Abstract
Analyte loss during syringe filtration is often overlooked in quantitative analyses. Bisphenol A and similar compounds are being extensively examined for their toxicity. This study serves to characterize common syringe filters and their effects on analytical results. The adsorbance of bisphenol A and similar compounds by filter media made from cellulose acetate, nylon, and polytetrafluoroethylene was examined using UV–Vis spectroscopy. Samples of bis(4-hydroxydiphenyl)methane, bis(4-hydroxyphenyl)sulfone, 2,2-bis(3-methyl-4-hydroxyphenyl)propane, 4,4′-(propane-2,2-diyl)diphenol, 4-cumylphenol, and phenol were studied. Solutions were analyzed before and after filtration. Filter composition significantly affected analyte concentrations in the filtrate solutions.
Introduction
Bisphenol A (BPA) is an organic compound most often used to manufacture polycarbonates. This molecule can leach out of products and be ingested by humans (Mercea, 2009). It mimics natural hormones causing adverse effects (Olea et al., 1996), even at low concentrations (vom Saal and Hughes, 2005). There have been over 90 studies examining the effects of BPA on humans, although the extent of symptoms within practical exposure limits still requires further evidence (Rochester, 2013). This increase in awareness of the effects of BPA is leading to discussions of stricter regulations (e.g., Barraza, 2013; Erler and Novak, 2010). In a response to the regulations, manufacturers are phasing out BPA (Flint et al., 2012; Lionti et al., 2013); it is anticipated that there will be an increase in the use of alternative types of bisphenols to accommodate emerging legislation. Similar to BPA, toxicity has been reported in these alternative bisphenols which contain various substituent groups (Chen et al., 2002).
Generally, sensitive instrumentation must be used to detect low levels of environmental health toxins. Thus, syringe filters are one convenient way to clear the sample of particulate matter before analysis (e.g., Martinelango et al., 2006). However, syringe filters affect the results of various analytical methods (Ahmad et al., 2001; Heimann and Jakobsen, 2007; Liu et al., 2012), and thus, filtration effects should be considered when designing methods for analysis (Hebig et al., 2014). In anticipation of further experimentation involving bisphenols, this study was undertaken to understand the adsorption behavior of BPA and similar compounds in aqueous solution when filtered using a variety of common syringe filters. We hypothesize that syringe filtration can lead to inconsistent data. The effects of filtration on BPA analysis have been mentioned previously (Salgueiro-Gonzalez et al., 2012); however, to our knowledge, this is the first study to compare the effectiveness of filtration across a selection of bisphenols and phenolic compounds.
Experimental
Materials and data analysis
All chemicals were obtained from Sigma-Aldrich (St. Louis, MO). A Shimadzu (Kyoto, Japan) UV-1650PC with UV Probe 2.10 software was used to determine relative concentrations using quartz cuvettes with a 1-cm path length (Precision Cells Inc., Farmingdale, NY). Data were normalized by comparing the absorbance of filtered versus unfiltered samples and reporting the values as percentages. By reporting the relative concentration instead of absolute concentration, the differences in solubility can be ignored.
Methods
Samples of bis(4-hydroxydiphenyl)methane (BPF), bis(4-hydroxyphenyl)sulfone (BPS), 2,2-bis(3-methyl-4-hydroxyphenyl)propane (BPC), 4,4′-(propane-2,2-diyl)diphenol (BPA), 4-cumylphenol, and phenol were weighed to make approximately 100 ppm solutions in deionized water. If necessary, solutions were diluted to get absorptions approximately between 0.5 and 1.5 on the UV–Vis spectrometer, in order to get an adequate analytical response. Dilution factors varied among each compound due to differences in analytical response for each compound. The adsorption differences between filter media were examined by filtering samples four times each at analytical wavelengths between 250 and 300 nm. Syringe filters used were 0.45 µm nylon (Thermo Scientific, Waltham, MA), 0.2 µm polytetrafluoroethylene (PTFE) (Fisherbrand, Pittsburgh, PA), and 0.2 µm cellulose acetate (National Scientific, Rockwood, TN). A new filter was used for each replicate. Samples were approximately 1.5 mL.
The effect of pore size was studied by analyzing four replicates of a 100-ppm BPA solution filtered through 0.2 µm nylon (Fisherbrand), 0.45 µm nylon (Thermo Scientific), 0.2 µm PTFE (Fisherbrand), 0.45 µm PTFE (Fisherbrand), 0.2 µm cellulose acetate (National Scientific), and 0.45 µm cellulose acetate (Thermo Scientific) filters. The same stock solution and filter types were used to observe the adsorption saturation across a 15-mL injection range (samples taken at every 1.5 mL, for a total of 15 mL). A single filter was used to filter 15 mL of sample; each data point consisted of an average of three replicates taken at every 1.5 mL. A new filter was used for each replicate at analytical wavelength 277 nm.
Results and discussion
Generally, PTFE filters adsorbed the least and the nylon filters adsorbed the greatest amount of analyte. However, this trend was not the case for phenol, which showed greater adsorption to the cellulose acetate filter media than the other filters tested (Figure 1). This suggests that it is the central portion and substituent groups of the bisphenol molecule that caused the adsorption trends. Additionally, the number of polar or nonpolar substituent groups appeared to affect the adsorption characteristics accordingly. For example, BPC has two more methyl groups than BPA (Table 1), thus indicating BPC was more hydrophobic overall which caused BPC to adsorb to the hydrophobic PTFE filter more than BPA. The compound 4-cumyl phenol has one less hydroxyl group than BPA (Table 1), indicating it is slightly more hydrophobic than BPA overall, thus adsorbing to the hydrophobic PTFE filter more than BPA. When compared to BPA, BPF has no central methyl groups, thus it is less hydrophobic than BPA, which caused it to adsorb less to the PTFE filter than BPA. The correlation of substituent groups affecting polarity and adsorption seems to be present for the nylon and cellulose acetate filters as well when examining BPC, 4-cumylphenol, and BPF to BPA. The trend was not as apparent for BPS. BPS is insoluble in water according to a literature search (Table 1); however, BPS was slightly soluble in water in this experiment. According to Figure 1, BPS adsorbed to the hydrophobic PTFE filter less than BPA, which would indicate that it is more hydrophilic than BPA, if the aforementioned trend above, holds exclusively true. However, when examining the nylon and cellulose acetate filters, BPS would be more hydrophobic than BPA. Although the trend is noteworthy, statistical significance cannot be applied among all compounds for the substituent-adsorbance trend. Phenol had adsorption characteristics unlike any of the bisphenols (Figure 1), and it is the most soluble in water (Table 1). This indicates that the adsorption of analyte by syringe filters cannot be an assumption that the most polar analyte adsorbs to the most polar filter media and vice versa. The solvent, number of adsorption sites, and solvent/analyte interaction are some sources of variability.
The percentage of compound detected in the filtrate after filtration compared to unfiltered standard solutions. Error bars are ± the standard deviation. Representative chemical structures and solubility notes of the compounds involved in the study. Each bisphenol compound showed at least some solubility in water at <100 mg/L at 25℃ because an adequate response was obtained on the UV–Vis spectrophotometer in order to obtain data for Figure 1.
The data suggest that the PTFE filters had low adsorption affinity for bisphenols, thus, using PTFE filters should minimize analyte loss for bisphenol analyses requiring syringe filtration in aqueous solutions. Figure 2 shows that the type of filter media has a greater effect on adsorbance than pore size. However, pore size did appear to have some effect on the saturation characteristics of the syringe filters, likely from differences among surface area (Figure 3). Additionally, filter saturation increased within the first few milliliters of injection which is likely another source of variability among bisphenol analyses. It is recommended to discard the first few milliliters of filtrate to saturate the adsorption sites before analyzing the sample (Hebig et al., 2014). However, the amount of solution it takes to saturate the filter can vary according to Figure 3. This suggests that when performing quantitative analyses, the volume filtered should be consistent among all samples and standards to minimize error from adsorption variability.
A comparison of the effect of filter pore size and filter media on the percentage of BPA detected in the filtrate after filtration compared to unfiltered standard solutions. The pore size is given in µm. Error bars are ± the standard deviation. The percentage of BPA detected in the filtrate after filtration compared to unfiltered standard solutions, over approximately a 15-mL range. Pore size is in µm. Each point is the mean of 3 replicates.
A standard curve was prepared, and BPA exhibited a linear response (Figure 4). The detection limit of BPA for the UV–Vis spectrophotometer (UV-1650PC, Shimadzu) was under 2 ppm, using the Limit of Detection (LOD) defined by the ACS Committee on Environmental Improvement (1980). Although only relatively high levels of BPA (∼2 ppm and above) can be detected using UV–Vis spectroscopy, we chose this method due to its environmental efficiency. It allowed a number of samples to be analyzed without using hazardous organic solvents required for other instrumentation, such as High Performance Liquid Chromatography (HPLC). Additionally, adsorption characteristics likely vary in other solvents. Further studies should examine the effects of various solvents and manufacturers. The concept of this study can be broadly applied to analytical methods for testing bisphenols. It is critical to ensure accurate results when analyzing samples, thus it is imperative that scientists be aware that analyte concentrations can be significantly affected during extraction processes.
Standard curve of BPA solutions between 2 and 16 ppm. Mean response signal of a water blank was 0.0102 ± 0.0004. Absorbance value for LOD was calculated to be 0.0114, using 
Footnotes
Acknowledgements
The authors thank Dr Michael Goldcamp for the advice, edits, and revisions.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors thank Wilmington College for supporting this work.