4. SOLID-PHASE EXTRACTION–GAS
4.2. Solid-Phase Extraction–Gas Chromatography–
Tandem Mass Spectrometry
In recent years, GC–ion-trap detection (ITD) systems that can perform tandem MS (MS–MS) on a routine basis have become commercially available [86]. Be- cause ITD provides good sensitivity as well as increased selectivity in the MS–
MS mode, an on-line SPE–GC–ITD system was optimized for the trace-level determination of polar and apolar pesticides [53]. The Autoloop interface (see section 3.2.3) was operated at an injection temperature of 90°C, which permitted the determination of thermolabile pesticides such as carbofuran and carbaryl.
With sample volumes of 10 to 30 ml and a copolymer SPE cartridge, linear calibration curves were obtained for several pesticides over the range of 0.1 to 500 ng/L. Fully satisfactory tandem mass spectra were obtained at levels as low as 0.1 ng/L level in tap and river water. The system was used to analyze water from European and Asian rivers, and the determination of microcontaminants at 8 to 16 ng/L levels did not cause any problems (Fig. 11). Relevant analytical data are presented in Table 7. One conclusion may be that, for this target-com- pound type of analysis, a sample volume of 1 ml or less will be sufficient to comply with governmental directives.
In another application [78], SPETD–GC–ion-trap MS–MS was optimized for alachlor and metolachlor. Appropriate precursor ions with a high m/z value were selected from the electron impact (EI) and positive chemical ionization (PCI) spectra, and the CID voltage optimized so that the highest abundance of a selective product ion was observed to achieve maximum sensitivity. Detection limits of 0.1àg/L were reported for alachlor and metolachlor for 100-àl samples.
As an example, Figure 12 shows the analysis of 100 àl of Rotterdam harbor water, in which metolachlor was suspected to be present. Figure 12a shows the extracted ion SPETD–GC–MS chromatogram of mass m/z 162 and the mass spectrum of the peak at the retention time of metolachlor. The result cannot be called satisfactory. If, however, the analysis was performed in the MS–MS mode (Fig. 12b), the sample background had completely disappeared, and identification of the compound as metolachlor was perfectly straightforward due to the much higher selectivity. Quantification on the basis of the response of m/z 162 gave closely similar results for MS and MS–MS detection, i.e., 1.3 and 1.2 àg/L, respectively. The presence of metolachlor was also confirmed by SPETD–GC–
PCI–MS(–MS).
In many monitoring programs, a mixture of compounds must be addressed, a number of which can be analyzed by means of GC–MS, while others require
186 Hankemeier and Brinkman
Figure 11 Total ion current and reconstructed ion chromatograms obtained after SPE–
GC–MS/MS of 10 ml Rhine River water at m/z 172 (desethylatrazine), 200 (atrazine), 160 (alachlor), and 162 (metolachlor). (From Ref. 53.)
On-Line Sample Preparation for Water Analysis 187
Table 7 SPE–GC–MS-MS of Pesticides in 10 ml Tap Water Linear range
Analyte (ng/l) R2 RSDb(%) LOD (ng/l)
Desethylatrazinea 2–200 0.9941 18 0.5
Atrazine 1–200 0.9993 10 0.2
Metolachlor 1–200 0.9997 6 0.04
Trifluralin 0.1–200 0.9993 6 0.01
Carbofuran 0.1–200 0.9981 7 0.1
Parathion-methyl 2–200 0.9991 7 1
Alachlor 0.1–200 0.9996 6 0.05
Fenitrothion 0.1–200 0.9961 7 0.1
Fenthion 0.1–200 0.9969 6 0.1
Parathion-ethyl 5–200 0.9993 6 2
Carbaryl 1–200 0.9979 6 0.1
aLess good results for this polar analyte mainly due to integration problems.
bRSD determined at 10 ng/l analyte concentration (n⫽7).
Source: Ref. 53.
an LC-based approach. If, on the LC side, a particle beam (PB) interface is used, the two techniques can be combined in one setup, sharing the sample-handling unit as well as the MS detector. With this so-called Multianalysis system [87,88], two subsequent runs are performed per sample. First, the analytes from an approx- imately 10-ml sample trace enrichment are desorbed and sent to the GC–MS; in the next run, a 100- to 200-ml sample (with the larger volume compensating for the lower sensitivity) is preconcentrated, desorbed, and analyzed by LC–diode array–UV detector–MS. In both instances, classical EI spectra are generated that can be searched by using any GC–MS library. In one example, nine test com- pounds, triazines, anilides, and organophosphorous pesticides, were added to tap water. Detection limits were 0.005 to 0.1àg/L for SPE–GC–MS (10-ml samples) in the full-scan mode, and 0.5 to 7àg/L for SPE–LC–PB–MS (100-ml samples) in the full-scan and 0.05 to 1 àg/L in the SIM mode. Obviously, there is an urgent need to improve the performance of the PB interface, but this is a topic outside the scope of the present discussion. When using negative chemical ioniza- tion (NCI) MS, with methane as a reagent gas, the detection limits for SPE–
GC–MS and SPE–LC–PB–MS could be improved 10- to 30-fold for most of the chlorinated pesticides [89]. The Multianalysis system was used to monitor the pollution at a number of sampling sites along the Nitra River (Slovak Republic), a tributary of the Danube, during a 2-year surveillance program.
188 Hankemeier and Brinkman
Figure 12 (a) SPETD–GC–MS ion chromatogram (m/z 162) of 100àl of Rotterdam harbor water. Insert shows the mass spectrum of the prominent peak. (b) SPETD–GC–
MS/MS daughter chromatogram (m/z 162; parent mass, m/z 238) of same sample. Insert shows the mass spectrum of the prominent peak. (From Ref. 78.)
On-Line Sample Preparation for Water Analysis 189