Inert Anode for Al Production

in

ge R&in 1000ia

a r t i c l e i n f o

Article history:Received 23 February 2014Received in revised form 27 April 2014Accepted 28 April 2014Available online 9 May 2014

a b s t r a c t

12Al2O3 34C Al

34CO2 1

This means that over tens of millions tons of coke will be consumedannually and this number keeps increasing with the rising demandfor aluminum on a global scale [4,5]. In addition, because carbon

ert anodes, whichce carbonith inert a

2Al2O3 4Al 3O2

By comparing with Eqs. (1) and (2), theoretically, the product at theanode will be O2 in place of CO2, which will largely eliminate carbonemissions during this production process. Likewise, the emissions ofother greenhouse gases, such as most of the CF4 and C2F6 [6], PAH(from the combustion of the coal tar pitch content in the carbonanode) [7,8], and SO2 will be eliminated as well. Moreover, the u-orides emitted along with the dust and heat will be considerably

Corresponding authors. Address: School of Materials Science and Engineering,Harbin Institute of Technology, Harbin 150001, China (N. Xie). Tel.: +86 45186412133; fax: +86 451 86414392.

E-mail addresses: [email protected] (N. Xie), [email protected] (W.Z. Shao).

Journal of Alloys and Compounds 610 (2014) 214223

Contents lists availab

Journal of Alloys a

.e land other greenhouse gas emissions [13]. Currently, the anodematerial for this electrolysis process is made of carbon. By usingthe carbon anode, the cell reaction is:

there is signicant interest in the utilization of inare non-consumable during electrolysis, to replain the electrolytic cells. If the cells are equipped wthe cell reaction becomes:http://dx.doi.org/10.1016/j.jallcom.2014.04.1980925-8388/ 2014 Elsevier B.V. All rights reserved.anodesnodes,

21. Introduction

Sustainability has become an increasingly important character-istic in most of the industrial engineering elds. Aluminum pro-duction by the HallHroult method is an energy-intensiveprocess that accounts for a globally signicant amount of CO2

anodes are degraded and replaced frequently, the electrolyte cellsare not able to be covered tightly. This requires a large increase inenergy consumption to maintain the temperature of the electrolyticcells.

With the increasing energy costs and heightened concernsabout the environmental footprint of the HallHroult process,Keywords:Inert anodeCermetAluminum productionGeometrical structureSustainableAs an energy-intensive process, aluminum production by the HallHroult method accounts forsignicant emissions of CO2 and some toxic greenhouse gases. The utilization of an inert anode in placeof a carbon anode was considered as a revolutionary technique to solve most of the current environmen-tal problems resulting from the HallHroult process. However, the critical property requirements of theinert anode materials signicantly limit the application of this technology. In light of the higher demandfor aluminum alloys than for pure aluminum, a partially inert anode was designed to produce aluminumalloys in a more sustainable way. Here, Cu2O/Cu cermet was chosen as the material of interest. The ther-mal corrosion behavior of Cu2O/Cu was investigated in Na3AlF6CaF2Al2O3 electrolyte at 960 C to elu-cidate the corrosion mechanisms of this type of partially inert anode for the production of aluminum oraluminum alloys. Furthermore, the effects of the geometrical structure of the Cu phase on the thermalcorrosion behavior of Cu2O/Cu cermet in the electrolyte were investigated as well. The thermal corrosionrate was evaluated by the weight loss method and the results show that the samples prepared withbranch-like Cu have higher thermal corrosion rate than those prepared with spherical Cu, and the corro-sion rate increases with decreasing size and increasing lling content of Cu phase. The calculated corro-sion rate was about 1.57.2 mg/cm2 h (1.89 cm/y) in the current testing procedure. The Cu contents inthe produced aluminum is less than 6.2 wt.%.

2014 Elsevier B.V. All rights reserved.Exploring Cu2O/Cu cermet as a partiallyaluminum in a sustainable way

Li-Chao Feng a,c, Ning Xie b,c,, Wen-Zhu Shao c,, Liana School of Mechanical Engineering, Huaihai Institute of Technology and Jiangsu Provincb School of Transportation Science and Engineering, Harbin Institute of Technology, Harbc School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 15d Physical Chemistry Department, Siberian Federal University, Krasnoyarsk 660041, Russ

journal homepage: wwwert anode to produce

Zhen c, V.V. Ivanov d

D Institute of Marine Resources, Lianyungang 222005, China50090, China1, China

le at ScienceDirect

nd Compounds

sevier .com/locate / ja lcom

reduced, which will greatly reduce the operators exposure to harm-ful environments during the anode changing procedure [9].

Although replacing the carbon anode with an inert anode isalways ideal for aluminum production, their application is still lim-ited due to some major challenges regarding the general propertyrequirements for inert anode materials. First of all, the inert anodemust have electrical and thermal conductivities as high as a con-ductors level; second, the exural and compressive strengths haveto be on the order of 102 MPa; and third, the corrosion rate of thematerials needs to be lower than 1.0 cm/year [10,11]. Unfortu-nately, it is still a great challenge to materials scientists to developa new material that can satisfy all these requirements.

So far, research on inert anode development is focused on threetypes of materials: ceramics, metal or metal alloys and ceramicmetal (cermet) composites. Typical ceramic inert anode materialsare Sb2O3, SnO2, CuO, NiFe2O4, NiO, CeO2 and their compounds[1214]. Themain advantage of the ceramic inert anode is its chem-ical stability in the highly corrosive environment at high tempera-tures; the metal or alloy inert anodes, on the other hand, havelower chemical stability but higher thermal and electrical conduc-tivities. Typical metal or alloy inert anode materials are FexCr1x,Ni, NiAl, CuNiFe, CuAlNiFe, NiFe, etc. [1518]. Cermetmaterials, which inherit advantages frombothmetals and ceramics,have been considered as another possible inert anode material foraluminum production [19]. Although the NiFe2O4NiOCu cermetshave been studied for decades, it is still not able to bewidely applieddue to some unsolved critical bottle neck problems.

In cermet materials, the addition of the metal phase in theceramics matrix aims to increase the electrical and thermal con-ductivities without sacricing chemical stability during the elec-trolysis process. Due to the chemical instability of the metalphase, its dosage should be minimized in order to minimize thecorrosion rate. The corrosion behavior of cermet materials as inertanodes for aluminum production has been investigated exten-sively. Three main corrosion mechanisms were introduced by for-mer studies: the chemical dissolution of oxides [20], the thermalreductive reaction with aluminum [21] and the metal phase disso-lution [22].

The studies on the chemical dissolution of oxides were focusedon the compounds of Ni and Fe. DeYoung [20] and Diep [23] inves-tigated the relationships between the solubility of Fe2O3, NiO andNiFe2O4 in NaFAlF3CaF2Al2O3 electrolyte. It was demonstratedthat the following reaction occurred:

3Fe2O3 2AlF3 6xNaF 6NaxFeOF1x Al2O3 3Although the solubility of aluminum in the electrolyte is limited, itwas also demonstrated that [22] the reaction between the alumi-num and the oxides is inevitable, and the reaction could be writtenas:

3MOn 2nAl 3M nAl2O3 4Dewing [21] gave an empirical relationship between the tempera-ture and the aluminum concentration in the electrolyte:

L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 214223 215Fig. 1. SEM morphologies of the (a) Cu2O, (b) branch-like Cu, (c) spherical Cu with size of 38 lm, and (d) 75 lm.

CAl 0:31759 0:01849CR 0:000344T

0:00407CaF2 0:000096CaF22 0:022853 Al2O32

5

where CAl is the mass concentration of the dissolved aluminum inthe electrolyte, [x] is the mass concentration of the compounds,CR is the molar ratio of NaF and total AlF3, and T is the temperaturein Celsius of the electrolyte. The metal phase dissolution phenome-non was investigated by Tarcy [22]. It was found that the metalphase, nickel, was removed from the surface of the anode duringthe electrolysis procedure, which led to a remarkable decrease ofthe electrical conductivity of the anode. Meanwhile, Windisch[24] found that the dissolution process of copper was different fromthat of nickel. The copper, instead of being dissolved directly, wasoxidized rst and subsequently dissolved by the electrolyte as cop-per oxides.

Due to the fact that the metal phase is the key component of thecermet to maintain a high electrical conductivity for the inertanode, the introduction of the corresponding metal impurities intothe nal aluminum is inevitable. However, thanks to the large mar-ket demands for aluminum alloys [4], the electrolytic cell equippedwith partially inert anode could be considered as a potential way toproduce aluminum alloys, which have much wider applicationsthan the pure aluminum. The International Aluminum Institute(IAI) reported that the ratio of secondary aluminum to primary alu-minum was only 17% in 1960, while it had increased to 33% in2006. IAI claimed that this value will be as high as 40% after

goal; however, the demands for AlNi or AlFe alloys are muchsmaller than for AlCu, AlSi, or AlZn alloys, which signicantlylimited the possibility of their widespread application. Copper,one of the major metals in cermet materials, has been proven tomaintain the high electrical and thermal conductivities. Unlike ironor nickel, copper acts as the main alloying element in most alumi-num alloys (from series of 1xxx to 7xxx). As a result, Cu2O/Cu cer-met could be considered as a candidate partially inert anodematerial to produce aluminum alloys due to Cu being the only ele-ment introduced into the product metal. Although the mechanicaland physical properties have been extensively investigated [2630], the thermal corrosion behavior of Cu2O/Cu cermet in Na3AlF6CaF2Al2O3 electrolyte at high temperatures (960 C) has neverbeen reported, and the high temperature corrosive mechanism isstill not clear.

In addition, although some light metal alloys have been suc-

216 L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 2142232040 [25]. In light of the tough challenge to develop a fully inertanode material, the investigation of a partially inert anode materialhas become a reasonable approach to mitigate the current environ-mental problems created during the production of aluminum andits alloys.

FeNi based cermets have been investigated as fully inert anodematerials for decades and showed a great potential to achieve the

Table 1Porosities of the Cu2O/Cu cermets prepared with various contents and structures ofCu phase.

Cu content, wt.% Branch like Cu 38 lm spherical Cu 75 lm spherical Cu

10 2.45 2.40 2.6115 2.47 2.58 2.3920 2.60 2.68 2.0725 2.68 2.55 2.5130 2.22 2.58 2.5735 2.14 2.26 2.64Fig. 2. Schematic illustration of the tcessfully prepared in molten salt system [31,32], and the corrosionbehaviors of the alloys in molten salt environment under oxidizingatmospheres have been studied [33,34], little work has beenreported on the preparation of aluminum alloys in a molten saltsystem. Therefore, it is necessary to develop a method for the alu-minum alloys preparation in Na3AlF6CaF2Al2O3 molten salt elec-trolyte at high temperatures (800960 C) and elucidate theirthermal corrosion behavior. Here, the thermal corrosion behaviorof Cu2O/Cu cermet was studied to elucidate the corrosion mecha-nisms of this type of inert anode for the production of aluminumor aluminum/copper alloys. Meanwhile, the effects of the geomet-rical structure of the copper phase on the thermal corrosion behav-ior of the Cu2O/Cu cermet in the electrolyte have been investigatedas well to help optimize the overall performance of this inert anodematerial.

2. Experimental

The Cu2O/Cu cermets were prepared via hot-pressing technology. The Cu2Opowder and Cu powders with different geometrical structures were directly mixedand then hot pressed. The Cu contents of the samples were 10 wt.%, 15 wt.%,20 wt.%, 25 wt.%, 30 wt.%, and 35 wt.%, respectively. The average particle size ofCu2O was about 10 lm. The Cu powders were branch-like or spherical with diam-eters of 38 lm and 75 lm. Fig. 1 shows the morphology of the raw materials. Part(a) shows the morphology of the Cu2O powder, which was irregular. The morphol-ogy of the branch like Cu particles is shown in Fig. 1(b), while the spherical particleswith diameters of 38 lm and 75 lm are shown in Fig. 1(c and d) respectively. Ascan be seen in this gure, the morphology of the Cu powder was either branch-likeor spherical, while the shape of the Cu2O powder was irregular. To diminish theoxygen from the Cu surfaces, the Cu powders were heated in H2 at 450 C for 2 hbefore preparing the composites.

In the hot pressing process, the Cu powders and Cu2O powder were ball-milledin dehydrated ethyl alcohol for 12 h, and subsequently dried in a vacuum furnace at80 C. After drying, the temperature was raised to 1050 C with a heating rate ofhermal corrosion testing device.

20 C/min and was followed by hot pressing with 25 MPa and a 40 min soak in agraphite mold. The furnace chamber was purged with 1.0 atm of argon gas fromthe start of the hot pressing procedure. The relative densities of the prepared sam-ples were tested according to the Archimedes method, and the porosities of the pre-pared samples are listed in Table 1. As can be seen in this table, the porosities of theprepared samples are all about 2.5%.

The schematic illustration of the thermal oxidation testing device is shown asFig. 2. The samples were immersed in the electrolyte (5 wt.%CaF2 + 5 wt.%Al2O3 +Na2AlF6) which was isolated in a covered graphite crucible. Samples with various

external dimensions were used to test whether the corrosion performances willbe affected by the size of the samples. The testing temperature was 960 C. The cor-rosion rate was evaluated by the weight loss method and was calculated from thefollowing equation:

Vs 2m1 m2S1 S2t 6

where Vs is the corrosion rate (g/cm2 h), m1, m2 and S1, S2 are the mass (g) and sur-face areas (cm2) before and after the corrosion test, respectively, and t is the corro-sion time (h). As samples will lose mass through both dissolution and mechanicalwear, and gain mass through formation, remnant electrolyte was removed fromthe samples before weighing to minimize error. Samples were immersed in moltenNaCl for 60 min, cooled and then ultrasonically cleaned in pure water to minimizeerror.

The phase structures were determined by X-ray diffraction pattern (XRD) on aRigaku D/max-rA X-ray diffractometer with Cu Ka radiation (k = 1.5406 ). The sur-face morphologies of the samples were observed by scanning electron microscopywith energy dispersive spectroscopy (SEM/EDS), performed on Hitachi S-4700 scan-ning electron microscopy. After that, the cross sectional optical microscopy wasobserved on Zeiss-MC80-DX.

The electrolysis experiment was performed in an electrolysis cell with a volumeof nearly 100 cm3 which was placed in a stainless steel container. Cu2O/Cu cermetanodes with a rectangular shape were used for testing. Alumina was added period-ically (manually) into the cell to maintain approximately the saturation level. Short-term (25 h) and long-term (100 h) electrolysis tests were used to measure theelectrochemical properties of the Cu2O/Cu cermets. An Al electrode was used asthe reference anode. The current density was 0.8 A/cm2, and the distance betweenthe anode and the cathode was about 4 cm. The electrolyte was Na3AlF6CaF2Al2O3, and electrolysis temperature was 960 C.

3. Results and discussion

The samples immersed in the electrolyte for 16 h have beendemonstrated their thermal stability in light of the minimal

Fig. 3. Macroscopic appearance of the Cu2O/Cu cermets with different Cu contentafter corrosion tests.

L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 214223 217Fig. 4. Cross sectional optical microstructures of Cu2O20 wt.% Cu cermet prepared with branch-like Cu after corrosion for (a) 2 h; (b) 4 h; (c) 8 h; and (d) 16 h.

variation of the external dimensions of all samples with varioussizes, shown in Fig. 3. The cross sectional optical microstructure(OM) of the samples immersed in the electrolyte for different dura-tions is shown in Fig. 4. As exhibited in part (a), (b), (c) and (d) ofthis gure, the reaction layers are able to be evidently observed inall samples. The thicknesses of the reaction layers are 12 lm,18 lm, 30 lm and 35 lm corresponding to the immersion timeof 2 h, 4 h, 8 h and 16 h, respectively. It increased 30 lm duringthe rst 8 h but only 5 lm during the next 8 h, suggesting thatthe growth rate of the layer is quicker in the initial stage duringthe whole corrosion process. Furthermore, due to the sinteringeffect and the increasing quantity of the CuAlO2, the density ofthe layer increases with increasing immersion time. The quantityof the micropores in the reaction layer reduced signicantly afterbeing immersed for 8 h. In addition, the Cu concentration at theinterface between the reaction layer and the cermet matrix varieswith increasing corrosion time. As demonstrated in part (a), after2 h of corrosion, the distribution of the Cu particles at the interface

image of the interface between the reaction layer and the cermetmatrix of the sample prepared with 20 wt.% of branch-like Cu

Fig. 6. Corrosion rate of Cu2OCu cermets prepared using branched Cu powderswith different Cu contents.

218 L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 214223exhibited little difference from the center of the sample; whilebeing corroded for 4 h, remarkable Cu accumulation at the inter-face area can be observed, shown as part (b). With an even longercorrosion time, shown as part (c) and (d), the content of the Cu par-ticles was sharply decreased and hardly be observed at the inter-face area. The high magnication cross section OM morphologyof the samples further demonstrated the remarkable difference ofthe samples with corrosion time of 4 h and 8 h, shown as Fig. 5.

Fig. 6 gives the corrosion rate as a function of Cu content of thesample prepared with branch-like Cu after being immersed for 8 h.As illustrated in this gure, the corrosion rate increases remarkablywith increasing Cu content. It was about 0.68 103g/cm2 h withCu content of 10 wt.% while it increased to 2.63 103g/cm2 hwith Cu content of 35 wt.%. This is not a surprising result becausethe Cu phase is much less stable than the Cu2O phase during thecorrosion process. The OM cross sectional morphology of the sam-ples with Cu content of 10 wt.% and 35 wt.% immersed in the elec-trolyte for 4 h are shown as part (a) and part (b) in Fig. 7. It can beseen from this gure that the thickness of the reaction layers aresimilar for these two samples, while the porosity of the samplewith 10 wt.% Cu is lower than the one with 35 wt.% Cu, whichresulted in a relatively lower corrosion rate.

Fig. 8 demonstrated the effects of the geometrical structure ofthe Cu phase on the corrosion rate of the cermet materials. It canbe seen from this gure that the sample prepared with branch-likeCu and spherical Cu with size of 75 lm have the highest and thelowest corrosion rate with values of 2.63 and 1.18 mg/cm2 h,Fig. 5. High magnication of the cross sectional optical microstructures of Cu2O/20 wtrespectively, and the sample prepared with spherical Cu with sizeof 38 lm has a medium value of 1.86 mg/cm2 h (the cross sectionaloptical microstructures of these samples are shown in Fig. 9). Com-paring with our former studies, it was found that the effects of thegeometrical structure of the Cu on the electrical or thermal con-ductivities are opposite to that of the corrosion rate. It was demon-strated that for the same Cu content, the samples prepared withbranch-like Cu have higher electrical and thermal conductivitiesthan the ones prepared with spherical Cu, and for those preparedwith spherical Cu, the increasing of the Cu size will lead to decreas-ing electrical and thermal conductivities [2628]. Consequently,another challenging problem yet to be solved is to optimize thegeometrical structure of the Cu phase to achieve the lowest corro-sion rate without sacricing the high electrical and thermalconductivities.

The crystallographic structures of the reaction layer, which wasremoved from the sample prepared with 20 wt.% of branch-like Cuimmersed in the electrolyte for 8 h, have been conrmed by XRDanalysis (Fig. 10). It clearly shows a pure CuAlO2 peak and littleimpurity peaks were observed. The corresponding SEM/EDS analy-sis further demonstrated the chemical compositions of the reactionlayer contain Al, Cu and O (shown in Fig. 11). The element mappingimage is exhibited in Fig. 12. In this gure, part (a) gives the SEM.% Cu cermet prepared with branch like Cu after corrosion for (a) 4 h and (b) 8 h.

u powders after corrosion for 8 h with Cu content of (a) 10 wt.% and (b) 35 wt.%.

nd Compounds 610 (2014) 214223 219Fig. 7. Optical microstructure of Cu2OCu cermets prepared using branched C

L.C. Feng et al. / Journal of Alloys aimmersed in the electrolyte for 8 h. Parts (b) to (g) are the elementmapping image corresponding to Cu, Al, F, Na, O and Ca, respec-tively. As was found in this gure, the concentration of Cu is homo-geneous in the cermet matrix and slightly lower in the reactionlayer; however, the uorine and aluminum were segregated outof the reaction layer and hardly observed in the matrix. Similarto the Cu concentration, the Na, O and Ca were also distributedin the matrix homogeneously and slightly higher at the reactionlayer area.

The corrosion of the Cu2O/Cu cermet in the Na3AlF6CaF2Al2O3electrolyte could be described by twofold mechanisms: the disso-lution of the cermet to form CuAlO2 passive layer and the physicalmigration of the Cu phase into the electrolyte. First, from the ther-modynamic perspective, the reaction between the Cu2O/Cu and theAl2O3 are described as follows:

12Cu2O 12Al2O3 CuAlO2 DG

1200 K 9:26 kJ=mol 7

Cu 12Al2O3 14O2 CuAlO2 DG

1200 K 50:47 kJ=mol 8

Apart from the reactions above, assuming the existence of O2, thefollowing reactions will also occur:

12Cu2O 14O2 CuO DG

1200 K 9:10 kJ=mol 9

CuO Al2O3 CuAl2O4 DG1200 K 9:59 kJ=mol 10

Fig. 8. The effects of the geometrical structure of the Cu phase on the corrosion rateof the Cu2O/Cu cermet.

Fig. 9. Cross sectional optical microstructures of Cu2O35 wt.%Cu cermet withdifferent geometrical structure of Cu after corrosion for 4 h (a) branch like Cu, (b)spherical Cu with size of 38 lm, (c) spherical Cu with size of 75 lm.

occur. In the initial stage of the corrosion, the growth direction of

duration, the dissolution of the Cu particles into the electrolyte willbe largely prohibited which results in the decreasing corrosion rate.

It should be noted that both the CuAlO2 reaction layer and theCu2O matrix will be dissolved during the electrolysis process. How-ever, the CuAlO2 layer has a lower corrosion rate, and therefore sig-nicantly retards the corrosion process.

According to the results obtained above, the yearly thermal cor-rosion thickness of the Cu2O/Cu cermet in the Na3AlF6CaF2Al2O3electrolyte at 960 C was calculated based on the followingequation:

C Vsq 24 365 13

where C is the annual corrosion depth, Vs is the corrosion rate, and qis the density of the samples. Table 2 lists the results of the calcu-lated annual thermal corrosion depth of the samples with differentgeometrical structures of the Cu phase. As a result of the thermalcorrosion rate, the sample prepared with spherical Cu with a diam-eter of 75 lm and branch-like Cu have the lowest and the highestFig. 10. XRD pattern of the reaction layer from the Cu2O20 wt.% Cu cermet

prepared with branch-like Cu after corrosion for 8 h.

220 L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 214223In this gure, the Cu particles that migrated into the pores of theCuAlO2 layer were shown in the yellow ellipses. The red ellipsesthe CuAlO2 crystals is normal to the interfaces, and form rod shapedcrystals (Fig. 13). Due to the increased thickness and the decreasedporosity of the CuAlO2 layer with prolongation of the corrosionCu Al2O3 12O2 CuAl2O4 DG1200 K 59:91 kJ=mol 11

CuAlO2 12Al2O3 14O2 CuAl2O4 DG1200 K 9:43 kJ=mol

12

In the above equations, the O2, whose concentration at the interfacebetween the Cu2O/Cu cermet and the Na3AlF6CaF2Al2O3 electro-lyte is governed by the parameters of the electrolysis process, playsan important role in determining the nal content of the CuAl2O4. Inthis thermal corrosion process, the main reaction product is CuAlO2and little CuAl2O4 was observed; consequently, Eq. (7) representedthe main reaction in this thermal corrosion process. Secondly, dis-lodgement and dissolution of the Cu in the electrolyte may alsoreveal the necking location which represents the dissolving process.

Fig. 11. EDS spectra of the layer on the surface of the Cu2O20wt.%annual corrosion rate with values of 1.53 and 3.42 cm/y (1.18 and2.63 mg/cm2 h), respectively, while the sample prepared withspherical Cu with a diameter of 38 lm has the medium value of2.41 cm/y (1.86 mg/cm2 h).

After the thermal corrosion tests, the short term (25 h) and thelong term (100 h) electrolysis tests were used to evaluate the per-formances of the Cu2O/Cu cermets. The residual current density, Ir,was measured with the reversible potential (2.25 V) of reaction (2),and the polarization resistance, dened as the slope of the poten-tial-current curve (dE/di) in the vicinity of stable potential, appearsto satisfy the following conditions:

Low residual current density: Ir < 50 mA/cm2. Relatively low polarization resistance: dE/di.

Moreover, these cermet anodes also satisfy:

High open circuit potential value: A > 2.0 V; Minimal variation of the external dimensions, and The electrical resistance of the materials did not increase withincreasing electrolysis time.

The long-term electrolysis tests showed that the relatively sta-ble performances of the cermets. The over voltage, dened as thedifference between the electrode voltage under a specic currentdensity and the balance voltage, is higher than that of the Pt anode.The residual current density was

nd CL.C. Feng et al. / Journal of Alloys arelatively high at the early stage of the test. At the beginning of thetest, the Cu2O/Cu anode contacted the electrolyte directly; as aresult, the relatively high corrosion rate will lead to a high residualcurrent density. With the formation of an external protective layer,the residual current density decreased with increase of testingtime. The anode potential was stable up to the critical potentialof 2.75 V. The corrosion rate was determined by the residual cur-rent density, and it was about 1.57.2 mg/cm2 h (1.89 cm/y) inthe current testing procedure. The Cu contents in the produced alu-minum is less than 6.2 wt.%.

Fig. 12. Element mapping of Cu2O/20 wt.%Cu cermet after corrosion for 8 h (a) bompounds 610 (2014) 214223 221The corrosion of the Cu2O/Cu cermet is a dynamic dissolutionprocess. Both the inert CuAlO2 layer and the Cu2O matrix will bedissolved with the development of the electrolysis process. In addi-tion, some of the Cu particles will be dislodged or oxidized thendissolved in the electrolyte; therefore, the 6% of Cu in the nal alloyis not only from the dissolution of the metal Cu phase but also fromthe electrolysis of the dissolved Cu2O and the CuAlO2. Although thechemical composition of the Cu2O/Cu cermet at the surface will besignicantly changed, the composition and microstructure atthe inside part of the bulk Cu2O/Cu cermet will not be changed

ackscattered electron image; (b) Cu; (c) Al; (d) F; (e) Na; (f) O; and (g) Ca.

ndFig. 13. SEMmorphology of the CuAlO2 layer on the surface of the Cu2O20 wt.% Cucermet prepared with branch like Cu after corrosion for 8 h.

Table 2Annual thermal corrosion rate of Cu2O/Cu cermet in Na3AlF6CaF2Al2O3 electrolyteat 960 C.

2222 L.C. Feng et al. / Journal of Alloys asignicantly during the electrolysis process. As a result, the electri-cal resistance of the bulk Cu2O/Cu cermet will not sharply changewith the development of the electrolysis process.

The oxidation behavior of the Cu2O/Cu cermet in the Na3AlF6CaF2Al2O3 electrolyte at high temperatures contains complicatedmechanisms that are challenging the high temperature oxidationtheories. Further studies must be performed to investigate theexisting passivation process.

4. Conclusions

In summary, for mitigating the signicantly negative environ-mental impacts which result from the current aluminum produc-tion process, the Cu2O/Cu cermet was prepared as the potentialinert anode material to prepare aluminum or aluminum/copperalloys. The thermal corrosion behavior of the Cu2O/Cu cermet withdifferent geometrical structures of the Cu phase was investigated.After analysis by SEM/EDS and XRD, it was found that only a pas-sive CuAlO2 layer was formed during the corrosion procedure.The thicknesses and density of the layer increased with increasingcorrosion time, and the rate of thickness and density increase atthe initial stage is higher than the late stage. Furthermore, the ther-mal corrosion rate was evaluated by the weight loss method andthe results show that the thermal corrosion rate of the samplesprepared with branch-like Cu is higher than for those preparedwith spherical Cu, and the corrosion rate increases with decreasingsizes and increasing lling contents of the Cu phase. The thermalcorrosion rate was about 0.68 mg/cm2 h with Cu content of10 wt.% while it increased to 2.63 mg/cm2 h with Cu content of35 wt.%. The sample prepared with branch-like Cu and sphericalCu with size 38 lm, 75 lm have thermal corrosion rate values of

the Cu82Al8Ni5Fe5 alloy in low-temperature aluminium electrolysis, Corros.

[19] Y. Zhu, Y. He, D. Wang, Fe30Ni5NiO alloy as inert anode for low-temperature aluminum electrolysis, JOM 63 (2011) 4549.

Geometrical structures of the Cu phase Corrosion rate, mg/cm h

Branch-like Cu 2.63Spherical Cu38 lm 1.86Spherical Cu75 lm 1.18[20] D.H. Deyoung, Solubilities of oxides for inert anodes in cryolite-based melts,Light Met. 2 (1986) 299307.

[21] E.W. Dewing, The chemistry of the alumina reduction cell, Can. Metall. Q. 30(1991) 153161.

[22] G.P. Tarcy, Corrosion and passivation of cermet inert anodes in cryolite-typeelectrolytes, Light Met. (1986) 309320.

[23] Q.B. Diep, E.W. Dewing, A. Sterten, The solubility of Fe2O3 in cryolite-alumnamelts, Metall. Mater. Trans. B 33 (2002) 140142.Sci. 70 (2013) 194202.[16] S. Helle, M. Pedron, B. Assouli, B. Davis, D. Guay, L. Rou, Structure and high-

temperature oxidation behaviour of CuNiFe alloys prepared by high-energyball milling for application as inert anodes in aluminium electrolysis, Corros.Sci. 52 (2010) 33483355.

[17] G. Goupil, G. Bonnefont, H. Idrissi, D. Guay, L. Rou, Consolidation ofmechanically alloyed CuNiFe material by spark plasma sintering andevaluation as inert anode for aluminum electrolysis, J. Alloys Comp. 580(2013) 256261.

[18] A. Allanore, L. Yin, D.R. Sadoway, A new anode material for oxygen evolution inmolten oxide electrolysis, Nature 497 (2013) 353357.2.63, 1.86 and 1.18 mg/cm2 h, respectively. After the electrolysistests, by using the sample prepared with 25 wt.% of branch likeCu as a reference, the calculated corrosion rate was about 1.57.2 mg/cm2 h in the current testing procedure. The Cu contentsin the produced aluminum is less than 6.2 wt.%.

Acknowledgements

This work was nancially supported by the National NaturalScience Foundation of China (No. 51371072), the Priority AcademicProgram Development of Jiangsu Higher Education Institutions, theLianyungang Scientic Plan-Industrial Program (CG1312), the Lian-yungang Scientic Plan-Industrial Program (CG1204) and the SixTalent Peaks Program of Jiangsu Province (2013-ZBZZ-032).

References

[1] C.A. McMillan, A.K. Gregory, Not all primary aluminum is created equal: lifecycle greenhouse gas emissions from 1990 to 2005, Environ. Sci. Technol. 43(2009) (1990) 15711577.

[2] Constance F. Acton, Paul C. Nordine, Daniel E. Rosner, High voltage pulsing of alaboratory aluminum electrolysis cell, Ind. Eng. Chem. Process Des. Dev. 15(1976) 285290.

[3] L. Cassayre, P. Patrice, C. Pierre, M. Laurent, Properties of low-temperaturemelting electrolytes for the aluminum electrolysis process: a review, J. Chem.Eng. Data 55 (2010) 45494560.

[4] J.M. Cullen, M.A. Julian, Mapping the global ow of aluminum: from liquidaluminum to end-use goods, Environ. Sci. Technol. 47 (2013) 30573064.

[5] G. Liu, E.B. Colton, D.B. Mller, Unearthing potentials for decarbonizing the USaluminum cycle, Environ. Sci. Technol. 45 (2011) 95159522.

[6] D.R. Worton, T.S. William, K.G. Laila, P.S. Keith, M. Patricia, E.O. David, P.H.Stephen, Atmospheric trends and radiative forcings of CF4 and C2F6 inferredfrom rn air, Environ. Sci. Technol. 41 (2007) 21842189.

[7] G.D. Breedveld, P. milien, S.L. Richard, C. Gerard, Sorption characteristics ofpolycyclic aromatic hydrocarbons in aluminum smelter residues, Environ. Sci.Technol. 41 (2007) 25422547.

[8] W. Rong, S. Tao, B. Wang, Y. Yang, C. Lang, Y. Zhang, J. Hu, J. Ma, H. Hung,Sources and pathways of polycyclic aromatic hydrocarbons transported toAlert, the Canadian High Arctic, Environ. Sci. Technol. 44 (2009) 10171022.

[9] H. Kvande, H. Warren, Inert anodes for Al smelters: energy balances andenvironmental impact, JOM 53 (2001) 2933.

[10] D.R. Sadoway, Inert anodes for the HallHeroult cell: the ultimate materialschallenge, JOM 53 (2001) 3435.

[11] R.P. Pawlek, Inert Anodes: An Update, Light Metals-Warrendale-ProceedingsTMS, 2002. 449456.

[12] Z.L. Tian, L.F. Huang, Y.Q. Lai, J. Li, Y.X. Liu, Effect of additive CaO on corrosionresistance of 10NiONiFe2O4 ceramic inert anodes for aluminium electrolysis,Light Met. (2008) 10591063.

[13] Y. Fu, X.J. Zhai, B. Bai, X.S. Zhang, Z.W. Wang, Molten salt electrolysis of SnO2-based inert anode, J Rare Earths 23 (Sp.) (2005) 8992.

[14] E.W. Dewing, G.M. Haarberg, S. Rolseth, L. Rnne, J. Thonstad, N. Aalberg, Thechemistry of solution of CeO2 in cryolite melts, Metall. Mater. Trans. B. 26B(1995) 8186.

[15] A.P. Khramov, V.A. Kovrov, Y.P. Zaikov, V.M. Chumarev, Anodic behaviour of

Compounds 610 (2014) 214223[24] C.F.J. Windisch, S.C. Marschman, Electrochemical polarization studies on Cuand Cu-containing cermet anodes for the aluminum industry, Light Met.(1987) 351355.

[25] K. Nakajima, O. Takeda, T. Miki, K. Matsubae, S. Nakamura, T. Nagasaka,Thermodynamic analysis of contamination by alloying elements in aluminumrecycling, Environ. Sci. Technol. 44 (2010) 55945600.

[26] N. Xie, W.Z. Shao, L. Feng, L. Lv, L. Zhen, Fractal analysis of disorderedconductorinsulator composites with different conductor backbone structuresnear percolation threshold, J. Phys. Chem. C 116 (2012) 1951719525.

[27] W.Z. Shao, N. Xie, L. Zhen, L.C. Feng, Conductivity critical exponents lower thanthe universal value in continuum percolation systems, J. Phys.: Condens.Matter. 20 (2008) 395235.

[28] W.Z. Shao, L.C. Feng, L. Zhen, N. Xie, Thermal expansion behavior of Cu/Cu2Ocermets with different Cu structures, Ceram. Int. 35 (2009) 28032807.

[29] L.C. Feng, W.Z. Shao, L. Zhen, N. Xie, Microstructure and mechanical propertyof Cu2O/Cu cermet prepared by in situ reduction-hot pressing method, Mater.Lett. 62 (2008) 31213123.

[30] W.Z. Shao, N. Xie, Y.C. Li, L. Zhen, L.C. Feng, Electric conductivity andpercolation threshold research of CuCu2O cermet, Trans. Non-Ferro. Metal.Soc. China SP15 (2005). 297241.

[31] M.L. Zhang, Y.D. Yan, Z.Y. Hou, L.A. Fan, Z. Chen, D.X. Tang, An electrochemicalmethod for the preparation of MgLi alloys at low temperature molten saltsystem, J. Alloys Comp. 440 (2007) 362366.

[32] M. Harata, K. Yasuda, H. Yakushiji, T.H. Okabe, Electrochemical production ofAlSc alloy in CaCl2Sc2O3 molten salt, J. Alloys Comp. 474 (2009) 124130.

[33] S.H. Cho, J.M. Hur, C.S. Seo, S.W. Park, High temperature corrosion ofsuperalloys in a molten salt under an oxidizing atmosphere, J. Alloys Comp.452 (2008) 1115.

[34] S.H. Cho, J.M. Hur, C.S. Seo, J.S. Yoon, S.W. Park, Hot corrosion behavior of Ni-base alloys in a molten salt under an oxidizing atmosphere, J. Alloys Comp. 468(2009) 263269.

L.C. Feng et al. / Journal of Alloys and Compounds 610 (2014) 214223 223

Exploring Cu2O/Cu cermet as a partially inert anode to produce aluminum in a sustainable way1 Introduction2 Experimental3 Results and discussion4 ConclusionsAcknowledgementsReferences