Solid-Liquid Extraction with Low Temperature Purification Coupled with Gas Chromatography and Mass Spectrometry for Determination of Polychlorinated Biphenyls in Sewage Sludge

Supplementary Information Solid-Liquid Extraction with Low Temperature Purification Coupled with Gas Chromatography and Mass Spectrometry for Determination of Polychlorinated Biphenyls in Sewage Sludge Mariana R. Maia, Ana Luíza P. Arcanjo, Gevany P. Pinho* and Flaviano O. Silvério Instituto de Ciências Agrárias, Universidade Federal de Minas Gerais, Avenida Universitária, 1000, 39404-547 Montes Claros-MG, Brazil


Introduction
Sewage sludge is the solid residue produced by the wastewater treatment plants (WWTP). This residue can be used in agriculture as a fertilizer complement or soil conditioner. The sludge has a high content of organic matter and essential elements to plants such as nitrogen and phosphorus. 1,2 However, sewage sludge may contain chemical contaminants which can be transferred to soil, plants, animals, rivers and groundwater. [3][4][5][6] Among the contaminants, polychlorinated biphenyls (PCBs) stand out for being highly persistent in soil, for their low biodegradability and high lipophilicity and the proven harmful effects on biota and carcinogenicity on humans. 7-10 Although these compounds have been banned by the 2001 Stockholm Convention, recently, 209 polychlorinated biphenyl congeners have been identified in sewage effluents. 11 The PCB congeners named 28, 52, 101, 138, 153 and 180 are known as indicators of environmental pollution due to their highly frequent detection in environmental and biological samples. [12][13][14] PCBs are quantified by gas chromatography and mass spectrometric detection (GC-MS) or electron capture detection (ECD). 15 Traditional techniques such as shaker table, 16 soxhlet extraction 17 and ultrasound 18 are the most used ones for PCB studies on sewage sludge. 19 Accelerated solvent extraction, a more sophisticated technique, has also been used. 20 Recently, solid-liquid extraction with low temperature purification (SLE-LTP) was used for quantification of polycyclic aromatic hydrocarbons (PAHs) and chlorobenzenes in sewage sludge. 21,22 The principle of the technique was proposed by McCulley and McKinley in 1964, 23 but only in 2007 the technique was modified for pyrethroid determination in water, which was initially named liquid-liquid extraction with lowtemperature partition (LLE-LTP). 24 Significant advances in this technique have been achieved for the monitoring of chemical contaminants and, in the last eight years, it has been optimized and validated for different matrices such as milk, 25 tomato, 26 human urine, 27 beverages, 28 human liver, 29 pineapple, 30 lettuce 31 and Caiman yacare eggs. 32 The principle of SLE-LTP is based on the addition of a homogenous mixture composed of water and organic solvent to the sample. The system is homogenized and cooled to −20 °C to freeze the matrix components and the aqueous phase. The organic phase, typically acetonitrile, remains liquid and extracts the compounds of interest. The technique enables extraction and cleaning of extracts in one step, but depending on the matrix complexity, the extract may be subjected to solid phase extraction (SPE) prior to chromatographic analysis. 33 This study aimed to optimize and validate SLE-LTP of six polychlorinated biphenyls in sewage sludge samples. Extracts were analyzed by gas chromatography-mass spectrometry (GC-MS). The validated method was applied to analyze sludge samples from the WWTP of Montes Claros City, Brazil, from May to November 2013.
The chromatographic analysis was performed in a 7890A gas chromatograph coupled to 5975C mass spectrometer (Agilent Technologies) using a DB5-MS capillary column (Agilent Technologies), stationary phase 5% phenyl and 95% methylsiloxane, (30 m length × 0.32 mm internal diameter × 0.25 µm film), helium (99.9999% purity) as carrier gas at 1.5 mL min -1 flow rate. The split/splitless injector was held at 280 °C. The temperature program was 70 °C, 12 °C min -1 to 140 °C, 25 °C min -1 to 190 °C, 7 °C min -1 to 280 °C. The injection volume was 1 µL in the split ratio 1:5, using a CombiPAL injector. The mass detector was operated in the electron impact mode (70 eV) and a quadrupole mass analyzer. The interface was kept at 280 °C and the ion source at 230 °C. Device control and data acquisition were carried out using the ChemStation software (E.02.02.1431 copyright © 1989-2011-Agilent Technology).
Analyses were performed in selective ion monitoring mode (SIM) and data acquisition was divided into five groups of ions. Group 1 (0 to 11.50 min) comprised ions 186

Standard solutions
Standard stock solutions from the six congeners were prepared separately in hexane in the concentration of 400 mg L -1 and stored at −20 °C. The working solutions were then prepared from the dilution of the stock solutions in the same solvent, containing all the six congeners in the concentration of 20 mg L -1 .

Sewage sludge samples
Biphenyl-free sludge samples were collected from the drying bed of the wastewater treatment plant of the municipality of Juramento City, Minas Gerais, Brazil. These samples were used during the steps of SLE-LTP optimization and validation. The moisture content was determined by weighing 2.0000 g of sample in five replicates. Samples were placed in an oven at 100 °C and were weighed every hour until constant weight.
The sludge samples were collected in the wastewater treatment plant in the municipality of Montes Claros-MG, from May to November 2013 for application of the method in real samples.

SLE-LTP
The parameters copper mass, extraction solvent composition, form and time of homogenization, change in ionic strength, water pH and elution solvent volume were optimized in the SLE-LTP (Table 1). Extraction percentages in each optimization condition were analyzed by the t-test (p < 0.05).
In the optimized method, 4.0000 g of sludge were added to 22-mL glass flask and after spiked with 100 µL of standard solution (62.5 µg kg -1 ). The system was maintained at rest for 3 h. In the sample was added 0.80 mL of an aqueous solution containing 0.1 g NaCl. To the flask was added 6.50 mL isopropanol, 1.50 mL ethyl acetate and 1.5000 g copper metal. The system was homogenized in ultrasonic bath for 15 min and incubated in a stirring water bath at 40 °C for one hour. The samples were then kept at −20 °C for two hours. The extracts were subjected to an additional clean up step in the manifold with flow rate of 1 mL min -1 . The cartridge was prepared with 2.0 g silica and 1.5 g sodium sulfate. The extract was eluted with isopropanol and collected into a 10.00 mL volumetric flask. Validation Selectivity, linearity, limit of detection (LOD), limit of quantification (LOQ), accuracy and precision were evaluated. 34 Linearity was evaluated by preparing the calibration curve with seven concentration levels. The concentrations 4, 10, 20, 30, 40, 50 and 60 µg L -1 were used with three independent replicates for each level. The least square linear regression was applied to the experimental data for estimating the regression parameters. The outliers were treated and confirmed by the Jacknife test with maximum exclusion of 22.2% of the data. Regression analysis was evaluated by the parameters normality (Ryan and Joiner test), homoscedasticity (Brown and Forsythe test) and independence (Durbin and Watson test). Analysis of variance (ANOVA) was applied to each solvent and matrix calibration curve to test the lack of fit of linearity. The matrix effect was assessed by calculating the percentage of variation of the chromatographic response [% = (Ā matrix − Ā solvent ) / Ā solvent × 100]. The means of areas obtained for each PCB in solvent and matrix were compared by the t-test.
The selectivity of the method was investigated by evaluating the PCB-free samples (blank) in six independent replicates. Other interferences such as solvent derivatives, glassware, adsorbents and reagents were investigated by reapplying the method without using sewage sludge samples.
Limits of detection (LOD) and limits of quantification (LOQ) were determined by fortifying the sludge samples with biphenyls in the lowest acceptable concentrations. LOD and LOQ were considered as three and ten times, respectively, the baseline noise signal obtained for the samples free of biphenyls (blank).
Accuracy was evaluated by fortification/recovery experiments. Three concentration levels were studied 4, 8 and 20 µg L -1 in three replicates. Recovery values for each level were considered acceptable when between 70 and 120%. 34 Precision, under repeatability conditions, was evaluated by the relative standard deviation (RSD) obtained from fortification/recovery experiments of biphenyls in sludge.
The tests were performed in the concentration of 8 µg L -1 in seven replicates. The acceptability criterion for RSD was ≤ 20%. 34

SLE-LTP optimization
Previous studies have shown that the optimum ratio of organic solvent-water mixture in SLE-LTP is 8 to 4 mL. 24,26 In this study, the sludge samples contained 80% water (m/m). Thus, 0.8 mL of water was added to 4 g of sludge, resulting in 4 mL of water.
The first SLE-LTP parameter optimized was the addition of copper metal to the sludge sample. Copper removes sulfur compounds that interfere with the chromatographic analysis. 35,36 Thus, different masses of powdered copper were added to the sludge. Addition of 1.5 g of copper removed interferents in the same retention time of the biphenyls and four intense signals named A, B, C and D ( Figure 1). Therefore, this mass was used in the other optimization steps.
Although acetonitrile is traditionally used in SLE-LTP as extraction solvent, 25 four less polar extractor mixtures were assessed for the recovery of biphenyls ( Table 2). The mixture isopropanol:ethyl acetate in the ratio 6.50:1.50 mL (v/v) had the highest recovery rates for the six biphenyls, with values greater than 50%, as compared by the t-test (p < 0.05).
To improve the percentage recovery of biphenyls, the system homogenization was tested using vortex and ultrasound in two different times ( Table 2). The results showed that homogenization in ultrasound for 15 min had higher percentage recovery for four biphenyls, as compared by the t-test (p > 0.05).
Previous studies have reported the salting out effect during the extraction of contaminants in environmental matrices using SLE-LTP or LLE-LTP. 21,37,38 Aliquots of 0.80 mL water containing 0, 0.1 and 0.2 g NaCl were added to the sludge samples during SLE-LTP. Higher amounts of NaCl promoted the separation of the organic and aqueous phases before the temperature lowering and hence they were not used. The higher recovery percentages were obtained with the addition of 0.1 g NaCl (p < 0.05) ( Table 2). This positive effect is associated with ion solvation by water molecules, facilitating the migration of PCBs into the organic phase.
Although the chemical structure of the biphenyls does not change under pH variation, the sludge matrix components can ionize or degrade, influencing biphenyl extraction. The pH of the water added to the sludge sample was adjusted to 3, 7 and 13. The highest recovery percentages were obtained at pH 7 for the six biphenyls (Table 2), as compared by the t-test (p < 0.05). Similar results were reported using SLE-LTP for extraction of pesticides in honey. 33 After the sample freezing in SLE-LTP, the volume of the organic phase reduced from 8 to about 5 mL, because part of the organic solvent was retained in the solidified phase. These 5 mL extracts were subjected to the additional step of cleaning with silica to remove remaining interferents. The isopropanol volume added as eluent changed the recovery percentages of the biphenyls ( Table 2). The highest recovery percentages were obtained with a final volume of 10 mL extract (p < 0.05). Although the addition of the highest volume of isopropanol had diluted the extract, there was   Figure S1).

Linearity, matrix effect and selectivity
After optimization, the method was validated by estimating the seven main figures of merit. The linearity of detector response was evaluated using seven concentration levels, including the limit of quantification, the maximum residue limit and the concentration used in the method optimization. Concentrations were equally spaced in the range evaluated. Replicates of each calibration point provided information on the inherent variability of the response measurements (pure error). The calibration data were obtained by linear regression and the determination coefficients (R 2 ) were greater than 0.99 (Table 3). The lack of fit was not significant (p > 0.05) for the analytical curves prepared in the solvent and matrix extract, indicating the linearity of response in the range 4-60 µg L -1 (Table 3).
Linearity was assessed by the ordinary least squares method (OLSM). The outliers were confirmed by the Jacknife residue test, with maximum exclusion of 22.2% in 21 replicates (Figure 2). The Ryan-Joiner test confirmed that the regression residuals were normally distributed, with correlation coefficients greater than the critical values and non-significant deviation from normality for the curves (Figure 3). Levene's test confirmed the homoscedasticity of residues, demonstrating the homogeneous distribution of residues. Independence of regression residuals with distribution of points without positive or negative trend was confirmed by the Durbin-Watson test (Figure 4). The results obtained after the tests confirmed the OLSM suitability to PCB 28 (Figures 2, 3 and 4) and similar results were found for the other biphenyls studied (data not shown). This complete system for linearity evaluation followed a procedure based on an acceptable and consistent statistics proposed by Souza and Junqueira. 39 The matrix effect was detected in the GC-MS analysis for the six biphenyls, with significant differences in the chromatographic responses in hexane and in the sludge extract obtained after SLE-LTP (p < 0.05). The matrix effect   Similar results were found for pesticide analysis in tomato using SLE-LTP. 26 The selectivity of the method was demonstrated by the absence of interferents in the retention time of the six biphenyls during the analyses of biphenyl-free sludge extracts (blank) ( Figure 5).

Detection and quantification limits
The LOD achieved for the six biphenyls was 1.3 µg L -1 for the matrix extract and 3.3 µg kg -1 for the sludge sample. The LOQ for the six biphenyls were 4.0 µg kg -1 and 10.0 µg kg -1 for the extract and the sludge sample, respectively. This value was less than the maximum residue limit established by Sewage Sludge Directive of the European Community 40 that establishes limit for biphenyl < 800 µg kg -1 .

Accuracy and precision
The accuracy evaluates the closeness of agreement between the measured values and the true value. Accuracy was assessed using recovery experiments at concentrations of 1 × LQ, 2 × LQ and 5 × LQ, corresponding to 10, 20 and 50 µg kg -1 of biphenyls in sewage sludge. The recovery percentages of the concentrations 10 and 50 µg kg -1 were in compliance with the recommendations of the European Commission 40 which establish for concentrations ≥ 10 µg kg -1 recoveries between 80 and 110%. These results are in accordance with International Union of Pure and Applied Chemistry (IUPAC) recommendations that suggest the range of 70 to 120%. 34 Precision tests evaluate the agreement between the results. Precision was evaluated under repeatability conditions in the concentration of 8 µg L -1 with satisfactory results, since RSD was less than 18% ( Table 4).

Application of the method
Of the seven samples tested, PCB 52 was detected in June, July and August, and quantified in September (60 µg kg -1 ), October (70 µg kg -1 ) and November (50 µg kg -1 ). Similar concentrations of this biphenyl  were found in a study conducted in Turkey in the period 2009-2010. 41 The three samples had concentrations within the acceptable limit for biphenyls in sludge according to the European Community legislation (< 800 µg kg -1 ). 40

Conclusions
Solid-liquid extraction with low temperature purification was optimized and validated for the determination of six PCBs in sewage sludge. The method is simple, easy to perform and has low consumption of solvents and samples. The technique is efficient with recovery rates higher than 80% in the limit of quantification for PCBs. PCB 52 was quantified in 40% of the analyzed sludge samples with values below the maximum acceptable limit established by the European Community legislation. The method is a viable alternative for monitoring PCBs in sewage sludge used for agricultural purposes.

Supplementary Information
Supplementary data are available free of charge at http://jbcs.sbq.org.br as PDF file.