1 Introduction

The archaeological investigations in a plot in 1 Katre Street on Kastelli hill of the town of Chania commenced in 2006, as Maria Andreadaki-Vlazaki reported (Andreadaki-Vlazaki 2010, 2014, 2023): “The old city of Chania has the great privilege of living, breathing and moving on the ruins of many cultures with strong traces and can be called one of the oldest cities in Europe, whose life began as early as the 4th millennium BC. The city has lived since the 4th millennium BC. The very modern settlement, which is the great obstacle to the unveiling of its past, is at the same time the charm of this city, which is marked everywhere, in its subsoil, its soil and its air, by the ancient aura. The excavations which took place on Kastelli Hill in the old city of Chania, confirmed our recently expressed views that with the systematic exploration of the hill and especially in the area of Katre Street, the palatial center and part of the palace complex of Chania and Kydonia from the Minoan and Cretan–Mycenaean periods are gradually being uncovered”. During the first excavation season, the research focused on preliminary large scale ground works. A thick twentieth-century discharge layer was removed, after which three moats (A–C) were opened. The upper layers date back to the Roman occupation years. Underneath them, Classical and Archaic architectural remains came to light. Deeper in the ground, pits of the Geometric period as well as layers of the LM IIIC phase were detected. Large volumes of pottery and a significant number of small finds were gathered and recorded.

In this study, it was deemed important to identify the construction period of the fortified Wall that was dated as Hellenistic (late fourth century BC) and its eventual repairs later on. Therefore, a comparison study with dated mortars from the same excavation was carried out using physico-chemical, mineralogical and multivariate analyses. This work represents a compositional comparison between mortars of known dating, as indicated in other analogous studies, rather than true dating with conventional techniques (Miriello et al. 2010).

2 Methods of analyses

14 mortars sampled from four moats from the walls of 1 Katre Street in Chania, Crete, Greece, with codes ΚΧ1 to ΚΧ14 were analysed using physico-chemical laboratory techniques. The laboratory techniques used such as grain size distribution, X-ray Diffraction (XRD), thin section study under the polarizing microscope, Fourier Transform Infrared Spectroscopy (FTIR), Energy-Dispersive X-ray Fluorescence (EDXRF), Differential thermal/thermogravimetric analysis (DTA–TG) accurately provide the qualitative and quantitative composition of the samples. The use of two or more complementary techniques confirmed the accuracy and repeatability of the results (Blaeuer and Kueng 2007; Elsen 2006; Maravelaki-Kalaitzaki et al., 2011; Middendorf et al. 2005).

Figure 1 illustrates the sampling locations from moats A, B, Z and I from Katre 1 street indicated within the circle in the Chania map, along with the samples coming from Sevax and Daskalogianni street. Mortars KX1, KX3, KX10 and KX11 come from 1 Katre Street, Chania, Moat A, and are joint mortar from the Wall. The mortar KX3 is a joint mortar from the crepis of the ancient Wall (Fig. 2). The mortars ΚΧ8 and ΚΧ9a and b come from Moat B of Katre Street (Fig. 3). The mortars ΚΧ4 and ΚΧ5 come from Moat Z, wall 8, from the upper surface and the side wall, respectively, which is dated as Roman and form the basis of comparison of other mortars (Fig. 4). ΚΧ6 and ΚΧ7 come from Moat I, wall 6, which is dated as Roman, namely ΚΧ6 from layer 1, while ΚΧ7 from floor 14, layer 2, group 2 (Fig. 4). The mortar KX12 is dated to the Middle Minoan IIIB period (end of 17th c. BC) and comes from an excavation building on Daskalogianni and Sifaka street. The mortars KX13 and KX14 come from newer repairs of the walls and were considered for comparison purposes only. Mortar KX2 comes from another excavation area (Sevax) close to that under study and it is a structural mortar dated to the Late Minoan IB period (15th c. BC).

Fig. 1
Fig. 1
Full size image

Documentation of sampling location

Fig. 2
Fig. 2
Full size image

Documentation of sampling location and samples from moat A of the fortified wall

Fig. 3
Fig. 3
Full size image

Documentation of sampling location and samples KX8, KX9a and KX9b from moat B of the wall

Fig. 4
Fig. 4
Full size image

Documentation of sampling location and samples from moats Z (a) (KX4, KX5) and I (b) (KX6, KX7a and KX7b), building walls 8 and 6, respectively

Mortars were classified into four groups, based on their common macroscopic characteristics (Table 1).

Table 1 Grouping of mortars based on macroscopic study
Full size table

The granulometric analysis aimed to determine the grain size distribution of the aggregates of the mortars and the binder to aggregates ratio for each mortar separately (Maravelaki-Kalaitzaki et al., 2005).

The qualitative analysis of the samples was conducted using XRD and FTIR. In the FTIR analysis pre-weighed samples of total mortar, aggregates and bonder were used to provide sample semi-quantification by comparing them to standard samples (Farmer 1974). The mortars were also subjected to wet chemical analyses to determine the soluble silicate components and the soluble calcium oxide that indicate hydraulic component formation in samples (Maravelaki-Kalaitzaki et al., 2011).

For the quantitative chemical analysis of the samples, in addition to the traditional wet chemical analysis, DTA–TG and EDXRF techniques were also used. DTA–TG was employed, considering specific thermal ranges such as 600–850 °C related to the decarbonation of carbonates and 200–600 °C associated with dehydroxylation and the release of structurally bonded water within the crystal structure of materials, along with the decomposition of any organic compounds present. EDXRF was carried out on samples appropriately shaped into pellets for the determination of elements with atomic numbers higher than Al, using radioactive irradiation sources Fe-55, Cd-109 and Am-241. In the EDXRF analysis the elemental composition (qualitative analysis), as well as the minimum detection thresholds were specified. From sodium to uranium, a broad spectrum of atomic elements can be measured using EDXRF analysis, which offers detection limits ranging from low parts-per-million (ppm) to high weight percent (wt%). For the quantitative analysis, the most appropriate type of quantification was verified based on standard targets used. The following were used as standards for EDXRF analysis: (i) Argillaceous limestone, Standard Reference Material 1c, National Institute of Standards and Technology (NIST), (ii) Phosphate rock, Standard Reference Material 120b, National Institute of Standards and Technology (NIST), (iii) Major and minor element constituents, Soil-5, International Atomic Energy Agency and, (iv) Trace elements in coal fly ash, Standard Reference Material 1633b, National Institute of Standards and Technology (NIST).

Principal Component Analysis (PCA) is a mathematical method used to deal with the large amount of data resulting from a chemical analysis and is usually used to reduce data variance. It is a mathematical process that identifies the main components PC1, PC2, etc. that are a linear combination of the original variables and that explain most of the variation in the data (Miller and Miller 2000; Gardiner 1997). In this statistical method, a multidimensional space is created with variables as axes. The analysis of the main components was applied for the chemical elements of the aggregates with the highest concentration and for some important trace elements identified during the chemical analysis.

3 Results and discussion

Initially, the granulometric analysis of the samples was performed. Figure 5 displays the granulometric curves of the mortars. As can be clearly seen, all mortars are characterised by a granulometric gradation containing fine grains, except for mortar KX7. The granulometric curve of KX7 clearly indicates that it is a mortar with coarser aggregates compared to the others. In KX7 mortar, the amount of aggregates with a diameter exceeding 2 mm is significant (around 20%), while in other mortars the corresponding percentage is less than 1.2%. However, KX7 showed a notable difference from the others regarding the percentage of aggregates with a diameter of 0.075 mm. While in all mortars this percentage ranges from 5 to 12%, in KX7 it exceeds 18%. It is also worth mentioning that the percentage of aggregates of KX11 reaches 23% with a grain diameter of over 0.85 mm. The corresponding percentage in other mortars ranges from 0.6 to 13%.

Fig. 5
Fig. 5
Full size image

Grain size curves of mortars. The curve of mortar KX7 appears to differ significantly

From the granulometric curve, the determination of the binder-to-aggregate ratio for all samples (Table 2) was achieved. Based on this ratio, the studied mortars can be divided into 4 categories. The first category includes the mortars ΚΧ1 and ΚΧ4 with a binder-to-aggregate ratio of 1:1, the second one includes only the mortar KX5, where the amount of aggregates is twice that of binder, the third one contains the mortars ΚΧ6 and ΚΧ7 with a binder-to-aggregate ratio of 2:3 and finally the fourth includes the remaining mortars with a binder-to-aggregate ratio of 1:3, which is the most commonly found in mortar analyses (Arizzi and Cultrone 2021; Válek et al 2019; Velosa et al. 2010).

Table 2 Results of the granulometric analysis of mortars
Full size table

A first conclusion arises from comparing macroscopic study with granulometric analysis. According to the classifications that have been made, mortars ΚΧ3, ΚΧ8, ΚΧ9, ΚΧ10 and ΚΧ11, which come from the fortified wall, exhibit macroscopic similarities in terms of the binder-to-aggregate ratio and the grain size of their aggregates. Therefore, the preliminary classification based on macroscopic properties and presented in Table 1 is further supported by the grain size distribution analysis and the result of the binder-to-aggregate ratios.

The samples were then qualitatively analyzed through infrared spectroscopy (FTIR) and X-ray diffraction (XRD), to characterize the chemical compounds and minerals of the mortar composition. The analyses showed that mortars are mainly of calcitic nature, with a significant amount of quartz and plagioclases. In more detail, the compounds contained in the samples are presented in Table 3 along with a short description of the thin section study under the polarized microscope.

Table 3 Compounds identified in mortars via X-ray diffraction (XRD), infrared spectroscopy (FTIR) and thin section study
Full size table

As mentioned above, the studied mortars are of calcitic nature with a high amount of Portlandite (non- carbonated lime), as well as aggregates that are fragments of minerals and rocks. Rock fragments belong to the categories of sedimentary rocks, such as carbonate rocks (limestone, fossiliferous limestone and conglomerate), as well as to metamorphic rocks such as quartzite, but in a very small percentage. In all specimens, many fossils from algae, shells of marine organisms, elasmobranchs, corals and echinoderms are observed (Fig. 6). In addition, ceramic fragments were identified in the sample ΚΧ7, and are sporadically present in the other samples.

Fig. 6
Fig. 6
Full size image

Thin sections of samples under a polarizing microscope, where: Fs fossils, Fd feldspars, Pt Portlandite, Cc calcite, Qz quartz, CFr ceramic fragment

As shown in Table 3, the crystal phases of the samples are silicate minerals such as quartz, carbonate minerals like calcite and aragonite, and plagioclases with albite composition. Clinochloro, epidotus and clinopyroxene were also detected in a small percentage. The percentage of metal oxides or other metal minerals is low. Some mortars contain iron oxides such as magnetite and hematite, titanium oxides with rutile composition, titanite and titanium-iron oxides such as ilmenite. Finally, very few sulfides, such as pyrite and sphalerite, were detected only in ΚΧ4 and ΚΧ5 mortars.

The elemental and quantitative analysis of the samples through EDXRF provided additional information on the different nature of the mortars. To better detect light elements, analyses were conducted in a controlled helium atmosphere (He). The bar graphs in Figs. 7 and 8 provide information on the content of the elements for each mortar separately and allow for comparisons concerning the quantification of mortars.

Fig. 7
Fig. 7
Full size image

%wt. content of major elements such as, Na, Mg, Al, P, S, Cl, K, Ti, Si and Ca according to EDXRF analysis

Fig. 8
Fig. 8
Full size image

%wt. content of elements with low content (trace elements), according to EDXRF

The amount of calcium contained in mortars ΚΧ2 and ΚΧ6 ranges from 30 to 35%, a percentage that classifies them in the category of calcitic mortars. On the contrary, KX7 mortar stands out from the rest due to its low concentration of calcium and its high concentration of silicon and aluminum (Fig. 7). This high content of aluminosilicate components is due to the presence of quartz, conglomerate fragments and ceramic fragments, as detected with XRD. The use of KX7 as a floor mortar justifies the high content of aluminosilicates. KX6 mortar contained a considerable quantity of magnesium compared to the other mortars due to dolomite (Fig. 7). The mortars ΚΧ2 and ΚΧ5 seem to contain the smallest amount of clay components compared to the rest, only 0.17%, confirming their calcitic nature and the absence of ceramic fragments.

In the bar graph of Fig. 8, the concentrations of trace elements are presented. Noteworthy is the concentration of zirconium (Zr) in all the mortars, reaching up to 0.023% in mortar ΚΧ7, which also contained the highest amount of crushed ceramic.

As mentioned above, the mortars underwent further analyses to determine the soluble silicates (SiO2) as well as soluble calcium oxide (CaO). The mass loss within the 200–600 °C interval indicated the chemically bonded water corresponding to the dehydration and dehydroxylation of hydraulic compounds and partly of aluminosilicate minerals. Using the above results, the percentage of CaO of hydraulic compounds of mortars was also determined. This percentage was obtained after subtracting the CaO corresponding to calcite determined from DTA/TG (> 600 °C) from the soluble CaO in HCl 2 M. Finally, the CO2/H2O ratio was calculated, where %CO2 corresponds to the decomposition of calcite, and the structural water (% wt.) chemically bonded to hydraulic compounds was determined by the loss on ignition in the intervals from 200–600 °C. Low values of this ratio indicate high hydraulicity of the mortar (Moropoulou, 2000; Maravelaki-Kalaitzaki et al. 2003). Based on the above values, the scatter diagrams of Figs. 9 and 10 were drawn.

Fig. 9
Fig. 9
Full size image

Classification of mortars based on their hydraulic components

Fig. 10
Fig. 10
Full size image

Classification of mortars based on calcite and insoluble solid residue

More specifically, in the plot of Fig. 9, mortars were classified based on their hydraulic components, as calculated by the methods described above. Mortars ΚΧ1, ΚΧ7 and ΚΧ11 contained the highest content of hydraulic components, followed by ΚΧ3, ΚΧ4, ΚΧ9 and ΚΧ10. On the other hand, in mortars ΚΧ2, ΚΧ6 and ΚΧ12, the CO2/H2O ratio is high, which classifies those mortars as aerial, i.e. mortars that harden only in the presence of air. Finally, regarding the diagram in Fig. 9, it is observed that hydraulicity decreases exponentially with increasing carbon dioxide, suggesting that calcite content and hydraulic compounds have an inverse proportional relationship. Therefore, mortars ΚΧ1, ΚΧ7, ΚΧ11 and ΚΧ3 are hydraulic, while ΚΧ12, ΚΧ6 and ΚΧ2 show moderate hydraulicity.

The diagram in Fig. 10 shows the ratio of %wt Calcite/%wt insoluble in HCl residue as a function of the (%wt) calcite content. The axes selected in this diagram correlate the total calcite content with components that are insoluble in hydrochloric acid; those components are primarily found in mortar aggregates. KX2 mortar contains the most calcite and the fewest insoluble HCl components. We observe that this classification is similar to the decrease in hydraulicity shown in Fig. 9.

In conclusion, mortars ΚΧ2, ΚΧ6 and ΚΧ12 differ in all indicators examined from the other mortars. Mortars KX1 and KX3, sourced from the fortified wall, are hydraulic mortars; however, KX1 contains a higher percentage of hydraulic components and also includes pieces of Portlandite that have not fully carbonized, as indicated by the XRD and thin section study. This last data shows that temperature and humidity conditions did not allow for the complete carbonation of lime, which explains the lower percentage of calcite found in this mortar compared to KX3 (Stefanidou et al. 2012). The hydraulic component and the acid soluble residue in HCl are indicators for the mortar classification from different construction phases and historic periods as documented in studies of structural mortars and plasters of monuments from different historic periods in Greece (Moropoulou 2000, 2005, Papayianni 2013, Stefanidou et al. 2014). Based on the indicators of the above chemical analyses, which have been used in mortar analyses of all historical periods from Greece, it follows that the mortars coming from the Wall resemble those that have been dated as Roman (Moropoulou, 2005).

Figure 11 shows the results from the mortar aggregate analysis. 13 chemical elements were selected as variables: PC1: Na, Mg, Si, K, Fe, Rb, Ba, Ni, Sr, As, Ca, Al and Zr, towards PC2: Si, K, Ca, Ti, Mn and Fe. This case study focused just on the first two main components, which accounted for a total explained variation of 75% (40% and 35%, respectively). This value is deemed appropriate for an adequate description of the entire dataset. It is observed that mortars ΚΧ2, ΚΧ6 and ΚΧ12 form one group, while the mortar that significantly differed from all the mortars, is the floor mortar ΚΧ7. The statistical analysis applied confirms the results obtained regarding the categorization of mortars by chemical methods. Although KX7 is clustered with KX1 and KX11 according to insoluble solid residue and hydraulic component results, it has been proven that it is significantly differed in the elemental composition from those samples and therefore the classification followed by grain size analysis and macro-microscopical observations corroborate the proposed grouping.

Fig. 11
Fig. 11
Full size image

Mortar classification based on elemental EDXRF aggregate analysis of mortars against two main components (PC1 and PC2)

4 Conclusions

The physicochemical study of mortars from the fortified wall and neighboring buildings on Katre Street aimed to classify chronologically and investigate the construction phases. By establishing appropriate indicators from the chemical analysis of mortars, mortar categorization was achieved, enabling further classification and comparisons to draw key conclusions regarding dating and manufacturing phases. Comparing the mortars under study with other, which had been dated according to the archaeological context, it was revealed that those from the wall belong to the Roman period, as they align well with the mortars of buildings identified from that time. This classification, derived from the conclusions of the chemical analysis, is further supported by the macroscopic characteristics of the mortars, the granulometric gradation of their aggregates and the multivariate analysis.

Physico-chemical analyses contribute to the extraction of valuable and reliable conclusions on the characterization of the construction phases and this comparative process allows to assign unknown mortars to a dated group with similar features. Roman mortars coming from the fortified wall are characterized as moderately hydraulic and might be restoration mortars of that period due to the severe earthquake that destroyed the fortifications. The hydraulicity is primarily due to the presence of hydraulic lime derived from the firing of marly limestones with aluminosilicate components that abound in the area. Ultimately, in some mortars hydraulicity is enhanced by the presence of crushed ceramic material that reacts with hydraulic lime and produces hydraulic compounds. Mortars are, therefore, characterized as medium strength natural hydraulic binder. The analyses provided valuable data on the composition of mortars that are compatible with those used in existing conservation interventions (Apostolopoulou and Moropoulou 2022). It is desirable for conservation mortar interventions hydraulic lime, fine-grained ceramic materials, and aggregates of a similar nature and granulometry to those found in the studied mortars to be used. The physicochemical and compositional analyses contribute to the understanding of technology and identification of the construction phases, and provide implications for designing conservation mortar with enhanced compatibility.