Tremolite-Asbestos Presence in Roman Archaeological Site of Micia, Romania

Abstract

This paper reports the first evidence of the presence of the mineral tremolite asbestos in Roman building materials from the Micia archaeological site (Romania), thus contributing to the understanding of the implications of ancient building materials. The Micia archaeological site includes both a fort and a civilian Roman military settlement that was inhabited by both civilians and soldiers from various Roman troops. Over time, since the late 2nd century AD, the settlement has undergone significant reconstruction, especially after some fires. Tremolite asbestos is a non-flammable mineral that, due to its fibrous properties, was used in the past in building materials, although it poses health risks when inhaled. To highlight it, several advanced and highly sensitive scientific techniques are used in this work to discover the presence of tremolite asbestos and to examine its structure, composition, and morphology inside the investigated samples. Tremolite asbestos is typically white to gray or greenish in color, characterized by thin, needle-like fibers that can easily become airborne and inhaled. It is a crystalline mineral that usually forms long, straight, sharp fibers. Under high magnification in optical microscopy or in scanning electron microscope images, correlated with other performant analytical techniques (XRD, WDXRF, FTIR, Raman, BET, TGA), tremolite asbestos appears as elongated, slender fibers—often bundled or intertwined—with smooth or slightly striated surfaces.

1. Introduction

Recent historical–archaeological studies of Roman buildings have led to an understanding of ancient construction techniques, as well as material use and resource supply [1,2,3].

The growing interest in characterizing Roman building materials, such as stone, mortars, and pigments, aims to uncover the methods used by the Romans in their architectural designs and helps to understand the materials and methods used in construction. Also, it plays an important role in the conservation and restoration of archaeological sites and in understanding how Roman materials have deteriorated over time and what techniques can be used to restore these sites authentically. For economic reasons, the ancient Romans tended to use locally available building materials (and cheap labor) whenever possible. The import and transport of building materials was limited to the bare necessities or to high-value luxury items such as marble. The most commonly used building materials by the Romans were wood, unburnt brick, stone, and concrete. Roman concrete was made from a mixture of sand, water, and pozzolanic ash. Core walls filled with a general mixture of mortar and rubble that included pieces of broken pottery were also used [4].

From an economic, military, and cultural point of view, Micia represents one of the largest camps of the auxiliary troops of Roman Dacia, along with Porolissum and Tibiscum, built on the south side of the Mureş River, about 3 km east of the Brăniscai Gorge, the narrowest point in the valley of this river (Figure 1). Established as a quasi-urban settlement, Micia has significant structures including a fort, an amphitheater, a thermal complex, craft areas, sacred areas, and two known necropolises, all in an area of approximately 25 hectares. The site is geographically bounded by the Mureș River to the north (approximately 2.5 km east of the Mureș Gorge), near the Poiana Ruscă Mountains, within or near the administrative areas of Vețel, Mintia, Herepeia, and Vulcez [5,6,7,8].

Micia was connected to one of the main Roman roads that crossed the territory of Transylvania, acting as a defense on the western borders of Roman Dacia, whose capital, the ancient city of Ulpia Traiana Sarmizegetusa, was located approximately 50 km away and the city of Apulum 80 km away. Another strategic mission of Micia was to protect the gold and silver mines in the Apuseni Mountains [7].

Micia was inhabited between the 2nd and 3rd centuries, the settlement being largely destroyed since antiquity, and following wars the population remained there, at least until the Hun invasions (5th century) when all Roman settlements were abandoned.

However, a large-scale process of destruction over time affected the ruins of Micia until the 19th century, when stone blocks from its walls were extracted by locals to repair the Brănişca road, devastated by the floods provoked by the River Mureș. The ancient city continued to deteriorate in 1869, when the construction of the Deva-Arad road started. Also, in the 1960s, when the Mintia thermal power plant was built, this contributed negatively to this deterioration process, while the Mureș River, whose course was diverted in 1967, submerged part of the ruins of Micia. In this area, andesitic tuffs from Pădurea Bejanului, Almașul Sec, and those south of the city of Deva (on the way to Almașul Sec) have been identified [9].

During the Roman period, the advancement of building materials technology was driven by events such as the Great Fire of Rome in July 64 AD. This is how fireproof stone for public buildings emerged. “Lapis Gabinus” was a popular type of quarry volcanic rock, used as fire protection in many public structures. This type of volcanic rock was full of intrusions, including basalt, which made it particularly resistant to fire. This is how Romans introduced asbestos into construction materials. At the Micia camp, asbestos was introduced from 170–175, when the destroyed fortifications were rebuilt. Large quadrangular blocks of Uroiu augustite–andesite (opus quadratum) were added and the foundation, up to 0.80 m deep, of quarry stone (micasite) bound with mortar (opus incertum) [7,8,10,11]. At this stage, asbestos was added to the mortar and used for the first time at the Micia monuments.

At that time, asbestos was brought from the Orșova area, Ponicova (Cazanelor reservation) where there is a recognized asbestos deposit on the Ciucaru Mare hill. In fact, the Orșova Mining Enterprise is famous for feldspar, mica, asbestos, talc, dolomite, and quartz. Considering the composition detected by analytical techniques, asbestos and mica could be extracted and transported from Orșova.

Having only a few sources of literature, based only on geographical studies [10] without sufficient details regarding the complete composition, structure, and weathering, a study regarding the application of asbestos is absolutely necessary. In this sense, the present study will first examine the composition of the samples collected from the Roman monuments identified at the Micia fort and will, for the first time at this site, highlight the presence of a form of tremolite asbestos in the mortars used in the second reconstruction that took place at the Micia fortification. Characterizing Roman building materials is a crucial endeavor that bridges archaeology, engineering, and cultural heritage. Because asbestos was found in some building materials extracted from the still-working archaeological site, some protective measures must be taken to make it accessible to potential visitors. High-performance and sensitive analytical techniques were used to highlight the tremolite-asbestos species and to identify the structure, composition, and morphology of these minerals, as follows: optical microscopy, stereomicroscopy and scanning electron microscopy (SEM), X-ray diffraction (XRD), wavelength-dispersive X-ray fluorescence (WDXRF), Fourier transform infrared spectroscopy (FTIR), Raman spectroscopy.

2. Materials and Methods

2.1. Materials

For this study, some samples of Roman-aged mortar have been collected from different building types belonging to the Micia archaeological site in Romania (

Table 1. Photos of samples collected and stereomicroscopy images.

No.	Investigated Sample	Stereomicroscopy Image
1
2
3
4
5
6

). Sampling was carried out under the guidance of archaeologists to ensure the collection of a sufficient number of specimens for obtaining statistically meaningful results. All the samples identified by the archaeologists are shown in Table 1. In this table are shown both the photos of the collected Roman samples and the stereomicroscopy images, before introducing the other analytical investigation methods [12].

2.2. Analytical Techniques

Optical microscopy (OM) was performed with a Novex Microscope BBS trinocular microscope (Euromex Microscopen B.V., Arnhem, the Netherlands) at various magnifications, a digital video camera (Axiocam 105, Zeiss, Göttingen, Germany), ZenPro software (ZEN 311) for a real-time data acquisition, and with analysis conducted using ImageJ 1.50.

Also, for concise and detailed images, an Edmund Optics C-Mount videomicroscope has been used (Edmund Optics Ltd., adresa:1 Opus Ave, Nether Poppleton, York YO26 6BL, UK), equipped with an AmScope 3.2MP MT9T001 CMOS C-Mount camera (3210 El Camino Real, Irvine, CA 92606, USA), an extension videotube, ZOOM ring, focus ring, a Zabber T-NA08A25 XYZ Micro linear actuator with a medium-intensity spot/coaxial light, an Edmund Optics 2X EO M Plan Apo Long Working Distance Infinity Corrected objective, and an Edmund Optics XYZ Stage.

When preparing samples for optical microscopy, the goal is to create a specimen that is thin enough and has sufficient contrast to observe features and is representative of the material being studied. Samples are mounted on glass slides or in resin blocks (especially for cross-sectional analysis).

Additionally, stereomicroscopy was performed using a Euromex trinocular stereomicroscope (Model 1903, EUROMEX Microscopen B.V., BD Arnhem, the Netherlands), with magnification capabilities ranging from 5× to 40×. Stereomicroscopy is used primarily for observing three-dimensional surface features of opaque or semi-transparent specimens at relatively low magnification (typically 5×–50×). Unlike optical (compound) microscopy, no sectioning or very thin slicing is needed because it relies on reflected light rather than transmitted light.

Environmental scanning electron microscopy (ESEM) was conducted using a FEI Quanta 200 microscope (Eindhoven, the Netherlands) to assess the morphology of the samples. The analysis was carried out under high-vacuum conditions, with magnification ranging from 50× to 100,000×. Prior to imaging, each specimen was coated with a 5 nm layer of gold using a Q150R-ES sputter coater (Quorum Technologies Ltd., West Sussex, UK) to minimize charging effects and enhance conductivity during SEM examination. Scanning electron microscopy (SEM) provides high-resolution, 3D-like images of surfaces by scanning a focused electron beam across the specimen. Because SEM uses electrons, not light, sample preparation needs a conductive coating (gold). Samples are fitted inside the SEM chamber (a few cm max) and are mounted to avoid movement.

X-ray diffraction (XRD) analysis has been achieved by using a Rigaku Ultima IV diffractometer (Rigaku Corporation, Tokyo, Japan), and the analysis was conducted using Cu Kα radiation (λ = 1.5406 Å), with an operating voltage of 40 kV and a current of 200 mA. Diffractograms were recorded over a 2θ range of 5° to 80°, with a scanning rate of 4° per minute. The samples were prepared by grinding, using an agate mortar and pestle or ball mill to reduce them to <10 μm particle sizes, and powder was pressed into a sample holder (often with a glass or zero-background slide).

For precise and detailed compositional analysis, the Rietveld refinement method was applied using a non-linear least-squares approach. Parameters such as peak shape, unit cell dimensions, crystallinity, and crystallite size were calculated [13].

Wavelength-dispersive X-ray fluorescence (WDXRF) was used with a Rigaku ZSX Primus II spectrometer (Rigaku, The Woodlands, TX, USA), which includes a Rh anode X-ray tube and a 4.0 kW power source, including a 30 µm front Be window, to conduct detailed analyses. An EZ scan method with Rigaku’s SQX software (ZSX version) was used for data processing, contributing to accurate and thorough spectroscopic results. The samples were prepared by the “pressed pellet method”, as follows: grind sample to <75 µm (typically using a ball mill or agate mortar), dry thoroughly (usually at 105 °C) to remove moisture, mix with a binder (e.g., 5% BOREOX), press into pellets using a hydraulic press (20–30 tons pressure).

Fourier transform infrared spectroscopy (FTIR) was performed using a Perkin Elmer spectrometer (Waltham, MA, USA) in attenuated total reflection (ATR) mode. In ATR mode, measurements were taken with a resolution of 4 cm⁻¹, accumulating 32 scans per spectrum, and covering a spectral range of 4000–400 cm⁻¹ to ensure comprehensive analysis. The samples were placed as a solid (powder or small piece) directly on the ATR crystal (usually diamond or ZnSe), pressure was applied to ensure good contact, and the spectrum was collected.

Raman spectra were acquired using a Rigaku Xantus-2 portable analyzer (Rigaku, The Woodlands, TX, USA), equipped with stabilized 785 nm and 1064 nm lasers. The analysis was performed with a resolution of 4 cm⁻¹ and a laser power of 252 mW. Data processing was carried out using Opus 7.0 software (Bruker Optics GmbH). The samples were placed directly under the objective lens (on a clean glass slide or holder), tweezers were used to avoid contamination, and the samples were gently flattened or spread thinly (for powders). The powders were compacted lightly to reduce laser light scattering.

Thermogravimetric analyses were conducted using a Pyris 1 TGA/DTG analyzer (Perkin Elmer, Waltham, MA, USA) over a temperature range of 50–800 °C, with a heating rate of 10 °C/min and nitrogen flow at 50 mL/min. The analysis was performed on ground samples of the binder, which had been mechanically separated from the larger aggregates of the mortars. Typical sample size was 5–20 mg, and the sample was fine, dry powder. It was spread evenly in pan (alumina, platinum, or ceramic) crucibles and cut or crushed into small, uniform pieces (<1 mm).

The textural properties of the samples were investigated using the Brunauer–Emmett–Teller (BET) method, which involved analyzing nitrogen adsorption and desorption isotherms at 77 K across a relative pressure range of 0.005 to 1.0. A NOVA2200e gas adsorption analyzer (Quantachrome Instruments, Boynton Beach, FL, USA) was employed for the measurements. Prior to analysis, the samples (0.1–1.0 g depending on material density and porosity) were degassed for 4 h at 180 °C under vacuum. The sample was placed as powder in a BET sample tube (typically glass). Data processing was performed using NovaWin version 11.03 software.

3. Results

3.1. Optical Microscopy Evidence of Tremolite-Asbestos Fibers

All samples collected from the archaeological site show multiple colors, from white to black, with a red area due to iron oxide samples. The black color may be due to andesite, as it is known that there was a large deposit of andesite in this area. The andesite is visible in the images obtained by stereomicroscopy (Table 1). The samples show a subrotund or tubular morphology and are predominantly siliceous, comprising fragments of monocrystalline quartz together with fragments of volcanic rocks characterized by trachytic and vacuolar textures. The grains in these samples fall within the fine to medium sand size range, from 0.125 to 0.5 mm. Their irregular, vacuolar appearance is attributed to fluids trapped during formation, with notable presence of tremolite–actinolite [14,15].

In general, asbestos refers to a group of six naturally occurring mineral fibers, which are classified into two main groups: the serpentine group, represented primarily by chrysotile, and the amphibole supergroup, which includes the asbestos-like varieties riebeckite, grunerite, anthophyllite, tremolite, and actinolite. These fibers show a good heat resistance and have a fibrous nature, making them useful in a variety of applications that require thermal stability and structural integrity, although their health risks are well documented due to their potential to cause serious respiratory diseases when inhaled. Actinolite is an intermediate member of a solid solution series between magnesium-rich tremolite, Ca₂(Mg_5.0₋_4.5Fe²⁺_0.0₋_0.5)Si₈O₂₂(OH)₂, and iron-rich ferro-actinolite, Ca₂(Mg_2.5–0.0Fe²⁺_2.5–5.0)Si₈O₂₂(OH)₂. Magnesium (Mg) and iron (Fe) ions can be freely exchanged within the crystal structure [16].

Tremolite asbestos was identified in the investigated samples from the archaeological site, most likely due to the reconstruction of these settlements after the fire. The asbestos can be identified either in the form of interconnected fibers and fibrils sometimes in the form of bundles or both as individual entities or as structures attached to andesite crystals. Also, the tremolite asbestos exhibits a silky luster and density and contains mobile and elastic fibers, as identified in our experiments. The fluffy white and striated appearance is clearly visible in all optical microscopy images, Figure 2.

Visual and optical microscopic examination of the investigated samples can easily put into evidence the presence of long fibers, translucid and flexible. The asbestos fibers are randomly oriented both as individual threads and as fiber bundles, which denotes the heterogeneous mixing of these fibers in the prepared mortar. Asbestos fibers are extremely fine, typically ranging in diameter from 100 nanometers to 1 micrometer, with lengths that can extend to several centimeters.

The particles visible in this figure have an elongated, fibrous texture in some areas, which is suggestive of tremolite-type morphology. This kind of structure could be consistent with fibrous asbestos-type tremolite, which is known for forming elongated needle-like or fibrous habits [17].

By visually analyzing the samples, three layers could be put into evidence. In our opinion, and based on the literature data, the middle layer could be a mixture of andesite with asbestos, while the exterior layer could be tremolite asbestos in the reconstruction mortars. All the subsequent investigation results show the existence of this layer of asbestos applied by the Roman population for the fire protection of these constructions.

These suppositions could be checked by optical microscopy images, as exemplified in Figure 3 showing a cross-section by optical microscopy (left) and more clearly by using ImageJ software (right).

3.2. Scanning Electron Microscopy (SEM)

According to the definition of “fiber” that we have used in this work (L > 5 μm, W < 3 μm and L/W > 3) [17,18], a fraction of the tremolite fibers found in the sample can be classified as tremolite asbestos. By SEM microscopy it was possible to clearly identify tremolite asbestos, characterized by elongated prismatic and fibrous crystals, as revealed through optical microscopy, Figure 4. The tremolite asbestos displays a lamellar structure, which indicates subsequent kinking and folding that occurred after its initial crystallization, highlighting a complex history of polyphasic deformation. Similar images have been reported by Rinaudo et al. [19].

These SEM images show features that resemble asbestos fibers, such as: elongated, needle-like shapes (visible centrally and to the right), high aspect ratio fibers (long and thin structures), and bundles or clusters of fibers embedded in a fragmented matrix. These are characteristic morphological traits of asbestos. The values shown in SEM images strongly support an asbestos habit, given the high aspect ratio, which is consistent with regulatory definitions of asbestos fibers.

These SEM images show some large, elongated structures, very high aspect ratio fibers (length ≫ width), and thin, needle-like, or splintery structures, with a tendency to appear as bundles or single fibrils. The L/W ratio varies between 5:1 and 46:1, which is higher than 3, consistent with elongated tremolite morphology. Regulatory bodies like NIOSH, OSHA, and EPA define asbestos fibers (e.g., tremolite asbestos) based on: length > 5 µm, width < 3 µm, L/W ≥ 3:1, but for clear asbestos classification, ≥ 20:1 and often ≥ 100:1 are used [20].

3.3. XRD/WDXRF Investigations

In the same manner, tremolite mineral has been identified by XRD techniques, Figure 5. In this figure, the collected samples have been analyzed, and tremolite, anorthite, cristobalite, quartz, and berlinite (aluminum phosphate) were put into evidence. Berlinite has been identified in similar Roman archaeological sites in Romania (Topolita, Deva, Rapoltu, and so on) [3].

XRD analysis indicates that the mineral paragenesis in the matrix of these samples consists primarily of tremolite, albite, and anorthite, suggesting a complex interplay of these minerals within the geological context. For exemplification, the XRD diagram of sample no. 2 is shown in Figure 6 after the Rietveld refinement method.

The Rietveld method minimizes the difference between an experimental model (observed data) and a model based on the hypothetical crystal structure and instrumental parameters (calculated model). All identified minerals could be observed and identified by the Rietveld method. Phase analysis was conducted through Rietveld refinement of both synthesized powders, using the TOPAS-Academic V3.1 software. In this figure, the XRD diagram presents the measured diffraction pattern (observed data) as a series of intensity peaks as a function of the 2θ angle. The overlaid calculated pattern, obtained from Rietveld refinement, is shown as a continuous line that closely matches the observed data. The vertical tick marks indicate the expected Bragg peak positions for each phase included in the refinement. The difference curve (observed–calculated) is shown at the bottom, highlighting the quality of the fit.

With this method, information can be obtained about a specific crystal structure, a hypothetical crystal structure, network parameters, and atomic positions, and their occupation can be confirmed/disproved. It allows obtaining information about a multiphase sample and helps to determine the relative amounts of each phase.

The Rietveld technique [21] is used to evaluate full profiles of all diffraction patterns and the QXRD-Rietveld method is the most accurate procedure for the quantification of crystalline mineral phases. For the materials containing an amorphous phase not exhibiting defined Bragg reflection, the QXRD-Rietveld method should be supplemented by using internal or external standards [22]. Recently, the identification and quantitative determination of amorphous and crystalline phases were performed by the QXRD-Rietveld method and synergic WDXRF, SEM, EDS, and multispectral image analysis [23]. However, the accuracy of the QXRD-Rietveld method depends on the crystallinity of the present phases, because “X-ray amorphous” phases cannot be directly measured [24].

X-ray amorphous phases identified in the samples had different values, as follows: Sample no. 4 (1%), Sample no. 5: 10.4%, and Sample no. 6: 3.6%. These samples have the highest silica and biphidite (phyllite) concentrations. The latter is a type of foliated metamorphic rock formed from slate that is further metamorphosed so that very fine-grained white mica achieves a preferred orientation [25,26]. The Rietveld refinement yielded a Rwp value of 10–12%, which is acceptable for a multiphase system of this complexity. The refinement model accounts for all major crystalline phases identified in the sample. The good agreement between observed and calculated patterns, as well as the low residuals in the difference plot, confirms the accuracy of the phase composition and structural parameters.

For the quantitative analysis via X-ray diffractometry the whole powder pattern fitting (WPPF) method was used. In this method, the profile fitting is carried out over a relatively broad angular range, incorporating information about the crystal system and lattice constants. The components identified in the mortar sample through XRD analysis include silica and alumina as the major components, iron oxide, and sodium and potassium alkali as minor components, Figure 7. Additionally, a small quantity of albite, a sodium aluminosilicate, was detected [27].

For the fragments with a possibly volcanic source, the chemical composition of the matrix identified by WDXRF (

Table 2. WDXRF analysis for all the samples collected from Micia Roman monuments.

	Sample No. 1	Sample No. 2	Sample No. 3	Sample No. 4	Sample No. 5	Sample No. 6
Oxide	Concentration, %
Na2O	3.4653 ± 0.31014	4.0862 ± 0.33715	3.4061 ±0.30668	4.0171 ±0.35643	3.2961 ±0.39452	3.0856 ±0.32632
MgO	1.3201 ± 0.14552	1.1072 ± 0.13904	1.0641 ±0.11972	0.6962 ±0.15198	0.9341 ±0.13627	0.9449 ±0.12918
Al2O3	21.9109 ± 0.1086	21.3876 ± 0.11212	21.8862 ±0.1027	22.3247 ±0.11406	20.6651 ±0.11073	21.5919 ±0.10741
SiO2	57.5004 ± 0.1559	60.7908 ± 0.1518	61.6323 ± 0.1518	58.7454 ± 0.1679	60.3688 ± 0.16446	58.0825 ± 0.15100
P2O5	0.4008 ± 0.0306	0.3748 ± 0.03480	0.2378 ± 0.02660	0.2765 ± 0.04121	0.6779 ± 0.02934	0.6080 ± 0.03077
SO3	0.5244 ± 0.02108		0.0368 ± 0.01854	0.0798 ± 0.02387		0.0936 ± 0.1969
Cl	0.0199 ± 0.00812		0.0190 ± 0.00710
K2O	1.9290 ± 0.03656	1.9094 ± 0.03322	1.9792 ± 0.02981	1.8778 ± 0.04308	2.0877 ± 0.04032	1.8126 ± 0.03567
CaO	6.0937 ±0.03497	5.4392 ± 0.0356	4.3263 ± 0.03147	5.9526 ± 0.03465	6.3984 ± 0.03871	6.1841 ± 0.03701
TiO2	0.4270 ± 0.05661	0.4307 ± 0.05087	0.3851 ± 0.04701	0.4274 ± 0.09215	0.3395 ± 0.05977	0.4334 ± 0.04599
MnO	0.1083 ± 0.0223	0.1330 ± 0.01790	0.1070 ± 0.01805	0.1087 ± 0.9177	0.1347 ± 0.02227	0.3161 ± 0.02124
Fe2O3	5.8688 ± 0.02479	3.9994 ± 0.01924	4.3533 ± 0.01483	4.7334 ± 0.02547	4.2710 ± 0.02626	5.7112 ± 0.02099
CuO				0.0323 ± 0.0153
Rb2O				0.0215 ± 0.00846
SrO	0.4315 ± 0.00806	0.3154 ± 0.00726	0.2733 ± 0.00640	0.4043 ± 0.00900	0.4012 ± 0.00882	0.3973 ± 0.00862
BaO			0.2935 ± 0.1188	0.3022 ± 0.13779	0.4029 ± 0.10928	0.6946 ± 0.21692
ZnO		0.0264 ± 0.00958			0.0227 ± 0.01117	0.0442 ± 0.01070
N.A.	2.033	1. 8538	1. 8524	1. 8174	1. 7796	1. 8689

) reveals variable amounts of CaO (4–6%), SiO₂ (57–62%), Al₂O₃ (20–23%), Fe₂O₃ (4–6%), K₂O (1–4%), MgO (0.6–1.4%), TiO₂ (0.3–0.5%), and Na₂O (3–4%). These results were similar for all the analyzed samples. Major element analyses contributed to the classification of the samples as tremolite as Hawthorne et al. reported [14]. Tremolite is a member of the amphibole group and is characterized by its specific chemical composition and structure, primarily consisting of calcium, magnesium, and iron in a silicate framework. Silica, SiO₂, could be identified with a proportion of 57 to 60%; meanwhile, alkali feldspar, together with an alkali metal oxide (Na₂O and K₂O) content of over 7%, was identified, which is responsible for the light-colored and aphanitic (fine-grained) texture, typically formed by the rapid cooling of lava or shallow intrusions enriched with silica and alkali metals.

These samples exhibit significant compositional heterogeneity in their aggregates with high concentrations of both limestone and volcanic fragments. The trace element concentrations (Table 2) indicate that these types of mortars contain volcanic fragments. Additionally, Sr and Ba are present in the investigated mortars, elements typically incorporated into the structure of less stable carbonates (such as aragonite) and sulfates. Table 2 also presents the mean concentrations (and standard deviations) of sulfur (S) and chlorine (Cl) for each mortar type. Mortars with the highest concentrations of sulfur and the lowest concentrations of chlorine may suggest the presence of gypsum in this area.

This is in good agreement with the observation of Marcus Vitruvius [28] and later confirmed by Deguara [29]. The samples show a broad range of mortar mixtures, primarily influenced by the local availability of raw materials during the Early Roman period. The presence of sulfur (S) and chlorine (Cl⁻), identified through chemical analysis, suggests the presence of saline phases, possibly halite [30].

The presence of S can be partially attributed to the volcanic content in the mortars [31] and anthropogenic contaminants. One possible scenario for the deliberate use of those mortars would have been preparation with marly binder and ceramic aggregate and different powders (marls, ceramic, and volcanic aggregates) which increased the pozzolanic activity.

The chemical composition of the samples analyzed by WDXRF shows the mortar types (middle layers) and their uses in the other materials present in this archaeological site. The samples have a low CaO content, which shows a close relationship with the SiO₂ + Fe₂O₃ + Al₂O₃ content, usually due to their aggregates (volcanic aggregates), Figure 8.

The carbonation process influences the microstructure of mortar by leading to alterations in its pore structure and density, which in turn affects the setting and hardening of the mortar. As carbon dioxide from the environment reacts with calcium hydroxide in the mortar, it forms calcium carbonate, impacting the overall strength and durability of the render through changes in moisture permeability and bond strength.

In the studied cases, the lowest CaO concentration and the highest (SiO₂ + Al₂O₃ + Fe₂O₃) concentration were found, most probably due to the carbonation process and the decrease in CaO. This could be evidence of the weathering process which could affect the mortar’s sustainability [32]. A high CaO content could suggest the use of lime-rich binders (Samples no. 1 and no. 5). A low CaO ratio (Sample no. 3) may indicate the use of volcanic ash, ceramics, or other pozzolanic additives. The middle CaO ratio (Samples no. 2, no. 4 and no. 6) could indicate a mixture between lime-rich binders and volcanic ash, ceramics and other pozzolanic additives.

3.4. FTIR

Figure 9 shows the representative FTIR-ATR spectra registered for the powdered fragments of the investigated samples. Pozzolana is identified in the spectra, characterized by the symmetrical vibration mode of the strong Si-O bands of silica at approximately 1080–1040 cm⁻¹, 795 cm⁻¹, and 462 cm⁻¹ (Si–O), which are assigned to silica gel (SiO₂), exhibiting a wide and intense peak [7], while the bands of calcium carbonate appeared at ≈1412 cm⁻¹ (CO₃²⁻ stretching) and 715 cm⁻¹ (CO₃²⁻ symmetrical bending), in line with the EDXRF results. The specific bands of Si-O at ~913, 1007, and 1091 cm⁻¹ and the lowest band at 1453 cm⁻¹ and 617 cm⁻¹ (Figure 9) are characteristic of asbestos and similar amphibole and pyroxene compounds [33]. Another silicate phase is relevant for the band at 465 cm⁻¹.

Saravanapavan and Hench [34] have identified Si-O-Si stretching vibrations in a calcium silicate structure. The mortar sample showed significant amounts of iron, complemented by the presence of the Fe-O bending peak for the mineral hematite at 540 cm⁻¹. Overlapping absorption peaks could complicate the identification of other mineral groups, which is expected due to the diverse inorganic minerals present in the raw materials used to make the mortars.

3.5. Raman Spectra

Raman spectroscopy effectively identifies and differentiates minerals within the amphibole group, but the peak assignment can be challenging due to the complex structural variations among them, leading to inconsistencies in peak/vibration assignments in different mortars. Tremolite shares similarities with other amphiboles like actinolite or anthophyllite, but small shifts in peaks (especially in the 670 cm⁻¹ and 1040 cm⁻¹ region) help distinguish it. Tremolite tends to have fewer Fe²⁺ substitutions than actinolite, which affects band position and intensity.

In the discussion of Raman spectra, it is important to consider all the potential processes that could occur in ancient mortars. These include the presence of calcium (alumino)silicate hydrate (C–(A–)S–H) and calcium silicate hydrate (C–S–H) gel, which exist as quasi-glasses in the form of nanometer-sized particles. The uptake of CO₂ by C–(A–)S–H lowers the pH of the pore solution, destabilizing the structure and causing deleterious cracking, which is generally responsible for the deterioration of mortars. The carbonation process also degrades the C–S–H gel, producing CaCO₃, silica gel, and water.

By using a 782 nm laser with Raman equipment, it was possible to put into evidence the main bands of asbestos (1000–2000 cm⁻¹). The most distinct peaks, Figure 10, are from 1200–1400 cm⁻¹ and are assigned to the Si-O-Si bridges (symmetric stretching vibrations) [19,35]. The obtained Raman spectra of the samples show typical features of the spectra of amphibole minerals [36,37]: the Raman spectra of the samples exhibited characteristic features between 300 and 600 cm⁻¹, corresponding to Mg–OH and Fe–OH vibrations, Si–O–Si bending motions, and OH– vibrations. Between 650 and 750 cm⁻¹, Si–O–Si symmetric stretching was observed, while peaks above 750 cm⁻¹ were attributed to O–Al–O symmetric stretching, as well as O–Si–O and Si–O–Si asymmetric stretching bands. The main feature in the low-wavenumber region, around 675 cm⁻¹, is the Si–O–Si symmetric stretching with Ag symmetry. This mode shifts from 675 cm⁻¹ in pure tremolite to 667 cm⁻¹ in Fe-rich actinolite due to the substitution of Mg²⁺ with the heavier Fe²⁺ [38].

In the investigated representative samples, the bands at 1280, 1358, and 1432 cm⁻¹ are specific to the middle layer. Well deconvoluted, the area between 1000 cm⁻¹ and 1600 cm⁻¹, specific to carbonates and silicates, belongs to tremolite. The other bands (1280 and 1358 cm⁻¹) could be assigned to the other minerals, amphibole and pyroxene compounds, confirming the FTIR results.

There are specific bands (1432 cm⁻¹) predominant in all the investigated selected samples, especially for middle and exterior layers, Figure 11, and assigned to tremolite asbestos, surrounded by other satellite bands at 1100, 1700, or 280 cm⁻¹.

3.6. Thermogravimetric Analysis

Temperature ranges (

Table 3. Weight losses (%) from TGA data.

Samples	<120 °C	120–200 °C	200–600 °C	>600 °C
1	0.45	0.27	0.5	0.08
2	0.43	0.32	0.3	0.1
3	0.23	0.18	0.2	0.13
4	0.17	0.2	0.05	0.14
5	0.1	0.03	0.56	0.2
6	0.26	0.07	0.87	0.09
Asbestos	3.59	1.2	4.23	10.15

) were selected as a function of the major thermal reactions suffered by mortars during heating: loss of adsorbed water (<120 °C), dehydration of salts as well as loss of zeolitic water and/or other hygroscopic compounds (120–200 °C), loss of structural water from hydraulic compounds like phyllosilicates, C-S-H, and/or C-A-H (200–600 °C), release of CO₂ by decomposition of calcium carbonate (600–800 °C), and other phenomena (>850 °C) such as decomposition of sulfates and/or loss of residual water and carbon dioxide. These thermal reactions were clearly recorded in TGA/DTA curves and confirmed by FTIR. A double peak of CO₂ emission likely due to the presence of Mg-bearing carbonates is also reported [39,40]. Results of the thermogravimetric and thermodifferential analyses (TGA-DTAs) of the investigated samples showed six main thermal processes: the first, at less than 100 °C, is due to the loss of weakly adsorbed water [41]. The bands around 550, 700, and 770 °C denote the bound water release and are assigned to the dehydroxylation of minerals from serpentine subgroup species.

The thermal analysis results of mortar samples show how different temperature ranges correspond to specific weight losses and chemical reactions. Temperatures up to 120 °C correspond to the dehydration of the sample, where water is lost from the material, likely from the surface or hygroscopic water. In the range 120–200 °C, the dehydration of hydrated salts could be identified by the weight loss, while in the range 200–600 °C, the weight loss is associated with the dehydration of hydraulic compounds (a measure of the extent to which the mortar has undergone hydraulic bonding–formation of compounds that harden when hydrated). Additionally, CO₂ could be generated from the reactions between calcium carbonate (CaCO₃) and silicates (XSiO₂), producing calcium silicates and carbon dioxide in the range 400–600 °C [42,43].

Above 600 °C, the decomposition of calcium carbonate (CaCO₃) (limestone or lime binder), leading to CO₂ release due to decarboxylation, indicates the transformation of the mortar and the mineral phases present. The calcium carbonate decomposition level is extremely low (the weight losses (%) after 600 °C are extremely low), meaning that the weathering of this rock is insignificant.

This analysis offers a detailed breakdown of the thermal stability and compositional changes in mortars over a broad range of temperatures, highlighting their dehydration and carbonation processes. The TGA/DTG method helps reveal how the material’s hydraulic properties evolve as a function of temperature, which is important for understanding the mortar’s performance and durability [40,44,45].

The most important thermal effects for accurate hydraulic classification are the weight loss of structural bound water (200–600 °C) and decomposition of carbonates (600–800 °C) with the release of CO₂ [46,47].

Specifically, the hydraulic reaction of the mortar may involve the following: lime and water reacting to produce hydrated lime, which is then consumed by reactive silicates and aluminosilicates (from pozzolana), forming C–S–H and C–A–S–H. Finally, calcium carbonate is formed through the reaction of CO₂ with unconsumed hydrated lime and/or as a result of the carbonation of hydrated phases [48].

The analyzed amphiboles could be calcic or sodic–calcic amphiboles with a tremolitic composition, indicating temperature metamorphic conditions: medium–low pressure and low to medium [49].

For the textural properties of the samples, the nitrogen adsorption/desorption measurements are shown. The results, specific surface area, total pore volume, and maximum values in the pore size distribution, are reported in

Table 4. Textural properties of Micia samples.

Sample	Surface Area (m2/g)	Pore Volume (cc/g)	Pore Diameter (nm)	K2O (%)	Na2O (%)
1	1.349	0.002	4.567	1.81	2.78
2	1.327	0.001	4.234	1.89	3.12
3	1.294	0.002	4.345	1.88	3.24
4	0.893	0.001	4.807	1.87	4.01
5	0.112	0.000	4.759	2.08	3.29
6	0.522	0.001	3.668	1.81	3.08

.

The BET method used revealed mesopores (2–50 nm) and provides general information on surfaces that include micropores and macropores if combined with other analytical techniques (SEM, XRD), presented in the above sections of our paper. The mesopores, defined as pores with a diameter of 2–50 nm (IUPAC), are the intermediate between micropores (≤2 nm) typical of zeolites (not identified in these samples) and macropores (≥50 nm) typical of porous glass. Silica is mesoporous, but this material does not possess an ordered mesopore structure and hence generally exhibits a wide distribution of pore diameters. Also, a quasi-glass structure could be identified due to calcium–aluminum–(magnesium) silicates being responsible for these pores’ distribution. Acid and alkali effects have important effects on mineral weathering.

High porosity may indicate low-temperature firing or degraded materials. An inverse or weakly inverse correlation is often observed: larger pores → smaller crystallites (due to less sintering, lower firing) and smaller pores → larger crystallites (well-fired, recrystallized phases). The tremolite sample exhibits a small average pore diameter of 4.7 nm and a relatively large crystallite size of 87 nm, suggesting a well-crystallized and compact fibrous structure, potentially indicative of high-temperature formation or minimal post-depositional alteration. Crystallites are coherent domains—so a size of ~87 nm means the mineral has undergone stable growth without major defects or breakdown.

Alkali attack induces marked changes in the crystal structure of aluminosilicate minerals due to dissolution of structural ions and/or rearrangement of the structure.

The textural properties of the samples were analyzed by nitrogen adsorption/desorption measurements, specific surface area, total pore volume, and maximum values in the pore size distribution. The impact of alkali treatment on minerals alters the chemical and structural properties of mineral surfaces. Alkali treatment makes minerals more polar, due to the formation of finely dispersed and highly energetic magnesium (Mg) or iron (Fe) oxide precipitates and/or the removal of outer silica sheets from mineral structures, leaving alumina sheets behind, which are more polar than silica. Also, alkali attack leads to the creation of new chemical groups and energetic centers on the mineral surfaces that can significantly alter how minerals interact with other substances.

However, alkali treatment may also alter the geometrical features of the mineral surfaces and the micropore structure, affecting the adsorption energies (the energy required to adsorb molecules onto the surface) and the distribution of adsorption sites.

In the cases studied in this paper, the specific surface area, total pore volume, and the pore diameters are more or less the same at constant Na₂O or K₂O concentrations. In this context, it could be presumed that the tremolite-asbestos samples identified above have similar textural properties at similar alkali concentrations.

Meanwhile, in Roman mortars K₂O and Na₂O have an important role. Alvarez et al. [50] reported that the addition of alkali as Na₂O and K₂O led to a higher durability, with a high repair capacity for built heritage. The higher the K₂O concentration, the lower the surface area value and the higher the Na₂O concentration, the higher the surface area value.

The reduced concentrations of K₂O and Na₂O in ancient Roman mortars lead to lower water absorption through capillarity, which enhances their durability against freezing and thawing cycles. These materials used in the middle layer could be considered as mortars and they exhibit textural properties characterized by the presence of small-diameter mesopores (2–5 nm), significantly smaller than those found in modern hydraulic mortars, which may contribute to their unique water adsorption characteristics and overall mechanical properties, resulting in higher durability and resistance.

4. Discussion

The Roman plastering technique effectively rendered walls waterproof, significantly enhancing their durability. A key technological advancement in the production of Roman mortars was the incorporation of both natural and artificial pozzolanic aggregates, which enhanced the material’s hydraulic properties, as well as its strength and durability under various environmental conditions [51]. In addition, this type of mortar was probably applied to improve the waterproofing.

Tremolite is a type of amphibole asbestos, a naturally occurring mineral composed primarily of calcium, magnesium, silicon, oxygen, and hydrogen. It is part of the amphibole group of silicate minerals and can occur in both fibrous and non-fibrous forms. When in its fibrous form, tremolite asbestos is particularly hazardous to health.

Tremolite asbestos is white to gray or greenish, with thin, needle-like fibers that can easily become airborne and be inhaled. It is crystalline and typically forms in long, straight, sharp fibers. In high-magnification or SEM images, tremolite-asbestos fibers usually look thin and elongated, often bundled or intertwined, with a smooth or slightly striated surface. Tremolite asbestos was identified in the wall layers, considering that this mineral was accessible in the area and was known for its magnificent properties. Or, during the reconstruction period, it was necessary to use a fireproof layer for the projection of the walls and especially of religious constructions. This is why, according to our expertise, the construction materials from these Roman monuments showed the presence of this asbestos mineral.

XRD analysis indicates that the mineral paragenesis in the matrix of these samples consists primarily of tremolite, albite, and anorthite, as major components, suggesting a complex interplay of these minerals within the geological context. The chemical composition of the middle layer materials put into evidence the presence of volcanic aggregates with high SiO₂ + Al₂O₃ + Fe₂O₃ values (higher than 80%) vs. the corresponding decreased CaO values (5–8%).

By using FTIR spectroscopy, a combination of bending (~600–700 cm⁻¹) and stretching (~1000–1100 cm⁻¹) of silicate units that forms a characteristic pattern was possible. FTIR can detect fibrous vs. non-fibrous phases and identify tremolite.

By Raman spectroscopy (782 nm laser), it was possible to put into evidence the main bands of asbestos (1000–2000 cm⁻¹). The most distinct peaks are from 1200–1400 cm⁻¹ and are assigned to the Si-O-Si bridges (symmetric stretching vibrations).

The interpretation of TGA is in correlation with the analyzed materials, taking into account both their composition (low CaO content), highlighted by the other analyses, and their age. It is observed that the investigated samples show no mass variation at temperatures higher than 600 °C, except few of them which show minor mass loss around 800 °C (measured up to 800 °C).

The tremolite sample exhibits a small average pore diameter of around 4.7 nm and a relatively large crystallite size of around 87 nm, suggesting a well-crystallized and compact fibrous structure, potentially indicative of high-temperature formation or minimal post-depositional alteration.

The characteristics of the Micia mortars are similar to those found at other Roman sites in the Mediterranean region, attesting to the continued use of the mortar manufacturing techniques outlined by Vitruvius, not only in Italy but also in Slovenia, Tunisia, Turkey, Malta, and Romanian territories [3,40,52,53,54].

These are black-gray materials that contain alkali ions, potassium ions (K⁺), and sodium ions (Na⁺) as predominant cations, while hydroxyl ions (OH⁻) play a crucial role in balancing the charge within the solution. The high abundance of K⁺ and Na⁺ contributes to the overall ionic strength and chemical properties of the pore fluid, with hydroxyl ions primarily serving to maintain charge neutrality.

Also, the alkalis present in the aggregates increase the hydroxyl ion concentrations, leading to high alkalinity the solution, which is responsible for the dissolution of silicate anions.

5. Conclusions

Highly sensitive and performant analytical techniques have been employed to identify the tremolite-asbestos species and analyze the structure, composition, and morphology of these minerals within the building materials of the Roman Micia monuments. These techniques include optical microscopy (OM), stereomicroscopy (SM), scanning electron microscopy (SEM), X-ray diffraction (XRD), wavelength dispersed X-ray fluorescence (WDXRF), FTIR and Raman spectroscopy, thermal analysis (TGA/DTA), and nitrogen adsorption/desorption measurements. After identifying the asbestos species, it could be presumed that this species was used after the fire in the Micia settlement in order to protect these monuments.

The microscopy investigation of these samples put into evidence the presence of long fibers, translucid and flexible, and elongated prismatic and fibrous crystals, with a wavy structure, belonging to tremolite asbestos. This species has been identified by XRD, XRF, and FTIR (by the bands from 1000–1600 cm⁻¹).

By using a 782 nm laser with Raman equipment, it was possible to put into evidence the main bands of asbestos (1000–2000 cm⁻¹). Also, the carbonated and silicate species have been prominently identified in some samples by FTIR and Raman spectra.

Tremolite-asbestos types were identified during the reconstruction period, as it was necessary to use a fireproof layer for the projection of the walls and especially of religious constructions. With a significant concentration of K₂O and Na₂O, the slightly hydraulic mortars were identified with an increased durability. The higher the K₂O concentration, the lower the surface area value and the higher the Na₂O concentration, the higher the surface area value.

The materials (repairing mortars) analyzed at Micia exhibit characteristics consistent with those observed at various Roman sites throughout the Mediterranean, indicating a sustained adherence to the mortar production methods outlined by Vitruvius. This continuity in technique underscores the influence of classical architectural principles on Roman construction practices across the region.