Introduction

The stable carbon (C) isotope ratio (δ¹³C) of dissolve inorganic carbon (DIC), such as carbonate (CO₃^2-) and bicarbonate (HCO₃^-), in soil extracts and water samples provides valuable information on the sources and biogeochemical C cycling (Atekwana and Krishnarmurthy, 1998). The natural background value of δ¹³C in uncontaminated environment (e.g., groundwater) ranges from ‒15‰ to ‒6‰ (Arneth and Hoefs, 1988; Mook, 2000; Grossman et al., 2002). However, increases in carbon dioxide (CO₂) in soil and water environment through aerobic decomposition of C3 organic matter (around ‒27‰) leads to decrease in the δ¹³C of DIC (Breukelen et al., 2003). Under anaerobic fermentation conditions, however, the evolved CO₂ is enriched in ¹³C (range: between ‒10 and +20‰) due to ¹³C isotope fractionation during methane (CH₄) generation, which produces ¹³C-depleted CH₄ (range: ‒110 and ‒50‰) (Grossman et al., 1989; Conrad, 2005; Mohammadzadeh and Clark, 2011; Wimmer et al., 2013).

There are typically two methods available for the measurement of δ¹³C of DIC; 1) gases DIC (i.e., CO₂) extraction after acidification of the sample solution followed by direct injection of CO₂ into stable isotope ratio mass spectrometer (SIRMS) (Atekwana and Krishnamurthy, 1998) and 2) precipitation of DIC as divalent metal-carbonate complexes followed by running on SIRMS equipped with elemental analyzer (EA) (Harris et al., 1997; Wimmer et al., 2013). Traditionally, the gaseous measurement δ¹³C of DIC requires a special headspace gas collection line, and thus it is not feasible for a large number of samples (Gillikin and Bouillon, 2007; Torres et al., 2005). Meanwhile, measurement of δ¹³C of DIC in precipitates using EA-SIRMS allows a direct combustion of the precipitates to liberate CO₂ into SIRMS through continuous gas flow; therefore, it is a more convenient method compared to the gases measurement for analyzing a large number of samples (Harris et al., 1997).

The carbonate precipitates for the analysis of δ¹³C of DIC are obtained by reacting the sample solution with addition of CaCl₂, BaCl₂, or SrCl₂ after raising the pH to shift the carbonate equilibrium (H₂CO₃ ↔ HCO₃^- ↔ CO₃^2-) to the right, which facilitates the formation of carbonate precipitates such as CaCO₃, BaCO₃, and SrCO₃ (Szynkiewicz et al., 2006). However, the effects of 1) pH adjustment (uncontrolled pH vs. raised pH) of the sample solution, 2) headspace (presence vs. absence of headspace) in the precipitation vessels, and 3) DIC in distilled water on the accuracy of the measurement of δ¹³C of DIC are not clearly investigated yet. As the contamination of the sample with atmospheric CO₂ and CO₂ dissolved in distilled water (δ¹³C below ‒10‰) in laboratories are critical concerns of the precipitation procedure, these factors need to be investigated.

This study was conducted to advance the carbonate precipitation method for precise analysis of δ¹³C of DIC by addressing the pH, headspace, and distilled water issue as mentioned above. We hypothesized that 1) raised pH may facilitate carbonate precipitation but it may fix atmospheric CO₂, resulting in a lowered δ¹³C, 2) the presence of headspace may also lower the δ¹³C due to contamination by atmospheric CO₂ depleted in ¹³CO₂, and 3) DIC in distilled water may also contribute to the erroneous measurement of the δ¹³C.

Materials and Methods

Overall experiment setting

In this experiment, we did not conduct mutual exclusive experiments by setting paired experiments to investigate the effects of pH, headspace, and distilled water separately. Rather than, we started the experiment following the results of Harris et al. (1997), who reported that SrCO₃ is more suitable for analysis the δ¹³C of DIC compared to BaCO₃ and CaCO₃. However, unfortunately, we have not obtained reliable results, and thus we tested the potential factors such as pH, headspace, and distilled water sequentially through a series of three experiments to achieve a reliable procedure (Table 1). All experiments were replicated three or five.

Materials

In this experiment, a reference DIC solution (100 mg C L^-1) was prepared by dissolving analytical grade (99.8%, Kanto Chemical, Japan) NaCO₃ in either double and tertiary distilled water purified with a water purification system (Human Power 1+, Human corporation, Korea). The δ¹³C of Na₂CO₃ determined using a continuous-flow stable isotope ratio mass spectrometer (SIRMS) (VisION, Isoprime Ltd, Cheadle Hulme, UK) was ‒7.45 ± 0.04‰ (n=5). The concentrations of DIC, dissolved organic C, and total C of distilled water were measured with a total organic C analyzer (Sievers 5310 C, GE Analytical Instruments, Boulder, CO, USA), and double distilled water had a higher (p < 0.001) C concentration than tertiary distilled water (Table 2). For carbonate precipitation, CaCl₂ (95.0%, Junsei, Japan), BaCl₂ (99.0%, Junsei, Japan), and SrCl₂ (99.0%, Junsei, Japan) were used. For the experiment, Na₂CO₃ solution (5 and 10 mg C L^-1) and precipitating agent solution (1 N concentration) was prepared immediately before (< 30 min) the experiment, and stored in air-tight bottles. To raise pH of the reaction solution, 1 N NaOH solution prepared from granular NaOH (97.0%, Junsei, Japan) was also prepared with other chemical reagents.

Experiment 1: headspace and double distilled water

In the Experiment 1 (Table 1), all the chemical reagents were prepared in double distilled water, and the mixture of SrCl₂ (2 mL) and BaCl₂ (2 mL) was used as a precipitating agent for each experiment vessel containing Na₂CO₃ solution. Initially, we have added 4 mL of SrCl₂ to the vessel, but precipitates were not formed, and thus we used the mixture of SrCl₂ (4 mL) and BaCl₂ (4 mL), which produced visible precipitates immediately after the addition. The Na₂CO₃ solution (200 mL) were transferred into a 250-mL centrifuge bottle (total volume was 282 mL) and the precipitating agent (4 mL) and 1 N NaOH (2 mL) was added. Therefore, there was approximately 76 mL of headspace (including bottle top space) in the bottle. The number of samples was 10 (two levels of Na₂CO₃ solution × five replicates).

Experiment 2: pH and double distilled water

In the Experiment 2 (Table 1), we changed the conditions of pH adjustment and headspace to solve the errors caused in the Experiment 1. Though precipitating agent was not initially considered, we have also changed the precipitating agent. All the chemical reagents were prepared in double distilled water, but CaCl₂ was used as a precipitating agent. A total of 12 centrifuge bottles were prepared (two levels of Na₂CO₃ solution × two pH treatments × three replicates). The Na₂CO₃ solution (250 mL) were transferred into a 250-mL centrifuge bottle and the precipitating agent (5 mL) was added. 1 N NaOH (2 mL) was added to half of the bottles to raise to pH from 10.2 to 12.2. After that, the remaining headspace were filled with the Na₂CO₃ solution to eliminate any headspace in the bottle.

Experiment 3: pH and tertiary distilled water

In the Experiment 3 (Table 1), we repeated the Experiment 2 by using tertiary distilled water to eliminate the potential interference caused by DIC of the double distilled water.

Precipitation and chemical analyses

The centrifuge bottle containing the samples were left at room temperature for 7 days to allow precipitation. The precipitation time was longer than 2 days (Bishop, 1990) and 6 days (Szynkiewicz et al., 2006) recommend by others. The bottles centrifuged at 5,000 rpm for 10 minutes, and a portion of supernatant (20 mL) was collected and the remaining was discarded. The precipitates were washed again by adding the corresponding precipitating agent (50 mL) followed by centrifugation.

The collected supernatant samples were analyzed for DIC concentration using the TOC analyzer. The precipitates were transferred a 50-mL conical tube, and dried under infra-red lamp. The dried precipitates were crushed to fine powder using a spatula. The δ¹³C of the powder was analyzed using the SIRMS.

Calculation

The recovery of carbonate (CO₃^2-) in precipitates (i.e., precipitation efficiency) was calculated as

Carbonate recovery (%) = [(DIC_i ‒ DIC_s)/DIC_i] × 100

where DIC_i and DIC_s are the concentration of DIC in the initial solution used and in the supernatant solution after precipitation reaction, respectively.

The δ¹³C were calculated as δ¹³C (‰) = [(R_sample/R_standard) ‒ 1] × 1000

where R is the atom % of ¹³C, and the standard was the Vienna Pee Dee Belemnite standards (R=1.12372%). Accuracy of the measurement by the SIRMS tested using IAEA-C6 (sucrose, −10.8‰) was better than 0.2‰.

Statistical analysis

Experimental data were tested for homogeneity of variance and normality of distribution with Levene’s test and Kolmogorov-Smirnov test, respectively, and results showed that the variance was homogeneous and the distribution was normal. The effects of experimental treatments (e.g., DIC level and pH adjustment) were assessed with the analysis of variance (ANOVA) using the general linear model using the IBM SPSS Statistics 23 (IBM Corp., Ammonk, New York, USA) (Table 4). When ANOVA is significant, pairwise comparisons of the means were conducted using the Duncan’s multiple range test. Statistical significance was set at α=0.05.

Results and Discussion

Precipitation efficiency

Precipitation efficiency of DIC was different across the three experiments (Fig. 1), and within each experiment the precipitation efficiency was affected by DIC concentration and/or pH adjustment (Table 3). At a given experimental condition, the precipitation efficiency (85.8 and 96.2% in low and high DIC concentrations, respectively) in the Experiment 1, which was processed by addition of mixture of SrCl₂ and BaCl₂ under raised pH in the presence of headspace, was greater than those in the Experiments 2 and 3 (37.4 - 70.0% at pH 10.2 and 77.0 - 94.1% at pH 12.2) (Fig. 1). This is due to the lower solubility of SrCO₃ and BaCO₃ compared to CaCO₃ (Stenger, 1996); for example, the water solubilities (at 20°C) of SrCO₃, BaCO₃, and CaCO₃ are 0.0011, 0.0022, and 0.0013 g 100 g^-1, respectively (Seidell, 1940).

Across the Experiments 1‒3, the precipitation efficiency was greater under high DIC concentration (Fig. 1 and Table 3) due to increased precipitation reaction through nucleation and crystal growth (Montes-Hernandez et al., 2009). An increase of pH from 10.2 to 12.2 in the Experiments 2 and 3 obviously increased the precipitation efficiency through a shift of carbonate equilibrium (H₂CO₃ ↔ HCO₃^- ↔ CO₃^2-) to the right (Szynkiewicz et al., 2006; Lim et al., 2012). Therefore, these results suggest that precipitates yield of DIC may be more efficient for samples with a high DIC concentration and by addition of SrCl₂ and BaCl₂ rather than CaCl₂.

Fig. 1

Recovery of carbonates in the precipitates (precipitation efficiency) under different experimental conditions. Vertical bars are standard errors of the mean (n=5 for Experiment 1 and 3 for Experiments 2 and 3). Different lowercase letters indicate that the means are statistically different among the treatments within each experiment (see Table 2).

Precision of δ¹³C

Though the precipitation efficiency was high enough for the Experiment 1, the δ¹³C of the SrCO₃·BaCO₃ was negative by more than 9‰ compared to the reference (‒7.45‰) (Fig. 2). In addition, the δ¹³C was not affected by DIC concentration (Table 4). Such errors could be ascribed primarily to the effect of ¹³C-depleted atmospheric CO₂ remained in the headspace and dissolved in the double distilled water as previously mentioned; Szynkiewicz et al. (2006) reported that the δ¹³C of contaminant DIC was ‒24.3‰. It may be also possible that the requirement of a high temperature for thermal decomposition of SrCO₃ and BaCO₃ causes C isotope fractionation, which produces ¹²CO₂ faster than ¹³CO₂ (Harris et al., 1997). Weast (1973) reported that thermal decomposition of SrCO₃ and BaCO₃ occurs at 1,100 and 1,300°C, respectively, while CaCO₃ thermally decomposes at 830°C.

In the Experiment 2, which was conducted by addition of CaCl₂ as a precipitating agent using double distilled water in the absence of a headspace, however, the precision of δ¹³C measurement was not improved (Fig. 2). This result strongly suggests that DIC dissolved in the double distilled water is the main reason of the poor precision of δ¹³C analysis. In the Experiment 3, which was conducted using tertiary distilled water under the same experimental conditions as the Experiment 2, the precision of δ¹³C measurement was greatly improved. The differences in the δ¹³C between the reference and the precipitates was less than 1.0‰, and the δ¹³C of the precipitates produced from the solution with a high DIC concentration (10 mg C L^-1) by adding CaCl₂ without pH adjustment (pH=10.6) in the absence of a headspace was the same as that of reference (Fig. 2). The slightly decreased δ¹³C of precipitates under a high pH (12.6) should be ascribed to the interference of the contaminant DIC from atmosphere; Szynkiewicz et al. (2006) reported that δ¹³C of DIC decreased by 0.03, 0.07, and 0.22 at pH 10.26, 11.38, and 12.11, respectively. Therefore, these results together with the precipitation efficiency data imply that pH increases may facilitate DIC precipitation, but there is a possibility of DIC contamination.

Fig. 2

δ¹³C of the reference (Na₂CO₃) and the precipitates obtained from the Experiments 1‒3. Vertical bars are standard errors of the mean (n=5 for the reference and Experiment 1 and 3 for Experiments 2 and 3). Different lowercase letters indicate that the means are statistically different among the treatments in the Experiment 3, and for the Experiments 1 and 2, there was no statistical significance (see Table 3).

Conclusion

Our results showed that DIC contaminant in double distilled water may cause unacceptable errors in the analysis of δ¹³C of DIC, highlighting the strong necessity of using DIC-free tertiary distilled water. In this context, though not directly examined in the present study, CO₂ in a headspace may also cause similar errors. However, as the effects of pH, headspace, and distilled water on the precision of δ¹³C analysis were not systematically investigated in the present study, we still lack of data on the quantitative errors caused by the conditions of pH, headspace, and distilled water for precipitation. Nevertheless, our results clearly showed that precipitation conditions for reliable δ¹³C measurement are DIC-free tertiary distilled water, samples with a high DIC concentration, CaCl₂ addition as a precipitating agent, pH 10.6, and the absence of a headspace.