Anda di halaman 1dari 8

This article was published in an Elsevier journal.

The attached copy


is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:

http://www.elsevier.com/copyright
Author's personal copy

Spectrochimica Acta Part B 62 (2007) 1123 – 1129


www.elsevier.com/locate/sab

Study of archaeological ceramics by total-reflection X-ray fluorescence


spectrometry: Semi-quantitative approach
R. Fernández-Ruiz a,⁎, M. García-Heras b,c
a
Universidad Autónoma de Madrid, Facultad de Ciencias, Servicio Interdepartamental de Investigación,
Modulo C-9, Laboratorio de TXRF, Crta. Colmenar, Km 15, Cantoblanco, E-28049, Madrid, Spain
b
Instituto de Historia-CSIC, C/Serrano, 13. E-28001, Madrid, Spain
c
CENIM-CSIC, Avda. Gregorio del Amo, 8. E-28040, Madrid, Spain
Received 28 December 2006; accepted 24 June 2007
Available online 23 August 2007

Abstract

Total-reflection X-ray fluorescence spectrometry has been compared with Instrumental Neutron Activation Analysis in order to test its potential
application to the study of archaeological ceramics in the archaeometric field. Two direct solid non-chemical sample preparation procedures have been
checked: solid sedimentation and solid chemical homogenization. For sedimentation procedure, total-reflection X-ray fluorescence allows the analysis
of the elemental composition with respect to the size fraction but not the average evaluation of the composition. For solid chemical homogenization
procedure, total-reflection X-ray fluorescence provides precise (from 0.8% to 27% of coefficient of variation) and accurate results (from 91% to 110% of
recovery) for 15 elements (Cr, Hf, Ni, Rb, Al, Ba, Ca, K, Mn, Ti, V, Cu, Ga, Y and Fe) with an easy sample preparation process of the solid clay and
without previous chemical treatment. The influence of the particle sizes has been checked by total-reflection X-ray fluorescence sample angle scans and
anomalous behaviors have been found for three additional detected elements: As, Sr and Zn, which can be attributed to interference effects of the mineral
grain sizes of their associated chemical phases in the total-reflection X-ray fluorescence interference region. The solid chemical homogenization
procedure produces data useful for archaeological interpretation, which is briefly illustrated by a case-study. Finally, the decantation procedure data can
be also useful for size chemical speciation and, consequently, for alternative environmental total-reflection X-ray fluorescence applications.
© 2007 Elsevier B.V. All rights reserved.

Keywords: TXRF; INAA; Archaeological ceramics; Size fraction; Compositional characterization

1. Introduction new information from emerging techniques. Nevertheless, the


wide use of this analytical method has not yet overcome, in some
The determination of the composition of ancient ceramic cases, drawbacks such as high cost, or difficulty of access to a
samples has a special relevance to the construction of heuristic suitable nuclear reactor for sample irradiation [2]. As alter-
models concerning the production and distribution of these natives, less expensive and more accessible methods have been
materials in the past. Currently, this kind of study constitutes employed such as traditional XRF, AAS or ICP-OES. One of the
about one third of all archaeometric research that is carried out challenges facing current archaeometric research on ceramics is
on an international scale [1]. Since the 1970s, the analytical tool the intra-regional differentiation of pottery production centers, a
most often used and accepted for the archaeological community subject that requires large and accurate sets of data on the
has been INAA, because of its ability to provide the required composition of ceramics. Therefore, the establishment of new
levels of accuracy, precision and detection limits for its analytical tools or the development of those already proven
application in archaeological ceramics studies. The large number requires both correct standardization and normalization with
of analyses already performed by several INAA laboratories also international accepted analytical techniques, such as INAA, in
provides an important data bank which can be compared with order to assure the validity of the analytical results [3–5].
TXRF main features and potential applications can be found in
⁎ Corresponding author. Fax: +34 914973529. the paper of Prange [6] or the excellent book of Klockenkämper
E-mail address: ramon.fernandez@uam.es (R. Fernández-Ruiz). [7]. Previous investigations, carried out by Klockenkämper et al.
0584-8547/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.sab.2007.06.015
Author's personal copy

1124 R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129

[8], Cariati et al. [9] or Fernández-Ruiz et al. [10–12], were the foil, adjusting the intensity so that a count rate of about 5000 cps
departure point for this investigation. The best approach to the was achieved and an acquisition time of 1000 s. TXRF 8030C
ideal TXRF thin film criteria for a solid particle deposition is that spectrometer (Cameca, France) was also used with the objective
the following three fundamental requirements are fulfilled: (1) of undertaking studies of angular dependence of signal intensity
chemical homogeneity of the solid particles, (2) average particle for some elements. This spectrometer combines a 3 kW X-ray
sizes around 1 μm and lower than 10 μm [13], and (3) tube with a Mo/W alloy anode with a W/C double-multilayer
homogeneous spatial distribution of the deposited particles on monochromator and in addition allows the variation of the
the sample carrier. incidence angle by tilting the sample holder unit.
To provide the TXRF validation for this type of studies, a The QELS system used in this study was the AutoSizer IIc of
reference clay material analyzed routinely by INAA was Malvern Instruments Ltd., equipped with a He–Ne 5 mW laser,
employed as reference material, Ohio Red Clay from Resco a photo-multiplier and a processing electronic system controlled
Products, Inc, Oak Hill, OH, USA (Ohio Red Clay-2 hereafter). by the Malvern AutoSizer computer package. The SEM
This sample was habitually used as batch control by the team of equipment used was the Philips XL-30 equipped with a W
the University of Missouri-Columbia in their INAA analyses source, detectors of secondary and backscattering electrons and
[14]. This work presents the results obtained in the comparison a vacuum working lower than 4 × 10− 4 Pa.
of both techniques (INAA and TXRF) for the Ohio Red Clay-2
reference sample and also, the investigations carried out for the 2.2. Procedure of sedimentation
samples preparation and optimization for TXRF analysis. The
application of the final TXRF procedure developed was applied 2.2.1. Samples preparation
for evaluating a real archaeological case. Keeping in mind the three main requirements mentioned in
the Introduction section, and as a first approach to the problem,
2. Experimental section it is possible to obtain different clay size fractions by the
sedimentation method. Three sets of samples (5 samples each)
2.1. Instrumentation were prepared from the Ohio Red Clay-2 reference sample.
First, the sample was ground for 30 min in an agate mortar.
Three main techniques have been used in this work. TXRF Then 100 mg of the ground sample was poured into a test-tube
for the analysis of the samples; quasi-elastic light scattering and mixed with high-purity water (Milli-Q, 18.2 MΩ) up to
spectroscopy (QELS) for the determination of the distribution 10 ml. Next, the test-tube was placed in an ultrasonic bath for
of particle sizes in suspension of the analyzed samples and 30 min in order to disaggregate and homogenize the sample.
scanning electron microscopy (SEM) for the investigation of the Once this process was carried out, the sample was left to settle
final size, distribution and homogeneity of particles deposited for 12 h. Finally, aliquots of 1 ml were taken, measured from the
on the flat carrier. base, at three different depths (15 cm in Set-1; 10 cm in Set-2;
The analysis by TXRF was performed by using a Seifert and 5 cm in Set-3) in order to obtain an average distribution of
EXTRA-II spectrometer (Rich Seifert & Co, Ahrensburg, different particle sizes (always lower than 10 μm and with an
Germany), equipped with a molybdenum X-ray fine focus average size around 1 μm as postulated in the second main
lines, and a Si(Li) detector with an active area of 80 mm2 and a requirement). Thus, the following three particle size sets were
resolution of 157 eV at 5.9 keV (Mn Kα). The measurements obtained: Set-1, lower than 2 μm; Set-2, between 0.05 and
were performed working at 50 kV and filtered with a 50 μm Mo 5 μm; and Set-3, between 0.1 and 10 μm. Fig. 1 shows the three
size distributions obtained by QELS.
From 2 to 5 μl of each fraction of particles was deposited on
a quartz sample support and dried on a ceramic hot plate. All
manipulations were made in an A-100 class laminar flow
chamber.

2.2.2. Results and discussion


In these conditions the samples were analyzed by TXRF.
Link Analytical AN-10000 computer package was used to
deconvolute the registered spectra. Once the element areas were
obtained, they were converted into relative mass units using the
equation

SFe Ax
mx ¼ mFe ð1Þ
Sx AFe

where m is the relative mass of a given element, A is the peak


area, S is the relative sensitivity and the index x and Fe denotes
Fig. 1. QELS spectra of aqueous sample distribution of Set-1, Set-2 and Set-3. each element analyzed and element chosen as reference
Author's personal copy

R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129 1125

Cu, Ga and Y, with respect to the particle size fraction analyzed


were found. In the case of the enrichments, the fraction of larger
particle sizes (Set-3, between 0.1 and 10 μm), was higher for
these elements. In the case of the impoverishments, the fraction
of lower particle sizes (Set-1, between 0.01 and 2 μm) was
higher for these elements. In the case of U-up behavior, the
fraction of medium particle sizes (Set-2, between 0.05 and
5 μm) was the lowest for this element. In the case of U-down
behavior, the fraction of medium particle sizes (Set-2, between
0.05 and 5 μm) was higher for these elements.
The behaviors observed experimentally can be explained
considering the heterogeneous nature of the clay. Clays are
multi-layered mineral mixtures in which there are differences in
mineral and chemical composition between particles of different
sizes. This fact implies that the densities of micro particles can
be different. This finding contradicted the first main require-
Fig. 2. Representative TXRF spectrum from Ohio Red Clay-2 reference sample. ment of our approach by lost of chemical homogeneity.
Therefore, from the quantitative point of view the method of
respectively. The S values are known since they were measured sedimentation can be rejected to analyze this kind of material by
by means of standard solutions, while the rest of the parameters TXRF. On the other hand, the procedure developed could open
are experimental ones [12]. the way to evaluate the chemical composition of a generic solid,
TXRF analyses were performed for the elements As, La, Nd, based on its size and it could have interest in environmental
U, Co, Cr, Cs, Hf, Ni, Rb, Sc, Sr, Zn, Al, Ba, Ca, K, Mn, Ti, V, pollution studies (e.g., chemical analysis of polluted soils or
Si, Cu, Ga, Y and Fe. From the 25 elements evaluated, a total urban aerosols to diagnose air pollution from traffic and
number of 18 were chosen for routine analysis. Si was rejected factories).
because the sample support was made of quartz, which involved Three elements, not analyzed by INAA in this material, Cu,
a Si random rise which was difficult to quantify. This fact could Ga and Y, have been determined by means of TXRF. From the
be avoided by using plastic sample supports. Nonetheless, it archaeological point of view, this fact is of great relevance since
was preferred to use a quartz sample support due to its lower an increase of the number of analyzable elements in an
background contribution. Co was also rejected because, in this archaeological material implies an increase of the number of
case, it was seriously interfered by the high intensity of Fe K variables able to solve its geographical origin.
lines signal. La, Nd, U, Cs and Sc were rejected because they
were close to the detection limits that TXRF presents for this 2.3. Procedure of solid chemical homogenization
matrix.
A relative mass value of 100 was associated to the Fe signal In order to resolve the phenomenon of segregation produced
(mFe). Fe was chosen because it is always present in clays and by the procedure of sedimentation and to comply with the first
ceramics of archaeological interest and it usually shows a
clearly differentiated peak in TXRF spectra as Fig. 2 shows.
The results of the Ohio Red Clay-2 reference sample
analyzed by INAA in the Archaeometry Laboratory at MURR
were expressed in ppm. For this reason, they were renormalized
in order to make the results comparable with those obtained by
TXRF. Thus, using the equation
cx
%vs Fex ¼ 100 ð2Þ
cFe
for the same group of elements it was possible to transform
quantitative data cx (ppm) into semi-quantitative results
expressed as % vs Fe.
Fig. 3 shows the TXRF results obtained for Set-1, Set-2 and
Set-3 compared with INAA measurements, all expressed as
relative concentrations in % vs Fe.
The results obtained in Fig. 3 show that different behaviors
were present for the elements evaluated. Enrichments of the
elements As, Al, Rb, Sr and K; impoverishments of the Fig. 3. Relative mass fraction for the 19 elements analyzed by TXRF for the
elements Mn and Ti; U-up behavior for the element Ba; and three different particle size fractions (Set-1, Set-2, Set-3) compared with INAA
finally, U-down behavior for the elements Cr, Hf, Ni, Zn, Ca, V, results. Error bars obtained with n = 5 for TXRF and n = 20 for INAA.
Author's personal copy

1126 R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129

Fig. 4. SEM micrographs of depositions according to different sample suspension agents. a–b: Ohio Red Clay-2 in high-purity water. c–d: Ohio Red Clay-2 in toluene.
e–f: Archaeological pottery sample in high-purity water.

main requirement, it was decided to prepare a new set (Set-CH) 2.3.2. SEM verification
in which the sedimentation problems were diminished. In order to verify the third main requirement, different kinds
of depositions were generated on quartz sample supports and
2.3.1. Sample preparation observed by SEM after gold metallization. The first one was a
First, the sample was ground by an agate mortar until it had a suspension in water, showing a distribution of concentric rings
particle size lower than 30 μm. Then it was ground again for 1 h in shape, probably associated with an edge effect in the liquid–
by using a vibration micro-pulverizer equipped with a ball and solid interface during the water evaporation process (Fig. 4-a,
base of agate. Afterwards, 1 ml of high-purity water was added. b). In order to obtain a distribution of particles that were isolated
Next, the mix was poured into a test-tube in which high-purity and homogeneous, we experimented with different agents of
water was added up to 10 ml. The sample was homogenized for different surface tensions, particularly with toluene and high-
1/2 h by ultrasonic disaggregation in order to disperse possible purity water. It was established that toluene provided the best
agglomeration of particles. Finally, the particle size distribution distribution (Fig. 4-c, d). Nevertheless, it was noticed that this
in suspension was checked by using QELS until it had the agent seriously distorted the analytical determinations. Probably
required distribution and, therefore, was in accordance with the the observed quantitative analytical distortion can be attributed
second main requirement. When the sample had this size to the mountain effect, at micro particle level, that can be
distribution, it was again homogenized and, maintaining the appreciated in Fig. 4-d. This effect produces the lost of the thin
agitation always constant, 5 μl of the suspension was placed on film geometry and the appearance of the energy dependent
the quartz flat carrier and dried on a ceramic hot plate. All matrix effect, which is in agreement with the experimental
manipulations were made in an A-100 class laminar flow observations. As a result, we finally decided to use high-purity
chamber. water, even though such a water did not strictly comply with the
Author's personal copy

R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129 1127

Table 1 unequivocally detected as the spectrum associated with the


Ohio Red Clay-2 elemental relative mass fraction expressed as % vs Fe for reference material shown (Fig. 1). Previous results (Fig. 3)
INAA and TXRF
present the same systematic increase in the relative concentra-
Element INAA σ (n = 20) CV (%) TXRF σ (n = 5) CV (%) Accuracy tions of As, Sr and Zn with respect to INAA values. This implies
% vs Fe % vs Fe Recovery that the systematic increase is independent of the average
(%) particle size fraction analyzed. One possible explanation can be
As 0.028 0.002 7.2 0.040 0.002 4.3 146 attributed to high variations in the sizes of mineral phases
Cr 0.18 0.01 5.6 0.180 0.010 5.8 101 present in the clays and associated to As, Zn and Sr as Prange et
Hf 0.015 0.001 6.9 0.013 0.004 27.0 91
al. suggest [15].
Ni 0.14 0.03 21.4 0.150 0.009 5.8 107
Rb 0.36 0.02 5.6 0.354 0.004 1.0 98 A deeper study of this discrepancy was carried out in order to
Sr 0.05 0.06 106.0 0.131 0.004 2.8 249 understand the differences found. By means of the TXRF 8030c
Zn 0.18 0.02 11.1 0.284 0.009 3.2 157 spectrometer, a sample-angle scan of one representative sample
Al 185 6 3.2 171 4 2.4 92 of the Set-QH was carried out. The scan was performed from
Ba 1.1 0.3 26.3 1.24 0.06 4.5 109
0.25 to 2.25 mrad, around the critical angle for quartz, with a
Ca 3.0 0.4 13.3 2.81 0.04 1.4 93
K 64 16 25.1 69.3 0.8 1.1 109 step of 0.1 mrad and a measurement time per step of 100 s. Fig.
Mn 0.51 0.01 2.0 0.50 0.01 1.9 98 5 shows the angular behavior of the following elements: Fe, Ti,
Ti 12.0 0.6 5.0 12.35 0.10 0.8 103 Rb, K, As, Sr and Zn. The different curves were normalized at
V 0.39 0.01 2.6 0.43 0.01 2.3 110 the critical angle for quartz, defined as the inflection point of the
Cu n.m a n.m n.m 0.040 0.005 12.7 n.e b
angular curve, located at around 1.87 mrad.
Ga n.m n.m n.m 0.050 0.002 3.6 n.e
Y n.m n.m n.m 0.08 0.01 12.5 n.e The angular behavior for the representative elements that
Fe 100 — reference 100 — reference present good accuracy (Fe, Ti, Rb and K) is very similar (see
a
Not measured. Fig. 5) and is in agreement with the analytical results (see
b
Not evaluated. Table 1). In contrast, the angular behavior for the inaccurate
elements (As, Sr and Zn) presents strong differences with
respect to the Fe behavior chosen as reference. The angular
third main requirement for clays. In practice, water offered the behavior of the elements As, Sr and Zn presents strong
best analytical results. oscillations in its signals. These oscillations are typical when the
The final objective of this study was the application to particle sizes are lower than 100 nm [15] and they are usually
ceramic materials. Due to this fact, we tested the behavior of a associated to the appearance of X-ray interference effects in the
pottery sample using water as suspension agent. Fig. 4-e and f interference region of TXRF. In this case, the solid particles
shows the aspect of an archaeological pottery preparation associated to As, Sr and Zn should be in the range between 20
suspended in water, where a homogeneous distribution of the and 50 nm to explain the observed modulation of their angular
deposition can be observed. dependence.

2.3.3. Results and discussion 2.4. Application to an archaeological case-study


Semi-quantitative results were obtained by replicating 5
analyses of the same Ohio Red Clay-2 reference material. From an analytical point of view, the primary objective of
Table 1 shows the results of the 18 elements chosen for routine this study was to assess the compositional variability of the
analysis in Set-CH. Data were acquired the same way as that for
Set-1, Set-2 and Set-3.
Results show that, in this case, the accuracy for Cr, Hf, Rb,
Ni, Al, Ba, Ca, K, Mn, Ti, Fe and V with respect to INAA values
was excellent, as is shown in Table 1. The accuracy was
evaluated by measuring the recovery percentage with respect to
the INAA values. For the previous 12 elements the recoveries
vary from 91% for Hf to 110% for V. The precisions obtained in
the measurements by TXRF were evaluated by measuring the
coefficients of variation (CV) expressed as percentages. TXRF
CVs vary from 0.8% for Ti to 27% for Hf, whereas in the
analysis by INAA CVs vary from 2% for Mn to 106% for Sr.
These analytical aspects do not imply that TXRF is a more
powerful tool than INAA, but that TXRF may offer both
precision and accuracy comparable with and, in some cases,
better than INAA.
It is important to note the disagreement between the INAA
and TXRF values for the elements As, Sr and Zn. For these Fig. 5. Sample angle scans versus integrated areas for Fe, Ti, Rb, K (circles) and
elements, TXRF presents a high sensitivity and they are Sr, Zn and As (triangles). Continuous lines are B-Spline interpolation.
Author's personal copy

1128 R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129

(Set-QH). We used such values in order to check the capability


of TXRF to discriminate groups of different composition.

2.4.1. Results and discussion


The cluster analysis employed the average linkage of the
mean Euclidean distances matrix. The resulting dendrogram
suggests that the data set had two parts, one of them constituted
by standard measurements, and the other constituted by the 24
samples analyzed. Within the latter, it may be observed that
samples are separated into two sub-groups. There are sherds
from Fabrics 2, 3 and 4 in both sub-groups, whereas only in the
first one sherds appear to belong to Fabric 1. Further clarity was
introduced as a result of the use of principal components
analysis. The three first components summarized 71.89% of the
total variation in the data. In Fig. 6a, the plot represents sample
scores with respect to the two first components. Three distinct
groups can be observed. All samples belonging to Fabric 1
appear on the left. The rest of the fabrics appear at the center.
The two standards plot on the upper part of the diagram.
Correlations between the variables responsible for these
associations may be seen in Fig. 6b. The first component
summarized 45.29% of the total variation. Four elements were
positively correlated with this one: Ba, V, Hf and Mn, whereas
Al, Ti, Rb and Cr had a negative correlation. The second
component summarized 18.48% of the total variation. Cu, Ni,
Rb and Cr had a positive correlation and Zn, Sr, K and Ga had a
negative correlation. Finally, the third component (not repre-
sented in the figures) had Y, Ni, Cr and Cu correlated in a
positive way, whereas Ca, As and K had a negative correlation.
Therefore, the discrimination of fabric 1 was related to the first
Fig. 6. (a) Plot of the first two principal components scores derived from analysis component (Fig. 6a).
of the 24 pottery EP samples plus results of the 2 standards. ST-1 INAA results To summarize, the statistical analysis showed two distinct
and ST-2 TXRF results for Set-QH. (b) Weight of the 17 variables taken into associations. One group was constituted by fabric 1, and one
account. group was formed by the remainder of the fabrics. The
compositional variability observed seems to reflect a high
homogeneity in the sherds analyzed, which might indicate a
pottery ensemble provided by some archaeological excavations, local provenance. The presence of two associations in the data
as a possible indicator of different sources or production units. set may be related either to the exploitation of two different
In this archaeological case, a study carried out by TXRF on sources of raw materials or to the presence of different
ceramic materials from the Celtiberian archaeological site of El production units. The percentages of fabric 1 in the total pottery
Palomar (Aragoncillo, Guadalajara, Spain) is briefly illustrated. ensemble, as well as typological forms associated to this fabric,
The settlement is located in the Upper Jalón Valley, adjacent to do not suggest a foreign origin for this type of manufacture.
the Upper Duero Valley in the north-eastern part of the Spanish Therefore, the local character of these productions is suggested
Central Meseta. A total of 24 representative samples from the in a preliminary way. This indicates that this pottery could have
different percentages of each fabric or type of ceramic been manufactured in the geographical area where El Palomar is
manufacture present in the settlement were analyzed (EP-1 to located. However, it is not possible to establish that pottery was
EP-4 Fabric 1; EP-5 to EP-8 and EP-13 to EP-21 Fabric 2; EP- elaborated in the archaeological site itself, particularly in view
22 to EP-24 Fabric 3; EP-9 to EP-12 Fabric 4). Analyses were of the absence of remains related to pottery production. Finally,
carried out by TXRF according to the solid chemical it is evident that the inclusion of analytical data for the standard
homogenization procedure mentioned above for Set-QH. Two Ohio Red Clay-2 (ST-1, INAA results and ST-2, TXRF results)
statistical procedures were used in the exploration of data: has revealed the ability of the TXRF technique to detect
cluster analysis and principal components analysis. Both were compositionally different groups.
carried out on log-transformed data on the concentrations of 18
elements determined, excluding Fe, in order to balance the 3. Conclusions
difference of magnitude between the major and trace elements
[16]. We also introduced the values of the standard in the The method of sedimentation can be rejected to analyze this
statistical analysis, either those obtained by INAA or by TXRF kind of material by TXRF for archaeological applications.
Author's personal copy

R. Fernández-Ruiz, M. García-Heras / Spectrochimica Acta Part B 62 (2007) 1123–1129 1129

However, the TXRF procedure developed could be a fruitful Acknowledgements


avenue to evaluate the chemical composition speciation of a
generic solid, based on its size and it could have interest in The authors wish to express their gratitude to Dr. Hector Neff
environmental pollution studies. and Dr. Michael D. Glascock (Archaeometry Laboratory at the
A reliable sample preparation process is needed for Missouri University Research Reactor Center) for giving them
application of TXRF to the analysis of archaeological ceramic permission to use a Ohio Red Clay-2 standard as well as data of
materials and the chemical homogenization approach complies their measurements by neutron activation.
with this requisite. This procedure implies that the ceramic
sample has chemical homogeneity, average particle size around References
1 μm and lower than 10 μm, and a chemically homogeneous
distribution of its particles. The final procedure can be [1] R. Djingova, I. Kuleff, Provenance study of pottery and glass by INAA,
summarized as follows: homogenization of the samples by a Neutron Activation Analysis and Plasma Emission Spectrometric Analysis
in Archaeology. Techniques and Applications, Occasional Paper, vol. 82,
high grinding process, suspension of the fine-ground mixture of British Museum, London, 1991, pp. 121–141.
the whole sample in high-purity water, ultrasonic disaggrega- [2] W.J. Bennett Jr., J.A. Blakely, R. Brinkmann, Ch.J. Vitaliano, The
tion of such suspension, and deposition with continuous provenience postulate: thoughts on the use of physical and chemical data in
agitation of an aliquot of that suspension on a flat carrier. The the study of ceramic materials, Analysis and Publication of Ceramics, Int.
more important aspects are the following: the chemical Ser., vol. 551, B.A.R., Oxford, 1989, pp. 31–44.
[3] G. Bronitsky, The use of materials-science techniques in the study of
homogenization procedure has proved to be quick, relatively pottery construction and use, Advances in Archaeological Method and
easy, and inexpensive and, in addition, does not require any Theory, vol. 9, Academic Press, New York, 1986, pp. 209–276.
chemical manipulation. [4] M. García-Heras, R. Fernández-Ruiz, J.D. Tornero, Analysis of archae-
The analysis was performed for 18 elements: Al, Ca, K, Ti, ological ceramics by TXRF, J. Archaeol. Sci. 24 (1997) 1003–1014.
Fe (majors); Mn, Zn, Sr, Rb, Cr (minors) and As, Hf, Ni, Ba, V, [5] A. Hein, H. Mommsen, J. Maran, Element concentration distributions and
most discriminating elements for provenancing by neutron activation
Cu, Ga and Y (traces). The results obtained show a high analyses of ceramics from Bronze Age sites in Greece, J. Archaeol. Sci. 26
accuracy for Cr, Hf, Rb, Ni, Al, Ba, Ca, K, Mn, Ti, Fe and V. (1999) 1053–1058.
The recoveries (accuracy) obtained vary from 91% for Hf to [6] A. Prange, Total reflection X-ray spectrometry: method and applications,
110% for V. The precisions obtained vary from 0.8% for Ti to Spectrochim. Acta Part B 44 (1989) 437–452.
27% for Hf, whereas in the analysis by INAA they vary from [7] R. Klockenkämper, Total Reflection X Ray Fluorescence Analysis, Wiley,
New York, 1997.
2% for Al to 106% for Sr. [8] R. Klockenkämper, A. von Bohlen, Survey of sampling techniques for
TXRF results for the elements As, Sr and Zn presented an solids suitable for microanalysis by total-reflection X-ray fluorescence
unexpected inaccuracy with respect to INAA values. These spectrometry, J. Anal. At. Spectrom. 14 (1999) 571–576.
unexpected values were studied more deeply by means of angle- [9] F. Cariati, P. Fermo, S. Gilardoni, A. Galli, M. Milazzo, A new approach
for archaeological ceramics analysis using total reflection X-ray
scan measurements and an anomalous behavior was found.
fluorescence spectrometry, Spectrochim. Acta Part B 58 (2003) 177–184.
Such behavior has been associated to the presence of particle [10] R. Fernández-Ruiz, V. Bermudez, Determination of the Nb/Ta mass ratio in
sizes between 20 and 50 nm corresponding to mineral phases LiNb1–xTaxO3 by TXRF, Spectrochim. Acta Part B 60 (2005) 231–235.
associated to As, Sr and Zn and therefore, the associated [11] R. Fernández-Ruiz, M. Furió, F. Cabello Galisteo, M. López Granados, R.
interference effects explain and justify the differences observed Mariscal, J.L.G. Fierro, Chemical analysis of used three way catalysts by
between TXRF and INAA values. total reflection X-ray fluorescence, Anal. Chem. 74 (2002) 5463–5469.
[12] R. Fernández-Ruiz, J. Capmany, Determination of the rare-earth/Nb mass
The archaeological case-study shows that TXRF is a useful ratio in doped LiNbO3 by the TXRF technique, J. Anal. At. Spectrom. 16
technique for the determination of the compositional attributes (2001) 867–869.
of ceramics. The comparison of these results with data [13] R. Klockenkämper, A. von Bohlen, Determination of critical thickness and
previously obtained from the clay reference material clearly the sensitivity for thin film analysis by total reflection X-ray fluorescence
shows the ability of the technique to distinguish groups of spectrometry, Spectrochim. Acta Part B 44 (1989) 461–469.
[14] J.W. Cogswell, H. Neff, M.D. Glascock, The effect of firing temperature
different compositional nature. On the other hand, TXRF has on the elemental characterization of pottery, J. Archaeol. Sci. 23 (1996)
provided analytical values comparable to those provided by 283–287.
INAA for 15 elements. Three additional elements (Cu, Ga and [15] A. Prange, U. Reus, H. Schwenke, J. Knoth, Optimization of TXRF
Y) present in the Ohio Red Clay-2 standard were determined by measurements by variable incident angles, Spectrochim. Acta Part B 54
(1999) 1505–1511.
TXRF, even though they were not analyzed by INAA.
[16] R.L. Bishop, H. Neff, Compositional data analysis in archaeology, in: R.O.
At the current stage of our research, TXRF provides semi- Allen (Ed.), Archaeological Chemistry IV, Advances in Chemistry Series,
quantitative results. This fact is not a serious drawback for vol. 220, American Chemical Society, Washington, 1989, pp. 57–86.
archaeological applications as the archaeological case-study
demonstrates. In any case, we are currently working in the
development of new TXRF strategies which will be certainly able
to quantitatively analyze this kind of materials in the next future.

Anda mungkin juga menyukai