Minerals Engineering 164 (2021) 106796

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Treatment of high-arsenic copper smelting flue dust with high copper

sulfate: Arsenic separation by low temperature roasting
Yujie Chen a, b, Shun Zhu d, Pekka Taskinen b, Ning Peng a, c, Bing Peng a, c, Ari Jokilaakso b,
Hui Liu a, c, *, Yanjie Liang a, c, *, Zongwen Zhao a, Zhongbing Wang e
a
School of Metallurgy and Environment, Central South University, Changsha, Hunan 410083, China
b
Department of Chemical and Metallurgical Engineering, School of Chemical Engineering, Aalto University, P.O. Box 16100, 00076 Aalto, Finland
c
Chinese National Engineering Research Center for Control & Treatment of Heavy Metal Pollution, Changsha, Hunan 410083, China
d
Sinosteel Tiancheng Environmental Protection Science & Technology CO., Ltd., Wuhan, Hubei 430223, China
e
School of Environment and Chemistry, Nanchang Hangkong University, Nanchang, Jiangxi 330063, China

A R T I C L E I N F O A B S T R A C T

Keywords: Arsenic is the most important hazardous trace element in copper concentrates. In order to separate arsenic from
Copper smelting the smelting processes, a pyrometallurgical treatment with additives was developed to deal with high-arsenic
Arsenic circulation copper smelter flue dust high in copper sulfate. Under the optimized conditions of mixing 30 wt% pyrite and
Pyrite
roasting at 550 ◦ C under an inert atmosphere for 30 min, more than 92% arsenic was volatilized as a high purity
Pyrometallurgy
Vaporization
arsenic trioxide by-product while most copper was retained in the calcine to be returned to smelting as a high-
In situ FT-IR copper feed. The high separation of arsenic is due to the transformation from the limited vaporization of arsenate
to volatile arsenic trioxide by pyrite at above 460 ◦ C. Two reaction paths of pyrite were identified as follows: one
followed the pattern FeS2 → Fe2O3 → FeAsO4 → Fe3O4/Fe2O3 and the other was FeS2 → FeS + S2 (g) → Fe3O4/
Fe2O3.

1. Introduction Chen et al. 2013) regarding the distribution of arsenic in pyrometal­

lurgical copper processing indicates that most arsenic transfers to the gas
Arsenic as a metalloid element can cause a great deal of detrimental phase in the smelting step and forms high arsenic flue dusts.
health effects to humans after acute and chronic exposure, e.g., hepatic, In industrial operations, the flue dusts are usually recirculated to the
cardiovascular, hematological, renal, pulmonary, genotoxic, respiratory feed-mixture, giving rise to arsenic enrichment in the flue dusts by
and carcinogenic effects (Mandal and Suzuki, 2002; Bowell et al., 2014; several orders of magnitude (Montenegro et al., 2013). When the
Wang et al., 2017). It is well known that copper smelting is one of the accumulation in some smelters reaches a level affecting the quality of
major anthropogenic sources of arsenic pollution (Monique and Frim­ anode copper, the circulation is discontinued and the dust treated
mel, 2013). separately with an arsenic bleed (Chen et al., 2010).
About 85% of copper in the world is smelted by pyrometallurgical As arsenic coexists with valuable metals (zinc, copper, lead etc.) in
processing (Liao et al., 2014), in which most of the arsenic enters the the metallurgical process, various investigations have studied flue dust
smelters in copper concentrate and is discharged into the environment in treatment including hydrometallurgical (Bakhtiatri et al., 2008; Xu
gas, liquid, and solid forms (Montenegro et al., 2013). This trend will be et al., 2010; Morales et al., 2010; Hou et al., 2011; Li et al., 2014;
unchanged in the coming years. Arsenic has significant implications in Alguacil et al., 2015; Lucheva et al., 2017; Gonzalez-Moreno et al.,
copper making, as it seriously affects the properties of copper products 2018) and pyrometallurgical processes (Li et al., 2018; Tan et al., 2018a)
and attracts increasing public awareness of environmental protection, according to differences in the water solubility and volatility of the el­
due to its volatility, toxicity, bioaccumulation and potential carcino­ ements to be recycled. Most studies were focused on the extraction of
genic properties (Liu et al., 2017; Sohn et al., 2004). Abundant literature one or two valuable metals by hydrometallurgical methods, after which
using thermodynamic modeling (Itagaki and Yazawa, 1982; Swinbourne the leaching residues were sent back to smelting. In these studies of
and Kho, 2012; Wang et al., 2017) and plant analyses (Kaur et al., 2010; hydrometallurgical treatment, less attention was paid to the recovery of

* Corresponding authors.
E-mail addresses: liuhuicsu@163.com (H. Liu), csulyj555@163.com (Y. Liang).

https://doi.org/10.1016/j.mineng.2021.106796
Received 29 July 2020; Received in revised form 25 September 2020; Accepted 9 January 2021
Available online 18 February 2021
0892-6875/© 2021 Published by Elsevier Ltd.
Y. Chen et al. Minerals Engineering 164 (2021) 106796

arsenic. This means that the arsenic in the flue dusts continues to be
circulated in the smelting processes. Chen et al. (Chen et al., 2012) used
NaCl-H2SO4 solution, NH3⋅H2O solution, NaOH solution, and HNO3-
NaCl solution to recover bismuth and arsenic from copper smelter flue
dusts after copper and zinc extraction. It has been shown that the
chemical nature of arsenic in a hydrometallurgical process prevents
arsenic removal in a one-step treatment (Xu et al., 2010; Yang et al.,
2017a); thus it will require an increased number of treatment steps and
produce other side effects. Therefore, removing arsenic from copper flue
dusts by hydrometallurgy is not an appropriate method.
In some papers (Tang et al., 2014; Zhong et al., 2017; Tan et al.,
2018a; Tan et al., 2018b; Li et al., 2016; Li et al., 2018), pyrometallur­
gical processes for removing arsenic from high arsenic flue dusts were
studied at low temperatures. In these studies, arsenic was mainly in the
form of As2O3 without sulfates in the flue dusts. According to the cal­
culations using HSC software, version 9.4.1 (Roine, 2018), gaseous Fig. 1. Flow chart and results of the sequential extraction of arsenic from
As2O3 can react with CuSO4 at temperatures above 230 ◦ C forming the ESPD.
copper arsenates that prevent arsenic volatilization. Liang et al. (Liang
et al., 2011) investigated the effect of temperature, roasting time, and with an X-ray fluorescence spectrometer 1800, XRF (Jindao, Japan), the
coke additive on the de-arsenification of flash smelting dusts using iron, sulfur, silicon, and oxygen contents were 43 wt%, 48 wt%, 5 wt%,
orthogonal experiments. The arsenic in the dusts mainly existed in the and 4 wt%, respectively. FeS2 was the major phase in the concentrate
form of arsenides and arsenates. The results showed that the de- and SiO2 the minor phase, as shown in Fig. S-1 (supplementary
arsenification yield was above 80% and the recovery of copper above material).
95%, under conditions of 1100 ◦ C roasting for 1 h with 12% coke ad­ The Na2S (Sinopharm Chemical Reagent Co., Ltd, China), C (Sino­
ditive. Therefore, the pyrometallurgical treatment of arsenic-containing pharm Chemical Reagent Co., Ltd, China), PbS (Shanghai Wokai
flue dusts needs to be formulated based on the characteristics of the Chemical Reagent Co., Ltd., China), and S (Tianjin Shentai Chemical
dusts. At present, there is no report on an arsenic removal method from Reagent Co., Ltd., China) used in this study were of analytical grade. The
copper smelting dust that contains high levels of arsenic and copper FeS used in this study was supplied by Alfa Aesar with a purity above
sulfate. 85%.
This work describes a novel and simple method for treating high-
arsenic copper smelter flue dusts with high copper sulfate by selective 2.2. Experimental approach
separation of arsenic through roasting with pyrite. Arsenic is volatilized
as trioxide at low temperatures which reduces the loss of other valuable The roasting experiments were carried out in a horizontal tube
metals from the calcine. The resulting calcine can be returned to the furnace (EHA 12/150, Carbolite Gero, UK) (Fig. 2). Its temperature was
smelting furnace. The volatilized high-purity arsenic trioxide can be controlled within ±5 ◦ C using an N-type thermocouple and a heated
reduced to elemental arsenic and used for synthesis of arsenic-bearing length of 150 mm was defined as the hot zone. The ESPD was carefully
precursors (Nazari et al., 2016; Long et al., 2012) and stored without blended with different additives in an agate mortar. The mixture was
environmental pollution. placed into an alumina crucible (45 mm-up/36 mm-down length × 22
mm-up/14 mm-down width × 15 mm height) and put into the furnace.
2. Experimental and analytical methods After that, the furnace was heated to the set temperature at a heating
rate of 10 ◦ C/min and stabilized there for a pre-set time (10–70 min).
2.1. Materials After each experiment, the furnace was cooled down to room tempera­
ture and the specimen was taken out for chemical analysis. The volatile
The industrial electrostatic precipitator dust (ESPD) used in this products were scraped off from the inside of the quartz work tube using a
study was from a bottom-blowing bath-smelting copper smelting plant long spoon. The exhaust gas was absorbed by NaOH solution. Nitrogen
in Hunan province, China. Its elemental composition was analyzed with or air was used as the atmosphere throughout the entire study at pre­
inductively coupled plasma-optical emission spectroscopy, Agilent 5100 determined flow rates (20–100 mL/min).
ICP-OES (Agilent, USA), which indicated copper as the main element. The volatilization degree of element i (ηi) is defined in the following
The extraction experiments showed that about 63% of the copper in the equation:
dust was in the form of copper sulfate. In addition, arsenic, lead and zinc
accounted for 16 wt%, 10 wt%, and 1 wt% of the dust, respectively
(Table 1).
The phase distribution of arsenic in the ESPD was obtained using a
sequential extraction (Yang et al., 2017b; Zhang, 1992). Its flow chart
and the results obtained are shown in Fig. 1a and b, respectively. As
shown in Fig. 1b, most of the arsenic coexisted with oxygen (91%). In
terms of the distinguishable arsenic speciation, trivalent and pentava­
lent arsenic accounted for 55% and 45% of the total, respectively.
The pyrite concentrate used in the experiments was received from a
mining area in Anhui province, China. According to chemical analysis

Table 1
Elemental composition of ESPD used in this study (wt.%).
Element Cu Fe S Si As Pb Zn Bi Fig. 2. Schematic diagram of reaction environment and experimental device.
(1) gas cylinder; (2) flowmeters; (3) thermocouple; (4) crucible; (5) quartz tube;
Content 19.3 2.6 7.3 0.2 16.4 9.9 1.1 0.7
(6) alkali absorbent.

2
Y. Chen et al. Minerals Engineering 164 (2021) 106796

ηi = (m0 × ωi − m1 × ωi ′ )/(m0 × ωi ) × 100% (1)

where, m0 and m1 are quantities of the specimen before and after

roasting; ωi and ωi ′ are the percentage contents of i in the specimen
before and after roasting, respectively.

2.3. Analytical methods

The particle size distributions of the flue dust samples were deter­
mined using a Mastersizer 2000 (Malvern, UK) laser diffraction particle
size analyzer. Alcohol was used as dispersant and the medium in the
measurements.
The microstructures of the samples were examined using a Nova
NanoSEM 230 FEI scanning electron microscope, SEM (Thermo Fischer
Scientific, USA), with an energy dispersive spectrometric analyzer
(EDS). Polished sections from selected dust samples were further
examined by SEM-EDS to characterize the cross sections of the ESPD
particles. The flue dust samples were mixed with resin and hardener and
Fig. 3. Evaporation degree of arsenic, lead and zinc from ESPD with different
were cast in a polypropylene mold (φ 30 mm) and cured at 55 ◦ C for 5 h.
additives at 550 ◦ C.
Then, they were polished with #300, #800, #1200, and #1500 emery
papers in sequential order along the longitudinal direction and washed
A high temperature and long roasting time would increase the cost of
with ethanol, followed by polishing.
the treatment. The optimal treatment conditions for selective volatili­
The crystallographic compositions of the samples were identified by
zation of arsenic from high arsenic flue dusts were obtained by mixing
a D/Max 2500 VB + 18 kW X-ray diffractometer, XRD (Rigaku, Japan),
30 wt% pyrite with 70 wt% ESPD and roasting for 30 min at 550 ◦ C
using Cu (40 kV, 300 mA) radiation, with steps of 0.02◦ and a scanning
under 80 mL/min N2. In these conditions, arsenic volatilization was in
speed of 10◦ ⋅min− 1 in the 2θ range from 10 to 80◦ .
excess of 92% and more than 95% copper was retained in the calcine. At
The total metals in the specimens were obtained by a wet digestion
the same time, less than 20% of lead and zinc was evaporated.
technique developed by the US Environmental Protection Agency
(USEPA, Method 3052). The dissolved metal concentrations were
determined by ICP-OES. Each specimen was divided into three to pro­ 3.2. Characterization of roasted products
vide three separate analyses per specimen. The average results were
presented to minimize the random errors. 3.2.1. Mineralogy of roasted products
The off-gas released during the roasting of the samples was analyzed Typical X-ray diffractograms of the ESPD studied and the calcines
using a TG apparatus (Thermogravimetry) coupled on-line with a obtained with 30 wt% pyrite roasted at 400, 500, 550, 600, 700, and
Fourier transform infrared spectrometer, FT-IR (PerkinElmer, USA) with 800 ◦ C for 30 min under flowing N2 atmosphere are shown in Fig. 5. As
4 cm− 1 resolution in the wave number range of 400–4000 cm− 1. seen in Fig. 5, PbSO4 was the predominant crystalline phase in the
original, untreated ESPD. When doped with 30 wt% pyrite, this main
3. Results and discussion peak of the calcine did not change when the temperature was increased
to 550 ◦ C, after which PbS became the main peak and PbSO4 exhibited
3.1. Selective removal efficiency for arsenic only a weak peak at about 2θ = 26◦ , which disappeared with a further
increase in temperature.
3.1.1. Arsenic volatilization promoter From the XRD pattern of the original ESPD, only small peaks of
The ESPD with additives (30 wt% + 70 wt% ESPD) was roasted at CuSO4⋅3H2O were identified. It was difficult to find distinct character­
550 ◦ C for 50 min in a 100 mL/min nitrogen flow. The ESPD without istic peaks for the copper phases, although copper was the main
additives was roasted in the above conditions in a 100 mL/min nitrogen elemental component in ESPD. This is because the copper compounds
atmosphere and a 100 mL/min air atmosphere, respectively. The were mostly in amorphous form, as discussed in our earlier results. In the
evaporation yields of arsenic, lead, and zinc from ESPD, with and calcines at 400 ◦ C and 500 ◦ C, it was also difficult to find distinct
without additives, are shown in Fig. 3. It can be seen that the three metal characteristic peaks for the copper phases, except for some small peaks
elements of interest were more volatile in the nitrogen atmosphere than that overlapped with others, i.e., CuS, Cu2S, CuO, Cu3(AsO4)2, and
in air, since a fraction of the metal was oxidized in the former to invo­ Cu4(As2O7)O2. Due to the weak peak intensities, they cannot be labelled
latile oxides. A similar phenomenon also took place when sulfur and on the presented X-ray diffractograms. When the temperature was
carbon were used as an additive. However, sulfides as additives had a increased further to 550 ◦ C, new phases of bornite Cu5FeS4 and copper
different effect on metal evaporation. They promoted arsenic volatili­ ferrite CuFe2O4 appeared in the calcine.
zation but had less effect on the evaporation of lead and zinc. The ef­ With respect to iron, due to the low iron content of ESPD (2.6 wt%),
fectivities of the studied additives on promoting arsenic volatilization only weak Fe3O4 and Fe2As4O12 peaks were identified in the XRD
were ranked as follows: FeS2 > FeS > Na2S > PbS > S > C. Since FeS2 pattern. After roasting at 400 ◦ C with 30 wt% pyrite, Fe3O4 disappeared
had the highest effectivity in promoting arsenic volatilization, it was and new phases of FeS2 and FeAsO4 appeared. In the calcines at 500 ◦ C,
chosen as the only additive in the detailed volatilization experiments. new phases of FeS and Fe4O3(AsO4)2 were identified. Compared to the
calcines at 400 ◦ C and 500 ◦ C, it was observed that FeS2 started to
3.1.2. Roasting process optimization decompose at 400–500 ◦ C, forming FeS. The iron arsenate phases dis­
In order to get a high volatilization yield of arsenic and low evapo­ appeared, while bornite Cu5FeS4 and base metal ferrites MeFe2O4 (Me =
ration rates of lead and zinc, ESPD mixed with pyrite was treated under Cu, Zn, Fe) appeared when the temperature was increased to 550 ◦ C.
different conditions. The effects of temperature (400–800 ◦ C), roasting With a further increase in roasting temperature to 800 ◦ C, the XRD
time (10–70 min), pyrite addition (10–50 wt%), and nitrogen flow rate pattern of the calcine was the same, although small differences in the
(20–100 mL/min) on the evaporation yields of arsenic, lead, and zinc peak positions were evident and, thus, the PDF cards of the Cu5FeS4
were investigated. The results obtained are summarized in Fig. 4. phase used for identification were different.

3
Y. Chen et al. Minerals Engineering 164 (2021) 106796

Fig. 4. Evaporation of arsenic, lead and zinc from ESPD mixed with pyrite in different roasting conditions studied in this work. (a:40 min, 30 wt% pyrite, 100 mL/
min N2; b: 550 ◦ C, 30 wt% pyrite, 100 mL/min N2; c: 550 ◦ C, 30 min, 100 mL/min N2; d: 550 ◦ C, 30 min, 30 wt% pyrite).

Fig. 5. XRD pattern changes of ESPD calcines after roasting with 30 wt% pyrite at different temperatures for 30 min under N2 atmosphere. 1, PbSO4 (82-1855); 2,
Fe3O4 (79-0416); 3, Fe2As4O12 (27-1166); 4, CuSO4⋅3H2O (22-0249); 5, FeAsO4 (21-0910); 6, FeS2 (71-1680); 7, FeS (89-6926)); 8, Fe4O3(AsO4)2 (85-1554); 9, PbS
(99-0053); 10, Cu5FeS4 (24-0050); 11, MeFe2O4 (Me = Cu, Zn, Fe); 12, Cu5FeS4 (25-1424).

3.2.2. Microstructures and particle size distributions weak peaks of arsenic-iron compounds, e.g., Fe2As4O12.
Back-scattered electron (BSE) micrographs from polished sections of Fig. 7a-d show the SEM images of the original ESPD and the roasted
the original ESPD and the element distribution maps of oxygen, sulfur, calcines at different temperatures. In Fig. 7a, it can be seen that the
iron, copper, zinc, arsenic, and lead by EDS are shown in Fig. 6. The BSE particles in ESPD had different morphologies; an uneven particle size
images show two types of particles. One fraction was composed of large and clear agglomeration are visible. When the pyrite-doped ESPD was
gray particles, enriched in lead, sulfur, and oxygen. The main phase of roasted, the agglomeration tendency of the calcine clearly disappeared.
this fraction according to the XRD analysis was PbSO4. The other frac­ With an increase in temperature, its surface pores first increased and
tion was composed of small and white bright (heavy) particles, con­ then decreased, due to sintering. Furthermore, the calcine particle size
taining mainly arsenic and iron. In addition, lead, sulfur and oxygen in gradually increased along with temperature, as shown in the particle
varying degrees were also abundant in this fraction. Copper was evenly size analyses in Fig. 7e and f, showing that at 550 ◦ C the particle size of
distributed between the two fractions, whilst zinc was less random. the calcine was visibly larger. However, when the temperature was
From the EDS analysis of the polished sections, it can be seen that further increased, the growth in particle size was not as pronounced as
arsenic was mainly wrapped by PbSO4 particles. Iron and arsenic earlier. A large particle size is beneficial to subsequent treatment of the
coexisted, identically according to the XRD results, which indicated calcine and reduces the environmental impact of handling the calcine,

4
Y. Chen et al. Minerals Engineering 164 (2021) 106796

increase volatility. To facilitate the characterization of the volatiles, the

condensed materials at 550 ◦ C and 600 ◦ C were collected, as shown in
Fig. 8. They were white, fine powders, and their XRD spectra proved the
peaks to be pure As2O3.

3.4. Roasting reaction mechanism

3.4.1. Roasting off-gas analysis in TG

TG-DTG (derivative thermogravimetry) analyses of roasted ESPD
and 70 wt% ESPD − 30 wt% pyrite mixtures under a flowing N2 at­
mosphere at a heating rate of 10 ◦ C/min are reproduced in Fig. 9a and b.
These figures show that the weight loss process in a temperature interval
from 30 to 800 ◦ C can be divided into four steps: I (30–113 ◦ C), II
(113–480 ◦ C), III (480–600 ◦ C) and IV (600–800 ◦ C).
As shown in Fig. 9d, thermal decomposition of pyrite occurs when
temperature exceeds approximately 430 ◦ C under air and approximately
460 ◦ C under an inert atmosphere. Under argon, FeS2 decomposes into
FeS and S2 (g). The weight loss of pyrite under argon at 30–480 ◦ C and
480–600 ◦ C was 2.13% and 5.62%, respectively. Assuming that pyrite
Fig. 6. An electron micrograph and EDS element maps of untreated ESPD. and ESPD do not react with each other and they volatilize separately, the
weight loss of the mixture in four steps should be: 4.93% (I), 8.43% (II),
which is caused by fugitive emissions of fine particles. 7.36% (III) and 14.23% (IV) (Fig. 10a and d). By comparing this with the
actual weight losses found in the four steps: 5.38% (I), 9.17% (II),
23.53% (III), and 11.31% (IV) (Fig. 9b), it is clear that pyrite promoted
3.3. Characterization of volatile products weight loss mainly within step III.
According to the TG curve of As2O5 shown in Fig. 9c, the total weight
According to Fig. 4a, the degree of arsenic volatilization at 500 ◦ C loss of As2O5 in nitrogen at 290 ◦ C was 10.2%, and it started at about
was 32.7%. However, as the temperature increased to 550 ◦ C, it abruptly 200 ◦ C. When the temperature increased to 400 ◦ C, the total weight loss
reached 93.1%. A further increase in temperature did not markedly was only 12.1%. Then the weight loss rate evened out (to a very small

Fig. 7. SEM secondary electron images of untreated ESPD (a), calcine with pyrite at 400 ◦ C (b), 550 ◦ C (c) and 600 ◦ C (d) for 30 min under N2 atmosphere and
cumulative size distributions (e) and size distribution frequencies (f) of calcines.

Fig. 8. Macrophotographs and XRD patterns of volatilized products after condensation.

5
Y. Chen et al. Minerals Engineering 164 (2021) 106796

Fig. 9. Thermodynamic analyses of ESPD (a), ESPD with pyrite (b), As2O3 and As2O5 (c) and pyrite (d) at a heating rate of 10 ◦ C/min.

weight loss rate) until the temperature reached about 700 ◦ C, where it 2009; Ahmmad et al., 2016).
increased sharply. An initial mass loss of 10.2% was attributed to the According to the literature summary above, the double bands at
release of H2O absorbed by As2O5. The subsequent weight loss could be around 1143 cm− 1 and 1313 cm− 1 in Fig. 10 were assigned to SO2, and
ascribed to the decomposition and volatilization of As2O5, which was the double bands at 1385 cm− 1 and 2493 cm− 1 to SO3. The characteristic
small below 550 ◦ C. It can be concluded that only a small fraction of peaks at 1020 cm− 1, 792–843 cm− 1, and 644–670 cm− 1 were assigned to
As2O5 in ESPD decomposed under argon at temperatures below 700 ◦ C. As2O3.
However, in the case of As2O3, when the temperature was raised In the 3D in-situ FT-IR spectra of off-gas at 30–800 ◦ C (Fig. 10a and
from 35 ◦ C to 400 ◦ C the weight loss by volatilization reached 93.7%, b), it was found that the peak intensities of SO2, SO3, and As2O3 in the
Fig. 9c, and at 484 ◦ C all of the As2O3 had volatilized, based on the initial off-gas released from ESPD, with and without pyrite, were different at
assay. The thermal behavior of As2O3 and As2O5 was in agreement with different temperatures. Fig. 10a1-a6 compare the FT-IR spectra of the
earlier observations (Martin, 1977; Helsen et al., 2004). Assuming that off-gas of ESPD at 113, 234, 400, 513, 550, and 600 ◦ C and Fig. 10b1-b6
all of the As2O3 in the sample evaporated, the corresponding degree of show those of the off-gas of ESPD with pyrite at 113, 231, 400, 537, 550,
arsenic removal with pyrite at 400 ◦ C should be 46.6%, instead of 22.6% and 600 ◦ C.
(Fig. 4a). This indicates that some arsenic was combined at low tem­ As shown in Fig. 10a1, when ESPD was roasted at 113 ◦ C only H2O
peratures with other elements to form non-volatile substances, thus was released into the off-gas, proving that the weight loss in stage I was
preventing evaporation. mainly due to the removal of humidity. During stage II, the FT-IR spectra
For a better understanding of the chemical processes at high roasting of gaseous products, shown in Fig. 10a2, proved that a small fraction of
temperature, an in-situ FT-IR tracing was performed of the off-gas As2O3, and SO2 was released at 234 ◦ C. When increasing the tempera­
composition generated by the roasting reactions in TG. The results ob­ ture, SO3 was produced, as shown in Fig. 10a3. When increasing the
tained are shown in Fig. 10. temperature further, the intensities of the SO2 and SO3 bands became
In Fig. 10, the bands in the wave number regions of 1400–1800 cm− 1 stronger (Fig. 10a4-a6). A similar process sequence occurred when
and 3400–3800 cm− 1 were attributed to the presence of H2O in the off- pyrite-doped ESPD was roasted: SO2 appeared in the gas already at
gas (Alvarez-Ayuso et al., 2008; Chindaprasrt et al., 2009; Li et al., 113 ◦ C (Fig. 10b1). However, during stages III and IV, the As2O3, SO2,
2015). The peaks related to gaseous CO2 were visible at 2250–2400 and SO3 bands were much stronger than without pyrite (Fig. 10b4-b6),
cm− 1 (Iskra et al., 2017). They were probably related simply to the indicating a more intensive release of As2O3, SO2, and SO3 into the off-
background CO2 of the IR spectrometer. gas.
Some researchers (Sodeau and Edward, 1980; Shen et al., 1996; Yang When ESPD was roasted, some As2O3 volatilized along with the
et al., 2005) have reported that three fundamental vibrational bands of increasing temperature, SO2 was also present. Some gaseous As2O3
SO2 existed in the FT-IR spectrum, the v1 (S– – O symmetric stretch) re­ clearly reacted with CuSO4, forming arsenates and releasing O2 and SO2.
gion at 1146–1190 cm , v2 (O– S– O deformation) at 530–570 cm− 1,
− 1 – – At low temperatures, SO2 and O2 can react and form SO3, or CuSO4 can
and v3 (S– – O asymmetric stretch) at 1280–1356 cm− 1, and four vibra­ react and release SO3. At the same time, As2O5 reacted with metal ox­
tional bands of SO3 observed at about 2438 cm− 1 (v1 + v3), 1386 cm− 1 ides, forming arsenates. As the roasting temperature was below the
(v3), 528 cm− 1 (v4), and 490 cm− 1 (v2). The most absorption-strong SO3 decomposition temperatures of arsenates, the amount of volatile arsenic
band lies at about 1386 cm− 1 and another SO3 band at about 2438 cm− 1 did not significantly increase at high temperatures.
is more attractive (Bonfim et al., 2017). Four prominent bands were When doping ESPD with pyrite, FeS2 can react with CuSO4, pro­
expected in the FT-IR spectrum of crystalline As2O3: the v1 (stretching ducing CuS, Fe2O3 and releasing gaseous SO2 and SO3 at low tempera­
vibrations) region at about 1050 cm− 1, v2 (symmetric stretching vi­ tures. With increasing temperature, the reaction path prior to
brations) at 625 cm− 1, v3 (doubly degenerated stretching vibrations) at decomposition of pyrite was similar to that of roasting pure ESPD. With
812 cm− 1, and v4 (doubly degenerated bending vibrations) at 495 cm− 1 a further increase in temperature, FeS2 decomposed, forming FeS and S2
(Lucovsky and Galeener, 1980; Ležal and Koňák, 1995; Gandhi et al. which reacted with arsenates and produced As2O3, thus promoting

6
Y. Chen et al. Minerals Engineering 164 (2021) 106796

Fig. 10. In-situ FT-IR spectra of off-gas released during TG roasting under flowing N2 at a heating rate of 10 ◦ C/min. (a-a6: ESPD; b-b6: ESPD with pyrite).

arsenic volatilization. The above process was consistent with the XRD
data of Fig. 5, as FeS was found in the calcine at 500 ◦ C, and all arsenates
disappeared at 550 ◦ C. A possible reaction sequence for the entire ESPD
mixed with pyrite is summarized and presented graphically in Fig. 11.

3.4.2. Thermodynamic analysis

The above mentioned reaction paths can be briefly represented with
the reactions shown in Table 2, in which Cu3(AsO4)2 and FeAsO4 were
merely representative of arsenates, and FeS and S2(g) represent the
decomposition products of pyrite.
A ΔGθ vs. T diagram of reactions (1)–(11) was obtained from the HSC
Fig. 11. A schematic release process of arsenic from ESPD by pyrite.
database, version 9.4.1 (Roine, 2018), under atmospheric pressure, as

7
Y. Chen et al. Minerals Engineering 164 (2021) 106796

Table 2 2018), as shown in Fig. 12b and c. The initial input components are
Main chemical reactions during the process of roasting pyrite-doped ESPD. shown in Table S-1 (supplementary material). It can be concluded that
Reaction ΔGθT − T/(kJ/mol – Eq. CuSO4 can react with arsenic oxides to form copper arsenates, whereas
◦
C) PbSO4 does not react and exists as a stable substance in the system
6CuSO4 + 2As2O3(g) = 2Cu3(AsO4)2 + 6SO2(g) + O2 ΔGθT = − 0.860 T + (1) (Fig. 12b). Fig. 12c further illustrates that releasing most of the As in
(g) 199.00 arsenates at 550 ◦ C is feasible in the experimental system. The gas phase
11CuSO4 + 8FeS2 = 11CuS + 4Fe2O3 + 16SO2(g) ΔGθT = − 3.366 T + (2) equilibrium compositions of the corresponding systems (Fig. S-2) show
277.84 that FeS2 can promote the release of arsenic in ESPD mainly in the form
2SO2(g) + O2(g) = 2SO3(g) ΔGθT = 0.188 T − (3)
146.42
of arsenic oxide.
2CuS + 3O2(g) = 2CuO + 2SO2(g) ΔGθT = 0.175 T − (4)
756.78 4. Conclusions
Fe2O3 + As2O5 = 2FeAsO4 ΔGθT = − 0.122 T − (5)
25.29
2Cu3(AsO4)2 + 3FeS2 = 3Cu2S + Fe3O4 + 3SO2(g) + ΔGθT = − 1.260 T + (6) A novel pyrometallurgical concept was developed to selectively
2As2O3(g) 326.31 separate arsenic from high-arsenic copper smelter flue dusts containing
10Cu3(AsO4)2 + 24FeS = 15Cu2S + 8Fe3O4 + 9SO2 ΔGθT = − 4.033 T + (7) high concentrations of copper sulfates, using pyrite as an additive. The
(g) + 10As2O3(g) 343.12
optimized process conditions in this work consisted of roasting for 30
40FeAsO4 + 25S2(g) = 16FeS + 8Fe3O4 + 34SO2(g) ΔGθT = -6.10 T + (8)
+ 20As2O3(g) 2724.53
min at 550 ◦ C under an inert atmosphere. During the roasting step, more
CuS + FeS = CuFeS2 ΔGθT = 0.021 T − (9) than 92% of the arsenic was volatilized and more than 95% of the copper
40.59 was retained in the calcine, indicating a sharp, one-step separation of
5CuFeS2 + 6O2(g) = Cu5FeS4 + 4FeO + 4SO2(g) ΔGθT = 0.159 T − (10) arsenic from copper in ESPD.
2204.23
The roasting mechanism was as follows: at low temperatures, As2O5
CuO + Fe2O3 = CuFe2O4 ΔGθT = − 0.023 T + (11)
11.97 reacted with metal oxides forming arsenates. Some gaseous As2O3
2FeS2 → 2FeS + S2(g) (12) reacted with CuSO4, forming arsenates above 157 ◦ C. When doping
ESPD with pyrite, the arsenates formed were reduced to As2O3 by pyrite
and its thermal decomposition products (FeS and S2(g)) above 460 ◦ C
presented in Fig. 12a. It indicates that within the corresponding tem­ which was then volatilized. After that, the reaction products in the
perature range mentioned, the reactions can proceed except for those calcine consisting of copper and iron sulfides, reacted to form chalco­
involving the decomposition products of pyrite (Eqs. (7)–(10)). As pyrite CuFeS2, which is easily oxidized into bornite Cu5FeS4 in the
shown in Fig. 10d, pyrite starts to decompose at about 460 ◦ C (Eq. (12)). presence of oxygen and copper oxides combined, as the iron oxides form
Therefore, arsenates can react with the products of pyrite at above copper ferrite CuFe2O4.
460 ◦ C. The treated copper flue dust calcine can be fed back into the smelting
On the other hand, the equilibrium phase compositions of simplified furnace for copper recovery, as a high purity arsenic trioxide byproduct
ESPD and ESPD mixed with FeS2 at 1 atm were also calculated using is captured from the off-gas. It can for instance be reduced to elemental
MTDATA version 7.1 and Mtox database 8.2 (Gisby et al., 2016; Roine, arsenic followed by synthesis of arsenic-bearing functional materials

Fig. 12. The calculation results of the studied system. (a: Standard Gibbs energy changes of possible reactions as a function of temperature in a temperature range of
0–800 ◦ C; b:equilibrium phase assemblages of CuSO4-PbSO4-As2O3-As2O5 reaction system; c: equilibrium phase assemblages of CuSO4-FeS2-As2O3-As2O5 reac­
tion system).

8
Y. Chen et al. Minerals Engineering 164 (2021) 106796

and chemicals. It can also be stored as a small volume waste, without the Gonzalez-Moreno, P., Inglesias-Gonzalez, N., Romero, R., Mazuelos, A., Carranza, F.,
2018. Recovery of zinc and copper from copper smelter flue dust. Optimisation of
risk of environmental pollution.
sulphuric acid leaching. Environ. Technol. 41 (9), 1093–1100.
Helsen, L., Van den Bulck, E., Van Bael, M.K., Vanhoyland, G., Mullens, J., 2004.
Credit authorship contribution statement Thermal behaviour of arsenic oxides (As2O5 and As2O3) and the influence of
reducing agents (glucose and activated carbon). Thermochim Acta 414 (2), 145–153.
Hou, X.G., Zhang, Y., Zhang, X., 2011. Leaching of copper and zinc from copper
Yujie Chen: Methodology, Formal analysis, Writing - original draft. converter dust with sulfuric acid. Hydrometall. China 30, 57–59.
Shun Zhu: Methodology, Resources, Investigation. Pekka Taskinen: Iskra, A., Gentleman, A.S., Kartouzian, A., Kent, M.J., Sharp, A.P., Mackenzie, S.R., 2017.
Infrared spectroscopy of gas-phase M+(CO2)n (M=Co, Rh, Ir) ion-molecule
Writing - review & editing, Software. Ning Peng: Validation. Bing complexes. J. Phys. Chem. A 121, 133–140.
Peng: Writing - review & editing, Conceptualization. Ari Jokilaakso: Itagaki, K., Yazawa, A., 1982. Thermodynamic evaluation of distribution behaviour of
Writing - review & editing, Funding acquisition. Hui Liu: Writing - re­ arsenic in copper smelting. Mater. Trans. (Japan) 23 (12), 759–767.
Kaur, R., Nexship, C., Wilson, M., George-Kennedy, D., 2010. Minor element deportment
view & editing, Funding acquisition. Yanjie Liang: Writing - review &
at the Kennecott Utah Copper Smelter. In: Copper 2010, Hamburg, Germany, DGBM,
editing, Funding acquisition. Zongwen Zhao: Methodology. Zhongb­ Clausthal-Zellerfeld, 2415–2432.
ing Wang: Validation. Ležal, D., Koňák, K., 1995. The characterization of the infrared absorption spectra of the
vitreous, cubic and monoclinic modification of As2O3. J. Non-Cryst. Solids 192–193,
187–190.
Lucheva, B., Iliev, P., Kolev, D., 2017. Hydro-pyrometallurgical treatment of copper
Declaration of Competing Interest converter flue dust. J. Chem. Technol. Metall. 52 (2), 320–325.
Lucovsky, G., Galeener, F.L., 1980. Local structure and vibrational spectra of v-As2O3.
The authors declare that they have no known competing financial J. Non-Cryst. Solids 37, 53–70.
Li, Q., Isabel, S.S.P., Zhao, Y., 2014. Sequential stepwise recovery of selected metals from
interests or personal relationships that could have appeared to influence flue dusts of secondary copper smelting. J. Cleaner Prod. 84, 663–670.
the work reported in this paper. Li, X., Liu, D., Wang, J., 2016. Experimental study on recovery of valuable metals from
copper smelting dust with high-content of arsenic. Mining Metall. 25 (6), 51–54 (in
Chinese).
Acknowledgements Li, X., Liu, D., Wang, J., 2018. Selective separation experiments of As from copper dust
containing As. Mater. Rev. B: Res. 32 (9), 3110–3115 (in Chinese).
The authors gratefully acknowledge financial support from the Na­ Li, Y., Liu, H., Peng, B., Min, X., Hu, M., Peng, N., Yuang, Y., Lei, J., 2015. Study on
separating of zinc and iron from zinc leaching residues by roasting with ammonium
tional Key R&D Program of China (grant 2018YFC1900302), the Na­ sulphate. Hydrometallurgy 158, 42–48.
tional Science Fund for Distinguished Young Scholars (grant 51825403), Liang, L., Li, L., Liao, C., Shi, Y., 2011. Study on de-arsenic from dust of flash smelting
the key project of the National Natural Science Foundation of China furnace. Nonferrous Met. (Extractive Metall.) 1, 9–11 (in Chinese).
Liao, L., Guo, X., Wang, Q., Tian, Q., Zhang, Y., 2014. Performance analysis of oxygen
(grant 51634010), the project of Innovation-driven Plan in Central bottom blowing copper smelting process using METSIM. Nonferrous Met. Sci. Eng. 5
South University (grant 20180018050001). The facilities provided by (5), 49–55.
Aalto University School of Chemical Engineering, are also Liu, H., Wang, C., Zou, C., Zhang, Y., Wang, J., 2017. Simultaneous volatilization
characteristics of arsenic and sulfur during isothermal coal combustion. Fuel 203,
acknowledged. 152–161.
Long, G., Peng, Y., Bradshaw, D., 2012. A review of copper–arsenic mineral removal from
copper concentrates. Miner. Eng. 36–38, 179–186.
Appendix A. Supplementary material
Mandal, B.K., Suzuki, K.T., 2002. Arsenic round the world: a review. Talanta 58,
201–235.
Supplementary data to this article can be found online at https://doi. Martin, J., 1977. Crystal Structure of As2O5. Angew. Chem. Int. Ed. 16 (5), 314–315.
org/10.1016/j.mineng.2021.106796. Monique, B., Frimmel, F.H., 2013. Arsenic- a review. Part I: occurrence, toxicity,
speciation, mobility. Acta Hydrochim. Et Hydrobiologica 31 (1), 9–18.
Montenegro, V., Sano, H., Fujisawa, T., 2013. Recirculation of high arsenic content
References copper smelting dust to smelting and converting processes. Miner. Eng. 49, 184–189.
Morales, A., Cruells, M., Roca, A., Bergo, R., 2010. Treatment of copper flash smelter flue
dusts for copper and zinc extraction and arsenic stabilization. Hydrometallurgy 105,
Ahmmad, S.K., Samee, M.A., Taqiullah, S.M., Rahman, S., 2016. FT-IR and Raman
148–154.
spectroscopic studies of ZnF2–ZnO–As2O3–TeO2 glasses. J. Taibah Univ. Sci. 10,
Nazari, A.M., Radzinski, R., Ghahreman, A., 2016. Review of arsenic metallurgy:
329–339.
Treatment of arsenical minerals and the immobilization of arsenic. Hydrometallurgy
Alguacil, F.J., Garcia-Diaz, I., Lopez, F., Rodriguez, O., 2015. Recycling of copper flue
174, 258–281.
dust via leaching-solvent extraction processing. Desalin. Water Treat. 56 (5),
Roine, A., 2018. HSC Chemistry 9.4. Outotec Research, Pori, Finland.
1202–1207.
Shen, H.J., Shen, M., Yu, C., Lee, Y., 1996. Isomers of SO3: Infrared absorption of OSOO
Alvarez-Ayuso, E., Querol, X., Plana, F., Alastuey, A., Moreno, N., Izquierdo, M., Font, O.,
in solid argon. J. Chem. Phys. 104 (15), 5745–5753.
Moreno, T., Díez, S., Vázquez, E., Bizinotto, M.B., 2008. Environmental, physical and
Sodeau, J.R., Edward, K.C.L., 1980. Photoxidation of sulfur dioxide in low-temperature
structural characterisation of geopolymer matrixes synthesised from coal (co-)
matrices. J. Phys. Chem. 84 (25), 3358–3362.
combustion fly ashes. J. Hazard. Mater. 154 (1–3), 175–183.
Sohn, H., Fukunaka, Y., Oishi, T., Asaki, Z., Song, H., 2004. Kinetics of As, Sb, Bi and Pb
Bakhtiatri, F., Zivdar, M., Atashi, H., Bagheri, S.A.S., 2008. Continuous copper recovery
volatilization from industrial copper matte during Ar+O2, bubbling. Metall. Mater.
from a smelter’s dust in stirred tank reactors. Int. J. Miner. Process. 86, 50–57.
Trans. B 35 (4), 651–661.
Bonfim, V., Castilho, R., Baptista, L., Pilling, S., 2017. SO3 formation from the X-ray
Swinbourne, D., Kho, T., 2012. Computational thermodynamics modeling of minor
photolysis of SO2 astrophysical ice analogues: FTIR spectroscopy and
element distributions during copper flash converting. Metall. Mater. Trans. B 43 (4),
thermodynamic investigations. PCCP 19, 26906–26917.
823–829.
Bowell, R.J., Alpers, C.N., Jamieson, H.E., Nordstrom, D.K., Majzlan, J., 2014. The
Tan, C., Li, L., Li, K., Zhong, D., 2018a. Separation of As from high As-Sb dust using Fe2O3
environmental geochemistry of arsenic- an overview. Rev. Mineral. Geochem. 79 (1),
as a fixative under O2–N2 atmosphere. Sep. Purif. Technol. 194, 81–88.
1–16.
Tan, C., Li, L., Zhong, D., Wang, H., Li, K., 2018b. Separation of arsenic and antimony
Chen, C., Zhang, L., Jahanshahi, S., 2010. Thermodynamic modeling of arsenic in copper
from dust with high content of arsenic by a selective sulfidation roasting process
smelting processes. Metall. Mater. Trans. B 41 (4), 1175–1185.
using sulfur. Trans. Nonferrous Met. Soc. China 28, 1027–1035.
Chen, C., Zhang, L., Jahanshahi, S., 2013. Application of MPE model to direct-to-blister
Tang, H., Qin, Q., Guo, Y., Zheng, X., Xue, P., Li, G., 2014. Pretreatment of high arsenic
flash smelting and deportment of minor elements. Copper 2013 Vol III (Book 2), Ch
and antimony smelting dust for arsenic removal using roasting process. Conservat.
IIM, Santiago, Chile, 289–304.
Utilizat. Mineral Resources 3, 35–38 (in Chinese).
Chen, Y., Liao, T., Li, G., Chen, B., Shi, X., 2012. Recovery of bismuth and arsenic from
US Environmental Protection Agency (EPA). Method 3052. Microwave assisted acid
copper smelter flue dusts after copper and zinc extraction. Miner. Eng. 39, 23–28.
digestion and organically based matrices. Washington DC., 1998.
Chindaprasrt, P., Jaturapitakkul, C., Chalee, W., Rattanasak, U., 2009. Comparative
Xu, Z., Li, Q., Nie, H., 2010. Pressure leaching technique of smelter dust with high-copper
study on the characteristics of fly ash and bottom ash geopolymers. Waste Manage.
and high-arsenic. Trans. Nonferrous Met. Soc. (China) 20, 176–181.
29 (2), 539–543.
Yang, S., Liu, Z., Mao, H., Hemley, R.J., 2005. High-pressure vibrational spectroscopy of
Gandhi, Y., Venkatramaiah, N., Kumar, V.R., Veeraiah, N., 2009. Spectroscopic and
sulfur dioxide. J. Chem. Phys. 122 (174511), 1–9.
dielectric properties of ZnF2–As2O3–TeO2 glass system doped with V2O5. Physica B
Yang, T., Fu, X., Liu, W., Chen, L., Zhang, D., 2017a. Hydrometallurgical treatment of
404, 1450–1464.
copper smelting dust by oxidation leaching and fractional precipitation technology.
Gisby, J., Taskinen, P., Pihlasalo, J., Li, Z., Tyrer, M., Pearce, J., Avarmaa, K.,
JOM 69 (10), 1982–1986.
Björklund, P., Davies, H., Korpi, M., Martin, S., Pesoen, L., Robinson, J., 2016.
Yang, W., Tian, S., Wu, J., Chai, L., Liao, Q., 2017b. Distribution and behavior of arsenic
MTDATA and the prediction of phase equilibria in oxide systems: 30 years of
during the reducing-matting smelting process. JOM 69 (6), 1077–1083.
industrial collaboration. Metall. Mater. Trans. B 48, 91–98.

9
Y. Chen et al. Minerals Engineering 164 (2021) 106796

Zhang, H., 1992. Chemical Phase Analysis of Ores and Industrial Products. Metallurgical Wang, Q., Guo, X., Tian, Q., Jiang, T., 2017. Effects of matte grade on the distribution of
Industry Press, Beijing, pp. 50–53 (in Chinese). minor elements (Pb, Zn, As, Sb, and Bi) in the bottom blown copper smelting process.
Zhong, D., Li, L., Tan, C., 2017. Separation of arsenic from the antimony-bearing dust Metals 7 (302), 1–11.
through selective oxidation using CuO. Metall. Mater. Trans. B 48, 1308–1314.

10