An HPLC method with UV detection, pH control, and reductive ascorbic acid for cyanuric acid analysis in water

Article abstract of DOI:10.1021/ac0005868

Every year over 250 million pounds of cyanuric acid (CA) and chlorinated isocyanurates are produced industrially. These compounds are standard ingredients in formulations for household bleaches, industrial cleansers, dishwasher compounds, general sanitizers, and chlorine stabilizers. The method developed for CA using high-performance liquid chromatography (HPLC) with UV detection simplifies and optimizes certain parameters of previous methodologies by effective pH control of the eluent (95% phosphate buffer: 5% methanol, v/v) to the narrow pH range of 7.2-7.4. UV detection was set at the optimum wavelength of 213 nm where the cyanuric ion absorbs strongly. Analysis at the lower pH range of 6.8-7.1 proved inadequate due to CA keto - enol tautomerism, while at pHs of < 6.8 there were substantial losses in analytical sensitivity. In contrast, pHs of > 7.4 proved more sensitive but their use was rejected because of CA elution at the chromatographic void volume and due to chemical interferences. The complex equilibria of chlorinated isocyanurates and associated species were suppressed by using reductive ascorbic acid to restrict the products to CA. UV, HPLC-UV, and electrospray ionization mass spectrometry techniques were combined to monitor the reactive chlorinated isocyanurates and to support the use of ascorbic acid. The resulting method is reproducible and measures CA in the 0.5-125 mg/L linear concentration range with a method detection limit of 0.05 mg/L in water.

Full text of DOI:10.1021/ac0005868

Anal. Chem. 2000, 72, 5820-5828
An HPLC Method with UV Detection, pH Control,
and Reductive Ascorbic Acid for Cyanuric Acid
Analysis in Water
Ricardo Cantu´,*,† Otis Evans,† Fred K. Kawahara,‡ Jody A. Shoemaker,† and Alfred P. Dufour†
National Exposure Research Laboratory, National Risk Management Research Laboratory, United States Environmental
Protection Agency, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268
chemically into its chlorinated derivatives.1 In addition, the
chlorinated isocyanurates are registered for use as disinfectants,
sanitizers, algaecides, fungicides, fungistats, bactericides, bac-
teriostats, microbiocides, and microbistats.6,7 The worldwide
production of chlorinated isocyanurates in 1996 was estimated to
be 292 million pounds (U.S. 195,8 Japan 52,8 and Western Europe
458). These wide applications and large chemical production
volume demand practical methodology to ensure these com-
pounds are used without posing undue hazard to human health
or the environment.
Cyanuric acid occurs as a mixture of keto (2,4,6-trioxo-s-
triazine) and enol (2,4,6-trihydroxy-s-triazine) tautomers in solution
(Figure 1, structures 1 and 2 , respectively).10-12 In alkaline
solution, cyanuric acid exists in the enolic form, while in acidic
solutions, the keto form is more stable.12 The keto tautomer 1 is
usually called isocyanuric acid. Cyanuric acid can undergo reaction
at either O- or N-triazine substituents depending upon its environ-
ment.13 The dichloro (DCCA, 4 ) and trichloro (TCCA, 5 ) isocy-
anurates are produced by stoichiometric reaction of CA, NaOH,
and chlorine gas in solution. Monochloroisocyanurate (3 ) is not
readily available because of its rapid disproportionation to CA and
DCCA. A considerable number of techniques such as titrimetry,4,14
turbidimetry,15 colorimetry,16,17 pulse polarography,18 gas chro-
matography (GC),19 high-performance liquid chromatography
(HPLC),15,20-33 and commercial devices16,17,34 have been developed
Every year over 2 5 0 million pounds of cyanuric acid (CA)
and chlorinated isocyanurates are produced industrially.
These compounds are standard ingredients in formula-
tions for household bleaches, industrial cleansers, dish-
washer compounds, general sanitizers, and chlorine
stabilizers. The method developed for CA using high-
performance liquid chromatography (HP LC) with UV
detection simplifies and optimizes certain parameters of
previous methodologies by effective pH control of the
eluent (9 5 % phosphate buffer: 5 % methanol, v/ v) to the
narrow pH range of 7 .2 -7 .4 . UV detection was set at the
optimum wavelength of 2 1 3 nm where the cyanuric ion
absorbs strongly. Analysis at the lower pH range of 6 .8 -
7.1 proved inadequate due to CA keto-enol tautomerism,
while at pHs of <6 .8 there were substantial losses in
analytical sensitivity. In contrast, pHs of >7 .4 proved
more sensitive but their use was rejected because of CA
elution at the chromatographic void volume and due to
chemical interferences. The complex equilibria of chlori-
nated isocyanurates and associated species were sup-
pressed by using reductive ascorbic acid to restrict the
products to CA. UV, HP LC-UV, and electrospray ioniza-
tion mass spectrometry techniques were combined to
monitor the reactive chlorinated isocyanurates and to
support the use of ascorbic acid. The resulting method is
reproducible and measures CA in the 0 .5 -1 2 5 mg/ L
linear concentration range with a method detection limit
of 0 .0 5 mg/ L in water.
(4) O’Brien, J. E. Ph.D. Dissertation, Harvard University, 1972.
(5) Petritsi, I. Ph.D. Dissertation. Yale University, 1964
(6) Registration Eligibility Document Facts. Chlorinated Isocyanurates; EPA
Document No.738-F-92-010; Washington, DC, 1992.
(7) Status of Pesticides in Registration, Reregistration, and Special Review
(Rainbow Report); Office of Pesticide Programs; USEPA, Washington, DC,
1998.
(8) Chlorinated Isocyanurates, Chemical Economic Handbook; SRI Interna-
tional: Menlo Park, CA, 1999.
(9) Manuscript in preparation.
(10) Cignitti, M.; Paoloni, L. Spectrochim. Acta 1 9 6 4 , 20, 221-218.
(11) Klotz, I. M.; Askounis, T. J. Am. Chem. Soc. 1 9 4 7 , 69, 801-803.
(12) Ito, M. Bull. Chem. Soc. Jpn. 1 9 5 3 , 26, 339-341.
(13) Quirke, J. M. E. In Comprehensive Heterocyclic Chemistry; Katritzky, A. R.,
Rees, C. W. Eds.; Pergamon Press: New York, 1984; Vol. 3, pp 457-530.
(14) Morales, R. Anal. Chem. 1 9 6 8 , 40, 1148-1149.
(15) Downes, C. J.; Mitchell, J. W.; Viotto, E. S.; Eggers, N. J. Water Res. 1 9 8 4 ,
18 (3), 277-280.
Cyanuric acid (CA) and chlorinated isocyanurates are added
as standard ingredients in formulations for household bleaches,
institutional and industrial cleansers, automatic dishwasher com-
pounds, and chlorine stabilizers, which are very well known for
preventing the total photolytic decomposition of chlorine disinfec-
tants.1-5 Most of the CA produced in industry is converted
* Corresponding author: (phone) (513) 569-7883; (fax) (513) 569-7757;
(e-mail) Cantu.Ricardo@epa.gov.
† National Exposure Research Laboratory.
‡ National Risk Management Research Laboratory.
(1) Encyclopedia of Chemical Technology, 4th ed.; John Wiley & Sons: New York,
1993; Vol. 7, pp 834-851.
(16) Rupe, C. O. to Environmental Test Systems, Inc. World Patent 8,906,355,
(2) Ullmann’s Encyclopedia of Industrial Chemistry, 5th ed.; New York, 1987;
July 13, 1989.
Vol. 8, pp 191-200.
(3) The Effect of Cyanuric Acid. a Chlorine Stabilizer, on Trihalomethane
Formation; EPA Document No. 600/ D-84-167; Cincinnati, OH, 1984.
(17) Rupe, C. O. to Environmental Test Systems, Inc. U.S. Patent 4, 855, 239,
Aug 8, 1989.
(18) Struys, J.; Wolfs, P. M. Anal. Chim. Acta 1 9 8 7 , 199, 173-176.
5820 Analytical Chemistry, Vol. 72, No. 23, December 1, 2000
10.1021/ac0005868 Not subject to U.S. Copyright. Publ. 2000 Am. Chem. Soc.
Published on Web 10/26/2000

Figure 1. Chemical structures of keto (1) and enol (2) tautomers of cyanuric acid and synthetic routes to their monochloro- (3), dichloro- (4),
and trichloro- (5) isocyanurate derivatives.
for CA analysis. Titration of CA as a monobasic acid is restrictive
because of interferences from other weak acids during end point
determinations. Application of potentiometry, amperometry, and
pulse polarographic techniques is frequently difficult, requiring
complex electrochemical methods with special electrode arrange-
ments, cell designs, supporting electrolytes, and organic solvent
requirements. Turbidimetry suffers from coprecipitation of inter-
ferences such as ammelide, ammeline, and melamine. GC
methods have proved functional, but the derivatization and
extraction steps require extra time for the total analysis. Finally,
HPLC methods with UV detection have been the most popular
because of their operational selectivity and sensitivity.
pHs (1-10), and settings for UV detection (195-226 nm). The
lack of agreement in these parameters warrants further examina-
tion of the conditions in these methods to extract their useful
properties and to define operational conditions for the measure-
ment of CA. Table 1 also includes various detectable levels for
comparative purposes.
The present work provides the necessary background regard-
ing existing methodology and presents significant improvements
and results for CA measurement. It is composed of five parts. In
part I, judicious criteria for selection of the separation phase are
applied based upon method preference, rapidity, toxicity, and
practicality. In part II, the UV spectrum of CA is discussed with
special emphasis on the restrictive keto-enol tautomerism af-
fecting the HPLC analysis of CA. In part III, the UV and HPLC-
UV techniques are combined to monitor the sensitivity and
retention of CA in the pH ranges of 1.94-7.41 (acidic) and 7.44-
10.06 (basic). The sensitivity trends are established by comparison
of the UV absorption peaks and integrated peak areas from the
HPLC. The retention trends are established with measured
capacity factors, k’s. Parts II and III delineate optimum conditions
for CA analysis. In part IV, the method is expanded to include
DCCA and TCCA whose reactivities in water were monitored
using HPLC-UV and electrospray ionization mass spectrometry
(ESI MS) techniques. Cyanuric acid calibration curves were
prepared for quantitative analyses and the method detection limit
(MDL) was calculated. Finally, part V examines several reductive
dechlorinators in regard to their spectroscopic and separation
compatibility to CA.
Table 1 summarizes HPLC methods for CA. It reveals a wide
variety of separation modes, columns, mobile-phase compositions,
(19) Bodalbhai, L. H.; Yokley, R. A.; Cheung, M. W. J. Agric. Food Chem. 1 9 8 8 ,
46, 161-167.
(20) Manual of Analytical Methods, Method 5030, 4th Ed.; National Institute for
Occupational Safety and Health: Washington, DC, 1994.
(21) Tucker, S. P.; Blade, L. M. Anal. Lett. 1 9 9 2 , 25 (12), 2265-2277.
(22) Chemical Information Manual; Instruction CPL 2-2.43A; Occupational Safety
and Health Administration: Washington, DC, July 1, 1991.
(23) Kontchou, C. Y.; Gschwind, N. Ecotoxicol. Environ. Saf. 1 9 9 9 , 43, 47-56.
(24) Schaber, P. M.; Colson, J.; Higgins, S.; Dietz, E.; Thielen, D.; Anspach, B.;
Brauer, J. Am. Lab. 1 9 9 9 , 31 (16), 13-21.
(25) Ghiorghis, A.; Talebian, A. H. J. Liq. Chromatogr. 1 9 9 1 , 14 (12), 2331-
2349.
(26) Pelizzeti, E.; Maurino, V.; Minero, C.; Carlin, V.; Pramauro, E.; Zerbinati,
O.; Tosato, M. L. Environ. Sci. Technol. 1 9 9 0 , 24, 1559-1565.
(27) Sugita, T.; Ishiwata, H.; Yoshihira, K.; Maekawa, A. Bull. Environ. Contam.
Toxicol. 1 9 9 0 , 44, 567-571.
(28) Allen, L. M.; Briggle, T. V.; Pfaffenberger, C. D. Drug Metab. 1 9 8 2 , 13 (3),
499-516.
The HPLC method developed has been adapted to demonstrate
its ruggedness in the analysis of real-world water samples during
field studies9 and extended to the analysis of CA in biological body
fluids for exposure studies.9 Common matrix interferences in water
were found to be strong absorbing inorganic anions such as
nitrates, phosphates, sulfates, chlorates, etc., while body fluids
contained additional interferences such as proteins and nonpolar
(29) Briggle, T. V.; Allen, L. M.; Duncan, R. C.; Pfaffenberger, C. D. J. Assoc. Off.
Anal. Chem. 1 9 8 1 , 64 (5), 1222-1226.
(30) Gonzalez, M. C.; Braun, A. M. Chemosphere 1 9 9 4 , 28 (12), 2121-2127.
(31) Fa¨rber, H.; Nick, K.; Scho¨ ler, H. F. Fresenius J. Anal. Chem. 1 9 9 4 , 350,
145-149.
(32) Guenu, S.; Hennion, M.-C. J. Chromatogr., A 1 9 9 4 , 665, 243-251.
(33) Pichon, V.; Chen, L.; Guenu, S.; Hennion, M.-C. J. Chromatogr., A 1 9 9 5 ,
711, 257-267.
(34) Pinsky, M. L. to FMC Corp., U.S. Patent 4,134,798, Jan 16, 1979.
Analytical Chemistry, Vol. 72, No. 23, December 1, 2000 5821

Table 1. HPLC Methods for Cyanuric Acid Analysis
detectability
(mg/ L)
mode
column
mobile-phase composition (v/ v, %)
pH UV
6.7 214
refs
normal phase
reversed phase octadecyl
amino
78/ 22 CH3CN/ KH2PO4-Na2HPO4
99/ 1 H2O/ H3PO4
0.02
0.25
4, 0.005
0.1
0.1
5
27
31
1
3
195
205
acidic H2O
32,33
95/ 5 Na2HPO4/ CH3OH
95/ 5 Na2HPO4/ CH3OH
95/ 5 H2O/ CH3CN
7.0 213
7.0 225
n/ a 226
28
20-22, 29
25
graphite
octadecyl
70/ 30 sodium phosphate/ CH3OH
KH2PO4/ H3PO4/ CH3OH
didodecyldimethylammonium bromide in phosphate buffer
didodecyldimethylammonium bromide in 50/ 50 H2O/ CH3OH
7
220
0.003
10
32-33
7.7 215
7.7 220
n/ a
23
ion pair
n/ aa
13
26
30
ion exchange
strong anion exchange tris(hydroxymethyl) methylamine sulfate
K2HPO4/ KOH
7.8 215
5
15
24
10
210
40b
a
n/ a, not available. b Obtained from HPLC and thermal gravimetric analysis, after heating 100.0 g of urea initially present.
compounds. The effective application of the method and sample
cleanup procedures will be described elsewhere.9
volumes were passed through the column for equilibration. Each
sample was injected three to five times to monitor reproducibility.
Peak areas from the chromatograms were computed using
Chromgraph Report version 1.2 (BAS). Final figures were dis-
played using Igor pro software version 3.14 (WaveMetrics, Inc.,
Portland, OR).
EXPERIMENTAL SECTION
Reagents. Cyanuric acid and chlorinated isocyanurate salts
were purchased from Aldrich Chemical (Milwaukee, WI). Stock
solutions of 530 mg/ L CA (98%), 520 mg/ L DCCA, sodium salt
(96%), and 535 mg/ L TCCA (97%) were prepared by dissolving
the solid in deionized water with stirring and gentle heating. All
of the solutions were passed through a 0.45-µm cellulose filter.
The standard solutions were prepared in the 0.5-125.0 mg/ L
concentration range in phosphate buffer. Their concentrations
were corrected to reagent purity specifications. The phosphate
buffers contained 0.005 M monobasic potassium phosphate (KH2-
PO4) and 0.0125 M dibasic potassium phosphate (K2HPO4). A
minimum of either acid or conjugate base solid was added to
achieve the exact pH within 1 pH unit of the buffer capacity.
Gradual changes in pH to acidic or alkaline conditions were
carried out by dropwise addition of either 1 M H2SO4 or 1 N
NaOH. The pH was measured using a pH Field System (Thomas
Scientific, Swedesboro, NJ) with a 0.01 pH resolution over the
pH range 0-14. Ascorbic acid (99+%) and sodium thiosulfate
(99%) were obtained from Aldrich Chemical and sodium sulfite
(98.9%) was obtained from Fisher Scientific (Pittsburgh, PA).
HP LC System and Conditions. The HPLC was performed
with a model 200B liquid chromatograph (Bioanalytical Systems,
W. Lafayette, IN) equipped with a Rheodyne (Cotati, CA) model
7125 injector employing a 20-µL sample loop and a built-in single-
channel UV-visible variable-wavelength detector. The UV detector
was set at the wavelength of 213 nm.
Electronic Spectra. Using a Beckman DU-640 spectropho-
tometer, the absorption spectra of solution samples were collected
with a special 1-cm quartz cell. The blank solution and sample
solutions were adjusted to the same pH. The resolution of the
diode array detector was set at 0.1 nm. Multiple scans (3-15)
were averaged to improve the signal-to-noise ratio.
FIA-ESI MS. A Finnigan LCQ DECA (San Jose, CA) ion trap
mass spectrometer was calibrated in the positive ion ESI mode
using an Ultramark 1621MRFA/ caffeine/ mixture and used in the
negative ion ESI mode. The heated capillary (350 °C), nitrogen
sheath gas (80 arbitrary units, AU), and nitrogen auxiliary gas
(30 AU) were optimized, in the negative ion ESI mode, using a
solution of 20 mg/ L benzoic acid infused at 0.8 mL/ min in 95:5
water/ methanol (%, v/ v). A Hewlett-Packard 1090 liquid chro-
matograph (Agilent Technology, Wilmington, DE) was used for
the flow injection analysis (FIA). In ESI MS, the instrument was
scanned from 110 to 300 amu with ESI needle voltage at -4 kV.
The samples at pH 7.3 were analyzed by FIA, using a 50-µL
injection loop into a 0.8 mL/ min flow of 95:5 water/ methanol
(%, v/ v).
RESULTS AND DISCUSSION
HP LC Method Selection and P reliminary Experiments.
Careful analysis of Table 1 reveals a preference for reversed-phase
separation mode in the HPLC analysis of CA. The high polarity
and hydrophilic characteristics of CA32 have been advantageous
in obtaining its rapid detection in reversed-phase liquid chroma-
tography (RPLC). Weak hydrophobic interaction between the
nonpolar octadecyl silica and the CA chromophore results in its
early elution from the analytical column. Elution at the void volume
has been prevented by maximizing water in the composition of
the mobile phase. The combination of the octadecyl silica column
and high water content mobile phases promotes maximum
retention in RPLC. Also, the analysis is relatively nontoxic, using
g95% phosphate buffer and e5% methanol (v/ v), which are not a
serious concern in analytical laboratories.
A 25 cm × 0.46 cm C18 Luna column end capped (Phenomenex,
Torrance, CA) with 5-µm particle size was used. The stability of
the column was monitored by measuring the number of plates
(N) of the CA peak over a 2-month period. Initial N were
11 600 ( 500 and subsequent usages after 2 months yielded N )
8200 ( 600. The plates were calculated from N ) 5.55 tr2/ w21/ 2
,
where tr is retention time of CA and w1/ 2 is the width at half-height
of the CA peak.
The HPLC was carried out under isocratic elution conditions.
Other parameters included a flow rate of 1 mL/ min and a data
acquisition rate of 150 points/ min. The solvents were purged using
high-purity helium gas (99.9%, Murray Hill, NJ). Twenty column
5822 Analytical Chemistry, Vol. 72, No. 23, December 1, 2000

Normal-phase liquid chromatography is not widely used due
to the limited solubility of CA in organic solvents. Cyanuric acid
exhibits high solubility in solvents such as dimethyl sulfoxide,
dimethyl formamide, and 1,4-dioxane. However, the first two
solvents can be ruled out for CA analysis because of their high
UV cutoffs of 268 nm35 and toxicity. Even though dioxane has a
UV cutoff of 215 nm,35 it is highly toxic and it is not a good solvent
choice versus the popular aqueous buffers used in RPLC.
Although ion pair chromatography appears appealing due to
rapid column equilibration with terbutylamine,36 its toxicity is very
high.37 Additionally, the pH of the eluent cannot be adjusted above
the pH of 8 because tertiary butylamine can undergo base-
catalyzed oxidation to the corresponding N-oxide.36 Alternative and
less toxic ion pair reagents consisting of long-chain tetraalkyl-
ammonium salts require long column equilibration times of several
hours.26 Ion exchange chromatography is not widely used because
it can give complicated chromatograms since the ionic CA has to
compete with other aqueous anions such as halides, chlorides,
hydroxyl, and nitrate. This is important because the nitrate ion
may be present in potable water at concentrations of up to 50
ppm.38 After reviewing Table 1, RPLC was chosen as the separation
technique for developing the method for CA.
Figure 2a shows the chromatogram of a 10 mg/ L CA standard
solution. The initial HPLC conditions employed an octadecyl
column and a 95:5 phosphate buffer/ methanol (%,v/ v) eluent.
The pH of the buffer was ∼6.94 while the detector was fixed
at 213 nm. A theoretical acid-ion pair distribution diagram was
constructed using CA’s first acid dissociation constant, pKa1
)
6.88,1,4,39 and the equilibrium expression Ka1 ) [H+][CA-1]/ [CA].
It is shown in Figure 2c and proved useful in this work. The use
of a high ionic strength buffer is assumed to give a constant ionic
medium, minimizing any influences from the activity coefficients.
The chromatogram in Figure 2a shows a striking splitting of the
CA peak, not previously reported in any of the methods from Table
1. This scenario is certainly a serious limitation for the qualitative
and quantitative analyses of CA where single peak identification
and symmetric peak area calculations are needed for accuracy.
As will be shown, this limitation was suppressed by effectively
controlling the pH of the mobile phase. However, prior to
establishing the effective pH(s), the pair of electronic transitions
(Figure 2b) exhibited by CA in solution were analyzed and the
most sensitive was chosen for UV detection in HPLC.
Electronic P roperties of Cyanuric Acid and Their Implica-
tion in Sensitivity and Retention in HP LC. Figure 2b shows a
typical UV spectrum of a CA solution at the pH ∼7. The spectrum
consist mainly of two absorption peaks, one near 200 nm, while
the other one is at 213 nm. The intensity of the first absorption
peak has been explained from π-π* transitions which originate
through the p overlap of CdO groups (keto tautomer) via the
nonbonding electrons of nitrogen,10 while the intensity of the
second absorption peak has been assigned to n-π* transitions
Figure 2. (a) HPLC, (b) UV spectrum, and (c) ion distribution
diagram of cyanuric acid in solution.
of nonbonding electrons residing on the triazine nitrogens (enol
form).10 The molar absorptivity coefficient (ꢀ) measures the
strength of the chromophore and it is directly responsible for the
observed sensitivity using a UV-visible detector.40 O’Brien4,39
measured ꢀ values for the neutral CA (keto form) and ionic CA
(enol form) which showed that the ionic species is a stronger
absorber of UV light, as evidenced by its molar absorptivity of
8.80 × 103 M-1 cm-1 at λ213(max) compared to 1.83 × 103 M-1 cm-1
at λ196(max) for the acidic keto tautomer. Assignment of the ionic
CA (peak 1) and the nondissociated CA (peak 2) in Figure 2a
does not agree with the ꢀ’s using retention arguments in
RPLC.10-13 There appears to be some uncertainty as to which form
predominates in solution to represent the correct structure of CA
at the pH ∼7. The coexistence of both keto and enol tautomers
in solution has been suggested.4,10-13
Ten of the 12 methods in Table 1 have chosen λ g 210 nm to
enhance the sensitivity of CA derived from the free resonance
conjugation of the enolate form since ꢀ213 . ꢀ196. However, Scho¨ ler
and co-workers31 selected UV detection at 195 nm using a buffer
at the pH ∼1. They reported additional retention of the CA peak
near 5 min compared to the tautomeric peak near 3 min shown
earlier in Figure 2a. A very acidic pH of the eluent can implicate
a drastic shift in the equilibrium of CA in solution. The equilibrium
(35) Snyder, L. R.; Kirkland, J. J.; Glajch, J. L. In Practical HPLC Method
Development, 2nd ed.; John Wiley & Sons: New York, 1997.
(36) Storm, T.; Reemtsma, T.; Jekel, M. J. Chromatogr., A 1 9 9 9 , 854, 175-185.
(37) Material Safety Data Sheet for Tertbutylamine; No. 10090333l; Sigma-Aldrich,
St. Louis, MO, 2000.
(38) Johnson, M.; Melbourne, P. Analyst 1 9 9 6 , 121, 1073-1078.
(39) O’Brien, J. E.; Morris, J. C.; Butler, J. N. In Chemistry of Water Supply,
Treatment, and Distribution; Rubin, A. J., Ed.; Ann Arbor Science: Ann Arbor,
MI, 1975; Chapter 14, pp 333-357.
(40) Cantu´, R.; Stencel, J. R.; Czernuszewicz, R. S.; Jaffe´, P. R.; Lash, T. D. Environ.
Sci. Technol. 2 0 0 0 , 34, 192-198.
Analytical Chemistry, Vol. 72, No. 23, December 1, 2000 5823

Table 2. pH Control Experiments and Cyanuric Acid Peak Characteristics Using HPLC-UVa
acidic pH
k′
peak area (%)
basic pH
k′
peak area (%)
6.88 ( 0.01b
6.03 ( 0.01
4.83 ( 0.01
3.88 ( 0.01
2.97 ( 0.01
1.94 ( 0.01
0.346 ( 0.002b
0.408 ( 0.004
0.471 ( 0.003
0.468 ( 0.002
0.473 ( 0.008
0.478c
40325 ( 2b
14315 ( 2
2921 ( 12
1344 ( 11
1045 ( 14
1123 ( 12
7.13 ( 0.01
7.41 ( 0.01
7.44 ( 0.01
8.07 ( 0.01
9.08 ( 0.01
10.06 ( 0.01
0.284 ( 0.020
0.217 ( 0.001
0.223 ( 0.003
0.134 ( 0.005
0.097 ( 0.001
0.057 ( 0.000
50871 ( 6
56957 ( 2
60245 ( 2
69449 ( 3
76336 ( 5
75559 ( 2
a
Data collected from HPLC with a 10 mg/ L cyanuric acid solution. b Standard deviation for pH (n ) 5), k′s (n ) 3), and peak areas (n ) 5) with
UV detection at 213 nm. c k′ from UV detection at 196 nm due to severe sloping.
in HPLC. Figure 3 (left panel) shows the UV spectrum of a 40
mg/ L CA solution at pH 7.24 (solid line) and the effects of
lowering solution pH on the intensity of the 213-nm absorption
peak. As the pH is lowered from 7.24 to 6.95, there is the first
drastic loss in intensity of this absorption attributed to the
reduction of ion content (aromatic nature) in solution (69 to 55%).
Further acidification to pH 6.59 perpetuated this trend as the ion
content dropped to 33%. Adjustments to pH 5.88 (9%) and 4.12
(0.2%) resulted in additional intensity reduction. At pH 3.40 (0.03%),
the isocyanuric ion peak has been nearly depleted. These UV
solution experiments proved useful for predicting the sensitivity
when the HPLC-UV technique was used; however, they are
inconclusive in explaining the tautomeric splitting of CA near pH
7 in HPLC (Figure 2a). Assignment of structures to these CA
peaks is complicated by the existence of the strong resonance of
both keto and enol tautomers. CA (keto form) with a six-
membered cyclic nonaromatic ring can be converted to cyanuric
ion (enol form) with alternative resonating double bond aromatic
structures. This electromeric polarization is induced in alkaline
solution at pHs greater than 7. Experiments are in progress to
explain this behavior.
The HPLC of a 10 mg/ L CA standard solution is shown in
Figure 3 (right panel). The UV detection was set at 213 nm and
the pH of the mobile phase ranged from 1.94 to 7.41. The k’s and
integrated peak areas with their relative standard deviations are
included in Table 2. A pH of 7.41 yielded good sensitivity (Figure
3a, area 56 957) but it provided the lowest retention (k′ ) 0.217).
As the pH was lowered to 7.13 (Figure 3b), the peak symmetry
became complicated, possibly from keto-enol CA tautomerism.
These influences are suppressed to a larger extent in Figure 3c
where the first peak is nearly gone. Single peak detection is
observed at pH 6.03 (Figure 3d) and other acidic conditions
(Figure 3e-h). A pH of 4.83 exhibits additional retention gains
with k′ ) 0.471 but with the continual loss of peak area (Figure
3e, area 2921). The chromatogram at pH 3.88 (Figure 3f) begins
to display void volume features, which become very dominant at
pH 2.97 (Figure 3g) and 1.94 (Figure 3h). None of these lower
pHs demonstrated any more retention as the k’s were all in the
range of 0.468-0.478 with very low sensitivity. The sensitivity of
CA (Figure 3h, area 1123) is lowered drastically at the benefit of
increased retention (k′ ) 0.408). The lack of reproducibility in
peak areas should be noted by their relative standard deviations
that are above 10% at pHs of <5 (Table 2).
Figure 3. Acidic influences on the UV spectrum of a 40 mg/L
cyanuric acid solution (left panel). Acidic effects on the HPLC of a 10
mg/L cyanuric acid solution (right panel).
can be driven from the enolic tautomer (most stable at neutral
and alkaline pHs) to the keto tautomer (stabilized in acidic pH).
This results in the less polar keto isocyanuric acid receiving more
retention benefits than the more polar enol cyanuric acid using
the nonpolar octadecyl column. Although this strategy proves
more effective in retaining CA, it is predicted to lack sensitivity
since ꢀ213 . ꢀ196. Also, the strong interference of molecular oxygen
near 200 nm can occur at their detection wavelength (Figure 3,
left panel). Low sensitivity can also occur from using a wavelength
distant enough from the optimum 213 nm. For example, Ghiorghis
and Talebian25 carried out HPLC with UV detection at 226 nm
and reported a detection limit of 5 mg/ L.
Electronic Spectra and Basic Influences in HPLC. Figure 4 (left
panel) shows a UV spectrum of a 25 mg/ L CA solution at pH
6.98 (bottom, dotted line) and the effects of increasing solution
Electronic Spectra of Cyanuric Acid with pH Control
Experiments in HP LC. Electronic Spectra and Acid Influences
5824 Analytical Chemistry, Vol. 72, No. 23, December 1, 2000

Figure 4. Basic influences on the UV spectrum of a 25 mg/L
cyanuric acid solution (left panel). Basic effects on the HPLC of a 10
mg/L cyanuric acid solution (right panel).
Figure 5. (a) UV solution spectra of 40 mg/L standard solutions of
cyanuric acid, dichloro-, and trichloroisocyanurates in water. Standard
calibration curves (b) before and (c) after release of residual chlorine.
pH to alkaline conditions on the intensity of the 213-nm electronic
band. The CA peak at pH 6.98 corresponds to 65% ionized
isocyanurate in solution and exhibits increased intensity as the
pH is changed slightly to 7.24, promoting further ionization (79%).
Enhancement is maximized by promoting additional ionization of
CA when the pH is adjusted to 7.81 (90%) and 8.40 (97%) as judged
by the intensity gain of the absorption peak. Note, four methods
in Table 1 used alkaline eluents at pH 7.7 and 7.8, which
correspond to 87 and 89% CA ionization, respectively (Figure 2c).
Three of these methods involved ion-exchange separation mech-
anisms while the fourth was carried out in RPLC using a graphitic
column (Table 1). UV detection was in the 215-220-nm range.
Wavelengths above the optimum 213 nm were possibly chosen
to avoid matrix influences or in anticipation of the second acid
dissociation of CA, which has been reported to produce a
bathochromic shift of the CA peak.39 In the absence of interfer-
ences, the UV spectrum of CA in phosphate buffer revealed no
such influences up to pH 8.40 (Figure 4, left panel). However,
the CA peak shifted to 215 nm in the presence of hypochlorite
ion (discussed in section Reactivity of Dichloro- and Trichloroiso-
cyanuric Acids and shown in Figure 5a).
using UV detection, a strong absorption at 192 nm for the hydroxyl
ion was observed at pH 10. This absorption became more
pronounced and shifted to give a broad and asymmetric peak at
195, 205, and 213 nm at pHs 11, 12, and 13, respectively (not
shown). HPLC was difficult using eluents at these pHs. Such high-
alkaline pHs are to be avoided because alkylation of CA can take
place to form O-alkyl derivatives13 and ring rupture of the triazine
structure has been reported to occur under alkaline conditions.41,42
Also, hypochlorite ion, commonly found in bleach, is known to
1- 43,44
cleave the triazine ring forming N2 and HCO3
.
Therefore,
the useful pH range is 7.2-9.9. A pH of 7.2 is chosen as the lower
limit because of CA tautomerism at 6.8-7.1 (Figures 2a and 3b,c)
and losses in analytical sensitivity at pHs of <6.8 resulting from
the reduction of ion content in solution.
The HPLC of a 10 mg/ L CA standard solution is shown in
Figure 4 (right panel). The measurements were conducted with
UV detection at 213 nm and the pH of the eluent in the range of
7.44-10.06. The k’s and peak areas are included in Table 2. The
peak area of CA is increased by adjusting the pH of the eluent
from 7.44 (Figure 4a, area 60 245) to the more alkaline pHs of
The pHs of 7.7 and 7.8 delineate the limit for ion-exchange
methods for CA analysis because more alkaline eluents can
catalyze ion pair reagents36 and interfere with UV detection by
the presence of the strong absorption of the hydroxyl ion. Careful
examination of Table 1 shows only one method with the mobile
phase at the pH of 10. When solutions of NaOH were analyzed
(41) Frazier, T. C.; Little, E. D.; Lloyd, B. E. J. Org. Chem. 1 9 6 0 , 25, 1944-
1946.
(42) Spencer Chemical Co. Brit. Patent 988,631, April 7, 1965.
(43) Carlson, R. H. to FMC Corp., U.S. Patent 4,075,094, Feb 21, 1978.
(44) Morris, J. C. J. Phys. Chem. 1 9 6 6 , 70 (12), 3798-3805.
Analytical Chemistry, Vol. 72, No. 23, December 1, 2000 5825

Table 3. Optimum HPLC Conditions for Cyanuric Acid
Analysis in Water
mode
reversed-phase HPLC
octadecyl C-18 (5-µm particle size)
25 cm (l) × 0.46 cm (w)
95/ 5 KH2PO4-K2HPO4 buffer:
CH3OH (%, v/ v)
stationary phase
column dimensions
mobile phase
eluent pH range
flow rate
7.2-7.4
1.0 mL/ min
temperature
35 °C
UV
213 nm
recommended dechlorinator
overall method advantages
ascorbic acid
rapid practical sensitivity,
intermediate retention,
reproducibility and low ring
structure reactivity
8.00 (Figure 4b, area 69 449), and 9.08 (Figure 4c, area 76 336).
At this last pH, 99% of CA has been ionized (Figure 2b for
reference) and no further gain in peak area is observed even at
higher pHs, e.g., pH 10 (Figure 4d, area 75559). Although a more
alkaline eluent at pH 9 favors sensitivity, it has low retention
(k′ ) 0.097). This is a disadvantage because many ionic interfer-
ences have low k’s on reversed-phase columns. Additional reten-
tion is gained at pH 8 (k′ ) 0.134) and nearly doubled at pH 7.4
(k′ ) 0.223), providing more retention. An eluent prepared at pH
7.4 is favored over the more alkaline pH of 8 or 9 because of
potential dissolution of the silica support and collapse of the
column bed over time.35 The alternative graphitic carbon column
is more effective in retaining CA; however, it lacks the higher
efficiency (high number of plates) that silica-based columns
possess. The carbon column requires longer equilibration time
and suffers from higher column back pressures. The conditions
for the rapid and simplified HPLC method for the analysis of CA
in water are summarized in Table 3.
Figure 6. (a) TIC of 5 mg/L solution samples of cyanuric acid (peak
1) and reacted dichloro- (peak 2) and trichloro- (peak 3) isocyanurates
in water after 5 min. (b) Selected ion monitoring (SIM) for cyanuric
acid ion at m/z 128 and (b) SIM for dichloro- and trichloroisocyanuric
ions at m/z 196. (d-f) Mass spectra of peaks 1-3 from TIC.
Reactivity of Dichloro- and Trichloroisocyanuric Acids.
Figure 5a shows the UV spectra of 40 mg/ L standard solutions
of CA, DCCA, and TCCA. The spectra are similar in peak position
with only 2-nm red shifts in the DCCA and TCCA. The major
difference is peak intensity, which is ∼50% lower for the chlori-
nated derivatives. HPLC-UV experiments yielded the calibration
plots as those shown in Figure 5b. The calibration curves for
quantitative analyses were Y ) 6319X + 4031 (CA), Y ) 3791X +
386 (DCCA), and Y ) 3527X - 51 (TCCA) with correlation
coefficients r2 g 0.9999. The MDL of CA was determined to be
0.05 mg/ L according to the equation MDL ) t(n-1,1-R)0.99) (S),
where t(n-1,1-R)0.99) ) the Student’s t value appropriate for a 99%
confidence level and a standard deviation estimate with n - 1
degrees of freedom and S is the standard deviation of seven
replicate analyses.45 Other detectable levels of CA are included
in Table 1 for comparative purposes. When either DCCA or TCCA
was injected, a peak was observed with the same retention time
as CA with no clear indication of unique peak identity, even when
eluents at different pHs were used (not shown). Reaction of TCCA
and water to produce CA has been suggested previously.20,21,46
The assumption of stoichiometric reactions of TCCA and DCCA
in water implies production of CA and the release of residual
chlorine. Figure 5c shows good agreement for the assumed
reaction by the good fit the former calibration points from the
DCCA and TCCA regressions have on the original CA plot. The
new calibration plots were Y ) 6460X + 385 (former DCCA) and
Y ) 6350X - 50 (former TCCA) with r2 g 0.9999, and the same
CA plot.
To add proof to these suggestions, FIA-ESI MS analyses were
conducted using solutions of 5 mg/ L CA, DCCA, and TCCA. The
ESI mass spectra (not shown) of CA (MW 129) and DCCA
(monoisotopic MW 197) produced [M - H]- ions at m/ z 128 and
196, respectively. The DCCA also fragmented to give a base peak
at m/ z 162 [M - Cl]-. In the case of TCCA (monoisotopic MW
231), ESI MS produced a spectrum (not shown) identical to DCCA
except that m/ z 196 [M - Cl]-is the base peak instead of m/ z
162 [M - 2Cl + H]-. Figure 6a shows the total ion chromatogram
(TIC) of sequential injections of 5 mg/ L solutions of CA, DCCA,
and TCCA reacted in water for 30 min at room temperature.
Monitoring m/ z 128 (CA) in Figure 6b and m/ z 196 (DCCA and
TCCA) in Figure 6c indicates DCCA and TCCA were converted
to CA. The mass spectra of the three peaks, shown in Figure 6d-
f, confirm the presence of CA and the absence of DCCA and
TCCA.
(45) Glaser, J. A.; Foerst, D. L.; McKee, G. D.; Quave, S. A.; Budde, W. L. Environ.
Sci. Technol. 1 9 8 1 , 15 (12), 1426-1435.
(46) Hammond, B. G.; Barbee, S. J.; Inoue, T.; Ishida, N.; Levinskas, G. J.; Stevens,
M. W.; Wheeler, A.; G.; Cascieri, T. Environ. Health Perspect. 1 9 8 6 , 69,
287-292.
5826 Analytical Chemistry, Vol. 72, No. 23, December 1, 2000

Table 4. Long-Term Reproducibility and Stability of
Cyanuric Acid in Distilled Water at 9.80 mg/La
analysis date
replicate no. (n)
mean ( SDb (mg/ L)
% RSDb
day 1
day 8
day 13
day 43
5
5
3
5
9.75 ( 0.13
9.44 ( 0.05
9.98 ( 0.11
9.81 ( 0.31
1.3
0.5
2.0
3.2
a
The standard solution was kept in a cold room at 5 °C at all times
except for 12-24 h time intervals when HPLC analyses were conducted
and the sample was allowed to equilibrate to room temperature.
b (R)SD, (relative) standard deviation.
Finally, the reproducibility of the method and the stability of
CA proved functional as the relative standard deviations (RSDs)
over 43 days for the analysis of a standard solution containing
9.80 mg/ L of CA were e3.2%, indicating good long-term reproduc-
ibility and sample stability (Table 4). The reproducibility is
comparable to other methodologies20 including the method
developed by Briggle and co-workers29 (7.61% RSD) or Ishiwata
and others (1.8% RSD).27
Dechlorination of Isocyanurates Derivatives. The reactivity
of the DCCA and TCCA to become CA was shown in the section
Reactivity of Dichloro- and Trichloroisocyanuric Acids. However,
water samples containing CA and excess chlorine are ideal for
CA and its chlorinated derivatives to act as chlorine reservoirs
giving rise to 10 potential species in solution.5,47-49 This scenario
adds complexity to the HPLC analysis because the samples need
to be dechlorinated first so the multiple chlorinated isocyanurates
are converted to CA. Failure to do this can possibly lead to lack
of reproducibility affecting the peak shape and the retention time
stability of CA.31 The dechlorination mechanism must (1) reduce
the free available chlorine in the water sample, (2) react with all
chlorinated isocyanurates to produce CA and to release additional
residual chlorine, and (3) further reduce the excess chlorine
formed after reaction. Figure 7 shows the UV spectra of several
dechlorinators used in the EPA drinking water methodology.50
The amount of dechlorination agent should be in excess to ensure
adequate dechlorination of any and all aqueous samples. Since
the proposed method employs UV detection at 213 nm, it is
necessary to employ a dechlorinator having minimum spectral
interference with CA. Sodium sulfite with absorption at 198 nm
exhibits a significant spectral overlap with the CA peak. Also, the
use of sodium sulfite resulted in baseline drift in the chromato-
gram and gave rise to a number of unidentified peaks (not shown).
Sodium thiosulfate is also inadequate due to the spectral matching
to the UV absorption of CA. A fresh solution of ascorbic acid shows
an absorption peak at 265 nm with minimal spectral interference
and is chromatographically compatible with CA (Figure 7b). Old
solutions of ascorbic acid are to be avoided because of its
spontaneous oxidation to dehydroascorbate.51
Figure 7. (a) Spectral compatibility of reductive dechlorinators and
cyanuric acid. (b) Separation of cyanuric acid from ascorbic acid in
HPLC.
CONCLUSION
Overall HPLC conditions for CA analysis in water were
established based upon method preference, rapidity, low toxicity,
and suppression of multiple isocyanurate species with the aid of
the reductive ascorbic acid. The simplified method developed
employs RPLC, which was the most popular separation phase of
the existing methodology. The analysis is rapid with the CA peak
eluting in 3 min. The optimum pH for the mobile phase and UV
detection wavelength were found at the pH range of 7.2-7.4 and
detection at 213 nm, respectively. This pH range is effective in
suppressing CA tautomeric behavior at 6.8-7.1 and avoids the
low sensitivity at pHs of <6.8. It was shown that eluents at pHs
of >7.4 are more sensitive; however, they are limited because of
the lack of retention of CA in the octadecyl column. The analysis
is relatively nontoxic using the popular phosphate buffers. The
method proved reproducible, and the samples were shown to be
stable over 43 days. Finally, the CA method was expanded to
encompass DCCA and TCCA. DCCA and TCCA reacted in water
to give the CA product using the reductive ascorbic acid, which
was found to be spectral and chromatographic compatible to CA.
The method measured CA in water in the 0.5-125 mg/ L range
with an MDL of 0.05 mg/ L.
(47) Matte, D.; Solastiouk, B.; Deglise, A. M. X. J. Can. Chem. 1 9 9 0 , 68 (2),
209-369.
ACKNOWLEDGMENT
(48) Solastiouk, B.; Deglise, A. M. X. J. Can. Chem. 1 9 8 8 , 66, 2188-2193.
(49) Matte, D.; Solastiouk, B.; Deglise, A. M. X. J. Can. Chem. 1 9 8 8 , 67, 786-
791.
The authors are deeply appreciative to Thomas D. Behymer,
Chief of the Chemical Exposure Research Branch, for his
(50) Methods for the Determination of Organic Compounds in Drinking Water,
EPA Document No. 600/ R-95/ 131; Supplement III, EPA, Washington, DC,
1995.
(51) Ru¨ melin, A.; Fauth, U.; Halma´gyi, M. Clin. Chem. Lab. Med. 1 9 9 9 , 37 (5),
533-536.
Analytical Chemistry, Vol. 72, No. 23, December 1, 2000 5827

invaluable contribution to this work. Matthew L. Magnuson from
the National Risk Management Research Laboratory is thanked
for helpful HPLC discussions. This paper was presented at the
32nd Central Regional Meeting of the American Chemical Society
in May 16-19, 2000, in Covington, KY. This paper has been
reviewed in accordance with the U.S. Environmental Protection
Agency’s peer and administrative review policies and approved
for publication. The mention of trade names or commercial
products does not constitute endorsement or recommendation for
use by the US Environmental Protection Agency.
Received for review May 22, 2000. Accepted September
20, 2000.
AC0005868
5828 Analytical Chemistry, Vol. 72, No. 23, December 1, 2000