Anna M. Semon1, Lindsay Bloch1, Mary Elizabeth Fitts1, and Rosanna Crow2 Ceramic Chemical Characterization Studies in the Southeast INTRODUCTION METHODS XRD The technique of X-ray diffraction (XRD) uses x-rays of a known wavelength to determine the lattice spacing of crystalline structures present in a sample. The resulting measurements are used to identify mineral compounds and their relative proportions. Compositional analysis of ceramics relies upon the principle that clay beds have elemental signatures specific to their parent material, and that these signatures are retained within the ceramic paste. There are many new and increasingly available techniques of ceramic chemical characterization that may be used to capture this patterned elemental variation. For archaeologists, this presents the possibility of attributing pottery to local or non-local clay sources. In this pilot study, the authors compared the results of XRD, ICP-MS, XRF, and pXRF techniques on three types of earthenware ceramic found in the southeastern US, encompassing different regions, time periods, and ceramic technology. The goal of each case study was to identify compositional groupings that represent discrete geographic origins for the ceramics, while also accounting the potential methodological or analytical considerations that may change according to the ware type and depositional history of the artifacts. This project developed from a growing interest in ceramic chemical characterization as a way of identifying compositional groupings for clays and archaeological ceramics recovered from our respective project areas. We tested the four methods through three case studies, comparing the limitations and advantages of each technique. Here are some preliminary conclusions: The technique of wavelength-dispersive X-ray fluorescence (WD-XRF) uses X-rays of known wavelength and intensity to excite the electrons of specific elements. As the electrons move back to a stable condition, they emit fluorescent (secondary) X-rays that are measured by a detector. Software is used to convert this information to parts per million (ppm), kilocounts per second (kcps), and mass% of elements and oxides. oOxides, major and minor elements were important for distinguishing samples in predictable ways: This study used a Rigaku Supermini wavelength-dispersive XRF spectrometer. Samples of at least seven grams were pulverized in a shatterbox and mixed with a paraffin binder at a ratio of 1 to 100 (binder to sample). The samples were machine-pressed into a 30mm-diameter pellet using a Carver 25-ton press for three minutes. The AGV-2 (Andesite) and MAG-1 (Marine Mud) standards were used for data calibration, and each sample was analyzed three times. Figure 1.4. Preparation of XRF pellet. Na2O, CaO, and K2O separated Catawba Valley samples according to feldspar composition In historic sherds, major elements such as Pb may represent production or post-depostional processes rather than clay source Rb and Zr differentiated Piedmont samples according to underlying geology Figure 1.2. Cutting sherds with trim saw. oXRD analyses identified mineralogical groups within the tested clays, information which was useful for structuring the analysis of elemental data pXRF ICP-MS Portable XRF (pXRF) technology is a newer method for archaeological sourcing studies. In this handheld device, an x-ray beam is directed at a sample. The beam causes atoms in the sample to fluoresce and emit both a broad and line spectrum that are characteristic of the atoms in the material. For this project we used a Bruker ASX Tracer III-V Portable EDXRF machine, containing a 40kV Rhodium anode X-ray tube with a PIN diode detector. The PIN diode detector was operated at 40 kV and 11.5 microamps using a filter composed of 6mil copper, 1 mil titanium, and 12 mil aluminum, to help look at the rare earth elements. Ceramic samples used in the three studies were larger than the beam aperture and each sample was tested in three different locations for 180 seconds per zap in order to test heterogeneity of the sample. of North Carolina – Chapel Hill and 2SIU - Carbondale DISCUSSION XRF We tested all samples using at least three of the following methods. Each method provided either mineralogical or elemental characterization of the ceramics or clays. All analyses were conducted by the authors at the University of North Carolina at Chapel Hill. Here, we briefly review the techniques and methodologies followed in this study. The results were evaluated using exploratory data analysis. This project used a Philips PW 1720 diffractometer, which emits Cu Ka radiation and has a graphite monochromator. Samples were pulverized in a mortar, and then a small amount of powder tapped onto double-sided tape attached to a glass microscope slide. Diffracted patterns were recorded between two theta angles of 2.5 and 70° with a scanning step of 0.02° and a measuring time of two seconds per point. 1University Figure 1.5. XRF pellets awaiting analysis. oXRF and ICP-MS techniques had high sensitivity and provided quantitative data not available through other methods; however, both techniques are time consuming and destructive Inductively coupled plasma-mass spectrometry (ICP-MS) is a technique that provides accurate and precise elemental concentrations for most isotopes. In solution ICP-MS, a small portion of powdered sample is digested in acid, then fed through a plasma torch that ionizes the solution. The sample then passes into an attached mass spectrometer, where it is separated according to atomic mass and charge. CONCLUSION For this project, we used a Varian 820-MS instrument. The samples were digested in an HNO3 solution for approximately 48 hours, then diluted to 2% HNO3 before being analyzed. Each was sampled for 3.5 minutes. Internal standards for each element, in solution form, were used to calibrate and reduce instrument drift. The resulting elemental measurements were calculated as parts-per-billion (ppb). Figure 1.3. Operating the pXRF. Figure 1.6. ICP-MS instrument. The goal of this pilot study was to determine compositional groupings that represent geographically discrete clays and archaeological ceramics, using multiple lines of evidence. The initial results are positive and indicate potential groupings. The authors intend to follow up on these findings to generate comparable and statistically rigorous data sets by increasing the number of samples from each case study. Additionally, further investigations will use published standards and increase the number of assays for every technique to address paste heterogeneity. Figure 1.1 Map of the overall project area, with the location of each case study indicated. Sr -0.4 -0.2 0.0 0.2 [5], d=3.3392(3), 2-theta=26 20 30 20 30 40 50 60 0.5 -0.5 0.0 0.5 Figure 2.4. PCA biplot of ICP-MS data (log-base 10 ppb) using Zr, Fe, Y, Nb, Mn, Ti, Sr. Results suggest site based differences in elemental concentration, but the presence of outliers indicates that more samples are needed to clarify this pattern. Rb Zn Y Nb 15 - keV - Figure 3.1 pXRF spectra of a Nassaw –Weyapee clay sample, and a sherd from the same site. In the PCA biplot (Figure 3.2), the clay and sherd separate along Axis 2. Table 3.1 XRD Results, Number of Clay Samples from Each Site with Identified Mineral Component. Figure 2.6. pXRF spectra and net area intensities for ceramic Figure 2.5. PCA biplot of pXRF data (log-base samples. The variation with Zr and Cu peaks indicate 10 net intensities) using Zr, Fe, Y, Nb, Rb, Ti, potentially meaningful variation. Sr, Zn, Cu. No significant patterning is apparent. Figure 2.7. A closer comparison of the net intensities for Cu indicates clear site based variation. 40 50 60 70 70 Figure 2.8. XRD spectra peaks show the various minerals within each sample, which include a mixture of clay minerals: kaolinite, illite, and smectite. There are subtle differences between the two spectra, but it appears that the South End sample have more illite. Conclusions: These analyses suggest the 2 clay samples tested are chemically and mineralogically different. As for the sherds, ICP-MS and XRF provide quantitative data that indicate compositional groups within the dataset. While pXRF does not indicate such groups, it produces interesting elemental variation, such as in Cu intensity, that should be investigated further by increasing the sample size. Results: X-ray diffraction identified a similar set of minerals in the Nassaw-Weyapee samples as previously had been found in Catawba clays (Table 3.1). Both plagioclase and K-feldspar are present, with albite being more common overall. The XRD results are useful for interpreting the principal components analysis of element and oxide data from X-ray fluorescence (Figure 3.2). Figure 3.3 Geology of the lower Catawba River Valley, USGS. Figure 3.2. PCA biplot of XRF data (log-base 10 ppm) for sherds and clays. Axis one differentiates samples containing more Kfeldspar and mica, which contain K2O, from those containing more plagioclase, which contains CaO, Na2O, and Sr. (Faure 2001). Axis two separates samples with higher Zr, Ti, and Na from those with Rb, Mg, and Fe. Conclusion: The characterization results can be explained with reference to the underlying geology of the lower Catawba River Valley (Figure 3.3). The Nassaw-Weyapee area contains mostly mafic and intermediate rock, while the later sites are situated on phyllonite mica with adjacent felsic deposits. The combined characterization data suggest that Catawba potters did not return to the clay pits around their former settlements after the small pox epidemic, focusing instead on Nisbet Bottoms and other nearby deposits. 0.8 0.6 Cu Ca 5 Zn66 0.0 0.5 1.0 1.5 Figure 4.1. Map of project area. Production Period Early 19th C. Henry Loy/J. Albright Alamance County Late 18th/early 19th C. Solomon Loy Alamance County Early 19th C. Thomas Dennis Randolph County 1812-1821 William Dennis Randolph County 1790-1832 1.5 1.6 0.4 0.2 Zr Na2O Rb K2O -0.6 Pb208 Axis 1 (74.9% variance) Location Person County 20 Figure 4.2. PXRF spectral comparison of Solomon Loy sherd (blue) and Joseph Loy sherd (red). Joseph Loy sample has higher peaks for iron, copper, and lead; while Solomon Loy sherd is higher in calcium and strontium. Ca44 Sr88 Site Name Joseph Loy 15 - keV - Fe57 -0.5 Sr 10 Zr90 -1.0 Pb 0 MnO 0.0 Sr TiO2 Fe2O3 -0.2 100 Joseph Loy Loy/Albright Solomon Loy Thomas Dennis William Dennis -1.5 Site Joseph Loy Loy/Albright Solomon Loy Thomas Dennis William Dennis -0.4 200 Site 0.0 10 300 0.5 0.5 0.0 5 2-theta (deg) 10 -0.5 Axis 1 (48.0% variance) [9], d=1.8154(3), 2-theta=50.214(10) [8], d=2.282(4), 2-theta=39.46(6) [7], d=2.53(5), 2-theta=35.4(7) 0 10 0.0e+000 0.2 Cu Relic Marsh [3], d=4.420(13), 2-theta=20.07(6) [4], d=4.22(3), 2-theta=21.05(13) [1], d=7.540(10), 2-theta=11.727(15) [2], d=7.20(3), 2-theta=12.29(6) 1000 0.4 Initial research found archaeologically-recovered clays to be chemically consistent with eighteenth-century Catawba ceramics (Crow 2011). Clay from a source used by modern Catawba potters also was found to be consistent with eighteenth-century materials. This study sampled three clays and eight sherds from the Catawba site Nassaw-Weyapee (ca. 1750-1759) to examine potential effects of community relocation on raw material use. Meas. data:SCI-RC-2/Data 1 Calc. data:SCI-RC-2/Data 1 Intensity (cps) [17], d=1.5399(2), 2-theta=60.028(10) [19],[20], d=1.3819(3), 2-theta=67.751(17) d=1.37332(12), 2-theta=68.234(7) [18], d=1.491(4), 2-theta=62.23(18) [15], d=1.8143(3), 2-theta=50.245(10) [16], d=1.6712(5), 2-theta=54.891(16) [13], d=2.1272(12), 2-theta=42.46(2) [14], d=1.9806(10), 2-theta=45.77(2) [11], d=2.2784(14), 2-theta=39.52(3) [12], d=2.2357(7), 2-theta=40.306(14) 2000 Zr Ti Ti Axis 1 (56.5% variance) [6], d=2.8130(17), 2-theta=31.784(19) South End [9], d=2.571(6), 2-theta=34.87(8) [10], d=2.4545(14), 2-theta=36.58(2) [3], d=4.468(10), 2-theta=19.85(4) [4], d=4.248(2), 2-theta=20.895(10) 2.0e+003 [2], d=7.11(3), 2-theta=12.43(5) 4.0e+003 [1], d=13.8(3), 2-theta=6.42(13) Intensity (cps) 6.0e+003 [5], d=3.565(14), 2-theta=24.96(10) [6], d=3.3855(9), 2-theta=26.302(7) [8], d=3.287(6), 2-theta=27.10(5) [7], d=3.3357(4), 2-theta=2 8.0e+003 Error Residual Meas. data:scirc1/Data 1 BG data:scirc1/Data 1 Calc. data:scirc1/Data 1 0.6 Zn -1.0 Axis 1 (67.9% variance) 1.0e+004 Fe Rb 0.4 Figure 2.3. PCA biplot of XRF data (log-base 10 ppm) using Ba, Ni, Rb, Sr, Y, Zr. PCA indicates samples group by site, based on general elemental concentrations. 9Li 207 has higher values for all elements, followed by 9Li1637, while 9Li21 has the lowest values. Nb Sr -1.0 -0.6 0.8 Cu Zr 1.0 Fe 0.0 Ti47 Nb93 -0.6 Zr -0.8 Sr88 Zr90 9Li1637 9Li207 9Li21 Sr 0.0 0.4 0.2 Mn55 9Li1637 9Li207 9Li21 Fe Axis 2 (17.0% variance) Ni -0.4 Fe57 x 1E3 Pulses -0.5 0.6 Y89 -0.4 Y 0.0 Axis 2 (30.8% variance) 0.6 0.4 0.2 0.0 Axis 2 (15.1% variance) Ba -0.2 Figure 21. Project sites on St. Catherines Island Rb Axis 2 (21.1% variance) Site 9Li1637 9Li207 9Li21 Site Y -0.2 0.8 0.8 (Thomas 2008, fig 2.4). Site x 1E3 Pulses 400 -0.5 Figure 2.2. Examples of Irene sherds and vessels Introduction: This case study compared kiln wasters from five historic earthenware production sites in the eastern Piedmont of North Carolina. Lead glazed coarse earthenware, often called redware, presents a challenge to provenience studies, because it was made in similar forms and decoration across a wide geographic and temporal range. However, wasters of known provenience can be compared chemically to establish site level or regional compositional markers. This pilot study of twelve sherds was conducted in order to determine the degree to which the products of multiple kilns could be differentiated based on elemental differences in clay bodies. Introduction: This case study examines the composition of Catawba Indian potsherds and clays, to investigate the raw material use of Catawba potters from the 1750s to the present. The Catawba left their early eighteenth-century settlements after a smallpox epidemic in 1759 and regrouped several miles downriver (Merrell 1989). After this move, Catawba potters developed a highly burnished trade ware for the colonial marketplace (Riggs 2010). Trade ware production continued for the next two centuries, becoming an art form that remains an important element of Catawba identity (Blumer 2004). -1.0 Introduction: This case study examines raw clays and Late Mississippian, Irene phase (AD 1300-1580) sherds recovered from St. Catherines Island, Georgia. The goal for this study was to determine discernible compositional differences of local resources from a relatively small defined area. The dataset consisted of 2 raw clay sources and 11 diagnostic sherds recovered from shell midden contexts at 3 Irene phase sites (Figure 2.1). The Irene phase ceramics are typically characterized by coarse grit temper; decorations of complicated stamping, incising, or burnish plain wares; and vessel forms of globular jars and carinated bowls (Figure 2.2). All four methods were used in this case study. XRD was conducted on the two visually different clay samples (South End and Relic Marsh) to determine potential crystalline structure differences. XRF, ICP-MS, and pXRF methods focused on determining compositional groups among the ceramics. North Carolina Piedmont Axis 2 (39.1% variance) Lower Catawba River Valley St. Catherines Island 1.7 1.8 Zr log-base 10 ppb Figure 4.4. ICP-MS values for zirconium, by sample (n=7). Concentration of this element is geographically variable, showing a general decrease from north to south within the study area. 1.9 2.0 Figure 4.3. PCA biplot of selected ICP-MS values (log-base 10 ppb). Component 1 is dominated by lead (Pb208) concentration, distinguishing the Joseph Loy sample from all others. Since this ware was lead glazed, lead content in these archaeological samples was likely affected by production or postdepositional processes. Component 2 is structured by (Ca44) and strontium (Sr88) concentrations. These elements leach from clays at the same rate during processes of chemical weathering, and thus variation of these elements within the samples is likely reflective of different clay sources. -1.0 -0.5 0.0 0.5 Axis 1 (49.5% variance) Figure 4.5. PCA biplot of selected oxides and trace element concentrations using XRF (log-base 10 mass % for oxides and log-base 10 ppm for elements). Northern samples differentiate from southern samples. Dennis samples have the highest manganese oxide values. Acknowledgements: The authors wish to thank the Research Laboratories of Archaeology for funding this research, and Dr. Vincas Steponaitis for his guidance of this project. Many thanks are owed to Dr. Drew Coleman, Dr. Allen Glazner, Ryan Frazer, and Sean Gaynor in the Geology Department, and Dr. Sohrab Habibi in the Mass Spectrometry Lab at UNC-CH for training and the use of their instruments. Thanks also to Bruker AXS and Bruce Kaiser for the loan of a Bruker Tracer III-V pXRF instrument and training to use it. Lindsay would like to thank Dr. Linda CarnesMcNaughton, Hal Pugh, Eleanor Minnock-Pugh, and the Research Labs of Archaeology for the loan of samples from their collections. Anna would like to thank the Edward John Noble Foundation, St. Catherines Foundation, and the American Museum of Natural History for the loan of samples for this project. Mary Beth and Rosie would like to thank Steve Davis and the Research Laboratories of Archaeology for the loan of samples, and Beckee Garris from the Catawba Cultural Preservation Project for providing the modern clay samples used in this project. References: 0.9 1.0 1.1 1.2 1.3 1.4 1.5 Rb log-base 10 ppm Figure 4.6. XRF values for rubidium, by sample (n=9). Concentration of this element is geographically variable, showing a general increase from north to south within the study area. Conclusions: The underlying geology of this region is complex. Though the sites studied here overlie similar geological formations, chemical characterization of the sherds suggest that the concentrations of several elements, especially zirconium (Zr), rubidium (Rb), and strontium (Sr,) vary predictably within the clay sources. With a larger sample, it may be possible to reliably differentiate ceramic products from northern and southern reaches of this area. Blumer, Thomas J. 2004 Catawba Indian Pottery: The Survival of a Folk Tradition. Tuscaloosa: University of Alabama Press. Crow, Rosanna 2011 Geochemical Analysis of Catawba Ceramics. Honors Thesis, Curriculum in Archaeology, University of North Carolina at Chapel Hill. Faure, Gunter 2001 Origin of Igneous Rocks: The Isotopic Evidence. Berlin, New York: Springer. Merrell, James H. 1989 The Indians' New World: Catawbas and their Neighbors from European Contact through the Era of Removal. Chapel Hill, NC: University of North Carolina Press. Riggs, Brett H. 2010 Temporal Trends in Native Ceramic Traditions of the Lower Catawba River Valley. Southeastern Archaeology 29(1):31-43.