INTRODUCTION
The radiocarbon value of dissolved inorganic carbon (DIC), which is one of the most powerful tools for dating groundwater samples, has been applied in many previous studies (Geyh Reference Geyh2000; Iwatsuki et al. Reference Iwatsuki, Xu, Itoh, Abe and Watanabe2000). DIC in water samples is usually extracted and collected by either a gas-strip method or a precipitation method. Both of these methods have frequently been used for groundwater studies (gas-strip method: Dorsett et al. Reference Dorsett, Cherrier, Martin and Cable2011; Stewart Reference Stewart2012; Dulinski et al. Reference Dulinski, Rozanski, Kuc, Gorczyca, Kania and Kapusta2013; precipitation method: Sültenfuβ et al. 2011; Stewart Reference Stewart2012; Corcho Alvarado et al. Reference Corcho Alvarado, Paces and Purtschert2013).
In the precipitation method, water samples are alkalinized with NaOH or ammonium followed by the addition of BaCl2 or SrCl2, after which DIC is precipitated as BaCO3 or SrCO3. These precipitation procedures are occasionally conducted during groundwater sampling in the field. This procedure enables the amount of samples to be reduced significantly, and facilitates the transport of the samples from the field to the laboratory. However, the precipitation might be contaminated by the alkaline solution because CO2 readily dissolves in alkaline solutions (Aggarwal et al. Reference Aggarwal, Araguas-Araguas, Choudhry, van Duren and Froehlich2014). In addition, it could be difficult to extract carbon from groundwater samples with very low DIC concentrations because sometimes precipitation cannot not be visually confirmed in such samples and this makes the collection of the precipitation difficult. Furthermore, precipitation may not quantitatively remove carbon due to interfering reactions (Minami and Takahashi Reference Minami and Takahashi2015). In such cases, the 14C might be fractionated. Possible contamination and potential 14C fractionation need to be quantitatively determined to assess the reliability of the precipitation method in groundwater DI14C dating.
In the gas-strip method, water samples are acidified by phosphoric acid and DIC is collected as CO2 (Atekwana and Krishnamurthy Reference Atekwana and Krishnamurthy1998). This method can be applied to a wide variety of groundwater samples, although they must be brought to the lab for processing. This can be the only method to collect DIC in cases where the precipitation method cannot be applied. To validate the method, the possibility of contamination during the gas-strip extraction and DIC recovery from solutions with low DIC concentration needs to be investigated.
In this study, the reliability of these two methods in the determination of 14C groundwater ages is examined, with special attention given to potential contamination. To accomplish this, (1) a preliminary comparison of 14C values obtained by the two preparation methods using a set of NaHCO3 solutions was made, (2) contamination in sample preparations was identified, and (3) the two methods were applied to natural groundwater samples.
METHODS
Preliminary Comparison of 14C Values from Two Preparation Methods Using NaHCO3 Solutions
A comparison of 14C results obtained by the two preparation methods for NaHCO3 solutions allows us to determine the precision and accuracy of each preparation method. This is accomplished by preparing 14C DIC standards, given the DIC concentrations of the solutions and 14C content of the NaHCO3 powder. Four types of NaHCO3 standard solutions were prepared (Table 1), and both the gas-strip and precipitation methods were applied to extract and collect DIC from them. The 14C content of the DIC extracted from the solutions was measured by accelerator mass spectroscopy (AMS). Solutions 1–4 were prepared to simulate the following: (1) an old (~25,000 BP) groundwater with low salinity, (2) younger (~10,000 BP) groundwater with low salinity, (3) younger groundwater with low DIC and salinity, and (4) younger groundwater with high salinity. For each solution, two samples, one obtained by gas-strip and one by precipitation, were compared.
Solutions for preliminary comparison of 14C values obtained by the two preparation methods.
* NaHCO3 used for Solutions 2 to 4 was prepared as follows: NaOH solution was bubbled with air to obtain precipitation of NaHCO3 and this NaHCO3 was mixed with the purchased solution.
All procedures except weighing the NaHCO3 powder were carried out in a glove box (GB) under an Ar atmosphere. A CO2 absorbent (Lithoryme: Allied Healthcare Products, St. Louis, MO, USA) was placed in the GB to ensure the CO2 concentration was low. Deionized water was placed in the GB and purged with Ar gas for at least 12 hr to reduce the DIC in the water. A fraction of this water was used for the DIC measurements after purging. NaHCO3 powder was subsequently added to the purged water and stirred until completely dissolved. A commercial NaHCO3 powder was used for Solution 1, while a second batch of NaHCO3 was prepared by bubbling air through NaOH for Solutions 2–4. In the case of Solution 4, a simulated seawater (SSW) was prepared by adding NaCl, KCl, MgSO4 7H2O, MgCl2 6H2O, and CaCl2 into the purged water and a NaHCO3 solution was prepared by dissolution of NaHCO3 into the SSW.
The four NaHCO3 solutions were sent to Beta Analytic (Miami, FL, USA) for analysis. Two sample bottles were used for the gas-strip method, while the remaining samples were extracted by the precipitation method for 14C analyses. Both the DIC collection and 14C measurements were conducted by Beta Analytic.
Identification of Contamination during Sample Preparation
A preliminary comparison of 14C in the NaHCO3 solutions showed the possibility of significant contamination with modern carbon during the precipitation procedure (see Results and Discussion). Therefore, a gas-strip line was employed to test the NaOH and SrCl2 solutions used in the DIC precipitation procedure. This step was taken to identify the source and amount of contamination associated with the precipitation method. The NaOH and SrCl2 solutions were prepared as follows. Dissolved gas in deionized water was purged with Ar in a GB, as described above. Two types of NaOH solutions were then prepared, one under atmospheric conditions and the other in the GB with the addition of granular NaOH to deionized water. SrCl2 solutions were subsequently prepared by adding the SrCl2 reagent to deionized water under atmospheric conditions.
DIC extraction using the gas-strip method was optimized by application of a vacuum gas-strip line (Figure 1) in the JAEA (Japan Atomic Energy Agency) and the following procedure was established. Prior to sample preparation, the air in the gas-strip line was removed under a vacuum of less than 0.1 Pa. Next, 2 mL of phosphoric acid was added to the water sample to convert the DIC to CO2 gas. CO2 gas was subsequently stripped from the water sample by bubbling with carrier gas (pure N2 or Ar) at a flow rate of 500 cc/min and collected in a cold trap of liquid N temperature. After 5 min of gas circulation, the carrier gas was evacuated and trapped CO2 was purified and directed into a glass tube at the gas sampling port. The yield of CO2 relative to DIC in the water samples was 90%. The DIC concentration in NaOH and SrCl2 solutions were estimated by the gas (CO2) pressure in the gas-strip line.
Gas-strip line to extract DIC from groundwater
Application of the Two Methods to Natural Groundwater Samples
Both the precipitation and gas-strip methods were applied to natural groundwater samples to identify 14C differences between each method. Groundwater samples were obtained from depths of 200–500 m in granitic rock at the Mizunami Underground Research Laboratory (MIU), Gifu Prefecture, Japan (Figure 2). The groundwater chemistry was Na-Ca-Cl or Na-Cl dominant with a salinity of less than 1 g/L (Iwatsuki et al. Reference Iwatsuki, Furue, Mie, Ioka and Mizuno2005, Reference Iwatsuki, Hagiwara, Ohmori, Munemoto and Onoe2015). Groundwater samples were collected into 1-L airtight glass vessels for precipitation. Next, 10 mL of 5N NaOH solution and 10 mL of 2N SrCl2 solution were added into the vessel to precipitate SrCO3, after which the samples were stored at 4°C. Precipitated SrCO3 was collected into small sample vessels by pipette in a CO2-free GB for freeze-dry storage. In parallel, groundwater samples for gas-strip method preparation were collected into 0.5-L airtight glass vessels and stored at 4°C, while CO2 gas was collected by the aforementioned gas-strip method for 14C measurement. In the case of natural samples, the DIC collection was carried out in the Tono Geoscience Center of JAEA and 14C was measured by AMS in JAEA (JAEA-AMS-TONO).
Groundwater sampling point at Mizunami underground research laboratory
RESULTS AND DISCUSSION
Preliminary Comparison of 14C Values from Both Preparation Methods Using NaHCO3 Solutions
14C and 13C values of the NaHCO3 powder used for Solution 1 were measured twice and the results were 0.5 pMC, –18.6‰ and 0.4 pMC, –17.1‰, respectively. The 14C and 13C values of NaHCO3 powder prepared for Solutions 2–4 were also measured twice and found to be 26.6 pMC, –18.9‰ and 26.1 pMC, –18.1‰, respectively. NaHCO3 solutions were prepared by dissolving NaHCO3 powders into deionized water or SSW. Therefore, the 14C in the DIC sample solutions can be calculated by Equation 1:
$$^{{{\rm 14}}} {\rm C}_{{\rm s}} {\equals}{{^{{{\rm 14}}} {\rm C}_{{\rm p}} {\times}[{\rm C}]_{{\rm p}} {\plus}^{{{\rm 14}}} {\rm C}_{{\rm w}} {\times}[{\rm C}]_{{\rm w}} } \over {[{\rm C}]_{{\rm p}} {\plus}[{\rm C}]_{{\rm w}} }}$$
where 14Cw is the 14C value (pMC) of DIC in the deionized water or SSW used for preparation of the NaHCO3 solutions; 14Cp is the 14C value of NaHCO3 powder; [C]p and [C]w are the concentrations of DIC (mg/L) provided from NaHCO3 powder and DIC in the deionized water, respectively; and [C]p is found from the difference between the DIC concentration of the sample solutions and [C]w.
14Cw is considered to be the 14C of CO2 in air. According to previous studies (Taylor Reference Taylor2004; Levin et al. 2013), the value of 14Cw is between 100 and 110 pMC. The [C]w and DIC concentration of sample solutions before the experiment (i.e. sum of [C]p and [C]w) were determined using a TOC analyzer and the measurements included a maximum error of 5%. The 14C concentrations of the powder NaHCO3 reagent were 0.4 and 0.5 pMC; thus, the correct value of 14Cw was considered to be between 0 and 1 pMC. Similarly, the correct values of 14Cw for prepared NaHCO3 Solutions 2–4 were assumed to be between 26 and 27 pMC. These values were put into Equation 1 and a possible range of 14Cs (correct value of 14C in the sample solution) was calculated. The possible ranges of 14Cs for Solutions 1–4 were 3.1 to 5.2, 26.0 to 28.0, 26.0 to 29.0, and 26.0 to 28.0 pMC, respectively. The calculated ranges of 14Cs are compared to the measured 14C values in Table 2.
Comparison of measured 14C and possible ranges of 14C for NaHCO3 solutions.
* Calculated by Equation 1.
Table 2 shows the 14C and 13C values of DIC measured by the gas strip (gas strip) and precipitation (precip) methods for NaHCO3 solutions. In Solutions 1 and 2, the differences between 14Cgas strip and 14Cprecip ranged from 4 to 7 pMC. However, the ages estimated by the gas-strip and precipitation methods differed significantly (by 7000 to 12,000 yr) for Solution 1, and by 1000 to 1200 yr for Solution 2. These results indicate that the uncertainty of 14C age becomes larger when the age of old groundwater is estimated, as expected. The difference between 14Cgas strip and 14Cprecip in Solution 3 was about 10 pMC, which was twice that of Solution 1 and 2. This result indicates the difference becomes more significant in the case of solution with low DIC. Furthermore, the difference in Solution 4 was 17 to 33 pMC, indicating a very large difference compared to other solutions. This result shows that the ionic strength and/or ion composition of the sample water can significantly affect 14C values of the precipitation method. There is also a possibility that the 14C values might be controlled by relative relationships between carbonate and the other ion concentrations.
As shown in Table 2, the 14Cgas strip values were close to the possible range of 14Cs values calculated from Equation 1, although measured 14Cgas strip values were always slightly (~1 pMC) higher. Conversely, 14Cprecip always showed much higher values relative to the calculated range of 14Cs values. Thus, the gas-strip method is shown to be reliable for many types of groundwater samples with minimal error. 14Cprecip always showed significantly higher values than 14Cs. Therefore, the precipitation method should be applied carefully to groundwater samples and potential errors in 14Cprecip results should be explicitly assessed.
The effects of DIC concentration were investigated by comparing the 14C values in Solutions 1, 2, and 3. The averaged differences between 14Cprecip and 14Cs in Solutions 1, 2, and 3 were 5.0, 5.1, and 10.4 pMC, respectively. These results clearly indicate that the relative difference in 14C estimated by the two methods increases with decreasing DIC concentration. These findings can be explained if we assume that a certain amount of modern carbon contaminated the sample during precipitation. If 0.6 to 0.7 mg of modern C contaminates 1 L of sample solution during the precipitation procedure, the observed differences between 14Cprecip and 14Cs will occur.
A comparison of Solutions 2 and 4 shows the effect of salinity on 14Cprecip. The difference between 14Cgas strip and 14Cprecip was greater in Solution 4 than in Solution 2. If contamination by modern carbon from the air was the dominant cause of this phenomenon, we would expect that the 13C values would be close to −8‰ (Taylor Reference Taylor2004). However, the 13Cprecip values in the Solution 4 test samples were −23.7 and −27.3‰, significantly lower than the 13C value of NaHCO3 powder (−18.5‰). Thus, the difference between 14Cgas strip and 14Cprecip in saline water cannot be explained by contamination of modern carbon from the air alone.
Identification of Contamination in Samples Prepared by the Precipitation Method
The preliminary comparison experiments described earlier indicate that contamination by modern carbon might occur during the precipitation procedure. Modern carbon contamination from air to alkaline and SrCl2 solutions was assumed to be the most likely source of contamination during the precipitation procedure. The background DIC concentrations in NaOH and SrCl2 solutions were analyzed using a gas-strip line to confirm this possibility. Table 3 shows the DIC concentrations of each solution. Degassed deionized water with H3PO4 contained 0.01 mg/L DIC. The SrCl2 solution is estimated to contain 0.04 mg/L DIC. Similarly, the DIC content of the NaOH solution (1 L of degassed deionzed water with 5 mL of 5N NaOH solution) was estimated to range from 0.2 to 0.3 mg/L, when 5N NaOH solutions were prepared under atmosphere conditions. On the other hand, NaOH solutions prepared under inert conditions contained approximately 0.2 mg/L DIC, which was the same as prepared in atmosphere condition. Accordingly, the modern carbon contamination of the solutions appears to originate from the NaOH granules. Such background contamination of modern carbon probably also occurs for other alkaline solutions, such as ammonium solutions. Contamination with modern carbon in preliminary experiments using NaHCO3 solutions could also originate from the NaOH solution used in the precipitation method.
Background DIC concentration of solutions used for precipitation method.
Conversely, modern carbon contamination might occur in the gas-strip method during the addition of H3PO4 into the groundwater prior to carrier gas circulation because deionized water with H3PO4 contained minor but measurable DIC. This would be expected to influence the 14Cgas strip result from the NaHCO3 solutions, which deviate by about 1 pMC from 14Cs (Table 2).
Application of the Two Methods to Natural Groundwater Sample
14C values of DIC in groundwater samples prepared by both the gas-strip and precipitation methods are shown in Table 4. 13C values of DIC are also listed in Table 4. 14Cprecip values were approximately 0.4–9.5 pMC higher than those obtained using the gas-strip method. By assuming that 14Cgas strip is less contaminated by modern carbon and the 14C value of contaminated modern carbon in samples obtained by the precipitation method is 100 pMC, the mass balance equation describing contamination in the precipitation methods (14Cprecip) can be expressed by the following equation:
$${\rm DIC}_{{{\rm gw}}} {\times}^{{14}} {\rm C}_{{{\rm gas\,\,strip}}} {\plus}{\rm DIC}_{{{\rm contami}}} {\times}100{\equals}\left( {{\rm DIC}_{{{\rm gw}}} {\plus}{\rm DIC}_{{{\rm contami}}} } \right){\times}^{{14}} {\rm C}_{{{\rm precip}}} $$
where DICcontami is the amount of DIC contamination with modern carbon (mg) and DICgw is the amount of DIC (mg) in groundwater samples analyzed before addition of NaOH solution. The amounts of modern carbon contamination during the precipitation procedure (described as DICcontami above) were estimated to be 1 mg for all groundwater samples, regardless of the 14Cgas strip and DICgw values (Table 4). These findings suggest that modern carbon contamination during preparation of the precipitation method significantly influenced the 14C value according to the DIC content and 14C concentration of sampled water.
Comparison of 13C and 14C of groundwater samples prepared by gas-strip and precipitation methods.
* δ13Cprecip, δ13Cgas strip, 14Cprecip, and 14Cgas strip are measured 13C or 14C of samples prepared by precipitation and gas-strip method, respectively.
** DICcontami is the amount of DIC contamination of modern carbon calculated by Equation 2.
The 14Cgas strip value of groundwater at depths of 200 to 500 m were estimated to range from 2 to 29 pMC, with a tendency to become lower with increasing depth. We were not able to collect the DIC in groundwater at 500 m depth (12MI33 borehole) by the precipitation method. The reason that the DIC did not precipitate is not clear, but is likely related to the low DIC concentration (4.9 mg/L) and high salinity (about 450 mg/L) of the sample. The groundwater residence time at that depth is estimated to be about 31,000 yr by the 14Cgas strip value. However, the groundwater flow around this large underground facility (MIU) has been influenced by water drainage for more than 10 yr (Iwatsuki et al. Reference Iwatsuki, Hagiwara, Ohmori, Munemoto and Onoe2015), and the analyzed 14C values might reflect mixed shallow and deep groundwater sources.
When we discuss the groundwater ages with 14C, the 14C values should be corrected considering the geochemical reactions that could affect to the 14C values (Clark and Fritz Reference Clark and Fritz1997; Kalin Reference Kalin2000). However, in this article groundwater ages are estimated without correction of geochemical reactions because our main purpose is to indicate the difference between the precipitation and gas-strip methods.
Applicability of the Precipitation Method for Groundwater Dating
As discussed earlier, contamination with modern carbon probably occurs during the DIC precipitation procedure. The effects of contamination varied significantly depending on the DIC and 14C concentrations of the groundwater samples. Equation 2 was applied to estimate the error in cases in which modern carbon contaminated groundwater with various DIC concentrations. For example, Figures 3a and 3b show 14C values after contamination of 1 L of groundwater with 0.5 or 1 mg modern carbon as estimated by Equation 2. The error from the true 14C value became more significant as the DIC and 14C concentrations of sampled groundwater decreased. Constraints for DIC and 14C of groundwater that should be analyzed within 10% error are estimated in Figure 3c. Because background contamination with modern carbon was estimated for each preparation method, the uncertainty in 14C analysis could be inferred by the DIC concentration and probable 14C content of the groundwater. As described, 1 L of NaHCO3 solutions and groundwater samples were contaminated with about 0.7 mg and 1 mg of modern carbon, respectively.
Error in 14C analysis for contamination by modern carbon for groundwater with initial DIC concentrations of 0–20 mg/L. 14C value of contaminated modern carbon is assumed to be 100 pMC. (a) Contamination by modern carbon is 0.5 mg/L. (b) Contamination by modern carbon is 1 mg/L. The shaded area shows error within 10% of the known 14C value. (c) Constraint for DIC and 14C of groundwater that should be analyzed within 10% error in the case of modern carbon contamination with 0.5 and 1 mg/L.
As shown in Figure 3b, if we want to control the 14C age error from the precipitation procedure within 10%, the concentration of DIC and 14Cgw contents should be higher than 20, 10, and 1 mg/L and 32, 47, and 83 pMC, respectively. If the error has to be controlled within 20%, the concentration of DIC and 14Cgw should be higher than 20, 10, and 1 mg/L and 19, 31, and 71 pMC, respectively.
Because our purpose was to determine the effects of both methods on estimation of groundwater age, the relationships between expected groundwater age estimated by 14C (without considering the effect of geochemical reaction) using the precipitation method and DIC concentrations are compared in Figure 4. The expected groundwater age estimated by 14C using the precipitation method was calculated by the following equations:
$$^{{{\rm 14}}} {\rm C}\,{\rm age}\,_{{{\rm ex}}} ({\rm y}){\equals}{{5730} \over {{\rm ln2}}}{\times}{\rm ln}\left( {{{^{{{\rm 14}}} {\rm C}_{{{\rm precip}}} } \over {100}}} \right)^{\!\!{(-1){\rm x}}}$$
$$^{{{\rm 14}}} {\rm C}_{{{\rm preip}}} {\equals}{{^{{{\rm 14}}} {\rm C}_{{{\rm gw}}} {\times}[{\rm DIC}]_{{{\rm gw}}} {\plus}^{{{\rm 14}}} {\rm C}_{{{\rm contami}}} {\times}1.0} \over {[{\rm DIC}]_{{{\rm gw}}} {\plus}1.0}}$$
Equation 3 is frequently used for estimation of groundwater age from 14C (Clark and Fritz Reference Clark and Fritz1997; Mook and Plicht Reference Mook and van der Plicht1999). 14C ageex is the expected 14C age after precipitation and 14Cprecip is the 14C concentration after precipitation with 1 mg modern carbon contamination. 14Cgw and 14Ccontami are the concentrations of DI14C in the groundwater samples and modern carbon, respectively. The value of 14Ccontami is set to 100 pMC. [DIC]gw is the concentration of DIC in the groundwater samples.
Relationship between concentration of DIC in groundwater and expected 14C ages (14C ageex) for samples prepared by the precipitation method where groundwater 14C (14Cgw) is 10, 50, and 80 pMC. 14C ageex values were calculated using Equations 3 and 4.
The acceptable error depends on the purpose of the study. When we applied the precipitation method to estimate groundwater age, the acceptable error for the study and conditions that affect the 14C age (DIC and 14C concentrations in targeted groundwater) had to be considered. The information provided in Figures 3 and 4 is useful to determine if the precipitation method is applicable for a targeted groundwater. According to the results of 14C values for NaHCO3 solutions and groundwater samples treated using the precipitation method, we could assume that the amount of modern carbon contamination to be about 1 mg. However, this value might depend on the atmosphere of the laboratory and reagents used for the precipitation procedure. Thus, quantitative estimation of contamination by modern carbon in each laboratory during the precipitation procedure should be evaluated if the precipitation method is chosen.
CONCLUSION
In this study, the differences in 14C values of samples prepared by the gas-strip and precipitation methods were compared in artificially prepared and natural groundwater samples. The 14C values obtained by the gas-strip method were very similar to the theoretically predicted 14C values in chemically distinct waters, indicating that this method can be applied to a variety of groundwaters. Conversely, the 14C value obtained by the precipitation method showed higher than predicted 14C values. In the case of natural groundwater samples, 14C obtained by the precipitation method always showed higher values than those from the gas-strip method. It was assumed that an average of 1 mg/L of modern carbon contaminated the samples during the precipitation procedure. This contamination affects the estimation of groundwater ages, especially when DIC concentrations and 14C contents are low. Thus, the applicability of the precipitation method should be considered according to the required accuracy and/or purpose of the study. The amount of contamination with modern carbon might differ among laboratories and should therefore be estimated in each laboratory to evaluate the applicability of the precipitation method.
For solutions with high salinity, the differences between 14C measured from precipitation samples and predicted 14C values increased. Thus, for groundwater with low DIC concentration and/or high salinity and old groundwater, the gas-strip method should be applied to estimate the groundwater age by DI14C.
ACKNOWLEDGMENTS
The authors thank T Yamaguchi, A Yokosuka, and M Yoshioka with CERES, Inc. and T Nishio with JAEA-AMS-TONO Laboratory for helping with the laboratory work. D Hood with Beta Analytic and S Matsuyama with Geo Science Laboratory are appreciated for their kind advice regarding 14C measurement and pretreatments. A portion of this study was performed under contracts awarded by the Ministry of Economy, Trade and Industry.
