Investigation of Anodic Gas Film Behavior in Hall Heroult Cell Using Low Temperature Electrolyte

Transcription

1 Investigation of Anodic Gas Film Behavior in Hall Heroult Cell Using Low Temperature Electrolyte MORSHED ALAM, WILLIAM YANG, KRISHNA MOHANARANGAM, GEOFFREY BROOKS, and YOSRY S. MORSI A novel one-fourth scale low temperature electrolytic model of the Hall Heroult cell was constructed to investigate the electrolytic bubble formation, coalescence, and movement under the horizontal anode surface. The principles of geometric and dynamic similarities were applied in developing this model electrolytic cell for the first time. A 0.28 M aqueous CuSO pct H 2 SO 4 solution was used as the electrolyte. Analogous to the Hall Heroult cell, copper (Cu) was deposited at the cathode and oxygen (O 2 ) bubbles were generated underneath the anode. The bubble generation mechanism, development, coalescence, and detachment underneath the anode were observed using a high speed camera. It was found that electrolytic bubbles were generated uniformly under the entire anode surface and grew through gas diffusion and coalescence. With increasing current density (CD) and anode inclination angle to the horizontal, bubble velocity increased underneath the anode surface. Moreover, the bubble layer thickness and bubble sizes decreased with an increase in anode inclination angle. Bubble coverage under the anode also decreased with increasing anode inclination angle, but was found to be insensitive to the change in CD. Finally, the calculated bubble resistance was found to decrease with increasing anode angle. DOI: /s x Ó The Minerals, Metals & Materials Society and ASM International 2013 I. INTRODUCTION ALUMINUM is produced by the Hall Heroult electrolytic process which was invented independently by Hall and Heroult in In this method, alumina (Al 2 O 3 ) is dissolved in a molten cryolite (Na 3 AlF 6 ) bath at around 1223 K (950 C) where it is reduced to produce liquid aluminum metal and oxygen ions. The liquid aluminum metal is slightly denser than the electrolyte and is continuously deposited at the bottom of the cell, while the oxygen reacts with the carbon anode to form CO 2. The overall cell reaction is 2Al 2 O 3ðsolutionÞ þ 3C ðsþ ¼ 4Al ðlþ þ 3CO 2ðgÞ : ½1Š The gas bubbles induce flow in the cell which plays an important positive role in homogenization of the alumina distribution and the temperature field in the electrolytic bath. Conversely, the gas bubbles increase the ohmic voltage drop underneath the anode surface, which in turn results in higher energy consumption for the smelting process. The phenomenon of bubble formation and sliding underneath the horizontal surface is complex due to the bubble shape, surface tension, and MORSHED ALAM, Postdoctoral Research Fellow, GEOFFREY BROOKS and YOSRY S. MORSI, Professors, are with the Faculty of Engineering and Industrial Sciences, Swinburne University of Technology, Hawthorn, VIC, Australia. Contact mmalam@swin. edu.au WILLIAM YANG and KRISHNA MOHANARANGAM, Research Scientists, are with the CSIRO Process Science and Engineering, Clayton, VIC, Australia. Manuscript submitted February 27, the anode surface characteristics. A number of studies have been carried out in the past on the bubble behavior under the anode surface and its effect on the electrolyte flow. Fortin et al. [1] used a full-scale water model where anodic gas evolution was simulated by passing air through a micro-porous polyethylene plate. The flow rate of air was selected from the current density (CD) and gas evolution correlation (four electrons are necessary to produce 1 mol of CO 2 ) which is 10 kam 2 = 2.71 Lm 2 s 1. The gas bubbles were found to nucleate at the porous sections on the anode surface and undergo spherical growth, lateral spread, mutual impingement, and coalescence to form larger bubble as shown schematically in Figure 1. The bubbles were then rolled along the anode surface and traveled around the anode edge. The measured gas bubble layer thickness was approximately 5 mm for a horizontal anode. The effects of CD, anode cathode distance (ACD), and anode inclination angle on anode coverage, bubble velocity, and gas release frequencies were investigated. An increase in CD increased the bubble size, thickness of the bubble front, average fraction of anode surface covered by bubbles, and the bubble velocity, while an increase in anode inclination was found to decrease these parameters except bubble velocity. The bubble release frequency was found to vary from 0.2 to 3.3 Hz depending on the anode inclination angle and was not influenced by the CD. Xiang-peng et al. [2] reported that the bubble detachment volume decreases and bubble sliding velocity increases with an increase in anode inclination angle, which is in agreement with the Fortin et al. [1] But, the bubble velocity was found to decrease when ACD was

2 Fig. 1 Bubble shape according to Fortin et al. [1] less than 4 cm, which is in contrast with the results of Fortin et al. [1] where it was reported that ACD has no effect on bubble velocity. Solheim and Thonstad [3] observed that bubble size decreases with the addition of i-propanol which inhibits the coalescence. It was reported that the smaller bubble size results in higher accumulated gas volume as well as higher resistivity in the bubble layer. Che et al. [4] reported that the bubble shape changes from ellipsoid to crescent with an increase in gas flow rates. In a later study, Perron et al. [5] also observed two different shapes of the bubbles underneath the anode surface. They attributed the existence of two distinct regimes, creeping bubble and the bubble on a wetting film, under the anode surface to the different shapes of the bubbles as shown in Figure 2. Bubbles a and b are in the creeping zone. In this zone, the buoyancy force is strong enough to overcome the surface force to deform the equilibrium shape of the bubble. The bubble travels along the entire surface maintaining a contact line. As the bubble continues to travel, its nose starts to climb on the liquid and formation of wetting film starts. Aussillous and Quere [6] also reported the formation of a liquid film between the bubble and solid. During the creeping motion, the longer axis lies in the direction of the displacement, while for the wetting film bubble, the longer axis lies perpendicular to the displacement. Das et al. [7] observed that bubbles lose their symmetric shape immediately after the detachment when the sidewall is at a close proximity. In their other studies, [8,9] it was reported that bubble size increases with an increase in liquid surface tension and decreases at a higher anode inclination angle. Shekhar and Evans [10] observed that the bubble layer becomes thinner in the case of a tilted anode, which in turn helps in lowering the bubbling effect. However, they reported the existence of a reverse flow in the lower half of the anode cathode gap in the case of a tilted anode configuration which could bring some oxygen bubbles in contact with liquid aluminum leading to its reoxidation. Solheim et al. [11] observed that liquid velocity under the anode surface decreased when the interpolar distance was less than 3 cm at constant CD. They [11] also calculated liquid velocity through a mathematical model. The results indicated that the circulation velocity increases with higher CD, higher angle of inclination, and reduced bath viscosity. The anode immersion depth was found to have no significant effect on the bath circulation which is in agreement with the observation of Xian-peng et al. [2] Cooksey and Yang [12,13] measured bubble-induced liquid flow in a full-scale water model of an aluminum reduction cell using the particle image velocimetry (PIV) technique. A recirculation zone was detected in both the center and the side channel of the cell. The area of high turbulence was located in the gas plume region near the end of the anode and at the liquid surface. Wang et al. [14] also observed a similar phenomenon through the laser Doppler velocimetry (LDV) technique in their physical modeling study. In the physical modeling studies that have been discussed so far, gas bubbles were generated by injecting air underneath the anode surface, which did not represent the actual bubble formation mechanism inside the Hall Heroult cell. The injected bubble has momentum and generates continuously at the point of injection, whereas the electrolytic bubble has zero momentum at the point of nucleation and can nucleate at any location under the anode surface if the electrolytic reaction takes place. In addition to that, some recent numerical studies [15,16] showed that the effect of magnetohydrodynamic (MHD) forces on the bath circulation cannot be ignored. The calculated bath velocity under the anode surface without the MHD forces was approximately 33 pct lower than that when both the bubble-induced motion and electro-magnetic force (EMF)-induced motion were considered. Consequently, the results obtained from the air water model cannot be applied directly to an electrolytic cell. In an attempt to generate the bubbles electrolytically, low temperature electrolytic models were developed by some researchers in the recent past. [17 19] Qian et al. [17,18] reported that the electrolytic bubbles are smaller compared with the injected bubbles and the anode surface was covered with a foamy layer of tiny bubbles. It was found that at equal CD or an equivalent gas generation ratio, the difference in bubble resistivity measurement was 20 pct for the two cases. However, in their study, bubbles were generated both at the anode and cathode surfaces, which also did not represent the true mechanism of the actual cell. In addition to cold modeling and low temperature electrolytic modeling, a number of bench-scale experimental studies [20 28] were also carried out by the researchers where bubbles were generated electrolytically and the electrochemical reactions were similar to those in an actual cell. Cell voltage fluctuations were measured using different data acquisition techniques during the electrolysis. Bubble-induced resistance and bubble layer thickness were calculated from the recorded cell voltage data. However, in those studies, the anode surface area was too small compared with the actual cells, except for the one by Aaberg et al. [21] As a result, measured bubble size may be different as it is known from the physical modeling that bubbles coalesce during movement under the horizontal anode. Also, it is very difficult to make a visual observation of the bubble growth mechanism due to the opaque electrolytic bath of bench-scale experiments. The first aim of this work was to build a novel low temperature electrolytic model of the Hall Heroult cell by applying the similarity principles discussed by Zhang et al. [29] Proper geometric and dynamic similarity between the model and actual cell is crucial to interpret the results obtained from the model cell. The second aim was to use the model to enhance the current

3 Fig. 2 Regimes of movement of the bubbles. [5] understanding of the electrolytic bubble formation, movement, and detachment characteristics under the horizontal anode surface. II. DESIGN OF ELECTROLYTIC CELL In designing the low temperature electrolytic model, emphasis was on maintaining the geometric and dynamic similarity between the real cell and the low temperature electrolytic model. A 1/4 th scale model of the Hall Heroult cell was built by maintaining complete geometric similarity. In order to maintain the dynamic similarity, five different dimensionless numbers were considered as reported by Zhang et al. [29] These are Inertial force Modified Froude number ¼ Buoyancy force ¼ q g q 2 q l q g gl Table I. Fluid Properties and Dynamic Similarity Analysis of the Present Model Real Cell Present Model Electrolyte cryolite 0.28 M CuSO pct H 2 SO 4 Electrolyte density [32] (kg/m 3 ) Electrolyte surface [33] tension (mn/m) Electrolyte viscosity [32] (kg/ms) Bubble cross-sectional 11 to 13 [25,27] 2to31 diameter before release (mm) Anode length (m) 1.35 [1] 0.35 Modified Froude number Eotvos number 19.3 to to 114 Morton number Modified Weber number¼ Inertial force Surface tension force Modified Reynolds number ¼ Inertial force Viscous force ¼ gq2l ¼q r pffiffiffiffiffiffiffiffi q l q g ql l Buoyancy force Eotvos number ¼ Surface tension force ¼ g q l q g L 2 r Morton number ¼ gl4 l q l q g q 2 : l r3 Here, q is the gas generation rate per unit surface area, g is the acceleration due to gravity, q l is the liquid density, q g is the gas density and r is the surface tension of the surrounding liquid, L is the characteristic dimension, and l l is the viscosity of surrounding liquid. From the literature review, it can be concluded that the buoyancy force and inertial force generated by the evolving gases are the main driving forces for the bubble movement and bath motion. Also, the Eotvos number together with the Morton number is used to characterize the bubble size and shape moving in a surrounding fluid medium. Hence, in the present study, the Modified Froude number, Eotvos number, and Morton number were considered for dynamic similarity analysis and are presented in Table I. Table I shows that the Eotvos number and the Morton number of the model and real cell have similar orders of magnitude. Hence, it can be said that the generated bubble shape in the model cell is likely to be similar to that of a real cell because these dimensionless numbers characterize the shape of bubbles. The bubble cross-sectional diameter before release was used as the characteristic dimension in the Eotvos number equation. Table I also shows that the Modified Froude number of the real cell is one order of magnitude higher than the model cell. This was because the gas generation rate in the present model cell was lower than the desired value. The gas generation rate for the model cell was calculated using Faraday s law as described in Appendix 1. In order to maintain the similar Modified Froude number, the ratio of the gas generation rate between the real cell and model should be 4.3 (described in Appendix 2), which means that the gas generation rate in the model cell should be q real cell ¼ 4:3 q model q model ¼ q real cell 4:3 ¼ 2:71 4:3 ¼ 0:63 Ls 1 m 2 : In the present study, the maximum gas generation rate used was 0:2 Ls 1 m 2 corresponding to the CD 0:3 Acm 2. If we increase the CD, we will get the desired bubble generation rate for our model. An increase in CD

4 increases the bubble release frequency, but the bubble volume at release is not influenced by the CD. [21] The main goal in this experiment was to generate bubbles electrolytically which are comparable with the bubbles of real Hall Heroult cell and then study the bubble formation, movement, and detachment under the horizontal anode. That is why emphasis was given on the Eotvos number and Morton number as these numbers characterize the bubble shapes. A. Experimental Rig In the previous low temperature electrolytic study of Qian et al., [17] bubbles were generated both at the cathode and anode. But, in the actual cell, no bubbles are generated at the cathode. Therefore, a separator was used to isolate the bubbles that generated at the cathode in order to minimize the effect of cathode bubbles on the anode. In the present study, the electrolyte was selected in such a way that no bubble should generate at the cathode. After studying the available aqueous solutions, it was found that CuSO 4 solution is an excellent candidate, and therefore was used in this study. The overall electrolytic reaction is CuSO 4 ðaqþþ2e þ H 2 O ¼ CuðÞþH s 2 SO 4 ðaqþþ 1 = 2O 2 ðgþ: ½2Š Figure 3 shows the experimental setup used in the present study. The inside dimensions of the model cell were mm 3. The anode and cathode dimensions were mm 2 and were placed parallel to each other. These dimensions were decided according to 1/4 th scale geometrical similarity analysis with the real cell. [1] The ACD was fixed at 50 mm. Lead alloy (6 mm thick) and stainless steel (3 mm thick) sheets were used as the anode and cathode, respectively. The anode immersion depth was fixed at 200 mm. The CuSO 4 solution was stored in a tank where it was heated and maintained at 323 K (50 C). The electrolyte was supplied into the model cell through the inlet at the bottom left corner of the cell. When the electrolyte level reached 250 mm from the bottom, it passed through the electrolyte overflow line into the heated tank from where it was recirculated again into the cell. A DC power supply (0 to 500 A) was used to supply current for electrolysis. But, current could not be increased over 125 A (0.3 A/cm 2 ) as the voltage drop was greater than 5 V which is the maximum voltage range for this power supply unit. In the current copper electrolytic process, the oxygen bubbles create acid mist due to its bursting action at the liquid surface which is hazardous for human health. Therefore, a 200-mm flexible reinforced PVC pipe was placed on top of the cell to extract the generated acid mist. As the cathode was opaque, it was impossible to record the bubble movement underneath the anode surface from the perpendicular direction. Therefore, bubble movement was recorded at a certain angle from the horizontal as shown in Figure 3(b). B. Experimental Procedure The electrolyte (0.28 M CuSO pct H 2 SO 4 ) was circulated continuously from the heated tank to the experimental rig to ensure that the temperature of the electrolyte remained around 318 K to 323 K (45 C to 50 C) inside the model cell, which is the requirement for the electrolysis of CuSO 4 solution. The temperature was monitored continuously through a digital thermometer (accuracy ±0.10 K) with a stainless steel probe. The circulation was turned off prior to the commencement of electrolysis so that the liquid velocity did not affect the bubble behavior. Then, the power supply was turned on and set to the desired CD to commence the electrochemical reaction. The experiments were run at different current densities and anode inclination angles to investigate the effect of these parameters on the bubble characteristics. The operating conditions are presented in Table II. The average bubble cross-sectional diameter and thickness (l and d in Figure 12 respectively) of the departing bubbles from the edge of the anode surface were measured using the high speed camera at 200 frames per second from two different locations: (a) Fig. 3 Schematics of experimental setup.

5 Table II. Operating Conditions Experiment No Angle of Inclination (deg) Current Density (A/cm 2 ) Fig. 4 Distorted image of the reference grid. Calculated Gas Generation Rate (Lm 2 s 1 ) perpendicular to the direction of motion of the bubbles and (b) inclined from the horizontal plane. The captured images were then processed using image processing software imagej and GIMP. C. Image Analysis The image distortion, created by the angle, was corrected using the perspective tool of GIMP image analysis software and with the help of a reference grid. At the beginning of the experiment, the reference grid ( mm 2 ) was placed underneath the anode surface. Then, the rig was filled with water. The camera was set in a suitable position to record the image of the reference grid as shown in Figure 4. After that, the grid was taken out and the rig was emptied. The CuSO 4 solution was then circulated from the heated tank and the DC power supply was switched on carefully to start the electrolytic reaction without disturbing the physical location of the camera. The images were recorded at 200 frames per second. Figure 5 shows the image of the reference grid after distortion correction. Each square in the corrected image of the reference grid is mm 2 and 1 mm = 6.55 pixels. Using a similar algorithm, the images at different operating conditions were corrected. After distortion correction, the generated electrolytic bubble diameter underneath the anode surface was Fig. 5 Image after distortion correction. calculated using the pixel information from the reference grid. The bubble velocity was measured from the images after distortion correction using the imagej software. First of all, two reference points were taken and the number of frames required for the bubble front to reach from one reference point to another was counted. Then, the frame numbers were divided by frame rate of recording to calculate the time taken by the bubble front to travel from one reference point to another. After that, the distance between two reference points was divided by the time to determine the velocity of the bubble. The bubble layer thickness was measured from the images recorded at a perpendicular position to the direction of bubble movement. From the bubble diameter measurements, the bubble coverage ratio on a mm 2 section of the anode was evaluated using the following formula: P n pd 2 n i¼1 4 u ¼ ½3Š where u is the bubble coverage and d n is the diameter of n th bubble. Using the value of bubble coverage and bubble layer thickness, the volume of bubble per unit surface area (cm 3 /cm 2 ) at release was calculated using the following equation [21] : V ¼ ud b ½4Š where d b is the bubble layer thickness. Total ohmic resistance of the electrolytic bath can be expressed as [21] R electrolyte with bubble ¼ 1 KA ð D d bþþ d b 1 u where K is the electrical conductivity of the electrolyte, A is the anode surface area, and D is the anode to cathode distance. With no bubbles present, d b =0 and u = 0. Hence, Eq. [5] becomes R electrolyte no bubble ¼ D KA : ½5Š ½6Š

6 Fig. 6 Initial bubble formation under the anode surface at 0.1 and 0.3 A/cm 2 CD and 1-deg angle.

7 Fig. 7 Bubble coalescence behavior 0.2 s apart (bubble motion is from left to right). Table III. Angle Bubble Diameter at Different Anode Inclination Angles Average Bubble Diameter Before Detachment from the Edge (mm) Maximum Bubble Diameter (mm) ± ± ± Using the experimental values of d b and u at different anode inclination angles and current densities, total resistance of the electrolyte was calculated using Eq. [5] at corresponding operating conditions. When no bubble was present, the electrolytic resistance was calculated through Eq. [6]. By subtracting the magnitude of Eq. [6] from Eq. [5], bubble-induced resistance was determined. III. RESULTS AND DISCUSSION Figure 6 shows the different stages of electrolytic bubble formation under the anode surface at two different current densities and 1-deg angle of inclination from the horizontal. After the start of the electrolytic process, the entire underside of the anode was covered by tiny bubbles as shown in Figure 6. Gradually, the immobile bubbles grew due to the gas diffusion and coalescence with the surrounding bubbles. There were no clear areas under the anode as compared to water modeling studies. [14,17,30] Bubbles remained stationary at the nucleation point until the component of the buoyancy force, parallel to the anode surface, was large enough to overcome the surface tension force and drag force. This occurred when the bubbles reached a certain volume. After that, the bubbles detached from the nucleation site and moved along the anode surface. The formation and movement mechanism was similar at both higher and lower current densities. The higher CD

8 Fig. 8 Change in bubble coverage, thickness, and resistance with anode inclination angle. Fig. 9 Change in bubble coverage with CD. Fig. 10 Bubble volume before release underneath the anode surface. Fig. 11 Bubble terminal velocity before detachment. speeds up the process as shown in Figure 6. At t = 4 seconds, the bubbles started to detach from the nucleation point in the case of 0.3 A/cm 2 CD, whereas the bubbles were only growing in the case of 0.1 A/ cm 2 CD. These figures also show that although small bubbles generate under the entire anode surface, the bubble grows bigger only at a limited number of nucleation sites which may depend on the morphology of the anode surface. This is a unique observation from this low temperature electrolytic model which was not observed from the previous water modeling studies. A separate study on this particular issue is required to understand the effect of anode surface properties on bubble characteristics. After detachment from the nucleation sites, the bubbles moved under the anode surface, coalesced with other bubbles along the way, and grew in size as shown in Figures 7(a) through (c). In Figure 7(a), two bubbles, 1 and 2, are traveling under the anode surface from left to right at different speeds.

9 Fig. 12 Bubble contact angle under the anode. After a certain period, coalescence of bubble 1 with bubble 2 resulted in a bigger bubble 3 as shown in Figures 7(b) and (c). Fortin bubbles [1] were not observed in the current study. This might be due to the low angle of inclination and CD used in this study. [31] Table III shows the average bubble diameters at different anode inclination angles. As expected, the bubble diameters were found to decrease with increasing anode inclination angle. This occurred due to the increase in buoyancy force at higher anode inclination angles. At higher buoyancy force, bubbles detach and travel comparatively faster underneath the anode surface and hence get less time to grow by gas diffusion and coalescence. As a result, bubble size decreases at higher anode inclination angles. From the table, it is seen that the value of a standard deviation of the average bubble size is very high, which means that the values of the bubble diameter were spread out over a large range. For a 1-deg anode angle, the range of measured bubble diameters was 2 to 31 mm. Table III also shows the maximum bubble diameter at different anode angles. The measured maximum diameter of the bubble at the 1-deg anode angle was found to be two times larger than that at a 3-deg anode angle. Figure 8 shows that at higher anode inclination angles, bubble coverage under the anode surface decreased, which is in agreement with previous researchers. [1,31] This occurred because bubbles escaped quickly from underneath the anode surface at higher inclination angles. At the horizontal anode condition, the calculated bubble coverage was 0.7 which is in the range of 0.65 to 0.9 as reported by Aaberg et al. [21] However, the bubble coverage was relatively unchanged with increasing CD as shown in Figure 9, which is in contrast with the previous studies, [1,31] but is consistent with Aaberg et al. [21] The measured mean bubble layer thickness in the case of the horizontal anode with no inclination was 0.4 cm which agreed well with the reported values of previous researchers. [21,25] Xue and Oye [25] estimated a maximum bubble layer thickness of 0.4 cm in their study. Aaberg et al. [21] reported that the bubble layer thickness ranges from 0.4 to 0.6 cm. Figure 8 also shows that the bubble layer became thinner at higher inclination angles, which is consistent with the observation of Shekhar and Evans. [10] Using the bubble coverage and layer thickness data, bubble resistance at different inclination angles was calculated using Eqs. [5] and [6] and also plotted in Figure 8. The figure shows that the bubble-induced resistance decreased with increasing anode inclination angle. Figure 10 shows the calculated bubble volume before release underneath the anode surface. When the anode was horizontal, the calculated bubble volume in the present study was 0.29 cm 3 /cm 2 which is in the range of the reported values of Aaberg et al. [21] and shown in Figure 10. One set of data from Aaberg et al. [21] showed very low value compared to the other three sets. They attributed the existence of a slight anode inclination angle as a possible reason for this lower bubble volume prediction. This set of data was in good agreement with the values calculated from the present experimental study at a 1-deg inclined anode as shown in Figure 10. It was observed that the bubble volume underneath the anode surface decreases with higher anode inclination angles, but remained relatively constant with increasing CD. Figure 11 shows the change in bubble terminal velocity before detachment from the anode surface with increasing CD. As expected, the bubble terminal velocity was found to increase with the increase in CD. This occurred because, at higher CD, the bubble generation rate increases underneath the anode surface. As the bubble penetration depth inside the electrolyte is limited, more and more bubbles slide under the anode surface and escape through the anode edge with high velocity if the CD is increased. The figure also shows that bubble terminal velocity increases with the increase of the electrode inclination angle. The component of the buoyancy force parallel to the anode plane increases at a higher inclination angle, which in turn accelerates the gas bubbles underneath the anode surface and the bubbles escape quickly from the edge of the anode. Wettability is an important parameter for investigating the bubble characteristics underneath the anode surface. The notion of wettability is based on the concept of an equilibrium state between the interfacial surface tension of three phases and the existence of an equilibrium contact angle. The contact angle (h) is the angle between the solid surface and the gas liquid interface as shown in Figure 12. The final contact angle of the bubble with the anode surface before departure was measured using the image analysis software imagej. The measured bubble contact angle was found to vary from 115 to 135 deg, which is in good agreement with the results of Xue and Oye [25] which ranged from 110 to 130 deg. IV. CONCLUSIONS A novel 1/4 th scale low temperature electrolytic model of the Hall Heroult cell was developed to investigate the

10 bubble characteristics under the anode surface. Principles of similarities were applied to maintain the geometric and dynamic similarity between the model and the actual cell. To the best of our knowledge, this is the first low temperature electrolytic model of the Hall Heroult cell where geometric and dynamic similarities were maintained. Geometric similarity confirmed that the model is of the same shape as the real cell and dynamic similarity ensured that the flow behavior inside the model is similar to the real cell, both of which are very important to interpret and use the model results. A 0.28 M CuSO pct H 2 SO 4 aqueous solution was used as the electrolyte which deposited Cu at the cathode and produced O 2 bubbles at the horizontal anode during the electrochemical reaction process similar to the phenomenon of the Hall Heroult cell. The behavior of the electrolytically generated bubbles was analyzed through a high speed camera. It was observed that just after the commencement of the electrolytic reaction, the entire anode surface was covered with tiny bubble foams. The immobile bubbles then grew through gas diffusion and coalescence until the buoyancy force was large enough to overcome the surface tension force at which point the bubbles started to slide along the inclined anode surface. The terminal velocity of the bubbles before departure was found to increase at higher current densities and anode inclination angles. The average bubble size, bubble layer thickness, and bubble coverage were found to decrease with an increase in anode inclination angle. As a result, the calculated bubble-induced resistance underneath the anode surface was lower in the case of inclined anode arrangements. The observed contact angle between the anode and the gas bubbles ranged from 115 to 135 deg, which was also in good agreement with the previous studies. From the present experimental study, it can be said that the inclined anode configuration has the potential to minimize the energy consumption rate of the Hall Heroult cell as it was found to effectively reduce the bubbleinduced resistance. Further study is recommended to understand the effect of anode surface roughness and electro-magnetic forces on the bubble formation and movement underneath the anode surface. CuSO 4 þ 2e þ H 2 O ¼ Cu þ H 2 SO 4 þ 1=2O 2 : Hence, 1 mol of Cu generates ½ mol of O 2. If the experiment is run at 120 A (0.112 A/cm 2 ), charge required per second can be calculated from Faraday s law: Q ¼ I t ¼ 120 A 1s¼ 120 A: Now, Q ¼ nf where F as the Faraday s constant = 96,500. n ¼ Q F ¼ ; 500 ¼ 1: : The number of electron exchange = 2. Hence, number of Cu moles = 1: ¼ 6: : From the reaction, each mole Cu generates ½ mol O 2. Therefore, number of O 2 mol generated per second is 6: ¼ 3: : At standard temperature and pressure, each mole gas occupies 22.4 L volume. At 323 K (50 C) temperature, it becomes 26.5 L. Hence, the rate of O 2 bubble generation is 3: :5 ¼ 8: Ls 1 ¼ 0:2 Lm 2 s 1 : " # q g q 2 q l q g gl APPENDIX 2 real cell q 2 real cell ¼ q l q g gl q model q l q g gl :44 ¼ :2 " # q g q 2 ¼ q l q g gl real cell model model q g model q g real cell ð Þ9:81 1:35 ð Þ9:81 0:35 1:2 0:44 ACKNOWLEDGMENTS The authors would like to acknowledge the CSIRO Minerals Down Under Flagship, Australia, for funding the project. APPENDIX 1 Cu electrowinning reaction can be written as follows: CuSO 4 þ 2e ¼ Cu þ SO 2 4 H 2 O ¼ 2H þ þ 2e þ 1=2O 2 q 2 real cell ¼ 18:5 q model q real cell q model ¼ 4:3: REFERENCES 1. S. Fortin, M. Gerhardt, and A.J. Gesing: Light Metals, TMS, Warrendale, PA, 1984, pp L. Xiang-peng, L. Jie, L. Yan-qing, Z. Heng-qin, and L. Ye-xiang: Trans. Nonferrous Met. Soc. China, 2004, vol. 14 (5), pp A. Solheim and J. Thonstad: Light Metals, TMS, Warrendale, PA, 1986, pp D.F. Che, J.J.J. Chen, and M.P. Taylor: CHEMECA, Auckland, 1990, pp

11 5. A. Perron, L.I. Kiss, and S. Poncsak: Light Metals, TMS, Warrendale, PA, 2005, pp P. Aussillous and D. Quere: Europhys. Lett., 2002, vol. 59 (3), pp S. Das, Y. Morsi, G. Brooks, W. Yang, and J.J.J. Chen: Light Metals, TMS, Warrendale, PA, 2011, pp S. Das, Y. Morsi, G. Brooks, W. Yang, and J.J.J. Chen: Colloids Surf. A, 2012, vol. 411, pp S. Das, Y. Morsi, G. Brooks, W. Yang, and J.J.J. Chen: 10th Australian Aluminium Smelting Technology Conference, 9 14 October, 2011, Launceston. 10. R. Shekhar and J.W. Evans: Metall. Mater. Trans. B, 1994, vol. 25B, pp A. Solheim, S.T. Johansen, S. Rolseth, and J. Thonstad: Light Metals, TMS, Warrendale, PA, 1989, pp M.A. Cooksey and W. Yang: Light Metals, TMS, Warrendale, PA, 2006, pp W. Yang and M.A. Cooksey: Light Metals, TMS, Warrendale, PA, 2007, pp Y. Wang, L. Zhang, and X. Zuo: Light Metals, TMS, Warrendale, PA, 2009, pp M.A. Doheim, A.M. EL-Kersh, and M.A. Ali: Metall. Mater. Trans. B, 2007, vol. 38B, pp S. Das, G. Brooks, and Y. Morsi: Metall. Mater. Trans. B, 2011, vol. 42B, pp K. Qian, J.J.J. Chen, and N. Matheou: J. Appl. Electrochem., 1997, vol. 27, pp K. Qian, Z.D. Chen, and J.J.J. Chen: J. Appl. Electrochem., 1998, vol. 28, pp M. Alam, Y. Morsi, W. Yang, K. Mohanarangam, G. Brooks, and J. Chen: Light Metals, TMS, Warrendale, PA, 2013, pp R. Dorin and E.J. Frazer: J. Appl. Electrochem., 1993, vol. 23, pp R.J. Aaberg, V. Ranum, K. Williamson, and B.J. Welch: Light Metals, TMS, Warrendale, PA, 1997, pp K. McGregor, G.A. Snook, A.J. Urban, and M.R. Lanyon: ECS Trans., 2010, vol. 28 (6), pp D. Kasherman and M. Skyllas-Kazacos: J. Appl. Electrochem., 1988, vol. 18, pp U. Utigard, J.H. Costa, P. Popelar, D.I. Walker, G. Cool, and P. Hoang: Light Metals, TMS, Warrendale, PA, 1994, pp J. Xue and H.A. Oye: Light Metals, TMS, Warrendale, PA, 1995, pp P.A. Solli, T. Haarberg, T. Eggen, E. Skybakmoen, and A. Sterten: Light Metals, TMS, Warrendale, PA, 1994, pp L. Cassayre, A. Torstein, U. Utigard, and S. Bouvet: JOM, 2002, vol. 54, pp X. Wang and T. Tabereaux: Light Metals, TMS, Warrendale, PA, 2000, pp W.D. Zhang, J.J.J. Chen, and M.P. Taylor: CHEMECA, Auckland, 1990, pp J. Zoric and A. Solheim: J. Appl. Electrochem., 2000, vol. 30, pp K. VéKony and L.I. Kiss: Metall. Mater. Trans. B, 2012, vol. 43B, pp D.C. Price and W.G. Davenport: Metall. Trans. B, 1980, vol. 11B, pp S.S. Djokić: Electrodeposition: Theory and Practice, Springer, New York, 2010.