ZERO-DIRECT-CARBON-EMISSION ALUMINUM PRODUCTION BY SOLID OXIDE MEMBRANE-BASED ELECTROLYSIS PROCESS
Shizhao Su1, Uday Pal*1, 2, Xiaofei Guan1, 2
1Division of Materials Science and Engineering, Boston University, Brookline, MA 02446
2 Department of Mechanical Engineering, Boston University, Boston, MA 02215 Keywords: Solid oxide membrane; Yttria stabilized zirconia; Aluminum production
Abstract
A zero-direct-carbon-emission solid oxide membrane (SOM) electrolysis process was designed and developed to produce high purity aluminum metal. An inert anode assembly containing liquid silver in a one-end-closed YSZ (yttria-stabilized zirconia) membrane tube and LSM (La0.8Sr0.2MnO3-δ)-Inconel inert anode current collector was immersed in an alumina containing molten fluoride flux. A proof-of-concept electrolysis experiment was performed to confirm the aluminum production by depositing liquid aluminum directly on a TiB2 cathode. An improved setup employing liquid aluminum cathode was subsequently used to produce high purity aluminum using the SOM electrolysis process. The membrane stability was confirmed using scanning electron microscopy and energy-dispersive X-ray spectroscopy. High purity aluminum (>99wt%) was produced and collected after the electrolysis.
Introduction
Aluminum is the third most abundant element in the earth’s crust, and the most abundant metallic element. It is an essential material in modern manufacturing. It is known for its low density, high strength, corrosion resistance and high electrical and thermal conductivities. It has been the second largest used industrial metal for the past fifty years. Demand for aluminum production is increasing in recent years, mainly due to the substitution of aluminum in the transportation sector and other lightweight structural applications[1].
The production process of aluminum was invented in 1886 by Charles Martin Hall and Paul L.T.
Héroult (i.e. the Hall-Héroult process) by electrolyzing a cryolite-alumina melt. While the Hall- Héroult process has been significantly improved during the past 100 years, resulting in cheaper aluminum and lesser energy consumption, the current energy consumption in primary aluminum production is still 13kWh/kg of aluminum while the theoretical minimum energy requirement of the process is 5.99kWh/kg[2]. The anode effects of the process results in perfluorocarbon emissions that has a global warming potential 6500 – 9200 times that of CO2. Use of graphite anode in the current process also causes 1.52kg CO2 emission per kg of Al produced (excluding CO2 emission introduced by electricity consumption) [2]. The market potential of aluminum and the environmental impact of the current aluminum production process justify research and development of alternative electrolytic processes for aluminum production that can both reduce the cost and eliminate adverse environment impacts.
Advances in Molten Slags, Fluxes, and Salts: Proceedings of The 10th International Conference on Molten Slags, Fluxes and Salts (MOLTEN16) Edited by: Ramana G. Reddy, Pinakin Chaubal, P. Chris Pistorius, and Uday Pal TMS (The Minerals, Metals & Materials Society), 2016
Solid oxide membrane (SOM) electrolysis is a novel metals extraction technique that is being developed for the production of several energy-intensive metals, such as Mg, Ti, Ta, Yb and Si[3]–
[11]. The SOM electrolysis process features the utilization of an oxygen-ion-conducting membrane, typically made of yttria-stabilized zirconia (YSZ), for directly electrolyzing metal oxides. The desired metal oxide is dissolved in a pre-selected non-consumable molten flux. The YSZ membrane separates an inert anode from the molten flux and a cathode. When applied DC potential between the anode and the cathode exceeds the dissociation potential of the desired metal oxide, the metal is reduced at the cathode and oxygen ions migrate through the YSZ membrane and are oxidized at the anode. A novel LSM (La0.8Sr0.2MnO3- δ)-Inconel inert anode current collector and liquid silver anode are used in SOM electrolysis process for oxygen evolution[5].
Thus, the process also produces pure oxygen gas as a value-added byproduct[7].
In this paper, a novel laboratory-scale process for aluminum production using SOM based electrolysis is presented. Aluminum oxide dissolved in a pre-selected molten fluoride flux, which is chemically compatible with the YSZ membrane, was reduced in a SOM cell at 1200℃. By employing the YSZ membrane based inert-anode assembly, this process produces and collects pure aluminum and oxygen separately. By removing the carbon anode and its manufacture, the direct emission of CO2 and other greenhouse gases are eliminated. The process enables the use of an insulated, corrosion-free lined steel vessel without frozen salt ledge and thus significantly reduces the heat loss and increases the energy efficiency. The electrochemistry of the process was monitored. The aluminum that was produced in the process was characterized by scanning electron microscope (SEM) and energy-dispersive X-ray spectroscopy (EDS). The YSZ membrane stability during the electrolysis was also studied in order to provide insights for long term aluminum production.
Experimental
A. SOM Electrolysis Setup
Figure 1 shows the schematic of a SOM electrolysis setup. The setup consists of a grade 304 stainless steel (SS-304) crucible that was heated to 1200˚C in forming gas (95% Argon-5%H2) to ensure an inert atmosphere. Inside the crucible, 450g of powdered flux (45 wt% MgF2 – 55 wt%
CaF2 containing CaO, Al2O3 and YF3) was used to form the molten electrolyte. 45 wt% MgF2 – 55 wt% CaF2 is a eutectic composition having a melting temperature of 974℃[12]. YF3 was added to prevent yttrium depletion from the YSZ membrane and increase the membrane stability[13], [14]. A one-end-closed 6 mol% YSZ membrane tube separated the flux from 44g of liquid silver enclosed inside the YSZ tube. An LSM-Inconel inert anode current collector was disposed in the YSZ tube and was submerged in the liquid silver anode. The structure of the anode assembly is described in previous research where it was successfully used in Mg-SOM electrolysis cell[5], [7].
An Al2O3 dielectric tube served as a cathode separator as well as a source of Al2O3. Because liquid aluminum is less dense than the molten flux, the aluminum produced in the electrolysis would float on top of the molten flux and be confined within the dielectric tube. This prevents the aluminum from spreading on the surface of the flux and shorting the cell. The Al2O3 dielectric tube was inserted in a stainless steel support welded to the bottom of the stainless steel crucible. The stainless steel support had multiple opening at the side to allow flux to go through the support and contact with the cathode. The powdered flux was pre-melted in the stainless steel crucible with the
Al2O3 tube inserted in the stainless steel support prior to the electrolysis. 2.30g of aluminum shots were added through the Al2O3 tube to form a liquid aluminum pool when melted. The liquid aluminum pool served as the cathode. The cathode current collector consisted of a stainless steel crucible welded to a stainless steel rod. The crucible contained a layer of liquid silver that formed a liquid contact with a piece of TiB2, which is a conductive ceramic that wets but does not alloy with liquid aluminum pool. The liquid silver layer ensured that the contact resistance between the stainless steel crucible and the TiB2 piece was minimized. A boron nitride cover supported by an alumina pin was tightly fit on the TiB2 bar to prevent aluminum from climbing up the TiB2 bar.
During the electrolysis, the TiB2 cathode current collector was inserted in the liquid aluminum but stayed above the molten flux (see the detailed cathode design in Figure 1). Two stainless steel reference electrodes were inserted in the molten flux to monitor the electrochemical behavior of the flux during the electrolysis. The two reference electrodes were electrically isolated by two alumina tubes. One of the alumina tubes also served as a gas stirring tube. To facilitate chemical homogeneity in the molten flux during the experiment, forming gas was passed through the stirring tube at 20 cm3/min. Forming gas was also purged in the reaction chamber at a rate of 1000 cm3/min during the experiment to ensure the SOM cell was in an inert environment.
A proof-of-concept electrolysis experiment using a slightly different setup was performed in order to demonstrate the feasibility of producing aluminum by SOM electrolysis. The proof-of-concept electrolysis setup had no aluminum shots added in the cathode alumina dielectric tube prior to the electrolysis. The alumina was reduced directly at the TiB2 cathode, which was inserted in the molten flux. Such an experimental setup confirmed that all the aluminum collected near the cathode was produced due to the dissociation of aluminum oxide.
Figure 1: Experiment setup of SOM electrolysis
B. Chemicals and Materials
Magnesium fluoride hydrate (MgF2·xH2O, min. 87% MgF2), calcium fluoride (99.5%), calcium oxide (reagent grade), aluminum oxide (99%), yttrium (III) fluoride (99.9%) and aluminum shots (99.5%) were supplied by Alfa Aesar (Ward Hill, MA). The powders were dried for 4 hours at 400℃ to remove moisture and crystallized water. The powders were dry mixed in a ball mill overnight according to the flux stoichiometry. The closed one end YSZ tube (6 mol% Y2O3, 1.90cm o.d., 1.27cm i.d., 60.96cm length) was supplied by McDanel (Beaver Falls, PA). The alumina stirring tubes and the cathode alumina sleeve were supplied by CoorsTek (Golden, CO).
The forming gas was supplied by Airgas (Billerica, MA). The boron nitride cover was machined from a boron nitride plate supplied by Saint-Gobain (Malvern, PA). The titanium diboride was provided by Infinium Inc. (Natick, MA) and machined by wire-cut electrical discharge machining (EDM) to the desired shape.
C. Electrochemical Characterization of Electrolysis Process
The electrochemical characterization of the SOM cell was performed between the LSM-Inconel inert anode current collector and the stainless steel rod cathode current collector. When an applied DC potential exceeded the dissociation potential of Al2O3, Al was produced at the cathode, and O2
evolved at the liquid silver anode. The overall cell reaction is shown as
Al2O3 = 2Al(l) + 3/2O2(g) (1) Potentiodynamic scans (5 mV/s) were performed to determine the dissociation potential of Al2O3
and to identify impurity cations dissociation. The potentiodynamic scans were performed using a Solartron SI 1280B electrochemical measurement system. Electrolysis runs were performed by applying a constant DC potential which exceeds the dissociation potential of Al2O3 between the cathode and the anode. The electrolysis runs were performed using an Agilent Technologies N5743A power source. The anode exit gas (oxygen) during the electrolysis passed through a FMA- 4305 digital flow meter (OMEGA Engineering) that measures its flow rate.
The electronic transference number of the flux was measured using the two stainless steel reference electrodes. Previous work has shown that the dissolution of metallic species in the fluoride flux decreases the SOM electrolysis current efficiency and provides a pathway for the applied potential to reduce the ZrO2 at the outer surface of the YSZ membrane[15], [16]. It is therefore important to monitor the electronic current by measuring the electronic transference number during the electrolysis. The electronic and ionic resistances of the flux can be modeled by two resistors in parallel. Therefore, the relationship between the total ohmic resistance (RT(flux)), the ohmic electronic resistance (Re(flux)), and the ohmic ionic resistance (Ri(flux)) of the flux can be expressed by the relation
1
𝑅𝑇(𝑓𝑙𝑢𝑥)=𝑅 1
𝑒(𝑓𝑙𝑢𝑥)+𝑅 1
𝑖(𝑓𝑙𝑢𝑥) (2) Electrochemical impedance spectroscopy (EIS) scans were performed between the two reference electrodes to determine RT(flux). The EIS scans were performed using a Princeton Applied research
20000 to 1 Hz with 20mV amplitude. RT(flux) was obtained from the value of the high-frequency intercept on the real axis of the Nyquist plot [17], [18]. The ohmic electronic resistance, Re(flux), was determined by performing a potentiostatic hold between the two reference electrodes with a constant small electric potential (~0.1V). Re(flux) was calculated by dividing the applied potential by the measured current. Ri(flux) was calculated from RT(flux) and Re(flux) according to Eq.(2). In the flux system, the electronic transference number te(flux) is the ratio of the electronic conductivity to the total conductivity as shown in Eq. (3), where σe is the electronic conductivity, and σtot is the total conductivity.
𝑡𝑒(𝑓𝑙𝑢𝑥)=𝜎𝑒(𝑓𝑙𝑢𝑥)𝜎
(𝑡𝑜𝑡) (3) The electronic conductivity is proportional to the inverse of electronic resistance, and the ionic conductivity is proportional to the inverse of ionic resistance; the cell constant does not change during the experiment, therefore the electronic transference number te(flux) can be calculated using the following equations [19]:
𝑡𝑒(𝑓𝑙𝑢𝑥)=
1 𝑅𝑒(𝑓𝑙𝑢𝑥) 1 𝑅𝑒(𝑓𝑙𝑢𝑥) + 1
𝑅𝑖(𝑓𝑙𝑢𝑥)
(4)
𝑡𝑒(𝑓𝑙𝑢𝑥)= 𝑅𝑖(𝑓𝑙𝑢𝑥)
𝑅𝑒(𝑓𝑙𝑢𝑥)+𝑅𝑖(𝑓𝑙𝑢𝑥) (5) D. Post-experimental Characterization
After the SOM electrolysis experiment, the setup was disassembled for detailed analysis. A piece of flux sample near the cathode containing the aluminum product and a piece of YSZ membrane were mounted in epoxy, polished, spray coated with carbon, and then characterized using a Zeiss supra 55VP field emission SEM. The purity of the aluminum was measured by EDS quantitative spectra analysis. The yttrium and zirconium content in the YSZ membrane were measured by EDS line scan to assess membrane stability after the electrolysis.
Results A. Proof-of-concept Electrolysis Results
The proof-of-concept electrolysis was performed with an applied DC potential of 2.75V between the liquid silver anode and the TiB2 cathode for 24 hours. The electrolysis was stopped and the current-potential relationship is characterized by a potentiodynamic scan between the anode and the cathode shown in Figure 2. The negative current seen in the potentiodynamic scan prior to alumina dissociation is due to reverse reaction of the electrolysis process, This is because the cathodic chamber after electrolysis has a lower O2 chemical potential (due to the presence of aluminum) compared with the anodic chamber (O2 environment). Thus the negative current prior to alumina dissociation is due to high aluminum activity at the cathode. The only dissociation potential identified from the potentiodynamic scan was at 2.10V which appears to correspond to the Nernst potential (2.08V) for the reaction Al2O3 = 2Al(l) + 3/2O2(g) at 1200ºC. This indicates that all impurity cations were dissociated during the first electrolysis. The leakage electronic
current before alumina dissociation was negligible, indicating that the flux was primarily ionic.
Subsequently, electrolysis with an applied potential of 2.75V for 3 hours were performed.
Figure 2: Potentiodynamic scan (5mV/s) after 1st electrolysis
After SOM electrolysis, aluminum was found inside the cathode alumina sleeve as shown in Figure 3. EDS quantitative analysis showed that the product was primarily Al-Mg alloy with 94wt% Al.
MgO activity in the flux is due to the presence of Mg2+ and O2-, and its reduction results in the observed Mg in the aluminum produced.
Figure 3: SEM image of aluminum metal produced in the proof-of-concept electrolysis Figure 4 shows the SEM image and the EDS line scan of the YSZ membrane after the electrolysis.
It is shown from the line scan that the flux did not attack the YSZ membrane during the electrolysis.
The yttrium and zirconium content was uniform along the YSZ membrane. The ratio of yttrium and zirconium was similar to as received YSZ membrane. It is confirmed that the YSZ membrane was chemically and electrochemically stable after 28 hours of electrolysis.
Figure 4: SEM image and EDS line scan of YSZ membrane after electrolysis. (a) Element analysis of EDS line scan. (b) Quantitative element analysis of atomic percentage of Y and Zr B. Electrochemical Characterization of the Electrolysis Process
The SOM electrolysis was performed with the experimental setup shown in Figure 1 employing the liquid aluminum cathode. A PDS was conducted between the inert anode current collector and the stainless steel current collector prior to the electrolysis, and the current-potential relationship is shown in Figure 5. The dissociation potential of the impurity oxides and Al2O3 were identified to be approximately 1.5V and 2.1V, respectively. The leakage current caused by the dissociation of impurity oxides was approximately 0.1A. A pre-electrolysis at the potential of 2V (less than the dissociation potential of Al2O3) was performed to dissociate the impurity oxides and reduce the leakage current. After 1 hour of pre-electrolysis, another PDS was conducted between the inert anode current collector and the stainless steel current collector. The current-potential relationship is also shown in Figure 5. The dissociation potential of Al2O3 was identified to be approximately 2.1V. Negligible leakage current indicates that there was no electronic current passing through the flux.
Figure 5: Potentiodynamic scans (5mV/s) showing the current-potential relationship before pre- electrolysis and after pre-electrolysis
C. SOM Electrolysis
After pre-electrolysis hold and scans, the setup was subjected to three electrolytic holds. The first electrolysis was performed at 3.5V for 2 hours; the second at 4V for 0.5 hours, and the third at 4.3V for 22 hours (see Figure 6). Total charge passed through the cell was 31254.75C, corresponding to production of 2.91g of aluminum assuming 100% current efficiency.
Figure 6: Current-time relationship of the three SOM electrolysis D. Electronic Transference Number of the Flux
Before the first electrolysis, and immediately after the pre-electrolysis and 1st electrolysis, the electronic transference number using the stainless steel reference electrodes were measured. Table I shows the measured resistances and the electronic transference number of the flux. It is seen that the electronic transference number was low and did not increase during electrolysis. This indicates that the flux was primarily ionic during electrolysis, and the electronic conductivity of the flux did not change due to dissolution of the metal.
Table I: The Measured Resistances and the Electronic Transference Number of the Flux Measurement RT(flux) (Ω) Re(flux) (Ω) Ri(flux) (Ω) te(flux)
Initial 0.780 18.186 0.815 0.043
After pre-electrolysis 0.320 11.120 0.329 0.029
After 1st electrolysis 0.284 16.046 0.289 0.018
E. Post-experimental Characterization
The furnace was cooled after the SOM electrolysis experiment, and the setup was disassembled.
The aluminum was mounted in epoxy, polished and characterized with EDS. Figure 7 shows the cross-section of the aluminum produced during electrolysis and the EDS spectra of the cross- section. Quantitative analysis of the EDS spectra shows the purity of Al was >99wt% with less than 1wt% Mg impurity. The liquid aluminum cathode is always in contact with the alumina dielectric tube, therefore the alumina activity at the cathode interface is high. This is expected to lower the cathodic polarization, and MgO reduction. Employing the liquid aluminum cathode also allows any Mg metal reduced during electrolysis to evaporate from the surface of the liquid aluminum. Thus the amount Mg impurity in the Al product was significantly reduced. However, the Mg that evaporated from the cathode was small and could have oxidized, making its detection difficult after electrolysis.
Figure 7: The cross-section of the Al collected near the cathode (left) and its EDS spectra (right) Conclusions
Aluminum has been produced by electrolyzing aluminum oxide dissolved in pre-selected molten fluoride flux using SOM electrolysis process. The flux system (eutectic MgF2 – CaF2 containing CaO, Al2O3 and YF3) was used in the electrolysis process at 1200℃. The dissociation potential of aluminum oxide was identified to be around 2.1V. The production of aluminum was confirmed by a proof-of-concept electrolysis experiment. The YSZ membrane was stable after the electrolysis.
By employing a liquid aluminum cathode, the aluminum produced in the process was with high purity (>99wt%). The electronic transference number of the flux remained small and did not increase during electrolysis indicating no metal dissolution occurred in the flux.
Acknowledgements
This project was financially supported by the Department of Energy (DOE) through ARPA-E REMOTE program. The authors would like to thank Alexander Kithes, Thomas Villalon Jr. and Matthew Mirek for assisting with performing some of the experiments and Dr. Adam Powell of Infinium Inc., Natick MA, for helpful discussions.
References
[1] “Global Nonferrous Metal Industry Overview.” [Online]. Available:
http://www.statista.com/statistics/242777/aluminum-demand-and-supply-growth-rates/.
[2] “US energy requirements for aluminum production,” U.S. Dep. Energy EERE Rep., 2007.
[3] U. B. Pal, D. E. Woolley, and G. B. Kenney, “Emerging SOM technology for the green synthesis of metals from oxides,” JOM, vol. 53, no. 10, pp. 32–35, Oct. 2001.
[4] A. Krishnan, U. B. Pal, and X. G. Lu, “Solid oxide membrane process for magnesium production directly from magnesium oxide,” Metall. Mater. Trans. B, vol. 36, no. 4, pp.
463–473, Aug. 2005.
[5] X. Guan, U. B. Pal, S. Gopalan, and a. C. Powell, “LSM (La0.8Sr0.2MnO3- )-Inconel Inert Anode Current Collector for Solid Oxide Membrane (SOM) Electrolysis,” J.
Electrochem. Soc., vol. 160, no. 11, pp. F1179–F1186, Sep. 2013.
[6] E. S. Gratz, X. Guan, J. D. Milshtein, U. B. Pal, and A. C. Powell, “Mitigating Electronic Current in Molten Flux for the Magnesium SOM Process,” Metall. Mater. Trans. B, vol.
45, no. 4, pp. 1325–1336, Aug. 2014.
[7] X. Guan, U. B. Pal, and A. C. Powell, “Energy-Efficient and Environmentally Friendly Solid Oxide Membrane Electrolysis Process for Magnesium Oxide Reduction: Experiment and Modeling,” Metall. Mater. Trans. E, vol. 1, no. 2, pp. 132–144, Jun. 2014.
[8] M. Suput, R. Delucas, S. Pati, G. Ye, U. Pal, and A. C. Powell IV, “Solid oxide membrane technology for environmentally sound production of titanium,” Miner. Process. Extr.
Metall., vol. 117, no. 2, pp. 118–122, Jun. 2008.
[9] A. Krishnan, X. G. Lu, and U. B. Pal, “Solid Oxide Membrane (SOM) technology for environmentally sound production of tantalum metal and alloys from their oxide sources,”
Scand. J. Metall., vol. 34, no. 5, pp. 293–301, 2005.
[10] Y. Jiang, J. Xu, X. Guan, U. B. Pal, and S. N. Basu, “Production of Silicon by Solid Oxide Membrane-Based Electrolysis Process,” MRS Proc., vol. 1493, pp. MRSF12–1493–E19–
89, Jan. 2013.
[11] X. Guan, S. Su, U. B. Pal, and A. C. Powell, “Periodic Shorting of SOM Cell to Remove Soluble Magnesium in Molten Flux and Improve Faradaic Efficiency,” Metall. Mater.
Trans. B, vol. 45, no. 6, pp. 2138–2144, Dec. 2014.
[12] P. Chartrand and A. D. Pelton, “Thermodynamic evaluation and optimization of the LiF- NaF-KF-MgF2-CaF2 system using the modified quasi-chemical model,” Metall. Mater.
Trans. A, vol. 32, no. 6, pp. 1385–1396, Jun. 2001.
[13] E. S. Gratz, J. D. Milshtein, and U. B. Pal, “Determining Yttria-Stabilized Zirconia (YSZ) Stability in Molten Oxy-Fluoride Flux for the Production of Magnesium with the SOM Process,” J. Am. Ceram. Soc., vol. 96, no. 32300, pp. 3279–3285, Jun. 2013.
[14] J. Xu, B. Lo, Y. Jiang, U. Pal, and S. Basu, “Stability of yttria stabilized zirconia in molten oxy-fluorite flux for the production of silicon with the solid oxide membrane process,” J. Eur. Ceram. Soc., vol. 34, no. 15, pp. 3887–3896, Dec. 2014.
[15] E. Gratz, S. Pati, J. Milshtein, A. Powell, and U. Pal, “Effect of Electronic Current on the Solid Oxide Membrane (SOM) Process for Magnesium Production,” in Electrometallurgy 2012, John Wiley & Sons, Inc., 2012, pp. 111–118.
[16] X. Guan, P. a. Zink, U. B. Pal, and A. C. Powell, “Recycling of Magnesium Alloy Employing Refining and Solid Oxide Membrane (SOM) Electrolysis,” Metall. Mater.
Trans. B, vol. 44, no. 2, pp. 261–271, Apr. 2013.
[17] K. R. Cooper and M. Smith, “Electrical test methods for on-line fuel cell ohmic resistance measurement,” J. Power Sources, vol. 160, no. 2, pp. 1088–1095, 2006.
[18] S. C. Britten and U. B. Pal, “Solid-state amperometric sensor for the In-situ monitoring of slag composition and transport properties,” Metall. Mater. Trans. B, vol. 31, no. 4, pp.
733–753, 2000.
[19] A. V. Virkar, “Theoretical analysis of the role of interfaces in transport through oxygen ion and electron conducting membranes,” J. Power Sources, vol. 147, no. 1–2, pp. 8–31,