Available online at www.sciencedirect.

com

CERAMICS
INTERNATIONAL
Ceramics International 40 (2014) 16409–16419
www.elsevier.com/locate/ceramint

A combined use of optical microscopy, X-ray powder diffraction and

micro-Raman spectroscopy for the characterization of ancient ceramic
from Ebla (Syria)
Paolo Balliranoa, Caterina De Vitoa, Laura Medeghinia, Silvano Mignardia,n, Vincenzo Ferrinia,
Paolo Matthiaeb, Danilo Bersanic, Pier Paolo Lotticic
a
Department of Earth Sciences, Sapienza University of Rome, P.le Aldo Moro 5, 00185 Rome, Italy
b
Department of Sciences of Antiquities, Sapienza University of Rome, P.le Aldo Moro 5, 00185 Rome, Italy
c
Department of Physics and Earth Sciences, University of Parma, Parco Area delle Scienze 7/a, 43124 Parma, Italy
Received 17 July 2014; received in revised form 29 July 2014; accepted 29 July 2014
Available online 7 August 2014

Abstract

Ancient ceramics from the archaeological site of Ebla (Syria), dating back between 2250 and 1800 B.C., have been characterized by a
combined use of optical microscopy (OM), X-ray powder diffraction (XRPD) and micro-Raman spectroscopy (micro-RS). Petrographic
observations indicate that different fabrics are present, in terms of microstructure, groundmass and inclusions. XRPD allowed the identification
and quantification of mineral phases of both unheated and heated samples at 950 1C, by means of Rietveld refinements. In particular, XRPD of
heated samples highlights significant differences among the mineralogical assemblages of the analyzed samples, suggesting that the relative
amount of carbonate and silicate minerals plays a key role, driving the reactions during the firing process. Furthermore, the mineralogical
composition of unheated samples suggests a firing temperature in the 800–850 1C range, excluding a sample fired at lower temperatures (400–
500 1C). Concerning the redox state of the firing atmosphere, the occurrence of hematite in some samples indicates that they were fired in
oxidizing conditions, whereas other ceramic artifacts containing magnetite were fired in reducing conditions. Micro-RS results highlight that the
mineralogical components of the decorated surfaces are hematite in the red areas and magnetite in the black ones.
& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.

Keywords: Bronze age; Pottery; Syria; OM; micro-RS; Rietveld analysis

1. Introduction Furthermore, the investigation of the micro-textural relationships

between paste matrix and mineral inclusions, and their variations
The common interest of scientists and archaeologists con- with increasing firing temperatures provide useful information to
cerning analyses of ancient ceramics is mainly related to the reconstruct the overall pottery production process.
identification of their mineralogical, petrographic and chemical The transformation process during firing depends on the
features. The aim is to answer questions about the chemical grain size, mineralogy and chemistry of both clays and
composition of ceramic, as well as to explore the technological inclusions of the raw material and their relative abundances
aspects (firing temperature and redox state of the firing [1,6–10]. In particular, the chemical composition seems to
atmosphere), and to define the nature and the provenance of drive the crystallization of new phases, such as gehlenite,
raw materials [1–5]. wollastonite and anorthite.
For these reasons, in the investigation of archaeological
pottery the main challenge is the choice of appropriate
n
Corresponding author. Tel.: þ39 06 4991 4155; fax: þ 39 06 445 4729. advanced analytical methods, in addition to those routinely
E-mail address: silvano.mignardi@uniroma1.it (S. Mignardi). used in the field of cultural heritage, to obtain qualitative

http://dx.doi.org/10.1016/j.ceramint.2014.07.149
0272-8842/& 2014 Elsevier Ltd and Techna Group S.r.l. All rights reserved.
16410 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

and quantitative data for the complete characterization of the 2. Geological settings
ceramic artifacts [11]. Indeed, while OM is essential in the
definition of the ceramic fabric, XRPD and micro-RS The archaeological site of Ebla, located in the north-western
provide information on structural parameters and molecular Syria, is built on Paleogene limestone. Its name, meaning “white
structures [12,13]. In particular, laboratory parallel-beam rock,” seems to originate from the color of the sedimentary
X-ray powder diffraction has been recently used for very sequences outcropping in the area. A brief description of the
accurate structural investigations also in the field of cultural geological framework of the Ebla area is reported in Fig. 1.
heritage [14,15]. The experimental setup requires the pre- Syria is part of Levantine Basin, representing the southeastern-
paration of the sample in a capillary, the data being collected most sector of the Mediterranean Sea. It occurs in a complex
in transmission mode. Such geometry allows an almost geodynamic setting (i.e., a continental margin), between the
complete removal of the parasitic effect of preferred Arabian and Levantine plates. These plates are separated by the
orientation. Dead Sea transform fault. The Levantine basin, a thinned
In this study, a multi-analytical approach including OM, continental crust, originated during the opening of the Neo-
XRPD and micro-RS has been applied to Bronze Age Syrian Tethys [16]. Geological and geophysical data suggest that the
potteries from the archaeological site of Ebla. The Rietveld Levantine basin is mainly composed by a Phanerozoic sedimentary
refinement performed on XRPD data acquired in transmission succession overlying an igneous-metamorphic basement [16].
mode allowed an accurate quantification of mineral phases. During the Cenozoic (mainly Neogene), an extensive
The nature of the pigments used for the decorations of the igneous activity occurred in the Mashrek region, Middle East,
pottery has been investigated. Selected samples containing producing alkaline mafic rocks and basalts [16,17].
different amounts of carbonate minerals were re-fired at In the area of the archaeological site of Ebla, limestones,
temperature well above the decarbonation temperature (800– marls, having intercalated lenses of clay materials, and cherts
850 1C for Ca-rich materials). These results allowed us to (Paleogene) are present. Moreover, in the western and eastern
explore how the different mineralogical composition of the raw sectors of this area also ophiolites (Maastrichtian) and basaltic
material drives the final mineralogical assemblage of the rocks (Neogene-Quaternary) are present in outcrops (Fig. 1).
pottery.
Moreover, the results provide information about the firing 3. Archaeological setting
process of the ceramics, from which the technological back-
ground of the ancient Syrian population can be inferred; in Since 1964 “La Sapienza Expedition to Syria,” directed by
addition, these data can be very useful to trace the provenance Prof. Paolo Matthiae, carried out a systematic exploration of
of the raw materials. the ancient site of Ebla [18].

Fig. 1. A schematic geological map of central Syria reporting the location of Tell Mardikh-Ebla (modified from Maritan et al. [29]).
 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419 16411

Ebla, an artificial hill (Tell) near the modern village of Tell temperature, within the furnace, in about 10 h. This thermal
Mardikh, is one of the most important historical centers in treatment has been applied to assure the complete decarbona-
Syria, at about 55 km from Aleppo. Ebla was located at the tion process of coarse-sized carbonate grains.
crossroads of the caravan routes from Mesopotamia to central Data analysis was performed by the Rietveld method using
Syria and the Mediterranean coast. The archaeological site Topas 4.2 [21]. This program implements the Fundamental
occupied an area of about 56 ha and was marked by an outer Parameters Approach FPA [22]. FPA is a convolution
defensive wall. approach in which the peak-shape is synthesized from a priori
This site is worldwide famous for the Ebla tablets, an known features of the diffractometer (i.e., the emission profile
archive of about 20,000 cuneiform tablets, dating back to of the source, the width of the slits, the angle of divergence of
around 2250 B.C. the incident beam) and the microstructural features of the
A large number of potteries have been unearthed and they specimen. Peak shape was modeled through FPA with the peak
were dated mainly between the half of the third millennium broadening that was assumed to follow a Lorentzian (size) and
and the half of the second millennium B.C.; samples dating to a Gaussian (strain) behavior [23]. Peak positions were cor-
the first millennium B.C. until the Hellenistic Period have been rected for sample displacement from the focusing circle. The
found on the Acropolis [18]. background was fitted with a Chebyshev polynomial of the
The ceramics here studied are dated between the third and first kind. Refined parameters included: absorption correction,
second millennium B.C., i.e., Early Bronze Age IV (EB IVB, scale factors, cell parameters, and peak shapes for all minerals.
2250–2000 B.C.) and Middle Bronze Age I (MB I, 2000–1800 Quantitative phase analysis is restricted to crystalline phases,
B.C.). The fragments have been collected from two areas of as the possible occurrence of glass would be masked by the
the archaeological site: area HH in the Lower Town south-east contribution of the glass of the capillary. An example of the
has provided a long architectural and stratigraphic sequence Rietveld plots of sample TM.06.HHCE.575/16 is reported in
spanning the main phases of the town of Ebla; and area H Fig. 2.
located across the Western Rampart, where a probe provided Micro-Raman analysis was performed using a Jobin-Yvon
EB IVB ceramic materials from the ashy fillings accumulated Horiba Labram instrument with focal distance of 30 cm, 1800
in the Middle Bronze Age ramparts. grooves/mm grating, 100 mm slit width,  1.5 cm–1 spectral
resolution (Department of Physics and Earth Sciences, Uni-
4. Instrumentation versity of Parma, Italy). The measurements were carried out on
about 20 selected spots on the external and internal surface of
Petrographic analysis of thin sections was performed by each sherd. The 632.8 nm line from a He-Ne laser was used for
polarizing microscopy according to Whitbread criteria [19,20] excitation. The laser power was set lower than 1 mW on the
using a Zeiss D-7082 Oberkochen polarized optical micro- sample by means of neutral density filters to avoid undesired
scope (Department of Earth Sciences, Sapienza University, heating effects. An Olympus microscope with 10x
Rome, Italy). (NA ¼ 0.25), 50x (0.75), ULWD 50x (0.55), 100x (0.90)
The XRPD analysis of potsherds was carried out on a objectives and motorized x-y stage was used to focus on the
parallel-beam Bruker AXS D8 Advance, operating in transmis- samples with maximum spatial resolution of  2 mm. The
sion in θ-θ geometry (Department of Earth Sciences, Sapienza system was calibrated using the most intense Raman peak of
University, Rome, Italy). The instrument is fitted with an pure silicon at 520.6 cm–1 before each experimental session.
incident-beam Göbel mirror, a position-sensitive detector Raman spectra were acquired with 30 to 100 s integration time
(PSD) VÅNTEC-1 set to a 61 2θ aperture. The optical system in 3–5 accumulation cycles, according to the signal to noise
includes a 0.8 mm divergence slit, 3.71 Soller slits on the ratio. The spectra were baseline corrected and filtered as
incident beam, and a radial Soller system along the diffracted needed with LabSpecs software in order to remove back-
beam placed just before the PSD. ground fluorescence and noise and were analyzed using the
For XRPD analysis of ceramic materials containing small existing databases [24].
inclusions (Table 1), a fragment of each potsherd with size of
about 2  2  2 mm3 (  20 mg) has been selected. To obtain
reliable results, for sherds with large inclusions (up to 2 mm), 5. Materials
about 150–200 mg of sample have been collected.
The samples were gently hand ground in an agate mortar Twenty pottery samples of the EB IVB and MB I phases
under ethanol (particle size o 20 mm). The powders were have been investigated. They were selected on the basis of
loaded in 0.5 mm diameter borosilicate-glass capillaries their macroscopic features, i.e., grain size, color, decoration
aligned onto a standard goniometer head. XRPD data were and age. The potsherds consist of medium and small size
collected on this apparatus in step-scan mode in the 3–1451 2θ fragments of Simple and Painted Simple Ware jars, bowls and
angular range (CuKα), using a step size of 0.02191 2θ and fragments of a tall cylindrical goblet, representing the most
counting time of 10 s. A further series of patterns were typical vessels shapes and the most diffuse classes in the
collected on powdered fragments of the same samples fired pottery repertoire of Western Inland Syria.
from room temperature to 950 ˚C in approximately 12 h. After The color of the ceramic fragments varies from orange-red,
2 h at peak temperature, the samples were cooled to room light brown to very pale brown; some samples exhibit a
 16412
Table 1
Macroscopic and microscopic characteristics of Ebla samples.

Samples Grain Color Decoration Class Age Porosity Matrix Inclusions Fabric
Size

TM.70.H.1166/1 Fine Light red Red painted Simple painted EBIVB Porosity o10% 90% 5% o 0.5 mm. Frequent: iron oxides nodules, grog; 1
(Figs. 3,5) Ware common: fragments of limestone, calcite
Meso-vesicles Calcareous matrix
TM.06.HHW.341/95 Medium Light Not painted Simple Ware EBIVB Porosity o10% 80% 10% o 0.5 mm. Frequent: iron oxides nodules, 1
brown grog; common: fragments of limestone, calcite

P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

Meso and Calcareous matrix
micro-vesicles
TM.06.HHW.383/34 Fine Beige Not painted Simple Ware EBIVB Porosity 5% 90% 5% o 0.5 mm. Frequent: iron oxides nodules; 1
Meso-vesicles Calcareous matrix common: fragments of limestone
TM.06.HHW.383/37 Medium Light Not painted Simple Ware EBIVB Porosity o10% 80% 10% o 0.5 mm. Frequent: iron oxides nodules, 1
brown grog; common: fragments of limestone, calcite
Meso-vesicles Calcareous matrix
TM.06.HHCE.994/21 Fine Light Not painted Simple Ware EBIVB Porosity 5% 90% 5% o 0.5 mm. Frequent: iron oxides nodules; 1
brown common: fragments of limestone, calcite
Micro-vesicles Calcareous matrix
TM.06.HHCE.994/25 Fine Light red Not painted Simple Ware EBIVB Porosity 5% 90% 5% o 0.5 mm. Frequent: iron oxides nodules, grog; 1
Micro-vesicles Calcareous matrix common: fragments of limestone
TM.06.HHN.822/37 Medium Beige Not painted Simple Ware MBI Porosity o10% 85% 10% o 0.5 mm. Frequent: iron oxides nodules, 1
Meso and Calcareous matrix grog; common: fragments of limestone, calcite
micro-vesicles
TM.06.HHW.323/37 Coarse Beige Black painted Simple painted EBIVB Porosity 10% 50% 40% o1.0 mm. Dominant: fragments of limestone, 2
(Figs. 3,5) Ware fossils; common: iron oxides nodules
Meso-vesicles Calcareous matrix
and -vughs
TM.06.HHCE.994/22 Coarse Light Not painted Simple Ware EBIVB Porosity 10% 50% 40% o1.0 mm. Dominant: fragments of limestone, 2
brown fossils; common: iron oxides nodules
Meso-vughs Calcareous matrix
TM.06.HH.989/46 Coarse Beige Combed Simple Ware EBIVB Porosity 10% 50% 30% o1.0 mm. Dominant: fragments of limestone, 2
Meso-vesicles Calcareous matrix fossils; common: iron oxides nodules
and -vughs
TM.06.HHCE.994/26 Coarse Light Not painted Simple Ware EBIVB Porosity 10% 50% 40% o 1.0 mm. Dominant: fragments of limestone; 2
brown common: iron oxides nodules; rare: fossils
Meso-vesicles Calcareous matrix
and -vughs

TM.06.HHN.822/29 Coarse Light red Not painted Simple Ware MBI Porosity 10% 50% 30% o 1.0 mm. Dominant: fragments 2
Meso-vughs Calcareous matrix of limestone; common: iron oxides nodules, fossils
TM.06.HHN.822/43 Coarse Light red Not painted Simple Ware MBI Porosity 10% 50% 30% o 1.0 mm. Dominant: fragments 2
Meso-vesicles Calcareous matrix of limestone, fossils; common: iron oxides nodules
and -vughs
TM.70.H.1154/13 (Fig.5) Coarse Beige Combed and black Simple painted MBI Porosity 10% 50% 40% o 1.0 mm. Dominant: fossils; common: 2
painted Ware iron oxides nodules, fragments of limestone
Meso-vesicles Calcareous matrix
and -vughs
TM.06.HHW.383/31 Coarse Beige Not painted Simple Ware EBIVB Porosity 10% 60% 30% o 2.0 mm. Frequent: fragment of limestone 3
Meso-vesicles and Calcareous matrix and fossils, grog; common: fragment of basaltic
-vughs dark clay pellets rocks, iron oxides nodules; rare: grains of quartz
TM.06.HH.989/47 Coarse Light red Not painted Simple Ware EBIVB Porosity 10% 70% 20% o 2.0 mm. Frequent: fragment of limestone 3
meso-vesicles Calcareous dark clay and fossils; common: fragment of basaltic rocks,
pellets grog, iron oxides nodules
TM.06.HH.989/44 Medium Light Red decoration Simple painted EBIVB Porosity 10% 70% 20% o 2.0 mm. Frequent: fossils and grog; 3
brown Ware common: fragment of basaltic rocks, fragment of
Meso-vesicles and Calcareous limestone, iron oxides nodules; rare: grains of quartz
-vughs
TM.06.HHW.383/43 Coarse Light Not painted Simple Ware EBIVB Porosity 20% 60% 20% o 2.0 mm. Frequent: fragment of limestone 3

P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

brown and fossils, grog; common: fragment of basaltic
Meso-vesicles Calcareous dark rocks, iron oxides nodules
clay pellets
TM.06.HHN.822/33 Medium Light red Black decoration Simple painted MBI Porosity 10% 60% 30% o 2.0 mm. Frequent: fragment of limestone 3
Ware and grog; common: fragment of basaltic rocks, iron
Meso-vesicles and Calcareous dark oxides nodules, fossils; rare: grains of quartz
-vughs clay pellets
TM.06.HHCE.575/16 Coarse Beige Combed decoration Simple Ware MBI Porosity 10% 60% 30% o 2.0 mm. Frequent: fragment of limestone 3
(Figs. 3,5) and fossils, grog; common: fragment of basaltic
Meso-vesicles and Calcareous dark rocks, iron oxides nodules
-vughs clay pellets

16413
16414 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

Fig. 2. Rietveld plots of sample TM.06.HHCE.575/16. Blue: experimental; red: calculated; below: difference. Vertical bars refer to the position of calculated Bragg
reflections. Color code: same as in quantitative phase analysis.

surface decoration consisting of simple black or red painting fragments of sedimentary rocks, mainly limestone (from
while others present combed decorations (Table 1). rounded to sub-rounded and size 0.2–1.0 mm), fragments of
fossils (sub-rounded to well-rounded and size 0.1 to 0.2 mm)
6. Results and discussion and iron oxides nodules (with rounded shape and size around
0.1 mm).
6.1. Petrographic analysis Fabric (3) (6 samples) includes ceramic fragments with
porosity ranging between 10 to 20% of the total of the volume
The mineralogical and petrographic features as well as the mainly represented by meso-vesicles and rare meso-vughs, not
percentage, grain size and distribution of inclusions were aligned to the margin of the sample. The groundmass is
determined by OM analysis. The nature of the inclusions, calcareous and heterogeneous with diffuse clay pellets and the
their packing and the mean size were used to distinguish three color varies from brown to beige. The inclusions, with an
petrographic fabrics (Table 1), according to the criteria of unimodal grain-size distribution, representing 20–30% of the
Whitbread [19,20] (Fig. 3). total of the volume, consist mainly of dominant micritic calcite
Fabric (1) (7 samples) is characterized by porosity lower (from sub-angular to sub-rounded in shape and 0.1 to 1.0 mm),
than 10% of the total volume, consisting mainly of meso- fragments of sedimentary rocks, mainly limestone (rounded in
vesicles (0.05 to 0.5 mm in size) not aligned to the margin of shape with size in the range 0.2–1.5 mm), fragments of fossils
the sample in a heterogeneous calcareous groundmass; the (from angular to well-rounded and size 0.2 to 1.9 mm), grog
color ranges from beige to brown. The inclusions, 5–10% of (sub-rounded ranging from 0.2 to 1.0 mm), iron oxides
the total volume, have an unimodal grain-size distribution, nodules (with rounded shape and size ranging from 0.1 to
mainly consisting of calcite (from very-angular to sub-angular 0.5 mm), fragments of basaltic rocks (angular in shape and
in shape and size ranging from 0.1 to 0.3 mm), diffuse iron ranging from 0.5 to 2.0 mm) and rare quartz (from sub-angular
oxides nodules (with rounded shape and size from 0.1 to to angular ranging from 0.2 to 0.6 mm).
0.5 mm) and grog (sub-rounded and sub-angular, sized about On the basis of the mineralogical assemblage, all analyzed
0.3 mm). In some samples, fragments of fossils (sub-angular samples are Ca-rich due to the presence of abundant calcite.
and angular and size 0.3 to 0.4 mm) also occur. In addition, some samples of fabric 1 contain aragonite as
Fabric (2) (7 samples) displays a porosity of about 10% of well, which is a common constituent of fragments of fossil
the total of the volume mainly consisting of meso-vesicles and shells.
rare meso-vughs (0.05 to 0.5 mm in size), not aligned to the The occurrence of different fabrics in the same pottery class
margin of the sample. The groundmass is generally hetero- could be due to compositional variations within raw material
geneous and calcareous with a color varying from orange- sources. This feature suggests that the paste used in the
beige to dark brown. The potsherds are characterized by production of Simple and Painted Simple Ware from Ebla
inclusions with an unimodal grain-size distribution and a was not selected by the potters for the different uses of the
packing ranging between 30 and 40%. The inclusions are artifacts.
 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419 16415

we discuss the results of three samples, which are representa-

tive of the petrographic fabrics and of the major variations
observed in all analyzed potteries.
Calcite, quartz, clinopyroxene, gehlenite, and hematite are
ubiquitous though with relevant differences in their relative
abundances. As an example, calcite content spans from less
than 20 wt.% (TM.70.H.1166/1) to more than 50 wt.%
(TM.06.HHW.323/37), whereas the content of clinopyroxene
from ca. 10 wt.% (TM.06.HHW.323/37) to ca. 50 wt.%
(TM.70.H.1166/1). A high content of calcite indicates that
the potteries were fired at temperature below 900 1C (the
temperature of complete decomposition of calcite) or for a
short time that did not allow the complete decomposition.
Moreover, sample TM.06.HHCE.575/16 has a relevant content
of feldspars, mainly anorthite (ca. 13 wt.%). Besides, sample
TM.70.H.1166/1 contains ca. 18 wt.% of aragonite.
Evaluation of the cell parameters of clinopyroxene indicates
strong similarities among the samples. In fact, the a-parameter
ranges from 9.734(1) to 9.742(6) Å, the b-parameter from
8.867(1) to 8.879(2) Å, the c-parameter from 5.286(1) to
5.3048(6) Å, and the β angle from 105.99(4) to 106.093(8)1.
The refined values are consistent with those of the ferrian
aluminian clinopyroxene also known as fassaite [1,2].
The formation of fassaite could involve chlorite or mixed-
layer illite/chlorite as shown by reactions (1) and (2):
(Mg,Fe2 þ )5Al(Al,Si)3O10(OH)8 þ 4CaCO3 þ 5SiO2-4Ca
(Mg,Fe2 þ )Si2O6 þ (Mg,Fe)Al2O4 þ 4CO2 þ 4H2O (1)
þ
K(Fe34-2 ,Mg1-3)Σ5Al2(Si5-7Al3-1O20)(OH)10 þ 2CaCO3-2Ca
3þ
(Fe ,Mg)(Al,Si)SiO6 þ KAlSi3O8 þ Fe2O3 þ MgAl2O4
þ 2CO2 þ 5H2O (2)

Evaluation of the tetragonal cell parameters of melilite, i.e.,

gehlenite, indicated minor variation of the cell parameters with
the a-parameter ranging from 7.7299(7) to 7.7406(4) Å and the
c-parameter from 5.039(2) to 5.0444(5) Å. These values are
consistent with those of a member of the åkermanite-gehlenite
solid solution slightly displaced toward the gehlenite end-
Fig. 3. Thin section photomicrographs of representative samples of the three member. In fact, the average cell parameters compare favor-
petrographic fabrics (mag 2.5 x and crossed polarizers): 1) sample TM.70.
H.1166/1 Fabric (1); 2) sample TM.06.HHW.323/37 Fabric (2); 3) sample
ably with those of the sample of composition Ca2(Al0.65
TM.06.HHCE 575/16 Fabric (3). Mg0.35)[Al0.65Si1.35O7] investigated by Merlini et al. [25].
The QPA carried out on the heated samples indicate the
Finally, the petrographic and mineralogical features found in occurrence of relevant modifications of the non-treated miner-
the potsherds revealed a strong affinity of the raw material with alogical assemblages (Table 2, Fig. 4). An obvious effect of
the local geological sources suggesting a supply from the firing is the dramatic reduction of the calcium carbonate
surrounding area. content that is counterbalanced by the crystallization of neo-
formation phases as wollastonite and subordinate larnite
Ca2SiO4. Sample TM.06.HHW.323/37 (Fig. 4a) shows calcite
6.2. XRPD analysis after heating, due to the occurrence of abundant calcite crystals
as well as large fragments of calcareous fossils characterized
XRPD provided quantitative phase analyses (QPA) of the by a slower kinetics of decarbonation. In the case of TM.06.
potsherds. No attempt to quantify the amorphous content by HHCE.575/16 (Fig. 4b), a content of approximately 6 wt.% of
the spike method was performed. In fact, petrographic analysis portlandite Ca(OH)2 has been found whereas sample TM.70.
did not reveal the occurrence of any significant amorphous H.1166/1 (Fig. 4c) contains 1.3 wt.% of lime CaO.
component within the studied ceramics (Table 2). Significant No significant variation of the cell parameters of both
differences were observed among the mineralogical assem- gehlenite and clinopyroxene was observed. The abundant
blages of the various analyzed samples. For this reason, here occurrence of gehlenite, along with wollastonite and larnite
16416 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

Table 2
Quantitative phase analysis from XRPD data by Rietveld method.

Mineral phases Fabric 1 Fabric 2 Fabric 3

TM.70.H.1166/1 TM.06.HHW.323/37 TM.06.HHCE.575/16

Unheated 950 1C Unheated 950 1C Unheated 950 1C

Quartz 7.7(1) 3.9(1) 6.1(8) 4.9(2) 6.1(1) 3.3(2)

Plagioclase – 9.2(7) Traces Traces 12.2(7) 10.8(16)
K-feldspar – – Traces Traces 1.4(3) 9.1(12)
Clinopyroxene 46.6(7) 61.2(16) 8.9(4) 12.0(2) 15.8(3) 18.3(11)
Calcite 18.9(4) – 57.4(3) 7.6(6) 35.4(4) –
Aragonite 17.8(4) – – – – –
Clay Traces – – – Traces –
Gehlenite 7.1(7) 9.9(5) 27.1(2) 44.4(17) 27.5(4) 34.9(12)
Hematite 1.9(3) 1.3(1) 0.5(1) 1.5(7) 1.6(1) 1.8(4)
Portlandite – – – – – 6.4(6)
Wollastonite – 5.5(6) – 20.2(14) – 15.4(11)
Lime – 1.3(1) – – – –
Larnite – 6.7(13) – 9.0(18) – Traces

Traceso1 wt%. The numbers in brackets refer to the corresponding standard deviations.

after firing of TM.06.HHW.323/37 sample (Fig. 4a), is mainly In addition, the excess of calcium and the presence of silica
due to the effect of mass loss arising from calcite decarbona- in the system stabilized also larnite, a typical product formed in
tion. The free CaO in the system reacted with silica and contact metamorphism of limestone as well as in cements,
alumina to form the above mineralogical assemblage. The according to the reaction (5):
observed minor variations of cell parameters of both gehlenite
2CaO þ SiO2- Ca2SiO4 (5)
and clinopyroxene indicate a fairly limited cation exchange
and/or partial recrystallization of those phases during firing.
Besides, sample TM.06.HHCE.575/16 (Fig. 4b) shows an For sample TM.06.HHCE.575/16, in addition to the reactions
almost perfect coincidence between the cell parameters calcu- (3) and (4), the formation of K-feldspar (sanidine) could have
lated before and after firing. occurred at expenses of plagioclase in the boundaries of the
The QPA performed on these three samples, considering grains of this mineral due to the substitution of Na by K [1,2].
both unheated and heated aliquots, suggest the following Moreover, in this sample the clinopyroxene content increases,
reactions in the thermal range of 800–950 1C: following the reactions (1) and (2), and portlandite forms.
(a) increase of gehlenite (for example, TM.06.HHW.323/37, The formation of K-feldspar is generally explained with
Table 2) according to reaction (3) reactions involving muscovite/kaolinite and calcite. However,
as in our case muscovite was not found in the unheated
Al2O3 þ SiO2 þ 2CaCO3-Ca2Al2SiO7 þ 2CO2 (3) samples, the occurrence of clay minerals although in traces in
several samples supports the hypothesis that illite reacted with
and (b) formation of wollastonite (for example, TM.06. calcite to form K-feldspar (see for example sample TM.06.
HHCE.575/16, Table 2) favored by the high content of HHCE.575/16 in Table 2). The unheated sample TM.
carbonate according to the following reaction (4): H.70.1166/1 presents an unusual mineralogical assemblage
involving the coexistence of aragonite, calcite and gehlenite
CaCO3 þ SiO2-CaSiO3 þ CO2 (4) (Table 2, Fig. 4c). The presence of aragonite and calcite
suggests a firing temperature below the 400–500 1C range in
which the complete transformation of aragonite to calcite
However, there is experimental evidence that gehlenite is
occurs. However, in sample TM.H.70.1166/1 as XRPD
stable mainly in the 850–950 1C thermal range [3,26–28]. The
analysis detected clay minerals only in traces, stable at these
reactants calcium, silica and alumina, involved in the gehlenite
temperatures, the persistence of aragonite at temperature over
forming reaction, are provided by the breakdown of clay
500 1C along with gehlenite, which forms at temperature
minerals and carbonate.
above 800 1C, could be due to the presence of coarse-sized
Therefore, the occurrence of this mineral in some samples
fragments of shells, reducing condition of atmosphere and a
from Ebla suggests firing temperature in the 800–850 1C
short duration of the firing. The use of shell fragments as
range. Conversely, samples in which gehlenite has not been
tempering material was common during Bronze Age [29].
found were fired at lower temperature, unfavorable to the
For the heated sample TM.H.70.1166/1, in addition to
crystallization of neo-formation phases. In some samples
reactions (1) to (5) the formation of plagioclase occurs
containing gehlenite and both primary and newly formed
following reaction (6):
clinopyroxene, the firing temperature probably reached
900 1C as previously observed by Rathossi and Pontikes [2]. Al2O3 þ 2SiO2 þ CaCO3- CaAl2Si2O8 þ CO2 (6)
 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419 16417

minerals? A preliminary mass balance (Table 2) seems to

support this hypothesis. Indeed, the observed transformations
combined with petrographic observations suggest that the
differences are mainly related to mineralogical composition
of the raw material along with grain size, firing temperature,
duration and kiln atmosphere [1–3].

6.3. Micro-Raman spectroscopy

Micro-RS has been useful in the identification of pigments

used in the surface decorations of fragments.
In particular, in the spectra acquired in selected spots of the
red areas, the Raman bands at 224, 245, 299, 411, 498 and
611 cm–1 have been attributed to iron oxide hematite, whereas
a strong band at about 660 cm–1 typical of magnetite has been
found in the black areas (Fig. 5) [30]. This finding suggests a
different redox-state of atmosphere, i.e., oxidizing for red
paintings and reducing conditions for the black ones.
However, in some samples the presence of hematite in the
internal body and that of magnetite in the black decorated areas
has been observed. In this view, the concomitant presence of
different Fe-oxides is related to uncontrolled reducing/oxidiz-
ing conditions during the firing cycle.
In addition to the previous mineral phases, micro-RS
identified also quartz, by the strong band at 465 cm–1,
a medium intensity band at 207 cm–1 and a weak one at
356 cm–1.
Calcite has been recognized by a strong band at 1086 cm–1
and other weak bands at 712 and 281 cm–1. In some samples, a
two-peaked feature at around 500 cm–1, typical of feldspars,
has been observed.
In selected spots of the internal body, hematite and anatase
have been also detected. Titanium oxides were used in the past
as markers to estimate the firing temperature; at low tempera-
ture, anatase is the most common polymorph of titanium
dioxide, stable in 600–900 1C range, whereas above this
temperature threshold anatase converts into rutile [31]. There-
fore, the presence of anatase may indicate a firing temperature
lower than 900 1C [32]. This role of titanium oxide phases is
however controversial, as several investigations revealed that
different origin and environmental conditions could interfere
on the temperature transition [32–34].
Fig. 4. Magnified view of the diffraction patterns: a) TM.06.HHW.323/37; b)
TM.06.HHCE.575/16; c) TM.70.H.1166/1. Red color: untreated material;
black color: fired material. Relevant reflections assigned to the various minerals 7. Possible provenance of the raw materials
are marked. Qz ¼ quartz, Pl ¼ plagioclase (anhortite), Kfs ¼ K-feldspar,
Cpx ¼ clinopyroxene, Cal ¼ calcite, Arg ¼ aragonite, Gh ¼ gehlenite,
Concerning our pottery, this study points out that the
Hem ¼ hematite, Por ¼ portlandite, Wo ¼ wollastonite, Lrn ¼ larnite.
ceramic artifacts from Ebla were made employing Ca-rich
raw materials, in which the carbonate minerals were naturally
Moreover, the cell parameters of both melilite and clinopyr- present in the clays and/or added as temper. In addition, other
oxene after firing are significantly different with respect to the mineral inclusions and rock fragments as well as fossil shells
corresponding ones of the raw material. This fact clearly were probably added to improve the workability of the paste.
confirms the different origin of the raw material. The addition of temper is generally marked by a bimodal
An obvious question arises at this point. Are the observed grain-size distribution of the inclusions in thin section [35]. In
differences among different samples, and those encountered in our case, although grog occurs in fabric 1, the very small
each sample differently treated (i.e., unheated and heated), quantities of inclusions (5–10%) do not univocally support the
mainly driven by the relative amounts of carbonate and silicate hypothesis of a voluntary tempering process of the paste.
16418 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419

Fig. 5. Representative pottery samples from Ebla and Raman spectra collected on the body and on decorated surfaces. Sample TM.70.H.1166/1 Fabric (1); samples
TM.06.HHW.323/37 and TM.70.H.1154/13 Fabric (2); sample TM.06.HHCE.575/16 Fabric (3).

Therefore, the potter used a natural clay for the production of 2) XRPD provided quantitative phase analyses of the mineral
the wares of fabric 1. assemblage of the potsherds, highlighting that Ebla pottery is
For fabric 2 the unimodal grain-size distribution of the Ca-rich as testified by the presence of abundant neo-formed
inclusions observed seems to exclude the intentional addition silicates of Ca such as gehlenite, wollastonite and larnite.
of temper materials. On the contrary, fabric 3 is characterized 3) The comparison between data obtained in unheated and
by an apparent unimodal grain-size distribution; however, a heated samples provided information on the evolution at
bimodal distribution cannot be excluded, suggesting a temper- high temperature of Ca-rich ceramic. In addition, it allowed
ing process of the paste. the identification of unusual mineral assemblages and
The comparison of petrographic, spectroscopic and XRPD possible mechanisms of their coexistence.
data with the geological setting of the area supports the 4) Micro-Raman analysis gave exhaustive information on the
hypothesis that raw material was supplied from sources next mineralogical phases of the decorated surfaces of Ebla
to the archaeological site of Ebla. In particular, both clayey ceramic; in particular, the red decorations areas contain
materials and limestones occur in the vicinity of the archae- hematite and the black ones magnetite.
ological site (Fig. 1). The sources of quartz and feldspars as 5) The mineralogical composition of ceramic from Ebla
tempers were probably sands, which derived from the erosive suggests a firing temperature in the range of 800–850 1C
process of igneous-metamorphic rocks occurring in the area, and for some samples up to 900 1C, excluding sample TM.
whereas the presence of primary clinopyroxene in all samples H.70.1166/1. The analyzed samples were fired in oxidizing
could be ascribed to materials deriving from volcanic rocks. or reducing conditions to attain a particular color of the
decorated surfaces.
8. Conclusions

This work has demonstrated that the combined use of XRPD

and micro-RS along with OM is a useful multi-analytical Acknowledgments
approach to characterize ancient ceramics.
On the basis of qualitative and quantitative data we reached Financial support was provided by Sapienza University of
the following main conclusions: Rome and University of Parma. P.B., C.D., L.M. and S.M.
analyzed and interpreted the experimental data and wrote the
1) OM allowed to identify three different fabrics and to define manuscript with inputs from the other authors. The authors
both mineralogical and petrographic features of the pottery. thank the archaeologists M. D'Andrea and A. Vacca for
 P. Ballirano et al. / Ceramics International 40 (2014) 16409–16419 16419

morphological and stylistic analyses of the sherds. Finally, [17] A.F.M. Abdel-Rahman, P.E. Nassar, Cenozoic volcanism in the Middle
anonymous reviewers are thanked for the useful comments and East: petrogenesis of alkali basalts from northern Lebanon, Geol. Mag.
141 (2004) 545–563.
suggestions.
[18] P. Matthiae, Ebla. La città del trono, Archeologia e storia, Einaudi,
Torino (2010).
References [19] I.K. Whitbread, The characterization of argillaceous inclusions in ceramic
thin sections, Archaeometry 28 (1986) 79–88.
[1] C. Rathossi, Y. Pontikes, Effect of firing temperature and atmosphere on [20] I.K. Whitbread, Greek transport amphorae: a petrological and archae-
ceramics made of NW Peloponnese clay sediments. Part I: Reaction ological study, Fitch Laboratory Occasional Papers, Athens, British
paths, crystalline phases, microstructure and colour, J. Eur. Ceram. Soc. School at Athens, 1995.
30 (2010) 1841–1851. [21] AXS Bruker, Topas V4.2: General profile and structure analysis
[2] C. Rathossi, Y. Pontikes, Effect of firing temperature and atmosphere on software for powder diffraction data, Bruker AXS, Karlsruhe, Germany,
ceramics made of NW Peloponnese clay sediments: Part II. Chemistry of 2009.
pyrometamorphic minerals and comparison with ancient ceramics, J. Eur. [22] R.W. Cheary, A. Coelho, A fundamental parameters approach to X-ray
Ceram. Soc. 30 (2010) 1853–1866. line-profile fitting, J. Appl. Crystallogr 25 (1992) 109–121.
[3] P. Riccardi, B. Messiga, P. Duminuco, An approach to the dynamics of [23] R. Delhez, T.H. de Keijser, J.I. Langford, D. Louër, E.J. Mittemeijer, E.
clay firing, Appl. Clay Sci. 15 (1999) 393–409. J. Sonneveld, in: R.A. Young (Ed.), In The Rietveld Method, Oxford
[4] M.A. Al Maadeed, N.J. Al Thani, M. Sadeq, R. Sobott, Pottery from University Press, 1993.
Halat Aobeer and Hazem Al-Jasrah, Qatar, Ceram. Int. 39 (2013) [24] L. Burgio, R.J.H. Clark, Library of FT-Raman spectra of pigments,
8301–8309. minerals, pigment media and varnishes, and supplement to existing
[5] A. Issi, A. Kara, A.O. Alp, An investigation of Hellenistic period pottery library of Raman spectra of pigments with visible excitation, Spectro-
production technology from Harabebezikan/Turkey, Ceram. Int. 37 chim. Acta A 57 (2001) 1491–1521.
(2011) 2575–2582. [25] M. Merlini, M. Gemmi, G. Cruciani, G. Artioli, High-temperature
[6] M. Maggetti, Phase analysis and its significance for technology and behaviour of melilite: in situ X-ray diffraction study of gehlenite–
origin, in: J.S. Olin, A.D. Franklin (Eds.), Archaeological Ceramics, åkermanite–Na melilite solid solution, Phys. Chem. Miner. 35 (2008)
Smithsonian Institution, Washington, 1982, pp. 121–133. 147–155.
[7] G. Cultrone, C. Rodriguez-Navarro, E. Sebastian, O. Cazalla, M.J. De La [26] M.M. Jordán, A. Boix, T. Sanfeliu, C. de la Fuente, Firing transforma-
Torre, Carbonate and silicate phase reactions during ceramic firing, Eur. tions of cretaceous clays used in the manufacturing of ceramic tiles, Appl.
J. Mineral 13 (2001) 621–634. Clay Sci. 14 (1999) 225–234.
[8] C. Tschegg, T. Ntaflos, I. Hein, Thermally triggered two-stage reaction of [27] M. Hajjaji, S. Kacim, Clay–calcite mixes: sintering and phase formation,
carbonates and clay during ceramic firing – A case study on Bronze Age Br. Ceram. Trans 103 (2004) 29–32.
Cypriot ceramics, Appl. Clay Sci. 43 (2009) 69–78. [28] B. Moroni, C. Conti, Technological features of Renaissance pottery from
[9] B. Fabbri, S. Gualtieri, S. Shoval, The presence of calcite in archeological Deruta (Umbria, Italy): An experimental study, Appl. Clay Sci. 33 (2006)
ceramics, J. Eur. Ceram. Soc. 34 (2014) 1899–1911. 230–246.
[10] M.J. Trindade, M.I. Dias, J. Coroado, F. Rocha, Mineralogical transformations [29] L. Maritan, C. Mazzoli, V. Michielin, D. Morandi Bonacossi, M. Luciani,
of calcareous rich clays with firing: A comparative study between calcite and G. Molin, The provenance and production technology of Bronze Age and
dolomite rich clays from Algarve, Portugal, Appl. Clay Sci. 42 (2009) Iron Age pottery from Tell Mishrifeh/Qatna (Syria), Archaeometry 47
345–355. (2005) 723–744.
[11] C. De Vito, L. Medeghini, S. Mignardi, D. Orlandi, L. Nigro, [30] D. Bersani, P.P. Lottici, A. Montenero, Micro-Raman investigation of
F. Spagnoli, P.P. Lottici, D. Bersani, Technological fingerprints of iron oxide films and powders produced by sol–gel syntheses, J. Raman
Black-GlossWare from Motya (Western Sicily, Italy), Appl. Clay Sci Spectrosc. 30 (1999) 355–360.
88-89 (2014) 202–213. [31] N. Bouzidi, A. Bouzidi, P. Gaudon, D. Merabet, P. Blanchart, Porcelain
[12] R. Scarpelli, R.J.H. Clark, A.M. De Francesco, Archaeometric study of containing anatase and rutile nanocrystals, Ceram. Int. 39 (2013)
black-coated pottery from Pompeii by different analytical techniques, 489–495.
Spectrochim. Acta A 120 (2014) 60–66. [32] F.C. Gennari, D.M. Pasquevich, Enhancing effect of iron chlorides on the
[13] V. Morra, A. De Bonis, C. Grifa, A. Langella, L. Cavassa, R. Piovesan, anatase-rutile transition in titanium dioxide, J. Am. Ceram. Soc. 82
Minero-petrographic study of cooking ware and pompeian red ware (rosso (1999) 1915–1921.
pompeiano) from Cuma (southern Italy), Archaeometry 55 (2013) 852–879. [33] M. Sendova, V. Zhelyaskov, M. Scalera, M. Ramsey, Micro-Raman
[14] P. Ballirano, A. Maras, Mineralogical characterization of the blue spectroscopic study of pottery fragments from the Lapatsa tomb, Cyprus,
pigment of Michelangelo's fresco The Last Judgment, Am. Mineral. 91 ca 2500 B.C, J. Raman Spectrosc. 36 (2005) 829–833.
(2006) 997–1005. [34] Y.S. Liou, Y. Chang Liu, H.Y. Huang, Micro-Raman spectroscopic study
[15] P. Ballirano, C. De Vito, S. Mignardi, V. Ferrini, Phase transitions in the of cord-marked pottery decorated with red coatings from Taiwan, ca
Mg-CO2-H2O system and the thermal decomposition of dypingite, 2600–1700 B.C, J. Raman Spectrosc. 42 (2011) 1062–1068.
Mg5(CO3)4(OH)2∙5H2O: Implications for geosequestration of carbon [35] W.K. Barnett, The identification of clay collection and modification in
dioxide, Chem. Geol. 340 (2013) 59–67. prehistoric potting at the early Neolithic site of Balma Margineda,
[16] G.L. Netzeband, K. Gohl, C.P. Hubscher, Z. Ben-Avraham, Andorra. In Recent Developments in Ceramic Petrology, A. Middleton
G.A. Diehghain, D. Gajewski, P. Liersch, The Levantine Basin-crustal and I. Freestone eds., British Museum Occasional Paper No. 81 (1991)
structure and origin, Tectonophysics 418 (2006) 167–188. 17-38.