Major Structural Components in
Freshwater Dissolved Organic
Matter
B U U A N L A M , † A N D R E W B A E R , †
M E H R A N A L A E E , ‡ B R E N T L E F E B V R E , §
A R V I N M O S E R , § A N T O N Y W I L L I A M S , § A N D
A N D R É J . S I M P S O N * , †
Department of Chemistry, University of Toronto Scarborough,
Toronto, Ontario, Canada, M1C 1A4, National Water Research
Institute, Environment Canada,
867 Lakeshore Road, P.O. Box 5050, Burlington,
Ontario, Canada, L7R 4A6, and Advanced Chemistry
Development Inc., 110 Yonge Street, 14th floor, Toronto,
Ontario, Canada, M5C 1T4
Received June 2, 2007. Revised manuscript received October
7, 2007. Accepted October 9, 2007.
Dissolved organic matter (DOM) contains a complex array of
chemical components that are intimately linked to many
environmental processes, including the global carbon cycle,
and the fate and transport of chemical pollutants. Despite its
importance, fundamental aspects, such as the structural
components in DOM remain elusive, due in part to the molecular
complexity of the material. Here, we utilize multidimensional
nuclear magnetic resonance spectroscopy to demonstrate the
major structural components in Lake Ontario DOM. These
include carboxyl-rich alicyclic molecules (CRAM), heteropolysac-
charides, and aromatic compounds, which are consistent
with components recently identified in marine dissolved organic
matter (1). In addition, long-range proton-carbon correlations
are obtained for DOM, which support the existence of material
derived from linear terpenoids (MDLT). It is tentatively
suggested that the bulk of freshwater dissolved organic
matter is aliphatic in nature, with CRAM derived from cyclic
terpenoids, and MDLT derived from linear terpenoids. This is in
agreement with previous reports which indicate terpenoids
as major precursors of DOM (2). At this time it is not clear in
LakeOntariowhether theseprecursorsareof terrestrialoraquatic
origin or whether transformations proceed via biological and/
or photochemical processes.
Introduction
Dissolved organic matter (DOM) is a complex, hetero-
geneous mixture found ubiquitously in nature. It comprises
a major mobile fraction of organic carbon on Earth and is
an intimate link between the terrestrial and aquatic environ-
ment (3, 4). Terrestrial and freshwater DOM experiences an
annual flux of approximately 0.4 × 1015 g C/year via riverine
discharge (5) to the marine environment. It is believed that
DOM plays a significant role in the enhanced solubility (6)
of chemical contaminants and may potentially be a shuttle
for the long-range transport of chemicals globally. Thus, the
cycling of DOM from freshwater to marine sources is not
only important in the global carbon cycle, but is a significant
mediator in the fate and transport of pollutants in the
environment. Despite this importance, there is still much to
be revealed regarding the structural components that make
up this complex environmental mixture and how these com-
pounds vary between freshwater and marine environments.
Here, dissolved organic matter from Lake Ontario, Canada
is studied in detail using multidimensional nuclear magnetic
resonance (NMR) spectroscopy. Lake Ontario covers just over
19000 km2, contains over 1600 cubic kilometers of freshwater
(7), and is part of the Great Lakes, which represent the world’s
largest freshwater lakes system. Recently a pivotal paper by
Hertkorn et al. (1) utilized a range of modern 1-D and 2-D
NMR approaches in combination with Fourier transform ion
cyclotron resonance mass spectrometry (FT-ICR-MS) to
identify carboxyl-rich alicyclic molecules (CRAM) in oceanic
DOM. This pioneering paper has been essential in providing
key assignments making further NMR based studies possible.
Here, we build upon the work of Hertkorn et al. (1) who have
reported on major structural and refractory components of
marine DOM, extending these initial findings to show that ma-
rine and freshwater DOM share many structural similarities.
Two dimensional solution-state NMR spectroscopy
(1, 8–15) and three-dimensional NMR (16, 17) are becoming
widely employed and very powerful techniques to study
structures and interactions in environmental chemistry.
Long-range proton-carbon correlations have been especially
useful for identifying structures in terrestrial derived materials
(9, 14, 18). However, collecting such data is extremely
challenging given the relatively low sensitivity of the experi-
ments and the fast relaxation in DOM. Combining recent
improved long-range NMR experiments with relaxation
optimized delays (19) permits weak long-range correlations
toberecordedforaquaticDOM.Thelong-rangeproton-carbon
correlations help confirm previous assignments of CRAM
and support the presence of an aliphatic material derived
from linear terpenoids in freshwater DOM.
Materials and Methods
Sample Preparation. Freshwater was pumped from a
depth of 50 cm and at a distance of 20 m from the Lake
Ontario shoreline (Darlington Provincial Park, Ontario,
Canada). Lake Ontario DOM (LO-DOM) was isolated as
described by Simpson et al. (20). Briefly, water from Lake
Ontario was prefiltered through 0.22 µm poly(vinylidene
difluoride) (PVDF) filters. DOM was isolated on diethylami-
noethyl-cellulose resin, recovered using 0.1 M NaOH, ion-
exchanged, (note: pH was adjusted to ∼6 after ion-
exchanging), and freeze-dried. Excess salts were removed
from the sample by extensive dialysis against double-distilled
water using 100 Dalton molecular weight cutoff cellulose
ester tubing. The sample was once again freeze-dried to
obtain a powder.
NMR Analysis. Sample (100 mg) was resuspended in 1
mL of deuterium oxide (D2O) and titrated to pH 13.1 using
NaOD (40% by weight) to ensure complete solubility. Samples
were analyzed using a Bruker Avance 500 MHz NMR
spectrometer equipped with a 1H-BB-13C 5 mm, triple
resonance broadband inverse probe. Average 1H relaxations
were estimated at ∼380 ms (T2) and ∼630 ms (T1) for the
LO-DOM. 1-D solution state 1H NMR experiments were
performed with 512 scans, a recycle delay of 3 s, 32768 time
domain points, and an acquisition time of 1.6 s. Solvent
suppression was achieved by presaturation utilizing relax-
* Corresponding author phone: 1-416-287-7547; fax: 1-416-287-
7279; e-mail: andre.simpson@utoronto.ca.
† University of Toronto Scarborough.
‡ Environment Canada.
§ Advanced Chemistry Development Inc.
Environ. Sci. Technol. XXXX, xxx, 000–000
10.1021/es0713072 CCC: $37.00  XXXX American Chemical Society VOL. xxx, NO. xx, XXXX / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 A
Published on Web 11/08/2007

ation gradients and echoes (21). Spectra were apodized
through multiplication with an exponential decay corre-
sponding to 1 Hz line broadening, and a zero filling factor
of 2. Diffusion-edited experiments were performed using a
bipolar pulse longitudinal encode-decode sequence (22).
Scans (1024) were collected using a 1.25 ms, 53.5 gauss/cm,
sine-shaped gradient pulse, a diffusion time of 50 ms, 8192
time domain points, 410 ms acquisition time, and a sample
temperature of 298 K. Spectra were apodized through
multiplication with an exponential decay corresponding to
10 Hz line broadening and zero filling factor of 2.
13C data were acquired using a 5 mm broadband probe
using WALTZ-16 decoupling. Scans (37888) were collected
using inverse gated decoupling, an 8 s recycle delay, and an
acquisition time of 186 ms. Multiplicity editing was carried
out using distortionless enhanced polarization transfer and
a quaternary only sequence (23). All 13C spectra were
processed with an exponential function corresponding to a
35 Hz line broadening.
Heteronuclear multiple quantum coherence (HMQC)
spectra were collected in phase-sensitive mode using echo/
antiecho gradient selection. 1024 scans were collected for
each of the 256 increments in the F1 dimension. 1024 data
points were collected in F2, a 1J 1H-13C value of 145 Hz, a
relaxation delay of 1 s was employed, and an acquisition
time of 77 ms. The F2 dimension was multiplied by an
exponential function corresponding to a 15 Hz line broad-
ening, whereas the F1 dimension was processed using a sine-
squared function with a π/2 phase shift and a zero-filling
factor of 2.
Heteronuclear multiple bond correlation (HMBC) were
carried out in phase-sensitive mode using echo/antiecho
gradient selection (19) and a relaxation optimized delay of
25 ms for the evolution of long-range couplings. 2048 scans
were collected for each of the 128 increments in the F1
dimension. 2048 data points were collected in F2, a relaxation
delay of 1 s, and acquisition time of 154 ms were used. The
F2 dimension was multiplied by an exponential function
corresponding to a 15 Hz line broadening, whereas the F1
dimension was processed using a sine-squared function with
a π/2 phase shift and a zero-filling factor of 2.
Spectral predictions were carried out using Advanced
Chemistry Development’s ACD/SpecManager and ACD/2D
NMR Predictor using Neural Network Prediction algorithms
(version 10.02). Parameters used for prediction including
spectral resolution, and base frequency were chosen to match
those of the real data sets as closely as possible. For
comparison of predicted versus real shifts, narrow line widths
of 2 Hz 1H and 5 Hz 13C were used to accurately evaluate the
predicted shift data. For HMQC, only one bond 1H-13C
couplings are considered, whereas in HMBC, only 2 and 3
bond couplings are considered, and in rare cases, 4 bond
couplings may also be included as a further investigative
tool. In Figure 4A the spectral line widths have been increased
to 10 Hz (1H) and 20 Hz (13C). This is for visualization purposes
only such that the crosspeaks are enlarged and easier to see.
Results and Discussion
Characterization of Lake Ontario DOM using 1D and
2DNMRdata. Figure 1 shows the 1H NMR data of freshwater
DOM from Lake Ontario (LO-DOM). Major structural com-
ponents present include aliphatics (I), carboxyl-rich alicyclic
molecules (CRAM) (II), carbohydrates (III), and aromatics
(IV). Further discussion is provided later in this paper. Signals
from larger macromolecular and/or aggregated species can
be further emphasized by the use of diffusion editing, which
retains only signals from macromolecular and/or aggregated
species (24). The diffusion edited spectrum (Figure 1B)
compared to that of the conventional 1H NMR spectrum
(Figure 1A) shows a generally similar profile, indicating that
the structures present as stable aggregates and/or macro-
molecular species have a generally similar composition when
compared to the sample as a whole. At this time, it is not
possible to distinguish whether the species are macromo-
lecular in nature or simply aggregated/associated due to the
high concentration of the sample, furthermore it is not
possible to provide any insight into how much of the sample
may be present in the form of aggregates/macromolecules.
Future studies based on diffusion ordered spectroscopy are
planned to address this aspect of DOM (16).
Figure 2 depicts the 13C spectrum (A) of the LO-DOM
sample, and the corresponding, scaled, multiplicity edited
spectra (B). The LO-DOM contains strong resonances from
CRAM (see ref 1 and later) and has a lower CH/CH2 ratio 1.6)
in comparison to oceanic DOM (ratio 2.3) isolated by
ultrafiltration (1). This higher relative abundance of CH2 vs
CH may be partially explained by the presence of materials
derived from linear terpenoids (MDLT). This component and
others present are discussed later in this paper.
Due to the large degree of overlap, extracting detailed
structural information from the 1D NMR alone is difficult.
2D NMR experiments provide increased spectral dispersion
as well as additional connectivity information, which permits
further characterization of the chemical functionalities
present in DOM. Figure 2C shows the HMQC NMR spectrum
for the LO-DOM sample. The HMQC experiment detects one
bond 1H-13C connectivities in an organic structure (13).
The HMQC NMR spectrum identifies a range of chemical
constituents present, including anomeric units in carbohy-
drates (1), functionalized and/or conjugated olefins (also see
“MDLT section” later for further considerations) (2), aromat-
ics (3), N-acetyl and/or O-acetyl, S-CH3 (4), aliphatics (5),
carboxyl-rich alicyclic molecules (6) (CRAM) ref (1) (see later
for discussion), methyl esters (7), methylene (CH2) from
carbohydrates (8), and methine (CH) from carbohydrates
(9). Note, assignments offered here are also consistent with
total correlation spectroscopy (TOCSY), nuclear overhauser
effect spectroscopy, and edited heteronuclear 2D correlations
(data not shown), as well as literature assignments (1). It is
interesting to note that the methoxy group from lignins, often
the most intense signal in soil organic matter (13), is not
present in LO-DOM, indicating that terrestrial inputs are
quickly transformed in Lake Ontario. The methyl ester region
(region 7, Figure 2C) should not be confused with the methoxy
from lignin which is not present in the LO-DOM sample. It
is important to add that elemental analysis (C:N ratio 20:1)
and TOCSY NMR data (not shown) were also collected for
the LO-DOM sample. TOCSY and HMQC (HMQC assignment
not discussed explicitly, due to its low abundance) contain
FIGURE 1. 1H NMR spectra showing (A) freshwater DOM from
Lake Ontario (LO-DOM), and (B) the diffusion edited spectrum
for LO-DOM. Resonances from I, aliphatics; II, CRAM; III,
carbohydrates; and IV, aromatics.
B 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. xxx, NO. xx, XXXX