Resources, Conservation & Recycling 157 (2020) 104746

Contents lists available at ScienceDirect

Resources, Conservation & Recycling

journal homepage: www.elsevier.com/locate/resconrec

Review

Pyrometallurgical removal of zinc from basic oxygen steelmaking dust – A T

review of best available technology
Daniel J.C. Stewarta, Andrew R. Barrona,b,c,*
a
Energy Safety Research Institute, Swansea University Bay Campus, Swansea, SA1 8EN, UK
b
Department of Chemistry and Department of Materials Science and Nanoengineerging, Rice University, Houston, Texas, 77005, USA
c
Environmental Engineering Programme Area, Faculty of Engineering, Universiti Teknologi Brunei, Brunei Darussalam

ARTICLE INFO ABSTRACT

Keywords: Approximately 20 kg of dust and sludges are produced per ton of liquid steel produced via the blast furnaces
Basic oxygen steelmaking (BF)/basic oxygen furnace (BOF) production route. Many of these dusts are recycled through the sinter plant or
Dust blast furnace route without issue, but high zinc content dusts are routinely landfilled. Hydrometallurgical
Iron techniques, such as alkaline leaching, that are often utilized to remove zinc from electric arc furnace dusts are
Pyrometallurgy
inappropriate for recovery of material from BOS dust due to the lower zinc concentration present and extra post-
Zinc
processing steps to utilize the separated iron product. Pyrometallurgical treatment through a rotary hearth
furnace (RHF), in processes such as FASTMET®, can currently be considered as the most commercially attractive
option for the processing and recovery of iron and zinc units when employed as part of an integrated steelworks.
The crude zinc oxide produced is suitable for sale to zinc smelters, and the direct reduced iron produced provides
process benefits through use, such as reduced blast furnace coking rates and increased productivity. The ad-
vantages and disadvantages of variations and alternatives are reviewed with regard to future developments.

1. Introduction zinc, the typical process routes become unsuitable. Because the blast
furnace process is remarkably sensitive to volatile metals most critically
An integrated steelworks, consisting of blast furnaces and a basic zinc these, zinc-containing materials have been ordinarily disposed of
oxygen steelmaking facility, is an example of effective materials effi- through landfill. The reason for the sensitivity to zinc is that once
ciency in 21st century manufacturing. The most common process for charged into a blast furnace, any zinc component is reduced to ele-
the production of steel is the basic oxygen steelmaking (BOS) process, mental Zn. Due to the low boiling point of the metal (907 °C) compared
accounting for well over a billion tons of steel produced globally (World to the furnace temperature range (1600–1650 °C), the vapor rises back
Steel Association, 2017a,b). A significant volume of BOS dust is created through the furnace stack and re-condenses (Singh, 2012), leading to
as a by-product of the production of liquid steel through the BOS pro- condensation of scaffolds (accretions) of zinc on the walls of the fur-
cess, which is generally recovered through the off-gas cleaning system. nace. These deposits can affect both solid and gas flow through the
BOS dust is also known as basic oxygen furnace (BOF) dust, BOF fume, furnace negatively impacting productivity and risking damage to the
or BOS slurry, and varies depending on the origin of the material. For furnace lining through burden slips. Zinc is also known to attack re-
simplicity, this paper defines BOS dust as the fine ferrous by-product fractories in the upper stack of the furnace and therefore potentially
produced during the blowing period in an oxygen converter/furnace at impact on campaign life (Narita et al., 1981). As such, the concentra-
a steelmaking facility, which is wet-scrubbed from the off-gas system of tion of zinc loaded to a blast furnace is tightly controlled, with levels
the steelmaking plant and will refer to the material as such when citing around 100–120 g/THM generally permissible (THM = tons of hot
references using differing nomenclature. metal). In context, for a plant producing 10 Mt of liquid steel per
Typically, BOS dust is highly metalized and has iron content (> 60 annum, this would allow for a maximum 1 kt of zinc to be charged to
% FeTot) comparable with good quality ores, and therefore could be a the furnace per annum if these limits are to be followed. The dilution of
valuable secondary source of iron units for an integrated steel plant. zinc bearing wastes and reintegrating them into existing steelmaking
While the recirculation of fine iron-bearing materials back into the blast processes are therefore not suitable for processing zinc-bearing wastes
furnace is commonplace, when these materials are contaminated with on a sustainable scale. The volume of zinc re-entering the steelmaking

⁎
Corresponding author at: Energy Safety Research Institute, Swansea University Bay Campus, Swansea, SA1 8EN, UK.
E-mail address: [email protected] (A.R. Barron).

https://doi.org/10.1016/j.resconrec.2020.104746
Received 11 October 2019; Received in revised form 3 January 2020; Accepted 3 February 2020
0921-3449/ © 2020 Published by Elsevier B.V.
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

process through galvanized scrap steel recycling is simply too much.

Zinc and iron are routinely married together through the hot-dip
galvanization process, with the zinc providing galvanic protection for
the steel. It has been postulated that without substantial improvements
to the recovery rate of zinc from its industrial uses (principally hot-dip
galvanizing) global zinc reserves will be outstripped by demand as early
as 2050 (Daigo et al., 2014). It is therefore clear, that without a go-
between process to economically remove zinc from the steel material
cycle, the process of hot dip galvanizing will be dependent on depleting
global reserves of zinc. A number of hydrometallurgical techniques,
such as alkaline leaching, are often utilized to remove zinc from electric
arc furnace dusts (Lin et al., 2017; Shawabkeh, 2014; Dutra et al.,
2006); however, these are inappropriate for recovery of material from
BOS dust due to the lower zinc concentration present, in the latter, and
Fig. 1. Global Steel production by process route over the past 30 years.
extra post-processing steps to utilize the separated iron product.
Data adapted from the World Steel Association annual production statistics
Nevertheless, ammoniacal leaching has been reported to have results
(World Steel Association, 2017a,b).
for wastes from BOS furnaces where the zinc content is 2.82 % (Gargul
and Boryczko, 2015).
There are several emerging novel techniques for the pyr- no external heat input due to the extremely exothermic nature of the
ometallurgical separation of zinc from steelmaking dusts, with one of chemical process. In the BOS process, molten pig iron is charged into a
the most promising being rapid microwave carbothermal reduction converter lined with basic refractory bricks, with scrap steel typically
(Sun et al., 2008) which has advantages such as rapid reaction times used as an additional iron source and coolant which can make up 25 %
(99.99 % Zn removal in 15 min at 1100 W) and feasible capture of a by mass of the charge (Singh, 2012). Oxygen is then blown into the
metallic zinc product. Hybridized pyrometallurgical/hydro- vessel at supersonic speeds (Brandt and Warner, 2005) through a water-
metallurgical techniques have also been studied such as chlorination cooled lance to oxidize impurities within the molten iron, which are
roasting followed by a leaching step (Jaafar, 2014). These have ad- then trapped in the molten slag or emitted from the vessel as a gas.
vantages in the lower processing temperatures required (750 °C) and The chemical reactions within the basic oxygen steelmaking process
hence lower energy input when compared to other pyrometallurgical are given in Eqs. (1)–(8), where the state of the reagents is signified by
techniques. However, the disadvantages of a poorly valorized ferrous the choice of braces, “{}”, to denote a gas phase reactant/product,
product and comparatively low Zn removal (97 %) mean upscale is square brackets, “[]”, to denote the reactant/product is dissolved in the
unlikely. These technologies are still in their infancy, having not bulk Fe(l) phase, and round brackets, “()”, denote the reactant/product
achieved pilot scale operation as of the time of writing, and as such will is dissolved within the liquid slag phase. The oxygen injected by the
not form the main basis of this review – instead focusing more on lance dissociates into the molten iron bath (Eq. (1)), which then acts to
proven scalable technology and how they might be applied to process rapidly oxidize impurities within the melt such as C, S, Si, Mn and P
BOS dust. which are either captured into the molten slag or escape the bath as gas
The need to adopt commercial routes to the re-use of zinc waste in the case of carbon. Fluxes such as CaO are also added as a processing
from BOS dust is part of a broader move to lower the environmental aid, to promote the removal of S and P from the liquid iron. These
impact of the steel industry (Allwood et al., 2010; Kim and Worrell, equations are an oversimplification of the complex equilibria between
2002) and ensure its economic sustainability (Fisher and Barron, 2019; the slag and metal interfaces but are useful nonetheless to describe the
Lobato et al., 2015). Thus, there are significant incentives to find ways overall sequestration of unwanted elements in the hot metal to the slag
to as much waste as possible. Herein, we review approaches to the re- or to the gas phase.
use of BOS dust, with particular focus on those that provide the best {O2} → 2[O] (1)
available technology to supplement an integrated steelworks in pro-
cessing its zinc-bearing by-products, and to make recommendations to [C] + [O] → {CO} (2)
the industry based off viability, technical feasibility and environmental
[Si] + 2[O] → (SiO2) (3)
considerations. The goal is to ensure that BOS dust is considered a
material with potential as resource rather than a waste. 2(CaO) + (SiO2) → (2CaO·SiO2) (4)
The objective of this paper is to provide a short to medium term
[Mn] + [O] → (MnO) (5)
outlook for the feasibility of pyrometallurgical separation techniques
for BOS dusts. The recycling of low zinc bearing materials is routinely 2[P] + 5[O] + 3(CaO) → (Ca3(PO4)2) (6)
performed at most steel plants through dilution in the blast furnace
burden. Very high Zn bearing materials such as Electric Arc Furnace [FeS] + (CaO) → (FeO) + (CaS) (7)
dusts are very well studied and reviewed (Lin et al., 2017; Walburga 2[P] + 5(FeO) + 4(CaO) → 4(CaO·P2O5) + 5[Fe] (8)
Keglevich de Buzin et al., 2017) due to the hazardous nature of the
material. Materials that are moderately contaminated with Zn such as
BOS dust are comparatively understudied, despite the fact they are 3. BOS dust
produced on a far greater scale.
The BOS process naturally generates co-products alongside the de-
2. The basic oxygen steelmaking (BOS) process sired liquid steel, and while the largest by mass of these co-products is
steelmaking slag (Fisher and Barron, 2019), the volume of ferrous dusts
Basic oxygen steelmaking is the most common production route for cannot be ignored. As a consequence of the injection of oxygen into the
steel in the world today (Fig. 1). Immediately following WWII, it be- melt at ultrasonic speeds as well as the turbulent conditions of the bath
came commercially viable to mass produce huge volumes of high purity caused by so-called ‘carbon boil’, a significant amount of fine material is
oxygen for industrial use. This availability of oxygen for use as an ejected from the bath. This fine material is then scrubbed out of the off-
oxidizing agent in the steel industry rapidly supplanted the outdated gas from the process through either wet venturi scrubber systems
Open-Hearth Furnace, as BOS plants are more productive and require (Onnen, 1972) or dry electrostatic precipitators (Henschen, 1968)

2
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

before the off-gas is suitable to either collect for caloric recovery or to molten iron free of the melt (Fig. 2f).
flare to the atmosphere. The BOS dust is then collected either as a filter From investigations into these two mechanisms it is abundantly
baghouse dust or a wet filter cake. clear that there is a relationship between the carbon content of the iron
bath and the volume of fumes generated. It has been shown in lab scale
(Turkdogan and Leake, 1959; Hoskins et al., 1966) and full production
3.1. Formation mechanisms of basic oxygen steelmaking dust scale at the steel plant in Port Talbot, UK (Steer et al., 2014) that at
higher carbon concentrations within the iron bath, the rate of fume
The mechanisms of formation of BOS dust have been well re- generation is higher. However, both mechanisms can be dependent on
searched, as the conditions within the furnace and off-gas system can the carbon levels so this alone is not enough to determine the pre-
have a significant effect on the chemistry of the dust product. In 1944 dominance of each mechanism. Tsujino et al. devised a method of de-
Kohlmeyer and Spandau theorized that the reaction of gaseous carbon termining the ratio of so-called ‘bubble burst’ particles to ‘fume’ parti-
monoxide formed from partial carbon oxidation was reacting with cles to elucidate the dominant production mechanism (Tsujino et al.,
elemental iron to yield iron pentacarbonyl, i.e., Eq. (9) (Kohlmeyer and 1989).
Spandau, 1944). Molybdenum and manganese are both typical alloying elements that
Fe0 + 5CO → Fe(CO)5 (9) may be found as part of the melt in a BOS converter, but have sig-
nificantly different boiling temperatures, 4612 °C and 1962 °C, respec-
Since Fe(CO)5 decomposes readily back to Fe° at comparatively low tively. By assuming molybdenum volatilization from the bath is negli-
temperatures (426 °C) (Santos et al., 1983), and therefore it’s unlikely gible it can be assumed that any molybdenum present within the BOS
that it would be able to form in the steelmaking environment at a dust must have been mechanically ejected from the bath via the bubble
temperature in excess of 1500 °C. It would also be expected that per- bursting mechanism rather than volatilized, while the manganese
forming a blow using carbon monoxide would also cause the char- concentration would indicate volatilization. This approach allowed for
acteristic visible fuming associated with the formation of BOS dust, but direct measurement of the ratio of the two particle types throughout the
in practice, this does not occur (Turkdogan and Leake, 1959). In reality, blow and showed that the initial dust produced in the blow was prin-
there are two competing mechanisms at play in the formation of BOS cipally through the bubble bursting mechanism, whereas at the con-
dust, and these are the vaporization or fuming mechanism, and the clusion of the blowing period the fuming/vaporization mechanism
bubble burst mechanism (Nedar, 1996). dominated.
Initially the vaporization of iron from a steel bath seems rather It should be note that the volumes of dust generated from BOS
counterintuitive because the standard boiling point of iron is 2861 °C converters are widely variable, usually estimated at around 15–20 kg/
and the temperatures in oxygen steelmaking are around 1000 °C lower. THM. There is clearly many dependent process variables that can affect
Nevertheless, Goetz described the phenomenon of localized ‘hot spot- the composition of BOS dust, including: lance height, oxygen injection
ting’ within the BOS process (Goetz, 1980), where the extremely exo- pressure, and slag volume (Okhotskii, 2007; Gritzan and Neuschütz,
thermic reactions at the gas/metal interface generate heat so rapidly 2001).
that it cannot be transported away effectively by convection into the
bulk of the bath. Temperatures between 2400–2600 °C, as measured on 3.2. The origin of zinc in the BOS process
industrial basic oxygen furnaces using optical pyrometry, have been
reported. This localized hotspot formation leads to volatilization and The use of galvanized strip steel has increasingly found use in au-
oxidation of iron from the bath in the form of a fine fume. tomotive body panels, due to its formability, corrosion resistance and
The second mechanism is the mechanical ejection of metal droplets good strength/mass ratio. Steel is the most widely recycled material on
from the bath due to the collapse of CO bubbles formed in situ (Eq. (2)). the planet (World Steel Association, 2017a,b) by mass and in order for
This reaction spontaneously forms bubbles of carbon monoxide within steelmaking to be truly sustainable, galvanized steel must also be re-
the steel bath and is critical to the speed of the BOS process, since it cycled at the end of its useful service life. Should any galvanized steel
provides excellent bath agitation and mixing. Once these bubbles have be included in the scrap charge to the BOS furnace at the beginning of
formed in the extremely high temperatures found in the molten iron, the process, any zinc present will volatilize almost completely from the
they are forced upwards due to their relative density before reaching basic oxygen furnace, upon addition of the pig iron.
the surface (Fig. 2a). Upon reaching the surface of the hot metal, the top Scraps with high zinc content are often cheaper than low zinc
of the bubble thins (Fig. 2b) until it's surface tension can no longer content scrap for this reason. Electric arc furnace (EAF) dusts are sub-
contain the hot gas, which break clear of the liquid and escapes to at- stantially higher in zinc content, along with other tramp elements such
mosphere (Fig. 2c). The resulting high pressure at the edges of the now as lead, due to their nature of having typically 100 % scrap charges,
ruptured bubble (Fig. 2d) form a jet (Fig. 2e) that ejects droplets of (Oustadakis et al., 2010). Integrated plants generally manage BOS dust
recycling through careful management of scrap chemistry inputted to
the furnace, then reintroducing the material to the sinter plant and then
the blast furnace but extremely tight limitations on zinc input mean
unexpectedly high zinc material cannot be processed in this manner.

3.3. Off-gas system design

One key design feature of a BOS plant’s off-gas system, that can have
a substantial effect on the chemistry of the dust product, is whether it is
a closed or open hood design. In a closed hood system, air ingress into
the off-gas system is extremely limited and as such degree of oxidation
as well as the formation of some spinel phases are suppressed. This is
Fig. 2. Schematic of the bubble bursting mechanism for the formation of BOS typical for a steel plant with an off-gas recovery system in place, as the
dust, showing (a) a CO bubble formed in the hot metal, (b) reaching the surface gaseous product of the BOS process has substantial caloric value and is
it thins, (c) breaking clear of the liquid and escaping to atmosphere, where (d) utilized as an energy source. BOS dust from these systems typically has
the resulting high pressure at the edges forms a jet (e) that ejects droplets of a black appearance. In an open hood system, the opposite occurs, the
molten iron free of the melt (f). environment of the system is much more oxidative and combustive. As

3
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

such higher iron oxides are expected and the off-gas dust has a more solidification of liquid iron droplets ejected via the bubble burst me-
reddish appearance (Goetz, 1980). chanism.
Using SEM Heinreich classified three distinct particle groups within
4. Characterization of BOS dust Port Talbot BOS dust (Heinrich, 2015). The first and most abundant is
described as ‘fines’, particles sized around 5 μm. These are consistently
As a potentially valuable ferrous resource, BOS dust from steel produced throughout the blow period, and zinc content is distributed
plants all around the world has been characterized using a number of evenly throughout this fraction of the powder. Iron spheres were then
different techniques. Although many different monikers are used to described as the next most abundant. These are much larger (up to
describe the material, it is defined in this work as the fine material that 500 μm) in diameter and contain less zinc than the finer fraction. The
is removed from the off-gas system from a BOS vessel, produced during sizing of these particles corresponds well to theoretical calculations of
the blowing period. the upper limit of the diameter of a liquid steel droplet to be ejected
from steel, which was in the region of 500 μm (Hahn and Neuschütz,
4.1. Particle size 2002) Fig. 4.

It has been reported that the distribution of particle size within BOS 4.3. Chemical analysis
dust samples is from 0.5 μm up to around 50 μm as determined by laser-
based granulometry. Dividing the dust into a fine (< 38 μm) and coarse The determination of the chemical composition of steelmaking
(> 38 μm) fraction, it was also found that the finer fraction was con- materials is carried out via X-ray fluorescence (XRF) spectroscopy or
siderably more Zn rich (Kelebek et al., 2004). Gritzan and Neuschütz atomic emission/absorption spectroscopy (AES/AAS). XRF is generally
subdivided BOS dust into four classifications by size fraction with used for analysis for non-trace elements, or in silaceous samples such as
maxima at 1, 12, 50 and 140 μm, with the largest fraction likely ex- in ore characterization, but for zinc determination in BOS dust, diges-
tending past the detection limit of the granulometry technique used tion of samples followed by inductively coupled plasma atomic emis-
(200 μm) (Gritzan and Neuschütz, 2001). However, it was also found sion spectroscopy (ICP-AES) is generally the preferred technique due to
through SEM observations that many of these larger particles were its sensitivity to low concentrations and accuracy (Orbaek and Barron,
agglomerations of the finer particles (Gargul et al., 2016). Along with 2015) with detection limits for solid samples as low as 2.5 ppm and
the fine and agglomerated particles, the spheres of iron ejected from the 0.01 ppm respectively being achievable.
melt were also found to be present around 50 μm (Gargul et al., 2016). Typical metal analysis of BOS dust from literature is shown in
A study on BOS dust particle size distribution between two different, Table 1, from which it is clear that there is a great deal of variability in
unnamed integrated steelworks in the USA showed differences in the the overall composition of samples. This is particularly true in the case
main fractions of the dust, with one plant having a substantially larger of zinc and iron content as these are the two key value drivers of metals
fine fraction < 75 μm than the other (Ma, 2016). An investigation on recovery of BOS dust. Zinc content of BOS dust can be controlled sub-
BOS dust produced through the Port Talbot works (UK) showed varia- stantially through the elimination of galvanized scrap’s inclusion in the
tion in the particle size analysis throughout the blowing process scrap charge to the furnace, will levels capable of reaching 0.1–0.2 wt%
(Heinrich, 2015). A greater proportion of finer particles were generated Zn but total elimination of zinc from the material is extremely chal-
during the blow than before or after. These findings were also in lenging because of the technical challenges of limiting zinc con-
agreement with that of Kelebek that showed the finer fraction of BOS tamination in merchant scrap and residual levels of zinc within the hot
dust contained a higher proportion of zinc than the coarser fractions metal itself (∼50 ppm).
(Kelebek et al., 2004). The implication of this observation combined EDX analysis also indicates magnesium and calcium are also present
with SEM measurements observations sheds some light on the inter- within BOS dust (Vereš et al., 2015), which is likely the result of fine
action of the ferrous material with the zinc vapor present immediately flux additions to the BOS vessel during processing blowing out through
above the BOS vessel. the off-gas system and being collected together with the metallic dust
These observations of Zn enrichment of the finer fraction and zinc (Vereš et al., 2015).
being present on the outer surface implies that the iron morphology is
dictated by the mechanisms while zinc is still in the gas phase (see 4.4. Phase determination
above). Once the gas stream has cooled sufficiently, Zn condenses onto
the surface of the material as ZnO and then potentially goes on to react X-ray diffraction (XRD) analysis is a common tool for character-
to form ZnFe2O4 at the Fe/Zn interface if the conditions are suitable. ization of the crystalline phases in steelmaking raw materials, products
and wastes. It is a relatively insensitive tool with detection limits of
4.2. Morphology around 1 wt%. Quantitative analysis can be challenging as the tech-
nique is very matrix dependent but is still very effective for rapid
Kelebek et al. described the morphology of BOS dust, from scanning crystallographic analysis. When determining effective recycling solu-
electron microscopy (SEM), as comprising a very heterogeneous mix- tions for BOS dust an important consideration is the chemical form of
ture of particles (Kelebek et al., 2004), with sizes ranging from 3 μm to the zinc within the material. The two likely candidates are ZnO (zincite)
100 μm (Fig. 3a). Many of the finer particles are agglomerated to larger and ZnFe2O4 (franklinite) and the relative abundance of these com-
particles, differentiated as spheroidal and non-spheroidal (Fig. 3b). pounds is of great significance. Unfortunately, Fe3O4 and ZnFe2O4 have
Energy dispersive X-ray spectroscopy (EDS) showed a relatively uni- very similar crystallographic unit cells (Fd 3 m) and the powder XRD
form distribution of zinc across these particles. This information was patterns are nearly indistinguishable (Fig. 5), which makes XRD a po-
used to determine that the Zn in BOS dust is localized almost entirely as tentially unreliable measure of the zincite:franklinite ratio present in
a coating around an iron core. This observation was confirmed through BOS dust and alternatives need to be explored. Nevertheless, powder
a cross-sectional analysis of a particle, showing a metallic iron core, XRD can still be a useful tool in exploring the phases present within BOS
surrounded by iron oxides and zinc oxide/zinc ferrite mixtures dust.
(Fig. 3c). These observations were confirmed by other researchers Reports of the phases present within BOS dust are tremendously
(Vereš et al., 2015), and are in agreement with the observations of two varied, which reflects how substantially the conditions within the BOS
separate and distinct mechanisms of formation: finer more oxidized vessel and off-gas system can affect the chemistry of the by-product.
particles produced through volatilization of iron and much larger, Table 2 shows a selection of phases identified by different authors from
highly metalized and globular particles generated from the steel plants across the world. It is clear from Table 2 that the

4
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Fig. 3. SEM images of (a) globular spheroid particle, (b) agglomerated particle consisting of submicron particles, and (c) cross sectional view of globular particle
showing metallic Fe core.
Adapted with permission from Kelebek et al. (2004), copyright Elsevier 2004.

Fig. 4. SEM images of ultrafine particles adhering to the exterior of larger, globular particles.
Adapted with permission from Vereš (2015), copyright Taylor and Francis 2015.

morphology of BOS dust is drastically variable from plant to plant and franklinite (Krishnamurthy et al., 1974).
even based on the age of the dust that’s analyzed and the environment It has been reported that dust from the Port Talbot (UK) plant, that
the material was stored in. has been stored outdoors for a number of years, showed significant
The difference between closed and open flue gas hood systems, re- oxidation of iron to Fe3O4 and Fe2O3 compared to a more recent sample
ferred to as suppressed and complete combustion, was investigated (Jaafar, 2014). Such ‘weathering’ is to be expected during outdoor
using powder XRD (Ray et al., 1997). The differences in the crystalline storage.
phases present are stark. In closed hood systems the dust is far more Calcium in the form of CaCO3 and CaO is added as a refining flux to
metalized, with the majority of iron present existing in the metallic the BOS process and calcium compounds are identifiable in BOS dust as
form (Fe°) or as FeO (i.e., FeII), while the calcium is present as CaCO3 a result. In the heat of the BOS vessel some CaCO3 decomposes to CaO
(calcite). In the open hood systems, where the BOS gas is combusted (Eq. (10)).
fully before dedusting (Goetz, 1980), the dust is as expected, much
CaCO3 → CaO + CO2 (10)
more oxidized. Iron exists principally as FeIII or mixed FeII/FeIII oxides
(Fe2O3 and Fe3O4, respectively), and the calcium tends to be present as In the off-gas system, any CaO reacts with CO2 present and reforms
CaO rather than the carbonate. It is likely that the conditions in an open CaCO3 via the reverse reaction. It is worth noting that this carbonation
hood off-gas system would favor the formation of ZnFe2O4 as high of CaO happens rapidly within the off-gas system and not as part of the
temperatures and oxidizing conditions are typical for the production of weathering process during storage. Heinrich evidenced this in-situ

Table 1
Elemental abundance in samples of BOS dust (wt%).
Source Ca Fe Mg Mn Pb Zn Reference

Unknown 7.40 59.0 2.1 1.48 – 1.48 Mikhail (1998)

Dofasco Hamilton – 61.0 – – – 1.59 Kelebek (2004)
Dofasco Hamilton – 50.16 – – – 2.4 Goetz (1980)
Tata Steel Port Talbot – – – – – 4.8 Steer (2014)
Tata Steel Port Talbot 3.0–8.80 50.0–80.0 0.20–5.0 0.40–2.20 0.20–1.80 1.7–6.5 Heinrich (2015)
ArcelorMittal Monlevade (Brazil) 4.18 50.65 1.49 – 0.07 4.37 Cantarino (2012)
U.S. Steel Kosice (Slovakia) 5.5 49.87 2.68 – 0.24 9.37 Vereš (2015)

5
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

iron cools the solubility of carbon decreases and a layer of graphite can
form on the surface of the liquid in a foliated dendritic structure (Liu
and Loper, 1991). This graphite is known in the steel industry as ‘kish’
and could be introduced into the BOS off-gas system during charging of
pig iron, as kish is prone to become airborne due to its low density.

4.5. Mössbauer spectroscopy

Mössbauer spectroscopy has historically been used to characterize

oxidation states of iron within iron ores (Salama et al., 2015), as well as
a range of other iron containing materials (Zboril et al., 2002; Kuzmann
et al., 2003; Rose et al., 2002). While providing insight into the oxi-
dation state and chemical environment of iron within a sample of BOS
dust, its principal use in the study of BOS dust is its ability to indirectly
differentiate ZnO from ZnFe2O4 (Vereš et al., 2015).
Fig. 5. Calculated powder XRD patterns (using VESTA software) for magnetite 57
Fe Mössbauer spectroscopy was used to confirm the chemical
#1011032 (Bragg, 1915) and zinc ferrite #9002487 (Pavese 2000) showing the environments in which iron was present in the sample. Franklinite was
similarity in diffraction patterns and the difficulty in differentiation. positively confirmed by the presence of a central doublet in the spec-
trum (Vereš et al., 2015). Total Fe and Zn content was confirmed
Table 2 through digestion and subsequent Atomic Absorption Spectroscopy.
Phases identified in BOS dust by powder XRD. The ratio of ZnO to ZnFe2O4 could then be calculated and for the BOS
Source Phases present Reference dust used in the study, the sample was 13.47 wt% ZnFe2O4 and 7.10 wt
% ZnO. This suggests that the ratio of zinc existing in the ZnO:ZnFe2O4
Dofasco Hamilton FeO, Fe2O3, ZnFe2O4 Kelebek (2004) forms is 60:40; however, it is critical to note that from the work in
Not known α-Fe, FeO, Fe3O4, Fe2O3, CaCO3, Gargul (2016)
question it is unclear whether the plant in Slovakia that the dust was
SiO2, ZnO
ArcelorMittal Méditerranée Fe3O4, Fe2O3, CaCO3, C Sammut (2008)
generated by has an open or closed hood off-gas system (Vereš et al.,
Port Talbot α-Fe, FeO, Fe3O4, Fe2O3, CaCO3, Jaafar (2014) 2015). The abundance of higher Fe oxides and lack of metallic iron
SiO2, ZnO, ZnFe2O4, C present within the sample, would suggest a combustive process, which
Port Talbot α-Fe, FeO, CaCO3, ZnFe2O4 Jaafar (2014) may lead to a greater degree of zinc existing in the franklinite form.
Not known Fe3O4, Fe2O3, CaCO3, Ca(OH)2, Mikhail (1998)
Mössbauer spectroscopy shows tremendous potential for the ana-
ZnO, ZnFe2O4
U.S. Steel Kosice (Slovakia) α-Fe, FeO, Fe3O4, Fe2O3, ZnO, Vereš (2015) lysis of zinc in steelmaking wastes as the ZnO/ZnFe2O4 ratio has huge
ZnFe2O4 implications for extractive process selection, see below.

5. Pyrometallurgical separation techniques

carbonization through the use of a novel slurry sampling system that
showed calcium in BOS dust produced moments prior was already in Based upon a thermodynamic modelling study (Hay and Rankin,
the form of CaCO3 (Heinrich, 2015). 1994) it was proposed that a range of operating regimes and choice of
Fig. 6 shows the calculated dominant phases at 1000 K (approx- process were feasible for the pyrometallurgical treatment of BOS dust to
imate BOS gas temperatures within the off-gas system). As the ratio of remove the zinc (and other volatile elements) and reuse the iron as hot
the partial pressures, PCO/PCO2, approaches zero as combustion of metal, metallized clinker or iron oxide clinker. As such pyr-
carbon monoxide nears completion the dominant zinc phase changes ometallurgical processes for removal of zinc from steelmaking by-pro-
from ZnO to ZnFe2O4. ducts are somewhat well established and offer some attractive features
SiO2 reported in some samples has likely been introduced into the such as good scalability and (generally) continuous processing. Pyr-
sample from the splashing of BOS slag during the process (Gargul et al., ometallurgical separation plants are, however, usually capital intensive,
2016). Graphitic phases have been reported in a number of studies with high setup costs and energy demand. Consequently, in the case of
(Jaafar 2014; Sammut et al., 2008) and these could correspond to the separation of iron and zinc in steel by-product recovery, carbothermal
introduction of ‘kish’ to the sample in the process. As carbon-saturated reduction and separation by volatilization is the industry standard. An
obvious downside of carbothermal reduction is the production of CO2
as an intrinsic product of the process, creating greater carbon emissions
to an industry already having to struggling to meet the challenges of
decarbonization (Worrell et al., 2001; Lin and Wang, 2015; Kim and
Worrell, 2002; Allwood et al., 2010; Gielen and Moriguchi, 2002).
Pyrometallurgical plants also generally produce a large amount of dust
and noise, which must be taken into consideration when assessing plant
feasibility.
The relatively high volatility of zinc relative to iron means that it
can be physically separated from BOS dust by heating under reducing
conditions. Carbon is gasified through the Boudouard reaction (Eq.
(11)) and then ZnO is reduced to elemental Zn vapor by carbon mon-
oxide (Eq. (12)).

C + CO2 → 2CO (11)

ZnO + CO → Zn + CO2 (12)

Fig. 6. Predominance diagram for the Zn-Fe-CO-CO2 system at 1000 K. Zinc in the vapor phase readily oxidizes in the presence of CO2 to
Calculated using FactSage 7.3 (Isobar (+) at 1 atm). reform ZnO via the reverse reaction. Zinc ferrite reduces in a similar

6
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

fashion, reacting with CO (Eq. (13)) or thermal decomposition of 5.1. The Waelz kiln
ZnFe2O4 to ZnO (Eq. (14)) between 719 °C and 1050 °C (Lee et al.,
2001). One of the oldest treatments for zinc containing dusts, the Waelz
process, involves the heating of dust mixed intimately with a carbon
3ZnFe2O4 + 4CO → 3Zn + 2Fe3O4 + 2CO2 (13)
fuel, then heated > 1000 °C in a rotating kiln, and fired by a burner at
ZnFe2O4 → ZnO + 2Fe2O3 (14) the exit end of the tunnel (Fig. 8). A typical Waelz kiln is around
50–70 m long with a diameter of 4–5 m. The zinc containing com-
Under hot, reducing conditions such as these, reduction of iron pounds are reduced, volatilized and consequently reoxidize in the gas
oxides also occurs, which is very favorable for ferrous recovery, because stream, thus, separating the zinc from the ferrous material.
a semi-metalized iron product from the separation process has con- The charge for the furnace is prepared by pelletizing and is charged
siderable benefits for an integrated steelworks. Direct reduced iron at the top (entry end) of the furnace, as the drum rotates the burden
(DRI), or sponge iron, can be directly charged into the iron making transits down the drum and reacts with the reducing gas generated from
process, thus reducing reductant rates within the blast furnace, dis- the carbon source. The process has several disadvantages, such as long
placing expensive coke, and improving the process economy retention times of up to 8 h that lead to very low productivity.
(Stephenson and Smailer, 1999). Typically, Waelz kilns found use in the recovery of zinc from EAF dust
DRI can be produced in a variety of ways, including countercurrent or zinc ores with a mineral content too low for more traditional mineral
shaft furnaces utilizing reformed natural gas, fluidized beds and others dressing (Porter, 1991). Waelz kiln processing is globally well estab-
(Anameric and Kawatra, 2007; Chatterjee, 2012), but the critical con- lished with over 1 million tons of capacity for EAF dust in the late 1990s
sideration for processing zinc-bearing materials is the direction of gas (Mager et al., 2003, 2013), which increased to the 3.4 million tons of
flow relative to the oxidizing gradient. In order to prevent condensation EAF dust capacity across 35 kilns currently operating globally (Masson
and recirculation of zinc within the furnace, the oxidizing gradient must and Briol, 1969). However, the Waelz kiln cannot economically process
be perpendicular to the material flow direction: to carry condensed zinc material lower in zinc than around 10 wt%, due to the relatively low
oxide away from the ferrous burden. This effectively rules out the value of the iron-bearing product and the thermal inefficiency of the
countercurrent natural gas reduction units such as the Midrex® process process.
often used to produce DRI at mine sites (Atsushi et al., 2010). Fig. 7 Table 4 shows a typical chemical analysis from a Waelz kiln of the
provides an outline of the process (Fruehan, 2005). input material (EAF dust), the ferrous product and the zinc-rich ‘waelz
Solid based reduction processes are much more effective in the oxide’ collected through the off-gas dedusting system. Often fluxes are
processing of zinc-bearing by-products. Common reductant sources in- added to the infeed to manage slag chemistry and prevent accretions or
clude coke breeze, pulverized coal, and carbon-bearing revert material ‘slag rings’ forming within the kiln at specific thermal zones. These
such as blast furnace dust (Ahmed, 2018). Table 3 compares gas direct rings or accretions can reduce productivity through disruption solid
reduction and solid fuel direct reduction to conventional blast furnace material flow through the furnace, attacking refractory and increase
pig iron production. Seeing as both the gas-solid and blast furnace are maintenance costs. Of course, the addition of fluxes in the form of
both unsuitable for Zn bearing by-products such as BOS dust this review mineral oxides are not a perfect solution however, as introducing
will only discuss in detail the solid-solid DRI producing processes that gangue material into the ferrous burden decreases its value in use for
can tolerate high zinc loadings. recycling purposes.

Fig. 7. Simplified schematic process flow for a two-stage natural gas DRI plant such as Midrex®.
Reproduced with permission from Fruehan (2005). Copyright Taylor & Francis.

7
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Table 3
Comparison of DRI production processes with traditional blast furnace production.
Use Blast furnace iron making Gas-solid direct reduction Solid-solid direct reduction

Reducing agent Coke, can be supplemented with pulverized coal/ Reformed natural gas, by-product gases Coal, coke breeze, carbon-bearing revert material
tar injection such as coke oven gas (COG)
Ferrous input Sinter, ore, DRI. Low volatile metal content is Limited to ore, volatile metal loading Flexible: ore, waste oxides, etc. Volatile metals are not
material critical similar to a blast furnace restricted
Input material form No fine material, pellets/lump ore. Must have Pellets, lump, some processes utilize fine Pellets or fine material, depending on the specific
sufficient strength to withstand furnace conditions material such as fluidized bed processes process
Furnace design Large vertical shaft furnace, charged from the top Typically, a vertical shaft furnace Rotating kiln or rotary hearth furnace
with a hot blast fed in through tuyeres at base
Product Molten pig iron Direct reduced iron Direct reduced iron or Waelz slag depending on the
process
Product quality Very high quality, complete separation of Fe and Usually, excellent quality depending on Dependent on feedstock material and coal quality.
gangue. Slag formed has some desulfurizing the quality of the ore and reducing gas Usually unsuitable for EAF/BOS use due to high S and
capacity gangue content
By-products Liquid slag, dusts, caloric off-gas which can be Caloric off-gas which can be combusted Off-gas fully combusted in situ, baghouse dust contains
reused to heat stoves to heat process, dusts ZnO product suitable for sale

Fig. 8. Schematic representation of the Waelz kiln process.

Reproduced with permission from Mager et al. (2003, 2013). Copyright Springer-Verlag.

Table 4 5.2. The rotary hearth furnace

Chemical analysis (Wt%) of the inputs and products of the Waelz kiln process
(Mager et al., 2003, 2013). Initially patented in the 1960s the rotary hearth furnace process
Element or compound EAF dust Ferrous product Waelz oxide takes advantage of the high reaction speeds afforded through the firing
of self-reducing composite pellets of iron oxide and carbon. A rotary
Wt% Wt% Wt% hearth furnace (RHF) consists of a circular turntable rotating inside of a
refractory lined tunnel (Fig. 9). Heat is supplied by natural gas or
Zn 18.0–25.0 0.2–2.0 55.0–58.0
Pb 2.0–7.0 0.5–1.0 7.0–10.0 pulverized coal burners, but the majority of the process heat comes
Cd 0.03–0.1 < 0.01 0.1–0.2 from the combustion of the carbon source within the charge itself. The
F 0.2–0.5 0.1–0.2 0.4–0.7 gas flow is countercurrent to the material flow so hot process gases can
Cl 1.0–4.0 0.03–0.05 4.0–8.0
preheat the freshly charged pellets, resulting in good heat exchange and
C 1.0–5.0 3.0–8.0 0.5–1.0
FeO 20.0–38.0 30.0–50.0 4.0–7.0
thermal efficiency.
Femet/Fe – 80.0–90.0 – RHFs can accommodate a far wider range of materials than a BF,
CaO 6.0–9.0 15.0–25.0 0.7–1.2 due to the lack of vertical loading of the furnace burden. In the BF,
SiO2 3.0–5.0 6.0–12.0 0.5–1.0 pellets nearer the bottom of the stack must support the mass of the
Na2O 1.5–2.0 1.2–1.6 0.1–0.2
burden above it. Typically blast furnace pellets must have a cold
K2O 1.0–1.5 0.7–0.9 0.1–0.2
compression strength of > 2450 N per pellet to withstand the com-
pressive load of the furnace (Lu et al., 2015). In the RHF, the burden

8
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Fig. 9. A schematic cross-sectional view of the zinc removal mechanism in the rotary hearth furnace (RHF).

layer is at most 3 pellets thick. As such it is permissible to charge ma- 5.3. Fastmet®
terials with far less compressive strength: > 40 N per pellet is accep-
table to be charged to the furnace. Developed by Kobe Steel, FASTMET® is one of the most commer-
There are many different commercialized processes for direct re- cially established RHF processes globally, with around a million tons of
duction using an RHF such as FASTMET®, DRyIron®, and INMETCO® steelmaking waste processing capacity spread across six plants in Japan
but all follow very similar process principals. The chemical process (Kobelco, 2019). Fig. 10 shows a schematic of the FASTMET® process
remains the same for each process (Table 5), the carbon source in the flow.
pellets begins to gasify to form a localized reducing atmosphere within Pellets are produced by the blending of iron bearing material with a
and proximal to the furnace burden. The CO then reduces the iron stoichiometric amount of carbon (coal, coke breeze, or carbon-bearing
oxides stepwise to metallic iron, as well as volatilizing any Zn and Pb revert material) with a binding agent. These pellets are charged in one
present. Solid state reduction also occurs between C and FexOy but this or two layers onto the hearth floor. The composite pellets are then
is slow and much less significant. Post-combustion of CO to CO2 is heated to around 1300–1350 °C and over the course of 8–16 min react
exothermic and supplies more thermal energy to the process to drive and are discharged via a water-cooled rotating screw.
the endothermic gasification and reduction reactions. Hot DRI is discharged from the exit end of the furnace which can
A key advantage of the RHF over the Waelz kiln is an increase in the either be held in nitrogen purged canisters for hot charging to the EAF
quality of the ZnO product due to the lower contamination with fine melt shop or turned into hot briquetted iron (HBI). HBI can then be
iron material, due to the large pellets not abrading against each other to more readily stored for later use in a blast furnace. An important ac-
generate excess iron fines. Excessive contamination of iron is undesir- knowledgment of the typical product chemistry (Table 6) is the sulfur
able for zinc reprocessing purposes. The addition of fluxes is usually content of the DRI produced from by-product material. Often DRI
also unnecessary allowing for increased furnace productivity and a produced from recycled materials is unfit for recharging directly into
higher value DRI product. the BOF due to sulfur limitations, instead of being routed to the blast
furnace. While this does indeed recycle the ferrous material into hot

Table 5
Principal chemical reactions occurring in a rotary hearth furnace.
Process stage Reaction Equation

Gasification of carbon Devolatilization of coal coal → H2, H2O, CO, CH4

Combustion C + O2 → CO2
Boudouard reaction C + CO2 → 2CO
Water gas shift CO2 + H2 → CO + H2O
Zinc removal Carbothermal zincite reduction ZnO + CO → Zn + CO2
Zinc ferrite thermal decomposition ZnFe2O4 → ZnO + 2Fe2O3
Zinc ferrite reduction 3ZnFe2O4 + 4CO → 3Zn + 2Fe3O4 + 2CO2
Reduction of zincite by hydrogen ZnO + H2 → Zn + H2O
Iron reduction Carbothermal reduction of iron Fe2O3 + CO → 2Fe3O4 + CO2
Fe3O4 + CO → 3FeO + CO2
FeO + CO → Fe + CO2
Reduction of iron by hydrogen Fe2O3 + H2 → 2Fe3O4 + H2O
Fe3O4 + H2 → 3FeO + H2O
FeO + H2 → Fe + H2O
Iron carburization Gas carburization 3Fe + 2CO → Fe3C + CO2
Solid carburization 3Fe + C → Fe3C

9
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Fig. 10. Simplified schematic of the FASTMET® process flow.

Table 6 Ellwood City, PA (Money et al., 2000) and has been in continuous
Typical chemistry of steelmaking by-product input and DRI output of a operation since 1976 processing stainless steel wastes. This plant is
FASTMET® RHF (Chatterjee, 2012). integrated into a submerged arc furnace unit that converts the DRI
Material Fe (wt%) FeO (wt%) C (wt%) S (wt%) Zn (wt%) FeMet/Fe produced into hot metal, which is subsequently cast into ingots suitable
as feedstock for the stainless steel industry. The Ellwood city plant
BF dust 40.1 19.2 31.0 0.42 0.01 – produces 20 kt of pig iron ingots per annum, with a raw material ca-
BF Sludge 33.0 10.6 31.4 0.49 0.14 –
pacity of 50 kt per annum (Hanewald et al., 1992). Zinc is effectively
BOS Sludge 63.2 64.4 0.74 0.1 0.43 –
Fastmet DRI 70.5 – 1.13 0.35 0.004 95 % eliminated from the output material partially due to the RHF reduction
step and also due to the iron smelting step in the submerged arc fur-
nace.
metal, DRI has a lower value in use when charged to the BF than the
BOF as a scrap replacement. 5.5. DRyIron™ process
Variants of FASTMET® such as FASTMELT® incorporate a DRI
melting unit to produce hot metal directly using a submerged arc fur- The DRyIron™ process is again, very similar to FASTMET® and
nace, however these units have an increased energy demand and would INMETCO. The fundamental difference is that DRyIron™ is fed ex-
likely be uneconomical for use within an integrated works. clusively by cold briquetted material (Fig. 12) and has found com-
mercial use preparing DRI from BOS dust and blast furnace dust. Global
5.4. INMETCO process capacity for this process is currently around 1.25 mt per annum
(Seetharaman, 2013), making DRyIron™ one of the most commercially
The INMETCO (International Metals Reclamation Company, Inc.) successful rotary hearth furnace processes. This capacity is almost en-
process is very similar to FASTMET®, in that it also utilizes cold bonded tirely concentrated in the East Asian steel industry in countries such as
carbon-iron oxide material composite pellets as feedstock to a rotary South Korea, Japan, and China following the closure of plants in the
hearth furnace. Typically for INMETCO 20–25 % volatile content coal is USA in the early 2000s.
used as a reducing agent. The reason for the favorability of the DRyIron™ process for handling
The process was initially developed to processing stainless steel- of zinc bearing wastes from an integrated steel plant is that cold bri-
making wastes for Ni and Cr recovery (Espinosa et al., 2004), but its quetting processes can utilize exceptionally fine material and produce a
scope also extends to Zn recovery from EAF dust. The key point of product with consistent size and density. With regard to zinc removal
difference between FASTMET® and INMETCO is the number of pellet this is critical as a tight size and distribution of in feed material ensures
layers charged to the furnace simultaneously (Fig. 11). INMETCO uti- consistent and homogenous zinc removal, which is vital to ensure the
lizes a multiple layer system around 30 mm in depth (and accordingly recyclability of the DRI produced to the blast furnace.
has a higher furnace residence time for the material) whereas FAS- DRyIron™ also does not utilize a binder in the briquetting process.
TMET® is operated on a monolayer basis. Multiple layers of pellets can Binderless briquetting has two key advantages: reduced preparation
lead to inhomogeneity of the DRI product, with the top layers reacting costs and reduced gangue in the DRI product. Metallization levels of 95
substantially quicker than the base layer; however, the thermal load on % are achievable in short residence times of 10–15 min and residual
the furnace hearth is lower than in monolayer processes. The process is levels of P and S in the DRI can be managed through reductant selection
also highly dependent on green pellet strength as not only must the (Chatterjee, 2012; Rinker, 2001).
pellets survive the manual handling required to charge them to the Unlike the FASTMET® process, material is dried prior to the ag-
furnace, but they must also have sufficient strength to handle the glomeration step rather than after forming. This can provide ad-
weight of pellets above them as well as the rapid vaporization of the vantages with the consistency of the forming process, as steelmaking
contained moisture and volatiles within the composites. by-product dusts can be tremendously variable depending on the
The only commercial plant operating the INMETCO process is in method by which they were scrubbed from off-gas (venturi scrubbers

10
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Fig. 11. Simplified schematic of the INMETCO process flow.

versus electrostatic systems) and their storage conditions. articles writing (Lehtinen, 2003). The Hoyt Lakes plant, known as
The produced DRI from DRyIron™ is around 80 % metallized and Mesabi Nugget, was based upon the commercial principal of producing
are suitable for hot charging to an EAF, or under controlled inert a high value-added iron product at close proximity to the mining site.
cooling conditions can be subsequently used in the blast furnace de- This provides obvious logistical advantages as the iron nuggets pro-
pending on raw material selection. duced are around 50 % lighter and 90 % more compact than the
equivalent ferrous load in ore. As shown in Table 7, these iron nuggets
5.6. ITmk3 are chemically very similar to blast furnace pig iron and were suc-
cessfully used to displace scrap in electric arc furnaces (Kikuchi et al.,
ITmk3 is an evolution of the FASTMET® and FASTMELT® processes 2010).
developed by Kobe Steel and is based on the rotary hearth furnace. The The ITmk3 process was developed for use on iron ore but the RHF
fundamental difference between ITmk3 and other rotary hearth fur- reactors ability to handle material that contains high quantities of vo-
naces, is that the furnace is run between 1400–1450 °C and through latile metals such as zinc could mean it is applicable for adding value to
management of slag chemistry, rapid in-situ smelting of iron and gen- steelmaking by-products (Zhang et al., 2013). There would be several
eration of a liquid slag occurs. This is a vast departure from the purely advantages to producing pig iron nuggets versus a DRI product, the
solid-state reduction that takes place within FASTMET® and the result is gangue content in revert DRI is relatively high and as such it is typically
nuggets of gangue free pig iron and a wholly separate slag that is me- recycled to the BF rather than as a scrap replacement in the BOF. An-
chanically separable. other key advantage of utilizing the by-products of an integrated works
There has only been one commercially operating ITmk3 plant for this process is that both the carbon source and iron source are ef-
worldwide, in Hoyt Lakes, MN (USA) through a joint venture between fectively free, as BF and BOS dust respectively could be used. There are
Kobe Steel and Steel Dynamics. This 500 kt per annum plant began also CO2 emission and energy saving advantages for ITmk3, when
operation in 2010 but is under a long-term mothballing as of this compared to a small-scale blast furnace, of 38.6 % and 14.3 % per ton of

Fig. 12. Simplified schematic for the DRyIron™ process flow.

11
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Table 7 was exceptional, with the residual levels of zinc within the iron pro-
Comparison of ITmk3 pig iron nuggets (Lehtinen, 2003) versus BF pig iron. duced being below the detection limit of the analysis used. This is ob-
Material C (wt%) Si (wt%) Mn (wt%) S (wt%) P (wt%) FeTot viously very promising for the Hi-QIP processes application in mana-
ging iron bearing wastes from an integrated works, but more research
Iron nugget 2.5-4.3 0.2 0.1 0.015 0.06 Bal. must be undertaken to reduce the sulfur and phosphorous residual le-
BF pig iron 4.76 0.58 0.3 0.005 0.08 Bal.
vels within the metal as they would likely render the iron product
produced from by-product materials totally unusable as a scrap re-
placement in the BOS process, which is very sensitive to input sulfur
pig iron produced, respectively, (Atsushi et al., 2010). This saving in
levels.
energy and emissions are afforded by the total combustion in-situ of the
Unfortunately, there are other drawbacks to the Hi-QIP process re-
process gas; however, the immensely variable chemistry of by-product
lating to the non-agglomeration of the furnace burden. The levels of
materials as well as contamination with very high melting point mi-
iron in the collected zinc bearing secondary dust in the process are high
neral oxides such as Al2O3 could render an ITmk3 process unfeasible for
(6.66 wt%) which may be unacceptable to zinc smelters without further
the recovery of iron from BOS dust.
refinement. An agglomerated bed also has poor gas permeability and
The optimal size of an ITmk3 plant in terms of benefitting from
heat transfer properties, which may offset any productivity bonuses
economies of scale appears to be around 500 kt per annum. This is a
given through denser hearth loading. There is also the potential for
very large volume with regards to steelmaking dust production con-
yield losses within Hi-QIP due to the separation stage, through solution
sidering a 10 mt per annum integrated steelworks would produce ap-
loss of iron as FeO into the molten slag phase; however, this can be
proximately 200 kt total BF and BOS dust annually. This lack of supply
mitigated through increased hold times and reductant levels and iron
of ferrous material at a single site for the process could be mitigated
yields of 97 % were shown to be achievable (Ishiwata et al., 2009).
through centralizing a large ITmk3 unit in a steel production intensive
Reduction temperatures in Hi-QIP are very high (1550 °C), even
region such as China’s Hebei province where a great deal of BOS and BF
compared to the relatively high temperature ITmk3 (1450 °C) process.
dust is generated in proximity. On the other hand, western countries
This will obviously have implications on the operational costs of the
where the steel industry has declined substantially, such as the United
plant, but the exceptional zinc removal and high value ferrous product
Kingdom, has left a legacy of large stockpiles of these materials which
may still prove attractive for reintegration of zinc bearing wastes to the
were simply unable to be economically recovered. The exact quantity of
iron material cycle.
material in these stockpiles is not accurately known but is estimated to
be extensive (MacKillop, 2009). Accurate quantification of a region’s
5.8. COMET process
ferrous assets locked away in these stockpiles will be critical for as-
sessing the feasibility of any zinc separation technology.
The COMET process is another variant on the RHF process devel-
oped by the Centre for Research in Metallurgy (CRM), utilizing a non-
5.7. Hi-QIP agglomerated iron ore mixture in alternating layers with a reducing
agent and a desulfurizing agent such as limestone (Steffen and Lüngen,
The high-quality iron pebble process (Hi-QIP) is chemically 2004).
speaking very similar to ITmk3, in that it involves direct smelting of Process temperatures in the COMET process are similar to those in
iron to generate so called ‘pebbles’ of iron and a separate slag phase. INMETCO and FASTMET®, but typically residence times are higher and
The key difference, however, is in Hi-QIPs use of a non-agglomerated productivity therefore lower, in the former, due to the non-agglom-
iron feed material mixed with fluxes and a reducing agent, contained eration of the feed. Following the reduction step, DRI is formed in a
within hollows that are depressed mechanically into a carbonaceous sheet of about 10 mm in thickness (relating to the thickness of the ore
bed on the furnace. This carbon layer has four key functions; the hearth layer in the original charge), which is broken apart and screened from
is protected from direct contact with aggressive molten slags, the the other discrete layers. The non-agglomeration provides a few key
thermal load on the refractories is lessened, and it acts as an auxiliary process advantages, which may make it attractive. Firstly, the removal
reducing agent. Finally, it also determines the geometry of the iron of the pelletization step removes an element of operational cost from
pebbles through control of the size of the hollows. The primary reactor the plant as well as removing the need for a binder. Secondly, the
in the Hi-QIP is also a rotary hearth furnace, heated to around 1500 °C discrete layers allow for the inclusion of a desulfurizing agent such as
with a residence time of 15−20 min. Critically, Hi-QIP has previously limestone, which is separable from the DRI product. As a result, sulfur
been deemed suitable for the processing of BF dusts at an integrated concentration in Sidcomet DRI are usually very low, and less dependent
works (Sawa et al., 2001). on reductant selection. This is an advantage for sulfur sensitive DRI end
A pilot plant operated in Japan in the early 2000′s with reasonable uses such as BOF or EAF steelmaking. The lack of intimate contact
results and capacity of 15 tpd. The chemistry of the produced iron between the carbon and iron layers also reduces the overall level of
pebbles is shown in Table 8. They are very chemically similar to ITmk3 gangue within the direct reduced iron, as coal ash is not introduced into
pig iron nuggets as well as blast furnace pig, but the sulfur level would the material.
be a cause for concern for reintroduction to the BOF (Sawa et al., 2001). Table 10 shows typical chemical analysis for COMET DRI produced
It is likely this high sulfur level is the result of reductant selection. The from virgin ore. COMET DRI is clearly of a superior quality to other coal
coal used in the study was 0.54 wt% sulfur with the intimate contact based RHF processes in terms of sulfur and gangue content, but this
between the carbonaceous bed and liquid iron causing sulfurization of advantage comes at a productivity cost, the material in the COMET
the hot metal. process has a residence time of approximately 80 min, whereas FAS-
Table 9 shows the chemical analysis of the BF dust used in com- TMET® operates at around a 12 min residence time (Holappa and
parison to the iron pebbles produced (Sawa et al., 2001). Zn removal Kekkonen, 2000).
COMET is also suitable for processing zinc and lead bearing by-
Table 8
products and the reducing agent layer is also suitable for recycling in an
Chemical analysis of Hi-QIP iron pebbles (Sawa et al., 2001). integrated steelworks through a sinter plant. The COMET process has
yet to be scaled to commercial operation, it seems unlikely that the
Material C (wt%) Si (wt%) Mn (wt%) S (wt%) P (wt%) FeTot
benefits of lower sulfur and gangue content in the DRI product, and the
Hi-QIP Pebbles 2.1 – 3.0 0.08 0.01 0.25 0.04 Balance lack of an agglomeration stage is substantial enough to offset the re-
duction in productivity.

12
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Table 9
Chemical composition (Wt%) of the BF dust alongside iron pebbles produced at 1550 °C for 555 s (Sawa et al., 2001).
Composition (wt%)

Material C FeTotal Zn Pb Na K Cl SiO2 Al2O3 CaO MgO Mn P S

BF dust 21.3 42.9 2.54 0.64 0.11 1.2 1.13 5.67 3.25 3.12 0.37 0.26 0.07 0.91
Slag – 5.8 0.02 – – – – 34.0 18.2 34.3 3.42 0.92 – –
Iron pebble 0.93 Bal. < 0.01 < 0.01 – – – – – – – 0.02 0.13 0.72
Off-gas dust 0.68 6.66 47.1 7.02 0.22 1.95 13.2 0.20 – – – – – 2.34

Table 10 The Horsehead Resource Development Co., Inc. patented a hydro-

Typical analysis of COMET DRI (Chatterjee, 2012). carbon-fuelled cyclone melting system for treatment of high zinc dusts
Material C (wt%) SiO2 (wt%) S (wt%) FeTot (wt%) FeMet (wt%)
(Keegel, 1996) and a 30,000 t/year plant was established in Texas in
1993. The plant involved injecting fine, dried steelmaking dusts with
COMET DRI 0.2 1.86 0.032 95.8 88.1 oxygen enriched air and hydrocarbon fuel into a cyclonic reactor. Re-
duction of volatiles such as zinc and lead would occur giving a crude Zn
oxide product and a ferrous slag (HRDC, 2020).
6. Zinc recovery focused pyrometallurgical extraction processes The Ausmelt process is a two-stage furnace process, involving a
smelting vessel and a reduction vessel. Dusts and lump coal are charged
All of the above-mentioned processes have focused principally on to the furnace and enriched air and powdered reductants are inputted
the removal of zinc from an iron product to allow for the iron to be through a cooled lance. ZnO fumes out of the melt and is collected for
recovered through steelmaking. As such, zinc in every process discussed sale (Assis, 1998).
thus far is recovered as a crude ZnO, usually slightly contaminated by The iron bearing dust recovery–zinc iron plasma process (IBDR-
iron dust. This section highlights technologies designed to produce ZIPP) is another crude oxide producing zinc recovery process.
metallic Zn as a primary product. Pelletized dusts and reductants are charged into a furnace where two
Due to the much higher zinc content of EAF dust compared with electrodes generate a high voltage arc. This forms a plasma, which
BOS dust, most of this research on value generation from producing forces the reduction of the raw materials. This process is differentiated
metallic zinc took place on EAF dust; however, the data provides insight from many of the other crude oxide generating processes by the fact it
to applicability for BOS dust. Electrothermic processes based on tradi- forms molten pig iron suitable for charging directly to a BOS vessel. The
tional zinc extraction from ore are a commercial reality, however, they IBDR-ZIPP process is energy and capital intensive but produces a stable
typically consist of repurposed facilities retrofitted to handle steel- slag, a useable iron product in the form of pig iron and a saleable zinc
making dusts. Due to the low zinc content of steelmaking dusts relative oxide meaning no further by-product processing is required within the
to virgin zinc ores and the capital outlay to set up and maintain them, it steelmaking operation prior to sale.
is highly unlikely that these types of processes will expand into BOS
dust recycling. Typically, these processes are characterized by very high
process costs either due to their discontinuous nature, high electricity 7. Next generation ironmaking technology
requirements or complex design.
The Laclede Steel process effectively utilized a modified, electric arc The ULCOS (Ultra Low CO2 Steelmaking) project’s key technological
furnace to reduce and volatize Zn from EAF dust. However, what made development in collaboration with Hismelt has been the HIsarna pro-
the process innovative was the inclusion of a splash condenser to cap- cess. This ironmaking technology is a huge paradigm shift from the
ture metallic zinc from the process gas rather than allowing oxidation of blast furnace iron production route, and offers numerous potential
the Zn product (Zunkel, 1996). The benefit of this is that a lead splash advantages including dramatic CO2 reductions, high energy efficiency
condenser can co-produce lead bullion, and lead is a major contaminant and, most importantly for the discussion in this paper, high raw ma-
in EAF dust. Yet the process was pained by production problems in terial flexibility (Meijer et al., 2013).
relation to the quality of zinc produced, it was not deemed high enough Currently undergoing pilot scale testing at Tata Steel IJmuiden in
for hot dip galvanizing use internally as intended. Laclede Steel was the Netherlands, the operating concept of this technology is a cyclonic
liquidated in 2002 amidst difficult economic steelmaking conditions. reduction section located directly above a final smelting reactor. Fine
The Mintek Enviroplas process (Schoukens et al., 1993) utilizes a DC material and flux are injected in at high speeds to the top of the reactor
arc furnace to produce a metallic Zn product through the condensation where it is pre-reduced by the reactive smelter gases and descends via
of Zn vapor in a lead splash condenser, but the extremely fine nature of gravity into the final bath smelter.
the dust particles processed can cause issues with splash condenser This process is designed to alleviate the need for environmentally
efficiency (Assis, 1998). and economically expensive feed preparation for the blast furnace
A hybridized RHF/electrical furnace plant was built in Arkansas (sintering, coking) as well as take advantage of the excellent mixing and
(USA) by AllMet around 1998. This plant utilized a standard RHF thermal efficiency of the HIsmelt smelter bath technology (Dry et al.,
configuration with crude oxide produced being re-fumed in an electric 1999). The product of the HIsarna process is hot metal, chemically
furnace and condensed using a molten zinc splash condenser but was comparable to that produced via the blast furnace. Typical hot metal
found to be untenable – the splash condenser was never truly opera- chemistry of HIsarna iron from the pilot plant in the Netherlands is
tional and the quality of DRI produced was poor (Southwick, 2010). shown in Table 11.
HIsarna based reactors are not as sensitive to zinc loading as a blast

Table 11
Typical analysis of HIsarna iron (Dry, 1999).
Elemental Composition (wt%)

C S Cr P Mn V Si Ti
HIsarna Iron 3.7–4.3 0.1–0.2 0.03–0.10 0.02–0.06 0.02–0.05 0.05–0.013 0.003–0.013 0.0–0.002

13
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Table 12
Key influencing properties of BOS dust on recycling process selection.
Property Comments

FeTot (wt%) A larger proportion of iron compared to non-ferrous minerals makes recycling more attractive due to higher throughput and lower latent heat loss
during recycling of the produced DRI in the ironmaking process.
Zn (wt%) For material < 0.1 Zn wt% recycling through dilution may be feasible and for very high Zn content material (> 20 wt%) hydrometallurgical
techniques may become commercially viable.
ZnO:ZnFe2O4 The proportion of zinc present in the ferrite form is critical when selecting a recycling route, ZnFe2O4 is highly resistant to hydrometallurgical
extraction under realistic industrial conditions.
OFe (wt %) The degree of oxidation of iron present within BOS dust is critical when determining the feasibility of a pyrometallurgical Zn process. The more
metallized (and therefore the lower the value of OFe) the lower the requirement for reducing agents in the process and the shorter the processing
time. This leads to a higher Fetot of process feed as well as reducing raw material costs.
CaO:SiO2 This ratio is often referred to as the ‘basicity’ of the produced DRI and has major implications for the value in use of the product in blast furnace/
basic oxygen steelmaking production, as well as for the mechanical strength of the DRI (Chaung et al.).
Halide content. F, Cl (wt %) The overall concentration of halides in the material to be recycled has an effect on the value of the separated zinc product due to its deleterious
effect on electrolytic zinc production

furnace, and a Zn rich dust fraction is produced within the cyclonic persists as a key materials efficiency and environmental issue for steel
section of the reactor. It may therefore be feasible to reintegrate high manufacturers. BOS dust in particular presents a challenge as the zinc
zinc dusts into a HIsarna based ironmaking process to enrich the zinc content renders it unable to be effectively ‘diluted out’ through re-
content to an extent whereby they become attractive raw materials for introduction to existing BF/BOS processing but the value of the zinc
zinc recyclers. content is substantially less than in EAF dust meaning many existing
The inclusion of high zinc materials such as BOS dust, EAF dust and processes use to passivate and/or generate value from EAF dust are not
BF dust into the HIsarna furnace burden has been studied (Peters, 2019) commercially viable for the integrated works.
and appears promising. The extremely high temperatures in the cy- The key commercial drivers for an integrate steelworks with regards
clonic reactor portion of the plant combined with the reducing atmo- to processing this Zn bearing material are;
sphere means that multi-stage enrichment of zinc in HIsarna flue dust
may be able to create a dust fraction high enough in Zn for direct sale to • High zinc removal, allowing full material reintegration into existing
zinc processors. processes
• Valorization of iron units present within the material
8. Further processing of the crude zinc oxide product • High capacity and good economies of scale
The product of almost every process described in this review is The necessity of a high value iron product effectively rules out hy-
crude zinc oxide powder, which is extracted from the off-gas system of drometallurgical recovery of Zn as the resulting iron-bearing sludge
the heat treatment unit. The exceptions to this such as Enviroplas, would be very low, likely to require drying and agglomeration via a
which utilize a liquid zinc condenser system to recover zinc in the sinter plant before iron can be recovered.
metallic form. Theoretically, an Enviroplas plant would be able to sell The morphology of the dust and its chemical composition have a
recovered zinc with minimal further processing to zinc end users. large impact on the viability of recycling routes. The main influencing
However, these condensers are not currently operated on a large scale factors are described in Table 12. Although the Waelz Kiln has been
and as the product requires almost no post processing, attention should surpassed by the RHF, as the principal reactor for the removal of zinc
instead be paid to the ZnO product from RHF and Waelz Kiln based from steelmaking by-products, it is unclear which of the current gen-
processes. This zinc rich dust generated from most of the technology eration of RHF technologies will dominate in the short to medium term,
discussed here is usually referred to as a ‘Waelz Oxide’ and is recyclable FASTMET® and DRyIron™ have clear advantages and are readily com-
through two major pathways; The Imperial Smelting Furnace (ISF), mercially available and proven technologies. Production unit sizes of
which is in effect a zinc producing blast furnace (Morgan and around 200–300 kt are ideally sized for managing the BOS dust output
Greenwood, 1968), and electrolytic zinc production which is by far the from a single plant and can easily feed DRI back into the blast furnace
dominant production pathway with around 90 % of zinc produced or basic oxygen steelmaking process Table 13.
globally being produced electrolytically (Antuñano et al., 2019). It seems unlikely that newer steel plants globally will be allowed to
The primary concern with regard to Waelz oxides recycling into an operate without a recovery pathway for zinc bearing wastes generated
electrolytic process comes from contamination of the oxide with ha- through the process, as legislative pressure to reduce environmental
lides, such as Cl and F. Halide contamination can cause rapid dete- impact and rising landfilling costs render previous disposal routes un-
rioration of electrodes and introduce impurities into the final metallic economical. The fact that RHF technology can utilize other waste
zinc during processing (Wu et al., 2014). Several processes have been products such as the carbon bearing BF dust is another advantage of
developed that are capable of dealing with the high levels of halide these reactors, displacing coals with a by-product and thus eliminating
impurities found in Waelz Oxide including the Modified Zincex® Pro- BF dusts need for landfilling simultaneously. As high-quality zinc ore
cess (Diaz and Martin, 1994) and Zinclor process (Nogueira and deposits become scarcer moving through the 21st century the value of
Cosmen, 1983). The value of the zinc oxide produced must be factored recovered zinc oxides will likely rise, which may make in-situ recovery
into determining the economics of separating zinc from BOS dust, one at a steel plant a more economically attractive option.
key advantage that the RHF has over the Waelz kiln is that without the The RHF appears set to be the best available technology for the first
tumbling action of the burden inside a Waelz kiln, mechanical ejection half of the 21st century, many different variations on this technology
of iron oxide material into the off-gas system is limited and hence the have achieved commercial success and their ubiquity, the high value in
value of the Waelz oxide produced is greater. use of the produced Direct Reduced Iron and lower technical com-
plexity than the obsolete Waelz Kiln make them very attractive for an
integrated steel plant.
9. Conclusions
Zinc focused recovery methods such as the Enviroplas process are
unlikely to see widespread use in the short to medium term and are
The reintegration of zinc bearing by-products of steelmaking

14
 Table 13
Summary of advantages and disadvantages of various zinc removal processes.
Process Products Process Description Advantages Disadvantages
D.J.C. Stewart and A.R. Barron

Waelz Kiln Waelz slag, ZnO Horizontal rotating cylinder fed by granular feed and fired with natural Well established and proven technology. Low productivity
gas/coal. 6+ hours at > 1000 °C
No need for pelletization of in-feed Low metallization
High Fe content of ZnO produced
High maintenance costs
FASTMET DRI, ZnO RHF fed with pelletized iron oxides and carbon source. 1300 °C for Most established commercial RHF process DRI has a higher gangue content than INMETCO
6–12 min etc.
High throughput
DRI useful in blast furnace
DryIRon DRI, ZnO RHF fed with binderless briquettes. 1300 °C for ∼15 min No need for expensive binder. No binder means unfired briquettes are
extremely fragile
High metallization % Not as established as Fastmet
ITmk3 Pig Iron Nuggets RHF fed with pellets, 1450 °C for 10–20 min Complete metallization and slag separation High refractory and energy operational
expenditure
Rapid reaction speeds Untested for Zn recovery from by-products
Extremely high value product Only one commercial plant that no longer
operates
INMETCO Direct Reduced Iron, ZnO Pelletized feed charged in 3 layers to an RHF. 1300 °C for > 60 min. Can Proven technology, industrial plant operating Long residence time and low throughput
be supplemented with a submerged arc furnace melter
Supplementing with melter unit can increase thermal Layered structure leads to non-homogeneity of

15
efficiency by utilizing latent heat of process DRI DRI
Melter units are energy intensive
COMET Direct reduced iron, ZnO Non-agglomerated feed charged to an RHF and fired for > 60 min at Addition of limestone layer can give very low sulfur DRI Low productivity per sq metre of hearth
1300 °C. Can be supplemented with a submerged arc furnace melter
No pelletization required. Limited DRI strength
Can be fueled by coke oven gas. Melter units are energy intensive
Can produce hot metal with addition of submerged arc
furnace.
Hi-QIP Pig Iron Nuggets, ZnO Powder feed charged onto a carbon bed within an RHF fired to 1500 °C for High value pig iron product Very high temperature requires expensive
10–20 min. Total separation of Fe and slag occurs refractories and high energy costs
Rapid reaction speed Not commercially implemented
Zinc removal has been previously explored using Hi-QIP Requires auxiliary carbon source
Laclede Steel Process Hot metal, metallic Zn, A modified EAF with a zinc/lead splash condenser Recovery of both zinc and lead provides high value Very high capital investment and operational
lead bullion products costs
More suited for EAF dust processing
Enviroplas Stabilized slag, metallic DC arc furnace with splash condenser High value metallic zinc product Low value ferrous product
Zn
More suitable for passivated hazardous EAF
dusts than BOS dust
Ausmelt Stabilised slag, ZnO Two stage smelting reduction reactor Small process footprint Low value ferrous product.
Not electrically powered Only economical for high zinc wastes
IBDR-ZIPP ZnO, hot metal, slag Agglomerated feed of material to a plasma heated reduction smelting High value products which can be used within the steel Very high expenses and capital cost
reactor plant
Only economical for high zinc wastes
Resources, Conservation & Recycling 157 (2020) 104746
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

much more suitable for processing EAF dust where the high zinc con- Metallurgiccal Society of CIM, Calgary, Canada August 1998. Paper 8423.
tent can be more effectively valorized. This may change in the far future Atsushi, M., Uemura, H., Sakaguchi, T., 2010. MIDREX processes. Kobelco Technol. Rev.
29 (12), 50–57.
however as cheap renewable energy becomes more readily available, Bragg, W.H., 1915. The structure of magnetite and the spinels. Nature 95 (7). https://doi.
technology that utilizes electrical energy such as IBDR-ZIPP and org/10.1038/095561a0. 561–561.
Enviroplas could become feasible if the value of the recovered zinc can Brandt, D.A., Warner, J.C., 2005. Metallurgy Fundamentals. Goodheart-Wilcox. Tinley
Park, IL ISBN:1605250805.
offset their high capital requirements and energy consumption. Cantarino, M.V., De Carvalho Filho, C., Borges Mansur, M., 2012. Selective removal of
High value-added technologies such as Hi-QIP and ITmk3 have not zinc from basic oxygen furnace sludges. Hydrometallurgy 111–112 (1), 124–128.
seen commercial use in recovery of volatile metals from by-product https://doi.org/10.1016/j.hydromet.2011.11.004.
Chatterjee, A., 2012. Sponge Iron Production by Direct Reduction of Iron Oxide, 2nd ed.
material but may in future, depending on economic conditions. Hi-QIP PHI Private Learning Limited, Delhi, India ASIN:B00K7BMK1Y.
and ITmk3 are less suitable for smaller scale application such as Daigo, I., Osako, S., Adachi, Y., Matsuno, Y., 2014. Time-series analysis of global zinc
managing the dust produced at a single integrated works but may po- demand associated with steel. Resour. Conserv. Recy. 82 (1), 35–40. https://doi.org/
10.1016/j.resconrec.2013.10.013.
tentially be viable for larger scale remediations of post-industrial
Diaz, G., Martin, D., 1994. Modified Zincex process: the clean, safe and profitable solution
countries where large legacy stockpiles of materials exist and require to the zinc secondaries treatment. Resour. Conserv. Recycl. 10 (1-2), 43–57. https://
remediation, such as the United Kingdom. doi.org/10.1016/0921-3449(94)90037-X.
In terms of removal of zinc and in the metallization of the product, Dry, R.J., Bates, C.P., Price, D.P., 1999. HIsmelt – the future in direct ironmaking.
Ironmaking Conference 58, 361–366 ISBN: 1-886362-32-7.
the molten stage processes such as Hi-QIP, ITmk3 etc. are outstanding, Dutra, A.J.B., Paiva, P.R.P., Tavares, L.M., 2006. Alkaline leaching of zinc from electric
able to reduce zinc content in the ferrous output to trace levels and arc furnace steel dust. Miner. Eng. 19 (5), 478–485. https://doi.org/10.1016/j.
obtaining complete metallization of the residual iron. The drawback mineng.2005.08.013.
Espinosa, D.C.R., Bernardes, A.M., Alberto, J., Tenório, S., 2004. An overview on the
however is the increased reaction temperatures required to drive the current processes for the recycling of batteries. J. Power Sources 135 (1-2), 311–319.
carburization and melting reactions and the operational costs that they https://doi.org/10.1016/j.jpowsour.2004.03.083.
incur. Fisher, L.V., Barron, A.R., 2019. The recycling and reuse of steelmaking slags – a review.
Resour. Conserv. Recycl. 146 (7), 244–255. https://doi.org/10.1016/j.resconrec.
It is to be expected that the price of zinc will rise as global high- 2019.03.010.
quality virgin ore reserves dwindle and demand for galvanized steel Fruehan, R.J., 2005. New steelmaking processes: drivers, requirements and potential
increases and thus the processing of lower zinc ferrous by-products will impact. Ironmak. Steelmak. 32 (1), 3–8. https://doi.org/10.1179/irs.2005.32.1.3.
Gargul, K., Boryczko, B., 2015. Removal of zinc from dusts and sludges from basic oxygen
become more economically favorable as it has with the higher zinc EAF furnaces in the process of ammoniacal leaching. Arch. Civ. Mech. Eng. 15 (1),
dusts. As such, the lower boundary of Zn wt% in dusts that can be 179–187. https://doi.org/10.1016/j.acme.2014.08.004.
economically processed is likely to fall in time. Legislative pressure may Gargul, K., Jarosz, P., Małecki, S., 2016. Alkaline leaching of low zinc content iron-
bearing sludges. Arch. Metall. Mater. 61 (1), 43–50. https://doi.org/10.1515/amm-
also have an impact, as landfilling fees are also likely to rise in devel-
2016-0013.
oped countries, making recycling more commercially attractive. For the Gielen, D., Moriguchi, Y., 2002. CO2 in the iron and steel industry: an analysis of Japanese
iron producer production of a reduced iron product from these by- emission reduction potentials. Energy Policy 30 (10), 849–863. https://doi.org/10.
products also has the economic incentive of displacing expensive re- 1016/S0301-4215(01)00143-4.
Goetz, F., 1980. The Mechanism of B.O.F. Fume Formation. Thesis, MEng. McMaster
ducing agents and iron units in further iron production. Ultimately, the University.
next generation of ironmaking technology, processes such as HISarna Gritzan, X.X., Neuschütz, D., 2001. Rates and mechanisms of dust generation in oxygen
will need to alleviate the tight raw material requirements of the blast steelmaking. Steel Res. 72 (9), 324–330. https://doi.org/10.1002/srin.200100126.
Hahn, I., Neuschütz, D., 2002. Ejection of steel and slag droplets from gas stirred steel
furnace and allow for high Zn materials to be processed without pre- melts. Ironmak. Steelmak. 29 (3), 219–223. https://doi.org/10.1179/
treatment. The steel industry must overcome these barriers to the cir- 030192302225004115.
cular usage of galvanized steel if it is to become truly sustainable. Hanewald, R.H., Munson, W.A., Schweyer, D.L., 1992. Processing steel wastes pyr-
ometallurgically at INMETCO. Min. Metall. Explor. 9 (4), 169–173. https://doi.org/
10.1007/BF03403430.
Declaration of Competing Interest Hay, S.M., Rankin, W.J., 1994. Recovery of iron and zinc from blast furnace and basic
oxygen furnace dusts: a thermodynamic evaluation. Miner. Eng. 7 (8), 985–1001.
https://doi.org/10.1016/0892-6875(94)90028-0.
The authors declare that they have no known competing financial
Heinrich, T., 2015. Thesis, PhD. Cardiff University.
interests or personal relationships that could have appeared to influ- Henschen, H.C., 1968. Wet vs dry gas cleaning in the steel industry. J. Air Pollut. Control
ence the work reported in this paper. Assoc. 18 (5), 338–342. https://doi.org/10.1080/00022470.1968.10469138.
Holappa, L.E., Kekkonen, M., 2000. Comparison of different coal based direct reduction
processes. Technical Report, Helsinki Univ. of Technology, Otaniemi (Finland). Mat.
Acknowledgments Sci. Metall ISBN:951-22-5134-5.
Hoskins, B.J., Draghici, I., Davies, H.C., 1966. A new look at the cause of fuming. JOM 18
Financial support was provided by Materials and Manufacturing (7), 803–810. https://doi.org/10.1007/BF03378469.
HRDC, Horsehead Resource Development Co., Inc. Flame Reactor Technology
Academy (M2A) that has been made possible through funding from the Description. https://clu-in.org/products/site/complete/democomp/horseres.htm.
European Social Fund via the Welsh Government, the Engineering and (accessed 24 September 2019).
Physical Sciences Research Council (EPSRC), and Tata Steel. Additional Ishiwata, N., Sawa, Y., Hiroha, H., Matsui, T., Murao, A., Higuchi, T., Takeda, K., 2009.
Investigation of reduction and smelting mechanism in the Hi-QIP process. Steel Res.
support is provided by the Reducing Industrial Carbon Emissions (RICE) Int. 80 (8), 523–529. https://doi.org/10.2374/SRI09SP040.
operations funded by the Welsh European Funding Office (WEFO) Jaafar, I., 2014. Chlorination for the Removal of Zinc From Basic Oxygen Steelmaking
through the Welsh Government. (BOS) By-product, Thesis, PhD. Cardiff University.
J.F. Keegel, Jr., 1996. Methods for recycling electric arc furnace dust. US Patent 5,538,
532.
References Kelebek, S., Yörük, S., Davis, B., 2004. Characterization of basic oxygen furnace dust and
zinc removal by acid leaching. Miner. Eng. 17 (2), 285–291. https://doi.org/10.
1016/j.mineng.2003.10.030.
Ahmed, H., 2018. New trends in the application of carbon-bearing materials in blast
Kikuchi, S., Ito, S., Kobayashi, I., Tsuge, O., Tokuda, O., 2010. Midrex Technologies,
furnace iron-making. Minerals 8 (12), 561. https://doi.org/10.3390/min8120561.
ITmk3 process. Kobelco Technol. Rev. 29, 69–77.
Allwood, J., Cullen, J.M., Milford, R.L., 2010. Options for achieving a 50% cut in in-
Kim, Y., Worrell, E., 2002. International comparison of CO2 emission trends in the iron
dustrial carbon emissions by 2050. Environ. Sci. Technol. 44 (6), 1888–1894. https://
and steel industry. Energy Policy 30 (10), 827–838. https://doi.org/10.1016/S0301-
doi.org/10.1021/es902909k.
4215(01)00130-6.
Anameric, B., Kawatra, K., 2007. Properties and features of direct reduced iron. Min.
Kobelco, Fastmet® Process. http://www.kobelco.co.jp/english/products/fastmet/. (ac-
Proc. Ext. Met. Rev. 28 (1), 59–116. https://doi.org/10.1080/08827500600835576.
cessed 20 May 2019).
Antuñano, N., Cambra, J.F., Arias, P.L., 2019. Hydrometallurgical processes for Waelz
Kohlmeyer, E., Spandau, H., 1944. Ueber die Vergasung Von Eisen. Arch. Eisenhüttenwes
oxide valorisaton – an overview. Process. Saf. Environ. Prot. 129, 308–320. https://
18 (1), 1–6. https://doi.org/10.1002/srin.194401336.
doi.org/10.1016/j.psep.2019.06.028.
Krishnamurthy, K.R., Gopalakrishnan, J., Aravamudan, G., Sastri, M.V.C., 1974. Studies
Assis, G., 1998. Emerging Pyrometallurgical Processes for Zinc and Lead Recovery From
on the formation of zinc ferrite. J. Inorg. Nucl. Chem. 36 (3), 569–573. https://doi.
Zinc-bearing Waste Materials. Zinc and Lead Processing Symposium. The
org/10.1016/0022-1902(74)80115-6.

16
D.J.C. Stewart and A.R. Barron Resources, Conservation & Recycling 157 (2020) 104746

Kuzmann, E., Nagy, S., Vértes, A., 2003. Critical review of analytical applications of doi.org/10.1080/10473289.1997.10463929.
Mössbauer spectroscopy illustrated by mineralogical and geological examples (IUPAC Rinker, F.G., 2001. Using the DRylron™ process to produce value-added iron units. Miner.
Technical Report). Pure Appl. Chem. 75 (6), 801–858. https://doi.org/10.1351/ Metall. Process. 18 (1), 45–48. https://doi.org/10.1007/BF03402870.
pac200375060801. Rose, J., Cortalezzi-Fidalgo, M.M., Moustier, S., Magnetto, C., Jones, C.D., Barron, A.R.,
Lee, J., Lin, C., Chen, H., 2001. Carbothermal reduction of zinc ferrite. Metall. Mat. Trans Wiesner, M.R., Bottero, J.-Y., 2002. Synthesis and characterization of carboxylate-
B 32, 1033–1040. https://doi.org/10.1007/s11663-001-0092-9. FeOOH nanoparticles (ferroxanes) and ferroxane-derived ceramics. Chem. Mater. 14
Lehtinen, L., 2003. The Mesabi Nugget Project: New Ironmaking Technology of the (2), 621–628. https://doi.org/10.1021/cm010583r.
Future. Direct from MIDREX Q2 3–8. . Salama, W., El Aref, M., Gaupp, R., 2015. Spectroscopic characterization of iron ores
Lin, B., Wang, X., 2015. Carbon emissions from energy intensive industry in China: evi- formed in different geological environments using FTIR, XPS, Mössbauer spectro-
dence from the iron & steel industry. Renew. Sust. Energ. Rev. 47 (7), 746–754. scopy and thermoanalyses. Spectrochim Acta. Part A Mol. Biomol. Spectrosc 136
https://doi.org/10.1016/j.rser.2015.03.056. (Part C), 1816–1826. https://doi.org/10.1016/j.saa.2014.10.090.
Lin, X., Peng, Z., Yan, J., Li, Z., Hwang, J.-Y., Zhang, Y., Li, G., Jiang, T., 2017. Sammut, M.L., Rose, J., Masion, A., Fiani, E., Depoux, M., Ziebel, A., Hazemann, J.L.,
Pyrometallurgical recycling of electric arc furnace dust. J. Clean. Prod. 149 (4), Proux, O., Borschneck, D., Noack, Y., 2008. Determination of zinc speciation in basic
1079–1100. https://doi.org/10.1016/j.jclepro.2017.02.128. oxygen furnace flying dust by chemical extractions and X-ray spectroscopy.
Liu, S., Loper Jr., C.R., 1991. The formation of kish graphite. Carbon 29 (4-5), 547–555. Chemosphere 70 (11), 1945–1951. https://doi.org/10.1016/j.chemosphere.2007.09.
https://doi.org/10.1016/0008-6223(91)90119-4. 063.
Lobato, N.C.C., Villegas, E.A., Mansur, M.B., 2015. Management of solid wastes from Santos, J., Phillips, J., Dumesic, J.A., 1983. Metal-support interactions between iron and
steelmaking and galvanizing processes: a brief review. Resour. Conserv. Recy. 102 titania for catalysts prepared by thermal decomposition of iron pentacarbonyl and by
(9), 49–57. https://doi.org/10.1016/j.resconrec.2015.05.025. impregnation. J. Catal. 81 (1), 147–167. https://doi.org/10.1016/0021-9517(83)
Lu, L., Pan, J., Zhu, D., 2015. Iron Ore, Chapter 16. Quality Requirements of Iron Ore for 90154-9.
Iron Production. Elsevier ISBN: 9781782421566. Sawa, Y., Yamamoto, T., Takeda, K., Itaya, H., 2001. New coal based process to produce
Ma, N., 2016. Recycling of basic oxygen furnace steelmaking dust by in-process separa- high quality DRI for the EAF. ISIJ Int. 41, 17–21. https://doi.org/10.2355/
tion of zinc from the dust. J. Clean. Prod. 112 (5), 4497–4504. https://doi.org/10. isijinternational.41.Suppl_S17.
1016/j.jclepro.2015.07.009. Schoukens, A.F.S., Shaw, F., Chemaly, E.C., 1993. The Enviroplas process for the treat-
MacKillop, F., 2009. The construction of ‘waste’ in the UK steel industry. J. Environ. Plan. ment of steel-plant dusts. J. South. Afr. Inst. Min. Metall. 93 (1), 1–7.
Manag. 52 (2), 177–194. https://doi.org/10.1080/09640560802666529. Seetharaman, S., 2013. In: Seetharaman, S., McLean, A., Guthrie, R., Sridhar, S. (Eds.),
Mager, K., Meurer, U., Wirling, K., 2003. Minimizing dioxin and furan emissions during Treatise on Process Metallurgy, Vol. 3: Industrial Processes, Part A. Elsevier, Oxford.
zinc dust recycle by the Waelz process. JOM 55 (8), 20–25. https://doi.org/10.1007/ Shawabkeh, R.A., 2014. Hydrometallurgical extraction of zinc from Jordanian electric arc
s11837-003-0099-6. furnace dust. Hydrometallurgy 104 (1), 61–65. https://doi.org/10.1016/j.hydromet.
Mager, K., Meurer, U., Garcia-Egocheaga, B., Goicoechea, N., Rutten, J., Saage, W., 2010.04.014.
Simonetti, F., 2013. Recovery of zinc oxide from secondary raw materials: new de- Singh, R., 2012. Applied Welding Engineering: Processes, Codes and Standards. Chapter
velopments of the waelz process. In: Stewart, D.L., Daley, J.C., Stephens, R.L. (Eds.), 5. Production of Steel. Elsevierhttps://doi.org/10.1016/B978-0-12-391916-8.
In Recycling of Metals and Engineered Materials, https://doi.org/10.1002/ 00005-4.
9781118788073.ch29. Southwick, L.M., 2010. Still no simple solution to processing EAF dust. Steel Times Int.
Masson, N., Briol, P., 1969. A Brief Summary of Zinc Oxide Processing Methods Available 43–45. Available from: www.steeltimesint.com/contentimages/features/
for the Bongará Deposit. Zinc One Resources Inc. EAFdustMar10.pdf. Accessed 24 September 2019.
Meijer, K., Zeilstra, C., Teerbuis, C., Ouwehand, M., van der Stel, J., 2013. Developments Steer, J., Griffiths, A., Heinrich, T., Thomas, A., Barnes, C., 2014. Correlation of BOS
in alternative ironmaking. Trans. Indian Inst. Met. 66 (5-6), 475–481. https://doi. process variables with dust mass formation and zinc content. Ironmak. Steelmak. 41
org/10.1007/s12666-013-0309-z. (7), 554–560. https://doi.org/10.1179/1743281213Y.0000000152.
Mikhail, S.A., Turcotte, A.-M., 1998. Thermal reduction of steel-making secondary ma- Steffen, R., Lüngen, H.-B., 2004. State of the art technology of direct and smelting-re-
terials: I. Basic-oxygen-furnace dust. Thermochim. Acta. 311 (1-2), 113–119. https:// duction of iron ores. Metall. Res. Technol. 101 (3), 171–182. https://doi.org/10.
doi.org/10.1016/S0040-6031(97)00430-9. 1051/metal:2004131.
Money, K.L., Hanewald, R.H., Bleakney, R.R., 2000. Processing steel waste pyr- Stephenson, R.L., Smailer, R.M., 1999. Direct Reduced Iron: Technology and Economics
ometallurgically at INMETCO. Recycling of metals and engineered materials. In: of Production and Use. Iron & Steel Society, London, UK ISBN:1886362319.
Steart, D.L., Stephens, R., Daley, J.C. (Eds.), The Minerals, Metals & Materials Society. Sun, X., Hwang, J., Huang, X., 2008. The microwave processing of electric arc furnace
Warrendale, PA, pp. 397–408 ISBN:0-87339-49-1. dust. JOM 60, 35–39. https://doi.org/10.1007/s11837-008-0132-x.
Morgan, S.W.K., Greenwood, D.A., 1968. The metallurgical and economic behavior of Tsujino, R., Hirai, M., Ohno, T., Ishiwata, N., Inoshita, T., 1989. Mechanism of dust
lead in the imperial smelting furnace. JOM 20 (12), 31–35. https://doi.org/10.1007/ generation in a converter with minimum slag. ISIJ Int. 29 (4), 291–299. https://doi.
BF03378767. org/10.2355/isijinternational.29.291.
Narita, K., Onoye, T., Satoh, Y., Miyamoto, M., Taniguchi, K., Kamatani, S., Sato, T., Turkdogan, E., Leake, L., 1959. Preliminary studies on the evolution of fumes from iron at
Fukihara, S., 1981. Effects of alkalis and zinc on the wear of blast furnace refractories high temperatures. J. Iron Steel Inst. 192, 162–170.
and the tuyere displacement. Trans. Iron Steel Inst. Japan 21 (12), 839–845. https:// Vereš, J., Šepelák, V., Hredzák, S., 2015. Chemical, mineralogical and morphological
doi.org/10.2355/isijinternational1966.21.839. characterisation of basic oxygen furnace dust. Miner. Process. Extr. Metall. 124 (1),
Nedar, L., 1996. Dust formation in a BOF converter. Steel Res. Int. 67 (8), 320–327. 1–8. https://doi.org/10.1179/1743285514Y.0000000069.
https://doi.org/10.1002/srin.199605497. Walburga Keglevich de Buzin, P.J., Heck, N., Vilela, C., A, C, F, 2017. EAF dust: an
Nogueira, E.D., Cosmen, P., 1983. The zinclor process: simultaneous production of zinc overview on the influences of physical, chemical and mineral features in its recycling
and chlorine. Zinc’ 83. In: Proceedings 13th Annual Hydrometallurgy Conference. and waste incorporation routes. J. Mater. Res. Technol. 6 (2), 194–202. https://doi.
Canadian Institute of Mining, Metallurgy and Petroleum, Montreal. pp. 1–22. org/10.1016/j.jmrt.2016.10.002.
Okhotskii, V.B., 2007. Formation of iron oxide dust in steel smelting. Steel Transl. 37 World Steel Association, 2017a. World Steel in Figures 2017.
(12), 979–981. https://doi.org/10.3103/S0967091207120017. World Steel Association, 2017b. Sustainable Steel - Indicators 2017 and the Future.
Onnen, J.H., 1972. Wet scrubbers tackle pollution. Environ. Sci. Technol. 6 (12), Available from www.worldsteel.org. Accessed 8 May 2019.
994–998. https://doi.org/10.1021/es60071a007. Worrell, E., Price, L., Martin, N., 2001. Energy efficiency and carbon dioxide emissions
Orbaek, A., Barron, A.R., 2015. ICP-AES analysis of nanoparticles. OpenStax CNX 24 reduction opportunities in the US iron and steel sector. Energy 26 (5), 513–536.
(March). http://cnx.org/contents/fb3818e2-cbcb-4582-bb7c-3a467a716cc0@21. https://doi.org/10.1016/S0360-5442(01)00017-2.
Oustadakis, P., Tsakiridis, P.E., Katsiapi, A., Agatzini-Leonardou, S., 2010. Wu, X., Liu, Z., Liu, X.L., 2014. The effects of additives on the electrowinning of zinc from
Hydrometallurgical process for zinc recovery from electric arc furnace dust (EAFD). sulphate solutions with high fluoride concentration. Hydrometallurgy 141, 31–35.
Part I: characterization and leaching by diluted sulphuric acid. J. Hazard. Mater. 179 https://doi.org/10.1016/j.hydromet.2013.09.007.
(1-3), 1–7. https://doi.org/10.1016/j.jhazmat.2010.01.059. Zboril, R., Mashlan, M., Petridis, D., 2002. Iron(III) oxides from thermal processes-
Peters, A.G.A., 2019. Zinc Vapourisation From Sludge Wastes Under Thermal Processing synthesis, structural and magnetic properties, Mössbauer spectroscopy characteriza-
Conditions – Zinc Enrichment in HIsarna Flue Dust, Thesis MSc. Delft University of tion, and applications. Chem. Mater. 14 (3), 969–982. https://doi.org/10.1021/
Technology. cm0111074.
Porter, F., 1991. Zinc Handbook: Properties, Processing, and Use in Design. International Zhang, Y.Y., Qi, Y.H., Zou, Z.S., Li, Y.G., 2013. Development prospect of rotary hearth
Lead Zinc Research Organization ISBN, 0824783409. furnace process in China. Adv. Mater. Res. 746, 533–538. https://doi.org/10.4028/
Ray, S.K., Chattopadhyay, G., Ray, A.K., 1997. Evaluation of dust generated from basic www.scientific.net/AMR.746.533.
oxygen furnace steel making. J. Air Waste Manag. Assoc. 47 (6), 716–721. https:// Zunkel, D., 1996. What to do with your EAF Dust. Steel Times Int. 20 (3), 46–50.

17