebound numbers obtained on weathered and unweathered surfaces. The differences between some of the Schmidt hammer rebound values can be explained by the differences in the values obtained from smooth surfaces, surfaces containing protruding grains and the fissured zone (Kolaitai and Papadopoulos 1993; Katz et al. 2000; Ericson 2004; Aydin 2009; Nazir et al. 2013; Sousa 2014). Furthermore, the values obtained from the Schmidt hammer test indicate changes in strength of the weathered and unweathered surfaces without any specific information about the weathering characteristics (Siedel et al. 2011).

Specifically, the surface strength decreases because of a loss of cohesion between grains. As a result, water can move easily within the pores and transport soluble salt, which is reflected by the flakes and surface deterioration (Mol 2014). The influences of water, porosity and mechanical strength changes result in the degradation of stones, which is reflected in the weathering forms. Finally, weathering generally affects limestone more than carbonate tufa because the weathering forms are controlled mainly by the lithology and the position of the stone within the Wall. These results were also reported by Stück et al. (2011). Thus, the higher decay on the north face may be due to water, and that on the west face may be due to a higher thermal influence (Eppes et al. 2016) through salt action.

Conclusions

In this study, limestone and carbonate tufa showed several weathering forms, including flaking, alveolisation, efflorescence, discolouration, white crust, fracture, loss of materials and powdering. These weathering forms were mainly produced by salt action (gypsum and halite) as the result of interactions of multiple factors with a contribution of rising damp. Both stones were analysed and characterised via laboratory tests to understand stone durability. Limestone and tufa present different mineralogical, petrological and petrophysical properties that explain their behaviour in the constructions.

The porosity of limestone is lower than that of carbonate tufa, which is characterised by a specific macropore volume. This pore structure property is reflected in the capillary absorption coefficient. Because porosity strongly influences the mechanical properties, the P wave velocity and porosity display a negative linear relationship as pore space hinders the propagation of ultrasonic waves. Compressive strength and porosity exhibit a negative exponential trend. Limestones and tufas present a medium to very low resistance ranging from 56 to 70 MPa for limestones and from 1.5 to 3 MPa for carbonate tufas. The salt crystallisation test estimated the weathering processes as well as the petrophysical evolution of the stone response. Carbonate tufas show an excellent durability against salt crystallisation compared to the limestones.

Degradation and an increase in fissures were found on the surface of the stones. Bioturbation influences the porosity and iron oxide content of the stones, inducing granular disaggregation and brownish colours after contact with water. These weathering forms are accentuated by salt action and create fractures, flaking and more granular disaggregation. Mechanical strength is important in explaining the degradation in carbonate tufa, which has a very low resistance to loads, when used in the lower part of the Byzantine Wall. For carbonate tufas, there is a loss of salt constituents as the pore size is enormous and salt crystals are easily transported out of the stones. The failure of these stones in construction is due to their low resistance to compression. These stones failed easily and very quickly when used at the foot of the Byzantine Wall and remained intact when used in the upper rows.

In the Byzantine Wall, the damage ranges from medium to severe, demonstrating the inadequate conservation of the Byzantine Wall and the need for urgent preservation measures. The replacement of severely degraded stone is needed during the first steps of the conservation. This step will decrease the hazard of the structurural loss of stability of the Byzantine Wall. To limit the salt influence of the Wall, it is essential to prevent the water action and the rising damp. Finally, future deterioration of the built heritage of Tebessa is expected to be more substantial for limestone than carbonate tufa.