Introduction
Intercomparisons between radiocarbon laboratories are crucial for ensuring the accuracy and precision of measurements. Several papers present intercomparisons in different contexts (Dumoulin et al. Reference Dumoulin, Comby-Zerbino, Delqué-Količ, Moreau, Caffy, Hain, Perron, Thellier, Setti, Berthier and Beck2017; Pauwels et al. Reference Pauwels, Lamberty, Schimmel, De Bièvre and Günzler1998; Scott et al. Reference Scott, Naysmith and Cook2018; Shen et al. Reference Shen, Tang, Wang, Qi, Li, Wei, Sasa, Shi, Zhang, Chen, Qi, Wang, Zhou, He, Zhao and He2022; Takahashi et al. Reference Takahashi, Minami, Aramaki, Handa, Saito-Kokubu, Itoh and Kumamoto2019). Quality assurance and control measures involve systematic examinations and global comparisons between laboratories. The reproducibility and quality of the results of radiocarbon measurements require intercomparisons between samples prepared in different laboratories. Radiocarbon measurements involve complex procedures with varying sample pretreatment and processing methods (Scott et al. Reference Scott, Harkness and Cook2016, Reference Scott, Cook, Naysmith and Staff2019).
Standard protocols and common reference materials are used to improve laboratory procedures, reduce contamination levels, and enhance the precision and accuracy levels of radiocarbon dating results, which should be a priority (Adolphi et al. Reference Adolphi, Güttler, Wacker, Skog and Muscheler2013; Kuzmin et al. Reference Kuzmin, Fiedel, Street, Reimer, Boudin and van der Plicht2018; Quarta et al. Reference Quarta, Molnár, Hajdas, Calcagnile, Major and Jull2021) for exemple, in elucidating the chronology of archaeological and historical events (Cuzange et al. Reference Cuzange, Delqué-Količ, Goslar, Grootes, Higham and Kaltnecker2007), developing methods to meet industrial expectations (Varga et al. Reference Varga, Hajdas, Calcagnile, Quarta, Major and Jull2023), or inspecting food products (Quarta et al. Reference Quarta, Hajdas, Molnár, Varga, Calcagnile and D’Elia2022). Dating protocols are so significant that some researchers have assessed whether differences in obtained dates arise due to geophysical conditions or to calibration errors (Miller et al. Reference Miller, Lehman, Wolak, Turnbull, Dunn and Graven2013).
Despite quality control procedures, some issues may still occur, which are often best detected through independent interlaboratory comparisons. Continuously conducting proficiency tests is fundamental for maintaining user confidence in radiocarbon dating techniques and in laboratory results (Jull et al. Reference Jull, Pearson, Taylor, Southon, Santos and Kohl2018; Scott et al. Reference Scott, Naysmith and Cook2017, Reference Scott, Naysmith and Cook2018). The quality control process also ensures that sample handling procedures are contamination-free and suitable for routine protocols (Sironić et al. Reference Sironić, Bronić, Horvatinčić, Barešić, Obelić and Felja2013), thus preventing age discrepancies resulting from sample preparation (Szidat et al. Reference Szidat, Bench, Bernardoni, Calzolai, Czimczik and Derendorp2013).
The first sample preparation laboratory for 14C-AMS analysis in Latin America was installed in 2009 at Federal Fluminense University (UFF) (Anjos et al. Reference Anjos, Macario, Gomes, Linares, Queiroz and Carvalho2013). However, prior to the definitive installation of the NEC Single Stage Accelerator Mass Spectrometry system, many collaborations were performed between UFF and international institutions (e.g., Barbosa et al. Reference Barbosa, Buarque, Gaspar, Macario, Anjos and Gomes2004; Lima et al. Reference Lima, Macario, Anjos, Gomes, Coimbra and Elmore2002; Macario et al. Reference Macario, Gomes, Anjos, Carvalho, Linares and Alves2013). Currently, UFF dominates all sample preparation stages, creating, defining, and establishing sample preparation protocols and improvements until the graphitization stage is achieved (Bragança et al. Reference Bragança, Oliveira, Macario, Nunes, Muniz and Lamego2021; Macario et al. Reference Macario, Oliveira, Carvalho, Santos, Xu and Chanca2015; Oliveira et al. Reference Oliveira, Macario and Pereira2016, Reference Oliveira, Macario and Silva2020, Reference Oliveira, Macario, Carvalho, Moreira, Alves and Chanca2021). The Federal University of Bahia recently initiated sample preparation for 14C dating via accelerator mass spectrometry (Santos Reference Santos2020). This initiative has been rooted in the history of the university since 1970, when proportional counters were used to quantify 14C in samples (Lôbo Reference Lôbo1972; Costa Reference Costa1997; Flexor et al. Reference Flexor, Lôbo and Rapaire1972). In this context, the UFBA implemented a sample preparation laboratory for 14C dating to address a need in the Brazilian archaeological field and in other research areas, especially hydrology (Cisneros et al. Reference Cisneros, Raja, Ghilardi, Dunne, Pinheiro and Regalado Fernández2022; Docio et al. Reference Docio, Rasbold, da Silva, Parolin, Caxambu and Pinheiro2021; Duin et al. Reference Duin, Toinaike, Alupki and Opoya2015; Green et al. Reference Green, Green and Neves2016), assisting in graphite production and 14C research, particularly in the northern and northeastern regions of Brazil.
The aim of this paper is to present the initial results obtained from the AMS 14C facility at UFF as part of an intercomparison exercise conducted in collaboration with the sample preparation laboratory for 14C dating at the Federal University of Bahia (LAPA14C – UFBA).
Methodology
Vacuum line
The samples were prepared using a vacuum line to purify and capture the CO2. A schematic diagram of the graphitization line installed at the Federal University of Bahia is shown in Figure 1. This line has two CO2 inlets for each sample: (i) inorganic samples-CO2 pass through a needle after acidic hydrolyzation, and (ii) organic samples-CO2 from thermal conversion are inserted into a flexible tube, where the exhaust tube is broken. Then, the samples are introduced into the line. After sample insertion, CO2 is purified by passing through a trap where water vapor was retained. Subsequently, the CO2 is transferred to a known-volume trap with liquid nitrogen, following quantification before the samples are purified and directed to the graphitization tubes.
Figure 1. Schematic diagram of the UFBA vacuum line.
To determine whether the amount of CO2 is sufficient to form the minimum required quantity of carbon that could be measured in the AMS system, it is necessary to measure the pressure of the purified CO2 trapped in the known volume by the calibrating the vacuum line. For this purpose, calcium carbonate (CaCO3) samples were weighed to various values, evacuated, and reacted with excess phosphoric acid. The samples were prepared according to Oliveira et al. (Reference Oliveira, Macario, Carvalho, Moreira, Alves and Chanca2021). Physical treatment of the samples, which involves the removal of visible contaminants, was optional because only standard samples were analyzed.
Inorganic standard preparation
The carbonate standard samples, C1 (Carrara marble) and C2 (carbonate dissolved in water), were converted into CO2 by acid hydrolysis at room temperature using 1 ml of 85% H3PO4 injected with a needle into vacuum tubes. The carbonate samples were stored in a septum-sealed vial. The tube was left for the necessary time until the reaction was complete, and it was then returned to the vacuum line to transfer the CO2 through the needle. Schematic diagrams of the (a) organic and (b) inorganic standard preparation processes are shown in Figure 2.
Figure 2. Schematic diagram of the steps of (a) inorganic sample preparation and (b) organic sample preparation.
Organic standard preparation
The typical treatment used for organic wood samples, such as C9 (wood fossil) and CENA 1126 (shell samples, secondary coal standard), is the acid-base-acid (ABA) protocol (see Figure 2 (b) for illustration). The selected sample fragments were placed inside a Pyrex® test tube with a length of 13 cm. Then, 1.0 M HCl was added for reaction cycles of 2 hours at 90ºC to hydrolyze the inorganic fraction. The process was repeated until the supernatant became transparent and no further gas was produced. Sodium hydroxide (NaOH) with a molarity of 1.0 M was added for 1 hour at 90ºC until the supernatant became transparent (light-straw colored). The different acid and alkali reagents were rinsed with Milli-Q water at least three times to neutralize the solution pH. After the last acidification, the sample was rinsed five times with Milli-Q water to a neutral pH and left in a furnace at 90ºC for 2 hours until it was removed (Oliveira et al. Reference Oliveira, Macario, Carvalho, Moreira, Alves and Chanca2021).
Finally, organic samples, including C6 (sugarcane sucrose) and OXII (oxalic acid), which did not undergo chemical treatment, were taken to the vacuum line to be evacuated in quartz tubes. All organic materials were converted into CO2 by combustion. In the UFBA laboratory, sample combustion occurred in sealed quartz tubes containing preheated cupric oxide (Synth, 0.01% of the carbon compounds) and silver wire (Synth, 99.5% of the quantity) at 900ºC for 3 hours. Subsequently, 9 mm or 6 mm tubes (depending on the amount of sample) were used for regular-sized standard samples.
Graphite conversion
After CO2 extraction, both the organic and inorganic samples followed the same methodology for conversion into graphite. Graphitization was performed in a muffle oven at 550ºC in batches of independently torch-sealed tubes. The graphitization tubes consisted of outer borosilicate glass tubes (outer diameter of 9 mm and length of 15 cm), 30–35 mg of Zn (Aldrich 99.995%) powder, 10–15 mg of TiH2 (Sigma–Aldrich; 98% purity and 325 mesh) and an inner tube (diameter of 4 mm and a length of 5 cm) containing 5 mg of Fe (Sigma–Aldrich; 98% purity and 325 mesh) (Macario et al. Reference Macario, Oliveira, Carvalho, Santos, Xu and Chanca2015; Xu et al. Reference Xu, Trumbore, Zheng, Southon, McDuffee, Luttgen and Liu2007).
14C measurement
After obtaining the graphite, the samples were sent to LAC-UFF for analysis. At the LAC-UFF, measurements were performed in a SSAMS system built by the NEC that was especially designed for 14C measurements. The SSAMS is a single-stage system, indicating that no acceleration occurred after the stripping process. With a high voltage deck at 250 kV, a sequential beam injection line, tuning magnets, two off-axis Faraday cups, and a solid-state detector, this system was capable of measuring both 14C/12C and 14C/13C ratios (Linares et al. Reference Linares, Macario, Santos, Carvalho, dos Santos, Gomes, Castro, Oliveira and Alves2015).
δ13C measurements
Graphite of the reference samples C1, C2, C6, and OXII were analyzed at the Stable Isotopes Laboratory at UFBA using an EA–IRMS system (Elemental Analyzer from Costec coupled with the Delta-V Isotope Ratio Mass Spectrometer from Thermo Finnigan). Elemental analysis was performed to quantify the carbon contents in the graphite samples of the reference materials and to determine the isotopic values.
Approximately 0.5 mg of graphite was placed in tin capsules and introduced into the elemental analyzer. In the autosampler, the samples were processed individually in an oxidation furnace at 1020ºC in the presence of oxygen. The oxidation reaction products were directed to a reduction reactor operating at 600ºC. Consequently, the generated CO2 was analyzed and quantified. The isotopic values were calculated and determined based on the reference standards USGS-40 and USGS-41 from the International Atomic Energy Agency, and the mass values were calculated based on the reference acetanilide standard (Costec). The δ13C values are expressed (‰) relative to the international reference standard for carbon isotopes Vienna Pee Dee Belemnite (VPDB) according to the following equation:
$${{\rm{\delta }}^{13}}{\rm{C}} = {\rm{\;}}\left( {{{{R_{sample}} - {R_{standard}}} \over {{R_{standard}}}}} \right) \times {10^3}(‰)$$
where R=13C/12C. Ninety standards were analyzed during the measurements. The accuracy of the analytical measurement process was approximately 0.1‰, while the error in mass determination was approximately 5%.
Results and discussion
We needed a line pressure reference corresponding to approximately 1 mg of carbon for graphitization. Calcium carbonate (CaCO3) samples with different weights were evacuated and prepared with excess phosphoric acid. After the reaction, the line was calibrated using the resulting CO2 pressure readings of the samples in a known standard volume for the different reacted calcium carbonate masses hydrolyzed with excess phosphoric acid. Figure 3 shows the pressure values (Torr) versus the amount of carbon in milligrams. The linear adjustment of the data was performed following the formula Pressure(Torr)=61.7*Carbon(mg)+5.3; that is, approximately 67 Torr corresponds to 1 mg of carbon. The graphitization line had to be calibrated before using the vacuum line to determine or adjust the ideal amount of carbon in the sample to graphitize to optimize the measurement values in the AMS.
Figure 3. Pressure (Torr) versus the amount of carbon in milligrams.
The laboratory was swiped for the presence of 14C tracer contamination. Some laboratory locations were analyzed for contamination via 14C analysis. The pMC values of the swipe samples from the muffle, floor, centrifuge, exhaust hood, vacuum line, and oven were 2.58 ± 0.05, 6.47 ± 0.07, 5.84 ± 0.07, 4.28 ± 0.06, 6.34 ± 0.07, 4.61 ± 0.06 and 9.73 ± 0.09, respectively. The average swipe value obtained for all places was 5.7 ± 2.2. The swipe results indicated that the laboratory was in operation and that there was no 14C trace contamination.
The pMC values obtained for the inorganic (C1, C2) and organic (C6, C9, and CENA1126) references are shown in Table 1, as are the standard values. The reference materials used in this intercomparison were as follows. C1 is Carrara marble, and its pMC equal to 0.00. C2 was obtained from carbonate dissolved in water, and its pMC equal to 41.14. C6 is sugarcane sucrose, and its pMC equal to 150.61. C9 is a sample of wood fossil with a pMC ranging from 0.12–0.21. CENA 1126 is a secondary coal standard with a pMC value of 47.61±0.04. Finally, OXII is oxalic acid, and its pMC equal to 134.08. The standard SRM 4990C was used to calculate and normalize sample ages.
Table 1. 14C activity measured at LAC-UFF and prepared either at UFBA or UFF compared to the reference values
* Combustion was carried out at UFF, and graphitization was carried out at UFBA.
Some of the samples were prepared totally and partially at LAPA14C-UFBA and another part at LAC-UFF. The aim of this exercise was to assess whether contamination occurred at any stage of preparation at the UFBA. Part of the preparation process started at UFBA and was interrupted to continue the preparation until graphitization at LAC-UFF. Materials were initially prepared at LAC-UFF and finalized at UFBA. The results are summarized in Table 1 and could be directly compared to reference values. The obtained values closely aligned with the reference values, demonstrating reproducibility and quality and indicating minimal contamination during chemical and graphitization processes. Some samples were entirely prepared in Bahia, while others were partially prepared at this location. Specifically, these samples underwent combustion and ABA treatment at UFF, with graphitization conducted at UFBA. Some standards are highlighted (*) in Table 1. Additionally, as indicated in the last column of the table, further samples were entirely prepared at UFF.
Wood (C9) was prepared to evaluate the ABA, combustion, and possible contamination characteristics at this stage. Other standard samples, C1, C2, C6, and OXII, were investigated for possible contamination, reproducibility, and quality. Table 1 shows that the observed percent modern carbon (pMC) values obtained at UFBA exhibited more significant standard deviations than those obtained at UFF. Notably, IAEA-C6 (sucrose) demonstrated a deviation of 3.6 pMC for samples subjected to combustion and graphitization at UFBA. Conversely, the sample subjected to graphitization at UFBA and combustion at UFF displayed a minor deviation of 1.8, whereas that fully prepared at UFF exhibited a minimal standard deviation of 0.38 pMC. Hence, even sample preparation processes involving chemical treatments, material washing, and other procedures had to be executed meticulously to minimize the standard deviation. Concerning the pMC values of the IAEA-C9 standards, it was observed that the measurements prepared at UFBA exhibited increases of approximately 1–2 pMC compared to those prepared at UFF, along with greater standard deviations. This discrepancy likely arose due to the chemical processes involved in ABA preparation.
With another part of the graphite, replicates were prepared for measurements of the δ13C in the elementary analyzer coupled to the isotopic ratio mass spectrometer (EA–IRMS) to study the isotopic fractionation and yield of the graphitization reaction. The isotopic ratio measurements of materials C1, C2, C6, C9, and OXII yielded graphite contents of <50%. These graphite production processes were accompanied by isotopic depletion, with an average difference of −9‰. These values are shown in Table 2.
Table 2. EA–IRMS measurements for δ13C (‰) standard references and graphite values
Xu et al. (Reference Xu, Trumbore, Zheng, Southon, McDuffee, Luttgen and Liu2007) found 69% to 95% yields. A yield of more than 95% would be ideal, and achieving this percentage is our perspective. To achieve this, we will increase the accuracy of measurements in the steps before graphitization. Some factors that may have contributed to the low yield in the graphitization process, and which should be monitored more rigorously, are i) quality of the reagents used, ii) the gas purification stage, iii) the stability of the muffle temperature; iv) variations in the weighing of reagents; and v) fractionation during measure in AMS.
The δ13C values obtained at UFBA for the reference materials C1, C2, C6, C9 and OXII were 0.37‰ (n = 3), –17.7‰ (n = 4), –20.1‰ (n = 3), –40.1‰ (n = 5) and –17.2‰ (n = 4), respectively. These values, compared with the reference values in Table 2, showed isotopic discrepancies of –2.1‰ for C1, –3.1‰ for C2, –14.7‰ for C6, –15.5‰ for C9 and –0.6‰ for OXII.
Figure 4 illustrates that isotopic fractionation varied nonlinearly. According to Xu et al. (Reference Xu, Trumbore, Zheng, Southon, McDuffee, Luttgen and Liu2007), the δ13C of graphite could be –8‰ lighter than the reference value under these graphitization conditions using Zn/Fe. OXII was removed from the graph because its fractionation was less than 1‰. The highest apparent fractionations occurred with the organic samples C6 and C9, followed by the inorganic sample C2 compared with their reference isotopic values. Regarding the standard value of C2, the combustion of carbonate was not complete, causing a decrease of 9.5‰ relative to the reference value. For C6 (sucrose) and C9 (wood), incomplete combustion could have generated fractionations of 9.3‰ and 16.2‰, respectively. Another factor still under investigation is that the muffle heating ramp could be influential. Macario et al. (Reference Macario, Oliveira, Carvalho, Santos, Xu and Chanca2015) reported that depending on the quantities of reagents and conditions (related to time and temperature), the difference reached 8‰. According to separate analyses, materials C2, C6, and C9 showed decreased yields. A likely cause of this phenomenon was the presence of water in the graphitization reaction, which produced methane and decreased the yield.
Figure 4. Plot of the δ13C of graphite versus known δ13C of the reference materials. This figure shows that fractionation is not constant.
Conclusion
Using the IAEA standards C1 and C2 allows the study of possible contamination sources that originate from the hydrolyzation process (introduction of phosphoric acid) and the addition of CO2 to the vacuum line. Using standards IAEA-C9, IAEA-C6 and OXII (SRM 4990C) enables the verification of CO2 preparation by the combustion of organic samples. Concerning C9, ABA chemical pretreatment is considered. OXII is used for normalization for all samples during the 14C AMS measurement, regardless of the material type. The sample preparation laboratory for 14C dating via the AMS technique at UFBA is set up to process specimens for radiocarbon dating. Concerning the errors associated with the measurements, the UFBA laboratory must enhance the accuracy of the results by refining both the chemical preparation and graphitization processes to prevent contamination. Such contaminants may arise during handling procedures or be linked to specific reagents utilized. The measurements of the stable isotopic ratio of materials C2, C6, and C9 provide results for the graphite yield (<50%), accompanied by isotopic depletion. To improve the graphite yield and consequently decrease isotopic fractionation, the quantities of reagents, humidity control, time, and temperature of the muffle furnace during the graphitization process need to be evaluated to optimize the UFBA protocol. The LAC-UFF protocol can be performed as an initial step; thus, it should be optimized for LAPA14C-UFBA.
This work shows the results of the steps performed for intercomparison and the results obtained from the standards analyzed in the intercomparison process. Initially, in 2019, when the sample preparation laboratory was installed, the samples were prepared in the UFBA. The graphite produced was measured in 2020 in the LAC-UFF. In this paper, we demonstrate our ability to produce graphite and show that the results of the standard graphite samples are consistent with those of the graphite samples produced by LAC-UFF.
Acknowledgments
The authors would like to thank the Brazilian financial agencies CNPq (Brazilian National Council for Science and Technology) 307771/2017-2, 309412/2019-6 and 317397/2021-4, INCT-FNA 464898/2014-5, FAPERJ (Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro) E-26/201.320/2022 and E-26/200.540/2023, and FAPESB (Research Support Foundation in the State of Bahia) 3931/2014 for their support.
