Environ. Sci. Technol. 2007, 41, 4895-4900
Characterization of the Polarity of Natural Organic Matter under Ambient Conditions by the Polarity Rapid Assessment Method (PRAM) F E R N A N D O L . R O S A R I O - O R T I Z , * ,† SHANE SNYDER,† AND I. H. (MEL) SUFFET‡ Southern Nevada Water Authority, 1350 Richard Bunker Avenue, Henderson, Nevada 89015, and Environmental Science and Engineering Program, University of California at Los Angeles, School of Public Health, Room 46-081, CHS, Charles E. Young Drive South, Los Angeles, California 90095
The polarity rapid assessment method (PRAM) characterizes the polarity of aqueous natural organic matter (NOM) by quantifying the amount of material adsorbed onto different solid-phase extraction (SPE) sorbents. The analysis is performed under ambient conditions resulting in the elimination of pretreatment steps that may alter the chemical characteristics of the NOM, allowing an accurate representation of its polarity as it exists in the environment. Additionally, analysis only requires 200 mL of sample and can be performed in 2 h. In this paper, the underlying theory of the method is presented, followed by its optimization, with emphasis on the development of conditions for the analysis of NOM in natural waters. A series of organic probe compounds showed that the most important physicochemical property describing the interaction between the NOM and the SPE sorbents was the hydrophobic surface area, allowing for the estimation of the hydrophobic character under ambient conditions. Evaluation of the effects of chemical concentration, pH, and ionic strength show that (1) concentration did not have an effect on PRAM characterization as long as the pH and ionic strength remained constant; (2) changes in pH and ionic strength resulted in considerable changes in PRAM characterization, as a result of the changes in configuration of the NOM; and (3) PRAM characterization of NOM can be completed in the concentration range of 0.50 correlations are reported). No estimate for k50 for C18 was performed due to the long times to reach initial breakthrough. The three nonpolar sorbents, C2, C8, and C18, were all designed for the extraction and analysis of hydrophobic analytes from aqueous matrices (13). As a result, it is expected to observe strong correlations between the retention onto the SPE and the hydrophobicity of a chemical or mixture of chemicals. For C8, the strongest correlation was observed with hydrophobic solvent available surface area (HPO SASA) with an r value of 0.87. Additionally, the correlation between k50 and log ko/w for C8 had an r ) 0.55. An inverse relation, described by an r ) -0.58 between k50 for C8 and Hansen solubility parameter H-bonding (HSP-HB) was also observed. The HSP-HB described the ability of a molecule to form hydrogen bonds. The correlation between C8 and carbon π area was -0.51, suggesting that the addition of aromaticity resulted in a decreased interaction. These results indicate that retention onto C8 and probably C18 is indeed dominated by the hydropbobicity of the material, indicating that these two sorbents will quantify the hydrophobic character of the NOM. The polar SPE sorbents (Diol, Silica, and CN) are used for the sorption of polar analytes from nonpolar and aqueous matrices. The retention mechanism for these SPE sorbents was described as polar interaction (13), in which dipoledipole and hydrogen-bonding would dominate the interaction between NOM and the sorbents. As a result it was expected that correlations between k50 and properties such as HSP HB and other measures of hydrophilic (HPI) SASA would be significant. However, the analysis revealed no such correlation between HSP HB and any other HPI measure, although a strong correlation between the same descriptors used for HPO SASA was observed. This is a surprising result since the SPE literature described these sorbents as used for the extraction of polar compounds (13, 20) and may be due to an unknown interaction related to a hypothesized water or silicic acid displacement mechanisms. Water must be displaced from the sorbent’s active sites before the interaction can occur, limiting the sorption. In natural water, silicic acid could be attached to NOM therefore the silicate ion would be the displaced ion for sorption to the polar sites, (21). Further studies will examine this alternate displacement theory. No correlation analysis was performed for NH2 and SAX. For these anionic sorbents, the primary mechanism is anion exchange, which has been well documented for the use with NOM (19, 22, 23). Results confirm that when charged probes 4898
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
FIGURE 4. Polarity characterization for IHSS NOM (RO isolates) samples. Conditions: 2.00 mgC/L, pH 2.5 and 8 with phosphate buffers (total ionic strength 0.01 M). are used, no retention is observed on any other sorbent other than the two anion exchangers, NH2 and SAX. The analysis of probes was done using k50 instead of RC, since the breakthrough curves did not follow the same model as the breakthrough curves for NOM samples. Table 2 shows good correlations between the k50 and chemical and physical properties. As a result, it could be stated that RC for specific materials and NOM would be dominated by the same mechanisms described for probes. Effect of Sample pH on PRAM Analysis. NOM standards at two pHs (2.5 and 8.0) were used to test the polarity results with UVA 254 nm detection. Figure 4 presents the results for both samples. There were significant differences between the results at the two pHs analyzed. At pH 2.5, the anionic functional groups on the NOM are protonated, allowing the material to interact with the SPE sorbents, as evidenced by RC > 0. Under these conditions, we observed a C18 RC of 0.65, indicating that around 65% of the material which absorbs UVA is nonpolar, compared to approximately 20% having polar character (Diol RC ) 0.2). At pH 8, the NOM is charged, causing the retention to be minimal. These data offer insight on the overall change in structure of NOM when it undergoes changes in pH. The polar and nonpolar parts of the NOM mixture observed at pH 2.5 are not observed at pH 8 indicating that the overall structure of the NOM at pH 8 is charged and allows minimal polar or nonpolar interactions. Effect of Sample Concentration and Ionic Strength. Due to the range of concentrations of NOM and ionic strengths found in natural waters, it is important to understand how these two parameters affect PRAM results. These two parameters were evaluated using two natural water samples (LVW and HR) for which 1:5 dilutions were prepared with and without ionic strength adjustment and the results were compared to the original sample. Figure 5a presents the results for the HR sample. The results for both the raw sample and the 1:5 dilution with ionic strength adjustment were within experimental error. The 1:5 dilutions without adjustment of the ionic strength to the original levels were lower for all the sorbents. The diluted sample showed different PRAM results than those observed for the raw and the diluted sample with ionic strength adjustment. Figure 5b presents the results for the LVW sample. Both raw samples and 1:5 dilutions with ionic strength adjustment were within experimental error. The diluted samples without ionic strength adjustment were within error from the raw and adjusted samples (except for Diol), indicating that the effect of ionic strength was less pronounced than for the HR sample. For natural samples, the changes in concentration had a minimum effect on the results as long as the ionic strength remained constant. For organic probes, there was a con-
FIGURE 5. UVA retention coefficients obtained for the (a) HR (DOC ) 10.0 mg C/L) and (b) LVW (DOC ) 9.6 mg C/L) samples at original conditions and at 1:5 dilutions with and without adjustment of ionic strength (with NaCl).
FIGURE 6. UVA- and DOC-based retention coefficients obtained for sample collected at the Sweetwater reservoir. Sample DOC ) 26.9 mg C/L. centration effect. The retention of resorcinol onto the SPE sorbents increased with concentration (see Supporting Information). Analysis of Natural Waters: UVA and DOC Detection. Figure 6 presents the results for the sample collected at the Sweetwater Reservoir, with UVA and DOC quantification. The DOC for this sample was 26.9 mgC/L. The evaluation was to ensure that subtraction of the background DOC levels from each SPE sorbent could be done for a comparison to UVA only. For this sample the DOC-based RCs were consistently higher than the UVA-based RCs, except for CN. DOC-based RC suggest around a 65% nonpolar character, and nearly 40% polar character for these samples. Based on the UVA data, we observed that nonpolar character dominates, evidenced by higher retention on C18 and C2. RC for
FIGURE 7. UVA and DOC based retention coefficients obtained for the (a) HR (DOC ) 10.0 mg C/L) and (b) LVW (DOC ) 9.6 mg C/L) samples. Silica and Diol were close to zero, indicating that the aromatic portion of the NOM is less polar. It should be noted that the RC for different SPE sorbents does not add up to unity. This indicates that structurally NOM has different domains of variable polarity. Therefore we cannot classify NOM solely as polar or nonpolar, since the same moiety may have both characters. As a result, characterization is performed by estimating the polar and nonpolar character by performing parallel evaluations with the SPE sorbents. This is advantageous since in some instances the overall interest resides on the characterization of NOM as a mixture including all sites with multiple polarity domains. This is also a distinction between the XAD method used and the PRAM, which makes any comparison impossible. The RCs for the HR sample are presented in Figure 7a for a DOC of 10.0 mgC/L. For nonpolar sorbents, the UVA-based RCs were larger than the DOC-based RCs. This effect could be explained by the possibility of smaller molecules within the NOM having higher UVA. These molecules may be relatively insignificant based on total mass but may significantly affect UVA. For polar SPEs the DOC-based RCs were higher than the UVA RCs, indicating that a higher proportion of polar NOM may have low aromaticity. It was observed that the DOC-based RC for CN was ,0, due to the chromatographic effect (15). In this case, the RC was reported as zero (see Supporting Information). Figure 7b presents the results for the characterization of the LVW sample. The same effect is observed, with the nonpolar sorbents having greater UVA-based RCs than DOC-based RCs. For the polar sorbents, the order is reversed, with the DOC-based RCs being larger. Applications of PRAM and General Considerations. The use of PRAM for the characterization of the polarity of DOM is advantageous since it allows the analysis of NOM under VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4899
ambient conditions. The procedure requires limited sample (approximately 200 mL of sample) and can be done on a regular basis, allowing the assessment of temporal and spatial variations in NOM properties. Furthermore, PRAM is currently the only characterization method which allows a multidimensional polarity characterization of the DOM under ambient conditions. The results are reproducible, with coefficients of variation less than 5%. Both pH and ionic strength are known to affect the structure of NOM (10). These two parameters will determine the configuration of the NOM in the environment, and therefore the results obtained should be reported with pH and ionic strength values. In addition, the breakthrough curves will be influenced by the overall concentration of NOM in water. It is recommended that PRAM characterization be done under the range of concentrations (
