Journal of the Science of Food and Agriculture
          3 Sci Food Agric 80:1798-1804 (online: 2000)
 Survey of  bottled waters  for  perchlorate by

 electrospray ionization  mass spectrometry

 (ESI-MS)  and ion  chromatography (IC)f

 Edward T Urbansky,1* Baohua Gu,2 Matthew L Magnuson,1 Gilbert M  Brown3 and

 Catherine A Kelty1
 1 United States Environrnental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory,
 Water Supply and Water Resources Division, Cincinnati, OH 45268, USA
 2Environmental Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA*
 ^Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA*


    Abstract: Perchlorate has been identified in ground and surface waters around the  USA including
    some that serve as supplies for drinking water. Because perchlorate salts are used as solid oxidants in
    rockets and ordnance, water contamination may occur near military or aerospace installations or
    defense industry manufacturing facilities. This ion has been added to the Environmental Protection
    Agency's Contaminant Candidate List and the Unregulated Contaminant Monitoring Rule. Concern
    over perchlorate has prompted many residents hi affected areas to switch to bottled water; however,
    bottled waters have not  previously been examined for perchlorate contamination. Should the EPA
    promulgate a regulation for  municipal water systems,  US  law requires  the Food and Drug
    Administration to  take action  on bottled water. Methods will therefore be required to determine
    perchlorate concentrations not only in tap water, but also in bottled waters. Ion chromatography (1C)
    is the primary technique used for its analysis in drinking water, but it does not provide a unique
    identification. Confirmation by electrospray ionization mass spectrometry (ESI-MS) can serve in this
    capacity. The ESI-MS method  can be applied to these products, but it requires an understanding of
    matrix effects, especially of high ionic strength that can suppress electrospray. When using methyl
    isobutyl ketone (MIBK) as the extraction solvent,  the ESI-MS method can reach lower limits of
    detection of 6ngml~1 for some bottled waters. However, dilution required to negate ionic strength
    effects in mineral waters can raise this by a  factor  of 10 or more, depending on the sample.
    Decyltrimethylammonium cation (added as the bromide salt) is used to produce an ion pair that is
    extracted into MIBK. After extraction, the sum of the peak areas of the ions C10H21NMej(Br)(ClO,,)~
    (m/s=380) and C^HjjNMe^ClO,^ (mfe=400) is used to quantitate perchlorate. Standard additions
    are used to account for most of the matrix effects.  In this work, eight domestic brands and eight
    imported brands of bottled water were comparatively analyzed by the two techniques. For comparison,
    a finished potable water known to contain perchlorate was also tested. None of the bottled waters were
    found to contain any perchlorate within the lower limit of detection for the 1C method. Recoveries on
    spiked samples subjected to the 1C method were >98%.
    Published in 2000 for SCI by John Wiley & Sons,  Ltd

 Keywords: bottled water; perchlorate; ion chromatography; electrospray ionization mass spectrometry; drinking
 water
 INTRODUCTION
 In  1997, perchlorate ion was found in sources of
 drinking water for much of the south western USA
 including the Colorado River.1"3 Improvements in ion
 chromatography for this moiety have resulted in lower
 limits of detection on the order of Sngml"1 (parts per
 billion) and made detection possible at these sites.2"10
 Perchlorate  has  since been found in ground  and
surface waters at concentrations ranging from 5ng
ml"1 to S.Tmgml"1.1  It is believed that  the per-
chlorate is largely derived from defense and aerospace
industry practices and military operations that took
place decades ago.
  Perchlorate is a strong oxidant, and this behavior is
obvious when   hot  concentrated perchloric acid
touches organic matter. Nonetheless, perchlorate
 * Correspondence to: Edward T Urbansky, United States Environmental Protection Agency, Office of Research and Development, National
 Risk Management Research Laboratory, Water Supply and Water Resources Division, Cincinnati, OH 45268, USA
 E-mail: Urbansky.Edward@EPA.gov
 f This article is a US Government work and is in the public domain in the USA.
 * Oak Ridge National Laboratory is operated under contract by University of Tennessee-Battelle LLC for the US Department of Energy.
 (Received 24 February 2000; revised version received 24 April 2000; accepted 31 May 2000)
 Published in 2000 for SCI by John Wiley & Sons, Ltd
                                                                                           1798

-------
                                                                 Survey of bottled waters for perchlorate by JSSI-JMS
reduction is encumbered by a high activation energy,
which precludes reaction under cold  and  dilute
conditions. The chemistry of perchlorate has been
reviewed  previously.1  As  a  consequence  of this
activation barrier, perchlorate is quite unreactive and
thus long-lived under the dilute  and relatively cold
conditions  encountered in natural bodies of water,
finished potable water supplies, and in biophysiology
(as when water is consumed by people or animals). In
terms of public health,  perchlorate interferes with
iodide uptake in the thyroid  gland because the two
anions are  similar in size.1'3'11'12 The EPA National
Center for Environmental Assessment has undertaken
studies to determine safe levels for exposure  to this
contaminant. Meanwhile, two provisional action levels
of 4 and ISngml"1, based on different assumptions
used in the original toxicology studies, remain in force.
  At  present,  the  EPA Office of  Water has not
established a national primary drinking water standard
for perchlorate; however, this ion has been added to
the Contaminant Candidate List (CCL)13'14 and the
Unregulated    Contaminants    Monitoring   Rule
(UCMR).15 If the EPA promulgates a drinking water
regulation, the Food and Drug Administration (FDA),
will be required to act on bottled water, pursuant to 21
USC 349.  Newspaper accounts have suggested that
many people in areas with perchlorate-contaminated
drinking water have  switched  to  bottled  water,
although bottled waters have not so far been tested
to our knowledge.
  Previously, we reported  on the determination  of
perchlorate in drinking  water using  electrospray
ionization  mass spectrometry  (ESI-MS).16"19  Per-
chlorate anion may be extracted into  organic solvents
using surfactant cations,  especially alkyltrimethylam-
monium ions (eg decyl, lauryl, myristyl or cetyl).17~19
Other investigators  also  report quantitation of per-
chlorate by mass spectrometric methods.20"23  In this
work, we examine the application of this approach as
well as the ion chromatographic method to ten brands
of bottled water. Ion chromatography can be a useful
screening tool. If a sample is chromatographed and no
peak is observed, a fortified sample can be tested. If
recovery of the spike is satisfactory, there is reasonable
assurance that the analyte is not present. One of the
weaknesses of any chromatographic  method is that
identification by retention time  is  not  necessarily
unique.  Confirmatory testing  can be accomplished
effectively by mass spectrometry because of the mass
of the ion and the resulting mlz ratio. Isotopic ratios
based on relative abundance (in this case 35C1 vs 37C1
with a 3:1 ratio) provide  additional evidence that the
identification is correct.
  Bottled  waters  can  vary  significantly in  ionic
strength, salt composition and dissolved gas content
(mainly carbon dioxide). Like municipal potable water
supplies, bottled waters vary in treatment techniques,
including disinfection. Some bottled  waters, such as
Pepsi's Aquafina or Coca-Cola's Dasani  are derived
from  local  sources — either municipal tap water  or
natural  springs. * Both  brands are then purified by
reverse  osmosis.  Other  bottled  waters,  such  as
Grayson Mountain Springs, Dannon, Polar Mountain
Water, Evian, San Pellegrino, Naya and Perrier are
taken from specific natural springs. Both Perrier and
San Pellegrino are naturally carbonated. Some inter-
nationally distributed bottled waters are taken from
specific  springs and therefore represent widespread
geographical  exposure  for  North American  and
European nations, even if the products are  consumed
by  a relatively  small  fraction  of the population.
Consumption of this nature presents  challenges for
risk assessment and  management. Products such as
Dasani and Aquafina represent a local  bottler's water
source and  thus a  localized  geographical zone  of
exposure.  Of course, they must be tested  on a local
scale as they represent the specific source for a bottler
and not for the brand overall.
EXPERIMENTAL PROCEDURE
Sample procurement and custody
Bottled waters were obtained by EPA staff. Manufac-
turer seals were broken by EPA staff in the laboratory;
1.2-dl portions were decanted into new polypropylene
or high-density polyethylene bottles, placed in sealed
packages to ensure chain of custody, and sent to the
Oak Ridge National Laboratory (ORNL) by overnight
carrier, where they were opened exclusively by ORNL
staff. Perrier and San  Pellegrino  were  allowed  to
de-gas CO2 prior to use. For comparison, a perchlo-
rate-tainted potable water sample (Southern Nevada
Water Authority) was also subjected to  the experi-
mental procedure. Tested waters are listed in Table 1.

Electrospray ionization  mass spectrometric
analyses
Reagents
High purity deionized  water  (HP DI water) was
prepared by polishing house deionized water (reverse
osmosis/UV-irradiation)  with a Barnstead EasyPure
system with organic-removal and deionizing cartridges
to  obtain  a  resistivity   >18MJicm. The  cationic
surfactant,    decyltrimethylammonium    bromide
(C10H21NMe3Br), was obtained from Fluka (Buchs,
Switzerland) and prepared at  0.20M (5.6gdr:) by
dissolving the  solid in HP  DI water. The choice of
surfactant involves  striking a balance  among three
main factors: the ability form extractable ion  pairs,
availability of the reagent in high purity, and resistance
to forming intractable emulsions, as described pre-
viously.18   HPLC   grade   4-methyl-2-pentanone
(methyl isobutyl ketone, MIBK) was obtained from
Spectrum Quality Chemicals (San Rafael, CA, USA)
and used  as  the  extraction  solvent.  Ammonium
perchlorate  (Aldrich, Milwaukee, WI,  USA) was
prepared at l.OOgT1 ClO^"  in HP DI water and
serially diluted as necessary.
JSdFoodAgric 80:1798-1804 (online: 2000)
                                                                                                  1799

-------
ET Urbansky et al
Table 1. Waters examined in this study
 Brand
                                                      Company, address
                                                         Source
Apollinaris Naturally Sparkling Spring Water  Liberty Richter (distributor), Saddle Brook, NJ 07663, USA   Germany
Aquafina

Crystal Geyser Natural Spring Water
Dannon

Dasani
Eureka Springs All Natural Spring Water
Evian
Fountainhead Natural Spring Water
Gerolsteiner Natural Mineral Water

Grayson Mountains
Naya
S Pellegrino
Perrier
Poland Spring Natural Spring Water
Polar Mountain Spring Water

Southern Nevada3 Water Authority

Volvic Natural Spring Water
PepsiCo bottler was Warrenton Products Inc, Warrenton,    Varies by location
  MO 63383, USA
CG Roxane LP, P O Box 249, Benton, TN 37307, USA      CG Roxane
Dannon National Spring Water, 208 Harbor Drive, Stamford,  Spring Piedmont,
                                                    Quebec, Canada
                                                  Varies by location
                                                  Municipal tap
                                                  Cachet Spring
  CT 06902, USA
The Coca-Cola Co, Atlanta, GA, USA
Saegertown Beverages, Saegertown, PA 16433, USA
SA Evian Co, Evian-Les-Bains, France
Fountainhead, White Water Falls Rd, Oconee, SC, USA
Gerolsteiner Brunnen GmbH & Co,
  D-54567 Gerolstein/Vulkaneifel, Germany
Grayson Mountain Water Co, Independence, VA, USA
Nora Beverages Inc, Mirabel, Quebec, JOV1ZO, Canada
San Pellegrino, Italy
Perrier, Vergeze, France
Poland Spring Water Co, Poland, ME 04274, USA
Crystal Springs Water Co, Mableton, GA 30059, USA

Las Vegas, NV, USA

Danone Group, Ste Volvic, at Volvic 63530, France
                                                  Sumter Natural Forest
                                                  Gerolstein spring
                                                    (near Rhine, Moselle)
                                                  Local spring
                                                  Local spring
                                                  Local spring
                                                  Local spring
                                                  Local spring
                                                  Crystal Spring,
                                                    Blue Ridge, GA
                                                  Lake Mead on
                                                    Colorado River
                                                  Clairvic Spring
a SNWA is a finished potable water from a municipal plant and not a bottled water. It is used for comparison as it is known to contain perchlorate ion.
Extraction
The following were quantitatively added to a 100-ml
volumetric flask:  96.0ml  of the  test water  sample,
1.0ml of 0.20M C10H21NMe3Br (aquesus), and 5.0ml
of MIBK.  The rationale for this procedure  is fairly
straightforward. The presence of small water droplets
cannot  be  tolerated  in  the ESI-MS analysis.  A
separately funnel will not work because MIBK is less
dense than water (coming out second); thus, residual
water droplets left on the funnel will be carried along in
the MIBK layer. Because of the large volume of water
to MIBK,  a  40-ml vial, such as that used for many
EPA drinking water methods also will not work. The
advantage  to a 100-ml volumetric flask is that the
solvent is constrained to the neck, where it is readily
drawn off once  the   extraction  is  complete.  We
emphasize that the flask is not used in its volumetric
capacity,  but  merely  as   a convenient  vessel  for
recovering  the extraction solvent. Duplicate samples
were run with the following fortifications  (spikes): 0
(unspiked),  10, or 20ngml~1 ClO^".  As a blank,
unspiked samples of the bottled waters were extracted
without the surfactant. The blank correction proce-
dure has been explained previously.18'19  Due to
limited sample volume of the first lot, Perrier and S
Pellegrino were not run in duplicate; however, another
lot was tested in duplicate.

Instrumentation
A Finnigan-Mat (San  Rafael, CA, USA) TSQ 700
                   quadrupole mass spectrometer (Q3) with a Finnegan-
                   Mat  electrospray interface  apparatus was  used for
                   analysis.  A  Waters  (Milford,  MA, USA)  MS-600
                   pump  delivered  pesticide  residue  analysis  grade
                   methanol  (Burdick & Jackson, Baxter Healthcare,
                   Muskegon, MI, USA) as the carrier; sheathing gas
                   pressure was 480 kPa (70 psi); capillary temperature
                   was 200°C; spray potential was 4.0kV.

                   Ion chromatographic analyses
                   Samples were  analyzed at ORNL as  received from
                   EPA following the method reported by Jackson et al. 8
                   Dionex (Sunnyvale, CA, USA) AG11 guard and ASH
                   analytical columns were used on a Dionex DX-500 ion
                   chromatograph for these analyses. In addition to the
                   test water samples, spiked samples containing 10 and
                   20ngml~1 CIO 7 were run to verify performance.
                  RESULTS AND DISCUSSION
                  ESI-MS data analysis
                  To correct for other species that form ions with mlz
                  equal  to  that of the analyte ions, the peak areas
                  obtained for an MIBK extract without detergent were
                  subtracted from those  extracted  with the cationic
                  surfactant.  Perchlorate  forms  complexes  with the
                  surfactant cation  of  the  form  C10H21NMe3(Br)
                  (ClO4r   (m/z=38Q)   and  C10H?1NMe3(ClO4)J
                  (m/z—400').  When extracted with dichloromethane,
                  municipal potable  water  samples  showed   linear
1800
                                                                     J Sci Food Agric 80:1798-1804 (online: 2000)

-------
                                                                     Survey of bottled waters for perchlorate by ESI-MS
               lOppb
                          20ppb
   .1?
                                   unspikcd
     JVAAJVJUAJL-JUVAJLJUVAJVJ?!'00
              5.0        100        is5zrTo
                     Time (min)

Figure 1. Negative ion ESI-MS injection peaks for Southern Nevada Water
Authority finished water, a supply known to contain perchlorate. The blank
value is obtained without surfactant (no C10H2,NMeJ). Along with the
cationic surfactant, standard additions of 10 and 20ngmr1 are used to
extrapolate to the perchlorate concentration in the water. Two ions are used
to quantitate the perchlorate: C10H21NMe3(Br) (CIO4)- (m/z=380) and
C10H21NMe3(CIO4)J (m/z=400). Note that perchlorate concentration is
related to peak area and not peak height. The relationship between
concentration (amount of analyte injected) and peak area is not
immediately evident because of the shape of the peaks, and integrated
areas must be used. Nearly all (>90%) of the perchlorate is extracted into
the solvent the first time as shown by successive extractions.
increase in peak area  with perchlorate concentra-
tion.18'19 For this work, however, MIBKwas chosen as
the  solvent to  minimize  exposure  of  laboratory
personnel  to  chlorinated solvents in keeping with
EPA's  goals  of laboratory  safety  in test method
development. Partitioning and ionization  of the ion
pairs in MIBK is  reduced  relative  to  methylene
chloride. This proceduce is sufficient for the purposes
of quantitation, but it requires more careful treatment
of the  raw data. The  difference in  partitioning is
heightened in waters of higher ionic strength because
there is more competition for  the  decyltrimethyl-
ammonium ion. The  best  results are  obtained by
summing the areas of the two  ions, C10H21NMe3(Br)
(ClO4r and C10H21NMe3(ClO4)-. While it may be
possible to mitigate the effect by using more surfac-
tant, there is a significant risk of forming emulsions
and the latter approach was not explored.
  Figure 1 shows the flow injection peaks observed for
a  water known to  contain  perchlorate  (SNWA).
Making standard additions and monitoring the sum
of the  peak areas at m/z = 38Q  and  400, we then
extrapolated to the abscissa. This gives the concentra-
tion of perchlorate  in  the  sample  in a reasonably
straightforward fashion (Fig 2).

Analytical  results
Perchlorate was found only in the water sample from
the Southern Nevada Water Authority. The concen-
tration  of  the  ion obtained using 1C and  ESI-MS
methods agreed within the limits  of experimental
error. None of the bottled waters tested contained
perchlorate above the lower limit of detection reported
for the  1C  method, 5ngml~1.  The 1C  method is well
established8 and straightforward.  Chromatograms for
Aquafina and Naya  are shown in Fig 3; recovery is
excellent as shown  by  the  chromatograms of the
fortified samples.  For all of the  waters, recovery  of
the  spike  by 1C  was  >98%. Perchlorate was not
detected in any brand of water. For comparison, Fig 4
              -50    5    10   15   20   25
              Perchlorate concentration (ng ml'1)

Figure 2. Standard addition plots for (Q) Aquafina, (A) Naya, and (V)
SNWA from ESI-MS analysis. The sum of the blank-corrected peak areas
for m/z=380 and 400 are plotted against the added perchlorate
concentration (duplicate spikes at 10 and 20ngmP1). The x-intercept
represents the concentration of perchlorate in the sample.
for SNWA  finished  water
         "1
shows  chromatograms
unspiked and spiked with lOngml
  Least squares parameters for the standard additions
used in the  ESI-MS method are  summarized for
bottled waters  in Table  2. The ESI-MS  method
corroborates the 1C result obtained for the SNWA
sample. For the non-carbonated bottled waters, the
ESI-MS results are consistent with the 1C findings that
perchlorate is below the lower limit of detection, of
approximately Sngml"1. The  two  naturally carbo-
nated waters show a curious effect: positive abscissal
intercepts,  which  correspond  to  negative  analyte
concentrations.  This non-sensical result is  caused by
the high ionic strength of these waters. Although they
were de-gassed prior to analysis to remove  CO2, they
contain high levels of dissolved inorganic salts. High
ionic strength interferes with electrospray ionization;
accordingly,  signal  intensity  is  diminished.  This
produces a negative peak relative to the background
signal of the methanol  carrier, which has an ionic
strength approaching zero. In some cases, the problem
can be eliminated by diluting the sample; however, the
lower limit of detection is raised by a factor equal to the
dilution factor. Consequently, there is a trade-off to be
made, and thorough characterization of the matrix is
required to apply this method optimally to waters with
high ionic  content. Even with 10% v/v dilutions, we
were unable to  successfully quantitate trace perchlo-
rate in  Perrier,  Pellegrino, or several other sparkling
mineral waters  that contain high levels of dissolved
inorganic salts (positive x-intercepts were obtained for
all). Although  perchlorate  concentrations >200ng
ml"1 can be  quantitated in mineral waters,  this has
little practical application.
  The lower limit of detection (LLOD) for the ESI-
MS  method using MIBK is engml""1, based on DI
water standards run at 0, 1.0, 2.0, 4.0, 6.0,  8.0 and
JSci Food Agric 80:1798-1804 (online: 2000)
                                                                                                        1801

-------
ET Urbansky et al
                                     0.8
                                                   Aqua Flna Bottlad Watar
                                       0246    8   10   12

                                 (3.)         Retention time (min)
                                                                                               Naya Bottlad Watar '
                                                             0246    8   10    12

                                                       (b)         Retention time (min)
Figure 3. Ion chromatograms for
bottled water samples that contain no
perchlorate: (a) Aquaflna and (b) Naya.
Chromatograms for the same samples
fortified with 10ngml~1 perchlorate:
(c) Aquafina and (d) Naya.
                                                    Aqua Flna Bottlad Walar
                    0   2    4   6    8   !0    12
              (C)         Retention time (m!n)
                                                                                               Naya Bottlad Watar
                                                                                                  Spiked to ppb CIO.'
                                                    0    2    4    6    8   10   12
                                              (d)         Retention time (min)
 lO.Ongml  l. Eight replicates at 6.0ngml  l gave an
 estimated standard  deviation of O^ngml"1. This
 leads to an EPA method detection limit (MDL)  of
 O.Tngml"1 (using Student's t = 3.500 for the 99%
 confidence interval and seven degrees of freedom).
 However, a threshold signal  distinguishable from the
 blank does not occur below 5ngml-1, and so we feel a
 minimum reporting  level of S-dngml"1  is the most
 appropriate.
   In addition to  the possibility of electrospray sup-
 pression, it is worth  noting that samples  of very high
 ionic strength may  suffer from  a  limiting  reagent
 problem.  The post-mixing  concentration  of the
 cationic surfactant is  2.0mM; thus, competition among
 anions for the surfactant must be considered. This is
 particularly true  for less-hydra ted  anions, such  as
 nitrate  or  bromide,  which  are extracted into the
 MIBK, paired with the surfactant cation. Competition
 for the surfactant - to the point of adversely affecting
 the  analysis  — is uncommon with  potable  water
                                      samples. However, the ESI-MS method cannot readily
                                      be used for beverages such as fruit juices or wines or for
                                      seawater.

                                      Applicability and conclusions
                                      While 1C is likely to be the predominant technique for
                                      determining perchlorate ion concentration in raw and
                                      finished drinking  waters  (including  bottled waters),
                                      ESI-MS provides a useful means of confirming the 1C
                                      identification, which is based on retention time alone.
                                      Capillary electrophoresis or Raman scattering spectro-
                                      metry24'23 may one day play a similar role, but these
                                      techniques are currently not sensitive enough without
                                      pre-concentration (eg with an anion exchange resin or
                                      sample stacking). ESI-MS may be applied to a variety
                                      of drinking water matrices as a confirmatory technique
                                      to support identifications made by 1C. It is reasonably
                                      rugged and reproducible  while  requiring a minimum
                                      of sample preparation and no separation step prior to
                                      analysis. As  with the 1C method, high  levels of
                05  0.4-
                   0.2
(a)
           5. Nevada Wator Authority

0.0 -]
  0246
             8    10    12
Retention time (min)
                                                         CO  0.4-
                                                            0.0-
                                                               0
                                                                      S. Nevada Water Authority
                                      (b)
                                                                    2     4    6    8    10    12

                                                                     Retention time (min)
 Figure 4. Ion chromatograms for SNWA water (a) unspiked and (b) spiked with 10ngml"1 perchlorate.
 1802
                                                                       JSciFoodAgric 80:1798-1804 (online: 2000)

-------
ET Urbansky et al
     Perchlorate in the Environment, ed by Urbansky ET, Kluwer/
     Plenum, New York, NY, USA (2000).
 11 Wolff J, Perchlorate and the thyroid gland. Pharm Revs 50:89-
     105 (1998).
 12 Clark J, Chapter 3. Toxicology of perchlorate, in Perchlorate in the
     Environment) ed by Urbansky ET, Kluwer/Plenum, New York,
     NY, USA (2000).
 13 Perciasepe R,  Part  III. Environmental Protection Agency.
     Announcement of the drinking water contaminant candidate
     list; notice. Fed Regist 63:10273-10287 (1998).
 14 Drinking Water Contaminant Candidate List, Feb 1998, EPA
     DocNo815-F-98-002.
 15 Browner C, Part II. Environmental Protection Agency, 40 CFR
     parts 9, 141 and 142, Revisions to unregulated contaminant
     monitoring regulation for public water systems; final rule. Fed
     Regist 64:50555-50620 (1999).
 16 Urbansky ET, Magnuson  ML,  Freeman D  and Jelks  C,
     Quantitation  of perchlorate ion by  electrospray  ionization
     mass spectrometry (ESI-MS) using stable association com-
     plexes with organic cations and bases to enhance selectivity. J
     AnalAtSpectrom 14:1861-1866 (1999).
 17 Urbansky ET and Magnuson ML, Chapter 8. Sensitivity and
     selectivity enhancement in perchlorate  anion quantitation
     using complexation-electrospray ionization-mass spectrome-
     try, in Perchlorate in the Environment, ed by Urbansky ET,
     Kluwer/Plenum, New York, NY, USA (2000).
 18 Magnuson ML, Urbansky ET and Kelty CA, Determination of
     perchlorate at  trace levels in  drinking water  by ion-pair
     extraction with electrospray ionization mass spectrometry.
     AnalChem 72:25-29 (2000).
 19 Magnuson ML,  Urbansky  ET   and Kelty CA, Microscale
     extraction of perchlorate  in drinking water with low level
     detection by electrospray-mass spectrometry. Talanta 52:285-
     291 (2000).
 20 Barnett DA, Guevremont R and Purves RW, Determination of
     parts-per-trillion levels of  chlorate, bromate, and iodate by
     electrospray ionizarion/high-field asymmetric waveform ion
     mobility  spectrometry/mass  spectrometry. Appl Spectrosc
     53:1367-1374 (1999).
21 Barnett DA and  Horlick G, Quantitative  electrospray mass
     spectrometry  of halides and  halogenic anions. J Anal At
     Spectrom 12:497-501 (1997).
22 Clewell RE, Chaudhuri S, Dickson S, Cassady RS, Wallner WN,
     Eldridge JE and Tsui DT, Chapter 6. Analysis of trace level
     perchlorate in drinking water and ground water by electrospray
     mass spectrometry, in Perchlorate in the Environment,  ed by
     Urbansky ET, Kluwer/Plenum, New York, NY, USA (2000).
23 Koester CJ, Beller HR and Halden RU, Analysis of perchlorate
     in groundwater by electrospray ionization mass spectrometry/
     mass spectrometry. Environ Sci Technol 34:1862-1864 (2000).
24 Kowalchyk WK, Walker PAIII and Morris MS, Rapid normal
     Raman spectroscopy of sub-ppm oxy-anion solutions: the role
     of electrophoretic preconcentration. Appl Spectrosc 49:1183—
     1188 (1995).
25 Miller  AG  and Macklin JA, Matrix effects on the Raman
     analytical lines of oxyanions. Anal Ghem 52:807-812 (1980).
26 Urbansky ET,  Magnuson ML, Kelty CA, Brown SK, Per-
     chlorate uptake by salt cedar (Tamarix ramosissima) in the Las
     Vegas Wash riparian ecosystem. Set Tot Environ 256:227-232
     (2000).
27 Schilt AA, Perchloric Acid and  Perchlorates, GFS Chemicals,
     Columbus, OH, USA, pp 3-4  (1979) and references therein.
28 Susarla  S,   Collette  TW,  Garrison  AW,   Wolfe NL  and
     McCutcheon  SC,  Perchlorate identification  in fertilizers.
     Environ  Sci  Technol  33:3469-3472  (1999) and references
     therein; see also correction Environ Sci Technol 34:224 (2000).
29 Logan  BE,  A  review of chlorate-  and perchlorate-respiring
     microorganisms. Biorem J 2:69-78  (1998) and reference^
     therein.
30 Coates JD, MichaelidouU, Bruce RA, O'Connor SM, Crespi JN
     and Achenbach LA,  The ubiquity and diversity of dissim-
     ilatory  (per)chlorate-reducing  bacteria.  Appl Env Microbiol
     65:5234-5241 (1999).
 1804
                                                                                  JSciFoodAgric 80:1798-1804 (online: 2000)

-------