Introduction
In the 1940s, Willard F. Libby proposed to measure radiocarbon as a technique for determining the age of samples (Libby Reference Libby1946). Since then, this approach has gained recognition in the scientific community. To date, radiocarbon dating and its instrumentation are being constantly improved (Hajdas et al. Reference Hajdas, Ascough, Garnett, Fallon, Pearson, Quarta, Spalding, Yamaguchi and Yoneda2021).
One of the measurement techniques used for radiocarbon dating is accelerator mass spectrometry (AMS). This technique combines particle acceleration with mass spectrometry by separating ions of different masses in an electromagnetic field (Jull and Burr Reference Jull and Burr2014). The radiocarbon technique is dedicated to samples of various materials containing carbon (e.g., bone, wood, shell, barley mash, and humic acid). The main advantage of the AMS technique is the relatively low sample mass required for measurement (∼1 mg of carbon or less) compared to other radiocarbon measurement methods. The small mass of the sample necessitates very clean conditions during laboratory procedures, as contamination by foreign carbon alters the radiocarbon age. Dating samples with ages at the limit of the range of the radiocarbon method (50–55 ka) is particularly difficult (Hajdas et al. Reference Hajdas, Hendriks, Fontana and Monegato2017). The appropriate choice of chemical preparation plays a major role with such samples. Due to low radiocarbon content the contamination, in particular by modern carbon, bears serious consequences for the determined 14C age.
The Gliwice Radiocarbon Laboratory was established in 1967 by Prof. Włodzimierz Mościcki, a pioneer in the 14C measurements with gas proportional counters (Mościcki, Reference Mościcki1953, Reference Mościcki1958). Since then, the unit has grown substantially, and at present, is officially known as the Radiocarbon and Mass Spectrometry Laboratory, a part of the Division of Geochronology and Environmental Isotopes at the Institute of Physics – Centre for Science and Education in the Silesian University of Technology. The laboratory is equipped with modern facilities for the chemical pretreatment of samples, graphitization, and radiocarbon measurements. While graphite targets have been prepared since 1999 (Czernik and Goslar Reference Czernik and Goslar2001; Piotrowska Reference Piotrowska2013) they were previously analyzed by external laboratories equipped with AMS spectrometers. In 2022, our laboratory was equipped with a MICADAS accelerator mass spectrometer (Synal et al. Reference Synal, Stocker and Suter2007) produced by the Swiss company IonPlus.
The significant asset of the Gliwice Laboratory is its scientific and technical staff. The team consists of physicists experienced in radiocarbon dating, including experts in dating various materials such as bones, wood, and geological sediments, as well as representatives of Earth and Environmental Sciences, and a dendrochronologist. Their combined expertise is valuable from the planning stage of sampling to the selection of chemical preparation methods, the choice of material fraction destined for measurement, data analysis, and processing.
The international intercomparison programs are well established within the radiocarbon community (Scott et al. Reference Scott, Naysmith and Cook2018). The concept of the project is to provide identical sample sets of various materials to as many radiocarbon dating laboratories as possible. Over 80 laboratories participated in the latest edition called GIRI (Glasgow International Radiocarbon Inter-Comparison; Scott et al. Reference Scott, Naysmith and Cook2022, Reference Scott, Naysmith and Dunbar2023). An outcome of participating in this comparison is the generation of information regarding the quality and reproducibility of the results obtained in the laboratory. Quality assurance and quality control processes are critical in terms of the reliability of the radiocarbon ages obtained, especially when a newly launched device is used for analysis. Additionally, such extensive intercomparison programs allow for the acquisition of new reference samples for routinely dated materials, for which the 14C content has been analyzed in a significant number of laboratories. The Gliwice Laboratory actively participated in previous programs (Pazdur et al. Reference Pazdur, Fogtman, Michczyński and Pawlyta2003).
In this article, we present the status report of the MICADAS at Radiocarbon and Mass Spectrometry Laboratory in Gliwice based on the results for Oxalic Acid II (NIST) standard, IAEA reference materials and background samples obtained over the last two years. We also show the 14C dating results of samples provided to our laboratory as part of the GIRI inter-comparison. We describe the chemical preparation methods for samples of different materials, the types of standards and measurement backgrounds used, and the results of radiocarbon dating compared with the preliminary consensus values for all radiocarbon laboratories participating in the GIRI program.
Reference materials in use in Gliwice Laboratory
Oxalic Acid II (OxII, NIST SRM 4990C; Mann [Reference Mann1983]) is used as the normalization standard. In addition, IAEA C-3, IAEA C-4, IAEA C-5, IAEA C-7 and IAEA C-8 are used as reference materials of known and certified 14C content to monitor the preparation and measurement process.
Since June 2023, the primary background used for organic samples is Phthalic Anhydride (Sigma-Aldrich), for detection of contamination at the combustion and graphitization steps. Secondary background samples are chemically processed blanks: wood (mainly well-preserved subfossil OLGA wood, provided by Leibniz-Laboratory for Radiometric Dating and Stable Isotope Research of Christian-Albrechts Universität, in Kiel, Germany), and coal, whose ages are beyond the range of radiocarbon dating. Due to differences in pretreatment processes, the background for carbonates and liquid fuels is determined separately from other samples, using measurements from appropriate materials, that is Gliwice Marble and Liquid Fuel Background – ON/UF-BC diesel (Baranyika et al. Reference Baranyika, Piotrowska and Michczyński2024), respectively.
Chemical pretreatment protocols in use in Gliwice Laboratory
The samples undergo chemical preparation to remove any contamination, and to extract the designated fraction for radiocarbon dating. All reagents used are of analytical purity grade, and the demineralized water produced with ion-exchange column, has conductivity less than 0.1 μS.
Wood samples
The samples are submitted to an ABA (acid–base–acid) protocol to remove potential contaminants caused by infiltration from sediment/soil. For this purpose, the samples are treated with a 4% hydrochloric acid solution for 1 hr at 75–85°C to remove carbonates. Next, they are rinsed to a neutral pH. Subsequently, the wood is processed with a 4% sodium hydroxide solution for 1 hr at 75–85°C to remove humic and fulvic acids. Again, the samples are rinsed until a neutral pH is reached. In the next step, 4% hydrochloric acid solution is added for 1 hr at 75–85°C to remove contamination with modern carbon dioxide that may be incorporated into the sample structure during the previous use of the alkaline solution. Then, samples are rinsed until a neutral pH is obtained (Michczyńska et al. Reference Michczyńska, Krąpiec, Michczyński, Pawlyta, Goslar, Nawrocka, Piotrowska, Szychowska-Krąpiec, Waliszewska and Zborowska2018; Němec et al. Reference Němec, Wacker, Hajdas and Gäggeler2010a).
When required, before the ABA protocol, the wood samples are cut into small pieces to increase the contact surface with the chemical reagents. They are then examined and appropriate preparation steps are applied, taking into account the state of preservation, possible contaminants, amount of available material, and expected age. Recent wood and materials suspected to be treated with paints, resins, waxes, or preservatives are subjected to Soxhlet extraction. The aim is to flush out resins and other compounds that can be removed with organic solvents. For this purpose, a mixture of 99.8% ethanol and 99.9% toluene is added to each sample in a 1:1 volume ratio. The procedure lasts 4 hr at 80°C. In the second step, the wood is washed with 99.8% ethanol, also for 4 hr at 80°C. Then, the samples are rinsed several times in boiling demineralized water (Sheppard and Thompson Reference Sheppard and Thompson2000; Southon and Magana Reference Southon and Magana2010).
After the ABA protocol, the final stage (Bleaching) involves the use of a 5% sodium chlorite solution acidified with hydrochloric acid to pH = 2, usually repeated twice, for 2 hr at 75–85°C in an ultrasonic bath. More repetitions may be needed to bleach some samples (Dee et al. Reference Dee, Palstra, Aerts-Bijma, Bleeker, De Bruijn, Ghebru, Jansen, Kuitems, Paul, Richie, Spriensma, Scifo, Van Zonneveld, Verstappen-Dumoulin, Wietzes-Land and Meijer2020; Hajdas et al. Reference Hajdas, Hendriks, Fontana and Monegato2017). As a result of this preparation, the lignin is removed, and only cellulose remains. Finally, the samples are rinsed to a neutral pH, dried, and weighed.
For wood samples with a radiocarbon age close to background, use of the ABA protocol alone is sufficient (Jędrzejowski et al. Reference Jędrzejowski, Michczyńska, Kłusek, Michczyński, Pawełczyk, Piotrowska, Wyss Heeb and Hajdas2024).
Bone samples
The bone preparation is conducted following a procedure described by Pawełczyk et al. (Reference Pawełczyk, Hajdas, Sadykov, Blochin and Caspari2022, Reference Pawełczyk, Niedziałkowska, Pawełczyk, Piotrowska and Sykut2024). The procedure begins with mechanical abrasion of the surface, cleaning in demineralized water in an ultrasonic bath, then drying and crushing in a hand mortar to ∼1 mm particles. The gelatine extraction is performed according to the Longin method (Longin Reference Longin1971), modified by Piotrowska and Goslar (Reference Piotrowska and Goslar2002), with the use of hydrochloric acid and alkali solution at room temperature. The bone is treated with 0.5 M hydrochloric acid to hydrolyze the mineral fraction, and the reaction was considered complete when the pH stabilized at <1, and no bubbles are visible. The residuum is rinsed and treated with 0.1 M sodium hydroxide for 30 minutes. Then, the sample is rinsed, and the residuum is subjected to gelatinization in an acidic solution (hydrochloric acid, pH = 3) for 17 h at 85°C. After gelatinization, the supernatant is centrifuged, and the sample is filtered through a pre-cleaned 9 mL Ezee Filter™ separator (Elkay), which are polypropylene tubes with sintered polyethylene filters. Finally, the sample is freeze-dried.
Before freeze-drying sample may be subjected to ultrafiltration. The sample is placed in ultrafiltration tube (Millipore Amicon Ultra-15), precleaned following the protocol of (Brock et al. Reference Brock, Ramsey and Higham2007). Then, it is centrifuged to collect the fraction >30 kD and remove the fraction <30 kD.
Organic remains
The organic remains are treated according to the ABA protocol, similar to one used in the wood preparation process, but the reagents used have a lower concentration. Firstly, the samples are treated with 0.5 M hydrochloric acid for 1 hr at 85°C. Then, they are rinsed with demineralized water until a neutral pH is reached. Afterwards, 0.1 M sodium hydroxide is added (1 hr, 85°C) to remove the humic compounds, followed by rinsing. Next, the samples are again subjected to reaction with 0.5 M hydrochloric acid for 1 hr at 85°C. After rinsing with demineralized water to a neutral pH, the samples are dried.
Carbonates
Carbonates are decomposed to CO2 in the vacuum line using 103% orthophosphoric acid, using the technique described by (Piotrowska Reference Piotrowska2013). The bottom part of a glass tube is filled with sample material and the glass tube fused at ca. 45° to the first, is filled with ca. 1 mL of H3PO4. After connection to a vacuum line the tube is pumped for ca. 1 hr. Afterwards, the tubes are tilted for the acid to reach the sample. The reaction is continued as long as bubbles are visible and can be accelerated by heating of the reaction tube. The high concentration of acid is intended to reduce the amount of water and thus speed up the pumping process of the vacuum system. The extracted CO2 is cryogenically purified and transferred into flame-sealed quartz tubes.
Liquid fuels
No chemical treatment is applied to liquid fuels (Baranyika et al. Reference Baranyika, Piotrowska and Michczyński2024). A volume of 2 µL is subsampled into a tin capsule for liquid samples and sealed using a sealing press. To prevent evaporation, these capsules are additionally closed in another capsule—the one for solid samples (Baranyika et al. Reference Baranyika, Piotrowska, Kłusek, Michczyński and Pawlyta2022, Reference Baranyika, Piotrowska and Michczyński2024; Gill et al. Reference Gill, Michczyńska, Michczyński, Piotrowska, Kłusek, Końska, Wróblewski, Nadeau and Seiler2022).
Graphitization and 14C concentration measurements
After chemical treatment, the dried fraction of material selected for measurement is weighed into tin capsules, combusted in a Vario Micro Cube Elementar elemental analyzer, and then graphitized using an AGE-3 system (Němec et al. Reference Němec, Wacker and Gäggeler2010b; Wacker et al. Reference Wacker, Němec and Bourquin2010c). The routine graphite mass is set to 1 mg of carbon. The CO2 from carbonate decomposition is transferred directly to AGE using a manual cracker, equipped with an additional drying agent. The oxalic acid (for OxII, C-7 and C-8) is weighed directly to tin capsules.
Before measurement, each graphite sample is pressed into a cathode. The radiocarbon concentration in the graphitized samples is then measured using the MICADAS accelerator mass spectrometer at the Gliwice Radiocarbon Laboratory (Figure 1), manufactured by IonPlus (Synal et al. Reference Synal, Stocker and Suter2007; Wacker et al. Reference Wacker, Bonani, Friedrich, Hajdas, Kromer, Němec, Ruff, Suter, Synal and Vockenhuber2010a). This compact device (3.2 m × 2.6 m) has a magazine that can hold up to 40 graphite targets. Typically, at least three backgrounds and five standards are included in a magazine containing samples of similar materials. Of these five standards, three are usually normalization standards (NIST Oxalic Acid II SRM4990C), and at least two are secondary standards (from IAEA). The latter are placed in the magazine to control the accuracy of the measurements and are analyzed as unknown samples (Wacker et al. Reference Wacker, Christl and Synal2010b).
Figure 1. The MICADAS spectrometer at the Gliwice Radiocarbon Laboratory.
The calculations of F14C, radiocarbon age and their uncertainties are performed in the BATS software (Wacker et al. Reference Wacker, Christl and Synal2010b), including the correction. The isotopic ratio of carbon to nitrogen (C/Nat) in collagen samples is determined based on the results from the Vario Micro Cube Elementar elemental analyzer, calibrated using aspartic (C = 36.09 ± 0.29%, N = 10.54 ± 0.06%) and glutamic acid (C = 40.77 ± 0.19%, N = 9.53 ± 0.12%) reference materials provided by Elemental Microanalysis.
Results of long-term measurements of reference materials and background samples
The results obtained for OxII, IAEA C-7, and IAEA C-8 (Clercq et al., Reference Clercq, van der Plicht and Gröning1997) at the Gliwice Radiocarbon Laboratory during the first two years of operation of the MICADAS are presented in Figure 2, for graphites weighing around 1 mgC. The average F14C value of the primary standard OxII is 1.3406 ± 0.0029 (n = 198; Figure 2a), while the results of the secondary standards are: IAEA C-7 (0.4950 ± 0.0028 F14C, n = 50; Figure 2b) and IAEA C-8 (0.1509 ± 0.0016 F14C, n = 42; Figure 2c).
Figure 2. The F14C values of a) OxII, b) IAEA C-7, and c) IAEA C-8 measured from September 2022 to July 2024 with the Gliwice MICADAS. For each reference material, the mean value and the band within the range of one standard deviation are marked.
The most common background samples used in the Gliwice Radiocarbon Laboratory are shown in the Fig. 3. The long-term mean F14C value for PhA is 0.0033 (n = 72) which corresponds to about 45,900 14C years. The other backgrounds mean values are as follows: 0.0039 for coal (n = 34), 0.0030 for Gliwice Marble (n = 11), 0.0043 for OLGA wood (n = 5) and 0.0034 for Liquid Fuel Background (n = 11).
Figure 3. The F14C values for background samples measured from September 2022 to July 2024 with the Gliwice MICADAS.
Intercomparison results
We received 17 samples of unknown ages for AMS dating (Table 1) from the Glasgow International Radiocarbon Inter-Comparison (GIRI) program. The only known information was the material type, and that samples I, J, and N were close to the background or the dating limit. Other information about the samples mentioned in this section is described in (Scott Reference Scott2003b, Reference Scott2003a; Scott et al. Reference Scott, Naysmith and Dunbar2023).
Table 1. GIRI samples. Abbreviations used to designate chemical preparation methods: A – hydrochloric acid, B – sodium hydroxide, Cl – sodium chlorite, S – pretreatment in the Soxhlet apparatus, UF – ultrafiltration. The radiocarbon results obtained at the Gliwice Laboratory compared with the mean values from the GIRI preliminary report (Scott et al. Reference Scott, Naysmith and Dunbar2023). Sample N is reported as Fm (without background correction). To facilitate comparison with the GIRI preliminary results, certain values are not rounded with the usual practice of reporting radiocarbon results. Coverage factor
$k$
is calculated between results obtained at the Gliwice Laboratory and GIRI preliminary results. The
$\left| k \right| \lt 1$
represents agreement at the 68.3% confidence level and results where
$\left| k \right| \lt 2$
represents agreement at the 95.4% confidence level. The agreement between results for all samples denotes a high level of consistency with the consensus results calculated for all laboratories participating in the GIRI program
The chemical pretreatment, dated fraction, and combustion mass for the GIRI samples are presented in Table 1. For samples N and J in the GIRI inter-comparison, described as close to the background limit, we used two preparation methods: the S+ABA+Cl (S – pretreatment in the Soxhlet apparatus; A – acid; B – base; A – acid; Cl – bleaching) protocol and only the ABA protocol. The first method was chosen to maintain consistency with other wood samples in the comparison. The second method was used because, in some cases where wood samples are close to the background limit, the amount of cellulose may be low, making cellulose extraction challenging. Thus, processing whole wood using the ABA protocol may suffice.
The results are listed in Table 1. 14C concentration in sample N was comparable to the background level. Therefore, we reported its concentration without background correction (as Fm value).
To show how our results correlated with those of other reports (Scott et al. Reference Scott, Naysmith and Dunbar2023), we calculated the F14C difference between the results obtained from the Gliwice Laboratory and the preliminary consensus values reported from the GIRI program by Scott et al. (Reference Scott, Naysmith and Dunbar2023), as shown in Figure 4. Additionally, we determined the coverage factor
$k$
(Table 1). The difference between reported F14C values varies from 0 (samples A, and O) to –0.0033 (sample L). A positive difference in value means that the radiocarbon content reported by the Gliwice Laboratory was higher than the mean for the GIRI program, while a negative result indicated the opposite.
Figure 4. The difference between the F14C of the GIRI samples measured in the Radiocarbon and Mass Spectrometry Laboratory in Gliwice (F14CGliwice) and the preliminary consensus values in GIRI program (F14CGIRI) in all participating laboratories (Scott et al. Reference Scott, Naysmith and Dunbar2023).
Discussion
The F14C value obtained for OxII and measured from September 2022 to July 2024 in the Gliwice Laboratory is consistent with the reference value 1.3407 ± 0.0005 (Mann, Reference Mann1983). Similarly, the C-7 and C-8 standards are consistent with the consensus values F14C: 0.4953 ± 0.0012 and 0.1503 ± 0.0017 (Clercq et al. Reference Clercq, van der Plicht and Gröning1997).
From September 2022 to May 2023, we used coal as background material. After testing PhA, we found that it is more stable and repeatable than coal. Consequently, since June 2023, PhA has been the main background (Figure 2). Other types of background samples are used in much smaller quantities because they provide a control background for specific types of samples or chemical preparation.
All our GIRI results are consistent with the preliminary consensus values reported by Scott et al. (Reference Scott, Naysmith and Dunbar2023). The analysis of the sign of the
$k$
coefficient as well as difference between the F14CGliwice and the F14CGIRI (Figure 4), show that for some samples the results obtained by the Gliwice Laboratory are slightly higher or lower than those reported by Scott et al. (Reference Scott, Naysmith and Dunbar2023). This indicates that the difference in the obtained values is only related to the random scatter of results and that no systematic error has occurred in the measurements.
The observed differences in radiocarbon concentration may result from measurement procedure. After several consecutive runs, trace carbon and cesium amounts can accumulate on the ionizer and lens in the ion source, leading to potential sample contamination, which may lead to higher background values, explaining the scatter in results, and increasing the F14C measurement background. To minimalize this influence, the ion source is cleaned after analyzing approximately 10 magazines or whenever the background level starts to rise.
For ensuring an appropriate quality of measurements in a typical setup, each AMS magazine includes 3 to 4 background samples, and the average background value measured for that magazine is subtracted from all other samples, including standards and unknowns, as detailed in Wacker et al. (Reference Wacker, Christl and Synal2010b). Similarly, measurement uncertainties are calculated according to Wacker et al. (Reference Wacker, Christl and Synal2010b).
Conclusions
In the summer of 2022, our laboratory installed a MICADAS spectrometer. From then until the summer of 2024, we have measured about 1800 graphite samples including samples <1 mg of carbon. The F14C values of the standards (OxII, C-7 and C-8) obtained by Gliwice Laboratory are consistent with their reference values. The primary measurement background we use is PhA. Its average radiocarbon content is 0.0033 F14C, which sets the limit of radiocarbon dating at the 45,900 14C years. Other materials used in our laboratory as background give the following average results: coal (0.0039 F14C), Gliwice Marble (0.0030 F14C), OLGA wood (0.0043 F14C) and Liquid Fuel Background (0.0034 F14C).
In 2021, the Radiocarbon and Mass Spectrometry Laboratory in Gliwice participated in GIRI. As a result, 17 samples of various materials (wood, cellulose, bone, humic acid, and barley mash) of unknown age were dated using the AMS technique. The radiocarbon results were compared with the GIRI preliminary report (Scott et al. Reference Scott, Naysmith and Dunbar2023), and results from the Gliwice Laboratory were consistent with those preliminary findings at a high-level of confidence.
Our laboratory carries out a comprehensive sample dating process, from sample chemical preparation, through combustion and graphitization, finishing with accelerator mass spectrometer measurements. The results outlined in this paper highlighted the effectiveness of the radiocarbon analysis techniques utilized at the Radiocarbon and Mass Spectrometry Laboratory in Gliwice. Hence, our laboratory is capable of dating samples across a wide spectrum ranging from contemporary to the limits of the radiocarbon method, achieving precision on par with that of other laboratories.
Acknowledgments
This study was supported by the Silesian University of Technology as part of the project 14/020/SDU/10-21-01. The acquisition of MICADAS within CEMIZ (Centre for Isotope Methods) project was financed by European Regional Development Fund (ERDF) with support from SUT internal funds. The open access publication fee is covered by project FESL.10.25-IZ.01-06C9/23-004.
Open access
