SAMPLE SELECTION, CHARACTERIZATION AND CHOICE OF TREATMENT FOR ACCURATE RADIOCARBON ANALYSIS—INSIGHTS FROM THE ETH LABORATORY

Abstract

Accurate radiocarbon (¹⁴C) analysis depends on a successful carbon separation relevant to the studied object. The process of ¹⁴C dating involves the following steps: characterization and sample choice, sample treatment, measurements, and evaluation of the results. Here, we provide an overview of conventional approaches to macromolecular samples and address specific issues such as detecting and removing contamination with roots, dolomite, and conservation products. We discuss the application of elemental analysis (%N, %C) in the preparation of bones and the infrared analysis in monitoring the contamination of samples. Our observations provide the basis for the discussions of the existing results and for planning the future sampling.

INTRODUCTION

All laboratories strive to achieve the highest precision and accuracy when performing radiocarbon (¹⁴C) analysis. Although these two terms are often exchanged to express the desire for the best chronological estimates, they are not synonymous. Modern measurement techniques, certificates, and intercomparison studies provide quality assurance. However, the most precise ages can also be inaccurate (Geyh 2008).

The accuracy of radiocarbon ages is dependent on various factors. First is the source of carbon built into the sample at the time of its formation, i.e., ¹⁴C age or the isotopic signal of the reservoir. Other factors are the stage of preservation or degradation of the sample and contamination with allochthonous carbon, which might be related or amplified by the degradation (van Klinken and Hedges 1998; van Klinken 1999). Last is the selection of the original sample and its purification before ¹⁴C analysis.

The wide range of materials and applications of radiocarbon analysis requires using protocols developed for different types of material (Hajdas 2008; Wood 2015; Hajdas et al. 2021a). In general, all laboratories follow standard procedures of ABA, cellulose separation, Longin method or Ultra Filtration for bones, but modifications of protocols are standard practice (Brock et al. 2007; Hajdas et al. 2007, 2009; Brock et al. 2010a, 2010b, 2013, 2018; Rubinetti et al. 2020; Pawelczyk et al. 2022). New opportunities for the separation of carbon suitable for radiocarbon dating arrived with the development of compund-specific radiocarbon analysis (CSRA) (Eglinton et al. 1996; Ingalls and Pearson 2005). The range of applications of CSRA expanded the field of RA, especially in studies of sedimentary records (for example Blattmann et al. 2020; McNichol and Lindauer 2022), and archeology, including dating pottery (Casanova et al. 2022) and bones (Deviese et al. 2018). Our overview concentrates on the preparation of macromolecular type of samples and conventional pretreatment (van Klinken and Hedges 1998).

In addition to the sample treatment (purification), material selection is essential. The assignment occurs during the fieldwork. Later, the refined choice of material or separation of the suitable fraction is performed in the laboratory after a visual investigation using binoculars. Dependening on the type of material, the selection of datable carbon by wet or dry sieving, separation of macro and micro remains, drilling, and cutting suitable pieces is chosen.

Binocular observation is most effective and paramount to the understanding of the sample. For example, a mixture of anthracite and charcoal has been observed in the samples of rock varnish, which allowed us to scrutinize and question the validity of the radiocarbon dating of rock varnish (Beck et al. 1998). Often, synthetic materials such as textiles can be easily identified. The most common problem detected using the microscope is contamination by roots (in situ) or anthropogenic contaminants such as dust, hair, and fiber. The latter, is somewhat random and difficult to deal with because there is no guarantee that all contaminants can be ever picked out of the samples. More common is contamination with roots observed in soils, peat, sediments, wood, and charcoal. Sieving removes roots from sediments and peat (Hajdas et al. 2021b); however, the infested wood and charcoal used for radiocarbon analysis can be contaminated even if visible roots are removed from the sample. Sieving is also used when macro and micro remains are selected for radiocarbon analysis. Separation of terrestrial macrofossils assures that the material is free of reservoir effect (hard water effect) (Hajdas et al. 1993, 1998, 2021a).

Another material that requires sieving and selecting suitable carbon fraction (grain size) is lime mortar (Lindroos et al. 2007). An alternative method is the cryo-breaking and ultrasonic separation (Nawrocka et al. 2005; Marzaioli et al. 2014; Michalska et al. 2017).

Radiocarbon dating of bone, tooth and antler requires the separation and purification of collagen (Weiner and Bar-Yosef 1990; Yizhaq et al. 2005). A degree of preservation determines the success of radiocarbon analysis (van Klinken 1999). Brock et al. (2010b) tested an assessment based on elemental analysis of %C and %N content, or C/N ratio of the original bone, and found that bones with %N< 0.76 are less promising. Another technique used by radiocarbon laboratories employs the infrared analysis to detect characteristic absorption lines of the bone to assess the preservation of collagen (D’Elia et al. 2007; Lebon et al. 2016; Cersoy et al. 2017; France et al. 2020; Leskovar et al. 2022).

More sophisticated methods of sample screening and preselection can be supported by the Fourier transform infrared (FTIR), thermal gravimetric analysis (TGA), scanning electron microscope (SEM), Raman, direct temperature-resolved mass spectrometry (DT-MS), Py-GC-MS, and other techniques. Identifying specific components using FTIR spectroscopy can be utilized in studies of bones, paintings, and charcoal (Alon et al. 2002). The infrared light absorbed (or transmitted) depends on the studied material’s molecular composition. Molecules with different types of vibration modes absorb characteristic wavelengths. The specific regions and peaks of absorption/transmission minima allow the identification of molecules/material, which is often supported by the existing databases of FTIR spectra (https://centers.weizmann.ac.il/kimmel-arch/infrared-spectra-library). In the preparation of radiocarbon samples, the detection of synthetic and conservation materials is critical. Most synthetic polymers, which are long-chain carbon molecules, are made of fossil carbon; therefore, the dead carbon contamination is significant. Often, such contamination requires modification of the standard treatment as well as additional control of the clean sample before combustion and AMS ¹⁴C analysis (Yizhaq et al. 2005).

The use of FTIR in radiocarbon laboratories is not limited to the detection of synthetic contaminants added during the conservation and preservation process, including the use of pesticides (Tiilikkala et al. 2010). As mentioned above, FTIR is useful in screening for well-preserved bones and in the characterization of mortars (Paama et al. 1998; Al Sekhaneh et al. 2020; Calandra et al. 2022). Moreover, the FTIR analysis of sediments can provide information about the carbonate content and, most importantly, indicate the presence of dolomite and other minerals. The standard treatment of acid-base-acid is insufficient to remove the dolomitic component. The effect of contamination with carbon-free dolomite is amplified by the fact that glacial and fluvial sediments have a very low organic %C, and dolomite is free of ¹⁴C. Also, organic carbon in soil and sediments might have old components trapped by minerals, such as clay (Scharpenseel and Becker-Heidmann 1992), which requires a different approach such as a stepped-combustion or Ramped pyrolysis/oxidation (McGeehin et al. 2001; Wang et al. 2016; Hemingway et al. 2019).

This paper presents an overview of the most common methods used to characterize and select material suitable for radiocarbon dating. Establishing and monitoring treatment efficiency is the key to accurate radiocarbon dating.

METHODS

The spectrum of sample material submitted and processed at the ETH radiocarbon laboratory is wide. Thus, the overview of methods used to select and purify material for radiocarbon dating is based on our observations gained during a couple of decades. Table 1 shows methods applied in a pre-screening process followed by sample preparation (Table 1 Supplementary Material) before the AMS analysis.

Table 1 Sample types, pre-screening methods and preparation methods. Details of pretreatment methods are summarized in Table 1 of the Supplementary Material.

Microscope/Binocular (Magnification 10×–50×)

A visual investigation and documentation of suspicious contaminants is the first step before selection. Some samples, such as macrofossils, foraminifera, and sieved fractions of mortar, are selected and identified by the researchers before submission. The microscope is indispensable for determining macrofossils (Hajdas et al. 1993, 1998) or foraminifera shells (Broecker et al. 1990) but also for removing contaminants such as exogenous fiber, roots, or remains of insects and hair.

Sieving

Sieves, mesh 125 μm (also 150 μm can be applied) are used to separate fine (roots-free) fractions from samples of sedimentary deposits (peat, soil, sediments). At ETH laboratory, we apply 100 mm diameter stainless steel sieves (Retsch) and a collection pan with a funnel to collect water and a fine fraction (Figure 1).

Figure 1 Setup for wet sieving of sediment and peat samples. The fine fraction (<125 μm or 150 μm) is collected in a glass beaker. The larger fraction (top sieve) can be investigated under binoculars.

It is essential to work on fresh or stored in a freezer (wet) material; samples should not be dried, crushed, or milled before the sieving and are soaked in MiliQ water to soften and disintegrate the bulk.

Stainless steel sieves (100 mm Retsch), mesh 45 μm and 63 μm, and a collection pan are placed on the Retsch dry-sieve shaker to sieve mortar. Before sieving, the sample is investigated, and if present, any lime lumps are collected from the bulk. Small sample fragments are then crushed, and if present, stones (aggregates of mortar) are removed, and the powder is sieved. The process is repeated to collect at least 100 mg of powder 45–63 μm. The smaller and larger fractions are also collected and archived.

Elemental Analysis

The carbon content of sedimentary deposits (TOC) varies greatly; therefore, %C analysis is performed on clean fractions before combustion and the AMS analysis. A few milligrams (5–10 mg) of pure material are weighed and packed in the aluminium (Al) ¹ or tin (Sn) cups for analysis with an Elemental Analyzer (EA). The measured %C is used to calculate the mass of the sample material, which contains 1 mg of C. For example, 10 mg of material with a C content of 10% needs to be combusted for the graphite target to contain 1 mg of C. Samples with poor %C (less than 1%) are planned for analysis with gas ion source (GIS) (Ruff et al. 2010; Haghipour et al. 2019) and the equivalent of ca. 100 μg of carbon is packed in Al cups.

Nitrogen and carbon content of original bones (%N and %C) can be measured in an EA. A small portion of the cleaned original bone (5–10 mg) is weighed and packed in Al cups for EA analysis. The values of %N, %C, and C/N ratio are saved into the database. Currently, at the ETH laboratory, the preparation of samples with very low %N (<1%) is stopped.

The C/N_at ratio of gelatin is obtained during combustion and graphitization (Nemec et al. 2010). This value is stored in the database, and the yield (mass gelatin/mass bone sample) indicates the gelatin’s purity/quality.

The C/N_at ratio can also be used to identify the type of material (Hajdas et al. 2014) and purity (Boudin et al. 2013).

Fourier Transform Infrared Spectroscopy

Characterization of sample material and detection of potential contamination is possible with the help of FTIR spectroscopy. No preparation is required for analysis using the ATR modus (for example, PerkinElmer Spotlight 200i used at the ETH laboratory), but multiple subsamples should be analyzed (heterogeneity of contamination). The transmission/absorption spectra are compared with the spectra of specific materials. Cross-check of clean samples is performed before combustion. Moreover, the FTIR can be applied to screen/assess the preservation of bones.

Sample Pretreatment

Different types of samples are treated differently after the characterization of the sample and selection of suitable fractions. For example, the methods described below are routinely applied at the ETH laboratory. The pretreatment details used at the ETH laboratory are summarized in Table 1 of the Supplementary Material.

Acid-Base-Acid (ABA)

The sequence of washes in acid and base is applied before the combustion of organic samples. The ABA procedure can be modified to adjust to sample contamination or preservation degree. Some steps, such as base, can be omitted (for example, when dating TOC of soils), or a stronger solution and longer treatment time is applied, for example, when dolomite is present and acid wash is extended to multiple days. Modifications are also applied in the case of a base step, which is destructive for wool and silk, therefore, such samples are subject to a short base step and performed at room temperature.

Poorly preserved charcoal can dissolve in the base, and only humic acid can be collected for radiocarbon analysis; however, one must be aware that it can be of mixed carbon sources. Also, peat and sediment fine fraction treatment can be modified to separate and date humin and humic fractions.

Solvents

A standard sequence of hexane, acetone and ethanol is applied when FTIR analysis indicates the presence of complex carbon molecules (oil, fat, waxes, conservation material). The Soxhlet apparatus is often applied to wash the sample in a clean solvent (Hajdas et al. 2004; Hajdas 2008). However, glass vials are used when chloroform treatment is necessary (Liccioli et al. 2017; Kessler et al. 2022) because samples float and might discharge via the siphon of the Soxhlet apparatus. The glass vials with samples and solvents are placed on a shaker for a few hours in the heated block (60ºC). The solvent is replaced, and the wash continues for one working day. The sample is left to dry overnight and checked with FTIR the next day. If required, the cleaning is repeated. It is worth noting that the use of glass vials and shaker tables is more sustainable and requires a lower quantity of solvents but more of pipetting out the liquid solvents, which must be done under the fume hood.

Ultra-Filtration of Gelatin

The treatment of bones and antlers requires the separation and purification of gelatin (Table 2). The main modification of the procedure (Hajdas et al. 2009) is a return to a dissolution of fragments of bones without crushing them (Hajdas et al. 2007). The modification was introduced following studies of Fewlass et al. (2017, 2019), who showed that demineralization of larger pieces of bones improves collagen recovery and allows radiocarbon analysis to be performed on much smaller samples of bone.

Table 2 Contamination (wavenumber cm^-1 in bold Italics) detected with the help of FTIR and the results of radiocarbon dating after additional treatment.

Sequential Dissolution Mortar

Mortar powder (45–63 μm; ca. 100 mg) is dissolved in condensed phosphoric acid. The CO₂ is collected and closed in a glass tube in four intervals, each 3 seconds long. Dependening on the amount of CO₂, the sample is sealed in a glass tube for graphitization (>200 μg) or for GIS (Hajdas et al. 2020a, 2020b).

Special Samples

Occasionally, unique samples such as paint (Hendriks et al. 2018), lime lumps (Lindroos et al. 2018), cremated bones (Lanting et al. 2001; Major et al. 2019), iron (Hüls et al. 2004), wine (Quarta et al. 2022) and other liquid samples can be subjects of radiocarbon analysis. Methods used to prepare such samples are summarized in Table 1 Supplementary material.

Dependening on C content, purified samples are graphitized (Nemec et al. 2010) or analyzed as gas samples (Ruff et al. 2010) at MICADAS (Synal et al. 2007).

RESULTS AND DISCUSSION

Effects of the applied screening and sample treatment are illustrated by the example of various materials analyzed and sometimes re-analyzed after additional treatment. Evaluation of results for samples measured with GIS (minimal C mass) considers constant mass contamination (Welte et al. 2018; Haghipour et al. 2019). The results reports include: Radiocarbon ages and F¹⁴C (Stuiver and Polach 1977; Reimer et al. 2004), δ¹³C measured during AMS analysis, C/N_at, (%C and %N from combustion prior graphitization), C mass of targets analyzed by AMS and, if available Yield=mass after treatment/mass start. The yield value provides information about sample conditions but sometimes cannot be provided (wet sample, small sample, missing notes).

In addition, for collagen samples, we might provide IRMS analysis C/N_at, δ¹³C, δ¹⁵N

Roots

Roots present in peat (Figure 2a) or sediment samples can be removed. Sieving was successfully applied in dating the sedimentary deposits in the Italian valley Val Ferret southeast of the Mont Blanc Massif (45°56′35″N 7°05′26″E) to clarify the controversial chronology of the 1717 avalanche (Hajdas et al. 2021b). However, contamination of wood samples or charcoal is often impossible to remove. Roots are growing deep into the sample (Figure 2b), and a web of roots is hidden inside the pieces of charcoal or wood.

Figure 2 (a) Sample of peat VF-28 (Val Ferret) (Hajdas et al. 2021b) was sieved to remove the visible roots (b) roots observed in the wood submitted to the laboratory could not be removed.

Dolomite

Contamination with calcium carbonates CaCO₃ (limestone) is removed from samples (any material) in the acid step of ABA treatment. Most samples are sufficiently treated in this step, and radiocarbon analysis provides accurate ages. However, sediment or soil samples from the regions with dolomite require stronger treatment because dolomite CaMg(CO₃)₂ reacts slowly with HCl. Table 2 shows F¹⁴C measured on bulk sediment treated with standard ABA, which resulted in ages outside of the expected range. The FTIR investigation of the remaining clean sample has shown that the dolomitic component is still present in the sample, showing absorption lines 1436, 888, 730 cm^–1 (Table 2).

Additional, extensive treatment lasting for a few days, with stronger HCl acid (1 M instead of 0.5 M) removed dolomite, and higher F¹⁴C values of the sample were measured. Higher F¹⁴C indicates the removal of contaminants such as dolomite and has values of F¹⁴C close to 0. The effect is also visible as a change in δ¹³C of the sample from ∼0 toward more negative values (Table 2).

Conservation Materials

Treatment of samples from heritage objects often depends on the contamination type. Table 2 shows examples of results obtained for a variety of samples. The radiocarbon age of the canvas, dated too old for the expected age, was treated with an additional solvent. The change in FTIR spectra of the clean sample was confirmed by radiocarbon analysis returning a higher F¹⁴C of the clean canvas, i.e., the contamination with fossil carbon was removed.

Results of radiocarbon analysis of wood treated with polyethylene glycol (PEG) is an example of difficulties in removing PEG. The FTIR absorption spectra of PEG-treated wood show strong absorption peaks of PEG (2888, 1467, 1367,1342,1147,1107,1058, 963, 842 cm^–1). These peaks are still in the spectra even after additional solvent treatment with ethyl acetate. The measured F¹⁴C increased from 0.269 ± 0.001 to 0.614 after treatment with solvents, but it is still inaccurate due to the detected PEG. The contamination appears to be heterogenous, and mechanical scraping affected the measured F¹⁴C (Table 2).

Collagen Preservation

Prescreening of bones for collagen preservation was only introduced at the ETH laboratory after the Elemental Analyzer was installed in 2012. However, at that time, the practice of checking the %C, %N analysis of the original bone was yet to be part of the standard protocol. Nevertheless, the data collected for over 1000 bones prepared in the ETH laboratory during the last nine years (2013–2022) indicate a wide range of values: %N between 0.05 and 7 and %C values of 0.15 to 28.

The majority of successfully dated bones (N=880) had %N>1. From 150 bones with %N<1, 80 bones failed, most (N=57) of which had %N< 0.76 i.e., below limit proposed by (Brock et al. 2010b). It is worth noting that some bones with %N>1 did not provide gelatin (Figure 3) showing limitations of this prescreening method thus a combination of methods can be of help (van Klinken 1999; Brock et al. 2010b). The poor preservation results in a very low yield but the low yield can also be due to sample handling, especially of fine grain samples. The quality of UF-purified gelatin is well illustrated by the C/N_at values of the graphitized sample (AGE system; Nemec et al. 2010) and the yield. Most samples with C/N_at between 3.1–3.4 show higher yield, while samples with C/N_at outside 3.0–3.5 range had very low yield (Figure 4). Evaluation of the success of sample decontamination remains difficult (van Klinken 1999) however the observed abnormal C/N_at values require check and repeated analysis. Observations from the ETH laboratory suggest that often C/N_at values closer to 3.2–3.3, which is the range for modern bones (Ambrose 1990), result in a better agreement with expected ages.

Figure 3 Results of elemental analysis %N and %C of 80 bones which gave no gelatin.

Figure 4 Correlation between the yield of gelatin and C/N_at ratio of the purified gelatin (%N and %C values from combustion prior to graphitization). The rectangle marks the samples with acceptable yield of >0.1% and C/N_at in the range 3.1–3.4.

SUMMARY AND CONCLUSIONS

The wide range of sample material submitted to research and service ¹⁴C laboratories requires the application of different protocols. Instrumental support is the key to monitoring visible and invisible contamination. The presented examples highlight some of the most frequent obstacles to accurate radiocarbon ages. Contamination with roots might be challenging to observe, and it is quite possible that samples of bulk measured in the past were contaminated by roots. Conservation can also be invisible; if undocumented, contamination with old carbon is inevitable. The FTIR spectra provide only qualitative information about the possible contamination, but the results obtained on clean material are satisfactory. Our application of elemental analysis (%N, %C) to access the preservation of bone collagen is possible because of the available equipment. However, using FTIR is an alternative method that can help save the time required to weigh the samples. Finally, all the observations and characterization results are crucial in evaluating the final results. Heterogeneous pieces of mortar and poorly preserved bones need careful evaluation of parameters such as C/N_at or the preparation yield. In conclusion, a holistic assessment of so-called “outliers” can help improve the accuracy of radiocarbon analysis.