On dioxin formation in iron ore sintering

Article abstract of DOI:10.1021/es026292g

Iron ore sintering is an important source of dioxins , polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/ Fs). This paper reports on attempts to identify materials, conditions, and mechanisms responsible for PCDD/F formation (i) by investigating salient properties of ores (viz., with respect to oxidation, condensation, and chlorination of model organics) and (ii) by mimicking the industrial process on a microscale with real-life materials. Principles of Design of Experiments (DOE) are employed. The reactivities of iron ores differ greatly. Limonite/goethite soft ore is a very active oxidation catalyst (e.g., for benzene and phenol), a property that may be useful in cleaning up crude sintering process offgases, whereas hematite/magnetite hard ore is not. The latter, however strongly promotes condensation of phenol to dibenzofuran. A newly built labmicroscale sintering facility could satisfactorily imitate the large-scale process, in part or as a whole. Results obtained with realistic feed mixtures point at dioxin formation in the sinter bed at levels significant enough to explain a major part of the outputs observed in the real-life process. With ～8 ppm (wt) of chloride added as NaCl, the PCDD/F output doubled, but with the same proportion of chlorine administered as C₂Cl₄, the dioxin output was over 2 orders of magnitude larger. The use of process reverts, etc. containing chlorinated organics should therefore be avoided. PCDD/F congener patterns are also reported and compared with those observed in practice.

Full text of DOI:10.1021/es026292g

Environ. Sci. Technol. 2003, 37, 3323-3331
offgas of one of the Dutch municipal solid waste incinerators
On Dioxin Formation in Iron Ore
Sintering
(MSWI) in 1977 (1), there has been a growing concern on
“dioxins” formed in this and in other thermal industrial
processes. In the course of time, dioxins in gaseous, solid,
or liquid effluents of other processes have been reported,
including the chemical industry (i.e., production of pesticides)
(2), paper manufacturing (3), and the metallurgy of iron and
non-iron metals (4). So far, worldwide attempts to close the
gap between PCDD/ F levels measured in the environment
and the recognized sources of these noxious compounds
have been unsuccessful (4). Nevertheless it is now sure that
metallurgy in general is an important dioxin source. Where
exactly in these processes and how dioxins arise are unknown.
Therefore, under aegis of the European Union, in 1998
university and industrial groups from Belgium, Germany,
Sweden, and The Netherlands joined to investigate pathways
and mechanism(s) of formation with the ultimate aim to
minimize outputs of PCDD/ Fs in thermal industrial processes
(MINIDIP).
M A R I U S Z K . C I E P L I K , † , ‡
J O S E P A S T O R C A R B O N E L L , §
C H R I S T I N A M U N˜ O Z , § S A R A H B A K E R , |
S O P H I E K R U¨ G E R , P E R L I L J E L I N D , #
S T E L L A N M A R K L U N D , # A N D
R O B E R T L O U W * , †
Center for Chemistry and the Environment, Leiden Institute
of Chemistry, Gorlaeus Laboratories, Leiden University,
P.O. Box 9502, 2300 RA Leiden, The Netherlands,
Universitat de Vale`ncia, 46100 Vale`ncia, Spain,
Grinnell College, Grinnell, Iowa 50112,
Universite´ de Strasbourg, 67000 Strasbourg, France, and
Chemistry Department, Environmental Chemistry,
Umeå University, SE- 901 87 Umeå, Sweden
This paper deals with iron ore sintering, a key large-
scale metallurgical process. Some preliminary results of
our work have been presented earlier (5). Below a brief
description is given of the process, the materials involved,
and the process parameters considered relevant for dioxin
formation.
Sintering Process. Sintering is a preliminary process in
iron smelting with the aim to prepare a raw material for blast
furnaces. The basic principle of sintering may be compared
to smoking of a cigarette or a pipe: sucking air through a
glowing material moves the burning front down. The design
of a traditional Dwight-Lloyd type sintering plant is very
simple and basically unchanged from its early beginning (see
Figure 1). Iron ores mixed with cokes and some additives (1)
are ignited by (natural or producers) gas-fueled burners (2)
situated at the beginning of the steel belt conveyer (3); air
is sucked through the sinter layer by means of wind legs (4)
and the fan (5), which moves the burning front down the
layer; the cooled sinter is ground (by means of rough rolls)
and collected.
Environmental concernswith tighter emission limits
imposed, not only for dioxinssresulted in a redesigning:
instead of one APC (6) unit at the end of the process, the
system was fitted with high-pressure scrubbers, air/ water
recycling loops, and AC (activated carbon) injection absorb-
ers. This improved design, called the Energy Optimized
Sintering (EOS), is used for example by Corus Nederland
(Figure 2). It leads to a considerable drop in (trace) organic
emissions at an overall higher energy efficiency.
Locally, temperatures in the sintering bed can become as
high as 1200-1250 °C, but due to its inhomogenity, the actual
maximum temperature may locally be less, maybe not higher
than 1000 °C. The temperature in the wind legs ranges from
ambient at the beginning through some tens of degrees higher
values while water is evaporated from the sinterfeed to 200-
600 °C when dry air breaks through and the burning front
reaches the bottom near the end of the sinter belt.
Materials. The following (raw) materials are usedswith
proportions typical for a plant operated in Western Europe:
Iron Ores. At least five different ores are employed.
Geologically they belong to the mineral groups limonite/
goethite (soft oresslow iron content, comprising ca. 15-20
wt % of the mix) or hematite/ magnetite (hard oresshigh
iron content, in total ca. 65 wt % of the sinterfeed).
Additives. This rather ill-defined group of materials
includes plant reverts from the whole of the iron metallurgical
process as well as mineral flux agents and binders.
Iron ore sintering is an important source of “dioxins”,
polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/
Fs). This paper reports on attempts to identify materials,
conditions, and mechanisms responsible for PCDD/F formation
(i) by investigating salient properties of ores (viz., with
respect to oxidation, condensation, and chlorination of
model organics) and (ii) by mimicking the industrial process
on a microscale with real-life materials. Principles of
Design of Experiments (DOE) are employed. The reactivities
of iron ores differ greatly. Limonite/goethite “soft” ore is
a very active oxidation catalyst (e.g., for benzene and phenol),
a property that may be useful in cleaning up crude
sintering process offgases, whereas hematite/magnetite
“hard” ore is not. The latter, however strongly promotes
condensation of phenol to dibenzofuran. A newly built lab-
microscale sintering facility could satisfactorily imitate
the large-scale process, in part or as a whole. Results
obtained with realistic feed mixtures point at dioxin formation
in the sinter bed at levels significant enough to explain
a major part of the outputs observed in the real-life process.
With ∼8 ppm (wt) of chloride added as NaCl, the PCDD/F
output doubled, but with the same proportion of chlorine
administered as C Cl4, the dioxin output was over 2 orders
2
of magnitude larger. The use of process reverts, etc.
containing chlorinated organics should therefore be avoided.
PCDD/F congener patterns are also reported and compared
with those observed in practice.
Introduction
Background. Ever since the discovery of polychlorinated
dibenzo-p-dioxins and dibenzofurans (PCDD/ Fs) in the
* Corresponding author phone: +31-71-5274289; fax: +31-71-
5274451; e-mail: r.louw@chem.leidenuniv.nl.
† Leiden University.
‡ Present address: Energy Research Centre of the Netherlands
(ECN), Westerduinweg 3, P.O. Box 1, 1755 ZG, Petten, The Nether-
lands.
§
Visiting student from the Universitat de Vale`ncia.
| Visiting student from Grinnell College.
Visiting student from Universite´ de Strasbourg.
# Umeå University.
9
10.1021/es026292g CCC: $25.00
Published on Web 06/27/2003
2003 Am erican Chem ical Society
VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 3 2 3

the 3rd field) (6). Little if any data are available on the total
chlorine balance of the sintering process, particularly when
considering organic chlorine.
Another important ingredient is water. The sinterfeed is
moistened by 6.3-6.5 wt % to prevent its falling apart while
it is put onto the sinter belt as well as to allow smooth burning.
It is important to consider the salinity and metal contents
of the (surface) water used. For example, the average salinity
of the Rhine River water is between 70 and 100 mg/ L (data:
RijkswaterstaatsDutch Water Management Council, Lobith
monitoring point). At some 5 million ton per year throughput
(70 Mt/ yr of sinter, 6.5 wt % of water), this gives an impressive
load of 350-500 t of NaCl per year.
Yet another source of chlorine may be the air used in the
process. Sintering plants can be located close to, or directly
on, the seashore. Sea saline aerosols are known to contain
relatively high levels of chloride (7), ca. 4-5 µg/ m3 . However,
at an estimated “consumption” of some 700 Mm3/ yr, this
gives a negligible amount of chlorine of ∼150 kg as compared
to the water-related input of hundreds of tons. The same
holds for the input of atmospheric HCl, in view of an average
HCl concentration in the industrial air (8) of around 0.2-3
µg/ m3.
FIGURE 1. Schematic representation of a Dwight-Lloyd iron ore
sinter plant: (1) feeder, (2) burner for ignition, (3) steel belt conveyer
with (4) wind legs, (5) fan, and (6) APC unit.
Other Elem ents. Apart from iron, the ores contain also
other metals; the most important are alkaline metals (Ca,
Mg, Na, and K up to 2 wt % as oxides), aluminum (0.1-3 wt
% as Al2O3), titanium (0.1-0.5 wt % as TiO2), and manganese
(0.1-0.2 wt %). Phosphorus and sulfur contents are ca. 0.1
wt % each. Interestingly, copper (a notorious catalyst in MSW
fly ashes) is not reported for any of the ores, whereas ESP
sinter plant ashes may contain up to tens of a percent of this
very subject-relevant metal (6).
FIGURE 2. Schematic representation of the Energy Optimized Sinter
(EOS) plant: (1) hopper, (2) burners, (3) sinter belt, (4) wind legs,
(5) fan, (6) stack, (7) rough rolls, (8) ESP, (9) high-pressure scrubber/
AC injection, (10) air/dust recycling hood, and (12) fresh/recycled
air supply.
Dioxins and the Sintering Process. Formation of PCDD/
Fs requires the presence of carbon-containing materialss
either specific organic compounds or a carbonaceous
matrixsand sources of chlorine and oxygen plus increased
temperatures (200-800 °C, at higher temperatres PCDD/ Fs
will rapidly decompose), whereas the presence of catalytic
metals can be essential at modest temperatures.
In the sinter bed, basically three layers can be recog-
nized: raw material (wet and cold), the burning front, and
the cool-down zone, consisting of sintered material. In the
context of dioxin formation, the thin border between the
burning front and the raw materials is actually the most
intriguing zone. In this region, the products of incomplete
combustion (PICs) surviving the heat of the burning front
may condense, while the temperature is high enough to
enable reactions with species in the raw materials acting as
catalysts. Also, the native carbon-containing materials may
react via the so-called de novo route (6).
Plant Reverts. the most interesting category is ball mill
reverts, consisting of finely ground iron ores soaked with
mineral oil. The oil content is relatively high (up to 10 wt %),
whichsgiven the proportion used in the sinterfeed, 10-15
wt %sresults in a high additional organic input. Upon
analysis, we have found the oil to consist of mostly linear
and branched long-chain aliphatics (octane through dode-
cane) with trace amounts of substituted cycloalkanes (mostly
cyclohexanes) as well as aromatics (naphthalenes and higher).
An important fact is that this oil-containing material is only
allowed to be recycled in this way in The Netherlands and
Belgium (for example, in Great Britain it is subject to
landfilling).
Other reverts include sludges and ashes from dedusting
units. These materials, especially the ESP ashes, have
substantial organic carbon contents as well as considerable
chloride levels.
Some 5 wt % of the sinterfeed is actually sinter itself. As
the bottom, it serves as a thermal insulator preventing the
conveyer belt from excessive heat. Fine sinter, not useful for
a blast furnace, is also returned, admixed with the “raw”
sinterfeed.
A common flux agent is olivine (2-3 wt %). It allows
melting the iron ores at temperatures of around 1000 °C
rather than over 1200 °C, as we have experimentally verified.
Olivine is a mixed iron/ magnesium silicate. Limestone (at
ca. 2 wt %) is another common additive.
Coke. In fact, the so-called “breeze”, the fines fraction
from the coking process, serves as the fuel in the sintering
process, employed at 3-3.5 mass %. The ash content of this
coke is ca. 10-13 wt % and consists mostly of silicon and
aluminum oxides. On this basis, the coke-related carbon input
is ca. 2-3% of the total mass. Once again, this shows the
importance of the organic carbon load from the plant reverts.
Chlorine Input. Nearly all the above-mentioned materials
are very poor in chlorine, typically ca. 0.05 wt %. Exceptionally
high levels of chlorine were only reported for ESP ashes (from
Obviously, during sintering conditions are met wherein
dioxins can be formed andsfor some partssurvive. Also
downstream, in the wind legs as well as in the APC devices,
dioxins may arise in a way not unlike that documented for
(waste) incinerators (6, 9). Worth mentioning here is the time
scale of the process. Very relevant for kinetic and mechanistic
discussion is that it takes some 40 min for the materials to
pass on, while in minutes the materials are subjected to a
favorable temperature window, as compared with perhaps
only a few seconds in the wind legs and minutes again in the
APC zone.
In this paper, the possible formation of dioxins during
sintering is considered, while other publications (10, 11)
discuss the formation of dioxins downstream of the process.
Next to orienting experiments with fixed beds of iron oress
to evaluate their reactivity with respect to model organic
compounds such as benzenesa new setup has been designed
and employed to mimic the sintering process as a whole on
a (micro) labscale. After its optimization with the aid of
Experimental Design, actual tests have been performeds
9
3 3 2 4 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

The originally all-quartz reactor was soon replaced by a
passivated stainless steel one as the melted sinter grains were
very difficult to remove, leading to destruction of the reactor.
However, stainless steel has another drawback: with too high
fuel contents the reactor wall can melt through, which
damages the heater and of course ends the experiment.
Procedure. After filling with a fresh portion of sinterfeed,
the reactor was placed in the oven. The PUF plug unit was
then connected, after which the system was checked for leaks.
Next, a background gas sample was analyzed to ensure proper
performance of the GC facility, eventually followed by the
ignition of the system. As mentioned above, the hydrogen
burner was only used in the first series of experiments. In the
course of the investigation it turned out that the water
produced by the burner not only clogged the cryostatted
PUF plug but also quenched the sintering. Therefore in later
experiments the temperature of the top oven was initially set
to 800 °C, allowing for self-ignition of the coke. Upon ignition,
the temperature of the top oven was lowered to 600 °C (or
a value designated by the ED spreadsheet), while monitoring
the advancing burning front by means of thermocouples
placed inside the top, middle, and next to the bottom layer
of the sinterfeed. The temperature of the bottom oven was
set initially at 150°C (or again, a value dictated by the ED)
to avoid condensation of water and organic products. When
the burning front passed the middle part of the reactor, the
temperature of the bottom unit was raised to 600 °C, while
the top temperature was set to ambient, allowing the sintered
material to cool in a natural way.
FIGURE 3. Schematic representation of the microscale sintering
facility: (1) ignition burner; (2 and 3) top and bottom oven,
respectively; (4) sinterfeed; (5) tubular quartz/stainless steel reactor;
(6) stainless steel grid; (7) filter/demister assembly; (8) Supelco
ORBO1000 cartridge; (9) polyurethane foam plug; and (10) thermo-
couples.
with offline analysesson dioxin formation, output, as a
function of important operational parameters. In an attempt
to detect (ultra)trace species leaving the sinter bed online,
the setup was coupled to a JET-REMPI (Resonance-Enhanced
Multi Photon Ionization) spectrometer at DLR, Stuttgart,
Germany built by Dr H.-H. Grotheer.
The experiment was finished when the burning front
passed the bottom thermocouple. The reactor was then
quickly removed and cooled in a stream of air. The PUF plug
cartridge was also removed and allowed to thaw and
subsequently extracted as described below.
Analysis. Details on online gas chromatographic analysis
of gas-bound products are described elsewhere (12).
Thawed PUF plugs were subjected to 18-h Soxhlet
extraction with 110 mL of a 10/ 1 (v/ v) mixture of n-hexane/
acetone. The extraction turnover was ∼10 cycles/ h. The
solution was concentrated on a rotary evaporator to a volume
of a few milliliters. Next, it was condensed to 0.50 mL by
means of TurboVap gas stream evaporator. The extracts were
analyzed for organic products in Leiden by means of an offline
gas chromatography (HP 5890 II series, split/ splittless inlet
capillary CP-SIL5CB/ MS column, HP6890 autoinjector,
carrier gas: He) combined with a mass spectrometer (HP
MSD 5972). Dioxin analyses were performed in Umeå by
means of HRGC/ HRMS in the following way: 10% of the
extract was taken to the cleanup process and spiked with
labeled internal standards (IS). The cleanup is based on
sequential use of open liquid chromatographic columns.
Three types were used for PCDD/ F analysis; first a multilayer
silica gel column; basic (SiO2-KOH, 20%), neutral activated
(SiO2, 130 °C), and acidic gel (SiO2-H2SO4, 40%) eluted with
n-hexane. This was followed by a super-grade alumina
column eluted with cyclohexane:dichloromethane (1:1), and
finally a carbopack C column (carbon:Celite) eluted with
toluene. Tetradecane was added as a keeper before each
concentration with a rotary evaporator. Finally, recovery
standard addition completes the cleanup, and the 30-µL
sample is ready for analysis by high-resolution gas chro-
matography coupled with high-resolution mass spectrometry
(HRGC/ HRMS). PCDD/ Fs were analyzed on a VG-250S (GC:
HP 5890 with an SP2330-column, 60 m, 0.32 mm i.d.).
Chem icals. The following chemicals were utilized: ben-
zene (Merck p.a., 99%, distilled), acetone (Baker, >99.5%),
naphthalene (Janssen Chimica, >99%), benzonitrile (Merck,
>99%), chlorobenzene (Baker Analysed, >99%), phenol
(Aldrich, >99%), phenanthrene (>99%), trichloroethane
(Baker Analysed, >95%), benzaldehyde (Baker Analysed,
Experimental Section
Setups. Setup A has been described earlier (12). It consisted
of a vertical quartz tubular reactor in an electrical oven,
equipped with supplies for gaseous and liquid reagents,
sampling facilities, and online GC. The reactor tube (length
∼30 cm, diameter 2 cm) has been modified to accommodate
a fixed bed of iron ore (typically ca. 200 g, bed height ∼10
cm) to investigate the properties of these materials.
The microscale sintering facility B is schematically
depicted in Figure 3. The system consisted of two Watlow
300W/ 60V ceramic heating elements in an insulating ceramic
casing, regulated by means of REX 100 temperature control-
lers. The tubular reactor, packed with the sinterfeed sup-
ported on a stainless steel grid resting on a Seger spring, was
directly connected to the air/ organic supply on the top and
to a Supelco ORBO1000 PUF plug sampler at the bottom, so
as to reduce the dead volume to a minimum. Gases were
supplied by means of electronic mass flow controllers (Brooks
5850TR). Optionally organics were injected into the gas stream
at the top of the reactor by means of a Hamilton gastight
syringe on a Microdialysis injection pump. In the top
assembly, a small homemade hydrogen burner was fitted,
used for ignition purposes only in preliminary experiments.
Stainless steel connectors, equipped with valves, in the filter/
demister assembly and the PUF plug cartridge enabled
sampling of the outflowing gases through heated sampling
lines by means of online gas chromatography as described
below. The PUF plug cartridge was kept at -25 °C by means
of a homemade cooling mantle and a cryostat.
The reactor system could be heated very rapidly while it
was also possible to cool it down (within ca. 5 min) by
quenching with countercurrent nitrogen. Moreover, the
reactor was mounted inside the oven in a way allowing
removing and opening it in a matter of minutes.
9
VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 3 2 5

>95%), dibenzofuran (Fluka, >95%), monobenzofuran
(synthesized, >95%), toluene (Merck p.a., >99%), 2-chloro-
phenol (Merck, >95%), 2,4,6-trichlorophenol (Aldrich, >95%),
tetrachloroethylene (Baker Analysed, >95%), hydrogen
(Air Products, 99.995%), methane (Air Products, 99.995%),
nitrogen (Air Products 99.995%), oxygen (Air Products
99.995%), and sodium chloride (Merck, >99%). Dioxin
analyses were performed using the following solvents:
cyclohexane, cyclopentane, dichloromethane, n-hexane,
tetradecane, toluene (all Fluka, HPLC/ GC-grade) and stan-
dard solutions for analysis: 13C-labeled and native with 16
2,3,7,8-substituted PCDD/ F congeners (CIL, >99%).
Iron ores, olivine, limestone, millscale, and plant reverts
were used as received from Corus Nederland. Genuine sinter
and cokes were ground prior to usescourtesy of Corus R&D
Department.
TABLE 1. DOE Spreadsheet for Evaluation/Optimization of the
Microsintering Setup
top bottom
oven oven
soft
N2 flow O flow temp temp
exp (mmol/h) (mmol/h) (°C)
organ- coke ore
2
a,b
(°C) burnera ics
(%) (%)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
4012
2643
4012
2643
4012
2643
2643
4012
2643
2643
4012
4012
2643
3328
3328
2643
4012
4012
3328
408
408
408
849
408
849
408
849
408
849
849
849
849
628
628
408
849
408
628
300
600
600
300
300
600
600
600
300
600
300
300
300
450
450
300
600
600
450
100
100
400
100
100
400
100
100
400
400
400
400
100
250
250
400
100
400
250
1
0
0
0
0
0
1
0
1
1
1
0
1
0
0
0
1
1
0
0
0
0
0
1
1
1
1
0
0
1
0
1
0
0
1
0
1
0
5
3
5
5
3
3
5
5
3
5
5
3
3
4
4
5
3
3
4
5
5
25
25
25
25
25
5
25
5
25
5
Results
5
Reactivity of Iron Ores. Using setup A (see Experimental
Section) benzene per se and solutions of phenol or dibenzo-
furan (1/ 9 mol in benzene) have been passed at a feed rate
of ca. 1.7 g/ h in 2.5-5 L/ h of synthetic air (20% O2 in nitrogen),
void residence times of the carrier gas about 4 s, over a fixed
bed (ca. 6 cm thick) of two different common iron ore
materials: a limonite (ca. 60% Fe, hereafter called the “soft
ore”) and a magnetite (ca. 80% Fe, the “hard ore”). The
temperature was varied from 200 to 700 °C. Salient results
are depicted in Figures 4 and 5. It is immediately clear that
the two types of ore differ greatly in their catalytic activity
and properties; the soft ore is much more activesas an
oxidation catalystswhile the hard ore promotes the con-
densation (of phenol) to dibenzofuran (DBF). Next to DBF,
traces of naphthalene and phenanthrene were detected in
the liquid catch of the runs with the hard ore.
15
15
5
25
5
15
a
b
Zero m eans absent, 1 m eans em ployed. Mixture of 14 com pounds
corresponding with a prim ary com bustion offgas (9), adm ixed with the
sinterfeed at a level of ca. 800 ppm (wt) (m ostly benzene), details in ref
14.
TABLE 2. Parameters in Runs Focusing on Dioxin Formation
exp
name/code
coke
(wt %)
hard ore
(wt %)
soft ore
(wt %)
chlorine
forma
In two more experiments an amount of Cl equivalent to
phenolsas aqueous HCl or as C2Cl4swas added to the feed
solution and reacted at 220 and 380 °C (∼50% conversion of
phenol) with the soft and the hard iron ores as the bed
material, respectively. This resulted in very limited outputs
of (mono)chlorophenols. No other chlorine-containing prod-
ucts could be identified.
When looking at the conversion of phenol over a soft ore
bed (Figure 4), one may see that complete conversion takes
place at temperatures of ca. 250 °C, while the T50% lies outside
the experimental range of temperatures, somewhere around
200 °C. Worth mentioning is the fact that next to unconverted
substrates no heavier products were detected in the liquid
samples. Also, the total COx yields are closely following the
conversion curves, although the output of carbon monoxide
is somewhat higher when using benzene per se.
Lab-Microscale Iron Sintering. Introductory Experiments.
The performance of the microscale sintering setup B has
been evaluated and optimized using the principles of the
DOE (13). Table 1 summarizes the binary experimental
parameters (variables) together with the experimental ranges.
Measured outputs (responses) included fuel (coke) burnout
efficiency, oxygen consumption, degree and quality of
sintering, and possible outputs of organics.
The results will not be presented here in detail as no
dioxin data have been obtained. By using the DOE technique,
the performance of the system could be optimized. Also, the
obtained results were easy to interpret by means of the
used DOE software (Umetrics Modde, tryout version). The
factors required for the sintering to occur and determining
the quality of the sinter have been identified, so providing
an optimized set of conditions used in the following
investigation.
Standard 1/I
Standard 2/II
no chlorine/III
organic Cl2/IV
hard ore/V
soft ore/VI
5% coke/VII
7% coke/VIII
3
3
3
3
3
3
5
7
15
15
15
15
30
5
20
20
20
20
10
30
20
20
NaCl in H2O
NaCl in H2O
none
C2Cl4
NaCl in H2O
NaCl in H2O
NaCl in H2O
NaCl in H2O
15
15
a
Chlorine was supplied in the “process water” or as tetrachloro-
ethylene adm ixed with the (200 g) sinterfeed at ca. 1.5 m g of
Cl/experim ent (7.5 ppm ).
the temperature of the top and bottom ovens was set at 600°C
while the gas flow was kept at 100 L/ h.
Runs I-IV focus on the role of chlorine (amount, form);
the respective total PCDD/ F outputs are shown in Figure 6.
Figure 7 depicts relative yields of dioxins in runs V-VIII with
different sinterfeed composition. In Figures 8-10, congener-
specific data are given.
Discussion
Properties of Iron Ores. Figures 4 and 5 substantiate that
the two ores used in the investigation have very different
properties regarding oxidation and condensation of organics.
The soft iron ore helps to smoothly oxidize benzene, and
even better phenol, at rates very much higher than in absence
of the ore. Remarkably, benzene admixed with (10%) phenol
reacts faster, as its conversion curve is shifted to 50-60 °C
lower temperatures. Such a synergistic effect, with two
compounds of different nature, is not uncommon and can
be beneficially used for example in the catalytic combustion
of chlorinated benzenes with added aliphatic hydrocarbons
(15). Anyway, it can be concluded that the soft iron ore is a
potent oxidation catalyst, comparable even with sophisticated
noble metal-based catalysts (15).
Dioxin Formation in Microsintering. Next, a series of eight
experiments at standard, optimized conditions has been
conducted with different sinterfeeds (Table 2). Throughout,
9
3 3 2 6 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

traces of (mono)chlorophenols have been detected, sug-
gesting that both ores have a small but distinct potential for
chlorination of organics.
Microscale Sintering. Even with a relatively large spread
between the two standard runs 1 and 2, dioxin outputs vary
more than 2 orders of magnitude when comparing experi-
ments with organic, inorganic, and without chlorine (Figure
6). With the same input levels in both “chlorine in” series,
it shows the high potential of organic-bound chlorine for
dioxin formation. Consequently, in practice, contamination
of the sinterfeed with even trace levels of organic chlorine
should be avoided. Interestingly, despite this large variation,
the toxicity factor of the resulting mixture stays relatively
constant; the ratio between the total masses of the formed
PCDD/ Fs and their TEQ is around 40.
Proper attention has been paid to the overall chlorine
balance. When feeding chlorine in the process water, some
65% was found back afterward in the sinter bed upon
extraction with slightly alkaline aqueous solution. This shows
that chloride is to a large extent immobilized by the materials
present in the sinterfeed, most likely by the alkaline minerals
(limestone) in addition to the iron oxides. The 65% mentioned
above should be considered a minimal value since a part of
the chlorine may have become occluded inside sinter particles
and, therefore, is inaccessible for the extraction. Only ca.
10% of the chlorine was found in the water condensate frozen
on the PUF plug. As other chlorine-containing organic
products could not be found, as yet some 25% in the chlorine
balance is unaccounted for.
FIGURE 4. Experimental results on runs with “soft” iron ore.
Figure 7 compares outputs of dioxins for various com-
positions of the sinterfeed, but with the same chlorine form
and amount. The overall DD/ DF ratio is remarkably constant,
with about twice as much DDs as DFs. The dioxin outputs
increase with 30-40% when the coke content is doubled.
This suggests an increased production of organic precursors,
not excluding formation of “complete” PCDD/ F structures.
Remarkably, the dioxins from runs 7 and 8 have ∼15 times
higher TEQ levels. This can be traced back to the congener
distributions (Figure 8). The overall lower chlorination degree
as observed in the runs with more fuel and in the case of the
hard ore has a pronounced effect on the overall toxicity of
the relatively constant output of the tetra- through octa-
DD/ DF congeners. This may be due to a limited chlorination
potential of the involved materials, which is consonant with
the results of our “reactivity” study. However, the question
remains how in such dilute chlorine streams and with such
“impotent” catalysts multiple chlorination occurs, leading
to even hepta- and octachloro derivatives.
In Figure 9, outputs of 2,3,7,8-substituted species are
depicted for the runs with the different chlorine levels and
forms. Surprisingly, the overall chlorination degree in the
run with only the “native” chlorine is much higher than in
those with HCl or organic chlorine in the feed. The latter
experiments gave much higher outputs, but with a roughly
comparable distribution. Figure 10 gives “Hagenmaier” charts
(with fractions of a specific congener in the total of 2,3,7,8-
chlorinated DD or DF family) together with that for the real-
life pattern from the work of Calcagni et al. (18). How
experimental data compare with those from a real sinter
plant is broader discussed in the next section.
After coupling of the microscale sintering facility with the
JET-REMPI analyzer of DLR in an attempt to observe online
species leaving the active sinter bed, an experiment analogous
to Standard 1, Table 2 led to the identification of a number
of trace products previously not seen in our low-resolution
analyses. These were phenol, monochlorophenols, dibenzo-
furan, (chloro)benzenes, and various polycyclic aromatic
hydrocarbons. Unfortunately, due to the instability of the
REMPI system, a proper quantification of the obtained data
was not possible. Nevertheless it has been proven that species
FIGURE 5. Experimental results on runs with “hard” iron ore.
The hard iron ore also exhibits some catalytic activity,
but of a different nature. For example the conversion of
phenol sets in around 300 °C, still some 150° less than in the
blank runsand needed for its gas-phase slow combustion
(16). Next to COx (now with much more CO than CO2),
relatively large amounts of the condensation product dibenzo-
furan are detected. Around 330 °C, the DBF output reaches
even some 35% on the converted phenol, much higher than
the 1-2% levels observed in gas-phase slow combustion (17).
Figure 5 shows a second peak in the output curve of DBF,
correlating with the onset of benzene conversion. This can
be understood if benzene is in first instance oxidized to a
phenolic entity, again subject to condensation. As mentioned
earlier, in runs with HCl added to the original mixture, only
9
VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 3 2 7

FIGURE 6. Summary of PCDD/F outputs in experiments with different chlorine form/quantity.
FIGURE 7. Relative outputs of PCDD/F as a function of the sinterfeed composition.
relevant for dioxin formation can survive the fierce conditions
of the sinter bed, staying available for catalytic reactions.
Experim ental Results versus Real-Life Data. With the
data at handsusing the average of the dioxin outputs in the
two standard runs with water-bound chlorinesan attempt
is made to compare these outputs to those of real-life
processes. (Table 3). Given that pertinent data on large-scale
sintering processes are confidential, hence unavailable,
several important parameters have been estimated.
The calculated total numbers of nanograms of TEQ per
ton of sinter produced in the (model) real facility and in our
microscale setup differ (only) by some 30%; indeed a close
match, also because dioxin outputs from various sinter plants
will not be identical, let alone that fluctuations can occur
within a plant (22).
In an excellent recent publication, Kasai et al. (20) also
favor dioxin formation in the sinter bed itself, although in
experiments with sinterfeed spiked with labeled dioxins the
answers were not clear. Obviously the observed PCDD/ Fs
are a net result of formation and destruction through catalytic
processes inside the bed.
Much work within the frame of the MINIDIP project (6,
10) and by others (11) has been done on the possibility of
dioxin formation in the air ducts and in the APC zone of a
common sintering facility. In the course of these investiga-
tions, it was found that some process-derived materials have
9
3 3 2 8 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

FIGURE 8. PCDD/F congener distributions as a function of feed composition.
FIGURE 9. PCDD/F congener distribution as a function chlorine form and concentration.
a very high PCDD/ F formation potential through de novo
reactions (i.e., conversion of carbonaceous matter on ash/
sinter particles at elevated temperatures). For example,
Buekens et al. (10) report on extremely high dioxin levels, in
the range of 10 000 ng/ g, obtained by reacting some
electrostatic precipitator ashes at temperatures of around
300 °C. In practice, however, the temperature in the ESP
never exceeds 170 °C, so this result, though very interesting,
must be treated with caution. Also, computer-generated
temperature/ dust concentration profiles in the sinter plant
air ducts (22) indicate that particles reside at temperatures
exceeding 300 °C for at best a few seconds. By combining
these results with experimental findings of Buekens et al.
(10), a model was generated indicating that the de novo
activity on fly ashes cannot be held responsible for the overall
dioxin outputs from sintering.
Finally, it should not go without comment that the
experimental PCDD/ F congener distributions (Figure 10) are
at (sometimes considerable) variance with the real-life charts.
The dibenzo-p-dioxins constitute almost 75% of the total of
formed PCDD/ Fs as compared to ∼70% of PCDFs in the
real-life samples; also the chlorination degree is different. It
should be pointed out that the real sinter plant and our lab
facility, based on an hourly sinter output, differ not less than
6 orders of magnitude in scale. This and the process
mechanics (flows, sintering efficiency, perhaps some wall
9
VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 3 2 9

FIGURE 10. PCDD/F toxic congener distribution charts, compared.
TABLE 3. Comparison of Experimental and Real-Life Data on Dioxin Formationa
real life
value
units
value
11
Leiden experimentsb
ng/exp
total PCDD/Fs in PUF plug
TEQ in PUF plug
ng TEQ/exp
ng TEQ/Nm 3
ng/Nm 3
0.25
TEQ m easured at stack (19)
PCDD/F concn at wind boxes (20) 190
concn after APC devices (20)
TEQ after APC (20)
daily output (sinter) (20)
43
73.3
PCDD/F concn
7.3
ng/Nm 3
0.04
ng TEQ/Nm 3
Mg/d, g/exp
Mg/d, g/exp
Nm 3/Mg of sinter
9 600
10 500
2 200
150
200
1 000
0.15
sinter output
sinterfeed input
extrapolated total air flow/ton of sinter
total air throughput
daily inputc
air flow/ton of sinter (21)
total daily air flow
21 120 000 Nm 3/d, Nm 3/exp
dioxin output/ton of sinterd
418 000
88
1.04
ng of PCDDF/Mg of sinter 73 333 extrapolated PCDD/F output/ ton of sinter
ng TEQ/Mg of sinter
TEQ/ton of sinter at the stacke
f
total TEQ wind boxes
ng TEQ/Nm 3
ng TEQ/Mg of sinter
1.67
1 667
TEQ in gas
total TEQ/ton of sinterg
2 290
extrapolated TEQ/ton of sinter
a
b
c
Estim ated values are in italics. Based on the “standard” experim ents (code I and II) with water-bound chlorine. Daily output + 6.5% water
d
e
+ 2% carbon (output +“volatilized weight”). (Concentration at wind boxes × air flow)/sinter output. (TEQ after EPC × air flow)/sinter output;
f (Total TEQ after APC)/(total PCDDF after APC)(PCDDF concentration at wind boxes). (Estim ated TEQ in wind boxes × air flow)/sinter output.
g
effects, etc.) may be comparable but not exactly the same.
Also, the sinterfeeds did not include (models for) substances
Literature Cited
(1) Olie, K.; Vermeulen, K. P.; Hutzinger, O. Chemosphere 1977, 8,
such as millscale oil-containing refuse, ESP ashes/ sludges,
etc. that are likely to have a distinct effect on the formation
of PCDD/ Fs. Moreover, the congener distribution from a
real plant depends strongly on the sampling point (18), as
can be seen in Figure 10 (compare Calcagni’s “scrubber”
and “baghouse” data).
455-459.
(2) Zook, D.; Rappe, C. In Dioxins and Health; Schecter, A., Ed.;
Plenum Press: New York, 1995.
(3) Santl, H.; Gruber, L.; Sto¨ hrer, E. Chemosphere 1995, 29, 1987-
1993.
(4) Brzuzy, L. P.; Hites, R. A. Environ. Sci. Technol. 1996, 30, 3647-
3648.
(5) Cieplik, M. K.; Louw, R.; Pastor Carbonell, J. Organohalogen
Compd. 1999, 41, 141-143.
Acknowledgments
(6) Cieplik, M. K.; Louw, R.; Hell, K.; Stieglitz, L. On catalytic
pathways to dioxins in the iron sintering process. Poster
presentation at the International NCCC-1, 2000 Meeting,
Lunteren, The Netherlands, January 17-19, 2000.
(7) McWilliams, E. L.; Sealy, R. L. Atmos. Environ. 1997, 21, 2661-
2665.
(8) Matusca, P.; Schwarz, B.; Bachmann, K. Atmos. Environ. 1984,
18, 1667-1675.
(9) Jay, K.; Stieglitz, L. Chemosphere 1995, 30, 1249-1260.
(10) Buekens, A.; Stieglitz, L.; Hell, K.; Huang, H.; Segers, P.
Chemosphere 2001, 42, 729-735.
We thank Ir. Marinus te Lindert and the helpful crew of the
R&D laboratory of Corus Nederland, IJmuiden, for supplying
sinterfeed materials; Dr. Kees Olie (University of Amsterdam)
for performing some of the dioxin analyses; and Dr. Horst-
Henning Grotheer and co-workers (DLR Stuttgart) for al-
lowing and helping with the JET-REMPI measurements.
Financial support of the European Union through MINIDIP
Project (ENV4-CT97-0587) is gratefully acknowledged.
9
3 3 3 0 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 15, 2003

(11) Xhrouet, C.; Nadin, C.; De Pauw, E. Environ. Sci. Technol. 2002,
36, 2760-2765.
(12) Cieplik, M. K.; Cairols Oviedo, M.; Louw, R. Chemosphere 2000,
40, 195-199.
(13) Eriksson, L.; Johansson, E.; Kettaneh-Wold, N.; Wikstro¨ m, C.;
Wold, S. In Design of Experiments: Principles and Applications;
Umeå Learnways AB: Stockholm, Sweden, 2000; ISBN 91-
973730-0-1.
(14) Cieplik, M. K. Ph.D. Thesis, Leiden University, Leiden, The
Netherlands, 2003 (in English).
(18) Calcagni, J.; Ciolek, M.; Mulrine, P. In Dioxin Emmissions from
U.S. Sinter Plants; U.S. Environmental Protection Agency, Office
of Air Quality Planning and Standards: Washington, DC, 1998.
(19) Broker et al. Organohalogen Compd. 1993, 11, 303.
(20) Kasai, E.; Hosotani, Y.; Kawaguchi, T.; Nushiro, K.; Aono, T. ISIJ
Int. 2001, 41, 93-97.
(21) Data from http:/ / www.chem.unep.ch/ pops/ POPs_Inc/
proceedings/ stpetbrg/ quass2.htm.
(22) Buekens, A. Free University of Brussels, Personal communica-
tion.
(15) Van den Brink, R.; Mulder, P.; Louw, R. Catal. Today 1999, 54,
101-106.
(16) Born, J. G. P.; Louw, R.; Mulder, P. Chemosphere 1989, 19, 401-
406.
(17) Wiater, I.; Born, J. G. P.; Louw, R. Eur. J. Org. Chem. 2000, 6,
921-928.
Received for review November 1, 2002. Revised manuscript
received April 28, 2003. Accepted May 6, 2003.
ES026292G
9
VOL. 37, NO. 15, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 3 3 3 1