Ca & PB Opacifier Usage in Roman Tesserae

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Appl. Phys.

A 92, 109–116 (2008) Applied Physics A


DOI: 10.1007/s00339-008-4456-8 Materials Science & Processing

s. lahlil1,u Rediscovering ancient glass technologies


i. biron1
l. galoisy2 through the examination of opacifier crystals
g. morin2 1 C2RMF Centre de Recherche et de Restauration des Musées de France, 14 quai François Mitterrand,
Palais du Louvre, Porte des Lions, 75001 Paris, France
2 IMPMC, Institut de Minéralogie et de Physique des Milieux Condensés, Universités Paris 6 et 7,
IPGP, 140 rue Lourmel, 75015 Paris, France

Received: 10 October 2007/Accepted: 15 January 2008 In this paper, we aim to understand how two groups of
Published online: 4 April 2008 • © Springer-Verlag 2008 ancient opaque glass products were obtained by comparing
their physical and chemical characteristics with the contem-
ABSTRACT The aim of the study is to understand how anti- porary opaque glass production (“corpo” technology). Ne-
monate opacifying crystals were obtained throughout history.
vers lampworking glass figures, dating from the 18th cent.
Two archaeological glass productions opacified with calcium
and lead antimonates are studied in this paper, in order to A.D. – coming mainly from the French “Musée des Arts
rediscover ancient opaque glass technologies: Roman mosaic Décoratifs” – are studied through seven white glasses opaci-
tesserae (1st cent. B.C.–4th cent. A.D.) and Nevers lampwork- fied with Ca-antimonates, and three green glasses opaci-
ing glass (18th cent. A.D.). The fine examination of crystalline fied with Pb-antimonates. Eight pink, blue and turquoise co-
phases and of the vitreous matrix is undertaken using vari- lored Ca-antimonate opacified Roman mosaic tesserae and
ous and complementary techniques. Results are compared with three Pb-antimonate opacified glasses (two yellow and green
a modern reference production, for which the technological tesserae, and one yellow decorated bead), dating from the 1st
process is well known. We demonstrate that Ca-antimonate cent. BC–4th cent. AD, are studied.
opacifiers in Roman mosaic tesserae, as well as in Nevers An approach based on the fine examination of the crys-
lampworking glass, were obtained by in situ crystallization. talline phases and of the vitreous matrices, using comple-
Nevertheless, Roman and Nevers glass would have undergone mentary analytical techniques, is proposed. Furthermore, we
different firing processes. We propose that the addition of pre- attempt to use the glass microstructure and the elements sub-
viously synthesized crystals or the use of “anime” could be the stituted in the antimonate crystals as indicators of the fabrica-
process used to obtain Pb-antimonate opacified glass, for both tion conditions and of the raw materials used.
productions studied. We demonstrate that CaO, PbO and Sb2 O3
Toproduce these crystals, three opacification processes are
concentrations in the bulk compositions and in the matrices, and
their evolution with the crystallinity ratio, offer robust criteria proposed: the addition to a translucent glass of natural crystals
for the distinction of the opacification process used. Also, the or crystals synthesized separately, the addition to a translucent
different crystalline structures help to provide information on glass of an opacifier-rich glass called “corpo”, and the in situ
the experimental conditions. crystallization (separation from the melt of the crystals). The
“corpo” technique is studied through the reference production
PACS 81.05.Kf; 81.10.Fq; 61.50.Lt; 61.05.cp; 82.80.Ej of contemporary mosaic tesserae made by the Orsoni Com-
pany in Venice. It consists of a three step process.
– A soda-lime-silica glass containing lead is prepared in
1 Introduction
a furnace at a temperature of 900 – 1000 ◦ C (antimony is in-
Glass is usually opacified by small crystalline par- troduced as oxidized stibnite). This glass, called “corpo”,
ticles, called opacifiers, dispersed in the vitreous matrix. is white due to the high amounts of Ca-antimonate crystals
Different types of opacifier have been used throughout his- formed in situ.
tory. Antimonates are especially interesting because they have – Independently, a translucent soda-lime-silica Pb-rich
been used for a long time, from Ancient Egypt to the present glass, colored or not, is prepared.
time, with periods of interruption. Two types of antimonate – To opacify this translucent glass, the “corpo” is then in-
can be distinguished: Ca-antimonate (white colored) and Pb- troduced in several steps at a temperature around 900 ◦ C,
antimonate (yellow colored). The crystal color may or may and its concentration is adjusted according to the tint de-
not induce a modification of the glass coloration. For example, sired. In the case of uncolored glass, a white opaque glass
the combination of a translucent turquoise matrix with Ca- is obtained; in the case of colored glass, the addition of the
antimonate crystals results in a lightened turquoise opaque “corpo” allows one to make different hues (the crystals act
glass, whereas the combination of a translucent turquoise ma- as clarifiers).
trix with Pb-antimonate crystals results in a green opaque In contrast, in the in situ crystallization process stricto sensus,
glass. Ca- or Pb-antimonate crystals are formed by the diffusion and
the combination of the elements present in the raw material
u Fax: +33-140202422, E-mail: sophia.lahlil@culture.gouv.fr mixture or in the melt. The quantity and the nature of crys-
110 Applied Physics A – Materials Science & Processing

tals formed depend on the chemical composition of the glass, coupled with the SEM (EDX), and electron microprobe ana-
on the oxidation state of antimony and on the experimental lyses (EPMA). The results in weight oxides were normalized
conditions used (temperature, duration, cooling rate, furnace at 100%. Samples presenting very small crystals (< 1 µm)
atmosphere and nature of the crucible). were also observed with a field electron gun (FEG; Gemini
15-25, Carl Zeiss instruments) at 5 kV (SE images) or 20 kV
(BSE images) accelerating potential.
2 Analytical methods
2.1 Microstructure and chemical analyses 2.2 Crystallographic measurements

Millimeter-sized fragments of glass were removed X-ray diffraction analyses (XRD) were performed
from all the samples studied. These fragments were embed- directly on Orsoni and Roman mosaic tesserae using a X-Pert
ded in a polyvinyl resin block and polished using diamond Pro diffractometer (Panalytical device) with vertical go-
pastes down to 0.25 µm, vacuum coated with carbon and ex- niometer under Bragg–Brentano geometry, with Cu K α Ni-
amined in a scanning electron microscope at 20 kV accelera- filtered radiation. Measurements were performed in a 5–120◦
ting potential (SEM; JEOL 840, Oxford Instruments). Ele- range, with a time step of 0.017 s. Suitable databases were
mental compositions of vitreous matrices and crystals were used for peak assignments using X-Pert Panalytical software.
determined using an energy-dispersive X-ray spectrometer Rietveld refinement was performed with FullProf software.

FIGURE 1 SEM-BSE images of Ca-antimonate crystals in a white contemporary Orsoni glass, in a white Nevers lampworking glass and in a blue Roman
mosaic tesserae
LAHLIL et al. Rediscovering ancient glass technologies through the examination of opacifier crystals 111

Microdiffraction (µDRX) measurements were performed in If the density, the morphology and the chemical homo-
the C2RMF, (Cu K α , 5–50◦ range) on Nevers lampworking geneity of the crystals are all dependent upon their formation
glass samples as they all measure less than 3 mm length. history, the observation of the microstructure is not always
sufficient to define the technological process used. Develop-
ment of an imaging treatment is in progress, in order to find
3 Results and discussion evidence of new features in the glass microstructure, charac-
3.1 Calcium antimonate opacified glass teristic of each production.

3.1.1 The microstructure as an indicator of the technological 3.1.2 Using CaO and Sb2 O3 concentrations to identify the
conditions. In the Orsoni, Nevers and Roman glasses studied, opacification process. In order to evaluate the crystal quan-
the microstructure is characterized by a rather homogeneous tity in each opaque glass, we define the crystallinity ratio
distribution of small geometrical crystals of Ca-antimonates, as the surface of crystals/surface of the amorphous phase in
with sometimes small aggregates of various shapes and sizes a sample. It is determined by the area ratio according to BSE
(Fig. 1). The amount of crystals varies from one sample to an- images and by theoretical calculations. We assume that CaO
other. Considering Orsoni glass, we notice that the addition and Sb2 O3 concentrations in the bulk compositions (vitreous
of a “corpo” to a translucent glass modifies neither the crystal matrix + crystals) and in the vitreous matrices and their evo-
distributions nor their morphologies. lution with the crystallinity ratio could be good criteria for the
Some differences can be noticed: large aggregates are distinction of different opacifying processes.
present in Nevers and Roman glasses, but not in the Orsoni In the reference samples of Orsoni contemporary glass
glasses (Fig. 2). The individual crystal size is generally lower (soda-lime-silica glass containing lead, Table 1), the amounts
in Nevers (mean size ∼ 3 µm) and Roman glasses (mean size of CaO and Sb2 O3 in the bulk compositions, as well as in the
∼ 2 µm), compared to Orsoni glass (mean size ∼ 4 µm). The vitreous matrices, are positively correlated (Fig. 3) and their
large aggregates can have very peculiar morphologies and values increase with the crystallinity ratio. This is due to the
relatively high sizes (from 7 to 70 µm) (Fig. 2). These ob- fact that they are opacified using the “corpo” process: during
servations are in good agreement with previous studies made the opacification, part of the “corpo” (matrix and crystals) can
on white and blue Roman glasses [1–3]. In turquoise Roman dissolve within the translucent glass to re-establish the ther-
tesserae, aggregates are especially different: they are very fre- modynamic equilibrium. Consequently, the more “corpo” is
quent, have generally high sizes (from 20 to 200 µm), and added, the higher the amounts of CaO and Sb2 O3 in the bulk
are constituted by small crystals compacted together (Fig. 2). compositions and in the vitreous matrices.
These noticeable differences could indicate variable techno- In Roman mosaic tesserae (soda-lime-silica glass, Table 1)
logical conditions and/or the possible influence of copper – and Nevers lampworking glass (mixed alkali-lime-silica glass
used as a glass colorant in these tesserae – on the precipitation with low lead contents, Table 1) the evolutions of the amounts
of Ca-antimonate crystals. of CaO and Sb2 O3 in the bulk compositions are very different

FIGURE 2 SEM-BSE images of Ca-antimonates aggregates in a white Nevers glass and in blue and turquoise Roman mosaic tesserae
112 Applied Physics A – Materials Science & Processing

Production Orsoni glass 21st cent. A.D. Nevers glass 18th cent. A.D. Roman glass 1st cent. B.C.–4th cent. A.D.
Opacifier Ca-antimonate Ca-antimonate Pb-antimonate Ca-antimonate Pb-antimonate
Color Corpo White Blue Turquoise White Green Pink Blue Turquoise Green Yellow
Nb samples 1 3 3 1 7 3 1 5 2 1 2

Na2 O 15.5 ± 0.7 13.1 ± 2.8 16.8 ± 1.8 14.5 ± 0.6 6.5 ± 1.6 9.1 ± 4.6 16.8 ± 0.7 15.3 ± 0.5 17.8 ± 0.5 18.3 ± 0.5 10.8 ± 4.3
MgO nd nd 0.3 nd 2.3 ± 0.5 1.7 ± 0.9 1.4 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.2
Al2 O3 nd nd 0.4 nd 0.5 ± 0.4 0.3 ± 0.2 1.7 ± 0.2 2.2 ± 0.2 1.7 ± 0.7 1.7 ± 0.4 0.9 ± 0.6
SiO2 47.3 ± 2.2 54.2 ± 4.1 51.9 ± 1.0 49.8 ± 0.6 58.1 ± 3.0 59.8 ± 4.0 67.3 ± 1.2 67.9 ± 0.6 64.8 ± 1.8 65.4 ± 1.3 55.6 ± 16.8
SO3 0.5 nd 0.6 nd 0.5 ± 0.1 0.4 ± 0.3 0.4 ± 0.1 0.4 ± 0.1 0.2 ± 0.4 nd 0.9 ± 0.2
K2 O 0.2 ± 0.1 0.1 ± 0.1 0.7 ± 0.9 0.6 ± 0.1 10.8 ± 1.3 9.3 ± 1.5 0.6 ± 0.1 0.6 ± 0.2 0.5 ± 0.1 0.7 ± 0.1 0.3 ± 0.5
CaO 5.5 ± 0.3 3.2 ± 1.8 1.6 ± 0.4 2.3 ± 0.1 8.6 ± 1.7 5.6 ± 1.4 6.2 ± 0.3 8.2 ± 0.1 4.5 ± 0.5 6.3 ± 0.3 5.3 ± 1.1
Fe2 O3 nd nd nd nd 1.0 ± 1.0 0.3 ± 0.2 0.8 ± 0.1 0.9 ± 0.2 0.8 ± 0.1 0.8 ± 0.3 2.5 ± 1.5
CoO nd nd nd nd nd 0.1 ± 0.1 nd 0.1 ± 0.1 nd nd nd
CuO nd nd 1.3 ± 1.1 1.9 ± 0.1 nd 3.8 ± 2.0 nd 0.2 ± 0.1 3.7 ± 0.2 2.7 ± 0.4 0.1 ± 0.1
SnO2 nd nd nd nd nd nd nd nd nd 0.3 ± 0.3 nd
Sb2 O3 10.6 ± 1.0 6.2 ± 3.4 3.0 ± 0.5 3.4 ± 0.2 6.4 ± 1.0 0.6 ± 0.2 3.8 ± 1.2 2.9 ± 0.7 4.1 ± 0.7 1.0 ± 0.8 3.5 ± 2.5
PbO 20.3 ± 1.6 22.8 ± 1.5 24.0 ± 1.0 25.4 ± 1.4 4.3 ± 2.5 8.7 ± 5.8 nd nd nd 2.5 ± 0.6 18.7 ± 17.4

nd = not detected
TABLE 1 Elemental compositions obtained by EDX measurements of three opaque glass productions expressed in weight percent oxide

from the contemporary glasses (Fig. 3). Both glass produc- courages us to think that Roman mosaic tesserae were made
tions would have been made using an opacifying process at higher temperature than Nevers and contemporary glasses
different from the “corpo” one. It seems that the proportions (T > 900 – 1000 ◦ C). Therefore, assuming that the same opaci-
of CaO and Sb2 O3 were adjusted according to the opacity de- fication process – in situ crystallization – was used, diffe-
sired, and consequently, that Ca-antimonates had crystallized rences observed in Roman and Nevers glass productions
in situ. could be due to different experimental conditions and/or to
The in situ crystallization of Ca-antimonates in Roman different glass compositions (Fig. 3). In order to test this hy-
glass has been suggested by some authors according to mi- pothesis and to evaluate the phase transition temperature of
crostructural observations and chemical analyses [1–7]. They the Ca2 Sb2 O7 –CaSb2 O6 system in vitreous matrices, synthe-
propose that antimony was added as stibnite in the raw ses of Ca-antimonate opacified soda-lime-silica glasses are in
form (Sb2 S3 gray mineral) [8], or as roasted stibnite (Sb2 O3 , progress in the laboratory.
Sb2 O5 ) [1, 2], to the raw materials or to a translucent glass. The presence of elements such as Na or Pb, in the vitreous
Foster and Jackson [7] obtained Ca-antimonate opacified matrix, could also play a role in the crystallization of Ca-
soda-lime-silica glass by in situ crystallization using two antimonates. Indeed, significant amounts of Na are detected
manufacturing methods: in large aggregates of Ca2 Sb2 O7 crystals, in turquoise Roman
tesserae (Table 2). These glasses have the same levels of Na2 O
– a one stage process where Sb2 O3 (6 wt. %) was introduced
and lower levels of CaO than the other Ca-antimonate opaci-
in the raw materials;
fied Roman glasses but contain copper as a colorant (Table 1).
– a two stage process, where Sb2 O3 (6 wt. %) or Sb2 S3
(6.99 wt. %) was added to a typical Roman soda-lime-
silica glass and the batches were re-melted.
The chemical analyses and the crystalline structure of Ca-
antimonates have not been investigated.

3.1.3 Using crystal structure to approach experimental con-


ditions (temperature, duration, cooling rate, ...). Generally,
Ca-antimonates are present in opaque glass as Ca2 Sb2 O7
(orthorhombic structure) and CaSb2 O6 (hexagonal struc-
ture) [1, 2, 8–10]. We suggest that the ratio of the contents
of Ca-antimonate phases CaSb2 O6 /Ca2 Sb2 O7 , could provide
information on the range of temperature used to produce
opaque glass. In fact, the synthesis of pure crystals of Ca-
antimonates shows that CaSb2 O6 is formed at higher tem-
perature than Ca2 Sb2 O7 . CaSb2 O6 crystallizes from 927 ◦ C at
the expense of Ca2 Sb2 O7 and becomes the major phase from
1094 ◦ C [11].
In contemporary and Nevers glasses, Ca-antimonate com-
pounds are mainly present as Ca2 Sb2 O7 with an orthorhom-
bic structure, whereas in Roman glasses, the two crystalline FIGURE 3 CaO vs. Sb2 O3 concentrations in the bulk compositions (vit-
phases are always detected: CaSb2 O6 hexagonal is the major reous matrix + crystals) of Orsoni glass (full squares), Nevers glass (full
phase and Ca2 Sb2 O7 orthorhombic is the minor phase. It en- triangles) and in Roman tesserae (hollow triangles) (EDX analyses)
LAHLIL et al. Rediscovering ancient glass technologies through the examination of opacifier crystals 113

FIGURE 4 SEM-BSE images of crystals distri-


bution in an opaque green Nevers lampworking
glass and in an opaque green Roman mosaic
tesserae

FIGURE 5 SEM-BSE images of yellow aggre-


gates of Pb-antimonate crystals in an opaque
green Nevers glass and in an opaque green Ro-
man mosaic tesserae

wt.% Orsoni glass 21st cent. AD Nevers glass 18th cent. AD Roman glass 1st cent. BC–4th cent. AD
Color White-blue-turquoise White Turquoise
Nb analyses ∼ 70 ∼ 40 ∼ 20

Na2 O nd nd 5.5–11.5
CaO 20.6–22.3 69.6–73.1 6.5–9.0
Sb2 O3 65.8–67.9 23.8–25.4 71.5–83.4
PbO 8–10 1–3 nd

nd = not detected
TABLE 2 Elemental compositions obtained by EDX measurements of Ca-antimonate crystals in the three opaque glass productions expressed in weight
percent oxide

In the same way, Ca is partially substituted by Pb – coming 2. the average concentration of lead in the crystals is very
from the vitreous matrices – in Ca2 Sb2 O7 crystals in all con- stable for a large number of analyses (Table 2),
temporary and Nevers glasses (Tables 1 and 2). These latter 3. the Rietveld refinement of XRD patterns is in good agree-
crystals undoubtedly contain Pb because: ment with the incorporation of Pb in the structure of Ca-
antimonate crystals.
1. the EDX and EMPA analyses are mainly undertaken on
large size crystals (∼ 10 µm > EDX and EPMA beam ra- This feature has not yet been reported in the literature. The
diation areas), presence of lead in the glass could possibly aid the precipi-
114 Applied Physics A – Materials Science & Processing

FIGURE 6 Fe2 O3 vs. Sb2 O3 and PbO concentrations in the bulk compositions of Pb-antimonate opacified Roman glass (EDX analyses)

tation of Ca-antimonate crystals as suggested by Foster and no stoichiometric formula has been established (PbO = 44–
Jackson [7]. 56 wt. %; Sb2 O3 = 18–35 wt. %; SnO2 = 9–20 wt. %). Fur-
thermore, isolated crystals of SnO2 are sometimes observed in
3.2 Lead antimonate opacified glass Nevers glass. Note that the presence of tin in Pb-antimonate
crystals of Roman mosaic tesserae has not been reported yet.
Orsoni glass could not serve as a modern reference For both productions, it is possible that tin was brought
in the study of Nevers and Roman Pb-antimonates opacified with raw materials, or was added deliberately, because only
glasses because yellow and green contemporary tesserae are traces of SnO2 are detected in the vitreous matrices of Ne-
now opacified with cadmium sulfide. vers and Roman glasses, and aggregates with a tin-rich center
Roman glass is soda-lime-silica glass whereas Nevers are present. The low concentrations of Sn in glass matrices
lampworking glass is mixed alkali-lime-silica glass. They could also reflect the fact that tin is less soluble than Sb in
both contain variable amounts of lead (Roman glass: PbO = glass. It would nucleate first and serve as a nucleus for the
2–32 wt. %, Nevers glass: PbO = 5–16 wt. %, EDX ana- antimonate crystals. Tin playing a role in Pb-antimonate crys-
lyses) (Table 1). Yellow glasses are colored by yellow Pb- tallization would explain why it is present in crystals of glass
antimonate crystals, whereas green glasses are colored by the productions from periods more than 1400 years apart. For, Ro-
same opacifiers and copper oxide dissolved in the vitreous man glass, tin seems to occur mainly in the opacifiers of the
matrix. green glasses and its concentration is higher in the green than
in the yellow glasses. This may be a result of the addition of
3.2.1 The microstructure as an indicator of the technological tin with copper in the form of bronze scale, a well known co-
conditions. The microstructure of Roman and of Nevers lorant in ancient glasses. The Sn : Cu ratios are right for a 10%
glasses is characterized by a rather homogeneous distribution Sn bronze (Table 1).
of geometrical crystals in the vitreous matrix (Fig. 4). In both Tin is commonly found with a mixture of Pb and Sb, in
productions, several aggregates formed by crystals closely other productions of opaque glass or in yellow glass-based
packed together are observed (Fig. 5). The morphology of pigments, dating from the 16th –late 19th cent. A.D. [12–18].
these aggregates differs notably in the two groups of glass Moreover, recipe books of Murano glassmakers, contempo-
studied. In Roman glass, aggregates are often constituted by rary with Nevers glass, dating from the 18th –19th cent. A.D.,
very small acicular or granular crystals (∼ 0.5 µm), whereas describe the preparation of a yellow pigment called “anime”,
in Nevers glass, aggregates are larger, more frequent, and con- from raw materials containing Pb, Sb, tin and sand [18–21].
stituted by regular geometrical crystals (mean size = 2 – 3 µm) Recipes report that in a first step, metallic lead and tin were
(Fig. 5). These noticeable differences could indicate variable calcinated together, in order to obtain an oxide of lead-tin yel-
technological conditions. low. This product was then mixed with Sb sulphide, minium
In both productions, opacifiers contain mainly Pb and Sb and sand, to obtain a glassy yellow pigment, called “anime”,
and correspond to a crystalline phase close to Pb2 Sb2 O7 with which was used as an opacifying product [18, 19]. In this pro-
a cubic pyrochlore structure. SEM-EDX and EPMA measure- cess, tin is introduced with lead, as lead-tin yellow (type I
ments show that tin oxide is sometimes present in aggregates Pb2 SnO4 or type II PbSnO3 ), or as residual cassiterite SnO2 -
of Nevers and Roman glasses. Heterogeneous aggregates are by-product of the tin calcination [19].
illustrated by BSE images (Fig. 5): white areas correspond to Differently from Nevers glass, the presence of iron in
a mixture of Pb, Sb and Sn oxides, whereas gray areas cor- the crystals of Roman glass is detected (Fe2 O3 = 3.85 wt. %;
respond mainly to tin oxide (SnO2 = 90 wt. %). While tin is mean value EDX analyses). It indicates that Pb2 Sb2 O7 opaci-
present only in aggregates in Roman glasses, it is also de- fiers, in the two productions, would not have been elabo-
tected in isolated crystals in Nevers lampworking glass and rated from the same raw materials. The correlations between
LAHLIL et al. Rediscovering ancient glass technologies through the examination of opacifier crystals 115

the amounts of Fe2 O3 , PbO and Sb2 O3 in the bulk com- “anime” was powdered and added to a translucent glass melt,
positions (to confirm with more analyses) let us think that at a relatively low temperature, in order to prevent the yel-
iron would come from Sb and/or Pb sources or would have low pigment from dissolving within the glass [18–21]. Thus,
been added on purpose during crystal elaboration (Fig. 6a such an opacification process consists in the addition of syn-
and b). Crystal stoichiometry in aggregates would be close to thesized crystals directly to the melt. Nevertheless, more ana-
Pb2 (Fe y Snx Sb2−y−x )O7−y−x/2 with x = 0–1 and y = 0–1. The lyses on new samples and XRD measurements, as well as
presence of iron in yellow opacifiers of Roman glasses has not syntheses of opaque glass in laboratory are in progress in
been reported yet in the literature. order to help us discovering how these glasses were made.
It is known that the substitution of Sb in Pb-antimonate
opacifiers can modify the crystal coloration: tin would slightly 4 Conclusions
brighten the yellow color, whereas iron produces a yellow-
orange color [17–19, 22]. The detailed study of the vitreous matrices and
crystals of opaque glasses is a very profitable approach for the
3.2.2 Using PbO, Sb2 O3 concentrations to identify the opaci- determination of the technological processes used to produce
fication processes. Considering Roman tesserae, the evolu- ancient opaque glass. Our results show that the in situ crys-
tion of PbO and Sb2 O3 concentrations in the bulk analyses tallization process was used to produce Roman tesserae and
show that they are positively correlated (Fig. 7). These preli- Nevers lampworking Ca-antimonate opacified glasses. Ne-
minary results show a general trend already observed in yel- vertheless, Roman and Nevers glass would have undergone
low and green Roman glasses [2]. If the same evolution with different firing processes. In green Nevers lampworking glass,
the crystallinity ratio is verified through further analyses, the yellow lead-tin-antimonate opacifiers could have been pro-
hypothesis that yellow and green Roman glasses were opaci- duced using the contemporary “anime” recipes (addition to
fied by the addition to the melt of natural or previously synthe- the glass of opacifiers synthesized separately). In the case of
sized crystals or by using a “corpo” process will be confirmed. Roman glass opacified with Pb-antimonates, preliminary re-
The interpretation of different studies based on microstruc- sults demonstrate that a similar process, but still different from
tural and chemical analyses tends to confirm this hypothe- Nevers glass, would have been used. Nevertheless, for both
sis. Mass et al. [2] suggest that natural Pb-antimonates were productions, more experiments and measurements should be
present in litharge sources resulting from the cupellation pro- done to confirm these hypotheses.
cess, and that once added at a relatively low temperature to Useful criteria to distinguish different opacification pro-
the melt, a further crystallization occurred during the cooling. cesses are proposed: assuming a homogeneous production,
Brun et al. [3] propose the addition of powdered Pb-Sb com- the correlation between the concentrations of CaO (or PbO)
pounds, which would decompose due to the temperature at and Sb2 O3 in the bulk compositions and in vitreous matri-
which it is added to the glass. At this stage of the study, we ces and their evolution with the crystallinity ratio can help
cannot discard any of these hypotheses. to differentiate a “corpo” technique, or an addition of crys-
Considering Nevers lampworking glass, the evolution of tals, from an in situ crystallization. Furthermore, in the case
PbO and Sb2 O3 concentrations in the bulk analyses are still of Ca-antimonates opacified glass, the presence of the two
not sufficient to determine the opacification process. How- crystalline phases CaSb2 O6 and Ca2 Sb2 O7 and their relative
ever, clues are provided by the study of ancient glassmaking proportions could provide information about the range of tem-
recipes and techniques contemporary to Nevers production perature used to produce opaque glass. Besides, elements sub-
(Muranese recipe books of the 18th –19th cent. A.D.). The stituted in the antimonate crystals could give clues about the
experimental conditions and/or the raw materials used and
can modify the crystals coloration.
To test the different hypotheses proposed and to under-
stand both the influence of the initial mixture composition
and the firing process on the crystallization, syntheses of
Ca- and Pb-antimonate opacified glass are in progress in the
laboratory.

ACKNOWLEDGEMENTS For allowing us the access to the


glass samples studied, we are very grateful to the following glassmakers,
scientists and curators: The Angelo Orsoni Company (Venice), C. Moretti,
J.-L. Olivié (Musée de Arts Décoratifs, Paris), P. Santopadre (Istituto Cen-
trale del Restauro, Rome), M. Verità (Stazione Sperimentale del Vetro,
Venice). Experiments were successfully performed only with the fruitful co-
operation of other scientists: P. Walter (C2RMF), M. Fialin and F. Couffignal
(Service Camparis, Université Paris 6), S. Poissonnet (CEA). Special thanks
to Y. Adda for his everyday support and advices. M. Menu, M. Verità, I. Free-
stone, P. Lehuédé and R. Pillay are gratefully acknowledged for their help.

REFERENCES

1 M. Bimson, I.C. Freestone, J. Glass Stud. 25, 55 (1983)


FIGURE 7 Sb2 O3 vs. PbO amounts in the bulk compositions of Pb- 2 J.L. Mass, R.E. Stone, M.T. Wypyski, Mater. Res. Symp. Proc. 462, 193
antimonate opacified Roman glass (EDX analyses) (1997)
116 Applied Physics A – Materials Science & Processing

3 N. Brun, M. Pernot, B. Velde, Actes du colloque de Namur Technique et 14 C. Salerno, S. Tommasi Ferroni, Art et Chimie: la Couleur (CNRS
science: les arts du verre, 97–109 (1989) éditions, Paris, 2000), pp. 43–48
4 J. Henderson, Mater. Res. Soc. Proc. 185, 601 (1991) 15 U. Santamaria, P. Moioli, C. Seccaroni, Art et Chimie: la Couleur
5 M. Verità, A. Renier, S. Zecchin, J. Cult. Herit. 3, 261 (2002) (CNRS éditions, Paris, 2000), pp. 38–42
6 M. Uboldi, M. Verità, J. Glass Stud. 45, 115 (2003) 16 C. Sandalinas, S. Ruiz-Moreno, Stud. Conserv. 49, 41 (2004)
7 H.E. Foster, C.M. Jackson, Glass Technol. 46, 327 (2005) 17 J. Dik, E. Hermens, R. Peshar, H. Schenk, Archaeometry 47, 593 (2005)
8 W.E.S. Turner, H.P. Rooksby, Glastechn. Ber. Sonderband: V. Int. 18 C. Moretti, S. Hreglich, Rivista della Stazione Sperimentale del Vetro 5,
Glaskongress 32K(VIII), 17 (1959) 28 (2005)
9 I. Biron, P. Dandrige, M.T. Wypyski, M. Vandevyver, L’oeuvre de Limo- 19 C. Moretti, S. Hreglich, Glass Technol. 25, 277 (1984)
ges (Edited by Réunion des musées nationaux Paris 1995), pp. 48–62 20 L. Zecchin, Vetro e vetrai di Murano III (Arsenale Venice, Venice,
10 A.J. Shortland, Archaeometry 44, 517 (2002) 1990)
11 K.H. Butler, M.J. Bergin, V.M.B. Hannaford, J. Electrochem. Soc. 97, 21 C. Moretti, C.S. Salerno, S. Tommasi Ferroni, Ricette vetrarie muranesi-
117 (1950) Gasparo Brunoro e il manoscritto di Danzica (Nardini, Florence,
12 R.J.H. Clark, L. Cridland, B.M. Kariuki, K.D.M. Harris, R. Withnall, 2004)
J. Chem. Soc. Dalton Trans. 2577 (1995) 22 I.N. Wainwright, J.M. Taylor, D.H. Rosamond, Artists’ pigments 1, 219
13 A. Roy, B.H. Berrie, IIC Dublin (1998), p. 160 (1986)

You might also like