Automated solid phase extraction (SPE) has been combined with HPLC and fluorescence detection for the analysis of polycyclic aromatic hydrocarbons in drinking water samples. This approach offers lower operating costs and improved productivity as sample numbers increase.


With numbers of samples passing through analytical laboratories increasing, understandable demands are placed upon the cost-effectiveness, productivity and reproducibility of any extraction method. As laboratory technology has advanced, sample preparation equipment has become available that enables users to automatically pre-treat and extract samples online. This instrumentation although complex, can be reconfigured and trained to perform a variety of automated tasks and can be coupled to final detection methods, providing users with a fully automated streamlined process requiring little or no human intervention.


Polycyclic aromatic hydrocarbons (PAHs) are a specific group of naturally occurring and anthropogenic chemicals. Selected PAHs possess carcinogenic, mutagenic and teratogenic potential, a PAHs toxicity is very structurally dependent, with isomers (PAHs with the same formula and number of rings) varying from being nontoxic to being extremely toxic. All PAHs have a similar structure based on two or more fused aromatic rings.

Anthropogenic PAHs are formed during the combustion of coal, oil, gas, wood and other organic substances. They are naturally present in crude oils, coal tars and petroleum derived products and are produced during some natural events such as forest fires and volcanoes, but generally they arise from combustion or oil related, man-made sources.

Initial concerns related to PAHs, mainly focused on their carcinogenic potential, but more recently focus has shifted towards their ability to cause interferences with hormonal systems and as a result their potential effect on reproductive systems. It has also been documented that PAHs have likely additive (synergistic) effect in the presence of other environmental contaminants. Published data suggests a particular PAHs potency may be increased several times in the presence of other contaminants, and their relative potency in complex mixtures remains largely unknown.

The monitoring of PAHs in the aquatic environment has increased significantly in its importance over the last few years for these reasons. As a consequence selected PAHs have been listed as priority contaminants in the Water Framework Directive (WFD). For a European Union member state this means that the monitoring of these compounds in the aquatic environment at appropriately low ng/L concentrations is necessary. For compliance with the Drinking Water Inspectorate (DWI) water quality regulations 2000, the methodology must be capable of achieving detection limits below 10 ng/L benzo[a]pyrene and 100 ng/L for the sum of the concentrations of the four following individual compounds, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(ghi)perylene and indeno(1,2,3-cd)pyrene.

Traditional methodology

There are several laboratory techniques already utilised by those organisations concerned with the determination of PAHs in the aquatic environment. For the initial extraction and pre-concentration of PAHs from aqueous samples there are generally two techniques used, liquid-liquid extraction (LLE) and solid phase extraction (SPE). Historically sample preparation has mainly been accomplished using LLE, but in recent years SPE has gained much popularity. Both methods of extraction present advantages and disadvantages.

Liquid-liquid extraction, alternatively known as solvent extraction or partitioning, is a method to separate compounds based on their relative solubilities in different immiscible liquids, in this case water (sample) and an organic solvent such as hexane or dichloromethane. Advantageously, this method extracts PAHs both dissolved in the water and adsorbed upon any suspended particles in the sample. Disadvantages are that, the method requires the use of relatively large volumes of solvents, and numerous pieces of laboratory glassware. The process is both time consuming and is not easy to automate.

Solid-phase extraction (SPE) is a separation process which removes compounds from a mixture based on their physical and chemical properties. In this case, SPE uses the affinity of dissolved PAHs for a solid sorbent through which the water (sample) is passed. PAHs are sorbed onto the solid sorbent and undesired components pass through unretained. The PAHs are then desorbed using an organic solvent. This method benefits from much reduced solvent consumption, and a great deal of selectivity can be achieved by choosing an appropriate sorbent for the compound class to be extracted. The method is amenable to automation and so requires no human intervention once the samples are loaded into the autosampler tray.

Automated Solid Phase Extraction

Automating the SPE process and coupling it to the detection provides a rugged, streamlined, cost effective analytical solution to the extraction and analysis of PAHs in drinking water samples. The automated extraction and pre-concentration method details are discussed briefly below.

A 30 ml aliquot of sample is manually transferred to a 40 ml EPA glass vial. All further sample manipulations are performed automatically by either the MPS 3XL or TE-100. Samples are modified by the addition of isopropanol (IPA) to provide a final concentration of 25 % IPA in each sample. This addition of the IPA enhances extraction efficiency and causes desorption of any analytes from vial walls. Deuterated internal standards are added automatically using the MPS. Using a large volume liquid handling syringe, 10 ml of sample is introduced to a stainless steel sample loop on the MPS 3XL.

Using a dedicated sample enrichment pump incorporating a solvent selection valve, the sample is concentrated using an on-line solid phase extraction cartridge. Following sample enrichment, the cartridge is eluted using one of the analytical pumps with 100 % acetonitrile, thus ensuring efficient desorption from the cartridge. Post cartridge, eluent flow is mixed with water from a second pump, using a high pressure mixing tee to provide the analytical gradient conditions for separation and detection.

During sample analysis the cartridge, loop, injector and sample lines are cleaned and equilibrated in preparation for the next sample. The Maestro software, which is fully integrated, contains comprehensive prep ahead functionality, which maximises sample throughout.

The TE-100 contains the functionality to automatically switch to alternative SPE cartridges, if and when the presently utilised cartridge becomes blocked. This functionality is based on a user configurable pressure threshold and a six position fourteen port switching valve which accommodates up to six extra cartridges, if required a further two valves may be added giving extra cartridge capacity.


Detection of the PAHs is performed using an Agilent 1200 G1321A fluorescence detector. Quantification of individual components, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene and benzo[ghi]perylene was carried out at 432 nm whilst fluoranthene and indeno[1,2,3-cd]pyrene was quantified at 510 nm.

Recovery and precision experiments were performed on all standards within the calibration range, 0.4 – 200 ng/L. The data was derived from spiked tap water samples and quantified against the solvent standards following any necessary background subtraction, thus taking into account any possible interfering matrix effects or contributing analyte signal from the initial sample.

1 – Benzo[b]fluoranthene 2 – Benzo[k]fluoranthene 3 – Benzo[a]pyrene 4 – Benzo[ghi]perylene 5 – Fluoranthene 6 – Indeno[1,2,3-cd]pyrene

Concentration (ng/L) Std1 Std2 Std3 Std4 Std5 Std6 Std7
Fluoranthene 1 2.5 5 12.5 25 50 125
Benzo[b]fluoranthene 0.4 1 2 5 10 20 50
Benzo[k]fluoranthene 0.4 1 2 5 10 20 50
Benzo[a]pyrene 1 2.5 5 12.5 25 50 125
Benzo[ghi]perylene 1.6 4 8 20 40 80 200
Indeno[123-cd]pyrene 1 2.5 5 12.5 25 50 125

Initial calibration curve data were generated between 0.4 and 200 ng/L. The curves exhibited linearity with correlation coefficients (R2) between 0.9997 and 0.9999. Following this determination of linearity data, tap water samples (∑=35) were spiked at each level (n=5) to give the concentrations indicated in the above table, these samples were then enriched and quantified to provide the recovery data illustrated in the table below.

Recoveries (%) Std1 Std2 Std3 Std4 Std5 Std6 Std7