Upload
olivier-schalm
View
219
Download
0
Embed Size (px)
Citation preview
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
1/20
M A N G AN E S E S TA I N I N G O F A R C H AE O L O G IC A L G L A S S :
T H E C H A R A C T E R I Z A T I O N O F M n - R I C H I N C L U S I O N S
I N L E A C H E D L A Y E R S A N D A H Y P O T H E S I S O F
I T S F O R M A T I O N *
O. SCHALM,1 K. PROOST,1 K. DE VIS,2 S. CAGNO,1 K. JANSSENS,1 F. MEES,3
P. JACOBS4 and J. CAEN2
1Department of Chemistry, University of Antwerp, Universiteitsplein 1, B-2610 Antwerp, Belgium2Conservation Studies, Artesis University College of Antwerp, Blindestraat 9, B-2000 Antwerp, Belgium
3Department of Geology and Mineralogy, Royal Museum for Central Africa, Leuvensesteenweg 13, B-3080 Tervuren, Belgium4Department of Geology and Soil Science, Ghent University, Krijgslaan 281/S8, B-9000 Ghent, Belgium
During the study of a large number of archaeological glass fragments, manganese-richinclusions in leached layers were observed in a limited number of cases. This phenomenon
occurs only in black-coloured leached layers. Since the formation mechanism of such
manganese-rich inclusions is still unclear, a combination of several analytical techniques was
used in order to investigate this phenomenon and, more specifically, to obtain more informa-
tion on (a) the composition and morphology of the inclusions, (b) the chemical state of Mn and
(c) the 3D morphology of the inclusions. A mechanism that might explain the formation of
these inclusions is proposed.
KEYWORDS:ARCHAEOLOGICAL GLASS, MANGANESE STAINING, HETEROGENEITIES,
LEACHED LAYERS
INTRODUCTION
Historic glasses contain usually a small amount of manganese (Sayre 1963). Most of this
manganese is present as the colourless Mn(II), while a small fraction may exist as the strongly
coloured Mn(III). Higher oxidation states are normally not present in glass because they are
unstable at the temperatures at which glass is produced. This element was introduced into the
glass batch from two different sources: (1) as an impurity in one of the raw materials, such as
wood ash (Misra et al. 1993; Stern and Gerber 2004); and (2) as a deliberate addition of
pyrolusite to the liquid batch, in order to decolorize the glass (Smith and Gnudi 1990). In the
liquid glass, the higher oxidation states of manganese transforms the colouring substance Fe(II)into the less colouring Fe(III).
The presence of Mn can also cause darkening of glass. It is known that the oxidation of Mn(II)
in colourless glass, initiated by sunlight, results in the formation of purple glass in the course of
time (Long et al. 1998; Kunicki-Goldfinger 2008). This phenomenon is known as solarization.
However, there is another mechanism in which glass can be discoloured by manganese. Several
authors have noticed that a darkening results from the intrusion of manganese from the environ-
ment into decaying glass (Newton and Davison 1989; Cox and Ford 1993; Domnech-Carb
et al. 2001; Silvestri et al. 2005). This effect is not only observed in medieval glass, but also in
Roman glass (Janssenset al. 2000). The brown areas in the corrosion layer contain manganese in
*Received 14 July 2009; accepted 29 January 2010
Archaeometry53 , 1 (2011) 103122 doi: 10.1111/j.1475-4754.2010.00534.x
University of Oxford, 2010
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
2/20
the 4+oxidation state (Cooperet al. 1993; Domnech Carbet al. 2006). According to Silvestri
et al. (2005), the occurrence of zones rich in Mn, Ca, P and Fe in corrosion layers are charac-
teristic for underground alteration. It does not occur for marine alteration.
In nature, manganese oxides and manganates are formed by either a chemical or a biological
oxidation of Mn(II), in a variety of aquatic environments (Mandernacket al. 1995). Since theformation mechanism of manganese-rich inclusions is still unclear, a number of historical glass
samples containing this type of inclusion were characterized by means of several analytical
techniques.
EXPERIMENTAL
The samples in this study were selected from a larger set of samples of 14th17th century window
glass that was analysed during a quantification campaign (Schalmet al. 2007). Most of the glass
fragments studied in that work were excavated in Raversijde, Antwerp and Namur (Belgium).
Several fragments were embedded together in one block of resin. The fragments in the resinwere orientated in such a way that the cross-sections, perpendicular to the original glass surface,
could be readily studied. The surface of the resin was ground flat with corundum paper and
polished with fine diamond pastes (down to a final diameter of 0.25 mm). In many cases, the
sample preparation of extremely deteriorated glass was impossible. The slightest pressure on
these glasses resulted in breakage or even pulverization. Only fragments that were heavily
deteriorated but that were mechanically strong enough could be embedded and polished. An
Olympus SZX12 binocular optical microscope was used to study colour fluctuations in the
leached layers at a microscopic scale.
A scanning electron microscope energy-dispersive X-ray system (SEMEDX) was used for
the compositional study of small zones (down to 100 mm3
) in samples and for visualization ofelemental distributions on a microscopic scale. The instrument was equipped with a thin-window
Princeton Gamma Tech energy-dispersive Si(Li) detector. The analyses were performed on the
samples after the resin blocks were coated with a thin layer of carbon. The system allows the
recording of X-ray emission lines for elements down to carbon; detection limits of approximately
0.1wt% for elements with an atomic number between 11 and 20 are obtained. For the quantitative
analyses, X-ray spectra were collected with a 2 nA electron beam at an energy of 20 keV. The
spectra were collected at a maximum magnification of 3000 over a time period of 400 s. Under
such circumstances, sodium migration was not observed during the measurements. The chemical
composition was calculated by using a thin-film elemental yield approach (Schalm and Janssens
2003).Synchrotron radiation was employed to obtain information on the oxidation states of manga-
nese in the samples. These experiments were performed at the micro-focus beam line L of
the Hamburger Synchrotronstrahlungslabor (HASYLAB) in Hamburg (Germany), using a
polycapillary-based set-up (Falkenberget al. 2001).A polycapillary half-lens produced a focused
beam on the sample surface with a diameter of approximately 20 mm (Proost et al. 2003). The
sample was mounted on a sample stage that could be remotely translated in the x,y,zdirections.
Fluorescent photons, produced by the sample, were detected by an energy-dispersive Si(Li)
detector in a 45 geometry. When performing microscopic X-ray absorption near-edge structure
(m-XANES) experiments on the manganese edge, the energy of the primary beam is varied in
small steps (1 eV) around the MnK edge at 6.435 keV. During such an energy scan, the numberof manganese fluorescence photons was recorded by the energy-dispersive detector during
typically 520 s per energy value, in order to collect a MnK XANES spectrum. The XANES
104 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
3/20
spectra of original glass, the leached layers and the dark Mn-rich inclusions glass were qualita-
tively compared to those of the following reference compounds for, respectively, Mn(II), Mn(III)
and Mn(IV)that is, MnSO4.H2O, Mn2O3and MnO2in order to obtain information on the
oxidation state of manganese in the corroded glasses. The origin of the energy scale of the
XANES spectra was arbitrarily set to the first inflection point of the MnSO4.H2O referencespectrum. The reference compounds were diluted with boron nitride (about 1% of manganese in
BN) and pressed into pellets in order to minimize self-absorption effects.
Three-dimensional information on the Mn-rich inclusions in some of the corrosion bodies was
obtained by absorption tomography, using a SKYSCAN-1072 high-resolution desktop micro-CT
instrument (www.skyscan.be). In this instrument, an X-ray tube with a tungsten anode produces
a beam of polychromatic radiation. The sample of interest, a heavily corroded glass fragment,
was rotated in the beam while a CCD camera behind the sample recorded the transmitted beam,
visualizing the attenuation by the sample under different rotation angles. A mathematical recon-
struction algorithm allows the calculation of the absorption in arbitrary slices within the object of
interest. This was done using Cone-Beam Reconstruction software delivered with the instrument.The X-ray tube was operated at 78 mA and 137 kV.
THEORETICAL BACKGROUND
Just beneath the surface of a glass object, the chemical composition and the structure of the glass
network are affected by reactions with substances from the environment surrounding the glass.
Moisture is the most important factor initiating and sustaining the different reactions leading to
glass decay. Usually, chemical deterioration is described as a combination of three simultaneous
partial processes (Silvestriet al. 2005; Pollard and Heron 2008). The processes, described below,
explain how a leached layer is formed and why the local concentration of mobile cations such asNa+, K+, Mg2+, Ca2+, Mn(2+,3+) and Fe(2+,3+) decreases.
The first process is the penetration of molecular water into the glass. This is due to molecu-
lar diffusion and/or through reversible hydrolysis and condensation reactions. When water
tagged with O18 diffuses into the glass, O18 exchanges with O16 atoms of the siliconoxygen
glass network. During this process, the silicate network is subject to structural transformations
(Bunker 1994; Doremus 1998). For buried medieval glass, the diffusion of ppm-amounts of
molecular water into the glass surface precedes the ion exchange (Sterpenich and Libourel
2006):
+ ( ) Si O Si H O aq 2 Si OH2
2 Si OH Si O Si H O2 glass + ( )
The process of ion exchange is based on the leaching of mobile cations from the glass and their
substitution with protons following the reaction given below (Scholze 1982; Schreiner 1991;
Melcher and Schreiner 2004):
+ + +( )+
( ) +
( )+
( )Si O M H Si O H Mglass aq glass aq
In this process, the silicate network remains intact. It is driven by the concentration gradient
between the glass and the environment. The mobility of the cations depends on their charge,
their size and on the composition of the initial, uncorroded glass. As a result, some glass
compositions are more resistant to leaching than others. For example, it is well known that(silica-rich) Roman glass is usually in a better condition than (silica-poor) medieval glass, in
spite of its more advanced age. Since in this process, (heavier) cations are replaced by (lighter)
Manganese staining of archaeological glass 105
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
4/20
protons, while the volume of the leached glass remains very similar, the density of the leached
layer becomes smaller than that of the original glass. The leached-out ions are able to form
weathering products such as sulphates, carbonates, hydro-carbonates, chlorides or nitrates on
top of the glass surface. These products are formed when the solubility products of these
substances present in the water film on the surface are exceeded (Schreiner et al. 1999;Woisetschlger et al. 2000; Domnech-Carb et al. 2006; Melcher and Schreiner 2006).
If small amounts of water in close contact to the glass are not regularly replenished, its pH will
increase as a consequence of the ongoing ion-exchange. This is the case when water is entrapped
in cracks or in surface scratches. When pH values exceed the value of 9, hydroxyl ions in solution
will attack the silicate network:
+ + ( )Si O Si OH Si OH Si Oaq
+ + ( )
( )Si O H O Si OH OH2 aq aq
A consequence of these three reactions is that a leached layer is formed, characterized by a
cation-depleted and structurally weakened SiO network. This so-called leached layer gradually
increases in thickness until the complete glass is transformed into a deteriorated product. The
leached layer can be transformed into a silica-rich film in which the number of SiOSi bonds has
been increased (Hench 1975).
Although the above-described reactions clearly fail to predict the transformation of homoge-
neous glass into heterogeneous leached layers, many types of heterogeneous leached layers have
been reported in the past (Cox and Ford 1993; Sterpenich and Libourel 2001). Alteration zones
show a typically finely laminated structure in which hydrated phosphate compounds might be
found. Freestone (1985) reported that Ca phosphate can be precipitated inside the leached
layer of archaeological glass, probably in the form of apatite (Ca5(PO4)3OH) or brushite
(CaHPO4.2H2O). Bioglass is another example that shows how glass containing P2O5is able tointeract with Ca-rich compounds from the environment. In that case, the P2O5in the glass is able
to form a strong and stable bond with newly growing bone. This is due to the formation of a stable
calcium phosphate film on the surface of the material (Hench 1975). A special form of hetero-
geneities in leached layers is the presence of manganese-rich inclusions. In the list of corrosion
products formed during precipitation reactions, manganese-rich compounds are usually not
mentioned.
RESULTS
Morphology of some heavily deteriorated glasses
Among the ~500 archaeological window glass fragments examined, several heavily or even
completely deteriorated samples containing dark inclusions were identified. This type of inclu-
sion was only encountered in blackened fragments. Moreover, the dark colour was always
restricted to the leached layers themselves, and was never present in the original, unaffected parts
of the glass. Optical and backscattered electron images of cross-sections of the samples that were
examined in detail are shown in Figures 1 and 2. A number of the heavily deteriorated glass
fragments still contained a core of original glass, with a chemical composition very similar to that
of other, less corroded, glass fragments. For buried archaeological glass, two types of manganese-rich inclusions could be distinguished: (1) dark inclusions in the low-density lamellae of stratified
leached layers, and in cracks such as encountered in a 15th century artefact excavated in
106 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
5/20
Raversijde and shown in Figure 1; and (2) dendritic invasion of the leached layer by brownish-
black material, such as found in the glass fragments recovered in Namur and Antwerp. Figure 2
(a) shows an optical image of a polished cross-section of a fragment of 16th17th century
window glass excavated in Namur, while Figure 2 (b) shows a cross-section of a 14th century
window glass sample excavated in Antwerp. The optical microscopy image of Figure 1 demon-
strates that the leached layer just below the glass surface has a brownish to black colour. While,
optically, no particular structure can be discerned in this dark-coloured area, the corresponding
backscattered electron microscopy image (Fig. 3 (b)) does reveal a layered microstructure in this
region. In Figure 3 (b), the light grey region at the bottom of the image is the original glass, the
dark grey region is the leached layer and the bright inclusions within the leached layer are thebrown-black coloured inclusions. The black area at the top of the image is the surrounding resin,
in which the samples are embedded. The stratified structure of the leached layer consists of
alternating layers of lower and of higher densities. This image suggests that the dark inclusions
were preferentially formed in the darker (i.e., low-density) lamellae. The optical photographs
shown in Figure 2, on the other hand, clearly show that here the blackbrown structures were not
formed in low-density lamellae of stratified leached layers but in more homogeneous leached
layers, forming dendritic structures.
The phenomenon of manganese staining was also found in a 15th century free-standing stained
glass window. In an originally colourless pane, manganese-rich inclusions were found in the
dark-coloured lamellae of the stratified leached layer. This is shown in Figure 4. Next to thesemanganese-rich inclusions, the leached layer contains regions rich in calcium and phosphorus as
well.
Figure 1 An optical image of a deteriorated region in a cross-sectioned 15th century sample excavated in Raversijde.
In this sample, Mn-rich inclusions can be found between the lamellae of the stratified leached layer. This can be seen in
Figure 3.
Manganese staining of archaeological glass 107
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
6/20
Compositional analysis of the glass samples
It is striking to note that almost all of the ~500 15th17th century window glass fragments that
were analysed have a composition that is very similar to the average composition, given in
(a)
(b)
Figure 2 An optical image of a dendritic invasion of manganese in cross-sectioned samples excavated in (a) Namur
(16th17th century) and (b) Antwerp (14th century).
108 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
7/20
Figure 3 X-ray images (a) and a backscattered electron image (b) of the cross-sectioned 15th century sample of
Raversijde. In this sample, manganese migrated along the cracks and between the low-density regions within the lamellarstructure.
Manganese staining of archaeological glass 109
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
8/20
Table 1. Window glass produced before the 15th century has a clearly different composition from
that of later periods, especially regarding its SiO2 content. Despite this great compositional
similarity of the original glass, among the analysed set were encountered:
(a) fragments with almost no signs of deterioration,
(b) heavily deteriorated fragments but still with a core of original glass, and
(c) fragments that no longer contained any original glass.
Clearly, in such a sample set, it was not possible to find a relation between the chemical
composition of the original glass and its resistance against corrosion. It can be noted in Table 1
that, on average, the original glass contains a small amount of MnO (0.85%). Although other
metals such as copper, zinc and lead are present in the inclusions (see below, Figs 3 and 5), theseconcentrations are too low to allow detection in the original, uncorroded glass by means of
SEMEDX.
Figure 4 A detail of the leached layer of a glass fragment from a 15th century free-standing stained glass window: (a)
optical image; (b) backscattered electron image; (c) X-ray images.
110 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
9/20
Figure 3 (a) shows SEMEDX X-ray images of the Raversijde fragment (shown in Fig. 1).
The dark inclusions that formed along the low-density lamellae in stratified leached layers are
rich in manganese. The backscattered electron image of the central part of the corrosion body (see
Fig. 3 (b)) demonstrates that manganese has penetrated the darker lamellae in the leached layer,
indicating that the (lateral) diffusion in these layers is much faster than in the high-density
brighter ones. There is also an Mn-enrichment in cracks perpendicular to the lamellae.Figure 5 shows a number of SEMEDX X-ray spectra collected from the Raversijde fragment.
Figure 5 (a) is a spectrum of the original glass, where an initial amount of ~0.8% manganese is
Table 1 The average composition of 396 15th17th century
calco-potassic window glasses
Oxide Concentration [min; max]
Na2O 1.55 10.06 [0; 5.7]
MgO 3.57 10.04 [0.04; 7.4]
Al2O3 3.58 10.05 [0.6; 5.7]
SiO2 58.7 10.1 [51.5; 67.4]
P2O5 2.33 10.04 [0.14; 4.7]
SO3 0.20 10.01 [0; 1.03]
Cl 0.29 10.01 [0; 0.81]
K2O 6.2 10.1 [1.4; 18.6]
CaO 21.3 10.1 [8.8; 26.3]
MnO 0.85 10.02 [0.01; 4.7]
Fe2O3 0.93 10.01 [0.14; 2.15]
BaO 0.50 10.01 [0.02; 1.42]
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
Energy (keV)
Relativeintensity
O
Al
Si
P
Pb-M
KCa
Ba
ZnCu
Mn FeMg
Na
(a)
(b)
(c)
Figure 5 SEMEDX X-ray spectra of (a) original uncorroded glass, (b) a leached layer and (c) a manganese-rich
inclusion from the Raversijde sample shown in Figure 3.
Manganese staining of archaeological glass 111
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
10/20
present, next to other elements that are common in glass, such as Al, Si, P, K, Ca and Fe. Figure 5
(b) shows the SEMEDX spectrum of the leached layer. It is clear that the intensity of the X-ray
lines of most elements decreased as a result of the leaching process, while that of Al and Si
remained as dominant lines. The spectrum in Figure 5 (b) also indicates that more or less the
same concentration of Mn is present in the leached layer as in the original glass. In Figure 5 (c),
the spectrum of one of the manganese-rich inclusions is presented. While a lower Si signal is
observed, the Mn peak is substantially more intense than in Figure 5 (b). The spectrum in
Figure 5 (c) demonstrates that inside the manganese-rich bodies there is still a considerableamount of silicon present, while other metals such as copper, zinc and lead also appear to have
co-precipitated together with the Mn. This sample suggests that manganese migrated from the
environment into the glass.
X-ray images collected by means of SEMEDX of the Namur and Antwerp fragments (cf.,
Fig. 2) are presented in Figures 6 and 7. These X-ray images demonstrate that the Mn is
concentrated in dark-coloured inclusions, of ~30200 mm in size. In Figure 6, just beneath the
surface, at the bottom of the image, the inclusions appear to be isolated and surrounded by the
leached glass layer. However, the images acquired with light microscopy suggest that the isolated
inclusions in the leached layer have a tree-like appearance, spreading out from a central position
at the surface of the leached layer. These images indirectly suggest that the inclusions are notisolated inclusions, but are part of a three-dimensional structure that has grown into the leached
layer from the surface.
Figure 6 SEMEDX X-ray images of the cross-sectioned sample shown in Figure 2 (a).
112 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
11/20
Internal or external source of manganese?
When a volume of 1 mm3 of healthy glass contains less manganese than 1 mm3 of corroded glass,
then it can be assumed that manganese has migrated from the environment into the glass (i.e.,
external source). In the opposite case, the formation of inclusions is caused by concentrating the
manganese from the glass itself into a limited number of inclusions (i.e., internal source).
In principle, the type of source causing the formation of manganese inclusions can be deter-
mined by measuring the volume fraction, Vinclusions, occupied by the inclusions and by the volume
fraction of the leached layer, Vleached, inside the corroded glass. This can be done by means of
tomography. Also, the average manganese concentrations in the healthy glass,Whealthy(Mn), in the
leached layer,Wleached(Mn), and in the inclusions,Winclusions(Mn), need to be determined. However,
the major problem with SEMEDX is that the quantitative analysis requires a normalization of100 wt%, which is hampered by the unknown amount of water in the corroded glass. Two
methods were used to overcome this problem:
SiO2content as an internal standard.The number of Si atoms per cm3 in the healthy glass and
in the leached layer can be considered as the same. For this reason, the concentration W(Mn) in
wt% is proportional to the ratio of peak intensities I(Mn)/I(Si). However, for the inclusions this
is not always the case, because they can be formed in zones that are poor in SiO2, such as cracks
or low-density lamellae. As a result of this, the concentration W(Mn) for the inclusions will be
overestimated.
Peak-to-background ratio for the MnKa peak.Although this ratio is only approximately
proportional to the manganese concentration, it is an easy way to suppress the effect of live time,beam current or self-absorption of X-ray peaks. This method is used for the quantitative analyses
of various mineral particles (van Grieken and Markowicz 2002):
Figure 7 X-ray images of a cross-sectioned 14th century sample excavated in Antwerp. This image illustrates the
dendritic appearance of the manganese invasion.
Manganese staining of archaeological glass 113
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
12/20
R V W V W
W=
( ) + ( )leached leached inclusions inclusions
health
Mn Mn
yy Mn( )
For one tomographed archaeological sample, the volume fraction occupied by the inclusionsinside the corroded glass is 0.085. When the SiO2content is used as internal source, the ratio R
is equal to 1.78. For the peak-to-background method, Ris equal to 1.29. Both methods suggest
that manganese originates from an external source. However, neither quantification method could
be validated with the samples that were analysed and, therefore, the results should be used with
care. Fortunately, from an experimental point of view, the X-ray spectra collected from the
Raversijde sample and shown in Figure 5 indicate that it is possible for manganese to migrate
from the environment into the corroded glass because the intensity of the Mn-Kapeak in the
leached layer is similar to that of the healthy glass.
Manganese speciation by means ofm-XANES
The dominant oxidation state of manganese in the original glass, in the leached layer and in the
inclusions can be determined by means ofm-XANES, because a positive relation exists between
the redox state and the position of the MnK edge (Farges 2005). Such a relation was also noticed
for the MnL3edge (Schofieldet al. 1995). In uncorroded glass, Mn is predominantly present as
Mn2+ and, therefore, typically XANES spectra are obtained resembling that of the MnSO4reference compound shown in Figure 8.
Figure 8 (a) shows MnK XANES spectra from the original glass and the leached layer of the
Raversijde sample, while Figure 8 (b) shows the XANES spectra of the Mn-rich inclusions in
comparison to the reference spectra. The XANES profiles of the original glass and the leachedlayer indicate that mainly Mn2+ is present. The fact that the spectra of original glass and leached
layer profiles are nearly identical indicates that during the leaching process, no changes in the
(a) (b)
Figure 8 (a) Manganese K-edge XANES profiles of the original glass core and the leached layer of a heavily corroded
glass, excavated in Raversijde. Leached glass: profiles A and B; original glass: profile C. (b) Manganese K-edge XANES
profiles of a manganese-rich inclusion between lamellae of the stratified leached layer, excavated in Raversijde and
shown in Figure 3.
114 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
13/20
oxidation state of the Mn took place. The XANES spectra of the manganese-rich areas show a
strong resemblance to the MnO2reference profile. However, on the left edge of the absorption
jump an additional shoulder is visible in spectra AD of Figure 8 (b), indicating that next to
Mn(IV) lower-oxidation states of manganese can also be present. This might be the result of the
presence of Mn(II) and Mn(IV) in close proximity. Between the Mn(IV) inclusions, insoluble
Mn(II) hydroxides or carbonates might be present as well (Tonner et al. 1999).In Figure 9, the XANES spectra from the original glass, the leached layer and the Mn-rich
inclusions of the Namur samples are displayed, showing similar spectra as found for the Raver-
sijde samples. The same applies to the XANES spectra recorded from the Antwerp fragment
shown in Figure 10.
It can be concluded that in all cases, Mn(II)-like profiles are observed in the original glass and
in the leached layer, while the XANES spectrum of the dark inclusions or inclusions mostly
resembles that of Mn(IV). This similarity suggests that with respect to the Mn redox state, there
are no fundamental differences among the Mn-rich inclusions found in the corrosion bodies
studied. Therefore, it can be assumed that both types of inclusions (Mn inclusions between
lamellae and dendritic-shaped inclusions) are formed by identical, or at least very similar,processes.
The three-dimensional form and structure of the manganese inclusions
Although the two-dimensional images collected by optical and electron microscopy (cf.,
Figs 17) suggest that in some cases the Mn-rich inclusions form complex three-dimensional
dendrites or other shapes, these techniques only allow the visualization of the cross-sections of
these structures. A suitable method for studying these structures in their entirety is X-ray
absorption computed micro-tomography (CMT). The resolution and sensitivity obtainable from
present-day desktop X-ray micro-tomographs is sufficiently good to permit a detailed visualiza-tion of the MnOxnetwork within the corrosion bodies without physically slicing them (Roemich
et al. 2005).
Figure 9 Manganese K-edge XANES profiles of heavily corroded glass, excavated in Namur: profile A, Mn-rich
inclusions; profile B, leached layer; profile C, original glass.
Manganese staining of archaeological glass 115
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
14/20
Figure 10 Manganese K-edge XANES profiles of dendritic Mn-rich inclusions present in the leached layer of the glass,
excavated in Antwerp (see Fig. 2 (b)): profiles A and B, Mn-rich inclusions: profile C, leached layer.
Figure 11 A tomographic slice of a sample, excavated in Antwerp. This cross-section clearly shows the dendritic
invasion. Crusts on the surface of the leached layer are also visible.
116 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
15/20
A typical absorption tomogram is presented in Figure 11, showing the density variation within
the Antwerp fragment. The dark area is the remaining original glass, featuring a higher density
(and therefore a higher linear absorption coefficient) than the surrounding leached layers. The
dark spots inside the leached layer and close to the surface are the manganese-rich inclusions. In
this image, the dendrites just below the surface, the cracks (i.e., the dark lines), and tubular
inclusions (elliptical inclusions in the middle of the leached layer) can clearly be observed.
Another advantage of the use of tomographic imaging in this context is the fact that, simulta-
neously with vizualizing the density variations induced below the surface by the leaching
process, the precipitate formation on top of the original glass surface can also be studied; inFigure 11, the crust-like material attached on the lower glass surface has a dark tone, indicating
a high density/strong X-ray absorption, probably due to the presence of the cations that were
leached out of the glass. This material is usually not observed during electron microscopy
analyses, as it disappears during sample preparation.
The manganese inclusions in the 3D reconstruction (the dark grey pixels in Fig. 11) were
isolated from the surrounding material (the light grey pixels) by removing all pixels that did not
correspond with the grey value of the Mn-rich inclusions. Small regions were selected from the
data cube in order to evaluate the shape of the inclusions. In this manner, different shapes of
manganese-rich inclusions can be distinguished, as shown schematically in Figure 12:
(a) dendrites,(b) planar shapes, and
(c) tubular structures.
Figure 12 Manganese inclusions in different shapes: (a) penetration of dendrites from the glass surface, seen from the
centre of the glass towards the surface; (b) manganese in planar shapes; (c) tubular manganese enrichments just below
the dendrites; (d) a schematic overview of the different shapes.
Manganese staining of archaeological glass 117
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
16/20
The dendrites are located just below the glass surface and have a maximal penetration depth of
~600 mm. The dendrites shown in Figure 12 are seen from the centre of the glass towards the glass
surface. Due to a lack of resolution, only the contour of the dendrites can be observed, resulting
in droplet-shaped dendrites. The planar and tubular-shaped Mn-rich regions (see Figs 12 (b) and12 (c)) are mostly located deeper in the glass and are often oriented parallel with the glass surface.
A hypothesis explaining inclusion formation
For all of the glass analysed, manganese-rich inclusions occur only in the leached layer and never
in the original remaining glass. This indicates that the formation of such inclusions occurs in two
steps, where the first step involves the formation of a leached layer, followed by the formation of
Mn-rich inclusions in the leached layer. The reason why these inclusions were systematically
found in the leached layer and never in the original glass might be explained by a much higher
mobility of ions in the low-density leached layer. In stratified leached layers, which are usuallyparallel to the glass surface, diffusion in the low-density lamellae is easier than in the higher-
density lamellae, resulting in planar manganese inclusions. The formation of such parallel planes
of Mn-rich inclusions might be explained using the mechanism shown in Figure 13. The Mn
inclusions are formed along the interface between the healthy glass and the leached layer. When
the leached layer grows, the Mn inclusions will be formed at the new interface, deeper into the
glass. This mechanism suggests that the formation of Mn-rich inclusions can occur simulta-
neously with the formation of the leached layer formation. In that case, the formation of Mn
inclusions is influenced by the slow movement of the border between the leached layer and the
healthy glass. This explains the occurrence of Mn precipitates in elongated cracks perpendicular
to the surface, as can be seen in Figure 3 (b). In other cases, these inclusions can also be formedat later times, when the leached layer is already present.
It is remarkable that the Mn compounds are concentrated in a limited number of inclusions,
surrounded by Mn-poor leached layer. For one tomographed sample, the inclusions occupied
only 8.5 vol% of the corroded glass. Such a local enrichment can be the result of the migration
of all the manganese in the corroded glass (i.e., the internal source) towards a limited number of
centres. This mechanism might explain the presence of black inclusions in stained-glass windows
(Perez y Jorba et al. 1980). However, in one archaeological inclusion-containing sample, the
leached glass and the original glass feature similar Mn concentrations. In that case, the presence
of inclusions can only be explained by an intrusion of soluble Mn2+ from the surrounding
environment (i.e., the external source), where the glass was buried, into the glass surface.Experiments in which highly sensitive glasses were immersed in a solution of Pb2+, Cu2+ and Mn2+
demonstrated that these cations are able to penetrate the leached layer but not the healthy glass.
Figure 13 The formation of specific morphologies when the movement of the border between the leached layer and the
healthy glass is hampering the formation of Mn inclusions.
118 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
17/20
However, during such experiments the pH close to the glass surface increased, which could result
in a precipitation of the soluble Mn2+ into Mn(OH)2. It should be remarked that this phenomenon
might have affected the conclusions of Watkinsonet al. (2005). It is clear that for both internal
and external Mn sources, the formation of darkened glass requires a migration of Mn2+ towards
the inclusions. This migration is needed in order to explain the heterogeneous distribution of
manganese throughout the leached layer.
The Mn-rich inclusions appear to consist of Mn4+. Also, for dendritic pyrolusite intrusions in
sandstone and for manganese rock varnish the Mn valences are between 3+and 4+(McKeown
and Post 2001). Since the solubility of oxides and hydroxides of Mn4+ in aqueous systems is very
low, migration of Mn4+ from the surrounding environment into the glass cannot explain thepresence of Mn4+-rich inclusions. Moreover, Mn4+ does not exist in the original glass and in the
leached layer. This means that during the formation of the inclusions, Mn2+ must have been
oxidized into Mn4+. Oxidation of Mn2+ requires oxygen gas and water. This means that a flux of
both constituents, and in some cases also Mn(II), migrates from the external environment into the
leached layer. At the same time, a flux of Mn(II) migrates out of the glass. The interdiffusion of
both fluxes causes an oxidation and precipitation of Mn(IV) compounds as a direct consequence.
This reaction mechanism is visualized in Figure 14.
This explains why the inclusions are always in contact with the glass surface. Although the
oxidation of manganese appears to be driven by a spontaneous chemical reaction, a biological
driving force is not excluded. It should be mentioned that some micro-organisms live on theoxidation of manganese (de la Torre and Gmez-Alarcn 1994):
2Mn O 2H Oleached glass 2 leached glass 2 leached glass2+
( ) ( ) (+ + ))
( )+
( )
+2MnO 4H2 leached glass leached glass
Once nucleation centres are formed, a depletion of manganese around these centres occurs and
a concentration gradient is set up. As a consequence, more Mn diffuses down the gradient towards
the nucleation centres. Most probably, only in these centres are the cations transformed into Mn 4+.
The presence of a limited number of nucleation centres suggests that earlier formed inclusions
catalyse the oxidation of Mn
2+
. It is known that reactiondiffusion systems result in the formationof complex patterns (Chopard et al. 1999). A particular example is the formation of mineral
dendrites such as pyrolusite intrusions in sandstone and manganese rock varnish. The dendrites
Figure 14 The formation of Mn-rich inclusions due to the interdiffusion of a flux of water, oxygen and, in some cases,
also Mn(II) from the external environment into the weathered layer and a flux of manganese(II) out of the fresh glass.
Manganese staining of archaeological glass 119
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
18/20
are formed by a diffusion of reagents coupled to a precipitation reaction. Aggregates are formed
when saturation is reached or when the precipitation is catalysed by aggregates that were formed
previously (Chopard et al. 1991). This mechanism appears to be very similar to the inclusion
formation in leached layers.
CONCLUSIONS
In (partially) corroded black archaeological glass fragments, manganese enrichments were exclu-
sively found in the leached glass and never in the original glass. They do not invade the leached
layer uniformly; instead, they are concentrated in well-defined inclusions. The formation of such
inclusions is probably a secondary degradation phenomenon, occurring after a leached layer has
been formed. The inclusions have a planar appearance in the low-density lamellae of stratified
leached layers and in cracks. In other cases, dendritic or tubular shapes were observed, gradually
developing from the glass surface into the deteriorated glass.
It was shown that the manganese-rich inclusions contain black-coloured Mn(IV)-rich com-
pounds. This oxidation state does not occur in the original glass or in the leached layer. Mobile
Mn(IV) ions do not occur in natural waters. Therefore, mobile Mn(II) ions originating in the
surrounding soil or in the leached layer itself must be oxidized in accumulation centres, forming
Mn(IV)-rich substance precipitates in the micropores of the leached glass. This explains why the
inclusions contain manganese-rich compounds as well as leached glass. Migration of Mn(II)
towards the inclusions is coupled with a redox reaction, leading to the formation of complex
patterns such as dendrites. Such a diffusion-reaction coupling is possible when previously formed
Mn(IV) nuclei catalyse the reaction.
REFERENCES
Bunker, B. C., 1994, Molecular mechanisms for corrosion of silica and silicate glasses,Journal of Non-Crystalline Solids,
179, 3008.
Chopard, B., Herrmann, H. J., and Vicsek, T., 1991, Structure and growth mechanism of mineral dendrites, Nature, 353,
40912.
Chopard, B., Droz, M., Magnin, J., Rcz, Z., and Zrinyi, M., 1999, Liesegang patterns: effect of dissociation of the
invading electrolyte, Journal of Physical Chemistry A, 103(10), 14326.
Cooper, G. I., Cox, G. A., and Perutz, R. N., 1993, Infra-red microspectroscopy as a complementary technique to
electron-probe microanalysis for the investigation of natural corrosion on potash glasses, Journal of Microscopy,
170(2), 11118.
Cox, G. A., and Ford, B. A., 1993, The long-term corrosion of glass by ground-water,Journal of Materials Science, 28,563747.
de la Torre, M. A., and Gmez-Alarcn, G., 1994, Manganese and iron oxidation by fungi isolated from building stone,
Microbial Ecology, 2 7(2), 17788.
Domnech-Carb, A., Domnech-Carb, M. T., and Osete-Cortina, L., 2001, Identification of manganese(IV) centers in
archaeological glass using microsample coatings attached to polymer film electrodes, Electroanalysis, 13(11),
92735.
Domnech-Carb, M. T., Domnech, A., and Osete, L., 2006, A study on corrosion processes of archaeological glass from
the Valencian region (Spain) and its consolidation treatment, Microchimica Acta, 154, 12342.
Doremus, R. H., 1998, Diffusion of water and oxygen in quartz: reactiondiffusion model,Earth and Planetary Science
Letters, 163, 4351.
Falkenberg, G., Clauss, O., Swiderski, A., and Tschentscher, Th., 2001, Upgrade of the X-ray fluorescence beamline at
HASYLAB/DESY,X-Ray Spectrometry, 30, 1703.Farges, F., 2005, Ab initioand experimental pre-edge investigations of the Mn K-edge XANES in oxide-type materials,
Physical Review B, 71(15), 155109.
120 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
19/20
Freestone, I., 1985, Retention of phosphate in buried ceramics: an electron microbeam approach, Archaeometry, 27,
16177.
Hench, L. L., 1975, Characterization of glass corrosion and durability, Journal of Non-Crystalline Solids, 19,
2739.
Janssens, K. H. A., Adams, F. C. V., and Rindby, A. (eds.), 2000,Microscopic X-ray fluorescence analysis, 299303, John
Wiley & Sons, Ltd, Chichester.Kunicki-Goldfinger, J., 2008, Unstable historic glass: symptoms, causes, mechanisms and conservation, Reviews in
Conservation, 9, 4760.
Long, B. T., Peters, L. J., and Schreiber, H. D., 1998, Solarization of sodalimesilicate glass containing manganese,
Journal of Non-Crystalline Solids, 239(13), 12630.
Mandernack, K. W., Post, J., and Tebo, B. M., 1995, Manganese mineral formation by bacterial spores of the marine
Bacillus, strain SG-1: evidence for the direct oxidation of Mn(II) to Mn(IV), Geochimica et Cosmochimica Acta,
59(21), 4393408.
McKeown, D. A., and Post, J. E., 2001, Characterization of manganese oxide mineralogy in rock varnish and dendrites
using X-ray absorption spectroscopy, American Mineralogist, 86, 70113.
Melcher, M., and Schreiner, M., 2004, Statistical evaluation of potashlimesilica glass weathering, Analytical and
Bioanalytical Chemistry, 379(4), 62839.
Melcher, M., and Schreiner, M., 2006, Leaching studies on naturally weathered potashlimesilica glasses, Journal ofNon-Crystalline Solids, 352, 36879.
Misra, M. K., Ragland, K. W., and Baker, A. J., 1993, Wood ash composition as a function of furnace temperature,
Biomass and Bioenergy, 4(2), 10316.
Newton, R., and Davison, S., 1989,Conservation of glass, 154, Butterworth, London.
Perez y Jorba, M., Dallas, J. P., Bauer, C., Bahezre, C., and Martin, J. C., 1980, Deterioration of stained glass by
atmospheric corrosion and micro-organisms,Journal of Materials Science, 1 5(7), 16407.
Pollard, M., and Heron, C., 2008, Archaeological chemistry, ch. 5, Royal Society of Chemistry, Cambridge.
Proost, K., Vincze, L., Janssens, K., Gao, N., Bulska, E., Schreiner, M., and Falkenberg, G., 2003, Characterization of a
polycapillary lens for use in micro-XANES experiments,X-Ray Spectrometry, 32(3), 21522.
Roemich, H., Lpez, E., Mees, F., Jacobs, P., Cornelis, E., van Dyck, D., and Domnech Carb, T., 2005, Microfocus
X-ray computed tomography (mCT) for archaeological glasses, inCultural heritage conservation and environmental
impact assessment by non-destructive testing and micro-analysis(eds. R. van Grieken and K. Janssens), 3747, A.A.Balkema, Leiden.
Sayre, E. V., 1963, The intentional use of antinomy and manganese in ancient glasses, inAdvances in glass technology,
part 2(eds. F. R. Matson and G. E. Rindone), 26382, Plenum Press, New York.
Schalm, O., and Janssens, K., 2003, A flexible and accurate quantification algorithm for electron probe X-ray microanaly-
sis based on thin-film element yields, Spectrochimica Acta Part B: Atomic Spectroscopy, 58(4), 66980.
Schalm, O., Janssens, K., Wouters, H., and Caluw, D., 2007, Composition of 1218th century window glass in Belgium:
non-figurative windows in secular buildings and stained-glass windows in religious buildings,Spectrochimica Acta
Part B, 62, 6638.
Schofield, P. F., Cressey G., Wren Howard, P., and Henderson, C. M. B., 1995, Origin of colour in iron and manganese
containing glasses investigated by synchrotron radiation,Glass Technology, 36(3), 8994.
Scholze, H., 1982, Chemical durability of glasses, Journal of Non-Crystalline Solids, 52 , 91103.
Schreiner, M., 1991, Glass of the past: the degradation and deterioration of medieval glass artifacts, Mikrochimica Acta(Wien), 11, 25564.
Schreiner, M., Woisetschlger, G., Schmitz, I., and Wadsak, M., 1999, Characterisation of surface layers formed under
natural environmental conditions on medieval stained glass and ancient copper alloys using SEM, SIMS and atomic
force microscopy, Journal of Analytical Atomic Spectrometry, 14(3), 395403.
Silvestri, A., Molin, G., and Salviulo, G., 2005, Archaeological glass alteration products in marine and land-based
environments: morphological, chemical and microtextural characterization, Journal of Non-Crystalline Solids,
351(1617), 133849.
Smith, C. S., and Gnudi, M. T. (eds.), 1990, The pirotechnia of Vannoccio Bringiccio: the classic sixteenth-century
treatise on metals and metallurgy, 112, Dover, New York.
Stern, W. B., and Gerber, Y., 2004, Potassiumcalcium glass: new data and experiments,Archaeometry, 46 , 13756.
Sterpenich, J., and Libourel, G., 2001, Using stained glass windows to understand the durability of toxic waste matrices,
Chemical Geology, 174
(13), 18193.Sterpenich, J., and Libourel, G., 2006, Evidence of water diffusion in silicate glasses under natural weathering conditions
given by buried medieval stained glasses, Journal of Non-Crystalline Solids, 352, 544651.
Manganese staining of archaeological glass 121
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122
8/13/2019 Article - Schalm2010 - Manganese Staining of Archaeological Glass
20/20
Tonner, B. P., Droubay, T., Denlinger, J., Meyer-Ilse, W., Rothe, J., Kneedler, E., Pecher, K., Nealson, K., Grundl, T., and
Lewandowski, Z., 1999, Soft X-ray spectroscopy and imaging of interfacial chemistry in environmental specimens,
Surface and Interface Analysis, 27(4), 11528.
Van Grieken, R. E., and Markowicz, A. A. (eds.), 2002,Handbook of X-ray spectrometry, 2nd edn, revised and expanded,
9068, Marcel Dekker, New York.
Watkinson, D., Weber, L., and Anheuser, K., 2005, Staining of archaeological glass from manganese-rich environments,Archaeometry, 4 7, 6982.
Woisetschlger, G., Dutz, M., Paul, S., and Schreiner, M., 2000, Weathering phenomena on naturally weathered
potashlimesilicaglass with medieval composition studied by secondary electron microscopy and energy dispersive
microanalysis,Microchimica Acta, 135, 12130.
122 O. Schalmet al.
University of Oxford, 2010, Archaeometry 53, 1 (2011) 103122