Cadmium (Cd) and Nickel (Ni) Distribution on Size-Fractioned Soil Humic Substance (SHS)

Soil humic substances (SHS) are heterogeneous, complex mixtures, whose concentration, chemical composition, and structure affect the transport and distribution of heavy metals. This study investigated the distribution behavior of two heavy metals [cadmium (Cd) and nickel (Ni)] in high molecular weight SHS (HMHS, 1 kDa–0.45 μm) and low molecular weight SHS (LMHS, <1 kDa) extracted from agricultural soils. The HMHS mass fractions were 45.1 ± 19.3%, 17.1 ± 6.7%, and 57.7 ± 18.5% for dissolved organic carbon (DOC), Cd, and Ni, respectively. The metal binding affinity, unit organic carbon binding with heavy metal ratios ([Me]/[DOC]), were between 0.41 ± 0.09 μmol/g-C and 7.29 ± 2.27 μmol/g-C. Cd preferred binding with LMHS (p < 0.001), while Ni preferred binding with HMHS (p < 0.001). The optical indicators SUVA254, SR, and FI were 3.16 ± 1.62 L/mg-C/m, 0.54 ± 0.18 and 1.57 ± 0.15, respectively for HMHS and 2.65 ± 1.25 L/mg-C/m, 0.40 ± 0.17, and 1.68 ± 0.12, respectively for LMHS. The HMHS contained more aromatic and lower FI values than LMHS. Multilinear regression showed a significant positive correlation between the measured predicted [Me]/[DOC] ratios (r = 0.52–0.72, p < 0.001). The results show that the optical indices can distinguish the chemical composition and structure of different size SHS and predict the binding ability of Me-SHS.

The SHS concentration (represented by C content); however, does not fully predict the mobility potential and binding strength of metals. The SHS chemical composition, structure, and molecular weight are important factors that affect the Me-SHS binding capacity. In strongly-binding Me-SHS complexes, metals are stable and will not readily release; this reduces the metal mobile potential and biological toxicity. However, in a weakly-binding Me-SHS complex, metals may release into the environment and affect ecological safety and security, and inhibit crop growth [1,7,[9][10][11][12]. Several studies have used the metal binding affinity [Me]/[DOC] ratio to understand the preferences, distribution, bioavailability, and mobile potential of metal binding with dissolved organic matters (DOM) and SHS [1,9,10,[13][14][15].
The SHS has a wide range of molecular weights [1,2]. The SHS composition and molecular weight are greatly influenced by the source of the organic matter and the biogeochemical processes in the soil [1,2,11,16]. The SHS molecular weight may play an important role in heavy metal binding. A molecular weight of 1 kDa is a common quantity used to distinguish high and low molecular weight of DOM and SHS. Hence, in this study, SHS molecular weight less than 1 kDa was designated as the low molecular weight SHS (LMHS), and the molecular weight higher than 1 kDa was designated as the high molecular weight SHS (HMHS). It is worth noting that the LMHS still contained various heavy metal species (free ion metals, binding with low molecular weight SHS) [17][18][19][20][21][22][23]. Moreover, the low molecular weight DOM and SHS had a metal and hydrophobic organic compound binding capacity [24][25][26].
Previous studies on the distribution and partition of metals between high and low molecular weight organic matter focused on the DOM in an aquatic environment [17,[21][22][23][27][28][29] and a few studies focused on soil solutions [18]. The optical indices were used to examine the influencing factors of the [Me]/[DOM] ratio, including a water/calcium chloride (CaCl 2 )-extracted soil solution and soil leachate [9,10] as well as bulk water DOM [13][14][15]. The widely-ranging SHS molecular weights had different chemical compositions and structure, which may have affected the Me-SHS binding behavior. Information on the distribution and partition of heavy metals in HMHS and LMHS separated by SHS is still lacking.

Site Description and Soil Sampling, Treatment, and Measurement
The study site was located in a paddy experimental field in southern Taiwan (location 23 • 12 39.5" N, 120 • 10 54.0" E). The soil samples were taken with an auger (ID 5 cm) to 60 cm depth and about 2.0 kg was taken for each sample. Eight soil samples were collected from the sampling site each time. In total, 24 soil samples were used to analyze the concentration and optical values of the SHS and heavy metals.
The soil samples were air-dried (about one week) in the laboratory. The soil was passed through a sieve mesh #10 (2.0 mm) and provided for subsequent tests. The basic properties of the soil were measured with the sieved soil (pH, and organic matter (OM)). The soil contained 46%, 33%, and 21% sand, silt, and clay, respectively and was classified as loam soil. The soil pH was 8.07 ± 0.38 and the OM was 2.28 ± 0.09%. The total metal concentrations (Fe, Mn, Cd, Cr, Cu, Ni, and Zn) were analyzed with an aqua regia digestion method. Three g soil was added in 30 mL aqua regia solution; after 16 h settling and 2 h digestion at 180 • C, the solution was measured as 100 mL. Alkaline metals (Ca, Mg, K, and Na) were extracted with a 1.0 M ammonium acetate solution at soil mass/volume = 1/10.

HS Extraction and Separation
Bulk SHS (BHS) was extracted with an alkaline solution. Briefly, 5 g soil was added to 0.1 N HCl 100 mL solution to remove alkaline metals and carbonate. The 100 mL of 0.1 N NaOH was added to the residual soil at the W/V = 1/20. After 24 h shaking, the soil solution was centrifuged at 4500 rpm for 30 min. The NaOH-extracted humic substance was identified as the bulk SHS (BHS) solution. A three-liter BHS solution was used to separate the HMHS and LMHS fractions. A cross-flow ultrafiltration system equipped with 1 kDa nominal molecular weight cutoff ceramic membrane cartridges (Filtanium, France, membrane area 320 cm 2 ) was used to separate the high molecular weight SHS (HMHS, 1 kDa-0.45 µm) and the low molecular weight SHS (LMHS, <1 kDa). The feed flow rate was 1.7-2.0 L/min and the penetration-flow rate was 25 mL/min. The cross-flow ultrafiltration working pressure was kept at 5 kg/cm 2 . In the concentration process, the volume concentration factor (C f ) was kept at 10 (Equation (1)), in which the retentate flow was directed back to the feed bottle; this was the HMHS solution at the end of the separation. The penetration flow was directed to another container to be collected as the LMHS (<1 kDa). The mass balances of the dissolved organic carbon (C content) and heavy metals were calculated by Equation (2). The HMHS and LMHS mass fractions were calculated following Equations (3) and (4).
where C f was the volume concentration factor. R(%) was the mass balance percentage of the separated mass (HMHS added to LMHS) divided by the feed mass (BHS) for DOC, Cd, and Ni. C HMHS , C LMHS , and C bulk were the measured metal concentrations and C content in the HMHS, LMHS, and BHS solutions, respectively. V HMHS , V LMHS , and V bulk were the metal and DOC volumes for the separation procedure for the HMHS, LMHS and BHS solutions, respectively. M HMHS and M LMHS were the mass fractions of DOC and metals in the HMHS and LMHS solutions. The membrane was cleaned and preconditioned before each SHS separation experiment as recommended by manufacture protocol. The three SHS fractions were measured via DOC, UV-vis, and fluorescence spectroscopy.

Dissolved Organic Carbon and Metals Measurement
The BHS, HMHS, and LMHS C contents were measured with a TOC-V analyzer (Shimadzu, Japan). High concentrations of heavy metals and alkaline metals (Ca, Mg, K, and Na) were measured with an atomic absorption spectroscopy (AAS) (Hitachi, Z-2300, Japan). When the metal concentration was lower than the AAS detection limit, the metal concentration was measured with graphite furnace atomization (Hitachi, Z-3000, Japan).

UV-Vis and Fluorescent Measurement
The BHS and sized SHS solutions were diluted to 5 mg-C/L with ultrapure water. The absorbance was measured with an ultraviolet/visible spectrophotometer (Hitachi, U-2900) and fluorescence spectra were recorded on a fluorescence spectrometer (Hitachi F-7000). The absorbance at 700-800 nm was set as the background value. The absorbance of the sample was subtracted from the average of the absorbance at 700-800 nm [30]. The UV-Vis spectrophotometric scanning wavelength was 800-200 nm.
The excitation/emission matrixes (EEMs) were generated by recording emission spectra from 250-550 nm at 2.0 nm steps for an excitation wavelength between 200-450 nm at 5 nm increments. The scanning rate was 2400 nm/min. The value of the blank sample was subtracted from the sample fluorescent data. The UV-Vis absorbance at 254 nm was <0.2 and the inner filter effect correction was ignored.

Optical Index and Metal Binding Affinity Calculation
The specific ultraviolet absorbance at 254 nm (SUVA 254 , L/mg-C/m) is the absorbance of the sample at 254 nm UV 254 (cm −1 ) divided by the C content of the SHS sample (mg-C/L) multiplied by 100 [36]. The UV-Vis spectral slopes in the 275-295 nm and 350-400 nm spectral ranges were determined and were reported as S 275-295 to S 350-400 . The slope ratio S R was calculated as the ratio of S 275-295 to S 350-400 . [30,37]. The Fluorescence Index (FI) is the fluorescence intensity ratio of Em = 450 to Em = 500 nm at Ex = 370 nm [31,32]. The metal binding affinity ([Me]/[DOC] ratio, µmol-Me/g-C) was calculated as the metal concentration divided by the SHS C content.

Statistic Analysis
In this study, linear regression, step-wise regression, and the difference tests used the S-Plus software (V 6.2, Insightful Corporation, Seattle, WA, USA) at significance levels at p < 0.05. Two group difference tests between HMHS and LMHS (such as, concentration, mass fraction, [Me]/[DOC] ratio, and indicators) were used in the t-test method. The three group difference tests used the ANOVA test method. Fluorescence indicators were calculated at R script developed by Lapworth and Kinniburgh [38]. The step-wise procedure, the measured [Me]/[DOC] ratio was set as the dependent parameter and the selected indicators were set as independent parameters. We then conducted the step-wise regression procedure to obtain the significant indicator and coefficients. Next, we used the predicted equation and significant indicators and coefficients to calculate the predicted [Me]/[DOC] ratio.

UV-Vis and Fluorescent Index
Optical indicators have been used extensively and are effective tools for examining the chemical composition and structure of DOM and dissolved SHS [3,9,10,16,39,40]. Table 1 lists the SUVA 254 , S R , and FI values of BHS, HMHS, and LMHS. The SHS optical indices were comparable to reported values in soil solution, where SUVA 254 values ranged from 0.38-7.18 L/mg-C/m [3,9,10,16,40], S R values ranged from 0.64-3.26 [16,39], and FI values ranged from 1.08-2.03 [3,16,31]. The SUVA 254 is positively correlated with the aromatic content [41]. When SUVA 254 is <3 L/mg-C/m, the SHS solution consists mainly of hydrophilic aromatic compounds, while when SUVA 254 is >4 L/mg-C/m, the composition of the SHS solution is dominated by hydrophobic compounds [34]. The average HMHS SUVA 254 value was higher than the average LMHS SUVA 254 value but insignificantly different (p = 0.23). The SUVA 254 value showed that the SHS solution mainly contained hydrophilic and low aromatic compounds. S R is the indicator of average molecular weight [30]. S R values show that HMHS was significantly greater than LMHS (p = 0.007). Although, Helms et al. [30] suggested that S R was negatively correlated with mean molecular weight, the S R of this study was significantly positively correlated with the SUVA 254 and significantly negatively correlated with the FI, respectively (r = 0.842 for SUVA 254 , r = −0.765 for FI, p < 0.001, n = 72). HMHS had higher S R values than LMHS (Table 1), which confirmed that higher S R values had higher aromaticity, molecular weight and terrestrial sources in this study.
FI is the relative contribution of terrestrial sources [3,31,32,42]. FI values less than 1.4 indicated the HS had strong terrestrial sources and when larger than 1.9, had microbial sources [31,32]. HMHS and LMHS both had median terrestrial sources and HMHS had a higher terrestrial contribution than LMHS (p = 0.006). FI values had a significantly negative correlation with SUVA 254 values (r = −0.735, p < 0.001). The optical indices showed that the HMHS had higher aromaticity, average molecular weight, and terrestrial sources than LMHS.
The BHS C content was 678 ± 247 mg/kg and the BHS-C content/TOC ratio was 3.7%. In four agricultural soil tests by Fernández-Romero et al. [3], organic matter ranged from 4.3-15.6%. The water extraction organic carbon (WEOC) was 80-620 mg/kg at room temperature; the WEOC/OM ratios were 0.12-0.40%. In another study, Nkhili et al. [40] reported soil was extracted with different water temperature (20, 60, and 80 • C). The WEOC was 310-401 mg/kg, and WEOC/TOC ratios were 0.93-1.16%. The alkaline-extracted soil organic carbon in the study was slightly higher than water-extracted soil organic carbon at room temperature. Table 2 lists the total concentrations of Cd, Cr, Cu, Ni, and Zn and the BHS concentrations of these heavy metals. The total metal concentrations are generally similar to the concentration of the uncontaminated agricultural soils [4,11,43,44]. The BHS concentration of Cu (1.19 mg/kg) was higher than the concentrations of the other metals; all other metals were below 1.0 mg/kg. The ratios of BHS to the total metal concentration shows that the ratios of Cd to Cu had higher ratios (15.4% and 12.1%, respectively) than the other metals Ni (1.07%), and both Cr and Zn < 1%. Cambier et al. [11] reported that "the ratios suggested that Cu and Cd preferentially combined with soil humic substances." Matong et al. [4] reported that three agricultural soils were sequentially extracted with acetic, ascorbic and hydrogen peroxide digestion. The exchangeable, and organic binding fractions were 52-61%, and 24-32% for Cd, and Ni, respectively. The fractions were similar to our results that showed Cd was higher than Ni by exchangeable and organic fraction.  Table 3 lists the C content and concentrations of Cd and Ni of BHS, HMHS, and LMHS. This study investigated the metals Cd and Ni because they had accepted metal mass balance. However, the number of mass balances within reasonable range (100 ± 25%) [22,26,27,45] were 15, 10, and 6 for Cr, Cu, and Zn, respectively for the separation. We ignored these data and the distribution of heavy metals Cr, Cu and Zn was not investigated in later discussions. The mass balances of DOC, Cd, and Ni in the separation process were calculated following Equation (2). The average mass balances were between 83-90%, which were within a reasonable range.
Wang et al. [18] used a mixture of acetic acid (formic, and malic acid) to extract soil solutions, which were separated into HMHS (1 kDa-0.45 µm) and LMHS (<1 kDa). Most metals had a high LMHS fraction (71% (Cd) and 82% (Ni)). Ilina et al. [47] used ultrafiltration to separate water-extracted soil solution; the HMHS were 65%, and 83% for Cd, and Ni, respectively. Hartland et al. [48] observed hyperalkaline cave drip-water, the HMW fractions (1 kDa-1.0 µm) were 64% for Ni, which were similar to the results in our study. In four treatment processes in a municipal wastewater treatment plant, Hargreaves et al. [23] reported the HMW fractions (1 kDa-0.45 µm) were 67-75% for Ni [23]. In sediment pore water, Dabrin et al. [22] reported the HMW fractions (5 kDa-0.45 µm) were 9-18%, and 5-20% for Cd, and Ni, respectively (estimated from Figure). The mass fractions of high and low molecular weights in water extracted organic matter and DOM are variations. The fractions may depend on the matrix, type of metal, extraction solvent, extraction method and solid/liquid ratio, and separation method and conditions [18,49].

Correlation between [Me]/[DOC] Ratios and Optical Indices
We used the Pearson correlation analysis to explore the SHS factors affecting the [Me]/[DOC] ratios. Table 4   The ratios of [Me]/[DOC] HMHS had no correlation with the HMHS optical indicators, which may be due to the complex composition of HMHS containing biopolymers [42]. It is believed that humic acid (HA) and fulvic acid (FA) contain various functional groups that have a strong metal-binding ability. The HA and FA molecular weights range from sub-kDa to several kDa [1,42]. The LMHS indices S R and SUVA 254 had a strong correlation with [Cd, Ni]/[DOC] ratios, which suggested that LMHS (<1 kDa), still containing low molecular weight HA and FA, have the potential to bind to Cd and Ni. Meanwhile, the ([Cd]/[DOC]) LMHS ratio had a negative correlation with the terrestrial sources index, which suggested that the HS high terrestrial sources had a high Cd-HS binding ability [50,51]. The ratios of [Cd, Ni]/[DOC] HMHS were insignificantly correlated with the indices, which implied a complex composition of HMHS, with aromatic content and molecular weight, cannot predict the Me-HS binding ability. The high molecular weight sources may contain biopolymers. Table 4 Table 5 shows two types of regression equations; one was predicted with the three selected indicators and the other equation predicted with the three indicators and the C content. When the C content was added as a predicted indicator, the determined coefficients were slightly increased compared to the regression without the C content. The C content coefficient was negative, which is the same as the result in Table 4 and is consistent with the previous study [9,10].   In the studied HS, Cd preferred combining with low molecular weight HS, so it is negatively correlated with the molecular weight indicator (SR) and positively correlated with aromaticity (SUVA254). However, Ni preferred combining with high molecular weight; therefore, it positively correlated with the molecular weight indicator (SR) and negatively correlated with aromaticity (SUVA254). The coefficient values of FI for both Cd and Ni were positive, which indicated that higher terrestrial sources correlated with lower Me-HS binding ability. The Cd and Ni preferred to combine with the SHS microbial source. The results of the [Me]/[DOC] ratio predicted by the indicators show that the combination of Me-SHS is affected by both the C content and the chemical structure of the SHS. In the studied HS, Cd preferred combining with low molecular weight HS, so it is negatively correlated with the molecular weight indicator (S R ) and positively correlated with aromaticity (SUVA 254 ). However, Ni preferred combining with high molecular weight; therefore, it positively correlated with the molecular weight indicator (S R ) and negatively correlated with aromaticity (SUVA 254 ). The coefficient values of FI for both Cd and Ni were positive, which indicated that higher terrestrial sources correlated with lower Me-HS binding ability. The Cd and Ni preferred to combine with the SHS microbial source. The results of the [Me]/[DOC] ratio predicted by the indicators show that the combination of Me-SHS is affected by both the C content and the chemical structure of the SHS.

Conclusions
Separating soil SHS into different molecular weights allows understanding of the different chemical structure and composition of HS and HS-Me binding ability between the HMHS and LMHS. The Cd and Ni were not evenly distributed in HMHS and LMHS. The separated SHS showed that Cd