Fully automated on-line liquid chromatographic separation system for polar pollutants in various types of water

Analytzca Chumca Acta, 268 (1992) 55-65 Elsewer Science Publishers B V , Amsterdam

55

Fully automated on-line liquid chromatographic separation system for polar pollutants in various types of water J Slobodmk and E R Brouwer Depatiment of Analytuzal Chemrstry, Free Unrvers~@,De Boeklaan 1083, 1081 HVAmsterdam (Netherlands)

R B Geerdmk and W H Mulder Instttute for Inland Water Management and Waste Water Treatment, P 0 Box 17, 82oOAA L.elystad (Netherlands)

H Lmgeman and U A Th Brmkman Department of Analykal Chemwy, Free Unrversity, De Boelaan 1083,108l HVAmsterdam (Netherlands) (Received 17th January 1992, revised manuscnpt received 4th May 1992)

Abstract

A fully automated column hqmd chromatographlc separation system using on-line trace enrichment, gradient elutlon and diode-array detection for the trace-level detennmation of polar pollutants IS described Automation of the system was achieved by means of an automated cartridge-exchange system (PROSPEKT) Relevant parameters such as pH, volume and iomc strength of the sample, flow-rate during the ennchment step and wavelengths and band widths dunng detection were optlmlzed for eighteen pollutants m vanous types of water at concentration levels below 5 pg 1-l The determmatlon hnut for all test compounds in liquid chromatographlc grade water was 0 1 pg I-‘, and Identtication, vra diode-array spectra, could be perfonned at the same level The mean relative standard deviations of the peak areas and the retention tunes for all the test compounds were 10% and 0 3%, respectively, at the 5 pg 1-l level for nver Rhme water Eceywords Liquid chromatography,

Sample preparation,

According to the Rhine Actlon Program of 1987, mltlated after a disastrous fire at a pharmaceutlcal company m Basle (Switzerland), the AQUALARM envuonmental momtormg and alarm system for the major Dutch river systems, which momtors the presence of orgamc and morgame mlcropollutants m the waters of the avers Meuse and Rhme, and the branches of the rover Rhine (Waal, IJsel), was developed and mstalled Correspondence to H Lmgeman, Department of Analytical Chemistry, Free Umversity, De Boekan 1083, 1081 HV Amsterdam (Netherlands)

Pesticides,

Polar pollutants,

Waters

[l] So far, no such system IS available for the determmatlon of polar pollutants In the context of both the Rhme Action Program and the Rhme Basm Program [2,3], attention has been devoted to the development of a fully automated and robust early warnmg system for the determmation of polar pollutants at trace levels (< 5 pg 1-l) wth a procedure that sunultaneously provldes structural mformatlon of the analytes [3,41 As a sign&ant number of polar pollutants cannot be determmed by gas chromatography (GC), column hqmd chromatography (LC) IS the separation techmque of first choice [5]. Advan-

0003-2670/92/$05 00 8 1992 - Elsewer Science Publishers B V All nghts reserved

J Slobodmk et al /Anal

56

tages of LC over GC are the slmphclty and robustness of the system However, compared with GC, LC has some hmltatlons with respect to selectlvlty and sensltlvlty These disadvantages can be solved by using mass spectrometrlc or dlodearray detectlon (DAD) m combination with online trace enrichment techniques based on small cartridges [2,3] or membrane extraction discs [6,7] The development of a non-automated on-line trace enrichment gradient LC-diode-array detectlon (DAD) system for the determination of polar pollutants m river Rhme water, has been reported recently [3] In this paper, a fully automated LC procedure using on-line trace ennchment and DAD for the ldentlflcatlon and determination of eighteen polar pollutants, mcludmg carbamates, tnazmes, chloroamhnes, phenylurea herblcldes, aromatlc chloroammes, hydroketolactones, chlorinated phenols and organophosphorus compounds, at trace levels 1s described The mam aim of this study was to investigate the various parameters that influence the sensltlvlty and the overall performance of such an automated system

EXPERIMENTAL Equipment

The LC analyses were performed with two Model 400 solvent-delivery systems (Applied Blosystems, Ramsey, NJ), a laboratory-made srxport mJectlon valve (Free Unmersrty, Amsterdam), equipped with a 30-~1 mjectlon loop, and a Model 1000 S diode-array detector (Applied Blosystems) For single-wavelength momtormg, the detector was set at 230 nm with a band width of 50 nm During recording of the absorbance spectra the diode width was 2 nm and the spectra were recorded from 190 to 370 nm Data from the detector were collected by a standard Model 80 286 mlcroprocessor-based computer with 1 Mbyte of extended memory, and subsequently handled by the LabCalc software (Galactic Industries, Salem, NH) All experiments were done at amblent temperature Trace enrichment of the analytes was done on 10 mm x 2 0 mm I d disposable trace enrichment

Chum Acta 268 (1992) 55-65

cartridges of the PROSPEKT (Spark Holland, Emmen, Netherlands) automated sample preparation unit The cartridges contained styrenedlvlnylbenzene copolymer (15-25pm PLRP-S, Polymer Laboratories, Church Stretton) with a pore size of 10 nm The cartridges were condotloned via a solvent-delivery umt (SDU) (Spark Holland), and sampling was done with a Model 300 preparatrve pump (Gynkotek, Munich) The PROSPEKT was used to condltlon the trace enrichment cartrldges and to control the whole system

Chemicals and reagents The stainless-steel analytical columns were a 100 mm x 2 1 mm 1 d Hypersll C,, column (Hewlett-Packard, Waldbronn), a 150 mm X 4 6 mm 1 d LlChrosorb RP-18 column (Chrompack, Mlddelburg, Netherlands) and a 250 mm X 4 6 mm 1 d Chromspher Pesticides column (Chrom-

pack) All columns were packed with 5-pm partlcles A 250 mm x 4 6 mm 1 d PLRP-S column (Polymer Laboratories) packed with 5-pm partlcles was also tested LC gradient-grade acetomtnle, methanol and water were obtained from J T Baker (Deventer, Netherlands) Unless mentioned otherwise, LCgrade water was used “M&-Q water” was prepared by ultraflltratlon usmg a Mllh-Q system (M&pore, Bedford, MA) Dlsodmm hydrogenphosphate, concentrated perchlorlc acid and phosphoric acid (85%) were supplied by J T Baker The various solutes were supplied by Hoechst (Frankfurt), Rledel-de Haen (Seelze), Promochem (Wesel), and Dr S Ehrenstorfer (Augsburg) They all were at least 95% pure The eighteen pesticides used and the numbering used m this study are as follows carbendazlm cl), metamltron (2), chlorldazon (31, aldlcarb (4), bromacll (5), slmazme (6), 2-chloroamhne (71, atrazme (81, dmron (9), monolmuron (lo), warfarm (10, hnuron (12), 3 3’-dlchlorobenzldme (13), barban (14), dmoterb (151, dmoseb (16), pentachlorophenol (17) and phoxlm (18) River Rhine water was collected at Loblth and obtained from RIZA (Lelystad, Netherlands)

J .Vobodnzk et al /Anal

57

Chun Acta 268 (1992) 55-65

Pnor to use, the samples were filtered over a 0 45-pm BA membrane filter (Schlelcher & &hull, Dassel) Procedures Stock standard solutions (1 mg m 5 ml) of the analytes m methanol were prepared every 6 weeks Workmg standard soluttons of the test compounds were prepared every 3 days by ddutmg the stock standard solutions with LC-grade, M&-Q, tap or rwer Rhme water to a concentration of 001-5 pg 1-l All stock and workmg standard solutions were stored at 4°C m the dark The gradient conditions for testing the analytlcal columns and using the LlChrosorb column (in the final procedure) were as follows solvent A, acetomtrde-0 01 M phosphate buffer (pH 3) (90 + 10, v/v) and solvent B, acetomtnle-0 01 M phosphate buffer (pH 3) (5 + 95, v/v) The gradlent profile was 0 mm, 100% B, changed linearly to 100% A (55 mm) and then linearly to 100% B (60 mm) The flow-rate was 10 ml mm-’ Prior to use, all eluent components were degassed usmg a vacuum ultrasonic bath (5 mm) Before startmg the first gradlent, the system was equlhbrated dally for 15 mm wxth solvent B (flow-rate 1 ml mm-‘) The trace enrichment cartndges were condotloned with 2 ml of methanol (flow-rate 2 ml

mm-‘) and then 2 ml of 0 001 M perchlorlc acid (1 ml mm-‘) using the SDU of the PROSPEKT Subsequently 100 ml, or m some mstances 150 ml, of sample were ennched on the cartridge at a flow-rate of about 4 ml mm-’ Desorptlon of the cartxxdge was performed (vvlth solvent A) by couplmg the cartridge on-lme Hrlth the analytical column and startmg of the gradient (see above) All steps of the procedure were automatically controlled by the PROSPEKT software In Table 1 the sample preparation programme (SPP) for the overall procedure 1s given

RESULTS AND DISCUSSION

Automutwn of the method Automatlon of on-lme trace ennchment systems can now be performed Hrlth commercially available systems Both the OSP-2 (Merck) and the PROSPEKT (Spark Holland) systems are equipped with a disposable cartrldge unit As an addltronal unit, the PROSPEKT contams a solvent-dehvery unit (SDU) for condltlonmg and washing of the trace enrichment cartridges In this study, a preparative LC pump was used for loading the samples The time sequence of the automated procedure IS given m Table 1 All steps of the sample

TABLE 1 Sample preparation programme of automated system a Tune (nun s) 0000 00 01 02 00 03 00 05 00 0700 3102 3130 3131 33 31 65 00

Solvent

Valve 1

Valve 2

PURGE

ELUTE

1 3 2 2 3 4 1 1

PURGE ELUTE PURGE ELUTE ELUTE

Flow-rate (ml mm-‘) 20 4 16 20 10 00

PURGE

Aux2

AlLx6

Other

CHC ON OFF ON OFF ON

20 00

OFF END

a Solvents 1, methanol, 2, 0001 M perchlonc acid, 3, sample, 4, eluent Valves 1 and 2, positIons accordmg to Fig 1 Aux 6, connection of PROSPEKT with 1000 S dmde-array detector, posltlons ON/OFF, Aux 2, connection of PROSPEKT wth LC pump, posItIons ON/OFF CHC, change cartridge END, end of analysis

J SIolbodnrket al /AnaL Chm Acta 268 (1992) 55-65

58

preparation programme (SPP) were programmed controlled by the on, and automatically PROSPEKT software The total set-up of the system is shown m Fig 1 The procedure starts by changmg the trace enrichment cartridge (00 mm 01s) After condltlomng of the cartridge (00 01-3102) the analysis 1s started by an awullary (Aux 6) slgnal to the detector (3130) In detad, the followmg steps take place 00 01, flushmg of the system caplllarres with methanol, 02 00, wetting of the cartrldge mth methanol and flushmg of the sample capillaries with the sample, 03 00, flushing of the system caprllarles with perchlorlc acid, 05 00, flushing of the cartridge with perchlonc acid to displace the methanol, 07 00, start of the sampling, 31 02, end of the sampling, 3130, desorptlon of the cartridge and start of the LC separation, 3131, flushing of the system caplllanes with methanol, 33 31, end of the flushmg, 65 00, end of SPP This SPP can be used to perform a series of analyses m a fully automated manner Therefore,

I

at the end of the fu-st run, the system munedlately starts with condltlonmg and sampling of the second cartridge The 60-mm gradient programme starts at 3130 of the first SPP and ends at 26 30 of the second run, which means that there are 5 mm to flush the system Hnth eluent B to prepare it for the next analysis at, again, 31 30 All data are automatically stored by the set option m the LabCalc software and are available at the end of the sequence The auxrhary (Aux 2) connects the PROSPEKT wth the sample loading pump As a result the exact flow-rate was 4 16 ml mm-’ m this mstance The atilary (Aux 6) of the PROSPEKT 1s connected wth the ground and event remote inputs of the detector The SPP 1s defined m a specific RUNPROGRAM which 1s used to run a series of analyses m a fully automated manner An analysis 1s started from the keyboard of the PROSPEKT by pressmg START and typmg the proper START and STOP RUNPROGRAM numbers

1

PURGE

g!dR -I

8

I

8

1

El.ulE

-tl

2

1

4

WA!STE

I

I

I SDU

Fig 1 Automated on-lme trace enrichment system for water samples 1, 2, 3 = High-pressure valves, 4 = trace ennchment cartridge,, 5, 6 = high-pressure gradient pumps, 7 = preparative pump for sample loadmg, 8 = analytxal column, 9 = detector, 10 = computer, 11= pnnter, 12 = pulse damper, 13 = purge pump, 14 = solenoid valve

J Sbbodmk et al /Anal

59

Chm Acta 268 (1992) 55-65

I

II I

J

0 014

AU

il

15

30

45

60

Time [mm] F@ 2 Chromatograms of mixture of test compounds (5 pg 1-l) m rwer Rhme water Ennched volume (a) 30 ml and (b) 100 ml For all other condltlons, see text

Optmuatwn of the analytccal system Several parameters that can be expected to have profound mfluence on the detectlon sensltlvlty of the test pollutants and/or the robustness of the total analytical system were studied Andytmzl column Four analytical columns were tested m combmatton urlth the 10 mm X 2 0 mm I d PLRP-S trace ennchment cartridges As the sorbent of the analytlcal column should possess a hydrophoblclty snmlar to that of the sorbent m the cartridge,, three C,,-bonded S&XI columns and a polymer column were tested Analysis of the test nuxture on the 5-pm Hypersd Cl, column gave a hardly satisfactory resolution, although good resolution was obtamed after loop 1nJectlon of the same nuxture l%s slightly surprrsmg behavlour was not mvestlgated further Completely satisfactory results were ob-

tamed with the 5qrn Chromspher Pestlcldes, S-pm LEhrosorb RP-18 and S-pm PLRP-S columns As the overall separation efficiency of the LKhrosorb column was better than that of the Chromspher Pesticides column, and the S-km polymer mater& caused a higher back-pressure than the 5qrn &bonded phases, the 150 mm x 4 6 mm I d LlChrosorb RP-18 column was used m all further work Wabsorbance detectron In order to optrrmze the DAD, experunents were performed at various wavelength-bandwidth combmatlons In each expernnent, 100 ml of a text nuxture 111LC-grade water (analyte concentration 0 5 pg 1-l) were ennched and analysed under the condtlons grven under Experimental The chromatograms were recorded at 230 nm usmg a band width of 50 nm, a large band width

.I Slobodmk et al /Anal

60

TABLE 2 Maximum absorbance wavelengths and breakthrough volumes of test compounds No

Compound

Maximum response wavelength (nm) a

Breakthrough volume (ml) b

1 2 3 4 5 6 7 8 9 10 11 12 13

Carbendazlm Metamltron Chlorldazon Aldlcarb Bromacll Simazme 2-Chloroamhne Atrazme Dmron Monohnuron Warfann Linuron 3,3’-Dlchlorobenzldme Barban Dmoterb Dmoseb Pentachlorophenol Phoxlm

220 210 220 210 210 220 210 220 245 210 210 210

25 15 15 25 40 50 20 90 65 75 >lOO > 100

210 210 210 280 210 280

> 100

14 15 16 17 18

> 100 > 100 >lOO >lOO >lOO

a Trace enrichment of 100 ml of LC-grade water at a concentratlon level of 0 5 pg 1-l usmg a detectlon band width of 5 nm b Breakthrough volumes on PLRP-S cartrldge usmg the fraction collection method [3] All other condltlons are given m the text

bemg advantageous only when a single wavelength has to be used for a wide variety of analytes However, as most of the compounds tested possess absorbance maxima at mutually divergent wavelengths, the response can be improved slgmfscantly by usmg snnultaneous detectlon at a number of carefully selected wavelengths These turned out to be 210, 220,245 and 280 nm [3,8,91, all at a band width of 5 nm as one 1s now workmg close to the absorbance maxuna For all analytes, at least one of the four wavelengths selected gave better results than the 230 nm-50 nm combmatlon, with a 2 5-fold increase bemg observed for aldlcarb, warfarm and pentachlorophenol The wavelengths selected are given m Table 2 Truce ennchment The breakthrough volumes of the test compounds were determined m Mdh-Q water at a splkmg level of 2 pg ml-’ by usmg the

Chrm Acta 268 (1992) 55-65

fraction collection method [3] The results are Included m Table 2 Actually, when using the overall analytical procedure, the UV response of all analytes showing early breakthrough increased on Increasing the sample volume from 30 to 100 ml This IS illustrated m Fig 2, where the LC-UV chromatograms obtamed after trace enrichment of 30 and 100 ml of spiked river Rhme water are shown For all pollutants, a 2-3-fold increase m the UV response IS seen, which can be explained by the lower concentrations used than during the study of the breakthrough volumes [21 Coquart and Henmon [lo] carried out trace enrichment from sample volumes of up to 500 ml Therefore, m this study, lOO- and 500-ml tap water samples spiked with 0 5 pg 1-l of all the test solutes except the early elutmg carbendazlm and metamltron were processed As the time needed for trace enrichment should, ideally, be essentially the same as the time required for the LC separation, m this mstance sample loading was performed at a flow-rate of 15 instead of 5 ml mm- ’ This did not cause any problems, which IS m agreement with early work m which flow-rates of up to 25 ml mm-’ were used without adverse effects [ll] As regards trace enrichment, the UV response of all stieen analytes was found to increase 2 5-5-fold on loading 500 instead of 100 or 150 ml of sample Although it should be emphaslzed that these are only prehmmary results, It 1s interesting that high-speed loading of large sample volumes 1s probably a viable approach m early warning procedures Sample treatment Two further aspects of mterest are the sample pH during trace ennchment, which may be a cntlcal parameter for basic analytes, and the removal of matrix Interferences In order to study the pH effect, spiked (1 pg I-‘) river Rhme water samples were adjusted to pH 3, 5 8, 6 8, 7 8 and 8 8 and lOO-ml ahquots were analysed For the present set of test compounds, the response over the whole pH range studled did not vary slgmficantly The matrix peak became shghtly larger at lower pH values, but this did not affect the performance of the system As the pH of river Rhine water (Loblth, Netherlands) varied between 6 2 and 8 2 m the period March 1990-June 1991, It will be possible to

J Skddnrk

et al /Anal

61

Chm Acta 268 (1992) 55-65

1

0 00065

0

15

30 Time [min]

45

AU

60

Fig 3 Chromatograms of rmxture of test compounds m LC-grade water Concentrations of the compounds (a) 30, (b) 50, (c) 100 ng I-’ Ennched volume IS 150 ml For all other condmons, see text

J Slobodmk et aL /Anal. Cham Acta 268 (1492) 55-65

62

process unbuffered river Rhme water samples m the present automated system In a previous study, attempts to remove matrix interferences from the PLRP-S trace enrichment cartridge by washing with, e g , acetomtnle-water (30 + 70, v/v> (pH 3) were not successful [12] Because the humlc and fulvlc substances may be expected to be negatively charged at high pH, which should facilitate their removal, washmg

with 0 001-O 1 M sodmm hydroxide solution was tested However, even at pH 13 the matrix peak decreased by only ca 30% Moreover, the decrease tended to vary conslderably between runs Finally, the addition of 0 1-O 5 mol 1-l of sodium chloride to river Rhme water samples, to test the influence of the ionic strength, did not have a positive effect on the response of the analytes as a result of a saltmg-out effect

I

O.OOSSAU

,

15

30

45

60

Time [r&n] Fig 4 Chromatograms of 150 ml of enrlched tap water (a) and correspondmg chromatograms after spllung with test compounds at the 0 1 pg 1-l (b) and 1 pg I-’ k) level For all other condltlons, see text

J Slobodmk et al /Anal

Chm

63

Acta 268 (1992) 55-65

Performance of the total system Usmg the optmuzed experunental condltlons dlscussed m the previous sections, the detectron hmlts of the eighteen pollutants m LC-grade wa-

ter were determmed by enrichment of 150-ml samples spked at the 30-100 ng 1-l level The chromatograms m Fig 3 demonstrate that the detection lumts are dlstmctly below 100 ng I-’

I

0.014 AU

80

45

Time [min]

I

lb

30

45

0 014 AU

80

Time (min]

l3g 5 Chromatogramsof mtiure of test compounds (5 pg I-‘) III(A) LC-grade water and (B) nver Rhme water Ennched volume 1s 100 ml For all other conrhtxms, see text

J Sbbodruk et al /Ad

64

for all compounds, even with single-wavelength (230 nm) detectlon Rehable spectral recogmtlon was possible m all instances at the 100 ng 1- ’ level As examples of real sample analyses, automated runs for blank and spiked (0 1 and 1 pg 1-l) tap water are shown m Fig 4 and for spiked (5 pg 1-l) river Rhine water and, for compartson, LC-grade water m Fig 5 It 1s interesting to add that, with river Rhine water, spectral recognition can easily be carried out at the 3 pg 1-l level, as 1s demonstrated for monohnuron m Fig 6 The detection limits for the analyses m spiked surface water and tap water typically are about 1 and 0 3 CLg1-l, respectively Cahbratlon graphs for all the test solutes were constructed using spiked river Rhme (100 ml, 05-5 pg 1-l) and Mllh-Q (150 ml, 005-2 pg 1-l) water For most analytes, the correlation coefficients are satisfactory (Table 3) and, with two exceptlons (metamltron and 2-chloroamlme, matrix interference), they are essentially equal for both types of sample The repeatablhty of the automated system was tested several times by enriching a hmlted number of loo-ml spiked (5 pg 1-l) river Rhine and other water samples, using a fresh PLRP-S cartrudge for each run The results m Table 4 are

Chm Acta 268 (1992) 55-6.5

TABLE 3 Correlatton coefficients of test compounds Mdh-Q water and river Rhme water a

after analysis m

No

Compound

rMw

raw

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Carbendazlm Metamltron Chlorldazon Aldicarb Bromacd Sunazine 2-Chloroamhne Atrazme Dmron Monohnuron Warfarm Lmuron 3,3’-Dlchlorobenzldme Barban Dmoterb Dmoseb Pentachlorophenol Phoxnn

0 9898 09946 0 9931 09801 0 9998 0 9.540 09907 0 9919 0 9849 0 9917 0 9967 0 9988 0 9631 0 9832 0 9961 0 9998 0 9944 _b

0 9863 -b 0 9929 0 9842 0 9693 0 9999 -b 0 9991 0 9971 0 9994 0 9992 0 9983 0 9964 0 9923 0 9976 0 9979 0 9995 0 9912

a rMw, analyses m M&-Q water (005-2 pg I-‘), enrlched volume 150 ml, rRw, analyses m river Rhme water (0 5-5 pg l-l), enrlched volume 100 ml Each cahbratlon graph was calculated from SIX data pomts, which were measured m tnphcate For all other condltlons, see text b Correlation coefflclent < 0 9 owmg to mobde phase or matrur mterferences

satisfactory as far as the retention tunes are concerned The peak-area data, with a relative standard devlatlon (R S D ) of less than 30% m all instances, are, by themselves, acceptable By usmg peak-height data, however, similar results were obtamed The wide gap between a series of very low and a series of high R S D values 1s not readily explained Other series of runs typlcally gave sumlar results, as shown m Table 4

I

235

280 Wavelength

325

a

[nm]

Fig 6 Absorbance spectra of monohnuron m rwer Rhine water at concentrations of (a) 3, (b) 1 and (c) 0 5 /~g I-’

Concluslo?Is A fully automated system for the trace-level determination of polar pollutants m surface and tap water IS described The system consists of an automated cartridge-exchange module (PROSPEKT) contammg small cartndges loaded with PLRP-S polymer packmg material, and a reversed-phase IX-DAD system All steps of the analytlcal procedure are controlled by the PROSPEKT software

J Slobodmk et al /Anal TABLE 4

Retention times and peak areas of test compounds (5 pg I-‘) m rwer Rhme water after a single use of four trace ennchment cartrldges a No

Compound

RTRw (RSD)

PA,(RSD)

1 2 3 4 5 6 7 8 9 10 11 12 13

Carbendazlm Metamltron Chlorldazon Aldlcarb Bromacd Slmazme 2-Chloroamhne Atrazme Dmron Monohnuron Warfarm Lmuron 3,3’-Dlchlorobenzldme Barban Dmoterb Dmoseb Pentachlorophenol Phoxlm

14 3 (0 1) 16 6 (0 4) 17 8 (0 1) 22 0 (0 1) 23 0 (0 2) 24 0 (0 1) 25 2 (0 2) 28 2 (0 1) 29 0 (0 1) 29 8 (0 1) 32 8 (0 3) 34 0 (0 4)

7 7 (27) 5 l(20) 5 3 (28) 3 5 (2 4) 8 7 (28) 113(04) 3 3 (22) 16 1 (27) llS(O3) 13 9 (0 4) 7 2 (0 3) 14 5 (3 4)

34 9 (0 1) 37 0 to 4) 39 6 (0 5) 40 3 (0 5) 40 8 (0 4) 42 5 (0 4)

10 6 (0 4) 115 (15) 7 l(0 1) 4 0 (11) 112(49) 3 2 (13)

14 15 16 17 18

65

Chm Acta 268 (1992) 55-65

= RT,w, retention time (mm) m river Rhme water, PA,,, peak area (arbitrary units) m rwer Rhine water, R S D, relative standard devlatlon (%) For other conditions, see text

The various parts of the system have been optlmlzed using eighteen polar pestlcldes as test solutes The automated system has been used to analyse 30-150~ml blank and spiked, surface and tap water samples Detection limits m units of concentration vary between about 0 1 and 1 pg l-‘, depending on the sample type and the characterlstlcs of the analyte Spectral resolution 1s generally possible at three fold higher concentrations The linearity and precision are acceptable evenatthelowpgl-l splkmg level The present, admittedly still hmlted, experience indicates that unattended operation of the total system does not encounter special problems With a slightly dlfferent fully automated system, m which only the ABI gradient and DAD system was replaced with the HP 1090 system, over 50 samples could be run unattended The analytical performance of the system essentially meets the prerequlsltes of the I-list of

the Dutch National Pohcy Document on Water Management 111, which states that it should be possible to detect 3 pg l- ’ of a pesticide m surface water and 0 1 pg 1-l m tap water This work was supported by the Institute for Inland Water Management and Waste Water Treatment (RIZA) (Lelystad, Netherlands) as part of the Rhine Action Program Part of this research will be incorporated m the Rhine Basin Program (Amsterdam/Waldbronn)

REFERENCES 1 Water m the Netherlands a Time for Actlon, National Policy Document on Water Management, House of Commons, Meeting penod 1988-89, The Hague, No 21250 (l-2) 2 E R Brouwer, I Llska, R B Geerdmk, P C M Frmtrop, W H Mulder, H Lmgeman and U A Th Brmkman, Chromatographla, 32 (1991) 445 I Llska, E R Brouwer, A G L Osthelmer, H Lmgeman, U A Th Brmkman, R B Geerdmk and W H Mulder, Int J Environ Anal Chem , 47 (1992) 267 EEC Dnnkmg Water Gmdehne, 80/779/EEC, EEC No L229/11-29, EEC, Brussels, 30th August, 1980 M W F Nlelen, R W Frel and U ATh Brmkman, m R W Frel and K Zech (Eds ), Selectlve Sample Handling and Detection m Liquid Chromatography, Part A, Elseveer, Amsterdam, 1988, Ch 1 E R Brouwer, H Lmgeman and U ATh Brmkman, Chromatographla, 29 (1990) 415 E R Brouwer, D J van Iperen, I Llska, H Lmgeman and UkTh Bnnkman, Int J Environ Anal Chem, 47 (1992) 257 R Reupert and E Ploger, HPLC Determination of 29 Controlled Herbicides m Water Supplies, Hewlett-Packard Application Note, Hewlett-Packard, Avondale, PA, 1990 R Reupert, R Zube and E Ploger, GIT Fachz Lab, 11 (1991) 106 10 V Coquart and M -C Henmon, J Chromatogr ,585 (1991) 67 11 R W Frel and U A Th Brmkman, Trends Anal Chem , l-2 (1981) 45 12 R B Geerdmk, A M B C Graumans and J Vrveen, Internal Report Determmatlon of Phenoxyacld Herblcldes In Water, Institute for Inland Water Management and Waste Water Treatment, Lelystad, Netherlands, 1990