Characterization of DOM as a function of MW by fluorescence EEM and HPLC-SEC using UVA, DOC, and fluorescence detection.

Abstract

To investigate the composition of dissolved organic matter (DOM) as a function of apparent molecular weight (MW) by rapid analytical methods, high performance liquid chromatography (HPLC)-size exclusion chromatography (SEC) was conducted with sequential on-line detectors consisting of UV, fluorescence, and quantitative DOC measurement. Fluorescence excitation-emission matrix (EEM) spectrophotometry was used to select wavelengths for the HPSEC on-line fluorescence system. The chosen peak maxima locations of excitation-emission wavelengths were 278-353 nm for protein-like substances and 337-423 nm for fulvic-like substances based on an analysis of EEM spectra for various samples and reference materials. This system provides quantitative and qualitative information on the specific MW components of DOM, including proportion of DOC (by DOC measurement), aromaticity (by comparison of UV and DOC measurements), and chemical properties (by fluorescence measurement). It further allows determination of organic matter characteristics (e.g., fulvic-like, protein-like, and polysaccharide-like substances) as a function of MW. Three types of samples (Irvine Ranch ground water (IRWD-GW), Barr Lake surface water (BL-SW), and Hawaii wastewater secondary effluent) were analyzed by the HPSEC-UVA-fluorescence-DOC system. These results were compared with fluorescence EEM for samples fractionated by HPLC-SEC. The DOM fraction in the high apparent MW range (over 10,000g/mol) consisted of polysaccharide-like substances for IRWD-GW and a mixture of polysaccharide-like/protein-like substances for BL-SW and wastewater secondary effluent. Minimal amounts of fulvic-like substances were found in the wastewater secondary effluent sample. The DOM fractions in a medium apparent MW range (5000-1000 g/M) showed higher aromaticity (fulvic in character) than any other fractions for all samples. For the DOM fraction in the low apparent MW range (below 680 g/M), additional aliphatic organic matter was found in IRWD-GW, while BL-SW contained protein-like processes. DOM plays an important role in drinking water and wastewater treatment processes. An enhanced HPSEC technique with multiple on-line detectors enables a better understanding of quantitative and qualitative DOM properties and can help to design and optimize water/wastewater treatment facilities.

Full text

Page 1

Water Research 37 (2003) 4295–4303
Characterization of DOM as a function of MW by fluorescence
EEM and HPLC-SEC using UVA, DOC, and
fluorescence detection
Namguk Hera, Gary Amya,*, Diane McKnighta, Jinsik Sohna,b, Yeomin Yoona,c
aCivil and Environmental Engineering, University of Colorado, ECOT 441, Boulder, CO 80309, USA
bKookmin University, Civil and Environmental Department, Seoul 136-702, South Korea
cArizona State University, Civil and Environmental Engineering, Tempe, AZ 85287, USA
Abstract
To investigate the composition of dissolved organic matter (DOM) as a function of apparent molecular weight (MW)
by rapid analytical methods, high performance liquid chromatography (HPLC)-size exclusion chromatography (SEC)
was conducted with sequential on-line detectors consisting of UV, fluorescence, and quantitative DOC measurement.
Fluorescence excitation-emission matrix (EEM) spectrophotometry was used to select wavelengths for the HPSEC on-
line fluorescence system. The chosen peak maxima locations of excitation-emission wavelengths were 278–353nm for
protein-like substances and 337–423nm for fulvic-like substances based on an analysis of EEM spectra for various
samples and reference materials.
This system provides quantitative and qualitative information on the specific MW components of DOM, including
proportion of DOC (by DOC measurement), aromaticity (by comparison of UV and DOC measurements), and
chemical properties (by fluorescence measurement). It further allows determination of organic matter characteristics
(e.g., fulvic-like, protein-like, and polysaccharide-like substances) as a function of MW. Three types of samples (Irvine
Ranch ground water (IRWD-GW), Barr Lake surface water (BL-SW), and Hawaii wastewater secondary effluent) were
analyzed by the HPSEC-UVA-fluorescence-DOC system. These results were compared with fluorescence EEM for
samples fractionated by HPLC-SEC.
The DOM fraction in the high apparent MW range (over 10,000g/mol) consisted of polysaccharide-like substances
for IRWD-GW and a mixture of polysaccharide-like/protein-like substances for BL-SW and wastewater secondary
effluent. Minimal amounts of fulvic-like substances were found in the wastewater secondary effluent sample. The DOM
fractions in a medium apparent MW range (5000–1000g/mol) showed higher aromaticity (fulvic in character) than any
other fractions for all samples. For the DOM fraction in the low apparent MW range (below 680g/mol), additional
aliphatic organic matter was found in IRWD-GW, while BL-SW contained protein-like processes.
DOM plays an important role in drinking water and wastewater treatment processes. An enhanced HPSEC
technique with multiple on-line detectors enables a better understanding of quantitative and qualitative DOM
properties and can help to design and optimize water/wastewater treatment facilities.
r 2003 Elsevier Ltd. All rights reserved.
Keywords: HPLC; SEC; Apparent MW; DOM; EEM
1. Introduction
Natural organic matter (NOM) present in ground and
surface waters is a complex heterogeneous mixture
composed of humic acids, fulvic acids, low molecular
weight (MW) organic acids, carbohydrates, proteins,
ARTICLE IN PRESS
*Corresponding author. Tel.: +1-303-492-6274; fax: +1-
303-492-7317.
E-mail address: gamy@spot.colorado.edu (G. Amy).
0043-1354/03/$-see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0043-1354(03)00317-8

Page 2

and other compound classes [1,2]. The understanding of
dissolved organic matter (DOM) properties as a func-
tion of size, in addition to the MW of DOM, is a crucial
factor to determine DOM treatability [1,3]. Recently,
aliphatic and partially aromatic DOM, in addition to
highly aromatic DOM, have been shown to be
important in water treatment. Specifically, polysacchar-
ides and proteins were found to be major foulants in
nanofiltration [4,5].
In spite of the importance of understanding the
characteristics of DOM, most chemical or physical
analyses have relied on traditional analyses (e.g., light
absorptivity, DOC concentration, aromaticity, fluores-
cence, XAD fractionation, and apparent MW) due to
the greater effort required for detailed structural
analysis. A previous study investigated the structure of
MW fractions after sample fractionation by dialysis
using numerous analytical approaches (NMR, pyrolysis-
GC/MS, IR, and thermogravimetric methods) [6].
However, this kind of analysis may not be generally
applicable because it is time consuming, expensive, and
requires considerable operator expertise. Therefore, it
would bevaluable to
methods that require only a small sample volume and
minimal pretreatment to aid in water and wastewater
treatment optimization. Of particular importance is the
ability to analyze bulk water samples as opposed to
samples subjected to isolation, fractionation, and/or
concentration.
DOM includes organic molecules with chromophoric
(light absorbing) and fluorophoric (light emitting)
moieties. Analysis of water samples by fluorescence
EEM is becoming increasingly widespread due to the
potential to interpret DOM fluorescence properties and
high instrument sensitivity [7,8]. Fluorescence spectro-
metry permits high optical resolution and the determi-
nation of excitation-emission matrices (EEMs) [8]. The
fluorescence in different spectral regions is associated
with different types of functional groups [9]. Therefore,
DOM fractions, alone or within a bulk sample, may
have their own characteristic EEMs.
The objective of this research was to develop a high-
performance size exclusion chromatography (HPSEC)
system combined with multiple detectors to characterize
DOM across a broad range of water samples as a
function of apparent MW. The HPSEC-ultraviolet
absorbance (UVA)-dissolved organic carbon (DOC)
detection system provides qualitative information (e.g.,
a specific UVA (SUVA) chromatogram as a function of
apparent MW by the relative ratio between UVA254and
DOC chromatograms) in addition to quantitative
information on size distribution [10,11]. However, this
system is not sufficient to identify chemical and physical
properties of the particular MW DOM components.
HPSEC fluorescence results obtained at different wave-
lengths may be used to discriminate the DOM
develop rapidanalytical
components as fulvic acids, humic acids, proteins, and
polysaccharides with the combination of UV and DOC
detection, all as a function of apparent MW.
2. Experimental methods
2.1. Fluorescence EEM
Fluorescence EEM spectra were measured by a JY-
Horiba/Spex Fluoromax-2 fluorometer with a xenon
lamp as the excitation source. EEM spectra are a
collection of a series of emission spectra over a range of
excitation wavelengths. In this experiment, EEM spectra
were collected with subsequent scanning emission
spectra from 250 to 600nm at 2nm increments by
varying the excitation wavelength from 200 to 500nm at
10nm increments. Spectral subtraction was performed
to remove blank spectra mainly caused by Raman
scattering.
DOM fractions can be categorized in terms of
carbohydrates, proteins, humic acids, and fulvic acids.
The fluorescence EEM spectra for four different organic
model compounds representative of DOM fractions
were measured: dextran (MW 40,000, no-aromaticity),
bovine serum albumin (MW 66,000, low-aromaticity),
Suwannee River humic acid (SRHA, with larger MW
and high aromaticity), and Suwannee River fulvic acid
(SRFA, with lower MW and high aromaticity). All
samples were filtered with a 0.45mm cartridge (nylon
filter) prior to analysis and 2mL sample volumes (1cm
cuvette) were used for the EEM spectra. Reference
samples were dissolved in HPLC eluent. The pH and
ionic strength of bulk (natural and wastewater) source
waters were adjusted to the same as the HP-SEC mobile
phase (pH 6.8 with phosphate buffer and ionic strength
0.1M) because EEM data would be used to set
wavelengths for measurement of fluorescence in the
HPLC-SEC procedure. Maintaining the same pH for all
samples prevents a spectral shift caused by changing pH
[12,13]. Solution concentrations were prepared from 1.5
to 2.0mg/L with dissolution (reference samples) or
dilution (bulk samples) to reduce primary and secondary
inner filtering effects [13,14]. The EEM spectra were not
significantly affected by changing ionic strengths of
samples dissolved in a Millipore Milli-Q (MQ) up to
0.1M (Na2SO4). These results were in agreement with
other researchers [12].
2.2. HPSEC-UVA-fluorescence-DOC
Separation by size exclusion was performed using a
TSK-50S (Toyopearl HW SOS, 30mm resin) column
prior to sequential on-line detectors consisting of UV
(SPD-6A Shimadzu UV detector), fluorescence (Waters
470 Scanning Fluorescencedetector),and DOC
ARTICLE IN PRESS
N. Her et al. / Water Research 37 (2003) 4295–4303
4296

Page 3

(Modified Sievers Total Organic Carbon Analyzer 800
Turbo). The instrumental setup of the assembly is
depicted in Fig. 1. The system has been described
previously except for the additional fluorescence detec-
tion [11]. The response of on-line fluorescence connected
to HPLC depends highly on the chosen excitation and
emission wavelengths. Through the results of EEM
spectrophotometry, the peak maxima locations for
protein-like substances and fulvic-like substances were
selected and applied to the HPSEC-fluorescence system.
A 20?200mm stainless steel HPLC column was
employed at room temperature with a phosphate buffer
eluent (0.0024M NaH2PO4+0.0016M Na2HPO4, pH
6.8) containing 0.025M sodium sulfate (Na2SO4),
producing an ionic strength of 0.1M. UVA, fluores-
cence, and DOC data were collected every 6s. The
intensity of the chromatograms for UVA at 254nm and
DOC were both measured in volts and calibrated to
general units as cm?1for UVA and mg/L for-DOC by
direct input of potassium hydrogen phthalate (PHP)
standard solutions (0.05, 0.1, 0.5, 1, 3, and 6mg/L) to
the HPLC. Potassium hydrogenphthalate (PHP) stan-
dards were used for calibration to quantify UV and
DOC peak heights or areas of chromatograms. Poly-
ethylene glycols (PEGs, 200–10,000g/mol) were used for
MW calibration of chromatograms. A relatively large
2mL sample volume was used to enhance DOC
detection and the flow rate was 1mL/min.
The pH (6.8) and ionic strength (0.1M) of each
sample was also adjusted with phosphate buffer and
sodium sulfate solutions (as similar to the mobile phase
as possible) before analysis to provide the same pH and
ionic strength for all samples and to reduce interactions
between NOM and the column packing material. Fig. 2
demonstrates the effects of sample conditions. The SEC
chromatogram of the SRHA dissolved in ultra-pure
(Milli-Q) water displays two peaks. However, the salt
boundary peak was significantly lowered or disappeared
for the SEC chromatogram of the SRHA dissolved in
eluent, indicating lower interactions between NOM and
the column resin. Therefore, a highly concentrated
solution of the mobile phase has been utilized to adjust
the ionic strength of the sample using a conductivity
meter.
In this study, DOM samples from two natural waters
(Irvine Ranch ground water (IRWD-GW), Barr Lake
surface water (BL-SW), and a secondary effluent from a
wastewater treatment plant in Hawaii were evaluated.
IRWD-GW in California was a blended sample that was
collected at five different depths (from 800 to 2000ft,
with different characteristics depending on depth (higher
DOC and SUVA with increasing depth)) from two wells;
this ground water contains humic material derived from
subsurface organic deposits. BL-SW is a hypereutrophic
water containing algae and extracellular products,
representing algal organic matter (AOM). The second-
ary effluent contains soluble microbial products, derived
from biological treatments as well as NOM from the
original drinking water source, representing effluent
organic matter (EfOM).
Using a fraction collector, fractions were also
collected from the HPSEC system (Fig. 1). Each
fraction was collected at the outlet of the UVA detector
to obtain EEM spectra for a comparison of data
between HPSEC-fluorescence and EEM. The Fluoro-
max-2 fluorometer allows adequate fluorescence EEM
sensitivity without pre-concentration for each fraction
sample, which was significantly diluted by the eluent
during HPSEC.
ARTICLE IN PRESS
Fig. 1. Schematic of HPLC-UVA-fluorescence-DOC.
Fig. 2. UVA and DOC peaks pf SRHA isolate diluted in MQ and eluent (TSK SOS Column (2?25cm), SRHA, Na2SO4eluent, ImL/
min of eluent, acid 2mL/min and oxidizer 2mL/min for DOC, phosphate buffer, ionic strength 0.1M).
N. Her et al. / Water Research 37 (2003) 4295–4303
4297

Page 4

3. Results and discussion
The EEM spectra of reference samples are shown in
Fig. 3. The intensity of EEM is represented by contour
lines. The intensity of fluorescence EEM for dextran was
negligible as expected with an erroneous response
(Fig. 3a). Albumin, SRHA, and SRFA (Fig. 3b–d) show
two pairs of peak maxima locations. The fluorescence
maxima of SRHA (Ex: 325nm-Em: 452nm) shows a red
shift (longer excitation and emission wavelengths)
compared to SRFA (Ex: 320nm-Em: 443nm), as
observed in previous studies of humic substances [12].
AOM, extracted from blue green algae, shows multiple
peaks (Fig. 3e), indicating that AOM is composed of
many components. In the case of the BL-SW, even
though it shows only one peak (Fig. 3f), the shape
(highly oriented to shorter excitation and emission
wavelengths) implies the existence of a peak at shorter
excitation and emission wavelengths.
Additional EEMs of reference samples were com-
pared in Fig. 4a and b to select representative peak
locations for protein-like and fulvic-Hke biopolymers.
AOM and secondary effluent, showing distinctive peak
maxima locations have been used for identifying both
biopolymers. For humic and fulvic acids, the peak
locations at longer excitation and emission wavelengths
were chosen from two peaks appearing in Fig. 3 because
some natural water samples did not show the peak at
shorter excitation and emission wavelengths. In the case
of protein-like biopolymers, peak maxima locations of
ARTICLE IN PRESS
Fig. 3. Fluorescence EEM of (a) dextran, (b) Albumin, (c) SRHA, (d) SRFA, (e) algal organic matter (AOM), and (f) BL-SW.
N. Her et al. / Water Research 37 (2003) 4295–4303
4298

Page 5

albumin and lysozyme were similar to those appearing in
natural waters and wastewater effluent (Fig. 4a). There-
fore, excitation and emission wavelengths for detecting
protein-like substances were specified as 278 and 353nm
(center of sample locations).
In the case of fulvic-like substances, the EEM peak
locations for natural and wastewater samples showed
significant differences from those of Suwannee River
(SR) isolates (Fig. 4b). The fluorescence EEM peak
maxima of SR isolates appeared at shorter excitation
and longer emission wavelengths than those of natural
water and wastewater samples. The excitation and
emission wavelengths chosen to detect fulvic-like sub-
stance were 337 and 423nm (center values). The results
of SR isolates were not used for determining center
values due to the significant deviation compared to bulk
water results.
Fig. 5 shows HPSEC-UVA-fluorescence-DOC chro-
matograms for IRWD-GW, BL-SW, and wastewater
secondary effluent. Two sample injections were per-
formed in series to obtain fluorescence chromatograms
at different wavelengths. All UVA, DOC, and SUVA
data in Table 1 were obtained based on the areas of
designated fractions in each chromatogram (values in
parentheses in Table 1 represent areas). IRWD-GW had
a DOC concentration of 11.9mg/L and a UVA of
0.79cm?1(SUVA: 6.59L/m-mg). The DOC concentra-
tion of BL-SW was 10.4mg/L and its UVA value was
low, 0.16cm?1(SUVA: 1.50L/m-mg). The DOC con-
centration of the secondary effluent was 12.9mg/L and
the UVA value was 0.29cm?1(SUVA: 2.21L/m-mg).
While the injected DOC concentrations were above
10mg/L, thehighest DOC
matograms shown in Fig. 5 were around 1mg/L
values inthe chro-
for all samples due to the greater than 10 fold
dilution by the eluent during size exclusion chromato-
graphy.
Based on the DOC chromatograms, five different
molecular size fractions were identified for each sample.
The apparent MW of fraction 1 was around 21,000g/
mol for all samples, believed to be polysaccharides [10].
However, the fraction 1 components show different
characteristics (Table 1) among the different samples.
For IRWD-GW, fraction 1 had a low SUVA value
(0.43L/m-mg) and nearly no fluorescence intensities of
protein-like substances obtained at EX: 278nm–EM:
353nm (Fluprotein, 0.04 area units) and fulvic-like
substances obtainedatEX:
(Flufulvic, 0.001 area units). The fraction 1 from BL-
SW had the lowest SUVA value (0.10L/m-mg) but
relatively high Fluprotein intensity (0.21 area units).
Fraction 1 in the secondary effluent had a relatively
high SUVA (0.99L/m-mg) with higher intensity of
Fluprotein (0.95 area units) and minimal intensity of
Flufulvic(0.034 area units). These results indicate that the
composition of fraction 1 is not the same for these three
samples. The fraction 1 of IRWD-GW mainly contains
a polysaccharide-like substance without other compo-
nents. The BL-SW fraction 1 appears to have a certain
amount of protein-like substances together with poly-
saccharide-like substances. In the case of secondary
effluent fraction 1, not only polysaccharide-like and
protein-like substances but also minimal amounts of a
fulvic-like substance were found. Through the chroma-
tograms portrayed by specific fluorescence (DOC-
normalized fluorescence obtained with fluorescence
height divided by DOC height at each MW), a higher
MW of protein-like substances was observed compared
337nm–EM:423nm
ARTICLE IN PRESS
Fig. 4. Peak maxima locations for (a) protein-like and (b) fulvic-like substances. AOM: algal organic matter extracted from blue-green
algae. Hawaii WW: secondary effluent of wastewater treatment in Hawaii (sampled in August 2000). IRWD-GW: Irvine Ranch Water
District ground water (sampled in April 2000). BL-SW: Barr Lake surface water (sampled in April 2001). Oise River: NF feed water
from Oise River in France (sampled in May 2001). Fl-GW: Florida ground water (sampled in April 2001). SRNOM: Suwannee River
NOM. Raand Rb: Upstream (Ra) and downstream (Rb) locations of Twizell Burn River in northeastern England (Data taken from [8]).
N. Her et al. / Water Research 37 (2003) 4295–4303
4299