Obsidian and Ancient Manufactured Glasses

By Ioannis Liritzis

This edited volume offers archaeologists and archaeometrists the latest technical information, the fundamentals of provenance studies, instrumentation used in these investigations, and strategies for the dating and interpretation of archaeological materials in glass studies. The contributors discuss recent advances in obsidian hydration dating, secondary ion mass spectrometry, and infrared photoacoustic spectroscopy, focusing on the application of these technologies to a variety of glass forms and incorporating studies that look at the social and economic strategies of past cultures.

With examples from Greece, the Middle East, Italy, Peru, Bolivia, Russia, Africa, and the Pacific region, provenance studies look at regional patterns of glass acquisition, production, and exchange, providing examples that use one or more instrumental methods to characterize materials from ancient societies.

Extensive figures and tables included.

• Language: English
• Publisher: University of New Mexico Press
• Release date: Apr 16, 2012
• ISBN: 9780826351616

Book preview

Preface

The impetus for this book came from our desire to produce an up-to-date and informative volume on current obsidian and man-made glass studies that will encourage further work in this field. To generate a wider appreciation of the project’s aim, the editors hosted a specialized international workshop at the European Cultural Center in the ancient Greek town of Delphi, Greece, in February 2008. At this venue, the invited authors presented first drafts of their manuscripts to an audience of their peers for comment and evaluation. Several additional contributions were also solicited directly from individuals who were unable to attend the workshop but had recently published new and relevant contributions. In addition, we are especially grateful to Dr. Masao Suzuki, whose Obsidian Summit meeting held during 2004 in Tokyo provided a venue and generous support from Rikkyo University for all those contributors who presented papers. The international Obsidian Summit provided a very useful stimulus for continuing research in obsidian studies leading to the Delphi meeting and the proceedings reported in this volume. Some of the summit’s unpublished studies now appear in this volume after the authors revised and updated their contributions.

Editors’ Introduction

About every 10 years it becomes necessary to prepare an edited volume by experts in the field of glass studies that reflects new developments and applications in absolute dating and provenance. The pace of analytical advances and instrument development is rapid, and now more than ever they are applied within the historical sciences. Dating and provenance are two topics often combined within a single publication since provenance studies are highly dependant upon age estimates if diachronic models of human behavior are desired. For this reason, we have again placed them together under the conventional format.

This effort continues the tradition of book publication that periodically updates archaeologists interested in glass studies. This tradition started in 1976 with the publication of Advances in Obsidian Glass Studies (R.E. Taylor, editor, Noyes Press) and was followed in 1998 by the most recent publication, entitled Archaeological Obsidian Studies: Method and Theory (M.S. Shackley, editor, Plenum Press). These books have been, and continue to be, highly referenced even 30 years later. This fact is supported by an important point made by M.S. Shackley in his volume, that in the last twenty years, the use of obsidian archaeometry has witnessed almost boundless expansion, partly due to the concomitant advances in computer technology and instrumental chemistry, and partly due to the recognition by archaeologists that archaeometrists provide much more than mere measurement (1998:2). As a result, we believe that the volume will continue to be of interest to a wide number of scholars in the years to come.

Archaeometric studies of natural and ancient manufactured glasses are conducted throughout the world as a means of looking at chronometric age and provenance and are widely applied more than ever before. Both of these topics are central aspects of archaeological investigations into past social systems. In this volume, the papers concerned with chronology look at recent advances pertinent to the method of obsidian hydration dating (OHD). Traditional approaches to this dating method using optical microscopy have become less favored and are being replaced by precise surface characterization techniques such as secondary ion mass spectrometry (SIMS) and infrared photoacoustic spectroscopy (IR-PAS). Archaeologists may be unaware of these developments, or intimidated by a presumed complexity of the science, such that applications of the new techniques are limited. In this volume we will detail the approaches and their application in a manner understandable and interesting to the archaeologist.

In the same vein, we look at the technology behind artifact provenance studies that includes instrumental neutron activation analysis (INAA), laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS), and X-ray fluorescence (XRF). All of these techniques are well-established technologies, and we will generally focus not on the instrumentation but on methods of application to a variety of glass forms. We have included papers that go well beyond descriptive applications of glass composition and provenance and incorporate studies that look at the social and economic strategies of past cultures.

This book contains 15 individual chapters by scholars from many parts of the globe. The areas of study include case examples from Greece, the Middle East, Italy, Peru, Bolivia, Russia, Africa, and the Pacific region. The book has been subdivided into two major sections. Part 1 discusses the analytical methods associated with current applications of OHD. It begins with a close look at the instrumentation used for SIMS analysis of glass surfaces. This technology has become the favored technique for acquiring hydrogen depth profiles on glass surfaces since it avoids the problematic issue of operator judgment during optical measurement at high magnification. Yet there are a number of instrument configurations, each with its own advantages and drawbacks. Sample surface conditions can also impact the shape of the profile and are thus problematic in application to archaeological specimens. SIMS is a central aspect of one especially important contribution that includes the results of a 26-year-long weathering experiment at ambient temperature. It is currently unique to the discipline and will likely not be repeated for many years. The results of long-term weathering experiments have contributed to our understanding of how water diffuses into glass at ambient temperature. The observations acquired here support the development of a new dating method known as SIMS-SS (SIMS–saturated surface). Dating by SIMS-SS has not seen wide application, but a series of publications, including the one presented here, has clarified the theory and application of the method. The SIM-SS dating method is highly advantageous because the hydrogen profile is a proxy for the impinging parameters of ground temperature and soil relative humidity. SIMS analysis is not always affordable, and for more conventional applications of OHD that require numerical estimates of temperature, we have included a chapter on estimating regional ground temperatures from meteorological data. This first section finishes with a review paper that brings the readers up to date on obsidian studies on the African continent, a region of the world unfamiliar to many of the world’s archaeologists.

Part 2 contains 10 papers on provenance studies that look at regional patterns of glass acquisition, production, or exchange. Each contribution provides a case example that uses one or more instrumental methods to characterize materials of an ancient society. We have subdivided these papers into the study of natural glasses (e.g., basalt or obsidian) or man-made glasses. The papers on manufactured glasses refer to the provenance and technology of Roman glasses in the ancient world, a compositional and technological study from Kenyan glass vessels and beads (AD 900–1750), and production techniques and provenance of Rhodian (Aegean) and Thebean (mainland Greece) glass fragments (seventh–fourth century BC), the latter tackling the issue of unusual raw materials.

Provenance studies on manufactured glasses are no less complex than the characterization of regional obsidian or basalt source distributions. Human experimentation and raw materials have been combined into a single compositional product, and the task is to distinguish not only how the glasses were made but where they came from as well. In such cases additional tools are brought to bear on these situations and include the use of isotopic studies. The impact of regional source material selection on glass composition can provide clues about geographic location as can the elements imparted by the use of specific types of plants specific to areas or regions.

As the methods of glass characterization become well validated, fewer papers are concerned with the specifics of instrumentation. Similarly, as regions become intensively studied a focus on the development of geochemical databases for natural glasses loses its place as the primary objective. Rather, as we see here, multiple forms of instrumentation are combined to invent new analytical protocols such as the combined use of portable-XRF and the more reliable techniques such as instrumental neutron activation analysis (INAA). This confidence in the accuracy and precision of analytical methods that form large databases now allows for broad regional syntheses of how raw materials were acquired, transported, and exchanged. Many of the chapters in this volume illustrate how populations interacted over time and space and provide much-needed detail on the dynamics of human interaction. Regional interaction studies are generally the norm and have become increasingly challenging as larger data sets from well-controlled archaeological contexts are required. However, significant challenges remain to acquire a representative picture of this interaction, and these tasks include the appropriate sampling of assemblages to capture the full variability in the flow of material culture that may not be incorporated into artifacts of one type or size category. A continued attention to problems of provenance and chronology will serve to make diachronic interaction studies a central, and informative, undertaking associated with the study of past cultural systems.

To close the book, we present a chapter written by M. Steven Shackley, whose long-term experience and historical perspective provide some insight to the trajectory of glass studies in archaeology. His chapter deals with the value of natural and manufactured glass studies in third-millennium archaeology, and brings to this field his perspective on archaeological science and social archaeology. He critically examines the areas of obsidian dating and provenance, referring to various analytical techniques (XRF, INAA, PIXE-PIGE, LA-ICP-MS) employed by authors in the volume, and discusses the artificial glasses studies covered in the volume and their future prospects.

PART I

OBSIDIAN

HYDRATION DATING

Chapter 1

Aspects of Secondary Ion Mass Spectrometry (SIMS) Depth Profiling for Obsidian Hydration Dating

STEVEN W. NOVAK AND CHRISTOPHER M. STEVENSON

ABSTRACT

The SIMS technique provides an ideal tool for measuring hydrogen (H) profiles for use in obsidian hydration dating (OHD). The measurement of a natural obsidian fracture surface by SIMS is not straightforward, however, and subject to several factors. Measurements have been collected using both magnetic-sector and quadrupole-based SIMS instruments, with comparable results. The quadrupole-based instrument may provide a somewhat simpler analysis in terms of charge compensation and requiring less sample preparation. Although early measurements were made using negatively charged oxygen (O-) bombardment with positive ion detection, modern instruments using cesium (Cs) bombardment with negative ion detection should provide superior backgrounds for H analysis. Analysis of natural obsidian surfaces may be subject to problems due to surface roughness, cracking, or the presence of natural inclusions or vesicles. Measurements by atomic force microscopy (AFM) indicate root mean square (RMS) roughness values as low as 5 nanometers (nm) for conchoidal fracture surfaces. Rippling that develops within the SIMS craters due to ion bombardment causes RMS roughness of about 40–50 nm. These values are similar to the rate of fall of the H profile for deep profiles. Sputtering rates measured for a suite of obsidians from archaeological sites show a variation of less than 4%, indicating that an external obsidian standard can be used to calibrate the sputtering rate within the typical measurement error. For the hydration layer thickness measurement, the important parameter is the thickness of the hydration layer, as measured at the ½ fall of the H profile. Repeat measurements indicate this thickness can be measured to within ±1% on some samples. Excellent correlation between SIMS and IR measurements indicate the ability of SIMS to provide highly quantitative measurement of H in obsidian.

1. INTRODUCTION

Over five decades ago Friedman and Smith (1960) noted that freshly exposed surfaces of obsidian artifacts begin to absorb water immediately, eventually forming a water-rich hydrated layer in the range of 1–10 micrometers (μm). If the ground temperature, soil relative humidity, and rate of water diffusion in obsidian are accurately known, an age may be calculated for the creation of the artifact from a measurement of the hydrated layer thickness (Friedman and Long 1976). The measurement of obsidian hydration rims for the purpose of dating has been undertaken for many years now. Initial efforts used physical cross-sections and optical measurement, but more recently a number of advanced analytical methods have been used including nuclear reaction analysis, sputter-induced optical emission, and secondary ion mass spectrometry (SIMS). Among these techniques, SIMS should provide the most accurate and detailed measurement of the diffusion profile shape, due to the nature of the profiling technique. In this chapter we will summarize work done to date on SIMS profiling of archaeological obsidian samples, discuss some aspects of the technique that may affect the accuracy of the measured profiles, and present some new data on sputtering rates of obsidian and the roughness of fracture surfaces by atomic force microscopy (AFM) (see also Liritzis and Laskaris 2011; Liritzis et al. 2008a, 2008b). We hope to give readers a means to understand what is required for a SIMS analysis of a sample and the parameters that affect the accuracy of the measurements. In addition to the measurement of artifacts, we have performed a number of laboratory hydration experiments on obsidian to better define three parameters: (1) What is the effect of glass density and composition on the calibration used to estimate crater depth? (2) What is the effect of surface roughness on the precision of SIMS measurements? and (3) Can the calibration between infrared photoacoustic spectroscopy (IR-PAS) absorbance measurements and depth be improved?

2. PREVIOUS WORK

Initial development of the obsidian hydration dating method used cross-sectional polishing of the glass and optical width measurement of the hydrated rim. For relatively thick hydration layers present on very old artifacts, this method at times produced good linear correlation between layer thickness and hydration age (e.g., Meighan et al. 1968; Hull 2001). The first modern surface analysis technique applied to measuring hydration rim thickness was by Lee et al. (1974) and Lanford (1977), who measured the hydrogen profiles using nuclear reaction analysis. This method showed good correlation between hydration layer thickness and hydration temperature and provided quantitative H values. Curiously, this high-quality technique was not widely applied to the measurement of hydrated obsidians, possibly due to the lack of high-energy accelerators required for the nuclear reaction measurement. In addition, the technique is useful only within about 2 microns of the surface, restricting its use for younger artifacts in which the hydrated layer is thinner than 2 microns. At nearly the same time, Tsong et al. (1978) demonstrated the use of sputter-induced optical emission for obsidian hydration profiling. This technique is similar to SIMS in that it uses an ion beam to sputter the sample; however, the detected signal is the light emitted during the bombardment process. In the years since its inception in the late 1970s, SIMS has become a highly developed measurement technique, particularly in the semiconductor industry, and it is widely applied to problems in metallurgy, ceramics, geology, and mineralogy. As pointed out by Tsong et al. (1978), the ability to sputter the sample with a high-current ion beam yields the ability to measure hydration profiles to 10 microns in a reasonable amount of time, making SIMS a versatile technique.

Because of the ease of cutting and polishing obsidian and measuring the hydration rims optically, a large number of obsidian hydration measurements have been performed in the 1970s through the present day (Jackson 1984; Meighan et al. 1974; Meighan and Vanderhoeven 1978; Meighan and Russell 1981; Meighan and Scalise 1988; Stevenson and Ayres 2000; Hull 2001). Although this method seemed to give good absolute ages for some studies (Stevenson 2000), practitioners of the measurement point to significant errors in other cases (Ridings 1996). The major problems seemed to be the subjective nature of determining the thickness of the hydration layer by eye, the low resolution of the optical microscope (Scheetz and Stevenson 1988), and the correlation between the optical image and the actual depth of water diffusion (Anovitz et al. 1999; Stevenson et al. 1989, 2002; Tokoyama et al. 2008). As these problems were being discovered, newer techniques such as infrared spectroscopy were being developed to measure the water content of hydrated obsidian (Stevenson and Novak 2011). Infrared spectroscopy has been used to measure the bulk water content of obsidian (Newman et al. 1986; Zhang et al. 1996) with the photoacoustic method (IR-PAS) used to quantify the depth of surface water diffusion (Stevenson et al. 2001). The latter approach to hydration layer depth assessment has been calibrated relative to SIMS and shown to have a high precision estimated at 0.05 μm.

3. SIMS MEASUREMENTS

SIMS has been used to measure trace element contents, including H, in minerals and glasses almost since the inception of the technique (Anderson and Hinthorne 1973; Delaney and Karsten 1981; Honig 1995), and there has been a significant proliferation of papers using SIMS in geochemistry in the last 15 years as the technique has become more accepted. Detailed descriptions of the SIMS technique can be found in Benninghoven et al. (1987) and Wilson et al. (1989). A review of SIMS as applied in geochemistry is given by Ireland (1995), and Compston and Clement (2006) detail the most recent developments in the highly developed SIMS instruments used for isotope ratio measurements. Because most instruments previously used for geochemical analysis are magnetic-sector instruments, and because we have used a quadrupole-based instrument, some detailed description of each technique will be given here.

There are basically two types of SIMS instruments in common use, those with magnetic-sector mass spectrometers (figure 1.1) and those with quadrupole-based mass spectrometers (figure 1.2). The major difference lies in the design of the mass spectrometer used to filter the secondary ion signals. Magnetic-sector instruments use a curved ion flight path through a uniform magnetic field to mass-resolve molecular ion signals. The design allows for high mass resolution to eliminate overlaps from interfering molecular ions, which are not important for analysis of H. Quadrupole-based instruments have lower extraction efficiencies for secondary ions, which make them less sensitive, and they have only unit mass resolution. Quadrupole-based instruments have not intrinsically been designed for different purposes; however, their design makes them easier to use for analysis of insulating materials, like glasses, and they have the ability to bombard the sample with lower primary ion beam energies. For the purposes of measuring H profiles in obsidian, the ability to analyze insulating samples is important, but other differences between the two types of instrument are not as important. Several previous studies have been published on measuring hydration of archaeological materials using a magnetic-sector instrument (e.g., Anovitz et al. 1999, 2004). Studies using quadrupole-based instrument include Stevenson et al. (2001, 2004). Recent studies of glass hydration using SIMS include Dowsett (2004) and Fearn et al. (2006). An extensive study measuring water contents of geological samples has been published by Hauri et al. (2002) using a modern CAMECA magnetic-sector instrument.

FIGURE 1.1. Diagram of a magnetic sector (CAMECA) SIMS instrument.

FIGURE 1.2. Diagram of a quadrupole-based SIMS instrument (Phi).

The SIMS technique consists simply of bombarding a solid with a focused ion beam within a vacuum chamber and detecting the ions emitted from the surface using a mass spectrometer. Most early analytical work on minerals and glasses used magnetic-sector CAMECA instruments. In order to minimize charge buildup on the sample, these analyses used a negatively charged oxygen beam and the sample was coated with a conductive layer, typically gold. More recent CAMECA designs allow the use of a Cs beam; however, a gold coat is still necessary. Analyses for H in quadrupole instruments typically use a Cs beam, and no conductive coating is needed. The primary ion beam having a current of several nA (nanoamperes) to 100’s of nA is focused to a beam size of microns to 10’s of microns. Usually this beam is raster-scanned across the surface of the sample to form a flat-bottomed crater, but for microanalysis the beam may be unrastered. For any depth profiling application, such as the hydrogen profiles acquired on hydrated obsidians, the beam must be raster-scanned, and secondary ions accepted into the mass spectrometer must be limited to those emitted from the central part of the crater. In magnetic-sector instruments, the ion acceptance is limited by apertures within the mass spectrometer. In quadrupole-based instruments, ion acceptance is done by electronic gating, in which ions are only counted by the mass spectrometer when the beam is in the central part of the rastered area. Typically less than 40% of the rastered area is used as the acceptance area for analysis.

One significant difference between the two types of SIMS instruments that may affect the ability to reproducibly analyze archaeological samples has to do with the design of the extraction lens and field in front of the sample. Magnetic-sector instruments have the extraction lens spaced 5 mm from the sample surface and a voltage of 4500 V is used to extract secondary ions into the mass spectrometer. It is this high voltage gradient that makes it difficult to compensate for charge buildup on insulating samples in a magnetic-sector instrument. In contrast, the quadrupole-based instrument may have the extraction lens several millimeters from the sample and an extraction potential of several hundred to 1000 V. The surface of archaeological obsidian samples, although generally smooth, is typically curved. The strong field gradient in magnetic-sector instruments is greatly affected by the tilt and shape of the analysis surface, and small differences in sample positioning may greatly affect the analysis (Lux 1990; Chi and Simons 1990). In quadrupole-based instruments the sample stage can be tilted to aid in sample positioning, and the lower field gradient means that there is less effect due to curved sample surfaces.

One additional difference between the two types of instruments germane to analysis of archaeological samples is that the sample stage in the magnetic-sector instrument is fixed in place and will only allow samples less than 25 mm in diameter to be held within the sample holder. In the quadrupole-based instrument the sample stage is movable in all three dimensions and can be tilted, allowing curved samples to be readily aligned beneath the beam. There is much more freedom in sample size due to this arrangement and samples much larger than 25 mm can be introduced without resorting to cutting smaller pieces from them.

Analysis of insulating samples is much more difficult than analysis of conductive samples in SIMS because the sample is bombarded with a charged ion beam (Werner and Morgan 1976; Wittmaack 1979). Because insulating samples do not conduct away the charge, it builds up in the area of ion bombardment. This buildup changes the energy of ions leaving the surface, which strongly affects ion transmission through the mass spectrometer. Severe charge buildup will deflect the primary ion beam, preventing analysis of the sample. Compensating for this charge buildup is usually done by coating the sample with a conductive layer, as is often done for SEM imaging, and by bombarding the sample with a secondary electron beam. For analysis with magnetic-sector instruments, it is generally necessary to coat the sample with a conductive layer, a procedure that is perhaps undesirable for valuable archaeological artifacts. Ion bombardment may be with negatively charged oxygen (Anovitz et al. 1999) or, more recently with Cs bombardment (Hauri et al. 2002; Stevenson et al. 2001). For quadrupole-based instruments it is much more straightforward to compensate for charge buildup with only electron bombardment, eliminating the need to coat the samples with a conductive layer.

Early comments on factors important for analysis of H by SIMS include Magee and Botnick (1981), who note the importance of a low chamber vacuum to reduce the H background. Analyses of H (or water) in glasses and methods for SIMS analysis have been published by Delaney and Karsten (1981), Hervig and Williams (1988), and Ottolini et al. (1995). These authors used magnetic-sector instruments with O- bombardment and positive ion detection. Optimal analyses are achieved by using high-energy secondary ions in order to filter abundant positive H ions from electron beam bombardment, which may cause a high H background. There have been continuous improvements in the instruments and methods over the succeeding years. More recently Hauri et al. (2002) describe a method of H analysis using Cs bombardment with negative ion detection in a CAMECA magnetic-sector instrument. Improved analytical conditions and instruments have allowed routine measurements of H in minerals and glasses to as low as 4 parts per million (ppm) (Koga et al. 2003; Rhede and Wiedenbeck 2006). These studies have also documented the improved backgrounds achieved with Cs bombardment when contrasted with O- bombardment and have shown that high electron beam energies result in high backgrounds due to electron-stimulated desorption of H. Although these effects should not affect the accuracy of hydration profiles in obsidian or glass, the very high H contents of hydrated obsidian surfaces may cause high backgrounds if the electron gun parameters are not optimized.

4. QUADRUPOLE SIMS DEPTH PROFILES

We have performed several studies of archaeological artifacts and laboratory hydrated glasses using quadrupole-based SIMS (Stevenson et al. 2001; Liritzis et al. 2004; Ambrose et al. 2004). These dynamic SIMS measurements were carried out using Phi 6300 and 6600 quadrupole-based SIMS instruments. Samples are held by clips and screws to minimize damage, but may also be held with silver paste. Samples held with silver paste are typically dried in an oven for 30–60 minutes at 60–80 °C. No gold or other conductive coating is applied to the samples, in contrast to methods published for magnetic-sector SIMS instruments (Anovitz et al. 1999; Hauri et al. 2002; Riciputi et al. 2002).

Profiles are acquired using Cs bombardment, typically with a 5-kiloelectron-volts (keV) impact energy and an incidence angle of 60° relative to the surface normal. Ion beam current is typically 500–800 nA. The system has a typical base pressure of less than 7 x 10-10 torr and an operating pressure of about 2 x 10⁹ torr. The system typically achieves operating pressure within about 15 minutes of sample introduction. Because the sample stage can be adjusted in X, Y, Z, and tilt, it is relatively easy to reproducibly position a reflective surface in the analysis position (figure 1.3).

Typical SIMS depth profile results for H profiles in obsidian artifacts are given in figure 1.4 (Stevenson et al. 2004). These profiles show similar profile shapes in having a thin surface layer (up to 200 nm) in which the H content falls sharply, followed by a more gradually sloped profile that may extend many microns into the sample depending on the age of the artifact. Finally the H profile falls sharply at some depth that is also dependant on the age of the artifact. The surface peak may, at least partially, result from the adsorbed hydrocarbon layer detected on every air-exposed surface, commonly referred to as the adventitious carbon layer. However even smooth, polished, laboratory-hydrated obsidian samples show this surface peak, suggesting it is a fundamental feature of water diffusion into obsidian. Figure 1.4 also shows hydration layer thicknesses estimated by standard optical measurements. Note that the optically determined thicknesses are always less than the depth at which the H profiles fall, indicating that the optical technique consistently underestimates the true hydration layer thickness.

5. CRATER DEPTH CALIBRATION AND SPUTTERING RATE DIFFERENCES

Conversion of sputtering time to depth in SIMS profiles is typically done by measuring the analytical crater depths using a stylus profilometer. Repeat measurements of calibration standards have demonstrated that the stylus profilometer itself is precise to within 1%. A round-robin study of profilometry measurements of SIMS craters in flat (i.e., polished semiconductor) samples shows that this procedure is accurate to within 5% (Simons 1997). For hydrated obsidians, repeat measurements of polished samples have shown an error of 3% for flat samples (Riciputi et al. 2002). However, this latter procedure requires cutting and polishing a piece of material to ensure accurate measurements. Direct profilometry measurements can be made of craters on artifact surfaces. However, roughness and tilt of typical artifact surfaces commonly have larger errors than flat, polished surfaces.

FIGURE 1.3. Photomicrograph of SIMS craters on an oriented obsidian surface. The craters are approximately 225 microns on a side and about 7 microns deep.

FIGURE 1.4. Quadrupole SIMS depth profile measurements of hydrated obsidians from a variety of sources. Arrows show hydration layer thicknesses measured optically (Stevenson et al. 2004).

In our studies we have chosen to calibrate the concentration and depth axes by using a deuterium-implanted high-silica nonhydrated obsidian (KSW-354) (Novak and Mahood 1986) as a standard material. The use of ion-implanted standard materials is common for quantitative SIMS analyses in the semiconductor industry (Wilson et al. 1989). Calibration of the depth scale simply involves using the known peak depth of the implant, previously established by stylus profilometry, to calculate the sputtering rate for a given analytical run. The standard is analyzed in each analytical run so as to correct for small variations in instrument conditions from run to run. The advantages of using this method include simplicity, avoiding cutting the sample, and good reproducibility. Using this method for depth calibration avoids errors that may be introduced by stylus profilometer measurements of rough artifact surfaces.

In order to assess the variation in sputtering rate for different glasses, we depth profiled polished samples of 6 glasses from widely distributed obsidian geological sources in a single analytical run. The results are given in table 1.1 and show that these glasses have a 1 sigma relative standard deviation (RSD) of 3.6% of the average rate for all glasses. Note that these glasses have OH+H2O contents that range from 0.08% to 1.54%, suggesting that this range of water contents has little effect on sputtering rate. The RSD of 3.6% is comparable to the error in stylus profilometer measurements alone. While this is only a small sample of natural glasses used for artifacts, the results suggest that use of an ion-implanted standard for calibration of the depth scale does not introduce any larger error into the depth axis than other calibration methods and may actually avoid errors introduced by direct measurements of craters on irregular archaeological surfaces.

TABLE 1.1. SIMS SPUTTERING RATES ON OBSIDIANS OF DIFFERENT STRUCTURAL WATER CONTENT

6. INFRARED PHOTOACOUSTIC DEPTH MEASUREMENT

Our initial study utilizing SIMS profiling was undertaken to compare hydration layer thicknesses as measured by the traditional optical method, photoacoustic infrared spectroscopy (IR-PAS), and SIMS depth profiling (Stevenson et al. 2001, 2002). The IR-PAS signal should quantitatively measure the total amount of hydrogen incorporated into the sample, as will the integrated H signal measured by SIMS. This study showed that hydrated layer thicknesses measured using the traditional optical method systematically underestimate the H diffusion depth when compared to SIMS (figure 1.4). We propose that the underestimation derives from a sharp decrease in the stress-induced birefringence of the glass that prevents the full extent of the hydrated layer from being optically visible. A similar explanation was proposed by Riciputi et al. (2002).

Systematically conducted hydration experiments support this explanation for optical errors. An example of such a study is given in figure 1.5 (Ambrose et al. 2004). In this experiment fresh obsidian flakes were hydrated at temperatures between 10 °C and 40 °C for 26 years. The high depth resolution of SIMS allows even the minimal diffusion of water at 10 °C to be measured accurately. Note that the slope of the hydration front becomes less steep with increasing depth and temperature. If these hydration layers were measured optically, it seems likely that the differing