Chemo-Mineralogical Changes of Six European Monumental Stones Caused by Cyclic Isothermal Treatment at 600 °C

1. Introduction

Thermal treatment of stone and mineral materials is a wildly studied phenomenon in many areas of research and application. It has been used to determine the effect of fire damage on civil and building structures [1,2], in concrete technology [3], geotechnical engineering [4], ceramics firing [5], or to study the dynamics of the earth’s interior [6], among others. In the built cultural heritage, it can be used to study dating of ancient fires [7], provenance of materials [8], materials mix-design [9], to combat microbiological growth on stone surfaces [10] or as an artificial ageing technique prior to study conservation treatments [11,12]. Physical changes of building stone (e.g., Young’s modulus, flexural strength, porosity, etc.) caused by thermal treatment are well-studied phenomena [13,14,15,16,17]. Chemo-mineralogical alterations (e.g., decomposition onset, mass loss in different conditions, chemical alterations, etc.) lack in research, especially regarding transition temperatures and transformation extent and consequent influences on the stone’s structural integrity [18,19]. Stones that share the same chemical and mineralogical compositions might differ from the point of genesis and therefore microstructure, which means that their resistance towards thermal treatment is also likely to vary. Moreover, stone understood as agglomerates of minerals, even when monomineralic, differs from single crystals in their susceptibility to thermal treatment [20]. The most common methods to study chemo-mineralogical changes induced through various forms of thermal treatment include simultaneous thermal analysis (STA), X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), scanning electron- and optical microscopy (SEM and OM), while physical changes can be studied with numerous petrophysical tests (e.g., compressive- and tensile strength, ultrasonic pulse velocity, etc.) [21].

The main mineral components of most building stones are either of carbonate (i.e., calcite and dolomite) and/or of silicate (i.e., quartz, mica, feldspars, etc.) nature. Furthermore, a significant number of stones comprise clay minerals like kaolinite, illite or smectite. Iron oxides, such as hematite and limonite, are the main colour giving minerals, which make up minor and trace amounts of natural building stones. These minor and trace components are important because they are most likely to influence the visual appearance of the building stone upon thermal treatment, a phenomenon that has been well studied [22]. Minerals present in a minor amount can still have a strong influence on the stones integrity as a recent study demonstrated [23], where pyrite oxidation was responsible for an explosion like decay that started already from 400 °C. All aforementioned minerals are susceptible to heat degradation by varying degree, and their abundance can influence chemo-mineralogical alterations of the texture. Thermal decomposition of kaolinite (Al₂Si₂O₅(OH)₄) can be initiated at temperatures >400 °C, depending on experimental conditions and the structural properties of the mineral [24]. When kaolinite is calcined, dehydration is followed by dehydroxilation which is irreversible when the crystal structure collapses to form amorphous, reactive metakaolin (Al₂Si₂O₇). If the temperature further rises, transformation of the intermediate amorphous Al₂O₃ and SiO₂ take place and a formation of the crystalline γ-alumina or a blend of amorphous silica with alumina-silicon spinel is obtained, whereas temperatures >900 °C cause further transformations into mullite (Al₆Si₂O₁₃) and α-cristobalite [25]. All quartz (SiO₂) bearing stones require a temperature of 573 °C to undergo a sudden change in volume due to the α-to-β phase conversion. This thermal expansion, while reversible, is the primary cause for permanent physical changes in silicates. If heated further (>870 °C), β-quartz that contains impurities forms the high-temperature polymorph α-tridymite, while the latter causes a neoformation of β-cristobalite [26,27]. If tridymite formation does not occur, the β-quartz forms directly cristobalite. As regards carbonates, thermal expansion of calcite and dolomite rises progressively from room temperature to its decomposition. The calcination of calcite (trigonal polymorph of CaCO₃) begins at 600 °C [28] while dolomite (CaMg(CO₃)₂) can decompose earlier, starting from 500 °C [29,30]. However, such processes will differ and are influenced by many extrinsic and intrinsic parameters. Major extrinsic influences include the atmosphere condition, the peak temperature and partial pressure, the residence time, nature and duration of exposure as well as the physical state of a material (i.e., powder, solid or aggregates). Intrinsic parameters include the mineralogical composition, content and chemical reactivity as well as the microstructural properties of the fabric, such as crystallinity, grain-size distribution, porosity, etc. The thermal decomposition of calcite is a well-studied phenomenon due to its industrial importance [31]. Calcite decomposes into its solid calcium oxide and gaseous carbon dioxide reaction products. The solid product of the reaction is highly reactive and re-carbonates under atmospheric conditions, under the influence of moisture and carbon dioxide, back to calcium carbonate, a process known as the lime cycle. If calcined at temperatures >900 °C and in presence of quartz, a formation of Larenite (Ca₂SiO₄) is possible [32]. Comparable to calcite, thermal decomposition of dolomite has been extensively studied with numerous thermoanalytical techniques. However, unlike calcite, the decomposition reaction of dolomite is a more complex, multi-step reaction [33,34,35].

Through history, heritage structures constructed out of natural stone have suffered fire damage around the world [36]. Following recent events at Notre Dame de Paris, the Glasgow School of Arts or the New York’s Saint Sava Cathedral [37], the topic of how to remedy the damage on stone monuments caused by higher temperature still remains. A recent study [38] investigated challenges that a post-fire restoration poses with a focus on Notre Dame de Paris. The same study assumed that the interior stone condition implied maximal temperatures of 500 to 600 °C. Fire related peak temperatures can vary in range and duration, however, 600 °C is a temperature high enough to reduce the stone´s soundness and is therefore a widely studied peak temperature in many heat-related studies. Such temperatures are important for investigations of potential phase changes while higher temperatures have also relevance in industrial use and the understanding of planetary dynamics. Siegesmund [39] describes that for thermal expansion in building stone temperatures of up to 120 °C are relevant, however, elevated temperatures are still resulting in much higher residual strain and therefore more damage. He further notes that with temperatures below 500 °C the pore radii do not change significantly and with the exception of very porous stones (>15 %), it can be said that the higher the temperature, the higher the residual strain. Therefore, at 600 °C mineral decomposition is likely, depending on the mineral constitutes present in the texture and their exposure conditions.

This experimental study analyses chemo-mineralogical changes of six silicate and carbonate European building stones caused by cyclic thermal treatment during a static peak temperature of 600 °C. The purpose of this study is to determine the extent of chemo-mineralogical changes that occur on solid macroscopic samples on the base of their most abundant minerals, that is, calcite, dolomite, quartz and kaolinite. A multi-analytical approach was used combining OM, in-situ XRD and colour measurements to study alterations on solid samples. Furthermore, ground samples were also analysed by means of STA, cycling thermogravimetric analysis (TGA) and FTIR to determine the main chemo-mineralogical modifications that occur.

2. Materials and Methods

2.1. Petrographic Characterisation of the Lithotypes

Figure 1a shows the characteristic texture of one of the most renowned white marbles, the Apuan Marble (AM), quarried from the Apuan Alps in northern Tuscany, Italy. This fine-grained metamorphic rock consists of calcite and contains traces of dolomite as well as quartz and is also known under the name Carrara Marble. Its granulometry is equi- to inequigranular and the grain contacts are polygonal to interlobate with an approx. size of 0.2 mm. The Apuan Marble is dense, which is why its porosity, determined by mercury intrusion porosimetry (MIP) amounts to approx. 0.7 %.

The porous, biogenic sedimentary rock, St. Margarethen (SM), is a biosparite, or calcarenite that is quarried in the Leitha Mountains at the border between Lower Austria and Burgenland in Austria. The colour of this detritic limestone ranges from yellow-brownish to light grey. The micrograph (Figure 1b) displays a texture that is composed of small fragments of fossils of coralline red algae, foraminifera, serpulides, ostracods and echinoderms, some of which can reach a size of up to few centimetres in hand specimens. Additionally, SM contains detritic quartz in traces. The grain contacts of this calcarenite are cemented with coarse crystalline (i.e., > 10 μm) calcite cement. Its total open porosity (measured by MIP) is approx. 22 %.

Ajrate Trevino in Spain is the home of two carbonate varieties that are located in the same quarry. The beige to yellowish Ajarte Dolostone (AD), exhibits distinctive microstructural differences due to the varying abundance of the binding matrix (Figure 1c and Figure 1d). Its total open porosity amounts to 25 %. The creamy greyish calcite variety of Ajarte (AC) is classified as a poorly-sorted biomicrite to biosparite. Figure 1e shows the carbonate matrix that consists of recrystalised fossils and shell fragments but the stone´s microstructure makes it difficult to distinguish between components and the matrix. Furthermore, this stone has a high amount of intercrystalline pores and in generally pores filled with sparite cement. The porosity of the calcite variety of Ajarte is approx. 23 %. A clear mineralogical and chemical distinction of the two Ajarte varieties extracted from the same quarry can be seen in Table 1. For the XRD diffractograms the reader is referred to Supplementary Information (Figure S1).

The clastic sedimentary rock Balegem (BL), located in the provenance of Oost-Vlaanderen, Gemeente Balegem in Belgium, is normally light greyish in colour and its main components are calcite, quartz, feldspar and many different foraminifera. This detritic limestone, also known as Lede stone, is classified as a fossilifereous, siliclastic arenite. Due to its almost equal ratio of calcite to quartz, this stone is often referred to as a sandy limestone. The siliceous clasts are predominantly composed of angular quartz being approx. 0.1 to 0.2 mm in size. Occasionally, larger grains of siltstone, inclusions of goethite as well as glauconite can be observed too. It has no preferred bedding orientation and the matrix between the aggregates is a microsparitic to sparitic cement. Figure 1f shows the homogeneous texture of the Balegem stone, but it should be noted that this lithotype is quite inhomogeneous and occurs in varying qualities and petrophysical properties [45]. The porosity by MIP for the Balegem stone used in the present study amounts to 10 %.

The coarse-grained Schlaitdorf sandstone (SQ) is located in Württemberg, Germany and is light to reddish in colour. Its detrital fraction consists of quartz, feldspar, rock fragments and a cement that consists of a sparitic dolomite and microcrystalline silica with smaller amounts of kaolinite and illite (Figure 1g). The cement is homogeneously distributed in the well-sorted, not layered homogenous texture. The average grain size is about 0.5 mm and reaches a size of up 1 mm while the grains are very angular to subangular and they exhibit an average sphericity. Schlaitdorf’s main accessory minerals are apatite, tourmaline, zircon and opaque minerals with less than 1 % occurrence [46]. The open porosity of the stone, measured by MIP is 16 %.

The second silicate variety, the Obernkirchen quartz sandstone (OQ), is located in Lower Saxony’s Brückeberge in Germany and its colour ranges from whitish over greyish to yellowish and orange. It is a fine-grained clastic sedimentary rock with a grain size averaging between 0.05 to 0.1 mm as can be seen in Figure 1h. This quartzarenite has no obvious layering nor a bedding orientation and the grain contacts are manly planer to lobated with a slight authigenic growth of quartz. The lithotype is grain supported containing approx. 5 % of homogeneously distributed kaolinite. The open porosity of OQ is 24 %.

The studied lithotypes played an important role as historical building stones in Europe´s build cultural heritage and are still actively quarried for purposes of restoration and reconstruction work [47,48,49]. The abundance of the main mineralogical components as analysed by XRD are listed in Table 1.

2.2. Calcination

Quarried stone specimens with a dimension of 50 x 50 x 25 mm have been exposed to three thermal cycles at a static peak temperature of 600 °C. Therefore, a Thermo Scientific Heraeus K 114 electrically heated furnace with 3.5 L internal volume was used. The heating rate was set to 40 °C min^-1 until the isothermal conditions were met and subsequently kept for one hour at 600 °C. Following a cooling phase to approx. 35 °C with an opened front door to overcome differential stresses caused likely by a rapid cooling, this procedure was repeated for three times overall. The residence time was established by experimental pre-screenings for all the studied lithotypes [40]. The chosen time showed sufficient strength of the fabrics to overcome thermal gradients and thus complete breakdown of the specimens by formation of larger cracks. Moreover, three cycles displayed the highest threshold for reduction of soundness examined in terms of physical changes by means of sound speed propagation and water absorption. Even though the stagnation of the reduction of soundness for all lithotypes was not comparable to the same degree, the goal was to reduce the soundness of the fabric to the lowermost degree and study the chemo-mineralogical changes at this transition temperature. However, it is important to note that residence time deviates with shape of the specimen used as well as the volume of the oven. This needs to be adapted accordingly to achieve the same degradation effect like in the present study. For some characterisation techniques, where the alterations were difficult to be analysed, additional calcination for a prolonged time at 600 °C and at 900 °C was also performed and used to cross-validate chemo-mineralogical transformations. All these changes in experimental conditions are indicated in the corresponding graphs.

2.3. Analytical Techniques

Simultaneous thermal analysis (STA) was carried out employing a Netzsch device type STA 409 PC Luxx. The fresh quarried and grinded samples were heated at a constant rate of 10 K min^-1 while the weight change (TGA: thermogravimetric analysis) as well as the heat flow (DSC: differential scanning calorimetry) were recorded simultaneously. For the analysis 50 mg of sample was weighted into a Pt-pan and heated from room temperature to 1000 °C. The controlled atmosphere consisted of air, in an amount of 50 ml min^-1, and N₂, in an amount of 15 ml min^-1. Prior to testing, the powdered samples were stored in a drying oven at 65 °C for 72 hours in order to remove the free water from the surface.

The rate of weight change as a function of temperature was recoded with a TA instrument type 2050 equipped with an EGA furnace (TA Instruments, New Castle DE, USA) while the analysis was carried out using the TA Universal Analysis 2000 Software. The thermogravimetric analysis (TGA) was done under static air conditions without the flow of gas. It should be noted that the type of the furnace (especially its volume) is important as the CO₂ partial pressure during carbonate decomposition will vary with the type of equipment used. The 50 mg weighted sample was placed in a Pt-pan and heated in two experimental conditions to study the difference between cyclic and constant exposure time, both at isothermal conditions. The two setups included (i) three cycles at isothermal conditions at 600 °C (3 x 60 min) to simulate the thermal conditions in the furnace during calcination, and (ii) one cycle at isothermal conditions at 600 °C employing the same exposure time (1 x 180 min).

The Fourier-transform infrared spectroscopy (FTIR) was done by placing the powdered samples between two diamond plates and applying pressure to obtain a transparent film of the solid. Prior to the analysis of the sample the diamond grid was analysed solely to subtract the background signal. The spectra were recorded using a Thermo Scientific Nicolet iN10 MX Microscope equipped with an MCT/A detector. The measurement range was set to 4000-675 cm⁻¹ while the spectral collection was made in transmission mode accumulating 16 scans in 3.07s at a resolution of 8 cm⁻¹, using an Ominc Picta Software (ThermoFisher Scientific, Waltham MA, USA). The stones were analysed in their freshly quarried and thermally treated conditions. The samples were analysed within 20 minutes after thermal treatment.

All samples used for analysis of STA, cycling TGA and FTIR were ground in the same manner. Prior to grinding the samples have been crushed manually to an approx. size of 1 cm³. Afterwards they have been placed in a Fritsch pulverisette 9 Vibrating Cup Mill for a duration of 4 minutes. Only the silicate-bearing stones have been milled for additional 2 minutes until the sample material was haptic powdery and no granularity was sensed.

For measuring the colour difference before and after thermal treatment a Konica Minolta spectrophotometer type CM700d was used with a spot diameter of 8 mm and the test was performed according to the EN 15886 [41]. Hereby a standard daylight illuminant D65 with a 10° observer was applied. The results were analysed in the CIE 1976 L*, a* and b* colour space to obtain the overall colour change (ΔE*). Three to five spots per sample area and condition were averaged to obtain the colour change.

Optical microscopy was used on polished thin sections of standard thickness (i.e., 30 μm) with an Olympus BX41 microscope. Furthermore, fluorescence light microscopy (Zeiss AxioSCOPE A1) was employed to study the crack inducement after thermal treatment. The sampling was done on a cube of 50 x 50 x 50 mm to assure similar structural features and divided into three prisms (25 x 25 x 50 mm) to analyse (i) the sound condition, (ii) thermally treated conditions (three cycles at isothermal conditions at 600 °C, 60 min per cycle), and (iii) 900 °C (one cycle at isothermal conditions at 900 °C for 60 min). A temperature of 900 °C was employed for purposes of highlighting certain effects like colour change, microcracks inducement, and transformation front, among others. As the samples heated at 900 °C experience severe changes, including volumetric expansion, one week passed before sample preparation was possible. The samples were embedded in a yellow (fluorescent) dyed epoxy resin and cured overnight at 40 °C.

In-situ XRD spectra were collected using a PANalytical X´Pert MPD Pro powder diffractometer (PANalytical B.V., The Netherlands). A BBHD mirror was used to select the CuK_α1/2 characteristic lines (λ_α1 = 1.54060Å, λ_α2 = 1.54443Å) for the experiment. The instrument was setup for Bragg-Brentano reflection geometry in θ/θ-mode. The working distance between the sample and the position sensitive 1-dim. X´Celerator detector was 200 mm. A high-temperature furnace (Anton Paar HTK 1200N), operating in air under ambient pressure, was used to track the thermal decomposition of disk-shaped stone specimens, 18 mm in diameter and approx. 2.5 mm thick. Following dedicated temperature steps, the diffraction diagrams were recorded in-situ. Therefore, we were able to observe in-situ the phase evolution with temperature. For each temperature step the data was continuously collected in a range from 15 to 90° diffraction angle 2θ with a speed of 0.01°∆2θ/s. Subsequently the data were analysed, including also the Rietveld refinement, by the HighScore Plus software package [42] using the Powder Diffraction File (PDF4+ [43,44]; International Centre for Diffraction Data in Philadelphia, USA). Rietveld refinement is a full-profile fitting method in which the difference between the experimental XRD diffraction profile and the calculated one is minimized by a least-squares procedure in order to match the patterns and account for variations, such as those induced by temperature changes. The samples were measured at room temperature followed-up by heating steps up to 600 °C (40°C min^-1 heating ramp) simulating the calcination conditions used beforehand in the laboratory. Six subsequent XRD diffraction patterns were collected under isothermal conditions by stopping the heating ramp for 10 minutes at the temperatures 100, 200, 300, 400 and 500 °C, respectively. At 600 °C we used a temperature hold of 60 minutes, which corresponded to six patterns collected. After the heat treatment, the sample was cooled down to room temperature. This heating/cooling cycle was repeated three times to produce the generation of a sufficiently extended surface layer of transition phases. Prior to in-situ XRD a conventional XRD Panalytical XPert Pro MPD diffractometer was also employed. The samples analysed with the latter instrument were grinded as described above.

3. Results and Discussion

3.1. Powdered Sample Analysis: Simultaneous Thermal Analysis, Cyclic Thermal Gravimetric Analysis and Fourier-Transform Infrared Spectroscopy

STA provided, in addition to XRD, additional information on the mineralogical composition of the stones and was able to indicate differences in mass change between the materials. The calcium carbonate´s prominent endothermic peak that corresponds to its decomposition occurs at 863 °C for AM, SM and AC despite their differences in genesis and thus microstructure (see Figure S2 to S4). However, the onset for the AC variety is lower and starts just above 600 °C while the two-remaining carbonate´s onset is initiated at 650 °C. This difference might be explained though the much finer calcite microstructure with more surface area in the variety AC, when compared to AM and SM, where a difference between the cement and carbonate microfossils is indistinguishable. Moreover, sample characteristics (e.g., amount, packing density, particle size, thermal conductivity, etc.) are known to influence thermoanalytical recordings [50,51], despite all stones have been grinded in the same manner regardless of their differing mineral habitus. The total weight loss of carbonates, is attributed to the release of gaseous carbon dioxide due to calcite thermal decomposition, as examined by means of TGA. The variety BL differs in both, the onset and decomposition peak from the remaining carbonates (see Figure 2a). A possibility for the early decomposition onset at 560 °C might be ascribed to its minor mineralogical contents, that is, the presence of glauconite, which can amount to approx. 3.4 wt. % [45]. Glauconite, similarly to mica and kaolinite, starts to eliminate chemically bound water between 300 °C to 400 °C while a change in structure occurs only above 550 °C [52]. The shift of the endothermic calcite peak to a lower temperature of 828 °C might be related to the overall heterogeneous mineralogical content and the fine-grained calcite matrix, similar as within AC. While glauconite could not be detected in the XRD spectra, the occurrence was visible in thin section (see Supplementary Information). The exothermic peak above 900 °C can be attributed to recrystallisation processes but it cannot be assigned to a certain phenomenon, even more so as an overlap is possibly present, indicated through the endothermic peak at 920 °C. At this temperature range glauconite is known to collapse, decomposition of micas crystalline structure and formation of spinel can occur or a formation of hematite is possible, all phenomena that might have occurred [53,54]. The formation of hematite could take place either through the decomposition of glauconite or a transformation of goethite, both minerals present in BL stone variety [22]. The presence of goethite was found in BL as a natural phase as confirmed by measurements of XRD (see Figure S5). The dolomite in AD exhibits two distinct endothermic peaks (Figure 2b). The first, a double peak corresponding to dolomite decomposition, occurring at 770 °C with a second peak located at 780 °C, while the second peak corresponds to calcite decomposition and a rapid decomposition rate that occurs at 840 °C. The present of dolomite in the cement of the silicate variety SQ also exhibits the prominent double peak (738/756 °C) but without the subsequent calcite decomposition, which is either absent or overlapping (Figure 2c). In either case, the dolomite double peak present in SQ occurs at lower temperatures when compared to that occurring in the dolostone AD. However, it cannot be excluded that the decomposition onset of dolomite present in SQ might have overlapped with the dehydroxilation of kaolinite, which would influence the readings concerning onset degradation temperatures. Comparing the kaolinite`s present in both silicate varieties, their dehydroxilation is initiated at 450 for SQ and 420 °C for OQ while the endothermic peak hill is located at 502 °C for SQ and 475 °C for OQ. Only the OQ variety allowed to extract information of the completed dehydroxilation that lies at approx. 685 °C (see the derivative of the mass change in Figure 2d). For SQ completed dehydroxilation could not be observed as dolomite decomposition arises, as stated previously. Regarding the kaolinite onset and peak temperatures, grain size and the structural order-disorder are known to influence the shift and form of the peak patterns [55]. The difference between a sharp endotherm as observed within a reference clay (KGa-2 Kaolinite [56]) and the broad endotherms as within the clayey matrices inside the quartz sediments can be viewed in Figure S6. Moreover, the SQ variety is composed of a rather variegated marly clay consisting mainly of kaolinite and small amounts of illite, so it exhibits more impurities in terms of mineralogical content as compared to the kaolinite present in OQ [49]. As concerns the quartz bearing stones, quartz inversion is always exhibited at 574 °C. This so called “quartz peak” is normally reported at a temperature of 573 °C and the recorded 574 °C falls in the range of experimental error. A comparison of the residual mass after heating to 1000 °C can be extracted from Table 3 along with the results in mass change as analysed periodically and non-periodic by means of TGA.

Cyclic TGA is in agreement with STA as it confirms that the carbonate varieties AC and BL are more prone to thermal decomposition then AM and SM (see Table 3 and Figure 3a and 3b). TGA and STA despite having different heating rates, atmospheric conditions and residence time, have a similar trend regarding the mass change with an exception for AM and SM which lie in close proximity as regards the cyclic TGA data. Experiments concerning cyclic (i.e., three cycles at 60 min) and non-periodic (i.e., one cycle at 180 min) thermal treatment by means of TGA differ slightly in total mass change but the trend is the same for all samples. However, cyclic treatment always exhibits a more pronounced effect on the mass change then the non-periodic treatment when employing same residence time for both experimental setups. The results further show that thermal cycling does not have the same effect on all stone varieties concerning weight loss per cycle applied. In the case of AM and SM the first cycle is where the weight change occurs while the remaining cycles can be characterised as fluctuations (Figure 3a). Nevertheless, in all cases the first thermal cycle is where the most severe change occurs following by a reduction of mass change by each subsequent cycle. The dolostone AD undergoes the most pronounced weight change having the lowest residual mass and the highest weight loss with each thermal cycle as can be viewed in Figure 3a and 3c. The susceptibility to thermal decomposition starting from the most susceptible stone follows the order AD < SQ < AC < BL < OQ < AM < SM in the case of both, cyclic and non-periodic exposure conditions. As concerns SQ and OQ, the entire weight reduction is associated with kaolinite and in the case of SQ also with dolomite decomposed. The exact amount of kaolinite and dolomite in the analysed sample is not known, which hinders statements on the completeness of the decomposition of the same.

FTIR showed the presence of dolomite and kaolinite in SQ (Figure 3d) and kaolinite in OQ after the thermal treatment. The most prominent vibrational modes that confirm the presence of dolomite are the broad ν₃ vibration at 1415 cm⁻¹ (asymmetric stretching of ${CO}_3^{2-}$) and the sharper ν₂ vibration at 875 cm⁻¹ (out of plane bending of ${CO}_3^{2-}$). The easiest indices of a kaolinite presence are the bands assigned to the hydroxyl groups at 3621, 3660 and 3706 cm^-1. Furthermore, a 913 cm^-1 shoulder (Al-OH bending and translation modes) also indicates the presence of kaolinite (Figure S7a). The formation of metakaolin, a reactive amorphous phase, results often in the disappearance of the 3621-3706 cm^-1 as well as the 913 cm^-1 bands. However, it should be noted that in few cases the absence of kaolinite and dolomite was recorded, which shows that due to the inhomogeneity of the powdered samples multiple spectra need to be recorded to exclude the presence and/or absence of phases. For cross-validation, solid samples have been subjected to thermal treatment for 10 hours of exposure at 600 °C, following by a grinding of their surface layers (< 1 cm³). Within the latter samples, the presence of dolomite was also confirmed (Figure S7b). These results confirm that cyclic isothermal treatment at 600 °C will not entirely decompose kaolinite and dolomite constituents in a building stone, two of the most susceptible minerals towards thermal decomposition. The latter is true for the employed thermal conditions and used stone volumes.

3.2.. Solid Sample Analysis: In-Situ X-Ray Diffraction, Colour Measurements and Microscopy

3.2.1. X-Ray Diffraction

Initially XRD was done on samples quarried freshly and subsequently heated at cyclic isothermal treatment at 600 °C. Due to the transition temperature of 600 °C, calcium oxide that originates is highly reactive and rapidly forms calcium hydroxide which subsequently re-carbonates during the cooling- and sample preparation process, which hindered a detection of eventual decomposition phases. The only building stone that showed permanent difference was the dolostone variety AD, where a calcium magnesium carbonate phase was developed that showed broad peaks next to dolomite peaks (see Figure S8). The broad peaks might be attributed to concentration gradients due to non-stoichiometry compounds of the newly created phase [57]. The recording of highly reactive phases (i.e., calcium- or magnesium oxides) was only possible through the use of in-situ XRD (Figure 4). Here a real-time analysis of phase changes at the surfaces of the building stones was made possible by using the Rietveld method [58]. Through the Rietveld method we were able to calculate the amount of decomposition and formation phases in wt.-% (see Figure 5).

The most intriguing transformation that could be observed is a retrograde reaction during calcite decomposition. This phenomenon was observed on all studied calcium carbonates but was most pronounced on surfaces of AM (Figure 5a). The activation energy for the retrograde reaction is provided by temperatures ranging from 400 °C to 500 °C during the second and third cycle of heating: The formation of calcium carbonate was detected without the intermediate calcium hydroxide phase. Such a phenomenon was already observed within another study [59]. Comparing the phase evolution with temperature for all calcium carbonate samples, it is evident that the onset for thermal decomposition starts earlier for the variety AM, followed closely by SM and AC. This is likely associated with the surface sample topography, as lithotypes whose grain-to-grain contact are capable to transfer heat more efficiently as compared to sparse grain contacts accompanied by porosity and surfaces roughness. These differences in topography are most likely also responsible for the amount of phase transformed. At this point, it should be noted that there is a distinctive difference between the decomposition of calcite when studied with solid and powdered samples. The illustration of this issue is easiest displayed between data obtained with in-situ XRD and TGA. Namely, when analysed by TGA, AM exhibits the lowest weight loss, which corresponds to degradation processes, while AC the highest among the calcium carbonate stones. In contrary, in-situ XRD displays an opposite trend with AM being the most prone to degradation while AC yields the best resistance towards heat decomposition. These opposing trends can be explained by sample conditions, that is, the use of powder versus solid material. AC and AM were grinded for the same period of time, which most probably resulted in different powder sizes and therefore reactivity with AC being the more reactive one when grinded.

While the Rietveld refinement was easily done on (predominantly) monomineralic stones like the carbonates (see the phase development with temperature displayed in Figure 6), marble veining cannot be quantified by Rietveld refinement due to fluctuations of the phases caused by thermal expansion of the substrate. Such a diffractogram of AM having a dolomite vein is shown in Figure S9. The dolomite vein is easily degraded and periclase is formed. As regards the silicate variety, the typical alpha- or low quartz that is stable until 573 °C and the beta- or high quartz that appears at the isothermal treatment at 600 °C can be seen in Figure 7. To explain this phenomenon, it is important to know that interatomic distances are correlated with unit cell lattice parameters and employing higher temperatures results in changes of these interatomic distances, which is reflected in the change of the peak position. The transition from low- to high quartz is a sudden phase transition (change in crystal structure due to displacement but no bonds are broken) accompanied by a linear expansion [60]. This linear expansion, along with the linear expansion of calcite, is what is responsible for the variations as analysed by Rietveld refinement (normally < 2 wt.%). The expansion of the minerals causes changes in the penetration depth of the X-rays, which adds to the fluctuation in qualitative analysis. As the quartz inversion is reversible, phase decomposition in the XRD pattern can only be seen on dolomite and kaolinite (see Figure S10). SQ exhibits a lowering of both phases, dolostone and kaolinite, but not a complete decomposition, which could also be confirmed by FTIR analysis. While a temperature of 600 °C is high enough to cause an irreversible loss of hydroxyl groups of kaolinite, this process depends strongly on the structural state of the kaolinite (e.g. particle size, packing density, pressure of water, experimental conditions, among others) [24]. For SQ, the complete breakdown of kaolinite was not observed, and as a consequence, it can be assumed that kaolinite present deeper within the fabric will also be preserved. This can partially be explained by its protection in the confined spaces between the quartz grains. Furthermore, these results show that a comparison between powdered and solid samples is not reliable in regard to onset transition temperatures for dehydroxylation processes.

In-situ XRD had the advantage of using macroscopic solid samples, which investigated truthful conditions that happen during heating under atmospheric conditions. The number of patterns collected as well as the used heating ramp allowed for a collection of decomposition and formation of phases but no intermediate phases like brucite or portlandite.

3.2.2. Colour Measurements

Visual alterations of natural stones that have encountered thermal treatments are among the widely studied modifications. Responsible minerals for colour changes are phyllosilicates, iron oxides, such as goethite and calcium hydroxide. Colour changes can be observed on all stones (see Table 4) and they range from yellow-reddish over greyish to white.

Yellow, reddish and greyish discolouration originate from iron bearing minerals, they are among the most prominent ones when it comes to visual changes due to thermal treatments. A white veil can be ascribed to decomposition of calcium carbonate and formation as well as subsequent carbonation of calcium hydroxide. The white surface present on a discoloured fabric is often the only indices that transformation have happened, a sensing with analytical techniques fails due to the high reactivity of the products and their fast carbonation. AM is the only lithotype that turned brighter, while the remaining stone varieties appeared darker after the thermal treatment (see the lightness coordinates ΔL*). All stone varieties indicated more redness with the only exception of SM, which indicted more greenness after the thermal treatment. SM was also the stone variety that underwent the most severe colour change after thermal treatment, as can be seen by ΔE* and indication of blueness.

3.2.3. Optical Microscopy

Analyses of textural changes after thermal treatment were done with the help of OM. The colour, phase and microstructural changes are the most prominent differences that could be observed by means of OM and it can be observed throughout the whole sample volume analysed. As colour measurements were done only at the surface of the stones, optical microscopy confirmed that the mineralogical changes affect the entire specimen volume. A darkening of the microfossils in e.g., BL seems to be present, which can only be seen by OM as colour changes will average the visual alterations on the surface. AM displays a whitish precipitate on its surface, which comes about through the subsequent carbonation process of the thermally decomposed calcium carbonate. The latter changes on AM are surface limited, which can be explained by the diffusion-limited escape of CO₂ that results from calcium carbonate decomposition. Namely, when CO₂ escapes during decomposition its concentration in pores is able to act back on the calcite lattice thus preventing the occurrence of phase transformations [61].

As concerns changes of the clayey matrix, interpretations of OM micrographs are limited because the clayey structure can be inhomogeneous due to their complex diagenetic history. Most often a darkening of the clayey matrix could be seen in thin section in plane-polarised light, an effect that could only be confirmed by employing the 900°C heating conditions. To confirm chemo-mineralogical transformations caused by the thermal treatment at 600°C, the stones were additionally heated at 900°C to make sure to have severe changes and be able to recognize alterations caused by lower temperatures. With the comparison of these two temperatures employed it is visible that the higher the temperature, the more striking the damage in the natural stones, especially the colour change and microcracks inducement, independent of stone genesis and mineralogy.

Microcracks were visible and are most noticeable within the dense stone variety AM, but also within the silicate substrates SQ and OQ (see Figure 8). Using UV-light on thin sections imbedded in a fluorescent resin, the extent of microcracks inducement is easily evident (see Figure A1 and A2). The higher the temperature employed the more severe the inducement of inter- and intragranular cracks. The inducement of microcracks was less pronounced on the porous carbonates SM, AC and AD, as well as on BL (see Figures in the Supplementary Information). This can be related to the stones´ structure. Namely, coarse grained porous stones can accommodate more stresses formed by the expansion of minerals. A denser stone that are grain supported are more susceptible to crack development as there is no space for expansion. Moreover, induced microcracks in quartz come with a sudden change in volume owned to the quartz inversion. Furthermore, fine-grained porous stones are prone to crack development that goes throughout the entire specimen (most often observable within OQ). The latter is the case when differential expansion surpasses the strength of the stone leading to a formation of large cracks and the breakdown of the specimen. Nonetheless, induced microcracks cause a reduction in mechanical strength and an increase in water absorption, leading to changes in mechanical and petrophysical properties [62].

At temperatures of 900°C microstructurally driven decompositions that depend on the studied stone were observed. At such high temperatures the changes in the fabric are caused by three key characteristics, namely (1) the thermal expansion and retraction of single aggregates, (2) the differential stresses in the whole specimen under study and, (3) the volumetric changes caused by the chemo-mineralogical transformation during the decomposition and subsequent carbonation. However, even though the stones were exposed to 900 °C, sound fragments in terms of mineralogy and physical appearance are still present. Prominent changes are visible on SQ and OQ with a pore-radii change, where larger cavities and pores are created due to the expansion and contraction stresses and a rearrangement of the single grains (see Figure 8). Microstructure rearrangement is also visible in the sedimentary carbonate stones but in form of a blending effect of the microcrystalline calcite cement, along with abundant crack formation though the microfossils (see Figure 9). The most intriguing alterations can be observed on the microfossilis of SM. A thermally induced pitting decomposition can be seen that presents itself as sharp round islands (Figure 10). These phenomena do not follow the typical degradation profile, being most severe at the surface with a continuous reduction towards the inner bulk. Instead, the preferential pitting decomposition raises along formed cracks and microfossils structural features. As for AD, specifically in the microcrystalline calcite cement, the pitting degradation is visible in form of blurry patches. These phenomena appear almost black in cross-polarized light (see Figure 10) caused by the very small, needle-like hydroxides (in case of the carbonates Ca(OH)₂) which exhibit due to their grain sizes a very low birefringence and thus almost an isotropic optical character. Additional micrographs of the studied conditions, also for the herein less described stone varieties AD, AC and OQ, can be viewed in Supplementary Information.

4. Conclusions

The present study analysed chemo-mineralogical changes occurring at transition temperatures of 600°C on six European monumental stones. TGA confirms that mass loss is more pronounced when cyclic treatment is employed as compared to non-periodic thermal conditions when considering the same residence time to heat exposure. FTIR was sensitive in detecting kaolinite and dolomite confirming a retention of these minerals inside the stone fabric even though their dehydroxylation and decomposition temperature was exceeded. As assessed by microscopy, apart from the formed microcracks (petro-physical change) and colour alterations (chemo-mineralogical alterations), no evident changes can be observed within the stone specimens when exposed to cyclic isothermal treatment at 600 °C. Only higher temperatures (i.e., 900 °C) allowed to observe decomposition phenomena in the texture by means of thin section analysis. In carbonates, microcracks inducement is more severe in dense metamorphic lithotypes as compared to sedimentary varieties. Grain-supported silicates generate larger cracks causing a macroscopic failure of the samples. In-situ XRD analysis reveals that temperatures of 600 °C will cause phase transformations on calcite while the threshold for dolomite decomposition is found to be 500°C. For the same minerals, phase transformation is not initiated at the same temperature. An earlier onset is favored on denser substrates exhibiting more grain-to-grain contacts. Kaolinite is preserved even though its phase transient temperatures are exceeded, which can be explained by its occurrence in the confined stone matrix that makes it more protected towards transformations. The use of powder and solid stone substrates cannot be correlated. While STA, TGA and FTIR give a needed overview of possible changes at certain conditions, in-situ XRD has the advantage of using macroscopic solid samples, which investigated truthful conditions that happen during heating of monumental stone under atmospheric conditions. Thin section microscopy is a powerful tool to describe heat-related alterations ranging from the surface to the inner bulk, especially when employing higher temperatures. The calcination protocol as well as analytical approach allowed a precise collection of decomposition- and formation of phases as well as microstructural alterations.