Identification of 1,3-dioxanes and 1,3-dioxolanes as malodorous compounds at trace levels in river water, groundwater, and tap water

Article abstract of DOI:10.1021/es9704085

A study of organic compounds imparting odor problems in river waters and groundwaters has been conducted. The Tordera aquifer located in Barcelona and Girona (NE Spain) is the water supply reserve for many seasonally crowded villages on the coast. Closed loop stripping analysis (CLSA) and flavor profile analysis (FPA) have been employed as analytical tools to identify the compounds responsible for the odor complaints. The feasibility of purge-and- trap (P and T) has also been evaluated. The 2-alkyl-5,5-dimethyl-1,3-dioxanes and 2-alkyl-4-methyl-1,3-dioxolanes were the most significant compounds identified in river water and groundwater with a threshold odor of 10 ng/L for 2-ethyl-5,5-dimethyl-1,3-dioxane (2EDD), the most malodorous compound. The analyses were carried out by HRGC/MS, and the synthesized 1,3-dioxanes and dioxolanes were characterized by CI-MS and EI-MS/MS techniques. A company, currently manufacturing saturated and unsaturated polyester resins, located in the upper course of the river, produced these compounds as byproducts during the synthesis of resins. The pollution by dioxanes and dioxolanes affected all the aquifer and slowly diminished to the ppt levels when the company was forced to correctly treat their wastewaters. Additional examples of the presence of dioxanes and dioxolanes in wastewaters of other resin plants and also tap water of Barcelona are shown. A study of organic compounds imparting odor problems in river waters and groundwaters has been conducted. The Tordera aquifer located in Barcelona and Girona (NE Spain) is the water supply reserve for many seasonally crowded villages on the coast. Closed loop stripping analysis (CLSA) and flavor profile analysis (FPA) have been employed as analytical tools to identify the compounds responsible for the odor complaints. The feasibility of purge-and-trap (P&T) has also been evaluated. The 2-alkyl-5,5-dimethyl-1,3-dioxanes and 2-alkyl-4-methyl-1,3-dioxolanes were the most significant compounds identified in river water and groundwater with a threshold odor of 10 ng/L for 2-ethyl-5,5-dimethyl-1,3-dioxane (2EDD), the most malodorous compound. The analyses were carried out by HRGC/MS, and the synthesized 1,3-dioxanes and dioxolanes were characterized by CI-MS and EI-MS/MS techniques. A company, currently manufacturing saturated and unsaturated polyester resins, located in the upper course of the river, produced these compounds as byproducts during the synthesis of resins. The pollution by dioxanes and dioxolanes affected all the aquifer and slowly diminished to the ppt levels when the company was forced to correctly treat their wastewaters. Additional examples of the presence of dioxanes and dioxolanes in wastewaters of other resin plants and also tap water of Barcelona are shown.

Full text of DOI:10.1021/es9704085

Environ. Sci. Technol. 1998, 32, 206-216
to water implies that it is unsafe nor the absence of
organoleptic effects guarantee the absence of undesirable
pollutants (1).
Identification of 1,3-Dioxanes and
1,3-Dioxolanes as Malodorous
The sources of contaminants producing taste and odors
in water can be attributed to both natural and anthropogenic
compounds. Geosmin, 2-methyl isoborneol (MIB), trans-
2-cis-6-nonadienal, 2,3,6-trichloroanisole, dimethyl polysul-
fides, and isopropyl methoxy pyrazine, among others, are
the most frequently reported compounds of biological origin
causing off-flavors in water at nanogram per liter levels (2-
8). Excellent reviews on this subject have been published (9,
10). The AWWA taste and odor survey conducted in 1989
showed that the major odor problems reported by the utilities
were characterized as chlorine, earthy, fishy, and medicinal,
whereas taste problems were reported as sour, metallic, and
chlorine (11). Disinfection byproducts formed in treatment
plants such as iodinated trihalomethanes (12), short-chain
aldehydes or fatty acids (13, 14), and others have been
identified as sources of taste and odor episodes. Leaching
chemicals such as phenolic antioxidants from polyethylene
pipes (15) or pipe joint lubricants used during the installation
of plastic and ductile iron water mains (16) may also impart
tastes or odors to water. Recently, several taste and odor
threshold levels of common water contaminants have been
reported (17). Industrial and sewage effluents and leachates
from poor waste disposals or landfills are, among others,
frequently mentioned as the source of tastes and odors from
anthropogenic compounds (18, 19).
Closed loop stripping analysis (CLSA) according to the
method first described by Grob (20) and others (21, 22) is by
far the most reported analytical tool in the literature used to
identify the compound or compounds responsible for taste
and odor episodes. The CLSA technique is suitable for the
analysis at nanogram per liter levels and can be used in
combination with sensory GC techniques to discriminate
compounds in complex chromatograms by their odor.
Recently, this combination (CLSA and sensory GC) has
allowed the identification of dicyclopentadiene, with an
estimated threshold odor level of 10 ng/ L, as the compound
causing odor problems in groundwater (23).
Since October 1993, odor incidents had been noticed in
river water and groundwater wells in the Tordera aquifer
that is located in the Barcelona-Girona area (NE Spain).
Many of these wells are used for drinking purposes in
important tourist villages located near and along the coast.
Preliminary analytical data showed the presence of antioxi-
dants, aromatic solvents, and chlorinated solvents that due
to their low concentration levels were discarded as the
potential origin of the odor event as were a large number of
unidentified compounds with no equivalents in mass spectral
libraries. These compounds were further identified as 2-alkyl-
5,5-dimethyl-1,3-dioxanes and 2-alkyl-4-methyl-1,3-dioxo-
lanes (see acronyms in Table 1) after comparison of their
mass spectra with the products published by Cocheo et al.
(24) in the air around a polyester resin plant and by Preti et
al. (25), who identified 2-ethyl-5,5-dimethyl-1,3-dioxane
(2EDD) as the malodorous compound in the treated water
of Philadelphia.
The 2-alkyl-5,5-dimethyl-1,3-dioxanes, 2-alkyl-4-methyl-
1,3-dioxolanes, and 2-alkyl-1,3-dioxolanes (see Figure 1) are
byproducts present in wastes from resin manufacturing
plants that use glycols (i.e., neopentyl gycol, polypropylene
glycol, ethylene glycol) as raw material for acid-catalyzed
condensations. The presence of these compounds in water
has only been reported to our knowledge in the incident of
Philadelphia (25-28), Barcelona (29), groundwater (30-32),
Compounds at Trace Levels in River
Water, Groundwater, and Tap Water
J O R D I R O M E R O A N D
F R A N C E S C V E N T U R A *
AGBAR, Aigu¨ es de Barcelona, P. Sant Joan 39,
08009-Barcelona, Spain
J O S E P C A I X A C H A N D J O S E P R I V E R A
Mass Spectrometry Laboratory, Department of
Ecotechnologies, CID-CSIC, J. Girona 18,
08034-Barcelona, Spain
L L U IÄ S X A V I E R G O D EÄ A N D
J O S E P M A . N I N˜ E R O L A
Junta de Sanejament, Generalitat de Catalunya,
Provenc¸a 204, 08036-Barcelona, Spain
A study of organic compounds imparting odor problems
in river waters and groundwaters has been conducted. The
Tordera aquifer located in Barcelona and Girona (NE
Spain) is the water supply reserve for many seasonally
crowded villages on the coast. Closed loop stripping analysis
(CLSA) and flavor profile analysis (FPA) have been employed
as analytical tools to identify the compounds responsible
for the odor complaints. The feasibility of purge-and-
trap (P&T) has also been evaluated. The 2-alkyl-5,5-dimethyl-
1,3-dioxanes and 2-alkyl-4-methyl-1,3-dioxolanes were
the most significant compounds identified in river water and
groundwater with a threshold odor of 10 ng/L for 2-ethyl-
5,5-dimethyl-1,3-dioxane (2EDD), the most malodorous
compound. The analyses were carried out by HRGC/MS,
and the synthesized 1,3-dioxanes and dioxolanes were
characterized by CI-MS and EI-MS/MS techniques. A
company, currently manufacturing saturated and unsat-
urated polyester resins, located in the upper course of the
river, produced these compounds as byproducts during
the synthesis of resins. The pollution by dioxanes and di-
oxolanes affected all the aquifer and slowly diminished
to the ppt levels when the company was forced to correctly
treat their wastewaters. Additional examples of the pres-
ence of dioxanes and dioxolanes in wastewaters of other
resin plants and also tap water of Barcelona are shown.
Introduction
Water companies devote considerable time and effort to the
identification and treatment of compounds causing odors
in water supplies. Consumer complaints concerning taste
and odor problems deal with the public perception of an
unsafe drinking water. Nevertheless, it must be noted that
neither the presence of compounds imparting taste and odors
* Author to whom correspondence should be addressed; fax: 34-
3-474-32-37.
9
2 0 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998
S0013-936X(97)00408-2 CCC: $15.00
1998 Am erican Chem ical Society
Published on Web 01/15/1998

TABLE 1. Mass Spectra (EI, CI) of the Main Dioxanes and Dioxolanesa
compound
ions (m/z) (% abundance)
2-Alkyl-5,5-dimethyl-1,3-dioxanes (EI)
R ) H (DMD)
) Me (TMD)
56(100); 41(32); 57(12); 86(9); 115(9) [M - H]+; 69 (8)
56(100); 41(31); 115(23); 43 (23); 69 (19); 129(14) [M - H]+
56(100); 115(68); 69(48); 41 (45); 57(41); 143(9) [M - H]+
56(100), 115(79); 41(79); 43(79); 69(73); 157(6) [M - H]+
115(100); 56(81); 69(80); 41 (50); 85(32); 171(9) [M - H]+
115(100); 56(88); 69(76); 41 (53); 85(31); 171(9) [M - H]+
) Et (2EDD)
) i-Pr (2iPMD)
) i-But (2iBMD)
) But (2nBMD)
2-Alkyl-5,5-dimethyl-1,3-dioxanes (CI)b
R ) H (DMD)
) Me (TMD)
) Et (2EDD)
) i-Pr (2iPMD)
) But (2nBMD)
117(100)[M + H]+; 69(35); 99(25); 115(25); 87(18)
67(100); 77(67); 83(59); 69(56); 79(49), 131(26) [M + H]+
87(100); 115(73); 145(62) [M + H]+; 69(43); 85(21)
115(100); 87(98); 159(92) [M + H]+; 69(76); 71(25)
117(100); 72(85); 87(55); 145(14); 173(12) [M + H]+
2-Alkyl-4-methyl-1,3-dioxolanes (EI)
R ) H
87(100)[M - H]+; 44 (99); 58(57); 45 (27); 43 (27); 59(19)
87(100); 43(78); 58(53); 59(46); 41(32); 101(8)[M - H]+
87(100); 59(67); 41(43); 57(39); 72(20); 115(2) [M - H]+
87(100); 59(41); 41(32); 56(20); 43(16); 129(2) [M - H]+
87(100); 59(36); 41(23); 85(12); 43(9); 143(2) [M - H]+
87(100); 59(35); 41(22); 71(10); 57(8); 143(2) [M - H]+
) Me
) Et (2E4MDL)
) i-Pr
) i-Bu
) Bu
2-Alkyl-1,3-dioxolanes (EI)
R ) H
) Me
) Et
73(100) [M - H]+; 44 (48); 45(22); 43(19)
73(100); 43(73); 45(70); 58(30); 87(10)[M - H]+
73(100); 45(32); 57(10); 42(7); 43(6); 101(4) [M - H]+
73(100); 45(24); 43(8); 56(7); 41(7); 115(2) [M - H]+
) i-Pr
a
b
Acronym s of dioxanes and dioxolanes are in parentheses. Only m /z >50 am u are represented.
unconfined deposits continue into the delta forming a lightly
confined aquifer, which is separated from the upper un-
confined layer by a thin silty-sand formation. The system
thickness reaches 50 m near the sea. The aquifer permeability
ranges from 170 to 400 m/ d, and the specific flow of the deep
aquifer varies from 10 to 43 L s-1 m-1
.
Figure 2 shows the geographical situation of the 48
sampling sites and the four river water samples analyzed.
Twelve wells (4-8, 13, 15-18, 24, and 30) are used by
industries; eight of them (23, 42, 45, 54, 60, 62, 66, and 71)
are used for agricultural uses; two are piezometers (14 and
64), and the rest are used for drinking water purposes. The
water demand can be evaluated as an average in 30, 36, and
55 × 103 m3/ d for agricultural, industrial, and drinking water,
respectively.
FIGURE 1. (a) 2-Alkyl-5,5-dimethyl-1,3-dioxanes; (b) 2-alkyl-4-methyl-
1,3-dioxolanes; (c) 2-alkyl-1,3-dioxolanes.
and the effluent from a sewage treatment plant of a fibreglass
and resin manufacturing company (33).
The aim of this paper is to elucidate the origin of the
contamination of the Tordera aquifer by dioxanes and
dioxolanes, its spreading along the river and groundwater,
and also to assess its impact on the water quality during 2
yr. In addition, it is also shown for some examples how the
unappropriate treatment by a waste management facility
for industrial wastewaters coming from a resin plant can be
a source of odor complaints in tap water. The analytical
methodology consisted of CLSA and GC/ MS identification
of odorus compounds although the feasibility of purge-and-
trap (P&T) was also proved. The synthesis of the main
dioxanes and dioxolanes and their characterization by CI-
MS and EI-MS/ MS techniques was also performed to
accurately quantify their presence in water.
Glassware. Amber glass bottles were washed with soapy
water and rinsed with tap water, with Milli-Q water, and
finally with reagent-grade acetone. Then, they were air-dried
and baked at 180 °C for 1 h. After this procedure, their mouths
were covered with aluminum foil to prevent the presence of
dust or other contaminants.
Sam pling. The samples from different wells (see Figure
2) were collected after water had run for several minutes in
order to eliminate the stagnant water. Amber glass bottles
(1 L) were rinsed several times with the water to be analyzed
and were afterwards filled until overflow to prevent the loss
of volatile compounds by the presence of headspace. The
water samples were transported and stored at 4 °C until their
analysis within 24-48 h.
Experimental Section
Description of the Site. The Tordera River is located 65 km
northeast from Barcelona. From its source, it flows southeast
for 60 km draining palezoic schists and granites and
impermeable tertiary formations to the Mediterranean Sea.
Its alluvial aquifer is the main water supply for the sur-
rounding population (140 000 inhabitants in winter; 500 000
in summer) as well as for industries (mainly chemical plants)
and agricultural areas. The aquifer system is formed by
alluvial quaternary sediments. The middle and lower valley
(500-1000 m wide) is filled with sand and gravel (5-30 m
thick) laying on impermeable materials. These permeable
The frequency of sampling for wells was initially every 2
weeks until June-July 1994 then was monthly until November
1994, and finally was bimonthly. Once factory A was
identified as the origin of the odor events, the frequency of
sampling for its wastewaters was weekly until they changed
their wastewater treatment. After that, for several months,
twice a week and then weekly samples from wastewaters of
company A were analyzed until no significant variations in
the concentration levels of dioxanes and dioxolanes were
measured.
9
VOL. 32, NO. 2, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 2 0 7

FIGURE 2. Study area and sampling site locations in the Tordera aquifer. Groundwater samples are numbered, and river water samples
are encircled. Wells 4-8, 13, 15-18, 24, and 30 are used by industries; wells 23, 42, 45, 54, 60, 62, 66, and 71 are used for agricultural
purposes; wells 14 and 64 are piezometers, and the rest are used for drinking water purposes.
River water samples from four different points taken along
the river course (see Figure 2) were collected as described
above for the wells. The unfiltered water samples were
transported and stored at 4 °C until analysis. The frequency
of sampling was irregular, but a set of samples was analyzed
every 6 months.
(J&W Scientific, Folsom, CA) was used for drinking water
samples containing high levels of trihalomethanes (i.e.,
chlorodibromomethane and bromoform) to avoid coelution
with DMD and 2EDD, respectively.
P&T analyses were carried out on a Fisons MD800 mass
spectrometer equipped with a Fisons 8060 gas chromato-
graph. Extracts were transferred directly to the chromato-
graph (transfer line 180 °C) onto a 75 m × 530 µm i.d. (3 µm
film thickness) DB 624 fused silica column (J&W Scientific,
Folsom, CA). The GC temperature program was 30 (5 min)
to 275 °C (10 min) at a rate of 3 °C/ min. Helium was used
as the carrier gas, and the linear velocity was set at 71 cm/ s
at 30 °C. The mass spectrometer was operated in electron
impact mode (70 eV). The source temperature was main-
tained at 200 °C. Mass spectra were acquired by scanning
from 35 to 450 Da with 1 s/ decade.
GC/ MS analyses were carried out on a VG Trio-1000 mass
spectrometer equipped with a Fisons 8060 gas chromato-
graph. The same chromatographic conditions as described
above for CLSA extracts were used. The mass spectrometer
was operated in electron impact mode (70 eV). Transfer line
and ion source temperatures were maintained at 200 and
250 °C, respectively. Mass spectra were acquired by scanning
from 35 to 450 Da with 1 s/ decade.
GC/ CI-MS analyses were performed on a VG TS-250 (VG
Instruments, U.K.) double-focussing mass spectrometer
coupled with a Konik (S. Cugat, Spain) KNK-3000 gas
chromatograph. We employed the same chromatographic
conditions reported before. The chemical ionization oper-
ating conditions were as follows: ion source and interface
temperatures, 180 and 250 °C, respectively; ionization energy,
45 eV; trap current, 190 µA. Isobutane was used as reagent
gas at an ion source manifold pressure of 1 × 10-4 mbar. The
resolution was kept at 1000, and the mass spectra were
acquired by scanning from 60 to 300 at 1 s/ scan and a 0.2-s
delay time.
Closed Loop Stripping Analysis (CLSA Method). Analyses
were carried out in a commercial CLSA apparatus (Brech-
bu¨ ler, Switzerland) according to the method developed by
Grob (20). Water samples (1 L or diluted with Milli-Q) were
spiked with 1-chloroalkanes (C5, C6, C10, C12, C16, and C18) to
give a final concentration of 800 ng/ L for each compound.
We used 5-mg activated carbon filters. The samples were
stripped for 1 h. Temperatures of 45 and 55 °C were used
for water bath and carbon filter, respectively. After stripping,
the carbon filters were spiked with C8 and C14 1-chloroalkanes
at the same concentrations of the spiked water samples. The
filters were then extracted with 40 µL of CS2 (20 + 20 µL).
Purge-and-Trap Method. A commercial Tekmar 2000
apparatus (USA) was used. Volatile compounds in 5 or 25
mL were purged for 11 min with helium (40 mL/ min) and
adsorbed onto a Tenax column at room temperature.
Organic compounds were desorbed at 180 °C during 4 min
and transferred directly to the GC/ MS system. Fluoroben-
zene, 1-bromo-4-fluorobenzene, and 1,2-diclorobenzene-
d4 (Aldrich, Germany) were used as internal standards.
Instrum ental Conditions. Analyses of CLSA extracts were
carried out on a Fisons 8560 (U.K.) gas chromatograph
equipped with a FID detector. Injections (1 µL) were made
cold on-column into a 50 m × 320 µm i.d. CP-Sil 19CB
(0.25 µm film thickness) fused silica column (Chrompack,
The Netherlands). A deactivated precolumn (2 m) was used.
The GC temperature program was 30 °C (5 min) to 280 °C
(10 min) at a rate of 3 °C/ min. Helium was the carrier gas,
and the linear velocity was set at 31 cm/ s at 30 °C. Nitrogen
was used as the make up gas (125 kPa). A DB-5 column
9
2 0 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998

EI-MS/ MS analyses were performed on an AutoSpec-Q
hybrid (EBEqQ) mass spectrometer (VG Instruments, U.K.)
and the operating conditions were the following: ion source
temperature, 200 °C; trap current, 300 µA; acceleration
voltage, 8000 V; and ionization energy, 70 eV. For the MS/
MS experiments, the instrument was calibrated with per-
fluorokerosene (PFK) by selected dissociation of precursor
ions and the monitoring of product-Q ions on a quadrupole
analyser. The product-Q ions spectra were obtained by
selecting the precursor ion by MS1 with a resolution of ca.
1500. The precursor ion collided with xenon in the collision
cell (Rf only quadrupole collision cell, q) and Q was scanned
at 5 s/ scan over the range m/ z 45-450. For precursor-ion
spectra, the product-Q ion was selected by MS2 with unit
resolution and the magnetic field scanned at 5 s/ decade.
Xenon was used as the collision gas at a pressure of 4 × 10-6
mbar and optimized the collision energy to 50-60 eV.
Synthesis of 1,3-Dioxane and 1,3-Dioxolane Derivatives.
The 2-alkyl-5,5-dimethyl-1,3-dioxanes were synthesized by
condensation of neo-pentylglycol (NPG; 2,2-dimethyl-1,3-
propanediol) with an excess of 10% of different aldehydes
(p-formaldehyde, acetaldehyde, propanal, n-butanal, iso-
butanal, and n-pentanal) and concentrated sulfuric acid
(0.5%) as acid catalyst (34). All the reagents were from Aldrich
Chemie (Germany). The mixtures were refluxed 1 h at room
temperature. Then water was decanted; the acetals were
dried over sodium sulfate and filtered. Purity was 90-97%
for different compounds. The same compounds were
synthesized at microscale employing the same procedure as
described by Staudte et al. (26), who used p-toluenesulfonic
acid as the catalyst and acetonitrile as the solvent. Briefly,
the glycol and the acid were weighted into a microreaction
vial. Acetonitrile was added followed by the corresponding
aldehyde, and the mixture held at 35 °C for 2 h. After coming
to room temperature, an aliquot (ca. 10 µL) was diluted to
1 mL with acetone to be spiked into water for recovery studies.
The 2-alkyl-4-methyl-1,3-dioxolane and 2-alkyl-1,3-di-
oxolane compounds were synthesized using the same
procedure as described above but using propylene glycol
(PPG) and ethylene glycol (EG), respectively. Mixtures of cis
and trans isomers (usually 60/ 40) were obtained for the
different 2-alkyl-4-methyl-1,3-dioxolane derivatives.
a
TABLE 2. EI-MS/MS Mass Spectra of 2EDD
m/z
Product-Q ionsa
Precursor ionsb
115
69
56
69
41
41
[M]•+c
115
73, 115, [M - H]+
a
b
Product-Q ion spectrum of the ion m /z115. Precursor-ion spectrum
c
of m /z 69. (See text for experim ental conditions). Experim ents were
perform ed with [M]•+ ) 144 for 2EDD and [M]•+ ) 158 for 2iPMD.
Quantitation. Dioxane and dioxolane quantitation was
performed by the internal standard method. For each
compound, the response factor is calculated relative to
1-chlorohexane at four calibration levels. The calibration
analyses were obtained by analyzing Milli-Q water samples
spiked with 1-chlorohexane and different spiked solutions
of each dioxane and dioxolane (0.01, 0.25, 1, and 10 µg/ L).
The unknown concentrations in the sample are then obtained
FIGURE 3. EI-MS/MS of 2EDD. (a) Product-Q ion spectrum of m/z
115. (B) Precursor-ion spectrum of m/z 69. See text for experimental
conditions.
product-Q ion spectrum of m/ z 115 and the precursor-ion
spectrum of m/ z 69 from 2EDD (see Figure 3). The origin
of the ions m/ z 56 (base peak) and 69 is depicted in Figure
4. Thus, the m/ z 69 ion is generated through m/ z 115 from
consecutive losses of CO and H2O. The base peak m/ z 56
is generated by oxonium ions (a,b) via m/ z 73 ion, although
the exact fragmentation mechanism is unknown, the most
probably pathway is through a stabilized tertiary carbonium
ion C5 (36, 37).
Study of Recovery Efficiencies. Once the different
compounds were synthesized, a study of the recovery
efficiencies by CLSA and P&T methods was performed in
order to obtain a more accurate determination of con-
centration levels and the feasibility of the two procedures.
The recoveries from Milli-Q water spiked with 0.25 ppb of
each compound using the CLSA method are displayed in
Table 3.
from the mean response factors using the formula Cz ) A Ci/
z
RzA where Cz and Ci are the concentration of the unknown
i
compound and the internal standard; A and A are the peak
z
i
areas of the unknown and the internal standard, respectively;
and Rz is the response factor for the target compound.
Results and Discussion
Mass Spectrom etric Characterization of 1,3-Dioxanes and
1,3-Dioxolane Derivatives. EI mass spectra of 2-alkyl-5,5-
dimethyl-1,3-dioxane were previously described by Cocheo
et al. (24) and Preti et al. (25). They are characterized by
small [M - H]+ ions and losses of the alkyl chain at position
2 (m/ z ) 115), both corresponding to the known well-
stabilized oxonium ions (35). On the other hand, CI spectra
clearly show intense [M + H]+ ions and the other peaks
already observed in the EI mode. Table 1 shows the most
abundant ions in the EI and CI modes for these compounds
as well as several 1,3-dioxolane derivatives.
The 2-alkyl-5,5-dimethyl-1,3-dioxanes had a range of 13-
100% recovery for different compounds using the CLSA
method, being higher for the derivatives with longer alkyl
chains as expected by their log Kow. An acceptable recovery
of 57% was obtained for 2EDD. On the other hand, the
2-alkyl-4-methyl-1,3-dioxolanes showed lower recoveries
A proposed fragmentation pathway provided by the MS/
MS experiments described in Table 2 was obtained from the
9
VOL. 32, NO. 2, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 2 0 9

FIGURE 4. Proposed fragmentation pathway for mass spectra of 2-alkyl-5,5-dimethyl-1,3-dioxanes.
than their homologue 1,3-dioxane derivatives. The most
abundant compounds in real samples, the cis and trans
isomers of 2E4MDL, were recovered in a limited extent (26%).
The worst recoveries by CLSA were obtained for 2-alkyl-
1,3-dioxolanes being the isopropyl, the only tested compound
that showed a relative good performance.
The partition coefficients for the 2-alkyl-5,5-dimethyl-
1,3-dioxanes were calculated by means of a commercial
9
2 1 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998

TABLE 3. Recovery Efficiencies for 1,3-Dioxane and 1,3-Dioxolane Derivatives by CLSA and P&T Techniques
CLSA
recovery ( SD
c
used filtera
new filterb
P&T
compound
unsalted
salted
unsalted
salted
recovery ( SD
2-alkyl-5,5-dim ethyl-1,3-dioxanes
R ) H
Me
13 ( 5
33 ( 5
57 ( 8
95 ( 7
79 ( 9
100 ( 12
99 ( 13
nrf
17 ( 5
40 ( 7
61 ( 13
95 ( 9
nte
100 ( 11
101 ( 8
nt
nt
31 ( 5
nt
37 ( 2
60 ( 4
88 ( 6
108 ( 10
105 ( 9
107 ( 9
104 ( 3
2 ( 1
52 ( 6
28 ( 3
26 ( 1
42 ( 3
71 ( 8
72 ( 2
92 ( 2
124 ( 5
15 ( 1
15 ( 1
90 ( 8
R
Et
i-Pr
Pr
i-Bu
Bu
103 ( 9
103 ( 10
101 ( 5
104 ( 9
114 ( 5
3 ( 1
17 ( 3
82 ( 8
103 ( 9
109 ( 3
O
O
2-alkyl-4-m ethyl-1,3-dioxolanesd
R ) H
Me
Et
i-Pr
i-Bu
nr
12 ( 1
R
26 ( 3
73 ( 3
88 ( 8
62 ( 4
100 ( 7
97 ( 2
33 ( 1
63 ( 2
nt
O
O
nt
2-alkyl-1,3-dioxolanes
R ) H
nr
<1
3 ( 1
33 ( 5
nt
nt
nt
nt
nr
nr
15 ( 2
68 ( 5
nr
nr
20 ( 4
89 ( 13
8 ( 1
16 ( 2
nt
Me
Et
R
O
O
i-Pr
nt
a
b
c
Average of six determ inations at 0.25 ppb level. Average of three determ inations at 0.25 ppb level. Average of four determ inations. Range
1-100 ppb. Purge vessel 5 m L. Recoveries for both cis and trans isom ers. nt, not tested. f nr, not recovered.
d
e
computer program (38) that breaks down the molecule and
assigns a value to each fragment according to its data base.
The final value is the sum of the partial values with a final
correction after considering the structure of the molecule.
All compounds showed values of log Kow < 3, which is
assumed to have a low probability for a bioaccumu-
lation potential (39). The calculated values ranged from log
Kow ) 0.82 (DMD) to 2.90 (2iBMD), being 1.86 for 2EDD.
These differences can explain the wide range of recoveries
found. The partition coefficients for 2-alkyl-4-methyl-1,3-
dioxolanes were lower than their homologue 2-alkyl-5,5-
dimethyl-1,3-dioxanes. As an example, 2E4MDL had a
calculated log Kow ) 0.82, which explains its moderate
recovery.
The results described above were obtained from several
used carbon filters that we considered as representatives of
still-in-use filters. However, when new, non-used carbon
filters were employed, recoveries of both dioxanes and
dioxolanes increased markedly, but their efficiency quickly
dropped to the values cited previously. The salted CLSA
according to the method described by Hwang et al. (22) was
also performed with both new and used filters. This method
enhanced recoveries of all compounds related to the unsalted
procedure, being, as expected, more significant with the new
filters. In comparing both methods, DMD, TMD, and
2E4MDL were significantly better recovered by the salting-
out method among the compounds found in real samples
by using either new or used filters. Nevertheless, we
discarded the use of the salted method due to the large
number of samples analyzed and the potential problems
FIGURE 5. General scheme of the formation of 1,3-dioxane and
1,3-dioxolane derivatives.
associated with this method (excessive amounts of sodium
sulfate, tedious preparation, clogging, etc.).
We also studied the matrix effects by spiking dioxanes
and dioxolanes into river water. The obtained resultssnot
shownsdid not present significant differences with the values
displayed in Table 3 for Milli-Q water being the maximum
difference less than a 10% lower recovery.
pounds with a longer alkyl chain. The recovery for 2EDD by
P&T was 42%, slightly worse than CLSA. Taking into account
the compounds present in the real samples, both methods
are feasible but CLSA shows better recoveries for the identified
compounds.
The same group of compounds was studied by P&T.
Recoveries were performed in the 1-100 ppb range using a
5-mL purge vessel, being practically constant in this interval.
Comparing both methods, it seems that dioxanes and
dioxolanes with small alkyl chains (R ) H, Me) are better
recovered by P&T. However, CLSA is preferred for com-
Dioxanes and Dioxolanes in Waters. Organic Com-
pounds in the Wastewater of Company A. The wastewaters
from company A were regularly analyzed from November
1993 to June 1996. The identified compounds by GC/ MS of
CLSA extracts were dioxane and dioxolane derivatives and
large amounts of terpenes coming from rosin oil residues
9
VOL. 32, NO. 2, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 2 1 1

TABLE 4. Concentration Range, Mean Values, and Frequency of Appearance of Relevant 1,3-Dioxanes, 1,3-Dioxolanes, and
a
1,4-Dioxanes from Wastewaters of Company A
compound
mean
2293
68.7
129.6
954.5
0.2
max-min
frequency
5,5-dim ethyl-1,3-dioxane
16320-<0.01
1390-<0.01
3900-<0.01
27500-<0.01
7-<0.01
20-<0.01
<0.01
228-<0.01
99-<0.01
31400-<100
86000-<0.01
82.8
48.6
71.4
71.4
2.8
5.7
0
51.4
48.6
48.6
88.6
2,5,5-trim ethyl-1,3-dioxane
2-ethyl-5,5-dim ethyl-1,3-dioxane
2-iso-propyl-5,5-dim ethyl-1,3-dioxane
2-n-propyl-5,5-dim ethyl-1,3-dioxane
2-iso-butyl-5,5-dim ethyl-1,3-dioxane
2-n-butyl-5,5-dim ethyl-1,3-dioxane
2,5-dim ethyl-1,4-dioxaneb
0.7
<0.01
16.7
9.3
6400
2985
2,6-dim ethyl-1,4-dioxaneb
1,4-dioxane
2-ethyl-4-m ethyl-1,3-dioxolaneb
a
b
Results are expressed in µg/L. Data from J anuary 1994-J uly 1995 (n ) 35). Expressed as sum of cis and trans isom ers.
a
TABLE 5. Percentage of Concentrations of Individual Dioxanes and Dioxolanes from Wastewaters of Company A
range (µg/L)
compound
<0.01-1
1-100
100-1000
>1000
5,5-dim ethyl-1,3-dioxane
17.1
65.7
82.9
88.5
97.1
94.3
100
62.9
68.6
ndb
37.1
25.7
8.5
2.9
2.9
5.7
0
31.4
31.4
nd
14.3
5.7
5.7
2.9
0
0
0
5.7
0
51.4
28.6
31.4
2.9
2.9
5.7
0
0
0
0
0
2,5,5-trim ethyl-1,3-dioxane
2-ethyl-5,5-dim ethyl-1,3-dioxane
2-iso-propyl-5,5-dim ethyl-1,3-dioxane
2-n-propyl-5,5-dim ethyl-1,3-dioxane
2-iso-butyl-5,5-dim ethyl-1,3-dioxane
2-n-butyl-5,5-dim ethyl-1,3-dioxane
2,5-dim ethyl-1,4-dioxane
2,6-dim ethyl-1,4-dioxane
1,4-dioxane
2-ethyl-4-m ethyl-1,3-dioxolane
48.6
14.3
14.3
42.8
a
b
Data from J anuary 1994-J uly 1995 (n ) 35). nd, < 100 ppb for 1,4-dioxane.
that are largely used in lacquers, inks, adhesives, etc. The
most abundant products were 2-alkyl-5,5-dimethyl-1,3-
dioxanes (R )H to i-But) and 2-alkyl-4-methyl-1,3-dioxolanes
(R ) Et). Additional minor compounds were observed by
P&T such as 2-alkyl-1,3-dioxolanes (R ) Me and Et), 2-alkyl-
4-methyl-1,3-dioxanes (R ) H, Me), and miscellaneous
compounds such as 1,4-dioxane, 1,3-dioxolane, 2,4,4-tri-
methyl-1,3-dioxane, 2,4,4,6-tetramethyl-1,4-dioxane, and
2,4,8,10-tetraoxaspiro[5,5]undecane (a condensation product
of pentaeritrithol).
and 2iPMD are formed by the condensation of NPG with its
disproportionation products.
Table 4 displays the concentration levels (in µg/ L) of
relevant 1,3-dioxanes, 1,3-dioxolanes, and 1,4-dioxane from
the wastewaters of the company A and their frequency of
appearance during January 1994-July 1995. This period
covers several months before and after the company began
to treat the wastewaters to reduce the amounts of compounds
dumped into the river. The displayed results and the
following were corrected for recovery efficiency.
The company A produces both saturated and unsaturated
polyester resins. Polyester resins are produced by poly-
merization of different glycols in acid-catalyzed condensa-
tions. Saturated resins are mainly formed by condensation
of NPG or PPG with terephthalic acid. Its production has an
approximate ratio of 6:1. The company also produces minor
amounts of other resins with some modifications (i.e.,
ethylene glycol, adipic or isophthalic acids). The unsaturated
polyester resins in company A are formed by condensation
of PPG and maleic anhydride. Modifications using diethylene
glycol and phthalic or isophthalic or adipic acids are used
in minor amounts.
The formation of byproducts such as 1,3-dioxanes and
1,3-dioxolanes could be explained by the reaction of different
glycols and the aldehydes formed by dehydration of vicinal
diols in acidic medium. The general scheme of the formation
of these byproducts is depicted in Figure 5. Thus, 2-alkyl-
5,5-dimethyl-1,3-dioxanes, 2-alkyl-4-methyl-1,3-dioxolanes,
and 2-alkyl-1,3-dioxolanes come from the condensation of
the different glycols (NPG, PPG, and EG) and aldehydes.
Cocheo et al. (24) hypothesized the origin of the most
common compounds. Thus, 2E4MDL is formed by con-
densation between two molecules of PPG; TMD is formed
by condensation of NPG and EG, while 2EDD is the result
of the condensation between NPG and PPG. Finally, DMD
From the table, it is inferred that the most concentrated
compounds were usually in the ppm range during the odor
incidents. These levels were dramatically reduced to the
low ppb level when the company was forced by the
administration to treat its wastewaters. However, for some
months the results were sharply different. Thus, whereas
some samples showed a total absence of dioxanes and
dioxolanes, others contained low-medium ppb levels of 2EDD
and higher values for DMD and TMD. These unexpected
results reflected the variety of treatment procedures per-
formed in the factory (i.e., adsorption onto GAC, ozone
oxidation with Fenton’s reagent, UV/ H2O2) to reduce the
levels of dioxanes and dioxolanes dumped into the river.
Table 5 displays the percentage of concentrations of
individual dioxanes and dioxolanes in the wastewaters of
the company for the same period. Concentrations were
grouped in four groups (<0.01-1, 1-100, 100-1000, and
>1000 µg/ L). The table shows the improvements made by
A. Thus, except for DMD and TMD, which present significant
percentages in the range 1-100 ppb (37.1 and 25.7%,
respectively), all the other compounds were bracketed into
the <0.01-1 ppb range. These average values are practically
1000-fold less concentrated than the maximum concentra-
tions measured during the beginning of the incidents in
October 1993.
9
2 1 2 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998

TABLE 6. Levels (in µg/L) of Dioxanes and Dioxolanes in Groundwaters (April 1994-July 1995)a
concentration range (µg/L)
sample
DMD
TMD
2EDD
2iPMD
2iBMD
2,5DMD
2,6DMD
nd
2E4MDL
2iP4MDL
1
2
ndb
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd-6
nd
nd
nd
nd
nd
0.04
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
0.03
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd-3
nd
nd
nd-0.19
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
nd
4
5
6
7
8
9
124-2111
nd-1.39
nd
2.6-26.7
nd-tr
1.59-45.59
12.55
1.12
nd-4.27
tr
nd
5.91
2.01-20.60
6.93
1.72-8.70
0.04-2.65
nd
nd
5.30
4.84-19.26
1.12
2.14
3.20-14.26
3.37
2.58-11.11
0.79-8.98
6.30
4.20-6.27
6.53-7.11
5.81
4.8-112
27.1-673
nd-0.28
nd
0.29-7.07
nd
nd-4.91
0.94
4.30
nd-0.65
nd
nd
0.70
0.26-7.40
1.27
0.08-7.77
0.03-0.12
nd
nd
0.23
0.10-2.52
0.08
1.10
0.42-1.76
0.21
0.26-0.62
0.48-1.13
0.33
0.40-0.86
0.76-1.85
0.51
101-457
nd-0.01
nd
0.44-4.88
nd
nd-7.38
2.16
1.86
nd-1.37
nd
nd
0.82
0.38-3.21
2.27
1.82-2.31
0.18-0.78
nd
nd
0.38
0.18-1.17
0.08
1.21
0.28-1.47
0.85
0.35-0.89
0.37-1.20
1.19
0.30-0.55
0.35-1.12
0.53
2.0-35
nd-0.04
nd
nd-0.33
nd
0.01-0.34
nd
tr
nd
nd
nd
tr
nd-0.70
nd
0.04-0.70
nd-0.36
nd
nd
nd
0.03-0.96
nd
tr
nd-0.85
0.03
nd-0.49
nd-0.40
nd
0.01-0.31
nd-0.15
0.03
0.2-29
nd-0.04
nd
nd-0.05
nd
nd-0.12
nd
tr
nd
nd
nd
tr
nd-0.50
nd
0.03-0.50
nd-0.35
nd
nd
nd
nd-0.85
nd
tr
nd-0.91
0.03
nd-0.38
nd-0.32
nd
nd-0.23
nd
0.03
0.01
nd-0.90
nd-0.41
nd-0.31
0.04
61-3405
nd
nd
nd
nd-0.36
nd
nd-0.72
0.76
0.19
nd-0.23
nd
nd
0.38
nd-1.60
0.45
0.07-0.53
nd-0.32
nd
nd
nd
0.04-1.27
0.34
0.24
nd-1.25
0.28
nd-0.75
nd-0.72
0.48
0.24-0.55
nd-0.13
0.38
nd
2.14-20.30
nd
0.46-7.63
2.52
0.92
nd-7.03
tr
nd
3.85
3.75-50.60
5.60
12.02-68.47
2.11-29.52
nd
nd
1.05
7.14-66.73
4.97
4.20
5.32-38.28
2.94
6.37-32.95
nd-16.60
1.40
6.80-15.79
5.32-11.55
2.59
10
11
13
14
15
16
17
18
20
21
22
23
24
25
26
27
30
31
32
33
34
35
37
38
39
40
42
44
45
52
53
54
55
58
59
60
62
64
66
71
2.34
0.20
0.39
0.50
0.01
1.40
2.94-16.63
3.12-18.14
1.24-7.01
5.40
5.30
0.64-3.47
2.75
1.43
1.43
1.03-9.08
0.87
0.35-2.95
0.68-14.54
nd
nd-0.19
0.01-1.16
nd-0.67
0.38
0.34
nd-0.20
nd
nd
nd
nd-0.40
0.13
nd-0.10
nd-0.73
nd
0.18-1.36
0.09-2.05
0.06-0.51
0.12
0.27
nd-0.72
0.12
0.11
nd
0.02-1.04
nd
nd-1.63
0.05-1.13
nd
0.35-1.07
0.11-1.75
0.05-0.33
0.13
0.35
nd-0.19
0.42
0.23
nd
0.03-0.36
0.02
nd-0.50
0.08-0.53
nd
nd-1.08
nd-0.56
nd-0.42
0.04
0.04
nd-0.17
nd
nd
nd
nd-0.27
nd
nd-0.03
nd-0.45
nd
3.71-36.23
3.17-14.67
1.60-8.05
2.94
2.59
nd-1.60
3.15
0.98
0.21
1.05-6.60
0.15
nd-2.14
nd-17.99
nd
0.04
nd-0.26
nd
nd
nd
nd-0.24
nd
nd-0.02
nd-0.45
nd
nd
nd
nd
nd
nd
nd
nd
a
Data from April 1994-J uly 1995. Num ber of analysis: n ) 7 for sam ples 4, 7, 9, 17, 20, 21, 25, 42, 44, 53, 59, and 64; (n ) 5) for sam ples 35
b
and 40; (n ) 4) for sam ples 2, 13, 32, 33, and 62; (n ) 3) for sam ples 1, 5, 8, 15, and 30; (n ) 1), for rem aining sam ples. nd, <0.005 µg/L. tr, trace.
Groundwaters. Forty-eight different groundwater supplies
collected on the left and right sides of the Tordera River were
analyzed. Sampling points are shown in Figure 2. The results
displayed in Table 6 correspond to April 1994-July 1995, the
most intensive sampling period. The number of analysis for
each point reflects both the importance of the well and how
the variation of the dioxanes dumped by company A affected
the closer wells.
1,3-dioxanes (DMD, TMD, 2EDD, and 2iPMD) were identified
in almost all samples. The 2-alkyl-4-methyl-1,3-dioxolanes
were exclusively (2E4MDL) cis and trans isomers with very
few samples containing the isopropyl derivative. The third
group of compounds were both 2,5- and 2,6-dimethyl-1,4-
dioxanes (2,5DMD and 2,6DMD) cis and trans isomers. As
CLSA is not adequate for the determination of 1,4-dioxane
at low ppb levels (recovery <1%), the results for this
compound are not displayed. As an example, Figure 6 shows
the GC chromatogram of a CLSA extract from groundwater.
Several water samples were analyzed by flavor profile
analysis (FPA) according to the method described elsewhere
(40). The obtained results were in agreement with those
reported by Philadelphia’s researchers that established a
threshold odor of 5-10 ppt for 2EDD (25), although in our
case the trained panelists described the odor of the samples
and 2EDD as olive water.
Table 6 contains the minimum and maximum concen-
tration levels (expressed in µg/ L) for each compound during
the campaign. This allows the knowledge of levels of dioxanes
and dioxolanes at each point before the company A had begun
to treat their wastewaters and to know how this treatment
lowered the concentration of these compounds. This reflects
the direct influence of these dumpings into the aquifer.
The identified compounds in the groundwaters were the
same byproducts of company A. The 2-alkyl-5,5-dimethyl-
9
VOL. 32, NO. 2, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 2 1 3

amount of 2E4MDL (mid ppb level) dumped into the aquifer
by the treated wastewaters of company A was significantly
superior than 2EDD (low ppb level) and could explain the
slow decrease of 2E4MDL.
The wells near the mouth of the river and farthest from
the dumping are the most used for drinking purposes. They
presented significant variations of contaminant concentra-
tions either along the studied period or between them, even
for wells a few hundred meters appart. This can be attributed
to a different piezometric levels of wells and a difference in
the degree of pumping. No trends for a different distribution
in both sides of the river could be observed from these results
although samples far from the river banks showed the lowest
values.
River Water. Four selected points (15, 83, 73, and 62,
encircled in Figure 2) of the Tordera River were analyzed
twice a year from October 1993 to October 1996. Table 7
shows the concentration ranges for the most relevant
compounds. The higher levels were measured, as expected,
in the closest point to the dumping (15). The maximum
value for 2EDD (173 ppb) was determined in June 1994,
decreasing dramatically from this value to levels <2 ppb in
the last two years. The same trend was observed for 2iPMD
with a maximum of 3030 ppb in June 1994 to 33 ppb, which
was last measured October 1996. This represents a reduction
higher than 99% for these two significant dioxanes into the
river. Conversely, DMD (84 ppb in June 94), although
presented the same tendency as described for 2EDD and
2iPMD, showed a less marked decrease. The limited number
of analyzed samples does not allow us to generalize on the
behavior of these compounds, but it seems that they follow
the same trends as those observed in groundwaters.
Resin Manufacturing Plants. We were interested in
knowing if two other resin manufacturing plants (B and C)
not related to this incident and located on the banks of other
two rivers generated the same byproducts detected in the
wastewater of company A and therefore if they could be a
potential source for odor complaints. In the wastewaters of
both factories dioxanes and dioxolanes could be detected.
Therefore, we concluded that these compounds can be
considered as markers of pollution for resin plants. The
concentration levels were as follows: DMD (44 ppb), TMD
(11 ppb), 2EDD (73 ppb), 2iPMD (1411 ppb), 2nPMD (6 ppb),
2iBMD (227 ppb), and 2E4MDL (227 ppb) in the wastewaters
of company B. We could only measure the amount of 2EDD
(1513 ppb) in company C since the presence of large amounts
of aromatic solvents prevented the accurate quantification
or the presence of other dioxanes and dioxolanes. The levels
detected in company B (grab sample) were lower than those
usually found in company A and reflect a different production
(i.e., 2iPMD and 2iBMD were the main compounds, the latter
was not found in company A) of both companies and
probably a better treatment of their wastewaters.
Barcelona’s Tap Water. Odor incidents were detected in
the raw water entering the water works plant of Barcelona,
and some complaints from consumers of Barcelona’s tap
water were noticed in November 1994 and January-February
1995. The odor present in tap water was unanimously
described as “olive water”, a closer reference for mediter-
ranean people than the varnish/ chemical descriptor reported
by Philadelphia’s consumers (25-28). Several tap water
samples showed levels of 2EDD ranging from 8 to 49 ng/ L
in this period whereas up to 312 ng/ L was detected at the
entrance of the water treatment plant. These results were
unexpected because there are no resin manufacturing plants
along the Llobregat River, which supplies raw water to
Barcelona. Also, there were no connections between these
samples and those of the Tordera River located more than
60 km away. The origin of the malodorous raw water in the
water treatment plant of Barcelona was established when
FIGURE 6. GC chromatogram of a CLSA extract from groundwater
(well 9). The numbered peaks correspond to the following identified
compounds: (1) toluene, (2) cis-2E4MDL, (3) trans-2E4MDL, (4) cis-
2,5DMD, (5) cis-2,6DMD, (6) trans-2,5DMD, (7) trans-2,6DMD, (8) DMD,
(9) TMD, (10) ethylbenzene, (11) m+p-xylene, (12) o-xylene, (13)
2EDD, and (14) 2iPMD. An asterisk (*) denotes internal standards
(1-chloroalkanes C , C , C , and C ).
5
6
8
10
Concentration levels were usually at the low ppb level
except for sample 4, which is the closer well to the dumping
and where the maximum concentration of 2EDD (673 µg/ L)
was reached. This compound (2EDD) was positively identi-
fied in almost all samples with maximum levels of 7.77 ppb
(sample 20) and in the 1-2 ppb range for the farthest samples
from the dumping. The seasonal variation of 2EDD in several
wells located along the river course is depicted in Figure 7.
The figure clearly shows that the amount of 2EDD into the
aquifer dramatically diminished after November 1994 in all
wells studied. DMD also shows the same trend described
for 2EDD in agreement with their common origin.
The most abundant compound in all samples was
2E4MDL with a maximum measured value of 3405 µg/ L in
sample 4. The concentration range of 2E4MDL from the
remaining samples varied from the low ppb level to 68.5 ppb
(sample 20). The seasonal variation of 2E4MDL for the same
wells did not follow the same general trend described before
for DMD and 2EDD. Significant differences for each well
were observed. Thus, 2E4MDL decreased markedly between
June and November 1994 in well 7 and then almost a steady
state was reached. The concentration levels in wells 42, 59,
and 64 presented their maximum in November 1994; after
this date, a moderate decrease was measured but less marked
than that observed for 2EDD and DMD. The behavior of
2E4MDL in well 17 was quite different; it showed their
maximum concentration during January-March 1995 to
decrease to the same levels measured at mid-1994. These
results represented a mixed behavior between those observed
in well 7 and wells 42, 59, and 64. The values obtained for
sampling point 20 increased from June to November 1994.
During the period March-May 1995, the concentration levels
remained practically constant, and finally a significant
decrease was detected in July 1995. An unexpected low level
of 2E4MDL as compared with those usually found in this
sample was measured in January 1995. We did not find any
explanation for this result because the other dioxanes and
dioxolanes did not show the same concentration shift.
The difference in behavior in the aquifer between 2EDD,
DMD, and 2E4MDL can be atributed to several factors such
as higher solubility or mobility of 2E4MDL. In addition, the
9
2 1 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998

FIGURE 7. Seasonal variation of selected dioxanes and dioxolanes in several wells along the Tordera River.
samples collected along the Llobregat River course during
an odor event led us to a sewer of a waste management
facility located approximately 15 km away from the water
treatment plant. This factory was devoted to the treatment
and elimination of industrial residues. The main identified
compounds in this sewer were 2EDD (17.3 ppm), 2E4MDL
(16.7 ppm), DMD (0.5 ppm), TMD (0.5 ppm), and 2iPMD
(0.1 ppm). We believe that these results can be explained
when wastewaters from a polyester resin manufacturing plant
were carried by tanker truck to the waste management facility
located on the banks of Llobregat River. Once the origin was
detected, corrective measures such as cleaning of the sewers
and prohibition to treat this type of waste were enough to
solve the problem.
The results shown here prove that the presence of dioxanes
and dioxolanes in water can be a source for complaints from
consumers due to their extremely low threshold odor. The
most malodorous compound (2EDD) has been identified in
river water, tap water, wastewater, and groundwater; it is a
characteristic byproduct from resin plants. This compound
9
VOL. 32, NO. 2, 1998 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 2 1 5

(16) Fadel, M.; Gammie, L. Proceedings of the AWWA Water Quality
Technology Conference, Miami, FL; AWWA: Denver, 1993; pp
1907-1914.
TABLE 7. Concentration Levels (in µg/L) of Dioxanes and
Dioxolanes in River Water (October 1993-1996)a
(17) Young, W. F.; Horth, H.; Crane, R.; Ogden, T.; Arnott, M. Water
concentration range (µg/L)
Res. 1996, 30, 331-340.
river
sample DMD
TMD
2EDD
2iPMD
2iBMD 2E4MDL
(18) Keith, L. H. Advances in the Identification and Analysis of Organic
Pollutants in Water; Ann Arbor Science Publishers: Ann Arbor,
MI, 1981.
(19) Albaige´s, J.; Casado´ , F.; Ventura, F. Water Res. 1986, 20, 1153-
1159.
(20) Grob, K. J. Chromatogr. 1973, 84, 255-273.
(21) Krasner, S. W.; Hwang, C. J.; McGuire, M. J. Water Sci. Technol.
1983, 15, 127-138.
15
83
73
62
2.6-99 nd-20 0.30-173 4.8-3030 nd-22 39-152
3.6-104 nd-4.9 0.32-123 3.6-41 nd-0.12 34-122
4.5-32 nd-2.1 0.08-10 0.21-19 nd-0.43 12-34
nd-30.4 nd-0.70 0.08-4.0 nd-1.78 nd-0.29 17-20
a
Data from October 1993-1996. For 2E4MDL only results from J une
and October 1996 are displayed. Num ber of analysis, n )7. nd, <0.005
µg/L.
(22) Hwang, C. J.; Krasner, S. W.; Moylan, R. S.; Dale, M. S. Environ.
Sci. Technol. 1984, 18, 535-539.
(23) Ventura, F.; Romero, J.; Pare´s, J. Environ. Sci. Technol. 1997, 31,
2368-2374.
(24) Cocheo, V.; Boaretto, C.; Quaglio, F.; Sacco, P.; Zanetti, R. Analyst
1991, 116, 1337-1342.
(25) Preti, G.; Gittelman, T. S.; Staudte, P. B.; Luitweiler, P. Anal.
Chem. 1993, 65, 699A-702A.
has one of the lowest threshold odor values reported in the
literature for anthropogenic compounds and makes the water
unacceptable for drinking purposes when 2EDD concentra-
tion levels are higher than 5-10 ng/ L.
(26) Staudte, P. B.; Gittelman, T. S.; Luitweiler, P. Proceedings of the
AWWA Water Quality Technology Conference, Miami, FL,
AWWA: Denver, 1993; pp 559-579.
(27) Gittelman, T. S.; Luitweiler, P. Proceedings of the AWWA Water
Quality Technology Conference, Boston, MA; AWWA: Denver,
1996 (CD Rom version).
(28) Gittelman, T. S.; Luitweiler, P.; Staudte, P. B. Proceedings of the
AWWA Water Quality Technology Conference, Boston, MA;
AWWA: Denver, 1996 (CD Rom version).
(29) Ventura, F.; Matia, L.; Romero, J.; Boleda, Ma. R.; Mart´ı, I.; Martin,
J. Water Sci. Technol. 1995, 31, 63-68.
(30) Mantica, E.; Botta, D.; Castellani Pirri, L. In Organic Micro-
pollutants in the Aquatic Environment; Bjorseth, A., Angeletti,
G. Eds.; D. Reidel Publishing Co.: Dordrecht, Holland, 1986; pp
89-106.
(31) Bruchet, A.; Khiari, D. In Advances in Taste and Odor Treatment
and Control; AWWA Research Foundation and Lyonnaise des
Eaux: Denver, CO, 1995; pp 321-349.
Acknowledgments
We wish to thank Carme Costa from Junta de Sanejament
for her effort in sampling and designing all the campaign.
We are indebted to Narc´ıs Salvatella, Raquel Carden˜ oso, and
their team of trained panelists for performing FPA. We are
also grateful to Montse Cortina (CID-CSIC) for their help in
performing the analysis of river water samples and to Carles
Planas and Jesu´ s Caban˜ as (CID-CSIC) for synthesizing a
second set of compounds and MS/ MS experiments, respec-
tively.
Literature Cited
(1) Zoeteman, B. C. J. Sensory Assessment of Water Quality;
Pergamon Series on Environmental Science Vol. 2; Pergamon
Press: Oxford, U.K., 1980; pp 83-92.
(2) Hayes, K. P.; Burch, M. D. Water Res. 1989, 23, 115-121.
(3) Burlingame, G. A.; Dann, R. M.; Brock, G. L. J. Am. Water Works
Assoc. 1986, 78, 56-61.
(4) Burlingame, G. A.; Muldowney, J. J.; Maddrey, R. E. J. Am. Water
Works Assoc. 1992, 84, 92-97.
(5) Cotsaris, E.; Bruchet, A.; Mallevialle, J.; Bursill, D. B. Water Sci.
Technol. 1995, 31, 251-258.
(6) Persson, P. E. Water Sci. Technol. 1995, 31, 9-13.
(7) Ju¨ ttner, F. Water Sci. Technol. 1995, 31, 69-78.
(8) Wajon, J. E.; Alexander, R.; Kagi, R. I.; Kavanagh, B. Chemosphere
1985, 14, 85-89.
(9) Mallevialle, J., Suffet, I. H., Eds. Identification and Treatment of
Tastes and Odors in Drinking Water; AWWA Research Founda-
tion and Lyonnaise des Eaux: Denver, CO, 1987.
(10) Suffet, I. H., Mallevialle, J., Eds. Advances in Taste and Odor
Treatment and Control; AWWA Research Foundation and
Lyonnaise des Eaux: Denver, CO, 1995.
(11) Suffet, I. H.; Corado, A.; Chou, D.; McGuire, M. J.; Butterworth,
S. J. Am. Water Works Assoc. 1996, 88, 168-180.
(12) Bruchet, A.; N’Guyen, K.; Anselme, C. In Identification and
Treatment of Taste and Odor Compounds; AWWA: Denver, CO,
1989; pp 125-141.
(32) Botta, D.; Dancelli, E.; Mantica, E. Environ. Sci. Technol. 1996,
30, 453-462.
(33) Williams, L., Jr.; Brown, D. W.; Myron Stephenson, J. H.; Little,
J. A. In Identification and Analysis of Organic Pollutants in Water;
Keith, L. H., Ed.; Ann Arbor Science Publishing: Ann Arbor, MI,
1977; pp 499-516.
(34) McGorrin, R. J.; Pofahl, T. R.; Croasmun, W. R. Anal. Chem.
1987, 59, 1109A-1112A.
(35) Porter, Q. N.; Baldas, J. In Mass Spectrometry of Heterocyclic
Compounds; John Wiley & Sons: Chichester, U.K., 1971.
(36) Pontani, J. F.; Aubert, C. Org. Mass Spectrom. 1994, 29, 698-
699.
(37) Mc Lafferty, F. W.; Turecek, F. In Interpretation of Mass Spectra,
4th ed.; University Science Books: Mill Valley, CA, 1993.
(38) PrologP 4.2. CompuDrup Chemistry Ltd.: Budapest, Hungary.
(39) Franke, C. Chemosphere 1996, 32, 1897-1905.
(40) APHA. Section 2170. Flavor Profile Analysis. In Standard Methods
for the Examination of Water and Wastewater, 17th ed. Suppl.;
APHA: Washington, DC, 1992.
Received for review May 12, 1997. Revised manuscript re-
ceived October 10, 1997. Accepted October 12, 1997.X
(13) Hrudey, S. E.; Gac, A.; Daignault, S. A. Water Sci. Technol. 1988,
20, 55-61.
(14) Karlsson, S.; Grimvall, A.; Boren, H. Water Sci. Technol. 1995,
31, 49-54.
ES9704085
(15) Anselme, C.; N’Guyen, K.; Bruchet, A.; Mallevialle, J. Environ.
Technol. Lett. 1985, 6, 477-488.
X Abstract published in Advance ACS Abstracts, December 1, 1997.
9
2 1 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 32, NO. 2, 1998