An improved method for isotopic and quantitative analysis of dissolved organic carbon in natural water samples Open Access

Dissolved organic carbon (DOC) is the major form of organic carbon (OC) exported from the watershed (David et al. 1992), and the flux of DOC transported to the ocean via the river system was about 200 Mt C yr⁻¹ (Kirschbaum et al. 2019). DOC is closely related to the acid–base chemistry of acid-sensitive freshwater systems, as well as mobility, persistence, and toxicity of metals and other pollutants (David et al. 1992). DOC carbon isotope (δ¹³C_DOC) has been widely used to track DOC sources, transport, and fate in the watershed (David et al. 1992). The measurement of DOC content and δ¹³C_DOC ratios were helpful to solve these problems and identify the impact of human activities in the aquatic environment (Zhang et al. 2020b; Li et al. 2024). Hence, accurate determination of DOC content and δ¹³C_DOC ratios were essential to satisfy these demands. Previous studies have conducted much research on the pre-treatment method for the above parameters to obtain stable and believable results (Keys 1934; Sharp 1973; Romankevich 1984; Skrzypek 2013).

High-temperature combustion (HTC) (Lang et al. 2007; Federherr et al. 2014) and wet chemical oxidation (WCO) (Osburn & St-Jean 2007; Zhou et al. 2015) were two predominate pre-treatment methods for determining DOC content and δ¹³C_DOC ratios. These methods involve oxidizing DOC to carbon dioxide (CO₂) (Federherr et al. 2014), which is then introduced into a stable isotope ratio mass spectrometer (IRMS) with helium to analyze the m/z 44 signal value and isotopic composition of CO₂ (Gandhi et al. 2004). The primary difference between these two methods is the oxidative pathways of DOC (Romankevich 1984; Lang et al. 2012). DOC was transferred to CO₂ via a catalyst and high-temperature furnace combustion in the HTC method (Skopintsev et al. 1966; Romankevich 1984; Federherr et al. 2014), but through chemical oxidants (K₂O₂, H₂O₂, KMnO₄, K₂Cr₂O_7, K₂S₂O_8, and Na₂S₂O₈) in the WCO method (Menzel & Vaccaro 1964; Skopintsev 1976; Mckenna & Doering 1995; Yu et al. 2015; Skrzypek & Ford 2019). While the WCO method may not achieve complete oxidation compared to HTC, it significantly saves time by eliminating laborious steps like freeze-drying. Moreover, its results are stable, and contamination and blank issues are controllable (Romankevich 1984; Lang et al. 2012). The persulfate oxidant has been established as the standard method for seawater DOC determination by Menzel and Wilson (Wilson 1961; Menzel & Vaccaro 1964). Later, Zhou developed the oxidation conditions of the persulfate wet oxidation method (Zhou et al. 2015) and found that the precision using Na₂S₂O₈ as persulfate oxidant was better than that using K₂S₂O₈ oxidant (Zhou et al. 2015). Meanwhile, the incorporation of silver nitrate catalyst and heating in a water bath at 100 °C for 1 h facilitated the full oxidation of OC (Zhou et al. 2015).

In addition, dissolved inorganic carbon (DIC) must be eliminated before the oxidation of DOC because the CO₂ produced by DIC will interfere with the DOC content and δ¹³C_DOC ratios (Yu et al. 2015; Skrzypek & Ford 2019). Several DIC removal methods have been developed and modified previously (Assayag et al. 2006; Yu et al. 2015). The conventional techniques included the direct precipitation of DIC into carbonates and evolved gas for acidifying DIC with phosphoric acid (Assayag et al. 2006; Zhou et al. 2015; Yu et al. 2015). Previous studies have demonstrated that the DIC removal is not complete with such methods (Lang et al. 2012). The CO₂ produced by the DIC of the sample is partitioned between the gaseous and dissolved phases in the vials and a carbon isotope fractionation between gaseous and dissolved CO₂ results from this partition (Lang et al. 2012). For example, Zhou et al. found that the testing accuracy of solid standard samples when establishing the method was better than 0.2‰, but once the method was applied to groundwater, the precision decreased to ∼0.35‰ (Zhou et al. 2015). Therefore, it is necessary to keep the same circumstance and procedure for substance used to compare impacts from variable factors on DOC contents and isotope measurement (Rock & Mayer 2006). This defect reminded us that dissolved solid reference would produce blank error from ultrapure water. Hence, if more water was pre-treated to elevate the CO₂ signals, the blank error due to the large volume of ultrapure water input should be considered. Therefore, it is necessary to investigate the effect of water volume and DIC removal effects on the (m/z 44) signals of the CO₂ generated by DOC.

In this study, we tried to increase the water volume to produce enough CO₂ to obtain better precision and accuracy of measurement by Gasbench II-IRMS. To remove the internal DIC, we used the nitrogen purging process under the ice bath. Organic and inorganic carbon solids with known δ¹³C values were dissolved into ultrapure water as DOC, DIC, and mixed standard solutions. These standard solutions were oxidized, and their measured δ¹³C values were compared with their real δ¹³C values to check the effectiveness of our improvements.

Recent research has demonstrated that incomplete oxidation of samples may cause isotope fractionations during WCO (Skrzypek & Ford 2019). To constrain this disturbance, four solid reagents were chosen as the laboratory reference substances for DOC, which included potassium hydrogen phthalate (KHP) and acetylaniline (ACE), sucrose (SUC), and L(+)-glutamic acid (GLU) (guaranteed reagent, GR, Shanghai Aladdin Biochemical Technology Co., Ltd). Their δ¹³C values were measured using an Elemental Analyzer (EA, Flash 2000 HT, Thermo) coupled with a MAT-253 IRMS (Thermo Fisher-Scientific, Bremen, Germany) (Zhang et al. 2020a) (Table 1), and international standard substance IAEA-600 (δ¹³C = −27.771‰) was used to calibrate their isotope values.

In the aquatic environment, bicarbonate was the dominant component of DIC in the stream water (Das et al. 2005; Barnes & Raymond 2009). Therefore, NaHCO₃ (guaranteed reagent, GR, Shanghai Aladdin Biochemical Technology Co., Ltd) was selected as the DIC standard material in the laboratory. The δ¹³C ratios of solid NaHCO₃ were obtained by the traditional phosphoric acid method (Krishnamurthy & Atekwana 1997). Four to six drops of phosphoric acid (density 1.91) were injected into a 12 mL Exetainer tube (Labco, High Wycombe, UK) where solid NaHCO₃ (about 100 μg) was installed and purged by helium gas (99.999%) in advance. The headspace CO₂ was transported and analyzed using MAT 253 equipped with an online Gasbench II interface (Du & Song 2020). International standard carbonate solid NBS 18 (δ¹³C = −5.014‰) was used to calibrate the δ¹³C ratios of our laboratory reference. The δ¹³C ratios of solutions were also determined by the phosphoric acid method (Spötl 2005). About 0.5 mL of phosphoric acid (85% (V/V)) was injected into a 12-mL Exetainer tube (Labco, High Wycombe, UK) and purged with Helium gas (99.999%) to remove residual CO₂. Then, a syringe injected about 0.5–2 mL solutions into the 12-mL Exetainer tube. The headspace CO₂ was transported and analyzed through MAT 253 coupled with an online Gasbench II interface (Du & Song 2020). There is no obvious alternation of δ¹³C ratios between solid NaHCO₃ and its solutions due to no DIC in the ultrapure water.

To eliminate the background DOC interference in ultrapure water, we chose four types of ultrapure water and compared their DOC contents followed by our established procedure. These four types of ultrapure water included commercial bottle water with the brand Wahaha, C'estbon, Watson, and our laboratory ultrapure water (Supplementary material, Table S1). No solid reference substance was added to these ultrapure water samples, and the CO₂ peaks and DOC contents from these ultrapure water samples were listed in Supplementary material, Table S1.

To obtain DOC concentrations of water samples tested, a relation between the known DOC concentration and produced CO₂ amount must be established. KHP was used as a DOC reference because it was not easy to oxidize by WCO and has high solubility (Table 1). If a good linear relationship could be obtained with KHP solutions, other similarly oxidized organic substances could also be used to establish this similar linear relationship. Three DOC concentration gradients of 3, 6, and 9 mg L⁻¹, respectively, with four water volumes of 10, 20, 30, and 40 mL, were made to establish the above relations (Supplementary material, Table S2).

DIC reference solutions were prepared using NaHCO₃ to limit the effect of different DIC removal methods on the finial δ¹³C value (Supplementary material, Table S3). In addition, to verify the feasibility of the improved DIC removal method, DIC/DOC mixed standard solutions were also prepared using NaHCO₃ and KHP (Supplementary material, Table S3).

The glassware used in the experiment was soaked with 1 mol L⁻¹ of sodium hydroxide (NaOH, GR, 99.9%, CAS: 1310-73-2, Aladdin reagent) overnight and was cleaned by ultrapure water, and then calcined at 500 °C in the Muffle furnace for 5 h to remove organic matter completely.

The reference solutions in Supplementary material, Table S2 were added 500 μL of 85% H₃PO_4, then shaken and still for 12 h. Two milliliters of oxidant (50 mL H₂O + 2 g Na₂S₂O₈ + 100 μL 85% H₃PO₄) and 100 μL catalyst made of 0.1 mol L⁻¹ AgNO₃ (≥99.8%, CAS: 7761-88-8, Sinopharm Chemical Reagent Co. Ltd) were added into the liquids (Zhou et al. 2015). High-purity helium of 99.999% was used to remove residual CO₂ at 20 mL min⁻¹ for 10 min. Ultimately, these bottles were heated on a graphite furnace under 100 °C for 1 h and then cooled to room temperature (Sharp 1973).

After acidification, the dissolved CO₂ in solutions should be removed to eliminate their disturbance on finial isotope ratios. This study did not include a control experiment to investigate the necessity of ice-bath conditions during the DIC purging process. Following previous research findings, lower temperatures are more conducive to the recovery of volatile organic carbon (VOCs) (Xu et al. 2022). Therefore, in this study, high-purity nitrogen gas (99.999%) was utilized at 4 °C in an ice bath to remove dissolved CO₂. The DIC and DOC/DIC mixed reference solutions were utilized to test the effectiveness of the DIC removal method (Supplementary material, Table S3).

River water, lake water, and rainwater were used to test our improved method. These samples were filtered using a pre-burned Whatman GF/F filter (burning at 450 °C for 5 h) with a porosity of 0.7 mm and stored in a pre-treated glass bottle (Ge et al. 2020). All samples were added with two drops of saturated HgCl₂ (Kang et al. 2019) and kept in the dark environment at ≤4 °C to limit microbial activity. The improved method pre-treated the natural water samples to obtain their DOC contents and δ¹³C_DOC ratios.

The DOC content was obtained by the linear relationship between DOC amount (μg) and the CO₂ peak (m/z 44, mV) of standard KHP solutions. The oxidized CO₂ was carried by Helium gas and reloaded to the MAT 253 through a GasBench II peripheral equipment.

In order to calibrate the contribution of program blanks to isotope results, the CO₂ peaks and δ¹³C_blank of blank samples were calculated by indirect method, and compared with the peak and δ¹³C values of blank samples determined directly. KHP and SUC at different concentrations were determined to calculate M_blank and δ¹³C_blank. Its isotope range is wide, and it can basically cover δ¹³C_DOC of most natural water samples.

The linear relationship between δ¹³C values and CO₂ peaks a reciprocal (1/M) is observed from Equation (2). According to the keeling plot theory, two standards linear equations of δ¹³C values and CO₂ peaks reciprocal are established respectively. The slope is expressed as k_KHP, k_SUC and the intercept is b_KHP, b_SUC.

Because ultrapure water had minor DIC contents, we directly oxidized them with the oxidant and catalyst after phosphoric acid acidification (part 2.3). From Supplementary material, Table S1, it was observed that distilled water yielded the lowest CO₂ peak values after the WCO process, indicating that DOC could be effectively removed through distillation. Therefore, in the subsequent experiments, Watson distilled water was used to address these solid references and obtain standard DOC solutions. Additionally, DOC blank samples prepared without any substances were also used to quantify the impacts of this distilled water. After reducing these distilled water effects, the DOC content and isotope ratios of reference materials and natural water samples were precisely obtained.

It can be seen from Equation (4) that the smaller the volume of the upper space, the more the δ¹³C_H value in the headspace bottle can represent the real carbon isotope value of the sample. When the volume of solution in the 60 mL headspace bottle is 30 mL or 40 mL, the peak value can be achieved, but the carbon isotope values are closer to the real values. Besides, the consistency of sample preparation was ensured to follow the same treatment and analysis principles as natural water samples. The improved method is better applied for the pre-treatment of DOC content and isotopic value determination of water samples than the method established to utilize solid reference materials. The final volume of wet oxidation water samples can be determined according to natural water with different DOC concentration ranges. In the subsequent experiments, the focus was primarily on the 40 mL water volume.

This result is similar to the reproducibility (<0.21‰) of the δ¹³C values obtained by liquid chromatography-IRMS for humic acid samples in the mixed solution (Federherr et al. 2014). In comparison to the pre-treatment method of freeze-drying used by Federherr et al., our pre-treatment method offers advantages such as controlled blanks, cost-effectiveness, and time savings (Federherr et al. 2014). Additionally, the precision achieved in this experiment was not only lower than the research results of humic acid samples using liquid chromatography-IRMS (<0.35‰) (Skrzypek & Ford 2019), but also better than the test results of phthalic acid samples using GasBench-IRMS (0.4‰) (Lang et al. 2012), indicating that the improved method is more effective to remove the extra DIC.

The parallelisms of δ¹³C values from the four reference samples were good, and no obvious memory effects were observed. It proves that the mutual interference between samples could be ignored in the GasBench-IRMS test. The significant error observed in the 160 μg L(+)-GLU measurements may be attributed to gas leakage in the headspace vial during the pre-treatment process (Figure 4(a)), resulting in isotopic enrichment (Figure 4(b)). Furthermore, compounds containing more nitrogen, such as ACE and L(+)-GLU, substantially differed between their δ¹³C_GB values and δ¹³C_EA values (Figure 4(b)). These differences probably arose from the conversion of nitrogen in the compound, which led to stable isotope fractionation (Skrzypek & Ford 2019). This phenomenon could potentially serve as a limitation of the present methodology.

Ultimately, the m/z 44 signal values and δ¹³C values produced by CO₂ in the headspace bottles could be measured through GasBench II-IRMS. No modifications were made to this procedure.

After the pre-treatment following the above procedure, the DOC contents and δ¹³C_DOC values of the raw water samples could be obtained using reference solutions with known DOC contents and δ¹³C_DOC values. It was worth noting that the water samples and reference solutions should be dealt with under the same volume of 40 mL. DOC blank examples with the distilled water were also prepared to get the CO₂ peak background values. After that, the CO₂ peak from reference solutions represented the mixing of DOC from reference materials and distilled water. Consequently, the linear relationship between DOC contents and CO₂ peak from only reference materials was established after excluding the CO₂ peak from distilled water. This linear relationship could be utilized to calculate the DOC content in natural water samples due to no distilled water in them. For δ¹³C_DOC values of water samples, a linear relationship between δ¹³C_DOC raw values and true values was established by two reference materials, including sucrose and potassium hydrogen phthalate with δ¹³C values of −12.9‰ ± 0.2‰ and −29.1‰ ± 0.2‰, respectively (Table 1). Similar to DOC content determination, the δ¹³C_DOC values of reference solutions covered the δ¹³C values of both solid references and distilled water, so the above linear relationship was obtained only when δ¹³C values of distilled water were excluded.

The effective application of stable isotope analysis technique relies on achieving high precision and accuracy in determining isotopic composition (Lang et al. 2012; Zhou et al. 2015). However, there are certain limitations in the pre-treatment methods for stable carbon isotopes. For instance, this study identified that nitrogen conversion in samples may lead to carbon isotope fractionation, thereby causing varying degrees of impact on the results. This limitation has not been sufficiently addressed and necessitates further investigation for resolution. Furthermore, potential errors stemming from sample preparation and handling procedures can pose challenges in interpreting the complexities of environmental factors and biological processes. Therefore, prior to conducting this research methodology, several crucial considerations need to be taken into account. First, it is imperative to exercise caution during the constant heating process at 100 °C to prevent loosening or fracturing of bottle caps, as this could lead to the ingress of atmospheric CO₂ into the bottles or the escape of bottle gases, potentially influencing the outcomes. Second, fluctuations in DOC levels may introduce deviations in DOC quantification within the test solution, with an increase in the test solution volume exacerbating the impact of background effects. Hence, ensuring a sufficiently low DOC content in the blanks and deducting the corresponding background values during data analysis are pivotal measures.

Due to the low accuracy of the determination of DOC contents and δ¹³C values of natural water with low DOC concentration, we increased the volume of water samples to be oxidized by the WCO method. Nitrogen gas purging in an ice bath within 2 h was used to remove DIC, and its effectiveness was guaranteed by DIC solutions and DIC/DOC mixed solutions with known δ¹³C values. The distilled water had low DOC contents and could dissolve solid standard references. After excluding the background values of DOC contents and δ¹³C values from distilled water, good linear relationships between DOC amount and CO₂ produced by WCO, and δ¹³C raw values and true values were established to determine the DOC contents and δ¹³C values of natural water samples under the same conditions of 40 mL solutions. Our improved pre-treatment methods obtained good accuracy and stability on natural water samples with low DOC concentrations of ∼0.5 mg L⁻¹. Moreover, this method saved much pre-treatment time and realized batch processing of natural water samples through three-way valves in the purge process.

This work was supported by the National Natural Science Foundation of China (NSFC, No. 42073009, 41573095). We thank Shuangshuang Zhu for her assistance in the collection of the water sample and Yongtao Wang for his assistance in the establishment of the pre-treatment methods. The authors would also like to appreciate Associated Editor Dr Lucy Ibbotson and three anonymous reviewers of an earlier version of this paper for their consistent and on-target suggestions.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.