First Check and Typical Physical Pretreatment

All samples, regardless of material, are examined under a binocular loupe. Working with MICADAS, i.e., with quantities of carbon as small as a few tens of micrograms, requires great precautions. A hair of a sweater, one millimeter of hair, is 5 µg of carbon. It might lead to a rejuvenation of ∼2500 ¹⁴C years on a 100 µg C sample of 30 kBP or of ∼200 ¹⁴C years on a 100 µg C sample of 10 kBP. Moreover, it turns out that examination at this scale sometimes allows the material to be dated to be discretized: a carbonaceous ensemble may turn out to be a mixture of charcoal and unburned seeds, or plant excrement may contain µcharcoal that could potentially come from upper stratigraphical layers and would be better dated separately. This first examination allows for the correct selection of the dating support. When needed, mechanical abrasion is performed using a Dremmel® tool.

Charcoal and Wood (Bulk) Preparation

Acid-alkali-acid (AAA) treatment is commonly performed using HCl 1N and NaOH 0.1N. The chemistry is performed in a glass centrifuge tube, and solution removal is done with a Pasteur pipette.

Charcoal is typically treated at room temperature, while wood is cleaned at high temperatures to allow the small bubbles to work their way into the wood’s pores and remove any contaminants more effectively. Rinses are repeated until the pH of the water is restored to the one of ultrapure water. The rinsing after the basic treatment is the most critical, as the subsequent acid treatment could precipitate “organic contaminant residues” (formerly called humic acids) and lock them in the sample mass. The duration and the number of steps differ with the sample. Several short alkaline treatments are preferable to one long treatment in the same solution as at the start. As soon as the color changes, the solution is removed, or collected to date humic acid, and replaced by a new one, sometimes at a lower concentration and after a rinsing step if the sample has reacted strongly. Samples that are a priori fragile are treated by gradually increasing the concentration of NaOH. The specific case (unbreakable ionic bond formed during the alkali step between some functional groups of organic sample and bicarbonate from modern carbon dissolution) reported by Hatté et al. (Reference Hatté, Morvan, Noury and Paterne2001a) remains very rare.

The charcoals pigments of rupestrian paintings have long been investigated by our lab (Valladas et al. Reference Valladas, Cachier, Maurice, Dequiros, Clottes, Valdes, Uzquiano and Arnold1992) and are the focus of attention. They are small and very fragile and are often associated with mineral particles and other material. The general line of the chemical treatment remains the same, i.e., AAA treatment, but the chemical concentration is adapted, and most often reduced. As the residue after chemistry sometimes still contains minerals, it is difficult to assess a priori the quantity of carbon that remains available for combustion and hence to divide it correctly between several capsules to be sent to the EA-GIS system, i.e., by ensuring ∼100 µg of carbon per capsule. In this case, the sample is oxidized with pure oxygen on the “µline” where the gas obtained can be conditioned in one or several µtubes of ∼100 µg C each to optimize the measurement.

To transfer the sample into the quartz tubes required for the µline, we use a specific container. It is a centrifuge tube extended by a quartz tube of about 1 cm, forming an excrescence (Figure 3b). The sample is directed with a minimum of water in the bottom of the excrescence. A very fine metal rod with a twist at one extremity is plunged into the bottom of the tube. The whole is frozen in liquid nitrogen. The water forms a block of ice, trapping the sample and the metal rod. By slightly heating the excrescence with the hand, the ice block melts a little and it is possible to extract it from the tube by pulling on the metal rod. The ice and the sample can then be transferred in a combustion tube of the µline. The sample is then dried in the combustion tube in the vacuum oven.

Figure 3 Some photos to illustrate the LSCE ¹⁴C laboratory: (a) pigment and varnish in a pre-combusted Aluminum cup; (b) selection of 200 µg of colophony varnish in a tube with excrescence; (c) collagen in centrifugation tubes from bones after a modified Longin protocol; (d) two Erlenmeyers, each with 15 samples and 5 reference materials in individual disposable Teflon bags; (e) set of hard steel microchisels with different sizes of tool tip 0.120 mm, 0.25 mm, 0.5 mm and handle; (f) automatic balance (the red arrow points to the interchangeable head (one head per reference material or iron), the red to the vibrating motor, the green to the container holder (here a AGE3 reduction tube).

The reference materials we use the most frequently are listed in table 1. We also punctually use other charcoal and wood reference materials made available by the intercomparison exercises (e.g., SIRI H). The results we obtained over the last two years on our blank reference materials (the “AfSud” charcoal and three different woods: “El Akarit”, “Chiloé”, FIRI A) are shown in Figure 4. To illustrate our reproducibility, individual results of two reference materials, a charcoal (Gif1423) and a wood (SIRI E) are shown in Figure 5.

Table 1 Results obtained for the most frequently used reference materials with chemistry at LSCE. Data are from the 1.1.21 to 31.10.22 period. The first seven columns give the identification of the materials (name, nature, origin and expected age). They were mostly made available either by intercomparison exercises or from the ß-counting laboratory samples archive. (a) is for Scott et al. Reference Scott, Naysmith and Cook2017; (b) is for Scott et al. (Reference Scott, Cook and Naysmith2010); (c) is for F¹⁴C retrieved from IntCal20 (Reimer et al. Reference Reimer, Austin, Bard, Bayliss, Blackwell, Ramsey, Butzin, Cheng, Edwards and Friedrich2020); (d) is for F¹⁴C retrieved from Bomb21 (Hua et al. Reference Hua, Turnbull, Santos, Rakowski, Ancapichún, De Pol-Holz, Hammer, Lehman, Levin, Miller, Palmer and Turney2022). The following columns are for the results obtained at LSCE lab: the analyzed fraction and the results according to the measurement mode (solid or gas sources) and the introduction methods (EA-GIS, CHS-GIS or cracking-GIS). Only samples larger than 400 µg C for the solid source and larger than 60 µg C for the gas sources are shown here. For every modality, the number of analyzed targets, the F¹⁴C ± 1 sigma, the corresponding age in BP are provided. If the reference material is associated to a finite expected F¹⁴C, the z-test result between the expected value and our results is provided: (*) is for a 1-sigma agreement, (**) is for a 2-sigma agreement, (***) is for a 3-sigma agreement.

Figure 4 F¹⁴C (±1σ), ranked in ascending order, obtained on blank reference materials for the 1.1.21–31.10.22 period. The figure shows the results obtained on the solid source (circle) either after AGE3 or Gégé graphitization, the results obtained on the gas source through EA-GIS (star) and tube cracking-GIS (diamond) introduction after either the µline or the semi-automated carbonate line. Only samples larger than 400 µg C for the solid source and larger than 60 µg C for the gas sources are shown here. Results are for the IAEA-C1 carbonate, the AfSud charcoal, the A2 mammoth bone collagen and cellulose and bulk organic matter of “El Akarit”, “Chiloé”, FIRI A woods (all together). We handle three blank woods that were preserved in different environments and thus were subjected to different contaminations. This makes it possible to better match the blank to the specificities of the sample set.

Figure 5 F¹⁴C (±1σ), ranked in ascending order, obtained on three reference materials commonly used at LSCE, for the 1.1.21 – 31.10.22 period. The figure shows the results obtained on the solid source (circle) after AGE3 (organic material) or Gégé graphitization (carbonate) and the results obtained on the gas source through EA-GIS (cross) introduction or cracking-GIS (diamond). Only samples larger than 400 µg C for the solid source and larger than 60 µg C for the gas sources are shown here. Results are for the IAEA-C2 carbonate, the Gif1423 charcoal, SIRI E wood and the VIRI F bone collagen.

Cellulose Extraction

Cellulose extraction was not a widespread activity in the lab until we decided to participate in the community effort to provide more and more annual ¹⁴C records from tree rings. It was then no longer possible to use our conventional, time and labor consuming procedure. We therefore investigated another solution. Our new procedure, called “batch procedure”, derives from the protocol used for stable isotope measurements at LSCE which in turn derives from the original protocol of Leavitt and Danzer (Reference Leavitt and Danzer1993). We also followed some recommendations made by Southon and Mangana (Reference Southon and Magana2010) and added our own “touch”.

The chemistry is done in a 1.5 L Erlenmeyer flask in which the samples and standards are pooled, packed in an individual Teflon bag. The protocol is divided into three main steps: removal of water-soluble compounds, removal of lignin by a two-step oxidative delignification and removal of hemicellulose by alkaline extraction.

1. o From 7 to 8 mg of wood (sample and reference material) are cut into small chips. The mass depends on the species which can show different extraction yields.

2. o The single use Teflon bag is made of a 47mm Teflon disc with 10-µm pores, a Teflon label with engraved number and Teflon string.

3. o The Teflon bags are boiled for 6 hr in ∼1 L ultra-pure water in a 1.5 L Erlenmeyer. The water level is adjusted to ensure at least ∼0.5 L of water in the recipient.

4. o The Teflon bags are then transferred into a 1 L Erlenmeyer, covered with watch glass, filled with 100 mL of ultrapure water, 2 mL of 0.5N HCl and 100 mL of 10%wt NaClO₂ where they stay overnight at 70–80ºC. The volumes are defined to reach the right level of acidity and oxidizing strength.

5. o The following morning 25 mL of water, 10 mL of 0.5N HCl, and 25 mL of 10%wt NaClO₂ are added. 10 mL of 0.5N HCl and then 2 mL of 12N HCl are added after 2 and 2 extra hours. This is to maintain an acidic and oxidizing environment throughout the reaction, and overnight at 70–80ºC.

6. o The following morning, after having been rinsed several times with ultrapure water using ultrasound, the Teflon bags are immersed in 1M NaOH for 1 hr at 70–80ºC, rinsed again, immersed in 0.5N HCl for 1hour at 70–80ºC and rinsed again. Rinsing must be sufficiently repeated for the pH of the water coming out of the bags to be at the ultrapure water pH value.

7. o Samples within the Teflon bag are dried in a vacuum oven overnight and kept in a dry cabinet until analysis.

Our protocol allows the preparation of up to fifteen samples and five standards at the same time, without cross-contamination, thus saving time in handling and reducing chemical consumption. It lasts 3 days. The reference materials we use to check the reliability of our protocol are the same as for wood bulk.

Carbonate Preparation

Before being installed either on “BCA” or CHS, carbonate is pre-treated to remove any possible recrystallisation. In a generic way, the chemical treatment can be summarized as “HNO₃ leaching” (Tisnérat-Laborde et al. Reference Tisnérat-Laborde, Poupeau, Tannau and Paterne2001) but in practice, it differs according to the sample type: thin or thick biogenic structural carbonate, biogenic excretion carbonate or geologic carbonate.

Thin biogenic structural carbonate, such as that from pteropods or foraminifera and especially benthic foraminifera, is treated with caution. Shell or test may be crushed to allow efficient chemical treatment on all surfaces. If required, such samples are first cleaned in ultra-pure water to get rid of sediment. The duration of this step is kept as short as possible as with a pH of 5 ultrapure water begins to destabilize carbonate (pKa₁ ∼ 6.4). A weak acidic and oxidizing treatment is then done with 0.01N HNO₃. A rinsing step is applied on the largest samples but not to the smallest ones. Samples are then connected to the “degassing bench” to be dried on line. The HNO₃ treatment is applied within the vial for samples in CHS.

Thick biogenic structural carbonate, such as coral and mollusk shell, is first pre-cleaned by sand blasting until elimination of secondary crystallization in skeleton pores or surficial contaminants (Paterne et al. Reference Paterne, Ayliffe, Arnold, Cabioch, Tisnerat-Laborde, Hatté, Douville and Bard2004). Once rinsed and ultrasonically cleaned, the sample is crushed in an agate mortar or sampled at a regular spacing with a MicroDrill. The chemical treatment is similar to those described above but may be longer or be performed with a higher concentration of HNO₃, depending on the sample (size and crystalline structure).

Calcium carbonate from plant parenchyma cells is very fragile, often presented in microsphere clusters, and is treated with the same precautions as benthic foraminifera.

Biogenic excretion carbonate such as earthworm granules (Moine et al. Reference Moine, Antoine, Hatté, Landais, Mathieu, Prud’Homme and Rousseau2017) and geological carbonate such as speleothems show a more robust structure and HNO₃ leaching is done with 0.1N HNO₃ with a longer contact time. The risk of losing too much targeted material is much lower and the procedure is thus intensified to ensure total dissolution of potential recrystallisation and more efficient surrounding organic matter oxidation. Particular attention is paid to speleothem powders resulting from a MicroDrill, which can lead to heating of the carbonate crystalline structure and therefore contamination by atmospheric ¹⁴CO₂ during sampling.

The most frequent reference materials are shown in Table 1. They are completed with a coral blank and a speleothem blank that we did not use during the period summarized in the table. We also ask sample providers for a blank from the marine or lacustrine core to be studied. It consists of carbonate from the same species as the series to be dated but that is beyond the ¹⁴C- dating limit and if possible, even older than 100 kBP. This series is currently expanding to provide a better temporal coverage and to be closer to the carbonate structure of samples. Note that if the reference material is carbonate but not of the same origin as the sample series, it undergoes the same chemical treatment as the samples. See Figure 4 and 5 for results on IAEA-C1 blank and IAEA C2, respectively.

Organic Matter from Soil and Sediment

The preparation of soil and sediment differs according to the objectives of the study, i.e., whether an understanding of paleoclimatology or of the carbon cycle is sought. Amino acid and amino sugar are important components of the carbon cycle and must be preserved during the sample processing, whereas they can be considered as contaminant from modern microorganisms growing on the sediment when it comes to paleoclimatology.

Sediment and soil samples for paleoclimatological purposes are decarbonated if necessary, according to Gauthier and Hatté (Reference Gauthier and Hatté2008). Particular care is taken during the settling phase (acid step and rinsing) to avoid losing the fine fraction which is the principal support of organic carbon, and the drying temperature is carefully regulated to avoid destabilizing organic matter that has been made more fragile by chemical treatment.

Current knowledge on organo-mineral interactions shows that the organic molecules pile up on each other with the mineral surface as a basic support (e.g., Kleber et al. Reference Kleber, Sollins and Sutton2007). If the clays in the sediment have had another life before being deposited at the core site (fluviatile or lacustrine deposits, loess…), it is likely that the mineral-bounded organic matter is not contemporaneous with the event to be dated but is a phantom of the previous context. If the transit time between resuspension and deposition of mineral grains is not long enough or is not conducted under sufficiently oxidizing conditions, the original mineral-bounded organic matter is not eliminated by biological consumption or oxidation and will be part of the bulk sediment organic matter, together with the organic matter that results from autotrophic production at the coring site. When subjected to an acid treatment, some organic matter in the periphery will be eliminated. These are mainly amino acids and amino sugars, which are the constituents of what was formerly called “humic acids”. “Last in, first out”, these eliminated organic matters are often the most recent. They leave behind a residue with an apparent age older than the initial total organic matter, which we call “perfect bulk”. By adding a basic treatment, the chemistry attacks deeper layers. The residue after chemistry, which used to be called “humin”, has an even older apparent age compared to the original age. That is why the previously named “humin” is very often older than “humic acid” and why the (real) “perfect bulk” is in between. So, when applying an alkali step, we caution that the fractions obtained in the chemistry room are unlikely to fit the different possible sources of carbon, but that the difference between ¹⁴C datings may just give an idea of the range of ages expected in the mixture.

Based on this new knowledge of organo-mineral interaction, loess dating as previously advocated (Hatté et al. Reference Hatté, Pessenda, Lang and Paterne2001b) is re-evaluated according to the geomorphological context. The protocol works well in the Rhine Valley where dust transport (fine fraction) is long enough to allow almost complete oxidation of the organic material that was originally associated to the dust. The loess organic matter, in this context, is thus that of the plants that trapped the dust. Hatté et al.’s (Reference Hatté, Pessenda, Lang and Paterne2001b) protocol cannot be used as such, however, for dating the Great Plains loess section (USA) because the moraines, which are the source of dust, are too close to allow a high predominance of organic matter contemporaneous with loess formation. Rather, it reflects both trapping of moraine-trapped ancient organic matter and loess formation (unpublished data).

Carbonaceous sediment and soil samples for carbon cycle understanding purposes are decarbonated in capsules without any rinsing step or removal of aqueous solution. Amino-acid and amino-sugar should be kept within the sample and the soluble part that goes into the acid solution must remain part of the analyzed sample after the leaching. Sediment or soil carbonate content is evaluated thanks to two carbon content and δ¹³C measurements: one on bulk sediment, the second on sediment leached in the capsule. Decarbonation is done under a binocular loupe. Based on the carbon content evaluation, the amount of material to achieve a proper δ¹³C evaluation in the best conditions is weighed in two Sn capsules. One is closed, while the other one is placed under a binocular loupe where decarbonation can be closely monitored: 10 to 20 µL of 1N HCl is added to the sediment. If the amount of sediment is large, water can be added to wet the sediment and ensure proton diffusion throughout the whole sediment. The 1N HCl addition is repeated if bubbling or suspicion of bubbling is noticed. About 30 min later, an additional 10 µL of 1N HCl is added. Capsule and solution are frozen then freeze-dried and closed. Decarbonation is then performed as soon as possible before the IRMS measurements. 1 to 2 days delay can be managed by storing the capsules in a dry cabinet under dry neutral gas flow or by leaving them in the freeze-dryer. HCl salt is hydrophilic and tends to capture water to reform liquid HCl that reacts with Sn to form SnCl and fragilize the capsule. Based on carbon content and δ¹³C before and after the decarbonation, the mass of material to be sampled to obtain 100 µg of C for EA-GIS or 1mg of C for AGE3 and the volume of 1N HCl to be added for complete decarbonation is evaluated. We repeat decarbonation in the capsule as previously done. The volume of added HCl is restricted to a minimum to bring three folds of the required protons. This is done to limit the quantity of Cl^–, which is a poison for graphitization and makes the capsule fragile and less easy to manipulate.

Reference materials do not exist for soil and sediment samples. We thus use standards without chemistry to check the combustion, reduction, and measurement.

Bone Treatment

Bones are prepared in two ways: (1) a modified Longin protocol to extract pure collagen, and (2) the ninhydrin protocol to extract one specific carbon from the proteins embedded in the bone structure (Tisnérat-Laborde et al. Reference Tisnérat-Laborde, Valladas, Kaltnecker and Arnold2003). The latter requires more material than the former and is much more time-consuming. Collagen extraction is therefore the most common protocol used in the lab. It derives from the original Longin protocol (Longin Reference Longin1971), appropriated a long time ago by the Gif ß-counting lab and adapted to the small amounts of material now run on ECHoMICADAS, benefiting from tips of our sister labs (Cersoy et al. Reference Cersoy, Zazzo, Rofes, Tresset, Zirah, Gauthier, Kaltnecker, Thil and Tisnerat-Laborde2017; Richardin et al. Reference Richardin, Porcier, Ikram, Louarn and Berthet2017).

Prior to any treatment, the bone organic carbon content is evaluated for nitrogen content. We typically run samples with nitrogen content higher than 0.5%wt. Using the Ambrose (Reference Ambrose1998)’s relationship: %C_th = 2.7 %N + 1.4, this corresponds to a theoretical organic carbon content higher than 2.7%wt.

The chemical treatment follows the cleaning step done to ensure the elimination of any material that may contain proteins and polypeptides. The bone sample is then crushed to a fine powder using a glass mortar and pilon; these are preferred to a planetary mill to restrict the risk of contamination and loss of powdered sample adhering on the bowls. The first chemical step consists in decarbonation of the bone powder with 0.6N HCl. Acid is regularly added and pH checked to ensure a pH=1–2 environment. The leaching is done at 4°C. This step can last for 2 to 4 days. The bone powder is then rinsed with ultrapure water until pH=5. The rinsing steps are performed with centrifugation. Particular care is taken during the solution removal step with a Pipette Pasteur to be sure not to remove proteins that may already be outside the mineral structure. An alkali step follows with 0.01 to 0.1N NaOH at room temperature for 30 min or at 4ºC for 2 days depending on the sample. Several short alkali treatments may be required for some bones, especially those from a humid environment and this is preferred to a long unique alkali step. A rinsing step to retrieve a pH of 5–6 then follows. The next step consists in protein hydrolysis. It is done with 0.001N HCl (pH=3) at 80–90°C for about 12 hr. The acidic solution contains the collagen proteins. If the sample treatment yields a lot of mineral residues (phosphate?), the solution is filtered on a pre-combusted GF/F filter folded in a glass funnel or is filtered using a syringe equipped with a pre-combusted APFC filter. As filtration often results in a loss of material, it may be decided to carry out the separation between collagen and residue after the freeze-drying. The solution is thus left as it is, frozen and freeze-dried. The separation between clean collagen filament and the powdered residue is then done manually under a binocular loupe. The carbon and nitrogen content of the collagen, a white to yellowish fluff, is then analyzed to specify its purity. δ¹³C and δ¹⁵N are not systematically measured and are only performed for a specific purpose. If there is enough material and scientific or analytical interest, Fourier-transform infrared spectroscopy (FTIR) evaluation can be done.

The reference materials in use are shown in Table 1, the individual results on blank in Figure 4 and on VIRI F in Figure 5.

Organic Carbon from Water Samples

Other Materials Considered for ¹⁴C Dating

Dissolved Inorganic Carbon (DIC) and specific compounds are also run at LSCE. No chemistry is required prior to DIC extraction; the modus operandi can be found in Coularis (Reference Coularis2016) deriving from Bard et al. (Reference Bard, Arnold, Toggweiler, Maurice and Duplessy1989). The radiocarbon dating of specific compounds began in 2008 at LSCE with the acquisition of a Preparative Gas Chromatography with Fraction Collector (Prep-GC-FC) used for the ANR Dynamos project (Mendez-Millan et al. Reference Mendez-Millan, Nguyen Tu, Derenne, Zeller, Derrien, Egasse, Balesdent and Hatté2014). It was involved in the recently launched intercomparison exercise (Casanova et al. Reference Casanova, Knowles, Mulhall, Sikora, Smyth and Evershed2021).