Determination of diphenyl-ether herbicides and metabolites in natural waters using hplc

Analytica Chimica Acta414(2000)79–94Determination of diphenyl-ether herbicides and metabolites in natural waters using high-performance liquid chromatography withdiode array tandem mass spectrometric detection Aldo Laganàa,∗,Giovanna Fago a,Laura Fasciani a,Alessandra Marino b,Matteo Mosso aa Department of Chemistry,‘La Sapienza’University,Piazzale Aldo Moro5,00185Rome,Italyb ISPELS/DIPIA,Via Urbana167,00184Rome,ItalyReceived27September1999;received in revised form11February2000;accepted23February2000AbstractA method for the identification and quantification of neutral diphenyl-ether(DPhE)(aclonifen,bifenox,fluoroglycofen, lactofen,oxyfluorfen)and acid metabolites(acifluorfen,bifenox acid,fomesafen)at a low nanogram per liter concentration level in natural water is presented.These herbicides and their metabolites were determined in drinking water,groundwater, and river water.The analytes were isolated from water samples by solid-phase extraction(SPE)on Carbograph-1cartridges and analyzed with reversed phase high-performance liquid chromatography(RP-HPLC)using UV detection at a wavelength of290nm.The isolation procedure separated the acid DPhE from neutral DPhE during the elution from Carbograph-1 cartridge using dichloromethane:methanol(80:20,v/v)for neutral compounds and dichloromethane:methanol(80:20,v/v) acidified with25mmol/l of formic acid for the anionic metabolites.Quantification was performed by generation of an external calibration curve.High recoveries(>85%)of extraction were obtained for all the compounds.The method detection limit ranged from4.1to8.7ng/l for drinking water,from5.1to9.5ng/l for groundwater and from17.5to36.2ng/l for river water. This method involves confirmatory analysis by tandem mass spectrometry(MS-MS)in selected reaction monitoring(SRM) mode.Conditions for MS-MS detection of characteristic daughter ions formed by collision-induced dissociation of the parent ion are described.Thefinal samples were analyzed by HPLC–MS-MS utilizing a heat-assisted electrospray interface(turbo ion spray)(TISP)for acid DPhE and a heated nebulizer(HN)interface for neutral DPhE.Quantification was performed by generation of an internal calibration curve.Excellent method precision was demonstrated with a relative standard deviation of less than18%for all analytes at all concentration levels.Application of the method for detecting DPhE herbicide residues in a real-world groundwater samples was demonstrated.©2000Elsevier Science B.V.All rights reserved.Keywords:Diphenyl-ether herbicide;Natural water;Tandem mass spectrometry1.IntroductionMolecules that inhibit protoporphyrinogen oxi-dase(Protox)have been among the most frequently∗Corresponding author.Fax:+39-6-446-2960.E-mail address:lagana@uniroma1.it(gan`a)patented class of herbicides over the past decade. Commercial Protox inhibitors can be classified in a major chemical group,the p-nitrodiphenyl-ethers, commercially known as the diphenyl-ethers(DPhE). DPhE are selective herbicides used for pre-and post-emergence residual weed control of a wide spectrum of annual broadleaf weeds and grasses in0003-2670/00/$–see front matter©2000Elsevier Science B.V.All rights reserved. PII:S0003-2670(00)00813-880 gan`a et al./Analytica Chimica Acta 414(2000)79–94soybeans,peanuts,and other large seeded legumes.This class of herbicides is mainly composed of es-ters but few compounds are acids or have an acidic behavior,with p K a comprised between 2.7and 3.8.Structure–activity studies have shown that the high-est biological activity of DPhE is generated by the sub-stituents 2-Cl,4-CF 3,4 -NO 2and various structures in the 3 -position (Fig.1)[1].As a consequence of applying pesticides,different environmental areas,as well as ground and surface water,are polluted with DPhE.When present,very low amounts of these herbicides in environmental waters are then to be expected.The primary objec-tive in pesticide residue analysis of environmental waters is to increase the method sensitivity,in compli-ance with the demands for drinking water quality.A second important aspect of performing trace residue is to assure a high degree of confidence in the identifica-tion of the compounds,to avoid false ing tandem mass spectrometric detection (MS-MS),frag-mentation of the initially formed deprotonated molec-ular ions is achieved by collision-induceddissociation Fig.1.Structures and common names of the investigated diphenyl-ether herbicides.in the collision cell between the first and second quadrupole.In the analysis of pesticides,the processes involved in pesticide degradation should be considered,since these processes play an important role in determin-ing the parent compound’s persistence and long-term toxicological effects.Although many commonly used pesticides have either a maximum contaminant level (MCL)or a health advisory level set for drinking water,only few of the corresponding metabolites have had such levels determined.Studies of herbicide metabolites are critical to understand the fate and transport of these compounds in soil.Toxicological studies of metabolites are underway to assess whether their presence is important to the total pesticide bur-den of surface water and groundwater.Continuous development of analytical methods for the analysis of metabolites is important for water quality studies [2,3].There are two main metabolites that arise from the degradation of the DPhE herbicides studied,bifenox acid from the hydrolysis of bifenox and acifluorfengan`a et al./Analytica Chimica Acta414(2000)79–9481from the degradation of lactofen andfluoroglycofen. The conversion of the ester to its acid is due to the metabolic activity of soil bacteria and fungi. Graphitized carbon black(GCB),commercially re-ferred to as Envirocarb or Carbograph-1,have shown to be valuable adsorbent materials for solid-phase ex-traction(SPE)of a variety of pesticides in waters.The unique property of the GCB adsorbent is that isolation of acidic compounds from neutral DPhE can be easily achieved by differential elution[3–7].When analyzing acidic compounds in complex aqueous environmental matrices,this class fractionation offers the advantages of making the analytical method more selective.The presence of active centers bearing a positive charge enables GCBs to behave as both nonspecific and an-ion exchange sorbents.It follows that anionic organic compounds are specifically adsorbed on GCB surface via electrostatics forces,and they can be desorbed only by adding a displacing agent to an organic solution. Recently,trifluoroacetic acid[4],acetic acid[5],and formic acid[6,7]have been used as displacing agents. There are relatively few methods for determining DPhE herbicides in environmental samples,the ma-jority of the analytical methods published to date re-port the determination of the lonely acifluorfen[8–10], oxyfluorfen[11,12],and bifenox[13,14],which were thefirst commercially available DPhE herbicides.To date,no analytical procedure is available for the simul-taneous determination of both neutral and acid DPhE herbicides in water samples.In this report,a new analytical method combining off-line SPE with LC–UV-DAD analysis,and with a confirmatory procedure using SRM–LC–MS-MS has been developed in environmental matrices.The objectives of this study were as follows:(i)to extract and fractionate thefive neutral and three acidic DPhE herbicides(included the two major metabolites) from natural waters by means of SPE with a GCB cartridge,(ii)to test the possibility of employing a LC method based on conventional instrumentation,such as LC with UV detection-DAD,(iii)to evaluate an LC–MS-MS,with different interfaces,confirmatory method as an independent mean to analyze the target herbicides at ng/l level to obtain the highest degree of confidence in positive results,and(iv)to evaluate the methods reliability to several groundwater samples to show the relative importance of this compounds in the aquatic environment.2.Experimental2.1.Reagent and chemicalsThe DPhE herbicides:acifluorfen(5-(2-chloro-␣,␣,␣-trifluoro-p-tolyloxy)-2-nitrobenzoic acid);aclonifen (2-chloro-6-nitro-3-phenoxyaniline);bifenox(methyl 5-(2,4-dichlorophenoxy)-2-nitrobenzoate);bifenox acid(5-(2,4-dichlorophenoxy)-2-nitrobenzoic acid);fluoroglycofen(O-[5-(2-chloro-␣,␣,␣-trifluoro-p-toly-loxy)-2-nitrobenzoyl]glycolic acid);lactofen(ethyl O-[5-(2-chloro-␣,␣,␣-trifluoro-p-tolyloxy)-2-nitrobe-nzoyl]-dl-lactate;oxyfluorfen(2-chloro-␣,␣,␣-triflu-oro-p-tolyl3-ethoxy-4-nitrophenyl ether);fomesafen (5-(2-chloro-␣,␣,␣-trifluoro-p-tolyloxy)-N-methylsul-fonyl-2-nitrobenzamide)were purchased from LabSe-rvice Analytica(Bologna,Italy),and4-benzoylbenzoic acid and2,2 -dinitrobiphenyl from Fluka(Buchs,SG, Switzerland).For HPLC,distilled water was purified by passing it through a Milli-Q RG apparatus(Milli-pore,Bedford,MA,USA).Acetonitrile and methanol of LC gradient grade were from Carlo Erba(Milan, Italy).Formic acid was purchased from Merck(Darm-stadt,Germany).All other solvents were reagent grade (Carlo Erba)and were used as received.Carbograph-1 was supplied by LARA(Rome,Italy).The parti-cle size range was37–150␮m.The Carbograph-1 extraction cartridges were prepared byfilling large diameter(6cm×1.3cm i.d.)syringe-like polypropy-lene tubes(Supelco,Bellefonte,PA,USA)with0.5g of adsorbing material.Polyethylene frits were placed above and below the adsorbent bed.Before process-ing the sample,the cartridge was washed with7ml of dichloromethane:methanol(80:20,v/v)acidified with formic acid25mmol/l,5ml of methanol,20ml of water acidified with hydrochloric acid10mmol/l, respectively.2.2.Standard preparationThe primary standard solutions at a concentration of1mg/ml were prepared by dissolving10mg of each DPhE in10ml of CH3CN.The internal standards were also prepared by dissolving10ml of solid in10ml of CH3CN.Appropriate volumes of these solutions were used to prepare the working standard solutions.All standards were prepared once a month and stored at 4◦C.82 gan`a et al./Analytica Chimica Acta414(2000)79–94Table1Characteristics of water used in fortificationType pH TOC a(mg/l)Hardness b(mg/l) Drinking water 6.6 1.029 Groundwater 6.1 1.226River water 6.81071a Total organic content,determined by US Environmental Pro-tection Agency method415.1(combustion infrared).b Determined by American Public Health Association(APHA) method2340B(sum of Mg2+and Ca2+determination by ICP).2.3.Source and sample collectionThree types of water were used in this report.Chem-ical characterization of water samples concerning pH, dissolved C(TOC)and dissolved Ca2+and Mg2+ are reported in Table1.Ground and river water sam-ples were collected near Rome,Italy.In order to take river(Tiber)and groundwater samples a portable,au-tomated sampler(4900Priority Contaminant Sampler, Manning Products,TX)was used.The river water was filtered through0.45␮m pore size Whatman GF/C glassfiber-pads(Maidstone,UK).Until analysis the filtered samples were stored in1l dark glass bottles at 4◦C.Prior to the analysis,the sample was allowed to reach room temperature.The drinking water samples were collected from the tap in the laboratory.To eliminate sediments and gas pockets in the pipes,this water was collected after flushing for about10min.The tap water was degassed with helium before fortification to eliminate chlorine, which can react with and degrade some of the com-pounds.For recovery studies,water samples were spiked with an appropriate working standard solution,then shaken vigorously and set aside for2h.2.4.Solid-phase extraction procedureWater samples(1l of ground and drinking water,0.5l of river water)were artificially contaminated witha suitable volume of the composite working standard solution.A Carbograph-1cartridge wasfitted into a side-armfilteringflask.Liquids were forced through the extraction device under vacuum from a water pump.Water was forced through the cartridge atflow rates of30–50ml/min.When samples were passing through the cartridge,the vacuum was reduced to a minimum.The cartridge was then washed with 7ml of water.Most of the water remaining in the cartridge was expelled under vacuum for ca.5min and the residual water content further decreased by slowly passing1ml of CH3OH through the cartridge. The neutral DPhEs were then eluted with8ml of a CH2Cl2:CH3OH(80:20,v/v)solution. Thereafter,a suitably drilled cylindrical teflon pis-ton with one conical indented base and a Luer tip was forced into the cartridge until it reached the up-per frit.The cartridge was turned upside-down.The acid DPhEs were back-eluted by passing through the cartridge10ml of CH2Cl2:CH3OH(80:20,v/v)acid-ified with formic acid,25mmol/l,at aflow-rate of ca. 5ml/min.Before the evaporation step40␮l of NH4OH (30%):CH3OH(50:50,v/v)were added.Solvent evaporation was carried out in a water bath at40◦C,under a gentle stream of nitrogen. When both neutral and acidic extracts reached vol-umes of about200␮l,the walls of the vials were washed with100␮l of CH3CN:H2O(50:50,v/v)and 100␮l of HCOOH-acidified methanol(25mmol/l), respectively.The extract containing neutral DPhE was further concentrated to ca.200␮l and thefinal volume was carefully measured.The fraction con-taining the acid DPhE was taken to dryness.The residue was reconstituted with250␮l of a solution CH3CN:CH3OH(50:50,v/v)/H2O(60:40,v/v)acid-ified with100mmol/l HCOOH.For both fractions, 100␮l offinal extract was injected onto the LC column.2.5.Instrumental conditions2.5.1.Liquid chromatographic analysisLiquid chromatography was carried out using a Perkin-Elmer series400liquid chromatography (Perkin-Elmer,Norwalk,CT,USA)equipped with a Rheodyne7125injector(Supelco)with a100␮l loop.The analytes were chromatographed on a 25cm×4.6mm i.d.columnfilled with5␮m(aver-age particle size)C-18packing Alltima(Alltech, Deerfield,IL,USA).The precolumn Supelguard, 2cm×4.6mm i.d.,filled with5␮m C-18packing was supplied by Supelco.Absorbance was measured con-tinuously in the range of210–350nm by diode array detection(Diode Array Detector235C,Perkin-Elmer)gan`a et al./Analytica Chimica Acta414(2000)79–9483following solute separation on the HPLC column. The peak spectra were quantified at a wavelength of 290nm.Besides the retention times,UV peak spec-tra were compared with a spectra library assembled from spectra of spiked and extracted distilled water. Data were acquired by using a LCI-100Laboratory Computing Integrator Perkin–Elmer.2.5.1.1.Acid fraction analysis.To fractionate the acidic DPhEs,the mobile phase was phase A, CH3CN:CH3OH(50:50,v/v)acidified with HCOOH 20mmol/l and phase B,H2O acidified with HCOOH 50mmol/l.Gradient elution was performed by lin-early increasing the percentage of the organic modi-fier from55to70%in30min,followed by5min at 100%.Gases in the solvents were removed by sparg-ing with helium.Theflow rate of the mobile phase was1ml/min.2.5.1.2.Neutral fraction analysis.To fractionate the neutral DPhEs,the mobile phase was phase A,CH3CN and phase B,H2O.Gradient elution was performed by linearly increasing the percentage of the organic mod-ifier from62to75%in30min,followed by5min at 100%.Gases in the solvents were removed by purging with helium.Theflow rate of the mobile phase was 1ml/min.2.5.2.Mass spectrometric analysisA PE Sciex(Concord,ON,Canada)triple quadrupole mass spectrometer API365interfaced via a Sciex turbo ion spray(TISP)or heated nebulizer (HN)probe with the HPLC above described was used for all LC–MS-MS analysis.MassChrom1.0soft-ware,from PE Sciex,was used for data acquisition and presentation on a Power MacIntosh G3.2.5.2.1.TISP interface.LC–TISP–MS-MS analysis was carried out utilizing the chromatographic con-ditions above reported for the acid fraction.Here, 200␮l/min(20%)of the column effluent was diverted into the TISP source.Data acquisition,performed in the negative ion mode(NI),was divided into two periods,one for the internal standard and the second for the three target compounds,where individual ion optics and MS-MS tuning parameters were optimized to provide maximum sensitivity for the individual DPhEs.The temperature of the TISP heater was set at 350◦C and the dwell time was400ms.The mass spec-trometer was programmed to admit the deprotonated molecules[M−H]−at m/z225(IS:4-benzoylbenzoic acid),437(fomesafen),326(bifenox acid),and360 (acifluorfen)via thefirst quadrupolefilter(Q1),with collision-induced fragmentation at Q2,and monitor-ing the product ions via Q3at m/z181,286,282,and 316,respectively.2.5.2.2.HN interface.LC–HN–MS-MS analysis was carried out utilizing the chromatographic condi-tions for the neutral fraction reported before.The HN probe was maintained at400◦C,and the gas-phase chemical ionization was effected by a corona dis-charge needle(−3␮A)using negative ion atmo-spheric pressure chemical ionization(APCI).Data acquisition was divided intofive periods and the dwell time was400ms.The mass spectrometer was pro-grammed to monitor the following selective reactions: m/z244→153(IS:2,2 -dinitrobiphenyl),263→156 (aclonifen),340→236(bifenox),446→269(fluoro-glycofen),360→285(oxyfluorfen),and460→269 (lactofen).3.Results and discussion3.1.Solid-phase extraction and fractionation of neutral and acid DPhEThe SPE of the neutral fraction with Carbograph cartridge was carried out according to a previous work [15],an improved procedure was developed for the acidic analytes.In order to achieve efficient and se-lective desorption of the acidic DPhE herbicides from interfering natural occurring acidic compounds,prob-ably fulvic acids(normally present in aqueous sam-ples),effect of acidification of the eluent phase with acetic or formic acid at different concentrations has been evaluated.This investigation was conducted by using ground water samples fortified with the analytes at the level of500ng/l,the water samples were an-alyzed in triplicate.After washing the cartridge with the solvent mixture designed to elute neutral DPhE, stepwise desorption of acid DPhE was performed by passing different eluent phases acidified with acetic or formic acid.The concentration of acid in the elu-ent phase was acetic acid between10and100mmol/l,84 gan`a et al./Analytica Chimica Acta 414(2000)79–94formic acid between 5and 50mmol/l.With aceticacid as modifier,unsatisfactory low mean recoveries of acidic analytes were obtained.Under the same con-ditions,but with formic acid as modifier,quantitative recoveries of the three acid DFE were obtained with a concentration of 25or 50mmol/l.The eluent phase acidified with formic acid,25mmol/l was preferred as,with a higher concentration,an increase in the peaks due to the fulvic acids was observed.It is important to note that partial neutralization of the eluent phase containing the acidic DPhEs,before the evaporation step,leads to quantitative recoveries.Without NH 4OH,especially at low concentrations,a low recovery of fomesafen was achieved.This phe-nomenon may find an explanation in the fact that fomesafen is more readily esterified by methanol in an acidic environment.To quantify both neutral and acidic DPhE in spiked and unspiked aqueous samples,the external standard quantification procedure was followed.Standard so-lutions were prepared at five levels using each of two working standard solutions,one containing neu-tral DPhE and the other containing acidic ones,in the respective solvent mixture used to elute the two classes of analytes from the extraction cartridge.Thereafter,the rest of the procedure described before was fol-lowed.For each analyte,calibration curves were con-structed by plotting peak areas from analyte against the amounts injected from the standard solutions into the LC column.The response of the DAD detector was linear up to injected amounts of each neutral and acidic DPhE of 180–200ng.Table 2Recoveries a ,precision and method detection limit (MDLs)in different natural water samples at low concentration (LC–DAD analysis)CompoundDrinking water (n =6)(1l)Groundwater (n =6)(1l)River water (n =6)(0.5l)Recovery (%)RSD (%)MDL b (ng/l)Recovery (%)RSD (%)MDL a (ng/l)Recovery (%)RSD (%)MDL a (ng/l)Fomesafen 95 2.9 4.197 3.5 5.194 3.117.5Bifenox acid 97 3.7 5.493 4.6 6.4102 4.930.0Acifluorfen 98 4.97.296 3.6 5.2100 4.325.8Aclonifen 96 4.2 6.0102 4.0 6.197 3.419.8Bifenox94 5.07.096 3.8 5.595 3.922.2Fluorglycofen 91 6.48.793 6.89.590 6.736.2Oxyfluorfen 91 4.7 6.495 4.4 6.396 5.531.7Lactofen935.98.2905.57.4856.231.6a Spiked amounts:Drinking water,50ng/ml;Groundwater,50ng/ml;River water,200ng/ml.bMDL defined as three times the standard deviation calculated at the fortified level considered.3.2.Analytical parameter evaluation(a)Extraction recoveries,(b)precision,and (c)method detection limit (MDL)were evaluated by us-ing 1l fortified drinking water or ground water,and 0.5l of river (Tiber)water with two different spiking levels and analyzing each type of aqueous matrix.(a)For each type of aqueous matrix considered,an-alyte recoveries were determined by adding known values of working standard solution.The entire view of the analytical parameter evaluation for the sam-ples with a low level of DPhE herbicides is shown in Table 2.In drinking water and ground water spiked with 50–400ng/l the recoveries ranged from 90to 102%for low spike levels and from 93to 99%for high spike levels.In river water fortified with 200–2000ng/l recoveries were between 85and 102%for low spike levels and between 90and 104%for high spike levels.The recoveries obtained for all DPhE herbicides,ex-cept for lactofen in river water,were better than 90%.This demonstrates that no effect of irreversible adsorp-tion was produced by extraction device and that were unaffected by the nature of aqueous matrix in which the analytes were dissolved.For river water samples with very low spike levels (200ng/l),recoveries of some substances appeared to be reduced (ctofen 85%).Some effects of adsorption on suspended parti-cles were presumably responsible for this loss.(b)Repeatability precision values,expressed as the mean relative standard deviation (R.S.D.),for each compound was calculated from six independent ex-traction of DPhE pesticides from each type of aqueousgan`a et al./Analytica Chimica Acta414(2000)79–9485matrix.At the spike levels considered,the repeatabil-ity ranged from2.9to7.7%for all type of sample water,indicating a good performance of the method developed in this work.(c)For MDLs,the following spike levels were con-sidered:50ng/l for drinking and ground water and 200ng/l for river water.The listed MDLs are reported in Table2and it follows that this methods could largely satisfy the stringent requirements imposed by the Eu-ropean Community Directive[16]setting100ng/l as the maximum admissible concentration of an individ-ual pesticide and500ng/l for their sum in drinking water.In groundwater and river-water higher detec-tion limits were observed,due to the interference of matrix compounds.Typical LC/UV chromatograms obtained by DPhE-fortified groundwater and river water are shown in Fig.2.3.3.LC–MS-MS assay developmentMS and particularly MS-MS is the ideal technique, to avoid false positives.In a search of the‘ideal’in-terface for DPhE herbicides,the response of the eight selected compounds was studied by direct on column injection,the amount of each pesticide injected was 1␮g under full-scan conditions.The mass spectrum of each compound was recorded with two interfaces in both negative and positive ionization mode.TISP in negative acquisition mode for acidic DPhE and HN in negative acquisition mode for neutral DPhE were selected.3.4.Assay using TISP interfaceThe difference between the pneumatically assisted electrospray,or ion spray(ISP)[17],and TISP inter-face is the addition of a heated stream of dry nitrogen to the ionization chamber.The solvent evaporation process is accelerated without extreme heating of the analyte molecules.More details about differences between ISP,TISP,and HN interfaces are provided in Refs.[18,19].At the beginning the acidic DPhE herbicides were investigated either in TISP-PI or in TISP-NI mode. Comparing the S/N values of three selected herbicides in the two different modes it was observed that the sensitivity of bifenox acid was very low when operat-ing in the positive acquisition mode.All of the analytes form intense deprotonated molecular ions[M−H]−.The product ions related for each of the DPhE were chosen from an on-line full-scan LC–MS-MS experiment.In Fig.3the typical full-scan TISP MS-MS mass spectra and global schema for the possible fragments for fomesafen is presented.These results were ob-tained by direct injection of each pesticides at a concentration of10ng/␮l into a carrier stream of H2O:CH3CN:CH3OH(50:25:25,v/v)acidified with 25mM HCOOH,utilizing a cone voltage of the ori-fice of−5V for bifenox and fomesafen,−20V for acifluorfen,and−10V for4-benzoylbenzoic acid. Fomesafen exhibits a peak at m/z316for the loss of the fragment O=C=N–SO2CH3from the molecular ion[M−H]−.The subsequent loss of NO•gives rise to an intense peak at m/z286.This fragment may originate a peak at m/z222that can be ascribed to a styrenic derivative formed through a double H rear-rangement.Bifenox acid and acifluorfen show a very intense peak that may be ascribed to the fragment [M−H−CO2]−at m/z282and316,respectively.The loss of NO•gives rise to a not very intense peak at m/z 286for acifluorfen and at m/z252for bifenox acid. Fig.4shows the reconstructed selected ion current chromatograms of acid DPhE fortified river extract. The chromatograms were obtained after optimization of MS-MS conditions for each acid DPhE compound using one tuning period for acid herbicides and one for IS.3.5.Assay using HN interfaceThe response offive neutral DPhE herbicides was comparable in the HN-PI and in the HN-NI mode.The NI signal intensities of some neutral DPhE were ap-proximately two times lower than in PI.For aclonifen the sensitivity in PI was very low.This suggests that the method that has to be developed for all selected compounds must be based on the NI mode. Regarding the spectra of investigated herbicides showing relatively high signal intensities in the NI mode,the HN-NI spectra of DPhE showed[M−H]−as a pronounced peak with various fragments.The full-scan HN–MS-MS mass spectra for a neutral86 gan`a et al./Analytica Chimica Acta414(2000)79–94Fig.2.UV chromatograms of groundwater spiked at level of100ng/l and river water spiked at level of250ng/l with(a)three acid DPhE and(b)five neutral DPhE.Peaks numbers:1,fomesafen;2,bifenox acid;3,acifluorfen;4,aclonifen;5,bifenox;6,fluoroglycofen;7, oxyfluorfen;8,lactofen.gan`a et al./Analytica Chimica Acta414(2000)79–9487Fig.3.Representative full-scan(200–450amu)product ion spectra and proposed fragmentation ions from TISP–MS-MS analysis of fomesafen at a concentration of10ng/␮l into a carrier stream of H2O:CH3CN:CH3OH(50:25:25,v/v)acidified with25mM HCOOH.88 gan`a et al./Analytica Chimica Acta 414(2000)79–94Fig.4.Reconstructed ion current chromatograms of a NI TISP–MS-MS run of river water extract spiked with 4-benzoylbenzoic acid (I.S.),fomesafen,bifenox acid,and acifluorfen at 6ng/l,14ng/l,2.8ng/l,and 7ng/l,respectively.DPhE (lactofen)at concentration of 10ng/␮l in a solution of acetonitrile:water (70:30,v/v)and the proposed fragmentation is presented in Fig.5.The HN–MS-MS data on product ion spectra for neutral DPhE and IS obtained at different collision energies are listed in Table 3.Analyzing these data it can be noted that,except for oxyfluorfen which shows two significantly different product ions,all the other compounds exhibit more complex spectra.Lactofen shows a peak at m/z 344due to the loss of the CH 3–CO–CO 2–C 2H 5fragment from [M −H]−.This fragment shows a peak at m/z 316for the loss of CO,otherwise the loss of NO 2and then HCO shows a peak at m/z 269.The fluoroglycofen having a struc-ture similar to that of lactofen has the same fragments at m/z 316,285,269.Oxyfluorfen exhibits a peak at m/z 285coming from the ion [M −H −NO −OC 2H 5]−.Fragmentation of bifenox molecular ion [M −H]−gives rise to the peaks at m/z 293due to the loss of NO 2and m/z 252corresponding to [M −NO −CO 2CH 3]−.Then the peak at m/z 236can be matched to the [M −NO 2−CO 2CH 3]−fragment.Aclonifen shows a。