Water Research 101 (2016) 510e518

Contents lists available at ScienceDirect

Water Research
journal homepage: www.elsevier.com/locate/watres

Quantitative detection of powdered activated carbon in wastewater

treatment plant efuent by thermogravimetric analysis (TGA)
ver, Julia Plattner, Thomas Wintgens*
Therese Krahnsto
Institute for Ecopreneurship, School of Life Sciences, University of Applied Sciences and Arts Northwestern Switzerland, Grndenstrasse 40, 4132 Muttenz,
Switzerland

a r t i c l e i n f o a b s t r a c t

Article history: For the elimination of potentially harmful micropollutants, powdered activated carbon (PAC) adsorption
Received 25 January 2016 is applied in many wastewater treatment plants (WWTP). This holds the risk of PAC leakage into the
Received in revised form WWTP efuent and desorption of contaminants into natural water bodies. In order to assess a potential
27 May 2016
PAC leakage, PAC concentrations below several mg/L have to be detected in the WWTP efuent. None of
Accepted 3 June 2016
Available online 6 June 2016
the methods that are used for water analysis today are able to differentiate between activated carbon and
solid background matrix. Thus, a selective, quantitative and easily applicable method is still needed for
the detection of PAC residues in wastewater.
Keywords:
Micropollutants
In the present study, a method was developed to quantitatively measure the PAC content in waste-
Wastewater water by using ltration and thermogravimetric analysis (TGA), which is a well-established technique for
Powdered activated carbon the distinction between different solid materials. For the sample ltration, quartz lters with a tem-
Thermogravimetric analysis perature stability up to 950
C were used. This allowed for sensitive and well reproducible measure-
ments, as the TGA was not affected by the presence of the lter. The samples mass fractions were
calculated by integrating the mass decrease rate obtained by TGA in specic, clearly identiable peak
areas. A two-step TGA heating method consisting of N2 and O2 atmospheres led to a good differentiation
between PAC and biological background matrix, thanks to the reduction of peak overlapping. A linear
correlation was found between a samples PAC content and the corresponding peak areas under N2 and
O2, the sample volume and the solid mass separated by ltration. Based on these ndings, various
wastewater samples from different WWTPs were then analyzed by TGA with regard to their PAC content.
It was found that, compared to alternative techniques such as measurement of turbidity or total sus-
pended solids, the newly developed TGA method allows for a quantitative and selective detection of PAC
concentrations down to 0.1 mg/L. The method showed a linearity coefcient of 0.98 and relative standard
deviations of 10%, using small water sample volumes between 0.3 and 0.6 L.
2016 Elsevier Ltd. All rights reserved.

1. Introduction micropollutants into surface waters are efuents from wastewater

An increasing range of chemicals and degradation products is
being detected in our water bodies (Bolong et al., 2009). These
substances, referred to as micropollutants, are present in very low
concentrations, but may still pose a risk to the aquatic environment
(Magdeburg et al., 2014; Leusch et al., 2014). As surface and ground
water bodies often serve as sources for drinking water, negative
effects on public health cannot be ruled out (Bolong et al., 2009;
Pasquini et al., 2014). One of the most important entry paths of
* Corresponding author.
E-mail address: thomas.wintgens@fhnw.ch (T. Wintgens).
treatment plants (WWTP). Several anthropogenic pollutants can
pass existing conventional WWTPs and may enter natural water
bodies (Zietzschmann et al., 2014; Bolong et al., 2009). For this
reason, additional treatment steps in WWTPs will have to be
developed and implemented in the near future (Zietzschmann
et al., 2014; Snyder et al., 2003; Jekel et al., 2013). For instance, in
2014 Switzerland has revised its Water Protection Ordinance that
now demands an upgrade of around 100 WWTPs to reduce their
emission of micropollutants by 80% until 2040 (Eidgeno ssisches
Departement fr Umwelt, Verkehr, Energie und Kommunikation,
2015). According to several studies, ozonation and activated car-
bon adsorption are the most feasible and effective process options
for the removal of micropollutants from aqueous solutions (Bolong

http://dx.doi.org/10.1016/j.watres.2016.06.008
0043-1354/ 2016 Elsevier Ltd. All rights reserved.
 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto
wenberg et al., 2014; Margot et al., 2013). Activated
et al., 2009; Lo
carbon is able to adsorb and thus remove micropollutants from
water (Zietzschmann et al., 2014) and can be applied either in its
granular form or as a ne powder with a particle size of 10e150 mm.
Powdered activated carbon (PAC) can be dosed with a high exi-
bility in relation to the water volume and its degree of contami-
nation and is commonly applied in doses of 10e20 mg/L (Bo hler
et al., 2012; Lo wenberg et al., 2016). Once the dosed PAC is
loaded, it has to be separated from the WWTP efuent in order to
prevent desorption of pollutants into natural water bodies
wenberg et al., 2014; Metzger et al., 2011; To et al., 2008). With
(Lo
increasing adsorption efciency and decreasing lter retention, the
amount of micropollutants adsorbed to leaking PAC can make up a
substantial part of the pollutants released from the WWTP. It was
estimated in the here presented study that at an adsorption rate of
80% and a PAC retention of 90%, the PAC release can contribute to
almost one third of the micropollutants present in the efuent.
There are currently several approaches to separate powdered
activated carbon from wastewater and retain it in the treatment
plant. Deep bed ltration is a widely used method, but also ultra-
ltration has been proposed for PAC retention (Lo wenberg et al.,
2014; Snyder et al., 2007; Altmann et al., 2015a, Altmann, et al.,
2015b). In any case, the PAC retention of a separation step has to
be examined in order to assess the PAC leakage into the efuent.
Several authorities and institutions are currently proposing limit
values for the emission of PAC into natural water bodies (Vu et al.,
2012; Abegglen and Siegrist, 2012). For this purpose, PAC concen-
trations below several mg/L have to be detected and quantied in
WWTP efuents.
Due to its low concentration, the detection of PAC in WWTP
efuent was found to be challenging and an easily applicable,
quantitative method is still needed (Abegglen and Siegrist, 2012). In
several studies, turbidity measurements have been performed in
order to estimate the PAC concentrations in wastewater. The results
were found to be qualitatively comparable, but the measurements
were close to the detection limit and no quantitative evaluation was
possible (Metzger et al., 2011). DIN 38 409 describes the measuring
of total suspended solids (TSS) as an alternative method, but ex-
periments showed that measurements below 5 mg/L are of only
limited validity (Metzger et al., 2011). Metzger et al. assessed the
PAC concentrations by ltrating water samples with membrane
lters and then visually comparing the color of the lter with
previously dened reference lters (Metzger et al., 2011). This
method was able to detect PAC concentrations down to 0.05 mg/L.
However, it is a highly subjective method which provides qualita-
tive rather than quantitative results. All proposed methods have in
common that they do not differentiate between activated carbon
and other solids contained in the wastewater, e.g. biological matrix.
However, as in many cases only the PAC leakage is of particular
interest, a material specic method is more favorable.
A well-established technique that allows for the distinction
between different solid materials is Thermogravimetric Analysis
(TGA). TGA is used in industrial and scientic applications to
evaluate the purity and properties of e.g. pharmaceutical and
petrochemical materials. With this method, changes in the mass of
a solid sample are studied while the sample is subjected to a
controlled heating program. The obtained TGA curves provide in-
formation relating to certain physical or chemical phenomena, such
as thermal decomposition and oxidative degradation (Coats and
Redfern, 1963; Brown, 2001). Due to their different physico-
chemical properties, sewage sludge and carbon show signicant
differences in their TGA curves (Folgueras et al., 2003; Otero et al.,
2007, Otero et al., 2008). Samples taken from WWTP efuent will
thus exhibit an intermediate behavior between these two, so that it
will be possible to evaluate their composition. This approach was
511
conrmed in a study where the detection of activated carbon in
activated sludge was performed by TGA (Vu et al., 2012; Vogel,
2013). However, the referenced study does only focus on the
composition of the analyzed solids obtained by vaporization. In
contrast, the here presented study aimed at quantifying the abso-
lute PAC concentration in mg/L in the analyzed water sample by
calculating an overall mass balance. Thus, a sample preparation
method was needed that yields a complete and quantiable re-
covery of suspended solids, and sample centrifugation and ltra-
tion were tested for this purpose in the here presented study.
Furthermore, the method proposed by Vu et al. (2012) relies on a
comparative measurement of a PAC-free reference sample con-
sisting of background matrix only. This drawback can only be
overcome by clearly differentiating between the fractions of PAC
and background matrix in one single TGA curve. As this differen-
tiation was not possible using the mass decrease curve under O2
atmosphere only, a combined heating method under N2 and O2 was
developed, and the mass decrease rate was analyzed rather than
the absolute mass decrease.
This method was then applied to a variety of wastewater sam-
ples from different WWTPs and the results were compared with
those of reference methods in order to assess the TGA methods
sensitivity and reliability.
2. Materials and methods
2.1. Method development
In the rst stage of this study, for method development pur-
poses, wastewater samples were taken from the municipal WWTP
Ergolz 1 in Sissach (Switzerland) (Table 1). PAC-free samples were
taken in the supernatant of the secondary sedimentation. In order
to simulate various PAC concentrations in wastewater, these sam-
ples were spiked with dened, specic amounts of PAC later on in
the laboratory. By doing so, measurement series with PAC con-
centrations CPAC between 0 and 2 mg/L were performed in tripli-
cates in order to standardize and validate the TGA method.
As TGA can only be applied to solid samples, all water samples
had to be concentrated and dried before measurement. Two
different methods were tested for sample preparation, namely
centrifugation and ltration. In the case of centrifugation, a dened
water volume in the range of 5e20 L, depending on its estimated
solids concentration, was lled into 50 mL conical falcon tubes and
then centrifuged at 4,000 rpm for 10 min (Eppendorf, 5810 R). The
solid-rich phase was dried at 80
C in a nitrogen atmosphere for 8 h
(Techne, Dri-Block DB-3D). The weighed and ground solids were
then used for TGA measurement. Alternatively, a dened sample
volume V in the range of 0.3e0.6 L was ltrated with a pre-dried
quartz ber lter (Macherey-Nagel, MN QF-10, temperature sta-
ble up to 950
C) in a vacuum ltration unit with an active ltration
area of 1 cm2. The lter with a layer of separated solids on top was
carefully removed and dried at 105
C for 24 h. From weighing the
lter in the dry state before and after usage, the mass of separated
solids mtot was calculated. The lter together with the solid layer on
top was then used for TGA measurements.
For sample analysis, a thermogravimetric analyzer with a top-
loading balance was used (PerkinElmer, TGA4000). Depending on
the sample preparation, either pure solid samples or pieces of the
solid-containing lters were weighed into ceramic sample pans.
Sample masses were around 5e10 mg and the initial weight of the
sample was registered before starting the thermogravimetric
heating method. As was reported by Vogel (2013), sample analysis
under O2 does not allow for a clear differentiation between PAC and
background signals. Therefore, a preceding heating step under N2
atmosphere was inserted, for selectively removing the background
512 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto

Table 1
Characteristics of the WWTPs for water sampling and PAC quantication.

WWTP Location Population Origin municipal: PAC PAC separation Plant type DOC concentration in the inuent to the PAC
equivalents industrial dosage treatment

Ergolz 1 Sissach 40,000 100:0 12 mg/L Flocculation deep bed Technical 6.7 mg/L
ltration scale
Flotation surface ltration Pilot scale 6.7 mg/L
Bachwis Herisau 35,000 50:50 15 mg/L Sedimentation deep bed Full scale 13.2 mg/L
ltration
Basel Basel 250,000 90:10 15 mg/L Membrane ltration Pilot scale 11.6 mg/L

matrix prior to oxidation (Table 2). The resulting two-stage method is the known ratio of oxidative and pyrolytic decomposition of
was compared to the conventional approach during method background matrix and with
development, and was then applied for sample analysis during the
method application stage.
mO2
FTGA (5)
The sample weight msample and temperature T were digitally mO2 mN2
logged during the whole TGA measurement and were then
analyzed with the TGA software (PerkinElmer, Pyris Software, being a quantitative variable that can be obtained from the TGA
version 2009). For method development purposes, the sample curve analysis, the samples PAC content can be expressed by
degradation was assessed using both the curve msample vs. tem- inserting (2)e(5) into (1), which yields the linear equation:
perature and its time-derivative m_ sample vs. temperature, in order to
XTGA 1 FMatrix $FTGA  FMatrix : (6)
compare both options. During method application, the samples
were analyzed using the derivative curve. On the other hand, with regard to the sample preparation, the
samples theoretical PAC content XSample can be calculated as:
2.2. Calculation of the PAC content of a sample.
CPAC $V
Xsample (7)
In order to quantitatively assess the PAC content of a sample, the mtot
curve m_ sample vs. time during combined N2 and O2 treatment was
with CPAC being the concentration of powdered activated carbon in
used. The mass loss was calculated by integration of the clearly
the water sample, V the ltrated water volume and mtot the mass of
identiable mass decrease peaks in the specic time interval. A
solids separated by ltration. Based on a mass balance, it is
dimensionless variable XTGA was then dened as an indicator for
postulated that the TGA parameter XTGA correlates with the PAC
the samples PAC fraction:
content Xsample in a water sample:
mPAC
XTGA (1) Xsample XTGA $FDegr ; (8)
mPAC mMatrix; O2 mMatrix; N2
where FDegr is a measure for the degree of sample degradation
with mPAC being the mass decrease caused by PAC oxidation, while
during TGA. Hence, by inserting (6) and (7) into (8), a water sam-
mMatrix; O2 and mMatrix; N2 are the background mass decreases under
ples PAC concentration can be calculated from a TGA curve as:
O2 and N2, respectively. While the pyrolytic mass decrease is
entirely caused by background matrix: FDegr $mtot
CPAC $1 FMatrix $FTGA  FMatrix  (9)
mMatrix; N2 mN2 ; (2) V
The values for FDegr and FMatrix were obtained by measuring
the oxidative mass decrease is composed of both activated carbon several samples of known composition, as described in the results
oxidation and the oxidation of background matrix (Fig. 4): and discussion section.

mPAC mMatrix;O2 mO2 : (3)

2.3. Method application
If

mMatrix; O2 In the second stage of this study, wastewater samples from three
FMatrix (4) Swiss WWTPs were analyzed in triplicates using TGA and
mMatrix; N2
comparative methods (Fig. 1). All of these plants comprise a me-
chanical pre-treatment, a biological sludge treatment and a

Table 2
Temperature programs of the TGA heating methods.

O2 method N2/O2 method

Step Temperature Atmosphere Step Temperature Atmosphere

1 Heat from 30
C to 900
C with 5
C/min 20 mL/min oxygen 1 Heat from 30
C to 350
C with 5
C/min 20 mL/min nitrogen
2 Cool down to 30
C 20 mL/min oxygen 2 Constant at 350
C for 300 min 20 mL/min nitrogen
3 Cool down to 30
C 20 mL/min nitrogen
4 Constant at 30
C for 300 min 20 mL/min oxygen
5 Heat from 30
C to 900
C with 5
C/min 20 mL/min oxygen
6 Cool down to 30
C 20 mL/min oxygen
 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto 513

PAC
Primary and Deep bed
FeCl(SO4)
secondary treatment filtration

Tertiary
A Wastewater
effluent
PAC stage

PAC
Polymer Belt
filter
Tertiary
effluent

PAC flotation

PAC
Primary and
FeCl(SO4)
secondary treatment Sedimentation

Tertiary
B Wastewater
effluent
PAC stage Deep bed
filtration
PAC recirculation

Primary and
PAC Membrane
secondary treatment Ozonation
filtration
Tertiary
C Wastewater
effluent

PAC stage
O3

Sampling point

Fig. 1. Flow schemes of the PAC processes on WWTPs Ergolz 1 (A), Bachwis (B) and Basel (C) with sampling points (grey).

Temperature [C]
0 100 200 300 400 500 600 700 800 900
6 0.0
5
-0.1
sample [mg/min]

4
msample [mg]

-0.2
3
pure PAC
-0.3
2
background + 2 mg/L PAC
1 -0.4
background without PAC
0 -0.5
0 20 40 60 80 100 120 140 160 180
Time [min]

Fig. 2. Typical TGA curves of samples with varying composition (centrifuged, O2 method).

Temperature [C] Temperature [C]

0 200 400 600 800 0 200 400 600 800
7 0.05
6
sample [mg/min]

0
msample [mg]

5
4 -0.05

3 -0.1
2
-0.15
1 A B
0 -0.2
0 50 100 150 0 50 100 150
Time [min] Time [min]

Fig. 3. TGA curves of triplicates prepared by centrifugation (A) and ltration (B) (2 mg/L PAC, O2 method).
514
T. Krahnsto
Pure PAC:
Sample weight
Weight derivative
7
6.5
msample [mg] 6
5.5
5
4.5
A B
4 -0.1
0 50
Time [min]
Fig. 4. TGA curves of samples analyzed with the combined method under N2 (A) and O2 (B) (ltrated).
secondary clarication followed by a PAC adsorption step. The PAC
processes of these plants mainly differ in their volume ow ca-
pacity, the origin of the treated wastewater and the method used
for PAC separation (Table 1). The PAC dosage was at 12e15 mg/L in
all cases.
Several alternative techniques were applied in order to assess
the suspended solids content of wastewater samples and to vali-
date and compare the TGA method developed in this study.
Turbidity measurements were performed in glass vials with a
portable turbidimeter (WTW, Turb 355 IR/T) and the concentration
of particles (>1 mm) was measured with an online particle counter
(PAMAS, OLS50P). The amount of total suspended solids was
determined according to DIN 38 409-2 using cellulose nitrate lters
(Macherey-Nagel, PORA NC) and a precision scale (Mettler Toledo,
XS 205). Additionally, the lter color after ltration and drying was
visually compared to reference lters, as proposed by Metzger et al.
(2011).
3. Results and discussion
3.1. Qualitative evaluation of TGA curves
The TGA with a heating step in oxygen atmosphere showed a
characteristic curve that is in accordance with the results of pre-
vious studies (Vu et al., 2012; Otero et al., 2008). The behavior of a
sample exposed to a TGA heating method is characterized by de-
creases in sample mass msample, which each correspond to a peak of
the mass loss rate m _ sample (Fig. 2). In many cases, the sample mass
decreases slightly during the rst minutes of measurement at
temperatures lower than 150
C. This effect is related to residual
moisture in the lter and can be clearly differentiated from any
other mass decreases at higher temperatures. A signicant decrease
is observed in the temperature range 200
Ce400
C. This can be
explained by the volatilization and subsequent oxidation of
wastewater background matrix. Wastewater background matrix
consists of carbohydrates, proteins and fatty acids which are char-
acterized by a high fraction of volatiles. Thermal treatment results
in a decomposition of macromolecules into smaller volatile mole-
cules (Basu, 2010). Evaporation of these molecules is detected as a
mass loss in the TGA curve. Another sharp mass decrease at 400
C
e 700
C is caused by the oxidation of powdered activated carbon.
Activated carbon consists of a network of carbon rings and contains
negligible amounts of volatiles, thus oxidation reactions are gov-
erned by solid-gas reactions (Marsh and Rodrguez-Reinoso, 2006).
At temperatures below 500
C, the reaction rate is kinetically
controlled by a high activation energy (Luo and Jean-Charles, 2006).
It is assumed that, at 600e700
C, reaction kinetics increase rapidly
and are then controlled by in-pore oxygen and carbon monoxide
diffusion which, however, increases at progressing burn-off. The
ver et al. / Water Research 101 (2016) 510e518
Water +2 mg PAC /L: Background matrix:
Sample weight Sample weight
Weight derivative Weight derivative
0
sample [mg/min]
-0.05
100 0 50 100 150
Time [min]
specic temperature of this mass decrease is found to vary
depending on the sample composition, but could be clearly iden-
tied in all cases. Finally, at temperatures above 700
C, the sample
mass reaches a nearly constant value which corresponds to the ash
content of the sample. In the case of a constant temperature
gradient, the temperature axis can be directly correlated to a time
axis. In order to take into account mass transport and reaction ki-
netics and allow for the evaluation of the mass loss rate, the time
axis is of particular interest for analyzing a TGA curve.
3.2. Methods for sample preparation
In the studies performed by Vu et al. (2012) and Vogel (2013),
the solids were concentrated from sewage sludge samples by
evaporation. However, those samples contained around 10 g/L solid
matter, while in the here presented study, the solids content of the
water samples was always below 0.1 g/L. In preliminary experi-
ments, it was found that vaporization does not allow for the com-
plete recovery and mass balance of solid matter in this
concentration range. Thus, two alternative methods were tested for
sample preparation, namely centrifugation and ltration, and were
then compared with regard to their practical feasibility and
reproducibility. The loss of solids during the decanting, centrifu-
gation and drying procedure was found to be considerable: On
average, the centrifugation of 2.2 L water yielded the same solid
sample mass as the ltration of 1.0 L. This incomplete solid sepa-
ration increases the workload and calls for an additional, reliable
determination of the samples solids content. In contrast, in the
case of ltration, a complete removal of solids was achieved and the
separated solid mass correlated well with the TSS, making an
additional determination redundant. Furthermore, it was observed
that the solid separation by centrifugation is not always consistent
for biological matrix and activated carbon. By qualitatively
comparing the TGA curves obtained with these methods, it was
found that a better reproducibility could be achieved if the water
samples were ltrated rather than centrifuged (Fig. 3). If the water
sample is ltrated, the solid sample partly consists of lter material,
which reduces the degree of sample decomposition and causes a
slight mass decrease above 700
C. However, at this temperature,
the mass decrease attributed to activated carbon and biological
background matrix is already completed. This means that the
oxidation of lter material can be clearly differentiated from the
sample oxidation and that the presence of the lter during analysis
does not affect the analyzability of the TGA curve.
3.3. TGA heating methods
For the analysis of solid samples, two different TGA methods
were applied. The O2 method, in which the sample is exposed to
 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto 515

only one heating step in oxygen atmosphere, was found to be 0.6

relatively robust. In the corresponding TGA curves, all mass de-
y = 0.9219x - 0.3253
creases related to water, background matrix and activated carbon 0.5 R = 0.9591
could be identied. Nevertheless, due to overlapping effects, it was
impossible to specically and quantitatively determine the PAC
0.4

Xsample [mg/L ]
fraction. For this reason, the N2/O2 method was developed, which
leads to partial pyrolysis in a rst step under N2. Due to its chemical
composition, PAC is not expected to decompose at the applied 0.3
conditions (Marsh and Rodrguez-Reinoso, 2006), which was
veried by the present experiments (Fig. 4 A). The mass decrease 0.2
caused by pyrolytic reactions can therefore be selectively attributed
to the decomposition of water background matrix. In contrast, the 0.1
simultaneous oxidation of biomass and PAC results in an over-
lapping mass decrease under O2 atmosphere (Fig. 4 B). Still, the 0.0
mass decrease in this second heating step can be used as a measure 0.0 0.2 0.4 0.6 0.8 1.0
for the amount of oxidable solids in the TGA, including both
FTGA [ - ]
background matrix and PAC. This allows to differentiate between
the solids extracted from the original water sample and the lter Fig. 5. FTGA and Xsample obtained from samples of known composition, tted to a linear
material, which does not degrade under these conditions. If both correlation.
heating steps are taken into account, the combined N2/O2 heating
method allows for the identication of lter material, background
matrix and PAC, and for the quantication of the samples 2.0
composition by simply determining the pyrolytic and oxidative
mass decreases during TGA.
1.5

CPAC calculated [ mg/L ]

3.4. Quantitative assessment of mass fractions

As shown in the materials and methods section, based on a TGA

1.0
curve, the PAC content of a water sample can be calculated as:

FDegr $mtot
CPAC $1 FMatrix $FTGA  FMatrix  (10)
V 0.5

with FTGA being obtained from the TGA curve analysis, mtot and V
being parameters of the sample preparation step and FDegr and
FMatrix being dimensionless characteristics of PAC and background
0.0
matrix, respectively. The values obtained from several samples of 0 0.5 1 1.5 2
known composition were found to correlate with a linear trendline CPAC dosed[ mg/L ]
(n 19). If this trendline is assumed to satisfy equation (10) with
Fig. 6. PAC detection by TGA (CPAC measured) in samples that contained a known
FDegr $1 FMatrix being the slope and FDegr $FMatrix being the Y-
amount of PAC (CPAC dosed).
intercept, it is possible to calculate the values (Fig. 5):

FMatrix 0:55
and
FDegr 0:60:
In order to validate the developed method, CPAC was calculated
for samples with known PAC concentrations between 0 and 2 mg/L
and compared to the actual PAC dosage (Fig. 6). The method
repeatability was assessed on the basis of the standard deviations
of triplicate measurements at each concentration. The average
standard deviation was 0.04 mg/L, which corresponds to 10% of
the actual concentration, and it was observed that the relative
deviation decreased with increasing PAC concentration. Taking the
background noise into account, the methods detection limit was
estimated to be around 0.1 mg/L. The average recovery of PAC was
83.8%, which indicates a satisfactory method accuracy. However,
the accuracy was notably inuenced by the sample preparation
step, where the complete and reproducible separation of solids is
the determining factor. It is expected that the increase of the
sample volume and the optimal selection and conditioning of the
lter could increase the PAC recovery in the sample preparation
step in particular. The calculation of FTGA from TGA peak areas was
found to be relatively robust, even in the case of slight variations in
TGA curves due to sample inhomogeneities. The calculations based
on the tted values for FMatrix and FDegr correlated well with the
known sample composition (R2 0.959) and showed a Pearson
correlation coefcient of 0.98, indicating a good method linearity.
However, it should be noted that the tted parameters could vary
for efuents from different WWTPs, e.g. depending on the nitrogen
content of the sewage sludge (Go mez-Rico et al., 2005). It is thus
recommended for future measurements on different WWTPs to
rst determine FMatrix and FDegr before applying the calculation
method to TGA curves.
3.5. Method application for PAC retention assessment
The TGA method was applied to assess and compare the PAC
retention of four different separation technologies (Fig. 1). The re-
sults give an outlook on the targeted application of the developed
method and on the range of PAC concentrations in WWTP efuents.
The PAC removal of the nal separation stage was above 99% in all
cases, resulting in efuent concentrations close to the detection
limit (Fig. 7).
On WWTP Ergolz 1, two process options for PAC removal were
compared. It was found that by applying otation, the PAC
516 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto

1.2

1.0
A B C 17.5

0.8
PAC [mg/L]
0.6

0.4

0.2

0.0
Secondary Filtration Flotation Flotation+ Secondary Sediment. Filtration Ozonation PAC Filtration
effluent effluent effluent filtration effluent effluent effluent effluent reactor effluent
effluent

Fig. 7. Characterisation of wastewater samples taken from the WWTPs Ergolz 1 (A), Bachwis (B) and Basel (C) by N2/O2 TGA (n 3).

concentration can only be reduced to 0.4 mg/L. Thus, in this case, an
additional nal ltration step is necessary to achieve the same
efuent quality as by deep bed ltration. The PAC removal was
above 99% for both process options (Fig. 7 A).
On WWTP Bachwis, the PAC separated in the sedimentation step
is fed back into the biological treatment and discharged together
with the waste activated sludge (Fig. 1). However, it was found that
some PAC residues pass through the secondary clarier, resulting in
a PAC concentration of 0.5 mg/L in the secondary efuent. As on
WWTP Ergolz 1, the efuent still contains PAC in concentrations
close to the detection limit, corresponding to a PAC removal of
above 99% (Fig. 7 B).
The secondary efuent on WWTP Basel is rst treated in a pilot
scale ozonation unit before entering the PAC stage. In the adsorp-
tion reactor, a PAC concentration of 17.5 mg/L was measured, which
is nearly consistent with the targeted PAC dosage of 15 mg/L. The
ultraltration unit ensured a full PAC removal to values below the
limit of detection (Fig. 7 C).
3.6. Method evaluation and comparison
In order to compare the developed TGA method to reference
methods, the wastewater samples from several WWTPs with PAC
treatment were analyzed not only by TGA, but also by particle
counting, turbidity and TSS measurements, as well as visually after
Metzger et al. (2011) (Fig. 8).
The visual analysis of laboratory lters showed no blackness at
all, which corresponds to the absence of PAC according to Metzger
et al. (2011). This was the case for all sampled PAC stage efuents
(Fig. 8), indicating that the visual method is not sensitive enough to
detect PAC traces in wastewater. The results are considered semi-
quantitative and of limited accuracy, so that this method is not
recommended for PAC detection in WWTP efuents.
The TSS removal of the studied processes was between 76% and
91% and always lower than the corresponding PAC removals (Figs. 7
and 8). On WWTP Ergolz I, the TSS and turbidity measurements
qualitatively conrmed that the otation and subsequent surface
ltration yields a similar efuent quality as the deep bed ltration
with direct PAC dosage. However, only 2% of the solids leaking
through the lter were attributed to PAC (Fig. 8A). In the deep bed
ltration of WWTP Bachwis, comparable reductions of TSS (72%)
and of turbidity (67%) were observed with regard to the sedimen-
tation efuent, while the TGA method revealed a PAC retention by
almost 95% (Figs. 7 and 8B). As the ltration step on WWTP Basel

Fig. 8. Characterisation of wastewater samples taken from the WWTPs Ergolz 1 (A), Bachwis (B) and Basel (C) by different methods (n 3).
 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto
was realized by ultraltration the TSS concentration of 3.8 mg/L in
the ltration efuent was surprisingly high (Fig. 8C). However, no
PAC was detected on the permeate side of the membrane unit. In all
separation processes, the percentage removal value of TSS was
found to be lower than that of PAC. This shows clearly that the
determination of standard parameters such as TSS or turbidity is
not sufcient to assess PAC leakage into WWTP efuent. The main
disadvantage of the reference methods is that they represent the
total concentration of solids in the water without differentiating
between PAC and background matrix. However, combining TGA
and TSS results reveals that the leaking solids on WWTP Ergolz 1
and on WWTP Bachwis are partially composed of PAC, while the
efuent of the pilot plant on WWTP Basel contains no PAC at all.
This example shows very well that conventional parameters are no
suitable indicators for PAC in wastewater, and that the PAC speci-
city of the developed TGA method is an important contribution to
assessing the PAC retention of ltration processes.
TGA curves obtained from samples of different WWTPs showed
the same characteristic behavior and intervals of mass losses,
indicating that the TGA measurement was not affected by the origin
or the characteristics of the sampled wastewater or the applied PAC
product. However, variations of FDegr and FMatrix may explain the
slightly higher standard deviations in Bachwis and Basel compared
to WWTP Ergolz I, the water of which was used for method
development.
From a practical point of view, the developed method was found
to be well workable under laboratory conditions. It takes around
18 h to carry out one measurement, which is mainly due to lter
drying (minimum 12 h, e.g. over night) and TGA analysis (5.5 h).
However, laboratory staff is only required for sample ltration, lter
weighing and TGA curve analysis, which add up to an effective
workload of 15e20 min per sample. In the development stage, the
proposed method required visual peak interpretation which,
however, most likely can be automized once more data sets are
available.
4. Conclusion
In the present study, a method was developed to quantitatively
measure the PAC content in wastewater by using thermogravi-
metric analysis (TGA). Prior to TGA, the samples were ltered using
a quartz lter with a temperature stability of up to 950
C, so that
the separated solids could be analyzed by TGA without the mea-
surement being affected by the presence of the lter. Based on the
conducted laboratory experiments, a mathematical correlation was
found for the calculation of a samples PAC content from the cor-
responding TGA curve. In comparison to alternative methods, such
as measuring turbidity, suspended solids or particle concentration,
the developed method combining vacuum ltration and TGA is able
to selectively differentiate between PAC and background particle
matrix, which allows for assessing the retention of processes with
regard to PAC in particular. Different separation processes were
shown to achieve PAC removals between 95% and 100%. The TGA
method yields quantitative results with a detection limit around
0.1 mg PAC/L. One measurement requires small sample volumes
around 0.5 L and an effective workload of less than 20 min. Given
the incomplex sample preparation and relatively low TGA system
prices, it seems manageable to include the analysis of PAC con-
centration by the proposed method in the regular monitoring
routine of WWTPs equipped with PAC stages for advanced
treatment.
Acknowledgement
The authors express their gratitude towards the Swiss Federal
517
Ofce for the Environment for funding this study under contract
number UTF 450.06.13/IDM 2004.2423.391. Special thanks go to
Felix Thiele, Antje Langbein and Jan Stemann for their helpful
collaboration and valuable advice.
References
Abegglen, C., Siegrist, H., 2012. Mikroverunreinigungen aus kommunalem
Abwasser. Verfahren zur weitergehenden Elimination auf Kla ranlagen. Bunde-
samt fr Umwelt, Bern.
Altmann, J., Ruhl, A.S., Sauter, D., Pohl, J., Jekel, M., 2015a. How to dose powdered
activated carbon in deep bed ltration for efcient micropollutant removal.
Water Res. 78, 9e17.
Altmann, J., Sperlich, A., Jekel, M., 2015b. Integrating organic micropollutant
removal into tertiary ltration: combining PAC adsorption with advanced
phosphorus removal. Water Res. 84, 58e65.
Basu, P., 2010. Biomass Gasication and Pyrolysis - Practical Design and Theoriy.
Elsevier, ISBN 978-0-12-374988-8.
Bohler, M., Zwickenpug, B., Siegrist, H., Hollender, J., Joss, A., Ternes, T., 2012.
Removal of micropollutants in municipal wastewater treatment plants by
powder-activated carbon. Water Sci. Technol. 66.10, 2115e2121.
Bolong, N., Ismail, A.F., Salim, M.R., Matsuura, T., 2009. A review of the effects of
emerging contaminants in wastewater and options for their removal. Desali-
nation 239.1e3, 229e246.
Brown, M., 2001. Introduction to Thermal Analysis. Kluwer Academic Publishers,
ISBN 978-1-4020-0211-3.
Coats, A., Redfern, J., 1963. Thermogravimetric analysis. A review. Analyst 88,
906e924.
Eidgeno ssisches Departement fr Umwelt, Verkehr, Energie und Kommunikation,
Revision der Gewa sserschutzverordnung fr bessere Wasserqualit at, [Online].
Available: https://www.news.admin.ch/message/index.html?langde&msg-
id55786. [accessed 25.09.15.].
Folgueras, B., Daz, R., Xiberta, J., Prieto, I., 2003. Thermograimetric analysis of the
co-combustion of coal and sewage sludge. Fuel 82, 2051e2055.
mez-Rico, M.F., Font, R., Fullana, A., Martn-Gullo n, I., 2005. Thermogravimetric
Go
study of different sewage sludges and their relationship with the nitrogen
content. J. Anal. Appl. Pyrolysis 74, 421e428.
Jean-Charles, X.R., 2006. Research of oxidation properties of graphite used in HTR-
10. Inst. Nucl. New Energy Technol. 20e45.
Jekel, M., Ruhl, A.S., Meinel, F., Zietzschmann, F., Pugmacher Lima, S., Baur, N.,
Wenzel, M., Gnir, R., Sperlich, A., Dnnbier, U., Bo ckelmann, U., Hummelt, D.,
von Baar, P., Wode, F., Petersohn, D., Grummt, T., Eckhardt, A., Schulz, W.,
Heermann, A., Reemtsma, T., Seiwert, B., Schlittenbauer, L., Lesjean, B.,
Miehe, U., Remy, C., Stapf, M., Mutz, D., 2013. Anthropogenic organic micro-
pollutants and pathogens in the urban water cycle: assessment, barriers and
risk communication (ASKURIS). Environ. Sci. Eur. 25, 1e8.
Leusch, F., Khan, S., Laingam, S., Prochazka, E., Froscio, S., Trinh, T., Chapman, H.,
Humpage, A., 2014. Assessment of the application of bioanalytical tools as
surrogate measure of chemical contaminants in recycled water. Water Res. 49,
300e315.
wenberg, J., Zenker, A., Baggenstos, M., Koch, G., Kazner, C., Wintgens, T., 2014.
Lo
Comparison of two PAC/UF processes for the removal of micropollutants from
wastewater treatment plant efuent: process performance and removal ef-
ciency. Water Res. 56, 26e36.
wenberg, J., Zenker, A., Krahnsto
Lo ver, T., Bo hler, M., Baggenstos, M., Koch, G.,
Wintgens, T., 2016. Upgrade of deep bed ltration with activated carbon dosage
for compact micropollutant removal from wastewater in technical scale. Water
Res. 94, 246e256.
Magdeburg, A., Stalter, D., Schlsener, M., Ternes, T., Oehlmann, J., 2014. Evaluating
the efciency of advanced wastewater treatment: target analysis of organic
contaminants and (geno-)toxicity assessment tell a different story. Water Res.
50, 35e47.
Margot, J., Kienle, C., Magnet, A., Weil, M., Rossi, L., de Alencastro, L.F., Abegglen, C.,
Thonney, D., Che vre, N., Scha rer, M., Barry, D., 2013. Treatment of micro-
pollutants in municipal wastewater: ozone or powdered activated carbon? Sci.
Total Environ. 461-462, 480e498.
Marsh, H., Rodrguez-Reinoso, F., 2006. Activated Carbon. Elsevier, ISBN 0-08-
044463-6.
Metzger, S., Ro ssler, A., Kapp, H., 2011. Erweiterung kommunaler Kla ranlagen durch
eine adsorptive Stufe zur Elimination organischer Spurenstoffe. Optimierung
der Pulveraktivkohleabtrennung durch Filtration als Grundlage zur
Anlagendimensionierung.
Otero, M., Go  mez, X., Garca, A., Moran, A., 2007. Effects of sewage sludge blending
on the coal combustion: a thermogravimetric assessment. Chemosphere 69,
1740e1750.
Otero, M., Calvo, L., Gil, M., Garca, A., Mor an, A., 2008. Co-combustion of different
sewage sludge and coal: a non-isothermal thermogravimetric kinetic analysis.
Bioresour. Technol. 99, 6311e6319.
Pasquini, L., Munoz, J.-F., Pons, M.-N., Yvon, J., Dauchy, X., France, X., Le, N.D., France-
Lanord, C., Go rner, T., 2014. Occurrence of eight household micropollutants in
urban wastewater and their fate in a wastewater treatment plant. Statistical
evaluation. Sci. Total Environ. 481, 459e468.
Snyder, S., Westerhoff, P., Yoon, Y., Sedlak, D., 2003. Pharmaceuticals, personal care
518 ver et al. / Water Research 101 (2016) 510e518
T. Krahnsto

products, and endocrine disruptors in water: implications for the water in- von Klaranlagen. NETZSCH-Gera tebau GmbH, pp. 6e8.
dustry. Environ. Eng. Sci. 20, 449e469. Vu, T.T., Vogel, A., Fglein, E., Platz, S., Menzel, U., 2012. Nachweis von Pulver-
Snyder, S., Adham, S., Redding, A., Cannon, F., DeCarolis, J., Oppenheimer, J., Wert, E., aktivkohle in Abwa ssern mithilfe der Thermogravimetrie. Korresp. Abwasser,
Yoon, Y., 2007. Role of membranes and activated carbon in the removal of Abfall 59, 208e218.
endocrine disruptors and pharmaceuticals. Desalination 202, 156e181. Zietzschmann, F., Altmann, J., Ruhl, A.S., Dnnbier, U., Dommisch, I., Sperlich, A.,
~ as, B.J., Snoeyink, V.L., Ng, W.J., 2008. Effect of strongly competing
To, P.C., Marin Meinel, F., Jekel, M., 2014. Estimating organic micro-pollutant removal potential
background compounds on the kinetics of trace organic contaminant desorp- of activated carbons using UV absorption and carbon characteristics. Water Res.
tion from activated carbon. Environ. Sci. Technol. 42.7, 2606e2611. 56, 48e55.
Vogel, A., 2013. Thermogravimetrischer Nachweis von Pulveraktivkohle im Ablauf