Anal. Chem.

2007, 79, 3794-3801

Glyphosate and AMPA Analysis in Sewage Sludge

by LC-ESI-MS/MS after FMOC Derivatization on
Strong Anion-Exchange Resin as Solid Support
Aline Ghanem, Philippe Bados, Lucien Kerhoas, Jacqueline Dubroca, and Jacques Einhorn*

INRA, Unité de Phytopharmacie et Médiateurs Chimiques, Route de Saint-Cyr, 78026 Versailles Cedex, France

An innovative analytical method has been developed for shown to occur at the microgram-per-liter level in wastewaters
the determination of glyphosate and aminomethylphos- with a significant contribution issuing from their urban use.4 They
phonic acid (AMPA), its major metabolite, in sewage are thus suspected to be transferred to sewage sludge during
sludge. This method involved an alkaline extraction from wastewater treatment. Glyphosate presents a lower acute toxicity
sludge and purification on a strong anion-exchange resin. than other herbicides but appears to induce endocrine effects.5
While the analytes remained fixed by ionic interactions, The ecological effects of glyphosate and its metabolite still need
an “on-solid support” derivatization by FMOC-Cl was to be documented.6,7 To assess the environmental risk of sludge
carried out. This versatile approach allowed a 10 min used as land amendment, an in-depth knowledge of its composition
reaction and simple elimination of the excess of reagent. is required before application. The presence of persistent pollut-
The resulting derivatives remained retained by ionic and ants such as heavy metals, PAHs, and PCBs is generally assumed,
hydrophobic interactions with the resin until being eluted and their levels are surveyed by European regulation (Directive
by a mixed NaCl water/acetonitrile, 70/30, v/v, solution. 86/278/CEE). Nevertheless, no data are available for pesticides,
After an appropriate dilution and adjustment of the pH, particularly for glyphosate, which is likely explained by the
the sample was concentrated on an Oasis HLB solid- difficulties of analysis of such contaminants in addition to the
phase cartridge. For quality analysis of traces in complex complexity of the matrix.
matrixes, LC-ESI-MS/MS in the multiple reaction moni- Highly polar and amphoteric, glyphosate often represents an
toring positive mode was used fulfilling the European analytical challenge. It has four pKa values, 0.7, 2.2, 5.9, and 10.6,8
Union requirements (Decision 2002/657/CE). To over- whereas three values, 0.9, 5.6, and 10.2, characterize AMPA.9 A
come the matrix effects, stable isotopically labeled stan- variety of methods have been developed for glyphosate and AMPA
dards were added to the sludge extracts as internal analysis in waters, soils, and sediments. The majority of these
standards and were thus derivatized during the procedure are tedious due to the numerous steps employed for purification
in parallel to the analytes. Mean recovery values were 70% and the derivatization often required for either GC10,11 or HPLC.
( 7% for glyphosate and 63% ( 3% for AMPA. Limits of Direct HPLC analysis is possible by ion chromatography with
detection (20 and 30 ppb dw) and limits of quantification electrochemical or mass spectrometry detection.12-14 However,
(35 and 50 ppb dw) for glyphosate and AMPA, respec-
tively, were sufficient to monitor samples taken from Ile- (3) Schweinsberg, F.; Abke, W.; Rieth, K.; Rohmann, U.; Zullei-Seibert, N.
de-France wastewater treatment plants where contami- Toxicol. Lett. 1999, 107, 201-205.
(4) Kolpin, D. W.; Thurman, E. M.; Lee, E. A.; Meyer, M. T.; Furlong, E. T.;
nation currently reached 0.1-3 ppm and 2-35 ppm dw Glassmeyer, S. T. Sci. Total Environ. 2006, 354, 191-197.
for glyphosate and AMPA, respectively. (5) Richard, S.; Moslemi, S.; Sipahutar, H.; Benachour, N.; Seralini, G. E.
Environ. Health Perspect. 2005, 113, 716-720.
(6) Cauble, K.; Wagner, R. S. Bull. Environ. Contam. Toxicol. 2005, 75, 429-
Glyphosate [N-(phosphonomethyl)glycine] is among the most 435.
widely used herbicides. Introduced by Monsanto in the early (7) Ghanem, A.; Dubroca, J.; Chaplain, V.; Mougin, C. Environ. Chem. Lett.
1970s, it is a nonselective, postemergence herbicide dedicated to 2006, 4, 63-67.
(8) Sprankle, P.; Meggit, W. F.; Penner, D. Weed Sci. 1975, 23, 229-234.
weed and vegetation control. Its degradation in the environment (9) Traas, T. P.; Smith, C. E. Environmental Risk Limits for Aminomethylphos-
mainly occurs under biological conditions yielding aminometh- phonic Acid (AMPA). Setting Integrated Environmental Quality Standards
ylphosphonic acid (AMPA) as the major metabolite. However, of Netherlands Public Health and Environment National Institute, Rijksinsti-
tuut voor Volksgezondheid en Miliev (RIVM): The Netherlands, 2003, 23,
AMPA can originate from the degradation of phosphonic acids in 10.
detergents.1 Glyphosate and AMPA have been frequently detected (10) Deyrup, C. L.; Chang, S. M.; Weintraub, R. A.; Moye, H. A. J. Agric. Food
in natural waters.2,3 The presence of both contaminants was also Chem. 1985, 33, 944-947.
(11) Börjesson, E.; Tortensson, L. J. Chromatogr., A 2000, 886, 207-216.
* To whom correspondence should be addressed. E-mail: einhorn@ (12) Zhu, Y.; Zhang, F.; Tong, C.; Liu, W. J. Chromatogr., A 1999, 850, 297-
[Link]. Phone: +33-1-30-83-31-20. Fax: +33-1-30-83-31-19. 301.
(1) Skark, C.; Zullei-Seibert, N.; Schottler, U.; Schlett, C. Int. J. Environ. Anal. (13) Sato, K.; Jin, J.-Y.; Takeuchi, T.; Miwa, T.; Suenami, K.; Takekoshi, Y.; Kanno,
Chem. 1998, 70, 93-104. S. J. Chromatogr., A 2001, 919, 313-320.
(2) Smith, N. J.; Martin, R. C.; St Croix, R. G. Bull. Environ. Contam. Toxicol. (14) Bauer, K.-H.; Knepper, T. P.; Maes, A.; Schatz, V.; Voihsel, M. J. Chromatogr.,
1996, 57, 759-765. A 1999, 837, 117-128.

3794 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007 10.1021/ac062195k CCC: $37.00 © 2007 American Chemical Society
Published on Web 04/06/2007
sensitivity and selectivity are usually reached via either pre- Saulx-les-Chartreux, France. (2-13C, 99%; 15N, 98%)-glyphosate and
column15-18 or postcolumn derivatization.19,20 Precolumn deriva- (13C, 99%; 15N, 98%; methylene-d2, 98%)-AMPA in water solutions
tization of glyphosate and AMPA with 9-fluorenylmethyl chloro- at 100 µg/mL were provided from LGC Promochem, Teddington,
formate (FMOC-Cl) is advantageous since it improves the U.K. Standard glyphosate-FMOC and AMPA-FMOC (99.5% and
chromatographic separation of the analytes and subsequently their 96% purity, respectively) were obtained from Dr. Ehrenstorfer.
detection either by fluorimetry15 or mass spectrometry.17,21 Solid- FMOC-Cl (98% purity) was obtained from Sigma-Aldrich, Stein-
phase extraction methods (SPE) using various kinds of phases heim, Germany. Potassium dihydrogen phosphate, sodium hy-
are also reported to allow concentration and/or purification of droxide, sodium chloride, and acids were obtained from VWR,
glyphosate prior to its analysis in waters17,20,22 or soils.21 Fontenay-sous-Bois, France. Dimethyldichloride silane (DMDCS)
In this study, it is proposed to perform the FMOC derivatization was purchased from Sigma-Aldrich.
of glyphosate and AMPA after fixation of the analytes on a solid Oasis HLB [poly(divinylbenzene-co-N-vinylpyrrolidone] reversed-
support conveniently chosen in the analytical strategy. Subse- phase SPE cartridges (120 mg, 810 m2/g) were purchased from
quently, the derivatized analytes will be detected by using LC- Waters, Milford, MA. LiChrolut EN [poly(styrene-co-divinylben-
ESI-MS/MS in the multiple reaction monitoring (MRM) mode. zene] SPE cartridges (200 mg, 1200 m2/g) and Bond-Elut
By this approach, it was intended to gain the following advan- octadecyl C18 SPE cartridges (200 mg, 500 m2/g) were obtained
tages: (i) minimize the number of steps for sample purification from Varian, Harbor City, CA. AG-1 × 8 [poly(styrene-co-divinyl-
and concentration, (ii) minimize the volume of “solution” to be benzene]-based trimethylammonium anion-exchange resin (100-
derivatized, (iii) reduce sample handling during elimination of 200 mesh, chloride form) was purchased from Bio-Rad Labora-
excess reagent and, finally, (iv) preserve the possibility of tories, Hercules, CA. The anion-exchange resin was converted into
automation. the HCO3- and OH- forms using 4 bed volumes of 1 N NaHCO3
The on-solid support derivatization will be the key step to and 20 bed volumes of 1 N NaOH solutions, respectively. It was
satisfy all these criteria. Because sludge samples are highly then rinsed with water until pH < 9. Sewage sludge samples were
complex and “dirty”, the first purification step will have to be collected from Plaisir and Elancourt wastewater treatment plants
carried out on low-pressure or atmospheric resin columns (not (WWTPs), Ile-de-France region, in the dehydrated state (20% dw).
cartridges) using diluted extracts. In the case of a conventional Analytical Procedure. For the radioactive tests used during
approach, recovery of the free analytes should be possible only development, an aqueous solution of 14C-glyphosate (9.25 kBq)
through the use of large volumes of eluent. This condition would was spiked on the sludge, and after 30 min, the sample was kept
then oblige one to carry out the liquid-phase derivatization on an at 4 °C for 1 night. Radioactivity was monitored on a Wallace
aliquot to complete the reaction in a reasonable time (maximum model 1410 (Turku, Finland) liquid scintillation counter. To
of overnight), thus compromising the sensitivity of the overall
evaluate effects of parameters, analyses were done by LC-UV diode
method. In our procedure, this inconvenience is circumvented,
array using an Aquasil C18 column (150 mm × 2 mm i.d., 3 µm,
further guarantying an easy elimination of the reagent in excess.
Thermo-Electron, Les Ulis, France) and linear gradient reversed
This is the first method that determines glyphosate and AMPA
chromatography with 5 mM ammonium acetate (solvent A) and
residues in sewage sludge.
acetonitrile (solvent B).
Extraction. A 5 g portion of fresh sludge previously homog-
EXPERIMENTAL SECTION
enized was extracted with 50 mL of 0.1 N NaOH using a horizontal
To prevent the high adsorption of glyphosate and AMPA, glass
shaker (300 shakes/min) for 1 h and then centrifuged 5 min at
materials were silanized using 5% dimethyldichloride silane
2500g. The supernatant was filtered on 10 g of hyflosupercel silica,
(DMDCS) in hexane. After 10 min of contact, they were rinsed
twice with hexane then with methanol before being dried. and 20 mL of 0.1 N NaOH was used for rinsing. Extraction was
Chemicals and Reagents. All chemicals were of analytical operated twice, and the supernatants were combined. A 10 mL
grade, and 0.45 µm filtered ultrapure water (resistivity of 18.2 MΩ/ aliquot was adjusted to pH 9 with 1 N HCl and was spiked with
cm) was used for solutions. Glyphosate and AMPA (98% and 99% the internal standards (0.1 and 0.5 µg of the stable isotopically
purity, respectively) were purchased from Dr. Ehrenstorfer labeled glyphosate and AMPA, respectively).
(Augsburg, Germany). Standard solutions containing 150 µg/mL Fixation and Derivatization on Strong Anion-Exchange (SAX)
of each analyte were prepared in water and stored at 4 °C. Resin. A 300 mm × 10 mm column was packed with 1 g of AG-1
Working standard solutions were prepared by further dilution. × 8 SAX resin using pH 9 ultrapure water. The 10 mL aliquot
N-(Phosphono[14C]methyl)glycine (glyphosate; specific activity sample was subsequently percolated through the resin, and 5 mL
2084 MBq/mM, purity >99%) was purchased from Dislab’System, of ultrapure water was applied onto the resin for rinsing. For
derivatization, the resin was first conditioned with 2 mL of mixture
(15) Glass, R. L. J. Agric. Food Chem. 1983, 31, 280-282. A constituted of 25 mM borate buffer pH 9.2/acetonitrile (30:70
(16) Miles, C. J.; Moye, H. A. J. Agric. Food Chem. 1988, 36, 458-461.
(17) Vreeken, R. J.; Speksnijder, P.; Bodeldijk-Pastorova, I.; Noij, Th. H. M. J.
v/v), and then 2 mL of the same mixture containing 20 mg of
Chromatogr., A 1998, 794, 187-199. FMOC-Cl was added. The reagent was left to react with the
(18) Nedelkoska, T. V.; Low, G. K.-C. Anal. Chim. Acta 2004, 511, 145-153. trapped analytes during 10 min. The resin was then rinsed and
(19) EPA Method 547; EPA/600/4-90/020; U.S. Environmental Protection
Agency: Cincinnati, OH, 1990, 63-80.
reconditioned with 15 mL of mixture B containing 25 mM borate
(20) Mallat, E.; Barcelò, D. J. Chromatogr., A 1998, 823, 129-136. buffer pH 9.2/acetonitrile (50:50 v/v). Another 20 mg of FMOC-
(21) Ibanez, M.; Pozo, O. J.; Sancho, J. V.; Lopez, F. J.; Hernandez, F. J. Cl in 2 mL of mixture A was loaded to complete derivatization.
Chromatogr., A 2005, 1081, 145-155.
(22) Patsias, J.; Papadopoulou, A.; Papadopoulou-Mourkidou, E. J. Chromatogr., Excess reagent and interferences (FMOC-OH, etc.) were elimi-
A 2001, 932, 83-90. nated by rinsing the resin, first with 15 mL of mixture B and
Analytical Chemistry, Vol. 79, No. 10, May 15, 2007 3795
second with 10 mL of water. The target derivatives were eluted elution of the FMOC analytes and SPE concentration would finally
with 30 mL of 1 N NaCl/acetonitrile (70:30 v/v) at pH 5, and yield a methanolic (or acetonitrile) sample solution to subject to
recovered into 270 mL of ultrapure water at pH 3. LC-ESI-MS/MS analysis. Other alternatives were tested prelimi-
SPE Concentration of the FMOC Analytes. Prior to the precon- narily but were not successful. They included the use of (i) a
centration step, the pH of the previous solution was readjusted to cation-exchange column instead of the anionic one to fix the
3 with 1 N HCl. Cartridges of 3 mL volume packed with 120 mg analytes or (ii) a hydrophobic phase to trap the reagent and allow
Oasis HLB were used on an Autotrace SPE workstation (Tekmar, its reaction with a percolating solution of the analytes. However,
Cincinnati, OH). They were conditioned by flushing 5 mL of the strategy based on the use of an anion-exchange support to
MeOH followed by 5 mL of water and 5 mL of phosphate buffer undergo the derivatization with FMOC-Cl was successfully
at pH 3. The 300 mL sample solutions obtained above were accomplished. The development and optimization of this strategy
percolated at 5 mL/min. The cartridges were then rinsed with 5 are described in Figure 1.
mL of water and dried for 30 min under a nitrogen flow. Elution Extraction. Analyte extraction was attempted in alkaline
was performed with two fractions of 3.5 mL of MeOH, which were medium (0.1 N NaOH) following the example of soil studies.16
then combined and concentrated to 0.5 mL under N2 before The efficiency of this medium was tested using sludge samples
analysis. fortified with radiolabeled [14C] and unlabeled glyphosate (9.25
Liquid Chromatography Tandem Mass Spectrometry (LC-MS/ KBq, 5 ppm fresh weight). High recoveries (>80%) were obtained
MS). A Waters HPLC pump Alliance 2695 (Milford, MA) equipped for the extraction of [14C]-glyphosate from sludges after 0-72 h
with an automatic injector model 717 was used for separation on of contamination. The alkaline aqueous medium was thus found
a Discovery HS C18 column (50 mm × 2.0 mm, 5 µm) (Supelco, strongly desorptive for glyphosate even from highly organic
Bellefonte, PA), with a mobile phase consisting of 5 mM am- matrixes. No further improvement was obtained using a more
monium acetate buffer and acetonitrile at a flow rate of 0.15 mL/ concentrated alkaline solution or adding sonication.
min at 24 °C. A linear gradient from 20% to 100% of acetonitrile Interactions between the SAX Resin and the FMOC
was used within 12 min, and the final composition was maintained Analytes. According to our strategy, it was essential to ensure
4 min. Samples of 5 µL volume were used for injection. The LC the fixation of the analytes on the SAX resin in order to purify
system was connected to a Quattro LC triple-quadrupole mass the sample and subsequently perform the FMOC-Cl derivatiza-
spectrometer (Micromass, Manchester, U.K.) with an electrospray tion. Since counterions with the lowest selectivity for anion-
as interface. Data acquisition and processing were performed by exchange chromatography are most readily exchanged by target
MassLynx NT 4.0. MRM was used with dwell times of 0.3 s/scan. anions, we converted the commercial chloride resin into the
Source voltages were as follows: capillary 3.2 kV, extractor 2 V. hydroxide or hydrogen carbonate form (the selectivity order being
The source block and desolvation gas were heated at 120 and
as follows: OH- < HCO3- < Cl-). The trapping efficiency for
400 °C, respectively. Nitrogen was used as the nebulization and
our analytes was examined by means of radioactivity experiments
desolvation gas (75 and 450 L/h, respectively). For MS/MS,
on both types of resin. A 10 mL aqueous solution (pH 10-13)
collisional-induced dissociation (CID) was performed with 2.5 ×
containing [14C]-glyphosate was applied onto a glass column filled
10-3 mbar of argon and 10-50 eV collision energy.
with 1 g of SAX support. No breakthrough was detected in the
Breakthrough Study. The performance of SPE for concentra-
effluents.
tion of the FMOC analytes present in aqueous solutions was
Experiments were also carried out to check if FMOC deriva-
evaluated by means of breakthrough curves. The latter were
tives could be fixed on the solid support under the alkaline
established using a home-made automated device23 that monitored
conditions and how they could be eluted further. The fixation of
the 263 nm UV signal after percolation of 0.8 mg/L glyphosate-
the FMOC products on the SAX resin was studied using unlabeled
FMOC solutions through precolumns filled with the solid phase
standards. Trapping was found efficient (0% breakthrough) on both
to be tested.
OH- and HCO3- forms of resin, with the sample pH ranging from
9 to 13. In contrast, some breakthrough was observed when using
RESULTS AND DISCUSSION
Cl- (>23% of glyphosate-FMOC and >43% of AMPA-FMOC).
The main idea of this work was to include an “on-solid support
Aqueous solutions of NaCl are often used for the elution of
derivatization” step in the procedure to facilitate the general
linked ionic analytes because of the high affinity of Cl- ions for
treatment of the sample, particularly during the reaction itself,
this type of resin.15 However, this eluent (1-2 N solutions) was
and the elimination of the excess of reagent. The proposed
not strong enough since it yielded only 7% and 13% recovery of
strategy consisted in the following sequence. After extraction in
glyphosate-FMOC from OH- and HCO3- resins, respectively. By
an aqueous medium and centrifugation, the obtained solution
contrast, the introduction of an organic modifier like acetonitrile
would be loaded on a SAX resin column. This should allow the
in the saline medium strongly improved the elution efficiency,
trapping of glyphosate and its metabolite on the solid support
indicating that both ionic and hydrophobic interactions had to be
through ionic interactions. It was then hoped that the fixed
involved. With 10-30% of acetonitrile, recoveries reached from
analytes could be derivatized by contact with FMOC-Cl under
30% to 51% (OH-) and to 75% (HCO3-). However, the difference
conditions of solvent and time to be specified. Such in situ
observed between both phases led us to suspect a possible
derivatization should make it possible to easily eliminate the
degradation of the analyte during the elution process. This
excess reagent by a convenient rinsing of the column. Subsequent
question will be addressed next. Higher percentages of acetonitrile
(23) Nélieu, S.; Perreau, F.; Guichon, R.; Seguin, F.; Bry, C.; Einhorn, J. Anal. were not tested to limit the extent of dilution necessary before
Bioanal. Chem. 2005, 382, 108-114. operating the SPE concentration. The influence of the pH on
3796 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007
Figure 1. (a) Analytical strategy involving the FMOC-Cl derivatization onto the SAX resin. (b) Chemical principle of the derivatization approach
(e.g., glyphosate).

elution was also studied, and pH 5 was found optimal. 78% ( 4% (HCO3-) versus 71% ( 3% and 32% ( 4% (OH-) for the
Stability of FMOC Derivatives on the SAX Resin during glyphosate-FMOC and metabolite-FMOC derivatives, respec-
Derivatization. The behavior of glyphosate- and AMPA-FMOC tively.
on the SAX resin was studied in conditions simulating the Requirements of the “On-Solid Support Derivatization”.
derivatization process. Aqueous solutions of these compounds The FMOC derivatization of glyphosate and AMPA is usually
were applied at alkaline pH (9 and 10) for fixation. The solid performed in solution (cf., in waters, soil, or sediment extracts).
support was conditioned and treated by FMOC-Cl as described The reaction requires (1) an alkaline medium (often pH 9) for
in the Experimental Section (see final procedure). Recoveries the reactivity of the amine function and (2) the presence of
through elution were evaluated by LC-UV. A significant loss of acetonitrile for reagent solubility. The following experiments were
FMOC analytes was observed particularly for AMPA-FMOC. performed to check whether resin-bound glyphosate and AMPA
Recoveries (n ) 3) were 90% ( 5% and 63% ( 6% (HCO3-) versus could react with FMOC-Cl and in which optimal conditions
62% ( 5% and 20% ( 3% (OH-) for glyphosate-FMOC and the (Figure 2).
metabolite-FMOC derivative, respectively. In the presence of pH Conditions. Samples (10 mL) spiked with analytes (150 µg),
OH- ions, the resin environment becomes more alkaline, probably simulating sludge extracts, were applied on the resin at pH values
promoting the hydrolysis of the linked derivatives. AMPA-FMOC ranging from 7 to 10. In these preliminary experiments, solutions
thus appears more sensitive to such reaction. This competitive were made from ultrapure water in order to avoid matrix effects.
effect was reduced when operating with a resin conditioned under The following sequence was tested: (i) sample percolation on the
the less alkaline HCO3- form. The influence of the pH of the SAX resin (HCO3-) followed by conditioning with mixture A (see
solution to be percolated before “derivatization” was also noted. the Experimental Section), (ii) loading of FMOC-Cl (10 mg) in
Hydrolysis was partly reduced when loading the sample at a lower mixture A for derivatization and 5 min of contact, (iii) support
alkaline pH (9 instead of 10). Recoveries became 87% ( 6% and rinsing with mixture B followed by water only, and (iv) elution
Analytical Chemistry, Vol. 79, No. 10, May 15, 2007 3797
Figure 2. Influence of various parameters on the yields of FMOC derivatization for glyphosate and AMPA fixed on a HCO3- resin (ultrapure
water solutions spiked with 150 µg of each analyte, two replicates, yields evaluated by HPLC-UV; b glyphosate, 2 AMPA).

with a 1 N NaCl solution in acetonitrile/water, 30/70, v/v, at pH for both resin conditioning and FMOC-Cl loading. In the 8.2-
5. The FMOC derivatization was found effective under these 10 interval, the pH of the buffer was found optimal at 9.2 and in
conditions: the obtained yields of the derivatives were 78% ( 7% each case more favorable to glyphosate. The best results were
and 27% ( 6% for glyphosate and AMPA, respectively. They did further obtained at the 25 mM concentration. The derivatization
not differ significantly (t test, four degrees of freedom, P < 0.05) was effective even in the absence of borate, probably because of
at pH values between 7 and 9 (three points, two replicates, n ) a residual alkalinity within the resin after percolation. At 100 mM,
6), In contrast, at pH 10 the yield for glyphosate was significantly derivatization was insufficient since borate ions can compete with
lower with 62% as a mean value (t test, five degrees of freedom, the analytes on the resin thus providing some breakthrough.
P < 0.05). The pH 9 value to fix the analytes was kept for the rest Excess of FMOC-Cl and Time of Reaction. Since during normal
of the study, and various other parameters were tested in order use the reagent can be rapidly hydrolyzed into inactive FMOC-
to improve the yields especially for the metabolite. OH, an excess of the reagent is generally recommended. Suspect-
Several compositions of borate buffer-acetonitrile (pH and ing the same in our conditions, FMOC-Cl was applied at different
borate buffer concentration, acetonitrile percentage) were tested doses. The optimal dose was found to be 20 mg, a higher excess
3798 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007
Figure 3. Positive and negative ESI-MS responses of glyphosate-FMOC (m/z 392 or 390) and AMPA-FMOC (m/z 334 or 332) for different
mobile phase compositions: A, B, C (2.5 mM NH4OAc), D, E, F (5 mM NH4OAc), and G, H, I (10 mM NH4OAc) with 25%, 50%, and 75% of
acetonitrile, respectively.

giving rise to precipitates in the column. Surprisingly, it was also for glyphosate and doubled for AMPA in ultrapure water. A third
observed that the time of contact between the reagent and the application of the reagent did not enhance the yields further. For
fixed analytes (even with sonication assistance) had no significant sludge samples in which analytes may compete with other amino
influence (in the 5-30 min interval) on the final yields of acids, the double-derivatization process gave even better yields
derivatization for both analytes. It was thus concluded that the (88% and 49% for glyphosate and AMPA, respectively).
reaction is quite fast and probably blocked rapidly. Further, the Excess Reagent Removing. FMOC-Cl in excess as well as
difference in the conversion yields for glyphosate and AMPA FMOC-OH, its hydrolysis product, can compete for trapping with
might be due to the deeper migration of the metabolite during target derivatives during the SPE concentration, which may result
percolation, protecting it to some extent from the subsequent in a lower efficiency of this step. Furthermore, they can interfere
introduction of the reagent. This hypothesis was supported by with analytes during the chromatographic separation used for
the presence of the single anionic function (phosphonate) in analysis. To selectively remove these interferences we tested
AMPA versus a double one (phosphonate and carboxylate) in the various mixtures of aqueous borate buffer (pH 9.2)-acetonitrile,
glyphosate molecule. This factor and a possible intrinsic difference with ratios ranging from 30/70, v/v, to 70/30, v/v, and volumes
of reactivity between both amines (secondary vs primary) should from V0 to 15V0. The efficiency of these compositions was
be considered together. evaluated by comparing peak areas of residual reagent in the LC-
Reagent Volume and Number of “Derivatizations”. In an effort UV chromatograms of the washes. The best results, i.e., allowing
to get a better distribution of FMOC-Cl within the resin, we tested a maximum elimination of FMOC-Cl and maintaining the fixation
several reagent volumes (from 1 to 9V0, V0 being the volume of of the FMOC analytes, were obtained using 15V0 of borate-
the resin employed). The derivatization yields remained consider- acetonitrile 50/50, v/v.
ably lower for AMPA than for glyphosate, and no significant effect Optimization of the SPE Preconcentration. The perfor-
of the reagent volume was observed. A 2V0 loading of the reagent mance of three reversed phases toward the fixation of FMOC-
solution was chosen as a compromise. These results suggest that glyphosate was evaluated by comparing their breakthrough
the reaction could be limited by a rapid fixation of FMOC-Cl by volumes (VB; see the Experimental Section). At pH 3 and 3% of
interaction with the polymeric backbone of the resin, hampering acetonitrile in the percolated solutions, recorded performances
the reagent to reach the linked AMPA molecules. To overcome were as follows: C18 (VB e 50 mL) < LiChrolut EN (VB e 220
this problem, we tested a “double derivatization” each time using mL) < Oasis HLB (VB e 500 mL). The EN polymeric phase is
the same parameters (10 min of contact, 20 mg of FMOC-Cl often more efficient than C18 due to the active aromatic sites that
loaded in 2V0 of mixture A). The solid support was rinsed and allow π-π interactions.24 Such interactions likely occur with the
reconditioned intermediately in order to release new active sites aromatic backbone of the FMOC derivatizing group of the analyte.
before carrying out the second reagent application. With the latter, Furthermore, the occurrence of the pyrrolidone group, a hydrogen
the reagent could be distributed more deeply within the phase, acceptor, in the Oasis HLB support makes it even more efficient
allowing reaction with the underivatized resin-bound analytes. (24) Thurman, E. M.; Mills, M. S. Solid Phase Extraction: Principles and Practice;
Indeed, the yields increased from 60% to 82% by operating twice Wiley: New York, 1998; Vol. 147.

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007 3799
Table 1. MS/MS Conditions and Performances of the MRM Method (ESI Positive vs Negative)

Tr cone collision C/Q LOD LOQ linearity

compd (min) ESI voltage (V) transition energy (eV) % (RSD)a (pg) (pg) (r2)
glyphosate-FMOC 3.2 positive 19 392 f 214 (Q) 9.5 62(4) 30 50 0.9995
392 f 170 (C) 10
negative 20 390 f 168 (Q) 10 39(2) 80 140 0.9968
390 f 150 (C) 25
AMPA-FMOC 8.9 positive 19 334 f 156 (Q) 8.5 36(4) 50 80 0.9997
334 f 112 (C) 10
negative 11 332 f 110 (Q) 10 38(9) 50 125 0.9972
332 f 136 (C) 15
a Average value calculated from 12 injections of standard solutions (three replicates, four concentration levels).

Figure 4. MRM chromatograms in the positive mode for quantitation of (a) glyphosate-FMOC and (b) AMPA-FMOC in a sewage sludge
sample (Plaisir, January 2005). Concentrations found in ppm dw: 1.8 (glyphosate) and 33.3 (AMPA). Analytes are confirmed by C/Q × 100
being within (20% for glyphosate and (25% for AMPA of the expected values (see Table 1; cf., European Commission Decision 2002/657/
CE). IS: isotopically labeled standard.

toward polar solutes, i.e., glyphosate-FMOC.24,25 The influence positive and negative modes. The MS procedure was first
of pH and acetonitrile percentage on VB was thus studied for investigated using direct infusion of methanolic solutions of
optimization in the case of Oasis HLB. The VB decreased standard FMOC derivatives. The cone voltage was optimized, and
dramatically when the pH was changed to 5 or 7 (ca. 10 mL in the variations of current ion intensities were evaluated for both
both cases). Acidic conditions appeared necessary to maintain a [M + H]+ (positive ionization) and [M - H]- ions (negative
zero net charge form (according to pKa values) of glyphosate- ionization). The results showed optimum values at 19 V for both
FMOC, the most favorable for retention. A low percentage of compounds in the positive mode. Concerning the [M - H]- ions,
acetonitrile could be admitted, e.g., 3% or 5% yielding with our optimal values were found at 20 and 11 V for glyphosate-FMOC
system VB values of 500 or 235 mL, respectively, but not 7% (VB and AMPA-FMOC, respectively. The effects of ammonium
< 2 mL). With the use of 120 mg Oasis HLB cartridges instead acetate and organic modifier percentage in the mobile phase were
of precolumns (cf., VB measurements, Experimental Section) 99% studied, these factors being known to influence the ionization
( 2% and 102% ( 3% recoveries were obtained for glyphosate- process.26 The analyte response increased with increasing aceto-
FMOC and AMPA-FMOC, respectively, using 300 mL spiked nitrile percentage and was always higher for the glyphosate
solutions (3% acetonitrile/water pH 3). On the basis of these derivative. The 5 mM concentration of ammonium acetate was
results, the eluates originating from the SAX resin (pH > 5, 30% found optimal for the [M + H]+ ions. In the negative mode, all
acetonitrile) should be acidified to pH 3 and 10-fold diluted before intensities were lower than in the positive mode (Figure 3).
SPE on Oasis cartridges with no risk of breakthrough. For MS/MS optimization, the fragmentation of both [M + H]+
LC-ESI-MS/MS Conditions. Glyphosate-FMOC and AMPA- and [M - H]- ions was investigated as a function of collision
FMOC were analyzed and compared in both electrospray (ESI) energy. Two ion transitions (parent ion f product ion) for

(25) Hennion, M.-C. J. Chromatogr., A 1999, 856, 3-54. (26) Mathis, J. A.; McCord, B. R. Forensic Int. 2005, 154, 159-166.

3800 Analytical Chemistry, Vol. 79, No. 10, May 15, 2007
quantitation (Q) and for confirmation (C) were chosen in each Application to WWTP Samples. Our method was applied to
case according to intensity and specificity criteria for application monitor sewage sludge contamination during 1 year (July 2004-
of the MRM mode. In agreement with the MS results, a better June 2005) in two WWTPs from Ile-de-France region. All samples
sensitivity in the positive mode than in the negative one was were found to contain both compounds. Glyphosate concentrations
observed for both analytes (Table 1).21 The possible matrix effects ranged between 0.2 and 2.9 ppm dw at Elancourt and between
were then investigated. The calibration curves obtained from real 0.5 and 1.8 ppm dw at Plaisir. Concentrations of the metabolite
sludge extracts with standard addition (R1) were compared with were always higher (2.3-20.9 ppm dw and 9.6-33.3 ppm dw at
those from external methanolic standards (R2) (seven concentra- Elancourt and Plaisir, respectively) than those of glyphosate
tion levels e6 to 5000 pg/µL g 5 µL injected). A satisfactory reflecting its various possible origins.27,28 Figure 4 shows an
linearity for both curves (R2 > 0.999) was noticed. But slope ratios example of the LC-ESI-MS/MS quantitation and characterization
of 3 (R1/R2) for glyphosate-FMOC and of 2 for AMPA-FMOC of glyphosate and AMPA in one of those samples (Plaisir, January
were observed, proving the existence of positive matrix effects. 2005).
Therefore, isotopically labeled standards were used for quantita-
CONCLUSIONS
tion instead of the laborious standard addition method. These In this work, we developed, optimized, and validated a method
internal standards (IS) were added to the extract before purifica- for glyphosate and AMPA analysis in sewage sludge. We proved
tion to undergo FMOC derivatization like the analytes. herein the possibility of FMOC-Cl derivatization on a SAX resin
Analytical Performances. To determine recovery rates, for bound glyphosate and AMPA. This approach has the advantage
sludge extracts were spiked at two concentration levels, 10 and of allowing a rapid reaction and easy sample handling since
100 ppm (dw), and the entire analytical procedure was applied. purification, derivatization, and elimination of excess reagent occur
Recoveries determined at 10 ppm (n ) 3), after subtracting sludge on the same support. After elution of the FMOC derivatives,
initial contamination and using standard addition calibration, were samples are concentrated on Oasis HLB cartridges via appropriate
70% ( 7% for glyphosate and 63% ( 3% for AMPA. The results aqueous dilution and pH adjustment. Followed by LC-ESI-MS/
obtained at 100 ppm were very similar. MS in the positive MRM mode, our procedure appears, despite
The method developed with internal calibration was then the complexity of the matrix, sufficiently sensitive and precise to
validated, by evaluating repeatability and limits of detection quantify glyphosate and AMPA residues in real sludge samples.
(LODs) and quantification (LOQs). The repeatability is expressed Monitoring of these analytes in two WWTPs of Ile-de-France
by the RSD of the analyte quantification in sludge samples. The region led us to observe a contamination of glyphosate which
sample used in the validation study was found to contain (n ) 5) rarely exceeded 2 ppm dw, whereas AMPA corresponding levels
both analytes at 1.5 ppm dw (glyphosate) and 16.6 ppm dw were 10-40-fold higher over all the year. According to the LODs
(AMPA), with RSDs of 8% and 2%, respectively. of our method (20 and 30 ppb, respectively), much lower levels
The LODs and LOQs were estimated as the analyte concentra- of the contaminants could be detected in sludges.
tions producing signal/noise ratios of 3:1 and 5:1, respectively, in
ACKNOWLEDGMENT
the MRM chromatograms. Their values were extrapolated from
The authors are grateful to the French Agency for Environment
signals exhibited by unspiked sludge samples since they were
and Energy Management (ADEME) and to the National Institute
initially contaminated at significant levels and found lower than for Agricultural Research (INRA) for the financial support of
50 ppb dw for both analytes (LOD 20 and 30 ppb, LOQ 35 and 50 Project 0275048 and the Ph.D. fellowship of A.G. They also thank
ppb for glyphosate and AMPA, respectively). These limits were the WWTPs of Elancourt and Plaisir for providing sludge samples.
close to those reported for soils,21 although this matrix is less F. Perreau is acknowledged for helpful discussions.
complex than sludge.
Received for review November 21, 2006. Accepted March
(27) Williams, G. M.; Kroes, R; Munro, I. C. Regul. Toxicol. Pharmacol. 2000, 1, 2007.
31, 117-165.
(28) Steber, J.; Wierich, P. Chemosphere 1987, 16, 1323-1337. AC062195K

Analytical Chemistry, Vol. 79, No. 10, May 15, 2007 3801