Elemental concentrations obsidian

FIGURE 22 Obsidian in the eastern Mediterranean Sea area. Studying the relative concentration of trace elements in obsidian makes it possible to identify the obsidian and to determine its provenance. Determining the relative amounts of barium and zirconium in ancient obsidian tools and in samples from different sources of the natural glass, for example, made it possible to identify the provenance of obsidian used in eastern Mediterranean Sea area sites (Renfrew and Dixon 1976). 

X-ray fluorescence is a rapid and low-cost method that can be performed on solid samples. However, the depth of penetration of X-rays in most solid samples is relatively shallow. High-precision XRF on geological samples such as obsidian requires preparation of homogeneous, powdered samples pressed into pellet form. If some loss of precision and accuracy due to irregular size, shape, and thickness of samples is acceptable, obsidian specimens can be analyzed non-destructively. Samples smaller than 1 cm in diameter or with element concentrations less than 5 ppm are generally not suitable for XRF. XRF can determine about 10-15 elements in obsidian (K, Ti, Mn, Fe, Zn, Ga, Rb, Sr, Y, Zr, Nb, Pb, and Th). Fortunately, many of the measurable elements are the incompatible elements which provide discrimination between sources. 

Table II. Element concentration means and standard deviations for obsidian sources from Peru of major archaeological importance.

Elemental concentrations and standard deviations, obsidian sources, Peru, 539, 540f-542f... 

Based on the analysis made for thousands of thick obsidian and pottery samples analyzed over a six-year period, the accuracy and precision of PIXE measurements for thin and thick sample analyses have been foimd to be as low as 1.6% for major elements with precision ranging from 5% to 10% depending on the elemental concentration (Cohen 2002). 

If elemental concentration rather than oxide concentrations are required, then the oxide conversion can be removed at this point. For obsidian (rhyolitic volcanic glass), using silicon as an internal standard yields results in very close agreement with the Gratuze approach, as shown in Table 37.1. 

Monitoring the amount of material removed by the laser and transported to the ICP is conqjlicated making normalization of data difficult Conditions such as the texture of the sanq>le, location of the sample in the laser cell, surface topography, laser energy, and other hictors affect e amount of material diat is introduced to the ICP torch and thus the intensity of die signal monitored for the various atomic masses of interest In addition, instrumental drift affects count rates. With liquid sanqiles internal standards typically are used to counteract instrument drift, but this approach is not feasible when material for the analysis is ablated from an intact solid sanqile. If one or more elements can be determined by another analytic technique, dien these can serve as internal standards. In the case of rhyolitic obsidian, which has relatively consistent silicon concentrations (ca. 36%), we have determined that silicon count rates can be normalized to a common value. Likewise, standards are normalized to their known silicon concentrations. This value, divided by the actual number of counts produces a normalization factor ftom i ch all the odier elements in that san le can be multiplied. A regression of blank-subtracted normalized counts to known elemental concentrations in the standards yields a calibration equation that can be used to calculate elemental concentrations in the samples analyzed. 

In the Mediterranean Sea and Middle East area, for example, there are obsidian outflows only in Italy, in some islands in the Aegean Sea, and in Turkey. Artifacts made of obsidian, however, are widely distributed over much of this vast area. Chemical analysis of many of these artifacts has shown that most of the obsidian used to make them originated in one or another of the outflows mentioned, but also in far-distant places such as Armenia and Iran. Plotting on a graph the concentration of selected elements in samples from obsidian sources against that in samples from sites where it was used, enables the identification of the source of the samples (see Fig. 22). Moreover, this type of analysis also makes it possible to trace the routes through which obsidian (and most probably other goods) were traded in antiquity (Renfrew and Dixon 1976). 

Both the Na and K intensities in the K-feldspar profile of Figure 4 are stable with depth indicating a previously documented lack of alkali mobility in the surface layers of feldspars at low temperature (7). In contrast, K increases and Na decreases with depth beneath the obsidian surface demonstrating substantial elemental mobility. The K loss near the surface corresponds to a concentration increase measured in aqueous solution. Sodium profiles in obsidian should exhibit even greater near-surface losses relative to K based on profiles measured by HF leaching (3) and sputter-induced optical emission studies (6). 

There is a large and growing literature on Middle Eastern obsidian studies that contains elemental data from the analysis of many hundreds of samples. Some effort is needed to reconcile the disparate terminology used to classify sources of both known and unknown provenance. The task is made difficult by the use of different analytical techniques of varying precision, the selection of different element sets in analysis and reporting, and the use of different secondary standards to quantify the concentration data. 

For each sample ten trace elements (Fe, Ti, Ba, Ca, K, Mn, Rb, Sr, Y, Zr) were determined in a concentration range of 40 - 1000 ppm. Each obsidian sample is therefore represented by a point in a 10-dimensional space. 

A similar technique of data analysis but without sophisticated pattern recognition methods was applied by Bird et. al. C19D. The concentrations of three elements (F, Na, Al) were determined by proton induced gamma-ray emission in 700 obsidian artefacts from 20 sources. Two-dimensional plots Al versus Na and F versus Na" showed distinct clusters-... 

NRA technique has been used to characterize the obsidian samples from different mineral sites in Mexico by Murillo et al. (1998) who have determined the oxygen concentration by means of the 0(d, p) 0 reaction. To determine the structure and composition of the patina through the depth profiles of the constituent elements like C, N, O, the (d, p) nuclear reactions have been employed using the external beam NRA measurements on copper alloys of archaelogical significance with 3MeV protons and 2MeV deuterons by loannidou et al. (2000). 

---