The Optimum Electrolyte Parameters in the Application of High Current Density Silver Electroreﬁning

metals

Article The Optimum Electrolyte Parameters in the Application of High Current Density Silver Electroreﬁning

Arif T. Aji, Jari Aromaa and Mari Lundström *

Department of Chemical and Metallurgical Engineering, School of Chemical Engineering, Aalto University, P.O. Box 16200, FI 00076 Aalto, Finland; arif.aji@aalto.fi (A.T.A.); jari.aromaa@aalto.fi (J.A.) * Correspondence: mari.lundstrom@aalto.fi

Received: 5 October 2020; Accepted: 24 November 2020; Published: 28 November 2020

Abstract: Increasing silver production rate has been a challenge for the existing refining facilities. The application of high current density (HCD) as one of the possible solutions to increase the process throughput is also expected to reduce both energy consumption and process inventory. From the recently-developed models of silver electrorefining, this study simulated the optimum electrolyte parameters to optimize the specific energy consumption (SEC) and the silver inventory in the electrolyte for an HCD application. It was found that by using [Cu2+] in electrolyte, both objectives can be achieved. The suggested optimum composition range from this study was [Ag+] 100–150 g/dm3, 3 2+ 3 2 [HNO3] 5 g/dm , and [Cu ] 50–75 g/dm . HCD application (1000 A/m ) in these electrolyte conditions result in cell voltage of 2.7–3.2 V and SEC of 0.60–1.01 kWh/kg, with silver inventory in electrolyte of 26–39 kg silver for 100 kg per day basis. The corresponding figures for the conventional process were 1.5–2.8 V, 0.44–0.76 kWh/kg, and 15.54–194.25 kg, in respective order. These results show that, while HCD increases SEC by app. 30%, the improvement provides a significant smaller footprint as a result of a more compact of process. Thus, applying HCD in silver electrorefining offers the best solution in increasing production capacity and process efficiency.

Keywords: high current density; silver electroreﬁning; energy consumption

1. Introduction Silver is used in many applications such as coins, jewelry, medicine, dentistry, plating, electrical technology, chemical equipment, catalyst, and photography [1]. The recorded production of silver from 1950 to 2018 shows a linear trend with an increasing rate of approximately 300 tons of silver product per year as can be seen in Figure1a [ 2]. In order to meet the increasing demand, increase in the existing capacity or higher kinetics of processing is required. Accordingly, developments in the existing silver electrorefining process were more focused on the increase of processing rate, reduction of metal inventory, and increase of metal recovery rate [3]. One of the developments was high current density (HCD), or the application of current density over 1000 A/m2 silver electrorefining, which offers a logical solution for the increasing capacity demand. Started in 1980 in Sweden, this process development path has recorded 11 installations of HCD silver electrorefining worldwide until 2014 [4]. Furthermore, the reported- and published research and development of the process shows that recent investigations [4–10] were focused on the implementation of higher current density as shown by Figure1b, making it the most renowned improvement in the recent years.

Metals 2020, 10, 1596; doi:10.3390/met10121596 www.mdpi.com/journal/metals Metals 2020, 10, x FOR PEER REVIEW 2 of 13

In conventional silver electrorefining free acid [HNO3] is within the range of 0–10 g/dm3 [8], meanwhile for HCD, even though it is not clearly stated, the electrolyte should be less acidic with pH > 2 [4,5]. With the assumption that HNO3 is the only pH regulator, the pH > 2 values correspond to [HNO3] of lower than 0.63 g/dm3. Thus, the electrolyte of HCD has higher [Ag+] and lower [HNO3] in Metalscomparison2020, 10, 1596to the conventional process. The high [Ag+] in electrolyte increases the Ag inventory2 of 14, being one of the disadvantages of the HCD application.

FigureFigure 1.1. ((a).). Annual production of silver in the period of 1950–20181950–2018 [22]] andand ((bb)) publishedpublished studiesstudies onon silversilver electroreﬁningelectrorefining showingshowing thethe correlationcorrelation ofof currentcurrent densitydensity toto thethe concentrationconcentration ofof silversilver inin thethe electrolyteelectrolyte [[1,41,4––1010].].

Similar to the conventional electrorefining, HCD silver electrorefining uses cast anodes and Another distinct difference of the conventional and HCD process is the current density with AgNO -HNO electrolyte with the arrangement of Moebius cell [4]. Nevertheless, the processes conventional3 silver3 electrorefining current density is within the range of 200–800 A/m2 as shown in difference in electrolyte parameters. Conventional process uses [Ag+] = 40–150 g/dm3 Table 1 [1,5–9,12] whereas the HCD current density targets to 1000 A/m2 [4,5]. As shown in the same (as AgNO )[1,6–12] while the HCD process uses higher concentrations of [Ag+] = 100–150 g/dm3 table, conventional3 cell voltage within the range of 1.5–2.8 V [1,7,8] leads to specific energy in the electrolyte [4]. In conventional silver electrorefining free acid [HNO ] is within the range of consumption of 0.6–0.8 kWh/kg [11,12]. Meanwhile, HCD cell voltage is approximately3 5 V [4] which 0–10 g/dm3 [8], meanwhile for HCD, even though it is not clearly stated, the electrolyte should be less equal to 1.27 kWh/kg. This higher energy consumption of the silver electrorefining process is another acidic with pH > 2 [4,5]. With the assumption that HNO is the only pH regulator, the pH > 2 values disadvantage of the application of HCD. 3 3 + correspond to [HNO3] of lower than 0.63 g/dm . Thus, the electrolyte of HCD has higher [Ag ] and + lower [HNO3] in comparisonTable 1. Conventional to the conventional silver electrorefining process. The process high [Ag parameters] in electrolyte. increases the Ag inventory, being one of the disadvantages of the HCD application. ParametersAnother distinct [1] di fference[5] of the conventional[6] and[7] HCD process[8] is the current[9] density[12] with conventional[Ag+] silver electrorefining50 150–200 current 65– density120 is150 within the30 range–150 of 200–80050 A/m2 as50 shown in Table[HNO1[ 13],5 –9,12] whereas10 the2.5 HCD current0.6–10 density 2–6.2* targets to0 1000–10 A/m2 [4,5-]. As shown10 in the same table,T (°C) conventional- cell voltage35–50 within35 the range32 of 1.5–2.825 V [1,7,8] leads45 to specific- energy consumptionwt%Ag (%) of 0.6–0.895 kWh/kg99.3 [11, 12]. Meanwhile,96.5 HCD86–92 cell voltage86–92 is approximately98 5 V [˃4 99] which equal to 1.27 kWh/kg. This higher0.04 energy– consumption of the silver electrorefining process is another wt%Au (%) 4 0.01 8–9 8–9 0.5 - disadvantage of the application0.07 of HCD. wt%Cu (%) 1 0.4–0.6 3 0.5–1 0.5–1 1 - j (A/m2) 400Table–500 1. Conventional1000 500 silver–800 electrorefining 300# process200– parameters.400 400 400–500 Vcell (V) 2.0–2.5 - - 2.7 1.5–2.8 - 2.0–2.5 Parameters [1][5][6][7][8][9][12] * calculated from the pH information, and # was unit conversion from amps/sq feet. [Ag+] 50 150–200 65–120 150 30–150 50 50 At the[HNO current3] state of10 increasing 2.5 industrial 0.6–10 demand, 2–6.2 improvement * 0–10 to the process - throughput 10 is getting moreT (◦C) important than - ever before. 35–50 On the 35 other hand, 32 the conventional 25 silver 45 electrorefining - processwt% itselfAg (%)has a very wide95 range 99.3 of operating 96.5 parameters. 86–92 Thus, 86–92 optimization 98 for the >process99 is requiredwt% toAu take(%) into account 4 both 0.04–0.07opportunity of 0.01 improvements 8–9 of increasing 8–9 the 0.5 process throughput - wt%Cu (%) 1 0.4–0.6 3 0.5–1 0.5–1 1 - and to havej (A/ mwell2) -designed400–500 process parameters 1000 500–800 for an optimal300 # process.200–400 400 400–500 TheV disadvantagescell (V) 2.0–2.5of high cell voltage - and high - energy 2.7 consumption 1.5–2.8 of the HCD - application 2.0–2.5 in comparison to the* calculated conventional from the process pH information, consequently and # was unitraised conversion a question from amps of whether/sq feet. HCD could be adapted in the existing silver electrorefining to increase the productivity. The process integration At the current state of increasing industrial demand, improvement to the process throughput is getting more important than ever before. On the other hand, the conventional silver electrorefining process itself has a very wide range of operating parameters. Thus, optimization for the process is required to take into account both opportunity of improvements of increasing the process throughput and to have well-designed process parameters for an optimal process. Metals 2020, 10, x FOR PEER REVIEW 3 of 13 probably has been the cause for the low implementation of HCD from the time it was introduced almost four decades ago. In this investigation, the decisive parameters for the HCD application are defined for the optimum silver electrorefining process based on the current industrial conditions.

Metals2. Methods2020, 10 , 1596 3 of 14 Investigation on the applicability of HCD process to silver electrorefining is conducted through the simulation of the published models as shown by Figure 2. The models in consideration are The disadvantages of high cell voltage and high energy consumption of the HCD application regression models from the following publications: in comparison to the conventional process consequently raised a question of whether HCD could a. Empirical model of electrolyte conductivity as a function of electrolyte concentrations and be adapted in the existing silver electrorefining to increase the productivity. The process integration temperature [13]. probably has been the cause for the low implementation of HCD from the time it was introduced b. Silver dissolution as a function of electrolyte concentrations and temperature [14]. almost four decades ago. In this investigation, the decisive parameters for the HCD application are c. Effect of Au content in anode due to anode passivation [15]. defined for the optimum silver electrorefining process based on the current industrial conditions. d. Electrode overpotential [16]. 2. MethodsAll of the models were constructed with the range of parameters according to the existing industrial practices. Conductivity model was developed with a full factorial design of electrolyte Investigation on the applicability of HCD process to silver electrorefining is conducted through the concentrations ([Ag+], [Cu2+], [H2SO4], [Pb2+]) and temperature. Measurements were conducted with a simulation of the published models as shown by Figure2. The models in consideration are regression Knick Portamess® 913 Cond conductivity meter (Knick Elektronische Messgeräte GmbH and Co KG, models from the following publications: Berlin, Germany). Meanwhile, the models of silver dissolution and the effect of Au-to-anode a.passivationEmpirical were model developed of electrolyte from the series conductivity of anodic as polarization a function of of electrolyte pure silver concentrations and Au-Ag binary and alloys temperature with known [13 ranges]. of Au content. And finally, the electrode overpotential model was b.developed Silver from dissolution galvanostatic as a function measurements of electrolyte with concentrations pure silver as and working temperature electrode [14]. at a current 2 c.densityE ff ofect 1000 of Au A/m content. The in tests anode were due toconducted anode passivation at different [15 ]. concentrations and temperature of electrolyte. Statistical analysis the developed models has demonstrated that every model presented d. Electrode overpotential [16]. in [13–16] fulfils the requirements of a good model.

High current density silver electrorefining [14] Electrolyte conductivity modelling Effect of Au [13]

Anode and Anode passivation due Silver impurities Kinetics to Au content in electrolyte (Au and Cu) modelling anodes properties [15]

Effect of Cu Energy consumption, IR drop modelling Electrode [16] overpotential [16]

Figure 2. Method of the investigation on the aspects for the high current density (HCD) application Figure 2. Method of the investigation on the aspects for the high current density (HCD) application (dashed box) is based on the anodic [14–16] and electrolytic [13,16] modelling. (dashed box) is based on the anodic [14–16] and electrolytic [13,16] modelling. All of the models were constructed with the range of parameters according to the existing With the strict policy of precious metal refineries, measurements using industrial electrolytes industrial practices. Conductivity model was developed with a full factorial design of electrolyte and anodes were not+ possible.2+ Throughout2 +the study, only a series of industrial tests was conducted concentrations ([Ag ], [Cu ], [H2SO4], [Pb ]) and temperature. Measurements were conducted with in a precious metal refinery to validate the developed the conductivity model from previous study a Knick Portamess®913 Cond conductivity meter (Knick Elektronische Messgeräte GmbH and Co KG, [13] in industrial electrolyte concentration shown in Table 2. The conductivity values measured and Berlin, Germany). Meanwhile, the models of silver dissolution and the effect of Au-to-anode passivation calculated by the model for the 20 samples taken at two temperatures of 25 °C and 40 °C. From the were developed from the series of anodic polarization of pure silver and Au-Ag binary alloys with validation, additional models of conductivity were also developed and defined as validation (validation) known ranges of Au content. And finally, the electrode overpotential model was developed from and industrial (industrial) models. Validation model was developed from the concentration coefficients galvanostatic measurements with pure silver as working electrode at a current density of 1000 A/m2. from the empirical model while the temperature coefficient was taken from the validation results. On The tests were conducted at different concentrations and temperature of electrolyte. Statistical analysis the other hand, industrial model was independently developed only from the industrial the developed models has demonstrated that every model presented in [13–16] fulfils the requirements measurements. The schematic of the conductivity modellings and their correlation are shown in of a good model. Figure 3. With the strict policy of precious metal refineries, measurements using industrial electrolytes and anodes were not possible. Throughout the study, only a series of industrial tests was conducted in a precious metal refinery to validate the developed the conductivity model from previous study [13] in industrial electrolyte concentration shown in Table2. The conductivity values measured and calculated by the model for the 20 samples taken at two temperatures of 25 ◦C and 40 ◦C. From the validation, Metals 2020, 10, 1596 4 of 14

additional models of conductivity were also developed and defined as validation (κvalidation) and industrial (κindustrial) models. Validation model was developed from the concentration coefficients from the empirical model while the temperature coefficient was taken from the validation results. On the other hand, industrial model was independently developed only from the industrial measurements. The schematic of the conductivity modellings and their correlation are shown in Figure3. Metals 2020, 10, x FOR PEER REVIEW 4 of 13 Table 2. Electrolyte conditions of the industrial practice. Table 2. Electrolyte conditions of the industrial practice. + 3 3 2+ 3 2+ 3 Parameters [Ag ], g/dm [HNO3], g/dm [Cu ], g/dm [Pb ], g/dm Parameters [Ag+], g/dm3 [HNO3], g/dm3 [Cu2+], g/dm3 [Pb2+], g/dm3 Values 97.9 8.3 0.3 0.1 56.9 1.9 1.6 0.1 Values 97.9 ±± 8.3 0.3 ±± 0.1 56.9 ±± 1.9 1.6 ± 0.1±

Synthetic electrolyte in Industrial electrolyte and Validation process laboratory measurements measurements

kempirical =Σci,emp· [Xi,emp] + CT,emp· T kvalidation =Σci,emp· [Xi,emp] + CT,val· T kindustrial =Σci,ind· [Xi,ind] + CT,ind· T

Figure 3. SchematicSchematic of the empirical, validation, and industrial models.

AAllll of the developed developed models were also used as the base base to to study study HCD HCD application application with with particular particular attention paid paid to tothe thecell voltage cell voltage and specific and energyspecific consumption energy consumption (SEC). Simulations (SEC). and Simulations comparison and of comparisonHCD and conventional of HCD process and conventional were conducted process based on were the electrolyte conductedconcentrations based on the from electrolyte literature. concentrationsFinally, an optimum from condition literature. of Finally, HCD silver an optimum electrorefining condition was established of HCD silver in order electrorefining to have both was low establishedenergy consumption in order and to have low inventory both low of energy silver in consumption the electrorefining and low process. inventory of silver in the electrorefining process. 3. Theoretical Background 3. TheoreticalThe evaluation Background of optimum parameters was designed mainly based on two calculated outcome of specificThe energy evaluation consumption of optimum (SEC) parameters and silver was inventory designed in themainly process. based on two calculated outcome of specific energy consumption (SEC) and silver inventory in the process. 3.1. Specific Energy Consumption 3.1. SpecificTheoretically, Energy theConsumption main consumable of an electrorefining process is the electrical energy. With the electrical energy consumption (E ) outlined in Equation (1), the SEC of a silver electrorefining process Theoretically, the main consumablecons of an electrorefining process is the electrical energy. With is defined as Equation (2) below, where Ecell is the cell voltage (V). With the data of the constants (n, F the electrical energy consumption (Econs) outlined in Equation (1), the SEC of a silver electrorefining and MAg) and the current efficiency of conventional process of 92–95% [9,11] and HCD of 98–99% [4], process is defined as Equation (2) below, where Ecell is the cell voltage (V). With the data of the the only variable left from Equation (2) is the cell voltage (Vcell). constants (n, F and MAg) and the current efficiency of conventional process of 92–95% [9,11] and HCD of 98–99% [4], the only variable left from Equation (2) is the cell voltage (Vcell). Econs(kWh) = E I t (1) cell· · 퐸푐표푛푠! (kWh) = 퐸푐푒푙푙 · 퐼 · 푡 (1) kWh Econs V F n SEC = = cell· · . (2) 푘푊kgℎ 퐸m푐표푛푠Ag 3600푉푐푒푙푙CE· 퐹M· 푛Ag 푆퐸퐶 ( ) = = · · . (2) Cell voltage is the sum of potentials푘𝑔 from the푚퐴푔 process.3600 Excluding· 퐶퐸 · 푀퐴푔 the external voltage drops such as busbarCell voltage or electrical is the contacts,sum of potentials cell voltage from components the process. in Excluding silver electrorefining the external processvoltage aredrops voltage such asdrop busbar due toor theelectrical conductivity contacts, of cell electrolyte, voltage components anode polarization in silver with electrorefining the effect of process gold, and are cathodevoltage polarization.drop due to the Thus, conductivity the cell voltage of electrolyte, can be written anode as polarization Equation (3) with the effect of gold, and cathode polarization. Thus, the cell voltage can be written as Equation (3)

Vcell = ηa + ηc + ηpass + IR (3) 푉푐푒푙푙 = 푎 + |푐| + 푝푎푠푠 + 퐼푅 (3) where ηa is the anodic overpotential, ηc is the cathodic, ηpass is the excess polarization due to anode passivation, and IR is the potential drop due to electrolyte resistivity. The effect of conductivity to the IR drop is shown in Equation (4).

퐼 푙푒푙 푙푒푙 퐼푅 = ∙ = 푗 ∙ (4) 푑푟표푝 퐴 휅 휅 where I is current (A), lek is the interelectrode distance (m), A is the surface area (m2) and  is the conductivity (S/m).

Metals 2020, 10, 1596 5 of 14

where ηa is the anodic overpotential, ηc is the cathodic, ηpass is the excess polarization due to anode passivation, and IR is the potential drop due to electrolyte resistivity. The eﬀect of conductivity to the IR drop is shown in Equation (4). I l l IR = el = j el (4) drop A· κ · κ 2 where I is current (A), lek is the interelectrode distance (m), A is the surface area (m ) and κ is the conductivity (S/m).

3.2. Silver Inventory One of the economic aspects of the silver electrorefining besides the energy consumption is the silver inventory in the process. Silver inventory as the result of HCD can be calculated from the effect of increasing process throughput to the process footprint, in this case the volume of electrolyte. From the Faraday law of electrolysis in Equation (5), the production rate of a plant as function of the electrolyte volume shown in Equation (6) I t MAg m = · · (5) Ag n F ! · mAg kg A MAg = j Vtot (6) t day ·V · n F · · where j is the current density (A/m2), A/V is the effective surface area of anode to electrolyte volume 2 3 (m /dm ), MAg is the molar mass of silver, n is the electron valence of silver, F is Faraday constant and V is the electrolyte volume (dm3).

4. Results and Discussions

4.1. IR Rrop Energy consumption in an electroreﬁning process is mainly due to the resistivity of the electrolyte. To estimate the IR drop in this system, an empirical conductivity model based on laboratory measurements shown in Equation (7) was developed and conﬁrmed to be the accurate for the + 3 2+ 3 system when electrolyte properties are [Ag ] of 40–150 g/dm , [Cu ] of 0–80 g/dm , [HNO3] of 3 0–15 g/dm and T of 25–45 ◦C[13] and unlike the other models, conductivity modelling of silver electrolyte had been conducted by previous study [17]. Thus, this model was statistically tested and compared to the previous model by Gordon and Davenport [17] and is proven to be a valid model and has better accuracy in comparison to the model of Gordon and Davenport [13]. h i κ = 10.844 + 0.277 [Ag+] + 3.268 [HNO ] + 0.700 Cu2+ + 0.204 empirical · · 3 · h 2+i + Pb + 0.573 T + 0.008 [Ag ] T + 0.049 [HNO3] T + 0.02. (7) · · ·h ·i · · Cu2+ T · ·

In Equation (7) κ is the conductivity of electrolyte (mS/cm) and T is temperature (◦C) and concentrations are given as g/dm3. Furthermore, validation of the model was conducted and according to the method in Figure3, the developed validation and industrial models are shown in Equations (8) and (9) respectively. As the industrial model was developed only from the steady condition of industrial electrolyte measurements, the model shows different behavior. As an example, the effect of [Ag+] is shown to be lower and [HNO3] has a negative effect to the conductivity. h i h i κ = 61.2044 + 0.277 [Ag+] + 3.268 [HNO ] 1.648 T + 0.700 Cu2+ + 0.204 Pb2+ validation · · 3 − · · · +0.008 [Ag+] T + 0.049 [HNO ] T (8) · · h i · 3 · +0.02 Cu2+ T · Metals 2020, 10, 1596 6 of 14

h i h i h i κ = 114 + 0.11 Ag+ 0.99 [HNO ] + 0.32 T + 0.025 Cu2+ + 2.22 Pb2+ . (9) Industrial · − · 3 · · · The accuracy of all models to the industrial measurement values are shown in Figure4. Figure4a–c show that the empirical model accurately predicts conductivities at lower temperature Metals 2020, 10, x FOR PEER REVIEW 6 of 13 while at higher temperatures the model over calculates the electrolyte conductivity. Validation model accuracya challenge which for the is ascalculat showning by conductivity Figure4d–e hasof wider the improved range electrolyte prediction concentrations of conductivity. All valuesmodels at in 2 higherEquations temperature (7)–(9) were are closerincluded to thefor measurements.further calculation On thein this other study hand,. the goodness of fit (R ) of the modelFrom has the become electrolyte significantly concentrations lower as in shown Table by2, with Figure the4f. exempt Finally, of the [Ag industrial+], the impact model of applying has the bestHCD fit calculated to the measurement with the three values models as can were be seen simulated in Figure with4g–i. the While assumption industrial of modelindustrial is the current most representative model for industrial electrolyte, narrow range of the electrolyte concentration creates a density of 420 A/m2 and lek of 5 cm [11] as shown in Figure 5. The simulation of IR drop increase as a challengefunction of for [Ag the+] calculatingdemonstrates conductivity the impact of of wider HCD range application electrolyte in the concentrations. range of [Ag+ All] from models 60–200 in Equationsg/dm3. (7)–(9) were included for further calculation in this study.

160 150 (a) Conductivity at 25°C (b) Conductivity at 40°C (c) Empirical model regression 200 Measured Measured y = 0.13 x + 118 Predicted by  Predicted by  2 150 empirical empirical 145 R = 0.91 180

160 140 140

Conductivity (mS/cm) Conductivity (mS/cm) 140

130 Measured conductivity (mS/cm) 135 0 5 10 15 20 25 0 5 10 15 20 25 140 160 180 200 Measurements Measurements Predicted conductivity (mS/cm) 160 160 160 (d) Conductivity at 25°C (e) Conductivity at 40°C (f) Validation model regression Measured Measured y = 0.48 x + 72   2 150 Predicted validation 150 Predicted validation 150 R = 0.58

140 140 140

Conductivity (mS/cm) Conductivity (mS/cm)

130 130 Measured conductivity (mS/cm) 130 0 5 10 15 20 25 0 5 10 15 20 25 130 135 140 145 150 Measurements 150 150 Measurements 150 Predicted conductivity (mS/cm) (g) Conductivity at 25°C (i) Industrial model regression Measured y = x + 0.01 145 145 145 2 Predicted industrial R = 0.918

140 140 140

(h) Conductivity at 40°C 135 135 135

Conductivity (mS/cm) Conductivity (mS/cm) Measured

Predicted industrial

130 130 Measured conductivity (mS/cm) 130 0 5 10 15 20 25 0 5 10 15 20 25 136 138 140 142 144 Measurements Measurements Predicted conductivity (mS/cm) Figure 4. Results of predicted and measured values of empirical model (a–c), validation model (d–f), Figure 4. Results of predicted and measured values of empirical model (a–c), validation model (d–f), and industrial model (g–i) for the industrial electrolyte conditions in Table2. and industrial model (g–i) for the industrial electrolyte conditions in Table 2. From the electrolyte concentrations in Table2, with the exempt of [Ag+], the impact of applying 2 HCDFrom calculated the Figure with 5, the increasing three models the existing were simulated industrial withcurrent the density assumption to 1000 of A/m industrial increases current the IR drop significantly2 from app. 1.6 V calculated by the empirical model to 2 V from the validation density of 420 A/m and lek of 5 cm [11] as shown in Figure5. The simulation of IR drop increase as a functionand industrial of [Ag +models.] demonstrates The increase the impact accounts of HCD for more application than 100% in the of range the initial of [Ag IR+] dropfrom of 60–200 the process. g/dm3. + The FigureFrom the also Figure demonstrates5, increasing that thehigh existing [Ag ] is industrial preferred current in order density to have to lower 1000 Aincrease/m2 increases of IR dr theop. 2+ IRAnother drop signiﬁcantly potential factor from is the app. effect 1.6 Vof calculated [Cu ] as it byis shown the empirical to increase model the toelectrolyte 2 V from conductivity the validation as 2+ andshown industrial by all of models. the models The outlined increase in accounts Equations for more(7)–(9). than In a 100% recent of study, the initial having IR drop[Cu of] in the electrolyte process. + 3 2+ Thewas Figurealso encouraged also demonstrates as the study that highsugge[Agst +electrolyte] is preferred compositions in order to of have [Ag lower]  60 increase g/dm , of[CuIR]drop.  60 3 3 + 2+ Anotherg/dm , and potential [HNO3] factor  7 g/dm is the [18]. eﬀect Thus, of [ Cufurther2+] as simulation it is shown of to the increase effect of the [Ag electrolyte] and [Cu conductivity] is required asto shownhave an by optimum all of the condition models of outlined electrolyte in Equations for a minimum (7)–(9). impact In a recentof HCD study, application having due[Cu 2to+] thein electrolyteincrease of wasIR drop. also encouraged as the study suggest electrolyte compositions of [Ag+] > 60 g/dm3,

Metals 2020, 10, 1596 7 of 14

2+ 3 3 + 2+ [Cu ] < 60 g/dm , and [HNO3] < 7 g/dm [18]. Thus, further simulation of the eﬀect of [Ag ] and [Cu ] is required to have an optimum condition of electrolyte for a minimum impact of HCD application Metals 2020, 10, x FOR PEER REVIEW 7 of 13 due to the increase of IR drop.

Figure 5. Increase of IR drop for the implementation of HCD operation calculated by (a). empirical Figure 5. Increase of IR drop for the implementation of HCD operation calculated by (a). empirical model, (b) validation model, and (c) industrial model. model, (b) validation model, and (c) industrial model. Calculation of the conventional process IR drop by using the Equation (4) from the data in Calculation of the conventional process IR drop by using the Equation (4) from the data in Table Table1 shows IR drop values within the range of 0.8–2.3 V which equal to 55–82% of the total cell 1 shows IR drop values within the range of 0.8–2.3 V which equal to 55–82% of the total cell potential. potential. As it can be expected, HCD process results in higher cell potential thus the role of electrolyte As it can be expected, HCD process results in higher cell potential thus the role of electrolyte conductivity is signiﬁcant. With the positive eﬀect of [Cu2+] to the conductivity of electrolyte, it is conductivity is significant. With the positive effect of [Cu2+] to the conductivity of electrolyte, it is important to take a study of the possibility of having [Cu2+] in smaller ratios of [Cu2+]/[Ag+] to achieve important to take a study of the possibility of having [Cu2+] in smaller ratios of [Cu2+]/[Ag+] to achieve both optimum conductivity and to avoid the co-deposition of Cu on the cathode. A comparison of the both optimum conductivity and to avoid the co-deposition of Cu on the cathode. A comparison of [Cu2+]/[Ag+] levels from previous studies is shown in Table3. the [Cu2+]/[Ag+] levels from previous studies is shown in Table 3. Table 3. [Ag+] and maximum [Cu2+] in silver electrolytes. Table 3. [Ag+] and maximum [Cu2+] in silver electrolytes. Concentrations [19][7][20][6] Concentrations [19] [7] [20] [6] [Ag+] 40 150 60–160 65–120 >60 [Ag+] 40 150 60–160 65–120 ˃ 60 [Cu2+] 35 max 45 60 max 50–100 < 60 2+ [Cu2+[Cu]/[Ag+] ] 0.87535 0.3max 45 0.5–160 0.77–0.83max 50–100 0.8–1.3 60 [Cu2+]/[Ag+] 0.875 0.3 0.5–1 0.77–0.83 0.8–1.3

From the suggested generalgeneral [Ag[Ag++] of > 6060 g/dm g/dm3 [6][6] and and in in HCDHCD ofof 100 100–150–150 g/dm g/dm3 [[44],], the the IRIR drop values in the range of [Ag +]] ofof 60 60–200–200 g/dm g/dm33 werewere simulated simulated by by using using the the models models in in Equations Equations (7) (7)–(9)–(9) 22++ ++ 3 at the rangerange ofof [Cu[Cu ]/[Ag]/[Ag ] ]asas shown shown in in F Figureigure 66 withwith thethe constantconstant valuesvalues ofof [HNO[HNO3]] of 5 gg/dm/dm at median temperature of 35 ◦°C.C. Figure 6a,b show large ranges of IR drop and the growing distances between the IR contour lines which indicates that as the concentrations of [Ag+] and [Cu2+] are higher, their effect to IR drop reduction becomes insubstantial. As can be seen, in each of both figures, a bold line was added showing the limit where increase of concentrations affect to the reduction of IR drop substantially. Meanwhile, Figure 6 c shows narrow IR drop range and uniform increase throughout the [Ag+] levels. In order to set the optimum condition electrolyte, the added dashed lines indicate the range of [Ag+] within 100–150 g/dm3 as suggested for HCD operation [4]. From Figure 6 a, the bold line cross the the dashed line at app. 0.9 at 100 g/dm3 of [Ag+] and app. 0.5 at 150 g/dm3 of [Ag+]. With the first value of [Cu2+]/[Ag+] at 0.9 has high risk of Cu co-deposition, the second value at 0.5 can be taken as the minimum value of the ratio for an optimum HCD.

Metals 2020, 10, 1596 8 of 14 Metals 2020, 10, x FOR PEER REVIEW 8 of 13

FigureFigure 6.6.The The eeffectﬀect of of[Ag [Ag++]] andand [Cu[Cu22++]] to the IR drop in thethe applicationapplication ofof 10001000 AA/m/m22 currentcurrent densitydensity calculatedcalculated with with ( a(a)) empirical empirical model, model, ( b(b)) validation validation model, model, and and ( c()c) industrial industrial model. model.

FigureWith the6a,b range show of large [Ag+] rangesof 100–150 of IRg/dmdrop3 and and [Cu the2+]/[Ag growing+] of 0.5, distances the IR drop between values the forIR 1000contour A/m2 + 2+ linesoperation which based indicates on the that models as the concentrationsin this study are of shown[Ag ] andin Table[Cu 4]. areIn the higher, table, their the eempiricalffect to IR modeldrop reductionshows the becomes lowest IR insubstantial. drop calculation As while can be the seen, industrial in each model of both has figures, the highest a bold value. line was added showing the limit where increase of concentrations affect to the reduction of IR drop substantially. + Meanwhile,Table 4 Figure. Optimum6 c shows [Ag+] and narrow [Cu2+] IRin silverdrop rangeelectrolytes and uniformfor HCD silver increase electrorefining throughout operation. the [Ag ] levels. In order to set the optimum condition electrolyte, the added dashed lines indicate the range of [Ag+] [Ag+], Optimum [Cu2+], IR drop (V) withinLevel 100–150 g/dm3 as suggested for HCD operation [4]. From Figure6 a, the bold line cross the the g/dm3 [Cu2+]/[Ag+] g/dm3 Empirical Validation Industrial dashed line at app. 0.9 at 100 g/dm3 of [Ag+] and app. 0.5 at 150 g/dm3 of [Ag+]. With the first value Min2+ + of [Cu ]/[Ag ] at100 0.9 has high risk of0.5 Cu co-deposition,50 the second2.75 value at 0.53.24 can be taken3.67 as the + minimum[Ag ] value of the ratio for an optimum HCD. Max + 3 2+ + 2 With the range150 of [Ag ] of 100–1500.5 g /dm and [Cu 75]/[Ag ] of 0.5,2.05 the IR drop values2.30 for 10003.52 A/ m + operation[Ag ] based on the models in this study are shown in Table4. In the table, the empirical model shows the lowest IR drop calculation while the industrial model has the highest value. As can be seen from Table 4, the difference of empirical with the industrial conductivity values is significant.Table 4. Optimum This was[Ag mainly+] and caused[Cu2+] inby silver the actual electrolytes concentration for HCD silver of the electrorefining industrial electrolyte operation. where more dissolved contents were presents, while the empirical model was built only with the effect of + 2+ 2+ IR Drop (V) majorLevel concentrations[Ag+], g/ dmsuch3 as Optimum[Ag ], [HNO[Cu23+]], /[Ag[Cu+]], and[Cu [Pb2+], g]/dm. V3alidation model connects empirical and industrial model with the correlation schematically shown in FigureEmpirical 3. As mentioned Validation Industrialin Section 2Min, the[Ag validation+] tests100 were conducted 0.5 to a type of industrial 50 electrolyte 2.75 only, thus 3.24 the model 3.67 has + questionableMax [Ag ] validity150 to be applied in 0.5 different types of electrolyte.75 With 2.05 these 2.30 weakness 3.52 of each model in correlation to the range industrial application, all the models were taken into account in predictingAs can the be seenoutcome from in Table terms4, theof s dipecificfference energy of empirical consumption with theof HCD industrial in the conductivity suggested electrolyte values is significant.condition from This this was study. mainly caused by the actual concentration of the industrial electrolyte where more dissolved contents were presents, while the empirical model was built only with the effect of + a c 2+ 2+ major4.2. Anodic concentrations and Cathodic such Overpotential as [Ag ], [HNO (η and3], [Cuη ) ], and [Pb ]. Validation model connects empirical and industrialFrom the model modelling with of the the correlation galvanostatic schematically results [15 shown], cathodic in Figure overpotential3. As mentioned when inoperating Section2 ,at 1000 A/m2 current density could not be measured and modelled reliably due to the dendritic deposition, nevertheless the measured cathodic overpotentials were 147–297 mV and decreasing with

Metals 2020, 10, 1596 9 of 14 the validation tests were conducted to a type of industrial electrolyte only, thus the model has questionable validity to be applied in diﬀerent types of electrolyte. With these weakness of each model in correlation to the range industrial application, all the models were taken into account in predicting the outcome in terms of speciﬁc energy consumption of HCD in the suggested electrolyte condition from this study.

4.2. Anodic and Cathodic Overpotential (ηa and ηc) From the modelling of the galvanostatic results [15], cathodic overpotential when operating at 1000 A/m2 current density could not be measured and modelled reliably due to the dendritic deposition, nevertheless the measured cathodic overpotentials were 147–297 mV and decreasing with increasing silver concentration. The anodic overpotential at 1000 A/m2 as function of electrolyte conditions was modelled as follows:

Metals 2020, 10, x FOR PEER REVIEW h +i h 2+i 9 of 13 ηa = 452.8 1.14 Ag 5.31 [HNO ] 1.21 Cu 4.13 T (10) − · − · 3 − · − · increasing silver concentration. The anodic overpotential at 1000 A/m2 as function of electrolyte 3 whereconditionsηa is the was anodic modelled overpotential as follows: (mV), concentrations in g/dm and temperature in degrees centigrade. Using the electrolyte concentrations+ in Table4, the anodic overpotential2+ is within the range 휂푎 = 452.8 − 1.14 · [퐴𝑔 ] − 5.31 · [퐻푁푂3] − 1.21 · [퐶푢 ] − 4.13 · 푇 (10) of 20–107 mV. where ηa is the anodic overpotential (mV), concentrations in g/dm3 and temperature in degrees Anodic Tafel slope (ba) model of Ag in AgNO3-HNO3 from the study of Ag kinetic dissolution [14] centigrade. Using the electrolyte concentrations in Table 4, the anodic overpotential is within the can be written as Equation (11) with model accuracy shown in Figure7a, while the di ﬀerence of range of 20–107 mV. both values are shown in Figure7b. As can be seen from the model and measurements, the anodic Anodic Tafel slope (ba) model of Ag in AgNO3-HNO3 from the study of Ag kinetic dissolution Tafel[14] slope can is be within written the as rangeEquation of ( 0.24–0.4511) with model V. Thus, accuracy by using shown the in F correlationigure 7a, while of current the difference density of and 2 2 anodicboth overpotential values are shown outlined in F inigure Equation 7b. As (12),can be increase seen from of currentthe model density and measurements, from 420 A/m theto anodic 1000 A /m requiresTafel small slope increaseis within ofthe anodic range of overpotential 0.24–0.45 V. Thus, of 0.09–0.2 by using V. the This correlation suggests of thatcurrent increasing density and current densityanodic to 1000 overpotential A/m2 does outlined not increase in Equation the anodic (12), increase overpotential of current substantially. density from 420 A/m2 to 1000 2 A/m requires small increase of anodic overpotentialh i of 0.09–0.2 V. This suggests that increasing 2 + current density to 1000b A/ma = 0.7 does 0.002not increaseAg the 0.01anodic[HNO overpotential] 0.0037 substantially.T (11) − · − · 3 − · + 푏푎 = 0.7 − 0.002 · [퐴𝑔 ] − 0.01 · [퐻푁푂! 3] − 0.0037 · 푇 (11) i2 ∆ηa = ba log (12) 푖2i 휂푎 = 푏푎log ( ⁄1 ) (12) 푖1

FigureFigure 7. (a )7. Predicted (a) Predicted vs measuredvs measured of of anodic anodic Tafel Tafel slopeslope values of of Ag Ag in in AgNO AgNO3-HNO3-HNO3 electrolyte3 electrolyte and (band) residual (b) residual values values ofthe of the prediction prediction to to the the measurements measurements results. results.

+ 휂퐴푢 = 11.25 + 1.01 푤푡%퐴푢 − 0.09 [퐴𝑔 ] − 0.60[퐻푁푂3] + 0.38 푇. (13) With the limitation of Au content to maximum of 6%, the effect of passivation overpotential is minimized when using the electrolyte concentrations in Table 4. This results in passivation overpotential of 0.2 V.

Metals 2020, 10, 1596 10 of 14

4.3. Passivation Overpotential (ηpass) Gold in silver anodes causes anode passivation. This is due to anode surface enrichment by gold along silver dissolution. In earlier studies, application of HCD in high [Ag+] electrolyte (min 100 g/dm3) was shown to be kinetically eﬃcient for maximum Au content less than 6% in the anode [15]. The passivation range due to Au content in anode can be calculated by using Equation (13) [16]. h i η = 11.25 + 1.01 wt% 0.09 Ag+] 0.60[HNO + 0.38 T. (13) Au Au − − 3 With the limitation of Au content to maximum of 6%, the eﬀect of passivation overpotential is minimized when using the electrolyte concentrations in Table4. This results in passivation overpotential of 0.2 V.

4.4. Circulating Electrolyte volume For the value of A/V set to a constant of 0.004, 1 m2 of anode surface per 250 dm3 of bulk electrolyte, taken from the general industrial practices for Moebius cell [11]. At the rate of production of 100 kg/day, electrolyte volume as a function of applied current density is shown by Equation (14).

5 3 2.59 10 Vtot dm = · . (14) j A m2

From above equation, to have the same productivity, higher current density process requires smaller volume of electrolyte.

4.5. Cell Voltage, SEC, and Silver Inventory For summary, the electrolyte concentrations in Table4 resulted in the total cell voltage components as follow:

IR drop • Empirical model = 2.05–2.75 V # Validation model = 2.30–3.24 V # Industrial model = 3.52–3.67 V # ηa and ηc = 0.32–0.40 V • η = 0.02 V • Au As can be seen from the components above IR drop is the most dominant by contributing 86–91% of the cell voltage. This was further supported by the anodic Tafel slope model which shows that increase of current density insubstantially affects the anodic overpotential. The increasing portion of IR drop to the cell voltage were simulated as shown in Figure8. The contribution of IR drop to cell voltage increases with increase in current density and IR drop contributes more than 85% of the cell voltage at 1000 A/m2 operation. In addition, operating at lower range of electrolyte concentration increases the IR drop portion to more than 90% of the cell voltage for electrolyte with lower [Ag+]. From the simulation, increase of electrolyte concentrations reduce the IR drop portion of cell voltage significantly while increase of temperature is shown to have less impact. While IR drop can be reduced by increasing both temperature and concentrations of the electrolyte, it cannot be avoided that IR drop contribution to cell voltage at 1000 A/m2 is significant as calculated by all the models of the current study. Metals 2020, 10, 1596x FOR PEER REVIEW 11 of 1413

95 95 95

90 90 90

85 85 85

80 80 80

75 75 75

70 70 70

IR drop/cell voltage (%) (a) Empirical model 25°C IR drop/cell voltage (%) (b) Validation model 25°C IR drop/cell voltage (%) (c) Industrial model at 25°C 65 100 g/dm3 [Ag+] 65 100 g/dm3 [Ag+] 65 100 g/dm3 [Ag+] 150 g/dm3 [Ag+] 150 g/dm3 [Ag+] 150 g/dm3 [Ag+] 60 60 60 400 600 800 1000 400 600 800 1000 400 600 800 1000 Current density (A/m2) Current density (A/m2) Current density (A/m2) 95 95 95

90 90 90

85 85 85

80 80 80

75 75 75

70 70 70

IR drop/cell voltage (%) (d) Empirical model at 45°C IR drop/cell voltage (%) (e) Validation model at 45°C IR drop/cell voltage (%) (f) Industrial model at 45°C 65 100 g/dm3 [Ag+] 65 100 g/dm3 [Ag+] 65 100 g/dm3 [Ag+] 150 g/dm3 [Ag+] 150 g/dm3 [Ag+] 150 g/dm3 [Ag+] 60 60 60 400 600 800 1000 400 600 800 1000 400 600 800 1000 Current density (A/m2) Current density (A/m2) Current density (A/m2) Figure 8. IR drop and cell voltage calculation for the electrolyte concentrations shown in Table4. Figure 8. IR drop and cell voltage calculation for the electrolyte concentrations shown in Table 4. Towards the potential application of HCD as an improvement to the current silver electrorefining practices,Table competitiveness5. Specific energy of theconsumption process key (SEC) figures and in silver comparison inventory to the comparison conventional of HCD process and is of conventional silver electrorefining. importance. Accordingly, the key figures of both processes are outlined in Table5. The suggested parameters from this study areHCD compared in This to the HCDHCD from from previous publicationConventional [4] and to the conventionalParameters silver electrorefining [Study7–12]. Previous Study [4] Silver Electrorefining [7– 12] [Ag+] (g/dm3) 100–150 100–150 30–150 [Cu2+] (g/dm3) 50–75 n.a 45–100 [HNO3] (g/dm3) 5–7 n.a 0–10 T (°C) 35–45 55°C 25–50 Current density (A/m2) 1000 1000 200–500 Cell voltage (V) a) Empirical model a) 2.39–3.09 5 V 1.5–2.8 b) Validation model b) 2.64–3.58 c) Industrial model c) 3.86–4.01 Current efficiency (%) 98–99 [4] 98–99 92–95 SEC (kWh/kg) a) Empirical model a) 0.60–0.78 1.27–1.28* 0.44–0.76 b) Validation model b) 0.67–0.90 c) Industrial model c) 0.98–1.01 Comparisons based on 100 kg of Ag product / day Electrolyte volume (dm3) 259 259* 518–1295* Silver inventory (kg) 25.9–38.85 25.9–38.85* 15.54–194.25* *calculated from the data in the references.

Metals 2020, 10, 1596 12 of 14

Table 5. Speciﬁc energy consumption (SEC) and silver inventory comparison of HCD and conventional silver electroreﬁning.

Conventional Parameters HCD in This Study HCD from Previous Study [4] Silver Electroreﬁning [7–12] [Ag+] (g/dm3) 100–150 100–150 30–150 [Cu2+] (g/dm3) 50–75 n.a 45–100 3 [HNO3] (g/dm ) 5–7 n.a 0–10 T(◦C) 35–45 <55◦C 25–50 Current density (A/m2) 1000 1000 200–500 Cell voltage (V)

a) Empirical model a) 2.39–3.09 <5 V 1.5–2.8 b) Validation model b) 2.64–3.58 c) Industrial model c) 3.86–4.01

Current eﬃciency (%) 98–99 [4] 98–99 92–95 SEC (kWh/kg)

a) Empirical model a) 0.60–0.78 1.27–1.28 * 0.44–0.76 b) Validation model b) 0.67–0.90 c) Industrial model c) 0.98–1.01

Comparisons based on 100 kg of Ag product / day Electrolyte volume (dm3) 259 259 * 518–1295 * Silver inventory (kg) 25.9–38.85 25.9–38.85 * 15.54–194.25 * * calculated from the data in the references.

Table5 shows that HCD in this study has competitive advantage in comparison to both previous HCD and conventional silver electrorefining process. When the current density of HCD is increased to more than double of the conventional process, the cell voltage is shown to increase only for 30–40%. This increase of SEC is a significant improvement from the 1.27–1.28 kWh/kg from the previous HCD [4] which accounts more than 60% increase of SEC in comparison to the conventional process. With the + 2+ same [Ag ] level, the utilization of [Cu ] an [HNO3] to boost the conductivity was the cause of significantly lower SEC. From Table5, electrolyte volume and silver inventory were calculated for the same throughput. Comparison of HCD to the conventional process on the basis of 100 kg/day production rate shows that the median value of HCD silver inventory of approximately 32 kg which is far less than the median value of approximately 105 kg of the conventional process [7–12]. The required electrolyte volume of HCD was also shown to be less than half of the conventional process. Thus, the HCD application could turn to be highly beneficial as it is not only increasing the capacity due to the high throughput of the process but also has the possibility to reduce the process inventory and the equipment footprint. This suggests feasibility also from the economic point of view of the process.

5. Conclusions This study simulates the impact of HCD, 1000 A/m2 current density operation in silver electroreﬁning through the optimization of the electrolyte composition. With the objective to increase the applicability and competitiveness of HCD in comparison to the conventional process, an optimum range of electrolyte composition was developed through simulation. With the prudential practice of precious metal reﬁneries, the study has the limitation in the measurements of industrial electrolytes and anodes with only a validation of conductivity model was conducted. Therefore, this study was mainly focused on the simulation of the developed models from previous publications. Metals 2020, 10, 1596 13 of 14

From the simulation, it was found that by increasing the electrolyte conductivity through the introduction of [Cu2+] would lower the cell voltage from the IR drop significantly. While conductivity is in general an important property in electrorefining process, the higher current density in an HCD operation amplifies the significance of IR drop. Its contribution to cell voltage increased to more than 85% which is higher than the 55–82% range in the conventional process. The optimum concentrations for HCD electrolyte were suggested being within the range of 100–150 g/dm3 for [Ag+] and 50–75 g/dm3 for [Cu2+]. While increasing electrolyte silver and copper concentrations reduce the IR drop substantially, the effect of temperature is shown to have less impact. As a conclusion, HCD operation in the suggested electrolyte results in 0.60–1.01 kWh/kg of SEC with silver inventory in the electrolyte within the range of 25.9–38.85 kg for a 100 kg/day production rate. While this study elaborated the potential improvement of silver electrorefining process, study of the economic and environmental impact of HCD will add values to this process improvement.

Author Contributions: A.T.A. and J.A. performed the experiments, data analysis and wrote the paper. M.L. reviewed the paper and provided the research funding and resources. All authors have read and agreed to the published version of the manuscript. Funding: Symmet project (project number of 2117441). Conﬂicts of Interest: The authors declare no conﬂict of interest.

References

1. Habashi, F. Handbook of Extractive Metallurgy: Volume III Part 6: Precious Metals; Wiley-VCH Verlag-GmbH: Weinheim, Germany, 1998. 2. The Silver Institute. World Silver Surveys. Available online: https://www.silverinstitute.org/all-world-silver- surveys/ (accessed on 1 April 2020). 3. Mostert, P.J.; Radcliffe, P.H. Recent advances in gold refining technology at Rand Refinery. Dev. Miner. Process. 2005, 15, 653–670. [CrossRef] 4. Maliarik, M.; Johansson, K.Å.; Ögren, B.; Berg, G.; Johansson, C.D.; Lindh, R.; Ludvigsson, B. High current density silver electrorefining process: Technology, equipment, automation and Outotec’s Silver Refinery Plants. In Proceedings of the 7th International Symposium of Hydrometallurgy, Victoria, BC, Canada, 22–25 June 2014; Volume 2, pp. 91–100. 5. Jaskula, M.; Kammel, R. Untersuchungen zur Verbesserung des Platinmetalausbringens bei der industriellen Silberraffination. Metall 1997, 51, 393–400. 6. Pophanken, A.K.; Friedrich, B.; Koch, W. Challenges in the Electrolytic Refining of Silver–Influencing the Co-Deposition through Parameter Control. Rare Metal Technol. 2017, 2, 327–380. 7. Leigh, A.H. Precious metals refining practice. In Proceedings of the 2nd International Symposium on Hydrometallurgy, Chicago, IL, USA, 25 February–1 March 1973; pp. 95–110. 8. Pletcher, D.; Walsh, F.C. Industrial Electrochemistry, 2nd ed.; Springer: Dordrecht, The Netherlands, 2012; pp. 238–242. [CrossRef] 9. Cornelius, G. Die Raffination von Gold und Silber durch Elektrolyse. In Elektrolyse der Nichteisenmetalle: 11. Metallurgisches Seminar GDMB; Verlag Chemie: Weinheim, Germany, 1982; Volume 37, pp. 215–226. 10. Shibasaki, T.; Tsubogami, H.; Nozaki, M. Recent Improvements in Silver Electrolysis at Mitsubishi’s Osaka Refinery. In Proceedings of the Precious Metals’89, TMS conference, Las Vegas, NV, USA, 27 February–2 March 1989; pp. 419–430. 11. Mantell, C.L. Industrial Electrochemistry, 2nd ed.; McGgraw-Hill: New York, NY, USA, 1940; pp. 253–254. 12. Brumby, A.; Braumann, P.; Zimmermann, K.; van den Broeck, F.; Vandevelde, T.; Goia, D.; Renner, H.; Schlamp, G.; Zimmermann, K.; Weise, W.; et al. Silver, silver compounds, and silver alloys. In Ullmann’s Encyclopedia of Industrial Chemistry; Wiley-VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2012; pp. 42–43. 13. Aji, A.T.; Kalliomäki, T.; Wilson, B.P.; Aromaa, J.; Lundström, M. Modelling the effect of temperature and free acid, silver, copper and lead concentrations on silver electrorefining electrolyte conductivity. Hydrometallurgy 2016, 166, 154–159. [CrossRef] Metals 2020, 10, 1596 14 of 14

14. Aji, A.T.; Aromaa, J.; Wilson, B.P.; Mohanty, U.S.; Lundström, M. Kinetic Study and Modelling of Silver Dissolution in Synthetic Industrial Silver Electrolyte as a Function of Electrolyte Composition and Temperature. Corros. Sci. 2018, 138, 163–169. [CrossRef] 15. Aji, A.T.; Halli, P.; Guimont, A.; Wilson, B.P.; Aromaa, J.; Lundström, M. Modelling of silver anode dissolution and the effect of gold as impurity under simulated industrial silver electrorefining conditions. Hydrometallurgy 2019, 189, 105105. [CrossRef] 16. Aji, A.T.; Aromaa, J.; Lundström, M. Specific Energy Consumption Modelling of Silver Electrorefining. In Proceedings of the European Metallurgical Conference (EMC), Dusseldorf, Germany, 23–26 June 2019; Volume 2, pp. 807–820. 17. Gordon, N.L.; Davenport, W.G. Electrical conductivities, densities and viscosities of silver refining electrolyte. Can. Min. Q. 1981, 20, 369–372. [CrossRef] 18. Maurell-Lopez, A.K. Verhalten von Kupfer und Palladium bei der Raffinationselektrolyse von Recycling-Silber; Shaker Verlag GmbH: Düren, Germany, 2017; ISBN 978-3-8440-5987-8. 19. Johnson, O. Refining processes. In Silver: Economics, Metallurgy, and Use; Butts, A., Coxe, C.D., Eds.; Van Nostrand Company, Inc.: Princeton, NJ, USA, 1967; Chapter 5. 20. Hunter, W. Electrolytic Refining. US Patent no. 3,975,244, 17 August 1976.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional aﬃliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).