ISSN 0036-0295, Russian Metallurgy (Metally), Vol. 2007, No. 4, pp. 268–275. © Pleiades Publishing, Ltd., 2007.

Original Russian Text © O.A. Teplov, I.G. Voropaev, V.G. Dyubanov, L.I. Leont’ev, 2007, published in Metally, 2007, No. 4, pp. 11–20.

Hydrogen Reduction Kinetics of Electrometallurgical Slime

O. A. Teplov, I. G. Voropaev, V. G. Dyubanov, and L. I. Leont’ev
Baikov Institute of Metallurgy and Materials Science, Russian Academy of Sciences,
Leninskii pr. 49, Moscow, 119991 Russia
e-mail: teplov@ultra.imet.ac.ru
Received July 26, 2006

Abstract—The interaction of hydrogen with the zinc-containing electrometallurgical slime of the Severstal’
metallurgical works has been studied. The sequence of transformations in the slime heated to 1100°C in hydro-
gen or air has been established. The experimental and calculated weight losses coincide. Some of the carbonates
are shown to decompose in the temperature range 300–700°C, and most iron and zinc oxides are reduced to a
metal. In the temperature range 650–850°C, zinc is almost completely sublimated. At temperatures above
800°C, complex oxides are reduced and calcium and magnesium carbonates and sulfates are likely to decom-
pose. Experimental digital data on the zinc sublimation rate are processed by the least squares method with
approximating equations used in thermal analysis. The kinetics of nonisothermal zinc sublimation is compre-
hensively analyzed using a unique procedure developed for taking into account the background of a peak and
the effect of accompanying processes. An equation for the calculation of the zinc sublimation rate under exper-
imental conditions (fluidized bed) is given and tested.
PACS numbers: 47.45.Ab, 51.10.+y
DOI: 10.1134/S0036029507040027

INTRODUCTION stored on a multichannel computer-assisted system for

Ferrous metallurgy works have accumulated and do data acquisition and processing. The sampling rate in
not use huge stores of dustlike technological waste
materials containing many valuable elements, includ-
ing germanium, zinc, iron, and so on. For example, the Table 1. Composition of the electrometallurgical slime
from the Severstal’ metallurgical works
zinc content after the utilization of automotive bodies
and equipment reaches 20% in the slime of arc-furnace Component* Content, % Error, %
plants. It is wise to use this accumulation of technolog-
ical waste materials in new high-tech production pro- Fetot 27.68 ± 8.860
cesses. To process metallurgical slimes, one has to Femet <1.00 –
experimentally study their complex phase component, FeO 4.95 ± 0.160
their structure, and the sequence of transformation
Fe2O3 34.08 ± 0.400
reactions during their interaction with artificial media
of various oxidizing potentials at various temperatures SiO2 8.51 ± 0.190
and partial gas pressures. Al2O3 0.60 ± 0.059
The purpose of this work is to experimentally study P 0.111 ± 0.031
the interaction of the zinc-containing electrometallurgi- Mn 2.23 ± 0.360
cal slime of the Severstal’ metallurgical works with
hydrogen to obtain reliable data on the sequence and MgO 12.69 ± 0.290
kinetics of decomposition, reduction, and sublimation CaO 8.39 ± 0.290
reactions. Cr 0.57 ± 0.090
Ni <0.05 –
EXPERIMENTAL Cu 0.14 ± 0.040
We examined the dustlike slime of the arc-furnace S 0.87 ± 0.014
plant at the Severstal’ metallurgical works. The slime l.o.c 8.41 ± 0.300
composition is given in Table 1. Oil 0.29 ± 0.070
Thermogravimetric experiments were performed on HM 0.10 ± 0.002
a GTDT 24AV (Setaram) thermoanalyzer in a hydrogen
Zn 16.88 ± 3.040
or air flow in the course of linear heating of the slime to
approximately 1100°C. The results obtained were *l.o.c is the loss on calcination; HM is hygroscopic moisture.

268
 HYDROGEN REDUCTION KINETICS OF ELECTROMETALLURGICAL SLIME 269

each channel was one measurement per second. The υw.l, %/s
experiments were carried out at various heating rates
vheat of the slime. 0.016 2
The choice of hydrogen as a reducing agent is
caused by the fact that it is a more active reducer com-
pared to carbon. For example, the reduction of magne-
tite to metallic iron begins at 400–450°C upon heating 0.008
in hydrogen [1] and at ~800°C during carbon reduction. 1
It was also experimentally found [2] that the addition of I 3
CO to H2 in the temperature range 600–1230°C II
decreases the pellet reduction rate. Moreover, the use of
hydrogen significantly simplifies the interpretation of a 0
thermogravimetric (TG) curve due to a number of rea- ~
~
0.004 III IV
sons.
The numerical data obtained (accumulated data
arrays) were processed with computer programs in sev- 0 200 400 600 800 1000 1200
eral steps. In the first step, the data were cleaned from t, °C
outliers and nondistorting digital filtration. We then
digitally differentiated the experimental data on a
change in the sample weight and transformed them into Fig. 1. Temperature dependences of the slime sample
weight loss during heating at vheat = 5 K/min in (I; peaks 1,
differential thermogravimetric (DTG) data arrays, i.e., 2, 3 in the slime reduction DTG curve) a hydrogen flow and
into the temperature dependences of the weight loss (II) an air flow. The temperature dependences of (III) the
rate vw.l. Figure 1 shows several DTG curves thus dehydration rate of pure iron hydroxide FeOOH in hydro-
obtained. gen and (IV) the hydrogen-reduction rate of pure iron oxide
Fe2O3.
The DTG curves were then processed to reveal char-
acteristic temperature points (the temperatures of the
beginning and end of a peak, inflection points, minimax tbeg B tend
C
points) that divide these curves into characteristic tem-
perature–time intervals. The sample weight loss was
determined in each interval. The experimental accuracy
of measuring the sample weight was better than 0.3%.
Figure 2 schematically shows a DTG fragment and A
illustrates the principle of its preliminary processing in
order to reveal the characteristic temperatures. The
temperatures of the beginning and end of a peak were
determined from the intersection of the tangents to the D
side lines of a peak. Figure 2 also illustrates the meth-
ods used to take into account the peak background in
simple (dashed line) and complex (segments AB and
BC) versions.
To determine the CO2 concentration in the flue gas
during annealing of the slime in helium, we employed
a Binos gas analyzer (Leybold–Heraeus).

EXPERIMENTAL RESULTS Fig. 2. Schematic diagram for the processing of a complex

AND THE INTERPRETATION
OF DTG CURVES
The thermogravimetric experiments with the slime
in a hydrogen flow were carried out at vheat = 5, 10, and
20 K/min. Figure 1 shows DTG curve I obtained at vheat =
5 K/min, which contains three clear peaks. Background
DTG curve II, which was recorded when the slime was
heated under oxidizing conditions in air at the same
heating rate, can be used to comprehensively interpret the
slime weight loss on calcination (see Table 1). Upon heat-
ing in air, the slime lost about 9% of its weight, which
DTG curve: fine lines are the tangents to the side lines of the
DTG peak; Tbeg and Tend are the temperatures of the begin-
ning and end of the peak determined by a standard proce-
dure; the dashed line is the simple background line of the
peak; segments AB and BC are the lines of the complex ver-
sion of the background; CD is the tangent line (continuation
of the side line of the peak); and the hatched region illus-
trates the portion of the peak that is not affected by the pro-
cesses on the left and right of the peak.
agrees completely with the slime weight loss on calci-
nation given in Table 1 (8.41%).

RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4

270 TEPLOV et al.

Table 2. Results of the preliminary processing of the data on thermogravimetric tests of the slime in hydrogen
Experimental data at vheat , K/min
Characteristic experi-
No. 5 10 20 mental points in the
DTG curve (see Fig. 1)
T, °C weight loss, % T, °C weight loss, % T, °C weight loss, %
1 322 1.89 299 2.11 324 1.83 Beginning of peak 1
2 507 5.94 514 4.59 ~573 5.00 Inflection point 1
3 566 9.20 ~613 ~9.70 649 8.72 First maximum
4 647 14.13 686 14.56 ~721 13.50 Beginning of peak 2
5 762 26.89 808 28.00 853 26.47 Second maximum
6 801 33.77 841 34.00 914 35.19 Inflection point 2
7 836 36.28 905 36.97 927 36.02 Beginning of peak 3
8 941 41.45 – – 1025 41.06 Third maximum
9 1041 44.82 1057 45.09 1080 44.76 End of the process

Table 3. Average slime weight loss in the characteristic temperature–time intervals

Slime weight loss (%) in the temperature–time interval
Gas atmosphere from the beginning
from 300°C
to 300°C peak 1 peak 2 of peak 3 to the end
to inflection point 1
of the DTG curve
Hydrogen ~2 ~3 ~9 ~22 ~9
Air ~2 ~3 ~1 ~0.5 ~2

The experiments performed in hydrogen at slime is mainly determined by the limiting stages of these
heating rates of 5–20 K/min indicate that the sequence processes.
and character of the processes remain unchanged. The The results of preliminary processing of three DTG
experimental curves are similar to each other, which curves recorded at slime heating rates of 5, 10, and
indicates that the data obtained are reproducible and 20 K/min are given in Table 2. Table 3 lists the calcu-
reliable. An increase in the heating rate leads to only a lated average slime weight losses in the characteristic
small natural shift in the peak positions toward high temperature–time intervals of the processes.
temperatures. The temperature shift in the curves Moreover, making allowance for the slime composi-
induced by an increase in the heating rate is caused by tion (see Table 1), we calculated the maximum weight
the natural delay of the development of real processes losses upon slime annealing in hydrogen at 1100°C. For
(chemical reaction or diffusion) in the slime. The shift the calculation, we assumed that Mn, P, Cu, S, and Zn
in the slime are present in the form of the oxides MnO,
P2O5, CuO, SO2, and ZnO, respectively. The calcula-
Table 4. Calculated weight losses during annealing of the tion results are presented in Table 4. The total calcula-
slime in hydrogen at 1100°C
tion losses (43.4%) coincide approximately with the
Slime components final experimental losses (Table 2; 44.82, 45.09,
Weight loss, % 44.76%). On average, the difference between the calcu-
before annealing after annealing lated and experimental losses is only ~1.5% and is
FeO Fe 1.10
caused by the variation and uncertainty of the slime
composition (see Table 1). For example, the error of
Fe2O3 Fe 10.25 determining the zinc content in the slime is ~3%, and
P2O5 – 0.25 the concentrations of a number of impurities are not
MnO Mn 0.65 given.
SO2 – 1.74 We then identified the chemical reactions that lead
ZnO – 21.0 (16.9 + 4.1) to the detected weight losses in the temperature–time
intervals presented in Table 3. To this end, we used the
l.o.c – 8.41 chemical composition and analyzed the slime weight
Total 43.40 losses upon heating in air; as a result, we were able to

RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4

 HYDROGEN REDUCTION KINETICS OF ELECTROMETALLURGICAL SLIME
divide the tabulated losses on calcination into tempera-
ture intervals, to specify them, and to interpret them
correctly.
Comparing the data in Tables 2–4, we were able to
unambiguously interpret the processes of dehydration,
oxide reduction, and zinc sublimation, i.e., all pro-
cesses up to peak 3 in the DTG curve (see Fig. 1).
The interaction of iron oxides with H2 is known to
begin at ~300°C [3]. During heating to 300°C, hygro-
scopic and crystalline-hydrate moisture (~2%) is
removed. Figure 1 shows curves III and IV recorded for
pure substances heated in H2, namely, for iron hydrox-
ide FeOOH and pure iron oxide Fe2O3. These curves
demonstrate the temperature intervals of the processes
of FeOOH dehydration and the reduction of the dis-
perse Fe2O3 oxide. The first broad peak in curve IV cor-
responds to the transformation of Fe2O3 into magnetite,
and the second peak, which begins at ~400°C, indicates
the reduction of magnetite into metallic iron.
The reduction kinetics of a thin ZnO layer in H2 at
500–600°C is described in [4]. It was shown that pure
zinc oxide should be primarily reduced to a metal by
the beginning of the second peak, i.e., to temperatures
of 647, 686, and 721°C (Table 2) corresponding to heat-
ing rates of 5, 10, and 20 K/min, respectively. For
example, at 600°C, ZnO is completely reduced within
~25 min, and the time of heating from 500 to 647°C at
vheat = 5 K/min is ~30 min. Thus, peak 1 in the DTG
curve of the slime can be reliably attributed to the
reduction of iron and zinc oxides to a metal. The calcu-
lated weight losses in this process are ~15.4% and, on
the whole, agree with the losses in the TG curve (~12%)
in the range from 300°C to the beginning of peak 2.
However, if the air background is taken into account
(~3%), the losses induced by oxide reduction are only
~9%. Therefore, the reduction of iron and zinc oxides
has passed to only a ~60% completion, and the reduc-
tion of the remaining oxides proceeds in parallel with
other processes in the range of peak 2 and further.
Peak 2 in the DTG curve is also unambiguously
interpreted: this is the sublimation of zinc into a gas-
eous hydrogen flow with a total background of ~3.5%
(~3% of them make up the background of the further
reduction of iron and zinc oxides, and the other 0.5% is
the air background). The total losses at this stage indi-
cate a virtually complete zinc removal from the slime
and, hence, a virtually complete reduction of zinc
oxides (~1.6%). As a result, the calculated losses at this
stage are 21.4% (16.9 + 3.5) and agree fully with the
experimental losses (~22%). The transition of zinc to a
gas phase ends at a temperature close to the boiling
temperature of metallic zinc (908°C), which is 200–
300°C lower than the temperature of the carbothermic
Waelz process in tubular rotary kilns [5].
To interpret peak 3 and the tail in the DTG curve up
to ~1080°C, we consider the remaining hard-to-reduce
iron and manganese oxides, impurities, sulfur, phospho-
rus, and the losses on calcination (see Table 1). The losses
RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4
271
in the last interval in the DTG curve (~9%, Table 3)
exceed the losses on calcination. Moreover, the latter
losses also contain moisture (hygroscopic and crystal-
line-hydrate moisture) losses and the weight losses
related to easily decomposing iron and zinc carbonates
[6]. In the DTG curve, these processes are observed at
temperatures up to 300°C and in the range 300–600°C,
respectively.
The special-purpose experiment on the measure-
ment of the CO2 content in helium at the exit of the ther-
mogravimetric device supported the fact that the tem-
perature of the onset of carbonate decomposition is
near 300°C.
By making allowance for the considerations for the
last stage, we obtain 3.4%, which is lower than the
experimental losses by ~5.6% and agrees with the
losses in air (~2.5%). This discrepancy can be attrib-
uted to the following accompanying factors: more com-
plete sulfur and phosphorus removal in hydrogen as
compared to air annealing during the determination of
the losses on calcination; a higher final heating temper-
ature compared to that used for the determination of the
losses on calcination (usually 1000°C); final reduction
under conditions of sintering residual divalent iron
from the strongest compounds, such as calcium ferrites
and zinc and iron silicates; and the reduction and
removal of impurities that are not given in the chemical
analysis. For example, the total weight losses in exper-
iment are higher than the calculated losses by only
~1.5%. As compared to the total losses (~45%), the
insufficiently interpreted weight losses (~5.6%) are low
and well explainable qualitatively.
Thus, we may state that the thermogravimetric data
are virtually completely interpreted.
ANALYSIS OF THE PROCESSES OCCURRING
IN THE TEMPERATURE–TIME INTERVALS
With the found sequence of the processes proceed-
ing in the slime, we can analyze the reactions and their
mechanisms at individual stages. We begin with the
stage of iron oxide reduction, leaving aside the stage of
slime dehydration.
Iron Oxides
As shown in Fig. 1 (curve IV), the reduction of
Fe2O3 in H2 to Fe3O4 begins at ~300°C, and the reduc-
tion of Fe3O4 to Fe, at ~400°C. The related peaks are
well resolved. The region of the reduction of iron
oxides in the slime DTG curve differs from the analo-
gous region characteristic of pure iron oxide. The initial
temperatures of these regions virtually coincide. How-
ever, a comparison of the DTG curves of the slime in
hydrogen and air indicates that reduction begins only at
~400°C, since the DTG curves are approximately the
same up to this temperature. This means that the con-
tent of free Fe2O3 in the slime is low and that the reduc-
tion of iron begins virtually from magnetite.
272 TEPLOV et al.
The slime contains several initial iron oxides. The
chemical composition indicates that, if the entire FeO is
converted into magnetite, the slime should contain at
least 16% Fe3O4 and 23% Fe2O3. In electric-arc fur-
naces, slime forms under complex nonequilibrium con-
ditions at various heat times in a rapidly cooling gas
atmosphere with variable CO and CO2 contents. There-
fore, apart from iron oxides, the slime can contain non-
equilibrium oxide solid solutions and compounds,
including zinc ferrites (ZnFeO4), carbonates, and so on.
In practice, this means that part of the iron is thermody-
namically bound more strongly than iron in its pure
oxides. In DTG curves, the processes of the reduction
of more strongly bound iron shift toward high tempera-
tures. The temperature shift is confirmed by the points
of the maximum rates. For the slime, the corresponding
points are 70–100°C higher than those of pure iron
oxides.
The curve of the slime contains inflection point 1 at
a temperature of ~510°C (see Table 2); after this point,
the weight loss rate increases. This indicates the begin-
ning of another process, which is significantly over-
lapped by the reduction of iron oxides.
State of Zinc in the Slime
Based on the slime composition and oxidizing con-
ditions in the chimneys of electric-arc furnaces, we
assume that zinc vapor is first oxidized to ZnO and that
various complex zinc oxides, sulfates, and carbonates
can then form during further cooling. The presence of
zinc carbonates in the slime is experimentally sup-
ported by the CO2 content measured in the flue gas at
the exit of the thermogravimetric device. The slime
should not contain zinc sulfides, since these are known
to transform into zincite ZnO upon annealing in an oxi-
dizing atmosphere at 600–900°C [7]. The slime compo-
sition indicates the presence of a larger amount of slag
dust in it. Therefore, we may exclude the formation of
pure refractory zinc aluminates and silicates. Complex
zinc oxides are likely to form mainly with iron oxides
during joint oxidation and cooling of metal vapor.
The temperature of the onset of the reduction of the
ZnO oxide in the slime should be about 500°C [4],
which is observed in experiment. For example, the tem-
peratures of the first inflection point in Table 2 for
experiments at heating rates of 5 and 10 K/min are 507
and 514°C, respectively. From this point, iron and zinc
oxides begin to be reduced in parallel, and the phases in
which these oxides are bound much more strongly are
then involved in the process. As noted above, about
40% of the unreduced zinc and iron oxides are retained
by the beginning of zinc sublimation. Presumably,
these are solid solutions of ZnO in magnetite and more
complex solutions and, partly, zinc ferrite ZnFe2O4.
In a hydrogen atmosphere, zinc sulfates, if they are
present in the slime, decompose to form ZnO [7].
Moreover, it should be noted that reduced iron does not
RUSSIAN METALLURGY (METALLY)
interact with ZnO at low temperatures. A special-pur-
pose thermogravimetric experiment shows that the
interaction between carbonyl iron and ZnO becomes
significant only at a temperature of ~900°C. Thus, by
the beginning of sublimation, part of the zinc is reduced
and the remaining part (~40%) is bound into oxides.
Processing of the Experimental Data
on Zinc Sublimation
Since the zinc sublimation peak in the DTG curve is
well resolved, we processed these data. The complexity
and limitations related to the processing of an individ-
ual peak are caused by the impossibility of the exact
determination of the base line from which the peak data
are measured, and the use of the background recorded
under oxidizing conditions during slime heating in air
is incorrect (see Fig. 1). The background in hydrogen
would be different because of incomplete reduction
processes. Therefore, for simplicity, the base line is
often taken to be the straight line connecting the begin-
ning and end of the peak. In the schematic diagram in
Fig. 2, the simple base line of the peak is indicated by a
dashed line.
In a more complex version of the background-line
determination, the effect of overlapping processes at
the beginning and end of the peak are taken into
account (see Fig. 2). In this case, the peak region is
bounded by the peak curve, segments AB and BC of the
background line, and line CD (which is tangent to the
side curve of the peak). The hatched region corresponds
to the undistorted part of the peak, whose processing
can yield more correct results. Nevertheless, the pro-
cesses on the right and left of the peak cannot be com-
pletely taken into account, which distorts the results of
further data processing to some extent. This effect can
be weakened but not eliminated. A unique computer
program helps one to selectively exclude and not to take
into account part of the initial data both at the beginning
of the peak and from its tail. As a result, the data pro-
cessing efficiency increases. However, when the body
of processed data decreases significantly, a region of
unstable solutions appear (as will be shown below).
The experimental data on the zinc sublimation rate
thus selected were processed by the least squares
method with the model
------- = K 0 exp ⎛ – -------⎞ ( 1 – α ) ,
dα E n
dτ ⎝ RT ⎠
where α is the degree of transformation, K0 is the pre-
exponential factor, E is the activation energy, τ is the
time, and n is the formal reaction order.
As a result of processing, we obtain the apparent
activation energy, the formal reaction order, the preex-
ponential factor, and the approximation dispersion.
Table 5 and Fig. 3 give the results of processing the
sublimation peak and final zinc reduction at a slime
heating rate of 5 K/min. The data shown in Fig. 3
Vol. 2007 No. 4
 HYDROGEN REDUCTION KINETICS OF ELECTROMETALLURGICAL SLIME 273

belong to a complex background. They demonstrate Table 5. Least squares processing of the peak of zinc subli-
(a) the interval of reliable results, (b) the region of mation from the slime
unstable solutions, and (c) the region of the effect of
α interval, Simple background line Complex background
accompanying processes on the calculation results. The fractions
total peak area corresponds to a 100% transformation of unity n E, kJ/mol n E, kJ/mol
(in our case, 100% zinc sublimation).
For the simple background (Fig. 2, dashed line), the 0.01–0.99 1.44 280 0.845 218.5
calculated formal reaction order is n > 1 and changes 0.05–0.95 1.68 314 0.828 202.5
significantly as the experimental data cutoff threshold 0.10–0.90 1.55 310 0.800 199.3
changes from 0.01 (1%) to 0.15 (15%) from either side 0.15–0.85 1.35 290 0.80 193.0
of the peak. For the complex background, the reaction
order remains virtually the same within the limits 5– 0.20–0.80 1.18 268 0.547 185.8
15%, which would be expected if the background is 0.25–0.75 1.60 300 0 168.3
correctly taken into account (Table 5). As the cut por-
tion of the peak increases, the data array to be processed
decreases further, the results calculated according to is 0.67 for a decreasing spherical surface of sources
three versions begin to diverge, the scatter of the calcu- having the same size [8]. In other words, the higher the
lated parameters increases, and solution stability is lost formal reaction order, the higher the rate of decrease in
(Fig. 3, region b). In the interval of the most reliable the reaction surface. Moreover, Delmon [4] showed
data (interval a), we obtained the following kinetic that, in the presence of a size distribution of reacting
parameters: the average apparent activation energy is particles, the formal reaction order increases. In our
201 kJ/mol (48 kcal/mol), and the formal reaction order case, a certain size distribution of droplets also takes
is about 0.82.
place. This distribution is likely to be close to the Gaus-
The zinc sublimation peaks recorded at heating rates sian distribution. We can easily show that, in this case,
of 10 and 20 K/min were processed analogously, and the total initial zinc droplet surface is larger than the
similar results were obtained. total surface of zinc droplets whose size is equal to the
average droplet size. This means that, in our case, the
Analysis of the Zinc Sublimation Data rate of change in the reaction surface, at least at the
The processes where the limiting stage is monomo- beginning of the process, is slightly higher. Therefore,
lecular decomposition (thermal decomposition of sol- the apparent reaction order is higher than 0.67. Thus,
ids, dissolution, etc.) usually have a reaction order n ≈ the zinc sublimation reaction order (0.82) calculated for
1 [8]. When the limiting stage of a process is diffusion,
the reaction order is zero [8]. Therefore, we may state
that diffusion does not control the process of zinc sub- n
limation under our experimental conditions. For exam-
ple, the calculation demonstrates that, when zinc (DTG 0.92
curve, peak 2) is removed for a given hydrogen flow
saturated by zinc vapor, the amount of removed zinc is
about 20 times larger than the zinc amount in the sam- c a b
ple. This calculation was performed at a heat of zinc
0.84
evaporation of 117 kJ/mol. At some temperatures, we
obtained the following pressures of saturated zinc
vapor over a pure metal:
T, °C 700 750 800 850 908 0.76 1
pZn , kPa 8 16 30 54 100
2
Hence, internal diffusion and external diffusion only 3
weakly retard sublimation. The data obtained agree 0.68
with our interpretation of the retardation of the chemi- 0 10 20 30 S, %
cal reaction of zinc evaporation and sublimation into a
hydrogen flow. It is known that, if a process is con- Fig. 3. Dependence of the formal reaction order n on the
trolled by a chemical step, the formal reaction order “cut” peak area S: (1) cutting of the beginning of the peak,
depends strongly on size factors. For example, the (2) cutting of the peak tail, and (3) uniform cutting from
zeroth order is observed at a constant reaction surface, either side. (a) The convergent set of solutions and the
region of reliable results, (b) the region of divergent results
e.g., evaporation from a capillary. The reaction order is and unstable solutions, and (c) the region where the pro-
equal to 0.5 for a reaction on a cylindrical surface, e.g., cesses on the right and left of the peak affect the calculation
for the evaporation of filament or fiber substance, and it results.

RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4

274
vz.s, %/s
0.016
0.008
2 in the temperature range 900–1050°C.
1
0 calcination are caused by unaccounted components,
6500 7500 8500
Fig. 4. Comparison of the (1) experimental and (2) calcu-
lated rates of zinc sublimation vz.s from the electrometallur-
gical slime.
the most probable background is logically grounded
and acceptable.
On the whole, we can state that, under the experi-
mental conditions, the controlling step of zinc sublima-
tion is zinc evaporation with an apparent activation
energy of ~200 kJ/mol.
Based on the data obtained for the experimental and
related technological conditions (where zinc sublima-
tion is not controlled by diffusion steps, e.g., for the
estimation of zinc vapor removal from a thin layer or a
fluidized bed), we can write the following equation for
the sublimation rate in the temperature range 500–
900°C:
------- = 2.44 × 10 exp ⎛ – ---------------------------------⎞ ( 1 – α ) . (1)
dα 7 24350
dτ ⎝ 900 + τdT /dτ⎠
In Fig. 4, we compare the experimental data and the
data calculated by Eq. (1). These data are seen to agree
well with each other, which is also indicated by a small
approximation dispersion (4.6%).
Processes in the Fourth Interval in the DTG Curves
The weight losses in the final portion of the DTG
curves, i.e., at temperatures above ~900°C, are 8–9%.
By this stage, the entire zinc has passed into a gas, and
all easily reducible oxides, including free iron oxides,
are completely reduced. Since the calculated and exper-
imental weight-loss balances are approximately the
same, the losses at the final stage should be related to
the tabulated slime composition, i.e., to hard-to-remove
elements (such as sulfur and phosphorus), hard-to-
reduce oxides, and (only partly) unaccounted impuri-
ties. To calculate the sublimate content and composi-
tion at the final stage, the obtained slime composition is
insufficient (see Table 1), and the current phase compo-
sition of the sample is required. If we assume that the
TEPLOV et al.
entire sulfur in the slime is concentrated in sulfates and
that sulfur goes from the slime in the form of SO2 dur-
ing the hydrogen process, this cannot cover even half
the losses. All strong slime oxides (CaO, MgO, Al2O3,
SiO2) are not reduced by hydrogen at 900–1100°C;
therefore, the hard-to-reduce oxides should mainly
consist of their compounds and solutions with iron
oxides (fayalite, etc.). Iron reduction from such com-
pounds and solutions can also be complicated by sinter-
ing of reduced metallic iron [1, 9]. Thus, there is good
reason to believe that the final reduction of iron occurs
It is also obvious that the large weight losses during
τ, s
namely, the strongest carbonates that should form dur-
ing slime formation in the presence of CO2. These
strongest carbonates are calcite CaCO3 and dolomite
CaMg[CO3]2, which decompose upon heating in the
temperature ranges 800–100°C and 870–1000°C,
respectively [6]. Other carbonates, such as siderite
FeCO3 and smithsonite ZnCO3 dissociate at ~500°C
[6], and they should not be taken into account. Their
amount in the slime should be small, since they can
form only at a temperature of ~450°C or below. Hence,
this process occurs at a very low rate under conditions
where slime particles have been formed.
Thus, the final stage consists mainly of the removal
of sulfur in the form of H2S and SO2, the final reduction
of residual iron from strong oxides and their solid solu-
tions, and the decomposition of strong carbonates.
CONCLUSIONS
0.82 (1) Given reliable data on the composition of a
material, the thermogravimetric method alone can be
used to successfully study processes in redox media
and the mechanisms of these processes in complex
objects such as metallurgical slimes.
(2) When the electrometallurgical slime of the Sev-
erstal’ metallurgical works is heated in a hydrogen flow,
the processes of zinc reduction and sublimation are
shown to be separated. The differential thermogravi-
metric curve of the slime consists of four temperature–
time zones and has been completely interpreted.
(3) Zinc removal begins at ~650°C and ends at
~900°C.
(4) Our calculations demonstrated that zinc subli-
mation is controlled by a chemical step. For a thin and
fluidized bed, we obtained an equation for zinc subli-
mation in a hydrogen flow to the boiling temperature of
zinc (908°C).
(5) The data obtained can be used to find the
sequence of transformations in the slime during its
interaction with carbon and a converted gas.
RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4
 HYDROGEN REDUCTION KINETICS OF ELECTROMETALLURGICAL SLIME
REFERENCES
1. O. A. Teplov, I. G. Voropaev, G. K. Demenskii, and
Yu. V. Tarasenko, “Hydrogen Reduction Kinetics of Sin-
gle-Crystal Magnetite,” in Theory and Practice of Direct
Iron Production (Nauka, Moscow, 1986), pp. 156–160.
2. N. Towhidi and J. Szekely, “Reduction of Haematite Pel-
lets with Low SiO2 by Mixed CO + H2 at the Interval
600–1234°C,” Iron Steel 8 (6), 237–249 (1981).
3. L. A. Petrov, “Production of Iron-Ore Superconcentrates
and Their Use,” in Theory and Practice of Direct Iron
Production (Nauka, Moscow, 1986), pp. 16–19.
4. B. Delmon, Introduction la Cinetique Heterogene (Tech-
nip, Paris, 1969; Mir, Moscow, 1972).
RUSSIAN METALLURGY (METALLY) Vol. 2007 No. 4
275
5. P. A. Kozlov, Waelz Process (Ruda i Metally, Moscow,
2002) [in Russian].
6. B. F. Kulikov, V. V. Zuev, and I. A. Vainshenker, Miner-
alogical Handbook of a Dresser Engineer (Nedra, Len-
ingrad, 1978).
7. V. Ya. Zaitsev and E. V. Margulis Metallurgy of Lead and
Zinc (Metallurgiya, Moscow, 1985) [in Russian].
8. N. D. Topor, L. P. Ogorodnikova, and L. V. Mel’chakova,
Thermal Analysis of Minerals and Inorganic Com-
pounds (MGU, Moscow, 1987) [in Russian].
9. O. A. Teplov, “Gas Permeability of Sintering Porous
Randomly Packed Media,” Izv. Ross. Akad. Nauk, Ser.
Met., No. 4, 28–33 (2003) [Russian Metallurgy (Met-
ally), No. 4, 313–317 (2003)].