Gas–Solid Reaction Kinetics of ZnFe2O4 Formation from 907 to 1100

Article pubs.acs.org/JPCA

Gas−Solid Reaction Kinetics of ZnFe2O4 Formation from 907 to 1100 °C Thomas Suetens, Muxing Guo, Karel Van Acker, and Bart Blanpain* Department of Materials Engineering, KU Leuven, Kasteelpark Arenberg 44, Box 2450, 3001 Heverlee, Belgium ABSTRACT: The reaction kinetics of Zn vapor with Fe3O4 (magnetite) were studied from 907 to 1100 °C using a new experimental setup that only allows contact between the reactants through a gas−solid reaction. Hematite was used to create the reaction pellets. Because of the reducing atmosphere in the setup, a magnetite layer is formed on the outside of the pellet, which in turn reacts with the Zn vapor. After reaction, Zn concentration profiles were measured in the reacted magnetite layer using field-emission gun electron probe microanalysis. The reaction was confirmed to be diffusion-controlled. The effect of both volume and grain-boundary diffusion was observed in each experiment. The temperature dependence of both the volume and grainboundary diffusion coefficients was obtained along with the activation energies of the diffusion coefficients. This study provides crucial information for the development of technologies that are dependent on the reaction. One example is the in-process separation technology for the separation of Zn vapor from electric arc furnace off-gas.

■

INTRODUCTION Feeding galvanized steel scrap to an electric arc furnace (EAF) results in EAF dust containing zinc. This dust is formed inside the off-gas treatment system and consists mainly out of Zn and Fe oxides.1,2 Because of the presence of hazardous impurities such as Pb, Cr, and Cd, the dust is considered a hazardous material and can no longer be landfilled in the USA and Europe without prior treatment. Processing of the dust to recover the Zn has been ongoing for the last 25 years, mainly by carbothermic reduction in Waelz Kiln furnaces.3 A radical alternative to the traditional post-treatment approach is inprocess separation (IPS) as proposed by Ma.4 Through IPS, pure zinc oxide could be recovered directly from the EAF by separating the zinc vapor from slag and iron oxide particles in the furnace off-gas; however, the presence of iron oxide particles in the off-gas and the intake of air in the off-gas ducts were overlooked in the first feasibility calculations.4 Previous work by the authors shows that the separation is more complex than originally suggested. Under current EAF operation conditions, the formation of ZnFe2O4 through the gas−solid reaction 1 is thermodynamically favorable in the off-gas ducts.5 3Zn(g) + 2Fe3O4 (s) + 2O2 (g) = 3ZnFe2O4 (s)

the EAF off-gas. A new experimental setup was designed to ensure the gas−solid nature of the reaction and is documented in this paper. The influence of temperature on the reaction kinetics was studied because this is the most crucial reaction parameter that continuously varies throughout the EAF off-gas treatment system. We are aware that the average Zn vapor pressures in EAF offgas systems are ∼10−3 atm instead of 1 atm as used in our experiments. One can reason that this causes a lower chemical potential of Zn at the Fe-oxide surface in the EAF system than in our experiments, leading to lower diffusion speeds; however, a FactSage calculation indicated that for Fe2O3 reacting with Zn at a constant Zn vapor pressure of 10−2 atm and 10−3 atm (with a small amount of Ar to fill up the rest of the gas phase and at 1000 °C), at equilibrium we would obtain a homogeneous ZnFe-oxide phase with 19 and 12 wt % Zn, respectively. However, the authors already observed fully saturated ZnFe2O4 particles (27 wt %) in real EAFD.5 We conclude that the actual diffusion process is an interdiffusion of precipitated ZnO with magnetite. As long as the conditions of the experiments provide this ZnO saturation on the magnetite surface, the same conditions as in the industrial situation apply.

(1)

■

Any Zn losses from the vapor phase to the solid particle fraction will reduce the efficiency of the IPS technology. To optimally separate zinc from the iron and slag particles, the gas treatment system should be operated with minimal ZnFe2O4 formation. This could be obtained by limiting the reaction time or by operating the off-gas system under gas compositions that are unfavorable for ZnFe2O4 formation. In this study, the kinetics of the ZnFe2O4 formation reaction were determined to better understand the Zn distribution in © 2015 American Chemical Society

EXPERIMENTAL PROCEDURE Experimental Setup. A quartz experiment ampule was designed to contain both a Sn−Zn alloy and a hematite (Fe2O3) pellet, separated by a physical barrier (Figure 1). This

Received: February 8, 2015 Revised: March 15, 2015 Published: April 6, 2015 4718

DOI: 10.1021/acs.jpca.5b01304 J. Phys. Chem. A 2015, 119, 4718−4722

Article

The Journal of Physical Chemistry A

mechanically reduced in size for easy use in experiment tube creation. Because the reaction of Zn with Fe3O4 requires extra oxygen and the only source for oxygen in the ampule would be the Fe3O4 itself, a Fe2O3 pellet was used for the experiments. Preliminary tests quickly showed that the decomposition of Fe2O3 into Fe3O4 and O2 happens faster than the studied reaction with the Zn vapor, and a layer of Fe3O4 was formed around a Fe2O3 core. It is this Fe3O4 layer that reacts with the Zn vapor. The diffusion of cations in magnetite depends strongly on the oxygen fugacity in the system.8 As long as the Fe2O3 core is present, it will act as an oxygen buffer. The Fe2O3 pellets were produced from high-purity (>99%) powder (Sigma-Aldrich). The powder was placed in long cylindrical balloons and pressed into pellets using a cold isostatic press. After cutting the long hematite sticks to pellet size (±1 cm), the pellets were sintered at 950 °C for 6 h under atmospheric conditions. The final pellets have an average diameter and height of 5 and 7 mm, respectively. The porosity of the pellets was reduced by the sintering step down to 3% (as determined using the Archimedes method). One of the flat surfaces of the pellet was polished with 1 μm polishing medium to ensure a smooth reaction surface. Kinetics Experiments. The quartz ampules were placed in a preheated box furnace at reaction temperatures from 907 to 1100 °C, for the duration of 30 to 120 min. Upon reaching the reaction time, the samples were taken out of the furnace and quenched in water to halt the reaction. Measurement of the Zn Concentration Profile. The quenched Fe2O3 pellets were removed from their reaction ampule and imbedded in Technovit 4004, a fast setting epoxy resin. The pellets were then cut along the longitudinal cross section. The newly obtained cross section was prepared by grinding and polishing down to a 1 μm polishing medium and coating it with a 1 nm Pt/Pd layer using a Quorum Q150T S sample coater. The Zn concentration profile in the reaction pellet edge was then measured with a 1 μm step by quantitative WDS using a probe current of 15 nA and an acceleration voltage of 15 kV on a field-emission electron gun microprobe analyzer JEOL JXA-8530F EPMA. Interpreting the Zn Concentration Profiles. The observed Zn concentration profiles strongly resemble the profiles reported by Sabioni et al. for the diffusion of Zn in polycrystalline ceramics.9,10 A typical concentration profile is shown in Figure 2. It shows the Zn content in a pellet that reacted for 90 min at 1000 °C, as measured by EPMA. The profile displays two fundamentally different regimes: volume diffusion at smaller penetration depths and grain boundary diffusion at larger penetration depths. All experiments lead to concentration profiles of this type. Near the surface, the zinc atoms have ample time to diffuse into the grains, and volume diffusion is the dominant diffusion mechanism. Therefore, the profile obeys Ficks law and can be described by the solution of diffusion into a flat semi-infinite surface with a constant surface concentration

Figure 1. Design for the quartz experiment ampule features a bridge in the middle to prevent physical contact between the two reactants. The dimensions are given in millimeters. By varying X and Y, the volume for the Sn−Zn alloy can easily be expanded according to experimental requirements. A and B indicate the two openings in the glasswork that are used to insert the reactants and impose the vacuum inside the ampule.

barrier prevents direct physical contact between the two reactants except through the vapor phase, ensuring that any observed ZnFe2O4 formation is the result of the gas−solid reaction between Zn vapor and magnetite (Reaction 1). The reactants are inserted through opening A. After sealing off opening A, opening B is mounted on a high vacuum pump. The narrow section in this tube facilitates sealing the tube under the vacuum. Both openings are sealed using an oxy-hydrogen torch. Materials. All experiments were performed in vacuumsealed quartz ampules. Any zinc vapor generation will lead to an increase in pressure. Because of the fixed volume, the pressure in the ampule will increase up to the vapor pressure of the zinc vapor source at the reaction temperature. To decouple the zinc vapor pressure from the reaction temperature, three different Sn−Zn alloys were prepared. For one given temperature, two alloys will have a different zinc vapor pressure (pZn). Using FactSage 6.4, the alloy composition that results in a zinc vapor pressure of 1 atm in the reaction ampule volume was calculated for each reaction temperature (Table 1). Table 1. Sn−Zn Alloy Compositions (wt%) to Obtain a pZn of 1 atm at a Given Experiment Temperature, As Calculated with FactSage 6.4 temperature (°C)

Zn

Sn

907 1000 1050 1100

100 27 17 11

0 73 83 89

The amount of alloy used for each experiment was determined by the alloy composition as the total amount of Zn in the ampule was kept constant at 1 g. This large amount of Zn alloy ensured that any losses of Zn vapor due to diffusion were replenished by the alloy without risking depletion. The presence of a constant Zn vapor pressure at the reaction surface corresponds to a Whipple-type reaction experiment.6,7 All Zn alloys were created in the lab from pure Zn and Sn metal (99.99%). The different metal ratios were melted together at 500 °C in quartz tubes sealed off under high vacuum to prevent oxidation. The obtained alloys were then

⎛ x ⎞ ⎟⎟ CZn = CZn,0 + CZn,max × erfc⎜⎜ ⎝ 2 Dvt ⎠

(2)

CZn,0 and CZn,max are the initial Zn concentration in the pellet and the maximum Zn concentration in ZnFe2O4 with values 0 and 27.12 wt %, respectively, x is the penetration depth in 4719

DOI: 10.1021/acs.jpca.5b01304 J. Phys. Chem. A 2015, 119, 4718−4722

Article

The Journal of Physical Chemistry A

Figure 2. Zinc diffusion profile measured in the outer (Zn,Fe)Fe2O4 layer of a Fe2O3 pellet after reaction at 1000 °C for 90 min. The erfc function that best describes the zinc profile at short penetration depths is shown by the solid line.

Figure 4. Volume diffusion results.

ampules are at room temperature when entering the furnace, the first minutes of the experiment do not lead to the studied reaction but to the heating of the sample to the reaction temperature. Therefore, the actual reaction time will be shorter than the experiment time. For higher reaction temperatures this theoretical heating time becomes relatively long (>15 min at 1050 °C). Because the reaction will already start at lower temperatures, not all of the zinc present in the sample can be attributed to the reaction at a high temperature. Therefore, the 30 min experiment time measurement was omitted from the diffusion coefficient determination of the 1050 °C experiment run. For the experiments at 1100 °C, a box furnace with a higher thermal mass was used, which resulted in shorter theoretical heating times. All obtained diffusion results can be found in Table 2. The temperature dependence of the diffusion coefficient D can be described by the following Arrhenius equation

meters, t is the reaction time in seconds, and Dv is the volume diffusion coefficient in m2 s−1. Diffusion along the grain boundaries inside the pellet becomes more important at larger penetration depths. The concentration profile becomes almost linear. The exact relation between penetration distance and concentration has the following form log CZn = k × x 6/5

(3)

⎛ 123 × 103 ⎞ D(m 2 s−1) = 1.4 × 10−9 exp⎜ − ⎟ ⎝ R(J mol−1 K−1)T ⎠

(5)

This relation is similar to that of the diffusion of other cations in magnetite with a hematite buffer as reported in literature,8 as shown in Figure 5. For the grain boundary diffusion coefficient, Dgb, the temperature dependence can be described as follows ⎛ 176 × 103 ⎞ Dgbδ(m 3 s−1) = 1.8 × 10−10 exp⎜ − ⎟ ⎝ R(J mol−1 K−1)T ⎠

Figure 3. Logarithm of a zinc concentration profile of Figure 2, set out over x6/5. The k constant can be obtained as the slope from the linear fit shown by the solid line.

(6)

The exact thickness of the grain boundaries in our system is unknown; however, a value of 1 nm is given as the approximated thickness of a grain boundary in literature.7 Even if our grain boundaries are not exactly 1 nm, the difference will not be in the order of magnitudes. This allows us to rewrite eq 6 as

where k is a constant (as shown in Figure 3). This constant is used in the Le Claire relationship11,12 to determine the product Dgbδ: ⎛ 4D ⎞1/2 Dgbδ = 0.66⎜ v ⎟ ( −k)−5/3 ⎝ t ⎠

⎛ 176 × 103 ⎞ Dgb(m 2 s−1) = 1.8 × 10−1 exp⎜ − ⎟ ⎝ R(J mol−1 K−1)T ⎠

(4)

Dgb is the grain boundary diffusion coefficient and δ is the grain boundary width.

■

(7)

Over the studied temperature range, the Dgb/D ratio was around 105 to 106. Regardless of this large difference, the amount of Zn introduced in the spinel through volume diffusion is not negligible. In fact, integration of the concentration profile in Figure 2 reveals that almost 70% of the Zn can be attributed to volume diffusion (area under the erfc function/total area). This number will be even higher for

RESULTS AND DISCUSSION From the data fit near the edge, a 2(Dt)1/2 value is obtained for each experiment. By plotting Dt over t, both the diffusion coefficients (slope) and theoretical heating times (intercepts with the t axis) were found (Figure 4). Because our experiment 4720

DOI: 10.1021/acs.jpca.5b01304 J. Phys. Chem. A 2015, 119, 4718−4722

Article

The Journal of Physical Chemistry A Table 2. Experimental Conditions and the Obtained Corresponding Diffusion Coefficients for the Diffusion of Zn in (Zn,Fe)Fe2O4 T (°C) 907 907 907 907 1000 1000 1000 1000 1050 1050 1050 1050 1100 1100 1100 1100

t (m) 30 60 90 120 30 60 90 120 30 60 90 120 30 60 75 90

2(Dt)1/2 5.41 8.02 10.19 11.56 7.74 12.10 15.38 17.84 10.91 13.98 18.31 21.64 13.08 19.16 22.25 23.97

error on fit

D (m2/s)

intercept (s)

0.03 0.05 0.11 0.04 0.05 0.05 0.16 0.07 0.07 0.09 0.08 0.08 0.41 0.21 0.25 0.25

276 276 276 276 542 542 542 542 1009 1009 1009 1009 308 308 308 308

4.79 4.84 5.07 4.82 1.19 1.20 1.22 1.20 3.77 1.89 1.91 1.89 2.87 2.79 2.95 2.82

× × × × × × × × × × × × × × × ×

−15

10 10−15 10−15 10−15 10−14 10−14 10−14 10−14 10−14 10−14 10−14 10−14 10−14 10−14 10−14 10−14

Dgbδ (m3/s) 6.86 4.69 3.01 2.54 3.66 4.87 9.79 1.02 9.62 1.86 1.88 2.02 3.20 3.25 3.51 3.85

× × × × × × × × × × × × × × × ×

10−19 10−19 10−18 10−18 10−18 10−18 10−18 10−17 10−18 10−17 10−17 10−17 10−17 10−17 10−17 10−17

Dgb/D 1.43 9.70 5.95 5.26 3.07 4.07 8.04 8.52 2.55 9.85 9.84 1.07 1.12 1.17 1.19 1.36

× × × × × × × × × × × × × × × ×

105 104 105 105 105 105 105 105 105 105 105 106 106 106 106 106

The information gathered in this work helps us to better understand the reaction between Zn vapor and magnetite and brings us a step closer to the development of the IPS technology.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +32 (0)16 321216. E-mail: bart.blanpain@mtm. kuleuven.be. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This study was conducted as a part of the research supported by CR3, Center for Resource Recovery and Recycling an Industry/University Cooperative Research Center (I/UCRC) supported by NSF. We thank all center members that have supported this study and Joel Hubick for proofreading the paper. We also gratefully acknowledge support from the Hercules Foundation (project ZW09-09), giving us access to FEG-EPMA.

Figure 5. Position of the diffusion coefficients determined in this work compared with the diffusion of other cations in magnetite at the magnetite−hematite buffer as obtained from literature.8

small spherical particles where the reacting outer shell (volume diffusion controlled) represents the majority of the particle volume.

■

■

ABBREVIATIONS EAF, electric arc furnace; FEG-EPMA, field emission gun electron probe microanalysis; IPS, in-process separation; WDS, wavelength dispersive spectroscopy

CONCLUSIONS A new experimental method to study a chemical reaction was used to determine the kinetics of the gas−solid reaction between Zn vapor and Fe3O4 in the temperature range of 907 to 1100 °C under a constant Zn vapor pressure. • Using a field-emission gun electron probe microanalysis (FEG-EPMA), high-resolution diffusion profiles were obtained that clearly displayed two diffusion regimes. From the profiles, both volume and grain-boundary diffusion coefficients were obtained. • The observed volume diffusion coefficients show a similar temperature dependence compared with the diffusion of other cations in magnetite. • The grain-boundary diffusion product D gb δ was determined using the Le Clair model for Whipple-type experiments. • The activation energies for the reaction were found to be 123 and 176 kJ/mol for volume and grain boundary diffusion, respectively.

■

REFERENCES

(1) Huber, J. C. La Formation Des Poussieres Dans Un Four Electrique Acierie. Ph.D. Thesis, Institut National Polytechnique de Lorraine, 2000. (2) Guézennec, A. G.; Huber, J. C.; Patisson, F.; Sessiecq, P.; Birat, J. P.; Ablitzer, D. Dust Formation In Electric Arc Furnace: Birth Of The Particles. Powder Technol. 2005, 157, 2−11. (3) Suetens, T.; Klaasen, B.; Van Acker, K.; Blanpain, B. Comparison Of Electric Arc Furnace Dust Treatment Technologies Using Exergy Efficiency. J. Cleaner Prod. 2014, 65, 152−167. (4) Ma, N. On In-Process Separation Of Zinc From EAF Dust. In EPD Congress 2011, 2011; John Wiley & Sons: Boston, MA, pp947− 952. (5) Suetens, T.; Guo, M.; Van Acker, K.; Blanpain, B. Formation Of The ZnFe2O4 Phase In An Electric Arc Furnace Off-Gas Treatment System. Journal of Hazardous Materials 2015, 287, 180−187. 4721

DOI: 10.1021/acs.jpca.5b01304 J. Phys. Chem. A 2015, 119, 4718−4722

Article

The Journal of Physical Chemistry A (6) Whipple, R. T. P. Concentration Contours In Grain Boundary Diffusion. Philos. Mag. 1954, 45, 1225−1236. (7) Atkinson, A. Grain-Boundary Diffusion: An Historical Perspective. J. Chem. Soc., Faraday Trans. 1990, 86, 1307−1310. (8) Van Orman, J. A., Crispin, K. L. In Diffusion in Minerals and Melts; Zhang, Y. X., Cherniak, D. J., Eds.; Reviews in Mineralogy & Geochemistry; Mineralogical Soc Amer: Chantilly, VA, 2010; Vol. 72, pp 757−825. (9) Nogueira, M. A. N.; Ferraz, W. B.; Sabioni, A. C. S. Diffusion Of The 65Zn Radiotracer In ZnO Polycrystalline Ceramics. Mater. Res. 2003, 6, 167−171. (10) Sabioni, A. C. S.; Daniel, A. M. J. M.; Ferraz, W. B.; Pais, R. W. D.; Huntz, A.-M.; Jomard, F. Oxygen Diffusion In Bi2O3-Doped ZnO. Mater. Res. (Sao Carlos, Braz.) 2008, 11, 221−225. (11) Le Claire, A. D. The Analysis Of Grain Boundary Diffusion Measurements. Br. J. Appl. Phys. 1963, 14, 351. (12) Philibert, J. Atom Movements: Diffusion and Mass Transport in Solids; Editions de Physique: Les Ulis, France, 1991.

4722

DOI: 10.1021/acs.jpca.5b01304 J. Phys. Chem. A 2015, 119, 4718−4722