Real-time detection and identification of aqueous chlorine transformation products using QTOF MS

Article abstract of DOI:10.1021/ac8000989

A screening technique has been developed that allows the rapid, real-time detection and identification of major transformation products of organic contaminants during aqueous oxidation experiments. In this technique, a target contaminant is dissolved in b

Full text of DOI:10.1021/ac8000989

Anal. Chem. 2008, 80, 4193–4199
Real-Time Detection and Identification of Aqueous
Chlorine Transformation Products Using QTOF MS
Brett J. Vanderford,* Douglas B. Mawhinney, Fernando L. Rosario-Ortiz, and Shane A. Snyder
Southern Nevada Water Authority, P.O. Box 99954, Las Vegas, Nevada 89193-9954
A screening technique has been developed that allows the
rapid, real-time detection and identification of major
transformation products of organic contaminants during
aqueous oxidation experiments. In this technique, a target
contaminant is dissolved in buffered water and chlori-
nated by the addition of sodium hypochlorite to give a free
chlorine residual of 3 mg/L. Solution from the reaction
vessel is combined with methanol and pumped directly
into the electrospray ionization source of a quadrupole
time-of-flight mass spectrometer (QTOF MS). The real-
time decay of the target contaminant and the formation/
decay of transformation products are then monitored
using the QTOF MS. Subsequently, accurate mass mea-
surements with internal mass calibration (<5 ppm mass
error) and product ion scans are employed to identify
these transformation products. Unlike other techniques,
it requires no liquid chromatography, derivatization, or
quenching of residual chlorine, all of which can interfere
with transformation product analysis. To validate the
technique, aqueous chlorination experiments were per-
formed on triclosan, a previously studied environmental
contaminant. Earlier research showing that triclosan
underwent chlorine addition to form mono- and dichlo-
rinated transformation products was successfully repro-
duced, demonstrating the feasibility of the technique. In
addition, the technique revealed the formation of a stable
oxygen radical-containing transformation product result-
ing from the oxidation of either mono- or dichlorinated
triclosan. This triclosan transformation product was de-
termined to have an empirical formula of C12H4O3Cl4 with
3.9 ppm mass error. Furthermore, atorvastatin, a com-
monly prescribed medication and environmental contami-
nant, was subjected to aqueous chlorination and studied
with the technique. Atorvastatin underwent hydroxylation
to form two transformation products with the empirical
formulas C33H34FN2O6 (1.8 ppm mass error) and
C26H29O5NF (2.9 ppm mass error).
supplies, their removal by drinking water treatment processes has
been investigated.7–10 Chemical disinfectants, such as chlorine and
ozone, can effectively remove many organic contaminants;10–12
however, removal entails chemical alteration of the contaminant,
producing transformation products. As some research has shown
the potential toxicity of these transformation products,13–19 their
identification and study has been a topic of recent work.
To study the transformation of an organic contaminant during
oxidative treatment, a compound is typically dissolved in a solution
of buffered water to which a disinfectant is added. At various time
intervals, a sample aliquot is taken from the reaction vessel and
quenched with a reducing agent to remove residual oxidant and
prevent further oxidation of the sample until analysis can begin.
The most common quenching agent currently in use for trans-
formation product studies is sodium thiosulfate;13,14,16,20–27 how-
(2) Vanderford, B. J.; Snyder, S. A. Environ. Sci. Technol. 2006, 40, 7312–
7320
.
(3) Snyder, S. A.; Kelly, K. L.; Grange, A. H.; Sovocool, G. W.; Snyder, E. M.;
Giesy, J. P. In Pharmaceuticals and Personal Care Products in the Environ-
ment: Scientific and Regulatory Issues; Daughton, C. G., Jones-Lepp, T. L.,
Eds.; ACS Symposium Series 791; American Chemical Society: Washington,
DC, 2001; pp 116-140.
(4) Snyder, S. A.; Villeneuve, D. L.; Snyder, E. M.; Giesy, J. P. Environ. Sci.
Technol. 2001, 35, 3620–3625.
(5) Richardson, S. D. Anal. Chem. 2007, 79, 4295–4323.
(6) Richardson, S. D.; Ternes, T. A. Anal. Chem. 2005, 77, 3807–3838.
(7) Petrovic, M.; Gonzalez, S.; Barcelo, D. TrAC, Trends Anal. Chem. 2003,
22, 685–696
(8) Westerhoff, P.; Yoon, Y.; Snyder, S.; Wert, E. Environ. Sci. Technol. 2005,
39, 6649–6663
(9) Heberer, T.; Reddersen, K.; Mechlinski, A. Water Sci. Technol. 2002, 46,
81–88
.
.
.
(10) Snyder, S. A.; Wert, E. C.; Lei, H.; Westerhoff, P.; Yoon, Y. Removal of EDCs
and Pharmaceuticals in Drinking Water and Reuse Treatment Processes;
American Water Works Association: Denver, Colorado, 2007.
(11) Huber, M. M.; Canonica, S.; Park, G.-Y.; von Gunten, U. Environ. Sci.
Technol. 2003, 37, 1016–1024
(12) Snyder, S. A.; Wert, E. C.; Rexing, D. J.; Zegers, R. E.; Drury, D. D. Ozone:
Sci. Eng. 2006, 28, 445–460
(13) Shah, A. D.; Kim, J.-H.; Huang, C.-H. Environ. Sci. Technol. 2006, 40, 7228–
7235
(14) Canosa, P.; Morales, S.; Rodriguez, I.; Rubi, E.; Cela, R.; Gomez, M. Anal.
Bioanal. Chem. 2005, 383, 1119–1126
(15) Rule, K. L.; Ebbett, V. R.; Vikesland, P. J. Environ. Sci. Technol. 2005, 39,
3176–3185
(16) Hu, J.-Y.; Cheng, S.; Aizawa, T.; Terao, Y.; Kunikane, S. Environ. Sci. Technol.
2003, 37, 5665–5670
(17) Bedner, M.; MacCrehan, W. A. Chemosphere 2006, 65, 2130–2137
(18) Oliveira, D. P.; Carneiro, P. A.; Rech, C. M.; Zanoni, M. V. B.; Claxton,
L. D.; Umbuzeiro, G. A. Environ. Sci. Technol. 2006, 40, 6682–6689
(19) Buth, J. M.; Arnold, W. A.; McNeill, K. Environ. Sci. Technol. 2007, 41,
6228–6233
(20) Mascolo, G.; Lopez, A.; James, H.; Fielding, M. Water Res. 2001, 35, 1705–
1713
(21) Hu, J.-Y.; Aizawa, T.; Ookubo, S. Environ. Sci. Technol. 2002, 36, 1980–
1987
.
.
.
.
.
Organic contaminants such as pharmaceuticals, personal care
products, and endocrine-disrupting compounds have been the
subject of study due to their presence in surface and ground
waters.1–6 Because many surface waters that contain these
contaminants are used as raw water sources for drinking water
.
.
.
.
* To whom correspondence should be addressed. E-mail: brett.vanderford@
snwa.com. Phone: 702-856-3659. Fax: 702-856-3647.
(1) Vanderford, B. J.; Pearson, R. A.; Rexing, D. J.; Snyder, S. A. Anal. Chem.
.
2003, 75, 6265–6274
.
.
10.1021/ac8000989 CCC: $40.75 2008 American Chemical Society
Published on Web 05/09/2008
Analytical Chemistry, Vol. 80, No. 11, June 1, 2008 4193

ever,sodiumsulfite,15,17,28ascorbicacid,14,28ammoniumchloride,19,23,24
and others have been utilized. Although these chemicals ef-
fectively quench residual oxidants, some have been shown to
negatively affect transformation product determinations. Dodd and
Huang23 reported that transformation products of sulfamethox-
azole could be reduced back to the parent compound by sodium
thiosulfate. Canosa et al.14 demonstrated that both sodium
thiosulfate and ascorbic acid interfered with the determination of
labile transformation intermediates of triclosan. Ascorbic acid was
also shown by Duirk and Collette28 to be problematic during the
study of chlorpyrifos in the presence of aqueous chlorine.
However, these reactions may be compound-specific. Shah et al.13
found that measurements to determine carbadox reaction kinetics
with free chlorine were unaffected by quenching with sodium
thiosulfate. Similarly, Bedner and MacCrehan17 concluded that
acetaminophen transformation products 1,4-benzoquinone and
N-acetyl-p-benzoquinone imine are reduced back to acetaminophen
by sodium sulfite, whereas chloro- and dichloro-4-acetamidophenol
are not.
After the sample has been quenched, it is usually analyzed
for parent compound degradation and transformation product
identification/formation by either liquid chromatography (LC)
coupledwithUVand/ormassspectrometric(MS)detection13,16,17,20–27,29
or gas chromatography (GC) with or without derivatization
coupled with MS.14,15,28 When using LC, a common method
employed to avoid the problem of quenching is to leave the
residual oxidant unquenched and immediately inject an aliquot
of the reaction mixture into an LC system.22–25 However, because
mobile phase modifiers frequently used in LC analysis contain
ammonia (e.g., ammonium acetate, ammonium formate), un-
quenched free chlorine can react to form chloramines.17 Further-
more, pH buffers (such as acetate) used in batch experiments
have been shown to form transformation products in the presence
of free chlorine,13 thereby making the determination of the source
of detected products less obvious. This phenomenon could also
be extended to mobile phase modifiers because many contain the
acetate anion. Therefore, the uncertainty with regard to the effect
of chemical quenching agents, mobile phase modifiers, and pH
buffers on transformation product analyses makes their selection
difficult.
Figure 1. Schematic of experimental setup.
polar or thermally labile compounds is time-consuming, labor-
intensive, and has the potential to degrade chemically unstable
transformation products. LC-MS has been found to be useful for
product identification through product ion analysis; however,
confirmation standards14,15,17 and separate identification tech-
niques such as NMR21,23,26,30 are often necessary to verify
transformation product identities. Liquid chromatography coupled
with quadrupole time-of-flight mass spectrometry (LC-QTOF MS)
has been used recently to successfully study environmental
unknowns31,32 and transformation products of triazine herbicides
in water.33 The main advantages of this technique are higher mass
resolution (>5000) and accurate mass measurements (<5 ppm
mass error) to determine empirical formulas; however, it has not
been attempted for the study of pharmaceutical transformation
products and may be susceptible to the problems associated with
LC modifiers.
The goal of this study was to produce a screening technique
capable of rapidly identifying major organic contaminant trans-
formation products for further study without being susceptible to
the aforementioned limitations. This was made possible by directly
coupling the reaction vessel to the electrospray (ESI) ionization
source of a QTOF MS with the addition of methanol as an
ionization promoter. This technique eliminates the use of LC with
its associated buffers/reagents and the need for quenching the
residual oxidant in the reaction vessel, both of which can lead to
undesirable effects on the chlorination experiments and their
interpretation. Additionally, it provides a real-time technique that
is much faster at identifying major transformation products and
less labor-intensive than traditional analysis techniques. Further-
more, it is capable of determining empirical formulas of major
transformation products using accurate mass measurements and
providing structural information through product ion analysis.
To validate the technique, aqueous chlorination experiments
were performed on triclosan, a commonly used antibacterial agent
found in hand soaps and toothpaste that has been previously
examined using traditional methods. Accurate mass measure-
ments using internal mass calibration and product ion analysis
were employed to detect previously reported major transformation
products of triclosan. In addition, atorvastatin, a commonly
In addition to limitations associated with quenching and LC,
identification of transformation products with traditional analysis
techniques can be challenging. UV-vis detection lacks the mass-
related information afforded by an MS detector and is restricted
to compounds that absorb light between 200 and 800 nm, making
difficult the identification of transformation products that no longer
absorb light in this region (i.e., through loss of conjugation). GC/
MS is primarily used for nonpolar and nonthermally labile
compounds. Derivatization to make GC/MS more amenable to
(22) Pinkston, K. E.; Sedlak, D. L. Environ. Sci. Technol. 2004, 38, 4019–4025
.
(23) Dodd, M. C.; Huang, C.-H. Environ. Sci. Technol. 2004, 38, 5607–5615.
(24) Dodd, M. C.; Huang, C.-H. Water Res. 2007, 41, 647–655.
(25) Dodd, M. C.; Shah, A. D.; Von Gunten, U.; Huang, C.-H. Environ. Sci.
Technol. 2005, 39, 7065–7076.
(30) Mehrsheikh, A.; Bleeke, M.; Brosillon, S.; Laplanche, A.; Roche, P. Water
(26) McDowell, D. C.; Huber, M. M.; Wagner, M.; Von Gunten, U.; Ternes,
Res. 2006, 40, 3003–3014
(31) Ibanez, M.; Sancho, J. V.; Pozo, O. J.; Niessen, W. M. A.; Hernandez, F.
Rapid Commun. Mass Spectrom. 2005, 19, 169–178
(32) Bueno, M. J. M.; Aguera, A.; Gomez, M. J.; Hernando, M. D.; Garcia-Reyes,
J. F.; Fernandez Alba, A. Anal. Chem. 2007, 79, 9372–9384
(33) Ibanez, M.; Sancho, J. V.; Pozo, O. J.; Hernandez, F. Anal. Chem. 2004,
76, 1328–1335
.
T. A. Environ. Sci. Technol. 2005, 39, 8014–8022.
(27) Sandin-Espana, P.; Magrans, J. O.; Garcia-Baudin, J. M. Chromatographia
2005, 62, 133–137.
.
(28) Duirk, S. E.; Collette, T. W. Environ. Sci. Technol. 2006, 40, 546–551
(29) Mascolo, G.; Lopez, A.; James, H.; Fielding, M. Water Res. 2001, 35, 1695–
1704
.
.
.
.
4194 Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

Table 1. Instrument Parameter Settings
parameter
all experiments
parameter
product ions of m/z 336
product ions of m/z 573
source temperature (°C)
pulse duration (µs)
500
10
collision energy (V)
collision gas
-22
5
-25
5
gas 1 - nebulizer gas
gas 2 - turbo gas
40
90
15
curtain gas
ion spray voltage (V)
declustering potential (V)
declustering potential 2 (V)
focusing potential (V)
-2000
-30
-5
-265
prescribed pharmaceutical for the reduction of cholesterol that
has not yet been studied, was subjected to aqueous chlorination
and investigated using the technique.
MS was allowed to stabilize. Once a stable signal was observed,
data collection on the QTOF MS began. Two minutes later, an
aliquot of a 1/10 dilution of the 6% sodium hypochlorite solution
was added to give a final free chlorine concentration of 3 mg/L
as Cl2. Free chlorine concentrations were measured using DPD
colorimetry,34 and free chlorine residuals were found to be
constant throughout the experiments. The reactions were allowed
to progress for 1 h, during which data was collected using the
QTOF MS.
QTOF MS Methodology. The QTOF MS employed in this
study was an Applied Biosystems QStar Elite (Foster City, CA).
This instrument is capable of operating in TOF MS mode and in
QTOF (MS/MS) mode, in which a precursor ion selected by the
quadrupole is fragmented and the product ions analyzed using
the TOF mass analyzer. The instrument was calibrated daily using
a solution of potassium bromate and PFOS in negative electrospray
ionization mode. Although this initial calibration provided mass
accuracy within ±10 ppm mass error over a typical workday, these
experiments were conducted using internal mass calibration. This
procedure produced a calibration for each spectrum collected
using internal calibrant ions and helped correct for calibration
changes caused by environmental factors such as fluctuations in
room temperature. In these experiments, the internal calibrants
were the negative ions generated from the ionization of potassium
bromate and PFOS. A methanol solution of these calibrants was
EXPERIMENTAL SECTION
Materials. Triclosan was obtained from Sigma-Aldrich (St.
Louis, MO), and atorvastatin was from Toronto Research Chemi-
cals (Toronto, Canada). Potassium bromate was purchased from
Grace Davison Discovery Sciences (Deerfield, IL), and perfluo-
rooctanesulfonic acid (PFOS) was from Wellington Laboratories
(Guelph, Ontario, Canada). Sodium bicarbonate and trace analysis
grade methanol were obtained from Burdick and Jackson (Muskeg-
on, MI). Stocks of chlorine were prepared from a commercial
solution of sodium hypochlorite (purified grade 4-6% NaOCl)
from Fisher Scientific (Pittsburgh, PA). Three N hydrochloric acid
was obtained from Red Bird Service (Osgood, IN). Reagent water
was generated from deionized water that had been passed through
a Millipore (Billerica, MA) Milli-Q water purification system.
Experimental Setup. A schematic showing the experiment
setup is shown in Figure 1. Fluid from the reaction vessel was
pumped to a flow splitter at 4 mL/min using an Agilent (Palo Alto,
CA) 1311A pump that had been modified to bypass the solvent
switching valve and pulse dampener. This minimized the amount
of time necessary for the solution to travel from the reaction vessel
to the mass spectrometer. Removing the sample tubing from the
beaker to introduce air into the system demonstrated that the
travel time from the beaker to the QTOF MS was ∼8 s. The flow
was split five ways using a flow splitter to achieve a flow rate of
approximately 900 µL/min after the flow splitter. This was
accomplished by varying the lengths of tubing exiting the flow
splitter that were plumbed to waste. The split flow was then teed
into a flow of 1.5 mL/min of methanol from an Agilent 1312A
pump. The methanol was used to promote ionization of the target
compounds and their transformation products. This combined flow
was directed into the ESI source of the QTOF MS.
Chlorination experiments were performed in a 600 mL glass
beaker containing 500 mL of a 1 mM sodium bicarbonate solution
in reagent water, adjusted to pH 7.0 with 3 N hydrochloric acid.
The pH of the reaction mixture was measured using a pH meter
at the start and end of each experiment to verify that the pH was
constant. Appropriate volumes of triclosan or atorvastatin stock
solutions in methanol were added to the reaction vessel to give a
final concentration of ∼200 µg/L, and the solution was allowed
to mix using a stir bar and stir plate. Blank methanol experiments
were conducted to verify that the amount of methanol added did
not exhibit a chlorine demand (data not shown). Both pumps were
started and the intensity of the target compound on the QTOF
Figure 2. Mass spectrum at 1 min postchlorination showing lock
masses bromate (m/z 127) and PFOS (m/z 499), triclosan (m/z 287),
and formation of mono- (m/z 321) and dichlorinated (m/z 355) triclosan
transformation products. (Nominal masses are shown for clarity.)
Analytical Chemistry, Vol. 80, No. 11, June 1, 2008 4195

Table 2. Parent and Transformation Product Accurate Mass Measurements
accurate mass measurement
proposed empirical
theoretical mass
mass error
(ppm)
compound
(m/z)
formula
(m/z)
Knowns
286.9430
triclosan
C12H6O2Cl3
C12H5O2Cl4
C12H4O2Cl5
C33H34O5N2F
286.9438
320.9049
354.8659
557.2457
2.9
3.7
3.7
3.9
monochlorinated triclosan
dichlorinated triclosan
atorvastatin
320.9037
354.8672
557.2479
Unknowns
335.8907
monochlorinated triclosan radical
monochlorinated triclosan radical product ion
hydroxylated atorvastatin
C12H4O3Cl4
C6HO3Cl2
335.8920
190.9308
573.2406
454.2035
3.9
2.6
1.7
2.9
190.9313
573.2396
454.2022
C33H34O6N2F
C26H29O5NF
hydroxylated atorvastatin with loss of
phenylaminocarbonyl group and
hydroxylated atorvastatin product ion A
hydroxylated atorvastatin product ion B
294.1286
C19H17ONF
294.1299
4.4
added with a syringe pump through a tee ahead of the ESI source
(Figure 1). The flow rate was adjusted between 5 and 25 µL/min
to maintain the intensity of the calibrants and ensure the accuracy
of the mass measurements.
The instrument software allowed the control of a variety of
instrument parameters that affected the intensity of the observed
ions. These included parameters such as the voltage applied to
the electrospray needle, the pressure of nebulizing gas, and the
collision energy applied during MS/MS experiments. The optimal
instrumental settings at the flow rates used in these experiments
were determined prior to data collection and are listed in Table
1. In addition, the rate of data collection used in these experiments
was set to 1 spectrum/s.
experiment and their mass-to-charge ratios bracketed the desired
calibration range.
Triclosan Experiments. Triclosan has been previously stud-
ied under various reaction conditions.14,15,35 These studies agreed
that triclosan reacted with free chlorine or chloramines to initially
form two monochlorinated triclosan derivatives (5,6-dichloro-2-
(2,4-dichlorophenoxy)phenol and 4,5-dichloro-2-(2,4-dichlorophe-
noxy)phenol). Subsequently, a dichlorinated derivative (4,5,6-
trichloro-2-(2,4-dichlorophenoxy)phenol) was formed via further
oxidation of the monochlorinated triclosan. To determine whether
the technique would provide results comparable to those found
by previous researchers, triclosan was selected for study.
As discussed in the Experimental Section, triclosan was spiked
into the buffered reagent water and allowed to equilibrate with
stirring. During this time, the QTOF MS was operated in negative
ESI mode and monitored for triclosan. A large peak for the
deprotonated form ([M - H]-) of triclosan was observed at m/z
286.9430 (Figure 2 and Table 2). Considering the theoretical m/z
of deprotonated triclosan is 286.9438, the mass error was calcu-
lated to be 2.9 ppm. After the signal for triclosan had stabilized,
hypochlorite was added and the QTOF MS was monitored for
triclosan transformation products. Over the next 10 min, two
dominant ion clusters were observed to increase and then
decrease in intensity. Figure 2 shows a mass spectrum corre-
sponding to 1 min postchlorine addition. Accurate mass measure-
ments were performed to obtain mass-to-charge ratios of 320.9037
and 354.8672 for the monoisotopic mass ions of each ion cluster.
Empirical formula calculations on these m/z values yielded
formulas of C12H5O2Cl4 and C12H4O2Cl5, respectively, both with
mass errors of 3.7 ppm (Table 2). These empirical formulas equate
to replacements of hydrogen(s) with chlorine(s) and are the
monochlorinated and dichlorinated triclosan species found by
other researchers.
RESULTS AND DISCUSSION
Experimental Conditions. During initial method develop-
ment, it was found that direct infusion of the reaction solution
into the QTOF MS did not produce enough sensitivity to reliably
make accurate mass measurements of the target compounds and
their transformation products. Methanol and acetonitrile were
tested as ionization promoters by combining their flow with the
flow from the reaction vessel using a tee. After several attempts,
it was determined that syringe pumps were incapable of providing
sufficient backpressure to prevent the solution from the reaction
vessel from being pumped into the body of the syringe being used
for solvent infusion. Therefore, a piston-driven pump was used
instead. After determining a suitable method of solvent introduc-
tion, it was determined that methanol produced greater signal
intensity than acetonitrile for both target compounds and therefore
was chosen for all experiments, even though acetonitrile produced
fewer extraneous background peaks.
To successfully make accurate mass measurements for empiri-
cal formula calculations, a continuously infused internal mass
calibrant (lock mass) was found to be necessary. This allowed
the mass calibration of each mass spectrum to be corrected for
any variation in calibration during an experiment. Due to the
presence of an oxidizing agent in the flow from the reaction vessel,
care had to be taken to find a lock mass that would not be oxidized
under those conditions. Initial attempts to use iodide as a lock
mass failed because it was found that iodide was oxidized by the
residual chlorine and consequently was unreliable. Therefore,
bromate and PFOS were chosen because they were unlikely to
be further oxidized under the conditions of the chlorination
In contrast to previous research, the degradation products of
the major monochlorinated and dichlorinated triclosan transforma-
tion products were not detected by this technique with sufficient
intensity to perform empirical formula calculations. In general,
as contaminants degrade, they tend to lose mass and become
more volatile, making them less suitable for LC-MS analysis. As
a result, they may become less amenable to detection by this
technique, and in the case of triclosan, degradation products such
4196 Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

Figure 4. Proposed structures for triclosan radical (left) and triclosan
radical product ion (right).
Figure 3. Mass spectrum of 2 min prechlorination subtracted from
10 min postchlorination showing an unknown transformation product
at m/z 336. The four-chlorine isotope pattern is shown for comparison.
(Nominal masses shown for clarity: (A) 5,6-dichloro-2-(2,4-dichlo-
rophenoxy)phenol; (B) 4,5-dichloro-2-(2,4-dichlorophenoxy)phenol;
(C) 4,5,6-trichloro-2-(2,4-dichlorophenoxy)phenol.)
as 2,4-dichlorophenol and chloroform that were detected in
previous studies are less likely to be identified.
To elucidate possible products of triclosan that had not been
previously published, the mass spectral data were further analyzed.
Mass spectra from the first 2 min of data collection (prior to
chlorination) were subtracted from the mass spectra collected
during the first 10 min after chlorination. This revealed the mono-
and dichlorinated transformation products mentioned previously
but also showed the presence of an unknown product at m/z
335.8907 (Figure 3). When the relative isotope ratios of a
hypothetical compound with four chlorines (inset in Figure 3) are
compared with those of the products, it was evident that the
unknown compound contained four chlorine atoms. In addition,
it was noted that the nominal mass of the compound was even.
According to the nitrogen rule, an even-numbered mass must
either have (1) an odd number of nitrogens and an even number
of electrons or (2) an even number of nitrogens and an odd
number of electrons. Because triclosan has no nitrogens and ESI
produces even-electron ions, it was concluded that the unknown
transformation product must have either gained a nitrogen atom
or contained an odd number of electrons (i.e., be a radical). With
the use of an empirical formula calculator, it was determined that
the unknown product was an odd-electron ion having a molecular
formula of C12H4O3Cl4 with a mass error of 3.9 ppm (Table 2).
This mass accuracy is in agreement with the previously discussed
measurements of triclosan and the mono- and dichlorinated
triclosan products. Furthermore, the substitution of nitrogen into
the formula yielded mass errors of >25 ppm (data not shown),
which is consistent with there being no obvious source of reactive
nitrogen in the system. The formula of C12H4O3Cl4 is therefore
consistent with the replacement of two hydrogen atoms by a
chlorine atom and an oxygen radical (Figure 4).
Figure 5. Plot of decay of triclosan (A) and formation/decay of 5,6-
dichloro-2-(2,4-dichlorophenoxy)phenol and 4,5-dichloro-2-(2,4-dichlo-
rophenoxy)phenol (B), 4,5,6-trichloro-2-(2,4-dichlorophenoxy)phenol
(C), and triclosan radical (D).
zero) and indicating the molecule had fragmented and rearranged
to yield an even-electron ion (Figure 4). The molecular formula
suggests that the oxygen radical was added to the hydroxylated
side of triclosan because it contains three oxygen atoms, with the
most likely position being para to the hydroxyl group.
With the use of this technique, the intensity of an organic
contaminant and its identified transformation products can also
be plotted over the course of an experiment, as shown in Figure
5. These data can be used to determine the sequence of oxidation
events with a high degree of resolution. In contrast to techniques
in which a small number of discrete samples are periodically taken
over the course of an experiment and analyzed, this technique
allows the acquisition of 3600 data points/h. This greater data
resolution can lead to a better understanding of the sequence of
events during the reaction.
In the case of triclosan, it can be seen in Figure 5 that the
formation of monochlorinated triclosan begins first and dichlori-
nated triclosan formation begins shortly thereafter, confirming the
reaction sequence proposed by Rule et al.15 The monochlorinated
triclosan radical, however, begins to form approximately 1 min
after the mono- and dichlorinated triclosan species, coinciding with
the maxima of the monochlorinated triclosan peak. Figure 5 also
shows that the mono- and dichlorinated triclosan species quickly
rise in intensity, reach a maximum, and rapidly degrade. Con-
versely, the intensity of the monochlorinated triclosan radical
slowly rises during the first several minutes of the reaction and
then stabilizes over the first 15 min of the experiment (Supporting
Information Figure S1).
To further study this radical, product ion experiments were
initiated. The radical was fragmented in the collision cell, and a
major product ion was found at m/z 190.9313. Because the nominal
mass was now odd, this indicated that the product ion most likely
had both an even number of nitrogens and electrons. The
molecular formula was determined to be C6HO3Cl2 with a mass
error of 2.6 ppm, confirming an even number of nitrogens (i.e.,
Atorvastatin Experiments. Although atorvastatin has been
detected in surface water,2 its potential to form transformation
products has not been previously examined. After spiking atorv-
Analytical Chemistry, Vol. 80, No. 11, June 1, 2008 4197

The product ion at m/z 294.1286 resulted in a molecular
formula of C19H17ONF with a mass error of 4.4 ppm. As this
indicates a loss of C7H12O4 from the product ion at m/z 454.2022,
the only possible neutral loss is that of the heptanoic acid moiety.
These product ion experiments suggest that the hydroxyl group
is added either to the pyrrole group, resulting in the substitution
of the phenylaminocarbonyl moiety, or the phenyl or fluorophenyl
rings that are bonded to the pyrrole group; however, the exact
site of hydroxyl addition was not determined.
Limitations. As discussed above, transformation products tend
to become less massive and more volatile over time as they
degrade during oxidation. This poses a challenge for analysis by
a single technique as most are suitable for only a limited set of
chemical properties. Hence, as contaminants degrade, they are
less likely to be detected and identified by this technique as QTOF
MS is typically more suitable for higher molecular weight,
nonvolatile compounds. In addition, only compounds that are
ionizable using ESI will be detected by the QTOF MS. Depending
on the changes that occur to the functional groups of the target
compound during oxidation, there is the possibility that its
transformation products will no longer be ionizable by ESI. This
is unlikely to prevent the identification of the initial transformation
products; however, it may affect the ability of the QTOF MS to
detect certain products further along in the pathway as compounds
are progressively degraded. Although it was outside the scope of
this paper, additional ionization techniques such as atmospheric
pressure chemical ionization (APCI) and atmospheric pressure
photoionization (APPI) have the potential to be used as alternatives
to ESI. Another limitation of the technique is its inability to
differentiate between isomers. As a screening technique, it will
be able to quickly identify major transformation products; however,
to determine the number of isomers, another technique such as
fraction collection with chromatographic separation will be
necessary.
Figure 6. Plot of decay of atorvastatin (A) and formation/decay of
its two transformation products at m/z 573 (B) and 454 (C) during
the first 10 min of reaction. Due to its large initial intensity, atorvastatin
is only partially displayed.
astatin into the reaction vessel, the QTOF MS was monitored for
its presence. It was found to deprotonate in negative mode yielding
an [M - H]- ion at m/z 557.2479 with a mass error of 3.9 ppm
(Table 2).
After the addition of hypochlorite, the intensity of atorvastatin
rapidly decreased while the intensity of two other ions at m/z
573.2396 and 454.2022 initially increased and then slowly de-
creased over the course of the experiment (Figure 6). Elemental
formula calculations on the more intense of the two ions at m/z
573.2396 generated an empirical formula of C33H34O6N2F with a
mass error of 1.7 ppm (Table 2). This formula differs from that of
atorvastatin by the net gain of one oxygen atom, probably via the
replacement of a hydrogen atom with a hydroxyl group. The
empirical formula of the second major ion at m/z 454.2022 was
determined to be C26H29O5NF with a mass error of 2.9 ppm. This
formula suggests the hydroxylation of atorvastatin via the substitu-
tion of C7H5ON, which, due to the molecular structure of
atorvastatin, can only result from a loss of the phenylaminocar-
bonyl group. From an enlarged version of Figure 6 (Supporting
Information Figure S2), it can be seen that the two products form
simultaneously, although the rate of increase is different.
A product ion scan was performed on the m/z 573 transforma-
tion product to attempt to determine the most likely location of
the hydroxyl addition. The fragmentation yielded product ions at
m/z 454.2022 and 294.1286 (Figure 7 and Table 2). The product
ion at m/z 454.2022 was identical to that of the main transformation
product at m/z 454, suggesting the phenylaminocarbonyl bond
is relatively weak. In addition, two previous fragmentation studies
using LC-MS/MS showed that atorvastatin in positive ionization
mode produced a fragment at m/z 440,2,36 corresponding to m/z
438 in negative mode and m/z 454 after oxidation. These studies
concluded that m/z 440 was due to the loss of the phenylami-
nocarbonyl group, adding further evidence of the tendency of the
molecule to lose that group.
CONCLUSIONS
The screening technique described above provides a rapid
method for the real-time identification of major transformation
products of organic contaminants during chlorination reactions.
It requires no quenching of the reaction, liquid chromatogra-
phy, or derivatization, all of which can interfere with transfor-
mation product analysis. The technique was able to detect and
identify two major triclosan transformation products and was
used to identify novel transformation products of triclosan and
atorvastatin, including a radical that probably would not be
detected by other techniques due to its reactivity. In addition,
the technique yielded a high degree of resolution for the
sequence of oxidation events due to the large number of data
points collected. It is anticipated that the information from these
data can prevent the incorrect assignment of formation maxima
that may result from an inadequate number of data points and
lessen the chance of not observing a short-lived transformation
product.
(34) Clesceri, L. S., Greenberg, A. E., Eaton, A. D., Eds. Standard Methods for
the Examination of Water and Wastewater, 20th ed.; APHA, AWWA, WEF:
Washington, DC, 1998.
It is also expected that the technique can be used to provide
information on chemical kinetics and oxidative reaction mech-
anisms. To accomplish this, the intensity of the parent would
need to be related to its concentration. This would allow the
determination of the rate constant for the reaction of the parent
(35) Greyshock, A. E.; Vikesland, P. J. Environ. Sci. Technol. 2006, 40, 2615–
2622
(36) Jemal, M.; Ouyang, Z.; Chen, B.-C.; Teitz, D. Rapid Commun. Mass Spectrom.
1999, 13, 1003–1015
.
.
4198 Analytical Chemistry, Vol. 80, No. 11, June 1, 2008

Figure 7. Product ion spectrum of hydroxylated atorvastatin showing product ions at m/z 454 and 294. Product ion structures shown for m/z
573, 454, and 294. (Nominal masses are shown for clarity.)
compound with the oxidant. Although it was outside the scope
of this paper, initial attempts at using the technique for this
purpose were promising; however, the degree of ionization
suppression caused by the addition of hypochlorite will be an
important consideration.
Additionally, it is likely that the technique can be used to detect
and identify the major transformation products that result from
the reaction of organic contaminants with other chemical oxidants
such as ozone, chloramines, and chlorine dioxide, as well as ozone
and UV-based advanced treatment processes. The only require-
ment of the technique is for the fluid in the reaction vessel to be
directly transferable from the vessel to the mass spectrometer.
Because most bench-scale work for these oxidants is performed
in reaction vessels similar to those used in this experiment,
adapting the technique to use a different oxidant should be
relatively simple.
ACKNOWLEDGMENT
The authors thank Eric Wert and Rebecca Trenholm of the
Southern Nevada Water Authority (SNWA) for their assistance
with experimental design and analytical support.
SUPPORTING INFORMATION AVAILABLE
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
Received for review January 14, 2008. Accepted April 9,
2008.
AC8000989
Analytical Chemistry, Vol. 80, No. 11, June 1, 2008 4199