Rangsivek Ropru

Removal of dissolved metals from storm water
runoff by zero-valent iron

vorgelegt von

Ropru Rangsivek, M.Sc.
aus Pattani, Thailand

von der Fakultt III
Prozewissenschaften
der Technischen Universitt Berlin
zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften
- Dr.-Ing.-

genehmigte Dissertation

Promotionsausschu:

Vorsitzender: Prof. Dr.-Ing. Sven-Uwe Geien
Gutachter: Prof. Dr.-Ing. Martin Jekel
Gutachter: Prof. Dr.-Ing. Matthias Barjenbruch
Gutachter: Prof. Dr.-Ing. Heiko Diestel

Tag der wissenschaftlichen Aussprache: 23 April 2010

Berlin 2010
D 83

Acknowledgements

This PhD thesis was written based on my research work carried out at the department of water
quality control at the Technical University of Berlin (TU-Berlin) from June 2003 to
December 2006 under the supervision of Prof. Dr.-Ing. Martin Jekel. I would like to express
sincere thanks to him for giving me this opportunity and also for providing me all the support
over these years. I would also like to express gratitude to Prof. Dr.-Ing. Heiko Diestel and
Prof. Dr.-Ing. Horst Borgmann without their help I would never had such a chance. Prof. Dr.
Gary Amy has given me valuable guidance, expertise in research and continuous support
during a sabbatical year in Berlin. Prof. Dr.-Ing. Matthias Barjenbruch and Prof. Dr.-Ing.
Sven-Uwe Geien for his suggestions, review and taking up the examination job. They all
have paved the way for my life today.

I would like to acknowledge Dr.-Ing. Mathias Ernst, Dipl.-Ing. Marco Schmidt, Dipl.-Geogr.
Katharina Teschner, Dr.-Ing. Kyung-Ho Kwon, M.Sc. Agueda Sollis Tellez, Dipl.-Ing.
Carsten Bahr, M.Sc. Xing Zheng and other colleagues at the department for having great
discussions both on the research and other out-of-the-room topics throughout my stay in
Berlin. A thousand thanks must be devoted to Dr.-Ing. Thomas Ludwig who has helped me in
several aspects of the research works. Dr. Anke Putschew, Gisela Sosna, Ulrike Frster,
Angelika Kersten as well as others have enlightened me in laboratory equipments and
methods. Special thanks to Werner Dumler and Thomas Thele for technical-assistance. Their
works are really amazing. Wolfgang Wichmann, Michael Markmann and Brbel Bleicke are
also acknowledged. Thank you to Karin von Nordheim for helping in documentation over the
years.

Parts of my work have been reviewed by Priv.-Doz. Dr.-Ing. Thorsten Reemtsma, Dr.-Ing.
Arne Genz, Dipl.-Chem. Dirk Bloem, M.Sc. Amnuay Wattanakornsiri. Several journal
reviewers have greatly helped in improving the works. This includes Dr. Chicgoua Noubactep
and his profound knowledge of Fe
0
and the constructive email discussions. Special thanks are
extended to Rebeca, Shawn, Babara, Phra William (William Leslie), Mai-Lan Ha and Katja
for editorial work.

The SEM and EDX analyses were carried out at Zentraleinrichtung Elektronenmikroskopie
(ZELMI), TU Berlin. The author wishes to thank Dr. Burkhard Peplinski at Federal Institute
for Materials Research and Testing (BAM) and Dr. Andreas Hsch, Institute of Applied
Geosciences, TU-Berlin, for kindly helping in XRD analysis. Further AES and XRD were
performed by Dipl.-Ing. Bettina Camin and Dipl.-Ing. Eric Wild, Metallische Werkstoff at the
TU Berlin (Prof. Dr. Walter Reimers). Thank to Frau Maike Mai from the department of
waste management, TU-Berlin for providing the XAD Resin apparatus.

There are many more people who should be acknowledged for this project both directly and
indirectly, e.g., a small party at the small corner of the lab in the late evening playing loud
German country songs has always been in my memory. Heartfelt thanks to my Thai friends
who are always together: P Gai U, P Orn, P Ja, PBaan, P Tum, A U, P Tay, Luang A

Payom, Luang A Boonma, I Den, E Tud Net; N. Cap and several others. Phra A. Tony, Phra
Maha Sorasak, Laung P Tep, Mae Rabaeb, PAew, P Buay for being dhamma friends.

I would like to thank to my parents, Papa and Mama, for persistent and endless support
through these long years. Thanks also to Irene and Ingo Lindner. Without them and their
daughter, Katja, this work would not have been possible. Living for a period in Germany was
so great with Katjas (also my German) family. Thank you also to Oleang for being a good
friend during a cold winter in Chiangmai and now. Last but not least, Donmei, my son even
though he is the smallest person here, he has been the greatest driving force that pushes me
through this work at this late period while he is in Denmark with his mother.

Abstract

This study investigates the novel treatment method for heavy metals from contaminated
runoff by employing Fe
0
. Batch kinetic, equilibrium and flow-through configurations were
thoroughly studied. Based on the investigation, it was found that Fe
0
achieves a comparable
capacity to a commercial adsorbent like granular ferric hydroxide (GFH). The removal of
Cu
2+
and Zn
2+
by Fe
0
takes place through an array of rate-limiting steps including iron
oxidation, iron (oxy)hydroxide precipitation and metal adsorption/co-precipitation. Cu
2+
is
directly or indirectly reduced by Fe
0
and/or dissolved, structured Fe
2+
, and forms Cu
0
and
Cu
2
O as reduction products. The removal of Zn
2+
is mainly due to the adsorption and co-
precipitation processes. Results of batch kinetic tests using showed that when the solution is
low in DO, T, pH, IS or reactions take place under high metal and DOC concentrations, the
removal rate of Zn
2+
dramatically decreases.

The breakthrough column tests showed that the exhaustion of zinc occurred more rapidly than
copper; yielding a metal loading capacity of an average 76 mg g
-1
Fe
0
for copper (with min: 55
and max: 96 mg g
-1
) and 55 mg g
-1
Fe
0
for zinc (min: 22, max: 69 mg g
-1
). Increased pH and
temperature generally favor faster retention rates of copper and zinc, whereas increases in DO
and conductivity of the solution hinder the removal of copper in the column system. Among
parameters studied, NOM shows to be the most influent compound that results in a
dramatically decreased metal uptake rate. Inhibition of NOM on the removal of metals by Fe
0

was due to competitive adsorption, metal-ligand complexes and the inhibition that was caused
by surface coverage of iron corrosion products. The NOM fraction that mainly interacts with
metals is a larger size hydrophobic fraction with high aromaticity.

With aims to upscale the treatment process and to be able to simulate the breakthrough curves
at various scenarios, a pore-surface diffusion model (PSDM) was attempted. Generally, the
PSDM can adequately capture most of the determined curves but at increased NOM
concentration, an over prediction of the curve was observed.

Following the test and modeling at various scenarios, optimization of the processes was
carried out by incorporating Fe
0
with calcite, dolomite and magnesite media. The
recommended system for on-site treatment of metal-contaminated runoffs consisted of
columns filled with spiral shaped irons followed by an aeration unit and post filtration of
pumice/dolomite. According to the test, about 60-100 and 30-80 % of metal concentrations
could be removed at an EBCT of 5 and 1 min EBCT, respectively. A longer EBCT will yield
a better performance. For practical operations, carbonate materials tend to be passivated by
various minerals and biofilms. An occasional backwash in order to reactivate the surface of
materials inside and to maintain the hydraulic conductivity of the system is required.

Finally, cost analysis showed that Fe
0
is a suitable medium for stormwater runoff treatment.

Zusammenfassung

In dieser Studie wurde eine neuartige Methode zur Aufbereitung von schwermetallbelasteten
Oberflchenabflssen untersucht. Dabei kam nullwertiges Eisen (Fe
0
) als Barrieresystem zum
Einsatz. In Batch- und Sulenversuchen wurden die Mechanismen und der Einfluss der Was-
serqualitt auf die Schwermetallentfernung in dem Fe
0
-Barrierensystem untersucht. Die
Untersuchungen haben gezeigt, dass mit Fe
0
eine vergleichbare Beladungskapazitt erzielt
wird wie mit granuliertem Eisenhydroxid (GEH). Die Entfernung von gelstem Kupfer und
Zink durch Fe
0
ist auf mehrere geschwindigkeitsbestimmende Schritte zurckzufhren, wobei
die Eisenoxidation, die Eisenhydroxidbildung, die Metalladsorption und die Mitfllung von
Bedeutung sind. Das gelste Kupfer wird direkt oder indirekt durch Fe
0
reduziert. Als Re-
aktionsprodukte wurden Cu
0
und Cu
2
O festgestellt. Die Zinkentfernung ist hauptschlich auf
die Adsorption und auf die Mitfllung zurckzufhren. Die kinetischen Batchversuche haben
gezeigt, dass die Zinkentfernung durch geringe Sauerstoffgehalte, niedrige Temperaturen und
niedrige pH-Werte sowie hohe DOC- und Metallgehalte verringert wird.
Die Sulenversuche mit Fe
0
zeigten, dass das gelste Kupfer besser zurckgehalten wird als
das gelste Zink. Dabei wurde eine mittlere Beladung des Fe
0
in Hhe von 76 mg g
-1
(mit
min.: 55 und max.: 96 mg g
-1
) fr Kupfer und von 55 mg g
-1
(mit min.: 22 und max.: 69 mg g
-
1
) fr Zink ermittelt. Es zeigte sich ein besserer Rckhalt bei erhhten pH-Werten sowie bei
einer erhhten Wassertemperatur. Demgegenber verringerte sich der Kupfer- und
Zinkrckhalt bei einer Zunahme der Gelstsauerstoffkonzentration bzw. der Leitfhigkeit.
Von den untersuchten Einflussgren auf den Metallrckhalt wurde der NOM-Gehalt im
Oberflchenabfluss als wichtigster Einflussparameter identifiziert, der zu einer drastischen
Verringerung der Metallrckhaltes. Die festgestellte Inhibierung des Metallrckhaltes in
Gegenwart von NOM kann auf die konkurrierende Adsorption, auf die Bildung von
metallorganischen Komplexen sowie auf die Hemmung der Eisenkorrosion zurckgefhrt
werden. Die im Rahmen der Versuche ergab sich, dass vorzugsweise die hhermolekularen,
hydrophoben NOM-Fraktionen mit der Eisenoberflche reagieren.
Zur Simulation der Durchbruchskurven unter verschiedenen Randbedingungen wurden
Modellrechnungen mit einem Poren-Diffusionsmodell durchgefhrt. Dabei konnten die
experimentell ermittelten Durchbruchskurven durch geeignete Wahl der Modellparameter
hinreichend abgebildet werden. Im Falle erhhter NOM-Gehalte im Oberflchenabfluss war
zur Modellierung der Durchbruchskurven allerdings eine weitere Anpassung der Modell-
parameter erforderlich.
Mit dem Ziel der Verbesserung der Ausfllung und der Metalladsorption an der Eisenoxid-
oberflche wurden Untersuchungen mit karbonatischen Materialien (Calcit, Dolomit,
Magnesit) durchgefhrt. Die Versuche haben gezeigt, dass bei mittleren Verweilzeiten von 1
bzw. 5 Minuten ca. 30 bis 80 % bzw. ca. 60 bis 100 % der Metallkonzentrationen entfernt
werden knnen. Mit einer Erhhung der Verweilzeit kann dabei ein verbesserter Rckhalt
erzielt werden. Das zur on-site Behandlung von schwermetallhaltigen Oberflchenabflssen
empfohlene Reinigungssystem besteht aus einer mit spiralfrmigen Eisenspnen gefllten
Sule gefolgt von einer Belftungseinheit und einer nachgeschalteten Filtereinheit aus Bims-
stein / Dolomitstein. Fr die praktische Anwendung ist zu beachten, dass die karbonatischen
Materialien durch Ausbildung von Biofilmen und Deckschichten passiviert werden. Zum
Erhalt der hydraulischen Leitfhigkeit sowie zur Reaktivierung der Oberflchen ist deshalb
eine gelegentliche Rcksplung des Systems erforderlich.

Die Kostenanalyse ergab, dass Fe
0
ein geeignetes Medium fr Aufbereitung von
schwermetallbelasteten Oberflchenabflssen ist.

Publication lists

Conference paper

Rangsivek, R., Amy, G.L. and Jekel, M.R. (2005), Removal of dissolved metals by zero-
valent iron (ZVI) under storm water conditions, presented for 1
st
International conference on
environmental science and technology(EST), 23-26 January, New Orleans, Louisiana, USA.

Rangsivek, R. and Jekel, M.R. (2005), Removal of dissolved metals by zero-valent iron
(ZVI): Kinetics, equilibria, processes and implications for storm water runoff treatment,
presented at 10
st
International conference on urban drainage, 21-26 August, Copenhagen,
Denmark.

Rangsivek, R. and Jekel, M.R. (2007), A column investigation of Fe
0
for treatment of copper
and zinc under varying quality of runoff, presented at the 6
th
International conference on
sustainable techniques and strategies in urban water management (NOVATECH), 25-28 June,
Lyon, France.

Poster presentation

Rangsivek, R. and Jekel, M.R. (2007), Interaction of natural organic matter (NOM) on Fe
0

treatment processes of metal contaminated-runoff, presented at the 6
th
NOVATECH, 25-28
June, Lyon, France.

Publication in Journals

Rangsivek, R. and Jekel, M.R. (2005), Removal of dissolved metals by zero-valent iron
(ZVI): Kinetics, equilibria, processes and implications for storm water runoff treatment, Wat.
Res., 39, 17, p. 4153-4163 (at the 20
th
hottest article on Science Direct)

Rangsivek, R. and Jekel, M.R. (2008), Natural organic matter in roof runoff and its impact on
the Fe
0
treatment system of dissolved metals, Chemosphere, 71, 18-29.

Rangsivek, R. and Jekel, M.R. (2007), Development of an on-site Fe
0
system for treatment of
copper- and zinc- contaminated roof runoff, international journal of environmental and waste
management (IJEWM): Metal ions removal from liquid effluents, accepted

Rangsivek, R., and Jekel, M.R. (2010), Removal of dissolved metals from roof runoff in Fe
0

columns: water quality impacts, processes, modeling and optimization, submited.

Content xi

Content

Acknowledgement
Abstract
Publication lists
Content

1 Introduction 1
1.1 New concepts of urban water management 1
1.2 Contaminants in urban stormwater runoff 1
1.3 Objectives of the study 3
1.4 Structure and outline of the thesis 4

2 Theory and background of zero-valent iron barrier 7
2.1 Zero-valent iron barrier (ZVIB) for water treatment 7
2.2 Iron corrosion 8
2.2.1 Corrosion processes and mechanisms 8
2.2.2 Corrosion products 9
2.2.3 Corrosion rate 10
2.3 Removal processes in a ZVI system 11
2.3.1 Removal mechanisms 11
2.3.2 Removal of inorganic pollutants 16
2.4 Fe
0
treatment of metal contaminated runoff 18
2.4.1 Processes of metal removal 18
2.4.2 Practical issues of Fe
0
barrier 19
2.5 Natural organic matter 19
2.5.1 Significance of NOM 19
2.5.2 Characterization of NOM 20
2.5.3 Impacts of NOM on iron corrosion processes 23
2.6 Modeling of the F
0
system 25

3 Materials and methods 31
3.1 Batch experiments 31
3.1.1 Materials 31
3.1.2 Kinetic batch tests 33
3.1.3 Equilibrium tests 34
3.2 Column experiments 34
3.2.1 Materials 34
3.2.2 Laboratory experiments 35
3.2.3 Model prediction 36
3.3 NOM experiments 37
3.3.1 Materials and characterizations 37
3.3.2 Kinetic batch and column experiments

38

Content xii

3.4 Optimization and Field experiments 39
3.5 Analysis 41
3.5.1 Iron, copper, zinc 41
3.5.2 DOC, NOM 42
3.5.3 Others 42
3.5.4 Solid precipitates 42

4 Results and discussion 43
4.1. Batch experiments 43
4.1.1 Kinetics and stoichiometry of metal uptake 43
4.1.2 Impact of the quality of stormwater runoff 45
4.1.3 Equilibrium study 48
4.1.4 Characterization of DOC 50
4.1.5 Solid phase characterization 52
4.1.6 Conclusion 55
4.2 Column experiments with ZVI 57
4.2.1 Impacts of water quality 57
4.2.2 Breakthrough behavior and process identification 61
4.2.3 Model prediction 64
4.2.4 Conclusion 68
4.3. NOM experiments 70
4.3.1 Characterization of source water 70
4.3.2 Batch kinetic study with NOM 73
4.3.3 Column study with NOM 75
4.3.4 Impact of NOM on copper and zinc removal in ZVIB 76
4.3.5 NOM analyses by LC-OCD 79
4.3.6 Conclusion 80
4.4. Column experiments with ZVI and carbonatic materials 82
4.4.1 Optimization of the ZVIB using carbonatic materials 82
4.4.2 Processes of metal removal in ZVI/carbonatic barrier 87
4.4.3 Implication of ZVI/carbonatic barrier 90
4.4.4 Conclusion 92
4.5 Design of a treatment system and application of results 94
4.5.1 Application of Fe
0
for highway runoff treatment 94
0
for roof runoff treatment 96
4.5.3 Design of ZVICB system 97

5 Summary, conclusions and recommendations 104
5.1 Summary and conclusions 104
5.2 Recommendations for further work 106

References 108

Content xiii

Appendix A: Relevant information 120
Appendix B: Supplement data for tables and figures 122
Appendix C: Fractionation method 167
Appendix D: The pore and surface diffusion model (PSDM) 171

Content xiv

Content xv

Content of Figures

Fig 1.1 Concentration of copper and zinc in roof and highway runoffs (c.f., Table
A2 and A3)
2
Fig. 2.1 Permeable reactive barrier for treatment of contaminantated groundwater 7
Fig. 2.2 Schematic of possible reactions relevant to inorganic (i.e., copper and
zinc) removal in Fe
0
barrier
11
Fig. 2.3 pH dependent sorption of metal cations on iron hydroxide (adapted from
Denver, 1997). The co-precipitation results of Cr
3+
and Zn
2+
and precipitation of
Cd
2+
were obtained from Crawford et al. (1993) and Smith (1996), respectively.
13
Fig. 2.4 Schematic of copper and zinc removal in Fe
0
system 17
Fig. 2.5 NOM fractionation methods 21
Fig. 3.1 SEM and EDX mapping of Fe
0
used in this study 32
Fig. 3.2 pH
PZC
of iron oxides produced by corrosion of iron 32
Fig. 3.3 Kinetic batch reactor 33
Fig. 3.4 Laboratory column experiment (c.f., text) 35
Fig. 3.5 a) The overall view and b) the detailed onsite experiments of Fe
0
system
for treatment of copper and zinc roof runoff. The systems include 3 treatment
steps; 1) Fe
0
treatment columns 2) aeration step using tipping counter and aeration
basin and 3) post-filtration of pumice and dolomite (well mixed using 800 g each)
41
Fig. 4.1 Plots of a) normalized Cu
2+
and Zn
2+
concentration as a function of time,
b) stoichiometry determination of Cu
2+
cementation process [Cu
i
2+
1 mg L
-1
, Zn
i
2+

5 mg L
-1
, 0.5 g L
-1
ZVI, pH
i
5.0 0.1, pH
f
5.83, DO
i
8.0 mg L
-1
, T 20 1.0 C,
150 rpm in SSWR]. Error bars represent 95% confidence intervals from duplicate
experiments
43
Fig. 4.2 Results from batch experiments with pulsed dosing Plots of a) Total Cu
vs. time b) ln(C
t
/C
0
) vs time c) Total Fe vs. time and d) Total Fe dissoluted vs.
Total Cu removed at Cu
0
5 mg L
-1
(re-spiked with 5 mg L
-1
), ZVI 0.5 g L
-1
(0.192
m
2
L
-1
), pH 7, DO 0 mg L
-1
, T 20C and 150 rpm. The below figures were the
results from identical experiment with Pb
2+
dosing
44
Fig. 4.3 Comparison of experimental results carried out in () limiting and () un-
limiting DO concentration at varying pH 4.0-7.0 (controlled). [Cu
i
2+
1 mg L
-1
,
Zn
i
2+
5 mg L
-1
, 0.5 g L
-1
ZVI, T 20 1.0 C, 150 rpm in DI]
45
Fig. 4.4 Examples of normalized concentration profiles of Cu
2+
and Zn
2+
vs time.
Experiments were conducted with various runoff solutions (c.f. Table 1).
Representative profiles are chosen based on the description in the text
47
Fig. 4.5 Comparison of Cu
2+
, Zn
2+
and DOC equilibrium uptake by ZVI and GFH.
ZVI system contained various stormwater runoff solutions; ( ) SSWR pH
f
5.14-
5.56; ( ) Lankwitz pH
f
5.00-5.67; ( ) TU-SW pH
f
5.00-5.76; ( ) SRFA pH
f

5.00-6.57; ( ) UFA pH
f
5.33-6.51; ( ) Halensee pH
f
5.59-6.70. GFH isotherms
were simulated based on the results obtained from (1) Steiner (2003) and (2)
Ludwig (2004) in DI solution
49
Fig. 4.6. LC-OCD and UVD signals of a) UFA and b) HS stormwater runoffs
before and after equilibrium treatment with doses of 0.05 (0.02 for Halensee) and
0.5 g ZVI L
-1

51

Content xvi

Fig. 4.7 a) SEM and b) EDX mapping of ZVI scale treated with solution [Cu
i
2+
10
mg L
-1
, Zn
i
2+
10 mg L
-1
, pH 7.0, DO
i
8-9 mg L
-1
, 20C, 150 rpm, SSWR]
52
Fig. 4.8 X-Ray Diffractograms of ZVI scales treated under a) pH 4.0 DO < 0.5 mg
L
-1
and b) pH 7.0 DO
i
8-9 mg L
-1
[Cu
0
2+
and Zn
0
2+
10 mg L
-1
, 20C, 150 rpm,
SSWR Solution]
52
Fig. 4.9 The Auger Spectra of precipitates after cementation process [Cu
i
2+
5 mg
L
-1
, ZVI dimension 2.1x1.05x0.5cm
3
, pH 6.0 (controlled), DO < 0.5 mg L
-1
, T 20
1.0 C, 150 rpm in SSWR]
53
Fig. 4.10 Mapping of SEM image to EDX spectra of iron after experiment [Cu
i
2+

10 mg L
-1
, Zn
i
2+
10 mg L
-1
ZVI 0.192 m
2
L-1(0.5 g Fe
0
L
-1
), pH 7.0 (controlled), T
20 1.0 C, 150 rpm in SSWR]. (c.f., supplement data in Appendix B)
54
Fig. 4.11 Breakthrough behavior of copper and zinc and the corresponding iron
released from column at varying pHs
57
Fig. 4.12 Breakthrough curves of copper and zinc and the corresponding released
concentration of iron from the columns experiment at oxygenated (6-8 mg L
-1
) and
deoxygenated (< 1.3 mg L
-1
) conditions [Cu
2+
5.0 mg L
-1
, Zn
2+
5.0-6.5 mg L
-1
, pH
i

5-6, 5 min EBCT,10% Fe
0
(v/v) mixed with pumice]
58
-1
) and
-1
) conditions [Cu
2+
5.0 mg L
-1
, Zn
2+
5.0-6.5 mg L
-1
, pH
i

5-6, 5 min EBCT, 10% Fe
0
59
Fig. 4.14 Breakthrough behavior of copper and zinc and the corresponding iron
released from column at varying temperatures. (line divided between runoff and
DI employed in the experiment, c.f., text)
60
Fig. 4.15 (a) SEM and (b) EDX of of media obtained from typical column test as
compared with the original pumice, (c) The present of nodular metallic copper on
iron surface in DO-limiting solution and (d) the corresponding EDX diagram, c.f.
text
63
Fig. 4.16 XRD diagram of media obtained from typical column test 63
Fig. 4.17 LC-OCD diagrams of the inlet and outlet water samples obtained from
different columns
64
Fig. 4.18 Calibration of PSDM with breakthroughs of copper and zinc employing
equilibrium and kinetic parameters illustrated in Table 4.2
66
Fig. 4.19 Calibration of the PSDM with breakthroughs of copper and zinc
obtained from column experiments under different NOM concentrations. The
obtained kinetic parameters are illustrated in Table 4.2
66
Fig. 4.20 Sensitivity analysis of the breakthrough curves of copper obtained from
a large column test calibrating with (a) K
F
, (b) k
f
and (c) D
S
(modeling parameters
are according to Table 1 for lower bound with varied 25%)
67
Fig. 4.21 Plots of model prediction of the breakthrough results of a large column
test and the model prediction for assessment of the up-scaling and optimizing of
the processes
68
Fig. 4.22 LC-OCD of bulk water NOMs and their corresponding XAD-8 isolated
and XAD-8 effluent fractions of a) TU-SW b) TU-FF c) SR-NOM and d) Cu-R
runoffs, c.f., Table 4.3.
71

Content xvii

Fig. 4.23 Copper titration of runoff NOM samples from different sources 73
Fig. 4.24 a) and c) the kinetic removal of Zn
2+
in TU-SW, TUFF, SR-NOM and
Cu-R runoff solutions carried out in the batch test at initial pH
i
2.50 (Control: Fe
0

0 mg L
-1
) (in order pH
f
4.540.13, 4.290.06, 4.260.03 and 4.610.04) and 5.50
(pH
f
6.92, 9.57, 9.29 and 7.95), respectively, and b) and d) a comparison of its
corresponding DOC removal in correlation with removal of Zn
2+
[Cu
i
2+
5.0 mg L
-
1
, Zn
2+
5.0 mg L
-1
, 0.5 g Fe
0
L
-1
, well mixed and at room temperature]
74
Fig. 4.25 A comparison of the breakthrough curves of copper and zinc obtained
from the columns tested at varying NOM concentrations of TU Runoff [System:
Cu
i
2+
5.0 mg L
-1
, Zn
i
2+
5.0 mg L
-1
, NOM 0 mg L
-1
(pH
i
5.64 0.38, pH
f

5.840.58), NOM 2.9-10.3 mg L
-1
(pH
i
5.85 0.59, pH
f
6.140.21) and NOM 5.0-
61.0 mg L
-1
(pH
i
5.69 0.34, pH
f
6.00.62), 5 or 20 min EBCT, 10% particle Fe
0

(v/v) supported with pumice]
75
Fig. 4.26 The concentration profiles of copper (c/c
0
), zinc (c/c
0
) and iron (total)
from the breakthrough column that was continually fed with TU runoff having
different NOM concentrations (~9 mg NOM L
-1
before 1000 BV and with 25 mg
NOM L
-1
thereafter) [Cu
i
2+
5.0 mg L
-1
, Zn
i
2+
5.0 mg L
-1
, pH
i
5-6, 20 min EBCT,
10% particle Fe
0
76
Fig. 4.27 LC-OCD diagrams of TU-SW and TU-FF measured for the raw samples
and the samples at end of batch experiment
79
Fig. 4.28 Metal removal profiles from composited columns tested at UDK site
during the first and second rain events. The columns received runoffs from a
copper roof in the downflow mode (during the measurement, the average flow was
60 mL min
-1
)
83
Fig. 4.29 Metal removal profiles from composited columns tested at TU site
during the first and second rain events. The columns received runoffs from a
bitumen roof with a zinc gutter in the downflow mode (during the measurement,
the average flow was 60 mL min
-1
)
84
Fig. 4.30 Breakthrough curves of copper and zinc from sequential columns of
Fe
0
/carbonatic materials; (a) controlled Fe
0
column; (b) calcite-Fe
0
column; (c)
DMS-Fe
0
column; (d) DML-Fe
0
column; (e) Magnesite-Fe
0
columns; and (f) Fe
0
-
DML column with corresponding pH measurement
86
Fig. 4.31 a) illustration of saturation indexes of relevant minerals related to Fe
0

treatment barrier of copper and zinc, b) plot of effluent metal concentrations
obtained from the test in equilibrium with metal hydroxides
88
Fig. 4.32 SEM and EDX mapping of solid precipitation obtained from selected
composited and sequential column tests
89
Fig. 4.33 Results of the implication of the optimized Fe
0
/carbonatic barrier 90
Fig. 4.34 Hydrualic characteristic of the optimized Fe
0
/carbonatic barrier over time.
The backwashing of the media yield sustainable hydraulic performance
92
Fig. 4.35 Applications of Fe
0
for treatment of highway runoff 95
Fig. 4.36 GFH/CaCO
3
barrier for treatment system of copper faade 96
Fig. 4.37 Catchment area characteristics for illustration of the design of treatment
barrier
97
Fig. 4.38 Conceptual design of the treatment system 98
Fig. 4.39 a) 3D view of the treatment barrier b) baffle reactor and withdraw units
of the excess sludge
101

Content xviii

Fig. 4.40 Model simulation of outlet concentration of zinc and copper from the Fe
0

barrier. Based on the research data, with propoer design of carbonatic post
treatment barrier, the effluent will meet a quality standard requirement.
102
Fig. B1 Results of batch test with pulse doses of Cu
2+
[Cu
2+
5.0 mg L
-1
with total
doses of 20 mg L
-1
, Fe
0
0.5 g L
-1
, T 20C at pH 4.0-7.0 (controlled) at 150 rpm]
138
Fig. B2 A comparison of batch results with pulse doses of Cu
2+
using Fe
0
and
rusted (corroded) Fe
0
at 0.5 g L
-1
each [Cu
2+
5.0 mg L
-1
with total doses of 20 mg
L
-1
, T 20C at pH 5.0 (controlled) at 150 rpm].
138
2+
using Fe
0
and
GFH at 0.5 g L
-1
each [Cu
2+
5.0 mg L
-1
with total doses of 20 mg L
-1
, DO 0 mg L
-
1
, T 20C at pH 5.0 (controlled) at 150 rpm].
139
Fig. B4 Mapping of SEM and EDX spectra of iron from typical batch experiment
in oxygenate solution
139
Fig. B5 SEM spectra of iron from batch experiment containg Pb
2+
[Pb
2+
5.0 mg L
-
1
, Fe
0
0.5 g L
-1
, T 20C, DO 0 mg L
-1
, pH 5.0 (controlled) at 150 rpm]
139
Fig. B6 Auger spectra of precipitates after experiments [Cu
i
2+
or Pb
i
2+
5 mg L
-1
,
Fe
0
dimension 2.1x1.05x0.5cm
3
-1
, T 20 1.0
C, 150 rpm in SSWR]
140
Fig. B7 SEM and EDX spectra of precipitates after experiments [Cu
i
2+
10 mg L
-1
,
Zn
i
2+
10 mg L
-1
, 0.5 g Fe
0
L
-1
, pH 7.0 (controlled), , T 20 1.0 C, 150 rpm in
SSWR]
141
Fig. B8 Breakthrough behaviors of copper and zinc from columns containing iron
from various sources. The Fe
0
-SL represent a spiral iron shape illustrated in Table
3.2 and the Fe
0
-LW is the iron used in the study of Ludwig (2007).
150
Fig. B9 Breakthrough behaviors of copper and zinc in a Fe
0
/PM column (50% v/v
47 g). The column was operated as following; Cu
2+
6.5 mg L
-1
, Zn 7.3 mg L
-1

using winter TU-runoff, 9.6 cm bed high, diameter 2.5 cm, EBCT 1.8 min. The
test was terminated due to clogging. Lines represent simulation.
150
Fig. B10 Breakthrough behaviors of copper and zinc in Fe
0
/PM column (10% v/v).
The column was operated as following; [Cu
2+
5 mg L
-1
, Zn 5-7 mg L
-1
using UFA
water collected in summer period, pH~6.2, DOC
in
5.9 and DOC
out
5.1 mg L
-1
,
EBCT 5 min].
151
Fig. B11 LC-OCD diagrams of NOM runoff from various sources before and after
Fe
0
treatment
157
Fig. B12 A comparison of LC-OCD diagrams of NOM runoff from various
sources
158
Fig. B13 Example of acidity determination of HOS fraction of runoffs 158
Fig. B14 Results of longterm performance of carbonatic/Fe
0
composited columns 160

Content xix

Content of Tables

Table 2.1 Species of iron oxides detected in the Fe
0
corrosion system 9
Table 2.2 Standard electrode potentials of contaminants relevant for Fe
0
barrier 12
Table 2.3 General properties of iron and iron oxides 14
Table 2.4 The solubility and K
s0
values of some minerals relevant to Fe
0
barrier 15
Table 2.5 Characteristics of NOM containing water from different sources 22
Table 2.6 Experimental and modeling approaches of inorganic removal by Fe
0
and
iron oxides
29
Table 3.1 Characteristics of source waters for batch investigation 31
Table 3.2 Chemical and physical parameters of media used in this study 40
Table 3.3 Experimetal setting and eperational parameters of the column tests 40
Table 4.1 Experimental conditions and kinetic parameters on the effects of DOC
and IS
46
Table 4.2 Operational and input parameters for modelling of the breakthrough
curves
65
Table 4.3 Characteristics of source waters for investigation of NOM impacts 70
Table 4.4 Estimated contents of carboxylic and phenolic groups 72
Table 4.5 Water quality of the influent to Fe
0
-aeration-DML/PM system 90
Table 4.6 The criteria and design results of the ZVIB 99
Table 4.7 Input parameters for simulation of the long term efficiency of ZVIB 102
Table 4.8 Cost comparison for ZVICB and GFH/Lime treatment systems 103
Table A1 Physical and chemical data of iron, copper and zinc 120
Table A2 Concentration of copper and zinc in roof, street and highway runoff
samples
120
Table A3 Concentration of metals in highway runoffs 120
Table A4 Defined processes of contaminant removal ounder various experimental
conditions
121
Table B1 Determination of pH
PZC
by titration method 122
PZC
by salt addition method 123
Table B3 Determination of surface area of Fe
0
124
Table B4 Determination of surface area of iron oxides (rust) 125
Table B5 Determination of surface area of corroded iron 126
Table B6 Determination of Pb
2+
removal in Fe
0
system with pulse dosing 127
Table B7 Results of batch investigation to determine the impact of pH and DO 128
Table B8 Impact of Suwanee river fulvic acid on the kinetic of Cu
2+
and Zn
2+

removal
129
Table B9 Example of monitoring results of pH, ORP and DO from batch tests 130
Table B10 Example of monitoring results of pH, ORP and DO from batch tests 131
Table B11.1 Results of equilibrium batch test of Cu
2+
and Zn
2+
removal by Fe
0
132
2+
and Zn
2+
removal by Fe
0
133
2+
and Zn
2+
removal by Fe
0
134
2+
and Zn
2+
removal by Fe
0
135
2+
and Zn
2+
removal by Fe
0
136
2+
and Zn
2+
removal by Fe
0
137

Content xx

Table B12 Determination of pH impacts on breakthrough curves 142
Table B13 Determination of DO impacts on breakthrough curves 143
Table B14 Determination of conductivity impacts on breakthrough curves 144
Table B15 Determination of temperature impacts on breakthrough curves 145
Table B16 Determination of NOM impacts on breakthrough curves 146
Table B17 Determination of NOM impacts on breakthrough curves 147
Table B18 pH before and after the columns test of breakthrough experiments 148
Table B19 pH before and after the columns test of breakthrough
experiments (Cont.)
149
Table B20 Model prediction of breakthrough curves at EBCT 40 min 151
Table B21 Model prediction of breakthrough curves at Fe
0
50% 153
Table B22 Characterization of runoff NOM in term of hydrophobicity 154
Table B23 Determination of NOM impacts by batch test at initial acidic pH
i
2.5 155
Table B24 Determination of NOM impacts by batch test at initial runoff pH
i
5.0 156
Table B25 Results of field study of composited column 159
Table B26 Long term monitoring of field study of composited columns 160
Table B27 Breakthrough of copper in sequential column tests 161
Table B28 Breakthrough of zinc in sequential column tests 162
Table B29 pH of the sequential column tests 163
Table B30 Field study of sequential columns 164
Table B31 pH of field study of sequential column 164
Table B32 Field study of Fe
0
-DML/PM system (Slow Flow rate) 164
0
-DML/PM system (Max flow rate) 164
Table B34 Parameter used for cost estimation 166
Table B35 Data for calculation of cost factor (cost in Euro) 166
Table B36 Estimated investment cost for GFH/Lime and ZVICB treatment barrier
(cost is simplied for comparison)
166

Content xxi

Nomanclarture

AES Auger Electron Spectroscopy
BET Brunauer Emmett Teller
BV Bed volume
BVT Bed volume treated
CPHSDM constant pattern homogeneous surface diffusion model
Cu-R Copper roof runoff
DMS Dolomite Magno Fielt
DML Dolomite Magno Dol
DI Deionised water
DO Dissolved oxygen
DOC Dissolved organic carbon
EBCT Empty bed contact time
EDX Energy dispersive X-ray analyzer
GFH Granular ferric hydroxide
HA Humic Acid
HPO Hydrophobic fraction of NOM
HPI Hydrophillic fraction of NOM
HSDM Homogeneous surface diffusion model
IHSS International humic substance society
IS Ionic strenght
LC-OCD Liquid chormatography and online carbon detection
MINEQL
+
Chemical equilibrium software
NOM Natural organic matter
ORP Oxidation-reduction potential
pH
PZC
pH at point of zero charge
PSDM Pore and surface diffusion model
SEM Scanning electron microscopy
SRFA Suwannee river fulvic acid
SR-NOM Suwannee river NOM
SSWR Simulated storm water runoff (1:5 of tab water and DI)
T Temperature
TPI Transphillic fraction of NOM
TU-FF First flush of TU roof runoff
TU-SW Runoff from TU roof
UDK University of art, Berlin
UVD UV detection
XRD X-Ray Diffractograms
ZVI, Fe
0
Zero valent iron
ZVIB Zero valent iron barrier
ZVICB Zero valent iron incorperated carbonatic materials

Content xxii

Symbol

K
F
Freundlich constant
k
f
Film diffusion coefficient
D
S
Surface/pore diffusion coefficient
K
s0
Solubility constant
Eh
0
Standard electrode potentials
K
a
s
equilibrium constants
Cu
2+
/Cu
0
Redox couple of Cu
2+
and Cu
0

pH
i
Inlet pH
pH
f
Effluent or outlet pH
q(c) The solid-phase concentration

Bulk density

Water content
R
f
Retardation factor
n Number of bed required
C
in
(Zn) Inlet concentration of zinc
Cu
out
(Zn) Outlet concentration of zinc
q
M
Load of metal uptake
C
in
(DO) Inlet concentraion of dissolved oxygen
Cu
out
(DO) Outlet concentraiton of dissolved oxygen
v
F
Flow velocity
r(DO) Iron corrosion rate
a
v
Volumetric surface area of Fe
0

Total removal rate

m
(Fe) Total iron mass

t
50
Time to reach 50% usage of iron
Fc(0x) Concentration of oxidized iron
c
S
Mass of sludge
H
Pc(0H)
2
Mass of iron hydroxides
H
Pc
Iron mass

1. Introduction

1 Introduction

1.1 New concepts of urban water management

The rapid and improper development of an urban catchment contributes to a significant
increase in sealed surface area, i.e., concrete and asphalt on building, pavement, street and
highway structures. As compared with a natural surface, impermeable surfaces destructively
impact the water balance in the cities. This results in an array of problems including an
increased volume of stormwater runoff and higher peak flow rates, which are regarded as the
main causes of flooding. Furthermore, the concept of transporting rainwater out of the
catchments through modern sewer systems as rapidly as possible also led to a great decline of
urban soil moisture and natural recharge of the groundwater.

The infiltration of roof, street and highway runoffs is a recent idea of urban water
management of increased interest. Infiltration of stormwater may be performed through
different strategies using swale, pit, and trench or through infiltration ponds and porous
pavement systems. Wherever possible, stormwater runoff shall be infiltrated onsite in order to
decrease the flooding potential, while the filtrate also sustains the declining groundwater
resource (Boller, 1997). Despite this positive fact, application of stormwater runoff
percolation remains questionable in many cases, primarily due to the occurrence and
increased load of dissolved and particulate contaminants in stormwater runoff.

1.2 Contaminants in urban stormwater runoff

The relevant contaminants present in urban runoffs are mostly associated with surface
materials that the runoff passes through. Major sources of contaminants for street and
highway runoffs are fuel combustion products, lubrication system losses, degradation of
automobile tires, loss of transported load, as well as paint and corrosion products. Airborne,
soil erosion, industry, litter, animals, vegetation are some of the other sources of stormwater
pollution that may be transported by rainfall and wind from other sites and deposited in the
highway catchment (Barbosa, 1999). The pollution load in highway runoff depends on the
average daily traffic, air quality and rainfall intensity and duration (Hvitved-Jacobsen et al.,
1994). Of the various kinds of pollutants, heavy metals generated by vehicle wear and
emissions are considered priority pollutants because of their toxicity (FHWA, 1996). These
heavy metal species, unlike organic pollutants, do not degrade in the environment (Barbosa,
1999). The most commonly detected heavy metals in street and highway runoffs are copper
(Cu), lead (Pb) and zinc (Zn).

In contrast to street and highway runoffs, roof runoff has been traditionally considered as a
clean water resource suitable for groundwater recharge. Roof runoff generally contains less
organic and inorganic particulates. However, when runoffs originated from large surfaces
have to be discharged into small- and middle-sized streams, they can be contaminated with

1.Introduction 2

concentrations of organic and inorganic pollutants, particularly dissolved heavy metals from
the roof surface. These may exceed the legal dissolved concentration level (Fig. 1.1).

When addressing groundwater-associated runoff problems, heavy metals including cadmium
(Cd), copper, chromium (Cr), lead and zinc should be ranked in priority because of their
toxicity and persistent characteristics. In urban runoff samples, Cu, Pb and Zn are detected in
over 90% of the samples, cadmium and chromium are found to a lesser extent (Pitt, 1996).
Due to fuel regulations, lead usage is generally declining. Simultaneously, significant
emissions of copper and zinc have been reported in runoff originating from copper and zinc
roofs, typical metallic components found in urban areas. For subsequent groundwater
recharge, metals which are bound to particles can be readily removed by surface filtration and
sedimentation (Pitt, 1996). Due to the accumulation of heavy metals, the upper soil layers are
considered as hazardous materials. Additionally, dissolved heavy metals require particular
consideration as they are highly mobile. As a consequence, percolation of the contaminated
runoff waters may not only lead to a diffuse accumulation of pollutants in the soil or subsoil
layers but they may also be considered as a potential cause of long-term deterioration of
groundwater quality. This especially concerns the recharge sites that contain mainly sandy
and loamy soils or sites that have been subjected to acidic conditions (Pitt, 1996). To assure
the quality of filtrate water, a pretreatment system for contaminated runoff is required.

Fig. 1.1 Concentration of copper and zinc in roof and highway runoffs (c.f., Table A2 and A3)
0,01
0,1
1
10
100
Bayreuth
1988
Munich
2001
Paris
1998
Norway USA Germany
Min 10 5 0,8 0,091 0,013 0,32
Max 85 23 38 0,74 0,173 0,62
C
o
n
c
e
n
t
r
a
t
i
o
n

[
m
g

L
-
1
]
Bern 2000
Osnabrck
1997
Karlsruhe
2004
Norway USA Germany
Min 0,8 0,5 4 0,01 0,01 0,058
Max 10 33 32 0,43 0,101 0,117

1.Introduction 3

1.3 Objectives of the study

The goal of this research work was to develop a novel treatment system of stormwater runoff
contaminated with heavy metals. A simple and cost effective system was targeted. Therefore,
a fixed-bed system containing reactive materials was primarily considered. The system can be
placed downstream of the contaminated-flow and requires no consumption of energy, making
it economically feasible.

In this study, zero-valent iron (ZVI, Fe
0
) was chosen as a promising treatment medium. Fe
0

has advantages over other media, e.g., granular ferric hydroxide (GFH) (Driehaus et al., 1998;
Boller and Steiner, 2002; Steiner, 2003), due to its low cost as well as environmental benefits
in terms of reuse of solid wastes. Furthermore, ZVI has been successfully implemented as a
fixed-bed barrier for an on-site remediation system (e.g., Morrison et al., 2002). The reactive
materials have been shown to be environmentally safe, possess stability and sufficient
permeability. Nevertheless, when Fe
0
has to be applied under runoff conditions, issues
governing mechanisms of metal removal, source variations of runoff that may affect the
performance of treatment barrier, quality of the treated-water and a potential clogging of the
Fe
0
barrier need to be assessed. The design of a Fe
0
barrier requires the knowledge of site
characteristics such as geotechnical data, nature of the contaminants and the investigation of
critical parameters such as the required contaminant/reactant residence time and the stability
of possible transformation products. Major issues that must be resolved before the installation
of a system are the long term performance when operating under high oxygenated and high
flow conditions.

For a complete study of processes and development of a final engineering design of
stormwater runoff treatment systems, this PhD research had the following objectives:

(1) to understand the processes, kinetics as well as the equilibrium and maximum uptake
loads of metal removal in Fe
0
system,

(2) to investigate the behaviour of metal removal by Fe
0
under flow-through conditions
and changing runoff qualities,

(3) to understand the nature of natural organic matter (NOM) in roof runoffs and its
impact on the treatment processes,

(4) to model the breakthrough of Fe
0
column allowing the prediction of the results and
further up-scaling of the process, and

(5) finally to further optimize the treatment barrier using carbonate bearing materials.

Based on the understanding of these processes involved, the final engineering design of
stormwater runoff treatment systems can be established.

1.Introduction 4

1.4 Structure and outline of the thesis

At the beginning of the study, the processes and mechanisms for copper and zinc removal
from stormwater runoff employing Fe
0
were investigated employing batch kinetic and
equilibrium tests (Section 4.1). The study employed stormwater runoffs taken from roof,
street and highway. The impact of stormwater characteristics, e.g., dissolved oxygen (DO),
pH, temperature (T), dissolved organic carbon (DOC), ionic strength (IS) and metal
concentration on metal removal were thoroughly studied. Following the batch test, Fe
0
was
further assessed using a flow-through system offering a closer condition to actual implication
of the iron barrier. The breakthrough, kinetics and mass flow analyses were performed using
Fe
0
particle supported with pumice columns. They were tested using runoff collected from a
roof with varying water quality parameters (Section 4.2).

A model capable of simulating the dynamic processes of copper and zinc removal by Fe
0
at
varying runoff conditions is important for the engineering design of the treatment barrier.
Generally, a modeling approach for metal removal in Fe
0
system is complex and requires
several parameters that need extensive studies. Section 4.2.3 demonstrates a simple approach
for modeling the metal breakthrough curves in a Fe
0
system. The model was based on a pore-
surface-diffusion model (PSDM) using the AdsorbDesign software. It was shown that the
model is able to predict breakthrough curves obtained from column tests adequately and
allows the prediction of breakthroughs of metals at various scenarios.

In the course of the study, it was found that removal of metals dramatically decreases in the
presence of natural organic matter (NOM). Currently, there are only a limited number of
investigations that lead to the understanding of NOM impacts on the removal of contaminants
using Fe
0
in literature. Therefore, one of crucial aims of this study was to assess the impacts
and mechanisms associated with NOM in the Fe
0
treatment system of Cu
2+
and Zn
2+
. In
section 4.3, the runoff samples from two roofs taken at various periods of time were
characterized exclusively using XAD-4/8 adsorption resins, copper complexation, acidic
capacity and liquid chromatography with online carbon detection. Batch kinetic experiments
and flow-through configurations were performed and the results of metal removal were
elucidated taking into account the characteristics of NOM.

Following the laboratory test and modeling, the iron system was optimized in order to ensure
the required removal of pollutants within an appropriate time frame. To enhance the removal
of metals of Fe
0
columns, carbonatic materials were added to the iron system and tested at
various configurations. The result showed that a higher efficiency in terms of metal
immobilization could be achieved. Nevertheless, this also brings about a high clogging
potential of the treatment barrier by extensive accumulation of precipitated products. The
most promising configuration consisted of a spiral shape of Fe
0
followed by aeration and a
fixed-bed pumice/dolomite filter installed on-site for treatment of copper and zinc roof runoff
(Section 4.4).

1.Introduction 5

In the last chapter (Chapter 5), a summary of the results of this work and a recommendation
for further up-scale of the treatment system is discussed.

2. Theoretical background and practices

2 Theory and background of zero-valent iron barrier

2.1 Zero-valent iron barrier (ZVIB) for water treatment

A permeable reactive barrier (PRB) is an engineered zone of reactive materials placed
downstream of contaminated water (Fig. 2.1). The PRBs have been successfully employed for
remediation of a variety of compounds such as chlorinated hydrocarbons (Gillham and
OHannesin, 1992; 1994), nitrate (Robertson and Cherry, 1995; Huang and Zhang, 2005),
metals (Shokes and Mller, 1999) and radionuclides (Cantrell et al., 1995). Depending on the
targeted compounds, PRBs may be named in different terminologies such as a de-nitrification
barrier for the removal of NO
3
-
, permeable reactive walls and biological reactive barriers
(Robertson and Cherry, 1995; Tombre et al., 1997). In groundwater contaminated with acid
mine drainage, the reactive barriers may be constructed primarily of limestone, which is
referred to as a limestone drain (Hedin et al., 1994).

Fig. 2.1 Permeable reactive barriers for treatment of contaminanted groundwater

The so called Zero-valent iron barrier is the most common type of a permeable reactive
barrier. In this barrier system, Fe
0
is the medium primarily used for the treatment of
contaminants. The environmental usage of a Fe
0
barrier derived from a discovery by
researchers at the University of Waterloo in 1992 (Gillham and OHannesin, 1992). The
application of the technology was attributed earlier to halogenated aliphatics such as
trichloroethylene (TCE), cis- and trans-dichloroethylene (DCE) and vinyl chloride (VC)
(Gillham and OHannesin, 1992). Later, this was expanded to the treatment of acid mine
drainage (Wilkin and McNeil, 2003), inorganic contaminant chromate (CrO
4
2-
) and
redionuclides such as Uranite and a range of inorganic heavy metals such as Cd
2+
, Co
2+
, Zn
2+
,
Pb
2+
, Mn
2+
, Ni
2+
and complex ions such as UO
2
2+
as well as SO
4
2-
, NO
3
-
, SeO
4
2-
, TcO
4
2-
, and
CrO
4
2-
(Cantrell et al., 1995; Blowes and Ptacek, 1992). Most recently, Fe
0
was investigated

2. Theoretical background and practices 8

and applied to the treatment of Arsenate contaminated drinking water resources (Bang et al.,
2005; Karchunker and Jekel, 2002) and stormwater runoff treatment (Rangsivek and Jekel,
2005; 2007; 2008; Ludwig and Jekel, 2007).

2.2 Iron corrosion

The fundamental knowledge on the corrosion processes of iron, corrosion products and their
behaviors on metal removals is crucial to the understanding of the remediation processes for
inorganic contaminants in the Fe
0
barrier system as presented in the following;

2.2.1 Corrosion processes and mechanisms

When Fe
0
is immersed in an aqueous solution, an electrochemical corrosion will take place.
The corrosion of iron can be described as the oxidation of iron and its corresponding
reduction reaction: electrons are released from the anodic region of iron thus forming a
soluble ferrous iron (Fe
2+
) that dissipates out. In concomitant with the oxidation, the reduction
of oxygen takes place at the cathodic site and in the bulk water phase this also yields an
aqueous OH
that leads to an increase in pH (Eq. 2.1).

The fate of Fe
2+
depends on the solution chemistry. In the presence of sufficient DO and other
related oxidants, Fe
2+
will be oxidized and form Fe
3+
, which is subsequently precipitated into
amorphous ferric hydroxides (Fe(OH)
3(s)
or ferrihydrite) according to equation 2.2 and 2.3.
The precipitates of these insoluble ferric hydroxides and oxyhydroxides may either localize
close to the metallic iron surface or dissipate out into the bulk solution (Smith, 1998; Farrell
et al., 2000). In the Fe
0
barrier, the pH of the system generally fluctuates; there is a
consumption of proton as corrosion progresses in contrast to the formation of Fe(OH)
3
that
leads to a decrease in pH.

Aerobic Corrosion
2Fc
0
+ 0
2
+2E
2
0 = 2Fc
2+
+ 40E
-
(2.1)
4Fc
2+
+ 0
2
+2E
+
= 4Fc
3+
+ 20E
-
(2.2)
Fc
3+
+ 20E
-
= Fc(0E)
3(s)
(2.3)

Anaerobic Corrosion
Fc
0
+ 2E
2
0 = Fc
2+
+E
2
+ 20E
-
(2.4)
Fc
2+
+ 20E
-
= Fc(0E)
2(s)
(2.5)
SFc(0E)
2
=Fc
3
0
4
+ E
2
+ 2E
2
0 (2.6)

In the absence of DO, the corrosion of iron takes place through different pathways which
results in the formation of hydrous ferrous oxide (FeO.nH
2
O) or ferrous hydroxide (Fe(OH)
2
)
and H
2
(2.4-2.6). The rate of hydrogen reduction is much slower than as compared to a high


DO condition. In the Fe
0
barrier, it can be anticipated that an anoxic or anaerobic corrosion
can occur at the flows downstream as dissolved oxygen is depleted during the transport
through the Fe
0
barrier. Moreover, this type of iron corrosion can take place on the inner rust
layer that has been previously formed following the complete exhaustion of dissolved oxygen
at the outer layer of oxides.

2.2.2 Corrosion products

Although the mechanism of iron corrosion has been intensively studied, there is no general
agreement on the conditions where a single corrosion product can be formed and stabilized.
This is partly because the composition of rust depends on the condition where corrosion
occurs and the resulting corrosion products usually change with time.

Table 2.1 Species of iron oxides detected in the Fe
0
corrosion system*
Sources Metrix Condition Detected iron oxide species
Phillips et al. (2000) G Aerobic/anaerobic Iron oxyhydroxides, akaganeit (-FeOOH),
goethite (-FeOOH), aragonite (CaCO
3
),
siderite (FeCO
3
), iron sulfide (FeS)
Roh et al. (2000) G Aerobic/anaerobic Green rust, lepidocrocite, goethite,
akaganeite, siderite, amorphous iron
hydroxides, magnetite, akageneite etc.
Furukawa et al. (2002) G Anerobicic Ferrihydrite, magnetite, green rust 1,
aragonite, calcite, mackinawite, greigite and
lepidocrocite
Odziemkowski (1998) sG Anaerobic Ferrous hydroxide, magnetite
Farrel et al (2000) sG Anaerobic Magnetite, maghemite
Manning et al. (2002) sG Anaerobic Lepidrocrocite, magnetite and maghemite
Liang et al. (2003) G Anaerobic Iron oxides, -carbonate, -sulfides
Su et al. (2004) sG - Magnetite, hematite and lepidrocrocite
Kohn et al. (2005) G Anaerobic Wstite (FeO), magnetite (Fe
3
O
4
), maghemit
(-Fe
2
O
3
), fayalite (Fe
2
SiO
4
)
Wilkin and McNeil (2005) sG Anaerobic Green rust 1 and 2, mixed valent of iron
hydroxides
Rangsivek and Jekel (2005) R Aerobic Lepidocrocite, goethite and hematite
* data complied by Ludwig (2007) with additional references, G: groundwater; sG: simulated groundwater, R: runoff water

Many factors may influence the speciation and morphology of the iron corrosion products
(Table 2.1). Compounds which are often reported as iron corrosion phases in the Fe
0
barrier
usually include goethite (-FeOOH), lepidocrocite (-FeOOH), magnetite (Fe
3
O
4
), siderite
(FeCO
3
), ferrous hydroxide (Fe(OH)
2
), ferric hydroxide (Fe(OH)
3
), ferrihydrite (5Fe
2
O
3

9H
2
O), and green rusts (Roh et al., 2000; Phillips et al., 2000; Furukawa et al., 2002; Wilkin
and McNeil, 2005). Under anaerobic conditions, amorphous ferrous hydroxide, Fe(OH)
2
, is
the first corrosion product that forms which can be oxidized to form magnetite (Fe
3
O
4
)
(Cornell and Schwertmann, 1996; Odziemkowski, 1998). Green rusts are often found in water
containing carbonate, chloride and sulfate and is stable only at low redox potentials. Their
oxidation commonly leads to the formation of maghemite (-Fe
2
O
3
) and lepidocrocite
(Cornell and Schwertmann, 1996). In contrast, the oxidation of the Fe
2+
ions in well-aerated


systems at neutral pH leads to the formation of ferrihydrite Fe(OH)
3
(s) (Cornell and
Schwertmann, 1996). As Fe(OH)
3
is dehydrated over time it loses water molecules and slowly
changes into other trivalent irons such as lepidocrocite, goethite and most of the hydrous
ferric oxides as ordinary rust and oxide scales (Cornell and Schwertmann, 1996). Both rust
and oxide scales are usually comprised of a mixture of iron oxides together with other Fe
(e.g., siderite (FeCO
3
)) and non-Fe compounds (CaCO
3
) which are often found in the
processes of steel pipe corrosion (Stumm, 1957; 1959; Smith et al., 1997). On pure iron, -
FeOOH transforms to become the more stable -FeOOH. With increasing time, goethite
converts to either maghemite or hematite, the later usually requires higher temperatures. It can
exist as the non-magnetic FeOOH, -Fe
2
O
3
or magnetic -Fe
2
O
3
(Cornell and Schwertmann,
1996).

2.2.3 Corrosion rate

A high corrosion rate of iron is an important factor in controlling the removal of contaminants
in natural water. The rate of iron dissolution depends on various physical and chemical
parameters; it increased with temperature and flow velocity as a result of a higher mass
transport through reactive surfaces. The high flow velocity transports Fe
2+
ions away from the
Fe
0
surface thus lowers the supersaturation of precipitation production (Noubactep, 2008).
The increase of flow rate may also breakdown the previously formed iron oxides layer on the
Fe
0
surface. Since iron solubility is very low at natural pH, a higher corrosion rate of iron can
also lead to a formation of passivating layer that may subsequently retard the rate of oxygen
transport and iron corrosion in the long-term. Besides these factors, the increase of ionic
strength was also found to increase corrosion rate of iron by inducing precipitation of iron
oxides away from the iron surface (Furukawa et al., 2002; Rangsivek and Jekel, 2005).

Iron corrosion rate may be indirectly determined with the balance of dissolved oxygen
consumption before and after passing the iron surface as illustrated by Kuch (1984), Merkel
(2002), Karschunke (2005) and Ludwig (2007). The application of balance of oxygen depends
on the assumption of known final corrosion products and iron surface area. In a drinking
water pipeline, the corrosion rate was reported to be approx. 20 g m
-2
d
-1
(T = 12 C, v = 0.5
m s
-1
) at the beginning, which significantly dropped to 4 g m
-
d
-1
after one week of operation.
With approx. 2 years operation, the final stable corrosion rate of approx. 0.1 g m
-
d
-1
was
found (Sontheimer et al., 1981). In the study of metal removal in cast iron barrier, Karschunke
(2005) and Ludwig (2007) assumed the final products for Fe
2+
, Fe
3
O
4
and Fe
3+
, which results
in corrosion rates between 0.5-2.4 g m
-2
d
-1
with oxygen consumption rate of 0.06-0.02 L m
-2

min
-1
. A significantly lower iron corrison rate was found in the batch investigation by
Rangsivek and Jekel (2005) which were 0.005-0.034 g m
-2
d
-1
. This was likely due to a larger
surface area of iron determined by BET method as compared to calculated values from
physical data in other works. Although, the BET method, involving adsorption isotherm of
inert gas onto solid pore, give reproducible results, it is known that validity of the method
depends on the assumption underlying the data analysis of the method (Dzombak and Morel,
1990).


2.3 Removal processes in a ZVI system

2.3.1 Removal mechanisms

Four most possible and important physico-chemical processes involving immobilization of
inorganic compounds in the iron barrier are reductive transformation, adsorption/co-
precipitation, and precipitation (Fig. 2.2 and 2.4).

Reductive transformation process
Reductive transformation consists of two partial redox-reactions, i.e., the oxidation of one
species and the reduction of another that can be characterized by oxygen or electron transfer.
Accordingly, several inorganic contaminants may be reduced into their insoluble forms. The
reductive transformation process is the so-called cementation process in hydrometallurgical
processes for recovery of metal from the waste solution (Lung, 1986). It is well-known that
the cementation apparently takes place under acidic conditions in the absence of dissolved
oxygen at condition where iron is relatively soluble. The general accepted stoichiometry of
the cementation process can be described according to the following equation (2.7).

nHc
m+
+ mN = nH +mN
n+
(2.7)

The schematic for the recovery of metallic copper is shown in the following equation;
(Nadkarni et al., 1967; Nadkarni and Wadsworth, 1967; Annamalai and Murr, 1979;
Strickland and Lawson, 1971; Biswas and Reid, 1972; Ku and Chen, 1992);

Cu
2+
+ Fc
0
= Cu
0
+Fc
2+
(2.8)
Eb
_
Cu
2+
Cu
0
]
0
= +u.S4 I
Fig. 2.2 Schematic of possible reactions relevant to inorganic (i.e., copper and zinc) removal in Fe
0

barrier (c.f., text)


Table 2.2 Standard electrode potentials (E
0
) of contaminants relevant for Fe
0
barrier
Reaction Eh(V)
2E
2
0 + 2c
-

2E
2
0 + 20E
-
-0.83
Fc
2+
(oq) + 2c
-

Fc
0
-0.44
CJ
2+
(oq) + 2c
-

CJ
0
-0.40
E
+
+ c
-

1 2 E
2
(g) -0.00
Cu
2+
(oq) + 2c
-

Cu
0
+0.34
Zn
2+
(oq) + 2c
-

Zn
0
+0.76
Fc
3+
(oq) + c
-

Fc
2+
(oq) +0.77
0
2
+2E
2
0 + 4c
-

40E
-
-0.81
2Eg
2+
(oq) + 2c
-

Eg
2+
+0.92
Fc(0E)
3(S)
+SE
+
+ c
-

Fc
2+
(oq) +SE
2
0 +0.93
PJ
2+
(oq) + 2c
-

PJ
0
+0.99
Fc
2+
(oq) +SE
2
0

40E
-
+0.81
Cr
2
0
7
2-
+14E
+
+ 6c
-

2Cr
3+
+7E
2
0 +1.33
Ref. Benjamin (2002); Karschunker (2005); Hamann (1975) ; chemiemaster.de; Farrel et al. (2001)

With a stoichiometry factor (m/n) equal to 1, hence, about 0.88 g of iron is required for 1 g of
copper removal. The cementation of copper takes place due to a strong driving force for Cu
2+

reduction [+0.34 V for Cu
2+
/Cu
0
; -0.44 V for Fe
2+
/Fe
0
]. Table 2.2 depicts the standard
electrode potential for various compounds relevant for Fe
0
barrier in water treatment
processes.

In natural aerated water, the electrochemical reduction of redox-sensitive compounds takes
place through other redox couples than Fe
0
. Ongoing researches have shown that the
reduction of copper in natural water conditions likely takes palce via the reduction of aqueous
Fe
2+
or by the adsorbed or structural Fe
2+
[-0.35 to -0.56 V for Fe
3+
/Fe
2+
] (Noubactep, 2009).
This is the case since it was found that dissolved oxygen is mostly consumed by Fe
2+
rather
than by the surface of metallic irons (Fe
0
) (Huang and Zhang, 2005). The reaction follows;

Cu
2+
+ 2Fc
2+
= Cu
0
+2Fc
3+
(2.9)

Adsorption process
Adsorption implies a process where chemical contaminants partition from an aqueous
solution onto a solid surface, resulting in an immobilization of contaminants. Among the
variable surfaces, iron (oxy)hydroxides produced by the corrosion of iron is the dominant
adsorptive sites in the iron barriers. Iron corrosion products are very effective adsorbents for
both organic and inorganic contaminants; e.g., the cation can be bounded on the solid surface
by an exchange of the protons as described in equation 2.10 (Smith, 1998; Benjamin, 2002).

Fc - 0E + Hc
2+
= (Fc0) - Hc
+
+ E
+
(2.10)

The charge on the iron oxides develops due to the dissociation of the surface hydroxyl groups,


which arises from the protonation and deprotonation of the functional group of the surface
(OH-group). The reactions can be expressed as acids/bases equilibrium by the following
dissociation reactions,

Fc0E
2
+
= Fc0E +E
+
K
u,1
S
(2.11)
Fc0E = Fc0
-
+ E
+
K
u,2
S
(2.12)

The equilibrium constants for these reactions are termed acidity constants,
S
a,1
K ,
S
a,2
K , with the
surface (Fe) and can be derived by the law of mass reaction:

K
u,1
S
=
{Pc0H]{H
+
]
|Pc0H
2
+
|
(2.13)
K
u,2
S
=
{Pc0
-
]{H
+
]
{Pc0H]
(2.14)

The affinity of adsorption is generally determined by pH
PZC
. The pH
pzc
is the pH at which a
positive charge by protonation exists in approximately equal numbers with negative charges
caused by the development of deprotonation on the iron surface (equations 2.15 and 2.16);

{ Fc0E
2+
] = { Fc0
-
] (2.15)
pE
pzc
= u.S (pK
u,1
S
+ pK
u,2
S
) (2.16)

Under an acidic condition or pH below pH
pzc
, protons are sorbed on the functional group that
causes the iron surface to have a net positive charge and, thus, inhibit cation adsorption. At

Fig. 2.3 pH dependent sorption of metal cations on iron hydroxide (adapted from Denver,
1997). The co-precipitation results of Cr
3+
and Zn
2+
and precipitation of Cd
2+
were obtained
from Crawford et al. (1993) and Smith (1996), respectively.


above pH
pzc,
the oxygenatom stays deprotonized and the surface prevails to have a net
negative charge, thus, enhancing cation adsorption (Benjamin, 2002; Merkel and Planer-
Friedrich, 2005). Other than pH, the adsorption is dependent on the media properties such as
surface area of oxides and other water quality parameters such as ionic strength, temperature,
and concentration of compounds in the water solution. General properties of iron oxides are
demonstrated in Table 2.3.The pH dependent sorption of metals on iron hydroxide is
illustrated in Fig. 2.3.

Most adsorption reactions are reversible and occur at relatively rapid rates. Some adsorption
reactions are specific with the attachment occurring preferentially at particular sites. Other
adsorption reactions are less specific and have more of a tendency to compete with other ions
for attachment sites. In addition, the adsorption can result from either inner-sphere or outer-
sphere complexation adsorption resulting from ion-pair bonding due to electrostatic forces
and hydration water that separates the solvated ion from the surface. Surface complexation
models have been used effectively to predict the reaction chemistry of adsorption processes
(Dzombak and Morel, 1990).

Co-precipitation process
Another fundamental mechanism that is involved in the uptake of contaminants in the Fe
0

barrier is known as co-precipitation. Co-precipitation occurs whenever species (e.g., iron)
precipitate (e.g., as hydroxide) co-exists with a trace amounts of foreign species. In this
process, the foreign species are simply entrapped in the mass of precipitates (co-
precipitation). Co-precipitation is a well-known unspecific removal mechanism of species
from the aqueous solution (Noubactep, 2008a) and independently takes place. There is no
general clear cut distinction between adsorption and co-precipitation (Crawford et al., 1993).
In the case of adsorption, the adsorptive surface is initially present prior to the addition of the
metal ion whereas the co-precipitation of metals takes place is in a way that the precipitating
surface is formed in the presence of the metal iron to be removed. Crawford et al. (1993)
showed that there is a large enhancement of Ni
2+
and Cr
2+
removal and smaller for Zn
2+
taking
place by the co-precipitation processes, where metal removal occurs at a lower pH than the

Table 2.3 General properties of iron and iron oxides
Substance Formula
b
Color
b
Density
b
[g
m
-3
]
pH
PZC
[-]
Surface area
b

[m
2
g
-1
]
Zero-valent iron
a
Fe
0
- 7.86
c
- 0.385
a

Wstite FeO black 5.9-5.99
Ferrihydrite 5Fe
2
O
3
9H
2
O red-brown 3.96 159-600
d

Magnetite Fe
3
O
4
black 5.18 8.0 4-100
Maghemite Fe
2
O
3
reddish-brown 4.87
Goethite -FeOOH red-brown 4.26 7.5 52
Akaganeit -FeOOH yellow-brown 3.56 7.9 178
Lepidocrocite -FeOOH orange 4.09 15-260
Haemetite Fe
2
O
3
red 5.26 8.6 43.5
Rust
a
-FeOOH,
-FeOOH, Fe
2
O
3

red-brown - 7.2 42
Ref: (a) This work; (b) Cornell and Schwertmann (1996); (c) Noubactep (2008); (d) Dzombak and Morel (1990)


adsorption and precipitation (Fig.2.3). It is evident from several investigations that co-
precipitation is one of the main removal mechanisms of organic and inorganic contaminants
in the Fe
0
barrier (Noubactep, 2008a; Wilkin and McNeil, 2003).

Chemical precipitation process
Precipitation of heavy metals is a commonly used process in wastewater treatment plants. The
precipitation of metals takes place when the solubility product of the solid is reached and, as a
result, the dissolved form of contaminants will be converted into its various insoluble species
such as hydroxides, sulfides and carbonates. Solubility of minerals can be expressed by
dissolution of metal hydroxides and the corresponding equilibrium constants as demonstrated
in the following;

Hc(0E)
2
(s) = Hc
2+
+ 2(0E)
-
(2.16)

K
s0
=
|Mc
2+
|{0H
-
]
2
{Mc(0H)
2
]

Hc(C0)
3
(s) + E
+
= Hc
2+
+ EC0
3-
(2.17)

K
s0
=
|Mc
2+
||HC0
3-
|
{McC0
3
]{H
+
]

Table 2.4 shows the solubility of calcium, copper, zinc and iron minerals relevant to the Fe
0

barrier. The precipitation of metals typically results in a sharp sigmoidal removal profile
occurring at a higher pH than adsorption/co-precipitation processes (Fig. 2.3). In contrast to

Table 2.4 The solubility and K
s0
values of some minerals relevant to Fe
0
barrier*
Mineral Reaction Log L
Calcite CaCO
3
(s) Ca
2+
+ CO
3
2-
-8.35
Dolomite CaCO
3
MgCO
3
(s) Ca
2+
+ Mg
2+
+ 2CO
3
2-
-1.70
Siderite FeCO
3
Fe
2+
+ CO
3
2-
-10.7
Smithsonite ZnCO
3
(s) Zn
2+
+ CO
3
2-
-10.78
Malachite Cu
2
(OH)
2
CO
3
(s) + 2H
+
2Cu
2+
+ CO
3
2-
+ 2H
2
O -5.80
Azurite Cu3(OH)
2
(CO
3
)
2
(s) + 2H
+
3Cu
2+
+ 2CO
3
2-
+ 2H
2
O -18.0
Portlandite Ca(OH)
2
(s) Ca
2+
+ 2OH
-
-5.19
Fe(OH)
2
(s) Fe
2+
+ 2OH
-
-15.10
Goethite -FeOOH(s) + H
2
O Fe
3+
+ 3OH
-
-40.4
Haematite 0.5Fe
2
O
3
(s) + 1.5H
2
O Fe
3+
+ 3OH
-
-42.7
Tenorite CuO(s) + 2H
+
Cu
2+
+ H
2
O -7.65
Cu(OH)
2
Cu
2+
+ 2OH
-
-19.4
Zinckite ZnO(s) + 2H
+
Zn
2+
+ H
2
O -11.33
ZnO(s) + H
2
O Zn
2+
+ 2OH
-
-16.89
Zn(OH)
2
(s) Zn
2+
+ 2OH
-
-1.6.4
References: * Data complied by Ludwig (2007) with modification of K
s0
in according with; Morel et al. (1993); Mann
(1980); Sigg and Stumn (1996); Schecher et al. (2001); Feiknecht (1963); Benjamin (2002); Schecher and McAvoy (2001)


reductive transformation and adsorption/co-precipitation processes, the precipitation of metal
hydroxide is a fast reaction and can proceed within a magnitude of a second under an alkaline
pH (Merkel and Planer-Friedrich, 2005). However, such a process implies that Fe
0
barriers
will sooner or later encounter different degrees of permeability losses, amour of reactive
surface and clogging by precipitated products (Simon et al., 2001; Cravotta and Watzlaf,
2002).

2.3.2 Removal of inorganic pollutants

A number of studies have been devoted to identifying the mechanisms involved in the
removal of inorganic compounds during transport through the reactive iron barrier.
Nevertheless, there has been greatly inconsistency among these works (c.f., Table A4 in
appendix). These inconsistencies can be attributed mainly due to the complexity of the iron
barrier itself, in which multiple reactions are possible, as well as the differences in the
employed experimental conditions.

Gu et al. (1998) conducted two batch experiments to compare the removal rate of iron and
iron oxides. In their experiments, a very high concentration of U
6+
per iron or iron oxides
mass ratio was tested under well mixed conditions. They observed the precipitation of
insoluble U
4+
on the iron surface and concluded that the uptake of U
6+
was primarily due to
the reductive transformation processes. Contrary to the shaking batch test, Noubactep et al.
(2003) performed the test using un-shook vials presumably closer to actual groundwater
conditions. The experiment was maintained undisturbed for over 50 days in darkness and
water quality parameters were measured at interval periods. Based on the observation; the
removal of U
6+
increased considerably from a mild pH range, i.e., from 18% at pH 3.9 to 94%
at pH 4.4 indicating the role of adsorption/co-precipitation processes because if the reductive
transformation was the main removal mechanism, there should have been a higher removal of
U
6+
at a mildly acidic pH (pH 3.9) compared to a higher pH. In addition to this work, Fiedor
et al. (1998) compared the reaction of U
6+
with Fe
0
under two distinguished conditions;
deoxygenated and oxygenated conditions and found that the adsorption is the dominant
removal mechanism. Qui et al. (2000) observed no transformation or only partial reduction of
metals (U
6+
, Cr
6+
and Se
6+
)

on the Fe
0
surface.

The batch investigation of Fe
0
treatment of heavy metals under an acid mine drainage
condition shows that most of the metals including Cu
2+
and Zn
2+
were removed via adsorption
and precipitation processes. The Hg
2+
was, however, removed by reductive transformation
because of its high driving force (Wilkin and McNeil, 2003). Gibert et al. (2003) reported a
formation of nodular shaped copper metals on the iron surface during passage through a
Fe
0
/organic-composed barrier and concluded that Cu
2+
was removed mainly through
cementation processes whereas Zn
2+
was removed via a metal sulfide-complex and the
adsorption onto iron (oxy)hydroxide and organic matter. Tranyek et al. (2005) reported the
rates of copper reductive transformation based on a range of batch tests. Lopez et al. (2003)
found that reductive transformation of copper prevailed only at a relatively high temperature
and at a high Cu
2+
per iron ratio. In their study, the reactive iron originated from short-blast


iron materials that contained only 7% metallic iron per its total unit mass. The reduction
process of Cu
2+
by Fe
0
has long been well established but generally limited in
hydrometallurgical processes under which growth of metallic copper on iron surfaces and a
sponge-like amorphous structure were often observed (Ku and Chen, 1992).

Based on these existing literatures, it is difficult to anticipate which reaction is prevalent at a
specific remediation site during Fe
0
treatment of heavy metals in the iron
barrier. The variability in physical and
chemical characteristics as well as the
strong ability of Fe
0
and related
corrosion products in regulating the
redox chemistry of the solution makes
the processes extremely complex.
Nevertheless, mechanisms that are
most important in the removal of
inorganic pollutants in an Fe
0
barrier
system can be named; (1) reductive
transformation, (2) adsorption/co-
precipitation and (3) chemical
precipitation (Cantrell et al., 1995;
Smith, 1996; Fiedor et al., 1998; Gu
et al., 1998; Shokes and Moller, 1999;
Blowes et al., 2000; Naftz et al., 2002;
Noubactep et al., 2003; Wilkin and
McNeil, 2003). Generally, the latter
two processes are favored at higher
pH in accordance with the corrosion
processes in the presence of dissolved
oxygen. The reductive
transformation or cementation
process, imply that redox-sensitive
compounds are reduced into insoluble
forms, has been generally shown to be
highly effective under acidic conditions in the absence of DO (Nadkarni et al., 1967;
Nadkarni and Wadsworth, 1967; Annamalai and Murr, 1979; Strickland and Lawson, 1971;
Biswas and Reid, 1972; Ku and Chen, 1992).


2.4 Fe
0
treatment of metal contaminated runoff

2.4.1 Process of metal removal

The schematic of reaction pathways possibly involved in the Fe
0
treatment barrier of Cu
2+
-
and Zn
2+
- contaminated roof runoff are demonstrated in Fig. 2.4. The cementation process
implies that redox-sensitive compounds such as copper are reduced into insoluble forms.
Although, it is well-known that the cementation reaction apparently takes place under acidic
conditions in the absence of dissolved oxygen (Nadkarni et al., 1967; Nadkarni and
Wadsworth, 1967; Annamalai and Murr, 1979; Strickland and Lawson, 1971; Biswas and
Reid, 1972; Ku and Chen, 1992), this process may not play significant role in an oxygenated
runoff. The reductive precipitation has been believed taking place via reaction with Fe
0

surfaces [0.34 V for Cu
2+
/Cu
0
; 0.76 V for Zn
2+
/Zn
0
, -0.44 V for Fe
2+
/Fe
0
], however, more
insight research has demonstrated that the indirect reduction with dissolved and structured
Fe
2+
in the vicinity of Fe
0
surface plays a major role in the process (Noubactep, 2008; Ghauch
et al., accepted).

When Fe
0
is placed in a flowing oxygenated runoff solution, Fe
2+
dissolutes out into the bulk
water phase and forms Fe
3+
which is subsequently precipitated into amorphous ferric
hydroxides (Fe(OH)
3(s)
or ferrihydrite) at an alkaline pH. The precipitates of ferric oxides are
either localized close to a metallic iron surface or dissipate out into the bulk solution (Smith,
1998; Farrell et al., 2000). The latter is more likely under stormwater runoff conditions due to
the low ionic strength of the runoff solution (Rangsivek and Jekel, 2005). The produced iron
oxides behave as a very effective adsorbent for several inorganic compounds, and thus, both
Cu
2+
and Zn
2+
can be removed via the adsorption reactions. Under the considered system, it is
also anticipated that the adsorbed heavy metals are slowly incorporated into the structure of
iron corrosion products while precipitating and transforming into other stable states such as
Goethite, Lepidocrocite and Hematite (-, -FeOOH and Fe
2
O
3
, respectively).

As a result of a production of OH
-
during iron corrosion, the runoff water can be naturalized
and if the increase of pH is high enough this may also lead to precipitation of various
hydroxides and carbonates of copper and zinc, e.g., CuOH
+
, CuCO
3
, ZnCO
3
and Zn(OH)
2
.
The solubility for copper and zinc equilibrium in a hydroxide-carbonate system are about 15-
50 g L
-1
in the pH range of 8-9 and 0.5-3.0 mg L
-1
at pH 9-11, respectively (Ludwig, 2007).
The lowest solubility of iron hydroxides is about 0.03 g L
-1
at pH 8-9. At concentrations up
to 0.5-2 mg L
-1
DOC, the solubility of copper may increase up to 0.92-1 mg L
-1
in the
equilibrium with malachite (Merkel, 2002). The copper solubility increases significantly from
approx. 0.14 mg L
-1
up to approx. 1 mg L
-1
in the presence of small quantities of DOC (0.5 to
2 mg L
-1
).


2.4.2 Practical issues of Fe
0
barrier

For simplicity and the desirable low cost of a runoff treatment system, a Fe
0
fixed bed is
considered feasible. However, the successful application of the system is usually limited by
two factors. The first is the systems efficiency in coping with the effluents quality at
variable runoff characteristics. Rangsivek and Jekel (2005) reported about twofold lower rates
of metal removal in the presence of a fraction of NOM. Also, the corrosion rates, phases and
morphology of the iron corrosion products can be strongly affected by water qualities and
operational parameters (Rangsivek and Jekel, 2005; 2008; Noubactep, 2008) and, thus,
influence the removal rate of metals. The second limitations for the actual implementation of
a Fe
0
system in especially oxygenated-water at high flow conditions of storm water runoff
conditions arise due to a loss of porosity and a decrease of the mass transport rate of
contaminants by generations of secondary minerals (Roh et al., 2000; Furukawa et al., 2002;
Kamolpornwijit et al., 2003). These problems are alleviated generally by mixing the Fe
0

media with sand.

2.5 Natural organic matter

2.5.1 Significance of NOM

Natural organic matter (NOM) is a complex mixture of a variety of polyfunctional organic
ligands, ranging from simple to more complex compounds of humic and fulvic acids. NOM is
highly heterogeneous, having a mixture of two major types of functional groups, carboxylic
and phenolic compounds, in which the pK
a
values are estimated to be 4.5 and 10, respectively
(Perdue, 1985). Based on these charateristics, NOM plays a crucial role in the transport and
fate of metals in aquatic environments (Aiken et al., 1985).

NOM is originated by different sources and can be classified into terrestrial zone, aquatic
zone, e.g., stream water, groundwater and seawater which are subsequently transported into
subsurface aquifers. NOM produced during sewage treatment processes is commonly referred
to as effluent organic matter (EfOM). Surface water NOM are believed to be the
degradation products of soil organic matter. In general, NOM are a mixture of by-products of
plants and microorganisms. It can exist in an aquatic environment as dissolved and particulate
compounds. NOM from different origins possesses independent characteristics and properties.
As compared with NOM in groundwater, the surface water is higher in hydrophobic
compounds and aromaticity (Table 2.5). The groundwater NOM is usually a residual of
surface water NOM, which has been adsorbed and degraded during transport to underground.
As a focal aim of this work, when rainwater falls into various surfaces, the generated runoff
may be enriched with natural organic matter and because of a wide variety of original material
on the catchment area, origin, age, chemical characteristics of the substances, can lead to a
variety of ways for its degradation and transformation. Thus, a complex structure and mixture
of the resulting substances may be produced.


Generally, NOM samples in the aquatic environment contain approximately 50%
hydrophobic and 30% hydrophilic fractions and the remaining 20% of total NOM
concentration are comprised of carbohydrates, amino acids, and other simple molecules
(Table 2.5). Among all fractions, the hydrophobic humic and fulvic acids are the most
reactive compounds which exhibit strong ligands. DOC may also include xenobiotic ligands,
like EDTA, NTA, or phosphonates in anthropogenically contaminated waters.

NOM are generally not toxic but direct problems arise with the presence of NOM with regard
to the drinkability resulting from aesthetic reasons such as undesirable smell, taste
impairments and color. A yellowish color is a result of humic acid in the drinking water.
However, the presence of NOM can also cause many problems in the drinking water
treatment process. During coagulation and flocculation reactions, NOM stabilizes dispersed
and colloidal particles in the water. NOM are known as an important precursor for potentially
hazardous Trihalomethanes (THM) or Halogenic acid (HAA) during the chlorination
processes. NOM is also considered a major cause of membrane fouling that is still a subject of
extensive studies. It is well known that NOM plays a crucial role in the transport of metals in
an aquatic environment. During the conventional treatment processes, NOM may compete for
adsorption sites with metals resulting in a decreasing capacity. It can significantly change
electrochemistry of the underlying layer of an adsorbent. For these reasons, it is anticipated
that NOM can strongly impact the removal of heavy metals in the Fe
0
treatment barrier.

2.5.2 Characterization of NOM

Due to the complexity of NOM, several techniques are required to completely characterize all
the essential information of NOM chemistry. Important characteristics of NOM include
composition, molecular size, molecular weight, structure and functionality. A brief review of
fractionation and characterization methods, which are employed in the present thesis, is
described in this section. The fractionation of NOM followed the XAD adsorption method
described by Aiken et al. (1985). The determination of molecular size/weight and acidity are
discussed according to gel permeation chromatography and potentiometric titration methods,
respectively. The copper complexation capacity is employed in order to gain the basic
knowledge of metal complexation capacity of NOM.

Natural organic matter exists in the environment both in dissolved and particulate forms.
However, when addressing NOM in the aquatic environment, the dissolved compound is
often the main interest to the study of aquatic chemistry. An effective means of isolation of
dissolved compounds from the total NOM samples can be performed by means of a
membrane procedure, a separation can be based on molecule size cutoff through 0.22 um.
However, in several cases this pore size of membrane pore size is too small and often results
in fouling and high operational cost. For general cases, filtration of NOM solution through
0.45 um membrane is suitable.


Fig. 2.5 NOM fractionation methods (Jarusuttirak, 2002)

In order to fractionate NOM, the state-of-art method or so called XAD-adsorption method
described by the international humic society is generally employed, which can be illustrated
in Fig. 2.5. The method is used to isolate humic substances from natural water. During the
procedure, the dissolved organic matter (DOM) was passed through XAD-4 and XAD-8 or a
combination of both columns. XAD resins are nonionic macropodous copolymers with large
surface areas. The hydrophobic effect is the principle driving force for organic sorption on
these resins. Sorption of organic acids such as humic substances is determined by the solutes
solubility and solution pH. Fundamentally, the hydrophobic fraction refers to the fraction that


is adsorbed onto the XAD-8 resin, which can be obtained by extracting the adsorbed
compounds using an alkaline extraction, i.e., NaOH. The same procedures were employed in
order to fractionate the hydrophobic and transphilic fractions. In the later cases, filtration was
made to XAD-4 and a combination of both columns, respectively. The fraction that can
neither be adsorbed on XAD-8 nor XAD-4 is designated as a hydrophilic, consisting of small
molecular polar materials as well as carbon- and amino acids and places 17 to 45% on the
DOC (Aiken et al. 1993). Before the XAD-fractionation, dialysis can lead to a separation of
colloids of existing NOM (Leenheer, 2004). To obtain fulvic acid out of humic substances,
the total fraction is precipitated at a low pH and passed through a desalting step using a cation
exchange. The humic acid and fulvic acid that are isolated from aqueous samples thus contain
only hydrophobic organic acids, but samples extracted from solid-phase materials include
both hydrophobic and hydrophilic acids (IHSS).

In a subsequent test for carbon-determination and UV-determination, the method is called
liquid chromatography (LC-OCD). The method divids total dissolved NOM into
polysaccharide, humic substances, building blocks, low molecular weight and amphilic
compounds (Huber and Frimmel, 1996). The measurement of the molecular size employed in
this work was based on gel permeation chromatography techniques (Huber and Frimmel,
1996). The principle of this technique is that the gel structure is perfused by a system of
pores, and the size of these is determined by the degree of cross-linking in the polymer. These
pores enable the gel to act as a chromatographic medium giving separations based on
differences in molecular size. When a solution, containing a mixture of molecules of varying
sizes, is applied to the top of a gel column and eluted with solvent, those molecules which
cannot enter the pores in the beads will pass between the beads and will be eluted first from
the column. Molecules smaller than the pore sizes of the gel will enter the pores and their
passage through the column will be retarded.

The most commonly encountered acidic functional groups in organic compounds include
carboxylic acids, phenols, ammonium ions, alcohols, and thiols. The most active organic
groups of NOM are carboxylic and phenolic compounds. In order to properly describe the
acid-base properties of a particular humic substance, it is essential that the identification and
quantification of acidic functional groups is accomplished. In this study, a direct titration

Table 2.5 Characteristics of NOM containing water from different sources
Source* pH Condc.

DOC UVA
254
SUVA HPO HPI TPI
(S cm
-1
) (mg L
-1
) (m
-1
) (m
-1
mg
-1
L) % % %
Natural water
a
7.1-8.5 183-384 2.7-8.4 5.7-29.5 1.4-3.5 39-50 24-35 22-28
Ground water
b
0.94 2.1 2.2 42
SRHA
b
6.8 300 10 74 7.4 94 1 5
AOM (Algal OM)
b
6.8 300 10 42 4.2 26 57 17
Influents to WWTP
c
6.5-7.4 4-4.9 43-53 37-40
EfOM
d
7.56-8.08 558-1195 5.9-14.3 16-26 1.82-2.71 32-40 40-45 20-23
Natural water
e
6.6-8.0 5.6-50.3 3.1-6.8 9-38 2.9-5.6 38-79 9-31 11-31
References: (a) Lee et al. (2004); (b) Amy 2004 (c) Korshin et al. (1997) (d) Jarusutthirak (e) Yoon et al. (2005)


technique is employed because it is a relatively simple, fast and commonly employed method
that will allow for comparison of the results. The potentiometric titration is operationally
defined by the carboxyl content as the concentration of functional groups that are neutralized
at a specified pH. The carboxylic group was determined in order to estimate carboxyl content
at the equivalence point for titration of pH around 7. The phenolic group was estimated at
about the double consumption of the base required for the carboxylic group (Oden et al.,
1993).

DOC was determined by means of thermal-catalytic oxidation using a high-TOC analyser. In
combination with spectroscopic methods, the aromaticity and the humic characteristic of the
water samples can be identified through the determination of spectral absorption at 254 nm
(UVA
254
), as well as the UVA/DOC ratio. The UV adsorption, conceptually linked to the
amount of aromatic carboxyl and phenolic groups. At the same time it is generally perceived
that specific UV radiation (UVA
254
/DOC) is an important proxy of aromaticity, total C-
content and molecular weight. By means of fluorescence and infrared-spectroscopic, single
structure features and functional groups can be identified (Senesi, 1990). The C13-NMR
(Nuclear magnetic resonance) represents a method to obtain information about the
functionality of the carbon atoms in the NOM-connections (Wong et al., 2002). A non-
destruction free method is the Pyrolyse-GC/MS; by using high temperatures to disaggregate
NOM into fleeting fragments, that can be identified by GC/MS-Analysis. Based on the
fragments, conclusions can be made on single material groups (protein, polysaccharide etc.)
(Abbt-brown et al. 1990; Crou, 2002). Other important characteristics of NOM involves the
determination of elemental content of the redissolved samples (DOC, Na, Ca, Cu, Fe, and
Mn), proton capacity and content of hydrolysable amino acids and carbohydrates.

2.5.3 Impacts of NOM on iron corrosion processes

Limited investigations are reported in the literatures that deal directly with NOM impacts on
the removal of contaminants using Fe
0
(Tratnyek et al. 2001; Dries et al., 2005). Nevertheless,
many previous studies that help in the understanding of the transport, fate and impact of NOM
were extensive. These works studied iron oxides or corrosion products, e.g., Davis and Leckie
(1978), Laxen (1985), Liang et al. (1993), Stevenson (1994), Gu et al. (2000), and Kieber et
al. (2005). Tratnyek et al. (2001) carried out the batch test to determine the impact of NOM
on Fe
0
reduction of several model organic compounds. In their study, NOM was found to
contribute to the inhibition effect on carbon tetrachloride and trichloroethylene (TCE),
presumably due to the adsorptive competition of the compounds for surface sites. The
increase in the reductive transformation rate of organic compounds was also observed in the
system containing juglone, lawsone, and anthraquinone disulfonate (modeled NOM
compounds), which was attributed to an electron mediation mechanism.

Dries et al. (2005) investigated the impact of Aldrich humic acid (HA) on the removal of zinc,
nickel and chromium in a Fe
0
system. In the batch test, the removal rate of Zn and Ni (5 mg l
-
1
) was lower in the presence of HA (20 mg l
-1
) than in the absence of HA. The column system


shows similar results, as a significant breakthrough of Ni and Zn occurred in the column
system fed with HA. The decrease removal of zinc and nickel was presumably due to the
formation of metal-humate complexes and binding of the adsorptive surface site. In both
systems, chromate was not affected significantly.

Laxen (1985) found that the adsorption of humic substances reduces the complexation of
Cd
2+
, Cu
2+
and Ni
2+
onto the ferric hydroxide surface. A ligand exchange has often been
described as an important mechanism between NOM removal in the presence of iron oxides
(Gu et al., 1998). The general reaction scheme can be expressed by the equations (Chi and
Amy, 2004);

Fc0E
2
+
+ S -C00
-
= Fc00C -S + E
2
0 (2.18)
Fc0E + S -C00 = Fc00C -S + 0E
-
(2.19)

where FeOH and FeOH
2
+
represent 1 mole of reactive hydroxyls and water, respectively,
bound to a surface of Fe mineral, and SCOO represents 1 mol of carboxylate groups on
NOM. On the contrary, Davis and Bhatnagar (1995) observed that HA accelerates the Cd
2+

adsorption onto Hematite. Davis and Leckie (1978) demonstrated that organic substances
could either enhance or diminish Cu
2+
adsorption on amorphous iron oxide by neutralizing the
surface charge of the adsorption sites. According to a number of studies, the oxidation of
ferrous to ferric iron can be affected by NOM. The oxidation rate is typically inhibited in the
presence of organic compounds at high DO, ranging from mild to strong effects. Humic
substance generally inhibits the oxidation rate hypothesized because of Fe
2+
ligand-complexes
(Kieber et al., 2005; Liang et al., 1993). However in the case of Fe
2+
oxidation under low DO
partial pressures, Liang et al. (1993) showed the oxidation rate is enhanced. They also found
that the iron hydroxides formed in a NOM containing solution have a predominantly negative
surface charge at natural pH. Besides these effects, NOM is known for having a reductive
property toward several inorganic contaminants (Stevenson, 1994). Gu et al. (2000) observed
that ferric iron hydroxide and CrO
4
2-
were reduced by different fractions of NOM. Davis and
Leckie (1978) gave a short summary stating that three mechanisms may be involved in the
adsorption of organic and metal on iron oxides: (i) binding of NOM to the oxide surface can
block the binding site and can also compete with metal removal (ii) NOM complexes with
metal remaining in the solution prevent the metals from being adsorbed at the oxide surface,
and (iii) the removal of metals is enhanced by the adsorbed NOM at the iron oxide surface.
Taken this into account and the fact that natural roof runoff NOM can vary greatly in
molecular weight, molecular size, organic functional group and other properties due to its
variable composition, it is likely that the impacts of NOM on the iron system are complex and
are governed by several reactions. Yet, a thorough understanding of the processes involved
has not been developed.


2.6 Modeling of the F
0
system

A model capable of simulating the dynamic processes of copper and zinc removal by Fe
0
at
varying runoff conditions is important for the engineering design of the treatment barrier. In
most of the reported works, the surface complexation model coupled with the mass transfer
kinetics has been employed (Table 2.3). Smith (1996; 1998) successfully describes the
removal of Cd
2+
, Pb
2+
and Zn
2+
by short blast iron using a dual mass transport incorporating a
triple layer model. Steiner et al. (2006) simulated copper removal by granular ferric hydroxide
using a Two Region Model, which considered film and pore diffusions as removal
mechanisms. Nikolaidis et al. (2003) and Tyrovola et al. (2007) employed a one dimension
advection-reaction-dispersion equation and considered the surface adsorption/co-precipitation
and the reductive transformation as removal pathways of Arsenic in Fe
0
column. In a more
complex modeling approaches, Wang and Salvage (2005) and Kouznetsova et al. (2007)
added terms describing the dynamic aging of iron which allow the simulation of treatment
processes of U(VI) and chlorinated ethenes, respectively, at a non-equilibrium condition.
Sperlich et al (2005; 2008) employed a constant pattern homogeneous surface diffusion model
(CPHSDM) to model the adsorption of arsenic by granular ferric hydroxide (GFH). In most of
the modeling approaches presented, both the kinetic and isotherm parameters required are
generally described in lump sum form of parameters (Sperlich, 2005) as these may have been
very difficult to determine at each experimental conditions (Smith, 1996; 1998; Sperlich,
2005; 2008). The following text discussed the adsorption equilibrium and adsorption dynamic
relevant to adsorption of metals in aqueous phase. It is aimed to apply these mathematical
models for describing the removal of metals in Fe
0
system. Further detailed discussions of the
model can be found in Sperlich et al (2005; 2008).

Adsorption Equilibrium
The adsorption equilibrium is defined as the amount of adsorbate which can be removed in
the treatment processes. It is the main parameter that is used for adsorption design, which is
described by the steady state conditions which exist between the aqueous-phase concentration
of the adsorbate and its solid-phase concentration. The equilibrium solid-phase concentration
depends on the properties of the adsorbent and the adsorbate, the solvent, temperature, and the
equilibrium aqueous-phase concentration.

The most common mathematical model used for describing the adsorption equilibrium are the
Freundlich (1906) and the Langmuir models (1918). The Langmuir model assumes a
homogeneous structure of an adsorption surface with a limited number of sorption, thus, its
application to a heterogeneous nature of adsorbents is limited. The Freundlich isotherm
describes adsorption equilibrium in terms of an exponential relation between the sorbed and
dissolved molecules. The shape of the Langmuir isotherm is a gradual positive curve that
flattens to a constant value whereas the Freundlich isotherm has a curved shape representing
in an exponential form. In contrast to the Langmuir isotherm which is derived from a
theoretical basis, the Freundlich isotherm has been developed empirically by fitting the curve
with experimental data. The Freundlich isotherm can be suggested as the following;


q(c) = K
P
c
n
(2.19)

where q(c) is the solid-phase concentration, KF is the Freundlich constant, c is the
concentration in the bulk water phase, and n is the Freundlich exponent. The parameters K
F

and n can be determined from an empirical correlation of batch adsorption experiments by
plotting logarithmic regression of the data and the linearized. The Freundlich components can
be calculated according to the following equation;

log q = log K
P
+ n log c (2.20)

The index n indicates whether the adsorption is favorable ( ) 1 < n or unfavorable 1) > (n . For a
given n, the value of K
F
directly reflects the adsorption capacity.

It is also possible to describe the transport of metals in the reactive barrier in terms of a
retardation factor,
f
R , of contaminants which is referred to the ratio between the rate of
water movement and rate of contaminant movement. This can be calculated by the equation;

R
]
= 1 +
p
0
n K
P
C
n-1
(2.21)

Where is bulk density and 0 is the water content

Adsorption dynamic
In order to describe the adsorption processes and effectively design the dynamic processes of
an adsorber, it is essential to know adsorption kinetics. The adsorption kinetics is expressed in
the breakthrough curves. The breakthrough of the column depends on the properties of the
adsorbate and adsorbent as well as the hydrodynamics of the system involving three steps of
transport, namely:

a) Film diffusion
Film diffusion describes mass transport of adsorbate molecules from the bulk to the outer
surface of the adsorbent. A linear concentration gradient is generally used to describe the
mass transfer resistance since the thickness of the boundary layer is not known. The film
diffusion coefficient can be determined by a slope of the plot of the metal concentration
against time and the linear curves in a semi-logarithmic concentration-time curve can be
obtained. The following equation describes the mechanism (Mertz et al., 1999):

n
P,
=
P,
(c
-c
) (2.22)

in which n
P,
is the mass transfer rate per unit of surface area,
P,
is a film diffusion
coefficient, c
is the adsorbate concentration in the bulk and c
is the adsorbate concentration

on the external surface of the adsorbent.

The film diffusion coefficient can also be estimated using the Gnielinski correlation as
following (Sontheimer et al., 1988);


k
]
=
|1+1.5(1-s)]
d
p
|2 +u.644Rc
1 2
Sc
1 3
] (2.23)

in which J
p
is the adsorbent particle diameter (m), D is the adsorbate gas or liquid phase
diffusivity (m
2
s
-1
) and e is the bed void fraction. Re and Sc refer to the Reynolds and Schmidt
number.

b) Intraparticle diffusion
The next step for the mass transfer interior of the adsorbent particles is the intraparticle
diffusion steps involving two kinetic mechanism; surface diffusion and pore diffusion. The
former describes the transport of adsorbate along the surface wall where the travel within the
liquid-filled pores in the particles is described as pore diffusion. The surface diffusion model
presumes a spherical particle of homogenous structure, although the adsorbent grains in
reality are very heterogeneous in structure. The mass transfer of surface diffusion can be
described by Ficks first law as follows;

n
S,
= p
P
S,
q
i
(2.24)

where n
S,
is the pore diffusion flux; p
P
is the particle density of adsorbent grain; q
is the
solid-phase concentration in the pores;
S,
is the pore diffusion coefficient. Analogous to
surface diffusion, the following equation describes the pore diffusion flux

n
P,
= e
P
P,
C
P,i
(2.25)

in which n
P,
is the pore diffusion flux; C
P,
is the adsorbate concentration in the pores;
P,
is
the pore diffusion coefficient; for the pore diffusion model, it is assumed that the pore fluid
concentration, C
P,
, is in equilibrium with the local solid-phase concentration.

The determination of the mass transfer coefficient can be done both by means of a batch
experiment and using empirical equations. The pore diffusion coefficient is related to the
adsorbate diffusitiy and the intraparicle physical properties as shown in the following
equation (Sontheimer et al. 1988);

p
=
:
p
(2.26)

in which D is the adsorbate gas or liquid phase diffusivity;
p
is the adsorbent tortuosity (-);
p
is the pore diffusion coefficient.

Generally, it is assumed that surface diffusion dominated over pore diffusion especially in
lower concentration range. The surface diffusion flux is also usually many times greater than
pore diffusion flux for the strongly adsorbed species. For natural waters, where both weak and
strong species exist, modeling that takes pore and surface mechanisms into account may
better describe the adsorption processes. Which transport mechanism controls the adsorption


processes can be determined using the surface to pore diffusion flux ratio (SPDRF) as
described by Sontheimer et al. (1988).

s
=
s
p
C
0
:
p
p
c
q
0
SPFR (2.27)

in which C
0
is the initial inlet concentration; is the adsorbate gas or liquid phase diffusivity;
S
is the surface diffusion coefficient; q
0
is the solid phase concentration in equilibrium with
C
0
for a sigle-solute equilibrium; e
p
is the void fraction (porosity) of the adsorbent (-); p
A
is
the apparent adsorbent density;
p
is the adsorbent tortuosity (-).


Table 2.6 Experimental and modeling approaches of inorganic removal by Fe
0
and iron oxides
Source Substance Conc.
[mg L
-1
]
NOM
[mg L
-1
]
pH
[-]
Flow
or EBCT
KF
[L
n
mg mg
-n
g
-1
]
1/n
[-]
Film diffusion
Coefficient k
f

[cm s
-1
]
Surface diffusivity, DS
(cm
2
s
-1
)
Modeling
Smith (1998) Pb
2+
, Cd
2+
, Zn
2+
5.0 - 4.0-5.5 70 h 18.5-28.9 0.66-0.71 1.5-5.0 1.0 x 10
12
TLM
Smith (2000) Pb
2+
10 - 4.0 0.35 cm s
-1
1.5 x 10
-3
1.5 x 10
-10
TLM
Nikolaidis et al. (2003) AsO
4
3+
0.35 8-11 6-6.3 6.4 cm min
-1
K
d
4,300 l kg
-1
ADT
Steiner (2006) Cu
2+
TRM
Sperlich (2005) AsO
4
3+
4.0-4.8 - 7.0 0.9-23 ml h
-1
63.2 0.17 1.11 x 10
-3
2.64 x 10
-9
HSDM
Tyrovola et al. (2007) AsO
4
3+
2.0 - 7-8 10 min K
d
160-210 l kd
-1
- 4.5 x 10
-4
-3.3 x 10
-3
HM1D, TLM
Wang and Salvage (2005) UO
2
2+
6-1134 - 5-6 1-72 h EQM*
Kouznetsova et al. (2007) TCE 0.9-245 - 0.5 m d
-1
111 0,59 3.57 x 10
-3
1.3 x 10
-10
RSNM
Genz et al. (2008) NOM 4.2-8.7 7.8 1.4-3.0 m h
-1
8-12 0.6-0.9 1.8-3.7 1.5 x 10
-10
m
2
HSDM
Yu et al. (2007) BTA 10 - 6-7 18-20 mL h
-1
0.79-2.0 0.67 - - ADT
Rangsivek and Jekel (prep.) Cu
2+
5 0-70 5-6 5 min 20-100 0.4-0.6 9.0 1.0 x 10
12
PSDM
Zn
2+
5-7 0-70 5-6 5 min 10 1.0 9.0 3.5 x 10
12
PSDM
TLM:Triple layer model; TRM: Two region model; HSDM: Homogenous surface diffusion model; ADT: Advection-Dispersion Transport; PSDM: Pore and surface diffusion model;
BTA: Benzotriazole; RSNM: Reactor-specific numerical models; EQM: based on equilibrium modeling that considers corrosion of iron, reductive precipitation, adsorption, speciation
using PHREEQC program;

3. Materials and methods

3. Materials and methods

3 Materials and methods

3.1 Batch experiments

3.1.1 Materials

Reagent and preparation
All chemicals are reagent grade (Merck, Germany). Metal stock solutions were prepared
using Cu(NO
3
)
2
.
3H
2
O, Pb(NO
3
)
2
.
3H
2
O and Zn(NO
3
)
2
.
6H
2
O in deionised water (DI).

Stormwater runoff
Model runoff solutions were either prepared in the laboratory using DI with/without
Suwannee river fulvic acid (SRFA, IHSS, USA, for DOC study) and simulated stormwater
runoff (SSWR) or collected in-situ from roof, street and highway catchments. Detailed
characterizations of model waters are shown in table 3.1.

Characterization of Fe
0

Fe
0
was prepared as scrap iron from a steel cylinder (ASTM A284 Steel grade C) using a
sawing machine. The iron contains approximately 98% Fe
0
, 0.24-0.36% C, <0.9% Mn,
<0.04% P, <0.05% S and <0.28% Si (Wt.%). The compositions of ZVI were obtained from an
available online-database (www.matweb.com) which corresponded to EDX examination. Its
particle size ranges between 0.4 and 1.25 mm with a surface area of 0.384 m
2
g
-1
using a
Brunauer Emmett Teller (BET) method (Gemini Analyser, Micromeritics) with argon gas. It
was pre-washed with acetone, dried and kept in an oxygen free environment until used. Visual
inspection, SEM and EDX analysis showed no oxides present on the iron particles (Fig. 3.1).

Table 3.1 Characteristics of source waters for batch investigation
Source* pH Condc.

Alk. Hardness

Cl
-
NO
3
-
SO
4
2-
DOC UVA
254
SUVA Color
(S cm
-1
) (mg L
-1
CaCO
3
) (mg L
-1
) (mg L
-1
) (m
-1
) (m
-1
mg
-1
L) (m
-1
)
SSWR
1
7.40 200 40.0 70 9.76 0.18 17.99 0.99 3.24 3.27 0.20
Lankwitz
2
6.88 82 23.8 45 5.39 11.83 5.30 1.29 3.65 2.83 0.12
TU-SW
3
5.35 80 1.90 15 2.41 3.33 5.50 3.49 11.73 3.36 0.42
SRFA
4
5.01 200 2.50 50 1.41 53.1 0.65 4.66 19.59 4.20 1.01
UFA
5
7.53 128 48.8 65 5.70 0.96 11.94 7.29 29.24 4.01 1.62
Halensee
6
7.42 1400 11.6 250 165 25.6 26.5 33.59 28.4 0.845 1.27
* Pre-filtrated using 1 m glass filter ** samples for Cl
-
, NO3
-
, SO4
2-
, DOC, UV were separated by 0.45 m membrane filter

1. SSWR: Simulated Stormwater Runoff prepared by dilution of 1:5 of Berlin tap water and DI;
2. Lankwitz: Roof and Street runoff from residential area in Berlin-Lankwitz;
3. TU-SW: Bitumen roof nearby the city center area of Berlin-Zoologischer Garten;
4. SRFA: prepared using Suwannee River Fulvic Acid (Cat. no. 1S101F) in DI;
5. UFA: Roof (with extensive greened roofing) and street runoff from residential area in Berlin-Tempelhof
6. Halensee: Runoff from separated stormwater sewer discharging highway runoff to the Halensee lake in Berlin Charlottenburg-Wilmersdorf

3. Materials and methods 32

Fig. 3.1 SEM and EDX mapping of Fe
0
used in this study

Fig. 3.2 pH
PZC
of iron oxides produced by corrosion of iron

Beside Fe
0
, iron oxides or rust and corrode iron produced by oxygen-corrosion of iron in
deionised water and air dried were also used in some experiments for this work. The pH
PZC
of
iron oxides was determined employing two methods, salt addition and acid-based titration,
according to Mullet et al. (1999) and Mustafa et al. (2002), respectively. Identical results
were obtained for both methods (Fig. 3.2). Table 3.2 shows the characteristic of the materials
employed in this work. In a comparasion study, Granular Ferric Hydroxide (GFH) obtained
from the same source studied by Genz et al. (2004) was used.

-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
2 4 6 8 10
C

m
-
2
pH
0,01M NaNO3 Mass Tritration
0,001 M NaNO3 Mass Tritration
0,001 M NaNO3 Salt Addition


3.1.2 Kinetic batch tests

In order to determine the rates and capacities of metal removal by Fe
0
, both kinetic and
equilibrium studies were performed. The kinetic investigation was carried out using a double
wall, gas-tight and thermostat controlled reactor with total volume of 3.4 L (Fig. 3.3). The
metal containing solution (3.0 L) was pre-equilibrated to achieve saturation using humidified
N
2
or air at a constant mixing rate (r), pH and temperature (T). The gas was then turned off.
Unless stated, the equilibrated solution contains 1 mg Cu
2+
L
-1
, 5 mg Zn
2+
L
-1
, pH
i
(initial) 5.0
0.1, ~50-200 S cm
-1
, 20C at 150 rpm. The Cu
2+
/Zn
2+
ratio was chosen as a general
concentration range detected in urban runoff. At the beginning of the experiment, an initial
control sample was taken, after which ZVI (0.5 g L
-1
) was introduced. Periodically, samples
were acquired from the reactor through an auto-sampler. They were filtered using a cellulose
nitrate membrane filter (0.45 m) for analysis of Cu
2+
, Zn
2+
and Fe
2+
. The pH was either
maintained constant using HNO
3
and NaOH or otherwise uncontrolled. The pH, DO and
redox potential (Eh) were measured and recorded by a computer throughout the run. The
usage of NO
3
-
background contributes insignificant interferences due to its much slower
reaction rate compared to that of

Cu
2+
(Meihr et al. 2004). Under various experimental
conditions in this study, the measured conductivity was relatively unchanged.

In the separate comparative experiments for determination of the removal rate of Cu
2+
and
Zn
2+
by ZVI, corroded ZVI and GFH, methods of pulse doses of metals were employed. This
method involved the addition of both metals to the reactor at interval times throughout the run
rather than only at the beginning (e.g., Fig. B1-B3, Appendix B). In this ways, both the kinetic
and total metal uptake load can be determined. The pulsed doses were at 5 mg L
-1
with 4

Fig. 3.3 Kinetic batch reactor

autosampler
out
in
pH, Eh, DO online collection
N
2
/ Air (manual) Computer/Data Logger
Gas out/
Refill
Plexi-glass stirrer
NaOH/HNO
3
T. control
gas tight reactor
Pump
Rotor
iron


pulsed and, thus, total concentration of metal added was 20 mg L
-1
. In some of the
experiments, Pb
2+
was also employed for comparison with Cu
2+
and Zn
2+
(Fig. 4.2).

3.1.3 Equilibrium tests

The equilibrium tests were performed by means of varying ZVI doses (0-1.5 g L
-1
) in a set of
model runoff volumes (50mL). They either contained no initial metal concentration or 2-7 mg
L
-1
of Cu
2+
or Zn
2+
alone, or in combination. pH and conductivity was adjusted initially to 5.0
0.1 and 200 S cm
-1
(except the Halensee sample which has 1400 S cm
-1
) using HNO
3
and
NaOH, respectively. The sets of vials were rotated at ~20C in darkness for 48 h (a pre-
determined time that sufficiently describes the equilibrium condition for stormwater runoff
treatment). Subsequently, samples were filtrated and analyzed for metals, DOC, UV
254
, color
(UV
436
) and pH.

3.2 Column experiments

3.2.1 Materials

Stormwater runoff
The model runoff (TU-SW) used in this work originated from a Bitumen roof with zinc gutter
at Technical University of Berlin (TU-Berlin).

Fe
0

Fe
0
was prepared as scrape iron particles (0.70-1.25 mm) having an average surface area of
0.348 m
2
BET g
-1
(Rangsivek and Jekel, 2005). This Fe
0
particle was mixed with granulated
pumice (PM) and particularly used in the investigation of laboratory column. Under field
investigation, the iron was also prepared in a spiral shape. According to the column
experiment using the same iron mass, both particle and spiral iron exhibited more or less the
same removal rates of metals (Fig. B8, appendix B).

Pumice
Pumice obtained from the Raab group (Germany) was mixed with Fe
0
to enhance the
hydraulic conductivity. It is characterized by a high porosity and has original compositions
consisting of non-fibrous, amorphous sodium-aluminium-silicate without considerable
crystalline silicic acid. As compared with Fe
0
, pumice is considered not reactive toward
metals. According to the provider, major compositions of pumices are 55% SiO
2
and 22%
Al
2
O
3
(Table 3.2, p. 40). Pumice was sieved to 0.40-1.25 mm and, subsequently, washed in a
flowing DI several times. The pumice was dried at 105C and kept at room conditions.


Carbonatic, activated carbon and GFH materials (see method in section 3.4)
For optimization of the Fe
0
barrier, i.e., to enhance the precipitation of metals, a pure calcium
carbonate (CaCO
3
) (hydro-calcit, Akdolit, Germany), a mixed mineral of Ca and Mg
carbonate (CaMg(CO
3
)
2
) (Magdo-Dol, Akdolit, Germany) and a Mg carbonate (MgCO
3
)
(Magnesia 314, Magnesia, Germany) were employed. The characteristics of these materials
are shown in Table 3.2. In addition, granular activated carbon (Chemron, Germay) and GFH
as employed in the study by Genz et al. (2004) were also trialed for this purpose. The media
were sieved to have a similar size to Fe
0
and used with no further treatment.

3.2.2 Laboratory experiments

Table 3.3 (p. 40) summarizes the breakthrough experiments both carried out in the laboratory
and at field conditions and the results of the tests are illustrated and discussed in section 4.2.
The laboratory breakthrough columns tests were conducted as a set of comparative tests,
usually consisting of 4-6 columns in each run. Unless otherwise stated, a column (2.5 cm
diameter, 25 cm length) containing Fe
0
particles (15 g) homogeneously mixed with PM (20 g)
(~10% Fe
0
v/v) was employed as a standard criteria (Col. 2 in Table 3.3). There were two
sampling ports with Teflon sieves at the inlet and the outlet ends of the column. The media
were supported by glass beads (diameter 2 mm) at both ends. To begin a set of breakthrough
experiments (Col.1 in Table 3.3), a pump (Ismatec, Ecoline, Switzerland) drove the runoff up
into the columns with a Tygon tube (diameter 2.79 mm) in an up-flow mode, while the other

pH, T, Cond.
Pump
drain out
metal stock
solution
10% iron/pumice
mixed
glass bead
out
in
Plexi-glass
stirrer
T. control
gas tight
reactor
Rotor
Reservoir
pH, T, Cond.
Pump
N
2
in
N
2
out
f
o
r

D
O
a
d
j
u
s
t
m
e
n
t
f
o
r

D
O
a
d
j
u
s
t
m
e
n
t
sampling
sampling
Thermostat
double wall

Fig. 3.4 Laboratory column experiment (c.f., text)


pump incremently injected the prepared-metal stock solution into the flowing stream (Fig.
3.4) giving a distinctly different characteristic of water quality parameters. In a typical run,
concentrations of copper and zinc in the runoff solution were each set at 5.0 mg L
-1
of Cu
2+

and 5.0-6.5 mg L
-1
of Zn
2+
, which was done by spiking a concentrated stock solution into the
runoff (Merck, Germany). The differences in concentration of copper and zinc were due to
variations in the natural occurrence of the zinc in the source runoff (i.e., 1-2 mg L
-1
)
(Rangsivek and Jekel, 2008). Temperature was controlled at 20C using a Haake thermostat
(Germany).

In the investigation, various water quality parameters were varied as follows: pH 4.2, 5.8 and
6.5; conductivity 80 and 200 S cm
-1
; dissolved oxygen (DO) <1.3 and 8 mg L
-1
; water
temperature 15, 20, 30C; natural organic matter (NOM) 0-61 mg L
-1
and empty bed contact
time (EBCT) of 5 and 20 min. These investigated water quality parameters are within the
range found for typical roof runoffs. In every set of the runs, a controlled column was
integrated. The runoff solutions were filled up daily and thus source variation of stormwater
could not be prevented. The obtained results from the comparative columns are attributed to
differences in the investigating water quality parameter.

To investigate the possibility of up-scaling of the iron barriers and to assess prediction of the
model (c.f., Table 3.2.4 and section 4.2.3), another breakthrough test was carried out. This
employed Plexiglas columns measuring 3.5 cm in diameter and 56 cm in length. The column
had 7 ports along the side wall separated at an eqal distance of 7 cm, allowing the
determination of the removal rate of metals after an elapsed distance along the flow. The
experiment was operated using 10%, 20% and 30% Fe
0
(v/v) columns mixed with pumice at
empty bed contact times (EBCT) of 2.25 min. After an extended period of time, clogging
prevailed and it was necessary to adjust the flow to a lower rate and, thus, only data from the
first two ports of 20%Fe
0
column has been obtained for comparison.

3.2.3 Model prediction

The equilibrium software MINEQL
+
was employed for the evaluation of the mechanisms that
involved the removal of copper and zinc within the column test (Schecher and McAvoy,
2001). To predict the breakthrough curves of the Fe
0
system, a pore surface diffusion model
(PSDM) was employed. This was modeled using a commercial software - AdDesign
(MichiganTech, USA) (Mertz et al., 1999) incorporating the following assumptions: plug-
flow condition in the bed, linear liquid-phase mass transfer, solid phase mass transfer by pore
and surface diffusion, constant hydraulic loading rate and diffusion coefficients, spherical
adsorbent grains and the Freundlich isotherm to describe the adsorption equilibrium.
Furthermore, the model assumes the transport of contaminants to the reactive surface
governed by film diffusion, intra-particle or pore and surface diffusion and internal surface
diffusion. Detailed derivation of the equations can be found in Friedman (1984), Crittenden
(1985a; 1985b) and Sontheimer et al. (1988) and is described in part in Appendix D.


In the modeling procedures, the adsorption equilibrium parameters (K
F
and n) were acquired
from batch investigations previously described in equilibrium batch test. In short, a varying
concentration of Cu
2+
and Zn
2+
(0-7 mg L
-1
) was equilibrated with doses of 0-1.5 g Fe
0
L
-1

using roof, street and highway runoffs. It was rotated in darkness for 48 h while the initial pH
5.0 was allowed to vary over the run. In this way, the resulting isotherm coefficients are
derived from the conditions under a wider range than roof runoff conditions, which can better
describe the breakthrough curves (i.e., cover the variability of the runoff qualities). The
remaining parameters, i.e., the film transfer coefficient k
F
, the surface diffusion coefficient D
S
,
and the pore diffusion coefficients D
P
, were obtained during the calibration process of the
model applying breakthrough results of the column tests. The modeling attempt was to
reasonably capture the curves. Empirically calculated values using, e.g., Gnielinskis
equation and previously reported values from other related works, e.g., from Smith (1998),
Sperlich et al. (2005; 2008) were used as a first approximation. The uptake of NOM in the Fe
0

system was modeled employing kinetic and adsorption equilibrium parameters databases that
was provided with the AdDesign Software.

Subsequently to the calibration of the model, the model was further validated utilizing results
obtained from the large column tests. Thereafter, the model was used to predict the
breakthroughs of copper and zinc at various scenarios, e.g., to test if the increase of EBCT or
iron ratio of 50% Fe
0
(v/v) is sufficient to improve the performance of the system.

3.3 NOM experiments

3.3.1Materials and characterizations

Stormwater runoffs
The runoff solutions originated from two roof types, a copper roof (UDK) and a bitumen roof
with a zinc gutter (TU) (Rangsivek and Jekel, 2005; 2008). The roof areas are located in the
same catchment (approx. 100 m distance), near the city centre of Berlin-Zoologischer Garten.
The runoff samples were collected from the reservoirs installed onsite. TU-FF and TU-SW
are representatives of the TU roof and considered as a first flush volume from the beginning
of Summer and the late Summer of the year 2005, respectively. The Cu-R sample was
collected from the copper roof in Autumn, operationally defined as an event mean volume.
After collection, the runoff samples were pre-filtrated using 0.45 m cellulose nitrate
membrane filters (Sartorius, Germany) and kept at 4C. In addition to that, Suwannee River
natural organic matter (SR-NOM, IHSS, USA) was also employed as a reference standard. It
was diluted in deionised water and then adjusted to have a similar characteristic to the realistic
runoff solutions. The detailed characterization of the source waters employed in this study is
described in Table 4.3.


XAD-Resin fractionations
Isolation of NOM was performed according to the method described by Aiken et al. (1992). In
the fractionation procedures, Amberlite XAD-8 and XAD-4 resin adsorption chromatography
were employed, for which three fractions including hydrophobic, hydrophilic and transphilic
fractions were obtained from the procedures. Because a smaller amount of isolated fraction
was obtained at each step of fractionation, only bulk runoffs and their corresponding
hydrophobic fractions were further characterized for copper complexation and acidity.

Copper complexation
The complexation of copper was carried out by measuring the free copper against the ion
selective electrode (ISE, Neolab, Germany) in an equilibrated solution, of which the Cu
2+

concentration was increased with the stepwise addition of a Cu
2+
stock solution. In the
titration experiment, a 100 mL of NOM solution in 0.015 M KNO
3
background was initially
purged with N
2
for 15 min. A Cu
2+
stock solution was introduced into equilibrating volume
using pipettes. The titrant was added every 5-20 min, which resulted in stable potential
readings (<0.5mV/5min). The pH was kept constant (0.02pH) by additions of 0.1 M HNO
3

and NaOH as necessary using a titrator with a syrine tip (T 80/50 Schott, Germany). The
added volume of titrant and solution for controlling pH was less than 10% of total NOM
volume.

Acidic density of NOM
The carboxylic and phenolic acid density of runoff NOM was performed by a potentiometric
titration. The titration experiment was performed using a 100 mL solution, previously
adjusted to obtain a constant ionic strength of 0.015 M NaCl. It was acidified to pH 2.8 and
purged with N
2
for 30 minutes. Subsequently, a calcium free 0.01 M NaOH solution was
added drop-wise into the solution under a N
2
atmosphere using a titrator with a syrine tip (T
80/50 Schott, Germany) whereas the pH was noted at an exact amount of base. The carboxylic
acidic group is defined by the consumption of NaOH from pH 3 to 8 whereas the phenolic
group required twice the amount of titrant as an estimation (Oden et al., 1993). For most
titrations about 60-70 additions were made in 2-5 min windows between additions to allow
for reactions. The pH usually became stable (<0.05 pH unit) in less than 2 min.

3.3.2 Kinetic batch and column experiments

All runoff solutions were adjusted to have a constant pH and conductivity of 300 S cm
-1

using NaOH and HNO
3
. The concentrations of Cu
2+
and Zn
2+
were brought to 5 mg L
-1
at two
initial pHs, e.g, 2.5 and 5.0. The batch tests were carried out by equilibrating 50 mL of the
runoff solutions with a Fe
0
concentration of 0.5 g L
-1
(0.186 m
2
L
-1
Fe
0
by method of BET
adsorption) for 200 h.

This Fe
0
has been characterized and used in previous investigations
(Rangsivek and Jekel, 2005). The pH was allowed to vary over the course of the run. The
experiment at pH
i
2.5 was not typical for roof runoff conditions; however, at this acidic pH,
evaluation of the mechanisms of NOM impact on the Fe
0
system could be observed in a
clearer manner. In addition, the final pH (pH
f
) of the solution ranged in a typical runoff
condition.


In order to determine the removal rate of metals, water samples of 1 or 2 ml were taken at
regular intervals. The samples were subsequently filtrated using a glass filter. Their volumes
were made up to 10 mL with a 5% HNO
3
solution and kept cool at 4C before analysis of
Cu
2+
, Zn
2+
and Fe
2+
. pH, UV
254
, UV
436
and LC-OCD sample analysis were taken at the
beginning and at the end of experiments. A comparison of samples filtrated using either 0.45
m cellulose or grass filter reveals similar result; therefore, the analyzed metal concentrations
were defined as dissolved substances.

The results of the column investigations to determine of the impact of NOM on the removal
of metals in Fe
0
system were evaluated (results from section 3.2.2). The experimental works
are described briefly; TU roof runoff spiked with heavy metals was fed into the column
containing Fe
0
and pumice. The columns were run in an up-flow mode at either 5 or 20 min
empty bed contact times (EBCT). The concentrations of NOM in the feeding solutions were
varied in order to examine the impact of EBCT and NOM concentrations on the treatment
process, respectively. Water samples were taken before and after passing through the
columns. They were analyzed for the total concentrations of copper, zinc and iron. The
differences in measured concentrations were attributed to the removal performance as a result
of a reaction with Fe
0
. A preliminary study showed that total metals (Cu
T
, Zn
T
and Fe
T
)
remain relatively dissolved in the runoff solution. Hence, bulk concentrations were defined as
the dissolved compounds.

3.4 Optimization and Field experiments

To optimize the Fe
0
columns, carbonate bearing materials were employed into the column
tests both as composited and sequential configurations (experiment Col. 2 and Col. 3 in Table
3.3). In the composited column, the carbonatic media was introduced into the Fe
0
and PM
columns at 10% (v/v) whereas the sequential columns integrate the carbonatic materials in a
separate column (10% media v/v) which was installed prior or after passing the Fe
0
column.
Unless otherwise stated, procedures of laboratory and field optimization experiments followed
those of breakthrough column tests previously described in section 3.2.2.

The onsite optimization studies were carried out both at the UDK and TU roof sites. The Fe
0

treatment barrier received copper and zinc roof runoff at realistic conditions. The large
columns sizes (3.5 cm diameter, 56 cm length) were employed. After successful optimization
of the column test established in the laboratory, it was found that spiral shaped iron showed
the most promising results. Thereafter, the on-site columns (Col. 4) were filled only with


Table 3.2 Chemical and physical paramters of media used in this study
Parameters Unit Fe
0
spiral Fe
0
Pumice Calcite Dolomite Magnesite
Source of materials - TU-Berlin TU-Berlin Raab group, DE Rheinkalk Akdolit, DE Rheinkalk Akdolit, DE Magnesia, DE
Name - ASTM grade C ASTM grade C - Hydro-calcit Magno Dol, Magno Filt Magnesia 314
Name in experiment - ZVI, Fe
0
spiral Fe
0
PM CaCO
3
DML and DMS MAG
Shape - particle spiral particle grain particle particle
Size mm 0.4-1.25 0.4-1.25 0.4-1.25 0.4-1.25 0.4-1.25
BET surface area m
2
g
-1
0.384 ~0.38* - - - -
Filled in column density kg m
-3
2037 204 272 1020 1020 1020
Chemcial data - 98% Fe
0
98% Fe
0
55% SiO
2
97% CaCO
3
46% MgCO
3
48%MgO
3

0.24-0.36% C 0.24-0.36% C Al
2
O
3
22% 2.0% CaO 54%CaCO
3
52.2%CO
2

0.04% Mn, 0.04% Mn 12%K2O+Na
2
O 0.6% MgCO
3

0.05% P 0,05% P Fe
2
O
3
, CaO, 0.2% Fe
2
O
3

0.28% S 0,28% S MgO, TiO
2
0,3% Al
2
O
3
and SiO
2

*based on the kinetic column test of two iron (Fig. B8 in Appendix B)

Table 3.3 Experimetal setting and eperational parameters of the column tests
Pameters Unit Standard columns Composite columns Sequential Columns Sequential Columns
Experiment - Col. 1 Col. 2 Col. 3 Col. 4
Matrix - TU-SW TU-SW and UDK TU-SW TU-SW with addition of Cu and DOC
Components - Cu+Zn Cu + Zn Cu/Zn/Cu+Zn Cu/Zn/Cu+Zn
Flow characteristic - manual natural drain/manual/down flow manual natural/manual
Cu mg L
-1
5.0 4.9-9.1 5.0 2.0
Zn mg L
-1
5.0-7.0 5.0-7.0 5.0-7.0 2 5.0
NOM mg L
-1
5 - 61 5 - 61 5-61 2.9 - 20
pH - 4.8 - 6.2 4.8 - 6.2 4.8 - 6.2 4.8 - 6.2
Conductivity S cm
-1
30-600 30 - 65 30-600 30 - 65
Column configuration - Fe
0
Calcite/Fe
0
, DML/Fe
0
, Mag/Fe
0
CaCO
3
-Fe
0
, DML-Fe
0
, DMS-Fe
0
, Mag-Fe
0
Spiral Fe
0
/aeration/filtration of DML/PM
Columns diameter cm 2.5 2.5 or 3.5 2.5 3.5
Columns length cm 15 15 or 56 15 56
Characteristic of iron - particle particle particle spiral
Mass of media
1
st
column: - 10% Fe
0
/PM (v/v) and 50% 10%, 20%, 30% Fe
0
/PM (v/v) 10% Fe
0
/PM (v/v) 2 x 10% Fe
0
/PM (v/v) ~ 100 g each
2
nd
column: - - - 10% carbonate media/PM (v/v) -
Post treatment system g - - - 800 g DML/800 g PM
Porosity - 80% 70% 80% 90%
Flow rate mL min
-1
15 15 and 108 15 540 and 108
EBCT (in Fe
0
bed) min 5 and some exp. 20 1 and 5 5 1 and 5
Col. 1: Breakthrough experiment, Col. 2: Laboratory and field condition, Col. 3: Sequential column test in the laboratory and Col. 4: Final design of breakthrough column tests


spiral shaped iron without any other media, followed by an aeration step and post-filtration
system of pumice and dolomite, homogenously mixed within the same fixed-bed filtration. A
detailed schematic of the onsite-treatment is illustrated in Fig. 3.5. The samples taken before
and after the were analyzed for copper, zinc and iron. A total concentration was measured. pH
and conductivity were measured following the description in section 3.5, respectively. The
flow of the water was measured by means of a tipping counter which transmits signal data to a
data logger (Amonit, Germany). To elucidate the mechanism involved in the treatment
processes, the solids precipitated after the columns test were analyzed employing SEM and
EDX together with an equilibrium calculation using MINEQL
+
software.

3.5 Analysis

3.5.1 Iron, copper, zinc

The samples were analyzed for copper, zinc and iron using a VARIAN (SpectrAA-300/400,
Australia) or a GBC atomic adsorption spectrophotometer (GBC 906AA, Australia). In the
batch test, the samples were filtered using a cellulose nitrate membrane filter (0.45 m) before
analysis of Cu
2+
, Zn
2+
and Fe
2+
. For the column test, the total samples were collected
andanalysed. A comparison of un-filtrated and filtrated samples (0.45 m, cellulose nitrate

Fig. 3.5 a) The overall view and b) the detailed onsite experiments of Fe
0
system for treatment
of copper and zinc roof runoff. The systems include 3 treatment steps; 1) Fe
0
treatment
columns, 2) aeration step using tipping counter and aeration basin, and 3) post-filtration of
pumice and dolomite


filter, Sartorius, Germany) revealed that the total concentration of copper and zinc results
were similar in both cases and they can be defined as dissolved concentrations.

3.5.2 DOC, NOM

DOC was determined by means of thermal-catalytic oxidation using a high-TOC analyser
(Elementar, Germany). UV
254
and colour were measured using a Lambda 12UV/VIS
spectrophotometer (Perkin-Elmer, Germany). SUVA (UV
254
/DOC) is an index of relative
aromaticity. For advanced DOC fractionation, liquid chromatography-organic carbon
detection (LC-OCD) was employed according to the method described by Huber and Frimmel
(1996). The method divided up DOC fractions into polysaccharides, humic substances,
hydrolysates (building blocks), low molecular acids and amphiphylics.

3.5.3 Others

pH and conductivity were measured using pH and conductivity electrodes (WTW, Germany),
respectively. Phosphate was measured using the FIA-Star 5000 Analyser (FOSS, Denmark).

3.5.4 Solid precipitates

At the end of selected experiments, iron samples were carefully transferred, and dried in an
inert nitrogen-purged glove box for further characterizations employing Scanning Electron
Microscopy (SEM), Energy Dispersive X-Ray analyzer (EDX) and X-Ray Diffraction (XRD).
For some particular cases, an Auger Electron Spectroscopy (AES) was also employed. The
scanning electron microscopy was carried out at 20 kV using a Hitachi S-2700 scanning
electron microscope, equipped with a LaB6 filament and a thin window X-ray PGT detector
for elemental analysis. All the powder samples were coated with a conductive layer of gold
prior to analysis. An X-ray diffractometer was used at 40 kV, 20 mA (CuKa radiation) with a
graphite monochromator in the diffracted beam path. Data were collected using a step width
of 0.02 with a count time of 5 sec per step. The XRD measurements were performed on a
Siemens D5000 (Bruker-AXS, Siemens) in Bragg-Brentano geometry using copper radiation
Kalpha (40 kV, 40 mA). The diffractogram was between 5 and 120 degrees with a step size of
0.02. The sample was used without sample spinning and with the V6 automatic aperture
through a diffraction angle of 6 mm. The AES were carried out under vacuum conditions at a
pressure of 5x10
-9
employing PHI660 (Physical electronics). The energy of the excited
electrons was 3keV.

For the purpose of analysis and to improve the detection in the batch test, solids were either
crushed or another preparation method was used to increase the metal load to iron surface
area. In this case, ZVI was prepared to have a certain geometry, i.e., 2.1x1.05x0.5 cm
3
was
used in the treatment process. The solid precipitates from column tests were powderized
before analysis.

4. Results and discussion batch tests

4 Results and discussion

4.1. Batch experiments

4.1.1 Kinetics and Stoichiometry of metal uptake

An example plot of normalized Cu
2+
and Zn
2+
concentration as a function of elapsed time
obtained from duplicate kinetic tests in SSWR under uncontrolled pH is depicted in Figure
4.1a). The experimental data follows a pseudo-first-order rate law,
t k
t
obs
e C C
=
0
, where C
0

and C
t
are metal ion concentrations at initial and at time t, respectively. By fitting to the rate
equation (6<n
c
<16, where n
c
is the number of correlated data), observed kinetic rate constants
(k
obs
) were calculated and used to interpret the experimental results. Accordingly, the results
in the Figure 4.1a) show that Cu
2+
uptake exhibited approximately twofold higher rates than
uptake of Zn
2+
, in which k
obs
of 0.275 0.019 h
-1
(t
1/2
2.52 0.17 h) and 0.123 0.001 h
-1
(t
1/2
5.62 0.05 h) were obtained, respectively.

Kinetically-determined parameters among the samples obtained in this study, cf. Table 4.2,
yielded comparable results to the findings in acid rock and acid mine drainage matrixes by
Shokes and Mller (1999) and Wilkin and McNeil (2003), owing to high variability of ZVI
reactivity (Meihr et al., 2004). In their studies, t
1/2
values of Cu
2+
uptake were reported of one
to two orders of magnitude smaller than Zn
2+
, varying in the range of 0.6 min - 2.0 h and 121
min - 8.2 h, respectively. Furthermore, it is generally agreed that the uptake of Zn
2+

essentially requires precipitation of iron oxides as sorption sites, whereas, reduction of Cu
2+

by Fe
0
is thermodynamically favoured. In hydrometallurgical processes the cementation of
Cu
2+
is suggested to follow as Cu
2+
+ Fe
0
Cu
0
+ Fe
2+
with k = 1.9 x 10
26
(Nadkarni et al.,
1967; Nadkarni and Wadsworth, 1967; Annamalai and Murr, 1979; Ku and Chen, 1992;

y = 5.167x - 0.171
R
2
= 0.984
0.0
1.0
2.0
3.0
4.0
0.00 0.20 0.40 0.60 0.80
Cu
2+
removed mg L
-1
F
e
2
+

d
i
s
s
o
l
v
e
d

m
g

L
-
1
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Time (hrs)
C
t
/
C
0

(
m
g
/
m
g
)
Cu2+
Zn2+

Fig. 4.1 Plots of a) normalized Cu
2+
and Zn
2+
concentration as a function of time, b)
stoichiometry determination of Cu
2+
cementation process [Cu
i
2+
1 mg L
-1
, Zn
i
2+
5 mg L
-1
, 0.5
g L
-1
ZVI, pH
i
5.0 0.1, pH
f
5.83, DO
i
8.0 mg L
-1
, T 20 1.0 C, 150 rpm in SSWR]. Error
bars represent 95% confidence intervals from duplicate experiments

4. Results and discussion batch tests 44

Lopz et al. 2003; Wilkin and McNeil, 2003). However, previous findings may not allow
sufficient interpretation of the results under stormwater runoff conditions. It is anticipated that
DO, pH, T, IS, DOC as well as other constituents in runoff conditions may exert a significant
influence on the stoichiometry and the role of each treatment process: cementation and
adsorption.

The stoichiometry for the Cu
2+
/Fe
0
redox-couple under stormwater runoff conditions was
experimentally evaluated in the present study. Depicted in Figure 4.1b) are the iron
concentrations that are dissolved into the solution plotted against copper ion that was removed
during the first 6 h run. Determination of stoichiometry factors (Fe
2+
/Cu
2+
) was obtained
from the regression, which demonstrated that about 5.0 mg of iron is required to remove 1 mg
of Cu
2+
(R
2
=0.984). This value was significantly greater than the typically reported one of
0.88 mg mg
-1
(Nadkarni and Wadsworth, 1967) under acidic pH and in the absence of DO,
which is presumably attributable to an excessive iron consumption by DO and accumulated
intermediate products, e.g., Fe
3+
(Nadkarni and Wadsworth, 1967; Bisward and Reid, 1972).
Generally, the factors observed for Fe
2+
/Cu
2+
are in a significantly higher range of 0.3-
12.0, depending highly on the water qualities and experimental conditions, cf. Table 4.2. It is
noteworthy, however, that the stoichiometry factors determined under the present conditions

Fig. 4.2 Results from batch experiments with pulsed dosing Plots of a) Total Cu vs. time, b)
ln(C
t
/C
0
) vs time, c) Total Fe vs. time, and d) Total Fe dissoluted vs. Total Cu removed at Cu
0

5 mg L
-1
(pulse dosing with 5 mg L
-1
), ZVI 0.5 g L
-1
(0.192 m
2
L
-1
), pH 7, DO 0 mg L
-1
, T
20C and 150 rpm. The below figures e) and f) were the results from identical experiment
with Pb
2+
and Cu
2+
dosing, respectively.


may not be solely caused by the cementation process but rather have been influenced by other
processes, i.e., iron hydroxide precipitation and reoxidation by accumulated products. To
verify this, experiments with pulse dosing of Cu
2+
and Pb
2+
were performed under
deoxygenated condition. It was found that the Fe
2+
/Cu
2+
factor of 1.1-1.7 (R
2
=0.93-0.966)
and 0.3 for Pb
2+
(R
2
=0.98) were obtained. These values followed (or are close to) the
stoichiometry of 0.88 and 0.27 mg mg
-1
although they were obtained from the regression of a
24 h run. Thus, to minimize the effects, the stoichiometry factors under stormwater runoff
conditions were calculated, based on the data obtained during an initial period.

4.1.2 Impact of the quality of stormwater runoff

Effect of DO and pH (Fig. 4.3)
Under oxygen limiting conditions (DO<0.5 mg L
-1
) in DI, regardless of pH, Cu
2+

concentrations gradually decreased from its initial values to nearly complete removal at 24 h

0.0
-0.5
-1.0
-1.5
-2.0
-2.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
-3.0
0 5 10 15 20 0 5 10 15 20 25
Cu
2+
- pH 4
Cu
2+
- pH 5
Cu
2+
- pH 6
Cu
2+
- pH 7
Zn
2+
- pH 4
Zn
2+
- pH 5
Zn
2+
- pH 6
Zn
2+
- pH 7
Time (hrs)
l
n
(
C
/
C
0
)

Fig. 4.3 Comparison of experimental results carried out in () limiting and () un-limiting
DO concentration at varying pH 4.0-7.0 (controlled). [Cu
i
2+
1 mg L
-1
, Zn
i
2+
5 mg L
-1
, 0.5 g
L
-1
ZVI, T 20 1.0 C, 150 rpm in DI]


Table 4.1 Experimental conditions and kinetic parameters on the effects of DOC and IS

Cu
2+

Zn
2+

Fe
2+
/ Cu
2+

Parameters Cui
2+
Zni
2+
pHf Ehf OUR kobs t1/2 R
2
kobs t1/2 R
2
m/n R
2

Solution/Unit mg/l mg/l - mV mg/l h h
-1
h - h
-1
h - g/g -
Effect of DOC
DI (DOC 0 mg L
-1
) 5.0 5.0 5.42 246 0.04 0.206 3.36 0.93 0.021 33.8 0.86 0.73 0.95
SSWR
Lankwitz
TU-SW
SRFA
UFA
Halensee
1.0
1.0
1.0
5.0
1.0
1.0
5.0
5.0
5.0
5.0
5.0
5.0
5.83
5.65
5.77
5.52
6.04
5.92
118
175
182
248
122
99
0.12
0.06
0.06
0.07
0.06
0.20
0.275
0.169
0.208
0.209
0.115
0.103
2.52
4.10
3.33
3.32
6.03
6.72
0.99
0.99
0.98
0.99
0.99
0.98
0.123
0.072
0.055
0.021
0.027
0.045
5.64
9.63
12.6
33.0
25.7
15.3
0.97
0.99
0.98
0.95
0.99
0.99
5.00
2.68
2.89
1.56
6.10
0.359
0.98
0.91
0.97
0.98
0.95
0.96
Effect of Ionic Strength
DI 1.0 5.0 5.54 179 0.06 0.330 2.91 0.99 0.058 11.9 0.99 1.17 0.99
1mM NaNO3 1.0 5.0 5.57 89 0.26 0.295 2.35 0.97 0.115 6.01 0.94 2.73 0.96
25 mM NaNO3 1.0 5.0 6.45 61 0.19 0.338 2.05 0.92 0.122 5.68 0.99 2.32 0.99
50 mM NaNO3 1.0 5.0 6.39 78 0.20 0.330 2.10 0.88 0.127 5.54 0.99 2.00 0.77
25mMNaCl 1.0 5.0 5.43 90 0.27 0.297 2.33 0.92 0.123 5.65 0.99 2.21 0.85
25 mM Na2SO4 1.0 5.0 6.10 81 0.21 0.358 1.94 0.85 0.087 7.96 0.99 1.43 0.65
ZVI doses of 0.5 g L
-1
in pHi (initial) 5.0 0.1 in all experiments, i is initial and f is final
OUR was calculated based on the slope of DO vs. time profiles. R
2
based on 6<nc<16, nc is the number of correlated data.
m/n is stoichiometry factor of cementation process (Fe
2+
/Cu
2+
), cf. text.

(>90%). Zn
2+
behaved differently, as within the initial phase no detectable losses were found,
however, when increasing elapsed time and pH some portions of Zn
2+
were removed,
presumably due to the adsorption and co-precipitation with iron oxide generated during
cementation of copper and during anoxic corrosion processes.

A comparison of experiments in oxygen-containing solutions (DO
i
8-9 mg L
-1
) showed that
significantly greater removal rates could be achieved for Zn
2+
, reflecting the strong
involvement of DO. This is according to the fact that corrosion of iron is accelerated in
presence of DO, subsequently promoting the cation adsorption. Calculation shows t
1/2
values
(between controlled pH 4.0 and 7.0)

of copper ion uptake decrease in a narrow range from
5.96-7.15 to 2.00-3.70 h while the values of

61.8-144.4

h

and

12.6-15.2 h were obtained for
Zn
2+
, corresponding to the experiments performed under deoxygenated and oxygenated
solutions, respectively. The experimental results demonstrated a modest dependency of pH on
Cu
2+
and Zn
2+
removal in the controlled pH system. The re-dissolution of Cu
2+
occurred at a
controlled pH of 4 after a 16 h run, which can be explained by the re-oxidation of deposited
copper with the accumulated Fe
3+
(Nadkarni and Wadsworth, 1967; Bisward and Reid, 1972).

In experiments where pH was allowed to vary over the course of runs, the final pH either
increased or decreased, depending on the initial pH setting (Smith, 1996). As an example for
the initial pH 5.0 1 . 0 typically employed in this study, pH rapidly drifted to the maximum
value within 2-3 h. It maintained or gradually decreased to the final range between 5.26 and
5.83, in accordance with the reduction in Eh and DO owing to the corrosion of iron (e.g., in
Table B9-B10, Appendix B). As a result of changes in redox chemistry, precipitation of iron
(oxy)hydroxide becomes favourable. It was, therefore, observed, in general, that the depletion
rates of heavy metals are higher in an uncontrolled pH system. This is especially significant in
a higher ionic strength solution whereby the iron dissolution rate dramatically increases.


Effect of Temperature.
Evaluation of temperature effects on the Cu
2+
and Zn
2+
uptake rate was performed by
changing equilibrated solution temperature between 5-35C. Based on the results obtained, a
higher temperature was observed to significantly increase the iron dissolution rate as well as
metal adsorption. Comparing the results between temperatures 5 and 35C, about a seven-fold
rate increase for Zn
2+
adsorption was evident. The varying temperature did not appear to
affect copper removal rate significantly since t
1/2
values varied in a narrow range of 3.14 to
1.99 h, respectively. The temperature dependency on Cu
2+
and Zn
2+
uptake rates by ZVI was
successfully determined by Arrhenius's equation. The calculated activation energy values of
5.23 kcal mole
-1
were obtained for Cu
2+
and 11.4 kcal mole
-1
for Zn
2+
, indicating that Cu
2+
and
Zn
2+
uptakes are surface or pore diffusion controlled reactions (Annamalai and Murr, 1979).

Effect of DOC
Runoff solutions containing varying concentration and characteristics of DOC exhibit
different impacts on the removal rates of Cu
2+
and Zn
2+
(Table 4.1., see also Figure 4.4). In
the Cu
2+
removal experiments, two distinct trends were observed. A higher rate (t
1/2
~3 h)
corresponded to the lower DOC content of less than 5 mg L
-1
, in which the removal rates at 0-
10 h elapsed time were identical in almost all experiments. However, in the tailing 10-24 h
phases, a non-removable metal fraction (~7-9%) was found in some samples. This perhaps
indicates the attribution of metal-DOC complexation effects. Higher DOC in UFA and
Halensee samples decreased the process rates about twofold (t
1/2
~ 6 h). These rates are
consistent with the values obtained in the system where DO is absent, thus, may lead to the
conclusion that DOC retardation is mainly associated with adsorption, i.e., complexation and
competitive adsorption, but not with the cementation process.

0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Time (h)
Z
n
2
+
t
/
Z
n
2
+
0

(
m
g
/
m
g
)
SSWR Halensee
LW UFA
0.0
0.2
0.4
0.6
0.8
1.0
0 5 10 15 20 25
Time (h)
C
u
2
+
t
/
C
u
2
+
0

(
m
g
/
m
g
)
DI SSWR Halensee UFA
Fig.4.4 Examples of normalized concentration profiles of Cu
2+
and Zn
2+
vs time.
Experiments were conducted with various runoff solutions (c.f., Table 3.1). Representative
profiles are chosen based on the description in the text


A similar trend with much greater impact of DOC on Zn
2+
removal rate was observed. As
compared with SSRW, an increasing DOC content of up to ~ 8 mg L
-1
in UFA hindered the
removal rate of Zn
2+
by 4.6 times. Corresponding to this, higher amounts of dissolved iron
were detected remaining in solution, suggesting the complexation effects of Fe
2+
by DOC,
which may subsequently retard the oxidation of ferrous to ferric iron. The removal rate of
Zn
2+
in the Halensee sample, on the contrary, was much less impacted although it had the
greatest DOC content. This could be explained by a tradeoff between DOC and a higher
amount of sorption sites generated during iron corrosion in a high ionic strength solution
(Furukawa et al. 2002; Kamolpornwijit et al., 2004). Also, the behavior and interactions of
DOC with iron oxides, however, could not be ruled out.

Effect of Ionic Strength.
The results in Table 4.1 indicated that IS has a minimal effect on removal rate of Cu
2+
.
Conversely, the increase in IS resulted in considerably higher Zn
2+
uptake rates. About 75%
of Zn
2+
could be removed in the DI solution, while a higher IS leads to an over 95% Zn
2+

achievable uptake. Types of salt background did not appear to affect the metal uptake rate,
although a slightly lower rate was found when using a Na
2
SO
4
background solution.

As previously discussed, the dependency of IS on the uptake of Zn
2+
can be explained by the
behaviour of iron oxide formation. Higher IS leads to iron oxide precipitation in the bulk
water phases diffusing away from the iron surfaces (Farrell et al., 2000). Consequently, the
iron oxide production takes place continuously.

Effect of Reactant Concentration.
A set of kinetic experiments was performed with varying concentrations of Cu
2+
and Zn
2+
;
absent, individual or in combination (0-10 mg L
-1
). Generally, it is found that copper and zinc
accelerated the iron corrosion and enhanced metal uptake capacities (equilibrium tests).
However, a higher Cu
2+
concentration was found to apparently hinder the Zn
2+
removal rate
(Table 4.1). This could possibly be attributed to the competitive adsorption effects. On the
contrary, higher concentrations of Cu
2+
and Zn
2+
between 1-10 mg L
-1
showed no influence
on the removal of dissolved copper. The result suggests that the enhancement or inhibition
effects depend individually on types and concentrations of each metal.

4.1.3 Equilibrium study

Figure 4.5 illustrates the metal and DOC uptake at equilibrium experimentally determined in
the ZVI system (model runoff solutions) in comparison to the simulated data of the GFH
system (DI solution) (Steiner, 2003; Ludwig, 2004). With the same initial pH setting, the
literature values of metal loaded on GFH reveal a large discrepancy. Such differences might
be due to the differences in experimental boundary scenarios.


Fig. 4.5 Comparison of Cu
2+
, Zn
2+
and DOC equilibrium uptake load by ZVI and GFH. ZVI
system contained various stormwater runoff solutions; ( ) SSWR pH
f
5.14-5.56; ( )
Lankwitz pH
f
5.00-5.67; ( ) TU-SW pH
f
5.00-5.76; ( ) SRFA pH
f
5.00-6.57; ( ) UFA pH
f

5.33-6.51; ( ) Halensee pH
f
5.59-6.70. GFH isotherms were simulated based on the results
obtained from (1) Steiner (2003) and (2) Ludwig (2004) in DI solution


Cu
2+
uptake in the ZVI system demonstrated a slightly lower performance than adsorption on
GFH. This may be caused by complexation effects of Cu
2+
caused by DOC in runoff matrixes,
and the generally lower pH in the data observed in the ZVI system, i.e., 5.0-6.70. However,
up to 100 mg Cu
2+
g
-1
ZVI could be achieved for both media at about 4 mg Cu
2+
L
-1

equilibrium concentration. A comparative kinetic study was carried out with pulse doses of
Cu
2+
, i.e., Cu
2+
was re-introduced into the systems every 6 h for a total 24 h run, showing that
adsorption of Cu
2+
on GFH (controlled pH 6.0 i.e. to enhance the adsorption) takes place
within 5-10 minutes. Metal uptake in the ZVI system (controlled pH 5.0, i.e., worse case for
adsorption) exhibits a slower rate. However, in the overall run, a superior performance in term
of its absolute metal loading capacity was observed for the latter system (Fig. B3 in Appendix
B).

The equilibrium capacities of zinc, associated with ZVI corrosion products, are shown to
agree well with the higher-lower boundaries of adsorption on GFH at pH 7.0, being
significantly higher than adsorption at pH 6.0 (Figure 4.5). About a 10 fold lower loading of
Zn
2+
than Cu
2+
were observed. In several data points, up to 50-100 mg Zn
2+
g
-1
ZVI could be
obtained.

Due to the heterogeneous and complex nature of DOC as well as the interactions between
DOC and metals, large variations of DOC loading were found (Figure 4.5). In the literature,
Teerman and Jekel (1999) reported a load of 30-60 mg DOC g
-1
GFH for fulvic and humic
acid adsorption (Sigma Aldrich, Roth). A similar loading of Suwannee River natural organic
matter (SRNOM) of up to 35 mg DOC g
-1
GFH with a non-absorbable fraction of 0.6 mg
DOC L
-1
has also been reported (Genz et al. 2004). For DOC removal from roof runoff on the
same media, Steiner (2003) found about 2.80 mg DOC g
-1
, with 0.4-0.5 mg L
-1
as a non-
absorbable DOC fraction. All of these outcomes are within the range determined in this study.
In most model runoff samples, about 0.8-1.0 mg DOC L
-1
was non-absorbable DOC fraction,
however, up to approximately 30 mg DOC L
-1
was determined for the Halensee solution.

4.1.4 Characterization of DOC

According to the LC-OCD analysis, stormwater runoff solutions from different catchment
areas revealed their distinct fractional characteristics. A larger molecular weight fulvic acid
with strong UV absorption constitutes as the main DOC fraction of UFA (Figure 4.6a) and
SRFA water samples. The signals follow similar patterns except that UFA contains a slightly
higher content of polysaccharides. TU-SW and Lankwitz samples contained approximately
equivalent fractions of humic substances and building blocks, either due to the low organic
sources in the catchments or the waters may have been biologically degraded. On the other
hand, a significant different characteristic of DOC was observed in the Halensee runoff. In
comparison to other samples, LC-OCD revealed strong signatures of low molecular weight
compounds in relation to a weak signal of humic substances (Figure 4.6b).


Fig. 4.6 LC-OCD and UVD signals of a) UFA and b) HS stormwater runoffs before and
after equilibrium treatment with doses of 0.05 (0.02 for Halensee) and 0.5 g ZVI L
-1

20 40 60 80 100
Time (min)
L
C
-
O
C
D

S
i
g
n
a
l
s
U
V
D

S
i
g
n
a
l
s
UFA Raw
ZVI 0.05 g/L
ZVI 0.5 g/L
20 40 60 80 100
Time (min)
L
C
-
O
C
D

S
i
g
n
a
l
s
U
V
D

S
i
g
n
a
l
s
Halensee
ZVI 0.5 g/L
ZVI 0.02 g/L
a)
b)


Depicted in figures 4.6a) and 4.6b) are the comparisons of LC-OCD and UVD signals of the
UFA and Halensee samples before and after treatment with doses of 0.02-0.5 g ZVI L
-1
. In
both samples, DOC and UV removal increases with higher doses of ZVI. Furthermore, larger
molecular weight fulvic acids were observed to be preferentially removed. This indicates that
the competitive adsorption of dissolved metals and DOC on iron oxide surfaces is impacted
with a certain fraction of DOC. In accordance with a study of Gu et al. (1994), the adsorption
of DOC on iron oxides involves a ligand exchange mechanism between carboxylic and
phenolic groups of humic substances and hydroxylic iron oxide surface groups. A similar
adsorption behavior has also been reported for groundwater DOC and SRNOM on GFH
media (Genz et al. 2004). Further investigation of NOM impact is demonstrated in section
4.4.

4.1.5 Solid phase characterization

Several iron samples after ZVI treatment processes under a variety of conditions were
characterized to understand the metal uptake mechanisms as well as to determine which
parameters govern the long-term treatment processes.

Fe
0
Cu
0
Cu
0
Cu
0 Cu
0
Cu
0
Fe
0 Fe
0
Fe
0
Fe
0
FeOOH
FeOOH
Fe
0
Fe
0
Fe
0
FeOOH
Cu
2
O
Cu
2
O
Cu
2
O
Cu
2
O

Fig. 4.8 X-Ray Diffractograms of ZVI
scales treated under a) pH 4.0 DO < 0.5 mg
L
-1
and b) pH 7.0 DO
i
8-9 mg L
-1
[Cu
0
2+
and
Zn
0
2+
10 mg L
-1
, 20C, 150 rpm, SSWR
Solution]

a)
b)

Fig. 4.7 a) SEM and b) EDX mapping
of ZVI scale treated with solution
[Cu
i
2+
10 mg L
-1
, Zn
i
2+
10 mg L
-1
, pH
7.0, DO
i
8-9 mg L
-1
, 20C, 150 rpm,
SSWR]

a)
b)


In the case of deoxygenated acidic conditions, the iron sample was, by visual inspection,
covered tightly with a black film. The strong intensive copper spectra could be detected by
EDX. This corresponds to X-ray diffractograms, revealing clear, intensive and narrow
reflections of two phases: metallic iron and metallic copper (Figure 4.8a). It is to be assumed
that the reflections of this copper phase are a result of electro-chemical separation of the
copper from the solution. The spectra indicates that copper is in a well-crystallised form.

Oxidation of copper takes place from Cu
2
O (Cuprites) to CuO (Tenorite); Cu
2
O has an
intensive black colour. This layer is, however, probably so thin (where appropriate also X-ray
amorphous) and could not be proved by XRD investigation. To verify this assumption,
additional samples were analysed using AES, confirming the existence of Cu
2
O (Fig. 4.9).
The presence of cuprous oxide layer could probably be attributed to re-oxidation of the
deposited metallic copper at the outer surface by Fe
3+
as previously discussed. Besides, it may
have been formed through a secondary pathway; a reduction of Cu
2+
by ferrous iron bounded
iron oxide surface as reported by Maithreepala and Doong (2004).

When DO was kept unlimited and pH was fixed at 7.0; iron was found predominantly covered
by iron oxide (Figure 4.7a, Fig. B4 in appendix B). Using EDX, an extensive amount of
copper and zinc were found to constitute at a typical iron surface and their existence of copper
and zinc shows a strong association with iron oxides (Fig. 4.10). Further characterisation
determined a relative strong appearance of copper precipitates at several bright spots on oxide
peaks. The preferential area where copper deposition takes place as depicted in Figure 4.7a),
-400
-300
-200
-100
0
100
200
300
0 500 1000
d
N
(
E
)
/
d
E
kinetic energy [eV]
AES-survey, Cu XPS Re2, MP2
149
474
488
507,5
592
650
702
Fe2O3
918,5
Cu
-500
-400
-300
-200
-100
0
100
200
300
400
500
0 500 1000
kinetic energy [eV]
Cu
0
5 mg L
-1
D0 0 mg L
-1
pH 6
42,5 50 57,5
473,5
487,5
507,5
591,5
649,5
701
Fe
2
O
3
837,5 917
Cu
2
O
Fig. 4.9 The Auger Spectra of precipitates after cementation process [Cu
i
2+
5 mg L
-1
, ZVI
3
-1
, T 20 1.0 C, 150 rpm in
SSWR]


Fig. 4.10 Mapping of SEM image to EDX spectra of iron after experiment [Cu
i
2+
10 mg L
-1
,
Zn
i
2+
10 mg L
-1
ZVI 0.192 m
2
L-1(0.5 g Fe
0
L
-1
), pH 7.0 (controlled), T 20 1.0 C, 150 rpm
in SSWR]. (c.f., supplement data in Appendix B)

reflected an electrical cell behaviour of iron, whereby iron dissolved at anodic surfaces and
copper precipitated at cathodic sites (Ku and Chen, 1992).

The separate sample for phase determination was covered by a very fragile reddish-yellow
layer. It was also observed that during batch experiments part of this layer had eroded out into
the bulkwater solution. Diffractograms of the iron sample, Figure 4.8b) show clearly
intensity-weak reflections in which four of these (in ca. 36.6, 42.3; 61.5 and 73.6) could
be isolated and assigned to a very low quantity of Cuprite (Cu
2
O). Another reflection of the
cuprites (at ca. 36.4) is also available, superimposed, however, with two reflections of the
lepidocrocites so that it has no additional supportive evidence. Two of the intensity-weak
reflections (at ca. 27.0 and 46.8) correspond to a fine crystallized form typical for the
lepidocrocite (-FeOOH), a common phase often found as a result of iron corrosion processes
in the presence of dissolved oxygen (Kamolpornwijit et al., 2004). The X-ray detection of
small amounts of cuprous oxides deposits may have, however, been interfered by the existing
lepidocrocite on an iron surface; hence, the quantitative comparison of copper depositions
between both deoxygenated and oxygenated conditions could not be established. Furthermore,
the phase of lepidocrocite (-FeOOH) indicated that the uptake pathway of metals under this
condition were, additionally, due to an adsorption process.


Regarding zinc, Figure 4.7b) illustrates some phase evident at ~1.0 and 8.6 keV. The former
fell within the copper phase and is not likely to be assigned. A much weaker EDX spectra of
zinc implied that uptake of zinc was less associated with ZVI corroding surface. How zinc is
located could not be established by means of SEM and EDX analysis, furthermore, no attempt
has been made to identify its phases. It is, nonetheless, reasonable to suggest in part that zinc
ions form surface complexes with iron oxides present on corroded ZVI surfaces.

4.1.6 Conclusion

The present section demonstrated that ZVI is an efficient media for capturing of heavy metals
(Cu
2+
and Zn
2+
) under typical runoff conditions. The mechanisms of metal uptake differ
considerably, depending on the metal species and experimental conditions. ZVI interacts
through reductive transformation and adsorption/co-precipitation processes. Experimentally
derived parameters suggested that stormwater runoff qualities exert a significant influence on
the removal of Cu
2+
and Zn
2+
. Compared with GFH, ZVI achieved equivalent loads while
providing that it could be obtained at no-cost. A potential disadvantage of Fe
0
in regard to the
immediate and long-term performance could be anticipated, as demonstrated in the following
lists. The summary also includes several important aspects which must be taken into account
for further implication as well as those related to the optimization approaches of ZVI
processes.

(i) A pre-treatment of iron may be required to remove substances covering the iron reactive
surface e.g. oil, when Fe
0
is obtained as solid waste.

(ii) The adsorption of metals on GFH takes place within 5-10 minutes. The cementation rate
of Cu
2+
by ZVI is relatively rapid, but is determined by a specific surface area and
morphology of reactive surfaces (Fe
0
and probably previously deposited Cu
0
) (Strickland and
Lawson, 1971; Annamalai and Murr, 1979). In contrast, Zn
2+
removal is kinetically slow with
a magnitude of several hours, governed through an array of rate-limiting reactions including
iron oxidation, iron (oxy)hydroxide precipitation and metal adsorption/co-precipitation. It is
noteworthy that when the solution is low in DO, T, pH, IS or reactions take place under high
metal and DOC concentrations, the removal rate of Zn
2+
dramatically decreases.

(iii) Whenever favoured, precipitation of copper and iron (oxy)hydroxides has positive
benefits in enhancing the cementation rate (Strickland and Lawson, 1971; Annamalai and
Murr, 1979) or behaving as adsorption sites for metals in runoff solution, respectively. The
precipitates may either localize on the iron surface or dissolute into the solution. This could,
therefore, negatively stimulate the failure of a treatment barrier by passivation of the reactive
Fe
0
surface as well as increase potential of pore blockage. The phases and morphology of
precipitates on the Fe
0
reactive surfaces are the key factors determining the magnitude of rate
retardation. In order to mitigate plugging of treatment barrier, Fe
0
has been typically mixed
with sand. In the following section, Fe
0
mixed with pumice is tested in a flow-through
configuration. The pumice possessing a high porosity might help in increasing hydraulic
loading rate and lower problems of clogging of the ZVIB.


(iv) The impact of dissolved organic carbon is significant. According to the results in this
study, there was sufficient evidence revealing that DOC inhibitions occurred via ligand
complexation, e.g., ligate with Fe
2+
and Cu
2+
, and competitive adsorption. However, DOC
may also directly inhibit Fe
0
reactivity by adsorbing on the solid surface. The electrochemical
modification of iron oxide surfaces by the adsorbed DOC have been extensively discussed,
playing an important role in the processes of metal mobilization in a natural environment. In
this regard, further investigations are required, specifically to determine the role of DOC in
the long-term performance of a ZVI treatment system.

(v) Based on the results of separate experiment, it was shown that Pb
2+
could be effectively
removed by ZVI which shows that the method is also applicable for removal of other
inorganic pollutants in stormwater runoff.

4. Results and discussion column tests 57

4.2. Column experiments with ZVI

The aim of this part was to assess the suitability of Fe
0
for the treatment of copper and zinc
under both close to- and realistic-conditions employing column test. The breakthrough curves
were determined using roof runoffs having varying characteristics. Furthermore, the PSD
adsorption model is applied to simulate the resulting breakthrough curves.

4.2.1 Impacts of water quality

Effect of pH
It was generally noticed that there was a better metal removal of Fe
0
/pumice with respect to
increases of pH (Fig. 4.11). The breakthrough curve of copper at inlet pH (pH
i
) 6.5 showed a
steep increase at the beginning but was subsequently replaced by a declining breakthrough
shape resulting in a final relative concentration (c/c
0
) of copper at approximately 0.5-0.7. The
columns operated at lower pHs showed a different trend; the eluted concentration constantly
increases and reached c/c
0
~ 1.0 at about 4,000 BV. As compared with the column at pH
i
5.8,
the pH
i
4.2 column shows slightly better performance despite the lower inlet pH. In the
treatment of zinc, an initial higher eluted concentration of c/c
0
= 0.3-0.5 was found (c/c
0
= 0.1
for copper). The breakthroughs at pH
i
6.5 show better performance compared with other pHs.
A lag time which indicates the time required for the formation of iron hydroxide for
adsorption sites was observed (i.e., 500-2,000 BV) for experiments at pH
i
4.2 and 5.8.

The higher outlet pHs of effluents as compared with initial pH coincide with an extensive
accumulation of iron hydroxides indicating the role of the adsorption/co-precipitation process.
The effluent pHs from columns of initial pH
i
4.2 (4.170.15), 5.8 (5.850.63) and 6.5
(6.460.16) were 5.720.91, 6.190.26 and 6.310.23, respectively. The outlet pH

(pH
f
) of
0 1000 2000 3000 4000 0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0
2
4
6
8
10
12
0,0
0,2
0,4
0,6
0,8
1,0
1,2
C
/
C
0

o
f

C
u
T
o
r

Z
n
T

[
m
g

m
g
-
1
]
BVT
pH 4.20
pH 5.80
pH 6.50
Zn
T
Fe
T Cu
T
Fig. 4.11 Breakthrough behavior of copper and zinc and the corresponding iron released from
column at varying pHs


column pH
i
4.2 was 6.04-6.22 at 0 1,406 BVT and gradually decreased to 4.84 at the end of
the run. There were intensive oxides deposited for column pH
i
5.8 whereas this was only
minor for pH
i
4.2, i.e., deposition of oxides occurred at the outlet end of the column and a
clear water/Fe
0
mixture with a high outlet concentration of iron, was observed. This suggests
that most of the metal and iron was more stabilized in the dissolved form at a lower pH
i
. On
the contrary, the decreased pH effluent of column pH
i
6.5 indicates the dominant process of
hydroxide precipitation over the production of OH
-
in the corrosion process.

The plot of Cu
2+
removed against dissoluted Fe
2+
shows a linear relationship with a slope of
1.38 (R
2
=0.91) corresponding well with the results of Rangsivek and Jekel (2005) suggesting
a dependency of the removal reaction with Fe
0
dissolution. Assuming an adsorption surface is
partially formed during the run, a repulsive force between iron oxide and cations would
prevent the adsorption reaction to take place at a lower pH than pH
PZC
, i.e., ~ 7.2 (Rangsivek
and Jekel, 2008). It is therefore reasonable that the observable removal of copper were partly
due to the reductive transformation process. The reductive precipitation has been believed
taking place through a direct reaction with Fe
0
surface; however, recently it was demonstrated
that the indirect reduction with dissolved and structured Fe
2+
in vicinity of Fe
0
surface plays a
major role in this process (Noubactep, 2008a; 2008b; Ghauch et al., accepted). In this study,
the pHs of the runoff fell below the range for chemical precipitation to occur (Rangsivek and
Jekel, 2008).

Effect of DO and conductivity
The breakthrough results of the columns operated at different DO levels demonstrated a
contrast in impact of the two metals (Fig. 4.12). The depletion rate of copper was hindered

-1
) and
-1
) conditions [Cu
2+
5.0 mg L
-1
, Zn
2+
5.0-6.5 mg L
-1
, pH
i

5.850.63 and pH
f
6.20.25 with 5 min EBCT,10% Fe
0

0 1000 2000 3000 4000 0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0
2
4
6
8
10
12
0,0
0,2
0,4
0,6
0,8
1,0
1,2
F
e
T
[
m
g

L
-
1
]
C
/
C
0

o
f

C
u
T
o
r

Z
n
T
[
m
g

m
g
-
1
]
BVT
DO 8 mg/L
DO 1.3 mg/L
Zn
T
Fe
T
Cu
T


with increases of DO, in contrast to the removal for zinc which was enhanced. The latter was
likely due to the higher availability of corrosion products in the presence of DO. A lag period
for the formation of iron oxides of about 1,000 BV was observed in concomitant with
decreasing iron at the outlet end, which could not be clearly seen for a low DO column. For
both columns, a drift of 0.32-0.34 pH units was observed and the decreased outlet
concentration of iron suggests that iron were mostly precipitated and deposited within the
columns. This formation of ferric hydroxides presumably resulted in a lower Fe
2+

concentration within the barrier which might contribute to a hindering effect for reductive
transformation of Cu
2+
.

0 1000 2000 3000 4000 0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0
2
4
6
8
10
12
0,0
0,2
0,4
0,6
0,8
1,0
1,2
F
e
T
[
m
g

L
-
1
]
C
/
C
0
o
f

C
u
T
o
r

Z
n
T
[
m
g

m
g
-
1
]
BVT
Cond. 80 uS/cm
Cond. 200 uS/cm
Zn
T
Fe
T
Cu
T

A similar effect of the conductivity to DO on the breakthroughs was observed which means
that the same inhibition pathways are involved. In the batch test, Rangsivek and Jekel (2005)
found the removal of copper and zinc increases with higher ionic strength of water, which was
attributed to a higher and spontaneous corrosion of iron. Here, the increasing removal of zinc
with respect to IS was observed only in the beginning of breakthrough study (<2500 BV), but
for all other periods similar removal ability for copper and zinc in both columns were
generally seen (Fig. 4.13). The inconsistent results between the batch and column tests might
be due to differences in system characteristics. As compared with the batch test, flow-through
configurations are characterized by a significantly higher flux of metals and dissolved oxygen
per unit mass of Fe
0
, and as IS increases, this can yield much higher iron oxide precipitates
within the columns at increasing iron dissolution rate. Thus, less iron is present in the
dissolved form causing lower activities of reductive precipitation. In addition, the
accumulated iron oxides in the column might hinder corrosion rate of iron though surface
passivation.

concentration of iron from the columns experiments at conductivities 80 and 200 S cm
-1

conditions [Cu
2+
5.0 mg L
-1
, Zn
2+
5.0-6.5 mg L
-1
, pH
i
5.850.63 and pH
f
6.140.21, 5 min
EBCT,10% Fe
0


Effect of temperature, NOM and EBCT
The impact of temperature on the removal of metals could not be verified when using runoff
water collected in the summer period (Fig. 4.14). An immediate breakthrough occurs after
only a few hundred of BVT, which was followed by a relatively stable breakthrough shape
with no clear distinction of temperature impact. The most plausible reason for this result
arises from the possession of a highly turbid and strong concentration of NOM in the summer
water (see discussion below). For analysis of the temperature impact, DI was substituted for
runoff water. The modeled water was prepared to resemble the characteristics of runoff water,
with the exception of organic matter that remained zero. According to this, temperature
increases from 15 to 30 C resulted in an increase of 30% metal removal. It is also important
to note that a higher elute iron concentration observed in the first period suggests that iron is
more mobile in NOM solution.

Further attempts to elucidate the impact of NOM on Fe
0
system were tested using runoff of
different NOM concentrations. The retention time of the runoff in the columns was set to 20
min to facilitate the examination. In addition, the effect of EBCT was to be investigated (c.f.,
Fig. 4.26). In the beginning of the test, runoff with ~9 mg NOM L
-1
was fed into the columns
and approximately 100% and 55% removal of copper and zinc were achieved, respectively.
The iron concentration which rationally indicates the activities of iron corrosion - was in a
range of 2 mg L
-1
and then rose to 5 mg L
-1
indicating an extensive corrosion of iron (at 0-
1,000 BV). After about 1,000 BV, the column was fed with 25 mg NOM L
-1
runoff solution;
this resulted in a substantially eluted concentration of metals occurring in concomitance with
the reduced concentration of iron. The lower removal performance of metals in presence of
NOM could be attributed to several reasons; i.e., a metal-ligand complex and a competitive

0 1300 2600 3900 5200 0 1300 2600 3900 5200 0 1300 2600 3900 5200 6500
0
2
4
6
8
10
12
0.0
0.2
0.4
0.6
0.8
1.0
1.2
F
e
T
[
m
g

L
-
1
]
C
/
C
0

o
f

C
u
T
o
r

Z
n
T

[
m
g

m
g
-
1
]
BVT
T 15C
T 20C
T 30C
Zn
T
Fe
T
Cu
T
Fig. 4.14 Breakthrough behavior of copper and zinc and the corresponding iron released from
column at varying temperatures (Lines divided between runoff and DI employed in the
experiment, c.f., text)


adsorption, or this may also cause a modification of the chemical property of the Fe
0
by
NOM coverage. A detailed investigation of NOM impacts on metal removal in Fe
0
is
discussed in section 4.3. When compared with the 5 min, a 20 min contact time demonstrates
a 60% and 50% better immobilization of copper and zinc, respectively.

4.2.2 Breakthrough behavior and process identification

Despite the high variability of runoff qualities, the breakthrough of copper and zinc from
different columns follows a similar behavior (Fig. 4.11-4.13, exclude those at high NOM
conditions of Fig. 4.14). The breakthroughs of copper were close to a sigmoidal shape. The
rapid exhaustion of zinc was similar to that in a high NOM solution indicating a low metal
retention capacity. In comparison to other related works, Smith (1996; 1998) reported the
removal of zinc, cadmium, and lead (10 mg L
-1
each at pH 4.0-5.5) by short-blast iron
followed a sigmoidal breakthrough shape and the breakthrough occurred at less than 1,000-
4,000 BV. Karschunke and Jekel (2002) reported the breakthrough of total and dissolved
arsenic removal in iron corrosion system occurred at about 5,000 and 10,000 BV using
simulated groundwater, respectively, but when testing with actual groundwater, immediate
breakthrough was observed. In contrast, Ludwig and Jekel (2007) observed that a steady state
between the uptake of contaminants (copper and zinc in roof runoff system) and the
production of adsorption sites exist where exhaustion could not be observed over 6,000 BVT
In their cases, a lower metal concentration of 2 mg L
-1
with 11 min EBCT was employed.
Here, probably the employed EBCT is too short to form such an equilibrium.

Another possible reason that could explain the observable high removal of metals in the
studies of Karschunke and Jekel (2002) and Ludwig and Jekel (2007) is the physical
characteristic of the system. In their study, spiral shaped iron without sand or pumice was
employed. With higher porosity within the ZVIB, a greater hydraulic gradient at the liquid-
solid interface is achieved leading to rust particles freely being removed, hence, less
passivation - both as the protective film on Fe
0
surface or within the pore of the barrier. In the
batch investigation with greater mixing intensity, iron oxides precipitate away from the Fe
0

surface resulting in spontaneous reactions (Rangsivek and Jekel, 2005). The important role of
mixing on the iron surface passivation by iron oxides and on the removal process of
contaminants by Fe
0
has been previously addressed (Ghauch et al., accepted).

To clarify whether the equilibrium of metal removal and corrosion of iron may exist and how
different iron shapes influence the breakthrough curves, additional breakthrough column tests
were carried out. For this, breakthroughs test using iron from Ludwig (2007)s studies was
compared with that of spiral shape and the particulate iron employed in this work.
Furthermore, the typically employed Fe
0
/PM column was tested using runoff water collected
from other sources (UFA sample). This water has a DOC content of about 5.9 mg L
-1
and pH
of 6.2. Finally, the result from breakthrough column filled with 50%Fe
0
(v/v) were discussed.


Based on the results, it was found that the equilibrium of metal removal may exist at various
conditions and this resulted in similar schematic of breakthrough developments as observed
by Ludwig (2007). In comparison, the spiral of iron from Ludwig (2007) exhibited a rapid
breakthrough and reached C/C
0
of approx. 0.7 for copper and 0.8 for zinc, respectively
(Fig.B8). A leveling off characteristic with continual flat curves for both metals could be seen.
This was also observed for spiral shape iron in this study, however, at much higher removal
rate likely due to higher surface area of iron. In contrast, the breakthrough curves of the
particle iron, which exhibited a similar rate to the spiral iron during 0-2,000 BVT shows a
trend to much higher outlet concentrations. This phenomenon might be caused by an
accumulation of iron corrosion products that may inhibit the transfer of contaminant to the
metal uptake sites. However, the reduction in removal of metals could not be observed in
UFA runoff (Fig. B10) and in the column filled with 50%Fe
0
(v/v) using TU runoff collected
from winter period and with the same Fe
0
media (Fig. B9). The latter might be due to lower
potential of precipitate formation in a rapid flow rate condition, i.e., iron concentration is
diluted by the incoming metal fluxes and thus lower oxides formation at the iron-water
interface. The observed leveling off schematic of the UFA breakthrough curves is likely a
result of a higher pH that favors an adsorption process. The impact of NOM on the treatment
efficiency should not be ruled out.

Depicted in Fig. 4.15-4.16 are the characterization results of SEM, EDX and XRD analyses of
iron samples taken from the column experiments. The iron samples taken from the columns
operated within a deoxygenated solution are covered tightly with a black film and resulted in
little or no reddish iron oxides deposition. A nodular shape precipitate corresponding to X-ray
diffractograms revealing an observed metallic copper phase (Figure 4.15(c)-4.16). The
presence of copper phases can be attributed to an electro-chemical separation of the copper
from the solution which agreed well with previous results of the batch studies. It is expected
that the resulting copper precipitates were due to reactions with Fe
0
surface and/or Fe
2+
(in a
structured or dissipated form).

For other cases, in contrast to DO limiting conditions, extensive iron precipitates, copper, zinc
and other phases such as oxygen, Al, Si, and K were found. These components distributed
over the Fe
0
/pumice mixtures when compared with the virgin pumice (Fig. 4.15b). The X-ray
detection reveals deposition of lepidocrocite (-FeOOH), hemathite (-FeOOH) and goetheit
(-FeOOH), implying that uptake of copper and zinc were associated with the adsorption/co-
precipitation processes and that aging of iron oxides had taken place.

The evidences of adsorption/co-precipitation processes were also observed by NOM removal
(Fig. 4.17). The adsorption of NOM was highest in pH
i
6.5 columns followed by pH
i
4.2,
which is due to an extensive accumulation of corrosion products induced by increases of pH
i

and a higher affinity of NOM adsorption at low pH, respectively. The lowest NOM adsorption
at pH
i
5.8 is likely due to the competitive adsorption between metal and NOM. The
adsorption of NOM was also observed for the columns in deoxygenated or/and higher
conductivity conditions.


a)
b)
c)
d)

Fig. 4.15 (a) SEM and (b) EDX of of media obtained from typical column tests as
compared with the original pumice, (c) The presence of nodular metallic copper on iron
surface in DO-limiting solution and (d) the corresponding EDX diagram, c.f. text

Fig. 4.16 XRD diagram of media obtained from typical column test


Fig. 4.17 LC-OCD diagrams of the inlet and outlet water samples obtained from different
columns

4.2.3 Model prediction

Results of model calibration
Three sets of modeling parameters are required to sufficiently describe the breakthroughs of
copper and zinc at different conditions (data from experiments using low NOM runoff
excluded that of Fig. 4.14). The model employed K
F
(and n) of 10-40 (0.6) for copper and 5-
20(1.0) for zinc, respectively, and this yielded an average k
F
of 0.4 x 10
-3
cm s
-1
for copper
and k
F
= 0.2 x 10
-3
cm s
-1
for zinc with higher and lower bound values demonstrated in Table
4.2. In the modeling of breakthroughs at increased NOM concentration, it was necessary to

30 40 50 60 70 80 90 100
Time (min)
O
C
D
Inlet
pH 4.2
pH 6.5
pH 5.8
DO < 1.3 mg/L
IS


Table 4.2 Operational and input parameters for modelling of the breakthrough curves
Parameters Cu
i
2+
Zn
i
2+
NOM D
i
L Fe
0
EBCT Cu
2+
Zn
2+

K
F
1/n k
f
K
F
1/n k
f

Conditions/Unit [mg l
-1
] [mg l
-1
] [mg l
-1
] [cm] [cm] [g] [ min] [x10
-3
] [x10
-3
]
Varying runoffs
a

Lower bound 5.0 6.5 2.0 2.5 15 15 4.9 40 0.6 1.5 20 1.0 0.35
Average
Upper bound
Effect of NOM
b

DI
NOM 3-10 mg/l
NOM 5-61 mg/l
5.0
5.0

5.0
5.0
5.0
6.5
6.5

6.5
6.5
6.5
2.0
2.0

0
4-9
9-70
2.5
2.5

2.5
2.5
2.5
15
15

15
15
15
15
15

15
15
15
4.9
4.9

4.9
4.9
4.9
20
10

100
20
20
0.6
0.6

0.4
0.6
0.6
0.4
0.5

0.9
0.4
0.2
10
5

60
10
5
1.0
1.0

0.2
1.0
1.0
0.20
0.09

0.13
0.20
0.10
Effect of EBCT
c

NOM 9 mg/l 5.0 6.5 9.0 2.5 15 15 20.0 20 0.6 0.9 10 1.0 0.08
NOM 25 mg/l 5.0 6.5 25.0 2.5 15 15 20.0 20 0.6 0.07 10 0.6 0.025
Large column
d
5.0 6.5 5.43 3.5 56 208 2.3 20 0.6 1.5 20 1.0 0.35

modify the kinetic coefficient to: k
F
0.2 x 10
-3
cm s
-1
for copper and k
F
= 0.1 x 10
-3
cm s
-1
for
zinc (Fig. 4.19). Furthermore, they were found to change with respect to changes of NOM
concentrations. For the calibration of the breakthrough curves, K
F
, n and k
F
were the most
influential parameters (e.g., Fig. 4.20), while varying the D
s
or D
p
did not significantly
improve the results. This indicates that the NOM may have inhibited the removal by surface
coverage leading to decreases in the mass transfer rate of metals. In all cases, the observed
higher kinetic coefficient of copper over zinc was presumably due to a higher mass transfer
rate reaction. The kinetic coefficients obtained from the calibration procedures in this work
are comparable to values obtained in the literature, i.e., an order of magnitude difference.
Smith (1998) reported k
f
= 1.5-5.0 x 10
-3
cm s
-1
and D
s
= 1.0-15 x 10
-11
cm
2
s
-1
for surface
adsorption of Pb
2+
, Zn
2+
and Cd
2+
onto short blast media. Sperlich et al. (2005) employed k
f
=
1.11 x 10
-3
cm s
-1
and D
s
= 2.64 x 10
-9
cm
2
s
-1
in the modeling of arsenic adsorption onto
GFH.

Integrating the breakthrough curves, which describes the mass of metal removed, and divided
them with mass of Fe
0
in the barrier yields a metal loading capacity of an average 76 mg g
-1
Fe
0
for copper (with min: 55 and max: 96 mg g
-1
, the lower line is excluded as it under
predicts the curve) and 55 mg g
-1
Fe
0
-1
). The presence of 9-61
mg L
-1
NOM concentration results in a much lower metal uptake load of 42 and 22 mg g
-1
,
respectively. Generally, the obtained coefficients in this study agree well with previous batch
works, as well as values reported in literature. In the Ludwig and Jekel (2007) studies, 50-300
mg g
-1
of copper and zinc uptake in an iron corrosion system were found. A smaller load of
metal uptake observed here as compared to Ludwig and Jekel (2007) might have been due to
a shorter EBCT employed or the physical and chemical characteristics of the system, i.e., the
clogging effect of the treatment barrier when pumice was filled into the columns. Smith
a) Breakthroughs from column experiment in low NOM runoff solution (Fig. 4.18)
b) Breakthroughs from column experiment at varying NOM concentration (Fig. 4.19)
c) Breakthroughs from determination of EBCT impact (Fig. 4.26 in supporting information)
d) Results from large column test (Fig. 4.20)
*For modelling of both copper and zinc, D
s
= 1.0 x 10
-10
cm
2
s
-1
and D
p
= 8.53 x 10
-16
cm
2
s
-1
were employed.


0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0.0
0.2
0.4
0.6
0.8
1.0
1.2
C
/
C
0
[
m
g

m
g
-
1
]
BVT
Upper bound
Average
Lower bound
Zn
T
Cu
T
0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0.0
0.2
0.4
0.6
0.8
1.0
1.2
C
/
C
0
[
m
g

m
g
-
1
]
BVT
NOM 9-70 mg/L
NOM 4-9 mg/L
NOM 0 mg/L
Zn
T
Cu
T
Fig. 4.19 Calibration of the PSDM with breakthroughs of copper and zinc obtained from
column experiments under different NOM concentrations. The obtained kinetic parameters
are demonstrated in Table 4.2

Fig. 4.18 Calibration of PSDM with breakthroughs of copper and zinc employing
equilibrium and kinetic parameters demonstrated in Table 4.2


(1995; 1996) reported an individual uptake load of Pb
2+
, Zn
2+
and Cd
2+
onto short blast media
of 60, 40, 20 mg g
-1
, respectively. Komnitsas et al. (2007) studied copper removal in acidic
pH 2.5-4.5 range and reported a maximum removal load of 13.3 mg g
-1
of iron. An
approximate loading capacity of 40 mg g
-1
was found in the adsorption of arsenate on GFH
(Sperlich et al., 2005). It must be, however, noted that the uptake load of metals per unit mass
of iron may not be representative values for comparasion with other media since iron oxides is
progressively and continually generated in the Fe
0
system, and thus, sorption site is increasing
over time.

Up-scaling and simulation at different scenarios
The PSDM can adequately predict the breakthrough curves in spite of differences in contact
times and column size employed of the large column test (Fig. 4.21a, 4.21b). Lower bound
coefficients were used in the model, whereas the average values over-predicted the results,
which might be due to the low NOM of the employed winter water (Rangsivek and Jekel,
2008).

The model simulation shows that the increase of EBCT to 30 and 40 min from the original 20
min can extend the time that copper begins to elute out of the Fe
0
columns of up to about
2,000 and 3,500 BVT, respectively, and the breakthrough follows a sigmoidal shape. To
improve the treatment efficiency of zinc, at least 40 min EBCT is required. However, it must
be noted that the considered system is the multi-component system where copper, zinc and
NOM present and competitive adsorption can not be prevented. In a system, where zinc is the
only component a better removal ability may be achieved. Fig. 4.21c and 4.21d illustrate
breakthrough results of the model employing a 50%Fe
0
column. The usage of higher iron ratio
shows to significantly improve the treatment performance; copper begins to elute out at about
8,000 BV while about 4,000 BV was observed for zinc. However, obviously the model fails to
predict the lag period of time required for producing adsorption sites for zinc as compared
with the experimental results of 50%Fe
0
, illustrated in Fig. B9 (appendix B).

0,0
0,3
0,6
0,9
1,2
0 2000 4000 6000 8000 10000 12000
C
/C
0
o
f
C
u
2
+
BVT
Model-Average KF
Model-lower KF
Model-higher KF
Experimental Data
0 2000 4000 6000 8000 10000 12000
BVT
Model-average kf
lower kf
higher kf
Experimental Data
0 2000 4000 6000 8000 10000 12000
BVT
Model-average Di
lower Di
higher Di
Experimental Data

Fig.4.20 Sensitivity analysis of the breakthrough curves of copper obtained from a
large column test calibrating with (a) K
F
, (b) k
f
and (c) D
S
(modeling parameters are
according to Table 4.2 for lower bound with varying of 25%)
a) b) c)


0,0
0,3
0,6
0,9
1,2
C
/
C
0
[
m
g

m
g
-
1
]
Cu Experimental data
EBCT 30 min
EBCT 40 min
model Cu
Zn experimental data
EBCT 30 min
model Zn
EBCT 40 min
0,0
0,3
0,6
0,9
1,2
0 2000 4000 6000 8000 10000 12000
C
/
C
0
[
m
g

m
g
-
1
]
BVT
Cu Experimental data
model 20%Fe0
model 50%Fe0
0 2000 4000 6000 8000 10000 12000
BVT
Zn Experimental data
model 20%Fe0
model 50%Fe0

Fig. 4.21 Plots of model prediction of the breakthrough results of a large column test and the
model prediction for assessment of the up-scaling and optimizing of the processes

4.2.4 Conclusion

The aim of this work was to develop a suitable system for the treatment of metal-
contaminated roof runoff. The assessment of the use of Fe
0
for the treatment of copper and
zinc was carried out under close to- and realistic conditions employing various column
configurations. Fe
0
was firstly evaluated in the laboratory column under changing runoff
characteristics. The study demonstrated that the breakthroughs of copper and zinc treated by a
Fe
0
system were either accelerated or hindered under different water characteristics.
Increased pH and temperature generally favor faster retention of metals whereas increased
DO and IS of the solution hinders the removal of copper in the column system. The
contrasting results between the batch and column tests were attributed to differences in the
conditions employed and the physico-chemical characteristics of solid precipitates involved in
the systems.

d
b a
C


(i) The difference in removal pathways brings about a distinguished transport rate and
breakthrough shape of copper and zinc. The uptake of zinc was mainly due to the
adsorption/co-precipitation processes, whose rates are determined by an array of reactions
including iron dissolution, iron precipitation and adsorption/co-precipitation. In contrast to
zinc, copper is mainly reduced to metallic copper. With extensive precipitation of iron oxides,
removal of copper via adsorption/co-precipitation processes could not be ruled out.

(ii) Any water qualities that tend to accelerate the rate of iron corrosion and precipitation of
its oxides is beneficial to the uptake of zinc due to increasing adsorption sites. This is,
however, not necessarily true in the case of copper, where accumulated products within the
columns showed to hinder the reduction rate. Over time, the adsorption may be the only
dominant reaction since reactive iron is fully covered. Among water quality investigated,
NOM holds the most influence leading to an abrupt releasing of metals out of the barrier.

(iii) A set of kinetic and equilibrium parameters was obtained from model calibrations of the
breakthroughs curves. These parameters allow the use of PSDM in designing a Fe
0
treatment
system with a wide range of conditions. When the model is to be employed, accuracy of
prediction depends strongly on the presence of NOM; hence, equilibrium isotherms and
kinetic data should be acquired from the runoff waters in the site of installation, were K
F
and
n are the most prominent parameters. Increasing of Fe
0
mass will improve the performance of
metal removal. It is, however, important to maintain high void area within the column, in
order to sustain sufficient hydraulic conductivity. Furthermore, coefficients required for the
modeling must be determined at specific site and conditions.

(iv) The model simulation suggests that sufficient iron is required to achieve a certain
performance of the treatment barrier, particularly for the adsorption of zinc. On the other
hand, a ratio of iron higher than 10% (v/v) can lead to failure both due to clogging or by
decreasing the mass transfer rate of metals. To achieve maximum performance of metal
removal by Fe
0
, the incorporation of carbonatic materials into the barrier may be promising.

4. Results and discussion - NOM impacts on Fe
0
system 70

4.3. NOM experiments

Among water quality parameters investigated, NOM shows most significant influences on the
removal rates of metals by Fe
0
. In this section, the characteristic of NOM in roof runoff,
mechanisms as well as the impacts of NOM on the Fe
0
treatment system are investigated.The
NOM in runoff waters was characterized using XAD-4/8 adsorption resins, copper
complexation, acidic capacity and liquid chromatography with online carbon detection. Batch
kinetic experiments and flow-through configurations were performed and the results of metal
removal were elucidated taking into account the characteristics of NOM.

4.3.1 Characterization of source water

The quality of TU runoff varied significantly over the 1 year experiment. In winter and spring,
TU runoff demonstrated a generally soft characteristic, low turbidity and ionic strength with
conductivity of about 30-65 S cm
-1
. They are mildly acidic with pH (min-max (mean value))
values of 4.806.26. The concentration of NOM was 2.86-10.33 (5.92) mg DOC L
-1
, whereas
the zinc concentration varied in a range of 1.76-2.30 (1.97) mg Zn
2+
L
-1
. Typically, the runoff
samples collected in winter and spring could be considered a dilute solution.

Contrary to the winter and spring samples, the waters collected during summer and autumn
are more concentrated with organic and inorganic substances. The waters are highly turbid,
smelly and possess much higher conductivity than the winter sample. After a long drought in
the beginning of summer, the dissolved NOM concentration was as high as 300 mg DOC L
-1

with conductivity of 600 S cm
-1
(before dilution to TU-FF, Table 4.3); however, that was a
small rain event. For typical runoff samples taken during summer, the concentration of NOM
is 5.07-61.0 (26.3) mg L
-1
(e.g., TU-SW, Table 4.3). The dissolved zinc concentration can
range up to 6.0 mg L
-1
in the first-flush sample but typically fell into a range of 1.46-3.54
(2.47) mg L
-1
as an event-mean concentration, which indicates that the release of Zn
2+
into
runoff water is more or less equally distributed over the year. The concentration of copper is
insignificant. Furthermore, phosphate is less than 0.1 mg L
-1
.

Table 4.3 Characteristics of source waters for investigation of NOM impacts
Source* pH Condc.

DOC UVA
254
SUVA HPO HPI TPI
(S cm
-1
) (mg L
-1
) (m
-1
) (m
-1
mg
-1
L) % % %
TU-SW 6.0 134 36 104 2.9 65 17 18
TU-FF 5.0 19 9.9 16.5 1.7 8 70 22
CU-R 6.4 148 9.8 29.9 3.1 45 28 27
SR-NOM 5.0 300 4.3 24.1 5.6 93 2 5
1. TU-SW: Bitumen roof runoff with zinc gutter at TU Berlin collected during Autumn
2. TU-FF : 1 to 30 dilution of the first flush of TU-SW after long drought period in the beginning of the Summer
3. CU-R: Copper roof runoff at the university of art in Berlin (UDK)
4. SR-NOM: prepared using Suwannee River NOM in deionized water (DI)

*Prefiltrated using 0.45 m cellulose nitrate filter; HPO hydrophobic fraction; HPI hydrophilic fraction
and TPI transphilic fraction

0
system 71

There was no attempt to analyze the annual quality of runoff from the copper roof. However,
a visual inspection of several samples from various periods show that the quality of runoff
followed a similar tendency of annual changes as that of TU roof owning to the same
catchment area. Based on a few collected samples, the concentration of copper and zinc were
in the range of 4.9-9.1 (7.0) mg L
-1
Cu
2+
, and 0.8-1.0 (0.9) mg L
-1
Zn
2+
, respectively. The
appearance of zinc in copper roof runoff might be because of the influence of vehicles from
adjacent streets or the existing zinc components on the roof.

Hydrophobicity and molecular size
A comparison between LC-OCD chromatograms of the runoff samples shows that NOM from
the same roof source, or catchment, has distinctly different characteristics (Fig.4.22). The TU-
SW and Cu-R are comprised mainly of humic substances and building blocks that have a
strong UV absorption, most likely derived from the decomposition of plants. In addition to
that, a minor content of polysaccharides have also been detected. Fulvic acid is the main part
of SR-NOM and the obtained LC-OCD diagram is similar to its corresponding fulvic fraction
as determined in a previous work (Rangsivek and Jekel, 2005). In all of these runoff samples,
humic substances constitute a significant proportion of the total organic carbon concentration.
Similarly, the surface water NOMs are composed of approximately 60% humic substances
(Ma et al., 2001).

U
V
D

S
i
g
n
a
l
s
TU FF Runoff
XAD-8 Isolate
XAD-8 Effluent
O
C
D

S
i
g
n
a
l
s
TU-SW
XAD-8 Isolate
XAD-8 Effluent
30 40 50 60 70 80 90 100
Retention Time (min)
O
C
D

S
i
g
n
a
l
s
Cu roof Runoff
XAD-8 Isolate
XAD-8 Effluent
30 40 50 60 70 80 90 100
U
V
D

S
i
g
n
a
l
s
SR NOM
XAD-8 Isolated
XAD-8 Effluent

Fig. 4.22 LC-OCD of bulk water NOMs and their corresponding XAD-8 isolated and XAD-8
effluent fractions of a) TU-SW b) TU-FF c) SR-NOM and d) Cu-R runoffs, c.f., Table 4.3

0
system 72

Table 4.4 Estimated contents of carboxylic and phenolic groups
Carboxylic
(mmol g
-1
C)
Phenolic
(mmol g
-1
C)
Total acidity
(mmol g
-1
C)
Overall acidity*
(mol L
-1
)
TU-SW 1.8 0.9 2.7 97
TU-FF 3.8 1.9 5.8 54
SR-NOM 9.1 4.5 13.6 95
Cu-R 6.4 3.2 9.7 58
* The overall acidity was calculated by taking into account the variability of DOC concentration of the raw water samples.

These waters generally exhibit hydrophobic and high aromatic characteristics as indicated by
a high value for SUVA (Table 4.3). In contrast to other samples, TU-FF exhibited primarily
fractions of small size range of low molecular weight acids and neutrals. The water
demonstrates a hydrophilic characteristic (70%). This could be attributed to the fact that
amphiphilics and neutrals possess a low content of organic functional groups (Fig. 4.22; Table
4.3). Corresponding to this, TU-FF has the lowest values of an index of aromaticity compared
to other runoff samples.

Acidity of source water
The results of calculation for the acidic capacity of the NOM samples are shown in Table 4.4
(c.f., Fig. B13). The compositions of carboxylic and phenolic groups decrease in the order
SR-NOM, Cu-R, TU-FF and TU-SW, respectively. TU runoff from the same source reveals
similarly low organic contents (2.72 and 5.76 mmol g
-1
C for TU-SW and TU-FF,
respectively), although they have different total concentrations. The generally low ligand
content in the TU-FF solution could be reasonably attributed to the composition of
hydrophilic compounds. Probably, most of the hydrophobic compounds in TU-SW have low
organic functional groups. However, taking into account the total concentration of dissolved
organic compounds, the TU-SW sample demonstrates the highest overall acidity capacity. In
contrast, the SR-NOM (13.61 mmol g
-1
C) and CU-R (9.65 mmol g
-1
C) consists of relatively
high organic functional groups. The carboxylic and phenolic contents in runoff samples
obtained in the present work are within the same range as the NOM samples from natural
surface water (Lu and Allen, 2002).

Copper complexation capacity
According to the results of titration experiments, the copper complexed by NOM runoffs are
only evident in the TU-SW (Fig. 4.23) and are minimal for others. Under experimental
conditions, i.e., at total 5 mg Cu
2+
L
-1
about 2 mg L
-1
is present as free copper in TU-SW.
Similar results were obtained in the titration of the hydrophobic fractions of the runoff NOMs.
In contrast to other studies of natural waters, however, the complexation capacity of the
runoff NOMs in the present work can be considered relatively low. He et al. (2001) found that
copper remained mostly in a complexed form. MacCarthy et al. (1975) employed Fourier
transform infrared spectroscopy and showed that carboxyl functional groups participate in the
complexation of Cu
2+
. In this work, TU-SW NOM possesses the highest content of total
acidity (Table 4.4). A low complexation capacity obtained in this study might have come
about because the titration experiments were run at a low pH 5.50. In an alkaline pH solution
the NOM-complexation of metals may be significantly greater (Cheng and Allen, 2006). A 10

0
system 73

Fig. 4.23 Copper titration of runoff NOM samples from different sources

fold increase of Cu
2+
and NOM complexation per pH unit was reported by Lu and Allen
(2002).

4.3.2 Batch kinetic study with NOM

In the experiments carried out at initial pH (pH
i
) 2.50 and 5.50, the Cu
2+
depletion rate took
place rapidly and a complete removal was achieved within 18 hours regardless of variation in
the source of roof waters and initial pHs. The first-order rate t
1/2
of copper removal was 2.90-
9.56 h. These results are comparable with that of a previous study (t
1/2
of 2.52-6.72 h at pH
i

5.0 and pH
f
5.42-6.04) using runoff from different surfaces including roof, street and highway
(Rangsivek and Jekel, 2005).

Removal rate of Zn
2+
was typically lower than that of Cu
2+
and showed a strong dependency
on pH
i
. The calculated results in Fig. 4.24a and 4.24c demonstrate that removal of Zn
2+
at pH
i

5.50 (final pH: pH
f
6.92-9.57) proceeded at higher rates than the reactions at pH
i
2.50 (pH
f

y = 1.021x + 0.018
R
2
= 0.998
0
2
4
6
8
10
12
0.00 2.00 4.00 6.00 8.00 10.00 12.00
Total [Cu
2+
] mg L
-1
[
C
u
2
+
]

m
e
a
s
u
r
e
d

b
y

I
S
E
,

m
g

L
-
1

SR NOM
TU SW
TU FF
Cu R
DI

0
system 74

4.24-4.63) for all of the tested runoff solutions, having t
1/2
of 14.1-40.8 h

and 87.7-267 h,
respectively. The higher removal rates at elevated pH values could be explained by a larger
number of iron oxides produced and their possession of negative surface charges that favor
adsorption processes. The t
1/2
of zinc removal is comparable with the previous batch test (t
1/2

of 5.64-33.8 h). The final pH
f
(pH
i
5.50) in this investigation ranged to much higher values,
which can be attributed to a longer batch time and, thus, allowing rate-limiting reactions such
as corrosion of iron to freely take place.

The depletion of NOM is dependent on the pH values, a higher pH is accompanied by a
decreased removal of DOC for all runoff solutions (i.e., compared between pH
i
2.50 and 5.50
in Fig. 4.24 and 4.24d) indicating that adsorption is a relevant reaction. In the adsorption
process of NOM, the surface-coordinated H
2
O or OH
-
groups on the hydroxylated iron oxide
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
0 50 100 150 200
Time [h]
C
/
C
0

[
m
g

m
g
-
1
]
TU SW TU FF
SR NOM Cu R
Control (no iron)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 50 100 150 200
Time (h)
C
/
C
0

[
m
g

m
g
-
1
]
TU SW
TU FF
SR NOM
Cu R
Control
0
25
50
75
100
TU-SW TU-FF SR-NOM CU-R
%

R
e
m
o
v
a
l
0.0
7.5
15.0
22.5
30.0
D
O
C

l
o
a
d

[
m
g

g
-
1
]
% Zn Removal
% DOC removal
DOC Load
0
25
50
75
100
TU-SW TU-FF SR-NOM CU-R
%

R
e
m
o
v
a
l
0
7.5
15
22.5
30
D
O
C

l
o
a
d

[
m
g

g
-
1
]
% Zn Removal (50h)
% DOC removal
DOC Load
Fig. 4.24 a) and c) the kinetic removal of Zn
2+
in TU-SW, TUFF, SR-NOM and Cu-R runoff
solutions carried out in the batch test at initial pH
i
2.50 (Control: Fe
0
0 mg L
-1
) (in order pH
f

4.540.13, 4.290.06, 4.260.03 and 4.610.04) and 5.50 (pH
f
6.92, 9.57, 9.29 and 7.95),
respectively, and b) and d) a comparison of its corresponding DOC removal in correlation
with removal of Zn
2+
[Cu
i
2+
5.0 mg L
-1
, Zn
2+
5.0 mg L
-1
, 0.5 g Fe
0
L
-1
, well mixed and at
room temperature]

a) b)
c) d)

0
system 75

surface (i.e., FeOH
2
+
and FeOH) dominate at low pH are replaced by anionic functional
groups (e.g., caboxyl and hydroxyl group) of NOM (Gu et al., 1994; 1995). The hydrophobic
organic acids generally possess higher affinity toward iron oxide surfaces and, due to this
basis, greater amounts of NOM for TU-SW, CU-R and SR-NOM runoff solutions were
removed than that of TU-FF runoff (Fig. 4.24b and 4.24d). In contrast to other runoff
samples, TU-FF NOM is characterized by a hydrophilic characteristic with low aromaticity. It
consists mainly of a smaller size range with low functional groups (see also in LC-OCD
characterization).

4.3.3 Column study with NOM

A comparison between the breakthrough curves of copper and zinc obtained from the column
experiments at varying organic content clearly shows the adverse effect of NOM on the
removal processes (Fig. 4.25); a higher NOM generally results in lower metal removal
efficiency. In deionised water, the outlet copper concentration gradually increases with bed
volume treated and at the end of run efficiency was still at approx. 60%, whereas, a much
steeper breakthrough was obtained for a solution with 5.90 mg NOM L
-1
. An immediate
breakthrough was even observed for a 26.2 mg NOM L
-1
solution (average concentration) and
yielded only 30% removal efficiency. As compared with copper, removal of zinc by the Fe
0
column appears significantly lower. This is due to the formation of adsorptive sites that are
responsible for the removal of zinc (Rangsivek and Jekel, 2005). A rapid breakthrough of zinc
with initial eluted concentration of c/c
0
= 0.3-0.5 followed by a much declining breakthrough
shape can be attributed to a required lag time for the formation of adsorption sites by iron
corrosion.

Fig. 4.25 A comparison of the breakthrough curves of copper and zinc obtained from the
columns tested at varying NOM concentrations of TU Runoff [System: Cu
i
2+
5.0 mg L
-1
,
Zn
i
2+
5.0 mg L
-1
, NOM 0 mg L
-1
(pH
i
5.64 0.38, pH
f
5.840.58), NOM 2.9-10.3 mg L
-1
(pH
i
5.85 0.59, pH
f
6.140.21) and NOM 5.0-61.0 mg L
-1
(pH
i
5.69 0.34, pH
f
6.00.62),
5 or 20 min EBCT, 10% particle Fe
0
0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000
0.0
0.2
0.4
0.6
0.8
1.0
1.2
BVT
C
/
C
0

[
m
g

m
g
-
1
]
NOM 5.0-61.0 (26.2) mg/L
NOM 2.9-10.3 (5.9) mg/L
NOM 0 mg/L
Zn
T
Cu
T

0
system 76

0.0
0.2
0.4
0.6
0.8
1.0
0 500 1000 1500 2000
BVT
C
/
C
0

[
m
g

m
g
-
1
]
0.0
2.0
4.0
6.0
F
e
T

[
m
g

L
-
1
]
CuT ZnT FeT

Depicted in Fig. 4.26 are the results of a flow-though column that was fed with TU runoff
with different NOM concentrations during continuous operation; the first 1,000 bed volume
(BV) was treated with ~9 mg NOM L
-1
and was subsequently fed with 25 mg NOM L
-1
runoff
solution. The retention time of the runoff in the column was set to 20 min in order to facilitate
the examination and also to investigate the effect of EBCT. In Fig. 4.26, approximately 100%
and 55% removal of copper and zinc were achieved, respectively. The iron concentration
which rationally indicates the activities of iron corrosion - was in a range of 2 mg L
-1
and then
increased to 5 mg L
-1
suggesting a more extensive corrosion of iron (at 0-1,000 BV). After the
fed solution was substituted, both metals were substantially eluted out of the column
occurring in concomitance with the reduced concentration of iron.

4.3.4 Impact of NOM on copper and zinc removal in ZVIB

The impact of NOM on metal removal in the Fe
0
system were elucidated based on the
correlation of the depletion loads of DOC and the removal profile of heavy metals under
different conditions. In the batch test carried out at pH
i
2.50, the removal load of NOM (Fig.
4.24b) is accompanied by the removal of zinc, ranging in order of SR-NOM (~57% Zn
removal, DOC load 5.9 mg g
-1
), TU-FF (58%, 7.4) and Cu-R (83%, 25.9), respectively. At
Fig.4.26 The concentration profiles of copper (c/c
0
), zinc (c/c
0
) and iron (total) from the
breakthrough column that was continually fed with TU runoff having different NOM
concentrations (~9 mg NOM L
-1
before 1000 BV and with 25 mg NOM L
-1
thereafter) [Cu
i
2+

5.0 mg L
-1
, Zn
i
2+
5.0 mg L
-1
, pH
i
5-6, 20 min EBCT, 10% particle Fe
0
(v/v) supported with
pumice]

0
system 77

this low pH (pH
i
2.50 to pH
f
4.24-4.63), the adsorption affinity of metals would have been
very low due to a repulsive force between positively charged iron oxide surfaces and cations
(pH
PZC
7.2). The equilibrium calculation using MINEQL
+
(condition: 2-Layer model using
Hydrous Ferric Oxides 300 m
2
L
-1
, 20C, pH 3.09.0) shows that no zinc could be removed at
any pH below 5.0. Rodda et al. (1993) also reported that zinc removal is insignificant at pH
lower than 6.0 with 0.20 g L
-1
goethite. In addition, Crawford et al. (1993) observed that the
co-precipitation and adsorption of Zn
2+
on iron oxides followed the same lines and were
insignificant at pH below 5. Therefore, one possible reason for such high removal of Zn
2+

might have been due to the previously adsorbed runoff NOM on the iron oxides surface.
Because of its permanently negative charge at all considered pH conditions, the adsorbed
NOM fraction can modify the electrochemical properties of the underlying surface layer
(Tombcz et al., 2004).

There is also evidence showing that metals may be complexed during batch reaction. In Fig.
4.24a, the kinetics of Zn
2+
removal in TU-SW (pH
i
2.5) showed a decreasing rate over time
(during 50-192 h) as compared with other runoffs indicating an inhibition occurs. Although, a
significant removal of DOC took place in TU-SW runoff, still 21 mg DOC L
-1
remained non-
absorbable, which might form complexes with the metals preventing Zn
2+
from being
adsorbed. In support of this, Dries et al. (2005) reported the formation of metal complexes
contributing to the 2.8 times reduction of Zn
2+
removal rates in a Humic Acid (HA)-Fe
0

containing batch system (HA 20.0 mg DOC L
-1
, 5.0 mg L
-1
Zn and Fe
0
6 g L
-1
.). Cheng and
Allen (2006) studied the binding characteristics of NOM from three surface waters and
reported that the Zinc-NOM complexes do not depend on the origin of surface waters. At pH
6, ionic strength 0.02 M and 10.0 mg L
-1
DOC, the same authors reported a total ligand
concentration of about 2.18 mmol g
-1
carbon. If this value is used to calculate the total DOC
concentration of 36 mg L
-1
(TU-SW) assuming a stoichiometry of complexation of 1:1
(NOM:Zinc), ~2.60 mg L
-1
Zn
2+
would be complexed in TU-SW runoff, which is comparable
to the remaining zinc concentration of ~2.18 mg L
-1
after 192 h in the batch test.
Alternatively, because complexation ability exists between different metals and Fe
2+
has the
highest affinity toward NOM as compared with Cu
2+
and Zn
2+
, it is more likely that a
significant portion of Fe
2+
forms complexes with NOM especially at high NOM
concentration. This results in a lower Fe
2+
oxidation rate, which subsequently hinders the
overall treatment rate of the Fe
0
system, as also discussed in a previous study (Rangsivek and
Jekel, 2005).

The complete removal of Zn
2+
was obtained in pH
i
5.50 solutions (Fig. 4.24c). The results of
50 h were plotted in relation to their NOM removal profiles, which are TU-SW (~70% Zinc
removal, NOM load per iron dosed 28.0 mg g
-1
Fe
0
, pH
f
6.92), SR-NOM (77%, 1.0, 9.29),
TU-FF (93%, 3.1, 9.57) and Cu-R (97%, 9.0, 7.95), respectively. Zn
2+
was removed in TU-
SW less than other runoffs, possibly attributed to the pH
f
, which was maintained below
pH
PZC
. At this pH<pH
PZC
condition, both ternary complex reaction and competitive
adsorption are taking place. For other runoff solutions in which the pH
f
values were above
pH
PZC
, both Zn
2+
and NOMs were removed concomitantly indicating an adsorption process.

0
system 78

Zn
2+
is, however, more prone to be adsorbed due to its preference toward the surface of iron
corrosion products at alkaline pH. In addition, a complete removal of metals indicates an
undersaturated condition exists in term of adsorption site for the adsorption of metals.

Beside other effects, the impact of NOM on the removal processes is the rise of metal
solubility that prevents metals and iron from precipitating as solids, i.e., as metal and iron
hydroxides, respectively. According to a calculation using MINEQL
+
, the solubilities for
copper and zinc equilibrium with hydroxide-carbonate system are about 15-50 g L
-1
in the
pH range of 8-9 and 0.5-3.0 mg L
-1
at pH 9-11, respectively. The lowest solubility of iron
hydroxides is about 0.03 g L
-1
at pH 8-9 (Ludwig, 2007). At concentrations up to 0.5-2 mg
L
-1
DOC, the solubility of copper may increase up to 0.92-1 mg L
-1
in the equilibrium with
malachite (Merkel, 2002). A much more significant increase in the solubility of copper and
zinc is anticipated in these high NOM containing runoffs employed in the present work.
Furthermore, the removal of metals mostly takes place at the initial phase, whereas, the non-
adsorbed concentrations of metals remaining in the solutions stay well below undersaturated
values. For this reason, the precipitation reaction of a metal solid as a result of an increase of
pH can be excluded.

The above depicted results of the column test demonstrate a lower iron concentration released
in concomitance with a higher amount of metals eluting out of the Fe
0
barrier at an increasing
NOM concentration of fed solution (Fig. 4.26). It was also observed that about a 20% total
concentration of DOC was removed during the Fe
0
reaction at each bed volume treated.
Hence, the decreased metal removal might have come about due to the fact that the adsorbed
NOM had blocked the reactive surface sites of iron, which also likely competes for adsorptive
sites with metals. In the study of Dries et al. (2005), a load of 4.5 mg HA g
-1
Fe
0
was found
not to interfere with the reactivity of Fe
0
toward contaminants (in deoxygenated water with
much higher EBCT). Specht et al. (2000) observed an insignificant influence of 5.0 mg NOM
g
-1
for clay at pH 5.0. In contrast, Tipping and Cooke (1982) reported that adsorbed humic
substances had a considerable effect on the electrophoretic mobility of goethite (40 mg g
-1
at
pH 7.0).

Another process possibly involved in the metal removal under the column experiment is
metal-NOM complexation. The dominant reaction is difficult to distinguish because of the
complex processes involved. However, if metal-complexation is considered to be significant,
an increasing amount of iron should have been observed at the outlet as an indication of metal
binding with NOM. This argument is only possible when the iron corrosion rate is noticeably
high (i.e., iron concentration at the outlet is high) and the aforementioned pathway (NOM
surface blockage of iron corrosion) is insignificant. For all these reasons and because of a
relatively fast contact time in the column test and low operational pH
inlet
of around ~5-6
(pH
outlet
is about 0.2-0.3 unit higher), it can be anticipated that there is a low involvement of
complexation of metals occurring in the Fe
0
system under the flow-through configuration.

0
system 79

4.3.5 NOM analyses by LC-OCD

The LC-OCD chromatograms of TU-SW and TU-FF bulk NOM show that larger molecular
weight compounds, specifically humic substances were preferentially removed during the
reaction with Fe
0
. This also corresponds to the UVD diagram demonstrating a favoring of
UV-absorbing fraction, e.g., aromatic moieties, being removed. The NOM of TU-FF runoff
decreased only to a small extent and the NOM fraction that was removed is located in a larger
size range. Gu et al. (1995) found the larger size hydrophobic fractions of NOM were
preferentially adsorbed on iron oxide surfaces over smaller size hydrophilic fractions. Genz et
al. (2005) reported that smaller and medium size fulvic acids are removed faster than other
low molecular weight fractions because of a higher diffusion rate. A ligand-exchange
mechanism between organic functional groups of humic substances and hydroxyl on iron
oxides surfaces governs the reactions of NOM adsorption (Gu et al., 1995). Further
elucidation of the adsorption behavior of organic compounds in the Fe
0
system was evaluated
based on the hydrophobic characteristics of NOM. In the TU-SW and TU-FF samples, when
the molecules are large, a non-humic fraction (XAD-8 effluent) is removed in approximately
the same degree as the hydrophobic fraction (XAD-8 isolate) (Fig. 4.27). The highly
hydrophobic NOM from Suwannee River exhibited almost a complete removal, whereas Cu-
R showed a similar result to that of TU-SW. In contrast to these results, the TU-FF NOM was
merely adsorbed. It mostly contains smaller molecular weight fractions of hydrophilic and
transphilic substances. These findings in overall showed that the steric arrangement of
carboxyl and hydroxyl functional groups play a greater role for the adsorption of NOM on
iron oxide surfaces than by chemical interactions between adsorbent surface groups and NOM
moieties. Furthermore, it was shown that the main fraction that was involved in the removal
of metals in Fe
0
the system is a larger size hydrophobic fraction.

Fig. 4.27 LC-OCD diagrams of TU-SW and TU-FF measured for the raw samples and the
samples at end of batch experiment
0
10
20
30
40
50
60
70
80
30 50 70 90
Time [min]
O
C
D
TU-SW raw
TU-SW 192 h
5
10
15
20
25
30 40 50 60 70 80 90 100
Time [min]
TU-FF raw
TU-FF 192 h

0
system 80

4.3.6 Conclusion

With an aim at better understanding the interaction of runoff NOM and the removal of metals
in the Fe
0
system, the present section employed extensive techniques for characterizing the
runoff-originated NOM samples. The results of characterization were subsequently used for
describing and explaining the removal of Cu
2+
and Zn
2+
at varying conditions during the batch
and flow-through systems. The following conclusions demonstrate the significant findings of
this study;

(i) It was shown that the characteristic of roof runoff from the same roof can be significantly
different. The concentration of NOM is significantly higher in the summer period as
compared to the winter samples. In general, NOM in roof runoff resembles natural waters
comprised mainly of humic substances with a relatively high aromaticity. This fraction of
NOM is the main compound that is actively involved in the removal of metals in the Fe
0

system. In contrast to this fraction of NOM, lower molecular weight and neutral NOM
fractions may only influence metal removal to a small extent.

(ii) The impact of NOM depends on the species of metal and the experimental conditions
being employed. In the batch investigation, NOM caused a minimal impact on the removal of
Cu
2+
, but either enhanced or inhibited the removal of Zn
2+
within the Fe
0
system. For the
batch treatment of runoff at acidic conditions, the adsorption of metals increased by a ternary
complex mechanism of NOM, metals and iron oxides. In some of the runoff solutions, it was
also evident that NOM complexation prevented metals from being adsorbed on the iron oxide
surface. However, under a typical runoff condition (i.e., at pH ~5.5), the main mechanism
responsible for removal of Zn
2+
and NOM is adsorption/co-precipitation, in which a
competitive adsorption between these two compounds exists. Whether competitive adsorption
or metal-NOM complexation is dominated at a specific condition depends largely on the
experimental pH. However, there is no clear distinction of these processes observed in this
study.

(iii) In the flow-through experiment, the breakthrough curves of Cu
2+
and Zn
2+
are
significantly influenced by the NOM concentration in water. The treatment efficiency can be
terminated under the presence of a relatively low NOM concentration, i.e., 5 mg NOM L
-1
.
This reduction in the removal of metal was presumably due to blockage of reactive iron
surfaces by the adsorbed NOM that also compete for adsorption sites. In contrast to the batch
test, the impact of metal-NOM complexes is considered insignificant under the conditions
studied.

(iv) NOM concentration is extremely high during summer and low during winter-time.
Nevertheless, this does not mean that the treatment of metal-contaminated runoffs by Fe
0
will
be worse in the summer. In fact, the frequency and amount of rainfall has to be taken into
account. A relatively small volume of the runoff in the summer makes the processes easier to
manage, e.g., by increasing the EBCT of the treatment column. It is also noteworthy that the

0
system 81

NOM-metal complex can reduce the potential impact of bioaccumulation and, thus, the non-
treatable fraction of contaminated water may not contribute to toxicity. Nevertheless, a
repeated exposure of the Fe
0
system by NOM may deteriorate the performance of the barrier
in the long run. During winter, the reactivity of Fe
0
and the treatment efficiency toward metals
can be greatly hindered by temperature. Even though NOM concentration is relatively low, a
constant releasing of metals from metal roofs generally coincides with the large volume of
runoff water making the process problematic.

4. Results and discussion ZVI and carbonatic barrier 82

4.4. Column experiments with ZVI and carbonatic materials

Results from previous chapter show that the Fe
0
/pumice barrier can barely cope with efficient
treatment of the metal-contaminated runoff within the desirable short time frame.
Furthermore, the Fe
0
barrier often failed during the long term operation due to a high
potential of clogging. In this part, the optimization of the processes is investigated employing
various carbonatic materials. They were tested using different configurations. It is aimed to
enhance the chemical precipitation of metals and iron oxides, while still offer sufficient
hydraulic loading permeability.

4.4.1 Optimization of the ZVIB using carbonatic materials

Composited carbonate/Fe
0
column (experiment. Col. 2)
A serie of experiments performed at TU and UDK roof sites clearly showed that the
Fe
0
/carbonate columns significantly enhance zinc removal efficiency as compared with the
Fe
0
columns (Fig. 4.28 and 4.29). During the first rain events at the UDK site (less than 50
BVT), zinc was merely eluted out of the Mag/Fe
0
and DML/Fe
0
columns. The Fe
0
/PM
column exhibited a slower kinetic and the control pumice column achieved a total 40-50%
metal removal the. The DML/Fe
0
and Mag/Fe
0
have a comparable treatment rate to zinc for
the first 10-15 cm but the latter exhibited a higher performance for the rest of the column. In
comparing the two sites, the removal performance of zinc was higher at the TU site, which
was likely due to a higher concentration of zinc that favored the removal processes. In
addition, a competitive adsorption effect of copper and zinc may be involved for the UDK
runoff. In contrast to zinc, the highest removal for copper was attained in the Fe
0
column,
followed by DML/Fe
0
and Mag/Fe
0
, respectively. This support the previous conclusion that
reductive transformation is the major removal pathway for copper rather than an adsorption
and co-precipitation, or chemical precipitation processes.

In the second measurement, the carbonate incorporated column at the UDK site experienced
clogging as well as preferential flow. This was found to a lesser extent for the pumice and
the Fe
0
columns. Visual inspection revealed that the clogging of the infiltration layer
occurred predominantly at the inlet end of the columns and was likely caused by particulates
in runoff waters. In addition, the clogging may have also been caused by the accumulation
of various hydroxide and carbonate precipitates within the treatment barrier. As a result of
flow reduction, the removal profiles of metals in the composited columns performed onsite
were only obtained at the TU site. For the second rain event at TU site (approx. at 400
BVT), the eluted concentration of 0.5 mg L
-1
from the Fe
0
column and breakthrough of the
pumice column were already observed. The observed low performance of Fe
0
likely
reflected a formation state of iron oxides. It was also likely due to a pH lower than pH
PZC

that did not favor the adsorption process. In contrast, the Mag/Fe
0
exhibited a complete
removal of zinc while the DML/Fe
0
showed a slightly increasing concentration front with
about 0.1 mg L
-1
zinc is the eluted concentration (Fig. 4.29).


0,0
1,0
2,0
3,0
4,0
5,0
6,0
0 20 40 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Cu
Pumice
Fe0
DML/Fe0
Mag/Fe0
0,0
0,2
0,4
0,6
0,8
1,0
1,2
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Zn
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
8,0
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Ca
0,0
1,0
2,0
3,0
4,0
5,0
6,0
7,0
8,0
9,0
10,0
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Mg
0,0
1,0
2,0
3,0
4,0
5,0
6,0
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Fe
T

Fig. 4.28 Metal removal profiles from composited columns tested at UDK site during the first and second rain events. The columns received
runoffs from a copper roof in the downflow mode (during the measurement, the average flow was 60 mL min
-1
)


0,0
0,2
0,4
0,6
0,8
1,0
1,2
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Zn - 40 BV Pumice
Fe0
DML/Fe0
Mag/Fe0
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Zn - 400 BV
0,00
0,50
1,00
1,50
2,00
2,50
3,00
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Ca - 400 BV
Fe0/DML
Fe0/Mag
0,0
0,5
1,0
1,5
2,0
2,5
3,0
3,5
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Mg - 400 BV
Fe0/DLM
Fe0/Mag
0,0
0,5
1,0
1,5
2,0
2,5
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Fe - 40 BV
0,0
0,2
0,4
0,6
0,8
1,0
1,2
1,4
0 10 20 30 40 50 60
C
o
n
c
.

[
m
g
/
L
]
Length [cm]
Fe - 400 BV

Fig. 4.29 Metal removal profiles from composited columns tested at TU site during the first and second rain events. The columns received
runoffs from a bitumen roof with a zinc gutter in the downflow mode (during the measurement, the average flow was 60 mL min
-1
)


The rest of the 4 month investigation of composited columns shows a significant loss of flow
permeability and finally ended at about 90% flow reduction. With the low flow rate and initial
copper and zinc concentrations on average of 5.36 mg L
-1
and 2.30 mg L
-1
the observed total
metal concentrations of the effluents were 0.035 and 0.47 mg L
-1
for dolomite and 0.025 and
0.07 mg L
-1
for magnesite composited columns, respectively (Fig B14). This accounts to 89-
99% removal efficiency.

The incorporation of carbonate bearing materials into a Fe
0
treatment barrier is by principal to
precipitate more dissolved metals into insoluble solids as well as to favor adsorption at an
alkaline pH via the addition of CO
3
2-
, HCO
3
-
, OH
-
, Ca
2+
and Mg
2+
. Based on the results of the
composit columns, approximately 4-7 mg L
-1
Ca
2+
and 7-9 mg L
-1
Mg
2+
were observed for the
first rain event and about two times lower for the second rain event (data from two sites were
used for comparison). As a consequence, the outlet pHs were usefully maintained at 8-11
which also results in an extensive deposition of greenish precipitates, supposely Malachite,
within the treatment columns of magnesite and DML. An extensive precipitation of iron
oxides was apparent in the DML/Fe
0
and Fe
0
columns, whereas this was not observed in the
Mag/Fe
0
column. Moreover, the concentration of iron along the Mag/Fe
0
column was minimal
suggesting that Mg
2+
may have inhibited the corrosion process of iron. It is therefore,
reasonable to conclude that the removal of metals in Mag/Fe
0
was solely caused by the
precipitation processes.

Sequential carbonate/Fe
0
columns (experiment. Col. 3)
A higher metal removal was generally observed for sequential carbonate/Fe
0
columns as
compared with the Fe
0
column (Fig. 4.30). Nevertheless, this also brought about a high
clogging potential similar to the composite configuration. In the Mag-Fe
0
columns, clogging
and preferential flow was evident after a few BVT. This was observed when a greenish solid
was increasingly precipitated within the column. The flow rate of the experiment was,
therefore, adjusted to 30% of the initial value. Corresponding to this, the effluent pH of Mag-
Fe
0
could be maintained at 8.1-11.1 and a complete removal of metals could be obtained
throughout the run (Fig. 4.30e). There was also no iron precipitates within the following Fe
0
column, supporting the previous assumption that Mg
2+
can inhibit the iron corrosion process.
Nevertheless, the results showed that column containing only Mag can effectively treat metal
from roof runoff. In comparison to the Mag-Fe
0
column, the effluent of the DML column was
shifted to about 9.5-10.7 with complete removal of metals observed only up to 342 BV. After
that DML gradually become inactive and the effluent concentration of metals increased. Due
to severe clogging within the Fe
0
column, the DML-Fe
0
column was terminated at 1,500
BVT. In other columns, the calcite-Fe
0
did not seem to clog the column but it was possible to
increase the pH up to 10.4 for only few BVs (Fig. 4.30b). A greenish color developed on the
surface of CaCO
3
that might lower the effectiveness of the media in increasing pH. The DMS
is considered inactive as there was no enhancement in term of pH buffering and removal of
metals. After about 2,000 BVT, the DMS was replaced by DML, and a better removal of
metals could be observed (Fig. 4.30c).


According to the results of the tested columns, it is evident that the major factor determining
the life span and overall efficiency of carbonatic-Fe
0
treatment barriers is the hydraulic
conductivity. Although, the increase of pH induced by the carbonate materials favor a higher
treatment rate of metals, this also resulted in a high potential for plugging and preferential
flow that brought about failure of the barrier. For this reason, another configuration of Fe
0

incorporated carbonatic materials, in which a sequential configuration of an Fe
0
column
followed by a filter bed containing dolomite/pumice column was examined. In this setup, the
system relies on the production of dissolved iron in the first Fe
0
column. The dissolved iron
and metals that passed through the first column will subsequently be transformed into iron
oxides in the second column of the filter bed. The incorporation of DML in the post filter
bed aimed to precipitate all the metals including Fe
2+
that may be persistent due to
complexation with NOM in runoff waters (Rangsivek and Jekel, 2005; 2008). Figure 4.30f
shows that the removal efficiency of copper and zinc in Fe
0
-DML/PM columns were 80-
100% and 50-100%, respectively. There was no significant clogging potential for up to
Fig. 4.30 Breakthrough curves of copper and zinc with corresponding pH measurement from
sequential columns of Fe
0
/carbonatic materials; (a) controlled Fe
0
column; (b) calcite-Fe
0
column; (c) DMS-Fe
0
column; (d) DML-Fe
0
column; (e) Magnesite-Fe
0
columns; and (f) Fe
0
-
DML column. The first line (above) represents output conc. of first column and the second
line (below): output conc. of second column. pHs are in reverse order of conc. profile.


2,000 BV. For this configuration, the pH of the filter bed could be maintained at 9.5-11. Its
buffering capacity gradually decreased when higher red oxides deposits were observed in
the filter column in concomitant with the reduction in the flow rate of the filter bed. The pH
at which the removal of metals become deficient was about 6.87-9.47 which occurred at
BVT ~2,000, the DML/PM media were taken out and washed before reloading into the
system. As a result, a complete removal of metals and a high hydraulic loading rate could be
re-attained suggesting that through backwashing, the system offers maintenance of the bed.

4.4.2 Processes of metal removal in ZVI/carbonatic barrier

Based on the results of composited and sequential column tests, it could be seen that removal
of metals depended strongly on the solutions pH suggesting complex processes of chemical
precipitation, adsorption and co-precipitation were involved. In order to assess the role of
chemical precipitation, the chemical equilibrium prediction using MINEQL
+
(Fig. 4.31) were
employed to elucidate the composition of solid precipitates in rational to SEM and EDX
analysis (Fig. 4.32).

In the composited column, the treatment media showed a visual mixture of various solid
precipitates, e.g., having reddish iron oxides, greenish and brownish colors. These phases
were generally fragilely deposited on the surface and pore of the pumice media. Using SEM
to examine the precipitates, nodular shaped solids formed tightly and distributed uniformly
over the reactive surfaces were observed (e.g., Fig. 4.32 of CaCO
3
-Fe
0
and DML-Fe
0
).
Employing EDX, the reflection could be assigned to phases of copper, zinc, Mg and Ca. The
metal phases usually co-exist with Ca and Mg, which are assumed to be a result of chemical
precipitation at an increasing pH. Results of chemical equilibrium calculation support this
assumption, in which precipitates that have a potential to precipitate were predicted as
mixtures of azurite, malachite, CuCO
3
, tenorite, Cu(OH)
2,
ZnO, Zincite, Zn(OH)
2
and ZnCO
3

(Fig. 4.31a). Other than these phases, Mg(OH)
2
and Siderite (FeCO
3
) may have also been
generated. In the case of the Fe
0
-DML sequential columns, the separate sample of Fe
0

columns (i.e., control column of Fe
0
with no influence of carbonate) was covered by a
reddish-yellow layer. Examination of media from the Fe
0
column reveals a nodular shape
growth with a strong spectrum of a copper phase. Zinc was also found to constitute at a
typical iron surface.

Further examination of the role of chemical precipitation was performed by means of plotting
of the effluent concentrations of Fe
0
, Mag/Fe
0
and DML/Fe
0
column tests against the
solubility of metal hydroxide and carbonates such as malachite and smithsonite. The
equilibrium model shows that super saturation exists for all the Fe
0
integrated carbonatic
columns and the observed effluent concentration of metals agree well with the solubility of
metal hydroxide and carbonate (Fig. 4.31b). Based on the results, the observed effluent
concentration of metals agreed well with the solubility concentrations of copper hydroxide
and malachite, which are approximately 15-50 g L
-1
at the pH range between 8 and 9 and at
0.5-3.0 mg L
-1
for zinc-hydroxides and smithsonite (ZnCO
3
(s)) at pH 9-11 (Ludwig, 2007).


Hidmi and Edwards (1999) reported a solubility of copper at 0.12 -0.043 mg L
-1
. The
observed metal concentrations significantly below the solubility at pH 5-7 were likely due to
the adsorption and co-precipitation processes. In this examination, the results of copper and
zinc concentrations from the field experiment illustrated in Fig. 4.33 were employed.

Besides metal hydroxides and carbonates, the equilibrium model also predicted that various
phases for voluminous precipitation of iron oxides would occur within the column test. The
oxides phases were assigned to lepicocrocite, maghemite, goethite, hematite and ferrihydrite.
Results of a model prediction correspond well with the XRD analysis performed in the batch
and columns tests previously shown in section 4.1 and 4.2. On the contrary, the solubility of

-40,00
-30,00
-20,00
-10,00
0,00
10,00
20,00
30,00
L
o
g

C
Azurite Malachite Tenorite
Cu(OH)2 Cupric ZnO
Zincite Zn(OH)2 ZnCO3:1H
CuCO3
a)
Lepidocrocite Goethite
Hematite Ferrihydrite
Maghemite Magnesite
b)
1,0E-03
1,0E+00
1,0E+03
1,0E+06
1,0E+09
1,0E+12
2 4 6 8 10 12 14
C
o
n
c
e
n
t
r
a
t
i
o
n

[

m
g

L
-
1
]
pH
Copper hydroxide
Malachite
Result of this study
Result from Merkel (2002)
c
min
= 15 g/L
Copper hydroxide
Malachite
c
min
= 50 g/L
c)
2 4 6 8 10 12 14
pH
Zinc hydroxide
Results from Mann et al. (1980)
Smithsonite
Results from Patterson et al. (1977)
This study
c
min
= 0,5 mg/L
Smithsonite
Zinc hydroxide
c
min
= 3 mg/L
d)

Fig. 4.31 a) and b) Illustration of saturation indexes of relevant minerals related to the Fe
0

treatment barrier of copper, zinc and iron oxides, respectively, c) and d) Plots of effluent
metal concentrations obtained from this work in equilibrium with metal hydroxides (the
shown data were complied by Ludwig, 2007)


Fig. 4.32 SEM and EDX mapping of solid precipitation obtained from selected composited and sequential column tests


magnetite was well below one and, thus, not precipitated under the stromwater runoff
conditions. It is well known that magnetite is only formed in deoxygenated conditions
(Cornell and Schwertmann, 1996). Due to a presence of these iron phases, it is suggested that
adsorption and co-precipitation process could not be ruled out and the performance of the
Fe
0
/carbonatic barriers was likely involved through a combined processes of adsorption/co-
precipitation and chemical precipitation.

4.4.3 Implication of ZVI/carbonatic barrier

Based on the results of laboratory and field experiments, it was evident that the column with
carbonatic materials exhibited a higher performance than the column filled only with Fe
0
.
Among media investigated, Mag possesses the highest removal rate of metals where metals

Table 4.5 Water quality of the influent to Fe
0
-aeration-DML/PM system
Rainfall data Paramter Zn Cu NOM
[mg L
-1
] [mg L
-1
] [mg L
-1
]
1. cycle Zn (natural runoff) 2 0-5 3
2. cycle Zn/Cu 5 5 3
3. cycle Zn/Cu 5 5 3
4. cycle

Zn/Cu/NOM 5 5 20
*There were some losses of metals through adsorption to particles in the storage tanks

0,0
1,0
2,0
3,0
+,0
5,0
6,0
Z
n
t
c
o
n
c
e
n
t
r
a
t
i
o
n

[
m
g

L
-
1
]
!nlet Conc. Effluent from Fe0 columns Effluent from filter bed
0,0
1,0
2,0
3,0
+,0
5,0
C
u
t
c
o
n
c
e
n
t
r
a
t
i
o
n

[
m
g

L
-
1
]
+
5
6
7
8
9
10
0 500 1000 1500 2000 2500 3000 3500 +000 +500 5000
p
H
BVT
Fig. 4.33 Results of the implication of the optimized Fe
0
/carbonatic barrier (Ex. Col. 4)


are immobilized purely by chemical precipitation process. Thus, when applying Mag in the
field, the following ZVIB may not be necessary however post sand filtration is required to
filter all metals carrying particles. The highest removal rate of Mag has a drawback due to
its greatest potentials for permeability losses and preferential flow of the treatment barrier.
The low hydraulic loading rate is preferentially required for actual operation. Alternative to
Mag, DML may be considered more suitable for stormwater runoff treatment due to its low
cost. The lower performance of the media can be justified by its ability to treat water at a
higher flow rate with less difficulty.

For further assessment of the Fe
0
-carbonatic barrier, a new configuration of Fe
0
/carbonatic
barrier was established based on the results of column tests. This treatment system consisted
of spiral irons in the first column (without pumice or sand), followed by aeration and a
fixed-bed system containing pumice and dolomite (or other carbonatic materials such as
magnesite). The spiral iron was used in the treatment process in order to increase void area
within ZVIB as it was found that building up of iron oxides inhibits the processes of metal
removal. The system principally relies on the production of dissolved iron in the Fe
0

column. The iron is subsequently transformed into iron oxides within the Fe
0
column or at
the later treatment steps. The aeration and post filter bed help in facilitating the oxidization
rate of dissolved iron. In this case, the metals are immobilized by means of a combined
process of adsorption/co-precipitation and precipitation with carbonate bearing materials
while still maintaining the reactivity of the iron column. The treatment barrier was installed
onsite receiving natural runoff from TU roof. Furthermore, the system was tested under
natural runoff and at fluctuating runoff quality where copper, zinc and NOM were added
artificially. The NOM soluion was originated from decomposition of reefs in the runoff
water. The parameters investigated are demonstrated in Table 4.5.

About 60-100 % of metal concentrations could be removed at EBCT of 5 min and about 30-
80% was achieved with 1 min EBCT in the ZVI/DML/PM barrier (ZVICB) (Fig. 4.33 and
Table B32). For this setup, there was also less effect of source variations of waters on the
treatment ability of metals. Noteworthy is that the presence of NOM in roof runoffs did not
influence the reaction rate of ZVICB as was found to be significant for Fe
0
column alone.
This could be due to an enhanced precipitation of metals and iron oxides by the DML filter
media. When closely examining the system, a larger part of copper was found to be removed
during Fe
0
passage whereby the removal of zinc mostly takes place during DML/pumice
filter bed. Overall, the outlet metal concentration in the runoff after passing through a
DML/pumice filter bed was kept below 1 mg L
-1
for over 4,000 bed volumes.

Major drawbacks of the ZVICB are iron oxide-passivation of dolomite media that reduces
the rate of carbonate dissolution, productivity of alkalinity as well as clogging from iron
oxyhydroxides and solid precipitates. In the system of Fe
0
-aeration-DML/PM filter, it was
possible to recover treatment efficiency and the hydraulic loading rate by backwashing the
post treatment media (Fig. 4.34). As could be observed, the efficiency of metal removal
begins to decrease after about 3,500 BV accompanying with clogging of the filter bed. After


backwashing, a similar high metal removal was attained. The pH of the effluent from the
filter bed and hydraulic loading rate is a good indicator of the requirement for backwashing
of the media. Furthermore, the pH uplifting capacity of DML media is the main factor that
determines overall effectiveness and thus the life span of the treatment barrier. If about 4 g
m
-3
of Ca and Mg are assumed to dissolve with the available 800 g of DML media, an
estimated 200 m
3
of water can be treated by this system.

Fig.4.34 Hydraulic characteristic of the optimized Fe
0
/carbonatic barrier over time. The
backwashing of the media yields sustainable hydraulic performance

4.4.4 Conclusion

Based on the experimental and modeling results, Fe
0
supported with pumice column can
barely cope with good requirements of water qualities. This section demonstrates the method
in order to optimize the Fe
0
barrier system. The method involved a combination of various
carbonatic media into the Fe
0
system both as composited and sequential columns. In this way,
the performance of the barrier will be increased by means of accelerating the rate of chemical
precipitation of metals. Incorporation of these materials also increases affinity of adsorption at
a higher pH. The columns were both tested in the laboratory and on-site receiving realistic
rain events. The results showed that Fe
0
/carbonatic columns significantly enhance metals
removal efficiency as compared with the Fe
0
column. By incorporating carbonate bearing
materials, the rate of chemical precipitation is increased and metals are precipitated into solid
and immobilized. Nevertheless, this method has its drawback due to the treatment columns
0
50
100
150
200
250
300
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
time (min)
F
l
o
w

(
m
L
/
m
i
n
)

o
r

m
L
0
20000
40000
60000
80000
100000
120000
140000
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
time (min)
A
c
c
o
m
u
l
a
t
e
d

f
l
o
w

(
m
L
)
Reality: flow decrease
over time
deal flow
back washing


higher potential for clogging and the final configuration of the Fe
0
treatment system was
therefore designed. The system consisted of columns filled with spiral shaped irons followed
by an aeration unit and post filtration of pumice/dolomite. The treatment unit was installed
onsite for treatment of metal-contaminated runoff under fluctuating conditions. It was found
that there was little impact of water quality parameters on the treatment efficiency of metals.
Nevertheless, the barrier requires a periodical backwash to retain a good removal efficiency
and also to gain an optimum hydraulic loading rate.


4.5 Design of a treatment system and application of results

This section demonstrates the concepts and guidelines for the ZVICB design. The design of
the treatment barrier takes into account the thorough understanding of physical-chemical
removal processes of metals derived from the research results. The knowledge of kinetic
removal rates of metals, conditions that suits immobilization processes, products and factors
determining the life span of the treatment system will be used to quantify the amount of media
required. It is important to consider the catchment and flow characteristics for the hydraulic
and structural design of the system aiming to minimize the potential risks of contaminants in
subsequent surface discharge or groundwater recharge. Furthermore, the chapter will illustrate
how the adsorption model is used for the assessment of the systems performance over the
long term period. Finally, the initial investment cost of the treatment media for ZVICB is
compared with the GFH/Lime system.

0
for highway runoff treatment

The conceptual design procedures for treatment of highway runoff are relatively well
established. In general, a wet detention pond system in combination with an infiltration basin
is applied (Hvitved-jacobsen et al., 1994). The immobilization processes of pollutants in the
detention pond are governed by sedimentation of particulates and adsorption of dissolved
compounds to sediments during the percolation period. Uptakes of soluble compounds by
plants through biological processes may also take place (Barbosa, 1999). To yield efficient
treatment of pollutants in the pond, physical characteristics of the treatment pond including
shape, water depth, side slope, and hydraulic retention time shall be designed to allow good
settling of particles and avoidance of a completely mixed condition. In general, a plug flow
characteristic is preferred. Designs of wet detention ponds for highway runoff are mostly
based on hydraulics and have a high return period of 10 to 100 years. Hence, during wet
weather periods, the overflowing runoff volumes above the treatment capacity of the pond as
well as remobilizing of the settled solids can result in lower treatment efficiency. Hence, an
advanced treatment method applying a fixed bed reactor in subsequence to a pond may be
integrated in order to enhance the overall treatment efficiency.

Fig. 4.35a-c illustrates the application of zero-valent iron for advanced treatment of highway
runoff. These conceptual ideas, though not for Fe
0
, are based on the research project initiated
for treatment and re-use of urban stormwater runoff in Denmark (http://www.life-
treasure.com). In the first illustrated method (Fig. 4.35a), Fe
0
may be incorporated into an
existing pond for production of iron salt as flocculants, which in proportional to the flow can
enhance the removal of particulates and dissolved compounds. The second method is to use
Fe
0
media to increase the removal of dissolved metals during the adsorption to sediments (Fig.
4.35b). In this case, Fe
0
is filled into the bottom sediments for production of iron oxides as
adsorption sites. The final method is to use ZVICB filter bed in the post treatment step of the
effluent (Fig. 4.35c). This method intends to remove more dissolved compounds than the
ponds capacity. Prior to filtration bed, passage through sand bed is crucial to reduce clogging


Fig. 4.35 Applications of Fe
0
for treatment of highway runoff (modified after Life-Treasure Project, http://www.life-treasure.com)

4. Results and discussion Design and application 96

of the barrier resulting in a longer life span of the treatment barrier. These methods can be
combined to achieve maximum treatment performance.

0
for roof runoff treatment

In comparison to high way runoff, roof runoff is characterized by a higher concentration and
constant flux of heavy metals. The metals in roof runoff are often present in a dissolved form
due to the low content of particulates and the acidic pH of water. As compared to a pond
system, a fixed-bed system is more advantageous for the design of the treatment system of
roof runoffs because of the required high metal uptake rate and the shortest contact time that
allows the system to use much less space and favours installation onsite. In several recent
studies, the fixed bed system was successfully demonstrated. Steiner (2003) employed a
GFH/CaCO
3
barrier for treatment of runoff from copper roof and faade for several sites in
Switzerland. An ion exchange barrier using clinoptilolite media was also proven
(Athanasiadis et al., 2005). Yet, a cost effective media with a high metal uptake capacity as
well as the long life span of the system is further aimed for actual implication in stormwater
treatment. Based on the results of this work, a novel fixed-bed system containing
Fe
0
/carbonatic material has been shown to possess a high removal efficiency with sufficient
hydraulic stability and, hence, may be further actualized. Depending on the treatment
configuration, application of ZVICB for treatment of roof runoff may be applied as illustrated
in Fig. 4.35 and 4.36.

Fig. 4.36 GFH/CaCO
3
barrier for treatment system of copper faade (Steiner, 2003)


4.5.3 Design of ZVICB system

To illustrate how the results of the research can be utilized at actual sites, this section
demonstrates the state-of-the-art design of the Fe
0
/carbonatic system. The following
assumptions have been drawn:

A catchment area consists of a bitumen roof with a zinc gutter and a copper roof (Fig.
4.37). Within the catchment area and already considering losses through evaporation
and adsorption to a permeable surface, it is assumed that rainfall is routed and
transformed into a runoff volume with a total runoff volume of 850.6 mm (Table 4.6).
The corrosion rates of metal roofs were assumed to be 3.6 g Me m
-2
a
-1
. The runoff
shall be treated before infiltration into the subsurface for purposes of groundwater
recharge.

The standards of water discharge are 0.05 mg L
-1
for copper and 0.5 mg L
-1
for zinc.
This corresponds to a required removal rate of 99 and 92%, respectively.

Copper Roof
Zinc Roof
Zinc Roof
Area
Hydrology
Soil
Location 10 km west of city
Use
Residental housing
Institute buildings
Parking lots
Roof 1,0 ha
Road 1,8 ha
Greenland 0,2 ha
Rain depth 620 mm / a
Summer
(Jun. Aug.)
75 %
Rest 25 %
Top Layer
Thickness: 9 m
Permeability: 6*10^-6 m/s
City center
Residential and University
area
Copper, Bitument Roofs
excluded
excluded
850.6 mm
75%
25%
Cathment

Fig. 4.37 Catchment area characteristics for illustration of the design of treatment barrier

In the design procedure, four steps are established;

(1) conceptual design of the treatment system of roof runoff
(2) determination of the media required and the treatment volume


(3) dimensioning the geometry of the treatment system, and
(4) estimation of the lifespan and the annual sludge waste

Conceptual design of the treatment system
The conceptual design of treatment system of roof runoff is illustrated in Fig. 4.38. The main
components of the treatment train are a storage tank, ZVICB and the recharge well. The
runoff from impervious area is transported through sewer lines and collected in the storage
tank through which runoff will then be fed into the treatment system. This storage tank is
designed not only to attenuate the flow but also to behave as settling step of particulates. The
treatment system may be constructed underground using concrete structure or above ground
as natural pond system that can be integrated into the recreation area.

Fig. 4.38 Conceptual design of the treatment system

Design of the treatment barrier
To design of the treatment system, the metal uptake capacities of 42 and 22 g kg
-1
Fe
0
for
copper and zinc have been used, respectively. These metal uptake loads were obtained from
the experiments at high NOM concentrations, which will result in a larger size of barrier for
reasons of safety factor (S.F.) in case water qualities are not known. An assumption of a total
50% iron till exhaustion was assigned. The design calculation sheet has been developed
(Table 4.6). The calculated results based on the work of Ludwig (2007) and Steiner (2003) are
also demonstrated for comparison. These require input parameters including the physical and
chemical data of iron and DO consumption rates.

Based on the calculation, it was found that about 18 and 34 kg of iron are required for the
treatment of copper and zinc, respectively. The higher value will be further used in the actual


implementation. This design assumes that reactive media is to be replaced annually. An
estimation of about 50% of total iron mass of 17 kg corrosion products should be classified as
toxic wastes and requires special disposal.

Table 4.6 The criteria and design results of the ZVIB

Ludwig Steiner This work
Paramter Symbol Unit (2007) (2003) Cu Zn
Catchment characteristics

Roof surface area A m
2
210 210 210 210
Corrosion rate r(Me) gMe m
-2
a
-1
3.6 3.6 3.6 3.6
Precipitation intensity r mm a
-1
850.6 850.6 850.6 850.6
Runoff coefficient - 1 1 1 1
Runoff volume V m
3
a
-1
179 179 179 179
or equal to L h
-1
20 20 20 20
Design criteria of ZVIB

Volumetric surface area of Fe
0
a
V
m
2
L
-1
0.8 - - -
Bulk density of Fe
0
media
s
kg L
-1
0.8 0.65
1
0.20
2
0.20
Bulk density of Ca media
c
kg L
-1

- 1.02 1.02
Oxygen uptake rate r(DO) L m
-2
min
-1
0.05 - - -
Metal uptake load q
M
mg Me g
-1
Fe
0
50 30 42 22
Inlet concentration - Me c
in
(Me) mg Me L
-1
6 6 6 6
Outlet concentration - Me c
out
(Me) mg Me L
-1
0.5 0.05 0.05 0.5
Design of the ZVI Barrier

Number of baffle n - 10.5 - - -
Total removal rate
% 92% 99% 99% 92%

Total iron mass m
(Fe) kg 40 25 18 34
Total cabonatic matials m
(Ca) kg - 25 3.2 3.2

Total volume of reactor V
L 50 38.5 95 175
Total HRT t
v,
h 2 -
3
-
4
-
Total height of bed L
m 5 - - -
Life span of the system

Time to reach 50% usage of iron t
50
a 1.0 1.0 1.0 1.0
Total flow treated Q
50
m
-3
180.6 180.6 180.6 180.6
1
bulk density of GFH mixed with Lime;
2
filled-in density of spiral Fe
0
;
3
designed to receive instant flow;
4
HRT of minimum
5 min is recommended

Design of the DML/PM barrier
Based on the results of the test, assuming 8 g m
-3
of Ca and Mg to dissolve into the runoff
water, an estimated 200 m
3
of water will require 1600 g of DML media. For a safety factor,
a 50% saturation is assumed and, thus, during a one year operation of the ZVICB, about 3.2
kg of DML media is required for one year operation. The equal amount of pumice is mixed
with DML to increase porosity and filterability.

Hydraulic and structural design
In term of engineering application and realization of the design in the field, about five times
amount of the treatment media will be used, which prolong the life span of the system up to
about five years. A larger system is preferred for installation onsite. This is also to reduce an


initial investment cost and ease of maintenance. The design was based on the assumption that
a total volume of runoff (90 or 180 m
3
a
-1
) is intercepted and collected for treatment at
constant and low flow rate with up to 6 h contact time over the period of one year. This
concept required sufficient storage space onsite.

An example for the hydraulic and structural design of Fe
0
treatment system for runoff has
been illustrated (Fig. 4.39a). The treatment barrier was designed as a single reactor, consisted
of three main components; (1) a sedimentation tank; (2) a baffle reactor tank consisting of an
Fe
0
barrier, aeration and DML/PM bed; and (3) the recharge basin. Conceptually, the
sedimentation tank will remove particulates as well as some metals bounded sediments from
entering the barrier. The majority of metals will, however, be removed in the baffle reactor
where the contaminated runoff will pass through the Fe
0
and carbonatic barriers with an up
and down flow characteristic. The baffle reactor was chosen to maximize the contact between
runoff and media preventing short circuit of the system. The water depth of less than 0.5-1.0
meter shall be designed to aerate water. Sufficient DO is required for complete processes of
iron corrosion and for subsequently oxidized the iron bound particles during the natural
aeration steps. In the last part, the baffle reactor consists of DML/PM that will fully oxidize
and precipitates all persistent compounds into solids. There are units for withdrawing excess
sludge from the bottom of the reactor, through gravity and a drainage pipe (Fig. 4.39b). The
outlet from the reactors will then be used to recharge into the subsurface. It is anticipated that
outlet water is basic so further aeration might be required before discharge. Based on the
study of Karchunke (2005), it may also be possible that the post filtration bed is constructed
only with sand. Nevertheless, in the case of stormwater runoff treatment at high NOM
concentration, a longer contact time during sand filtration is recommended to allow full
oxidation of iron into particulates. The treatment system can be constructed as a natural soil or
concrete structure depending on the cost factor. It is recommended that the treatment system
should be lined with HDPE in order to prevent metal diffusion into the infiltration layer. In
case of constructing with natural soil, a metal-transport zone where packed soil/Fe
0
at 10-25
cm above natural soil at the bottom of adsorption zone shall be installed. After a period of
time, the soil from this zone is considered toxicant wastes and must be withdrawn. It can be
dried to decrease its volume for further disposal.

As an alternative to the configuration described above, the ZVICB can be constructed onsite
and installed directly beneath the roof or faade surface as illustrated in Fig. 4.36 by Steiner
(2003). In that case, the ZVICB shall be designed to intercept the runoff volume originating
from a limited surface area and to treat the water at the shortest contact time as possible, i.e.,
less than 1 min in order to infiltrate the water underground immediately. The system requires
no storage basin but the design volume of media must take into account a highly fluctuating
flow rate, kinetic of removal, water qualities and metal influxes of the runoff at each rain
event. The adsorption model can be used for dimensioning treatment barrier.


Fig. 4.39 a) 3D view of the treatment barrier b) baffle reactor and withdraw units of the
excess sludge

Assessment of the treatment efficiency
The adsorption model as demonstrated in section 4.2.3 is employed for simulating the long
term performance of the ZVIB (Fig.4.40). The adsorption model allows for a closer
assessment of the system where the simulation of the treatment process at various scenarios,
i.e., the rainfall data can be used to assess the treatment efficiency over the life span of the
barrier. For modeling, the metal fluxes were simulated by multiplying the corrosion rates of
metals with precipitation volumes at each rain event, where a total volume of the runoff is
equal to 180 m
3
a
-1
as used in the design criteria. The treatment contact time was assumed to
be 1 h and by using an iron barrier with 180 kg Fe
0
in the barrier. It is aimed at demonstrating
the application of the model to examine the treatment efficiency for further design of the
barrier. In this example, it appears that the ZVIB does not capture metals sufficiently.
a)
b)


Nevertheless, it is expected that an incorporation of DML/PM media in the post treatment step
will ensure the quality of discharge in reaching the standard of effluent.

Table 4.7 Input parameters for simulation of the long term efficiency of ZVIB
Parameters D
i
L Fe
0
EBCT Cu
2+
Zn
2+

K
F
1/n k
f
K
F
1/n k
f

Conditions/Unit [m] [m] [kg] [ min] [x10
-3
] [x10
-3
]

Final design 3 10 180 60 20 0.6 0.2 5 1.0 0.10

0
10
20
30
+0
50
60
70
80
C
/
C
0
f
o
r

Z
n
0
10
20
30
+0
50
60
70
80
0 50 100 150 200 250 300 350 +00
C
/
C
0
f
o
r

C
u
days
0
10
20
30
+0
50
60
70
80
T
o
t
a
l

c
o
n
c
e
n
t
r
a
t
i
o
n

[
m
g

L
-
1
]

Fig. 4.40 Model simulation of outlet concentration of zinc and copper from the Fe
0
barrier.
Based on the research data, with a proper design of carbonatic post treatment barrier, the
effluent will meet a quality standard requirement.

Cost estimation
Table 4.8 demonstrates that the initial investment cost of treatment media for ZVICB is much
lower than for the GFH/Lime system, especially when scrap iron can be utilized. The relative
investment cost for the treatment media for ZVICB is likely to further decrease with an
increasing scale of the system (Table B36 in Appendix B). Nevertheless, since the GFH/Lime
requires a smaller foot print than the ZVICB because of a higher removal rate and the requires
high void volume (>90%) of ZVICB. A complete comparison must take into account the cost
of system construction. For this, total volume of reactors were estimated by dividing the
required mass of treatment media by the filled-in bulk density of the media and the cost of
construction can be calculated. Several factors influence the cost of treatment structure,
including frequency of refilling the treatment materials, capacity, type of construction
*For modeling of both copper and zinc, D
s
= 1.0 x 10
-10
cm
2
s
-1
and D
p
= 8.53 x 10
-16
cm
2
s
-1
were employed.


materials, and location. Steel structures can be constructed on ground-level, while concrete
structures are normally installed at or below grade. Below grade structures, e.g., lining type,
require higher excavation and backfill costs. A comparison of of GFH/Lime and ZVIB system
constructed with concrete, lining or bared soil (without structure) types for a range of roof
surface areas are shown in Table B36. The results were used to calculate the total investment
cost. The calculation shows that the investment cost of 0.39-0.66 Euro m
-2
y
-1
for GFH/Lime
and 0.01-0.62 Euro m
-2
y
-1
for ZVICB system were obtained. It must be noted that the cost
estimation did take only a limited requirement of factors into account and a lump-sum cost per
unit workload were used for calculation. Furthermore, e.g.,the required storage tank for
ZVICB was not included into the calculation, the data presented here may not be directly
applicable for estimation of the actual project.

Table 4.8 Cost comparison for ZVICB and GFH/Lime treatment systems
Parameters ZVICB GFH/Lime
Scrap iron Commercial ZVI
Initial cost for treatment media
ZVI or GFH (Euro t
-1
) 45 325 3,620
Magno-dol or Lime (Euro t
-1
) 341 341 272
Mass of iron required (kg) 180 180 125
Cost of ZVI or GFH (Euro) 8.1 58.5 452.5
Mass of DML/PM or Lime (kg) 16 16 125
Cost of post treatment media (Euro) 5.5 5.5 34
Total investment of media (Euro) 13.6 64 486.2

5. Summary

5 Summary, conclusions and recommendations

5.1 Summary and conclusions

A comprehensive study, including batch kinetic, equilibrium and column configurations, for
development of the novel treatment system of stormwater runoff has been demonstrated in
this work. The batch test was carried out to understand the kinetics of metal removal under
changing runoff conditions. The equilibrium tests were tested to determine the maximum
uptake load of metals. Runoff waters from various sources including roof, street and highways
runoff were used for examination. Following the batch test, Fe
0
was evaluated under a flow-
through configuration and the behaviour of the breakthrough of the Fe
0
column was observed
at actual operating conditions. Furthermore, the mechanisms involved in the removal of
metals in the presence of NOM were studied. A modelling approach that allows for the
prediction of the breakthrough results was validated and thus allowed for further evaluation of
the up-scaling process. Finally, a method for optimizing the Fe
0
barrier and cost of the barrier
were also demonstrated. The following summarizes the significant findings of this work;

Processes of copper and zinc removal
The results of batch and column tests reveal that the removal of Cu
2+
and Zn
2+
take place
independently through different processes. The majority of Cu
2+
is removed via a reductive
transformation process and the insoluble forms of Cu
0
and Cu
2
O were found as reaction
products. The reduction rate of copper is closely linked to the surface area of Fe
0
, where
accessity of contaminant toward the Fe
0
surface plays a significant role for the process to
occur. During the course of this work, the ongoing research has revealed that reduction of
contaminants in the Fe
0
system under a natural pH range likely take places at the vicinity of
Fe
0
surface. Unlike copper, the removal of zinc involves the adsorption and co-precipitation
processes. The removal of zinc is strongly associated with the availability of freshly
precipitated iron oxides as a result of an array of reactions including iron oxidation, iron
(oxy)hydroxide precipitation and metal adsorption/co-precipitation. The adsorption/co-
precipitation processes are also believed to play a major role in a long-term treatment of both
copper and zinc, since over time, the Fe
0
surface would be passivated.

The understanding of the mechanisms involved in copper and zinc removal contributes to the
process design selection of the treatment system. The treatment barrier for zinc removal
should allow a spontaneous corrosion of iron whereby rust particles can be freely removed
from the iron surface. On the contrary, the protective film of the Fe
0
surface might be
beneficial for reductive transformation of copper at the inner layer of rust where a ferrous
form of iron is available.

Impact of water quality parameters
Investigations under various water quality conditions demonstrated that the removal of copper
and zinc is strongly impacted by water quality parameters. Generally, any factors that tend to

5. Summary 105

decrease iron corrosion rates, also hinder the removal rates of Zn
2+
. Increasing pH, DO, IS
and temperature generally favored faster retention rates of zinc in favor of production of
sorption sites. In the batch test, it was found that increasing DO and IS of the solution increase
the retention of copper but this was not the case for the flow-through column, where the
removal of copper was hindered. This discrepancy between the batch and column results were
due to differences in the experimental conditions employed and the physico-chemical
characteristics of solid precipitates involved in the system. Among different water quality
parameters investigated, NOM showed the most influence on the dramatically decrease of
metal removal ability.

Equilibrium loading
The results of batch tests demonstrated that the equilibrium loadings of copper and zinc are 2-
100 mg g
-1
Fe
0
and 0.2-100 mg g
-1
Fe
0
for copper and for zinc, respectively. These values
agreed well with column investigation revealing that about 76 mg g
-1
Fe
0
for copper (with
min: 55 and max: 96 mg g
-1
) and 55 mg g
-1
Fe
0
-1
) could be
removed. The presence of 9-61 mg L
-1
NOM concentration resulted in a much lower metal
uptake load of 42 and 22 mg g
-1
, respectively.

NOM impact on Fe
0
treatment system
Based on the findings, it was shown that NOM influences the removal of metals in Fe
0

through several complex pathways including metal-ligand complexes, competitive adsorption
and a hindrance of iron corrosion caused by NOM surface coverage of Fe
0
. The NOM fraction
that mainly interacts with the removal of metals is a larger size hydrophobic fraction with
high aromaticity. The continuous input of NOM causes a short term effect on the treatment
efficiency of the Fe
0
system, while the effectiveness over the long term cannot be concluded.
Since NOM can degrade over time, the presence of NOM might contribute a smaller impact
than expected. Moreover, the modification of the surface underlying the NOM might also
contribute to an enhancing effect of metal removal in the long term. For the toxicology
analysis and fate of the metals of runoff water, these must be assessed through the
complexation capacity and biodegradation of NOM.

Modeling
The breakthrough curves of dissolved metals in the Fe
0
system are possibly predicted by the
PSDM. In terms of application, a set of modeling parameters have been demonstrated for
further use. The parameters were assessed in runoff solutions having different concentration
ranges of NOM. The K
F
(and n) of 10-40 (0.6) for copper and 5-20(1.0) for zinc and an
average k
F
of 0.4 x 10
-3
cm s
-1
for copper and k
F
= 0.2 x 10
-3
cm s
-1
for zinc may be used for
modeling of breakthrough in typical diluted solutions of roof runoff. The application of the
models must be used with care, particularly, in circumstances where the runoff quality is
highly polluted and their impacts on the breakthroughs seem significant, then a lower K
F
may
be applied. For more precise model predictions, kinetic and equilibrium parameters should be
quantified from the solution collected onsite. It must be considered that the characteristic of
iron and treatment configuration can tremendously affect the modeling results. As an

5. Summary 106

alternative to the modeling assessment, a small scale column test may be conducted to
evaluate the performance of the treatment barrier for further up-scaling purpose.

Optimization of the barrier and final design
Optimization of the processes has been carried out using various carbonatic materials.
According to these results, the recommended system for on-site treatment of metal-
contaminated runoffs consisted of a spiral shaped iron barrier followed by an aeration unit and
post filtration of pumice/dolomite may be applied. About 60-100 and 30-80 % of metal
concentrations could be removed at EBCT of 5 and 1 min EBCT, respectively. Longer EBCT
will yield a better performance. For practical operations, carbonate materials tend to be
passivated by various minerals and biofilms. An occasional backwash in order to reactivate
the surface of materials inside and to maintain the hydraulic conductivity of the system is
required.

Investment cost
It has been shown that the investment cost of ZVICB is much lower than that of GFH/Lime
system, with increasing differences at larger scale site. The simplified cost analysis shows that
the total investment for GFH/Lime and ZVICB systems were about 0.39-0.66 and 0.01-0.62
Euro m
-2
roof surface
y
-1
, respectively, depending on the types of structures. A higher investment
cost may be expected for actual scale sites, in which the yearly operational costs such as
maintenance and replacement should be included.

5.2 Recommendations for further work

Although, the mechanisms involving copper and zinc removal by Fe
0
system have
been extensively investigated in this work, it is still not completely understood
whether Cu
2+
is reduced by Fe
0
, Fe
2+
or structured Fe
2+
. According to the results, it is
obvious that the oxide layer on the surface of iron or that is accumulated within the
reactive barrier play a significant role on the treatment ability of heavy metals. The
impact occurs both physically and chemically. In addition, it is also difficult to
distinguish between the role of reduction, adsorption and co-precipitation processes
under a wide range of conditions. Further work on these aspects may help in the
engineering design and operation of the system in the long-term.

The PSDM has been introduced in this work for prediction of breakthrough curves.
Nevertheless, the parameters obtained were mostly derived using an empirical basis to
determine the best fit. The model described the complex processes of metal removal in
the Fe
0
system by a few lump sum kinetic and equilibrium parameters, which in
terms of application, may not pass rigorously validation specifically when different
types, shapes of iron and treatment configuration is employed. With an aim to design
an up-scaled system as well as in the prediction of the long term performance of the
barrier, further work on modeling is beneficial.

5. Summary 107

In terms of applicability, the results with the carbonatic material show that an
additional ZVI barrier may not necessarily be mandatory in the treatment chain. This
would simplify the required treatment system. Nevertheless, as it was observed that
the surfaces of carbonate materials tend to be passivated over time by various minerals
and biofilms and that may reduce its effectiveness overtime. The design parameters
and long term efficiency of the carbonate barrier require further investigation.

Consider that the iron material used for the barrier may be "contaminated" itself by
heavy metals such as copper, chromium and nickel. They may be released into the
water during corrosion of iron, but may be caught by the corrosion products. The
significance and fate of these contaminations should be evaluated in future research
work. In similar regards, it is also interesting to study the remobilization of the metals
that have been previously uptaken by iron oxides.

References

References

Abbt-Braun, G., Frimmel, F. and H. Schulten (1990) Strukturelle Charakterisierung isolierter
aquatischer Huminstoffe - Anwendbarkeit, Grenzen und Vergleich ausgewhlter
Methoden. Vom Wasser ,74, 325-338.

Aiken, G.R., Mcknight, D.M., Wershaw, R.L. and P. Maccarthy (1985) Humic substances in
soil, sediment, and water: geochemistry, isolation and characterization. John Wiley & Sons,
New York, 691.

Aiken, G., McKnight, D., Thorn, K. and E. Thurman (1992) Isolation of hydrophilic organic
acids from water using nonionic macroporous resins. Org. Geochem., 18, 4, 567573.

Amy, G. (2004) Fundamental insights into natural organic matter (NOM) fouling of
membranes. Presentation at the department of water quality control, Technical University of
Berlin, Germany.

Annamalai, V. and L.E. Murr (1979) Influence of deposit morphology on the kinetics of
copper cementation on pure iron, Hydrometallurgy, 4, 57-82.

Athanasiadis, K., Helmreich B. and P. Wilderer (2003) Entfernung von Zink aus
Dachablufen eines Zinkdaches durch Klinoptilolith. Regenwasserversickerung - Eine
Mglichkeit dezentraler Regenwasserbewirtschaftung. 31. Abwassertechnisches Seminar.
Berichte aus Wassergte- und Abfallwirtschaft TU Mnchen. Hrsg.: Prof. Dr. P. Wilderer.

Atkinson, R.J., Posner, A.M., and J.P. Quirk (1967) Adsorption of potential-determining ions
at the ferric oxide-aqueous electrolyte interface. J. Phys. Chem., 71, 550-558.

Bang, S., Johnson M.D., Korfiatis G.P. and X. Meng (2005) Chemical reactions between
arsenic and zero-valent iron in water. Wat. Res., 39, 5, 763-770.

Barbosa, A.E. (1999) Highway runoff pollution and design of infiltration ponds for pollutant
retention in semi-arid climates. Ph.D.Thesis, Environmental Engineering Laboratory, Aalborg
University, Denmark.

Bardin, J.P., Cres, F.N., Touzo, A. and J.L. Verjat (1996) Pollution measurements in a
stormwater settling and infiltration basin. The 7
th
International Conference on Urban Storm
Drainage, Hanover, Germany, 3, 1611-1616.

Benjamin, M.M. (2002) Water chemistry, McGraw-Hill, Newyork.

Biswas, A.K. and J.G. Reid (1972) Investigation of the cementation of copper on iron. Proc.
Aust. Inst. Min. Met., 242, 37-45.

Blowes, D.W., Ptacek, C.J., Benner, S.G., McRae, C.W.T., Bennett, T.A. and R.W. Puls
(2000) Treatment of inorganic contaminants using permeable reactive barriers, J. Contam.
Hydrol., 45, 123-137.

References 109

Blowes, D.W. and C.J. Ptacek (1992) Geochemical remediation of groundwater by permeable
reactive walls: removal of chromate by reaction with iron-bearing solids. Subsurface
restoration conference, U.S. Environmental protection agency, Kerr Laboratory, Dallas, TX,
214-216.

Boller, M.A. (1997) Tracking heavy metals reveals sustainability deficits of urban drainage
systems. Wat. Sci. Tech., 35, 9, 77-87.

Boller, M.A. and M. Steiner (2002) Diffuse emission and control of copper in urban surface
runoff. Wat. Sci. Tech., 46, 6-7, 173-181.

Bundesbodenschutzgesetz (BBodSchG) vom 17.03.1998. BGBl. 1998, Teil 1, Nr. 16.

Dannecker, W., Au, M. and H. Stechmann (1990) Substance load in rainwater runoff from
different streets in Hamburg. Sci. Total Environ., 93, 385-392.

Cantrell, K.J., Kaplan, D.I. and T.W. Wietsma (1995) Zero-valent iron for the in situ
remediation of selected metals in groundwater. J. Hazard. Mat., 42, 2, 201-212.

Cheng, T. and H.E. Allen (2006) Comparison of zinc complexation properties of dissolved
natural organic matter from different surface waters. J. Environ. Management., 80, 222229.

Cornell, R.M. and U. Schwertmann (1996) In: Bck, B. (Ed.), The iron oxides: structure,
properties, reactions, occurrence and uses. VCH Press, Weinheim, Germany.

Cravotta, C.A. and G.R. Watzlaf (2002) in Handbook of groundwater remediation using
permeable reactive barriers: applications to radionuclides, trace metals, and nutrients (editor:
Naftz, D.L., Morrison, S.J., Fuller, C.C., Davis, J.A.), Elsevier Science, Amsterdam.

Crawford, R.J., Harding, I.H., and D.E. Mainwaring (1993) Adsorption and coprecipitation of
single heavy metal ions onto the hydrated oxides of iron and chromium. Langmuir, 9, 3050-
3056.

Crittenden, J.C., Luft, P. and D.W. Hand (1985) Prediction of multicomponent adsorption
equilibria in background mixtures of unknown composition, Wat. Res. 19,12, 1537-1548.

Crittenden, J.C., Luft, P., Hand, D.W., Oravitz, J.L., Loper, S.W. and M. Ari (1985)
Prediction of multicomponent adsorption equilibria using ideal adsorbed solution theory,
Environ. Sci. Technol. 19,11,1307-1043

Crou, J. (2002) Approaches to characterizing NOM. Conference proceedings: NOM1-
Characterisation and treatment of natural organic matter, School of Water Sciences,
Cranfield University, U.K.

Chi, F.H. and G.L. Amy (2004) Kinetic study on the sorption of dissolved natural organic
matter onto different aquifer materials: the effects of hydrophobicity and functional groups. J.
Colloid Interf. Sci., 274, 380391.

Davis, J.A. and J.O. Leckie (1978) Effect of adsorbed complexing ligands on trace metal
uptake by hydrous oxides. Environ. Sci. Technol., 12, 1309-1315.

References 110

Denver, J.I. (1997) The geochemistry of natural waters. Surface and groundwater
environments, 3
rd
edition-Prentice Hall, New Jersey.

Dries, J., Bastiaens. L., Springael, D., Kuypers, S., Agathos, S.N. and L. Diels (2005) Effect
of humic acids on heavy metal removal by zero-valent iron in batch and continuous flow
column systems. Wat. Res., 39, 35313540.

Driehaus, W., Jekel, M. and U. Hildebrandt (1998) Granular ferric hydroxide a new
adsorbent for the removal of arsenic from natural water. J. Water SRT- Aqua., 47 (1), 1-6.

Dzombak, D. A. and F.M. Morel (1990) Surface complexation modeling: hydrous ferric
oxide, John Wiley & Sons, New York.

Farrell, J., Kason, M., Melitas, N. and T. Li (2000) Investigation of the long-term
performance of zero-valent iron for reductive dechlorination of trichloroethylene. Environ.
Sci. Technol., 34, 514-521.

FHWA (1996) Federal highway administration, evaluation and management of highway
runoff water quality, Report No. FHWA-PD-96-032, U.S. Department of Transportation.

Feitknecht, W. and P. Schindler (1963) Solubility constants of metal oxides, metal hydroxides
and metal hydroxide salts in aqueous solution. Pure and Applied Chemistry, 6, 132-199.

Fiedor, J.N., Bostick, W.D., Jarabek, R.J. and J. Farrell (1998) Understanding the mechanism
of uranium removal from groundwater by zero-valent iron using X-ray photoelectron
spectroscopy. Environ. Sci. Technol., 32, 1466-1473.

Frster, J. (1993) Dachflchen als Interface zwischen atmosphrischer Grenzschicht und
Kanalsystem: Untersuchungen zum Transportverhalten ausgewhlter organischer
Umweltchemikalienan einem Experimentaldachsystem. Dissertation an der Universitt
Bayreuth.

Freundlich, H. (1906) ber die Adsorption in Lsungen. Zeitschrift fr physik. Chemie 57,
385-470.

Friedman, G. (1984) Mathematical modelling of multicomponent adsorption in batch and
fixed-bed reactors, Mater Thesis, Michigan Technological University.

Furukawa, Y., Kim, J.-W., Watkins, J. and R.T. Wilkin (2002) Formation of ferrihydrite and
associated iron corrosion products in permeable reactive barriers of zero-valent iron. Environ.
Sci. Technol., 36, 5469-5475.

Genz, A., Baumgarten, B. and M. Jekel (2004) A new treatment process for removal of NOM
by adsorption and regeneration of granular ferric hydroxide (GFH). WEMS, Water
Environmental Management Series (2
nd
IWA Leading Edge Conference of Water and
Wastewater Treatment Technolgies), 107-116.

Ghauch, A., Assi, H.A. and A. Tuqan (accepted) Investigating the mechanism of clofibric
acid removal in Fe
0
/H
2
O systems, J. Hazard. Mater.

References 111

Gibert, O, de Pablo, J., Cortina, J.L. and C. Ayora (2003) Evaluation of municipal
compost/limestone/iron mixtures as filling material for permeable reactive barriers for in-situ
acid mine drainage treatment. J. Chem. Technol. Biot., 78, 489-496.

Gieska, M., Tanneberg H. and R.R. van der Ploeg (2000) Lokal erhhte
Schwermetallkonzentrationen in urbanen Bden durch Versickerung von Dachabflssen.
Wasser und Boden, 52, 3, 41-45.

Gillham, R.W. and S.F. OHannesin (1992) Metal-catalyzed abiotic degradation of
halogenated organic compounds. In Modern Trends in Hydrogeology, IAH conference,
Hamilton, Ontario, Canada.

Gillham, R.W. and S.F. OHannesin (1994) Enhanced degradation of halogenated aliphatics
by zero-valent iron. Ground Water, 32, 958-67.

Gromaire-Mertz, M.C., Garnaud S., Gonzales A. and G. Chebbo (1999) Characterisation of
urban runoff pollution in Paris. Wat. Sci. and Tech. 39, 2, 1-8.

Gu, B., Chen, J., Vairavamurthy, M.A. and S. Choi (2000) Chemical and biological reduction
of contaminant metals by natural organic matter. Extended abstract submitted to symposium
on "Chemical-biological interactions in contaminant fate", the 220
th
American chemical
society national meeting, 20-25 August, Washington, DC.

Gu, B., Liang, L., Dickey, M.J., Yin, X. and S. Dai (1998) Reductive precipitation of uranium
(VI) by zero-valent iron. Environ. Sci. Technol., 32, 3366-3373.

Gu, B., Schmitt, J., Chen, Z., Liang, L. and J. McCarthy (1994) Adsorption and desorption of
natural organic matter on iron oxide: Mechanisms and models. Environ. Sci. Technol., 28, 38-
46.

Gu, B., Schmitt, J., Chen, Z., Liang, L. and J. McCarthy (1995) Adsorption and desorption of
different organic matter fractions on iron oxide. Geochim. Cosmochim. Acta, 59, 2, 219-229.

Hamann, C.H. and W. Vielstich (1975) Elektrochemie I und II. Taschentexte Nr. 41 und 42.
Verlag Chemie, Weinheim.

Hamilton, R.S., Revitt, D.M., Warren, R.S. and M.J. Duggan (1987) Metal mass balance
studies within a small highway dominated catchment, Sci. Total Environ., 59, 365-368.

He, W., Wallinder, I.O. and C. Leygraf (2001) A laboratory study of copper and zinc runoff
during first flush and steady-state conditions, J. Corros. Sci., 43, 1, 127-146.

Hedin, R.S., Watzlaf, G.R. and R.W. Nairn (1994) Passive treatment of acid mine drainage
with limestone. J. Environ. Qual., 23, 1338-1345.

Hesse, O. (2005) Untersuchung von Regenwasser und Ablufen von Kupferdchern und
Aufzeigen von Mglichkeiten zur Verminderung schdlicher Einflsse auf die Umwelt am
Beispiel einer Therme in Belzig, Diplomarbeit an der Technische Universitt Berlin, Fakultt
III Institut fr Technischen Umweltschutz Fachgebiet Siedlungswasserwirtschaft.

References 112

Hewitt, C.N. and M.B. Rashed (1992) Removal rates of selected pollutants in the runoff
waters from a major rural highway, Water Res., 26, 3, 311-319.

Huang, Y.H. and T.C. Zhang (2005) Effects of dissolved oxygen on formation of corrosion
products and concomitant oxygen and nitrate reduction in zero valent iron systems with or
without aqueous Fe
2+
. Wat. Res., 39, 1751-60.

Huber, S. and F. Frimmel (1996) Size-exclusion chromatography with organic carbon
detection (LC-OCD): a fast and reliable method for the characterization of hydrophilic
organic matter in natural waters. Vom Wasser, 86, 277-290.

Hvitved-Jacobsen, T. and Y.A. Yousef (1991) Highway runoff quality, environmental impacts
and control in Highway Pollution, Studies in Environmental Science 44, R.S. Hamilton and
R.M. Harrison eds., Elsevier, 166-208.

Hvitved-Jacobsen, T., Johansen, N.B. and Y.A. Yousef (1994) Treatment systems for urban
and highway runoff in Denmark, Sci. Tot. Environ., 146/147, 499-506.

Jarrusutthirak, C. (2002) Fouling and flux decline of reverse osmosis (RO), nanofiltration
(NF) and ultrafiltration (UF) membranes associated with effluent organic matter (EFOM)
during wastewater reclamation/reuse. PhD Thesis, University of Colorado at Boulder.

Jia, Y., Aagaard P. and G.D. Breedveld (2007) Sorption of triazoles to soil and iron minerals.
Chemosphere, 67, 250-58.

Kamolpornwijit, W., Liang, L., Westa, O.R., Molinea, G.R. and A.B. Sullivana (2003)
Preferential flow path development and its influence on long-term PRB performance: column
study, J. Cont. Hydrol., 66, 161 178.

Kamolpornwijit, W., Liang, L., Moline, G.R., Hart, T. and O.R. West (2004) Identification
and quantification of mineral precipitation in Fe filings from a column study. Environ. Sci.
Technol., 38, 5757-5765.

Karschunke, K. and M. Jekel (2002) Arsenic removal by iron hydroxides, produced by
enhanced corrosion of iron, Water Supply, 2, 2, 237245.

Karschunke, K. (2005) Nutzung der Eisenkorrosion zur Entfernung von Arsen aus
Trinkwasser. Dissertation an der Fakultt III der TU Berlin.

Kieber, R.J., Skrabal, S.A., Smith, B.J. and J.D. Willey (2005) Organic complexation of
Fe(II) and its impact on the redox cycling of iron in rain. Environ. Sci. Technol. 39, 1576-
1583.

Kohn, T., Livi, K.J.T., Roberts, A.L. and P.J. Vikesland (2005) Longevity of granular iron in
groundwater treatment processes: corrosion product development. Environ. Sci. Technol., 39,
8, 2867-2879.

Korshin, G.V., Benjamin, M.M. and R.S. Sletten (1997) Adsorption of natural organic matter
(NOM) on iron oxide: effects on NOM composition and formation of organo-halide
compounds during chlorination. Wat. Res., 31, 7, 1643-1650.

References 113

Komnitsas, K., Bartzas, G., Fytas, K. and I. Paspaliaris (2007) Long-term efficiency and
kinetic evaluation of ZVI barriers during clean-up of copper containing solutions, Minerals
Engineering, 20, 200-1209.

Kouznetsova, I., Bayer, P., Ebert, M. and M. Finkel (2007) Modelling the long-term
performance of zero-valent iron using a spatio-temporal approach for iron aging. J. Cont.
Hydrol., 90, 1/2, 58-80.

Ku, Y. and C.H. Chen (1992) Kinetic study of copper deposition on iron by cementation
reaction. Sep. Sci. and Tech., 27, 10, 1259-1275.

Kuch, A. (1984) Untersuchungen zum Mechanismus der Aufeisenung in
Trinkwasserverteilungssystemen. Ph.D. Dissertation, University of Karlsruhe.

Langmuir, I. (1918) The adsorption of gases on plane surfaces of glass, mica and platinum. J.
Amer. Chem. Soc., 40, 1361-1403.

Laxen, D.P.H. (1985) Trace metal adsorption/coprecipitation on hydrous ferric oxide under
realistic conditions. Wat. Res., 19, 10, 1229-1236.

Lebeth, F. (1999) Erfordernisse des Gewsserschutzes bei der Versickerung von
Dachflchenwssern. sterreichische Wasser- und Abfallwirtschaft, 51, 1/2, 18-28.

Lee, N., Amy, G. Crou, J.-P. and H. Buisson (2004) Identification and understanding of
fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM). Wat.
Res., 38, 4511-4523.

Leenheer, J. A. (2004) Comprehensive assessment of precursors, diagenesis, and reactivity
towater treatment of dissolved and colloidal organic matter. Conference proceedings: NOM
research: Innovation and applications for drinking water treatment, Victor Harbor, Australia.

Lygren, E., Gjessing, E. and L. Berglind (1984) Pollution transport from a highway, Sci. Total
Environ., 33, 147-159.

Liang, L., Sullivan, A.B., West, O.R., Moline, G.R. and W. Kamolpornwijit (2003) Predicting
the precipitation of mineral phases in permeable reactive barriers. Environ. Eng. Sci., 20, 6,
635-653.

Liang, L., McNabb, J.A., Paulk, J.M., Gu, B. and J.F. McCarthy (1993) Kinetics of Fe(II)
oxygenation at low partial pressure of oxygen in the presence of natural organic matter.
Environ. Sci. Technol., 27, 1864-1870.

Lpez, F.A., Martn, M.I., Prez, C., Lpez-Delgado, A. and F.J. Alguacil (2003) Removal of
copper ions from aqueous solutions by a steel-making by-product. Wat. Res., 37, 16, 3883-
3890.

Lu, Y. and H.E. Allen (2002) Characterization of copper complexation with natural dissolved
organic matter (DOM) link to acid moieties of DOM and competition by Ca and Mg. Wat.
Res., 36, 5083-5101.

References 114

Ludwig, T. (2004) Adsorption isotherms of Cu
2+
and Zn
2+
on granular ferric hydroxide
(GFH). Unpublished data, department of water quality control, Technical University of
Berlin, Germany.

Ludwig, T. (2007) Kupfer- und Zinkentfernung aus Niederschlagsabfluss von Dchern in
einem Eisen-Korrosionssystem (in English: Removal of copper and zinc from roof runoff in
an iron-corrosion system). Ph.D. Dissertation, Institute of process engineering, Technical
University of Berlin, Germany.

Lung, T.N. (1986) The history of copper cementation on iron-the worlds first
hydrometallurgical process from medieval china. Hydrometallurgy, 17, 113-129.

Ma, H., Allen, H.E and Y. Yin (2001) Characterization of isolated fractions of dissolved
organic matter from natural waters and a wastewater effluent. Wat. Res., 35, 985-996.

MacCarthy, P., Mark, H.B., Jr., and P.R. Griffiths (1975) Direct measurement of the infrared
spectra of humic substances in water by fourier transform infrared spectroscopy. J. Agr. Food
Chem. 23, 600-602.

Maike, M. (2005) Method for fractionation of NOM, Chair Waste Management and
Environmental Research, Technical Universtiy of Berlin.

Maithreepala, R.A. and R.-A. Doong (2004) Reductive dechlorination of carbon tetrachloride
in aqueous solutions containing ferrous and copper ions. Environ. Sci. Technol., 38, 260-268.

Mann, A.W. and D.L. Deutscher (1980) Solution geochemistry of lead and zinc in water
containing carbonate, sulphate, and chloride ions. Chem. Geol., 29, 293-311.

Manning, B.A., Hunt M.L., Amrhein C. and J.A. Yarmoff (2002) Arsenic(III) and
arsenic(V) reactions with zerovalent iron corrosion products. Environ. Sci. Technol., 36, 24,
5455-5461.

Meihr, R., Tratnyek, P.G., Bandstra, J.Z., Scherer, M.M., Alowitz, M.J. and E.J. Bylaska
(2004) Diversity of contaminant reduction reactions by zero-valent iron: role of reductate.
Environ. Sci. Technol., 38, 139-147.

Merkel, T. (2002) Untersuchungen zu den chemischen Reaktionen bei der Flchenkorrosion
des Kupfers in Trinkwasserinstallationen (in English: Investigation of the chemical corrosion
of copper surface in drinking water installation). Ph.D. Dissertation, University of Karlsruhe,
Germany.

Merkel, B. J. and B. Planer-Friedrich (2005) Groundwater Geochemistry: A practical guide to
modeling of natural and contaminated aquatic systems, (Edited by Darrell Kirk Nordstorm),
Springer-Verlag, Berlin Heidelberg.

Mertz, K.A., Gobin, F., Hand, D.W., Hokanson, D.R. and J.C. Crittenden (1999) Manual
adsorption design software for windows (AdDesignS
TM
), Michigan technological university.

Morel, F.M.M., and J.G. Hering (1993) Principles and applications of aquatic chemistry.
Wiley-Interscience, New York.

References 115

Morrison, S.J. Metzler, D.R. and B.P. Dwyer (2002) Removal of As, Mn, Mo, Se, U, V and
Zn from groundwater by zero-valent iron in a passive treatment cell: reaction progress
modeling. J. Contam. Hydrol., 56, 99-116.

Mullet, M., Fievet, P., Reggiani, J.C. and J. Pagetti (1997) Surface electrochemical properties
of mixed oxide ceramic membranes: zeta-potential and surface charge density. J. Membrane
Science., 123, 255-265.

Mustafa, S., Dilara, B., Nargis, K., Naeem, A. and P. Shahida (2002) Surface properties of the
mixed oxides of iron and silica. Colloids. Surface., A 205, 273-282.

Nadkarni, R.M., Jelden, C.E., Bowles, K.C., Flanders, H.E. and M.E. Wadsworth (1967) A
kinetic study of copper precipitation on iron: part I. Trans. Met. Soc. AIME., 239, 581-585.

Nadkarni, R.M. and M.E. Wadsworth (1967) A kinetic study of copper precipitation on iron:
part II. Trans. Met. Soc. AIME, 239, 1066-1074.

Naftz, D.L., Morrison, S.J., Fuller, C.C. and J.A. Davis (2002) Handbook of groundwater
remediation using permeable reactive barriers: applications to radionuclides, trace metals, and
nutrients. Elsevier Science.

Nikolaidis, P.N., Dobbs, G.M. and J.A. Lackovic (2003) Arsenic removal by zero-valent iron:
field, laboratory and modelling studies, Wat. Res., 37, 1417-1425.

Noubactep, C., Meinrath, G., Dietrich, P. and B. Merkel (2003) Mitigating uranium in
groundwater: prospects and limitations. Environ. Sci. Technol., 37, 4304-4308.

Noubactep, C. (2008a) Processes of contaminant removal in Fe
0
-H
2
O systems revisited: the
important of co-orecipitation, The open Environmental Journal, 1, 9-13.

Noubactep, C. (2008b) A critical review on the process of contaminant removal in Fe
0
-H
2
O
systems, Environ. Technol., 29, 909-920.

Oden, W. I., Amy, G.L. and M. Conklin (1993) Subsurface interactions of humic substances
with Cu(II) in saturated media. Environ. Sci. Technol., 27, 6, 1045-1051.

Odziemkowsi, M.S., Schumacher, T.T., Gillham, R.W. and E.J. Readon (1998) Mechanism of
oxide film formation on iron in simulating groundwater solutions: raman spectroscopic
studies. Corrosion Science., 40, 2/3, 371-389.

Perdue, E.M. (1985) Acidic functional groups of humic substances - in humic substances in
soil, sediment, and water: geochemistry, isolation and characterization. John Wiley and Sons,
New York NY.

Phillips, D.H., Gu, B., Watson, D.B., Roh, Y., Liang, L. and S.Y. Lee (2000) Performance
evaluation of a zerovalent iron reactive barrier: mineralogical characteristics. Environ. Sci.
Technol., 34, 4169-4176.

References 116

Pitt, R. (1996) Groundwater contamination from stormwater infiltration, Ann Arbor Press,
Inc., Chelsea, Michigan.

Polkowska, Z., Grecki, T., and J. Namiesnik (2002) Quality of roof runoff waters from an
urban region (Gdansk, Poland). Chemosphere, 49, 1275-1283.

Priggemeyer, S. and S. Priggemeyer (1999) Metallkonzentrationen in Dachablaufwssern.
Metall, 53, 4, 204-205.

Qiu, S.R., Lai, H.-F. and M.J. Roberson (2000) Removal of contaminants from aqueous
solution by reaction with iron surfaces. Langmuir, 16, 2230-2236.

Rangsivek, R. and M.R. Jekel (2005) Removal of dissolved metals by zero-valent iron (ZVI):
Kinetics, equilibria, processes and implications for stormwater runoff treatment. Wat. Res.,
39, 4153-4163.

Rangsivek, R. and M.R. Jekel (2007) Development of an on-site Fe
0
system for treatment of
copper- and zinc- contaminated roof runoff. accepted for publication in International Journal
of Environment and Waste Management (IJEWM): Metal ions removal from liquid
effluents.

Rangsivek, R. and M.R. Jekel (2008) Natural organic matter (NOM) in roof runoff and its
impact on the Fe
0
treatment system of dissolved metals, Chemosphere, 71, 18-29.

Reardon, E.J. (1995) Anaerobic corrosion of granular iron: measurement and interpretation of
hydrogen evolution rates, Environ. Sci. Technol., 29, 2936-2945.

Revitt, D.M., Hamilton, R.S. and R.S. Warren (1990) The transport of heavy metals within a
small urban catchment. Sci. Total Environ., 93, 329-337.

Robertson, W.D. and J.A. Cherry (1995) In instu denitrification of septic-system nitrate using
reactive porous media barriers: Field trials. Ground Water, 33, 99-111.

Rodda, D.P., Johnson, B.B., and J.D. Wells (1993) The effect of temperature and pH on the
adsorption of Copper(II), Lead(II) and Zinc(II) onto goethite. J. Colloid Interf. Sci., 161, 57-
62.

Roh, Y., Lee, S.Y., and MP. Elless (2000) Characterization of corrosion products in the
permeable reactive barriers, Environ. Geol.,40, 1-2, 184-194.

Senesi, N. (1990). Molecular and quantitative aspects of the chemistry of fulvic acid and its
interactions with metal ions and organic chemicals. Part II. The fluorescence spectroscopy
approach. Analytica Chimica Acta, 232, 77-106.

Sigg, L. and W. Stumm (1996) Aquatische Chemie. B.G. Teubner, Stuttgart.

Schecher, W.D. and D.C. McAvoy (2001) Thermodynamic data used in MINEQL
+
Version
4.5. Environmental Research Software. Hallowell ME 04347.

References 117

Simon, F., Meggyes, T., Tunnerrneier, T., Czurda, K., and K.E. Roehl (2001) Long-term
behaviour of permeable reactive barriers used for the remediation of contaminated
groundwater. In 8
th
International Conference on Radioactive Waste Management and
Environmental Remediation, Sep 30Oct 4, Bruges, Belgium: American Society of
Mechanical Engineers.

Shokes, T.E. and G. Mller. (1999) Removal of dissolved heavy metals from acid rock
drainage using iron metal. Environ. Sci. Technol., 33, 282-287.

Smith, E.H. (1996) Uptake of heavy metals in batch systems by a recycled iron-bearing
material. Wat. Res., 30, 10, 2424-2434.

Smith, E.H. (1995) A recycled iron material for fixed-bed adsorption of heavy metals from
industrial wastewaters. Proceedings, 68
th
Water Environ. Fed. Annual Conf., Water Environ.
Fed., Washington, DC, 3, 559-570.

Smith, E.H. (1998) Surface complexation modelling of metal removal by recycled iron
sorbent, J. of Environ. Eng., 124, 10, 913-920.

Smith, S.E., Bisset, A., Colbourne, J.S., Holt, D.M. and B.J. Lloyd (1997) The occurrence
and significance of particles and deposits in a drinking water distribution system. Journal of
the New England Water Works Association, 111, 2, 135

Sontheimer, H., Crittenden, J.C. and R.S. Summers (1988) Activated carbon for water
treatment. DVGW-Forschungsstelle am Engler-Bunte-Institut der Universitt Karlsruhe,
Karlsruhe. Second edition in english of adsorptionsverfahren zur wasserreinigung.

Specht, C.H., Kumke, M.U. and F.H. Frimmel (2000) Characterization of NOM adsorption to
clay minterals by size exclusion chromatography. Wat. Res., 34,16, 4063-4069.

Sperlich, A., Werner, A., Genz, A., Amy, G., Worch, E. and M. Jekel (2005) Breakthrough
behavior of granular ferric hydroxide (GFH) fixed-bed adsorption filters: modeling and
experimental approaches, Wat. Res., 39, 6, 1190-1198.

Sperlich, A., Schimmelpfenning, S., Baumgarten, B., Genz, A., Amy, G., Worch, E. and M.
Jekel (2008) Predicting anion breakthrough in granular ferric hydroxide (GFH) adsorption
filters. Wat. Res., 42, 2073-2082.

Steiner, M. (2003) Adsorption von Kupfer aus Niederschlagsabflssen an granuliertes
Eisenhydroxid. PhD. Dissertation, Swiss Federal Institute of Technology Zurich (ETHZ),
Switzerland.

Steiner, M., Pronk, W. and M.A. Boller (2006) Modeling of copper sorption on to GFH and
design of full-scale GFH adsorbers, Environ. Sci. Technol., 40, 1629-1635.

Stevenson, F.J. (1994) Humus chemistry: genesis, composition and reactions. John Wiley &
Sons.

References 118

Stotz, G. (1987) Investigations of the properties of the surface water run-off from federal
highways in the FRG, Sci.Total Environ., 59, 359-373.

Strickland, P.H. and F. Lawson (1971) The cementation of metals from dilute aqueous
solutions. Proc. Aust. Inst. Min. Met., 237, 71-78.

Stumm, W. (1957) Corrosion studies. Public Works, 88, 12, 78.

Stumm, W. (1959) Estimating corrosion rates in water. Ind. and Eng. Chem., 51,12, 1487.

Su, C. and R.W. Puls (2004) Significance of iron(II,III) hydroxycarbonate green rust in
arsenic remediation using zerovalent iron in laboratory column tests. Environ. Sci. Technol.,
38, 19, 5224 - 5231.

Teermann, I. and M. Jekel (1999) Adsorption of humic substances onto -FeOOH and its
chemical regeneration. Wat. Sci. Tech., 40, 9, 199-206.

Tipping, E., and D. Cooke (1982) The effects of adsrobed humic substances on the surface
charge of goethite (-FeOOH) in freshwaters. Geochim. Cosmochim. Acta, 46, 75-80.

Tombcz, E., Zsuzsanna, L., Ills, E., Majzik, A. and E. Klumpp (2004) The role of reactive
surface sites and complexation by humic acids in the interaction of clay mineral and iron
oxide particles. Org. Geochem. 35, 257-267.

Tombre, M.S., Thomson, B.M. and L.L. Barton (1997) Use of a permeable biological reaction
barrier for groundwater remediation at a uranium mill tailings remedial action (UMTRA) site.
In International containment technology conference prodeedings Febuary 9-12, St.
Petersburg, FL., pp. 744-750.

Tratnyek, P.G., Scherer, M.M., Deng, B. and S. Hu (2001) Effects of natural organic matter,
anthropogenic surfactants, and model quinines on the reduction of contaminants by zero-
valent iron. Wat. Res. 35, 18, 4435-4443.

Tyrovola, K., Peroulaki, E. and N.P. Nikolaidis (2007) Modelling of arsenic immobilization
by zero valent iron. Eur. J. Soil Biol., 43, 356-367.

Voskamp, M., Abbt-Braun, G. and F.H. Frimmel (2005) Metallische Dachmaterialien als
Quelle fr die Schwermetalle Kupfer und Zink im Regenwasserablauf. Vortrag auf der 71.
Jahrestagung der Wasserchemischen Gesellschaft in Bad Mergentheim.

Wang, Y. and K. Salvage (2005) Immobilization of uranium in the presence of Fe
0(s)
: model
development and simulation of contrasting experimental conditions. Appl. Geochem., 20,
1268-1283.

Wilkin, R.T. and M.S. McNeil (2003) Laboratory evaluation of zero-valent iron to treat water
impacted by acid mine drainage. Chemosphere, 53, 715-725.

Wong, S., Hanna, J.V., King, S., Carroll, T.J., Eldridge, R.J., Dixon, D.R., Bolto, B.A., Hesse
S., Abbt-braun, G. and F.H. Frimmel (2002) Fractionation of natural organic matter in

References 119

drinking water and characterization by 13C cross-polarization magic-angle spinning NMR
spectroscopy and size exclusion chromatography. Environ. Sci. Technol. 36, 3497-3505.

Yoon, Y., Amy, G., Cho, J. and N. Her (2005) Effect of retained natural organic matter on
NOM rejection and membrane flux decline with nanofiltration and ultrafiltration.
Desalination, 173, 209-221.

Appendix A

Appendix A: Relevant information

Table A1 Physical and chemical data of iron, copper and zinc.
[Ref: http://www.lenntech.com/periodic/elements; Ludwig, 2007]
Chemical properties unit iron copper Zinc
Atomic number 26 29 30
Atomic mass g mol
-1
55,85 63,55 65,37
Density kg L
-1
7,8 8,92 7,11
Melting point C 1536 1083 420
Boiling point C 2861 2595 907
Oxidation number 0, 2, 3 0, 1, 2 0, 2
Electronic shell [Ar]3d
6
4s
2
[Ar] 3d
10
4s
1
[Ar]3d
10
4s
2

Energy of first ionization kJ mol
-1
761 743,5 904,5
Vanderwaals radius nm 0,126 0,128 0,138
Ionic radius nm 0,076 (+2) 0,069(+3) 0,074(+2)
0,064(+3) 0,096(+1)
Standard potentials mV -440(Fe
2+
/Fe
0
) +522(Cu
+
/Cu
0
) -763(Zn
2+
/Zn
0
)
+770(Fe
3+
/Fe
2+
) +345(Cu
2+
/Cu
0
)

Table A2 Concentration of copper and zinc in roof, street and highway runoff samples*
Year City Characteristic Copper conc.
[mg L
-1
]
Zinc conc.
[mg L
-1
]
Reference
1988 Bayreuth Zinc roof 10-85 Frster (1993)
1991 Wien N.A. 0.2 1.5 Lebeth (1999)
1997 Osnabrck Copper roof 0.5-33 Piggemeyer et al. (1999)
1998 Paris Zinc roof 0.003-0.25 0.8-38 Gromaire et al. (1999)
2000 Halle N.A. 0.03 2.58 Gieska et al. (2000)
2001 Dbendorf Copper roof 0.8-10 Steiner (2003)
Copper faade 0.9-100
2001 Mnchen Zinc roof 5-23 Athanasiadis et al. (2003)
2004 Karlsruhe Copper and 4-13 4-32 Voskamp et al. (2005)
Zinc roof
2005 Belzig Copper roof 1.52-7.61 0.04-0.25 Olaf (2005)
2008 Berlin Copper and zinc roofs 4.9-9.1 0.8-6 Rangsivek and Jekel (2008)
1994 Denmark Urban runoff 0.005-0.04 0.3-0.5 Hvitved-Jacobsen (1994)
Highway runoff 0.125-0.4 Barbosa (1999)
* Parts of data were complied by Ludwig (2007)

Table A3 Concentration of metals in highway runoffs (data complied by Barbosa, 1999)
Country Site characterisitcs ADT Lead Zinc Copper References
[g L
-1
] [g L
-1
] [g L
-1
]
Germany A81-Pleidelsheim 41000 202 360 97 Stotz (1987)
A6-Oberaisesheim 47000 245 620 117
A8/B10; Ulm/West 40600 163 320 58
Norway E6 Jessheim 8000 62-690 91-740 10-430 Lygren et al. (1984)
UK Road in residential area 720 28.10 16.6 6.5 Hamiton et al. (1987)
USA Maitland Interchange - 30-379 13-173 10-101 Hvitved-Jacobsen and Yousef (1991)
UK M6 motorway 37600 181 63 Revitt et al. (1990)
Germany Street in residential area 500 122 165.6 75.9 Dannecker et al. (1990)
UK Chilwell Gardens 150 0.6-150.5 0.7-65.3 0.1-14.3 Hewitt and Rashed (1992)
France Urban and industrial area, Lyon - 5.0-90.0 177-68.1 9.0-49.0 Bardin et al. (1996)

Appendix A 121

Table A4 Defined processes of contaminant removal under various experimental conditions
S
G - simulated groundwater; T tabwater; S simulated water; AMD acid mine drainage;
S
AMD simulated AMD; R runoff water;
S
R simulated R
L laboratory; B batch study; C column study
Processes Matrix System Components conc. Fe
0
conc. T pH
i
Condition Uptake load Refs.
[mg L
-1
] [g L
-1
] [m
2
L
-1
] [C] [-] [mg g Fe
0
]
Reduction
S
G L, B UO
2
2+
0-18,000 200 16-196 - 5 50 1
Co-precipitation T L, B U
6+
20 15 - 20 7.2 - - 2
Adsorption, Reduction

S L, B U
6+
6-8 ~5/9 in dia.
1/16 in. thick
6
5.7
Oxic,
Anoxic
-
-
3
Adsorption, Reduction S SS U
6+
/Cr
6+
/Se
6+
- Iron foil - - 8-8.5 Anoxic - 4
Adsorption, Co-precipitation AMD L, B As, Ni, Cu, Hg,
Mn, Zn
5-65 20-230 1-700 22 2.5-
4.5
- - 5
S
AMD L, C Zn, As, Cu, Cd 2-20 - - 23 3 - - 6
S
G L, B, A Cu 2,000 100 0.43 20 5 0.1 7
80 40
Reduction S L, B Cu 100 1-2.5 - 15-25 2-8 Anoxic - 8
Adsorption S L, B Zn 0.1-0.4 0.05-0.8 1.3 - 4 13 9
7 17
Adsorption, Reduction AMD L, B+C Cu 150 33 0.005 25 2.1 4.4 10
Zn 400 0.4
Co-precipitation S L, B Clofibric acid 10 15, 40 23 4.0 Oxic, Anoxic 11
Adsorption T L, B, C As 0.62 m
2
g
-1
20-25 7.5 Oxic, 4-60 12
Adsorption
s
R, R L,C Cu, 2 20 0.058 m
2
17-29 5.3-8 Oxic, 50-292 13
Zn 2 35-170
Reduction S L, C Cu 5,
50
1050 g 0.048 m
2
g
-1
24 2.5-4.5 - 0.93-2.68
6.03-13.33
14
Adsorption
s
G L, B Zn, Ni, Cr(VI) 5 2-6 1.5-4.5 - 7.0 Anoxic 0.69-2.79 15
C 5-10 102 76 m
2
2.0
Ref:
1
Gu et al. (1998);
2
Noubactep et al. (2003);
3
Fiedor et al. (1998);
4
Qui et al. (2000);
5
Wilkin and McNeil (2003);
6
Gibert et al. (2003);
7
Lopez et al. (2003);
8
Ku and Chen (1992);
9
Smith (1996);
10
Shokes and Moller (1999);
14
Ludwig (2007);
11
Ghauch et al. (accepted);
12
Karshunke (2005);
13
Ludwig (2007);
14
Komnitsas et al (2007);
15
Dries et al (2005)

Appendix B

Appendix B: Supplement data for tables and figures
[The usage of ,and . in appendix B is based on the European numerical system]

PZC
by titration method (Fig. 3.2)
Electrolite 0.01 M NaNO3 0.051045
Tritrants 0.2 g
mL 0.05M mol/dm
-3
M
NaOH added pH Acid Base mol l
-1
NO
3-
Na
+
OH
-
H
+

0 0 2,44 0,005 0,00000 0,001 0,00000 0,00000 0,00363 0,00137 0,58716
0,12 0,07 2,45 0,005 0,00009 0,001 0,00109 0,00000 0,00355 0,00136 0,58430
0,19 0,14 2,46 0,005 0,00018 0,001 0,00118 0,00000 0,00347 0,00135 0,58063
0,26 0,21 2,46 0,005 0,00027 0,001 0,00127 0,00000 0,00347 0,00126 0,54232
0,32 0,27 2,47 0,005 0,00034 0,001 0,00134 0,00000 0,00339 0,00127 0,54333
0,38 0,33 2,48 0,005 0,00042 0,001 0,00142 0,00000 0,00331 0,00127 0,54357
0,44 0,39 2,49 0,005 0,00050 0,001 0,00150 0,00000 0,00324 0,00127 0,54306
0,51 0,46 2,5 0,005 0,00059 0,001 0,00159 0,00000 0,00316 0,00125 0,53634
0,58 0,53 2,51 0,005 0,00068 0,001 0,00168 0,00000 0,00309 0,00123 0,52890
0,64 0,59 2,52 0,005 0,00075 0,001 0,00175 0,00000 0,00302 0,00123 0,52623
0,7 0,65 2,53 0,005 0,00083 0,001 0,00183 0,00000 0,00295 0,00122 0,52288
0,77 0,72 2,54 0,005 0,00092 0,001 0,00192 0,00000 0,00288 0,00120 0,51338
0,84 0,79 2,55 0,005 0,00101 0,001 0,00201 0,00000 0,00282 0,00117 0,50322
0,9 0,85 2,57 0,005 0,00108 0,001 0,00208 0,00000 0,00269 0,00122 0,52479
0,97 0,92 2,58 0,005 0,00117 0,001 0,00217 0,00000 0,00263 0,00120 0,51275
1,04 0,99 2,59 0,005 0,00126 0,001 0,00226 0,00000 0,00257 0,00117 0,50012
1,11 1,06 2,6 0,005 0,00135 0,001 0,00235 0,00000 0,00251 0,00114 0,48690
1,18 1,13 2,61 0,005 0,00144 0,001 0,00244 0,00000 0,00245 0,00110 0,47312
1,25 1,2 2,62 0,005 0,00153 0,001 0,00253 0,00000 0,00240 0,00107 0,45877
1,32 1,27 2,64 0,005 0,00162 0,001 0,00262 0,00000 0,00229 0,00109 0,46676
1,39 1,34 2,65 0,005 0,00171 0,001 0,00271 0,00000 0,00224 0,00105 0,45082
1,46 1,41 2,66 0,005 0,00180 0,001 0,00280 0,00000 0,00219 0,00101 0,43436
1,53 1,48 2,68 0,005 0,00189 0,001 0,00289 0,00000 0,00209 0,00102 0,43828
1,6 1,55 2,69 0,005 0,00198 0,001 0,00298 0,00000 0,00204 0,00098 0,42037
1,66 1,61 2,71 0,005 0,00205 0,001 0,00305 0,00000 0,00195 0,00100 0,42694
1,76 1,71 2,72 0,005 0,00218 0,001 0,00318 0,00000 0,00191 0,00091 0,39125
1,81 1,76 2,74 0,005 0,00225 0,001 0,00325 0,00000 0,00182 0,00093 0,40067
1,88 1,83 2,76 0,005 0,00234 0,001 0,00334 0,00000 0,00174 0,00093 0,39748
1,94 1,89 2,78 0,005 0,00241 0,001 0,00341 0,00000 0,00166 0,00093 0,39819
2,01 1,96 2,79 0,005 0,00250 0,001 0,00350 0,00000 0,00162 0,00088 0,37608
2,09 2,04 2,81 0,005 0,00260 0,001 0,00360 0,00000 0,00155 0,00085 0,36360
2,16 2,11 2,83 0,005 0,00269 0,001 0,00369 0,00000 0,00148 0,00083 0,35519
2,23 2,18 2,85 0,005 0,00278 0,001 0,00378 0,00000 0,00141 0,00081 0,34543
2,3 2,25 2,87 0,005 0,00287 0,001 0,00387 0,00000 0,00135 0,00078 0,33438
2,37 2,32 2,89 0,005 0,00296 0,001 0,00396 0,00000 0,00129 0,00075 0,32211
2,45 2,4 2,92 0,005 0,00306 0,001 0,00406 0,00000 0,00120 0,00074 0,31521
2,52 2,47 2,94 0,005 0,00315 0,001 0,00415 0,00000 0,00115 0,00070 0,30010
2,59 2,54 2,97 0,005 0,00324 0,001 0,00424 0,00000 0,00107 0,00069 0,29466
2,66 2,61 3 0,005 0,00333 0,001 0,00433 0,00000 0,00100 0,00067 0,28702
2,73 2,68 3,03 0,005 0,00342 0,001 0,00442 0,00000 0,00093 0,00065 0,27734
2,8 2,75 3,06 0,005 0,00351 0,001 0,00451 0,00000 0,00087 0,00062 0,26574
2,86 2,81 3,09 0,005 0,00359 0,001 0,00459 0,00000 0,00081 0,00060 0,25784
2,92 2,87 3,13 0,005 0,00366 0,001 0,00466 0,00000 0,00074 0,00060 0,25567
2,99 2,94 3,17 0,005 0,00375 0,001 0,00475 0,00000 0,00068 0,00057 0,24534
3,07 3,02 3,21 0,005 0,00385 0,001 0,00485 0,00000 0,00062 0,00053 0,22707
3,14 3,09 3,25 0,005 0,00394 0,001 0,00494 0,00000 0,00056 0,00049 0,21203
3,22 3,17 3,3 0,005 0,00405 0,001 0,00505 0,00000 0,00050 0,00045 0,19447
3,3 3,25 3,36 0,005 0,00415 0,001 0,00515 0,00000 0,00044 0,00042 0,17843
3,37 3,32 3,42 0,005 0,00424 0,001 0,00524 0,00000 0,00038 0,00038 0,16427
3,45 3,4 3,49 0,005 0,00434 0,001 0,00534 0,00000 0,00032 0,00034 0,14477
Appendix B 123

3,52 3,47 3,58 0,005 0,00443 0,001 0,00543 0,00000 0,00026 0,00031 0,13243
3,59 3,54 3,69 0,005 0,00452 0,001 0,00552 0,00000 0,00020 0,00028 0,11936
3,67 3,62 3,83 0,005 0,00462 0,001 0,00562 0,00000 0,00015 0,00023 0,09971
3,74 3,69 4,04 0,005 0,00471 0,001 0,00571 0,00000 0,00009 0,00020 0,08572
3,81 3,76 4,4 0,005 0,00480 0,001 0,00580 0,00000 0,00004 0,00016 0,06945
3,87 3,82 5,03 0,005 0,00487 0,001 0,00587 0,00000 0,00001 0,00012 0,04969
4 3,95 7,3 0,005 0,00504 0,001 0,00604 0,00000 0,00000 -0,00004 -0,01741
4,06 4,01 8,21 0,005 0,00512 0,001 0,00612 0,00000 0,00000 -0,00012 -0,04982
4,13 4,08 9,07 0,005 0,00521 0,001 0,00621 0,00001 0,00000 -0,00020 -0,08519
4,21 4,16 9,5 0,005 0,00531 0,001 0,00631 0,00002 0,00000 -0,00029 -0,12320

PZC
by salt addition method (Fig. 3.2)

Salt
0
0,001 NaNO3

Salt
f
0,1 NaNO3

pH
0
pH
f
pH
f
-pH
0
charge
4,61 5,26 0,65
6,77 6,89 0,12 0,0256
7,04 7,09 0,05 0,0099
7,33 7,28 -0,05 -0,0124
7,81 7,6 -0,21 -0,1102

-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
4 6 8
p
H
f
-
p
H
0
pH
0
Appendix B 124

Table B3 Determination of surface area of Fe
0
(Table 3.2)
Started: 24.10.2003 10:21:53

Sat. Pressure: 762.40 mmHg
Completed: 24.10.2003 14:22:11 Meas. Freespace: 0,1983 cm
3

Report Time: 27.10.2003 10:11:35 Sample Weight: 0,1014 g
Evac. Rate: 500.0 mmHg min
-1
Evac. Time: 0,500000 minutes
Analysis Mode: Equilibration Equil. Interval: 5 secs
Summary Report
Sing point surface area at P/P
0
0,29699657: 0,3118 m
2
g
-1

BET surface area: 0,3844 m
2
g
-1

Langmuir surface area: 0,5893 m
2
g
-1

Single point total pore volume of pores less
than 166,8517 nm diameter at P/P
0
0,98827072: 0,000004 cm
3
g
-1

Average pore diameter (4V/A by BET) 0,0381 nm
Langmuir Surface Area Report
Langmuir Surface Area: 0,5893 0,0909 m
2
g
-1

Slope: 7,387357 1,139686
Y - Intercept: 0,094211 0,219888
b: 0,012753
VM: 0,135366 cm
3
g
-1
STP
Correlation Coefficient: 8,902320E-10
Molecular Cross-sections: 0,1620 nm
2

Relative Pressure Vol Adsrobed 1/[VA x (P
0
/P)]
0 [cm
3
g
-1
STP]
0,059417626 0,0672 0,883601
0,079230059 0,0877 0,903582
0,098996585 0,1027 0,96381
0,118815577 0,1338 0,887783
0,138582103 0,1487 0,931899
0,158387977 0,1472 1,076319
0,178154502 0,1510 1,179594
0,197940712 0,1494 1,32523
0,217740031 0,1530 1,422981
0,237539350 0,1512 1,571348
0,257312430 0,1384 1,858534
0,277124858 0,1147 2,415569
0,296891394 0,1019 2,914595

0,300000000
0,350000000

0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
1
/
[
V
A

x

P
0
/
P
)
]
Relative Pressure (P/P
0
)
Langmuir Surface Area Plot
Appendix B 125

Table B4 Determination of surface area of iron oxides (rust) (Table 3.2)
Started: 03.11.2003 10:53:06 Sat. Pressure: 755,8 mmHg
3

-1
Summary Report
0
0,29699657: 44,6559 m
2
g
-1

2
g
-1

2
g
-1

0
0,98827072: 0,187834 cm
3
g
-1

Langmuir Surface Area Report
Langmuir Surface Area: 48,7576 0,9909 m
2
g
-1

Slope: 0,087195 0,001782
Y - Intercept: 0,002087 0,000344
b: 42,775239
VM: 11,200416 cm
3
g
-1
STP
2

0
/P)]
0 [cm
3
g
-1
STP]
0,059427099 8,2391 0,007669
0,079233923 9,2137 0,00934
0,099001056 10,0321 0,010953
0,118821119 10,7556 0,012537
0,138614714 11,4477 0,014057
0,158288464 12,1083 0,015543
0,178168835 12,6121 0,017189
0,197955819 13,1154 0,018819
0,217775972 13,5552 0,020539
0,237549611 13,8702 0,022463
0,257389530 14,2151 0,024383
0,277176513 14,3727 0,02668
0,296996566 14,5919 0,028952

0,300000000

0,350000000

0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
1
/
[
V
A

x

P
0
/
P
)
]
0
)
Langmuir Surface Area Plot
Appendix B 126

Table B5 Determination of surface area of corroded iron (Table 3.2)
Started: 29.10.2003 10:24:05 Sat. Pressure: 748,3 mmHg
3

-1
Summary Report
0
0,29699657: 4,1863 m
2
g
-1

2
g
-1

2
g
-1

0
0,98827072: 0,012535 cm
3
g
-1

BET surface area report:
BET surface area: 4,3106 0,0876 m
2
g
-1

Slope: 1,010515 0,020142
Y - Intercept: -0,000644 0,003888
b: -1569,007623
VM: 0,990225 cm
3
g
-1
STP
2

0
/P)]
0 [cm
3
g
-1
STP]
0,05948149 0,9509 0,06651
0,07927903 1,0269 0,83848
0,099057866 1,0884 0,101018
0,118876124 1,1352 0,118847
0,13866097 1,1783 0,136618
0,158412405 1,2178 0,15456
0,178223985 1,2462 0,174025
0,198028865

1,2746

0,193735
0,217853811 1,2991 0,214399
0,237625291 1,3201 0,236111
0,257430172 1,3337 0,259943
0,277201652 1,3472 0,284674
0,296979820 1,3679 0,30882

0,300000000
0,350000000

0.0
0.1
0.1
0.2
0.2
0.3
0.3
0.4
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
1
/
[
V
A

x

P
0
/
P
)
]
0
)
BET Surface Area Plot
Appendix B 127

Table B6 Determination of Pb
2+
removal in Fe
0
system with pulsed dosing (Fig. 4.2)
Metrix: Pb
2+
in model roof runoff
Pb
2+
: 5 mg L
-1

pH: 5,0 controlled
Conductivities: 80-200 S cm
-1

Temperature: 20 C

Fe
0
bed: 0,5 g Fe
0

C/C
0

Time h Pb
2+
Fe
2+
Fe conc. Pb Pb removed Total Pb remove
0 0,0 5,00 0,08 0 1,00 0,00 0,00
0,05 0,0 5,00 0,13 0,05 1,00 0,00 0,00
0,35 0,4 4,60 0,16 0,08 0,92 0,40 0,40
1,05 1,1 4,15 0,2 0,12 0,83 0,85 0,85
1,35 1,4 3,85 0,27 0,19 0,77 1,15 1,15
2,05 2,1 3,50 0,33 0,25 0,70 1,50 1,50
2,35 2,4 3,10 0,38 0,3 0,62 1,90 1,90
3,05 3,1 2,98 0,45 0,37 0,60 2,02 2,02
0 0,4 6,95 0,55 0,47 1,00 0,00 2,02
0,05 0,5 6,88 0,58 0,5 0,99 0,07 2,09
0,35 0,8 6,48 0,83 0,75 0,93 0,47 2,49
1,05 1,5 5,89 0,98 0,9 0,85 1,06 3,08
1,35 1,8 5,49 1,17 1,09 0,79 1,46 3,48
2,05 2,5 5,45 1,26 1,18 0,78 1,50 3,52
2,35 2,8 5,05 1,35 1,27 0,73 1,90 3,92
3,05 3,5 4,74 1,46 1,38 0,68 2,21 4,23
0 0,1 9,15 1,51 1,43 1,00 0,00 4,23
0,05 0,2 9,13 1,54 1,46 1,00 0,02 4,25
0,35 0,5 8,42 1,68 1,6 0,92 0,73 4,96
1,05 1,2 7,70 1,82 1,74 0,84 1,45 5,68
1,35 1,5 7,21 1,89 1,81 0,79 1,94 6,17
2,05 2,2 6,77 1,96 1,88 0,74 2,38 6,61
2,35 2,5 6,43 1,98 1,9 0,70 2,72 6,95
3,05 3,2 6,03 2,06 1,98 0,66 3,12 7,35
0 0,8 11,05 2,15 2,07 1,00 0,00 7,35
0,05 0,9 11,04 2,19 2,11 11,04 0,01 7,36
0,35 1,2 11,01 2,56 2,48 11,01 0,04 7,39
1,05 1,9 10,25 2,7 2,62 10,25 0,80 8,15
1,35 2,2 9,61 2,67 2,59 9,61 1,44 8,79
2,05 2,9 8,99 2,72 2,64 8,99 2,06 9,41
2,35 3,2 8,53 2,88 2,8 8,53 2,52 9,87
3,05 3,9 8,22 3 2,92 8,22 2,83 10,18

Appendix B 128

Table B7 Results of batch investigation to determine the impact of pH and DO (Fig. 4.3)
Metrix: Cu
2+
/Zn
2+
in DI water
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 4-5 mg L
-1

Conductivity: 80 S cm
-1

Temperature: 20 C

DO
i
: 0 and 7-8 mg L
-1

Fe
0
: 0,5 g Fe
0

c/c
0
of copper
h pH 4 pH 4 DO 0 pH 5 pH 5 DO 0 pH 6 pH 6 DO 0 pH 7 pH 7 DO 0
0,05 1,00 1,00 1,00 1,00 1,00 1,00 1,00 1,00
1,05 1,00 0,93 0,90 0,98 1,00 1,00 0,88 1,00
2,05 0,78 0,85 0,72 0,89 0,76 0,87 0,71 0,91
3,05 0,64 0,82 0,52 0,78 0,51 0,72 0,38 0,70
4,05 0,53 0,73 0,34 0,70 0,36 0,62 0,24 0,58
5,05 0,45 0,66 0,24 0,62 0,25 0,54 0,18 0,49
6,05 0,38 0,57 0,17 0,57 0,20 0,48 0,15 0,33
8,05 0,31 0,55 0,14 0,51 0,15 0,42 0,12 0,28
10,05 0,22 0,43 0,08 0,43 0,12 0,29 0,09 0,25
12,05 0,15 0,35 0,07 0,34 0,12 0,25 0,06 0,18
14,05 0,13 0,28 0,05 0,35 0,08 0,20 0,06 0,16
16,05 0,10 0,25 0,03 0,26 0,07 0,17 0,06 0,12
18,05 0,09 0,18 0,03 0,21 0,05 0,15 0,06 0,11
20,05 0,10 0,14 0,03 0,17 0,05 0,14 0,06 0,09
22,05 0,11 0,11 0,03 0,15 0,03 0,11 0,06 0,09
24,05 0,14 0,10 0,02 0,11 0,03 0,09 0,06 0,07
c/c
0
of zinc
h pH 4 pH 4 DO 0 pH 5 pH 5 DO 0 pH 6 pH 6 DO 0 pH 7 pH 7 DO 0
0,05 1,00 1,01 1,00 1,01 1,00 1,00 1,00 1,00
1,05 0,95 1,05 0,94 1,03 0,96 1,00 0,98 1,00
2,05 0,94 1,01 0,89 1,03 0,90 0,99 0,92 1,00
3,05 0,89 1,03 0,78 1,00 0,83 0,97 0,88 0,99
4,05 0,84 0,99 0,77 0,99 0,77 0,95 0,82 1,00
5,05 0,82 1,01 0,70 0,99 0,71 0,94 0,78 0,98
6,05 0,77 0,99 0,66 0,97 0,66 1,00 0,74 0,94
8,05 0,68 0,99 0,59 0,95 0,58 1,00 0,68 0,94
10,05 0,60 0,96 0,51 0,92 0,50 0,96 0,62 0,91
12,05 0,52 0,95 0,45 0,95 0,45 0,91 0,57 0,88
14,05 0,47 1,01 0,40 0,96 0,40 0,89 0,52 0,82
16,05 0,43 0,99 0,36 0,91 0,36 0,85 0,48 0,86
18,05 0,40 0,97 0,34 0,89 0,32 0,81 0,44 0,85
20,05 0,40 0,95 0,30 0,86 0,29 0,79 0,40 0,82
22,05 0,39 0,92 0,27 0,81 0,26 0,78 0,37 0,79
24,05 0,43 0,90 0,26 0,75 0,23 0,75 0,34 0,79

Appendix B 129

Table B8 Impact of Suwanee river fulvic acid on the kinetic of Cu
2+
and Zn
2+
removal (batch
system, Table 4.2)
Metrix: Cu
2+
/Zn
2+
in roof runoff
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

Conductivities: 80-200 S cm
-1

DOC: 5 mg L
-1

Temperature: 20 C

Fe
0
bed: 0,5 g Fe
0

h Cu Zn Fe c/c
0
of Cu c/c
0
of Zn Cu removed
0 4,55 4,83 0,03 1,00 1,00 0,00
0,05 4,54 4,83 0,07 1,00 1,00 0,01
1,05 3,72 4,73 1,21 0,82 0,98 0,83
2,05 2,92 4,54 2,21 0,64 0,94 1,63
3,05 2,33 4,3 3,07 0,51 0,89 2,22
4,05 1,85 4,17 3,67 0,41 0,86 2,70
5,05 1,46 3,98 4,14 0,32 0,82 3,09
6,05 1,24 3,96 4,89 0,27 0,82 3,31
8,05 0,83 3,73 5,57 0,18 0,77 3,72
10,05 0,58 3,55 6,34 0,13 0,73 3,97
12,05 0,42 3,41 6,98 0,09 0,71 4,13
14,05 0,32 3,3 7,46 0,07 0,68 4,23
16,05 0,26 3,24 8,05 0,06 0,67 4,29
18,05 0,22 3,24 8,88 0,05 0,67 4,33
20,05 0,19 3,12 9,14 0,04 0,65 4,36
22,05 0,17 3,06 9,65 0,04 0,63 4,38
24,05 0,13 2,92 7,76 0,03 0,60 4,42

Appendix B 130

Table B9 Example of monitoring results of pH, ORP and DO from batch tests
(example of Table 4.2)
System: Batch investigation at pH
i
= 5,0
Metrix: Cu
2+
/Zn
2+
in DI water
Cu
2+
: 4-5 mg L
-1
-1

Zn
2+
: 4-5 mg L
-1
Temperature: 20 C
DO: 7-8 mg L
-1
Fe
0
: 0,5 g Fe
0

Time [h]
pH
[-]
ORP
[mV]
DO
[mg L
-1
]

0,00 5,02 308 8,50
0,43 6,16 122 7,40
0,85 6,29 78 6,54
1,27 6,25 75 6,11
1,70 6,31 65 5,79
2,12 6,31 67 5,84
2,54 6,28 71 6,11
2,96 6,31 69 6,17
3,38 6,26 67 5,47
3,80 6,24 72 6,00
4,22 6,25 74 5,36
4,64 6,34 69 5,68
5,06 6,25 69 5,26
5,48 6,30 75 5,31
5,91 6,29 76 5,68
6,75 6,28 79 4,83
7,17 6,29 78 5,26
7,59 6,26 77 4,61
8,01 6,26 78 4,40
8,42 6,29 76 4,51
8,84 6,25 79 4,29
9,27 6,22 80 4,45
9,69 6,28 79 4,56
10,11 6,26 74 4,56
10,53 6,24 77 4,35
10,95 6,20 74 3,97
11,37 6,22 78 4,40
11,79 6,17 80 3,76
12,21 6,20 77 3,54
12,63 6,26 77 3,54
13,05 6,21 74 3,81
13,48 6,17 79 3,54
14,74 6,20 84 3,70

15,16 6,25 80 3,44
15,58 6,24 82 3,76
16,00 6,13 79 3,11
16,42 6,16 83 3,11
16,84 6,16 82 3,01
17,27 6,15 77 3,11
17,69 6,11 84 3,11
18,11 6,14 86 2,90
18,53 6,16 77 3,33
18,95 6,19 86 3,11
19,37 6,14 77 3,44
19,79 6,16 79 3,33
20,21 6,15 87 3,60
20,63 6,14 82 2,85
21,06 6,11 83 2,69
21,48 6,14 79 2,69
21,90 6,11 82 2,69
22,32 6,10 91 2,47
23,16 6,10 80 3,01
24,00 6,11 80 2,95

0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
0 5 10 15 20 25
p
H
Time [h]
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 5 10 15 20 25
D
O

[
m
g

L
-
1
]
Time [h]
0
50
100
150
200
250
300
350
0 5 10 15 20 25
O
R
P

[
m
V
]
Time [h]
Appendix B 131

Table B10 Example of monitoring results of pH, ORP and DO from batch tests (e.g., for
Table 4.2)
System: Batch investigation at controlled pH 7,0
Metrix: Cu
2+
/Zn
2+
in DI water
Cu
2+
: 4-5 mg L
-1
-1

Zn
2+
: 4-5 mg L
-1
Temperature: 20 C
DO: 7-9 mg L
-1
Fe
0
: 0,5 g Fe
0

Time [h]
pH
[-]
ORP
[mV]
DO
[mg L
-1
]
0,00 6,95 210 9,20

0,43 6,95 171 8,68
0,85 6,95 187 8,68
1,27 6,95 188 8,68
1,70 6,95 171 9,32
2,12 6,95 191 9,11
2,54 6,95 195 8,47
2,96 6,95 185 9,16
3,38 6,95 187 8,95
3,80 6,95 169 8,68
4,22 6,95 163 8,47
4,64 6,95 197 8,41
5,06 6,95 180 9,00
5,48 6,95 192 8,84
5,91 6,95 210 9,06
6,75 6,95 169 8,68
7,17 6,95 173 8,79
7,59 6,95 193 9,11
8,01 6,95 169 8,31
8,42 6,95 195 8,36
8,84 6,95 191 7,82
9,27 6,95 169 8,36
9,69 6,95 167 8,47
10,11 6,95 170 8,47
10,53 6,95 194 7,77
10,95 6,95 169 8,52
11,37 6,95 178 8,20
11,79 6,95 175 7,82
12,21 6,95 193 7,66
12,63 6,95 179 8,63
13,05 6,95 188 7,88
13,48 6,95 169 7,72
14,32 6,95 170 8,36
14,74 6,95 176 7,82
15,16 6,95 187 7,72
15,58 6,95 196 7,77
16,00 6,95 175 7,82
16,42 6,95 173 8,20
16,84 6,95 194 7,61
17,27 6,95 194 7,56
17,69 6,95 192 7,40
18,11 6,95 188 7,99
18,53 6,95 176 8,20
18,95 6,95 180 7,61
19,37 6,95 195 7,34
19,79 6,95 194 7,29
20,21 6,95 172 7,82
20,63 6,95 178 8,31
21,06 6,95 165 8,36
21,48 6,95 179 8,31
21,90 6,95 181 8,15
23,16 6,95 176 8,20
24,00 6,95 197 7,82

0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
0 5 10 15 20 25
p
H
Time [h]
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
9.0
10.0
0 5 10 15 20 25
D
O

[
m
g

L
-
1
]
Time [h]
0
50
100
150
200
250
0 5 10 15 20 25
O
R
P

[
m
V
]
Time [h]
Appendix B 132

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: TU-SW DOC Cu
2+
Zn
2+

Samples DOC ZVI[g L
-1
] UV
254
UV
436
initial Eq. [mg g
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 2,38 0,514 0,0462 0,004 3,94 1,12 5,49 5,12 0,16 9,65
S1.2 2,26 0,49 0,0555 0,0032 4,03 1,04 6,10 4,92 0,19 9,65
S1.3 2,08 0,498 0,0605 0,0074 4,66 2,74
S1.4 1,81 0,506 0,0405 0,0023 4,02 1,11 5,74 4,92 0,08 9,57 5,170 0,20 9,82
S2.1 2,02 0,442 0,0659 0,0037 3,94 0,85 7,00 5,12 0,51 10,43
S2.2 2,42 0,44 0,1282 0,0126 4,03 1,24 6,34 4,92 0,95 9,02
S2.3 2,52 0,398 0,1644 0,0182 4,66 2,36 5,78
S2.4 1,96 0,436 0,0489 0,0013 4,02 1,09 6,71 4,92 0,26 10,69 5,170 0,69 10,28
S3.1 1,91 0,412 0,0403 0,0016 3,94 0,94 7,29 5,12 1,09 9,78
S3.2 2,18 0,408 0,0887 0,0125 4,03 3,57 1,13 4,92 1,63 8,06
S3.3 2,04 0,44 0,1009 0,0092 4,66 2,92 3,95
S3.4 1,79 0,404 0,0445 0,0016 4,02 1,13 7,15 4,92 0,32 11,39 5,170 0,73 10,99
S4.1 1,66 0,348 0,0337 0,0006 3,94 1,16 7,99 5,12 0,98 11,90
S4.2 2,31 0,34 0,1298 0,0131 4,03 2,76 3,74 4,92 1,41
S4.3 2,21 0,358 0,1082 0,0097 4,66 3,75 2,55
S4.4 2,67 0,352 0,0713 0,0036 4,02 1,15 8,14 4,92 1,05 5,170 1,94 9,18
S5.1 2,21 0,284 0,0657 0,0034 3,94 1,25 9,47 5,12 1,28 13,52
S5.2 2,62 0,282 0,1379 0,0128 4,03 3,64 1,40 4,92 2,8 7,52
S5.3 2,55 0,28 0,1791 0,0155 4,66 4,04 2,20
S5.4 1,82 0,29 0,0402 0,0004 4,02 1,10 10,07 4,92 0,31 15,90 5,170 0,78 15,14
S6.1 2,27 0,22 0,0716 0,0038 3,94 1,23 12,33 5,12 1,55
S6.2 2,47 0,228 0,1224 0,0088 4,03 3,90 0,57 4,92 3,5 6,23
S6.3 2,33 0,25 0,1619 0,0156 4,66 4,14 2,06
S6.4 2,29 0,218 0,0587 0,0024 4,02 1,26 12,64 4,92 0,61 19,77 5,170 1,81 15,41
S7.1 2,38 0,188 0,1024 0,01 3,94 3,62 1,72 5,12 3 11,28
S7.2 2,55 0,192 0,1207 0,0091 4,03 3,70 1,72 4,92 3,86 5,52
S7.3 2,18 0,184 0,0751 0,0046 4,66 4,23 2,35
S7.4 2,01 0,186 0,0534 0,0013 4,02 3,48 2,92 4,92 1,75 17,04 5,170 3,110 11,08
S8.1 2,06 0,148 0,0532 0,0013 3,94 3,71 1,59 5,12 3,15
S8.2 2,76 0,158 0,1261 0,0084 4,03 4,23 4,92 3,95
S8.3 2,63 0,158 0,1975 0,0187 4,66 4,39 1,72
S8.4 2,11 0,148 0,0622 0,0026 4,02 3,82 1,35 4,92 3,12 12,16 5,170 4,470 4,73
S9.1 2,30 0,106 0,0768 0,0034 3,94 3,75 1,79 5,12 3,51 15,19
S9.2 2,66 0,102 0,0943 0,0046 4,03 4,29 4,92 4,82
S9.3 2,71 0,096 0,1002 0,0109 4,66 4,43 2,35
S9.4 2,50 0,1 0,0741 0,0025 4,02 3,49 5,30 4,92 1,76 5,170 3,92
Cu ConT1 2,82 0,1304 0,0073 5,37
Cu ConT2 3,10 0,1271 0,0052 4,86
Zn ConT1 3,14 0,121 0,0047 4,82
Zn ConT2 3,13 0,1141 0,0043 5,02
DOC ConT1 3,34 0,1136 0,004
DOC ConT2 3,23 0,121 0,0048
CuZnConT1 3,04 0,1233 0,0044 4,79 5,16
CuZnConT2 3,40 0,1172 0,0041 5,05 5,18
*ContT = controlled vials

Appendix B 133

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: Lankwitz DOC Cu
2+
Zn
2+

Samples DOC ZVI[g L
-1
] UV
254
UV
436
initial Eq. [mg g
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 1,21 1,492 0,0464 0,0226 1,28 1,21 0,05 4,85 0,18 3,13 0,67 0,08 0,39
S1.2 0,85 1,512 0,0533 0,025 1,35 0,85 0,33 5,86 0,04 3,85
S1.3 0,95 1,504 0,0353 0,0153 1,31 0,94 0,24 0,64 0,04 0,40
S1.4 0,78 1,506 0,0354 0,0184 1,30 0,78 0,34 4,87 0,11 3,16 5,84 0,17 3,76
S2.1 0,73 0,81 0,0454 0,0254 1,28 0,72 0,69 4,85 0,1 5,86 0,67 0,04 0,78
S2.2 1,04 0,79 0,0567 0,0285 1,35 1,04 0,39 5,86 0,07 7,33
S2.3 0,89 0,792 0,0304 0,0118 1,31 0,88 0,54 0,64 0,02 0,78
S2.4 1,05 0,808 0,0761 0,0206 1,30 1,05 0,31 4,87 0,41 5,52 5,84 1,14 5,82
S3.1 1,08 0,492 0,0578 0,0225 1,28 1,08 0,42 4,85 0 9,85 0,67 0,10 1,16
S3.2 0,82 0,512 0,0445 0,0262 1,35 0,82 1,04 5,86 1,18 9,15
S3.3 0,79 0,498 0,0567 0,0357 1,31 0,78 1,06 0,64 0,05 1,19
S3.4 0,73 0,52 0,0537 0,0196 1,30 0,72 1,11 4,87 0,17 9,04 5,84 0,93 9,45
S4.1 1,09 0,314 0,078 0,0271 1,28 1,08 0,64 4,85 1,16 11,75 0,67 0,27 1,29
S4.2 0,79 0,306 0,0442 0,0256 1,35 0,78 1,84 5,86 2,05 12,46
S4.3 0,94 0,296 0,0353 0,0175 1,31 0,94 1,26 0,64 0,05 2,01
S4.4 0,81 0,312 0,0686 0,0178 1,30 0,80 1,59 4,87 0,12 15,24 5,84 0,62 16,74
S5.1 0,71 0,198 0,033 0,018 1,28 0,70 2,93 4,85 0,94 19,74 0,67 0,28 1,99
S5.2 0,82 0,196 0,0475 0,031 1,35 0,82 2,70 5,86 2,91 15,07
S5.3 0,94 0,204 0,0529 0,0335 1,31 0,94 1,81 0,64 0,23 2,03
S5.4 0,75 0,214 0,0764 0,0169 1,30 0,75 2,57 4,87 0,13 22,17 5,84 1,71 19,32
S6.1 1,01 0,092 0,0296 0,0114 1,28 1,00 3,04 4,85 2,85 21,73 0,67 0,50 1,90
S6.2 0,96 0,102 0,0377 0,0144 1,35 0,96 3,82 5,86 4,7 11,42
S6.3 1,24 0,112 0,0391 0,0206 1,31 1,24 0,62 0,64 0,30 3,08
S6.4 0,89 0,1 0,0473 0,0192 1,30 0,89 4,11 4,87 2,62 22,55 5,84 4,19 16,55
S7.1 1,08 0,054 0,0364 0,0124 1,28 1,07 3,89 4,85 2,1 50,92 0,67 0,47 3,79
S7.2 1,01 0,05 0,0567 0,008 1,35 1,01 6,73 5,86 5,33 10,70
S7.3 0,92 0,054 0,0408 0,0158 1,31 0,92 7,28 0,64 0,42 4,16
S7.4 0,91 0,052 0,041 0,01 1,30 0,91 7,53 4,87 0,89 76,63 5,84 3,41 46,82
S8.1 0,87 0,04 0,0353 0,0153 1,28 0,87 10,34 4,85 3,34 37,75 0,67 0,54 3,37
S8.2 1,03 0,038 0,0485 0,0032 1,35 1,03 8,33 5,86 5,52 9,07
S8.3 0,89 0,046 0,0343 0,0149 1,31 0,89 9,20 0,64 0,41 5,10
S8.4 0,80 0,042 0,0453 0,0195 1,30 0,80 11,86 4,87 2,45 57,73 5,84 4,38 34,88
S9.1 1,37 0,022 0,0602 0,0073 1,28 1,36 4,85 4,12 33,18 0,67 0,58 4,31
S9.2 1,15 0,02 0,0448 0,0022 1,35 1,15 9,83 5,86 5,43 21,75
S9.3 0,95 0,018 0,0552 0,0093 1,31 0,95 20,18 0,64 0,55 5,27
S9.4 0,89 0,016 0,0458 0,0202 1,30 0,89 25,72 4,87 2,39 5,84 4,37
Cu ConT 1 1,14 0,0445 0,0015 4,65 0,68
Cu ConT 2 1,23 0,0441 0,0014 5,05 0,67
Zn ConT1 1,31 0,037 0,0015 0,07 5,67
Zn ConT2 1,39 0,0384 0,0013 0,01 6,06
DOC ConT1 1,22 0,0347 0,001 0 0,63
DOC ConT2 1,41 0,0382 0,0013 0 0,66
CuZnConT1 1,10 0,036 0,0014 4,7 5,67
CuZnConT2 1,50 0,042 0,0013 5,05 6,02

Appendix B 134

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: SRFA DOC Cu
2+
Zn
2+

Samples DOC ZVI[g L
-1
] UV
254
UV
436
initial Eq. [mg g
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 1,12 1,504 0,0182 0,0005 3,94 1,12 1,87 5,12 0,16 3,29
S1.2 1,05 1,516 0,0342 0,0083 4,03 1,04 1,97 4,92 0,19 3,12
S1.3 2,74 1,504 0,8653 0,1326 4,66 2,74
S1.4 1,12 1,498 0,0196 0,0008 4,02 1,11 1,94 4,92 0,08 3,23 5,17 0,20 3,31
S2.1 0,85 0,832 0,0178 0,0018 3,94 0,85 3,71 5,12 0,51 5,54
S2.2 1,25 0,848 0,0441 0,0126 4,03 1,24 3,29 4,92 0,95 4,68
S2.3 2,36 0,8 0,5027 0,0745 4,66 2,36 2,87
S2.4 1,10 0,828 0,0213 0,0013 4,02 1,09 3,53 4,92 0,26 5,62 5,17 0,69 5,41
S3.1 0,94 0,464 0,0142 0 3,94 0,94 6,47 5,12 1,09 8,68
S3.2 3,58 0,522 0,4655 0,0547 4,03 3,57 0,88 4,92 1,63 6,30
S3.3 2,92 0,51 0,6672 0,0943 4,66 2,92 3,40
S3.4 1,13 0,478 0,0188 0,0002 4,02 1,13 6,04 4,92 0,32 9,62 5,17 0,73 9,28
S4.1 1,17 0,292 0,0228 0,001 3,94 1,16 9,51 5,12 0,98 14,17
S4.2 2,77 0,284 0,4177 0,063 4,03 2,76 4,47 4,92 1,41
S4.3 3,75 0,28 0,8847 0,1144 4,66 3,75 3,25
S4.4 1,16 0,298 0,0228 0,0009 4,02 1,15 9,61 4,92 1,05 5,17 1,94 10,83
S5.1 1,26 0,206 0,0298 0,0014 3,94 1,25 13,05 5,12 1,28 18,64
S5.2 3,64 0,192 0,3762 0,0362 4,03 3,64 2,05 4,92 2,8 11,04
S5.3 4,05 0,19 0,6568 0,0775 4,66 4,04 3,23
S5.4 1,10 0,196 0,0249 0,0007 4,02 1,10 14,89 4,92 0,31 23,52 5,17 0,78 22,39
S6.1 1,23 0,094 0,0291 0,001 3,94 1,23 28,85 5,12 1,55
S6.2 3,91 0,098 0,3212 0,0291 4,03 3,90 1,33 4,92 3,5 14,49
S6.3 4,15 0,11 0,6351 0,0741 4,66 4,14 4,68
S6.4 1,27 0,09 0,0324 0,002 4,02 1,26 30,62 4,92 0,61 47,88 5,17 1,81 37,33
S7.1 3,62 0,056 0,3348 0,0308 3,94 3,62 5,76 5,12 3 37,85
S7.2 3,71 0,062 0,2666 0,0229 4,03 3,70 5,32 4,92 3,86 17,09
S7.3 4,23 0,056 0,4826 0,0519 4,66 4,23 7,71
S7.4 3,48 0,046 0,2608 0,0228 4,02 3,48 11,80 4,92 1,75 68,91 5,17 3,11 44,78
S8.1 3,71 0,014 0,366 0,036 3,94 3,71 16,85 5,12 3,15
S8.2 4,23 0,018 0,2632 0,0222 4,03 4,23 4,92 3,95
S8.3 4,39 0,026 0,2028 0,0137 4,66 4,39 10,46
S8.4 3,82 0,026 0,2205 0,0173 4,02 3,82 7,67 4,92 3,12 69,23 5,17 4,47 26,92
S9.1 3,76 0,026 0,2613 0,0217 3,94 3,75 7,28 5,12 3,51 61,92
S9.2 4,30 0,014 0,1883 0,0118 4,03 4,29 4,92 4,82
S9.3 4,44 0,008 0,2065 0,014 4,66 4,43 28,16
S9.4 3,49 0,01 0,2324 0,0195 4,02 3,49 52,96 4,92 1,76 5,17 3,92
Cu ConT 1 4,01 0,2077 0,0122 5,37
Cu ConT 2 3,89 0,1941 0,0111 4,86
Zn ConT1 3,72 0,1858 0,0099 4,82
Zn ConT2 4,04 0,1859 0,0098 5,02
DOC ConT1 4,40 0,196 0,0099
DOC ConT2 4,93 0,1958 0,0102
CuZnConT1 3,88 0,2022 0,012 4,79 5,16
CuZnConT2 4,17 0,2066 0,0127 5,05 5,18

Appendix B 135

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: UFA DOC Cu
2+
Zn
2+

Samples DOC ZVI[g L
-1
] UV
254
UV
436
initial Eq. [mg g
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 2,24 1,528 0,0642 0,0 3,94 1,12 1,84 5,12 0,16 3,24
S1.2 2,91 1,474 0,1832 0,1 4,03 1,04 2,02 4,92 0,19 3,2
S1.3 2,87 1,468 0,1153 0,0 4,66 2,74
S1.4 2,79 1,52 0,0823 0,0 4,02 1,11 1,91 4,92 0,08 3,18 5,17 0,20 3,3
S2.1 2,67 0,82 0,071 0,0 3,94 0,85 3,77 5,12 0,51 5,62
S2.2 3,56 0,794 0,2331 0,1 4,03 1,24 3,51 4,92 0,95 5,0
S2.3 3,29 0,794 0,0873 0,0 4,66 2,36 2,89
S2.4 1,91 0,804 0,0765 0,0 4,02 1,09 3,63 4,92 0,26 5,79 5,17 0,69 5,6
S3.1 2,74 0,504 0,0684 0,0 3,94 0,94 5,95 5,12 1,09 7,99
S3.2 4,22 0,492 0,5916 0,1 4,03 3,57 0,93 4,92 1,63 6,7
S3.3 3,54 0,512 0,1588 0,0 4,66 2,92 3,39
S3.4 2,93 0,5 0,0701 0,0 4,02 1,13 5,77 4,92 0,32 9,20 5,17 0,73 8,9
S4.1 3,01 0,402 0,0771 0,0 3,94 1,16 6,91 5,12 0,98 10,29
S4.2 6,35 0,422 0,4987 0,0 4,03 2,76 3,01 4,92 1,41
S4.3 4,14 0,422 0,6396 0,1 4,66 3,75 2,16
S4.4 2,73 0,418 0,0667 0,0 4,02 1,15 6,85 4,92 1,05 5,17 1,94 7,7
S5.1 5,12 0,268 0,3964 0,0 3,94 1,25 10,0 5,12 1,28 14,32
S5.2 6,77 0,288 0,5181 0,0 4,03 3,64 1,36 4,92 2,80 7,4
S5.3 6,23 0,316 0,6893 0,1 4,66 4,04 1,94
S5.4 2,77 0,304 0,0668 0,0 4,02 1,10 9,60 4,92 0,31 15,16 5,17 0,78 14,4
S6.1 5,68 0,21 0,4315 0,0 3,94 1,23 12,91 5,12 1,55
S6.2 6,98 0,214 0,3417 0,0 4,03 3,90 0,60 4,92 3,50 6,6
S6.3 7,01 0,182 0,4862 0,0 4,66 4,14 2,83
S6.4 3,60 0,212 0,1056 0,0 4,02 1,26 13,00 4,92 0,61 20,33 5,17 1,81 15,8
S7.1 6,28 0,096 0,3682 0,0 3,94 3,62 3,361 5,12 3,00 22,08
S7.2 7,07 0,118 0,3497 0,0 4,03 3,70 2,79 4,92 3,86 9,0
S7.3 7,39 0,11 0,543 0,0 4,66 4,23 3,92
S7.4 5,92 0,126 0,3862 0,0 4,02 3,48 4,30 4,92 1,75 25,15 5,17 3,11 16,3
S8.1 6,99 0,056 0,3278 0,0 3,94 3,71 4,21 5,12 3,15
S8.2 7,29 0,046 0,3107 0,0 4,03 4,23 4,92 3,95
S8.3 6,95 0,056 0,4146 0,0 4,66 4,39 4,85
S8.4 6,96 0,062 0,3352 0,0 4,02 3,82 3,22 4,92 3,12 29,03 5,17 4,47 11,3
S9.1 6,90 0,026 0,3205 0,0 3,94 3,75 7,28 5,12 3,51 61,92
S9.2 7,31 0,02 0,3 0,0 4,03 4,29 4,92 4,82
S9.3 7,26 0,018 0,3305 0,0 4,66 4,43 12,51
S9.4 7,11 0,012 0,302 0,0 4,02 3,49 44,13 4,92 1,76 5,17 3,92
Cu ConT 1 6,62 0,319 0,0 5,37
Cu ConT 2 7,41 0,315 0,017 4,86
Zn ConT1 7,55 0,296 0,0 4,82
Zn ConT2 7,88 0,295 0,016 5,02
DOC ConT1 7,30 0,292 0,0
DOC ConT2 7,15 0,292 0,016
CuZnConT1 7,52 0,312 0,0 4,79 5,16
CuZnConT2 6,24 0,313 0,017 5,05 5,18

Appendix B 136

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: SSWR Cu
2+
Zn
2+

Samples ZVI[g L
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 1,522 4,78 0,20 3,01 0,04 0,07
S1.2 1,498 4,82 0,24 3,06 5,08 0,96 2,8
S1.3 1,502 5,13 0,21 3,3
S1.4
S2.1 0,828 4,78 0,66 4,98 0,04 0,01
S2.2 0,834 4,82 0,22 5,52 5,08 0,27
S2.3 0,792 5,13 0,92 5,3
S2.4
S3.1 0,524 4,78 0,54 8,09 0,04 0,00
S3.2 0,526 4,82 0,58 8,06 5,08 1,57 6,7
S3.3 0,464 5,13 1,72 7,3
S3.4
S4.1 0,266 4,78 1,67 11,69 0,04 0,03
S4.2 0,298 4,82 1,09 12,52 5,08 2,63 8,2
S4.3 0,292 5,13 1,51 12,4
S4.4
S5.1 0,186 4,78 1,94 15,27 0,04 0,01
S5.2 0,17 4,82 1,39 20,18 5,08 3,23 10,9
S5.3 0,206 5,13 2,02 15,1
S5.4
S6.1 0,096 4,78 2,15 27,40 0,04 0,03
S6.2 0,112 4,82 1,28 31,61 5,08 3,32 15,7
S6.3 0,096 5,13 2,67 25,6
S6.4
S7.1 0,046 4,78 3,00 38,70 0,04 0,05
S7.2 0,056 4,82 2,83 5,08 4,29 14,1
S7.3 0,048 5,13 4,32 16,9
S7.4 0,052 4,82 2,37 47,12 5,08 4,06 19,6
S8.1 0,016 4,78 3,58 0,04 0,05
S8.2 0,026 4,82 3,01 69,62 5,08 4,55 20,4
S8.3 0,02 5,13 4,47 33,0
S8.4 0,014 4,82 3,36 104,29 5,08 4,83 17,9
S9.1 0,014 4,78 3,87 0,04 0,06
S9.2 0,01 4,82 3,62 120 5,08 4,86 22,0
S9.3 0,012 0,05 5,13 5,07 5,0
S9.4 0,028 4,82 3,02 5,08 4,82 9,3
Cu ConT 1 4,78 0,05
Cu ConT 2 4,78 0,04
Zn ConT1 5,13
Zn ConT2
DOC ConT1 4,82 5,08
DOC ConT2
CuZnConT1
CuZnConT2

Appendix B 137

2+
and Zn
2+
removal by Fe
0
(Fig. 4.5)

Source of water: Halensee DOC Cu
2+
Zn
2+

Samples DOC ZVI [g L
-1
] UV
254
UV
436
initial Eq. [mg g
-1
] initial Eq. [mg g
-1
] initial Eq. [mg g
-1
]
S1.1 30,11 1,6 0,0182 0,0005 32,66 30,11 1,59 4,95 0,08 3,04
S1.2 25,50 1,576 0,0342 0,0083 21,12 25,50 5,43 0,62
S1.3 23,44 1,604 0,8653 0,1326 19,00 23,44
S1.4 28,29 1,612 0,0196 0,0008 34,53 28,29 3,89 4,96 0,04 3,05 5,46 0,21 3,25
S2.1 28,19 0,964 0,0178 0,0018 32,66 28,19 4,64 4,95 0,07 5,06
S2.2 27,43 0,964 0,0441 0,0126 21,12 27,43 5,43 2,18
S2.3 20,03 0,922 0,5027 0,0745 19,00 20,03
S2.4 28,87 0,906 0,0213 0,0013 34,53 28,87 6,25 4,96 0,07 5,39 5,46 0,10 5,91
S3.1 28,09 0,51 0,0142 0 32,66 28,09 8,95 4,95 0,16 9,39
S3.2 27,96 0,504 0,4655 0,0547 21,12 27,96 5,43 4,21 2,42
S3.3 23,48 0,492 0,6672 0,0943 19,00 23,48
S3.4 28,57 0,504 0,0188 0,0002 34,53 28,57 11,82 4,96 0,19 9,46 5,46 0,69 9,46
S4.1 30,49 0,216 0,0228 0,001 32,66 30,49 10,04 4,95 1,79 14,63
S4.2 24,08 0,194 0,4177 0,063 21,12 24,08 5,43 4,59 4,33
S4.3 25,66 0,208 0,8847 0,1144 19,00 25,66
S4.4 27,66 0,228 0,0228 0,0009 34,53 27,66 30,10 4,96 1,8 13,86 5,46 3,46 8,77
S5.1 30,24 0,1 0,0298 0,0014 32,66 30,24 24,18 4,95 2,06 28,90
S5.2 22,47 0,096 0,3762 0,0362 21,12 22,47 5,43 5,12 3,22
S5.3 20,65 0,104 0,6568 0,0775 19,00 20,65
S5.4 27,93 0,11 0,0249 0,0007 34,53 27,93 59,92 4,96 1,22 34,00 5,46 2,97 22,63
S6.1 29,97 0,066 0,0291 0,001 32,66 29,97 40,63 4,95 2,37 39,09
S6.2 19,83 0,064 0,3212 0,0291 21,12 19,83 20,08 5,43 5,11 5,00
S6.3 20,05 0,056 0,6351 0,0741 19,00 20,05
S6.4 28,58 0,056 0,0324 0,002 34,53 28,58 106,22 4,96 2,31 47,32 5,46 4,11 24,10
S7.1 29,98 0,026 0,3348 0,0308 32,66 29,98 103,01 4,95 2,68 87,30
S7.2 20,03 0,024 0,2666 0,0229 21,12 20,03 45,35 5,43 5,15 11,66
S7.3 18,99 0,022 0,4826 0,0519 19,00 18,99 0,76
S7.4 30,27 0,02 0,2608 0,0228 34,53 30,27 212,75 4,96 3,37 79,50 5,46 4,67 39,50
S8.1 31,36 0,014 0,366 0,036 32,66 31,36 92,50 4,95 3,49 104,28
S8.2 21,34 0,008 0,2632 0,0222 21,12 21,34 5,43 5,28 18,75
S8.3 18,54 0,013 0,2028 0,0137 19,00 18,54 35,39
S8.4 31,68 0,016 0,2205 0,0173 34,53 31,68 178,02 4,96 3,5 91,25 5,46 4,76 43,75
S9.1 32,49 0,012 0,2613 0,0217 32,66 32,49 13,47 4,95 3,78 97,50
S9.2 20,80 0,008 0,1883 0,0118 21,12 20,80 5,43 5,26 21,25
S9.3 18,10 0,018 0,2065 0,014 19,00 18,10 50,19
S9.4 31,93 0,006 0,2324 0,0195 34,53 31,93 431,9 4,96 3,66 216,66 5,46 4,78 113,33
Cu ConT 1 31,58 0,2077 0,0122 4,95
Cu ConT 2 33,73 0,1941 0,0111
Zn ConT1 20,38 0,1858 0,0099 5,51
Zn ConT2 21,86 0,1859 0,0098 5,34
DOC ConT1 18,10 0,196 0,0099
DOC ConT2 19,91 0,1958 0,0102
CuZnConT1 33,58 0,2022 0,012 4,89 5,32
CuZnConT2 35,47 0,2066 0,0127 5,03 5,6

Appendix B 138

Fig. B1 Results of batch test with pulse doses of Cu
2+
[Cu
2+
5.0 mg L
-1
with total doses of 20
mg L
-1
, Fe
0
0.5 g L
-1
, DO 0 mg L
-1
, T 20C at pH 4.0-7.0 (controlled) at 150 rpm].

2+
using Fe
0
and rusted
(corroded) Fe
0
at 0.5 g L
-1
each [Cu
2+
5.0 mg L
-1
-1
, T 20C at pH
5.0 (controlled) at 150 rpm].

0,0
0,2
0,4
0,6
0,8
1,0
0 5 10 15 20 25
C
t
/
C
0
Time (hrs)
rusted iron
ZVI

y = 1,131x + 0,794
R = 0,977
y = 3,603x + 0,112
R = 0,985
0,0
10,0
20,0
30,0
40,0
50,0
60,0
0,0 5,0 10,0 15,0 20,0
d
i
s
s
o
l
u
t
e
d

F
e
T
m
g

L
-
1
Cu
T
mg L
-1
(removed)
ZVI
Rusted ZVI
Appendix B 139

2+
using Fe
0
and GFH at 0.5 g L
-
1
each [Cu
2+
5.0 mg L
-1
-1
, DO 0 mg L
-1
, T 20C at pH 5.0
(controlled) (pH 6.0 for GFH) at 150 rpm].

Fig. B4 Mapping of SEM and EDX spectra of iron from typical batch experiment in
oxygenate solution

Fig. B5 SEM spectra of iron from batch experiment containg Pb
2+
[Pb
2+
5.0 mg L
-1
, Fe
0
0.5 g
L
-1
, T 20C, DO 0 mg L
-1
, pH 5.0 (controlled) at 150 rpm] (in supplement to Fig. 4.2)

0.0
2.5
5.0
7.5
10.0
0 5 10 15 20 25
C
u
T
Time (h)
Adsorption on GFH
Cementation
Appendix B 140

-250
-200
-150
-100
-50
0
50
100
150
0 500 1000
d
N
(
E
)
/

d
E
AES Analysis- Pb
0
5 mg L
-1
D0 0 mg L
-1
, pH6, MP2
89,5
Pb2O3?
148,5
S
510
487,5
268
C ads
702,5
Fe2O3
-200
-150
-100
-50
0
50
100
150
0 500 1000
AES-analysis, Pb
0
5 mg L
-1
, D0 0 mg L
-1
, pH6, MP1
89,5
Pb2O3?
148
S
267,5
C ads
473,5
488
508
649
702
Fe2O3
-400
-300
-200
-100
0
100
200
300
0 500 1000
d
N
(
E
)
/
d
E
AES-survey, Cu XPS Re2, MP2
149
474
488
507,5
592
650
702
Fe2O3
918,5
Cu
-500
-400
-300
-200
-100
0
100
200
300
400
0 500 1000
AES-survey, Probe Cu-XPS-Re2, MP1
41 50 56,5/ 58,5
474
487,5
507,5
591
649,5
702
Fe
2
O
3
919
Cu
-400
-300
-200
-100
0
100
200
300
400
0 500 1000
d
N
(
E
)
/

d
E
kinetic energy [eV]
AES-survey, Probe Cu5 D00 S1, MP 2
41,5 50 56,5/ 57,5
473
488
507,5
591,5
649
702,5
Fe
2
O
3
836
916,5
Cu
2
O
-500
-400
-300
-200
-100
0
100
200
300
400
500
0 500 1000
kinetic energy [eV]
Cu
0
5 mg L
-1
D0 0 mg L
-1
pH 6
42,5 50 57,5
473,5
487,5
507,5
591,5
649,5
701
Fe
2
O
3
837,5 917
Cu
2
O

Fig. B6 Auger spectra of precipitates after experiments [Cu
i
2+
or Pb
i
2+
5.0 mg L
-1
, Fe
0

3
-1
, T 20 1.0 C, 150 rpm
in SSWR] (supplement data for Fig. 4.9 and B4).
Appendix B 141

Fig. B7 SEM and EDX spectra of precipitates after experiments [Cu
i
2+
10 mg L
-1
, Zn
i
2+
10 mg L
-1
, 0.5 g Fe
0
L
-1
, pH 7.0 (controlled), , T 20 1.0
C, 150 rpm in SSWR] (supplement data for Fig. 4.10).

Appendix B 142

Table B12 Determination of pH impacts on breakthrough curves (Fig. 4.11 and Fig. 4.18)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
with pumice, 5 min EBCT
c/c
0
for Cu
2+
c/c
0
for Zn
2+
Fe
2+

BV pH 4.2 pH 5.8 pH 6.5 pH 4.2 pH 5.8 pH 6.5 pH 4.2 pH 5.8 pH 6.5
37 0,06 0,09 0,10 0,41 0,49 0,33 0,84 0,60 0,32
183 0,09 0,11 0,12 0,54 0,58 0,41 0,84 0,52 0,32
257 0,16 0,10 0,22 0,67 0,76 0,63 0,77 0,42 0,10
403 0,16 0,19 0,22 0,69 0,64 0,64 0,71 0,37 0,06
477 0,17 0,18 0,21 0,67 0,62 0,63 0,67 0,34 0,05
550 0,19 0,19 0,27 0,66 0,55 0,64 0,66 0,35 0,02
746 0,18 0,19 0,41 0,62 0,50 0,66 0,63 0,39 0,08
819 0,17 0,18 0,37 0,56 0,48 0,63 0,60 0,37 0,08
1039 0,17 0,22 0,38 0,53 0,50 0,67 0,66 0,36 0,09
1112 0,19 0,21 0,37 0,58 0,49 0,68 0,62 0,24 0,09
1332 0,23 0,22 0,34 0,62 0,50 0,68 0,60 0,31 0,05
1406 0,24 0,18 0,44 0,65 0,56 0,73 0,67 0,20 0,07
1626 0,32 0,31 0,46 0,67 0,60 0,75 0,54 0,26 0,03
1699 0,37 0,42 0,48 0,68 0,65 0,72 0,51 0,23 0,04
1919 0,38 0,45 0,47 0,68 0,64 0,69 0,50 0,18 0,03
1992 0,36 0,40 0,40 0,68 0,60 0,68 0,44 0,15 0,04
2212 0,42 0,53 0,42 0,73 0,73 0,72 0,44 0,14 0,05
2286 0,44 0,64 0,43 0,75 0,78 0,70 0,42 0,16 0,06
2506 0,51 0,68 0,41 0,84 0,81 0,70 0,48 0,11 0,05
2579 0,52 0,59 0,36 0,83 0,76 0,73 0,38 0,10 0,04
2799 0,56 0,62 0,41 0,83 0,81 0,74 0,35 0,10 0,05
2872 0,59 0,70 0,43 0,85 0,86 0,76 0,36 0,10 0,06
3092 0,62 0,77 0,53 0,85 0,86 0,74 0,35 0,12 0,07
3166 0,75 0,78 0,50 0,90 0,89 0,82 0,32 0,08 0,04
3386 0,75 0,80 0,53 0,92 1,01 0,84 0,31 0,06 0,06
3459 0,77 0,91 0,44 0,98 1,03 0,81 0,34 0,07 0,05
3679 0,89 0,91 0,44 1,03 1,02 0,79 0,26 0,08 0,07
3753 0,89 0,88 0,50 1,03 0,99 0,87 0,24 0,09 0,06
3973 0,94 0,87 0,69 1,07 0,96 0,93 0,23 0,08 0,05

Appendix B 143

Table B13 Determination of DO impacts on breakthrough curves (Fig. 4.12 and Fig. 4.18)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
DO: ~1,3 and 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
c/c
0
for Cu
2+
c/c
0
for Zn
2+
Fe
2+

BV DO 8 DO 1,3 DO 8 DO 1,3 DO 8 DO 1,3
37 0,09 0,04 0,49 0,65 5,96 2,59
183 0,11 0,06 0,58 0,73 5,21 2,67
257 0,10 0,09 0,76 0,77 4,23 3,63
403 0,19 0,10 0,64 0,82 3,73 3,66
477 0,18 0,09 0,62 0,81 3,41 3,23
550 0,19 0,10 0,55 0,71 3,50 3,28
746 0,19 0,11 0,50 0,72 3,87 3,66
819 0,18 0,11 0,48 0,69 3,68 3,72
1039 0,22 0,14 0,50 0,78 3,58 4,32
1112 0,21 0,14 0,49 0,70 2,44 4,30
1332 0,22 0,20 0,50 0,77 3,09 3,98
1406 0,18 0,18 0,56 0,72 1,97 4,02
1626 0,31 0,22 0,60 0,82 2,59 3,61
1699 0,42 0,26 0,65 0,71 2,26 2,99
1919 0,45 0,26 0,64 0,75 1,82 3,04
1992 0,40 0,23 0,60 0,76 1,54 2,49
2212 0,53 0,32 0,73 0,95 1,39 2,57
2286 0,64 0,37 0,78 0,86 1,59 2,34
2506 0,68 0,50 0,81 0,96 1,06 1,94
2579 0,59 0,51 0,76 0,93 1,03 1,57
2799 0,62 0,60 0,81 1,11 0,97 1,48
2872 0,70 0,51 0,86 0,93 1,00 1,11
3092 0,77 0,59 0,86 0,97 1,20 1,28
3166 0,78 0,89 0,80 0,95
3386 0,80 0,79 1,01 0,90 0,62 0,82
3459 0,91 0,74 1,03 1,09 0,71 1,44
3679 0,91 0,73 1,02 1,05 0,82 1,65
3753 0,88 0,99 1,01 0,90 1,83
3973 0,87 0,96 1,10 0,77 1,91

Appendix B 144

Table B14 Determination of conductivity impacts on breakthrough curves (Fig. 4.13
and Fig. 4.18)
Metrix: Cu
2+
/Zn
2+
in roof runoff
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

Conductivity: 80 and 200S cm
-1

Temperature: 20 C
DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
c/c
0
for Cu
2+
c/c
0
for Zn
2+
Fe
2+

BV 80 200 80 200 80 200
37 0,09 0,09 0,49 0,31 5,96 5,06
183 0,11 0,15 0,58 0,40 5,21 5,46
257 0,10 0,32 0,76 0,66 4,23 3,29
403 0,19 0,31 0,64 0,67 3,73 3,15
477 0,18 0,30 0,62 0,64 3,41 3,15
550 0,19 0,31 0,55 0,60 3,50 2,99
746 0,19 0,34 0,50 0,56 3,87 3,05
819 0,18 0,33 0,48 0,56 3,68 2,69
1039 0,22 0,28 0,50 0,56 3,58 2,69
1112 0,21 0,42 0,49 0,57 2,44 2,23
1332 0,22 0,40 0,50 0,62 3,09 1,81
1406 0,18 0,50 0,56 0,68 1,97 1,60
1626 0,31 0,48 0,60 0,69 2,59 1,22
1699 0,42 0,60 0,65 0,77 2,26 0,97
1919 0,45 0,61 0,64 0,76 1,82 0,83
1992 0,40 0,55 0,60 0,74 1,54 0,74
2212 0,53 0,64 0,73 0,82 1,39 0,77
2286 0,64 0,71 0,78 0,85 1,59 0,87
2506 0,68 0,70 0,81 0,86 1,06 0,83
2579 0,59 0,62 0,76 0,81 1,03 0,87
2799 0,62 0,63 0,81 0,84 0,97 0,78
2872 0,70 0,68 0,86 0,87 1,00 0,84
3092 0,77 0,74 0,86 0,90 1,20 1,04
3166 0,78 0,75 0,89 1,09 0,80 1,27
3386 0,80 0,91 1,01 1,00 0,62 0,77
3459 0,91 0,82 1,03 0,97 0,71 0,89
3679 0,91 0,87 1,02 1,03 0,82 1,05
3753 0,88 0,79 0,99 0,99 0,90 1,28
3973 0,87 0,89 0,96 1,02 0,77 0,82

Appendix B 145

Table B15 Determination of temperature impacts on breakthrough curves (Fig. 4.14)
Metrix: Cu
2+
/Zn
2+
in roof runoff (high NOM) and DI

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
c/c
0
for Cu
2+
c/c
0
for Zn
2+
Fe
2+

BV 15 20 35 15 20 35 15 20 35
37 0,14 0,05 0,01 0,90 0,81 0,76 0,89 1,09 0,71
183 0,51 0,26 0,22 0,81 0,91 0,92 2,45 2,86 1,06
257 0,54 0,32 0,28 0,86 0,93 0,91 2,17 1,96 0,78
403 0,67 0,52 0,54 0,95 0,98 0,95 1,99 1,56 0,79
477 0,67 0,61 0,68 0,97 0,98 0,98 1,85 1,64 0,81
550 0,71 0,64 0,87 1,00 1,00 1,00 1,47 1,36 0,55
746 0,69 0,62 0,69 0,97 1,00 0,98 1,68 1,47 0,72
819 0,74 0,65 0,69 1,04 1,00 1,03 1,00 1,17 0,63
1039 0,68 0,59 0,68 1,02 1,04 1,01 1,22 1,25 0,71
1112 0,68 0,56 0,69 1,05 0,95 0,95 1,43 1,46 0,77
1332 0,64 0,62 0,70 0,99 0,96 0,95 1,39 1,78 0,84
1406 0,55 0,72 0,63 1,05 0,94 0,93 2,09 2,19 1,08
1626 0,84 0,81 0,85 1,17 1,01 1,04 2,81 2,84 2,30
1699 0,84 0,84 0,82 1,15 1,04 1,05 2,01 3,15 2,75
1919 0,83 0,81 0,88 1,12 1,00 1,04 2,34 2,82 2,59
1992 0,92 0,92 0,86 1,16 1,05 1,08 1,74 3,05 2,88
2212 0,89 0,65 0,76 1,07 0,92 0,93 1,87 2,40 1,66
2286 0,88 0,86 0,63 1,07 0,96 0,98 1,70 2,44 2,64
2506 0,85 0,81 0,56 1,04 0,92 0,94 1,39 2,55 2,58
2579 0,90 0,83 0,80 0,99 0,92 0,91 0,28 1,54 1,86
2799 0,86 0,86 0,62 1,00 0,91 0,88 0,59 0,24 0,03
2872 0,89 0,86 0,77 1,01 0,94 0,99 0,80 0,22 0,06
3092 0,85 0,80 0,67 0,98 0,89 0,89 1,39 1,71 1,24
3166 0,87 0,85 0,74 1,02 0,91 0,97 1,35 1,85 2,79
3386 0,72 0,81 0,70 0,89 0,84 0,89 1,40 1,95 2,85
3459 0,56 0,50 0,42 0,76 0,69 0,84 0,18 0,63 0,49
3679 0,82 0,63 0,60 1,05 1,07 0,19 0,43 3,79
3753 0,90 0,78 0,55 1,06 0,95 0,43 0,20 0,47 0,42
3973 0,86 0,81 0,68 0,92 0,94 0,85 0,18 0,59 0,53
4046 0,94 0,78 0,69 1,00 0,91 0,79 0,20 0,62 0,20
4266 0,92 0,83 0,70 1,00 0,89 0,80 0,24 0,65 0,16
4339 0,88 0,76 0,68 0,95 0,83 0,69 0,16 0,48 0,62
4559 0,97 0,85 0,64 1,00 0,89 0,70 0,19 0,71 0,58
4779 0,95 0,77 0,67 0,93 0,81 0,70 0,23 0,64 0,55
4999 0,95 0,74 0,62 1,04 0,83 0,69 0,25 0,65 0,62
5219 0,86 0,67 0,59 0,97 0,80 0,69 0,20 0,67 0,45
5439 0,94 0,73 0,65 1,01 0,83 0,67 0,16 0,54 0,31
5659 0,84 0,66 0,62 0,89 0,78 0,67 0,18 0,54 0,28
5879 0,89 0,69 0,68 0,98 0,81 0,76 0,15 0,63 0,29
6099 0,89 0,73 0,68 0,97 0,83 0,75 0,22 0,56 0,22

Appendix B 146

Table B16 Determination of NOM impacts on breakthrough curves (Table 4.2 and
Fig. 4.25)
Metrix: Cu
2+
/Zn
2+
in roof runoff
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

Conductivities: 80-200
Temperature: 20 C
Fe
0
bed: 15 g Fe
0
BV DI water BV NOM 2,9-10,3 BV NOM 5,0-61,0
C/C0 Cu Zn Cu Zn Cu Zn
37 0,013 0,318 37 0,088 0,311 0 0,028 0,489
306 0,108 0,583 183 0,151 0,402 186 0,184 0,709
419 0,117 0,620 257 0,315 0,663 342 0,031 0,715
566 0,128 0,633 403 0,309 0,672 553 0,317 0,759
713 0,130 0,623 477 0,304 0,639 818 0,586 0,885
872 0,132 0,673 550 0,309 0,602 935 0,676 0,975
978 0,134 0,644 746 0,343 0,562 1104 0,777 1,032
1165 0,145 0,690 819 0,334 0,562 1235 0,656 0,862
1274 0,153 0,676 1039 0,276 0,560 1384 0,586 0,878
1455 0,151 0,639 1112 0,415 0,571 1539 0,647 0,884
1772 0,144 0,578 1332 0,401 0,619 1714 0,470 0,847
1858 0,094 0,571 1406 0,503 0,676 1833 0,486 0,882
2420 0,168 0,706 1626 0,476 0,695 2057 0,805 0,931
2628 0,178 0,614 1699 0,598 0,766 2383 0,671 0,902
2762 0,212 0,625 1919 0,605 0,761 2615 0,750 0,915
2909 0,221 0,633 1992 0,552 0,735 2701 0,685 0,915
2995 0,223 0,632 2212 0,643 0,816 2860 0,663 0,977
3068 0,251 0,625 2286 0,705 0,846 2994 0,654 1,157
3215 0,234 0,649 2506 0,696 0,858 3169 0,794 1,366
3361 0,332 0,682 2579 0,622 0,807 0,936 1,357
3512 0,321 0,713 2799 0,629 0,840 1,452
3581 0,370 0,709 2872 0,684 0,868 1,545
3817 0,319 0,756 3092 0,745 0,901
3940 0,783 3166 0,749 1,086
3386 3386 0,909 1,004
3459 3459 0,817 0,969
3679 3679 0,872 1,026
3753 3753 0,786 0,989
3973 3973 0,886 1,022
4046 0,805 0,952
4266 0,884 1,022
4339 0,798 0,954
4559 0,812 0,958

Appendix B 147

Table B17 Determination of NOM impacts on breakthrough curves (Table 4.2 and Fig. 4.26)
Metrix: Cu
2+
/Zn
2+
in roof runoff
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
BV Cu
T
Zn
T
Fe
T

37 0,052 0,811 1,09
183 0,262 0,910 2,86
257 0,320 0,931 1,96
403 0,519 0,978 1,560
477 0,612 0,978 1,640
550 0,636 0,998 1,360
746 0,618 1,000 1,470
819 0,645 1,002 1,170
1039 0,587 1,037 1,250
1112 0,562 0,953 1,460
1332 0,616 0,959 1,780
1406 0,722 0,935 2,190
1626 0,812 1,008 2,840
1699 0,842 1,039 3,150
1919 0,806 1,002 2,820
1992 0,916 1,053 3,050
2212 0,648 0,923 2,400
2286 0,860 0,959 2,440
2506 0,810 0,919 2,550
2579 0,828 0,919 1,540
2799 0,855 0,906 0,240
2872 0,860 0,941 0,220
3092 0,801 0,888 1,710
3166 0,853 0,911 1,850
3386 0,808 0,843 1,950
3459 0,499 0,688 0,630
3679 0,632 1,066 0,430
3753 0,779 0,947 0,470
3973 0,808 0,939 0,590
4046 0,776 0,912 0,620
4266 0,830 0,893 0,650
4339 0,761 0,827 0,48
4559 0,849 0,889 0,71
4779 0,768 0,808 0,64
4999 0,744 0,835 0,65
5219 0,666 0,800 0,67
5439 0,726 0,835 0,54
5659 0,658 0,784 0,54
5879 0,695 0,811 0,63
6099 0,726 0,827 0,56

Appendix B 148

Table B18 pH before and after the breakthrough column experiments (Fig. 4.11-4.14)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C

Fe
0
bed: 15 g Fe
0
pH data Inlet Outlet
BVT pHi 4 pHi 6,5 pHi 5,8 DO 0 Cond. 200 pHi 4 pHi 6,5 pHi 5,8 DO 0 Cond. 200
37 6,06 6,23 6,18 6,24 6,14
183 4,3 6,44 5,59 5,62 5,6 6,22 6,5 6,21 6,26 6,17
257 6,07 6,3 6,2 6,24 6,12
403 6,16 6,42 6,26 6,34 6,18
477 6,08 6,26 6,2 6,22 6,13
550 4,24 6,58 5,85 5,85 5,85 6,18 6,21 6,27 6,19 6,16
746 6,17 6,19 6,21 6,21 6,1
819 4,33 6,35 5,05 5,05 5,05 6,04 6,2 5,98 6,17 5,98
1039 5,91 5,97 5,93 5,79 5,89
1112 4,11 6,36 5,71 5,78 5,63 6,09 6,45 6,09 6,08 6,03
1332 6,05 6,15 6,13 6,12 6,02
1406 4,14 6,35 5,65 5,65 5,65 6,09 6,3 6,13 6,15 6,12
1699 4,16 6,47 5,67 5,93 5,86 5,99 6,33 6,2 6,29 6,26
1992 4,15 6,41 5,98 6,03 6,2 5,89 6,24 6,17 6,18 6,1
2212 5,92 6,3 6,1 6,26 6,07
2286 4,14 6,43 5,83 6 6,12 5,84 6,29 6,02 6,2 6,11
2506 5,65 6,24 6,11 6,16 5,98
2579 4,11 6,46 6,19 6,25 6,17 5,54 6,45 6,16 6,13 6,28
2799 5,51 6,32 6,25 6,3 6,27
2872 5,42 6,5 6,35 6,43 6,28
3092 4,13 6,51 6,11 5,96 5,77 5,32 6,38 6,38 6,2 6,2
3166 4,12 6,46 6,16 6,08 6,08 5,31 6,22 6,17 6,2 6,11
3386 6,6 5,86 5,84 5,27 6,6 6,36 6,37 6,29
3459 4,11 6,45 6,02 6,05 6,03 4,89 6,32 6,14 6,1 6,11
3679 4,95 6,31 6,35 6,32 6,23
3753 4,16 6,56 6,26 6,09 6,09 4,88 6,38 6,32 6,37 6,23
3973 4,84 6,28 6,34 6,08 6,28
average 4,17 6,46 5,85 5,87 5,85 5,72 6,31 6,19 6,21 6,14
conductivity Inlet Outlet
BVT 1 2 3 4 5 1 2 3 4 5
37 107,4 107 95,5 90,6 239
183 99,5 105,3 82,3 87 213
477 96,5 106,5 79,6 80,6 200
746 81,3 75,4 66,1 70 181,1
1112 121,7 93,1 68,9 69,9 197,6 87 89 68,1 69,5 198
1406 105,9 90,1 71,5 76,3 192,8 85,9 87,8 71,8 76 183,4
1699 104,2 90,9 77,6 76,6 193,3 88,1 88,2 77,2 77,8 186,9
1992 108 94 82 85 203 86,2 89,6 81,7 84,3 203
2286 120,3 94,3 82,9 85,7 202 95,6 91,3 82,8 85,7 202
2579 122,4 94,4 83,7 90,1 192,5 96,3 91,8 83,2 84,1 193,2
3092 127,6 97,8 87,1 88,4 225 104,3 94,1 86,5 90,4 216
3166 123,2 98,6 96,7 88,8 224 99,1 94,1 95,4 90,1 218
3459 124,4 100,9 90,2 90,6 236 110,5 97,8 89,8 92,7 237
average 117,52 94,90 82,29 83,49 207,36 95,21 93,68 81,54 82,98 205,43

Appendix B 149

Table B19 pH before and after the breakthrough column experiments (Cont.)
Metrix: Cu
2+
/Zn
2+
in roof runoff
Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
Fe
0
bed: 15 g Fe
0
Inlet Outlet
BV BV 30C 15C high NOM 20C DI Int. NOM 30C 15C high NOM 20C DI Int. NOM
37 9 6,26 6,36 6,34 6,45 6,66 6,48 6,42 6,52 7,03 6,5 6,79 6,78
183 46 6 6,22 6,39 6,2 6,51 6,34
257 64 6,1 6,35 6,39 6,3 6,63 6,43
403 101 6,14 6,18 6,16 6,12 6,54 6,25 6,14 6,43 6,41 6,3 6,65 6,43
550 138 6,06 6,15 6,18 6,09 6,55 6,28 6,1 6,33 6,37 6,21 6,56 6,3
819 205 6,05 6,12 6,27 6,24 6,63 6,31 6,14 6,32 6,34 6,17 6,51 6,29
1112 278 6 5,94 6,12 5,91 6,24 5,93 6 6,22 6,35 6,06 6,3 6,07
1406 351 6,02 5,99 5,99 6,02 6,26 6,15 6,02 6,33 6,09 6,08 6,35 6,19
1699 425 5,15 5,08 5,85 5,14 5,26 5,1 5,55 5,61 5,76 5,5 5,9 5,53
1992 498 4,98 5,02 5,65 4,98 5,08 4,98 5,37 5,38 5,34 5,27 5,4 5,24
2286 571 5,11 5,07 5,19 5,11 5,79 5,15 5,54 5,38 5,76 5,54 5,46 5,5
2579 645 5,28 5,06 5,73 5,26 5,17 5,34 5,57 5,41 5,73 5,5 5,44 5,52
2872 718 4,75 4,83 5,68 4,9 5,01 4,55 5,32 5,2 5,58 5,25 4,97 4,79
3166 791 4,91 4,93 5,77 4,98 5,07 5 5,4 5,16 5,53 5,32 5,13 5,23
3679 920 5,42 5,54 5,83 5,38 5,04 5,04 5,85 5,79 6,05 5,76 5,24 5,44
3973 993 5,59 5,6 5,71 5,69 5,12 5,07 5,71 6,01 5,95 5,52 5,37 5,32
4413 1103 5,37 5,65 5,65 5,41 5,18 5,21
4853 1213 5,48 5,5 5,79 5,34 5,12 5,08 5,38 5,61 5,98 5,63 5,26 5,33
5293 1323 5,67 5,59 5,87 5,69 5,03 5,01 5,61 6 6,15 5,21 5,4 5,27
5733 1433 5,83 5,89 6,01 5,76 5,09 5,08 5,78 6,18 5,87 5,43 5,35 5,34
6173 1543 5,81 5,79 6,01 5,95 5,08 5,1 5,79 6,16 5,9 5,47 5,36 5,29
average 5,55 5,56 5,89 5,59 5,74 5,55 5,83 5,92 6,07 5,84 5,95 5,84

Appendix B 150

0,0
0,2
0,4
0,6
0,8
1,0
1,2
0 700 1400 2100 2800 3500
C
/
C
0
BVT
0 700 1400 2100 2800 3500
BVT
FE0SL
Fe0SL/PM
Fe0LW
FE0particle/PM control
Cu
2+ Zn
2+
Fig. B8 Breakthrough behaviors of copper and zinc from columns containing iron from various
sources. The Fe
0
-SL represent a spiral iron shape illustrated in Table 3.2 and the Fe
0
-LW is the
iron used in the study of Ludwig (2007).

Fig. B9 Breakthrough behaviors of copper and zinc in a Fe
0
/PM column (50% v/v 47 g).
The column was operated as following; Cu
2+
6.5 mg L
-1
, Zn 7.3 mg L
-1
using winter TU-
runoff, 9.6 cm bed high, diameter 2.5 cm, EBCT 1.8 min. The test was terminated due to
clogging. Lines represent simulation.
0.0
0.3
0.5
0.8
1.0
0 1000 2000 3000 4000 5000
C
/
C
0
BVT
Zinc - model Cu 50%ZVI
Zn 50%ZVI Copper - model

Appendix B 151

Fig. B10 Breakthrough behaviors of copper and zinc in Fe
0
/PM column (10% v/v) in UFA
runoff. The column was operated as following; [Cu
2+
5 mg L
-1
, Zn 5-7 mg L
-1
, the UFA water
was collected in the summer period, pH~6.2, DOC
in
5.9 and DOC
out
5.1 mg L
-1
, EBCT 5
min].

0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 1000 2000 3000 4000 5000
C
/
C
0
BVT

Appendix B 152

Table B20 Model prediction of breakthrough curves at EBCT 40 min (example of Fig. 4.21)
PSDM Results -- Filename = "c:\etdot10\ads\examples\trial big bed.dat"
Time BVT Usage Rate Copper-Middel Zinc-middel DOC
Minutes - m
3
kg
-1
C, mg L
-1
C, mg L
-1
C, mg L
-1

0,00 0,00 0,00 0,00 0,00 0,00
2880,00 576,00 1,49 0,00 1,29 0,86
5760,00 1152,00 2,98 0,03 2,03 2,35
8640,00 1728,00 4,48 0,14 2,79 2,41
11520,00 2304,00 5,97 0,44 3,53 2,22
14400,00 2880,00 7,46 1,03 4,20 2,12
17280,00 3456,00 8,95 1,90 4,77 2,06
20160,00 4032,00 10,44 2,85 5,23 2,03
23040,00 4608,00 11,94 3,71 5,60 2,02
25920,00 5184,00 13,43 4,35 5,88 2,01
28800,00 5760,00 14,92 4,75 6,10 2,00
31680,00 6336,00 16,41 4,97 6,28 2,00
34560,00 6912,00 17,90 5,07 6,42 2,00
37440,00 7488,00 19,40 5,09 6,53 2,00
40320,00 8064,00 20,89 5,09 6,63 2,00
46080,00 9216,00 23,87 5,05 6,75 2,00
48960,00 9792,00 25,36 5,04 6,79 2,00
51840,00 10368,00 26,86 5,03 6,83 2,00
54720,00 10944,00 28,35 5,03 6,86 2,00
57600,00 11520,00 29,84 5,02 6,87 2,00
60480,00 12096,00 31,33 5,02 6,89 2,00
63360,00 12672,00 32,82 5,02 6,91 2,00
66240,00 13248,00 34,32 5,01 6,92 2,00
69120,00 13824,00 35,81 5,01 6,94 2,00
72000,00 14400,00 37,30 5,01 6,94 2,00
74880,00 14976,00 38,79 5,01 6,95 2,00
77760,00 15552,00 40,28 5,01 6,96 2,00
80640,00 16128,00 41,78 5,01 6,97 2,00
83520,00 16704,00 43,27 5,00 6,97 2,00
86400,00 17280,00 44,76 5,00 6,98 2,00
89280,00 17856,00 46,25 5,00 6,98 2,00
92160,00 18432,00 47,74 5,00 6,98 2,00
95040,00 19008,00 49,24 5,00 6,98 2,00
97920,00 19584,00 50,73 5,00 6,99 2,00
100800,00 20160,00 52,22 5,00 6,99 2,00
103680,00 20736,00 53,71 5,00 6,99 2,00
106560,00 21312,00 55,20 5,00 6,99 2,00
109440,00 21888,00 56,70 5,00 6,99 2,00
115200,00 23040,00 59,68 5,00 6,99 2,00
118080,00 23616,00 61,17 5,00 6,99 2,00
120960,00 24192,00 62,66 5,00 7,00 2,00
123840,00 24768,00 64,16 5,00 7,00 2,00
126720,00 25344,00 65,65 5,00 7,00 2,00
129600,00 25920,00 67,14 5,00 7,00 2,00
132480,00 26496,00 68,63 5,00 7,00 2,00
135360,00 27072,00 70,12 5,00 7,00 2,00
138240,00 27648,00 71,62 5,00 7,00 2,00
141120,00 28224,00 73,11 5,00 7,00 2,00
144000,00 28800,00 74,60 5,00 7,00 2,00

Appendix B 153

Table B21 Model prediction of breakthrough curves at Fe
0
50% (example of Fig. 4.21)
Time BVT Usage Rate Copper-Middel Zinc-middel DOC
Minutes - m
3
kg
-1
C, mg L
-1
C, mg L
-1
C, mg L
-1

0 0 0,00 0,00 0,00 0,00
5760 320 0,33 0,00 0,00 -0,01
11520 640 0,67 0,00 0,00 0,01
17280 960 1,00 -0,01 0,01 0,00
23040 1280 1,33 -0,01 0,00 0,00
28800 1600 1,67 0,00 0,00 0,01
34560 1920 2,00 0,00 0,00 0,02
40320 2240 2,33 0,01 0,01 0,46
46080 2560 2,67 0,01 0,01 3,06
51840 2880 3,00 0,02 0,04 3,28
57600 3200 3,33 0,02 0,06 2,70
63360 3520 3,66 0,01 0,09 2,35
69120 3840 4,00 0,01 0,12 2,36
74880 4160 4,33 0,00 0,17 2,62
80640 4480 4,66 -0,01 0,26 2,84
86400 4800 5,00 -0,01 0,37 2,85
92160 5120 5,33 -0,01 0,51 2,68
97920 5440 5,66 0,00 0,70 2,49
103680 5760 6,00 0,01 0,93 2,37
115200 6400 6,66 0,05 1,59 2,34
120960 6720 7,00 0,06 2,01 2,35
126720 7040 7,33 0,06 2,49 2,30
132480 7360 7,66 0,05 3,03 2,24
138240 7680 8,00 0,04 3,63 2,20
144000 8000 8,33 0,04 4,31 2,17
149760 8320 8,66 0,07 5,04 2,16
155520 8640 9,00 0,24 5,78 2,13
161280 8960 9,33 0,64 6,46 2,09
167040 9280 9,66 1,37 7,01 2,07
172800 9600 9,99 2,40 7,36 2,03
178560 9920 10,33 3,57 7,52 2,01
184320 10240 10,66 4,55 7,53 2,00
190080 10560 10,99 5,17 7,43 2,00
195840 10880 11,33 5,43 7,30 2,00
201600 11200 11,66 5,47 7,16 1,99
207360 11520 11,99 5,33 7,05 2,00
213120 11840 12,33 5,16 6,97 2,00
218880 12160 12,66 5,02 6,90 2,00
224640 12480 12,99 4,94 6,89 2,00
230400 12800 13,33 4,92 6,90 2,00
236160 13120 13,66 4,91 6,91 2,00
241920 13440 13,99 4,93 6,92 2,00
247680 13760 14,33 4,95 6,94 2,00
253440 14080 14,66 4,96 6,95 2,00
259200 14400 14,99 4,97 6,95 2,00
264960 14720 15,33 4,98 6,96 2,00
270720 15040 15,66 4,99 6,97 2,00
276480 15360 15,99 5,00 6,97 2,00
282240 15680 16,32 5,00 6,98 2,00
288000 16000 16,66 5,00 6,98 2,00

Appendix B 154

Table B22 Characterization of runoff NOM in term of hydrophobicity (Table 4.3)
Hydrophilic Hydrophobic
Column Blind 1 HPI TPI Blind 2 ( HPI x 1)-B1 TPI-B2 HIX + HOS Total DOC Total DOC
TUSW 0,52 6,76 7,71 1,77 6,24 5,94 12,18 12,975 12,975
CUR 0,57 3,25 3,14 1,37 2,68 1,77 4,45 5,7 5,7
TUFF 0,55 7,73 3,07 0,62 7,18 2,45 9,63 9,5 9,5
SSWR 0,76 0,97 2,01 1,75 0,21 0,26 0,47 0,39 0,39

TUSW CUR TUFF SSWR
DOC total [mg C] 1,04 0,46 0,76 0,03
HIX [mg C] 0,50 0,21 0,57 0,02
HIS [hydrophilic acid]
HIN [hydrophilic neutral]
HIB [hydrophilic base]
HOS [hydrophobic acid] 0,30 0,09 0,12 0,01
HIX [hydrophilic fraction] 48,09 47,02 75,58 53,85
HIS [hydrophilic acid]
HIN [hydrophilic neutral]
HIB [hydrophilic base]

HOX [hydrophobic Fraction] 51,91 52,98 24,42 46,15
HOS [hydrophobic acid] 28,61 19,41 16,12 41,67
HON [hydrophobic neutral] 23,29 33,57 8,30 4,49

6,74 3,02 2,32 0,18
6,24 2,68 7,18 0,21

36,41 9,74 10,25 4,27
18,50 31,01 22,63 4,22

17,14 27,52 70,05 4,92
64,80 45,30 7,60 93,40

100,44 103,82 100,28 102,53

100,00 100,00 100,00 100,00
18,06 27,18 22,35 1,68
*The method of calculation was described in Appendix C

Appendix B 155

Table B23 Determination of NOM impacts using batch test at initial acidic pH
i
2.5 (Fig. 4.24)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

Conductivity: 80-200

Temperature: 20 C
Fe
0
bed: 0,5 g Fe
0

Zn removal at low pH

Cu removal

source/h 0 6 18 51 192 source Conc. Initial B1 B2 Average Error
%
Removal
TU FF 1 1,00 1,20 1,07 0,82 0,49 TU-SW 4,68 0,16 0,01 0,09 0,15 98
TU FF 2 1,00 1,24 1,00 0,76 0,46

TUFF 5,27 1,30 1,13 1,22 0,17 77

average 1,00 1,22 1,03 0,79 0,48 SSWR 5,14 0,33 0,23 0,28 0,10 95
SD 0,00 0,02 0,03 0,03 0,02

CUR 6,30 0,03 0,02 0,03 0,01 100

Error 0,00 0,04 0,07 0,06 0,03

TU SW 1 1,00 1,06 0,67 0,61 0,58 Zn removal
TU SW 2 1,00 1,00 0,67 0,60 0,47 source Conc. Initial B1 B2 Average Error
%
Removal
average 1,00 1,03 0,67 0,60 0,53

TU-SW 10,22 5,75 4,65 5,20 1,10 49

SD 0,00 0,03 0,00 0,00 0,05 TU-FF 5,74 2,45 2,33 2,39 0,12 58
Error 0,00 0,06 0,01 0,00 0,11

SR-NOM 5,41 2,33 2,36 2,35 0,03 57

Cu R 1 1,00 0,58 0,52 0,40 0,16 CU-R 6,29 0,89 1,28 1,09 0,39 83
Cu R 2 1,00 0,71 0,53 0,44 0,24
average 1,00 0,65 0,53 0,42 0,20 DOC removal
SD 0,00 0,07 0,01 0,02 0,04

source Conc. Initial B1 B2 Average Error
%
Removal Load
Error 0,00 0,13 0,01 0,04 0,08

[mg L
-1
]
SSWR 1 1,00 0,86 0,75 0,65 0,48

TU-SW 22,69 7,40 8,39 7,89 1,00 65 29,6
SSWR 2 1,00 0,90 0,93 0,69 0,49 TUFF 12,66 8,97 8,99 8,98 0,02 29 7,4
average 1,00 0,88 0,84 0,67 0,48

SR NOM 3,80 0,77 0,95 0,86 0,18 77 5,9
SD 0,00 0,02 0,09 0,02 0,00 CUR 16,48 3,62 3,48 3,55 0,14 78 25,9
Error 0,00 0,04 0,18 0,05 0,01
Control 1 1,00 1,10 1,10 1,12 1,12

Control 2 1,00 1,09 1,02 1,12 1,09
average 1,00 1,09 1,06 1,12 1,10
SD 0,00 0,00 0,04 0,00 0,02

Error 0,00 0,00 0,07 0,00 0,03

Appendix B 156

Table B24 Determination of NOM impacts using batch test at initial runoff pH
i
5.0 (Fig. 4.23)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

Conductivity: 80-200

Temperature: 20 C
Fe
0
bed: 0,5 g Fe
0

Cu removal
source/h 0 6 18 51 192
%
Removal
TU-SW 1,00 0,37 0,08 0,02 - 98
TUFF 1,00 0,19 0,02 0,02 - 98
SSWR 1,00 0,08 0,00 0,00 - 100
CuR 1,00 0,38 0,00 0,02 - 98
Zn removal
vials/h 0 6 18 51 192
%
Removal
TU SW 1,00 0,81 0,54 0,30 0,03 70
TU FF 1,00 0,59 0,39 0,07 0,00 93
SR NOM 1,00 0,53 0,41 0,24 0,02 76
Cu R 1,00 0,52 0,20 0,04 0,00 96
DOC removal
vials/h 0 50 c/c
0

%
Removal Load [mg g
-1
]
TU SW 28,60 14,62 0,51 49 28,0
TU FF 12,20 10,67 0,87 13 3,1
SR NOM 5,82 4,31 0,74 26 3,0
Cu R 9,74 4,98 0,51 49 9,5

Appendix B 157

0
10
20
30
40
50
60
70
80
30 40 50 60 70 80 90 100
L
C
-
O
C
D

S
i
g
n
a
l
s
TU-SW 0 h
TU-SW 200 h
0
5
10
15
20
25
30
35
40
45
50
30 40 50 60 70 80 90 100
L
C
-
O
C
D

S
i
g
n
a
l
s
CuR 0 h
CuR 200 h
0
5
10
15
20
25
30 40 50 60 70 80 90 100
L
C
-
O
C
D

S
i
g
n
a
l
s
Retention time (min)
SR NOM 0 h
SR NOM 200 h

Fig. B11 LC-OCD diagrams of NOM runoff from various sources before and after Fe
0

treatment (in supplement to Fig. 4.22 and 4.27)

Appendix B 158

30 40 50 60 70 80 90 100
O
C
D

S
i
g
n
a
l
s
TU-SW
CU-SW
TU-FF
SR-NOM

Fig. B12 A comparison of LC-OCD diagrams of NOM runoff from various sources
(in supplement to Fig. 4.22)

Fig. B13 Example of acidity determination of HOS fraction of runoffs (in supplement to
Table 4.4)
2
3
4
5
6
7
8
9
10
11
0.00 2.00 4.00 6.00 8.00
NaOH- added
p
H
DI Control
SRNOM HOS
TU New HOS
TU FF HOS
Copper Roof HOS

Appendix B 159

Table B25 Results of field study of composited column (Fig. 4.28-4.29)
Metrix: Cu
2+
/Zn
2+
from roof runoffs
Fe
0
bed: 100 g Fe
0
Site: copper roof runoff at UDK Site: zinc roof runoff at TU
Date: 26.07.2005

Date: 25.07.2005
Day of exp. 1 Day of exp. 1
Copper concentration Zn concentration
Length/Column 1 2 3 4 Column 1 2 3 4
0 4,82 4,87 4,845 4,845 0 0,96 0,89 0,925 0,925
7 4,13 1,26 1,2 2,66 7 1 0,64 0,39 0,33
14 3,68 0,55 0,86 1,71 14 0,87 0,43 0,24 0,23
21 3,42 0,44 0,73 1,02 21 0,78 0,32 0,17 0,47
28 3,13 0,34 0,66 0,8 28 0,8 0,24 0,13 0,08
35 2,85 0,28 0,57 0,55 35 0,78 0,23 0,16 0,06
42 2,41 0,19 0,44 0,35 42 0,7 0,17 0,11 0,05
49 2,4 0,13 0,39 0,27 49 0,58 0,36 0,22 0,02
56 1,83 0,13 0,38 0,18 56 0,49 0,07 0,04 0,03
Date: 26.07.2005 Date: 28.07.2005
Zinc concentration Zinc concentration
Column 1 2 3 4 Column 1 2 3 4
0 0,82 0,79 0,805 0,805 0 0,65 0,61 0,63 0,63
7 0,74 0,52 0,31 0,26 7 0,61 0,49 0,25 0,3
14 0,65 0,5 0,24 0,2 14 0,55 0,44 0,2 0,06
21 0,64 0,48 0,2 0,11 21 0,62 0,39 0,15 0,05
28 0,62 0,48 0,2 0,1 28 0,57 0,32 0,12 0,01
35 0,61 0,47 0,2 0,07 35 0,62 0,25 0,16 0
42 0,58 0,4 0,12 0 42 0,59 0,19 0,09 0
49 0,53 0,23 0,09 0,02 49 0,61 0,2 0,09 0,02
56 0,58 0,39 0,1 0,01 56 0,59 0,21 0,05 0
Date: 26.07.2005

Date: 28.07.2005
Day of exp. 1

Day of exp. 4
Ca concentration Ca concentration
0 4,51 3,59 4,05 3,85 0 1,61 1,61 1,61 1,61
7 4,08 4,25 3,61 4,48 7 1,34 1,25
14 3,87 4,49 3,59 5,08 14 1,74 1,17
21 4,03 4,6 3,68 5,08 21 2,17 1,14
28 4,04 4,4 3,86 5,03 28 2,28 1,15
35 4,22 4,58 4,17 5,4 35 2,35 1,17
42 4,33 4,83 3,99 5,71 42 2,21 1,09
49 4,57 4,86 3,72 6,4 49 2,43 1,2
56 4,5 4,37 4,49 7,24 56 2,39 1,44
Site: copper roof runoff at UDK

Site: zinc roof runoff at TU
Date: 26.07.2005 Date: 26.07.2005
Iron concentration Iron concentration
0 0,01 0,01 0,01 0,01 0 0 0 0 0
7 0,05 2,8 1,27 0,15 7 0 0,67 0,11 0,01
14 0,03 2,64 1,15 0,09 14 0,05 2,1 0,13 0,19
21 0,01 2,56 1,05 0 21 0,04 1,88 0,72 2,31
28 0 2,02 0,81 0 28 0,07 2,14 1,84 0,05
35 0 4,27 0,73 0 35 0,02 1,91 0,12 0,05
42 0 3,06 0,59 0 42 0,01 1,4 0,1 0
49 0,07 5,12 0,43 0 49 0 1,4 0,5 0
56 0,02 3,17 0,47 0,01 56 0 1,98 0,1 0

Appendix B 160

Table B26 Long term monitoring of field study of composited columns (Fig. 4.28-4.29)*

System: Field composited column test System: Field composited column test
Metrix: UDK Metrix: TU roof
Cu
2+
: 5,4 mg L
-1

Cu
2+
: 0,0 mg L
-1

Zn
2+
: 0,0 mg L
-1

Zn
2+
: 2,3 mg L
-1

columns: Fe
0
with carbonate bearing materials columns: Fe
0
with carbonate bearing materials

column Cu [mg L
-1
] average error column Zn [mg L
-1
] average error
PM 3,75 5,355 3,21 PM 2,35 2,295 0,11

6,96

2,24

Fe
0
/PM 0,6 0,5 0,20 Fe
0
/PM 0,46 0,495 0,07
0,4 0,53

Fe
0
/PM/DML 0,01 0,035 0,05 Fe
0
/PM/DML 0,6 0,47 0,26
0,06 0,34
Fe
0
/PM/Mag 0,02 0,025 0,01 Fe
0
/PM/Mag 0,08 0,07 0,02

0,03

0,06
* The long term monitoring of metal concentration. 100 mL each were collected for 12 rain
events to total samples and they were measured.
0,00
1,00
2,00
3,00
4,00
5,00
6,00
7,00
8,00
9,00
PM Fe0/PM Fe0/PM/DMLFe0/PM/Mag
C
u

C
o
n
c
.

[
m
g
/
L
]
0,00
0,50
1,00
1,50
2,00
2,50
3,00
PM Fe0/PM Fe0/PM/DMLFe0/PM/Mag
Z
n

C
o
n
c
.

[
m
g
/
L
]

Fig. B14 Results of longterm performance of carbonatic/Fe
0
composited columns

Appendix B 161

Table B27 Breakthrough of copper in sequential column tests (Fig. 4.30)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C
DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
with pumice and carbonate materials, 5 min EBCT (see describtion in chapter 3)
BVT Fe
0
Cu Af CaCO
3
Cu Af DMSCu Af DMLCu Af MagCu Af DMLCu
0 0,01 0,01 0,03 0,01 0,00 0,00
186 0,11 0,06 0,14 0,06 0,00 0,00
342 0,12 0,15 0,11 0,03 0,00 0,01
553 0,13 0,22 0,10 0,02 0,00 0,03
818 0,13 0,21 0,08 0,06 0,00 0,03
935 0,13 0,16 0,08 0,08 0,00 0,05
1104 0,13 0,15 0,08 0,07 0,00 0,06
1235 0,15 0,16 0,10 0,02 0,00 0,06
1384 0,15 0,16 0,12 0,07 0,00 0,14
1539 0,15 0,20 0,13 0,07 0,00 0,18
1714 0,14 0,18 0,18 0,01 0,10
1833 0,09 0,24 0,32 0,01 0,14
2057 0,17 0,30 0,25 0,04 0,15
2383 0,18 0,42 0,32 0,01 0,00
2615 0,21 0,49 0,04 0,01 0,00
2701 0,22 0,01 0,01 0,02
2860 0,22 0,01 0,01 0,04
2994 0,25 0,01 0,02 0,02
3169 0,23 0,02 0,00 0,00
3287 0,33 0,02 0,09 0,00
3450 0,32 0,02
3584 0,37 0,00
3742 0,32 0,03
4167 0,02
BVT Af ZVI Cu Af ZVI Cu Af ZVI Cu
Af ZVI
Cu Af ZVI Cu
0 0,17 1,08 0,03 0,00 0,03
186 0,62 0,94 0,34 0,01 0,18
342 0,74 0,88 1,32 0,17 0,03
553 0,75 0,91 0,00 0,01 0,32
818 0,79 0,89 0,00 0,04 0,59
935 0,83 0,92 0,00 0,02 0,68
1104 0,86 0,93 0,00 0,00 0,78
1235 0,81 0,99 1,12 0,00 0,66
1384 0,82 1,01 1,42 0,00 0,59
1539 0,76 0,91 1,71 0,01 0,65
1714 0,90 1,39 0,02 0,47
1833 0,91 0,00 0,03 0,49
2057 1,24 1,08 0,76 0,30 0,81
2383 0,98 1,08 1,12 0,15 0,67
2615 1,04 1,05 2,66 0,72 0,75
2701 1,29 1,04 1,76 0,05 0,69
2860 1,00 1,05 1,87 0,24 0,66
2994 0,79 1,13 1,47 1,21 0,65
3169 0,82 1,09 5,95 0,86 0,79
3287 0,81 1,00 0,93 0,64 0,94
3450 1,03 0,00
3584 1,08 0,00
3742 0,97 0,00
4167 0,00

Appendix B 162

Table B28 Breakthrough of zinc in sequential column tests (Fig. 4.30)
Metrix: Cu
2+
/Zn
2+
in roof runoff

Cu
2+
: 4-5 mg L
-1

Zn
2+
: 5-7 mg L
-1

-1

Temperature: 20 C

DO: 7-8 mg L
-1

Fe
0
bed: 15 g Fe
0
with pumice and carbonate materials, 5 min EBCT
BVT ZVI Zn Af CaCO3Zn Af DMSZn Af DMLZn Af MagZn Af DMLZn
0 0,32 0,00 0,20 0,00 0,00 0,49
186 0,58 0,51 0,64 0,08 0,00 0,71
342 0,62 0,66 0,65 0,04 0,00 0,71
553 0,63 0,73 0,63 0,42 0,00 0,76
818 0,62 0,77 0,70 0,65 0,00 0,89
935 0,67 0,87 0,74 0,88 0,00 0,98
1104 0,64 0,86 0,79 0,96 0,00 1,03
1235 0,69 0,78 0,82 0,54 0,00 0,86
1384 0,68 0,74 0,82 0,63 0,00 0,88
1539 0,64 0,68 0,82 0,86 0,00 0,88
1714 0,58 0,59 0,76 0,87 0,00 0,85
1833 0,57 0,67 0,78 0,71 0,01 0,88
2057 0,71 0,71 0,53 0,49 0,04 0,93
2383 0,61 0,71 0,83 0,67 0,01 0,90
2615 0,62 0,82 0,40 0,84 0,02 0,91
2701 0,63 0,75 0,43 0,92 0,02 0,91
2860 0,63 0,56 0,45 0,99 0,02 0,98
2994 0,62 0,49 1,11 0,06 1,16
3169 0,65 0,53 0,85 0,14 1,37
3287 0,68 0,10 0,55 1,00 0,11 1,36
3450 0,71 0,82 0,56 0,00 0,02 1,45
3584 0,71 0,74 0,00 0,01 1,55
3742 0,76 1,01 0,00
4167 0,78 0,00
BVT Af ZVI Zn Af ZVI Zn Af ZVI Zn Af ZVI Zn Af ZVI Zn
0,00 0,13 0,98 0,03 0,00 0,00
186 0,86 0,98 0,41 0,01 0,00
342 0,91 1,00 1,22 0,05 0,08
553 0,93 0,96 1,82 0,04 0,09
818 0,94 0,99 1,32 0,05 0,14
935 1,01 1,03 1,82 0,03 0,13
1104 1,05 1,03 1,23 0,02 0,14
1235 0,98 1,05 1,14 0,01 0,11
1384 0,90 1,08 1,04 0,01 0,24
1539 0,91 1,00 1,20 0,02 0,27
1714 0,90 0,96 1,18 0,02 0,26
1833 1,07 0,00 0,31
2057 1,05 1,05 0,85 0,31 0,47
2383 1,03 1,08 1,03 0,16 0,00
2615 1,13 1,05 1,35 0,80 0,01
2701 1,21 1,05 1,36 0,07 0,05
2860 1,18 1,04 1,39 0,30 0,09
2994 0,91 1,05 1,52 1,17 0,05
3169 0,92 1,07 1,82 0,95 0,01
3287 0,92 1,00 0,90 0,80 0,01
3450 0,89 0,95 0,85 0,40 0,01
3584 0,88 1,04 0,79 0,59 0,01
3742 0,87 0,86 0,40 0,01
4167 0,88 0,84 0,42

Appendix B 163

Table B29 pH of the sequential column tests (Fig. 4.30)
pH of the influent
BVT CaCO3/ZVI DML/ZVI DMS/ZVI Mag/ZVI ZVI/DML ZVI
0 6,16 6,31 6,1 6,08 6,03 5,96
37 5,77 5,77 5,80 5,75 5,81 5,70
419 5,23 5,21 5,26 5,25 5,32 5,22
713 5,82 5,76 5,82 5,73 5,81 5,76
978 5,57 5,63 5,66 5,52 5,73 5,64
1274 5,71 5,67 5,67 5,40 5,70 5,56
1772 5,78 5,62 5,74 5,71 5,72 5,62
2420 5,63 5,62 5,60 5,58 5,65 5,50
2628 5,82 5,79 5,79 5,68 5,84 5,74
2909 5,53 5,48 5,52 5,47 5,60 5,48
3215 5,46 5,22 5,42 5,35 5,43 5,43
3512 5,88 5,86 5,88 5,78 5,76 5,86
3817 5,85 5,85 5,73 5,65 5,67 5,77
pH of the effluents of carbonate columns
0 10,35 10,67 6,35 11,12 6,36

37 6,38 9,53 5,92 10,58 5,93
419 6,17 6,42 5,41 10,42 5,63

713 5,94 6,21 5,93 10,34 5,91
978 6,00 6,33 5,90 10,52 5,93

1274 5,84 6,14 5,65 10,58 5,88
1772 5,84 5,96 5,71 10,62 5,93

2420 5,82 5,98 5,78 9,96 5,83
2628 5,98 5,92 5,68 9,82 6,16

2909 5,82 5,85 5,53 10,01 5,76
3215 5,83 5,83 5,64 9,68 5,66

3512 6,38 6,11 5,99 9,28 5,98
3817 6,26 6,24 5,90 8,13 5,94
pH of the effluents of Fe
0
columns (except that of ZVI- DML/PM column)
0 10,77 10,85 6,9 11,33 11,05 6,3
37 6,23 10,09 5,96 10,67 10,74 5,82
419 6,09 6,36 5,81 10,54 10,26 5,74
713 6,22 6,38 6,19 10,29 10,23 6,04
978 6,05 6,27 5,97 10,03 9,4 5,57
1274 6,27 6,49 5,91 9,2 6,04
1772 5,95 6,22 5,98 10,57 9,47 5,56
2420 5,94 6,01 5,95 10,11 6,87 5,93
2628 5,96 5,86 5,92 9,78 11,44 5,6
2909 5,8 5,8 5,98 10,05 10,48 5,76
3215

5,78 5,9 9,82 10,58 5,46
3512 6,23 9,39 10,52 6,24
3817 6,15 8,8 10,58 6,24

Appendix B 164

Table B30 Field study of sequential columns (Fig. 4.33)
Metrix: Cu
2+
/Zn
2+
from roof runoffs
Cu
2+
: - mg L
-1

Zn
2+
: - mg L
-1

Fe
0
bed: 100 g Fe
0
with pumice, 1(MF) -5(SF) min EBCT
Inlet Outlet Filter bed
date L water BVT Cu Zn Cu Zn Cu Zn
1 350 650 1,83 0,76 0,15
2 525 974 2,83 4,305 0,52 2,405 0,02 0,97
3 613 1137 4,83 3,165 0,765
6 963 1786 3,95 4,36 1,53 3,985 0,41 1,23
7 1050 1949 2,24 1,87 0,385
8 1137 2110 4,32 3,125 0,18
9 1225 2274 2,53 5,08 1,24 4,74 0,02 0,515
15 1400 2598 4,28 3,15 0,96
27 1400 2598 2,82 1,83 1,8 0,83 0,97 0,2
41 1663 3086 4,305 2,46 1,02
42 1750 3248 2,92 4,83 0,49 3,21 0,02 0,835
43 1838 3410 4,36 3,99 1,42
44 1925 3573 3,87 2,24 1,43 1,83 0,03 0,37
47 2013 3735 4,32 3,14 0,21
49 2100 3898 5,08 5,2 0,62
50 2188 4060 2,41 4,28 1,2 3,01 0,02 0,97
51 2275 4222 1,83 0,87 0,18
55 2363 4385 2,78 4,305 1,6 2,4 1,3 0,99

Table B31 pH of field study of sequential column (Fig. 4.33)
BVT pH
inlet
pH
outlet
pH
after filtration

650 4,79 5,87 9,25
974 4,96 6,05 9,78
1137 4,90 5,95 7,86
1299 6,24 6,36 9,32
1786 5,68 5,76 7,21
1949 6,16 6,43 9,41
2111 5,41 6,22 9,56
2274 5,10 5,73 9,08
2598 5,33 5,93 8,59
3086 5,63 5,81 9,15
3248 6,13 5,59 8,82
3410 5,65 5,61 6,04
3573 6,14 6,07 6,87
3898 5,93 6,00 9,02
4060 6,13 6,17 8,90
4222 5,64 5,72 8,84
4385 5,92 5,97 9,08

Appendix B 165

0
-DML/PM system (Slow Flow rate) (in supplement to Fig. 4.33)
Inlet concentration
Date Length Column1 Column 2 Average SD Error C/C
0

1 0 1,76 1,9 1,83 0,07 0,14 1,00
2 0 4,31 4,3 4,305 0,005 0,01 1,00
3 0 4,73 4,93 4,83 0,1 0,2 1,00
6 0 4,26 4,45 4,355 0,095 0,19 1,00
7 0 2,18 2,3 2,24 0,06 0,12 1,00
8 0 4,3 4,34 4,32 0,02 0,04 1,00
9 0 5,06 5,1 5,08 0,02 0,04 1,00
15 0 4,36 4,2 4,28 0,08 0,16 1,00

Outlet concentration

Length Column1 Column 2 Average SD Error C/C
0

56 0,7 0,82 0,76 0,06 0,12 0,415

56 2,73 2,08 2,405 0,325 0,65 0,559
56 2,86 3,47 3,165 0,305 0,61 0,655
56 3,69 4,28 3,985 0,295 0,59 0,915

56 1,85 1,89 1,87 0,02 0,04 0,835

56 3,33 2,92 3,125 0,205 0,41 0,723
56 4,61 4,87 4,74 0,13 0,26 0,933
56 3,37 2,93 3,15 0,22 0,44 0,736

Filtrate concentration
Filter column Column1 Column 2 Average SD Error C/C
0

out.fil 0,16 0,14 0,15 0,01 0,02 0,08
out.fil 0,49 1,45 0,97 0,48 0,96 0,23

out.fil 0,91 0,62 0,765 0,145 0,29 0,16

out.fil 1,24 1,22 1,23 0,01 0,02 0,28
out.fil 0,36 0,41 0,385 0,025 0,05 0,17
out.fil 0,09 0,27 0,18 0,09 0,18 0,04

out.fil 0,71 0,32 0,515 0,195 0,39 0,10

out.fil 0,9 1,02 0,96 0,06 0,12 0,22

0
-DML/PM system (Max flow rate) (in supplement to Fig. 4.33)
Inlet concentration
Date Length Column1 Column 2 Average SD Error C/C
0

1 0 1,82 1,86 1,84 0,02 0,04 1,00
2 0 4,39 4,5 4,445 0,055 0,11 1,00
9 0 4,95 5,04 4,995 0,045 0,09 1,00
15 0 4,55 4,47 4,51 0,04 0,08 1,00

Outlet concentration

Length Column1 Column 2 Average SD Error C/C
0

56 1,44 1,35 1,395 0,045 0,09 0,758152
56 3,66 2,96 3,31 0,35 0,7 0,744657
56 5,09 5,14 5,115 0,025 0,05 1,024024
56 4,22 4,24 4,23 0,01 0,02 0,937916

Filtrate concentration

Filter column Column1 Column 2 Average SD Error C/C
0

out.fil 0,44 0,44 0,44 0 0 0,24
out.fil 2,58 2,25 2,415 0,165 0,33 0,54
out.fil 1,31 0,94 1,125 0,185 0,37 0,23
out.fil 1,51 2,62 2,065 0,555 1,11 0,46

Appendix B 166

Table B34 Parameter used for cost estimation
Corrosion rate 3.6 mg Me m
-2
y
-1
Estimated cost*
Runoff 850.6 mm (i) Structure cost per unit volume 500 Euro m
-3

GFH loading 30 mg g
-1
(ii) Lining cost per unit volume 70 Euro m
-3

ZVI loading 42 mg g
-1
(iii) Excavation cost per unit volume

10 Euro m
-3

Period of changing media 20 years (iv) media cost per unit volume GFH/Lime 3.62 Euro kg
-1

year of operation 20 years (vi) ZVICB 0.045 Euro kg
-1

* Estimation of construction and media cost take into account the decrease of cost for larger project based on the data in Table B35

Table B35 Data for calculation of cost factor (cost in Euro)
Volume (m
3
) (i) (ii) (iii) (iv) (vi) Volume (m
3
) (i) (ii) (iii) (iv) (vi)
25 500 70 19.2 3.62 0.045 5000 150 45 5.8 3.2 0.04
2500 200 60 7.7 3.5 0.043 10000 130 40 5.5 3 0.04

Table B36 Estimated investment cost for GFH/Lime and ZVICB treatment barrier (cost is simplied for comparison)
GFH/Lime
Roof surface Media Vol. of Reactor Total construction cost [Euro] Total construction cost [Euro m
-2
roof surface
y
-1
]*
[m
2
] [kg] [m
3
] Concrete type Lining type Soil type Concrete type Lining type Soil type
100 240 0.36 1327 887 849 0.66 0.44 0.42
300 720 1.10 3590 2570 2466 0.60 0.43 0.41
700 1680 2.58 5721 4215 4049 0.56 0.42 0.40
1100 2640 4.06 7786 5840 5613 0.54 0.41 0.40
1500 3600 5.5 9807 7450 7165 0.52 0.41 0.39
1900 4560 7.01 11795 9050 8707 0.51 0.40 0.39
2300 5520 8.49 13758 10641 10241 0.51 0.40 0.39
ZVICB
100 171 1.17 1230 163 53 0.62 0.08 0.03
300 514 3.52 2897 429 128 0.48 0.07 0.02
700 1200 8.23 4313 675 193 0.40 0.07 0.02
1100 1886 12.94 5607 911 254 0.36 0.06 0.02
1500 2571 17.64 6821 1139 312 0.34 0.06 0.02
1900 3257 22.35 7976 1362 368 0.32 0.06 0.02
2300 3943 27.05 9086 1581 421 0.31 0.06 0.01
*Concrete type = structure + excavation + media; Lining type = lining + excavation + media; Soil type = excavation + media and the cost take into account the decreasing cost at larger volume

Appendix C

Appendix C Fractionation method in German

Fraktionierung von DOC ber XAD-8 (Adsorberharz), AG-MP 50
(Kationenaustau-scherharz), und Duolite A-7 (Anionenaustauscherharz) (Mai,
2005)

Allgemeine Hinweise zur Apparatur

Grundstzlich drfen die Harze nie trockenlaufen, und fr eine optimale Fraktionierung
drfen keine Luftblasen im System enthalten sein.

Deshalb mssen beim Anschlu einer neuen Vorratsflasche an die Schlauchpumpe immer erst
die Schluche entlftet werden, ehe der Weg auf die Sulen freigegeben wird. Das heit, es
wird zuerst der obere Hahn so eingestellt, da nur der Weg zur Abfallflasche frei ist, dann
wird solange Flssigkeit gepumpt, bis keine Luftblasen mehr auf dem Weg bis zum Hahn
enthalten sind; anschlieend kann der Hahn so eingestellt werden, da die Flssigkeit in die
Sulen gelangt

Die Vorratsflaschen sollten immer gut gefllt sein, damit das Einsaugrohr immer in die
Flssigkeit hineinragt und somit keine Luftblasen angesaugt werden.

Beim Splen mit Methanol treten am XAD-8 durch die nderung der Oberflchenspannung
Luftblasen auf. Diese mssen anschlieend wieder entfernt werden. Hierzu mu man das
System ffnen, wobei darauf zu achten ist, da der Auslauf der jeweiligen Sule gesperrt ist,
sonst luft die Flssigkeit beim ffnen des Systems aus. Am Schliff zwischen Sule und
Hahn wird das System geffnet. Dann knnen die Luftblasen mit einen gedrehten Draht
entfernt werden. Beim Schlieen des Systems mu wiederum darauf geachtet werden, da
beim Zusammenstecken der Schliffverbindung keine Luftblasen zurckbleiben.

Generell ist immer darauf zu achten, da die Schlauchpumpe nur eingeschaltet wird, wenn ein
Weg fr die Flssigkeit geffnet ist, damit nirgends ein Schlauch abplatzt.

Die Fluraten der Schlauchpumpe von 1 bzw. 4 ml/min mssen ab und zu berprft und neu
eingestellt werden, da sich die Eigenschaften der Pumpenschluche mit zunehmender
Einsatzzeit ndern.

Durchfhrung der Fraktionierung

Vor und nach jeder Fraktionierung mu das gesamte System mit verschiedenen Flssigkeiten
zur Reinigung gesplt werden (s. 1.-5.). Hierbei ist zu beachten, da alle Lsungen nur mit
bidest. H
2
O anzusetzen sind.

Nach der Reinigung der Harze werden die Blindwerte fr die Fraktionen 1-3 gewonnen (s. 6.-
8.), deren DOC-Gehalt mglichst vor der Fraktionierung der Probe bestimmt werden sollte.
Die Differenz zwischen DOC der zugefhrten Lsung (= bidest. H
2
O) und DOC des
Blindwertes sollte mglichst klein sein. (In der Literatur wird von <0,4 ppm gesprochen; dies

Appendix C 168

konnte an dieser Apparatur jedoch bisher nicht eingehalten werden.) Sollten die Blindwerte 2
ppm bersteigen, mu nochmals die Reinigung duchgefhrt werden.

(Die Glasgefe, in denen die Lsungen aufgefangen und der DOC-Gehalt bestimmt wird,
sollten mit 0,1M HCl und bidest. H
2
O imUltraschallbad gereinigt und anschlieend mehrfach
gesplt werden.)

Sind die Blindwerte niedrig genug, kann mit der Fraktionierung der Probe begonnen werden
(s. 9.-13.). Die Probe mu membranfiltriert sein, ihr DOC-Gehalt ca. 15 ppm und der pH-
Wert 2,0 betragen. Die zur Elution der hydrophoben Suren verwendete 0,1 M NaOH (s. 13.)
sollte frisch angesetzt sein, damit der Anteil an absorbiertem CO
2
mglichst gering ist.

Ist die Fraktion 4 der Probe abgenommen, wird weiterhin 0,1 M NaOH durch das XAD-8-
Harz gepumpt, um den Blindwert der 4. Fraktion zu gewinnen (s. 14.). Alle hydrophoben
Suren wurden vorher mit 50 ml NaOH desorbiert und somit kann der zugehrige Blindwert
auch nach der Probenfraktionierung abgenommen werden; das erspart das Eluieren mit NaOH
und anschlieendes Splen mit bidest. H
2
O vor der Probenfraktionierung.

Danach folgt wieder die Reinigungsprozedur (s. 15./16. u. 1.-5.)

Alle gesammelten Lsungen werden angesuert und bis zur DOC-Bestimmung, die innerhalb
von 24 Stunden erfolgen sollte, dunkel und khl gelagert. Fr die Fraktionen 1-3 ist es
gnstig, den NPOC zu ermitteln, bei der NaOH-Fraktion kann auch die TC-IC-Differenz
gnstiger sein.

Berechnung der Anteile der einzelnen Fraktionen am Gesamt-DOC

Angabe in mg C (in 80 ml Probenlsung):

DOC
gesamt
[mg C] = DOC
Probe
[mg/l] * 80/1000
HIX [mg C] = (Frakt. 3 - Blw
Frakt. 3
) [mg/l] * 80/1000
HIS [mg C] = ((Frakt. 2 - Blw
Frakt. 2
) - (Frakt. 1 - Blw
Frakt. 1
)) [mg/l] * 80/1000
HIN [mg C] = (Frakt. 1 - Blw
Frakt. 1
) [mg/l] * 80/1000
HIB [mg C] = ((Frakt. 3 - Blw
Frakt. 3
) - (Frakt. 2 - Blw
Frakt. 2
)) [mg/l] * 80/1000
HOS [mg C] = (Frakt. 4 - Blw
Frakt. 4
) [mg/l] * 50/1000

Angabe in %

HIX [%] = HIX [mg C] * 100% / DOC
gesamt
[mg C]
HIS [%] = HIS [mg C] * 100% / DOC
gesamt
[mg C]
HIN [%] = HIN [mg C] * 100% / DOC
gesamt
[mg C]
HIB [%] = HIB [mg C] * 100% / DOC
gesamt
[mg C]

HOX [%] = 100% - HIX [%]
HOS [%] = HOS [mg C] * 100% / DOC
gesamt
[mg C]
HON [%] = HOX [%] - HOS [%]
(HOB = vernachlssigbar = <1% von DOC
gesamt
)

Appendix C 169

Abkrzungen

HIX = hydrophile Fraktion (= HIS + HIN + HIB)
HIS = hydrophile Suren
HIN = hydrophile Neutralstoffe
HIB = hydrophile Basen
HOX = hydrophobe Fraktion (= HOS + HON + HOB)
HOS = hydrophobe Suren
HON = hydrophobe Neutralstoffe
HOB = hydrophobe Basen

80/1000 und 50/1000 = Faktor zur Umrechnung von mg/l auf mg/80ml

Appendix D 170

Appendix D

Appendix D The pore and surface diffusion model (PSDM)
Mertz et al. (1999)

The PSDM is a dynamic fixed bed model that incorporates the following assumption
Constant flow rate
Plug-flow conditions exist in the bed
Linear driving force describes the local bulk phase mass flux at the exterior surface of
the adsorbent particle
Local adsorption equilibrium exists between the solute adsorbed onto the GAC
particle and the solute in the intraaggregate stagnant fluid
Intraparticle mass flux is described by surface and pore diffusion
Adsorption equilibrium of individual compounds can be represented by the Freundlich
isotherm equation and IAST describes the competition between the compounds
There are no interactions between adsorbing compounds during the diffusion process.

From the mass balance on the bulk phase and adsorbent phase, two partial differential
equations are obtained for each adsorbing component. A coupling equation between the bulk
phase and the adsorbent pahse is obtained from assuming local equilibrium at the exterior of
the adsorbent particle. In cases where pore and surface diffusion are present, local equilibrium
is assumed along the pore walls. These equations are non-dimensionalized and solved
numerical methods. The orthogonal collocation method is used to convert the partial
differential equation into a set of ordinary differential equations. The set of ordinary
differential equations is solved using the backward differentiation method formular, also
known as Gears stiff method. The partial differential equations describing the adsorber
dynamics in a fixed-bed adsorber is given below. Crittenden et al. (1986) and Friedman
(1984) have presented the derivation of the equations and the computer algorithms that were
used to solve the model equations.

The liquid phase mass balance for component i is given by:

C
i
t
(z, t) + I
C
i
z
(z, t) + S
k
],i
(1-s)
sR
|C
(z, t) -C
p,
(r = R, z, t)] = u (B-1)

The initial condition for equation D-1 is

C
(z, t) = u at u z , t = u

The boundary condition for equation D-1 is

C
(z, t) = C
0,
at z = u, t > u

The intraparticle phase mass balance for component i is given by:

1
r
2
s,
q
i
(r, z, t) +
p,i
s
p
p
c
C
p,
(r, z, t) =
t
q
(r, z, t) +
s
p
p
c
C
p,
(r, z, t) (B-2)

Appendix D 172

The initial condition for equation D-2 is

q
(r, z, t) +
s
p
p
c
C
p,
(r, z, t) = u at u r R, t = u

The first boundary condition for equation D-2 is:

r
q
(r, z, t) +
e
p
p
u
C
p,
(r, z, t) = u ot r = u, t u

The second boundary condition for equation D-2 is:

s,
p
u
q
r
(r = R, z, t) +
p,
e
p
C
p,
r
(r = R, z, t) = k
],
|C
(t) - C
p,
(r = R, z, t)]

The equation coupling the aqueous phase concentration of component i within the adsorbent
pores to the adsorbent phase concentration of component i is given by:

C
p,
(r, z, t) =
q
(r, z, t)
q
k
(r, z, t)
m
k
q
k(,z,t)
m
k
n
n
i

The above equation is derived assuming that the adsorption reaction rate is much faster than
the mass transfer rate (assumption of local equilibrium).

Nomenclature:

C
0,
initial concentration
C
(z, t) adsorbent concentration in bulk phase

C
p,
(r, z, t) adsorbate concentration in adsorbent pores
p,
pore diffusivity
s,
surface diffusivity
k
],
film transfer coefficient
bed length
m number of components
q
(r, z, t) adsorbent phase concentration

R average adsorbent particle radius
I loading rate

Greek letters

e
P
particle void fraction (porosity)
p
A
apparent adsorbent density

index i refers to component number i

Rangsivek Ropru

Uploaded by

Copyright:

Available Formats

Rangsivek Ropru

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Rangsivek Ropru

Uploaded by

Copyright:

Available Formats

Removal of dissolved metals from storm water

runoff by zero-valent iron

Total removal rate

(Fe) Total iron mass

that leads to an increase in pH (Eq. 2.1).

is the adsorbate concentration in the bulk and c

is the adsorbate concentration

% 92% 99% 99% 92%

(Ca) kg - 25 3.2 3.2

(z, t) adsorbent concentration in bulk phase

(r, z, t) adsorbent phase concentration

You might also like