Infrared spectra of mixtures of heated and unheated clay: Solving an interpretational conundrum

Fourier transform infrared (FTIR) spectroscopy is frequently used for archaeological studies related to fire, allowing, among other things, researchers to distinguish between unheated and heated clay minerals. However, heat signatures are not always clear‐cut in infrared spectra of bulk sediments, as spectra occasionally appear with ambiguous absorbance bands attributed to hydroxyl (OH) in clay minerals. This paper presents an experimental study addressing this interpretational problem by considering the effect of mixtures of heated and unheated clay, a phenomenon expected in archaeological sites. After creating experimental mixtures and testing them using bulk FTIR spectroscopy, our results indicate that even a relatively small amount of unheated clay—only ca. 5%–10% mixed into a fully heated deposit—will result in ambiguous infrared spectra that are difficult to interpret. For comparison, ambiguous bulk FTIR spectra from two archaeological contexts—an ashy fill within a pit installation and a hearth—were studied with FTIR microspectroscopy, which demonstrated the presence of unheated clay within a largely heated deposit. Micromorphological observations explain the mixed nature of the investigated archaeological contexts, in this case, primarily via bioturbation. Our results thus emphasize the importance of microcontextual analysis of clay minerals. Furthermore, these results indicate that heated deposits are likely missed altogether in some archaeological contexts where only bulk FTIR analyses have been conducted.


| INTRODUCTION: A GEOARCHAEOLOGICAL TECHNICAL INTERPRETATIONAL PROBLEM
Fourier transform infrared (FTIR) spectroscopy has been a common practice in archaeological research for more than 30 years (Monnier, 2018).The study of archaeological materials by FTIR spectroscopy was primarily developed using the KBr transmission method for bulk material analysis (Weiner, 2010), while FTIR microspectroscopy has also recently been applied to archaeological materials (Berna, 2017).One of the main advantages of FTIR spectroscopy is the detection of heat-induced changes to the molecular structure of clay minerals in geosciences ( Van der Marel & Beutelspacher, 1976, pp. 36-43) and in archaeology (e.g., in pottery- Shoval & Beck, 2005;in mudbricks-Forget et al., 2015; in sediments- Berna et al., 2007;Butler et al., 2020).The early research by Berna et al. (2007), which was the first systematic application of this concept to geoarchaeology, showed that the most pronounced changes in the infrared spectrum of various clay minerals are noted when clays are heated to 450-500°C and above.These spectral changes include a gradual change in the position of the main clay absorption band, from 1032 cm −1 in unheated clay to 1092 cm −1 in clay heated at 1100°C.Additionally, Berna et al. (2007) showed that the absorption bands associated with hydroxyl (OH) groups in clay minerals, at 3625 and 3695 cm −1 and the OH deformation band (Al-Al-OH) at 915 cm −1 , disappear when clay minerals are heated above 500°C.Furthermore, the smectite-kaolinite Si-O-Al band at 525-530 cm −1 decreases with heating at increasingly higher temperatures until it disappears at about 800°C (Berna et al., 2007, fig. 5).This approach, used on bulk samples, has been successfully applied globally, as indicated by over 300 citations of Berna et al.'s (2007) publication, according to Google Scholar (as of March 2023).
However, interpretational problems of clay spectra sometimes arose when studying sediments from archaeological sites.The problem occurs in spectra where the OH bands (at 3625, 3695, and 915 cm −1 ) appear as weak shoulders-indicating unheated claywhile the main clay absorbance band is shifted to higher wavenumbers, sometimes as high as 1040 cm −1 and often accompanied by the weakening or disappearance of the band at 525-530 cm −1indicating heated clay.Some scholars interpret such spectra as including clay heated to temperatures around 400-450°C (e.g., Friesem et al., 2014, Fig. 9d,e), while others have interpreted similar spectra as indeterminate due to the mixing of clay types, intrusion of other materials (opal) (e.g., Shahack-Gross et al., 2009, p. 178), or exposure to temperatures lower than 500°C (Laugier et al., 2021, p. 6).Shahack-Gross et al. (2009)  subsequent publications) did not help solve this conundrum because published materials (e.g., Friesem et al., 2014, fig. 5) focus on distinct changes in infrared spectra of clay minerals, especially those heated at temperatures above 450°C.We therefore set out in this study to test the possibility that such spectra obtained using bulk FTIR analysis via the KBr method, in fact, represent mixtures of unheated and heated clay minerals.

| EXPERIMENTAL: HEATED-UNHEATED CLAY MIXING
About 20 g of geogenic unheated local sediment from the vicinity of 'En Esur was used for the mixing experiment.Ten grams were heated in a muffle furnace (Thermolyne F6000; Thermo Fisher Scientific) for 4 h at 700°C.Mixtures of heated and unheated sediment were prepared by weighing and calculating the weight % of the heated and unheated proportions (Table 1).The dry mixtures were homogenized in 50 mL plastic tubes by vortexing for 2 min.All homogenized mixtures were then analyzed by bulk FTIR using the KBr method.
Spectra were obtained between 4000 and 400 cm −1 with a Nicolet iS5 (Thermo Fisher Scientific) spectrometer using Omnic 9.3 software.

| RESULTS: BULK MIXING EXPERIMENT
The unheated sediment from 'En Esur is composed of quartz and clay (with the presence of calcite) (Figure 2b | 825 As the experiment results indicate that spectra with contradicting evidence for heated clay may occur due to mixing, we further tested this in two archaeological case studies, using FTIR microspectroscopy and micromorphology.Notably, previous studies that used FTIR microspectroscopy in archaeology identified heated and unheated clay in close proximity, yet they focused on the understanding of broad contextual archaeological questions (e.g., Villagran et al., 2019, fig.3); the study reported here uses FTIR microspectroscopy to focus only on the understanding of mixtures of heated and unheated clay in bulk FTIR spectra.
From 'En Esur, an Early Chalcolithic II (ca.5200-4800 B.C.E.) stonelined and plastered pit installation was studied, where previous bulk FTIR spectra were ambiguous regarding heating (as explained above; Figure 2a), and a corresponding micromorphological thin section was collected.From Ein Ziq (Negev Highlands, Israel), an Intermediate Bronze Age (ca.2450-2200 B.C.E.) hearth was studied (Dunseth et al., 2017).The hearth (L.15/J/14) appeared in an excavated profile and was characterized, from top to bottom, to be composed of an ash-rich layer; a charcoal layer; and a rubified layer (Figure 1c).The bulk FTIR spectra from the hearth deposits were interpreted to include clay heated at low temperatures (400-500°C) due to the appearance of shoulders of the bands at 914, 3625, and 3695 cm −1 (Figure 2a).A micromorphological thin section was also produced from this hearth.Both thin sections were ground to the standard 30 μm thickness and mounted on a glass slide of 1 mm thickness.
A Nicolet iN10 infrared microscope (Thermo Fisher Scientific) attached to a spectrometer operated by Omnic Picta software was used to study the deposits in both thin sections.Spectra were collected with 128 scans using a cooled detector (MTC-A).Using a 50-μm-diameter aperture, particles were analyzed in transmission with a Reflectocromat ×15 objective between 4000 and 2400 cm −1 at 8 cm −1 resolution, which is the most suitable to study the effect of heat on clay minerals, where the OH absorbance bands of clay minerals appear (Berna et al., 2012(Berna et al., , 2017;;Villagran et al., 2017Villagran et al., , 2019) ) (Figure 2c).Unlike other studies, we used only transmission to avoid analyzing absorbance bands that may be affected by other minerals common in archaeological deposits, such as quartz, calcite, and carbonated hydroxylapatite; these do not absorb infrared radiation in the 3600 cm −1 region, as clay minerals do.An area of 6 × 2.5 mm in the thin section from the fill of the 'En Esur pit installation was scanned at 150 μm spatial resolution totaling 882 spectra, and an area of 13.5 × 2.5 mm in the thin section of the Ein Ziq hearth was scanned at 200 μm spatial resolution totaling 784 spectra.The difference in the scanned area and the spatial resolution between the two thin sections is due to the different contexts: the pit-installation fill is homogeneous, requiring more coverage, while the hearth is characterized by well-differentiated layers.Using Omnic Picta software, a photomosaic was generated for each scanned area followed by an automatic collection of FTIR spectra according to the determined spatial resolution and a compilation of all spectra into an intensity map for the OH absorbance bands of clay minerals.
Heated clay was identified based on the absence of OH bands (3625 and 3695 cm −1 ), while unheated clay was determined based on their presence (Figure 2c).The percentage of heated and unheated clay was calculated based on the proportion of spectra collected within the scanned areas: the entire scanned area from the 'En Esur pit installation and the ash and charcoal layers in the Ein Ziq hearth.
Calcite was noted based on an absorbance band at 2500 cm −1 .
Where calcite was noted, it appeared without OH bands and was thus eliminated from calculating the percentages of heated clay minerals.
Results from 'En Esur pit installation indicate that 13% of the scanned area is composed of unheated clay (Figure 3a,c and Table 2).
Micromorphologically, the lower part scanned is the pit-installation lining, composed of chalk and flint gravel that shows only small amounts of heated clay.The deposit above this lining, the pit fill deposit, is highly disturbed due to bioturbation.Calcified root channels suggest plant turbation in this archaeological deposit (Figure 4a,b).
T A B L E 1 Heated (700°C/4 h) and unheated sediment mixtures utilized in the bulk FTIR experiment.The deposits from the Ein Ziq hearth show an overall 8% of unheated clay (Figure 3b,d and Table 2), yet its distribution accords with its layered structure: primarily unheated clay below the rubified layer, primarily heated clay in the rubified layer and the lower part of the ash layer, and the presence of unheated clay at the upper part of the ash and charcoal layer.Micromorphological observations show a generally intact deposit in the lower part of the ash layer, with only occasional disturbance features such as an insect passage feature (Figure 4c).In contrast, the top part of the ash layer is more disturbed and includes large voids (Figure 4d).

| DISCUSSION AND CONCLUSIONS
The two archaeological contexts associated with possible firing activities (a fill within a pit installation and a hearth) produced ambiguous bulk FTIR spectra that did not allow us to determine whether or not the deposits were heated, and if they were heated then to what temperature (Figure 2a).This study shows that in a deposit containing quartz and clay, it is enough to have 10% (by weight) of unheated sediment mixed with a primarily heated sediment to obtain a bulk FTIR spectrum that will cause interpretational difficulties and possibly even mistakes.Therefore, we propose that such mixing calibrations should be conducted per studied site using prevailing regional sediments.We note that in the study reported here, we did not control for the composition of the sediment used for calibration versus the archaeological deposit; in this case, there was more quartz in the sediment used for experimental calibration than in the archaeological deposits.Therefore, we recommend that future studies should first determine the composition of the archaeological deposits to be studied and then conduct the calibration experiment of a local deposit having a similar composition.
The mechanism responsible for mixing heated and unheated clay in the archaeological contexts studied here is primarily plant and insect turbations.Other mixing mechanisms that may impact interpretation could be postdepositional clay translocation (i.e., illuviation), and possibly also cryoturbation (Stoops et al., 2018, p. 226).
and, over a decade later, Laugier et al. (2021) raised the need for experimental work to solve this conundrumB.C.E.) fill deposits in pit installations at the site of 'En Esur (also known as 'Ein Asawir; Figure 1a,b), we obtained spectra in which the absorption bands indicating heated and unheated clay appeared together, problematic for interpreting heating in the studied deposits (Figure 2a,b).This ambiguity hampered our understanding of the function of these installations.Consulting previously published clay heating experiments (e.g., Berna et al., 2007 and F I G U R E 1 (a) Location of the archaeological sites mentioned in the text.The inset (bottom left) shows the location of the region in the Mediterranean basin (red rectangle).(b) Early Chalcolithic pit installation from 'En Esur (red circle; photo by A. Peretz, courtesy of the Israel Antiquities Authority).(c) Hearth 15/J/14 at Ein Ziq.Note the black charcoal layer and ash layer just above it (photo by Z. C. Dunseth).
Infrared spectra from 'En Esur (upper) and Ein Ziq (lower), showing contradicting attributes.'En Esur: on the one hand, a slightly shifted main clay absorbance band (1036 cm −1 ) and a lowered band at 525 cm −1 indicating heated clay, and on the other hand, shoulders of bands indicating unheated clay at 3625, 3695, and 915 cm −1 .Ein Ziq: on the one hand, an unshifted main clay absorbance band (1034 cm −1 ) and a clear band at 525 cm −1 indicating unheated clay, and on the other hand, minute bands at 3625, 3695, and 914 cm −1 unexpected in unheated clay, thus suspected to originate from slightly heated clay.The band at 3533 cm −1 is from gypsum.(b) Changes in the spectra of mixtures comprising various proportions of unheated and heated sediment.See text for details.(c) Spectra obtained using Fourier transform infrared microspectroscopy: unheated clay (top) and heated clay (bottom).The high absorbance around 2900 cm −1 is due to the polyester resin used in the preparation of the thin sections.
,1).The main clay band in this sediment is located at 1036 cm −1 , and all bands accompanying unheated clay are present: 525-530 (here at 512 cm −1 position affected by the presence of the quartz band), 915, 3625, and 3695 cm −1 .The spectra of mixtures containing less than 60% (by weight) of heated sediment are similar to that of the unheated sediment.The mixture containing 60% (by weight) of heated sediment differs from unheated sediment by the smaller shoulder at 915 cm −1 (Figure2b,2).Spectra of mixtures between 60% and 90% (by weight) of heated sediment still include absorbance bands at 512, 3625, and 3695 cm −1 , while the main clay band is affected by the presence of the main quartz band at 1082 cm −1 (Figure2b,3).Spectra of mixtures containing 91%-93% (by weight) of heated sediment show a lowering of the bands associated with OH at 3625 and 3695 cm −1 (Figure 2b,4,5), and only spectra of mixtures containing 93% (by weight) of heated sediment completely lose the distinctive OH bands (Figure 2b,6).The results of this experiment indicate that a heated deposit including merely 10% (by weight) of unheated clay-containing sediment can change the bulk infrared spectrum so that the studied sample will either be identified as unheated, heated to temperatures between 400°C and 450°C, or merely indeterminate.F I G U R E 3 (a) Thin-section scan of the 'En Esur pit installation.The dashed line separates between the pit installation's constructed surface (below) and a brown-gray fill (above).The area scanned by FTIR microspectroscopy is noted by the red square.(b) Thin section scan of the Ein Ziq hearth.The dashed line separates between the yellow-brown deposits on which the hearth was constructed (below) and the hearth components (above) that include from bottom to top: rubified, charcoal, and ash layers.(c) Composite image (left) and intensity map (right) of the FTIR microspectroscopy scanned area in the 'En Esur thin section.Note the presence of heated (light blue) and unheated (green) clay in the fill (upper part of the image) within the pit installation.(d) Composite image (left) and intensity map (right) of the FTIR microspectroscopy scanned area in the Ein Ziq thin section.Note the dominance of unheated clay (red) in the sediment below the hearth and the presence of some unheated clay within the largely heated clay mass (green) within the hearth component.The intensity maps in (c) and (d) indicate the presence or absence of the 3695 cm −1 absorbance band calculated automatically by the software relative to the baseline.OGLOBLIN RAMIREZ ET AL.

F
I G U R E 4 (a, b) Microphotographs from the 'En Esur thin section demonstrating the presence of a calcified root channel (red arrow) in plane-polarized light (a) and in cross-polarized light (b), indicating plant turbation.(c, d) Microphotographs in plane-polarized light from the Ein Ziq thin section showing a passage feature in the ash layer (red arrow in c) as well as abundant voids in the top part of the ash layer (d), both indicating bioturbation.microspectroscopy analysis in combination with micromorphology when possible.An important implication of this study is that where archaeological deposits studied by bulk FTIR spectroscopy are identified having unambiguous signals for heated clay, these deposits are likely relatively undisturbed.
Percentage of spectra identified with unheated clay, heated clay, other materials (primarily calcite, flint, quartz), and voids.
This study shows in both experimental and real-world archaeological contexts that less than 10% of unheated sediment mixed with heated sediment is enough to cause difficulties interpreting the heat status of clay minerals in bulk FTIR spectra.To overcome such interpretational difficulties, we recommend conducting FTIR T A B L E 2