Asymmetrical flow field-flow fractionation (AF4) has been used to characterise the size and organic carbon concentration of colloids in environmental samples, without extensive pre-treatment. The addition of a true online organic carbon detector (OCD) to the combination of UV, fluorescence and multi-angle light scattering detectors provides a unique tool for a better understanding of the composition of dissolved and particulate organic matter (OM) present in environmental samples. Polystyrene sulfonate sodium salt polymers were used as standards. Colloidal material was characterised in surface water, road runoff from a residential area and the effluent from a recirculating aquaculture system. All samples contained organic or inorganic colloids associated with organic carbon to various degrees and with various sizes. The OCD clearly added another layer of information, allowing for detection of organic colloids independent of their ability to scatter light or absorb UV-light. The recovery of carbon varied between 10% and 100% depending on the nature of the sample.

INTRODUCTION

Organic matter (OM) is ubiquitous in the environment, but size, form and character may vary, ranging from truly dissolved (size below 1 nm), colloidal (size between 1 nm and 1 μm) or particulate (size greater than 1 μm). In addition, dissolved organic matter (DOM) is frequently characterised as being smaller than 0.45 μm. It can be classified by its hydrophobic, transphobic or hydrophilic character. Colloidal and particulate OM can have gel-like structures, forming nano-, micro- and macro-gels as well as transparent exopolymer particles (TEP) (Verdugo et al. 2004; Bar-Zeev et al. 2015). Aquatic OM originates from a wide variety of sources (i.e. leaching from soil, decay of organic material, microorganisms), and is degraded and transformed in many different ways (i.e. biological, chemical, photocatalytic), leading to a complex structure and mixture of the resulting substances (Frimmel 1998).

The role of OM in the environment is complex, but may have implications for human health and water treatment. OM may be associated with the removal, mobility and bioavailability of metal ions in the environment (Tang et al. 2014), and can be a carrier of hydrophobic organic compounds (Sillanpää 2014). OM can cause colour, taste and odour problems, react with oxidants to produce harmful disinfection by-products and interact with membranes to cause fouling. Furthermore, OM may contribute to corrosion and act as a substrate for bacterial growth in water distribution systems (Sillanpää 2014).

Due to the complexity of OM, its broad size-range and often low concentration, advanced and sensitive analytical techniques are needed for characterisation. Liquid chromatography with organic carbon detection, a combination of size exclusion chromatography (SEC) and organic carbon detection (OCD), has been successfully applied to the characterisation of DOM (Huber et al. 2011). The OCD has a very high sensitivity with a detection limit in the lower ppb range (Huber & Frimmel 1991). However, the application of SEC columns limits the sample range to constituents that do not interact with the gel in the column, and thus has limitations when characterising colloids. The gel may alter the samples due to shear forces and loss of certain fractions during analysis may occur. Samples are commonly pre-filtered using a 0.45 μm filter, a procedure that may significantly alter the sample composition. SEC is rather inflexible, since a change in separation capability can only be achieved by a change of the column material.

Alternatively, asymmetric flow field-flow fractionation (AF4) may be chosen, being a standard technique for the analysis of proteins and polymers in biochemistry and pharmaceutical research (Fraunhofer & Winter 2004; Liu et al. 2006). The high resolving power and the gentle treatment of particles during the size separation process, without the need of sample pre-treatment, make AF4 an ideal technique for characterisation of colloids in water (Gimbert et al. 2003; Boller & Kaegi 2011). However, its application in a water treatment context is still rare. Beckett et al. (1987) have demonstrated that the molecular weight distribution of humic and fulvic acids can be determined using the flow field-flow fractionation technique when combined with UV detection. Moon et al. (2006) characterised natural organic matter (NOM) as nanoparticles using flow field-flow fractionation and fluorescence detection in addition to UV detection. Kammer et al. (2011) published an extensive overview of studies characterising nanoparticles in complex food and environmental samples, none of which employed online carbon detection.

In contrast to SEC, the full colloidal range can be investigated by AF4, including macrogels and TEP, provided that organic carbon can be measured directly. However, this was difficult to achieve until recently. Reszat & Hendry (2005) modified a total organic carbon (TOC)-analyser to allow analysis of organic carbon at the low flow rate normally used in AF4. With a channel flow rate of 1.5 mL min−1, fractograms could be recorded with one data point every 4 seconds. The study claimed high detector sensitivity in the lower ppb range, but lacked proof and presented only an analysis of samples with dissolved organic carbon (DOC) higher than 18.6 mg/L.

This paper shows that AF4 combined with the online OCD provides sensitive organic carbon measurements. Combined with UV, fluorescence and multi-angle light scattering (MALS) detectors, it can be used for the characterisation of environmental samples without extensive pre-treatment. The addition of the OCD to the AF4 system provides a novel and powerful tool for characterisation of the different fractions of DOC and colloids in environmental samples.

MATERIALS AND METHODS

AF4 system

An AF2000 Multi Flow FFF-system from Postnova Analytics (Landsberg, Germany) was employed for sample fractionation in this study. Separation takes place in a channel consisting of a solid top cover, a spacer and a permeable membrane. A laminar flow carries the sample through the separation chamber and a separation field, referred to as cross flow, is applied perpendicular to the channel flow. Due to the higher diffusion coefficient of smaller particles relative to the larger ones and the parabolic flow profile of the channel flow, smaller particles reach higher and thus faster flow rates, allowing them to elute first. An extensive review of the technology applied for separation of colloids and polymers can be found elsewhere (Baalousha et al. 2011; Yohannes et al. 2011).

For the fractionation, a 1 kDa polyether sulfone membrane was used with a pore size of approximately 1 nm. A 350 nm spacer was used together with 10 mM NaCl as eluent. A linear cross flow gradient was applied and the detector flow rate was 0.5 mL/min. The AF4 system was coupled with several online detectors including UV 254 nm, fluorescence, MALS and OCD. The UV signal was measured with a Prominence UV/VIS detector SPD-20A (Shimadzu, Japan). Fluorescence was measured with a Jasco FP-920 detector, with excitation at 330 nm and emission at 450 nm, in order to target humic acid-like material (Chen et al. 2003). The MALS signal was measured with a PN3621 (21-Angle MALS, Postnova Analytics, Landsberg, Germany), and the radius of gyration was calculated based on weight averaging. The OCD detector (DOC Labor, Karlsruhe, Germany) oxidised organic carbon in a Gräntzel thin-film reactor and CO2 was monitored by infrared detection. At the inlet of the OCD, the solution was acidified to convert carbonates to carbonic acid. Detector signals were recorded with the AF2000 Postnova Analytics software version 1.2.0.19.

Polymer standards

The particle size was determined by comparison with standards as well as by direct MALS measurements. Polystyrene sulfonate sodium salt polymer standards (PSS) of 3 kDa, 33 kDa and 152 kDa (Postnova Analytics, Landsberg, Germany) were used to demonstrate the separation capacity.

Environmental samples

Sample selection consisted of surface water containing NOM, road runoff during a rain event and water from a recirculating aquaculture system (RAS). Grand River was used as the surface water sample for this study, being representative of a municipally and agriculturally impacted drinking water source used by North American water utilities. The sample was taken after roughing filtration pre-treatment, which was used to reduce peak concentration of suspended material prior to biofiltration. The concentration of TOC and DOC of the sample were respectively 6.26 and 6.24 mg/L. The road runoff sample was collected in a residential area in Trondheim city centre within the first hour of a rain event. The sample contained runoff from the actual road, but also from the sidewalk and a nearby park, and had very high TOC and filtered chemical oxygen demand (COD) of 363 mg/L and 3,678 mg/L O2, respectively. The RAS sample was effluent from a rearing tank, originating from a conventional system containing post-smolt and had a TOC of 1.5 mg/L. Both samples were pre-filtered by a 5 μm nylon filter (Spectra Mesh®, VWR Norway).

RESULTS AND DISCUSSION

The potential of AF4 coupled with online OCD detection, in addition to UV, fluorescence and MALS, for characterising the colloidal organic fraction of environmental samples was explored.

Separation of organic polymer standards

A solution containing polystyrene sulfonate polymers of 3, 33 and 152 kDa was separated and characterised using UV and OCD. Figure 1 shows that molecules with sizes starting from 3 kDa can be separated with AF4. It can be seen that the OCD adds a layer of information to the method, since it is now possible to directly measure the organic carbon content of a sample.
Figure 1

Separation example for polystyrene sulfonate, 3 kDa, 33 kDa and 152 kDa (separation conditions: injection volume 1 mL, focusing time 10 minutes, cross flow 3.2 mL/min).

Figure 1

Separation example for polystyrene sulfonate, 3 kDa, 33 kDa and 152 kDa (separation conditions: injection volume 1 mL, focusing time 10 minutes, cross flow 3.2 mL/min).

The OCD is very sensitive and the signal corresponds well with the UV-signal. While PSS is visible by UV254 absorption and also by fluorescence, the OCD has a clear advantage if organic substances are analysed which lack chromophores or fluorophores, such as polysaccharides or TEP. One disadvantage of the OCD may be the tendency of peak broadening, a consequence of the large reactor volume for oxidation and subsequent CO2 transport. It should be noted that the peaks in Figure 1 are corrected for time shift but not for peak broadening.

A natural surface water analysed by OCD and MALS

Figure 2(a) presents a typical AF4 fractogram obtained from Grand River water and is composed of two fractions. Colloidal components elute from the separation channel in order of increasing size, containing an organic carbon concentration of 0.18 mg/L. The first peak (0–24 minutes), identified by the OCD and UV detectors, has the lowest size according to the AF4 separation theory and a strong UV signal. The organic carbon content of the first fraction is 0.04 mg/L. The small MALS signal suggests a low concentration of colloidal matter in this fraction, therefore the size determination of this fraction with MALS is not available. However, comparison with the PSS standard indicates that a low molecular weight material of 3 kDa is present in this fraction. The high UV signal intensity suggests the presence of aromatic structure such as NOM. A second fraction was identified by the OCD, UV254 and MALS detectors. The organic carbon concentration of this fraction is 0.13 mg/L and the UV signal has significantly decreased, suggesting a different nature for the particulate material. Figure 2(b) presents the radius distribution and the corresponding UV signal for the elution time between 36 and 42 minutes. The calculated radius of gyration of this fraction is 469 ± 49 nm.
Figure 2

AF4 fractogram of (a) drinking water source and (b) size distribution (separation conditions: injection volume 5 mL, focusing time 30 minutes, cross flow 3.0 mL/min).

Figure 2

AF4 fractogram of (a) drinking water source and (b) size distribution (separation conditions: injection volume 5 mL, focusing time 30 minutes, cross flow 3.0 mL/min).

Road runoff from a residential area

The main goal of coupling the AF4 with OCD was to characterise complex environmental samples, such as urban runoff, as shown in Figure 3. As observed with the natural sample, the runoff sample may also contain several kinds of OM. The sample can be divided into three main fractions. The first fraction (0–15 minutes) contains OM with a low molecular weight of around 3–6 kDa (compared to PSS standards), possibly NOM, as indicated by the strong UV and fluorescence response. The DOC concentration was 14.2 mg/L.
Figure 3

AF4 fractogram of road runoff in a residential area in Trondheim (separation conditions: injection volume 1 mL, focusing time 20 minutes, cross flow 2.5 mL/min, sample dilution 1:10).

Figure 3

AF4 fractogram of road runoff in a residential area in Trondheim (separation conditions: injection volume 1 mL, focusing time 20 minutes, cross flow 2.5 mL/min, sample dilution 1:10).

Fraction 2, ranging from 15 to 28 minutes, is characterised by a strong OCD signal (corresponding to 15.6 mg/L) accompanied by a much lower UV signal compared to Fraction 1. The MALS signal is increasing, hinting at increased particle size. Fraction 3, ranging from 28 to 55 minutes, is dominated by the MALS signal, while the OCD and UV are fading out. The carbon content of this fraction was estimated to be 9.10 mg/L. MALS analysis revealed an average radius of gyration of 154 ± 35 nm around the peak maxima of Fraction 2, which increases to 424 ± 41 nm at the overlap between Fractions 2 and 3, and further to 809 ± 22 nm at the maximum of Fraction 3. From an initial TOC of 363 mg/L, only 38.9 mg were recovered, corresponding to 11%.

Organic colloids in effluent from a RAS

The RAS sample showed a different pattern compared to the two previous samples (see Figure 4). The sample shows weak signs of NOM-like substances with a rather small UV and fluorescence intensity around 5 minutes elution time. However, a strong OCD signal was found, showing a first maximum around 11 minutes and a second peak appearing at 35 minutes. Based on an evaluation of the peak shape, it may be assumed that there are several organic fractions combined within the first carbon peak, accumulating up to 1.1 mg/L organic carbon. The second OCD peak is accompanied by an increased MALS signal which indicates particles with a radius of gyration of 637 ± 56 nm and a carbon concentration of 0.39 mg/L. Compared to the initial TOC of the sample, about 100% of the carbon was recovered within the colloidal fraction.
Figure 4

AF4 fractogram of effluent from RAS (separation conditions: injection volume 2.5 mL, focusing time 30 minutes, cross flow 2.5 mL/min).

Figure 4

AF4 fractogram of effluent from RAS (separation conditions: injection volume 2.5 mL, focusing time 30 minutes, cross flow 2.5 mL/min).

The RAS sample clearly demonstrates the benefits of coupling the OCD and the AF4. In a conventional configuration without an OCD detector, it may be a challenge to confirm the presence of organic colloids or organic carbon associated with inorganic matter. The additional layer of information provided by the OCD detector is important in order to be able to draw conclusions about the impact of the presence of such colloids on the system they are contained in, i.e. a RAS tank, the fate of the colloids (and possibly associated pollutants) in the environment, or treatment required to remove such colloids.

While the OCD has been successfully used in combination with SEC, demonstrating high oxidation and recovery rates, one may argue that this may not be the case as particle size is increasing, once the whole colloidal range is considered. For dissolved material smaller than 0.45 μm in size, Huber et al. (2011) concluded that the OCD gives quantitative or close to quantitative results, except for triazines and some N-heterocyclic compounds. For larger materials, such analysis is lacking at this time.

CONCLUSIONS

This study shows that the addition of a sensitive online OCD to the AF4 detector pool offers a novel approach, a strong improvement of the technique, and extends significantly the areas of application. For the first time, this detector combination allows the direct measurement of organic carbon, adding a new layer of information. Therefore, the relevance of flow field-flow fractionation is extended for the characterisation of environmental water samples.

The addition of OCD allows analysis and characterisation of colloidal organic carbon, independent of its capability to scatter light, fluoresce or absorb UV. This technique can illuminate the grey zone between dissolved and particulate organic matter. However, future research is required in order to gain more information about the oxidation efficiency of the OCD for colloidal matter.

REFERENCES

REFERENCES
Bar-Zeev
E.
Passow
U.
Castrillon
S. R.-V.
Elimelech
M.
2015
Transparent exopolymer particles: from aquatic environments and engineered systems to membrane biofouling
.
Environmental Science & Technology
49
(
2
),
691
707
.
Boller
M.
Kaegi
R.
2011
Characterization of aquatic particles
. In:
Handbook on Particle Separation Processes
(
Nieuwenhuijzen
A.F. Van
Van der Graaf
J.
, eds).
IWA Publishing
,
London
,
pp. 15–60
.
Chen
W.
Westerhoff
P.
Leenheer
J. A.
Booksh
K.
2003
Fluorescence excitation – emission matrix regional integration to quantify spectra for dissolved organic matter
.
Environmental Science & Technology
37
(
24
),
5701
5710
.
Fraunhofer
W.
Winter
G.
2004
The use of asymmetrical flow field-flow fractionation in pharmaceutics and biopharmaceutics
.
European Journal of Pharmaceutics and Biopharmaceutics
58
(
2
),
369
383
.
Frimmel
F. H.
1998
Characterization of natural organic matter as major constituents in aquatic systems
.
Journal of Contaminant Hydrology
35
(
1–3
),
201
216
.
Gimbert
L. J.
Andrew
K. N.
Haygarth
P. M.
Worsfold
P. J.
2003
Environmental applications of flow field-flow fractionation (FIFFF)
.
TrAC Trends in Analytical Chemistry
22
(
9
),
615
633
.
Huber
S. A.
Frimmel
F. H.
1991
Flow-injection analysis of organic and inorganic carbon in the low-ppb range
.
Analytical Chemistry
63
(
19
),
2122
2130
.
Kammer
F. von der
Legros
S.
Hofmann
T.
Larsen
E. H.
Loeschner
K.
2011
Separation and characterization of nanoparticles in complex food and environmental samples by field-flow fractionation
.
TrAC Trends in Analytical Chemistry
30
(
3
),
425
436
.
Moon
J.
Kim
S. H.
Cho
J.
2006
Characterizations of natural organic matter as nano particle using flow field-flow fractionation
.
Colloids and Surfaces A: Physicochemical and Engineering Aspects
287
(
1–3
),
232
236
.
Sillanpää
M. E. T.
2014
Natural Organic Matter in Waters: Characterization and Treatment Methods
.
Elsevier
,
Waltham, MA, USA
.
Tang
W. W.
Zeng
G. M.
Gong
J. L.
Liang
J.
Xu
P.
Zhang
C.
Huang
B. B.
2014
Impact of humic/fulvic acid on the removal of heavy metals from aqueous solutions using nanomaterials: a review
.
Science of the Total Environment
468–469
,
1014
1027
.
Verdugo
P.
Alldredge
A. L.
Azam
F.
Kirchman
D. L.
Passow
U.
Santschi
P. H.
2004
The oceanic gel phase: a bridge in the DOM–POM continuum
.
Marine Chemistry
92
(
1–4
),
67
85
.
Yohannes
G.
Jussila
M.
Hartonen
K.
Riekkola
M.-L.
2011
Asymmetrical flow field-flow fractionation technique for separation and characterization of biopolymers and bioparticles
.
Journal of Chromatography A
1218
(
27
),
4104
4116
.