Content uploaded by M.s. Tite
Author content
All content in this area was uploaded by M.s. Tite on Dec 07, 2017
Content may be subject to copyright.
The invention of lustre: Iraq 9th and 10th centuries AD
T. Pradell
a,
*, J. Molera
b
, A.D. Smith
c
, M.S. Tite
d
a
Department Fı´sica i Enginyeria Nuclear, UPC, campus Baix Llobregat. ESAB. Av. Canal Olı´mpic, 08860 Castelldefels, Spain
b
GRMT, Department Fı´sica, Universitat de Girona, 17071 Girona/GRMA, Universitat de Vic, 08500 Vic, Spain
c
STFC, SRS Daresbury Laboratory, Keckwick Lane, Warrington WA4 4AD, UK
d
Research Laboratory for Archaeology and the History of Art, Dyson Perrins Building, South Parks Road, Oxford OX1 3QY, UK
Received 12 March 2007; received in revised form 6 August 2007; accepted 21 August 2007
Abstract
In this paper the study of four significant lustre samples covering 9th century AD polychrome and 10th century AD monochrome lustre from
Iraq is presented. The samples selected are representative of the earliest known lustre productions. The data obtained from the study of the
medieval samples are compared to laboratory reproductions and gives important clues about the invention, perfection and success of lustre dur-
ing this period. The change from polychrome to monochrome lustre decorations and the increase in the lead content of the glazes are the key
parameters in the success of obtaining a golden lustre.
Ó2007 Elsevier Ltd. All rights reserved.
Keywords: Early Islamic; Lustre; Technology; Electron microprobe; SR-micro-XRD; EXAFS; UVevis spectroscopy
1. Introduction
1.1. Historical
The first reduced copper and silver pigments were ap-
plied on clear glass objects and date back to 8th and 9th
century AD (Fustat and Basra, respectively) (Caiger Smith,
1991; Carboni, 2001). In 1970, Brill (1970) suggested that
these decorations were produced by ionic exchange between
the alkalis of the glass and the copper and silver of the lus-
tre paint such as in lustre ceramics (Pradell et al., 2005).
However, they never show the characteristic metallic shine
but instead appear as iridescent stains (Carboni, 2001).
The first lustre decorations applied on glazed pottery were
found in the Caliphs’ palace in Samarra (836e883 AD), al-
though they were most probably produced earlier, in the time
of Harun-al-Rashid (766e809 AD). Abbasid lustres were found
as well in the Mosque of Kerouian in Tunisia, the court of the
Hammanid princess in Qal’a (Algeria), and also in the court
of Ahman ibn Tulun (Fustat, Egypt) who was the administrator
of the Samarra court posted to Egypt in 868 AD, and who rebuilt
the city of Fustat in imitation of the luxury displayed in Sa-
marra. The lustre decorated pieces of the present study corre-
spond to this latest group (Caiger Smith, 1991; Mason, 2004).
The pottery was glazed and highly coloured and decorated
to resemble glass and metal vessels. The earliest Islamic
glazes were of the alkali-lime type, slightly underfired to re-
tain small bubbles and other unreacted particles, and thus to
produce a white substrate on which the decorations show up.
Then lead was progressively introduced with cassiterite simul-
taneously added to improve the opacity (Caiger Smith, 1991;
Mason, 2004). The earliest production of lustre is contempo-
rary to the beginning of mixed alkali-lead tin opacified glazes.
The 9th century AD Abbasid lustres are polychrome show-
ing two, three and sometimes four lustre colours, the designs
are intricate and the effect striking. Typical colours are olive
green; brown and amber; orange, yellow and crimson and
also extremely dark, almost black, red. Most of the time,
they do not show metallic shine but instead appear as iridescent
stains. Sometimes, metallic and non-metallic lustre colours of
different composition appear mixed in the same piece. Some
* Corresponding author. Tel.: þ34 935521129; fax: þ34 935521001.
E-mail address: trinitat.pradell@upc.edu (T. Pradell).
0305-4403/$ - see front matter Ó2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jas.2007.08.016
Journal of Archaeological Science 35 (2008) 1201e1215
http://www.elsevier.com/locate/jas
lustre colours such as copper reds need to be more strongly re-
duced than others and, in particular, silver based green metallic
golden lustre is quite easily produced and could occur acciden-
tally. Soon this golden lustre was appreciated and by the begin-
ning of the 10th century AD an olive-green monochrome
golden lustre production started. Sometimes they did not suc-
ceed in producing the golden metallic lustre but greenish-
yellow non-metallic lustres. These new lustres are associated
with the production of new shapes (cups, beakers, jugs, jars)
as well as out-rimmed bowls and plates with new designs con-
sisting of a central single animal (gazelles, elephants and birds)
or figure (e.g. lute players) with a background of contour panels
with a texture recalling woven textiles (Caiger Smith, 1991).
In this paper we selected from the Ashmolean Museum a set
of four pieces of lustre representative of the polychrome and
monochrome lustres. Two of the pieces correspond to poly-
chrome lustre from the 9th century AD; one with non-metallic
dark brown and amber lustre forming small flowers (p67); and
a second piece of non-metallic amber and olive-green golden
metallic lustre with the typical decorations of alternated amber
and green lines and circles (p51). The other two correspond to
the olive-green monochrome lustres from the 10th century AD,
a flat out-rimmed plate successfully golden (p32) and a non-
metallic small out-rimmed bowl showing a single central ani-
mal, a gazelle (p37). The front and rear decorations of these
pieces are shown in Fig. 1. These same pieces have already
been studied by Mason (2004) in a broader study of Islamic
glazed pottery dated from 8th to 14th centuries AD from
Iraq, Iran, Syria and Egypt, who gave full characterisation of
the ceramic pastes and glazes. The pastes are all dense, homo-
geneous and creamy known as ‘‘Samarra body’’ (Caiger Smith,
1991). The ceramic paste was studied by Mason (2004), who
calls it ‘‘Basra clay’’, and corresponds to a highly calcareous
clay (50e55% SiO
2
,20e25% CaO, 12% Al
2
O
3
, 6% MgO,
6% Fe
2
O
3
,1e2% Na
2
O). The glazes are mixed alkali-lead
tin opacified glazes with various additions of lead and tin show-
ing similar composition and microstructures for all the lustres
indicating a common production technique.
The lustre microstructure results from the use of a sophisti-
cated technique of production that involves a high level of em-
pirical knowledge of the behaviour of materials and the need
to control the temperature and atmosphere of the kiln during
firing. This empirical knowledge was mainly achieved during
the early production of lustre in Iraq and thus, the study of the
early Iraqi lustre is fundamental to understanding the progress
and success of lustre technology.
1.2. What is lustre?
Lustre is a nanosized metal-glass composite made of metal
copper and/or silver nanoparticles embedded in the glassy
matrix (Perez-Arantegui et al., 2001). The metal particle sizes
range between 5 nm and 50 nm and form a layer of varying
thickness of between 100 nm to 1 mm(Friederick et al., 2004;
Perez-Arantegui et al., 2004). The colour and transparency
of the layers are due to the absorption and scattering of light
in the layer. A continuous metal layer is transparent for
wavelengths below a given value (the Plasmon frequency)
and absorbs most of the light for larger wavelengths. Typically
the plasmon frequencies for metals lie in the ultraviolet, making
the metals colourless and opaque, although for some metals like
gold or copper they lie in the visible, 2.4 eV (520 nm) and
2.1 eV (590 nm) respectively, giving the characteristic yellow
and red colours. Metal nanoparticles are transparent for all
wavelengths with the exception of the enhanced absorption at
the surface plasmon resonance frequency. The position and
width of this depends on the nature and size of the nanoparticles
and the nature of the glassy media (Kreibig and Vollmer, 1995).
Typically, Cu and Ag are both present in the lustre layer, which
varies in composition from 100% silver to 100% copper (Borgia
et al., 2002; Padeletti and Fermo, 2003, 2004; Pe
´rez-Arantegui
et al., 2001, 2004; Darque-Ceretti et al., 2005; Padeletti et al.,
2006). The lustre colour is directly related to the Cu/Ag ratio
and the nanostructure. Moreover, the presence of Ag
þ
,Cu
þ
and Cu
2þ
(Padovani et al., 2003, 2004; Smith et al., 2003,
2006), either dissolved in the glassy matrix or forming crystal-
line compounds, also affects the final colour shown by the lustre
layers. Broadly speaking, silver lustre is yellow or green, and
copper lustre is amber, brown and red. However, nanocrystal-
line cuprite also produces a yellow lustre, Cu
2þ
dissolved in
the glass is green and the silver rich lustre then sometimes
appears dark brown.
Another peculiar optical property of lustre is its capability
of reflecting the light like a continuous metal surface. The
specular reflected light for a metal surface is negligible below
the plasmon wavelength and about 90% above the plasmon
wavelength. For the case of metal nanoparticles in a glassy
media, the increase in the reflectivity is related to the transition
from incoherent scattering from individual clusters into geo-
metric-optical transmission and reflection. This is due mainly
to the interference among the scattered electromagnetic waves
of all the particles. This phenomenon is linked to the presence
of a high density of metal nanoparticles and is also favoured
by the bigger size of the nanoparticles (Dusemund et al.,
1991; Farbman et al., 1992; Kreibig and Vollmer, 1995;
Collier et al., 1997). Copper rich lustres may reflect the light
like metal copper and silver rich lustres like metal gold. Pre-
vious reproduction studies of lustres indicate that the metal
reflectivity in the lustre layers is straightforwardly obtained
when the lustre layers are applied over lead-containing glazes
(Pradell et al., 2006; Molera et al., 2007; Pradell et al., in
press). This result was found to be consistent with archaeolog-
ical findings where low lead and lead-free glazes do not show
metallic shine ei.e. some early 9th century AD polychrome
productions from Iraq, 13th century AD Syrian Raqqa ce-
ramics and early lustre decorated glass objects from Egypt
(Caiger Smith, 1991; Carboni, 2001; Mason, 2004).
Finally, lustre decorations quite often show blue irides-
cences. Berthier et al. (2006) and Reillon and Berthier
(2006) have simulated the specular reflection for silver lustres
and demonstrated that its colour may change from yellow to
blue dependant upon the viewing angle and with a colour
range that varies with the density and size of the particles
and with the lustre layer thickness.
1202 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
Most relevant is that, although the colours and appearance
of lustres from several places and times may seem very dif-
ferent, their study has consistently shown that they all share
the same microstructure and composition and consequently
a common technology of production (Borgia et al., 2002;
Friederick et al., 2004; Padovani et al., 2003, 2004; Padeletti
and Fermo, 2003, 2004; Pe
´rez-Arantegui et al., 2001, 2004;
He
´lary et al., 2005; Darque-Ceretti et al., 2005; Colomban
and Truong, 2004; Colomban and Schreiber, 2005; Padeletti
et al., 2006).
front back
P67
P51
P37
P32
Fig. 1. Pictures taken from the front and back sides of the four lustres from Iraq corresponding to 9th century AD polychrome lustres (p67 and p51) and to 10th
century AD monochrome lustres (p32 and p37). For the interpretation of the colour in this figure, the reader is referred to the web version of this article.
1203T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
1.3. Lustre technology
Lustre is produced by the application of a raw paint over
the glass or glaze which is fired at temperatures of about
550 C in a reducing atmosphere. After cooling, the remaining
paint is washed off revealing the lustre beneath (Caiger Smith,
1991; Pradell et al., 2004). Lustre is thus a reaction layer ob-
tained from the interaction between the paint and the glaze
surface; Cu
þ
and Ag
þ
diffuse into the glassy surface where
they are transformed into metal nanoparticles by means of a re-
ducing process. The mechanism responsible for the penetra-
tion of Cu
þ
and Ag
þ
is their ionic exchange with the Na
þ
and K
þ
ions contained within the glass (Pradell et al., 2005).
The composition of the lustre paint is therefore different
from that of the final lustre layer. Lustre paint recipes are
very diverse but in all the cases include clay, copper or/and sil-
ver compounds and a sulphur-containing compound (Caiger
Smith, 1991; Allan, 1973; Pradell et al., 2004). The use of cop-
per sulphides, silver and sulphur are described in the early
14th century AD Persian treatise by Abu
¨’l-Qa
¨sim and cinnabar
(HgS) was found in the 13th century AD Hispano Moresque
workshop from Paterna (Valencia) (Molera et al., 2001); the
decomposition of cinnabar produces a sulphur reducing atmo-
sphere, which at temperatures between 500 C and 600 C
forms copper and silver sulphates and sulphides and reduces
Cu
2þ
to Cu
þ
(Pradell et al., 2004). These compounds are sim-
ilar to those used in recipes described in the Persian treatise.
Reproduction of lustre in laboratory conditions, but using
the same materials (paint mixtures) and similar firing temper-
atures and atmospheres to those used in medieval times was
successfully attempted by Pradell et al. (2006), Molera et al.
(2007) and Pradell et al. (in press). The sulphur reducing con-
ditions developed in the paint itself allow the formation of the
compounds responsible for the ionic exchange. Further exter-
nal reduction is necessary to form the metal nanoparticles.
Oxidising/neutral conditions allow the formation of cuprite
nanoparticles giving a yellow colour to the layer but further
reduction is necessary to produce metal copper nanoparticles
and the red ruby lustres. Silver results in the formation of a yel-
low/green lustre layer consisting of metal silver nanoparticles.
Silver tends to form aggregates from a few microns to some
hundreds of microns across that appear brown. The size of
the nanoparticles and lustre layer thickness was found to de-
pend on the glaze composition. Metal nanoparticles of about
12 nm and layers of about 700e800 nm thick were obtained
on a glaze produced from a commercial alkali-borate frit,
glaze-a, (48.5% SiO
2
, 12.0% Na
2
O, 7.1% K
2
O, 23.2% B
2
O
3
,
7.2% Al
2
O
3
, 0.9% CaO, 0.1% MgO, 0.1% Fe
2
O
3
and
<0.5% PbO) and metal nanoparticles up to 30 nm and layers
of about 250e300 nm were obtained on a mixed glaze pro-
duced from the mixture of the commercial alkali-borate frit
with PbO, glaze-m, (43.5% SiO
2
, 3.5% Na
2
O, 2.9% K
2
O,
11.0% B
2
O
3
, 5.7% Al
2
O
3
, 0.9% CaO, 0.1% MgO 0.1%
Fe
2
O
3
and 31.8% PbO). For firing temperatures of about
550 C or higher, metal-like reflectivity is always attained
for the lead-containing glaze. The metal-like shining lustre
layers are thinner and have a higher volume fraction of metal
nanoparticles, up to 15% (Pradell et al., 2007). This indicates
a lower diffusion of Cu
þ
and Ag
þ
into the glaze and an
enhanced nucleation and growth of the metal nanoparticles.
Early polychrome Iraqi lustre contains significant amounts
of both copper and silver. These mixed lustres were not stud-
ied in our previous work (Pradell et al., 2006; Molera et al.,
2007), and only one sample was obtained in Pradell et al.
(in press). Here we report the results for mixed copper and
silver lustres obtained using a mixed paint (half and a half
of copper and silver). The lustre layer colour, reflectivity,
chemical composition and crystalline nanoparticles thus pro-
duced are determined.
2. Analysis and characterisation
A full analysis and nanostructure characterisation of the
lustre layers is best achieved by a combination of different tech-
niques. In this paper we report the results obtained by: electron
microprobe (WDX), to assess ionic exchange; Micro-XRD, to
determine the size and type of nanoparticles present in the layer;
UVevis diffuse reflectance to obtain information about the
type, density, size and growth habits of the silver nanoparticles;
and XANES and EXAFS to establish the valence and local
structure of Cu and Ag in the lustre layers.
Chemical analyses of the lustre surface and of a section of
the glaze were obtained by Electron Microprobe, Cameca S-50
(WDX), experimental conditions being 1 mm spot size, 15 kV
and 10 nA probe current except for Na and K for which the
probe current was reduced to 1 nA and the spot size increased
to about 5 mm. Broadly speaking, the penetration depth of the
microprobe is greater (about 2 mm) than the lustre layer thick-
ness (about 0.5 mm), so that the analyses give information on
the overall composition. The mechanisms involved in lustre
formation are ionic exchange and diffusion, and since the dif-
fusion of elements involves a greater depth than the lustre
layer thickness itself, the analysis of a larger region gives a bet-
ter indication of the changes involved in the lustre formation.
However, the chemical composition of the lustre layer itself is
only indirectly assessed. Backscattering images of the glaze
sections and of the lustre surfaces were also obtained to deter-
mine the microstructure of the glazes and the homogeneity of
the lustre layers.
The crystalline species forming both glazes and lustre
layers were determined by synchrotron radiation micro X-ray
diffraction (SR-mXRD) analysis of the glaze cross-sections
and of thinned preparations from the lustre layers. Glaze sec-
tions were prepared by embedding the sample in resin and cut-
ting and polishing of the surfaces. Thin preparations (between
50 mm and 100 mm) of the lustre layers were obtained by
grinding and polishing. SR-mXRD was performed on beamline
9.6 at the SRS, Daresbury Laboratory in transmission geome-
try, using 0.87 A
˚wavelength, 200 mm spot size and recorded
using a CCD detector.
UVevis diffuse reflectance (DR) was measured directly on
the surface of the lustre layers. A small circular spot of 5 mm
was used to collect the data. The data are presented as log(1/
DR) which is equivalent to absorption for highly absorbing
1204 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
materials such as metal. The characteristic SPR absorption
peaks associated with the metal nanoparticles producing lustre
appear in the UVevis. The study of the corresponding UVe
vis spectra gives information on the type and size of the nano-
particles (Kreibig and Vollmer, 1995).
Finally, L-edge Cu X-ray absorption near edge Spectros-
copy (XANES) was obtained in beamline 3.4 at SRS Dares-
bury laboratory with a beam spot of 2 mm. L-edge XANES
probes the transition from the 2p level to the valence band
and can provide detailed information on the unoccupied den-
sity of states of the ligand and metal ions, and thus distinguish
between oxidation states of copper. The spectra were acquired
in fluorescence mode using a Ge solid-state detector. As the
layer of metal nanoparticles forming the lustre is very thin
(<1mm), the ‘thin-concentrated’ behaviour gave good fluores-
cence data. Silver L
3
edge (at 2.98 keV) EXAFS was also
collected on beamline 3.4. The narrow range of 173 eV only
equates to a k-range of 7, but due to the high scattering
from neighbouring silver atoms, structural information could
be obtained. EXAFS, extended X-ray absorption fine structure
(Stern, 2001) measurements at the Cu K-edge were collected
on the microfocus beamline 9.2 also at SRS Daresbury labora-
tory in fluorescence mode for amber areas of sample p51. The
sub-50 mm focal spot available on this instrument enabled the
selection of different locations showing different colours and
chemical composition.
3. Results
3.1. Replication studies
The replications were obtained following the same proce-
dure described by Pradell et al. (2006) and Molera et al.
(2007) using the same basic lustre paint recipe, but containing
equal amounts of copper and silver. Two glaze compositions
without lead (glaze-a), and the same glaze with added lead bi-
silicate (glaze-m); and two firing protocols, the first at 550 C
using a single oxidising or neutral gas, and the second also at
550 C combining first a oxidising/neutral gas and then
a reducing gas, were used. A summary of the results obtained
is shown in Fig. 2.
Without an applied reducing atmosphere amber and yellow-
ish-orange colours are obtained. Following a reducing protocol,
brown and dark brown-golden colours result. Golden metallic
shine is only achieved when the lustre is applied over a lead-
containing glaze and following the oxidising/neutral-then-
reducing protocol. The lustre layers are silver rich and form
highly concentrated regions. Chemical data indicate that the
mechanism of lustre development are by ionic exchange. SR-
mXRD analysis showed the presence of nanocrystalline silver
in all the cases. Typical sizes of 15e20 nm for the lead-free
glaze (j19 and j17), and 18e35 nm for the lead-containing
glaze (j94 and j93). The sizes measured are also slightly bigger
than the nanoparticles obtained for pure silver lustres in the
same conditions. However, metallic copper nanoparticles are
not found and, only under the reducing protocol is cuprite
observed. Therefore, copper is mainly dissolved in the glassy
matrix as Cu
þ
or Cu
2þ
. This result is consistent with Cu K-
edge micro-EXAFS data taken on medieval silver and copper
containing lustres (Padovani et al., 2003, 2004; Smith et al.,
2006); copper is fully oxidised in the silver rich areas of the de-
signs and reduced to metal only in the areas where silver is not
present. The explanation for this is that silver is much more
easily reduced to metal than copper. Therefore, silver is
reduced from Ag
þ
to metal and at the same time hampers the
reduction of Cu
þ
to metal. In some cases the reduction of silver
can lead to the oxidation of copper to Cu
2þ
. Moreover, as the
presence of copper helps silver to reduce to metal, the silver
metal nanoparticles formed are bigger than in the case of
a pure silver lustre.
The UVevis spectra corresponding to the lustres obtained
using the oxidising/neutral protocol shows SPR absorption
peaks characteristic of Ag
0
nanoparticles, although broader
and red shifted, corresponding to bigger sized silver metal
nanoparticles (25 nm and 40 nm, respectively). The dark
brown lustres obtained following the oxidising/neutral-then-
reducing protocol show a flatter shape. This is due to presence
of an exceptionally red shifted and extremely broad silver SPR
peak, which may be related not only to presence of large and
concentrated Ag
0
nanoparticles but also of nanocrystalline cu-
prite and Cu
2þ
dissolved in the glaze. Cu
2þ
is known to give
a characteristic bluish/green colour to the glaze and has a broad
absorption band between 600 nm and 900 nm (Gonella, 1996).
Moreover, the long tail at large wavelengths, shown by the lus-
tre produced over the lead-containing glaze, is also related to
the metal shine shown by the layer. This increase in the absorp-
tion at large wavelengths may also result from the coalescence
between metal nanoparticles (Kreibig and Vollmer, 1995).
Finally, it is important to note that the darker the colour, the
higher the total amount of silver and copper present in the
layer; when a thicker amount of paint is applied a dark brown
lustre is obtained while for thinner paints the colours are yel-
low-amber, even for oxidising/neutral-then-reducing protocols
(Pradell et al., in press). This is important because both col-
ours, brown and amber, may then be obtained simultaneously
without modifying the oxy-reducing conditions but by apply-
ing higher or lower amounts of copper and silver in the lustre
paint, or by controlling the thickness of the applied paint layer.
3.2. Islamic lustres
The materials studied are two polychrome lustres dating
from the second half of the 9th century AD (p67 and p51)
and two monochrome lustres (p32 and p37) dating from the
10th century AD. Fig. 1 shows the pictures of the four pieces
from which small samples were cut for analysis.
Average chemical analyses of the glassy matrix measured
from the polished cross-section of the glaze and also directly
from the lustre surface are shown in Table 1. The chemical
composition measured on both surfaces and either on the lustre
decorations or in the white areas is quite similar to the compo-
sition of the glassy regions in cross-section. However, the PbO
measured directly from the surfaces quite often gives a lower
concentration than in cross-section. Lead is known to be easily
1205T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
reduced and highly volatile, and when firing the glazes in a re-
ducing atmosphere at high temperatures (w1000 C) can lead
in a depletion of lead at the glaze surface during glaze firing.
However, this is much less pronounced at temperatures of
550e600 C, such as used in our reproductions and for the lus-
tre obtained in a wood feed kiln by a modern day Valencian
potter. Additionally, lead losses can also happen similar to
the loss of alkali during burial. With the exception of sample
p51, the lustres analysed show similar lead and sodium content
when comparing the cross-section and surface measurements,
especially considering that when measuring the surface we
also include the crystallites of the glaze and not just the glassy
matrix. The differences in this case may be related to either the
firing or the burial conditions.
The pastes are all dense, homogeneous and creamy and cor-
respond to a highly calcareous clay (50e55% SiO
2
,20e25%
CaO, 12% Al
2
O
3
, 6% MgO, 6% Fe
2
O
3
,1e2% Na
2
O) named
after Mason (2004) as Basra clay. The glazes show similar
composition and microstructures for all the lustres indicating
a common production technique. A backscattered electron im-
age of a glaze cross-section is shown in Fig. 3. The glazes are
thick, but the thickness varies widely from one point to an-
other and especially between the front and back surfaces of
the ceramic, being thinner on the latter, typical values range
from 100 mmto500mm. Their chemical compositions indicate
an alkaline-based glaze, with some PbO and SnO
2
added. The
amount of PbO varies considerably from sample to sample,
from 4% to 15%. The glaze appears very heterogeneous and
contains bubbles and important amounts of crystallites of dif-
ferent origin and nature. Electron microprobe and micro X-ray
diffraction data taken from different points of the glaze cross-
sections, served to identify rounded grains of unreacted quartz,
skeletons of the original cassiterite grains and recrystallized
cristobalite grains, relics from the original raw glaze com-
pounds; aluminosilicates and silicates such as diopside/pyrox-
ene floating all over the glaze with nepheline, wollastonite and
aluminium rich pyroxenes concentrated near the glaze-ceramic
interphase as a result of the glaze-ceramic paste reaction during
firing. The microstructure indicates the direct application of the
raw glaze mixture (plant ashes, quartz grains, PbO and SnO
2
)to
the fired ceramic surface at temperatures approaching 1000 C.
3.2.1. P67: Polychrome dark brown and amber lustre
This sample, which is one of the earliest examples of lustre
production, is a polychrome lustre from the 9th century AD,
profusely decorated and reproducing small flowers. The
main characteristics are the lack of metallic reflectivity and
the deliberate use of at least two colours, sometimes three:
dark brown, amber and green. Dark brown is combined with
amber and in the amber region, some spots of unintentional
green are also observed. A large dark brown area from the
front side and the corresponding amber area from the back
glaze
ID ID firing
protocol colour metal
shine
average
wt%
crystalline
comp.
Size
(nm) UV-Vis
6.4(1.9) Ag Ag015.1 (1.9)
neutral amber no
2.8(0.8) Cu cuprite?
33.4(6.4) Ag Ag021.3(2.2)
glaze-a
j17
j19
j94
j93
neutral-
then-
reducing
brown-
reddish no
1.5(0.3) Cu n.d.
6.5(6.5) Ag Ag017.7(2.5)
neutral yellow-
orange no
2.5(1.8) Cu cuprite?
12.2(8.5) Ag Ag035 (10)
glaze-m
neutral-
then-
reducing
very dark
brown
golden &
silvery
edges 1.5(0.8) Cu
300 40
05
00 600 700 800
λ(nm)
0.7
0.8
0.9
1
1.1
1.2
log(1/DR)
43
.5
32
.5 2
E(eV)
J19
J17
300 400 500 600 700 800
λ(nm)
0
0.4
0.8
1.2
log(1/DR)
43
.5
32
.5 2
E(eV)
J94
J93
Fig. 2. Summary of the data obtained for the samples reproduced in the laboratory: firing protocol, colour, copper and silver determined by WDX electron
microprobe, crystalline compounds and size of the nanoparticles determined by micro-XRD and UVevis data.
1206 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
Table 1
Chemical analysis of the Islamic glazes and luster layers
NNa
2
OK
2
OAl
2
O
3
SiO
2
CaO MgO FeO TiO
2
Cu Ag PbO SnO
2
Polychrome
9th AD
P-67(back) Glaze
cross-section
250e300 mm 5 6.42(0.29) 5.90(0.08) 1.82(0.36) 64.17(2.08) b.l.d 4.00(0.28) 0.30(0.08) 0.14(0.04) b.l.d b.l.d 4.33(0.29) 1.01(0.69)
Orange
amber lustre
Non-metallic 32 2.34(0.32) 3.60(0.21) 1.49(0.37) 66.52(0.87) 4.12(0.56) 3.91(0.78) 0.71(0.14) 0.10(0.02) 4.12(0.56) 3.91(0.78) 3.47(0.23) 1.22(0.54)
P-67(front) Glaze
cross-section
300e400 mm 10 7.16(0.28) 4.36(0.25) 2.34(1.03) 64.04(1.46) b.l.d 3.91(0.38) 0.25(0.05) 0.13(0.02) b.l.d b.l.d 6.86(0.65) 2.63(1.74)
Dark
brown lustre
Non-metallic 41 0.75(0.41) 1.61(0.32) 1.33(0.29) 61.32(2.83) 9.34(1.06) 11.55(4.66) 0.65(0.14) 0.09(0.02) 9.34(1.06) 11.55(4.66) 4.80(0.29) 2.11(1.08)
P-51 Glaze
cross-section
5 4.83(0.30) 4.83(0.25) 2.92(0.89) 60.94(2.73) b.l.d 2.40(0.20) 0.22(0.03) 0.07(0.03) b.l.d b.l.d 13.87(2.13) 2.02(0.48)
Green lustre Metallic 21 1.65(0.16) 2.75(0.12) 1.62(0.30) 72.30(1.02) 2.39(1.25) 4.11(1.63) 0.75(0.22) 0.10(0.04) 2.39(1.25) 4.11(1.63) 5.62(0.41) 1.93(0.46)
Amber lustre Non-metallic 24 1.86(0.31) 2.83(0.12) 1.67(0.36) 74.23(1.45) 4.06(0.44) 0.61(1.61) 0.84(0.25) 0.11(0.04) 4.06(0.44) 0.61(1.61) 5.61(0.37) 2.24(0.96)
Monochrome
10th AD
P-37 Glaze
cross-section
300e400 mm 5 6.47(0.13) 4.62(0.10) 2.61(0.34) 63.90(1.18) b.l.d 3.01(0.18) 0.25(0.03) 0.11(0.03) b.l.d b.l.d 7.98(0.43) 1.94(0.58)
Glaze surface 24 4.21(0.34) 4.68(0.28) 2.07(0.26) 67.78(1.57) b.l.d 2.74(0.16) 0.87(0.13) 0.10(0.03) b.l.d b.l.d 5.60(0.53) 2.50(0.64)
Green lustre Non-metallic 29 3.51(0.43) 4.42(0.20) 1.81(0.29) 67.29(1.58) b.l.d 3.40 (1.32)(0.16) 0.85(0.09) 0.12(0.06) b.l.d 3.40(1.32) 5.18(0.35) 2.59(0.83)
P-32 Glaze
cross-section
300e400 mm 5 4.98(0.11) 4.29(0.17) 2.08(0.54) 60.18(1.32) b.l.d 2.46(0.21) 0.21(0.04) 0.08(0.02) b.l.d b.l.d 14.86(0.64) 2.21(0.41)
Glaze surface 5 2.20(0.15) 3.88(0.06) 2.25(0.21) 69.54(0.61) 0.13(0.04) 0.22(0.14) 0.62(0.09) 0.09(0.02) 0.13(0.04) 0.22(0.14) 10.66(0.39) 2.98(0.39)
Green luster Metallic 35 1.44(0.21) 3.14(0.22) 2.19(0.22) 67.35(1.25) 0.36(0.13) 4.03(1.04) 0.61(0.08) 0.09(0.03) 0.36(0.13) 4.03(1.04) 10.01(0.49) 2.66(0.48)
All the data are given in wt%.
The data are averaged over Nmeasurements and the corresponding standard deviations appear in brackets. Below limit of detection (b.l.d).
1207T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
side of the ceramic were analysed. Chemical data are shown in
Table 1 and cross-sectional chemical profiles are shown in
Fig. 4.UVevis spectra, L
2,3
-edge Cu XANES and L
3
-edge
Ag EXAFS are shown Fig. 5aec, respectively.
Chemical analysis indicates that the ionic exchange of Cu
þ
and Ag
þ
with both Na
þ
and K
þ
is the mechanism responsible
for the penetration of silver and copper into the glaze for both
luster colours, as can be seen in Fig. 4. Both lustres show high
concentrations of Cu and Ag, 4.1% Cu and 3.9% Ag for the
amber lustre and 9.3% Cu and 11.6% Ag for the dark brown
lustre. Both lustres, amber and dark brown, contain nanocrys-
talline silver particles but no crystalline copper species have
been detected by SR-mXRD. The main differences between
both dark brown and the amber colours are that the former
is thicker, contains more Ag and Cu and the silver particles
are bigger (20e30 nm).
UVevis spectra of both lustre layers show the presence of
the typical surface plasmon resonance (SPR) absorption peaks
at about 452(1) nm characteristic of nanocrystalline silver
particles in a silica glassy matrix. The peak position is very
red shifted compared to the typical values obtained for nano-
crystalline silver in a silica based glassy matrix (405e430 nm)
and it is also very broad. The UVevis spectrum corresponding
to the dark brown lustre is also very flat with high absorption
in all the visible range and shows a long tail at large
wavelengths. This may be attributed either to the presence
of Cu
2þ
in the glaze structure or to the growth and coalescence
of silver nanoparticles.
The Cu L
2,3
-edge XANES shows the presence of Cu
2þ
and
Cu
þ
for the dark brown lustre, a higher Cu
þ
/Cu
2þ
ratio and
some copper metal for the amber lustre, Fig. 5b. However,
the amount of Cu
þ
in both lustres is higher than Cu
2þ
because,
for an equivalent amount, the intensity of the Cu
2þ
is 25 times
more intense than for the Cu
þ
peaks (van der Laan et al.,
1992). The oxidation of copper to Cu
þ
and, sometimes
Cu
2þ
, and its dissolution in the glaze when silver is exten-
sively present in the lustre layer has already been observed
in other cases (Pradell et al., in press; Smith et al., 2006)
and must be attributed to the higher reducing capability of
silver compared to copper. The analysis of the Ag L
3
-edge
EXAFS data indicate that for both amber and brown lustres
we have silver metal, Fig. 5c. However, the spectrum shows
a reduced amplitude for the amber lustre which seems to imply
either a smaller cluster size or a lower crystalline order than
for the brown lustre (Stern, 2001).
Moreover, some areas corresponding to the dark brown
decorations show blue iridescences. The top right flower
from the front side of sample p67 shown in Fig. 1 shows
this kind of blue iridescence. The same kind of blue irides-
cence was obtained in Pradell et al. (in press), when a mixed
silver/copper lustre was produced over a lead-free glaze at
temperatures of about 600 C. The lustre was also dark brown
and non-metallic showing a very spectacular blue iridescence
in the silver richer areas. The accumulation and/or coalescence
of silver and the presence of oxidized copper may be respon-
sible for this peculiar optical behaviour, although the exact
combination of sizes, density and thickness of the lustre layers
responsible for this are still not clear.
The warmer colours (amber and brown), the large silver
contents, high concentrations and big size of the silver nano-
particles, oxidised copper and the lack of metal shine corre-
spond well to the colours, shine and composition obtained in
the reproduction of mixed Cu/Ag lustres following oxidis-
ing/neutral-then-reducing protocols in lead-free glazes and
modifying the paint thickness. We must mention that, although
the Islamic glazes are not lead free, the lead content in this
glaze (less than about 7% PbO) was probably too low to pro-
duce a metallic lustre under these conditions. Finally, these
lustres do not need the strong reducing conditions necessary
for the production of red and coppery lustres, less intense
reducing conditions are enough to produce dark brown and
amber lustres.
3.2.2. P51: Polychrome green, brown and amber
This polychrome lustre is profusely decorated with typical
patterns of alternated amber and green lines and circles. It
shows three colours, that is, olive green, metallic golden lustre
alternated with amber brown and some circles of a darker
brown. These darker brown circles are made of a thicker
amber lustre. The olive green designs are very homogeneous
although they do show some amber-brown spots as seen in
Fig. 6a. Similarly, the amber-brown designs are also very
Fig. 3. SEM backscattering image from a polished cross-section from the glaze
of p51.
1208 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
homogeneous, but quite often show olive green contours as
shown in Fig. 6b. Chemical analysis of the brown spots in
the green areas, and of the green contours of the amber-brown
areas, indicate that in all the cases the amber brown contains
higher Cu (4% Cu and 0.6% Ag) than the green area which
contains higher Ag (2.4% Cu and 4.1% Ag). The back side
shows also some grey silvery brush marks. Two pictures taken
from the amber and green lustre decorations, along with corre-
sponding pictures taken by tilting the surface to detect the
specular reflected light are shown in Fig. 6c and d, respec-
tively. Metal shine is observed in the green lustre while the
amber lustre lacks it.
(a) (b)
0 400 800 1200 1600
d (microns)
0
1
2
at%
0
1
2
at%
0
10
20
30
at% Si
Fe
Mg
Ca
Na
K
Al
Si
Cu
Sn
Pb
Ag
0 400 800 1200 1600
d (microns)
0
1
2
at%
0
1
2
3
4
at%
0
10
20
30
at% Si
Fe
Mg
Ca
Na
K
Al
Si
Cu
Sn
Pb
Ag
012345
at%(Na+K)
0
1
2
3
4
5
at%(Cu+Ag)
02468
at%(Na+K)
0
2
4
6
8
at%(Ag+Cu)
Fig. 4. Chemical surface profiles and %(Ag þCu) versus %(Na þK) correlations for (a) amber lustre area from the back side and (b) dark brown lustre area from
the front side of sherd p67. For the interpretation of the colour in this figure, the reader is referred to the web version of this article.
1209T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
In these lustres a good chemical anticorrelation of (Cu þAg)
versus Na is obtained indicating that in this case Cu and Ag
enter the glaze by ionic exchange with Na. Ionic exchange is
preferentially produced with Na although, if there is not enough
Na available, K is also involved in the process (Molera et al.,
2007).
UVevis spectra, shown in Fig. 6e, were taken from two dif-
ferent green areas, one showing a metal golden shine and a sec-
ond region from the back surface of the sample which does not
show metal shine. We can see that the green area shows a broad
peak at about 434(5) nm and a width of about 0.98 0.4 eV.
The green golden area shows a double peak (350 nm and
450 nm, respectively). A transition from a single SPR peak
to the double peak was found in the silver lustre reproductions
showing a metal golden shine (Pradell et al., 2006; Molera
et al., 2007), and was attributed to the formation of close
packed chains of silver nanoparticles (Quinten and Kreibig,
1993; Kreibig and Vollmer, 1995). In fact the UVevis spec-
trum corresponding to the green non-metallic area already
shows the extra flattening of the top observed just before the
SPR peak splits. The formation of a golden green lustre layer
was found to be temperature dependent, that is, at tempera-
tures below 540 C the layer appeared yellow-green but they
did not show metal shine whilst at higher temperatures it al-
ways showed golden shine. Simultaneously, the corresponding
UVevis spectrum changed from a single broad top flat peak to
a double peak (Molera et al., 2007).
Cu K-edge micro-EXAFS data taken from the brown spots
in the green lustre and in the brown designs are shown in
Fig. 6f; the data fits well to the first CueO coordination shell
of cuprite, from which we deduce that all the copper is Cu
þ
.
There is no evidence of any further atomic structure which
would be expected from a crystalline material; in particular
the first CueCu shell of crystalline cuprite, at a distance of
3.02 A
˚, and which would be expected to be very prominent
is not present. This result clearly indicates that crystalline
cuprite is not formed and that Cu
þ
is dissolved in the glaze
structure.
The green golden is similar to the silver lustres produced
over lead-containing glazes (Molera et al., 2007), and the
mixed silver and copper amber lustres are also similar to the
colours obtained in the reproductions when a single neutral/
oxidizing protocol is followed.
3.2.3. P32 and P37: Monochrome olive green
Both samples correspond to the same type of monochrome
green lustre from the 10th century AD with a central single an-
imal or figure on a background of contour panels and a texture
recalling woven textiles (Caiger Smith, 1991). However, one is
a flat out-rimmed plate with a successful metallic golden lustre
940 960 980
Photon energy (eV)
0
0.002
0.004
0.006
0.008
0.01
I(a.u)
Cu2+
Cu+
Cu0
amber
dark brown
(b)
03 710
R (Å)
0
4
8
12
I(a.u)
Amber
dark brown Ag-Ag
1st shell
2nd shell 3rd-4th shell
9865421
300 400 500 600 700
0.6
0.8
1
1.2
1.4
log(1/DR)
4 3.5 3 2
E(eV)
amber
dark brown
(a) 2.5
246
k (Å-1)
-1
0
1
2
3
I(a.u)
Amber
dark brown
(c)
753
λ(nm)
Fig. 5. (a) UVevis spectra, (b) Cu L
3
-edge and L
2
-edge XANES and (c) Ag L
3
EXAFS spectra (top) and the corresponding Fourier transform (bottom) from the
lustre colours amber and dark brown from p67.
1210 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
(p32), see Fig. 7, while the other is a non-metallic small out-
rimmed bowl showing a single central gazelle (p37). In both
cases, the lustre has a green homogeneous colour. However
under the microscope, silver always appears heterogeneously
distributed forming spots of some tens of microns size, with
a higher concentration of silver in the centre as shown in
Fig. 7. Sometimes, these silver rich regions appear darker
(brownish) as shown in the magnified image of sample p37
in Fig. 7, and in some cases, when copper is present, they
may show a blue iridescence. The average content of silver
(a) (b)
(e) (f)
300 400 500 600 700
λ(nm)
0.8
1
1.2
1.4
log(1/DR)
43
.5
32
.5 2
E(eV)
green
green golden
0123456789
10
R(Å)
0
2
4
6
8
10
I(a.u)
2468
10 12
k (Å-1)
-1
0
1
I (a.u.)
brown spot
Cu-O (1st shell)
(c) (d)
Fig. 6. Colour variations and metallic shine in the green and amber lustres corresponding to p51: (a) green lustre showing some brown spots; (b) brown lustre
showing green edges; (c) and (d) palm tree like decoration made of green and brown lustres, the sample surface has been tilted with respect to the illumination
to show the gold like metallic shine of the green lustre and the lack of metallic shine of the brown lustre. (e) UVevis spectra corresponding to a green golden area
from the front side and a green without metallic shine area from the back side of p51. (f) Micro Cu K-edge EXAFS spectrum (top) and the corresponding Fourier
transform (bottom) from a brown spot in the green lustre layer from p51. (For interpretation of colour in this figure, the reader is referred to the web version of this
article.)
1211T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
in the lustres is about 3.4% Ag for p37 and 4.0% Ag for p32,
and in the brownish spots silver reaches 7% Ag. Both show
very low amounts of copper, 0.1% Cu for p37 and 0.4% Cu
for p32, respectively. Ion exchange between Ag and Na is
the main mechanism, although K is also exchanged albeit in
lower amounts.
The composition of the glazes is very similar although
sample p32 has higher amounts of PbO (14.9% PbO) than
p37 (8.0% PbO). The higher lead content of sherd p32 clearly
favours the development of the golden shine.
UVevis spectra corresponding to both samples are shown
in Fig. 8a. The spectrum corresponding to p32 shows the char-
acteristic splitting of the peak related to the gold metal shine,
while p37 does not, although the peak is quite broad and flat.
Fig. 8b shows the Ag L-edge spectra for both samples
which indicate that it is silver metal. Both spectra show
reduced amplitude compared to pure metal silver. They are
comparable to that seen in the amber colour of p67, and can
also be interpreted as resulting from either smaller cluster
sizes or lower crystalline order.
4. Discussion
The analysis of the pastes suggests that both polychrome
and monochrome lustre ceramics correspond to a common
geographic area which had a continued production for nearly
two centuries. The pastes are highly calcareous, dense, homo-
geneous and creamy. The glazes are of the mixed alkali-lead
tin opacified type that started to be used on this time of vari-
able thickness, very heterogeneous and containing lots of crys-
talline phases. They were most probably obtained by firing the
raw glaze mixture (quartz, cassiterite with plant ashes and
PbO) directly applied over the ceramic surface.
For these early Islamic lustres, ionic exchange is also the
mechanism of penetration of the metal ions into the glaze sur-
face. The lustre paint was applied over the glaze in a subsequent
firing and the range of temperatures could not be much higher
than 600 C as there is no evidence of glaze softening.
The orange amber and brown colours of the lustres is due to
the presence of Cu
þ
and Cu
2þ
dissolved in the glaze, together
with the precipitation of large amounts of silver nanoparticles.
Both colours could be obtained either using the same paint
composition and varying the thickness applied (Pradell et al.,
in press), or by changing the silver and copper content in the
paint and applying the same thickness. For brown either
a thicker paint layer, or a paint with a higher metal content is
required than for the amber. The colours are due to the contri-
butions in the UVevis absorption spectrum of the SPR peak of
big silver nanoparticles and of cuprite and Cu
þ
and Cu
2þ
ions
dissolved in the glaze. In order to obtain these colours both
Fig. 7. Bottom left : chemical map corresponding to Ag from the square area marked on the top left image of sample p37. Silver appears heterogeneously
distributed forming spots of 10e50 mm across. Right : images of sample p32, in the bottom right image the surface is tilted to catch the specular reflected light
and the metallic shine. For the interpretation of the colour in this figure, the reader is referred to the web version of this article.
1212 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
copper and silver have to be present in the lustre paint. Amber
and brown colours are then obtained following a neutral/
oxidising-then-reducing protocol. The presence of both copper
and silver ions in the glaze favours the precipitation of silver
nanoparticles while copper stays dissolved in the glaze as
Cu
þ
and Cu
2þ
.
Yellow/green and green colours are due to the presence of
small silver nanoparticles. In this case the SPR peak is sym-
metric with the maximum absorption below 440 nm. If the sil-
ver nanoparticles are closely packed, the UVevis spectrum
shows the typical double peaked and flatter spectrum charac-
teristic of the transition from isolated single particles to a col-
lective behaviour (from dielectric to metal like), and therefore
the lustre layers then show a high absorption and metal-like
reflectivity. For this to happen, the metal nanoparticles have
to be densely packed. Our reproduction studies have shown
that this happens more easily in the case of silver lustres under
less reducing conditions. In contrast, very strong reducing
conditions are needed to obtain both red ruby and red coppery
lustres. Moreover, our replications and previous studies have
demonstrated that the presence of PbO in the glaze favours
the reduction and formation of thinner, densely packed lustre
layers, and thus the metal-like shine of the lustre layers. We
have observed that in the early polychrome lustres from
Iraq, when the glaze contains higher PbO (above 10% PbO)
the silver rich green lustres show this metal shine. It is possible
to surmise that the potters realised that the use of recipes of
lustre paints that were richer in silver resulted in the gold
metal shine. Thus they moved to introduce silver rich recipes
and consequently to produce monochrome lustres. However,
the success was also linked to the introduction of higher
amounts of PbO in the glaze, however low control of the
lead content in the glaze composition sometimes still resulted
in the production of non-metallic shining green lustres. Lead
was being added to glaze mixtures generally at this time as
the thermal expansion coefficients of lead-containing glazes
are better matched to ceramics, leading to the eradication of
glaze cracking. It is probable that lustre ceramicists of the
time also noted the beneficial effect this addition has in pro-
ducing metal-like lustre finishes.
5. Conclusions
Ninth century AD Iraqi lustres were obtained by firing
a mixture of Cu and Ag compounds onto an alkali-lead tin
opacified glaze. The amber and dark brown colours come
out when large amounts of silver nanoparticles and Cu
þ
and
Cu
2þ
dissolved in the glaze are present in the lustre layer.
(a) (b)
300 400 500 600 700
1.1
1.2
1.3
log(1/DR)
4 3.5 3 2.5 2
E(eV)
p37
p32
012345678910
R (Å)
0
2
4
6
8
I(a.u)
p32
p37
234567
k (Å-1)
-1
0
1
2
I(a.u)
p32
p37
Ag-Ag
1st shell
2nd shell 3rd-4th shell
λ(nm)
Fig. 8. (a) UVevis spectra corresponding to a green golden area from p32 and to a green without metallic shine area from p37. (b) Ag L
3
-edge EXAFS spectra
(top) and the corresponding Fourier transform (bottom) from p32 and p37, respectively and showing similar features.
1213T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
The green colours characteristic of some 9th century AD lus-
tres and of the 10th century AD monochrome Iraqi production
are obtained for Cu-free Ag rich lustres (or at least very low
Cu-containing lustres). To obtain these Cu/Ag mixed or Ag
rich lustres, less reducing conditions are needed than for pure
red copper lustres and this resulted in a wide variety of colours.
Previous studies on single Cu or Ag lustres showed that
metal reflectivity is obtained when the lustre is applied on
a lead glaze and when a combined oxidising/neutral-then-
reducing firing protocol is followed. This also holds for the
mixed Cu/Ag lustres. The analysis of the early Islamic lustres
from Iraq shows that metal reflectivity is obtained for silver
rich green lustres but not for the mixed Cu/Ag amber and
brown lustres on relatively low PbO containing glazes (10e
15%). If the glaze contains lower amounts of lead, the metal
reflectivity, even for the silver rich green lustres, does not
appear.
The Cu/Ag mixture gave a large variety of colours. The ear-
liest polychrome 9th century AD productions correspond to
this type, although it is a period when the potters probably es-
sayed all kinds of lustre recipes. The introduction of lead in
the glaze formulation favoured the formation of gold shining
green silver rich lustres, and the change to a green mono-
chrome production in the 10th century AD.
Acknowledgements
T. Pradell wants to thank Professor J.W. Allan and the Ash-
molean Museum for providing the Islamic luster samples. T.
Pradell is funded by CyCIT grant MAT2004-01214 and
Generalitat de Catalunya grant 2005SGR00201. We also ac-
knowledge funding by the European Community through the
Research Infrastructure Action under FP6 ‘‘Structuring the
European Research Area’’ program project numbers 43025
and 45229 at SRS, Daresbury Laboratory. Dr. J. Molera is
funded by the program Ramon y Cajal and UdG research pro-
ject number 9104071.
References
Allan, J.W., 1973. Abu
¨’l-Qa
¨sim’s treatise on ceramics. Iran IX, 111e120.
Berthier, S., Padeletti, G., Fermo, F., Bouquillon, A., Aucouturier, M.,
Charron, E., Reillon, V., 2006. Lusters of renaissance pottery: experimental
and theoretical optical properties using inhomogeneous theories. Appl.
Phys. A 83, 573e579.
Borgia, I., Brunetti, B., Mariani, I., Sgamellotti, A., Cariati, F., Fermo, P.,
Mellini, M., Viti, C., Padeletti, G., 2002. Heterogeneous distribution of
metal nanocrystals in glazes of historical pottery. Appl. Surf. Sci. 185,
206e216.
Brill, H., 1970. Chemical studies of Islamic luster glass. In: Berger, R. (Ed.),
1970. Scientific Methods in Medieval Archaeology. UCLA Center for Me-
dieval and Renaissance Studies, Contributions, 4. University of California
Press, Berkeley.
Caiger Smith, A., 1991. Lustre Pottery. New Amsterdam Books, New York.
Carboni, S., 2001. Glass from Islamic Lands, Thames & Hudson in association
with the al-Sabah Collection, Dar al-Athar al-Islamiyyah. Kuwait National
Museum, London, UK.
Colomban, P., Truong, C., 2004. Non-destructive Raman study of the glazing
technique in lustre potteriesand faience(9e14th centuries): silver ions, nano-
clusters, microstructure and processing. J. Raman Spectrosc. 35, 195e207.
Colomban, P., Schreiber, J.D., 2005. Raman signature modification induced by
copper nanoparticles in silicate glass. J. Raman Spectrosc. 36, 884e890.
Collier, C.P., Saykally, R.J., Shiang, J.J., Henrichs, S.E., Heath, J.R., 1997.
Reversible tuning of silver quantum dot monolayer through the metal-
insulator transition. Science 277, 1978e1981.
Darque-Ceretti, E., He
´lary, D., Bouquillon, A., Aucouturier, M., 2005. Gold
like lustre: nanometric surface treatment for decoration of glazed ceramics
in ancient Islam, Moresque Spain and Renaissance Italy. Surf. Eng. 21
(5e6), 352e358.
Dusemund, B., Hoffmann, A., Salzmann, T., Kreibig, U., Smidth, G., 1991.
Cluster matter: the transition of optical elastic scattering to regular reflec-
tion. Z. Phys. D 20, 305e308.
Farbman, I., Levi, O., Efrima, S., 1992. Optical response of concentrated
colloids. J. Chem. Phys. 96 (9), 6477e6485.
Friederick, P., He
´lary, D., Schryvers, D., Darque-Ceretti, E., 2004. A TEM
study of nanoparticles in lustre glazes. Appl. Phys. A 79 (2), 283e288.
Gonella, F., 1996. Characterisation of CueNa ion-exchanged glass wave-
guides. Appl. Phys. Lett. 69 (3), 314e315.
He
´lary, D., Darque-Ceretti, E., Aucouturier, M., 2005. Contemporary golden-
like lustres on ceramics: morphological, chemical, and structural properties.
J. Am. Ceram. Soc. 88 (11), 3218e3221.
Kreibig, U., Vollmer, M., 1995. Optical Properties of Metal Cluster, 25.
Springer Verlag. Springer.
Mason, R.B., 2004. Shine like the sun. Lustre-painted and associated pottery
from the Medieval Middle East, Bibliotheca Iranica: Islamic Art and
Architecture series, 12. Mazda Publishers, Inc., Costa Mesa Canada.
Molera, J., Mesquida, M., Pe
´rez-Arantegui, J., Pradell, T., Vendrell, M., 2001.
Lustre receipts from a medieval workshop from Paterna. Archaeometry 43
(4), 455e460.
Molera, J., Baye
´s, C., Roura, P., Crespo, D., Pradell, T., 2007. Key parameters
in the production of medieval lustre colours and shines. J. Am. Ceram. Soc.
90 (7), 2245e2254.
Padovani, S., Sada, C., Mazzoldi, P., Brunetti, B., Borgia, I., Sgamellotti, A.,
Giullvi, A., D’Acapito, F., Battaglin, G., 2003. Copper in glazes of Renais-
sance luster pottery: nanoparticles, ions, and local environment. J. Appl.
Phys. 93, 10058e10063.
Padovani, S., Borgia, I., Brunetti, B., Sgamellotti, A., Giullvi, A., D’Acapito, F.,
Mazzoldi, P., Sada, C., Battaglin, G., 2004. Silver and copper nanoclusters in
the Luster decoration of Italian Renaissance pottery: an EXAFS study. Appl.
Phys. A 79 (2), 229e233.
Padeletti, G., Fermo, S., 2003. How the masters in Umbria, Italy, generated
and used nanoparticles in art fabrication during the Renaissance period.
Appl. Phys. A 76 (2), 515e525.
Padeletti, G., Fermo, S., 2004. Production of gold and ruby-red lusters in
Gubbio (Umbria, Italy) during the Renaissance period. Appl. Phys. A 79
(2), 241e245.
Padeletti, G., Ingo, G.M., Bouquillon, A., Pages-Camagna, S.,
Aucouturier, M., Roehrs, S., Fermo, P., 2006. First-time observation of
Mastro Giorgio masterpieces by means of non-destructive techniques.
Appl. Phys. A 83, 475e483.
Pe
´rez-Arantegui, J., Molera, J., Larrea, A., Pradell, T., Vendrell, M., Borgia, I.,
Brunetti, B.G., Cariati, F., Fermo, P., Mellini, M., Sgamellotti, A., Viti, C.,
2001. Luster pottery from the 13th to the 16th century: a nanostructured
thin metallic film. J. Am. Ceram. Soc. 84 (2), 442e446.
Pe
´rez-Arantegui, J., Larrea, A., Molera, J., Pradell, T., Vendrell, M., 2004.
Some aspects of the characterisation of decorations on ceramic glazes.
Appl. Phys. A 79 (2), 235e239.
Pradell, T., Molera, J., Vendrell, M., Pe
´rez-Arantegui, J., Pantos, E.,
Roberts, M., DiMichiel, M., 2004. The role of cinnabar in lustre produc-
tion. J. Am. Ceram. Soc. 86 (6), 1018e1023.
Pradell, T., Molera, J., Roque, J., Smith, A.D., Crespo, D., Pantos, E.,
Vendrell, M., 2005. Ionic-exchange mechanism in the formation of
medieval lustre decorations. J. Am. Ceram. Soc. 88 (5), 1281e1289.
Pradell, T., Molera, J., Baye
´s, C., Roura, P.,2006. Luster decoration of ceramics:
mechanisms of metallic luster formation. Appl. Phys. A 83, 203e208.
Pradell, T., Molera, J., Pantos, E., Smith, A.D., Martin, C.M., Labrador, A. Tem-
perature resolved reproductionof medieval luster. Appl. Phys. A., in press.
doi:10.1007/s00339-007-4226-z.
1214 T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
Pradell, T., Climent-Font, A., Molera, J., Zucchiatti, A., Ynsa, M.D.,
Roura, P., Crespo, D., 2007. Metallic and non-metallic shine in luster:
an elastic ion backscattering study. J. Appl. Phys. 101, 103518-1e
103518-8.
Quinten, M., Kreibig, U., 1993. Absorption and elastic scattering of light by
particle aggregates. Appl. Optics 32 (30), 6173e6183.
Reillon, V., Berthier, S., 2006. Modelization of the optical and colorimetric
properties of lustred ceramics. Appl. Phys. A 83, 257e265.
Smith, A.D., Pradell, T., Molera, J., Vendrell, M., Marcus, M.A.,
Pantos, E., 2003. MicroEXAFS study into the oxidation states of copper
coloured Hispano-Moresque lustre decorations. J. de Physique IV 104,
519e522.
Smith, A.D., Pradell, T., Roque, J., Molera, J., Vendrell-Saz, M., Dent, A.J.,
Pantos, E., 2006. Colour variations in 13th century Hispanic Lustre- an
EXAFS study. J. Non-Cryst. Solids 352, 5353e5361.
Stern, E.A., 2001. Musing about the development of XAS. J. Synch. Rad. 8,
49e54.
van der Laan, G., Pattrick, R.A.D., Henderson, C.M.B., Vaughan, D.J., 1992.
Oxidation state variations in copper minerals studied with Cu 2p X-ray
Absorption Spectroscopy. J. Phys. Chem. Solids 53, 1185e1190.
1215T. Pradell et al. / Journal of Archaeological Science 35 (2008) 1201e1215
