Hydrogen Reduction of Zinc and Iron Oxides Containing Mixtures ROGE´RIO NAVARRO C. DE SIQUEIRA, EDUARDO DE ALBUQUERQUE BROCCHI, PAMELA FERNANDES DE OLIVEIRA, and MARCELO SENNA MOTTA Zinc is a metal of significant technological importance and its production from secondary sources has motivated the development of alternative processes, such as the chemical treatment of electrical arc furnace (EAF) dust. Currently, the extraction of zinc from the mentioned residue using a carbon-containing reducing agent is in the process of being established commercially and technically. In the current study, the possibility of reducing zinc from an EAF dust sample through a H2 constant flux in a horizontal oven is studied. The reduction of a synthetic oxide mixture of analogous composition is also investigated. The results indicated that the reduction process is thermodynamically viable for temperatures higher than 1123 K (850 °C), and all zinc metal produced is transferred to the gas stream, enabling its complete separation from iron. The same reaction in the presence of zinc crystals was considered for synthesizing FeZn alloys. However, for the experimental conditions employed, although ZnO reduction was indeed thermodynamically hindered because of the presence of zinc crystals (the metal’s partial pressure was enhanced), the zinc metal’s escape within the gaseous phase could not be effectively avoided. DOI: 10.1007/s11663-013-9951-4 Ó The Minerals, Metals & Materials Society and ASM International 2013

I.

INTRODUCTION

REDUCTION of oxidized mineral concentrates is a well-established extractive process and has been used in the production of iron and several other metals, as tin and tungsten. The reducing agent is normally a carbonbearing material (coke, charcoal, etc.), hydrogen, or aluminum, the latter for the most expensive and refractory metals, for example, niobium.[1] Pyrometallurgical or, to a greater extent, hydrometallurgical processes, can also be applied to recover metals of technological importance from industrial residues such as slags and dusts.[2–5] Moreover, it is thought that the metal’s recovery could be established in the alloy form. In this context, particular combinations of iron and zinc can be very interesting as they form a variety of intermetallic compounds and alloys with relative importance in some industrial applications, especially protective coatings.[6] Therefore, based on economic and environmental perspectives, interest lies on the recovery ROGE´RIO NAVARRO C. DE SIQUEIRA, Assistant Professor, and EDUARDO DE ALBUQUERQUE BROCCHI, Associate Professor, are with the Materials Engineering Institute, PUC-Rio Rua Marqueˆs de Sa˜o Vicente, 225, Ga´vea, Rio de Janeiro, RJ Cep 22451-900, Brazil. Contact e-mail: [email protected] PAMELA FERNANDES DE OLIVEIRA, Undergraduate Chemical Engineering Student, is with the Materials Engineering Institute, PUC-Rio Rua Marqueˆs de Sa˜o Vicente, and also with the Chemical Institute, PUC-Rio - Rua Marqueˆs de Sa˜o Vicente. MARCELO SENNA MOTTA, Materials Engineer, formerly with the Materials Engineering Institute, PUC-Rio - Rua Marqueˆs de Sa˜o Vicente, is now with the Basck Ltd., 3 Charles Babbage Road, Cambridge, UK. Manuscript submitted May 13, 2013. Article published online October 17, 2013. 66—VOLUME 45B, FEBRUARY 2014

of Zn and Fe from electrical arc furnace (EAF) dust (mainly Fe2ZnO4), and it is worthwhile to mention that, in thesis, this recovery can be attained either through the separation of these metalls or through their simultaneous reduction leading to FeZn alloy formation. The latter is difficult due to the well-known high vapor pressure of zinc even at relatively low temperatures. This characteristic of zinc hinders production of its alloys in open reaction systems, as the process ideally requires the simultaneous reduction (or at least at somewhat similar rates) of the desired metals. Working at higher temperatures tends to be essential to keep the reaction rates high enough to avoid segregation of a particular metal. Therefore, the alloy formation process needs to be controlled to guarantee the fast reduction of the involved metal oxides, while keeping the metal with highest vapor pressure without significant evaporating. On the other hand, the difference between the vapor pressures of Fe and Zn can be successfully exploited for the separation and recovery of these metals from EAF dust. Industrially, this idea has motivated the development of the Wa¨elz-Kiln process,[7] where zinc is separated from the residue as a gas and precipitates in the oven cold end as ZnO. The process uses petroleum coke as highly efficient reducing agent but it also produces CO2 as a by-product. Other possibilities are given by hydrometallurgical routes,[8,9] which are carried out at low or ambient temperatures, but are also associated with the production of liquid residues. Selective chlorination process was also studied as a possible procedure for Zn recovery from EAF dust.[10] The handling of Cl2 and gaseous chlorinated products may also be related to environmental concerns. So, the hydrogen reduction, which seems to be environmentally attractive, is a METALLURGICAL AND MATERIALS TRANSACTIONS B

subject to be further explored. Indeed, some earlier studies have generated contributions in this field, both in relation to the thermodynamic modeling of the reduction process,[11] as well on what touches the development of experimental studies regarding the viability[11] and kinetic modeling.[12–14] According to these investigations, the reduction of pure Fe2ZnO4 by hydrogen is thermodynamically possible, a fact that was experimentally verified in the temperature range between 773 K and 1073 K (500 °C and 800 °C).[11] On what touches the kinetic studies, considerable volume of data has been already published in the range between 773 K and 1373 K (500 °C and 1100 °C). Significant conversions can be achieved through use of different reducing agents, for example, H2 or CO, demonstrating the viability of separation of zinc and iron. Moreover, depending on the reducing nature of the reaction atmosphere, iron can be produced in different oxidation states, while zinc is totally transported to the gas phase.[12–14] It should indeed be noticed that all mentioned studies were based on synthetic samples (pure Fe2ZnO4 or solid mixtures). Additionally, although all of them converge for the viability of separation of zinc and iron, none have addressed the possibility regarding the formation of Fe-Zn alloys, a material with large applications in the steel industry.[6] In this context, the purposes of the current study are to study the hydrogen reduction of real EAF samples (as part of a cleaner process for the recovery of zinc and iron) and the synthesis of Fe-Zn alloys through employing a synthetic mixture of iron and zinc oxides and a similar methodology as the one already described in literature for synthesizing alloys in other metallic systems.[15,16]

II.

THEORETICAL BACKGROUND

In Figure 1 the partial pressures of zinc and iron present in a gas phase containing N2 in equilibrium with the pure metal. During the calculations, the Gibbs energy models from the SSUB3 and SSOL databases[17] were employed together with software Thermocalc-M for the equilibrium calculations. The same databases were used for all other calculations performed in the study that follows. The comparison between the equilibrium pressures of the two elements shows that the partial pressure of Zn(g) (Figure 1(a)) is orders of magnitude higher than for Fe(g) (Figure 1(b)). Therefore, data on Figure 1 suggest that the process of zinc production through gas–solid reduction reactions should be associated with significant zinc losses to the gas phase. Indeed, the equilibrium calculations conducted for the reduction of Fe2ZnO4 with H2 show this tendency very clearly. The thermodynamic viability of the high-temperature reduction of iron and zinc oxides containing mixtures can be simulated through the calculation of the equilibrium between one mole of Fe2ZnO4 and a gas phase containing H2 at 1143 K (870 °C) (Figure 2(a)) and 1243 K (970 °C) (Figure 2(b)). The gases are considered to be ideal, and all oxide phases are treated METALLURGICAL AND MATERIALS TRANSACTIONS B

as pure substances. The database also includes the possibility of formation of a Fe-Zn alloy, which could be found in the solid (BCC) or liquid state. As shown, the reduction of Fe2ZnO4 with H2 is thermodynamically possible at both temperatures studied. The Fe2ZnO4 phase transforms to Fe3O4 and ZnO. Next, Fe3O4 is converted to FeO, and then reduced to metallic iron. Therefore, based on the thermodynamics of the system (Figures 2 and 3), the concomitant reductions of iron and zinc are possible in the temperature range between 1143 K and 1243 K (870 °C and 970 °C) (Figure 2). Moreover, a rise in temperature means higher losses of zinc to the gaseous state, hindering, in this way, the formation of Fe-Zn alloys (Figure 3).

III.

EXPERIMENTAL METHODOLOGY

A. Fe-Zn Oxides Reduction with H2 Mixtures containing zinc and iron oxides were prepared based on the nitrate thermal decomposition method.[15,16] Known masses of 98 pct pure Zn(NO3)2Æ6H2O (9.25 g) and Fe(NO3)3Æ9H2O (25.135 g) were mixed and heated up to 673 ± 5 K (400 ± 5 °C). After the entire water evaporates and the nitrate precipitates, the thermal decomposition settles in (Eq. [1]). The powder material produced is submitted to a further heat treatment 1273 ± 10 K (1000 ± 10 °C) for a duration of 10 to 24 hours for stimulating the formation of the spinel Fe2ZnO4—one of the main constituents of the EAF dust.[18,19] FeðNO3 Þ3 ¼ 0:5Fe2 O3 þ 3NO2 ðgÞ þ 0:75O2 ðgÞ ZnðNO3 Þ2 ¼ ZnO þ 2NO2 ðgÞ þ 0:5O2 ðgÞ:

½1

This synthetic oxide mixture was reduced with H2 in a horizontal tubular oven at 1143 ± 5 K (870 ± 5 °C). The atmosphere was controlled by establishing definite fluxes of commercial grade N2 and H2, respectively, equal to 0.11 and 0.18 L/min. During these experiments, N2 was used so as to create an inert atmosphere during successive sample mass determinations—for each weight measurement, the sample was let to come into contact with the N2 atmosphere for a period of 2 minutes. The total mass of oxide sample is measured as a function of time, and the reaction is conducted until it achieved a constant value. The sample was next reinserted in the reaction oven and allowed to react further with H2 for a reaction time of 10 minutes so as to minimize the effects of a possible re-oxidation during the final weight measurement. Finally, the reduced oxide mixture was cooled in N2 to 458 ± 2 K (185 ± 2 °C). Using the same equipment and similar procedures, an EAF dust sample was reduced with H2 with the temperature fixed at 1123 ± 5 K (850 ± 5 °C) and the H2 flux fixed at 0.18 L/min. Ultrapure argon was used rather than N2 at a flux of 0.11 L/min for maintaining the inert atmosphere during successive sample mass determinations and during the cooling of the sample to 345 ± 2 K (72 ± 2 °C). The overall conversion (maximum weight loss) was calculated according to Eq. [2] VOLUME 45B, FEBRUARY 2014—67

Fig. 1—Partial pressures of zinc (a) and iron (b) as a function of temperature.

Fig. 2—Phase number of moles (Np) in equilibrium for 1 mole of Fe2ZnO4 as a function of H2 activity at 1143 K (870 °C) (a) and 1243 K (970 °C) (b).

where mi is the initial sample mass, and mf the final sample mass.   mi  mf DmðpctÞ ¼ 100 : ½2 mi The total conversion, as measured by the amount of metallic iron expected, was calculated according to Eq. [3].    mexp  mFe;t ; ½3 Dr ðpctÞ ¼ 100 1  mFe;t where mFe,t stands for the theoretical mass calculated based on the stoichiometry of the individual reduction reactions (Eq. [4]), and mexp represents the final constant experimental mass achieved.

68—VOLUME 45B, FEBRUARY 2014

3Fe2 O3 þ H2 ðgÞ ¼ 2Fe3 O4 þ H2 OðgÞ Fe3 O4 þ H2 ðgÞ ¼ 3FeO þ H2 OðgÞ ZnO þ H2 ðgÞ ¼ Znðl; gÞ þ H2 OðgÞ

½4

Fe2 ZnO4 þ 4H2 ðgÞ ¼ 2Fe þ Znðl; gÞ þ 4H2 OðgÞ: The samples were finally characterized by X-ray diffraction (XRD) for determining the phases present and their mass fractions. Two different diffractometers were used: (i) SIEMENS-D5000, which works with powdered samples in the Bragg–Brentano geometry, Cu-Ka source, and graphite monochromator; and (ii) Paralytical model Xpert-Pro, which also works with powdered samples in the Bragg–Brentano geometry and Cu-Ka source with a multipoint XCelerator detector and graphite monochromator. For the additional characterization

METALLURGICAL AND MATERIALS TRANSACTIONS B

the metal required be added. To ensure that during the whole experiment the zinc crystals are present, a mass higher than the minimum value was used (2.7129 g).

IV.

RESULTS AND DISCUSSION

A. Fe-Zn Reduction with H2

Fig. 3—Moles of zinc present in BCC solid solution at 1243 K and 1143 K (970 °C and 870 °C).

of the EAF dust and reduced samples, energy dispersive spectroscopy (EDS) and scanning electron microscopy (SEM) were used (HITASHI-TM3000, backscattered mode, working at 15 kV). B. Fe-Zn Alloy Synthesis In order to explore the possibility of Fe–Zn alloys production, reduction experiments were accomplished at 973 ± 6 K (700 ± 6 °C) with a sample of the synthetic oxide mixture in the presence of metallic zinc crystals (Figure 4). The volatilization of part of the zinc added would create an atmosphere of higher P(Zn) in the neighborhood of the sample, thereby compensating for the mass loss associated with the transport of gaseous zinc together with the reaction gases. During this experiment, pure H2 at a flux of 0.18 L/min was employed, and ultrapure argon at a flux of 0.11 L/min was used for making the atmosphere inert between successive mass determinations. The amount of metallic zinc required to be added was calculated based on two independent tests developed with pure samples of zinc metal and the synthetic mixture, both being carried out at 973 ± 6 K (700 ± 6 °C). In the case of the test with pure zinc, the sample was exposed to an atmosphere containing ultrapure argon at a flux of 0.11 L/min. The mass of the sample was determined as a function of time, and the volatilization rate was then determined. For the experiment with the pure synthetic mixture, the sample was reacted with pure H2 (0.18 L/min), and ultrapure argon (0.11 L/min) was employed for obtaining an inert atmosphere. The sample mass as a function of time was then recorded and the reduction rate determined. Therefore, based on the knowledge of the synthetic oxide mixture mass to be reduced (0.2001 g) and its reduction rate determined as described before, the expected reaction time was determined. This information was used together with the volatilization rate of pure zinc for the calculation of the minimum amount of METALLURGICAL AND MATERIALS TRANSACTIONS B

The compositions of the iron and zinc oxides mixture used for this part of the study can be seen in Figure 5(a). The sample is mostly composed of the oxide Fe2ZnO4 (94.02 pct) with small amounts of Fe2O3 and ZnO. This is a typical result of a mixture containing oxides coformed obtained by a chemical decomposition method, whereas the precipitation of the oxides occur simultaneously, thus avoiding their segregation.[15,16] Similarly, the composition of the EAF dust sample used is presented in Figure 5(b). The major difference between the two samples is the presence of Fe2O3 in the synthetic sample and the absence of this oxide together with a significant amount of Fe3O4 in the EAF sample. Moreover, the presence of other elements besides Fe, Zn, and O in the EAF dust sample should explain the presence of two peaks located between the Bragg angles 25 deg and 30 deg not described during the Rietveld modeling (Figure 5(b)). This sample was then analyzed by SEM–EDS confirming the presence of significant amounts of Si, Ca, Mg, and Mn (Table I). The synthetic oxide mixture was reacted [1143 K (870 °C), 200 minutes] in pure H2 atmosphere according to the procedure explained on Section III. The maximum weight loss reached, as determined from Eq. [2], was equal to 48.7 pct. The composition of the solid reaction product was obtained by XRD (Figure 6(a)). Interestingly, significant amounts of FeO and Fe3O4 are present in the sample, with FeO being the predominant iron oxide. The presence of Fe3O4 can be explained by the re-oxidation of the sample after it has been removed from the reactor, either during the last weight measurement or after cooling. In the case of FeO, its presence in the final sample can be explained by the reaction with the H2O contained in the commercial N2 at the reaction temperature imposed. It is well known that the reduction of iron at temperatures lower than 853 K (580 °C) should proceed from Fe3O4 directly to Fe, but at higher temperatures, FeO should be first formed.[20] The former situation is associated with the re-oxidation of the sample once removed from the reactor, and the second to the in situ oxidation promoted by humidity at the imposed reaction temperature. Using a similar procedure as applied to the synthetic oxides, the reduction of the EAF sample [1123 ± 5 K (850 ± 5 °C), 135 minutes] resulted in a maximum weight loss (Eq. [2]) very similar to the one obtained for the synthetic oxide mixture—47.3 pct. The resulting XRD pattern is presented in Figure 6(b). The presence of residual Fe3O4 can be explained by the reaction between the sample and air after it has been cooled to 343 ± 2 K (70 ± 2 °C) and removed from the reactor. It is interesting to see that a similar amount of Fe3O4 was found to be present on the XRD pattern of the VOLUME 45B, FEBRUARY 2014—69

Fig. 4—Experimental set-up for the reduction experiments in the presence of pure Zn.

Fig. 5—XRD patterns of the synthetic oxide sample (a) and EAF dust sample (b) used for the reduction test.

Table I.

Composition of the EAF Dust Sample Determined Through EDS Microanalysis

Element Fe Zn O Mg Si Al P S Cl K Ca Cr Mn

Weigh (pct) 52.849 19.441 17.901 0.931 1.0018 0.42 0.146 0.453 0.874 0.902 1.444 0.344 3.292

reduced synthetic oxide mixture (Figure 6(a)). Indeed, the samples were removed from the oven at very close temperatures [343 K (70 °C)—synthetic mixture, 453 K (170 °C)—EAF sample]. However, the use of ultrapure argon in replacement of commercial N2 inhibited the EAF sample’s re-oxidation through its reaction with water (one of the reaction products), which, according to what was said before, should be responsible for the 70—VOLUME 45B, FEBRUARY 2014

FeO formation. Besides, it is interesting to note that a higher reaction time was required for the synthetic sample to reach a constant mass. This bears a relation with the fact that before Fe2ZnO4 reduction, its decomposition into ZnO and Fe3O4 must take place. As a result, one more reaction step is required for the process development (Eqs. [5a] to [5d]). Therefore, the sample with the highest initial Fe2ZnO4 mass fraction (synthetic mixture) must require a higher reaction time for its full reduction. The achieved iron conversions calculated according to Eq. [3] are presented on Table II. An analogous behavior among the tested samples is again observed. The lower conversion obtained with the synthetic sample is a result of the re-oxidation stimulated through flowing of commercial N2 before the last sample mass (mexp) was determined. Moreover, the reaction of the sample with the atmosphere during its weighing appears to have a minor effect, as in the case of the EAF sample, whose calculated conversion was equal to 100 pct. This fact reinforces the importance of letting the sample to react with H2 at the reaction temperature for at least 10 minutes after the previous weight determination. The oxides formed through re-oxidation by air are reconverted into metallic iron. Regarding the zinc reduction, there is no notable presence of this element in the metal form, pure or as METALLURGICAL AND MATERIALS TRANSACTIONS B

part of the BCC phase. If present, zinc would cause a distortion of the BCC lattice which would be noted as a peak shift toward lower angles in the XRD patterns in Figure 6. This was further confirmed by the Rietveld analysis of the XRD data, where the metallic phase proved to be composed exclusively of iron atoms. Indeed, the lattice parameters calculated after the implementation of the Rietveld method have proven to be equal to 0.28686 nm (synthetic mixture) and 0.28691 nm (EAF sample), which are very close to the literature reported value—0.28665 nm.[21] The nonidentification of zinc in neither of the samples is consistent with the thermodynamic evaluation previously shown (Section II). At 1143 K (870 °C) and pure H2 atmosphere, the Fe2ZnO4 phase should decompose into Fe3O4 and ZnO. In turn, ZnO would completely reduce to metallic zinc which is then stabilized in the gas phase as Fe3O4 is progressively transformed to become pure Fe (Figure 2). As the reaction system is open, all zinc metal formed is removed from the reactor. According to the thermodynamic consideration of Section II, a possible mechanism for explaining the reduction process is defined by Eq. [5]. ðaÞ Fe2 ZnO4 ¼ 2=3Fe3 O4 þ ZnO þ 1=6O2 ðgÞ; ðbÞ Fe3 O4 þ H2 ðgÞ ¼ 3FeO þ H2 OðgÞ; ðcÞ FeO þ H2 ðgÞ ¼ Fe þ H2 OðgÞ; ðdÞ ZnO þ H2 ðgÞ ¼ ZnðgÞ þ H2 OðgÞ:

½5

Therefore, based on the current results, the recovery of zinc from EAF dusts by H2 reduction could be considered as an environmental-friendly alternative to the already implemented industrial routes which release CO2 or require further treatment of toxic waste. B. Fe-Zn Alloy Synthesis The H2 reduction method could also be explored for the formation of a BCC Fe-Zn alloy phase if zinc vapor can be retained in the sample. Given that EAF dust samples contain other impurity metals (Table I), the method could be stretched further for the formation of more complex alloys and crystalline phases. In Figures 2 and 3, it is shown that a narrow window exists, in which the reduction of Fe2ZnO4 into Fe(s) and Zn(s) is possible, opening the way for Fe-Zn alloy synthesis. Moreover, it can be shown through similar calculations that this window should be larger at lower temperatures. Based on this information, samples of the synthetic oxide mixture (Figure 7) were then reduced in pure H2 at 973 K (700 °C), one of them in the presence of zinc crystals (Figures 8(b) and 11). The quantitative evaluations of the volatilization progresses of pure zinc and zinc content in the synthetic oxide mixture are presented in Figures 8(a) and (b). In both cases, a linear behavior was observed. In each case, according to mechanism defined by Eqs. [5a] to [5d], the global reaction rates can be calculated by Eq. [6], and

Fig. 6—XRD patterns for the reduced EAF (b) dust sample and synthetic oxide mixture sample (a). METALLURGICAL AND MATERIALS TRANSACTIONS B

VOLUME 45B, FEBRUARY 2014—71

are only a function of temperature if in the atmosphere P(H2) is fixed, for example, equal to one, as in the case of the experiments conducted in the current study. dnZnO ¼ kd ðTÞðPH2 Þnd ¼ kd ðTÞ dt dnFe2ZnO4 ¼ ka  kb  kc  kd ðPH2 Þnb þnc þnd dt ¼ ka  kb  kc  kd ðTÞ:

½6

In Eq. [6], ka, kb, kc, and kd are the rate constants for reactions represented by Eqs. [5a] to [5d], respectively. Also, nc, nb, and nd are their respective global orders in relation to H2. Therefore, for a fixed temperature, the reduction rates of both materials should be constant. It can indeed be observed that pure zinc ‘‘vaporized’’ at a rate equal to 1.87 9 102 g/min, while the measured Table II. Sample Synthetic EAF dust

Maximum Iron Conversions for Reduction of the Synthetic and EAF Samples mFe,t (g)

mexp (g)

Conversion (pct)

0.493 0.276

0.541 0.267

90.30 100

rate of mass loss of the synthetic mixture is one order of magnitude lower (1.4 9 103 g/min). This difference could be explained by the fact that the reduction of Fe2ZnO4 should develop in more than one stage. Indeed, by repeating the experiment with the synthetic oxide mixture at the same conditions as used for obtaining the results presented on Figure 8(b) and collecting samples thereof at 4 and 15 minutes (Figures 9(a) and (b)), the composition of the material varies in agreement with the proposed mechanism. After 4 minutes of reaction, the mass fraction of Fe2ZnO4 is lower than the value observed for the fresh sample (Figure 7), which is consistent with the starting of the Fe2ZnO4 decomposition. As a consequence, Fe3O4 and ZnO are generated, resulting in higher mass fractions for these oxides in comparison with the values associated with the nonreacted sample. It is interesting to note that based on the proposed mechanism, no Fe2O3 should be produced during the decomposition process. Indeed, the mass fraction of this oxide is very similar to the one found in the nonreacted sample (Figures 7 and 9(a)). Moreover, no traces of pure iron could be detected in the sample removed from the reactor at 4 minutes, suggesting that the reduction process had not started yet. Therefore, all compositional variations during the first 4 minutes of reaction can only

Fig. 7—XRD pattern of the synthetic oxide mixture used for the reduction experiments at 973 K (700 °C).

Fig. 8—Mass loss as a function of time at 973 K (700 °C): pure Zn sample (a) and synthetic oxide mixture (b). 72—VOLUME 45B, FEBRUARY 2014

METALLURGICAL AND MATERIALS TRANSACTIONS B

Fig. 9—Reduced synthetic oxide mixture samples—4 min (a) and 15 min (b).

Fig. 10—Number of moles of phase (Np) in equilibrium for 1 mole of Fe2ZnO4 as a function of H2 activity at 973 K (700 °C)—Fe2ZnO4 first decomposition (a), BCC solid solution formation (b). METALLURGICAL AND MATERIALS TRANSACTIONS B

VOLUME 45B, FEBRUARY 2014—73

be assigned to the decomposition of Fe2ZnO4. As time evolves, the mass fraction of Fe2ZnO4 reduces and that for ZnO grows (Figure 9(b)), suggesting that the decomposition process develops further. On the other hand, on what touches Fe3O4, the mass fraction measured at 15 minutes is lower than the one observed at 4 minutes, indicating that its reaction with H2 had already begun. Indeed, at 15 minutes, significant con-

Fig. 11—Total mass loss for the reduction of the synthetic oxide mixture with H2 under the presence of zinc crystals.

centrations of FeO and pure iron could be detected, which are, respectively, generated according to the reactions represented by Eqs. [5b] and [5c]. In the case of Fe2O3, the higher mass fraction found can only be explained through the re-oxidation of the sample by atmospheric air, once removed from the reactor at 973 K (700 °C). By employing an analogous methodology as applied for the thermodynamic modeling of Fe2ZnO4 reduction (Section II), but this time fixing temperature equal to 973 K (700 °C), the variations observed for the number of moles of each phase present in equilibrium agree with the mechanism discussed so far. First, Fe2ZnO4 decomposes into Fe3O4 and ZnO (Figure 10(a)). Next, Fe3O4 is converted to pure iron and ZnO to gaseous zinc (Figure 10(b)). During the entire process, no Fe2O3 is formed. In an attempt to synthesize Fe-Zn alloys, a sample of synthetic mixture (Figure 7) was reduced as described on Section III–A. Given the linear rate of volatilization of pure zinc measured (1.87 9 102 g/min), these crystals should form an atmosphere with a constant excess of partial pressure of Zn(g) during the reduction experiments. Indeed, the rate of mass loss during the reduction of the synthetic mixture in the presence of zinc crystals (Figure 11; 1.028 9 103 g/min) was lower than the one measured for the experiment without zinc

Fig. 12—XRD pattern of the synthetic oxide mixture reduced with H2 at 973 K (700 °C) under the presence of metallic zinc.

Fig. 13—Local EDS analysis of a cluster of metallic iron particles. 74—VOLUME 45B, FEBRUARY 2014

METALLURGICAL AND MATERIALS TRANSACTIONS B

addition (Figure 8(b); 1.40 9 103 g/min). However, the similar magnitude of the rate values indicates that the presence of the zinc crystals was not able to effectively block the loss of zinc metal to the gas phase. The retaining of zinc in the final reduced sample was confirmed through the XRD pattern of Figure 12, where a relatively small amount of ZnO can be identified with the rest of the sample being completely reduced to the metallic BCC phase. The presence of ZnO can be explained through the lowering of the thermodynamic driving force for the ZnO reduction reaction (Eq. [5d])—as P(Zn) in the atmosphere in contact with the sample achieves higher values, equilibrium is dislocated in the direction of ZnO regeneration. The Rietveld analysis of the XRD data obtained for the reduced sample (Figure 12) confirmed that the metallic phase is composed only of BCC iron, with a lattice parameter being equal to 0.286901 nm, which is very similar to the literature value of 0.28665 nm.[21] The absence of zinc in the metallic phase was further evidenced through the SEM–EDS results presented in Figure 13. Kinetically speaking, Fe-Zn alloys formation will depend on the balance between the ZnO reduction rate, the subsequent volatilization of metallic zinc, and the zinc and iron interaction to form the desired solid solution. It is thought that the two first steps develop faster than the third one, making impossible, under the studied experimental conditions, the Fe-Zn alloy formation, which is supported by XRD and SEM results (Figures 12 and 13).

V.

CONCLUSIONS

The reduction experiments conducted with synthetic Fe and Zn oxides mixture and EAF dust samples enabled iron conversions to be greater than 90 pct, while zinc was completely reduced, attesting the viability of the desired reactions in the temperature range between 1123 K and 1143 K (850 °C and 870 °C). It is in accordance with the results obtained during the thermodynamic evaluation. However, at the imposed temperatures, zinc was removed from the sample and transported away by the gas phase. This explains its absence in the metallic phase formed, lattice parameter of which was in all cases, indeed, very similar to the one published in the literature for pure iron (0.28665 nm), from EAF—0.28691 nm and from synthetic mixture—0.28686 nm. In relation to the EAF dust sample, the results indicate that zinc can be selectively extracted through reaction with H2 at temperatures in the range between 1123 K and 1143 K (850 °C and 870 °C). The high volatility of zinc metal, although interesting for separating it from the iron contained in the initial oxide mixture, brings difficulties for the production of Fe-Zn alloys. The higher partial pressure of zinc achieved during the reduction experiment in the presence of zinc crystals makes the thermodynamic driving

METALLURGICAL AND MATERIALS TRANSACTIONS B

force for the ZnO reduction lower. However, the XRD data indicated that no zinc was present in the metallic phase, suggesting that the higher P(Zn) value around the samples atmosphere was not able to compensate for the transport of the volatilized metal, thereby making the alloy formation impossible at the implemented experimental conditions. These results were supported by the XRD data which indicated the iron presence in the reduced material (0.28690-nm lattice parameter very close to the literature value of 0.28665 nm for pure iron) and through SEM/EDS observation of pure iron clusters. Based on the three steps suggested, the Fe-Zn alloy formation could not be attained.

ACKNOWLEDGMENTS The authors are especially grateful to the X-Ray Diffraction and Crystallography Laboratory of CBPF (Brazilian Center of Physics research) for the XRD analysis of the produced samples and also to CNPq, FAPERJ, and CAPES for financial support and scholarships.

REFERENCES 1. H. Fahti: Handbook of Pyrometallurgy, 2nd ed., Metallurgie Extractive, Quebec, 2002. 2. S. Huiting and E. Forberg: Waste Manag., 2002, vol. 23, pp. 933– 49. 3. I. Gaballah and E. Allain: Resour. Conserv. Recycl., 1994, vol. 10, pp. 75–85. 4. M. Kasonde, M. Tshikele, and M. Ilunga: J. Hazard. Mater., 2009, vol. 164, pp. 856–62. 5. S. Anand, R.P. Das, and P.K. Jena: Hydrometallurgy, 1981, vol. 7, pp. 243–52. 6. W. Xiong, Y. Kong, D. Yong, Z.-K. Liu, M. Selleby, and W.-H. Sun: CALPHAD, 2009, vol. 33, pp. 433–40. 7. G. Strohmeier and J.E. Bonestell: Iron Steel Eng., 1996, vol. 73, pp. 87–90. 8. H.-X. Li, Y. Wang, and D.-Q. Cang: J. Centr. South Univ. Technol., 2010, vol. 17 (5), pp. 967–71. 9. R.A. Shawabkeh: Hydrometallurgy, 2010, vol. 104 (1), pp. 61–65. 10. C.A. Pickles: J. Hazard. Mater., 2009, vol. 166, pp. 1030–1042. 11. S. Polsilapa, D.R. Sadedin, and P. Wangyao: High Temp. Mater. Process. Lond., 2012, vol. 30 (6), pp. 587–92. 12. L.F. Tong and P. Hayes: Miner. Process. Extr. Metall. Rev., 2007, vol. 28 (2), pp. 127–57. 13. L.F. Tong: Trans. Inst. Min. Metall. C, 2001, vol. 110, pp. 14–24. 14. L.F. Tong: Miner. Process. Extr. Metall. C, 2001, vol. 110 (3), pp. 123–32. 15. E.A. Brocchi, F.J. Moura, and D.W. de Macedo: Trans. Inst. Min. Metall. C, 2009, vol. 118, pp. 44–48. 16. P.K. Jena, E.A. Brocchi, and M.S. Motta: Metall. Mater. Trans. B, 2004, vol. 35B, pp. 1107–12. 17. P. Spencer and I. Ansara: SGTE Casebook—Thermodynamics at Work, Materials Modeling Series, The Institute of Materials, London, 1996. 18. C.A. Pickles and M. Robert: J. Hazard. Mater., 2010, vol. 179 (1– 3), pp. 309–317. 19. R.L. Nyirenda: Miner. Eng., 1991, vols. 7–11, pp. 1003–1025. 20. Y.K. Rao: Stoichiometry and Thermodynamics of Metallurgical Processes, Cambridge University Press, London, 1985. 21. R. Kohlhaas, P. Duenner, and N. Schmitz-Pranghe: Z. Angew. Phys., 1967, vol. 23, pp. 245–49.

VOLUME 45B, FEBRUARY 2014—75