3.1. Radiocarbon Results

The complete set of CaCO₃ precipitate ¹⁴C results as well as those of reference materials are reported in Tables S1 and S2 (supplementary material). Target sizes (in mg C) vary significantly, e.g., 0.22 to 0.97 mg C. Notably, minimum CO₂ yields were obtained for pure calcite and coral standard, where particles loaded in quartz tubes were significantly coarser than the fine powders of Nydal’s CaCO₃ precipitates.

Literature searches show that carbonate decomposition should occur at temperatures as low as 850°C (Stern Reference Stern1969). Even so, absolute pressure in the calciner environment (Maya et al. Reference Maya, Chejne, Gómez and Bhatia2018), size of the CaO crystallites formed under heat, mineral and organic impurities (Galwey and Brown, Reference Galwey and Brown1999) can also play a role in CaO reactivity and CO₂ release. Here, carbonate samples were loaded in an evacuated quartz tube with no partial pressure, until CuO was finally activated. Reactivity of CaO and its sintering against inner quartz tubes walls may have diminished the access to carbon during the CaCO₃ decomposition reaction. Plus, we cannot rule out CO₂ recombination with the CaO leftover and/or CuO, once quartz tubes returned to room temperature after combustion. We also used two temperature settings for sealed quartz tubes combustion and time durations (1000°C per 6 hr and/or 900°C over 3 hr), and distinct sample weights (8–12 mg; see Tables S1 and S2). All of the above, may have affected the CaCO₃ decomposition response during sealed tube combustion step.

Nonetheless, lower graphite mass (as mg C) and its impact on ¹⁴C results were addressed by background mass balance correction, using blanks treated in the same fashion as the other samples. For that, we used 7 blanks (2 calcites and 5 USGS coals) of different masses (Table S2). Their ¹⁴C concentration (relative to modern standard) and masses (as mg C) were plotted to derive the constant mass of contamination introduced during the entire sample processing (step-combustion, seal-off combustion, graphitization, and pressing) plus measurement at the spectrometer (Figure S1). A consistent value of 1 μg C modern blank was found, as ¹⁴C results of calcites and coals overlapped. Thus, this constant blank mass was used for ¹⁴C results background correction, following the equations in Santos et al. (Reference Santos, Southon, Griffin, Beaupre and Druffel2007).

As mentioned earlier, we also determined AMS online-δ¹³C values derived from ¹²C and ¹³C loop-by-loop measured from the same graphite targets (not shown here). The AMS online-δ¹³C values based on Nydal’s CaCO₃ precipitates were significantly lighter (e.g., δ¹³C = –20‰ in average) than what one would expect for a CaCO₃ matrix. The NaOH-static method, the one used by Nydal to capture atmospheric CO₂ during several days, tend to introduce large isotope fractionation in ¹³C (e.g., –15 to –25‰, according to Turnbull et al. [Reference Turnbull, Mikaloff Fletcher, Brailsford, Moss, Norris and Steinkamp2017]) due to the high alkalinity of NaOH. In Nydal and Lövseth (Reference Nydal and Lövseth1983), the CaCO₃ precipitates δ¹³C values reported ranged from –25 to –27‰ and were used for the intended purpose of correcting ¹⁴C data. When δ¹³C was not measured, its value was then estimated based on multiple samples (see details in Nydal and Lövseth Reference Nydal and Lövseth1996). Since at KCCAMS/UCI, isotopic fractionation correction to ¹⁴C data is performed by using the AMS online-δ¹³C values, the effect of chemical reaction shifts, machine and size dependence (if any) were completely addressed and removed. For more details on data analysis, see Santos et al. (Reference Santos, Southon, Griffin, Beaupre and Druffel2007, Reference Santos, Alexandre, Coe, Reyerson, Southon and De Carvalho2010).

Overall, duplicated ¹⁴C results as well as those of reference materials support the reliability of the measurements. The carbonate samples, L26 and FIRI-C, overlap in ±2σ of each other (Table S1). Paired ¹⁴C results of the samples L22B, L24, and L28 showed larger differences, e.g., more than ±2σ. Note that carbonate samples did not undergo chemical leaching by a weak HCl treatment (Santos et al. Reference Santos, Southon, Druffel-Rodriguez, Griffin and Mazon2004) before step-combustion, and this may have played a role. Lower temperature step-combustion is expected to remove just OCs, and not secondary carbonates. Another possibility is that the CaCO₃ precipitates contamination is not homogeneous, and distinct initial combusted masses resulted in uneven ¹⁴C signatures (see further discussion in 3.4 section). Replicated ¹⁴C results of the recalcitrant fraction of the Rice char overlap with each other as well as accurately reproduce its expected consensus ¹⁴C value (Table S2). Paired FIRI-C turbidite ¹⁴C results also overlapped with each other.

In figure 2 we compared UCI ¹⁴C results of CaCO₃ precipitates as F¹⁴C with those reported by Nydal and Lövseth (Reference Nydal and Lövseth1983) and those recently measured at NTNU by Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023). In the latter, we chose the set of ¹⁴C results from CaCO₃ powders subjected to just flash-combustion using an elemental analyzer for comparisons. Both NTNU and Nydal’s values were also reproduced in Table S1 to serve as reference. For a complete overview of chemical treatments attempts to remove contaminants from the CaCO₃ powders stored at NTNU, please refer to Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023).

Figure 2 Radiocarbon results as F¹⁴C values for both the 1963 (panel A) and 1980 (panel B) series from different works are shown. Note that panels show very distinctive y-axis scales. Color labels inserted in distinct plot areas discriminate the results from UCI (this study), NTNU (Seiler et al. Reference Seiler, Grootes, Svarva and Nadeau2023), Nydal (Nydal and Lövseth Reference Nydal and Lövseth1983), Fruholmen (Nydal and Lövseth Reference Nydal and Lövseth1996) and/or Schauinsland (Levin et al. Reference Levin, Kromer, Schoch-Fischer, Bruns, Münnich, Berdau, Vogel and Münnich1985).

For simplicity’s sake, in Figure 2 we averaged the UCI ¹⁴C results of CaCO₃ precipitates produced in duplication and used the greater error of the two, individual error or standard deviation of the paired ¹⁴C results, as error bar. According to Nydal’s runlog books, the 1963 samples measured here, and by Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023), were indeed measured by ¹⁴C before moving to storage in vial type 1 (Figure 1). Regarding the CaCO₃ precipitate 1980 series, they were not measured by Nydal in the past. Following Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023) approach, atmospheric ¹⁴C values of Schauinsland (48ºN) (Levin et al. Reference Levin, Kromer, Schoch-Fischer, Bruns, Münnich, Berdau, Vogel and Münnich1985) are loosely used here, as reference for the ¹⁴C results of the CaCO₃ precipitate 1980 series. We also added data from Fruholmen (71ºN) (Nydal and Lovseth 1996) during the same period, as a way to bracket the ¹⁴C results of Lindesnes (58ºN) site.

For the most part, current CaCO₃ precipitate ¹⁴C results obtained at UCI (this study) and at NTNU (Seiler et al. Reference Seiler, Grootes, Svarva and Nadeau2023) are in agreement with each other, especially those belonging to the 1963 series (Figure 2A). For the CaCO₃ precipitates 1980 series (Figure 2B), L358 ¹⁴C results between UCI and NTNU are visibly apart from each other (standard deviation = 0.6%). For detailed differences in ¹⁴C data and assemblages, see Table S1. Radiocarbon data agreements between UCI and at NTNU sets are remarkably good for samples that undergo different handling procedures, spectrometer measurement setups, and data analyses, especially when the contamination embedded into CaCO₃ powders was clearly not removed. As Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023) already pointed out, the ¹⁴C results of CaCO₃ powders from the 1980 series are higher than those from the Schauinsland (48ºN) site reported by Levin et al. (Reference Levin, Kromer, Schoch-Fischer, Bruns, Münnich, Berdau, Vogel and Münnich1985) (Figure 2B). The Schauinsland (48ºN) site is located 10º further south than Lindesnes (58ºN), and therefore, the geographical provenance of the air parcels transporting ¹⁴CO₂ reaching those sites may explain the differences observed here. As aforesaid, we also plotted the dataset of Fruholmen (Nydal and Lövseth Reference Nydal and Lövseth1996) for comparisons (Figure 2B). This geographic location is approximately 10º north of the Lindesnes (58º N) site. The fact that our ¹⁴C results fell between ¹⁴CO₂ signatures of both Schauinsland (48ºN) and Fruholmen (71ºN) could be somewhat promising. Yet, the results of a series of archaeological samples, archived in the same type 2 containers (Figure 1), indicate potential atmospheric contamination (Seiler et al. Reference Seiler, Grootes, Svarva and Nadeau2023). Without further atmospheric ¹⁴C results in very close proximity to the Lindesnes (58ºN) site, it cannot be substantiated whether the ¹⁴C results of the CaCO₃ precipitates of the 1980 series of either Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023) or this study are correct.

As per the reference materials (Table S2), sealed tube combusted step-combustion carbonates FIRI-C turbidite yielded a ¹⁴C age of 18,380 ± 57 (n=2) yrs BP, while coral standard a F¹⁴C of 0.9435 ± 0.0015 (Table S2). Both are in reasonable agreement with expected consensus values (reported in section 2.1), especially coral standard. FIRI-C showed a small ¹⁴C-age difference of about 200 yrs between sealed tube step-combustion at 285ºC (this study) and its consensus value, which is based on the standard acid hydrolyze procedure (Scott et al. Reference Scott, Boaretto, Bryant, Cook, Gulliksen, Harkness, Heinemeier, McGee, Naysmith, Possnert and van der Plicht2004). This difference is significantly smaller than that reported by Bush et al. (Reference Bush, Santos, Xu, Southon, Thiagarajan, Hines and Adkins2013), e.g., 2–4 kyrs offsets, when mixed carbon fractions of this turbidite were directly measured by ¹⁴C-AMS when loaded into aluminum target holders. Bush et al. (Reference Bush, Santos, Xu, Southon, Thiagarajan, Hines and Adkins2013) also attempted to remove the ¹⁴C effect of the FIRI-C organic fraction by heating this turbidite at 500°C in air. While the authors failed to reproduce expected ¹⁴C results of Scott et al. (Reference Scott, Boaretto, Bryant, Cook, Gulliksen, Harkness, Heinemeier, McGee, Naysmith, Possnert and van der Plicht2004), they demonstrated that the FIRI-C turbidite powder is highly active, and can reabsorb CO₂ from ambient, once powders are allowed to cool off. Santos et al. (Reference Santos, Alexandre, Coe, Reyerson, Southon and De Carvalho2010) demonstrated a similar effect when heating fine powders at 160°C. Thus, this issue seems to be related to absorption properties of materials in particulate form, their porosity level, and particulate surface area available, rather than just temperature settings for the purpose of cleansing.

Here, all samples (in the form of coarse particulates or powders, carbonates or organics) were not allowed to cool off after 285ºC treatment per 24 hr. Thus, we have no knowledge if reabsorption effects would be different between distinct types of particulate samples. While our FIRI-C two step-combustion oxidation age-value of 18,380 ± 57 (n=2) yrs BP is somewhat older than that reported in Scott et al. (Reference Scott, Boaretto, Bryant, Cook, Gulliksen, Harkness, Heinemeier, McGee, Naysmith, Possnert and van der Plicht2004), it is not significantly different. Radiocarbon result of the recalcitrant fraction of organic Rice char yielded F¹⁴C of 1.0655 ± 0.0020 (n=4), and is in perfect alignment with its expected ¹⁴C value (section 2.1). We can then conclude that nearly all OC have been removed from both FIRI-C and Rice char upon step-combustion treatment, as described in section 2.2.1.

3.2. FTIR Results

FTIR spectroscopy was performed to help determine possible CaCO₃ precipitate impurities (Figure 3). FIRI-C, a carbonate/clay mixture, was also analyzed in view of its heterogeneous properties. Typical FTIR of simple molecules, such as calcium carbonate (CaCO₃), water (H₂O), sodium hydroxide (NaOH) and sodium chloride (NaCl) are shown in the supplementary material (Figure S2) to assist in discussions.

Figure 3 FTIR spectrum of untreated (panel A) and heat-treated (panel B) CaCO₃ powders from 1963 and 1980, as well as FIRI-C turbidite (carbonate/clay mixture). The window was set between 4000 to 600 cm⁻¹.

All FTIR CaCO₃ precipitate profiles of untreated (Figure 3A) and step-combustion treated (Figure 3B) samples show the presence of the usual CaCO₃ profile peaks between 1500 and 700 cm⁻¹ (Figure S1), e.g., the pronounced and broad peaks (1400, 865, and 714 cm⁻¹). Those are characteristic peaks of CaCO₃ molecules and calcite crystals (Gupta et al. Reference Gupta, Singh, Kumar and Khajuria2015; NIST webbook online database—Figure S2). The FIRI-C untreated and heated-treated samples showed the IR band at 1022 cm⁻¹ due to Si-O-Si stretching vibration characteristic of sediments (Liu et al. Reference Liu, Colman, Brown, Minor and Li2013).

FTIR spectrum indicate that three of the 1963 untreated CaCO₃ powders were still slightly wet, possibly due to CaCO₃ powders exposed to moist air during storage. While no post-treatment procedures are detailed in Nydal and Lövseth (Reference Nydal and Lövseth1983), we assume that carbonate precipitate washing was sufficient to remove Na residues from incomplete removal of supernatant (NaOH) and reaction byproducts (NaCl). Nonetheless, we added here the transmittance peak profiles of pure H₂O as well as NaOH and NaCl (Figure S2). Their transmittance peaks most likely overlap in the same fingerprint broad region (2500–3750 cm⁻¹), and therefore, they would be difficult to disentangle in any FTIR profile, especially if they are subtle.

The 1980 series had little to no peaks within the 2500–3750 cm⁻¹ region before and after heat-treatment (Figure 3B), indicating that these samples were somewhat drier, than those of the 1963 series (Figure 3A). The rationality behind peak differences between 1963 and 1980 series samples may be explained by types of containers and lids (Figure 1). As Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023) noted, cap type 2 seems to provide better closure than cap type 1.

The FIRI-C untreated and heated-treated samples also showed FTIR profile differences by the reduction of the broad area of the single bond between 2500–4000 cm⁻¹. This area is known to contain aromatic C-H and C-C bonds, as well as overlapping νH-O (H₂O). A reduction of this broad band, but not its complete removal, would imply that some of the recalcitrant C-H and C-C bonds are still present after heat treatment. Unremoved recalcitrant fraction of FIRI-C may corroborate the small ¹⁴C-age difference of about 200 years from sealed tube step-combustion (this study). Even though in our procedure samples were transferred hot to vacuum line for evacuation and sealing (so that ambient-air CO₂ reabsorption over fine powders could be avoided), any unremoved elemental/refractory carbon would be combusted in conjunction with CaCO₃ in the presence of CuO. The evolved CO₂ (and its associate ¹⁴C value) would therefore be biased by this sediment-like carbon contribution.

3.3. Nydal’s Stored CaCO₃ Precipitates and Possible Sources Exogenous C Contamination

Nydal’s CaCO₃ precipitates have been stored in a room at NTNU without climatic condition controls (variable temperature and relative humidity). While containers with CaCO₃ precipitates (retested in this study) appeared to be properly closed at first glance, there is no guarantee that this was the case for the whole storage period. Moreover, the lid from the type 1 container appears to be less air-sealing effective than that of type 2 (Figure 1). The possibility of decades of exposure to moisture, increased air-CO₂ and variable temperatures, leads us to consider some theories on how CaCO₃ precipitates have been contaminated.

First, ambient CO₂ that seeped into containers could be adsorbed as such in CaCO₃ precipitate inter- and intra-particle voids. While we have not conducted porosimetry analysis of Nydal’s CaCO₃ precipitate, it is fair to assume that the powder is highly porous, as judged by bio- and industrially produced CaCO₃ precipitates that have been heavily studied for their permeability characteristics and CO₂ storage capabilities (e.g., Moore and Wade Reference Moore and Wade2013; Yoon et al. Reference Yoon, Major, Dewers and Eichhubl2017). Moreover, the large specific surface area of all types of finer powders and the interstitial gaps created by particles of different sizes tend to naturally entrap gasses. Nonetheless, adsorbed ambient CO₂ within and between particles is normally very volatile and sensitive to heat treatments as low as 160ºC (Santos et al. Reference Santos, Alexandre, Coe, Reyerson, Southon and De Carvalho2010). If the main contaminants of Nydal’s stored CaCO₃ precipitates are interparticle OCs and/or other gasses, they should have been mostly removed upon step-combustion treatment, as we demonstrated by ¹⁴C measurements of step-combustion Rice char and FIRI-C samples (Table S2). Moreover, partial OC removal of Nydal’s stored CaCO₃ precipitates by step-combustion would at least lead to measurable differences between ¹⁴C step-combustion treated CaCO₃ results and those of Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023). We have not observed these differences, thus solo ambient air-CO₂ contamination does not seem to explain the mismatch between current and Nydal’s ¹⁴C results.

A second hypothesis involves re-carbonation of ambient-CO₂ and formation of new layers of carbonate over and within intraparticle voids of the existing CaCO₃ powders, a type of CO₂-water-rock interaction at room-temperature (small amounts of hydrate lime—Ca(OH)₂, for example). For a significant secondary formation of CO₂ to CaCO₃ microspheres into more stable CaCO₃ crystalline forms at room temperature, pre-calcination of CaCO₃ to CaO at higher temperatures is required (Erans et al. Reference Erans, Nabavi and Manović2020). While we do not believe Nydal’s CaCO₃ precipitates undergo calcination, there is very little information on how they were handled. Among the numerous factors that can affect the CaCO₃ precipitation process and secondary species formation, Febrida et al. (Reference Febrida, Cahyanto, Herda, Muthukanan, Djustiana, Faizal, Panatarani and Joni2021) stress the presence of various foreign ions or molecules depending on the aqueous solution used from which the carbonate precipitates. For sample collection, Nydal used rainwater (Nydal and Lövseth Reference Nydal and Lövseth1983), which can be loaded with several ionic compounds (Carol Reference Carol1962). Therefore, the combination of high air-CO₂ pressures found in buildings (as far as 2500 ppm—Erans et al. Reference Erans, Nabavi and Manović2020), CO₂ solubility property in liquid water or vapor (Zeman and Lackner Reference Zeman and Lackner2004), and the fact that the Nydal’s CaCO₃ powder containers were found in a room with moisture variability for 30+ years of storage may have played a role in further air-CO₂ entrapment, soluble calcium bicarbonate formation (H₂CO₃), and CaCO₃ recrystallization filling up CaCO₃ pores, especially if foreign ions were present in precipitates (Sanz-Pérez et al. Reference Sanz-Pérez, Murdock, Didas and Jones2016; Giacomin et al. Reference Giacomin, Holm and Mérida2020; Toffolo Reference Toffolo2020; Febrida et al. Reference Febrida, Cahyanto, Herda, Muthukanan, Djustiana, Faizal, Panatarani and Joni2021). Thus, ambient CO₂ adsorbed to CaCO₃ powders may just have promoted a dynamic and continuous gas-solid exchange process with CO₂ bonding to the lattice. Dissolved CO₂ reacts with crystals through a CO₂-water-rock interaction, where H₂CO₃ and HCO₃ ^– secondary species are formed (Yoon et al. Reference Yoon, Major, Dewers and Eichhubl2017; Hanein et al. Reference Hanein, Simoni, Woo, Provis and Kinoshita2021; Huang et al. Reference Huang, Zhang, Santos, Rodríguez, Holden, Vetro and Czimczik2021). Once in the interior of minute particles, this exchanged CO₂ may no longer be easily removed by either heating at lower temperatures (this study) or selective leaching (Seiler et al. Reference Seiler, Grootes, Svarva and Nadeau2023). The latter may explain why the ¹⁴C results of CaCO₃ precipitates in this study corroborate so well with those in Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023). Finally, a simpler mechanism may just involve adsorption of air CO₂ to surface of powder crystals followed by desorption of a CO₂ molecule of the original crystal lattice, followed by subsequent crystal lattice diffusion. In every case, gas-solid exchange equilibrium followed by significant ¹⁴C changes to CaCO₃ original signal would require at least (a) the presence of very small crystals, and (b) longer duration of exposure to CO₂.

Even though we detected a difference in overall CaCO₃ precipitate mass before and after heat treatments, our oxidative decomposition of carbonate by step-combustion (possibly more active on powder surfaces, than their interior) most likely removed just H₂O and interparticle adsorbed CO₂. Any other adsorbed CO₂, either deeply trapped in CaCO₃ pores or bonded to the lattice, would contribute to the final ¹⁴C results we obtained.

We may never know for certain what is the mechanistic process of how ambient CO₂ has altered Nydal’s CaCO₃ precipitates, and when this occurred during the 30+ years storage. Either way, several attempts to remove this exogeneous carbon, e.g., after HCl and H₂O₂ treatments, or chemically untreated flash-combustion (Seiler et al. Reference Seiler, Grootes, Svarva and Nadeau2023) as well as ¹⁴C two-step combustion oxidation method (this study) were insufficient to bring the 1963 and the 1980 series to expected ¹⁴C values.

As far as one can tell, all of Nydal’s CaCO₃ precipitates published in the literature yielded correct ¹⁴C values once measurements were completed and corrections were applied (Nydal and Lövseth Reference Nydal and Lövseth1983, Reference Nydal and Lövseth1996). After 30+ years of storage, those same CaCO₃ precipitates are yielding inaccurate ¹⁴C values. Neither Seiler et al. (Reference Seiler, Grootes, Svarva and Nadeau2023) nor this study introduced artifacts to data to justify the differences detected. Hence, we can only assume that the current Nydal’s CaCO₃ precipitate archive is of no use to reproduce atmospheric ¹⁴CO₂ signatures, until the contamination issue can be effectively resolved.

Carbon dioxide sequestration in the form of CaCO₃ is a useful way to store atmospheric CO₂ for further analyses. But it requires proper storage conditions, ideally within a hermetically sealed vial, such as evacuated flame-sealed glass ampoules.