Determination of Bisphenol A in Drinking Water

Introduction

The level of bisphenol A (BPA) detected in foods and beverages has gained media attention over the last several years. Specifically, this interest is related to two types of containers:

• Re-useable rigid containers made of polycarbonate plastic; commonly used for water bottles, baby bottles, plastic mugs, carboys, and storage containers
• Metal cans with an internal epoxy-based lacquer coating; used to keep the foods or beverages from directly contacting the metal

BPA is one of the chemicals used to make both polycarbonate plastics and epoxy-based lacquers. Research has shown that small amounts of BPA can migrate into the foods and beverages enclosed in these types of containers. Furthermore, the transfer of BPA from the container to the contents increases if the containers are exposed to elevated temperatures, such as when baby bottles are heated and when metal cans are filled while the food or beverage is still hot^1-3.

Experimental

The focus of the work presented here was to demonstrate the extraction and analysis of BPA from drinking water. A sample was spiked with BPA to a level of 200 ng/mL prior to extraction. Sample processing using solid phase extraction (SPE) was selected to demonstrate the ability of this technique to perform both extraction and concentration tasks. Supelclean™ ENVI™- 18 SPE material inµglass tubes with PTFE frits was used. This hardware eliminates the possibility of introducing compounds (such as phthalates) which may be solvent-leached from alternative hardware (for example, polypropylene tubes with polyethylene frits).

The structure of BPA is shown in Figure 1. GC may be a more sensitive technique for this analyte, but it requires that BPA undergo derivatization prior to analysis. In addition to increasing the sample processing procedure by several steps, artifacts may be introduced during the derivatization steps. Therefore, HPLC was selected as the analytical technique for this work to minimize interferences. An Ascentis^® Express C18 column was utilized to obtain a fast HPLC analysis. The system used for this work was equipped with two detectors in series, ultraviolet (UV) and fluorescence (FL). The system was calibrated with several standards and a response factor for BPA was generated for each detector. This allowed recovery data of the spiked sample to be calculated. 
Figure 2 shows chromatograms of the 1 µg/mL calibration standard. Chromatograms of the spiked sample are shown in Figure 3. This figure also includes a complete description of the sample preparation steps.

HPLC Conditions

column: Ascentis Express C18, 10 cm x 2.1 mm I.D., 2.7 µm (Product No. 53823-U); mobile phase: water:acetonitrile (60:40); flow rate: 0.4 mL/min.; pressure: 3268 psi (225 bar); column temp.: 35 °C; detector: UV (230 nm); FL (Ex 225 nm, Em 310 nm); injection: 1 µL; sample: bisphenol A at 1 µg/mL in acetonitrile

HPLC Conditions

sample/matrix: Drinking water spiked with bisphenol A to a 0.2 µg/mL level;
SPE tube: Supelclean ENVI-18, 500 mg, 6 mL glass tube, PTFE frit (Product No. 54331-U);
condition: 1 mL 1% formic acid in acetonitrile, 1 mL DI water; sample addition: 5 mL spiked water; sample elution: 2 mL 1% formic acid in acetonitrile eluate; post- treatment: 1 mL evaporated, then reconstituted to 0.5 mL with acetonitrile
Conditions (except for sample) and peak IDs are the same as Figure 2.

Results and Discussion

A better signal-to-noise ratio was obtained with the FL detector. Also of note is the slightly longer retention time and broader peak shape observed on the FL chromatograms. These are caused by the extra system volume contributed as the sample passes through the UV cell as well as the tubing connecting the detectors. The removal of the UV component and shortening the tubing connecting the column to the FL detector would eliminate these phenomena.

As shown in Figure 3, a fast analysis was obtained in which the analyte is free of interference. The procedure results in a calculated 1 µg/mL BPA level in the final spiked sample extract. This works out to a 5-fold concentration (0.2 µg/mL in the spiked sample was concentrated to 1 µg/mL in the extract). Using the calibration factor generated for the FL detector, a recovery of 88% was calculated.

Conclusion

A comprehensive sample preparation and analytical procedure was developed for determining BPA in drinking water. This fast procedure employed materials and techniques selected in part for speed, but also those that would not contribute unwanted artifacts. The use of SPE allowed BPA to be extracted plus concentrated, which may result in greater method sensitivity compared to simple headspace or direct injection methods.