^a^-nioar
Quantitation of perchlorate ion by electrospray ionization mass
spectrometry . (ESI-MS) using ^stable assbcialbn Complexes with •
organic cations and bases to i enhance ^ selectivityt =- r
Edward T. Urbarisky,* Matthew L.Magnuspn? David Freemari arid Christopher Jelks : '
United States Environmental Protection Agency (EPA), Office bf Research : and Development - ' ' :;
National Risk Management Research Lahoratnrv Wntpr Simnh, n^A^^ta^ pjan,~Zii« '."':'' •-"•"
^
V- ^'Analytical "";.",.
.-;• Atomie* - .;/,
:- vSpectrametry; :;
i-' •- ._t- j '.: : ---.•~;-J.r': • . '
- - ' - ':-•- - ' •'-'.; --,.
.-••'-•' - : .' J . .r \ -.;,-
                                                   ,
  Division '..'Treatment Technology • Valuation 'Branch, 'Cincinnati
  E-mail: urbansky.edward@epamail.epa.goy; magnuson.matthew@epamaiiepa.goy,

  Received 14th July 1999, Accepted 28th September 1999
        ••' ''--'' -'.'.--,'"..  "i-i*-:." , ./. „'«/  '.;  •'r' ""•" J. '! -, ' :~r.^ " .* -• ;'•- "' :                                .              " j f '
  Quantitation of trace levels of perchlorate ion m^watef has become a key issue since this species was discovered
  in water supplies around the United States. Although ion chromatographic methods presently offer the lowest  ~
  limit.pf detection; ^40, nk (4 rig ml ,1), c,hromatographic; retention times-are not considered to be unique    •
  identifiers and.often cannot be used in legal-proceedings without confirmatory-testing. Mass spectromefery- can   '"
  provide such confirmation; However, detection capabilities can impbse a practical hmitatiori 6il;its^se;::'  --'    *
  Moreover, quadrupole mass spectrometers 'cannot provide sufficient i accurac/ and precision ^ in w/z4o Identify" ' ' '
  conclusively an ion as perchlofate! when samples are run directly without prior chromatographic or  '  ' ;"   "" '
  electrophoretic separation: We report on the 'abilities of (1) tetralkylatnmoniuni cations an4 (2) minimally"  ' : " " "
  nucleophilic, sterica'lly hindered organic bases to increase selectivity iii the electrpspray ionization mass "'-"'""' ?  ':>>>
  spectrometric (ESr-MS) determination of perchlorate ion without concomitant loss of sensitivity. Selectivity  ' "
  anses from the formation of a stable association'cbmpiex between,a base^olecule and -a perchlorate ariion f
  The best results were obtained using 10 UM chlorhexidine in methanohc solution; the lower limit of detection '   -'""
  (LLOD) for S/NSS2 was less  than or equalto 6.10 UM (10 ngml-1). This compares favorably with the LLOD •'" ""
  determined for perchlorate in the absence of any complexing agents ( « 0.05 UM = 5 ng mT '). For the other
  bases, which were diazabicyclo compounds (DBN, DBU, DBO), sensitivity was lower by 90% or more The
  chlorhexidine-perchlorate cdmplex-(w/z=<505) can be observed even in t&:Presehce ofequiformal nitrate"  -''"" ' '
 mtnte, hydrpgensulfate, chloride,  bromide, 'bromate, and chlorate (all together) dowiTfo approximately 1 'JIM- '   ':
 thus, themethpd is rugged;enbugh to fed appUcation to 'systems containing multiple morgamc anibris: ' " ' '"'"r


 Introduction.     ,.,''.: "'^.VL^ '"". '!? ";-;-;jEr-';i i:'-,"I-"r^';'''.',:

 Perchlorate ion was identified in ground and surface waters of
 the western United States in'l997, :arid has'since been found at
 sites around 'the country: It may  be  found- in the 'ground or
 surface" waters- near wherever- perchlorate salts -have been
 manufactured;- stored,: o'r used. ;Such;sites -are ofteir associated
 viith defense:pr •aerospace prograins '(of supporiingiridu^tries)
 since perchlorate salts find use- as ;s6lid 'bx'idants or energetics
 boosters m'rotkets Mi ^missiles. The anaiytical "chemistry of
 perchlorate and the significance of quantifying t'his coritami-^
 nant have been reviewed and described in detail elsewhere!1'2"'
   Because of the low concentrations  of perchlorate found at
 most of the contaminated :sites (5-50 ng ml"1),- the;=primary
 technique used .for water analysis has been ion chroniatography
 (1C): However,  chromatographic 'retention time  is not cpnT
 sidered_to be a unique identifier, and additibnal confirmation "is
 required to initiate legal actiori. Although 1C may be'used for
 routine monitoring, periodic  confirmatory testing must  be
 carried  out.  As, a consequence,  techniques such  as  mass
 spectrbmetry can  be expected to find a role  in "secondary
 confirmation, even if the  instrumentation is not generally
 available on-site to potable water utilities, responsible parties,
 or regulatory governmental agencies for primary environmen-
tal monitoring purposes.           -:'   -
  There  are  essentially  six  different  ways  of  assaying
perchlorate. Gravimetry and titrimetry only apply to :standar-
fUS Government copyright.
 dization  of fairly/concentrated' 'laboratory  (5=0.01 M)":-"spite"
 tibns. Ion-selective electrodes suffer from a number of anionic
 interferences as well as lower1 limits of .detection (LLODs) of
 ttl (11^=70ngml"1,1 \which is at least; twice;  what fhe'no
 observable /adverse effects level is predicted to'; be based on
 current.reseafch.3 Although Hauser et al report a detection
 limit of 10pgml"1 for an^ion-selebtive'-efectrode, this" was in
 cPnjunctiph with a capillary electrophores'is (CE)'separation^
 Barnett and Horlick used electrospray ionization mass spectrS-
 metry (ESI-MS) to obtain an LLOD of 0.050 JIM (5 ng itnl"1)'.5
 While spectrpphotometric determinations are reported to have
 similar-LLODs  to the; ESI-MS method,6  they  are:'not
 sufficiently?seleetive,'and dye purity can^be an issue.  Without
 priot  analyte  separation,  quadrupole -mass Spectrometers
 cannot measure mlz ratios  accurately' and precisely enough
 to••' identify  conclusively  an ion  Mthi a' mass of  99 u as
 perchlorate.                                           ;
  Despite the selectivity generally regarded as associated with
 mass spectrometric identification,-a large;number of small mass
 (.< SOQ.u) hydrophilic inorganic, or organic anions  can be found
 in natural water sources as; well as disinfected potable water
 supplies. -Mass spectrometry offers less ambiguity than  1C,
 but  with  a  reduction  in "sensitivity. Introduction/ionizatioh
 techniques for the involatile perchlorate ion are 'most likely
 constrained to either electrospray, which was used by Barnett
 and-Horlicfc,? or thermospray, which has not been reported for
perchlorate.  Extraction of  perchlorate salts from aqueous
solution is possible, but can be difficult to take advantage of
analytically.  Use of large cationic organic dyes, e.g., Brilliant
Cresyl Blue or Brilliant Green, is reported for spectrophoto-
                                                                     J.-Anal. At. Spectrom.,'1999,14, 1861-1866
                                  This journal is © The Royal Society of Chemistry 1999
                                                    1861

-------
                                                                          300 i
                                               J]   r
              0     100    200    300    400     500

             51   i  i  i  I  I' I   I  I  I  I  I  I  I   I  I
               0         100        200      '  300
                   Perchlorate concentration/uM '-•„•

Fig. 3 Calibration graphs (peak area versus perchlorate concentration)
for ESI-MS determination of perchlorate: (a) top to bottom, in (O)
10nM,(D)l HM,(A)0.1 UM chlorhexidine, MeOH solution; (b)(D,—)
1 mM DBU, (0,—) 10 mM DBN, (V, --) 1 HIM DBO. All lines are based
on least-squares regression.

to sec an unresolvable system of mixed anion complexes in the
mass spectrum or perhaps nothing at all, but this was not the
case. Instead, we found molecular ions of the form [chlorhex-
idine-X~]  for perchlorate,  chloride,  nitrate,  and  bromide.
However, there was no evidence for chlorhexidine  complexes
with nitrite, hydrogensulfate, acetate, bromate, or chlorate. It is
worth pointing out that we have never seen a  peak  that
corresponds to an association complex of acetate with any of
the reagents. We assume  that a peak at m/z=586 corresponds
to a chlorhexidine-hydrogensulfite complex that formed from"
sulfate reduced by the electrospray process, and we suspect that
chlorate, bromate, and nitrite are', reduced as  well. Further
exploration of these peaks and'the fate of the,Qther anions was
beyond  the scope of this work' and was not pursued.   .
   The peaks  for the chlorhexidine  complexes of chloride,
nitrate,  and bromide  are larger than the"  peak for the
perchlorate  complex; nevertheless,  Fig. 5 does show that the
perchlorate  response remains visible even under competition
for  the chlorhexidine.  Although the  sensitivity is reduced
relative to  perchlorate" in the absence of the other inorganic
anions,  a non-linear response curve can be generated, as shown
in Fig. 6. Attenuation of the signal at high'concentration is
probably due to reduced electrospray efficiency1 resulting from
ionic strength effects.            "- -:-- -•   '•'"•- •'•• '• -'' -"'
   Based on Fig. 3 and 6, we believe that this method could be
 applied under a variety  of conditions to both  analytical
 solutions  and  real water samples. Some  variation may-be
 needed  for optimization, such as changing pH or chlorhexidine
 concentration. Preconcentration may be  advised .for some
 samples, and we expect that standard approaches would work,
 e.g., simple evaporation, lyophilization, or onVcolumn relent
 tion (using a strong anion-exchange resin).: The suitability "of
 these would depend on both analyte concentration, and matrix
/
DO

50

A

^;T:\J£
- / 1000
/ 500
". 0. . 	 ....
:i. , , •" •: :. . .
1 i- -1 -•'• i-r--. I.-; i


' 1 ' 1 ' 1 ' 1 ' <
^o. .'.-..: ,;~
5,' 1 ,:-| -, 1 , I"5, -
) 1 ''2 3 ' '4'
1 ,- 1 'if -1

:_J
"
;

5 -

,_-/  :.---.•.--,. ,-0,0,  .,2.0 •' 4.0,;  6.0,.; 8.0  ; 10.0'   •'•<.:  ; - -.
   ,_:., -•:  ;-.,.:.,.,;,-  Perchlorate cpncentratiOn/|O.M .".    ,,<    ...  '

fig. 4 Calibratio^grapn (peaic area versus^^ perchlorate concentration)
for ' ESI-MS  determination  of perchlorate' iri 10 |M chlorhexidine,
MeOH solution. First six points used for least-squares regression line.
Presumably, points 7 and 8:suffer--from;loss of electrospray efficiency
due to  ionic strength." Inset: .Calibration graph for perchlorate anipn
without any  complexing agents,'MeOH solution, provided- for
comparison. Non-linear response.at .higher perchlorate concentration
is'observe'd with and without complexing agents. \'       ;   .  ., ,  ,
                                                                                                584
                                                                 ,.-.80-
 "H SP-
 §
 I40'
 0)
 CC -:---
" '-20;
                                                                                          567
04W-
500
                                     603   605
              525
                     >550
                            .575.
                            "m/z
                                       600
                                               ,625,
                                                     . .65,0-.
 Fig.-5 Negative  ion ESI;mass spectrum :for' I.UM' '(equifofmal)
 ISTBUClOife  NaClO3, NaCl,:,NaBrO3,,NaBr, NaND2,:.'NaNO3,,and
 NaHSQ,} jn 10 ,UM chlorhexidine, MeQH solution. In this case, the only
 perchlorate-chlorhexidine association complex .observed, is that con-
 taining one chlorhexidme,niolecule. and one'perchiorate ion, .See, text
 for'additional discussion:'    ,  "'"'   ,.,'•',',~,   r:,   ,,',,,. ,-V-."

 constituents'.'Given the ionic content of the test solutions, we
 expect that matrix effects co,uld readily be accounted for in
 drinking water by using the method of standard additions; but
 this would require validation and optimization on/a per .case
 basis. Fig. 5 suggests that .chlorhexidine might  even prove
 useful for selectively enhancing  a number of different anions,
 but further exploration ofl the phenomenon was  beyond  the
 scope of this work.      ;   , ;         . '          .'",•'. ,, .

   Applicability to potable water analysis.  Because of the high
 ionic strength of potable.water; 'which includes sodium chloride
 at  considerably  higher  concentration   than the  analyte,
 suppression of the electrospray signal" is to be expectecl: Two
 different potable waters: were tested: Cincinnati tap  water and
 Tri-Township  Water  (TTW)7'The sdurce of Cincinnati  tap
 water is the Ohio River.;TTW.is supplied by a well in Dearborn
 County, Indiana, and probably experiences some infiltration
 from the Whitewater River. The TTW sample was collected
 from a residential faucet in Logan Township.         ." - -   '.  .>
    A'20 ml sample of TTW water was evaporated to dryness at
 60 °C and reconstituted in an 'eqiial volume of 10% v/v HaO-
 MeOH:' Aqueoui*:;chlor'rie'xidrne'>was?::added'.; to giye':'i'"final
 chlorhexidine  concentration :<5'f. 1.6 IXM .iii ."the..'reconstituted
 1864    /. Anal. At. Spectrom., 1999,14, 1861-186.6

-------
             0.03
I I  [ \Ll |  I t i  I 1 I  I I  I / /f\
 5       10     15     20 50
        [ClO4-]/uM
 Pig. 6 Perchlorate response (peak area versus concentration) for ESI-
 MS determination in 10 JJ.M chlorhexidine in the presence of seven other
 equiformal anions,  MeOH solution; see Fig. 5 caption. This is not
 a true calibration graph, because the sample matrix (but not the
 chlorhexidine) varies in concentration, too. Inset: Calibration graph for
 perchlorate anion without any complexing agents, MeOH solution,
 provided  for comparison.  Non-linear  response  at  higher anion
 concentration is observed, presumably due to ionic strength effects
 on electrospray efficiency.                -  •

 sample.  It is important to  note that the mineralized residue
 resulting from  evaporation does not redissolve in the final
 solution.  In  order  to promote  as  much  dissolution of
 perchlorate salts as possible, the reconstituted samples were
 permitted to sit for 20-40 min before decanting the solvent. A
 signal distinguishable from that of unspiked  samples  was
 obtained at 20 and 40 UM perchlorate with a response factor of
 16 000 area units per 20 UM spiked into the sample. Much of the
 chlorhexidine peak (m/z=5Q5 u) is lost from the  supernatant,
 presumably due to adsorption  onto  the mineralized residue.
 For  this reason, spiking  the  supernatant  with  additional
 perchlorate after decanting  from the evaporation vessel gives
 no increase in signal. Nonetheless,  if  the chlorhexidine
 concentration is returned to 10 JJ.M '(as  indicated  by the peak
 at  7M/z=505u), further perchlorate  spikes give  appropriate
 response.
   Because  of  the electrospray suppression, large  dilution
 factors are required if the waters  are to be  run  without
 preconcentration. We found that a 5% v/v dilution of tap water
 samples gave the best results. Of course, this would mean that
 the perchlorate concentration in the water sample would have
 to be 20  times greater than the post-dilution concentration. In
 both  TTW and Cincinnati samples, a post-dilution concentra-
 tion  of  250 nM perchlorate  (1.25UM  pre-dilution)  gave  a
 response of about 9200-9500 area units above the unspiked
 blank. We must point out that the response is not linear in these
 matrices, and quantification of perchlorate under such cond-
 itions would require careful construction of a calibration graph
 in the matrix under study.
   Simple dilution is adjudged not to be a generally  suitable
 approach  for quantifying  perchlorate in the   presence  of
 multiple  other inorganic anions. Ideally,  a preconcentration
 technique would select .for perchlorate over other anions.  We
 speculate that preconcentration using certain highly selective
 resins, such as  those recently  developed  by the Oak Ridge
 National  Laboratory,7  would  dramatically  improve  the
 performance of  this  method and could reduce the LLOD by
 a factor of 100 or more.  .-:'"--

  Methanol versus propan-2-ol.  We compared the signal of
 50 ugmT1 C1O4~ standards with DBU and chlorhexidine in
propan-2-ol versus methanol.  No  measurable difference was
found. In addition,  we were  concerned  that  clusters of
(MeOH)2(H2O)2 might  be  responsible for units of A(mf
z)=+100  rather  than HC1O4. Since these  peaks are  also
 observed in the propan-2-ol solutions but not in perchlorate-
 free methanolic blanks, we conclude that the perchlorate is in
 fact responsible,  rather  than some cluster of solvent  or
 spectator species.              -

 Quaternary cations and other species       •<•'•.

 Performance of the tetraalkylammonium ions as complexing
 agents was so  much poorer than  the organic  bases that it
 suffices to say that these will be of little analytical utility, No
 signal was  detected for the molecular ions of the tetramethy-
 lammonium, tetraethylammonium,  or tetrapropylammonium
 cations, [(NR4+)(C1O4~)2]. Triethanolarnmonium and betha-
 nechol gave  observable signals, but these did  not  compare
 favorably   with  chlorhexidine or  the  diazabicyclo  bases.
 Tetrabutylammonium,  tetrahexylammonium,  and tetraocty-
 lammonium gave signals whose magnitude increased with the
 length  of the .carbon chain for equiformal [NR4+]. However,
 these  salts  are limited  by their  solubilities in  water and
 methanol  (or combinations thereof), and  did. not  compare
 favorably with the results obtained for the organic bases above.
 No signal  was found  at the mlz  ratio predicted for  the
 molecular ion of tetraphenylarsonium cation associated with a
 perchlorate anion  and either another perchlorate anion or a
 chloride anion. No signal was detected for a protonated nitron
 associated with perchlorate anion;  neutral nitron is zwitter-
 ionic, and it is usually protonated with acetic acid to solubilize
 it and to form the precipitant cation. In positive ion mode ESI-
 MS, however, all  of the expected cations could be identified;
 Consequently, we feel confident in saying that there is  a lack of
 the complexation behavior characteristic of the other  bases
 rather than an inability of .the cations to be carried  into the
 vapor phase.    ....... ; •     .                .         "

 Cationic organic dyes                                  ,

 We did not observe any evidence for the presence of complexes
 of perchlorate with cationic dyes in the ESI mass spectra when
 directly injecting the methanolic dye-perchlorate solutions or
 the reconstituted  extracts.  While we are uncertain  whether
 the extractions were successful,, we suspect that the failure Of the
 direct  injection is  probably  due  to  the   inability  of the
 electrospray apparatus to nebulize  and ionize these species
 effectively. Regardless, this particular strategy does not appear
 to be worth pursuing further, even though it may be suitable for
 the reported spectrophotometric determinations.6


 Conclusions

 Chlorhexidine gave the greatest improvement  in selectivity
 without loss  of sensitivity for perchlorate detection  and
 quantification, and it was the only base to show resistance to
 matrix  effects,  even  at much  lower  base concentration.
 Although the non-nucleophilic  organic bases  (i.e., DBO,
 DBN, DBU) gave acceptable increases in selectivity, losses of
 sensitivity were rather high. Surprisingly, cationic precipitants
 used in perchlorate gravimetry, i.e.,  tetraphenylarsonium and
 nitron,  gave  no signal at all. Because the organic bases
 substantially elevate the mass—by a factor of 3-8—the peak is
 completely separated from the low mass area where other
 anions are commonly found. In addition, complexes with water
 or methanol molecules are not seen in the mass spectrum; thus,
 peaks of species such as H2O;81Br~ (m/z=99), which was'seen
 by  Barnett and Horlick,5 are conveniently absent. The Use of
 complexing  reagents therefore permits ESI-MS analysis with-
 out separation on account of its increased selectivity. None-
 theless,  use  of 1C or CE is not precluded,  and hyphenated
 techniques (i.e., LC-MS  or CE-MS) may find a use for  such
reagents.The thoughtful use of complexing  agents such  as
chlorhexidine will perhaps provide a valuable tool in terms of
                                                                     ; J. Anal-At. Spectrom., 1999, 14, 1861-1866    1865

-------
increased selectivity without diminished sensitivity in the
analytical determination of the perchlorate anion.

Acknowledgements

The financial support of EPA's summer mentoring program is
recognized by D.F. and C.J.

References
1  E. T. Urbansky, Biorem. J., 1998,2,81, and references cited therein.
2  E. T. Urbansky and M. R. Schock, J. Environ Manag., 1999,56,79.
3  US Environmental Protection Agency, Perchlorate Environmental
   Contamination: Toxicological Review and Risk Characterization
   Based on Emerging Information, Review Draft, December 31, 1998,
   Document No. NCEA-1-503.
4  P. C. Hauser, N. D. Renner and A. P. C. Hong, Anal. Chim. Acta,
   1994, 295, 181.
5  D. A. Barnett and G. Horlick, /. Anal. At. Spectrom., 1997,12, 497.
6  A. A. Ensafl and B. Rezaei, Anal. Lett., 1998, 31, 167.
7  B. Gu,  G. M. Brown, S. D. Alexandratos, R. Ober and V. Patel,
   Selective Anion Exchange Resins for the Removal of Perchlorate
   CIO4~ from Groundwqter, Oak Ridge National  Laboratory, Oak
   Ridge, TN, February, 1999, Doc. No. ORNL/TM-13753, Environ-
   mental  Sciences Division Publication No. 4863.

                                            Paper 9105721H
                                                                                                                                 .A~
-------