ooooi
  Determination of Perchlorate at Trace
   Levels in Drinking Water by Ion-Pair
  Extraction with Electrospray lonization
	Mass Spectrometry

     Matthew L. Magnuson, Edward T. Urbansky, and
              Catherine A. Kelty
 National Risk Management Research Laboratory, Water Supply and
Water Resources Division, Treatment Technology Evaluation Branch
 United States Environmental Protection Agency, 26 West Martin
        Luther King Drive, Cincinnati, Ohio 45268
    JN4LYTIG1L
    C H  E  M I S  T  R Y
              Reprinted from
         Volume 72, Number 1, Pages 25-29

-------
Anal. Cham. 2000, 72, 25-29

Determination  of  Perchlorate  at Trace  Levels  in
Drinking Water by  Ion-Pair  Extraction  with
Electrospray  lonization Mass  Spectrometry

Matthew L. Magnuson,* Edward T. Urbansky, and Catherine A. Kelty
National Risk Management Research Laboratory, Water Supply and Water Resources Division, Treatment Technology
Evaluation Branch, United States Environmental Protection Agency, 26 West Martin Luther King Drive,
Cincinnati, Ohio 45268
Perchlorate has been added to the U.S. Environmental
Protection Agency's Drinking Water Contaminant Candi-
date list (CCL). The present work describes the analysis
of perchlorate in water by liquid-liquid extraction fol-
lowed by flow injection electrospray mass spectrometry
(ESVMS). Cationic surfactants, mostly alkyltrimethyl-
ammonium salts, are used to ion-pair aqueous perchlo-
rate, forming extractable ion pairs. The cationic surfactant
associates with the perchlorate ion to form a complex
detectable by ESI/MS. The selectivity of the extraction and
the mass spectrometric detection increases confidence in
the identification of perchlorate. The method detection
ILtnitfor perchlorate based on 3.14
-------
   interfering species at the m/z of the most abundant ion, m/z 99,
   i.e., by hydrated bromide Br(H20)-. The mass of perchlorate is
   also  in  the  region  of "chemical noise", the  m/z range  with
   abundant naturally  occurring low-mass ions.  Observation  of
   interfering ions in electrospray mass spectrometry without re-
   course to chromatography or another on-line separation technique
   is common, and considerations for experimental protocol have
   been discussed in some detail.10 Consequently, to increase the
   selectivity for perchlorate, we recently investigated the detection
   of perchlorate using selective associative complexes of perchlorate
   with organic bases and other substances.11 By observation of a
   complex at a mass > 300 units higher than that of perchlorate,
   the classical chemical noise region was avoided. The complexation
   increased selectively, did not significantly decrease sensitivity
   (LOD ~ 10 /ig/L-1),  and was relatively free of common spectro-
   scopic interferences. The quanthation of species, via association
   of the ions in the electrospray process, has also been reported
  for other analytes such as Cr3*, which has been electrosprayed
  as a negative chloro  complex.12 Horlick and co-workers refer to
  electrospray  of  the  complex  as  the "intermediate"  mode of
  observation.13 Recently, in this laboratory, haloacetic acids were
  analyzed by electrospray mass spectrometry in a similar fashion.14
    The use  of a complexing  agent for  perchlorate increases
  selectivity11 but does not necessarily improve sensitivity. One
  approach to increasing sensitivity  for perchlorate determination
  is to add an ion-pairing agent to the aqueous solution and extract
  the ion pair with an organic solvent6-7 Conventionally, the ion pair
  is formed with a good chromophore, such as  a dye, and  the
  perchlorate may be   determined spectrophotometrically.6-7 It
  seemed reasonable  to combine  the  enhanced selectivity of
  complexation in ESI/MS with the enhanced sensitivity of solvent
  extraction. In the present work,  cationic surfactants  (alkyltri-
  methylammonium salts) were investigated for  their dual role in
  the formation of a  solvent-extractable ion pair and an electro-
  sprayable selective  complex  for mass spectrometric detection.
  Contaminated water from southern Nevada was analyzed by ion-
 pair extraction with ESI/MS detection, and the results compared
 favorably with those from ion chromatography.

 EXPERIMENTAL SECTION
    Reagents.  Brilliant cresyl blue [81029-05-2  (CAS Registry
 number)], brilliant green [633-03-4], and crystal violet [548-62-9]
 were obtained from Spectrum  (New Brunswick, NJ). Octyltri-
 methylammonium bromide  (C8)  [2083-68-3],  decyltrimethyl-
 ammonium bromide (CIO) [2082-84-0], dodecyltrimethylammo-
 nium bromide (C12) [1119-94-4], tetradecyltrimethylammonium
 bromide (C14)  [1119-97-7], and tributylheptylammonium bromide
 (THAB) [85169-31-9] were used as  received from Fluka  (Buchs,
 Switzerland). The organic solvents were obtained from Fisher
 (Fairlawn, NJ)  and were of Optima or similar quality. Aqueous
 perchlorate fortifications were made with ammonium perchlorate

 (10) Sharp, B. L.; Sulaiman, A. B.; Taylor, K A.; Green, G. NJ. Anal. At Spectrom.
    1997, 12, 603-609.
 (11) Urbansky, E. T.; Magnuson, M. L.; Freeman, D.; Jelks, C. /. Anal. At.
    Spectrom., in press.
 (12) Gwizdala, A. B.; Johnson, S. K; Mollah, S.; Houk, R S./. Anal. At. Spectrom.
    1997,12, 504-506.
 (13) Agnes, G. R.; Horlick, G. Appl. Spectrosc. 1994, 48, 649-655.
 (14) Magnuson, M. L.; Kelly, C. A; Jelks, C. In preparation.
26 Analytical Chemistry, Vol. 72, No. 1, January  1, 2000
   Table 1. Summary of Experimental Conditions
   acquisition mode
   applied ESI spray potential (optimized)
   interface capillary temperature
   sheath gas pressure
   injection mode/injection volume
   carrier liquid
   flow rate
   extractive ion-pairing agent
     (optimized)
   concentration of ion-pairing agent
     in solution
   extraction solvent
   preconcentration factor
   reconstitution solvent
negative ESI/MS
4.0 kV
200 °C
70psi(480kPa)
flow injection/50.0 ftL
methanol/dichloromethane
   (70/30 v/v)
0.3 mL/min
decyltrimethylammonium
   bromide (CIO)
1.0 mM

dichloromethane
500/1
methanol/dichloromethane
   (70/30 v/v)
  [7790-98-9] (Aldrich, Milwaukee, WI). Dilutions were made with
  water deionized through reverse osmosis.
     Synthetic Tap Water. A soft synthetic tap water was prepared
  by adding appropriate ACS reagent grade salts to deionized water.
  The synthetic tap water, which represents an extreme tap water
  in terms of ionic strength, was prepared to contain the following
  anion concentrations: 5.6 mM (200 mg/L) chloride, 0.97 mM (60
  mg/L) nitrate, 53 fM (1 mg/L) fluoride, 0.078 ftM (10 fig/L)
  bromate, 0.12 /iM (10/ig/L) chlorate, 0.10 mM (10 mg/L) sulfate,
  0.16 mM (10 mg/L) carbonate, 63 fM (5 mg/L) bromide. These
  concentrations were selected to be greater than the average
  concentrations found in many source waters.15 Sodium salts were
  used, except for bromide and bromate, which were prepared from
  potassium salts.
    Apparatus. Injections were made with a Rheodyne (Rohnert
 Park, CA) model 7725 injector having a 200 ftL loop. The pump
 for the carrier liquid was a Waters 600 (Waters, Milford, MA).
 The mass spectrometer was a Finnigan MATTSQ-700 (Finnigan,
 San Jose, CA) equipped with a Finnigan electrospray interface.
 Mass spectra were acquired in the negative-ion mode by scanning
 Q3 over appropriate mass ranges. Other experimental parameters
 are listed in Table 1.
    Procedure. A volume of 500 mL of the aqueous sample,
 cationic surfactant, and  100 mL of the  extraction solvent were
 shaken together vigorously. A separatory funnel was used to
 collect the organic phase. The organic phase was then reduced
 through rotary evaporation to dryness at 60 °C (bath temperature).
 The residue was redissolved in 5-7 mL of dichloromethane, and
 the  mixture was  transferred to a  disposable test tube  and
 re-evaporated at 45-50 °C in a heater block The localized residue
 was then reconstituted in 1.00 mL of the chosen solvent, and the
 solution was transferred  to a 1.8 mL glass vial.  Injections of 50
/J.L of this solution were  then analyzed  by flow injection  (FD-
ESI/MS.

RESULTS AND DISCUSSION
   1. Extraction Conditions. The extraction solvent for these
experiments was dichloromethane. Methyl isobutyl ketone (MffiK),
ethyl acetate, and fert-butyl methyl ether were also investigated.'

(15) U.S. Environmental Protection Agency. Chemical Analysis of Interstate Carrier
    Water Supply Systems; EPA Document No. 430/9-75-005; GPO: Washington,
   JJL^, iy/5.

-------
6S
^

B.
O
>
16
14
12
10
 8
           CJ
s
u
Q
                           
-------
               100
               80
I
• !•(
0>

               60
              20
                               C10-Bi*ClO«
                                 380
                       C10-Br-Br
                        360
                                     ClO-CICVCIO,
                                        388
                340
                        360
                                 380
                                m/z
                                         400
                                                 420
 Figure 2.  Mass spectrum of 100 /
-------
which is often seen in the electrospray of solutions with many
ionic species.16 Thus, the synthetic tap water, with its high ionic
content, would have a more suppressed signal. Careful use of the
separatory funnel did not alleviate this problem. A third possible
source of matrix effects is the species that are probably extracted,
e.g., sulfate, but are not detected by ESI/MS. Although not a
spectroscopic interference, their presence in the electrospray may
chemically interfere with the association and/or electrospray of
the perchlorate complex. Likewise, the presence of other detected
species, namely the chloride complex and the bromide complex,
may interfere with the association and/or electrospray of the
perchlorate complex. Whatever the cause of the matrix effect, the
slopes of the calibration plots (Figure 4) vary between waters, so
accurate determination of perchlorate must be made through the
use of standard additions. The high correlation coefficients of the
calibration plots suggest that standard addition should result in
accurate perchlorate determination.
   For the plots in Figure 4, a blank was subtracted from the data.
The background is probably due to the presence of natural organic
matter (NOM) in the water. This material typically has a large
m/z distribution range but is present in very small quantities and
is not normally detected. Due to the large concentration factor
 (500-fold) in this experiment, the NOM is concentrated and results
in a false positive of 0-2 /tg/L  depending on the water. Since the
types  and quantities of NOM  vary with source water,  it is
necessary to determine  the blank for each source water. This
determination  can be made on the basis of the following
considerations: The cationic surfactant is essential for the extrac-
tion of perchlorate. Because the cationic surfactant (ion-pairing
agent) is present below its critical micelle concentration,19 it was
 assumed that the surfactant  should not  sufficiently affect the
 extraction of the natural organic matter. Experimentally, the
 background was determined to be the same with  and without  the
 surfactant Therefore, when the extraction procedure is performed
 (18) For the reduction, 5 g of zinc dust and 100 fiL of glacial acetic acid were
    added to 500 mL of the water sample. The sample was allowed to react
    12-24 h, and the zinc dust was then filtered off at 0.45 fM (cellulose
    acetate). These quantities were optimized and mass spectrometrically
    observed to eliminate the nitrate signal. Acetic acid performed better than
    mineral acids at the same molarity.
 (19) Mukcrjcc, P., Myscls, K. J., Eds. Critical Micelle Concentrations of Aqueous
    Surfactant Systems; U.S. Department of Commerce,  National Standard
    RcTercnce Data System, National Bureau of Standards: Washington,  DC,
     1971; Vol. 36, passim.
without the cationic surfactant, the blank value for the water is
obtained (Figure 4).
   To demonstrate the capabilities of ion-pair extraction with ESI/
MS detection for determining perchlorate in contaminated water,
drinking water was obtained from a source in southern Nevada.
The perchlorate concentration in this drinking water was deter-
mined using ion chromatography at the water utility to be 8-9
ftg/L. According to the procedure outlined above, the concentra-
tion of the perchlorate was determined to be 8.4 ± 0.2 fig/L (n =
3) by standard addition. The agreement between the two inde-
pendent techniques increases confidence in the results, namely
that the peak in the ion chromatograph is for perchlorate and not
for an interfering species. The FI-ESI/MS determination, made
at ~40 times the detection limit, shows higher precision than the
ion chromatography determination, made at 2-3 times the 1C
detection limit

CONCLUSION
   A sensitive technique for  the analysis of  perchlorate is
demonstrated for measuring trace levels  in a variety of water
matrixes. The use of an ion-pairing agent for extraction reduces
the spectroscopic interference  for the nonchromatographically
separated  sample.  The  results for  a drinking water  sample
obtained by FI-ESI/MS compare well to those obtained by another
technique, ion chromatography with conductivity detection. The
ion-pairing/extraction agent is  selective for perchlorate among
common ions. The method detection limit is among the lowest
reported in the literature. Therefore, results  from this technique
could be compared with those from ion chromatography, as  well
as other emerging low-level techniques, for  the  purpose of
increasing the confidence in the accuracy of the reported  per-
chlorate concentration.

 ACKNOWLEDGMENT
    The authors thank microbiologist Peggy Roefer of the Southern
 Nevada Water Authority for supplying samples of raw and finished
 waters used in this work.
 Received for review  August  12,  1999. Accepted  October
 21, 1999.
 AC9909204
                                                                         Analytical Chemistry, Vol. 72, No. 1, January 1, 2000   29

-------