Introduction

Optimization of the extraction of the compounds of interest from, for example, the biological matrix is an important step in the development of an analytical method, because it will affect the overall sensitivity and selectivity of the method. Sample preparation will not only lead to highly concentrated extracts, but can remove potentially interfering matrix compounds, resulting in enhanced selectivity and a more reproducible method independent of variations in the sample matrix. This is of particular importance in LC–MS analysis, in which suppression or enhancement of the mass spectrometric signal can occur when matrix components coelute with an analyte. Despite all these advantages, sample preparation is still regarded as time-consuming, laborious work, and there is much interest in simplifying this step.

Conventionally, liquid–liquid and solid-phase extraction (LLE and SPE) are chosen for non-volatile compounds whereas headspace extraction is the state of the art for volatile compounds. Increasing attention has recently been devoted to reducing sample volume, analysis time, and cost, and to eliminating the use of (chlorinated) solvents; automation is also encouraged to reduce the workload. Methods such as solid-phase micro-extraction (SPME) and liquid-phase micro-extraction (LPME), and use of restricted access materials (RAM) are, therefore, also increasing in popularity in analytical toxicology. This paper reviews sample-preparation techniques used in forensic and clinical, but not environmental, toxicology published since 2000, and discusses the trends in this important analytical step. SPME will not be covered because a critical review of this technique is presented elsewhere in this issue. The basic principles of the reviewed extraction techniques will not be discussed here and we refer the reader to comprehensive textbooks on solid-phase extraction and supercritical-fluid extraction techniques [1–3]

Extraction procedures for non-volatile compounds

Liquid–liquid extraction (LLE)

LLE using organic solvents

Liquid–liquid extraction is still frequently used in analytical toxicology, especially for (urgent) screening purposes, when analysis of a wide range of (unknown) compounds rather than target analysis is the objective [4–12]. Development of an LLE procedure is, moreover, not time-consuming, although it is difficult to automate, requires high-purity solvents, and can result in the formation of emulsions with incomplete phase separation, the last of which leads to impure extracts. Safe disposal of toxic solvents may also be problematic and expensive [13]. Liquid-phase microextraction (also referred to as solvent microextraction, liquid–liquid microextraction and single-drop microextraction) was therefore introduced in 1996. It is based on suspension of a single droplet of organic solvent, from the end of a microsyringe needle, in an aqueous solution. The droplet, containing analytes extracted on the basis of passive diffusion, is injected into the GC or HPLC. He and Kang recently developed a method for three-phase liquid–liquid–liquid microextraction of methamphetamine and amphetamine from urine in which organic solvent containing the acceptor droplet was applied on top of the sample [14]. These methods are not always robust, however, because the droplets may be lost during extraction. To overcome this, membrane extraction techniques have been developed. Microporous-membrane liquid–liquid extraction entails use of a two-phase system in which one aqueous phase and one organic phase are separated by a microporous hydrophobic membrane. The organic solvent fills the pores of the membrane and enables direct contact of the two phases without their mixing and, therefore, without emulsion formation. The supported liquid membrane extraction (SLM) consists of a three-phase system in which analytes in one aqueous phase can be extracted into another aqueous phase through an organic phase held between the aqueous phases by a porous, hydrophobic membrane [15–18]. Pedersen-Bjergaard and Rasmussen developed a SLM device in which the organic solvent is immobilised within the pores of a disposable polypropylene hollow fibre; the analytes are extracted from biological samples and trapped in a protected micro-extract (5–50 μL). As a result, consumption of organic solvents is low and clean and highly concentrated extracts are obtained [19]. Pedersen-Bjergaard and Rasmussen have reviewed the use of LPME for extraction of a wide range of drugs from plasma, whole blood, urine, and human milk [19].

Miyaguchi et al. reported another miniaturised procedure, based on a microchip liquid–liquid extraction, for analysis of amphetamine-type stimulants in urine [20]. This miniaturisation is based on pressure-driven manipulation of liquids in fluoroalkylated microchannels in a glass microchip. The organic solvent (1-chlorobutane) and the fortified urine sample were pumped through the microchip, resulting in an extraction of the compounds into the organic solvent, which was collected at the end of the microchannels and analysed by GC–FID. Another development in LLE is the extent of automation used to reduce laboratory workloads. Several papers describe the application of robotics in combination with 96-well format micro-tubes to (semi-)automatic LLE of drugs such as paclitaxel, ondansetron, limepiride, cyclosporin, and dextromethorphan in biological samples [21–26].

Supercritical-fluid extraction (SFE)

A compound becomes a supercritical fluid when maintained at a temperature and pressure above its critical point. Under such conditions the compound is neither a gas nor a liquid, and the state is best described as intermediate between the two extremes. Supercritical fluids have the solvent properties of liquids yet can be transported like gases. SFE is, thus, a method of extraction that eliminates the use of (hazardous) organic solvents while still having good solvating characteristics. The solvating power of the supercritical fluid can be adjusted by changing the pressure or temperature, or by adding a solvent modifier, for example methanol. Two modes of extraction are used—dynamic and static. In the dynamic mode fresh supercritical fluid is supplied continuously; in the static mode the sample is extracted with a fixed amount of supercritical fluid contained in a closed extraction vessel. The analytes extracted are isolated by trapping on an adsorbent or in a solvent after depressurisation, or by thermal trapping (depressurized in a cooled container) [27]. SFE using CO2 modified with methanol is widely used in analytical toxicology for analysis of hair; several SFE methods have been used for analysis of drugs of abuse, for example cocaine, codeine, and morphine [28] or amphetamines [29], in hair. Other forensic applications include analysis of compounds on the surface of the hair, because these are indicative of subject characteristics such as age, race, gender, or use of hair products [30].

Solid-phase extraction (SPE)

SPE is used to extract and concentrate analytes from a liquid matrix by partitioning the compounds between a solid phase and a liquid phase. The objective of SPE is to remove interfering compounds and to concentrate the analytes with good recovery and reproducible results. It should also facilitate rapid and efficient simultaneous processing of many samples [31]. SPE has disadvantages which include poor reproducibility, because of differences between batches of adsorbents, the difficulty of standardizing the use of a vacuum, and the variable nature of drying steps. With the development of new (polymeric) extraction supports, inter-batch reproducibility has been improved, however. SPE material is, moreover, expensive and optimisation of an SPE procedure is not always straightforward. An SPE procedure consists of four consecutive steps—column conditioning, sample loading, column washing, and elution. When such a procedure is being developed, a suitable adsorbent material and suitable washing and eluting solvents must be selected, in accordance with the characteristics of the analytes and the matrix, and the purpose of the analysis (screening or target analysis). The final extract should also be compatible with the final analytical procedure (HPLC or GC). Several types of SPE material are available. These are usually packed into syringe barrels which used in combination with SPE vacuum manifolds. The volume of the syringe barrel selected depends on sample volume, and the amount of adsorbent (between two frits) determines the sample capacity, which is approximately 5% of the mass of the adsorbent [13]. SPE disks or membranes were developed to avoid the drawbacks of classical barrels, for example reduced retention of analytes because of channelling through the adsorbent bed, which disturbs solvent flow, slow sample processing because of the small cross-sectional area, which can be blocked by matrix compounds, and variation in adsorbent density. They also require smaller volumes of conditioning and eluting solvents. This trend of minimizing the volume of the adsorbent bed, resulting in a small elution volume and thus highly concentrated extracts, is also apparent from the micro-scale formats of the SPE pipette tip and the solid-phase micro-extraction microfibre [13, 31, 32].

SPE adsorbents and their modes of interaction

Different types of interaction can occur between the analytes of interest and the active sites on the adsorbent [13]. These include both hydrophobic interactions, for example Van Der Waals forces, and hydrophilic interactions, for example dipole–dipole, induced dipole–dipole, hydrogen bonding, and π–π interaction. Other mechanisms include electrostatic attraction between charged groups on the compound and a charged group on the adsorbent surface, and molecular recognition mechanisms [33]. Reversed-phase, normal-phase, ion-exchange, and immunoaffinity adsorbents, and molecularly imprinted polymers (MIPs), exploit one of these interactions only, whereas mixed-mode adsorbents combine several interaction mechanisms. Restricted-access-matrix adsorbents (RAM) combine hydrophobic, ionic, or affinity interactions, and large matrix components (e.g. proteins) are excluded by appropriate selection of pore size or by use of chemical repulsion (by applying an appropriate hydrophilic coating to the adsorbent surface). The mode of interaction selected depends on the demands of the method, for example screening or target analysis, the sensitivity required, and the final composition (for example organic extract for GC or aqueous extract or HPLC).

Not only are the chemical characteristics of the functional groups on the adsorbent relevant, but also the characteristics of the backbones to which these functional groups are attached. The backbones are either silica or polymer-based. Silica-based columns tend to contain a limited number of underivatised or “free” silanols, resulting in polar, acidic patches on the adsorbent surface. This results in secondary hydrophilic or ionic interactions when reversed-phase and ion-exchange columns are used. These secondary interactions could be of interest, but are not reproducible, because the extent of end-capping, and thus the number of free silanols, can change from batch to batch. For normal-phase adsorbents, secondary hydrophobic interactions can occur, because of the small alkyl chains that support the functional groups. Silica-based adsorbents with a wide range of functional groups are available, are relatively inexpensive, and are stable within a pH range of approximately 2 to 7.5. Polymeric adsorbents (e.g. styrene–divinylbenzene) are more hydrophobic, more retentive, stable within the pH range 0 to 14, and no secondary interactions are observed. Newer polymers, for example Oasis HLB, combine hydrophilic and hydrophobic interactions. They do not always require conditioning and can be used to extract analytes over a large range of polarity. Because of their hydrophilic character, however, they can retain water, necessitating a longer drying step, especially if gas chromatography (with derivatisation) is used as the final analytical method [31, 34].

Reversed-phase adsorbents

Reversed-phase adsorbents are mainly used for extraction of apolar compounds from polar matrices, for example plasma, by use of hydrophobic interaction mechanisms. The compounds are eluted with a less polar organic solvent that disrupts the Van Der Waals forces. Apolar adsorbents, such as C18, are widely used in clinical and forensic toxicology because they are broad-range adsorbents which can be used to extract a wide range of compounds from a variety of biological samples (e.g. plasma, urine, brain tissue) [35–39]. Qi and Danielson recently developed mini reversed-phase SPE tubes and extracted dicyclomine and cyclopentolate from serum. The mini-tubes (0.5 and 1 cm long with an internal diameter of 1.55 mm) were used to furnish mini-extracts for their nano-ESI MS system [40].

Adsorption stationary phases and normal-phase adsorbents

Adsorption stationary phases are unmodified adsorbents such as pure silica, magnesium silicate, diatomaceous earth [41–44] or alumina [45–47]. Because silica and alumina adsorb water they should be kept dry before use and aqueous matrices should not be used. Water will deactivate hydrogen-bonding sites, resulting in reduced retention of the analytes and variable recovery. Diatomaceous earth is a porous material that can act as a support for the aqueous phase; this mechanism of extraction, called supported-liquid extraction, is related to conventional LLE [48, 49], but problems such as emulsion formation are avoided. Because the extraction is based on partition and not adsorption, there is still discussion about whether use of diatomaceous earth should be classified as LLE or SPE. Extraction with diatomaceous earth is used in systematic toxicological analysis (screening) [41]. Ninety-six-well diatomaceous earth plates have been used for extraction of indolocarbazole and carboxylic acid-based protease inhibitors from biological samples [42–44].

Normal-phase adsorbents are created when polar functional groups, for example cyano, primary amine, or diol functionality, are bonded to the silica surface. These adsorbents can be used to extract polar compounds from an apolar matrix, as a result of hydrophilic interactions. Elution with a polar organic solvent is necessary to disrupt the hydrophilic interactions. In clinical applications these adsorbents are mostly used to purify apolar extracts (e.g. in hexane) of solid matrices such as body tissues [13, 33].

Ion-exchange adsorbents

Extraction with ion-exchange adsorbents exploits ionic interactions between the analytes of interest and the functional groups on the adsorbent. By use of this mechanism, negatively (anion exchange) and positively (cation exchange) charged compounds can be extracted from a biological matrix. When this mode of extraction is used, pH control during the loading, washing, and elution steps is important. During loading on an anion (cation)-exchange adsorbent the pH should be two units higher (lower) than the pK a of the compound and two units lower (higher) than the pK a of the adsorbent. Under these conditions more than 99% of the functional groups are charged. During elution the pH is chosen so that the functional groups on the compound and/or adsorbent are not charged, resulting in disruption of the ionic interactions. Anion-exchange adsorbents contain quaternary amino groups of pK a>14 or aliphatic aminopropyl groups of pK a 9.8 bonded to a silica surface. For cation exchange the choice is between a weak (carboxylic acid, pK a 4.8) or strong (sulfonic acid, pK a<1) exchangers. The ionic strength of the buffer and the flow-rate should also be controlled. Ion-exchange adsorbents are used to isolate groups of either basic or acidic drugs from biological matrices. Several SPE procedures using ion-exchange have been developed [50–65]. Mixed-mode adsorbents combining cation or anion exchange with a hydrophobic or polymeric adsorbent are of interest in drug screening in forensic toxicological analysis [51]. Because the substances present are not known in advance, the extraction procedure must isolate a range of compounds and remove interfering substances from the matrix. These mixed-mode adsorbents combine several interaction mechanisms and, by use of different elution conditions, extracts containing neutral, basic and acidic drugs can be obtained separately [12, 41, 66–71].

Molecular recognition mechanisms

Use of immunoaffinity adsorbents and molecularly imprinted polymers (MIPs) exploits molecular-recognition mechanisms for selective extraction of trace amounts of the analytes of interest from matrices containing large amounts of interfering compounds. Because immunoaffinity adsorbents and MIPs are “tailor-made” with high selectivity for a target molecule or structural analogues, these adsorbents are not useful for toxicological screening; they are, instead, used for highly sensitive and specific target analysis. Several procedures in which molecular recognition mechanisms are used are listed in Table 1 [72–85].

Table 1 Molecular recognition mechanisms: immunoaffinity and molecularly imprinted polymer adsorbents
Full size table

Immunoaffinity adsorbents consist of biological antibodies covalently linked to silica, controlled-size glass particles, or agarose or other soft gels, and result in high affinity and selective antigen–antibody interactions. The preparation of these adsorbents and optimisation of immunoextraction techniques have been reviewed elsewhere [86–88]. The antibodies can be compound-specific but usually have some cross-reactivity with structural analogues [87]. They can also be denatured by contact with an organic solvent, or by fungi if stored incorrectly. The adsorbent should therefore be preserved in buffer solution containing sodium azide and stored in a cooled environment (4°C). These adsorbents should also be used under mild SPE conditions with regard to pH and organic content, are expensive, and only a limited number is commercially available [13, 34].

As a result of these disadvantages MIPs were produced by using template molecules to create cavities in a polymer which will “recognise” the target molecule. Specific cavities are created with the same dimensions as the analyte of interest and with the same mechanisms of interaction. The imprint is obtained by polymerisation of functional and cross-linking monomers in the presence of the template molecule [89]. MIPs are stable in organic solvents, and at higher temperatures and over a wider pH range than immunoaffinity adsorbents. They are prepared rapidly and easily [90]. It may, however, be difficult to remove the target molecules completely and residual template material may leak into sample extracts, leading to incorrect quantification, especially at trace levels [89]. This problem can be overcome by using a structural analogue as template. The polymer backbone of the adsorbents can also lead to non-specific hydrophobic binding of matrix compounds, especially when aqueous matrices are loaded. The analytes are therefore retained by a mixed-mode mechanism involving both selective affinity binding with imprints and non-specific physicochemical adsorption on the polymer surface [82]. In such circumstances the washing step will have a critical effect on the selectivity of the extraction. Caro et al. have reviewed the application of MIPs to the extraction of several drugs in a variety of biological samples, for example plasma, urine, and serum [91].

Restricted-access matrix adsorbents

Restricted-access matrix (RAM) adsorbents are used in clinical toxicology to exclude large molecules such as proteins and to extract low-molecular-mass analytes by use of hydrophobic, ionic, or affinity interactions. Access to the retentive part of the adsorbent by high-molecular-weight compounds is restricted by either a physical or a chemical diffusion barrier [89]. The pore diameter is the physical barrier and the chemical barrier is a protein or polymer network [92]. Because RAM adsorbents do not retain or denature proteins they enable direct injection of plasma or serum samples on to a chromatographic column (in HPLC) without rapid deterioration of chromatographic performance or clogging of the column. Use of RAM adsorbents in pre-analytical and/or analytical columns thus facilitates automation [92, 93].

If compounds are highly protein bound, the active sites of the compounds that would normally interact with the adsorbent are not available for this interaction. As a result, the drug is carried through the adsorbent bed by the protein instead of being retained on the column. The sample should therefore be passed slowly through the column; this results in a switching equilibrium between free and protein-bound compounds. Addition of salt or modification of the pH can also affect the extent of protein binding. Under these conditions, however, exact measurement of the initial amounts of physiologically bound and unbound material is no longer possible [13, 34].

Souverain et al. have reviewed commercially available RAM adsorbents and the methods in which they were applied before 2004 [92], and Cassiano et al. reviewed the history of the development of RAM adsorbents and their applications [94]. Methods used after 2004 are listed in Table 2. The alkyldiol silica (ADS) RAM is the RAM with a physical barrier most often applied. The macromolecules that are excluded are not adsorbed on the phase because of the hydrophilic groups on its surface. Materials with different internal surfaces are available; the interactions range from hydrophobic (C4, C8, C18; RP-ADS)[84, 93, 95–98] to weak and strong cation-exchange (XDS) [99–101]. Rbeida et al. used RAM adsorbents containing hydrophilic diol and diethylaminoethyl groups, with weak anion-exchange capacity, to extract acidic compounds such as naproxen, ibuprofen, and diclofenac from plasma [100]. Another cation-exchange restricted access material (diol groups combined with sulfonic acid groups) was developed by the same researchers for analysis of basic drugs such as atropine [101]. The most important RAMs using a chemical barrier are the semi-permeable-surface (SPS) adsorbents which use a polymer as the barrier, protein-coated silica, for example bovine serum albumin (BSA)-coated adsorbents [102], and the shielded-hydrophobic-phase (SHP) adsorbent, a silica-based material coated with a hydrophobic network of poly(ethylene oxide) with embedded hydrophobic phenyl groups, commercialised as Hisep.

Table 2 Restricted-access material (RAM)
Full size table

Extraction procedures for volatile compounds

The headspace technique is the best method for extraction for volatile compounds because it minimizes use of organic solvents, eliminates non-volatile interferences from the sample matrix, and is not labour-intensive. The technique can be used in two modes—static and dynamic [103]. In the static mode, partition and eventual equilibrium of the analyte between the sample and the gas phase occurs in a closed system. After equilibrium is achieved the vial is pressurised and the headspace is sampled and injected into the gas chromatograph. Static headspace extraction is still popular for volatile compounds such as fluorinated inhalation anaesthetics, γ-hydroxybutyric acid, ethanol, methanol, acetone, and isopropanol in analytical toxicology [104–108]. This “traditional” form of headspace analysis has been modified to enhance the sensitivity, by using concentration effects (HS-SPME, purge and trap, SPDE, HS-SME). Dynamic headspace extraction (“purge and trap”) is accomplished by passing carrier gas continuously above or through the sample and trapping evaporated volatile compounds in a cryogenic and/or adsorbent trap. The trap is then heated to release the analytes, which are transported to the chromatographic column. Examples of this procedure have been described for trichloroethane, trichloroacetic acid, benzene, toluene, ethylbenzene, xylene, and styrene [108–111]. In solid-phase dynamic extraction (SPDE) the headspace is passed through a needle coated with polydimethylsiloxane, used as an extraction and preconcentration medium. Musshoff et al. and Lachenmeier et al. published methods using SPDE to extract drugs of abuse from hair samples [112, 113]. Headspace solvent micro extraction (HS-SME), in which a single drop of a high-boiling solvent is suspended from an injection needle in the headspace above the sample in a closed vial, has recently been developed [114].

Another means of enhancing sensitivity is to increase the injection volume by combining a headspace technique with a cooled inlet or with a cooled GC column, also called cryogenic focusing or oven trapping, respectively [108, 110, 115–117]. Of all the HS techniques, however, headspace solid-phase micro extraction (HS-SPME) is the method attracting most interest in forensic and clinical toxicology [108, 118–122]; this technique is discussed elsewhere in this issue.

Automation

Increasing attention is being devoted to the possibility of automating analytical methods. Automation results in greater sample throughput, improved precision and accuracy, and a minimum of operator intervention, leading to safer sample handling and time-saving procedures. Carry-over and less control over the system can result in systematic errors, however, and analyte stability must be investigated when samples are analysed sequentially [123].

Several papers describe the application of robotics in combination with 96-well format micro-tubes to (semi)automatic LLE of drugs in biological samples [21–26, 124, 125]. For SPE analysis robotics are incorporated to reduce the sample workload [126], and several new SPE formats and adsorbents have been developed for greater compatibility with automation or for greater sample throughput. Solid-phase-extraction adsorbents that do not need a conditioning step and new SPE formats such as the 96 or 384-well plates [51, 127–138] have been developed for rapid sample preparation, resulting in high-throughput analysis [13]. Fully automated liquid-chromatographic methods using column-switching, are also possible, especially with developments such as RAM (see section “Restricted-access matrix adsorbents”) and large particles [92, 93, 139] on which biological samples can be injected without pre-treatment. Gas chromatography or on-line supercritical-fluid extraction [140] in combination with headspace extraction and SPME are more robust than when used in combination with, for example, SPE.

Relevant advantages of the on-line technique include reduced risk of sample contamination, suitability for full automation, and analysis of the complete sample (the whole extract is analysed), leading to improved sensitivity. When automating a SPE method one must keep in mind that a manual procedure cannot be transferred without minor modifications. Pressures, flow-rates, and solvent compositions (stability!) must be optimized again.

On-line systems for liquid chromatography

Although Mauri-Aucejo et al. have described on-line liquid–liquid extraction of amphetamine from urine [141], on-line solid-phase extraction seems to be gaining popularity. The most common on-line SPE system for liquid chromatography (LC) involves use of a small precolumn located in a six-port, high-pressure switching valve. This method is referred to as “column-switching”. When the sample is injected, it is preconcentrated on a precolumn that is pressure resistant and of small dimensions, to avoid band broadening. The interferences are then washed from the precolumn and flushed to waste. The valve then switches to elute the analytes on to the analytical column. To avoid band-broadening, a back flush desorption mode can be used, although this may result in blockage of the analytical column when biological samples are analysed. The column-switching method has limitations, for example the need to use small pre-columns containing a limited amount of adsorbent, resulting in a small breakthrough volume [98, 142–149]. An alternative is use of a six-port valve as an interface for switching arrangements. In this system the precolumn is not placed in the valve and preconcentration can be achieved in low or high-pressure mode. The advantage of this method is that larger pre-columns can be used [90, 150]. Unlike in off-line SPE, in which different types of adsorbent are preferred during extraction and separation, the pre-column adsorbent and the analytical column should be complementary in on-line mode, because they must endure the same mobile phases. The choice of adsorbents is, therefore, restricted.

On-line SPE-LC is very well suited for analysis of biological fluids such as urine and plasma, especially in combination with RAM [98, 143, 144, 151] and large-particle-size (LPS) [142, 147–149] adsorbents. These adsorbents can be used in a pre-column or can function as pre-column and analytical column. RAM supports can exclude macromolecules such as proteins by use of physical and chemical barriers whereas LPS adsorbents consist of large particles (30 or 50 μm) of silica or polymeric material, resulting in interaction with the analytes and passage of protein at high mobile-phase flow rates without generating a high back pressure in the system.

On-line systems for gas chromatography

Whereas on-line coupling of SPE with LC is robust, on-line coupling of SPE with GC is more complicated, because of the incompatibility of the aqueous solvents used for SPE with the stationary phases used for GC. The need for derivatisation and small injection volumes also makes on-line SPE–GC more complex and can limit the development of on-line GC procedures in clinical and forensic toxicology. Several systems have been developed, these use either six-port valves in combination with a drying gas or a solvent-vapour exit. In large-volume transfer the solvent evaporation technique is critical to obtaining efficient chromatography. On-column and loop-type interfaces are usually used with a retention gap and fully concurrent solvent evaporation techniques, respectively (Fig. 1). The on-line configuration is of interest for biological samples, because of the high sensitivity and small amount of sample required. Compared with applications in environmental analysis, however, use in the biomedical sciences is much less widespread, mainly because of the need for derivatisation [152]. On-line derivatisation in the GC column inlet has, however, already been reported [152]. Hyötyläinen and Riekkola have reviewed on-line SPE–GC methods for plasma, urine, and tissue samples. These methods have better detection limits and repeatability, and shorter analysis time and lower solvent consumption, than off-line procedures [152].

Fig. 1
figure 1

System for automated on-line SPE–GC–MS. 1, solvent channels; 2, purge leak restriction; 3, waste; 4, single-piston LC pump; 5, SPE precolumn; V1–V3, Prospect valves; SDU, solvent delivery unit; SVE, solvent vapour exit; OCI, on-column injector. (Reprinted from: Louter AJH et al. (1996) J Chromatogr A 725:67–83; Copyright (1996), with permission from Elsevier.) Samples are concentrated on the SPE precolumn between valves V1 and V2. The elution solvent is delivered by means of the syringe pump, and the elution phase is injected on-column, on to the retention gap connected with a retaining precolumn. This column is connected to the analytical column and a solvent-vapour exit, which is opened just before introduction of the eluent into the retention gap and closed just before full evaporation of the eluent, leaving the volatile analytes in a small zone before entering the analytical column

Full size image

For practical reasons, on-line SPE combined with GC is not always robust. Other techniques, for example SFE, headspace extraction, SPME [153], and microextraction in a packed syringe (MEPS) [154] thus seem more suited to on-line GC methods.

Conclusion

Trends in sample preparation in analytical toxicology are to reduce the use of solvents, to simplify manipulations, and to reduce the time necessary for sample preparation. Miniaturisation, high-throughput systems, and automation are of interest.

In liquid–liquid extraction the volume of hazardous solvents is reduced by such techniques as liquid-phase microextraction and membrane techniques. Use of supercritical fluids can even lead to extractions without organic solvents.

This miniaturisation is also seen in SPE, with the development of the SPE pipette tip or the SPME technique. Increasing attention is also being devoted to automation, because it leads to safer sample handling, higher throughput, and time-saving. These new automated methods are not frequently used in routine applications, however, because of the need for further optimisation, higher instrumental costs, and increased system complexity. Off-line automatic workstations are often used because they reduce workload and offer flexibility. This flexibility can be relevant in avoiding problems during derivatisation or because of matrix precipitation during analysis. The high protein content of plasma and blood can lead to problems such as precipitation after contact with buffers or solvents, resulting in blocked SPE tubes and analytical columns. The development of RAM and LPS adsorbents can therefore lead to increased use of on-line methods for analytical toxicology.

Thus, although off-line automation is frequently used in analytical toxicology, on-line procedures should become more robust and less complex, and thus become more valuable in everyday routine analysis. Recent developments, however, seem very encouraging for the future.