New Insights of Historical Mortars Beyond Pompei: The Example of Villa del Pezzolo , Sorrento Peninsula

: The topic of this study is the archaeometric characterization of mortars from Villa del Pezzolo , a Roman Villa located in Seiano (Napoli-Campania, Italy), dated between the 1st century B.C. and the 3rd century A.D. Mortars were analyzed by means of a multi-analytical approach (polarized optical microscopy, X-ray powder di ﬀ raction, scanning electron microscopy and energy-dispersed spectrometry, thermal analyses and mercury intrusion porosimetry) according to existing recommendations. Analytical results evidenced the use of local geomaterials composed of sedimentary and volcanic aggregates in the mix design and conﬁrmed the three distinct building phases identiﬁed by archaeologists. Volcanic tu ﬀ fragments, identiﬁed in the 1st building phase can be ascribed to Campanian Ignimbrite formation, widely cropping out in the Sorrento Peninsula, as conﬁrmed by the presence of glassy shards, partially devitriﬁed and replaced by authigenic feldspar, a typical feature of welded grey ignimbrite lithofacies (WGI). Volcanic aggregates in samples of the 2nd and 3rd building phases show, instead, the presence of leucite-bearing volcanic scoriae and garnet crystal fragments related to Somma-Vesuvius products. Study of these mortars allowed us to: (1) understand the production technologies; (2) highlight use of materials with hydraulic behavior, such as volcanic and ﬁctile fragments; (3) conﬁrm the three building phases from compositional features of mortars and (4) highlight the change over time of the volcanic aggregate for mortars mix-design.


Introduction
The Bay of Naples (Figure 1a) is characterized by the presence of several worldwide known archaeological sites [1][2][3][4] and references therein, some of them are very interesting and well preserved, among which Pompeii represents, by far, an almost unique example of a still-visible ancient town. Nevertheless, minor, but not for importance, sites (villas, cisterns, thermae, necropolis, etc.) are widespread all over the area.
Villa del Pezzolo is a Roman villa, located in Seiano, which represents a good example of a Patrician villa (Figure 1a,b). The villa was inhabited throughout four centuries, between the 1st century B.C. to 3rd century A.D. During that period, there were more reconstruction phases following catastrophic The site is very interesting both for archaeologists and geologists, as it also represents one of the few locations along Sorrento Peninsula coastline where the consequences of A.D. 79 Somma-Vesuvius eruption are clearly visible, despite the distance from the eruptive center [5].
Like other remarkable examples existing in the Bay of Naples, this villa has been preserved until now, even though located in an aggressive environment by the seaside, thanks to the construction expertise of Roman workers. In particular, thanks to the use of hydraulic mortars. Ancient mortars are composite materials with both crystalline and amorphous phases, containing inerts and reactive aggregates in a calcitic binder [6], which were in use until the end of the 19th century, when Portland cement first appeared [7]. The greatest interest on these artificial geomaterials is certainly due to ancient Roman builders and, mainly, to their skills in using well-defined materials such as pozzolana and fictile fragments in the mix design [8].
The main source of information about the Roman technical ability for manufacturing these materials is the treatise "De Architectura" by Marcus Vitruvius Pollio (80-15 B.C.), dedicated to emperor Augustus. This treatise, namely the second book, describes in detail the utilized materials in the construction sector, including those necessary for mortars production.
The excellent conservation state of many products made in the Roman period demonstrates the high technological level achieved by these builders. Roman builders knew that, thanks to the combination of the lime with special volcanic deposits (pozzolana), mortar became hydraulic, and it means allowing underwater hardening and also acquiring a greater mechanical strength [8][9][10]. Another very important skill was the possibility to use, whenever volcanic materials were not available, fragments of artificial fictile fragments, such as bricks, pottery or shattered tiles, which have the same hydraulic properties of pozzolana [8].
Studies on Roman mortars are very interesting to understand the "secrets" of such enduring resistance because, as of today, the Villae or other Roman buildings have resisted over two thousand years the impact of waves and weathering due to seawater [11][12][13][14][15]. Thus, Roman mortars represent a tangible example of a long-lasting transformation product of geological raw materials.
Nowadays, the concept of "durability" is still of great interest. The modern concrete technologists try to constantly improve their formulation to replicate Roman concrete features such as strength and resistance against aggressive agents (i.e., seawater chemical and mechanical action, acid rains, etc.; [16]).
This research aims to: (a) characterize the mortars used in the construction of Villa del Pezzolo in order to improve knowledge about Roman construction material manufacturing by means of a detailed microstructural and compositional examination of the cementiceous binding matrix and aggregates, (b) identify the mix design and provenance of raw materials; and (c) highlight the minerogenetic secondary processes.

Geological and Archaeological Background
Villa del Pezzolo is placed in a pocket beach along the north shore of the Sorrento Peninsula in a tiny bay called Marina di Equa (Seiano), in the northern flank of Lattari Mounts, constituting the southern margin of the Campanian plain (Figure 2a), [5,17].
The geological evidence of the Campanian plain are mostly related to tectonic extensional movements and volcanoclastic events [18]. The Lattari Mounts are placed 20 km south of the Somma-Vesuvius volcano and 40 km south of Naples (Italy). This mountain ridge is composed by Mesozoic limestones covered by late Quaternary ashfall deposits belonging to the A.D. 79 eruption of Somma-Vesuvius (Figure 2a), [19]. detailed microstructural and compositional examination of the cementiceous binding matrix and aggregates, (b) identify the mix design and provenance of raw materials; and (c) highlight the minerogenetic secondary processes.

Geological and Archaeological Background
Villa del Pezzolo is placed in a pocket beach along the north shore of the Sorrento Peninsula in a tiny bay called Marina di Equa (Seiano), in the northern flank of Lattari Mounts, constituting the southern margin of the Campanian plain (Figure 2a), [5,17].
The geological evidence of the Campanian plain are mostly related to tectonic extensional movements and volcanoclastic events [18]. The Lattari Mounts are placed 20 km south of the Somma-Vesuvius volcano and 40 km south of Naples (Italy). This mountain ridge is composed by Mesozoic limestones covered by late Quaternary ashfall deposits belonging to the A.D. 79 eruption of Somma-Vesuvius (Figure 2a), [19].
On the archaeological site, the impact of the A.D. 79 Somma-Vesuvius eruption is clearly visible. This geographical area was affected by of two phases with relative different deposits: primary, the sedimentary records related to base surge events which caused the formation of pyroclastic deposits (Figure 2b), [18]; secondary, the alluvial delta, called "Durece deposition" [19] by the local population, caused by alluvial events, following the volcanic eruption. This event caused the erosion of volcanic deposits [20] and the consequent burial of the villa. Later, the alluvial delta was broken down creating a 6-8 m high "Durece" cliff, still present on the beach (Figure 2c) [19]. On the archaeological site, the impact of the A.D. 79 Somma-Vesuvius eruption is clearly visible. This geographical area was affected by of two phases with relative different deposits: primary, the sedimentary records related to base surge events which caused the formation of pyroclastic deposits (Figure 2b), [18]; secondary, the alluvial delta, called "Durece deposition" [19] by the local population, caused by alluvial events, following the volcanic eruption. This event caused the erosion of volcanic deposits [20] and the consequent burial of the villa. Later, the alluvial delta was broken down creating a 6-8 m high "Durece" cliff, still present on the beach (Figure 2c) [19].
Moreover, the marine erosion processes seriously damaged the building, making it is impossible to recognize the original construction plan. However, analysis of geological stratigraphy [5] and masonry techniques confirms the archaeological hypothesis of three building phases (Figure 3), [21].
The first building phase was dated back from the 1st century B.C. to the 79 A.D. (Somma-Vesuvius eruptive event; Figure 3b) and is identifiable by the presence of masonry works in opus vittatum and opus reticulatum. The first opus became widespread during the Augustan age while the second one was used in southern Italy between the 1st century B.C. and the 1st century A.D. [18,22].
The villa was built on a limestone flank with a staircase, descending to the seaside, deeply damaged and buried by the eruptive products at the end of the first phase.
In the second phase, dated to the first half of 2nd century A.D. (Figure 3c), after a series of flood events, the villa was completely buried and rebuilt and some rooms were restored using the opus reticulatum with small elements [18][19][20][21][22]. This led to the creation of new spaces, different in form and orientation with respect to the original core structure (Figure 3c) [18][19][20][21][22][23]. This information is supported by the evidence that the walls of the second phase are not in line with the remnants of the first phase structures (Figure 3a) [5]. Moreover, the marine erosion processes seriously damaged the building, making it is impossible to recognize the original construction plan. However, analysis of geological stratigraphy [5] and masonry techniques confirms the archaeological hypothesis of three building phases (Figure 3), [21].
The first building phase was dated back from the 1st century B.C. to the 79 A.D. (Somma-Vesuvius eruptive event; Figure 3b) and is identifiable by the presence of masonry works in opus vittatum and opus reticulatum. The first opus became widespread during the Augustan age while the second one was used in southern Italy between the 1st century B.C. and the 1st century A.D. [18,22].
The villa was built on a limestone flank with a staircase, descending to the seaside, deeply damaged and buried by the eruptive products at the end of the first phase.
In the second phase, dated to the first half of 2nd century A.D. (Figure 3c), after a series of flood events, the villa was completely buried and rebuilt and some rooms were restored using the opus reticulatum with small elements [18][19][20][21][22]. This led to the creation of new spaces, different in form and orientation with respect to the original core structure (Figure 3c) [18][19][20][21][22][23]. This information is supported by the evidence that the walls of the second phase are not in line with the remnants of the first phase structures (Figure 3a) [5].
After this phase, new spaces were damaged and, in some cases, destroyed by sea action, probably due to their position on an alluvial delta [18]. Finally, a third phase (3rd century A.D.; Figure  3d) consisted in the building of a new access to the sea and in restoration of rooms resting on the terrace [18].
The decoration of the roman villa is documented by some sporadic marble findings, now kept at the National Archaeological Museum of Naples.  After this phase, new spaces were damaged and, in some cases, destroyed by sea action, probably due to their position on an alluvial delta [18]. Finally, a third phase (3rd century A.D.; Figure 3d) consisted in the building of a new access to the sea and in restoration of rooms resting on the terrace [18].
The decoration of the roman villa is documented by some sporadic marble findings, now kept at the National Archaeological Museum of Naples.

Materials and Methods
Sampling (20 mortars) was performed on several walls of the archaeological complex (Figure 3), i.e., thanks and with the assistance of archaeologists of the former Soprintendenza per i Beni Archeologici della Campania in order to take samples that were representative of the construction. The sampling was carried out by hammer and criteria such as the type of mortar, the good state of aggregation, the available quantities and little invasivity.
The collected samples consist of eighteen bedding mortars and two floor mortars (Figure 3a), and were grouped following the building phases (group A-1st building phase, group B-2nd building phase and group C-3rd building phase; Figure 3a; Table 1). The collected samples were characterized according to [24] European standard. Macroscopic characteristics, such as color, presence of aggregates, and others, were observed in order to identify materials and to plan the analytical approach, performed at DiSTAR (Dipartimento di Scienze della Terra, dell'Ambiente e delle Risorse, Università di Napoli Federico II) laboratories and at the Group Technical Center (CTG) of the company Italcementi Heidelberg Group in Bergamo (Italy). Color was evaluated by comparison with the Munsell Soil Color Chart [25].
Polarized optical microscopy (POM) on thin sections was performed using a Leica DFC280 microscope (Leica, Wetzlar, Germany) to observe the textural features and the petrographic composition of samples. The percentage of binder and aggregate was measured by modal analysis, counting at least 1500 points for each section, using Leica Q Win image analysis software (version 3.2.1, Leica, Wetzlar, Germany). This analysis can be considered representative since the maximum uncertainty of percentage for a total amount of 1500 points is about 2.8% [26].
Mineralogical determinations were performed on different fractions: binder, aggregates and fictile fragments, according to [24]. Mortars were gently disaggregated and separated in different fractions using a binocular microscope (WILD M38, Heerbrugg, Switzerland). Each fraction was powdered in an agate mortar to obtain a homogeneous sample to acquire (particle size< 200 µm) avoiding orientation related problems, primary extinction or crystallite size errors [27,28]. Qualitative mineralogical analyses were performed by X-ray powder diffraction (XRPD) using a PANalytical X'Pert PRO 3040/60 PW diffractometer (Malvern PANalytical, Almelo, The Netherlands) CuKα radiation, 40 kV, 40 mA, scanning interval 4-70 • 2θ, equivalent step size 0.017 • 2θ, with a step counting time of 120 s). To identify the mineral phases, the PANalytical Highscore Plus 3.0e software (Malvern PANalytical, Almelo, The Netherlands) with PDF-2/ICSD databases was used.
Micro-textural observations and quantitative micro-chemical analyses were carried out by Scanning Electron Microscopy coupled with Energy Dispersive Spectroscopy (SEM/EDS; JEOL JSM-5310 (Jeol Ltd., Tokyo, Japan) coupled with a Oxford Instruments Microanalysis Unit (Oxford Instruments plc, Abingdon, Oxfordshire, UK) equipped with an INCA X-act detector (ETAS group, Stuttgart, Germany). Measurements were performed with an INCA X-stream pulse processor (ETAS group, Stuttgart, Germany) using a 15 kV primary beam voltage, 50-100 µA filament current, variable spot size, from 30,000 to 200,000× magnification, 20 mm WD and 50 s net acquisition real time. The INCA Energy software (ETAS group, Stuttgart, Germany) was employed, using the XPP matrix correction scheme and the Pulse Pile up correction. The quant optimization was carried out using cobalt (FWHM-full width at half maximum peak height-of the strobed zero = 60-65 eV). The following standards from the Smithsonian Institute and MAC (Micro-Analysis Consultants Ltd. St Ives, UK) were used for calibration: diopside (Ca), fayalite (Fe), San Carlos olivine (Mg), anorthoclase (Na, Al, Si), rutile (Ti), serandite (Mn), microcline (K), apatite (P), fluorite (F), pyrite (S), sodium chloride (Cl), benitoite (Ba) and pure vanadium (V), [29]. The Kα, Lα, Lβ, or Mα lines were used for calibration, depending on the specific element.
High-resolution imaging of surface morphology (backscattered images) was generated by secondary electrons using the same instrument.
Micro-chemical analyses (EDS) were performed to determine chemical composition of binder and lime lumps and to study volcanic fragments identified among aggregates (point data). The Hydraulicity Index (HI) of binder and lime lumps was calculated according to Boynton [30] using (SiO 2 + Al 2 O 3 + The thermal behavior of mortars was carried out with a Mettler Toledo TGA/SDTA 851e instrument (Mettler Toledo, Columbus, OH, USA). The samples were heated from 40 • C to 1000 • C, with a heating rate of 10 • C/min in nitrogen atmosphere (flow rate 60 mL/min), with the main goal of determining the total (binder plus aggregates) hydraulic features of these materials. TG and DSC curves were acquired and then processed with Mettler Toledo STARe SW 7.01 software (Mettler Toledo, Columbus, OH, USA). Thermal analyses were performed on powered bulk samples and on their finer sieved fraction (<63 µm); the latter can be considered as a binder-enriched specimen [31].
The pore system of samples was investigated by means of Mercury Intrusion Porosimetry (MIP). Due to the scarce amount and small dimensions of samples, four samples, approximately 1 cm 3 in size, were selected on the bases of macroscopic and microscopic features. Analyses were performed on three fragments for each sample, and average results are reported. These fragments were dried in an oven for 24 h at 105 • C and then analyzed by Thermo Scientific equipment PASCAL 140 (Thermo Fisher Scientific, Waltham, MA, USA) with a maximum injection pressure of 0.4 MPa and PASCAL 240 with a maximum injection pressure of 200 MPa. These instruments, used consecutively, allowed us to assess (1) total volume of pores with a radius between 3.75 nm and 800 µm (mm 3 /g), (2) open porosity (vol.%), (3) bulk density (g/cm 3 ), (4) apparent density (g/cm 3 ), and (5) specific surface (m 2 /g); graphical and numerical representation of the distribution of pore size was also provided.

Texture and Optical Microscopy of the Mortar
Mortar samples appear to be macroscopically intact and quite hard ( Figure 4).
Bedding mortars made evident yellow to dark greyish brown colors (Munsell 2.5Y 7/3 to 10YR 4/2), with coarse aggregates ranging from 2 mm up to 3 cm (Figure 4a). The aggregates were made of volcanic and carbonate fragments and, only in samples of the group C, fictile fragments were recognized ( Figure 4b). Floor mortars showed a variability in color from very pale brown to reddish (Munsell 7.5YR-7/3 to 5YR-5/6). The aggregates mainly comprised of fictile fragments and subordinately by volcanic and carbonate fragments, variable in size from 1 mm to 2.5 cm. All mortars contained lime lumps (2-4 mm) that generally consisted of unreacted lime [31].
Minerals 2019, 9, x FOR PEER REVIEW 7 of 23 (Munsell 7.5YR-7/3 to 5YR-5/6). The aggregates mainly comprised of fictile fragments and subordinately by volcanic and carbonate fragments, variable in size from 1 mm to 2.5 cm. All mortars contained lime lumps (2-4 mm) that generally consisted of unreacted lime [31]. The floor mortars VP19 and VP20 samples, belonging to group C, were identified as cocciopesto, also called opus signinum, a typical building technique used in the ancient Rome for making floor and waterproofing cisterns. The binder phase was composed mainly of micritic matrix (25.4 vol.%) and lime lumps were also present (2.2 vol.%; Figure 5l).
Fictile fragments are different from each other, even if included in a single mortar sample. In particular, fictile matrices can be either characterized by high (Figure 5g) or low (Figure 5l) optical activity. In some fictile fragments, inclusions are mainly represented by tiny crystals of quartz, biotite, clinopyroxene (Figure 5m), whereas, in other fragments, they are characterized by a prevailing volcanic component (pumice and scoriae; Figure 5n).
Modal analysis (Table 2) highlighted the differences in the aggregate/binder ratio, particularly evident in group C. Mortars of this group showed a binder/aggregate ratio lower than 1, indicating a higher percentage of aggregates than the binder phase. The floor mortars VP19 and VP20 samples, belonging to group C, were identified as cocciopesto, also called opus signinum, a typical building technique used in the ancient Rome for making floor and waterproofing cisterns. The binder phase was composed mainly of micritic matrix (25.4 vol.%) and lime lumps were also present (2.2 vol.%; Figure 5l).
Fictile fragments are different from each other, even if included in a single mortar sample. In particular, fictile matrices can be either characterized by high (Figure 5g Modal analysis (Table 2) highlighted the differences in the aggregate/binder ratio, particularly evident in group C. Mortars of this group showed a binder/aggregate ratio lower than 1, indicating a higher percentage of aggregates than the binder phase.

Mineralogical Analyses
The collected samples were separated into binder, aggregates, and fictile fragments, according to [24] recommendation (mortar characterization) and then analyszd by XRPD. Results are shown in Table 3 and in Figure 6.   The XRPD results showed that, in binder, calcite is the main phase with subordinate amounts of gypsum. Regarding aggregates, sanidine, plagioclase, analcime, mica, clinopyroxene and calcite were recognized; chabazite is also present in some samples of group A and leucite in samples of groups B and C (Table 3; Figure 6).
Fictile fragments, only present in mortars of group C, showed the occurrence of calcite, quartz, mica, sanidine, clinopyroxene, plagioclase and hematite. All analyzed samples showed the presence of halite (Table 3; Figure 6).
XRPD analyses also allowed for revealing the presence of amorphous fraction, recognized by the rising of pattern background between 18 • and 40 • of a 2θ angle. This fraction is probably related to (1) volcanic glass component (pumice and scoriae) and (2) C-A-S-H phases (calciumaluminium-silicate-hydrate).  In order to obtain further information about the binder composition, SEM-EDS microanalyses were carried out on polished thin sections both on binder and on lime lumps (Tables 4 and 5). Moreover, HI was calculated, according to Boynton's formula [30], to achieve information on hydraulicity grade reached by mortars (Tables 4 and 5 HI of lime lumps is always lower than 0.10%, while HI of binder ranges between 0.10% and 0.25% (Figure 7c; Tables 4 and 5). VP19 and VP20 (floor mortars) samples showed, for binder, the highest HI of 0.22% and 0.25%, thus confirming their differences from all other samples.

Micro-Morphology and Chemical Analysis
As far as volcanic aggregates are concerned, SEM-EDS analyses highlighted for group A samples the presence of glassy shards, partially devitrified and partially coated by micrometric sanidine crystals (Figure 8a). Moreover, chemical composition of glassy shards and pumice (Supplementary Materials 1-S1) allowed us to classify them as trachyte according to the TAS diagram (Total Alkali vs. Silica) [36] (   crystal fragments (Figure 8b), both in pumice and binder. In this case, the chemical composition of pumice (Supplementary Materials 2-S2) reported in the TAS diagram (Figure 8c), allowed us to classify them as phonolite (Figure 8c). Additional information was obtained by chemical composition of garnets that were classified as a solid solution between andradite (48.98-58.38 mol%) and grossular (25.91-30.46 mol%), (calculated following Locock, [37]; Supplementary Materials 3-S3).

Thermogravimetric Analysis
Thermal analyses evidenced, for all samples, presence of: (1) carbon-dioxide content, derived from the dissociation of carbonatic phases and (2) structural bound water (SBW), which represents the amount of OH groups occurring in the silicate framework and inside the thinnest ultramicropores [29].
The highest SBW content was reported for sample VP17 (10.6%), while the lowest SBW was reported in samples VP19 and VP20 (3.94-3.45% respectively), also showing the lowest total LOI (Loss On Ignition; 23.76-24.09%); the latter samples were, in fact, floor mortars different from other bedding mortars.

Mercury Intrusion Porosimetry
The porosity was determined on fragments of selected mortars (VP1, VP8, VP11, VP20) due to the scarcity of available material. Table 7 summarizes the parameters, such as cumulative volume, bulk density, apparent density, open porosity and specific surface, provided by MIP. For each sample, the representative pore size distribution is also shown ( Figure 10). As regards the relative volume curves, pore radii mainly range between 5 and 100 nm, except for VP20, where a displacement of the pore access radius towards larger sizes (1000-10,000 nm) was recognized ( Figure 10).
The open porosity is very similar for all samples, ranging from 42.46 to 44.65 vol.% (Table 7) with unimodal and broadened shape of the cumulative pore size distribution. Once again, the curve for sample VP20 is more irregular (Figure 10).

Discussion
This research represents a first archaeometric study on mortars from Villa del Pezzolo and provides some useful information about mix-design, provenance of raw materials and reconstruction of building phases.
Analytical results allowed for identification of two different mix designs, for bedding and floor mortars, respectively. Bedding mortars were characterized by a mixture of lime, fine volcanic materials, different type of aggregates and water, while floor mortars represent the cocciopesto, due to the addition of fictile materials both as binder and aggregates. Cocciopesto, also known as opus signinum, represents the typical mix design used during the Roman age for waterproof mortars in water tanks, thermal pools, and in the caverns of aqueducts and for flooring [8].
As far as provenance of raw materials is concerned, the minero-petrographic and chemical analyses, along with the surrounding geological setting, confirmed a local origin and also the hypotheses of the three building phases of Villa del Pezzolo [18,21].
Volcanic fragments used as aggregate in the first building phase can be related to Campanian Ignimbrite formation, cropping out in the same geographical area (Figure 11). This hypothesis was also confirmed by the presence of specific zeolitic phases (chabazite and analcime; Table 3) and of glassy shards, partially devitrified and replaced by authigenic feldspar (Figure 8a), a typical feature of the welded gray Campanian Ignimbrite lithofacies (CI-WGI; [39]).
Furthermore, chemical composition of glassy shards and pumice fragments follows the compositional trend of CI (Figure 8c).

Discussion
This research represents a first archaeometric study on mortars from Villa del Pezzolo and provides some useful information about mix-design, provenance of raw materials and reconstruction of building phases.
Analytical results allowed for identification of two different mix designs, for bedding and floor mortars, respectively. Bedding mortars were characterized by a mixture of lime, fine volcanic materials, different type of aggregates and water, while floor mortars represent the cocciopesto, due to the addition of fictile materials both as binder and aggregates. Cocciopesto, also known as opus signinum, represents the typical mix design used during the Roman age for waterproof mortars in water tanks, thermal pools, and in the caverns of aqueducts and for flooring [8].
As far as provenance of raw materials is concerned, the minero-petrographic and chemical analyses, along with the surrounding geological setting, confirmed a local origin and also the hypotheses of the three building phases of Villa del Pezzolo [18,21].
Volcanic fragments used as aggregate in the first building phase can be related to Campanian Ignimbrite formation, cropping out in the same geographical area (Figure 11). This hypothesis was also confirmed by the presence of specific zeolitic phases (chabazite and analcime; Table 3) and of glassy shards, partially devitrified and replaced by authigenic feldspar (Figure 8a), a typical feature of the welded gray Campanian Ignimbrite lithofacies (CI-WGI; [39]).
Furthermore, chemical composition of glassy shards and pumice fragments follows the compositional trend of CI (Figure 8c).
Regarding the volcanic aggregates used in the mix design of the second and third building phases (after A.D. 79 Somma-Vesuvius eruption), some differences can be evidenced, mostly related to the mineralogical composition and aggregates shape (from angular to sub-rounded). The presence of leucite-bearing scoriae and crystal fragments of garnet, both in binder and pumice, allowed us to relate these materials to eruptive products of Somma-Vesuvius. Chemical composition of garnets (solid solution between andradite 52.99-57.29 mol.% and grossular 25.64-28.65 mol.%) is fully compatible with Somma-Vesuvius garnets (andradite 46-70 mol.% and grossularia 16-45 mol.%; [41] and unpublished garnets analyzed by Melluso from intrusive Somma-Vesuvius rocks; S3). Moreover, the chemical analysis of pumice, reported in the TAS diagram (Figure 8c) is likely that of Somma-Vesuvius products. Regarding the volcanic aggregates used in the mix design of the second and third building phases (after A.D. 79 Somma-Vesuvius eruption), some differences can be evidenced, mostly related to the mineralogical composition and aggregates shape (from angular to sub-rounded). The presence of leucite-bearing scoriae and crystal fragments of garnet, both in binder and pumice, allowed us to relate these materials to eruptive products of Somma-Vesuvius. Chemical composition of garnets (solid solution between andradite 52.99-57.29 mol.% and grossular 25.64-28.65 mol.%) is fully compatible with Somma-Vesuvius garnets (andradite 46-70 mol.% and grossularia 16-45 mol.%; [41] and unpublished garnets analyzed by Melluso from intrusive Somma-Vesuvius rocks; S3). Moreover, the chemical analysis of pumice, reported in the TAS diagram (Figure 8c) is likely that of Somma-Vesuvius products.
Furthermore, the presence of sub-rounded shaped aggregates allowed us to also hypothesize the use of "Durece" formation, an alluvial delta lithofacies formed by the products of debris-and mud-flows emplaced after strong rainfalls following the Somma-Vesuvius eruption [5].
The samples from the third building phase showed the presence of fictile fragments that are absent in the other building phases. POM observations evidenced that the fictile fragments are very different from each other in terms of matrix, textural and petrographic features. For these reasons, their provenance was hard to define; however, these differences suggest a recycling of building materials. These fictile fragments also have a technological role because they were used to provide hydraulicity to mortars [8,42], as also confirmed by the HI evaluation.
There is no specific information about the provenance of lime and carbonatic aggregates can be provided, but it is reasonable to suggest a local provenance due to the geographical position of the villa, very close to the northern flank of the Lattari Mounts (Mesozoic limestones; Figure 2a).
Composition of the cementiceous binding matrix, in some samples, showed the ubiquitous presence of lime lumps, gypsum and C-A-S-H gel, the latter derived from lime and pozzolanic material reactions (volcanic and fictile materials, [43]); its formation testifies to the achievement of mortar hydraulicity [23,44] Lime lumps are likely connected to not-well-reacted lime, the origin of which has to be searched in the properties of slaked lime (calcium hydroxide) and in the water/quicklime ratio [10,45]. Chemical analysis of lime lumps (CaO more than 92%) testified to the good nature of limestone used for lime production.
Gypsum can be ascribed to late sulphation of calcite or even reaction between sulfate and hydrated lime [46,47]. Halite in all samples can be easily related to interaction with seawater due to site location. Furthermore, the presence of sub-rounded shaped aggregates allowed us to also hypothesize the use of "Durece" formation, an alluvial delta lithofacies formed by the products of debris-and mud-flows emplaced after strong rainfalls following the Somma-Vesuvius eruption [5].
The samples from the third building phase showed the presence of fictile fragments that are absent in the other building phases. POM observations evidenced that the fictile fragments are very different from each other in terms of matrix, textural and petrographic features. For these reasons, their provenance was hard to define; however, these differences suggest a recycling of building materials. These fictile fragments also have a technological role because they were used to provide hydraulicity to mortars [8,42], as also confirmed by the HI evaluation.
There is no specific information about the provenance of lime and carbonatic aggregates can be provided, but it is reasonable to suggest a local provenance due to the geographical position of the villa, very close to the northern flank of the Lattari Mounts (Mesozoic limestones; Figure 2a).
Composition of the cementiceous binding matrix, in some samples, showed the ubiquitous presence of lime lumps, gypsum and C-A-S-H gel, the latter derived from lime and pozzolanic material reactions (volcanic and fictile materials, [43]); its formation testifies to the achievement of mortar hydraulicity [23,44] Lime lumps are likely connected to not-well-reacted lime, the origin of which has to be searched in the properties of slaked lime (calcium hydroxide) and in the water/quicklime ratio [10,45]. Chemical analysis of lime lumps (CaO more than 92%) testified to the good nature of limestone used for lime production.
Gypsum can be ascribed to late sulphation of calcite or even reaction between sulfate and hydrated lime [46,47]. Halite in all samples can be easily related to interaction with seawater due to site location.
A common feature of the investigated mortars is the hydraulicity features, confirmed by SEM-EDS and thermal analyses. The evaluated HI index allowed us to consider all mortars as hydraulic exclusively thanks to the addition of materials with pozzolanic activity (volcanic and fictile materials). The role of pozzolanic materials was evident by the HI values evaluated on lime lumps (<0.10%) and binder (0.10-0.25%) that can be classified as quicklime and weakly-moderately hydraulic materials [30,48].
Differences among HI index values (Tables 4 and 5; Figure 7c) further confirmed three building phases, evidencing a low value of HI (0.11-0.14%) for mortars of group A, average values (0.15-018%) for mortars of group B, and high values (0.15-0.25%) for group C. The higher HI index of group C is due to use of fictile fragments that improve the hydraulicity of mortars.
Hydraulicity of mortars was also further established by the results of thermal analyses, which showed weight loss of Structural Bound Water (SBW, 200-600 • C), generally related to the presence of hydrated compounds such as C-A-S-H, and the decomposition of calcite and other carbonates, usually occurring between 600-850 • C with a consequent release of CO 2 (Table 6; [29,49]).
According to the literature data [31,50], the binary diagram CO 2 /SBW vs. CO 2 (Figure 9) classified all analyzed mortars as natural pozzolanic. The high concentration of secondary mineralogical phases, the C-A-S-H gel, was supported by porosity tests, which showed a maximum in pore radius distribution between 5 and 100 nm and an open porosity between 40-45 vol.% ( Figure 10; Table 7). Small dimension of pore radius is due to the formation of C-A-S-H gel that fills the pores enhancing bonding in pumice clasts and improving resistance to weathering [9,34,51]. According to Sutter [52], the pore radius range is important for material durability: the range between 5 and 100 nm is considered not detrimental for durability, while the range 1000-10,000 nm, detected for sample VP20, can promote weathering [53,54].
The analyses of the mortars used in the construction of Villa del Pezzolo revealed important evidence with regard to reconstruction of building phases according to the hypothesis performed by archaeologists and geologists [5,19,21].
Generally, this analytical approach can provide important information about techniques used in the preparation of ancient mortars, and thus they can be used as references for future comparative studies, also involving other buildings in coeval constructions.
This research demonstrates that studying minor sites such as Villa del Pezzolo is very important and useful as, similarly to Pompeii, they experienced the effects of the A.D. 79 Somma-Vesuvius eruption and of debris and mud flows generated after this event. It is also very important to focus the attention to this kind of archaeological sites in order to improve the conservation state and fruition by the public.