ANALYST, SEPTEMBER 1993, VOL. 118 1117 Determination of Derivatized Urea Herbicides in Water by Solid-phase Extraction, Methylation and Gas Chromatography With a Nitrogen-Phosphorus Detector Steven Scott Essex Water Co., Trace Organics Laboratory, South tianning field, Chelms ford, Essex, UK Four urea herbicides, isoproturon, chlorotoluron, linuron and diuron, were determined by gas chromato- graphy (GC) with a nitrogen-phosphorus detector (NPD) after derivatization, with detection limits of 0.035, 0.039, 0.041 and 0.036 pg 1-1, respectively. The concentrations of all analytes were linear over the range 0.1-8.0 pg 1-1, with recoveries in excess of 75% from spiked potable waters. In their standard, underivatized form the herbicides were found to be thermally unstable on passage through a GC column.After derivatization, by methylation using iodomethane and a strong base, the resulting compounds were found to be stable at elevated temperatures, and so could be determined by GC. The derivatized herbicides were also analysed by GC-mass spectrometry, in order to elucidate the structures of the derivatized compounds. Each compound yielded a different product with a different retention time. The reaction was of the type typical of nucleophilic displacement, with the methyl group attacking the nitrogen of the amide group, forming a stable tertiary amide and hydrogen iodide gas. This method was found to be more selective than the Standing Committee of Analysts' method owing to the nature of the analysis. Firstly, GC, compared with high- performance liquid chromatography, offers better resolution.There are many ultraviolet absorbers in water which can be detected by the standard method, but use of a specific detector, such as an NPD, offers better selectivity. The method was also applied to other urea herbicides, including monuron, methabenzthiazuron and tebuthiuron, which were also successfully determined, although no quantitative data have been obtained. Keywords: Urea herbicide; gas chromatography with a nitrogen-phosphorus detector; meth ylation; water; solid-phase extraction As stated in Guidance on Safeguarding the Quality of Public Water Supplies ,1 urea herbicides2 are persistent in water and most damaging to the aquatic environment3 and are likely to reach water supplies. The maximum values for isoproturon and diuron within the Essex Water Company boundary of supply4 were 0.39 and 0.14 pg 1-1, found in potable waters in weeks 4 and 35 of 1992, respectively. Raw water samples reached 1.6 pg 1-1 for isoproturon, 0.36 pg 1-1 for chlorotolu- ron and 0.5 pg 1-1 for diuron during 1992.The Water Supply (Water Quality) Regulations 1989 state that individual pesti- cides and related products must not exceed 0.1 pg 1-1, or 0.5 pg 1-1 for total substances. As these limits are not uncom- monly exceeded, it is necessary to have a consistent method with good resolution and that is free from interference. The standard Standing Committee of Analysts' (SCA) method of analysis5 [double solvent extraction, one in alkaline conditions and the other in acidic conditions, followed by detection by high-performance liquid chromatography (HPLC) with ultraviolet (UV) absorbance at two wavelengths] suffers interference from other UV absorbers, and from a poor resolution of analytes (Figs.1 and 2). The proposed method is intended to provide a better alternative method of analysis. It involves derivatization of the urea herbicides by methylation of the amide group followed by separation by capillary gas chromatography (GC) and detec- tion by a nitrogen-phosphorus detector (NPD). The method was assessed for consistency of results, limits of detection, rates of recovery and linearity over the specified range, together with limits of determination of the derivatized products by GC-mass spectrometry (MS). Experiment a1 Reagents and Glassware All solvents were of HPLC grade (FSA Laboratory Supplies, Loughborough, Leicestershire, UK), with the water and methanol filtered through 0.45 pm membranes.The sodium hydride was in 80% dispersion with mineral oil (Aldrich Chemicals, Gillingham, Dorset , UK), the iodomethane (99%) was stabilized with copper (Aldrich) and the dimethyl sulfoxide was also of HPLC grade (Aldrich). A I B I ' \ \r e - Fig. l Spiked recovery from potable water by SCA method. (a) 0.1 pg ml-1 standard and (b) 0.1 pg 1-l spiked sample. Peak A = isoproturon and peak B = chlorotoluron1118 ANALYST, SEPTEMBER 1993, VOL. 118 The extraction cartridges were Octadecyl CIS (1 g) reversed- phase columns (J. T. BV Baker, Deventer, The Netherlands) fitted to a vacuum chamber (Supelchem UK, Saffron Walden, Essex, UK), with a vacuum pump (Charles Austin Pumps, Weybridge, Surrey, UK).Glassware for sample collection was rinsed with acetone, then with tap-water followed by ultra-high purity (UHP) water (Elgastat double ion-exchange resin and activated I I 1 I I I 1 lime - Fig. 2 Spiked recovery from potable water by SCA method. (a) 0.1 pg ml-l standard and (b) 0.1 pg 1y1 spiked sample. Peak A = linuron and peak B = diuron carbon columns, Elga, High Wycombe, Buckinghamshire, UK) and then with the sample. Glassware for analysis was washed with detergent and heated to 100 "C, cooled and rinsed with approximately 10% HCl, washed in acetone and finally in UHP water before baking for 1 h at 100 "C. Extraction Fix the solid-phase cartridges to a suitable vacuum chamber capable of reaching a maximum of 20 in (50.8 cm) of Hg vacuum.Condition the cartridges with 6 k 0.2 ml of filtered methanol followed by 6 k 0.2 ml of filtered water, making sure that the cartridges do not become dry. With vacuum off, apply 3 ml of UHP filtered water to the column. Add 1.0 1 k 10 ml of sample via poly- (tetrafluoroethylene) (PTFE) tubing and suitable PTFE connectors. Aspirate at a flow rate of 15 k 2 ml min-1. Wash the column with 3 k 0.5 ml of UHP filtered water. Dry under vacuum for 5 min. Time - Fig. 3 Mixed urea herbicide standard. 1.0 pg ml-1 standard. A, Isoproturon; B, chlorotoluron; C, linuron; and D, diuron L Temperature programme 190 "C 8 min 140°C L 0 . C min-' 10 min A Time - Fig. 4 Urea herbicides by proposed method (2.0 pg ml-1 standard).A, Monuron; B, isoproturon; C, chlorotoluron; D, linuron; E, diuron; F, methabenzthiazuron; and G, tebuthiuron il. I Time - Fig. 5 Spiked potable water (1 1). (a) 0.1 pg ml-l standard and (b) spiked sample A B D C c.-c. I I I 1 35.00 40.00 45.00 50.00 55.00 Time/min Fig. 6 Derivatized standard by GC-MS in SIM mode. Mixed standard, 5.0 pg ml-1. A, Isoproturon; B, chlorotoluron; C, linuron; and D, diuronANALYST, SEPTEMBER 1993, VOL. 118 1119 Table 1 Analysis of variance: proposed method Recovery (% 1 89.2 83.1 87.4 82.3 91.3 84.5 75.7 85.6 90.9 87.5 79.8 82.9 93.6 88.6 79.2 88.9 Degrees of freedoms 18.9 13.9 18.6 12.1 18.9 14.6 18.2 15.6 16.1 15.4 14.5 13.8 10.9 13.9 14.1 15.9 SW* 0.0122 0.0106 0.0129 0.0099 0.1625 0.1148 0.1815 0.1249 0.0105 0.0118 0.0125 0.0108 0.0726 0.1046 0.1211 0.1157 sbi o.Ooo1 0.0106 0.0129 0.0099 o.oO01 0.1141 0.0796 0.1057 0.0081 0.0102 0.0125 0.0123 0.1453 0.1141 0.1311 0.0907 Linford water 0.041 0.035 0.043 0.033 Mean Isoproturon 0.089 Chlorotoluron 0.083 Linuron 0.087 Diuron 0.082 Isoproturon 0.913 Chlorotoluron 0.845 Linuron 0.757 Diuron 0.856 Isoproturon 0.091 Chlorotoluron 0.088 Linuron 0.081 Diuron 0.083 Linford waterlo.1 pg 1-1 spike- Linford waterll.0 pg 1-1 spike- UHP waterlO.1 jig 1-1 spike- UHP waterll.O pg 1-1 spike- Isoproturon 0.936 Chlorotoluron 0.886 Linuron 0.792 Diuron 0.889 st* 0.0122 0.0159 0.0135 0.0183 0.1625 0.1618 0.1982 0.1637 0.0132 0.0156 0.0177 0.0164 0.1624 0.1547 0.1784 0.1471 UHP water 0.035 0.039 0.041 0.036 Limits of detectionlpg 1-1- Isoproturon Chlorotoluron Linuron Diuron * sw = Within-batch standard deviation.t s b = Between-batch standard deviation. * st = Total-batch standard deviation. @ Ref. 6. Table 2 Analysis of variance: SCA method Recovery (% 1 Degrees of freedoms Mean Isoproturon 0.076 Chlorotoluron 0.076 Linuron 0.073 Diuron 0.078 Isoproturon 0.768 Chlorotoluron 0.815 Linuron NDY Diuron ND Isoproturon 0.067 Chlorotoluron 0.071 Linuron 0.059 Diuron 0.076 Isoproturon 0.781 Chlorotoluron 0.825 Linuron 0.657 Diuron 0.661 Linford waterlO.1 pg I-' spike- Linford waterll.O pg I-' spike- UHP waterlO.1 pg I-' spike- UHP water11 .O pg 1- spike- SW* Sbt 0.009 0.011 0.018 0.023 0.014 0.006 0.019 0.019 0.017 0.012 0.026 0.029 76.1 76.1 72.5 77.7 12.1 16.8 4.7 5.4 76.8 81.5 ND ND 0.103 0.073 ND ND 0.103 0.073 ND ND 0.103 0.094 ND ND 19.1 15.9 ND ND 0.007 0.014 0.014 0.013 0.007 0.005 0.001 0.014 0.009 0.015 0.015 0.019 67.1 71.1 59.4 76.1 14.5 18.4 6.8 4.9 0.064 0.081 0.244 0.086 0.067 0.086 0.224 0.001 0.093 0.117 0.331 0.086 78.1 82.5 65.7 66.1 14.3 14.1 5.1 6.8 Limits of detectionlpg 1-I- Linford water 0.031 0.036 0.059 0.076 UHP water 0.023 0.046 0.046 0.043 Isoproturon Chlorotoluron Linuron Diuron * sw = Within-batch standard deviation. t sb = Between-batch standard deviation.* st = Total-batch standard deviation. @ Ref. 6. 7 ND = Not determined.1120 132 176 1, ' y 1177 ANALYST, SEPTEMBER 1993, VOL. 118 12031 50 45 40 - v) C 35 4- .- z g k 30 (II I & 25 X Q) m U 7 g 20 2 15 U 10 5 0 40 36 32 - v) c, .- 5 28 2 2 U 24 0 Y 0 20 Q) 5 16 -0 3 9 12 8 4 0 51 65 60 * '2 (a) 220 i' 100 140 180 220 174 c3702N 88 I I I 60 100 140 180 40 36 - 32 4- .- C 3 2 28 2 n & 24 c, .- - d z x 2o 8 + 16 C 3 12 8 4 0 5156 + 60 TH60N M+ I 226 I 100 140 180 220 '2 CBHGON / 2 c I M+ 60 100 140 180 220 m/z Fig.7 Mass spectra after derivatization. (a) Isoproturon, (b) chlorotoluron, ( c ) linuron and (d) diuron Elute the urea herbicides with 6 k 0.2 ml of filtered methanol, at a flow rate of 6 ml min-1, into a 10 ml centrifuge tube. Derivatization Evaporate the solvent residue to incipient dryness under a stream of nitrogen, while warming to approximately 40 "C. The nitrogen supply should be adjusted such that the surface of the solvent is just indented and no splashing occurs. Remove the tube from the nitrogen supply. Place approxi- mately 0.2 g of sodium hydride in a separate round-bottomed centrifuge tube, followed by 1 f 0.2 ml of diethyl ether.Evaporate to dryness under a stream of nitrogen. Add 1.5 k 0.2 ml of dimethyl sulfoxide to the tube and mix to form a suspension. One tube is required for three samples. Place approximately 300 f 50 yl of the sodium hydride suspension in the sample tube using a disposable pipette, and immediately add 50 k 5 yl of iodomethane. Leave standing in a fume cupboard for 10 min. In a fume cupboard add 1 k 0.2 ml of UHP water to the samples, taking care not to add too much too quickly as the sample can effervesce vigorously. Add 8 f 0.2 ml of diethyl ether to the sample, stopper the tube and shake vigorously for 2 min. Once the two phases haveANALYST, SEPTEMBER 1993, VOL.118 1121 separated, remove the supernatant organic phase with a clean disposable pipette and place in a clean, pointed centrifuge tube. Evaporate the extract to incipient dryness, accurately dilute to 1 ml with diethyl ether, and stopper the tube tightly. An analysis of variance (ANQVA)6 was carried out over 10 d to ensure at least ten degrees of freedom to assess the performance of the method (Table 1). Instrumentation The GC analyses were carried out using an AMS 92 gas chroma tograp h (Analytical Instrumentation , Cambridge, UK), fitted with an NPD, with the injector operated in split mode at a ratio of 30 : 1. The column was a fused-silica BP1 (25 m x 0.22 mm i.d., 0.25 pm film thickness) capillary column [SGE (UK), Milton Keynes, Buckinghamshire, UK].The operating parameters were: injector temperature 250 "C, detector temperature 250 "C, and column temperature 150 "C (isothermal). The total run time was approximately 8 min. The NPD was operated with the following gas flow rates: carrier (helium) 25 ml min-1, hydrogen 3 ml min-1, and air 150 ml min-1. The output was linked to an SP4400 Chromjet integrator (Spectra-Physics Analytical, Fremont, CA, USA). The GC-MS analysis was performed using an HP 5980 Series I1 gas chromatograph (Hewlett-Packard, Avondale, PA, USA) with an HP 5971A mass-selective detector; splitless injection was carried out for 1.5 min using an HP 7673 auto-injector. The fused-silica capillary column was a DB 1701 (60 m x 0.32 mm i.d. , 0.15 p,m film thickness) column (J & W Scientific, Folsom, CA, USA).Data acquisition and process- ing were performed by Microsoft MS-DOS 3.3 (Microsoft, Redmond, WA, USA], with mass calibration based on perfluorotributylamine (PFTBA). All mass spectra were obtained under electron-impact conditions (70 eV). The mass spectra were continuously scanned over the mass range 50-450 u at 0.5 s decade-'. Results Under the conditions described, separation of the four derivatized products was achieved (Fig. 3). With slight temperature programming, seven derived herbicides could be resolved (Fig. 4). Spiked recoveries were carried out on UHP water and a well water supply (Linford Well, Essex, UK) (see Table 1). A spiked recovery was also performed on a river-derived potable water (Hanningfield Final Water, Essex, UK), which shows few interference peaks (Fig.5). Calibration was over the range 0.1-8.0 pg 1-1 within which all analytes provided a linear response passing through the origin. Derivatized standards of 5.0 p,g 1-1 were run on the mass spectrometer as described (Fig. 6). From the data obtained it was possible to elucidate the structures of the derivatized products (Figs. 1 and 2). Discussion It can be seen that interference peaks and relatively poor recoveries make the previous methods for the determination of herbicides unsuitable for low-level analysis of river and potable waters (Figs. 7 and 8). The proposed method is based on methylation of the amide NH group to form a stable tertiary amide (Fig. 8). The reaction is a typical nucleophilic substitution reaction of the type shown below, where X is a readily displaced, stable leaving group R-X Sample-NH - Sample-N-R + HX Base 1 1 lsoproturon MI mlz= 148.2 M2 mlz= 72.2 1 1 Chlorotoluron MI mlz = 154.2 M2 mlz= 72.2 1 1 CI- L2 Linuron MI mlz= 175.1 M2 mlz = 88.2 Cl' L 2 Diuron MI mlz= 175.1 M2 mlz= 72.2 Fig.8 Structures of derivatized herbicides showing major fragmen- tation For urea herbicides, the leaving group is iodine, with the sodium hydride acting as a strong basic catalyst, and the reaction being achieved at ambient temperature. Efferves- cence of the acrid hydrogen iodide can be seen during the reaction. Separation of the four urea herbicides is achieved easily without the need for temperature programming, and although other peaks can be detected, they do not generally have the same retention time as the analytes and are of lower frequency than those for analysis by HPLC with UV detection.The limits of detection, calculated by multiplying the standard deviation of the mean of the low spikes by 3.3 (WRC, NS30, Revised June 1989)6 (Table l), are, in reality, much lower than the values stated, with peaks of concentration 0.02 pg 1-1 being seen for all analytes. The blank samples yielded no response and so could not be used for determining the limits of detection. For further confirmation of the derivatized product, the samples can be analysed by GC-MS in the selective-ion mode (SIM), searching for the base peak and molecular ion (Fig. 6). Comparison With Standard SCA Methods The standard SCA methods for the determination of urea herbicides involves sample concentration by extraction, evap- oration of the extract to incipient dryness, dissolution of the residue in the mobile phase, and analysis, using polar solvents, by HPLC with UV detection.1122 ANALYST, SEPTEMBER 1993, VOL. 118 In order to determine the four urea herbicides mentioned here, it is necessary to perform two separate extractions, one in alkaline conditions for isoproturon and chlorotoluron, and the other in acidic conditions for linuron and diuron.To gain the best response, they should be analysed at two separate wavelengths, viz., 240 and 220 nm, respectively. Although reasonable separation and response can be achieved by using this standard method (Figs. 1 and 2), at low levels of analyte in water (CO.1 pg 1-I), background peaks are significant and, without additional quantitative data, should be regarded as maxima (Figs. 1 and 2).Phenoxy acidic herbicides are a known contaminant, and, when present, the ureas have to be re-analysed at 270-280 nm. Because of the nature of UV detection and the fact that river and potable waters contain many UV absorbers, it is clear that analysis by GC-NPD will be much more selective. Recoveries performed on the same waters, and at the same concentrations, are generally about 10% lower for the SCA method, with comparable limits of detection (Tables 1 and 2). References 1 Department of the Environment Welsh Office, Guidance on Safeguarding the Quality of Public Water Supplies, HM Sta- tionery Office, London, 1990, pp. 99-102. 2 Worthing, C. R., and Hance, R. J., The Pesticide Manual, British Crop Protection Council, Farnham, Surrey, 9th edn., 1991, pp. 161, 322,507 and 520. 3 Ivens, G. W., The UK Pesticide Guide, British Crop Protection Council, Cambridge, UK, 1992, p. 18. 4 Essex Water Co., Water Quality Register, Chelmsford, Essex, 1992. 5 Standing Committee of Analysts, The Determination of Carba- mates, Thiocarbamates, Related Compounds and Ureas in Water, HM Stationery Office, London, 1987, pp. 15-21. 6 Water Research Centre, NS30-A Manual on Analytical Quality Control for the Water Industry, Marlow, Buckingham- shire, 1989, pp. 57 and 131. Paper 2104602 E Received December 14, 1992 Accepted February 25, 1993