Petrographic characterization and durability of carbonate stones used in UNESCO World Heritage Sites in northeastern Italy

This paper presents the petrographic and textural characterization of some ornamental limestones widely used in UNESCO World Heritage Sites in northeastern Italy, and the assessment of the main decay factors present in the environment where they are employed. Eleven carbonate building materials have been here considered, all commonly present in the built environment of northeastern Italy: two different varieties of Vicenza Stone (Nanto and Costozza), of Verona Stone (Red and Brown Verona), of Asiago Stone (Pink and White Asiago), and of Chiampo Stone (Ondagata and Paglierino), the Istria Stone (Orsera), the Aurisina Stone, and the Botticino Stone. The Carrara marble is also considered, and used as a reference material for the determination of the grain-size distribution. Stone durability was measured by accelerated aging tests which reproduced freeze–thaw and salt crystallization cycles, among the main causes of deterioration in the region. Petrographic and textural features of these carbonate rocks as well as their porosity resulted to strongly influence their deterioration rate, and their knowledge is, therefore, essential when trying to predict stone decay as a function of the local environmental forcings.


Introduction
Rocks are important building raw materials used as dimension and ornamental stones since prehistoric times. They are susceptible to deterioration by various physical, chemical, and biological agents. Damage may also derive from human contact, such as walking on stone floors, or from accidental or deliberate actions. Stone deterioration rate is controlled by environmental factors (e.g., rainfall, relative humidity, temperature, air pollution) and petrographic features, i.e., mineralogical composition and texture (Salvini et al. 2022, and references quoted therein). Restoration and conservation practices often employ rock types that are esthetically similar to the original building stone, when this is unavailable, the source is unknown, or the replacement stone is more resistant to deterioration at current environmental conditions (Graue et al. 2011;Apostolopoulou et al. 2019). Hence, a good knowledge of stone resources and their characteristics (petrography, mineralogy, historical use, decay behavior, exploitation techniques) is required. The available published data on the petrographic and physical-mechanical properties, and on the durability of building stones are copious. These materials differ for their composition (e.g., carbonate vs. silicate), geological origin (e.g., sedimentary, igneous, metamorphic), and textural features (e.g., porosity, grain size, anisotropy due to foliation, stratification, etc.) which all affect their bulk behavior (e.g., Sousa et al. 2005;Molina et al. 2013;Freire-Lista et al. 2015, 2021Germinario et al. 2015;2017;Sousa et al. 2018;2021;Çelik and Sert 2020;Zalooli et al. 2020;Sitzia et al. 2021;Pötzl et al. 2022). They are generally either part of the material culture of a specific region or they experienced, and still have, a worldwide diffusion as building materials, such as numerous Egyptian stones, Greek, Turkish, and Italian marbles, just to give a few examples (Lazzarini 2006(Lazzarini , 2019Harrell 2008;Antonelli and Lazzarini 2015;Al-Bashaireh 2021). The durability of stone building materials is determined by accelerated aging experiments, most commonly freeze-thaw and salt crystallization tests (Molina et al. 2013;Germinario et al. 2017;Sousa et al. 2018). In some cases, the thermal properties and the resistance to stresses related to specific applications such as cladding have been also assessed Coletti et al. 2021;Pires et al. 2022).
Since the prehistoric age, carbonate rocks have been the most diffused dimension stones in the built environment in many regions of the world due to their wide availability and ease of work compared to silicate rocks, often as locally exploited, exceptionally, and generally as decorative stones, transported over long distances (Siegesmund and Snethlage 2014). Starting from the nineteenth century, due to the development of motorized technologies for stone processing and the improvement of transportation capabilities, granitoid and other silicate rocks, less subjected to weathering than carbonate rocks, conquered an increasing market share.
Transportation costs have always discouraged the commercialization of dimension stones over long distances so that often specific rock types characterize certain cities or specific cultural landscapes (Siegesmund and Snethlage 2014) such as the Carrara marble in northern Tuscany (central western Italy), the Lecce Stone in Salento (Apulia, southern Italy), the Globigerina Stone in Malta, the Rosso Verona Stone in northeastern Italy, etc., for which some have been designated (Carrara marble, Maltese Globigerina Limestone; Primavori 2015; Cassar et al. 2017) by the Executive Committee of the International Union of Geological Sciences (IUGS), through its sub-commission on Heritage Stones (HSS), or nominated (Lecce Stone, Rosso Verona marble) as suitable Global Heritage Stone Resources (Primavori 2019; 2020).
This represents an important aspect of the cultural heritage that should be safeguarded from excessive contamination from imported stones, making deterioration issues of local dimension and ornamental stones even more critical. Nowadays, the use of locally available limestones for cladding on modern buildings is growing as designers are considering these materials as part of the local cultural identity and a link between the present and the past (Mariani et al. 2018;Kotradyová 2019), as well as a way to sustain a ecofriendlier local economy by reducing carbon footprint which is considerably higher for stone imported due to the impact of transport (Crishna et al. 2011;Capitano et al. 2017).
In the whole southern Europe, local stones have been widely employed as building and ornamental materials for historical and modern constructions, such as church embellishment, and urban decor (monuments, paving, fountains, façade coverings, etc.). In northeastern Italy, various carbonate stones with suitable characteristics crop out in the Alpine foothills (Southern or Italian Prealps) and have been quarried since antiquity, the most representative of which have been considered in this work.
Considering as a broad overview of the climate conditions the map of the Köppen-Geiger climate types updated and digitalized by Kottek et al. (2006), northeastern Italy is characterized by a temperate mild-to-warm fully humid climate with warm-to-hot summers (Cfab). Yearly average rainfall is about 1000 mm, with precipitation even during the driest month. According to Brimblecombe (2010), the climate conditions relevant to the weathering of architectural stones under this Köppen-Geiger climate type are dump conditions with variation in humidity, favoring salt crystallization cycles, and occasional temperature below freezing point in wintertime, with potential for frost weathering. In addition, dump conditions and stone wetting connected to direct wind-driven rain, fog or vapor condensation, favor uptake and reaction of air pollutants on stone, triggering different deterioration mechanisms including stone dissolution, salt mobilization, and formation of crusts on the stone surface (Scherer 1999;Orr et al. 2018;Camuffo 2019;Sesana et al. 2021). These correspond to the main environmental factors identified by Sabbioni et al. (2012) as more relevant for carbonate stone deterioration in Europe. In the frame of the NOAH'S ARK Project, starting from annual maps of precipitation, precipitation frequency, wind-driven rain, and frost, Sabbioni et al. (2012) modeled the changes from recent past to near and far future of salt crystallization frequency, wet-frost index (number of rainy days with temperature below freezing point), biomass accumulation according to different future climate scenarios, and consequent surface recession and thermoclastism of low-porosity carbonate stones. Knowledge of stone behavior after freeze-thaw and salt crystallization cycles is, therefore, essential to evaluate ability of the dimension stone to endure and resistance to decay under the environmental conditions typical of the temperate regions. Ice and salt growth within the pore system have a detrimental impact on the material through crystallization pressure, causing stress perturbations and fracture propagation, leading to deterioration of stone mechanical properties and eventually affecting its esthetic quality; these processes are significantly controlled by specific textural features of the rock such as porosity, pore-size distribution, and content in clay minerals (Benavente et al. 2004(Benavente et al. , 2007a(Benavente et al. , 2007bMolina et al. 2013;Martínez-Martínez et al. 2013;Benavente et al. 2018;Salvini et al. 2022), the effectiveness of which needs to be determined to reliably evaluate stone vulnerability under specific environmental forcing through specifically designed laboratory programs of accelerated aging tests and field exposure tests (Germinario et al. 2015Vidorni et al. 2019;Salvini et al. 2022).

Materials
Twelve lithologies have been chosen among the limestones commonly used in the built heritage of northeastern Italy ( Fig. 1): Vicenza Stone (Nanto and Costozza variety), Istria Stone (Orsera variety), Chiampo Stone (Paglierino and Ondagata variety), Red and Brown Verona Stone, Pink and White Asiago Stone, Aurisina Stone, Botticino Stone and Carrara marble. These stones come from different geological formations cropping out in northeastern Italy, Croatia (Istria Stone), and in the Tuscany region (Italy) (Fig. 2). A summary of the main geological outlines and lithological classification of all the considered rock types is reported in Table 1. In this study, the Carrara marble is used as a reference material to test the reliability of optical microscopy in determining the grain-size distribution by comparing results with those obtained by EBSD (Mineralogy).

Vicenza Stones (Nanto and Costozza varieties, NA and CO)
Vicenza Stone (Pietra Tenera, or "Soft Stone", of Vicenza) is a general name used to indicate various biocalcarenitic limestones quarried in the Berici Hills, extending over an area of about 165 km 2 south of Vicenza, characterized by light colors with ivory, light grey, straw yellow, warm light brown hues. Lithostratigraphically, Vicenza Stone belongs to the early-to-middle Eocene Nummulitic Limestone Formation and the Oligocene Castelgomberto Limestone Formation (Cattaneo et al. 1976;Benchiarin 2007;Cappellaro et al. 2012). They derive their popular local name of "Soft Stone" from their soft texture, which makes them easy to cut and carve, favored by the high porosity and the water content when freshly extracted. Visible macropores (> 200 μm, according to the nomenclature reported in Siegesmund and Dürrast 2011) are abundant and the content in large bioclasts is high. The typical weathering forms consist in dissolution and pulverization, accompanied by an often-important biological colonization; humidity marks and black crusts are often present. Due to its high specific surface determined by porosity and the presence of thin clay layers containing swelling micas such as montmorillonite, Vicenza Stone is liable to deteriorate (Marchesini et al. 1972;Cattaneo et al. 1976;Fassina and Cherido 1985;Ginevra et al. 1999;Benchiarin 2007;Benchiarin et al. 2012;Di Benedetto et al. 2015).
In the present work, the Nanto and the Costozza varieties (samples NA and CO, respectively) have been considered.
Nanto Stone is a type of Vicenza Stone quarried near the town of Nanto, along the eastern margin of the Berici Hills, from the layers placed at the bottom of the Nummulitic Limestone Formation (Massari et al. 1976;Benchiarin 2007). This variety is also known with the commercial name of Giallo Dorato (i.e., "Golden Yellow") due to its warm yellowish-brown hue. Its texture is characterized by isoriented large bioclasts (mainly macroforaminifera) in a golden sand-sized matrix rich in planktonic foraminifera.
Costozza Stone is the traditional light ivory variety of White Vicenza Stone. It is a packstone rich in encrusting coralline algae, large foraminifera, bryozoans, and echinoids with variable amounts of miliolids, which create its thin and flowery texture. It is quarried from the stratigraphically higher layers of the Castelgomberto Limestone Formation, which includes back-reef tidal channel deposits of Oligocene age.
Vicenza Stone was largely used in historical masonry and in statuary, as admirably testified by the Olympic Theatre, the Palladian Villas and the Lodges of the Basilica Palladiana in Vicenza (all part of the UNESCO World Heritage Site "City of Vicenza and the Palladian Villas of the Veneto"), the 78 statues located in Prato della Valle square in Padua (Braga 2004;Cornale and Rosanò 1994) (Fig. 1S), the Cathedral of Modena and the Ghirlandina (belonging to the UNESCO World Heritage Site "Cathedral, Torre Civica and Piazza Grande, Modena"). This stone was also used in the Mysterious Baths by De Chirico in Milan, and in numerous twentieth century buildings, both for internal and external cladding, internal floors and staircases, and other building components. Thanks to its fire resistance, it has been traditionally used in fireplaces (Rodolico 1953).

Aurisina Stone (AU)
Aurisina Stone (sample AU) is a general term used to indicate several varieties of biomicritic and fossiliferous limestones (from grainstone to wackestone) quarried in the Aurisina quarrying district since the Roman Age (e.g., from the top portion of the so-called "Roman Quarry"; Maritan et al. 2003), some hundred meters from the coastline, about 10 km north of Trieste. These limestones crop out in the Classical Karst Region of northeastern Italy and southwestern Slovenia, and belong to the carbonate sequences of the northwesternmost part of the External Dinarides. Stratigraphically, the Aurisina Limestones belong to the late Cretaceous Sežana (late Turonian-early Santonian) and Lipica (Santonian-Campanian) Formations (Cucchi et al. 1987(Cucchi et al. , 2015Jurkovšek et al. 2016), the latter including the most economically relevant layers, constituted by massive beds with homogeneous texture of biomicrite and biosparite with abundant rudist bioclasts, deposited in a shallow-water platform from low-energy littoral and lagoonal to high-energy open shelf environments.

Istria Stone (Orsera variety, OR)
The Istrian peninsula is part of a vast mainly Mesozoic carbonate platform with limestones ranging in age from late Jurassic to Eocene, and structurally forming an anticline, the core of which crops out in mid-western Istria, where the Kirmenjak unit (late Tithonian) transgressively lies on the late Kimmeridgian-early Tithonian Rovinj bauxite (Durn et al. 2003). The Kirmenjak unit consists in three members: a basal member of clay, breccia, and mudstone, followed by a second member of thick-bedded stylolitic mudstone, and a third member made of fenestral mudstone. Stylolitic mudstones were deposited in a low-energy subtidal to low intertidal environment, forming a 65 m thick course which provided the best quality and most valuable "Istrian limestone". Although thousands of quarries are known from the Istrian peninsula, the most representative material used in the Venetian architecture derives from this course, which was thoroughly exploited in the two largest known quarries near Vrsar (Orsera) and Rovinj (Lazzarini 2012). In the present study, we consider the Orsera variety (Kirmenjak Stone, from the Montraker quarry) (sample OR), characterized by a salt white-pale grey color, very compact texture, low porosity, and conchoidal fracture, with compressive strength and density approaching those of marble (Geometrante et al. 2000).
The use of the Istria Stone in Roman times is limited to Istria, with rare examples not further than Aquileia (Lazzarini 2012). It is unknown when the Istria Stone was first used in Venice, but quarrying activity was already well established in the thirteenth century when numerous statues were realized It has been calculated that about 80% of stone employed in Venice buildings derived from Istria region (Dunda and Kujundžić 2004). The technical features of this stone, consisting in its high durability, extremely low porosity and water absorption, high compressive strength, and resistance to the Venetian salt-rich environment, made this stone the ideal building material to be used for the foundations directly on top of the timber boards surmounting the timber piles. This stone layer ended up with a continuous course of stones called "cadene" above the average sea level (Foraboschi 2017), with the stylolitic layers oriented horizontally truncating the capillary water paths, and creating a multilayer moisture barrier (Šimunić Buršić et al. 2007;Lazzarini 2012). In addition, like any other compact limestone, Istrian Stone is difficult to be carved (even more than marble) because it tends to splinter easily and presents conchoidal fracture, especially when worked parallel to the joints; on the other hand, this stone can be easily polished (Lazzarini 2012). In the city of Venice (UNESCO World Heritage Site "Venice and its Lagoon"), numerous buildings are completely covered with Istria Stone, such as the Doge's Palace (fourteenth century), the Rialto Bridge and the Bridge of Sighs (end of the sixteenth century), while a modern example is that of Santa Lucia railway station, built in the first half of the twentieth century. Orsera Limestone was also used in Dubrovnik (Croatia) together with the varieties quarried in Vrnik and Korčula (Crnković and Jovičić 1993), for the Basilica Euprasiana in Poreč, Croatia (sixth century), for the Mausoleum of Theodoric in Ravenna (sixth century) (all UNESCO World Heritage Sites), for the Cathedral of Fermo (Marche region, central Italy; reconstructed at the beginning of the thirteenth century), for the Tomb of Dante in Ravenna (a neoclassical monument built in the second half of the eighteenth century), for the Monument to the Fallen of the First World War in Ancona, for the ancient "kažuni", traditional stone cottages of the Istrian peninsula that served as shelter for shepherds ( Fig. 2S), just to mention some remarkable examples (Dunda and Kujundžić 2004). The use of the Istria Stone has been documented also in Turin, where some ornamental elements, and interior and exterior facings of the San Vincenzo block of via Roma are made of this stone (Borghi et al. 2015).

Chiampo Stones (Ondagata and Paglierino varieties, ON and PA)
Chiampo Limestone refers to a compact facies of the Nummulitic Limestone Formation (early-to-middle Eocene), which is ~ 30 mt thick, and crops out in the Venetian Prealps, more specifically in the area of the eastern Lessini Mountains, along both the flanks of the Chiampo Valley (Papazzoni and Sirotti 1995; Matteucci and Russo 2005), intercalated among late Paleocene-Eocene mafic tuffs, some of which containing an important fauna of three-dimensionally preserved siliceous sponges (Frisone et al. 2020).The commercial names, Chiampo Marble or Chiampo Stone, refer to different varieties of this sedimentary rocks which generally consist in rudstones rich in macroforaminifera, calcareous algae, rare scleractinian corals, echinoids, and mollusks, with light brownish-grey to pale pink color. This study considers the Paglierino and Ondagata varieties. Chiampo Stone was largely appreciated by the Italian architect Vincenzo Scamozzi (1548-1616) because of its whiteness, granularity (similar to that of Vicenza Stone), strength, and compactness.
The Ondagata variety (sample ON) is characterized by a pinkish hue and evident wavy stratification. It can be perfectly polished even if the stone texture is not uniform as it is often composed of an aggregate of fossil remains, especially Nummulites and Discocyclina. Currently, the most precious type is the light brownish-grey Paglierino variety (sample PA), which is compact and homogeneous, with barely visible stratification. Chiampo Limestone has been used mainly for small decorative elements (stairs, wells) but also for monuments like the two monolithic columns in the "Piazza dei Signori" city square of Vicenza (Rodolico 1953). Between the 1920s and the 1950s, the Chiampo quarries have been extensively cultivated for the production of the headstones used in the first World War CWGC (Commonwealth War Graves Commission) cemeteries in Italy, the Niguarda Hospital in Milan, the Exchange Palace (Palazzo della Nuova Borsa Valori) in Genoa, the Courthouse and the Finance Palace in Bolzano, the "Palazzo Grande" in Livorno, the Shrine of the Báb and the International Archives, two of the buildings in the Bahá'í World Centre in Haifa (Israel) (Fig. 3S).

Verona Stones (Red Verona and Brown Verona marble, RV and BV)
Red and Brown Verona Marbles (samples RV and BV, respectively) are commercial names used to indicate limestones with different reddish hues pertaining to the Rosso Ammonitico Veronese (RAV) Formation (middle-late Jurassic), which crops out extensively in the Venetian Prealps. This work considers both the Red and Brown Verona varieties. RAV is a very distinctive lithostratigraphic unit in the Mesozoic deposits of the Trento Plateau, a submarine structural highly characterized by very low sedimentation rates (Winterer and Bosellini 1981;Martire et al. 2006;Lukeneder 2010Lukeneder , 2011. Sometimes, due to its high request, the stone is also quarried from the Scaglia Rossa Formation, which is less valuable. Petrographically, the Verona Stone is a nodular ammonite-bearing limestone characterized by the juxtaposition of light-colored nodules and anastomosing darker layers richer in clay and crossed by dissolution seams. Grain size is fine, and the color is variable from red to yellow or white. The biogenic component is prevalent, with skeletal rests mainly of planktonic and benthic foraminifera, radiolarians, calcispheres, and thin-shelled bivalves (Benchiarin 2007;Martire et al. 2006).
This type of stone, also called "Veronese marble" or "Verona Broccatello limestone", was well known by Leonardo da Vinci, who reported that "Truovasi nelle montagne di Verona la sua pietra rossa mista di tutti i nichi convertiti in essa pietra" (In the mountains of Verona, there is a red stone mixed with all the shells converted in the same type of stone). This stone has been widely used as ornamental and dimension stone since ancient times. For example, in Verona, the Roman amphitheatre known as Arena, Ponte Pietra (Stone Bridge), and the Roman Theatre have been erected using this stone. In Venice, the Red Verona Stone has been used for architectural details such as capitals, columns, floor slabs, architraves, cornices, and doorframes, such as the internal floor and architectural elements of the portals on the western façade of St. Mark's Basilica, and well curbs such as in the courtyard of Ca' d'Oro (early fifteenth century). In association with Istrian Stone, Red Verona Stone provided the characteristic two-toned external ornamentation of the Doge's Palace and many other buildings in Venice. The use of architectural elements in Red Verona Stone is common in many of the major cathedrals of Northern Italy such as in those of Cremona, Parma, and Bologna. The lion sculptures supporting columns at the entrance of the Basilica of Santa Maria Maggiore in Bergamo ( Fig. 3S) have been also realized with Red Verona Stone. In the last century, its diffusion in the global market allowed the Red Verona Stone (sometimes as Rosso Ammonitico Stone) being enumerated among the stones of interest for urban geoturism itineraries, contributing to the geodiversity in urban cultural spaces in cities from other continents (e.g., in Rio de Janeiro; Polck et al. 2020).

Asiago Stones (Pink and White varieties, RO and BI)
The typical lithofacies of the Asiago Stone pertains to the Maiolica Formation (also reported as Biancone Formation in older literature), formed in the early Cretaceous on the Trento Plateau, and stratigraphically lying on the Rosso Ammonitico Formation. It consists in pelagic thin-bedded white micritic limestone containing nodules and lenses of cherts, rich in microfossils of radiolarians, ostracods, echinoids, sponge spiculae, and foraminifera (Martire 1996;Lukeneder 2010Lukeneder , 2011. Stylolites are common parallel to layering. Macroscopically, some facies are similar to Red Verona, but normally the color is paler and texture is less nodular. This research considers both the Pink and White Asiago Stone varieties (samples RO and BI, respectively).
The stone is more frequently used for small decorative elements in residential buildings of the Veneto Region (windowsill, doorframe, tiling), but there are examples of use for "noble purposes" like the new altar in the Basilica of Saint Anthony in Padua, where the Pink Asiago is coupled with Orsera Stone, and the Asiago War Memorial (Fig. 3S).

Botticino Stone (BO)
Botticino Stone (sample BO) is an ornamental stone quarried in the Botticino, Rezzato, and Mazzano area, east of Brescia (northern Italy), mostly from open-pit sites. It consists in partially dolomitized stylolitic compact micritic limestones forming thick ivory, grey or cream-colored beds. Lithostratigraphically, Botticino Stone belongs to the Corna Formation (early Jurassic), and comprises carbonate deposits formed in a subtidal platform environment with episodes of temporary emersion marked by up to centimetric illite-rich layers influencing cultivation operations. These limestones show a variety of textures ranging from desiccated mudstone to floatstone and wackestone containing oncolites, dasycladacean algae, and red algae, and other bioclasts which are generally hard to determine due to the intense dolomitization (Schirolli 1997;Di Battistini et al. 2005;Vernia et al. 2005;Borghi et al. 2015;Masetti et al. 2017). Quarrying activity in the area also involved limestones now formalized in the overlying "Ercinite di Rezzato" and "Corso Rosso di Botticino" stratigraphical units, representing the progressive drowning of the Corna platform, with increasingly pelagic and deep-sea facies, consisting in cream-colored and then reddish mudstones and wackestones containing peloids, crinoids, echinoids, and increasing amounts of sponge spicules, with frequent chert nodules (Schirolli 2007).
The variety considered in this study is that of the Classical Botticino Marble (in commercial names, limestones are commonly called marbles, although strictly incorrect geologically speaking), the most diffuse type. This stone has been used since Roman times, especially in Brescia (in antiquity Brixia), where the quarries are located. Here, Botticino Stone was used in the construction of the Roman Temple of the Emperor Vespasian (Capitolium of Brixia) and the Theater (first century), the Church of Our Ladie of Miracles (fifteenth century), the Loggia Palace (fifteenth century), and the New Cathedral (seventeenth-nineteenth century) (Clerici and Meda 2005). At the end of nineteenth century, the use of this building stone began to spread in the whole Italy and abroad (Fig. 4S): the Vittoriano monument in Rome, the Brussels-North railway station, the Victory Monument in Bolzano, the Palace of the Commercial Italian Bank in Milan, the White House in Washington, the Grand Central Terminal and the basement of the Statue of Liberty in New York, the Chamber of Commerce in Osaka, the Red Sea Hospital in Jeddah, the One International Place in the Financial District of Boston, are all realized using this stone.

Carrara marble (M)
White Carrara marble is the only metamorphic rock considered in this research. We used this stone as a reference material to evaluate the reliability of determining the grainsize distribution by optical microscopy (OM), by comparing with results obtained by electron backscattered diffraction (EBSD).
Carrara marble has a large number of commercial varieties. These rocks derive from original Mesozoic carbonate sequences of the Tuscan Nappe which underwent low-grade metamorphism, and now they crop out in the Apuan Alps (Carmignani et al. 1978;Meccheri et al. 2007;Borghi et al. 2015). The Apuan marbles are the most famous Italian marbles. Although they have homogeneous chemical and mineralogical composition, different Apuan white marbles show different behavior when exposed to external environment (Cantisani et al. 2009).
White Carrara marbles have a pearly white color, fine-tomedium grain size; their color is homogeneous or spotted with small patches and gray veins, irregularly distributed due to microcrystalline pyrite. Carrara marble has been extensively used since the Roman Age. Important monuments of Ancient Rome as well as Renaissance sculptures are made of this material, such as the Pantheon, Trajan's (113 CE) and Marcus Aurelius' Columns (180-192 CE), Michelangelo's Pietà (1499 CE) and David (1501-1504 CE) (Fig. 4S). The importance of these monuments and the availability of raw material made this stone a primary choice for relevant monuments and buildings all over the world. The Marble Arch in London (nineteenth century), innumerable sarcophagi of nobles around Europe, the Cathedral of Manila, the Opera House in Oslo, the Finland Hall in Helsinki, the Prem Mandir Hindu Temple in Vrindavan (India), the Sheikh Zayed Grand Mosque in Abu Dhabi, Harvard Medical School in Boston, various other buildings in the USA and Canada, and the fountain Bassin de Madrid in Paris are on the long list of prominent buildings which largely utilized this marble. As other stones, Carrara marble has been largely used also in the territories close to the quarry district (i.e., the Apuan district and Tuscany). For example, the Cathedrals of Massa and Carrara as well as the Ducal Palace of Massa are entirely decorated with Carrara marble cladding; in the whole Tuscany, churches are often decorated combining Carrara marble with other colored ornamental stones (e.g., the Cathedral, the Baptistery, and the Giotto's bell tower in Florence).

Analytical methods
All the samples were analyzed under a petrographic viewpoint using a polarizing optical microscope (OM) Zeiss® AxioScope.A1 coupled with an AxioCamMRC5 camera connected to a PC running the AxioVision Red 4.8 software package. This equipment was also used to estimate the average size of the calcite grains in the different textural elements recognized in thin section (e.g., micrite, coarsegrained micrite, sparite, bioclasts), by measuring crystal diameter on a number of grains between 10 and 712. Each grain was manually selected and its boundary recognized in cross-polarized light from the difference in optical orientation and cleavage traces direction of the adjacent grains, or from the interface refraction at the grain boundary. A thin section of Carrara marble was Syton polished and analyzed with an electron backscattered diffraction (EBSD) system, installed on a CamScan MX 2500 scanning electron microscope (SEM) and equipped with a NordlysNano detector and the HKL-Channel5 acquisition and post-processing software (Oxford Instruments), to determine the grain-size distribution and compare these results with measurements obtained by optical microscopy on the same thin section. The hit rate of the automatically acquired EBSD map of Carrara marble was 97.6%, and grain-size measurements were determined automatically using the "grain size" component of the post-processing software, where grain-size has been determined on the basis of misorientation exceeding 10° between adjacent grains. Size of the grains has been measured both as area (m 2 ) and diameter (m).
Mineralogical composition was determined by X-ray powder diffraction (XRPD) on a Philips X-Pert PRO diffractometer, in Bragg-Brentano geometry, equipped with a Cu X-ray tube, operating at 40 kV and 40 mA and a solidstate detector (X'Celerator). The powders (about 1 g) of the 12 studied samples were mounted on a circular sample holder (32 mm Ø). Scans were performed in the range 3° to 80° 2θ with an integrated step size of 0.017° 2θ and a counting time of 1 s per step. Identification of minerals, quantification, and cell parameter determination were performed using the reference intensity ratio (RIR) method implemented in High Score Plus and the ICSD database (PANalytical).
Mercury intrusion porosimetry (MIP) was performed using a PoreMaster 33 system (Quantachrome Instruments ® ) with the following parameters: sample cell is 1.0 × 3.0 cm in size and 2 cm 3 in volume, pressure range is 0.5-33.000 psi, contact angle (θ) of mercury is 140°, and surface tension (σ) of mercury is 0.48 N/m (480 dyn/cm), pore size ranges from 0.0064 to 950 μm. The value of the density of materials was measured through a Ultrapic 1200 pycnometer (Quantachrome Instruments ® ). Before measurements, samples were dried at 40 °C for 24 h and then about 2 g (i.e., a little core with diameter of 8 mm and 2 cm long) of material were analyzed.
Water absorption (WA) was tested on six cubic samples (50 mm edge) for each rock type (UNI EN 13755, 2008). The test consisted in determining the percentage of water absorbed by the mass of the sample over time. Samples were periodically weighed (every 24 h) until they reached constant mass.
Open porosity (na) was calculated as.
where M 0 is the mass of the dried sample, Ms is the mass of the sample saturated with water under vacuum, and MH is the hydrostatic weight of the sample saturated with water under vacuum. Capillarity rise was qualitatively determined on the wetted level on a side of cubic-shaped samples (50 mm edge) for each rock type by visual observation of time lapses images (duration = 30 min).
Uniaxial mechanical tests were carried out on six cubic samples (50 × 50x50 mm) for each rock type on an ADVANTESTS9 (Controls Italia s.r.l.) press with a load force of 1 MPa/s, according to UNI EN 1926(2007. Frost resistance was evaluated by the freeze-thaw test on six cubic samples (edge: 50 mm) for each rock type through 50 cycles of 24 h (UNI EN 12371, 2003).
Durability of stones was determined by salt crystallization (UNI EN 12370, 2001) and freeze-thaw cycles (UNI EN 13755, 2008). Resistance to salt crystallization was determined on six cubic samples (edge: 50 mm) for each rock type. Samples have been subjected to 15 cycles of 24 h each (UNI EN 12370, 2001). During both the aging tests, weight loss was daily measured on samples after each cycle according to: in which W x is the weight after x cycles and W i is the initial weight.
Ultrasound propagation velocities Vp (compression pulse) were measured on the rock cubes at regular intervals during the freeze-thaw (every five cycles) and salt crystallization (every two cycles) tests, to measure Vp values on the three orthogonal directions (Vp 1 , Vp 2 , and Vp 3 ), to evaluate salt and ice damage to the pore system and texture, to determine the variation in structural anisotropy, and to identify any changes in the degree of compactness during and after the aging tests (Molina et al. 2013;Coletti et al. 2018). Measurements were performed using a portable MATEST C369N pulser-receiver coupled with transducers of 55 MHz over a circular contact surface of 3 cm in diameter. A viscoelastic couplant (an ultrasound eco-gel) was used for good coupling between the transducers and the rock surfaces.

Optical microscope observation and grain-size estimation
The analysis under optical microscope pointed out a great heterogeneity among the types of stone considered here. Photomicrographs of the main textural features recognized in the various samples and characterized by different grainsize are reported in Figs. 3, 4. Synthetic results about the grain size of the different textural elements recognized in thin section are reported in Table 2.
Two different varieties of Vicenza Stone have been analyzed: the Nanto Stone (sample NA), and the Costozza Stone (sample CO). Nanto Stone (sample NA) is petrographically classified as a packstone or a rudstone (Dunham 1962), when grains above 2 mm are abundant (biomicrite, according to Folk 1959). It is characterized by a clastic-organogenic, and grain-supported texture. The matrix is mainly intergranular, with grain size of about 0.5 μm (fine-grained micrite) and 6.2 μm (coarsegrained micrite). In thin section, this rock appears rather compact, with pore spaces partially filled with pseudomatrix, more rarely with cement, suggesting compaction processes during diagenesis. The skeleton is mainly represented by bioclasts, where Nummulites, Discocyclina, and Operculina are prevalent over red algae and other benthic organisms (bryozoans, echinoids). Nummulites are characterized by the radial fibrous structure of the thick hyaline walls, with fibers 3-4 μm wide. Skeletal grains of benthic organisms are often micritized, with grain size below 1 μm. In this sample, a sparitic cement (average grain size = 45 μm) is rare (see Figs. 3a-b).
Costozza Stone (sample CO) is petrographically classified as a grainstone (biomicrite, according to Folk 1959) and it displays a clastic-organogenic grain-supported texture. Large bioclasts above several mm in length are common. Porosity is high and large pores are more frequent than in Nanto variety, partially filled by a thin layer of an isopachous calcite cement with an average grain size of 38 μm. Encrusting red algae (Rhodophyta) and bioclasts of benthic organisms (bryozoans, echinoids) prevail over Nummulites, Amphistegina, Operculina, and planktonic foraminifera (Miliolids). This suggests that the depositional environment was an inner and middle Aurisina Stone (sample AU) can be described as a rudstone, or a grainstone, when referring to the finer grain-size facies. Aurisina Stone is characterized by the abundance of large fragments of rudists (Hippuritidae and Radiolitidae), associated to finer grained bioclasts of red and green algae (Thaumatoporella spp.), and foraminifera, typically Accordiella conica, Bolivinopsis sp., Cuneolina sp., Dicyclina schulmbergeri, Keramosphaerina tergestina, Moncharmontia apenninica, Murgella lata, Rotorbinella scarsellai, although difficult to identify when micritized, more rarely miliolids and textularids, and cyanobacteria (Decastronema Micrite is present, with average grain size between 0.5 μm (fine-grained micrite) and 3.5 μm (coarse-grained micrite). Echinoids are also frequent and partially micritized. Sparry calcite is within the grain-size range of 25-35 μm (Fig. 3e). Textural features, and the presence of bioclasts of different origin, suggest alternation of environmental conditions from low-energy lagoon to high-energy pertitidal and open platform.
Istria Stone, and specifically the Orsera variety (sample OR), is a mudstone characterized by a compact micritic texture determined by fine-grained micrite where calcite crystals are generally smaller than 0.7 μm. This rock type is characterized by stylolitic discontinuities, and rare bioclasts of difficult identification. Small pores observable at the microscopic scale are very rare. Veins and irregularly shaped cavities are filled with sparite with average grain size of the calcite crystals of about 40 μm (Fig. 3f).
Both the varieties of Chiampo Stone here considered, Ondagata (sample ON) and Paglierino (sample PA), can be classified as rudstone, due to their grain-supported texture determined by large bioclasts of macroforaminifera of the genera Nummulites and Discocyclina, Miliolids, Rotaliids, red algae, echinoids, bryozoans, crustaceans, mollusks, and rare corals. The structure is compact and characterized by low porosity, with clotted peloidal micrite filling most of intergranular spaces. Micrite is made of grains generally smaller than 1.5 μm; some domains are characterized by coarse-grained micrite, with crystals having an average diameter of about 4-5 μm. Nummulites are constituted by characteristic radial carbonate fibers across the thick test walls. Sparite can be found as secondary filling phase within bioclasts or as cement further reducing most of residual intergranular porosity, forming coarse aggregates with average grain size of 41 μm (sample ON) or 45 μm (sample PA) (Figs. 3g-l).
The two varieties of Verona Stone here analyzed are the Red Verona (sample RV) and the Brown Verona (sample BV) ones. Petrographically, they are both classified as wackestone (or as sparse biomicrite). The matrix is composed of a micrite with calcite grains in average 1 to 3 μm thick, with domains characterized by a coarse-grained micrite of about 5 μm. Matrix also contains clay minerals sometimes concentrated along specific layers favoring accelerated deterioration mechanisms. Bioclasts consist in protoglobigerinids, thin-shelled bivalves, calcified radiolarian molds, Saccocoma, and sparse Calpionellids up to 80 μm in size (in the Red Verona Stone). Thin-shelled bivalves display a spine shape section about 1 mm long and 30 μm thick. These fossils optically behave like monocrystalline individuals (Figs. 4a-e).
Pink (sample RO) and White (sample BI) Asiago, the two varieties of Asiago Stone here considered, are mudstone or wackestone, depending on abundance of bioclasts, and are characterized by a fine-grained micritic matrix made of calcite crystals with an average dimension of 0.4 μm. Some domains present a coarser micrite, with calcite grains of about 4.3 μm. Bioclasts are dominated by Calpionellids with individuals of about 45-50 μm in size (Figs. 4f-h), while ostracods calcified radiolarians and ghosts of micritized pelagic foraminifera are sparse.
Botticino Stone (sample BO) is an intensely dolomitized floatstone. In this, thin section is characterized by large dolomite crystals with an average dimension of 127 μm forming domains with a crystalline structure surrounded by a micritic matrix in which fine-grained micrite portions with calcite crystals of about 1 μm are juxtaposed to coarsegrained micrite portions where calcite grain size is about 6 μm (Figs. 4i-j).
Carrara marble (sample M) is an almost pure calcite low-grade metamorphic rock with low amounts of accessory minerals (e.g., muscovite). It presents a homogeneous texture made of finely crystalline subhedral to euhedral calcite crystals (Fig. 4l). Microscopic measurements of grain dimension on 712 randomly selected calcite crystals show that grain size of calcite ranges from 2 μm to 498 μm with a mean value of 154 μm. To evaluate the reliability of optical microscopy in determining the grain size and the grainsize distribution, the same thin section was also analyzed by electron backscattered diffraction (EBSD). In comparing the two sets of data obtained by these two different analytical approaches, we decided to consider a threshold grain size of 25 μm, which is close to the thickness of a thin section. This is because the number of smaller grains that can be indexed and measured by EBSD is much higher than that reliably measurable by optical microscopy. The cumulative frequency grain-size distribution curves obtained by these two different analytical techniques resulted to be very similar (Fig. 5), indicating that the determination of the grain-size distribution by optical microscopy in thin section of a rock or of a specific texturally homogeneous domain within a thin section is sufficiently reliable, provided it is based on a statistically representative number of measurements.

Mineralogy
Mineral composition was determined by XRPD to evaluate the possible influence of different contents in minor mineral phases on stone vulnerability to deterioration processes. Results are reported in Table 3. Because all the rock types here considered are carbonate rocks, the main mineral phase is obviously calcite, which always resulted to be comprised between 95 and 98%, with the exception of Botticino Stone (sample BO) which shows the typical composition of a dolomitic limestone, with 57% of calcite and 47% of dolomite. Analyzing the data obtained for the two varieties of Vicenza Stone, Nanto Stone has a higher insoluble residue than Costozza Stone, as already reported in the literature (Benchiarin 2007). This may be due to the different environmental conditions of deposition and to the different terrigenous contribution at the time of formation of these sediments. The residual fraction of the Nanto stone is composed by illite (5%).
As expected, the mineralogical composition of Orsera Stone (sample OR), which is a white and compact limestone is completely calcitic (100% calcite), with minor components below detection limit. Also, the Aurisina Stone (sample AU) resulted to be completely constituted by calcite. As it concerns the Chiampo Stone (samples ON and PA), both varieties show similar composition, with calcite between 61 and 66%, Mg calcite between 34 and 38%, and traces of palygorskite. The presence of this clay mineral can encourage the detachment of bioclasts during freeze-thaw cycles or wet-dry cycles, amplifying stone vulnerability under environmental forcings.
The two varieties of Verona Stone (Brown Verona, sample BV, and Red Verona, sample RV) show a different composition regarding the minor mineral components. Red Verona variety is basically constituted by calcite only, while Brown Verona variety contains minor mineral phases (quartz and illite) which are sometimes arranged in layers, determining a considerable heterogeneity of the material.
Both varieties of the Asiago Stone (Pink Asiago, sample RO, and White Asiago, sample BI) show a small amount of quartz as a minor component (2-3%), significantly lower than in Verona Stone. This is congruent with the macroscopic texture of Asiago Stone, where discontinuities related to the concentration of clay minerals are less evident than in Brown Verona Stone (BV).
As expected, Carrara marble is completely constituted by calcite, with traces of muscovite below detection limit for the XRPD but recognized in thin section.

Chemical composition
Bulk chemical composition of the samples was determined by XRF, and results are summarized in  the Costozza Stone, probably due to the presence of higher amount of illite and other clay minerals. In general, there is a relation between trace elements concentration (Table 5) and depositional environment (Benchiarin 2007). For example, Sr content is helpful in understanding the origin and diagenesis of carbonates, since dolomitic limestones generally contain less Sr than calcitic limestone. As expected, in the Botticino Stone, Sr concentration is low. A relatively high value of Sr/Mg ratio (observed in Asiago Stone, Aurisina Stone, Vicenza Stone) suggests sedimentation in shallow-water environment of a coastal platform domain. Other significant markers are redox-sensitive trace elements (V, Cr, Ni, Co, Zn, Cu), widely used as indicators of paleo-environmental conditions on the seabed. In particular, the Ni/Co ratio is considered an index of paleooxidizing conditions. In our samples, the values of this ratio are always < 5, indicating that the sediments were deposited under an oxic water column. Considering the relative Ni enrichment of Nanto Stone (sample NA) and Brown Verona Stone (sample BV), we may argue that these rock types were deposited under less oxic conditions than the others.

Physical and mechanical properties
Porosity is a major factor in rock weathering, controlling not only the mobility of the fluids through a material, but also the effectiveness of the deterioration processes involved. Stone vulnerability to deterioration by salt and ice crystallization, chemical weathering and hydroxylation changes significantly depending on the amount, shape, size, and distribution of pores, which control capillarity, water flow, and condensation/evaporation within the pore system. In particular, capillary rise is an important cause of decay in historical monument, and it is referred to be particularly effective on stones with high porosity in the pore range 0.1-10 μm (Benavente et al. 2021). For these purposes, MIP and water absorption results allowed to study the open porosity and the water behavior in relation to the pore-size distribution.
Results obtained from these two techniques show quite similar estimation of the total porosity (%), although MIP tends to overestimate pores with diameters below 0.1 μm (Table 6). This behavior, observed in particular for the Pink and White Asiago Stones (samples RO and BI), and for the Aurisina and Botticino Stones (samples AU and BO, respectively), can be explained by the ink-bottle effect, which can cause the collapse of small irregularly shaped voids at high pressure during mercury intrusion, causing the overestimation of their pore-size fraction (Moro and Böhni 2002;Giesche 2006;Galaup et al. 2012;Coletti et al. 2016).
Based on total porosity derived from MIP and water absorption, the studied stones can be grouped into three classes (Table 6): (i) highly porous stones (total porosity above 20%): Costozza Stone (sample CO) and Nanto Stone (sample NA) belong to this group; ii) stones with low porosity (total porosity between 1 and 5%): Pink and White Asiago Stones (samples RO and BI), Botticino, Aurisina and Ondagata Stones (samples BO, AU, and ON, respectively) belong to this second group; iii) stones with very low porosity (total porosity below 1%): compact and wellcemented limestones such as Brown and Red Verona Stones (samples BV and RV, respectively), Paglierino and Orsera Stones (samples PA and OR) belong to this third group. Porosity from MIP data, within specific pore-size intervals relevant for different deterioration processes, has been investigated and compared among the samples. These pore intervals were selected according to the main critical throat dimensions relevant for water circulation and for generating effective mechanical stresses during salt and ice crystallization. For this purpose, the following pore-throat distribution classes were considered: pores with diameter below 0.1 μm (micropores), where the Kelvin effect is dominant in controlling pore saturation by condensation as a function of relative humidity; pores with diameter between 0.1 μm and 1 mm (capillary pores), which are those within the critical size range for capillary rise and water solution transport; pores with throat size interval between 0.1 and 10 μm, that is the Table 6 Summary of total porosity (%) from water absorption (WA, na%), and mercury intrusion porosimetry (MIP, P%); pores fractions (%) in different pore-size ranges (%) obtained from MIP: pores with diameter < 0.1 µm (micropores); pores with diameter > 0.1 µm (capillary pores); pores within the range: 0. range often considered as the most critical for the generation of crystallization stresses during salt and ice growth (Benavente et al. 2021); and pores with dimension above 10 μm, less affected by deterioration mechanisms connected to capillary rise and crystallization processes, but promoting high water absorption rates. The Nanto Stone (NA) shows the highest number of capillary pores accounting for almost half of the total porosity (above 15% of the 27.18% total pore volume), while 11.90% is made by pores falling in the range below 0.1 μm. In the Costozza Stone (sample CO), porosity has a bimodal distribution, represented both by small pores in the range below 0.1 μm (13.34%), and larger pores with dimensions above 10 μm (13.84%) ( Table 6). Qualitative capillarity test ( Fig. 6 and Fig. 5S) verified the close relation between water rise and the relative abundance of capillary pores (> 0.1 μm) determined by MIP, confirming that the higher susceptibility to deterioration of the Vicenza Stones (CO and NA; Fig. 6a, b) is strongly correlated to their higher fraction of pores within this range (Table 6). Other lithologies displayed low water absorption by capillary rise due to their low total porosity and low porosity in the pore-size range critical for capillarity. Moreover, some lithologies displayed preferential water absorption along stylolitic discontinuities due to their higher content of clay minerals, as in the Asiago (samples BI and RO; Fig. 6c) and Verona Stones (samples RV and BV). In general, these types of sedimentary structures tend to increase capillarity suction, enhancing capillary rise when discontinuities are vertical, thus increasing deterioration rate of structures in contact to the ground. Other samples display more homogenous structures; consequently, capillary rise turns out to be uniform, such as in Botticino Stone (Fig. 6d). In these samples, the effectiveness of capillary suction in promoting deterioration mechanisms is strongly influenced by the total porosity, especially within the critical pore range for capillary, but not by stone orientation. Although mechanical properties depend not only on the effective porosity, in the studied samples, there is a strong correlation between the effective porosity and uniaxial compressive strength measurements and P-wave velocities (Vp). Uniaxial and ultrasound tests show that as porosity increases, load resistance and Vp consistently decrease. The relationship between uniaxial compressive strength and compressional wave velocity is reported in Tables 6 and 7. As expected, the rocks with high porosity display low uniaxial compressive strength (σ = 30.20 and 32.05 MPa, for Fig. 6 Capillary rise in the studied samples after 30 min, marked by a dotted yellow line. For each stone, two specimens have been tested, one orienting vertically discontinuities (e.g., stratification, stylolitic surface), in such a way that capillary rise is parallel to the discontinuity (specimen on the left), the other positioning them horizontally, so that capillary rise is perpendicular to anisotropy (specimen on the right). The following samples are here represented: a Costozza Stone; b Nanto Stone; c Pink Asiago Stone; d Botticino Stone. The full set of samples is reported in Fig. 5S of the supplementary materials  (Table 7). These data coincide with the measurements of the Young's modulus. All stones, indeed, have elastic modulus comprised between 50 and 60 GPa, with the highest values for the Orsera (sample OR) and the Botticino (sample BO) Stones (29.04 and 72.22 GPa, respectively), while the weakest values are found for the most porous rocks, with the Nanto (sample NA) and the Costozza (sample CO) Stones reporting a value of 9.97 and 10.54 GPa, respectively (Table 7).

Decay behavior
Porosity and pore-size distribution are essential in determining the effectiveness of the deterioration mechanisms, especially in the presence of salts, in areas subjected to significant fluctuations of the relative humidity, or keen to freezing events during the winter. All these conditions are typical of the northeastern regions of Italy, although with significant variations from lagoon and coastal environments, and inland and piedmont areas, both in daily and seasonal variations. For this reason, it is important to determine the behavior of these materials when subjected to freeze-thaw and salt crystallization cycles. Micritic limestones, characterized by compact and fine-grained matrix, showed the highest durability to freeze-thaw cycles. On the contrary, limestones characterized by high porosity, coarse grain size, and presence of large bioclasts resulted to be significantly weaker. These different trends are well documented by visual observations during the tests, reporting on development of macroscopic fractures and material loss, quantified by weighing the samples after each aging cycle, and graphically represented by the differential weight curves reported in Fig. 7. While compact limestones before and after 50 cycles have apparently the same aspect and minimum material loss, Vicenza Stones (samples NA and CO), characterized by high porosity and low compactness, at the end of the freeze-thaw aging test resulted to be considerably damaged, registering significant loss of material from the corners and the edges Fig. 7 Aging tests: a Freeze-thaw curves during 50 aging cycles; b ultrasound monitoring during freeze-thaw cycles; c salt crystallization curves during 15 aging cycles; d ultrasound monitoring during salt crystallization cycles of the specimens. Specifically, Nanto Stone (sample NA) suffered a progressive accelerated decay staring from the 30th cycle (Fig. 7a), with 15% of material loss after the 50th cycle (Table 7). Costozza Stone (sample CO) followed a similar trend with final material loss of 10.46% after the 50th cycle (Table 7). Also, the Pink Asiago Stone (sample RO), although displaying a compact and micritic texture, registered significant material loss. Indeed, this stone, despite visually showing a general good response to ice crystallization, registered 1.37% of weight loss at the 25th cycle, when a chunk of material detached along a stylolitic discontinuity (Fig. 7a). Weakness of the Vicenza Stones (samples NA and CO) under freeze-thaw cycles is confirmed by the variation in total porosity, which resulted to be 4.66% for the Nanto Stone and 7.1% for the Costozza Stone (Table 8). This is due to the pressure exerted by the ice crystals formed during freezing that damaged the pore structure causing its collapse and the consequent increase of pore size (Molina et al. 2013).
Ultrasound monitoring recorded a general decrease of wave transmission velocity (see Vp values in Table 8), especially after the 45th cycle (Fig. 7b). The slowest wave velocities were detected for samples with the highest porosity (samples NA and CO), but an overall decrease of wave transmission velocity was verified in all the materials (Fig. 7b). Patterns in Fig. 7b reflect the changes occurred in the pore system; the highest difference in porosity (ΔP%) measured before and after the freeze-thaw aging test has been registered in samples with the most significant decrease in ultrasonic wave transmission velocity. Table 8 shows that the highest ΔP% value is found in the Costozza Stone (ΔP% = 7.1%), followed by the Nanto Stone (ΔP% = 4.66%), and the Asiago varieties (White Asiago: ΔP% = 1.67%; Pink Asiago: ΔP% = 0.77%). On the contrary, when ΔP% is small, wave transmission velocity resulted to be almost constant, such as in the Brown Verona Stone (ΔP% = 0.47%) and the Orsera Stone (ΔP% = 0.41%) ( Table 8).
The effect of salt crystallization on the stones resulted to be similar to what observed after the freeze-thaw cycles. Although weight variations are generally low (Fig. 7c), often registering an increase in weight due to salt accumulation within pores visual observation allowed identifying different deterioration patterns. Material loss in the White Asiago Stone (sample BI) was extremely limited, with salt accumulation exceeding material loss (negative value for weight loss in Table 8), and consisting in peeling of the external surface. The Paglierino variety (sample PA) showed a sudden decrease in weight at the 11th cycle, due to the detachment of a fragment (Fig. 7c) along a stylolite structure, without further significant weight variation until the end of the test. The Vicenza Stones (samples NA and CO) show a totally different behavior, reflecting their different pore structure (Table 6). In these two types of stone, due to the sudden rise of deterioration effects, it was impossible to measure the ultrasonic wave transmission velocity after the 3rd cycle, due to the progressive decay which lead to total disintegration after 15 cycles, with a final weight loss of 83.09 and 94.04%, respectively (Table 8; Fig. 7d). In this case, loss of material preferentially affected the fine-grained fraction, with large bioclasts being progressively isolated by the disaggregation of the surrounding matrix.
Ultrasound P-wave monitoring highlighted that, although other limestones are apparently unaffected by salt crystallization as they registered no material loss during the tests, some of them suffered some form of internal damaging such as development of microcracks or fissuring due to salt crystallization within the pores. This behavior is also particularly evident in the Paglierino variety (sample PA) and the Aurisina Stone (sample AU), in which the P-wave transmission velocities are lower (Vpe = 4903 and 4522 m/s, respectively) than those measured on micritic limestones (Table 8). The negative values of ΔP% reported in Table 8, apparently suggesting a decrease in porosity, are actually due to the residual salt trapped inside the pore system, although all samples were accurately washed after the salt crystallization tests, following UNI EN 12370 (2001) recommendation. Despite Vicenza Stones (samples NA and CO) have ΔP% positive values, indicating an increase in porosity, these values probably represent an underestimation of the real porosity variation, due to a certain amount of salt accumulation inside the pores. Although Costozza Stone (sample CO) and Nanto Stone (sample NA) display some differences in the pore-size distribution (Table 6), they recorded similar damage after freeze-thaw (10.46% and 15.10% weight loss, respectively) and salt crystallization tests (94.04% and 83.09% weight loss, respectively). These two rock types display a similar pore fraction above 0.1 μm (15.18% and 15.28%, respectively). In the Nanto Stone, the majority of these pores (15.11%) fall in the interval 0.1-10 μm while only 0.17% is above 10 μm (Table 6). Conversely, porosity in the Costozza Stone is mainly above 10 μm (13.84%) and only a minor fraction falls within the 0.1-10 μm interval (1.34%). Following the common assumption that the 0.1-10 μm pore-size interval is the most critical for damage related to salt and ice crystallization (Benavente et al. 2021), we would expect a greater resistance of the Costozza Stone (CO) compared to the Nanto Stone (NA) after freeze-thaw and salt crystallization tests. Instead, our data significantly disagree, suggesting that this assumption is too simplistic, that also the pore fraction above 10 μm is fully involved in the deterioration mechanism, and that the development of frost and salt crystallization damage is a more complex process, determined by the simultaneous occurrence of pores of different size, mutually interconnected in such a way that supersaturation conditions will develop in those pores connected to saturated smaller pores (Steiger 2005a(Steiger , 2005bCamuffo 2019).

Discussion and conclusion
In this paper, we studied a number of carbonate rocks widely used in the cultural heritage of northwestern Italy, in many cases part of UNESCO World Heritage Sites. Results demonstrate that depending on the different textural properties (e.g., grain-size distribution of calcite crystals, porosity, pore-size distribution), they exhibit different durability against frost and salt crystallization cycles. Frost and salt damage to these carbonate monumental stones has been verified to be associated with two principal mechanisms of deterioration. The first produces the development of microcracks with formation of fissures changing the inner pore structure, significantly affecting the mechanical properties, and locally causing the surface to chip off. This is particularly evident in the Vicenza Stones (samples NA and CO) and less in the Asiago Stones (samples BI and RO), which record the highest increase in porosity and reduction in the P-wave velocities (Vp) ( Table 8) and the highest weight loss during the aging tests (Fig. 7). The second mechanism causes crumbling of the matrix, locally determining the detachment of larger grains such as bioclasts, or peeling, with progressive loss of material from the stone surface. This is the main behavior observed in all the other stones, which generally record a considerably lower weight loss apparently indicating a greater resistance to salt and ice crystallization. However, ultrasound tests revealed, also in these cases, a significant reduction of the P-wave velocities, suggesting hidden structural damages inside the samples. These mechanisms will produce, in the long term, specific deterioration patterns which are indeed very similar to those commonly observed in historical areas of northeastern Italy. Deterioration effects are enhanced where large and small pores are combined, as well as in the presence of large bioclasts and/or in the case of coarse-grained stones. This can be observed in the Vicenza Stones (samples NA and CO) and in the Aurisina Stone (sample AU), which also display a significant reduction of the P-wave velocities after the aging tests (Table 8). The simultaneous occurrence of pores within what is commonly assumed as the critical size range for salt crystallization (0.1-10 μm) and larger pores (> 10 μm) creates a geometrical combination of interconnected pores of different throat that plays an important role in deterioration during freezing, salt transport, and precipitation (Camuffo 2019). Indeed, a larger pore will experience supersaturation conditions when connected to a smaller pore, exerting a crystallization pressure against the pore wall that is function of the difference in the throat diameters of the two pores (Steiger 2005b).
The presence of pores with critical size for water absorption (> 0.1 μm) was also confirmed by observing the differences in capillary rise among the stones (Fig. 6 and Fig. 5S). In general, the Vicenza Stones (samples NA and CO) registered the greatest water absorption due to the abundance of pores which encourage capillary rise. The observation of the capillary absorption and the results of the aging tests highlighted also the preferential water absorption and development of decay in correspondence to stylolites, as in the Asiago Stones (samples BI and RO) and in the Red and Brown Verona Stones (samples RV and BV). Therefore, low compactness and high porosity make the Vicenza Stones the most vulnerable among the stones studied here, while the presence of stylolites is the main factor producing visible damaging effects in the more compact stones, which otherwise produce microcracks with little macroscopic effects. Chiampo Stones, Botticino Stone, Aurisina Stone, and Orsera Stone demonstrated to be the most durable to freezing and salt crystallization. Nonetheless, the decline in their mechanical properties suggest the formation of microcracks which probably still present little connection at this deterioration level, with negligible effect on water flaw within the capillary system. This does not necessarily mean that buildings made of these stones are safe, but that the threshold beyond which we will observe an acceleration of the deterioration rate has not been reached during the experiments. Interaction with rainwater, extreme events, growth of gypsum crusts, salt accumulation, and stone cleaning can still lead to decay acceleration with potential catastrophic effects (Smith et al. 2008;Benavente et al. 2018), determining loss of esthetic value of the cultural heritage. Therefore, the knowledge of the factors influencing the deterioration rate of stones can help to understand the level of stress under specific environmental conditions, and subsequently to mitigate the damaging mechanisms by the prediction of decay as a function of environmental parameters.